From 7fd9bc741a2c853ed56928f22376d6259c799ccf Mon Sep 17 00:00:00 2001
From: sebffischer <sebf.fischer@gmail.com>
Date: Fri, 7 Feb 2025 09:22:44 +0000
Subject: [PATCH] =?UTF-8?q?Deploying=20to=20gh-pages=20from=20@=20mlr-org/?=
 =?UTF-8?q?mlr3torch@4148c9e97bd7066497066e5fa3edf32b9eb3af96=20?=
 =?UTF-8?q?=F0=9F=9A=80?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 dev/LICENSE.html                              |   2 +-
 dev/articles/callback_list.html               |   2 +-
 dev/articles/callbacks.html                   |   4 +-
 dev/articles/get_started.html                 |   2 +-
 dev/articles/index.html                       |   2 +-
 dev/articles/internals_pipeop_torch.html      |  61 ++--
 .../figure-html/unnamed-chunk-37-1.png        | Bin 123412 -> 123444 bytes
 .../figure-html/unnamed-chunk-42-1.png        | Bin 123448 -> 123444 bytes
 .../figure-html/unnamed-chunk-46-1.png        | Bin 123680 -> 123724 bytes
 .../figure-html/unnamed-chunk-48-1.png        | Bin 123270 -> 123160 bytes
 .../figure-html/unnamed-chunk-50-1.png        | Bin 123276 -> 123374 bytes
 dev/articles/layer_list.html                  |   2 +-
 dev/articles/lazy_tensor.html                 |  10 +-
 dev/articles/loss_list.html                   |   2 +-
 dev/articles/optimizer_list.html              |   2 +-
 dev/articles/pipeop_torch.html                |   2 +-
 dev/articles/preprocessing_list.html          |   2 +-
 dev/articles/tabular_learner_list.html        |   2 +-
 dev/articles/task_list.html                   |   2 +-
 dev/articles/vision_learner_list.html         |   2 +-
 dev/authors.html                              |   6 +-
 dev/index.html                                |   4 +-
 dev/news/index.html                           |  36 ++-
 dev/pkgdown.yml                               |   2 +-
 dev/reference/DataDescriptor.html             |   6 +-
 dev/reference/ModelDescriptor.html            |   2 +-
 dev/reference/PipeOpPreprocTorchTrafoNop.html |   2 +-
 .../PipeOpPreprocTorchTrafoReshape.html       |   2 +-
 dev/reference/Select.html                     |   2 +-
 dev/reference/TorchCallback.html              |   4 +-
 dev/reference/TorchDescriptor.html            |   2 +-
 dev/reference/TorchIngressToken.html          |   2 +-
 dev/reference/TorchLoss.html                  |   2 +-
 dev/reference/TorchOptimizer.html             |   2 +-
 dev/reference/as_data_descriptor.html         |   6 +-
 dev/reference/as_lazy_tensor.html             |   2 +-
 dev/reference/as_lr_scheduler.html            |   2 +-
 dev/reference/as_torch_callback.html          |   2 +-
 dev/reference/as_torch_callbacks.html         |   2 +-
 dev/reference/as_torch_loss.html              |   2 +-
 dev/reference/as_torch_optimizer.html         |   2 +-
 dev/reference/assert_lazy_tensor.html         |   2 +-
 dev/reference/auto_device.html                |   2 +-
 dev/reference/batchgetter_categ.html          |   2 +-
 dev/reference/batchgetter_num.html            |   2 +-
 dev/reference/callback_set.html               |   2 +-
 dev/reference/equals-.lazy_tensor.html        |   2 +-
 dev/reference/index.html                      |   2 +-
 dev/reference/is_lazy_tensor.html             |   2 +-
 dev/reference/lazy_tensor.html                |   2 +-
 dev/reference/materialize.html                | 264 +++++++++---------
 dev/reference/materialize_internal.html       |   2 +-
 dev/reference/mlr3torch-package.html          |   2 +-
 dev/reference/mlr3torch_callbacks.html        |   2 +-
 dev/reference/mlr3torch_losses.html           |   2 +-
 dev/reference/mlr3torch_optimizers.html       |   2 +-
 dev/reference/mlr_backends_lazy.html          |   2 +-
 .../mlr_callback_set.checkpoint.html          |   2 +-
 dev/reference/mlr_callback_set.history.html   |   2 +-
 dev/reference/mlr_callback_set.html           |   2 +-
 .../mlr_callback_set.lr_scheduler.html        |   2 +-
 dev/reference/mlr_callback_set.progress.html  |  14 +-
 dev/reference/mlr_callback_set.tb.html        |   2 +-
 dev/reference/mlr_callback_set.unfreeze.html  |   2 +-
 dev/reference/mlr_context_torch.html          |   2 +-
 dev/reference/mlr_learners.mlp.html           |   2 +-
 dev/reference/mlr_learners.tab_resnet.html    |   2 +-
 .../mlr_learners.torch_featureless.html       |   2 +-
 dev/reference/mlr_learners.torchvision.html   |   2 +-
 dev/reference/mlr_learners_torch.html         |   2 +-
 dev/reference/mlr_learners_torch_image.html   |   2 +-
 dev/reference/mlr_learners_torch_model.html   |   2 +-
 dev/reference/mlr_pipeops_module.html         |   2 +-
 .../mlr_pipeops_nn_adaptive_avg_pool1d.html   |   2 +-
 .../mlr_pipeops_nn_adaptive_avg_pool2d.html   |   2 +-
 .../mlr_pipeops_nn_adaptive_avg_pool3d.html   |   2 +-
 dev/reference/mlr_pipeops_nn_avg_pool1d.html  |   2 +-
 dev/reference/mlr_pipeops_nn_avg_pool2d.html  |   2 +-
 dev/reference/mlr_pipeops_nn_avg_pool3d.html  |   2 +-
 .../mlr_pipeops_nn_batch_norm1d.html          |   2 +-
 .../mlr_pipeops_nn_batch_norm2d.html          |   2 +-
 .../mlr_pipeops_nn_batch_norm3d.html          |   2 +-
 dev/reference/mlr_pipeops_nn_block.html       |   2 +-
 dev/reference/mlr_pipeops_nn_celu.html        |   2 +-
 dev/reference/mlr_pipeops_nn_conv1d.html      |   2 +-
 dev/reference/mlr_pipeops_nn_conv2d.html      |   2 +-
 dev/reference/mlr_pipeops_nn_conv3d.html      |   2 +-
 .../mlr_pipeops_nn_conv_transpose1d.html      |   2 +-
 .../mlr_pipeops_nn_conv_transpose2d.html      |   2 +-
 .../mlr_pipeops_nn_conv_transpose3d.html      |   2 +-
 dev/reference/mlr_pipeops_nn_dropout.html     |   2 +-
 dev/reference/mlr_pipeops_nn_elu.html         |   2 +-
 dev/reference/mlr_pipeops_nn_flatten.html     |   2 +-
 dev/reference/mlr_pipeops_nn_gelu.html        |   2 +-
 dev/reference/mlr_pipeops_nn_glu.html         |   2 +-
 dev/reference/mlr_pipeops_nn_hardshrink.html  |   2 +-
 dev/reference/mlr_pipeops_nn_hardsigmoid.html |   2 +-
 dev/reference/mlr_pipeops_nn_hardtanh.html    |   2 +-
 dev/reference/mlr_pipeops_nn_head.html        |   2 +-
 dev/reference/mlr_pipeops_nn_layer_norm.html  |   2 +-
 dev/reference/mlr_pipeops_nn_leaky_relu.html  |   2 +-
 dev/reference/mlr_pipeops_nn_linear.html      |   2 +-
 dev/reference/mlr_pipeops_nn_log_sigmoid.html |   2 +-
 dev/reference/mlr_pipeops_nn_max_pool1d.html  |   2 +-
 dev/reference/mlr_pipeops_nn_max_pool2d.html  |   2 +-
 dev/reference/mlr_pipeops_nn_max_pool3d.html  |   2 +-
 dev/reference/mlr_pipeops_nn_merge.html       |   2 +-
 dev/reference/mlr_pipeops_nn_merge_cat.html   |   2 +-
 dev/reference/mlr_pipeops_nn_merge_prod.html  |   2 +-
 dev/reference/mlr_pipeops_nn_merge_sum.html   |   2 +-
 dev/reference/mlr_pipeops_nn_prelu.html       |   2 +-
 dev/reference/mlr_pipeops_nn_relu.html        |   2 +-
 dev/reference/mlr_pipeops_nn_relu6.html       |   2 +-
 dev/reference/mlr_pipeops_nn_reshape.html     |   2 +-
 dev/reference/mlr_pipeops_nn_rrelu.html       |   2 +-
 dev/reference/mlr_pipeops_nn_selu.html        |   2 +-
 dev/reference/mlr_pipeops_nn_sigmoid.html     |   2 +-
 dev/reference/mlr_pipeops_nn_softmax.html     |   2 +-
 dev/reference/mlr_pipeops_nn_softplus.html    |   2 +-
 dev/reference/mlr_pipeops_nn_softshrink.html  |   2 +-
 dev/reference/mlr_pipeops_nn_softsign.html    |   2 +-
 dev/reference/mlr_pipeops_nn_squeeze.html     |   2 +-
 dev/reference/mlr_pipeops_nn_tanh.html        |   2 +-
 dev/reference/mlr_pipeops_nn_tanhshrink.html  |   2 +-
 dev/reference/mlr_pipeops_nn_threshold.html   |   2 +-
 dev/reference/mlr_pipeops_nn_unsqueeze.html   |   2 +-
 dev/reference/mlr_pipeops_preproc_torch.html  |   2 +-
 dev/reference/mlr_pipeops_torch.html          |   2 +-
 .../mlr_pipeops_torch_callbacks.html          |   2 +-
 dev/reference/mlr_pipeops_torch_ingress.html  |   2 +-
 .../mlr_pipeops_torch_ingress_categ.html      |   2 +-
 .../mlr_pipeops_torch_ingress_ltnsr.html      |  48 ++--
 .../mlr_pipeops_torch_ingress_num.html        |   2 +-
 dev/reference/mlr_pipeops_torch_loss.html     |   2 +-
 dev/reference/mlr_pipeops_torch_model.html    |   2 +-
 .../mlr_pipeops_torch_model_classif.html      |  28 +-
 .../mlr_pipeops_torch_model_regr.html         |  14 +-
 .../mlr_pipeops_torch_optimizer.html          |   2 +-
 .../mlr_pipeopsaugment_center_crop.html       |   2 +-
 .../mlr_pipeopsaugment_color_jitter.html      |   2 +-
 dev/reference/mlr_pipeopsaugment_crop.html    |   2 +-
 dev/reference/mlr_pipeopsaugment_hflip.html   |   2 +-
 .../mlr_pipeopsaugment_random_affine.html     |   2 +-
 .../mlr_pipeopsaugment_random_choice.html     |   2 +-
 .../mlr_pipeopsaugment_random_crop.html       |   2 +-
 ...pipeopsaugment_random_horizontal_flip.html |   2 +-
 .../mlr_pipeopsaugment_random_order.html      |   2 +-
 ...lr_pipeopsaugment_random_resized_crop.html |   2 +-
 ...r_pipeopsaugment_random_vertical_flip.html |   2 +-
 .../mlr_pipeopsaugment_resized_crop.html      |   2 +-
 dev/reference/mlr_pipeopsaugment_rotate.html  |   2 +-
 dev/reference/mlr_pipeopsaugment_vflip.html   |   2 +-
 .../mlr_pipeopstrafo_adjust_brightness.html   |   2 +-
 .../mlr_pipeopstrafo_adjust_gamma.html        |   2 +-
 .../mlr_pipeopstrafo_adjust_hue.html          |   2 +-
 .../mlr_pipeopstrafo_adjust_saturation.html   |   2 +-
 dev/reference/mlr_pipeopstrafo_grayscale.html |   2 +-
 dev/reference/mlr_pipeopstrafo_normalize.html |   2 +-
 dev/reference/mlr_pipeopstrafo_pad.html       |   2 +-
 dev/reference/mlr_pipeopstrafo_resize.html    |   2 +-
 .../mlr_pipeopstrafo_rgb_to_grayscale.html    |   2 +-
 dev/reference/mlr_tasks_cifar.html            |   2 +-
 dev/reference/mlr_tasks_lazy_iris.html        |   2 +-
 dev/reference/mlr_tasks_melanoma.html         |   2 +-
 dev/reference/mlr_tasks_mnist.html            |   2 +-
 dev/reference/mlr_tasks_tiny_imagenet.html    |   2 +-
 .../model_descriptor_to_learner.html          |   2 +-
 dev/reference/model_descriptor_to_module.html |   2 +-
 dev/reference/model_descriptor_union.html     |   2 +-
 dev/reference/nn.html                         |   2 +-
 dev/reference/nn_graph.html                   |  34 +--
 dev/reference/nn_merge_cat.html               |   2 +-
 dev/reference/nn_merge_prod.html              |   2 +-
 dev/reference/nn_merge_sum.html               |   2 +-
 dev/reference/nn_reshape.html                 |   2 +-
 dev/reference/nn_squeeze.html                 |   2 +-
 dev/reference/nn_unsqueeze.html               |   2 +-
 dev/reference/pipeop_preproc_torch.html       |   2 +-
 dev/reference/replace_head.html               |   2 +-
 dev/reference/t_clbk.html                     |   2 +-
 dev/reference/t_loss.html                     |   2 +-
 dev/reference/t_opt.html                      |   2 +-
 dev/reference/task_dataset.html               |   2 +-
 dev/reference/torch_callback.html             |   2 +-
 dev/search.json                               |   2 +-
 185 files changed, 452 insertions(+), 417 deletions(-)

diff --git a/dev/LICENSE.html b/dev/LICENSE.html
index 3654b936..9053510f 100644
--- a/dev/LICENSE.html
+++ b/dev/LICENSE.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/articles/callback_list.html b/dev/articles/callback_list.html
index 20cddada..cbdd725b 100644
--- a/dev/articles/callback_list.html
+++ b/dev/articles/callback_list.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/articles/callbacks.html b/dev/articles/callbacks.html
index 4349e570..4f4554f5 100644
--- a/dev/articles/callbacks.html
+++ b/dev/articles/callbacks.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -240,7 +240,7 @@ <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: 0x562772bf68b8&gt;</span></span>
+<span><span class="co">##   Parent env: &lt;environment: 0x560291776688&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 e43b8b2a..45bf7ee4 100644
--- a/dev/articles/get_started.html
+++ b/dev/articles/get_started.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/articles/index.html b/dev/articles/index.html
index be7d031f..6e3c98fc 100644
--- a/dev/articles/index.html
+++ b/dev/articles/index.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/articles/internals_pipeop_torch.html b/dev/articles/internals_pipeop_torch.html
index 2184f29a..0d600329 100644
--- a/dev/articles/internals_pipeop_torch.html
+++ b/dev/articles/internals_pipeop_torch.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -119,8 +119,8 @@ <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://torch.mlverse.org/docs/reference/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;  0.5186  0.0007 -0.5029</span></span>
-<span><span class="co">#&gt;  0.3135  1.2928  1.4883</span></span>
+<span><span class="co">#&gt;  0.2528 -0.3532 -0.2106</span></span>
+<span><span class="co">#&gt;  0.4273 -0.6144  0.1054</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
@@ -132,8 +132,8 @@ <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; -0.0060  0.5752 -0.1935 -0.0095</span></span>
-<span><span class="co">#&gt; -1.0660  1.3540 -1.2291 -1.1676</span></span>
+<span><span class="co">#&gt;  0.4542 -0.2453  0.4121  0.1119</span></span>
+<span><span class="co">#&gt;  0.2307 -0.3919  0.6421 -0.0936</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>
@@ -149,8 +149,8 @@ <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;  0.5381  0.2576  0.2043</span></span>
-<span><span class="co">#&gt;  0.5026  0.2841  0.2133</span></span>
+<span><span class="co">#&gt;  0.2538  0.3499  0.3963</span></span>
+<span><span class="co">#&gt;  0.2639  0.3569  0.3792</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>
@@ -185,8 +185,8 @@ <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">$</span><span class="fu">train</span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/list.html" class="external-link">list</a></span><span class="op">(</span><span class="va">input</span><span class="op">)</span><span class="op">)</span><span class="op">[[</span><span class="fl">1</span><span class="op">]</span><span class="op">]</span></span>
 <span><span class="va">output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt; -0.0060  0.5752 -0.1935 -0.0095</span></span>
-<span><span class="co">#&gt; -1.0660  1.3540 -1.2291 -1.1676</span></span>
+<span><span class="co">#&gt;  0.4542 -0.2453  0.4121  0.1119</span></span>
+<span><span class="co">#&gt;  0.2307 -0.3919  0.6421 -0.0936</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,4} ][ grad_fn = &lt;AddmmBackward0&gt; ]</span></span></code></pre></div>
 <p>Note we only use the <code>$train()</code>, since torch modules do
 not have anything that maps to the <code>state</code> (it is filled by
@@ -211,8 +211,8 @@ <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;  0.5381  0.2576  0.2043</span></span>
-<span><span class="co">#&gt;  0.5026  0.2841  0.2133</span></span>
+<span><span class="co">#&gt;  0.2538  0.3499  0.3963</span></span>
+<span><span class="co">#&gt;  0.2639  0.3569  0.3792</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
@@ -260,8 +260,8 @@ <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;  0.5381  0.2576  0.2043</span></span>
-<span><span class="co">#&gt;  0.5026  0.2841  0.2133</span></span>
+<span><span class="co">#&gt;  0.2538  0.3499  0.3963</span></span>
+<span><span class="co">#&gt;  0.2639  0.3569  0.3792</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">
@@ -378,8 +378,8 @@ <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; -0.2818  0.9781 -0.2101 -0.0965</span></span>
-<span><span class="co">#&gt; -1.0439 -0.2501  0.6842 -0.5891</span></span>
+<span><span class="co">#&gt;  0.2274  0.2261 -0.4887 -0.1051</span></span>
+<span><span class="co">#&gt;  0.1440  0.2850 -0.3803 -0.2081</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">
@@ -443,9 +443,9 @@ <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; -1.7275 -0.8973  2.2747  0.1551</span></span>
-<span><span class="co">#&gt; -1.5510 -0.6123  2.0240  0.3031</span></span>
-<span><span class="co">#&gt; -1.6239 -0.7723  2.0707  0.1194</span></span>
+<span><span class="co">#&gt; -1.6916  1.1106  3.0460  3.0379</span></span>
+<span><span class="co">#&gt; -1.4601  0.9800  2.7398  2.9830</span></span>
+<span><span class="co">#&gt; -1.5414  1.0404  2.7707  2.8034</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">
@@ -471,8 +471,8 @@ <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;  0.4332  0.2392  0.3276</span></span>
-<span><span class="co">#&gt;  0.4445  0.2243  0.3311</span></span>
+<span><span class="co">#&gt;  0.1791  0.4450  0.3759</span></span>
+<span><span class="co">#&gt;  0.1817  0.4483  0.3700</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">
@@ -686,15 +686,14 @@ <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; -3.3103</span></span>
-<span><span class="co">#&gt;  -3.3103</span></span>
+<span><span class="co">#&gt; -0.4005</span></span>
+<span><span class="co">#&gt; -0.3776</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; $nn_softmax.output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt;  0.3040  0.3772  0.3189</span></span>
-<span><span class="co">#&gt;  0.3040  0.3772  0.3189</span></span>
+<span><span class="co">#&gt;  0.0855  0.3679  0.5466</span></span>
+<span><span class="co">#&gt;  0.0941  0.3845  0.5214</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,3} ][ grad_fn = &lt;SoftmaxBackward0&gt; ]</span></span></code></pre></div>
 <p>The first linear layer that takes “Sepal” input
 (<code>"linear1"</code>) creates a 2x4 tensor (batch size 2, 4 units),
@@ -703,14 +702,14 @@ <h3 id="building-more-interesting-nns">Building more interesting NNs<a class="an
 <code class="sourceCode R"><span><span class="va">iris_module</span><span class="op">$</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; $output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt; -3.8395 -3.5382 -1.5806 -0.5244</span></span>
-<span><span class="co">#&gt; -3.6195 -3.1991 -1.7572 -0.6355</span></span>
+<span><span class="co">#&gt;  2.7784  1.0401 -2.5610  1.4984</span></span>
+<span><span class="co">#&gt;  2.5208  1.0563 -2.5473  1.5691</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,4} ][ grad_fn = &lt;AddmmBackward0&gt; ]</span></span>
 <span><span class="va">iris_module</span><span class="op">$</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; $output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt; -1.0179 -0.4841  0.5960  0.6553 -0.2139</span></span>
-<span><span class="co">#&gt; -1.0179 -0.4841  0.5960  0.6553 -0.2139</span></span>
+<span><span class="co">#&gt; -0.1651  0.7799  0.3149 -0.0282  0.6354</span></span>
+<span><span class="co">#&gt; -0.1651  0.7799  0.3149 -0.0282  0.6354</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,5} ][ grad_fn = &lt;AddmmBackward0&gt; ]</span></span></code></pre></div>
 <p>We observe that the <code>po("nn_merge_cat")</code> concatenates
 these, as expected:</p>
@@ -718,8 +717,8 @@ <h3 id="building-more-interesting-nns">Building more interesting NNs<a class="an
 <code class="sourceCode R"><span><span class="va">iris_module</span><span class="op">$</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; $output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt; -3.8395 -3.5382 -1.5806 -0.5244 -1.0179 -0.4841  0.5960  0.6553 -0.2139</span></span>
-<span><span class="co">#&gt; -3.6195 -3.1991 -1.7572 -0.6355 -1.0179 -0.4841  0.5960  0.6553 -0.2139</span></span>
+<span><span class="co">#&gt;  2.7784  1.0401 -2.5610  1.4984 -0.1651  0.7799  0.3149 -0.0282  0.6354</span></span>
+<span><span class="co">#&gt;  2.5208  1.0563 -2.5473  1.5691 -0.1651  0.7799  0.3149 -0.0282  0.6354</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,9} ][ grad_fn = &lt;CatBackward0&gt; ]</span></span></code></pre></div>
 </li>
 </ol>
diff --git a/dev/articles/internals_pipeop_torch_files/figure-html/unnamed-chunk-37-1.png b/dev/articles/internals_pipeop_torch_files/figure-html/unnamed-chunk-37-1.png
index cace4a61fb56655e9efd3c398d1802160c3a60ae..2dbc7983116cf987a9c48bc32a3a855d22e9db55 100644
GIT binary patch
delta 107550
zcmX_IWmr~S*F;ee5m35Aq@@L<Un!-fyOHkhBOMYd-AZ?NOLup7cQ@Y#eZF6KUEX@{
zv(H|$X3fkdtQo$f8NQ&G6CxmxE9JJh=49TewMP|s_T<xg?V0+M=Ev*B-=4lD_V9a+
zi7qH;LhPYPIhZ0!QS$vWA0h(To0PNJ(8_w=$CpD-KIx&NF5c#xj5^ghpYmMIR@ShX
z^~TB2tv1IfFcw-gn6HZ|BPFPx@=|xsoh{2&pUM7cr&${>Smc)=oREM%s7q9C^_m5(
zbDN+hQ`Fun3|-Z)j#htvMPwKG!`^<&HEZWg<r#_B%eyy7>fi_8vxCI7u|gsQ#xXyh
z^7yow#C~?k<H{YFbNg)#wbYH}(XwX)`sD+LjjtO~TqWl>GYv$)Ju0v<@vt5uj8dwW
zR#p~}5P5SDc`Efily4u}V}4)1h2~T1_$0N&YddV~-LYX%co4to>>w{X7Dj=|g6yf;
z?IS^{e(ad}S)*t2z2c1T3R8$_{KMKq-w62@?D}48E1t3R)m|WR!*R<_rOsKai{BI$
z6p?Opz6v;GVqh9FC9^XWCb7gi(=^cKW9FTJ6u|Rjh^G9Z@p=jIJtMh2syfAMOAzm>
zs8~a+spbwAk-Ol}G3yYo{&Nz##lXaxs#Z?I^7+dOdo@dv<&m9$SG-=|;zD!W+*8Vm
za(uCB_|>KPkw>?sw(eymQSG7i)=o!Lx)dJ^9L+>kUAu5c)UJYay{j`6?ewVkB=LGz
z5GfeS3er#Z+23tJCokYyS6AA|HAr>0T=EDtTJ(#F$jJ{;QFyO*Yw4>n4d21p>@xef
zs<^$exl<^?IVh8Gq=SE(T*2Vm>)NWRt$eG52wqFlsq9T`Hh4SW&nIHBbe;y2MQ9t0
zYtK0M#YfO6pHqz11eB%t{RpX(rYM7r+P6_6rC+_4kcEs`)Cg|^KICsV|JG`=>QRF!
zZT;RFPxR%uTK~X^4b#*Rp%Ck~d3ZWN3#Ha2zbpT<p)ez&K2JM%2n&;8f5yFp<dp^m
zYLi{Q36Yfq$voYLE)|A%g}OU~j@Ok3SI(-oJ9uaHYy$=_DhiBAl!O(s@e?BnMcYTu
zzShstkU-!>4HFE`V?3<<&XGDiAYJ|LaZu~mU#0n|SXJqVe*zjecvn$vok)@C61euq
z@mLfLFy5KX?&ia`R*Byt&$-ddjgQ~3@#`R7N?gRl<$iqpYrvqOAPdje$W-y*)tV$>
z!)Q~;pP@4?y(Rg~3b<QR?EOQ`VO-b<*K1U$@W<ueM`vEk6O+yi|9mOc?>%+-LQ>!3
zRX4_yuEkDWZ(yEIbi8nx$H8SYSMUsfeqRDu|E>42!RlPNz8>8N73IbRdf!)7=8OSl
z$H9z(7t#3Y2T%0NYeNeCaf-2S+$&Sfoa5xn`_{w+q|kBgot`8nNGedtr$W+0R~s4(
zrB);qH1VEOP~LB98hU10e{o)Vfn3IvxiBPbToqZW*;FN>$v_w1!9i5Fi`87*tfJkR
z^F{cQ4JN<9b5Rr5deV+D&t2S5LhA%mA{%ltCTya5Ixc3N1MUpj`y(Mo_OEzJ%y79*
z+0(jj2{A0RPu{UY^;0ojkGk5s{H~n8Cw(V$cK>dAq{N&3{^>{-3=A=hu)qgd1ci_r
zuUt=>J=gL?As$ns+|#@KjEt`Z#ciK6SxWG(+$Gpwi02w`TOq?3NA_g*9@<!GPD!FK
zugni^;pD`{ZiXeV{2Bw#NN#Z_Bps@qUD;SAp3JDZ+8AU5?XU&M4O5o6_D93ci4$K=
z+lhQTyu(#cmj1NeGV|mu6xeg592)&nVk*zm@v4ge8#j7*k>LK37VVb5v)sQ8P{4^F
zbP`j4OB%bW9#b!@1zX#U6vJR+$Li?n@qe&mn*90tmx-}2PZK%w#mduhc6qbvug+R~
zt1FZNn9!KsO@WO;igf~cJkd(?rs11b_4JgKXRxrzMma@B9*@gl=4W;+<E^o=qX!C_
zJXI&_y+S{hF*<BpkI2os{%IbL!$BN9SZm!=Dn3?wEqN=k+%eacr}X}-h~Ni_=akQ3
zR^`jM&X-jJU5+t^jl@kO=)jWwIKRW?KNKtU@@{b`i<yX)xAS8D<_Y0=d*%F13VG}I
zNWyQarQE%{L55?ll-zJU0-mk3M%B|+y-#<ZHjI=fsyhaDn^^7}NgkdM=cztxEEf;i
zq|EO`7hIT3mBEftT<kCKRIi+Nxw1CAL(Nm;kP{!7(U3J_;cu1v-B-|oKT{n%|20<?
zLgE${GI|`4q`9{)#GlA&v(5685SA#5Sta;klTRN|66~?SQ(0<^+xW|+xlCRdGBUE|
zwKXj*t&6dRFSc^ZR$MTvWk|=pU-Y&fjd}W2x@C^y)?8F7p<}G-><r?Ei`1*FcAId+
ze3JCf4<6iAA=kq^COUa@^N|}u_?k5G_zRNe{hOTD`2TMsGkkZ}HMUliIIi%K^im!<
zH)(`3G7mlVa&JvuLP`j?8Z4>#X3Oouhx+~qNbRX2A+4U*A;lrieza;tvMOdSD6*uy
zx`4__Z1DDp%@TM@)$~$R-v}K|m7AlXo>M5{WQIkA{eq<3TN_@&uff^uutoSLx?+){
zG1snAQB!LI0K=cFcMm(q3kyU4P5B%bmj7+;j=_-QOB7x<6n*=dvxM5Y366@#-(v}r
z?ekRFxo|Hw<3$zpFh{{5{T};DrHnU+FlbigOvnBR2jSuQ_LlPvz3<=qZsEG$=T%99
zRw%HIqwpX-q!QQZXHKB(u+M5rYO^)f^0Rt3X5YxxtHY5Rip^`uj&IKo&-X{9xHywY
z;LC0VoJ1a{lC@yRC?~Np{7$Lx^NKPyQ7--N?`m?s(S^g&cXISoNo?D+r!^eWuinyo
zaglErt`&S-N?BS=$mlh^{dIR3<7{^#2NdX06K7O2boDVC*%ihQh4Ys}0>C}#*OI@b
zGYuz;WLRX=@WMp|ll4p~d;4CWv>Mi0%{Y+<#gyg-4?Yy1QGCu{^sev?-kWQPv?fX}
zAVxuSrgyoj7lXZfrO57>EN$E$fq4&3c>1i?8x5$Z=4u@qI@M7p6Of*ZvTfeZVVAI9
zZEIi$K}Uzk$5iQ|l&&|p_HO9<i3tGyz9xMJR6%u}kd8=Q!e!iz_3oNVKX`97G`x+b
zHl`dv(S)SknzEvjbziEIxr-A|kaO3ED8=_4B7(wNbp8q<FD<oM#)9v<H*`21ofKht
z)E)89e0q!z8Hd>+&o=J90Kxpwmu}yuOANYqY^J`QIqfT>n<9YPV_;&+T&2!VlL3u^
zhp2r-C%fRxC|@RcaPcuHb>#o~CXi+obKSpMKV?!Hhrh4($I=zseh1zX)!X*up7zSX
z8{#;5PxZFRR2=G3IM7~SI_KM!5F8OM@dh+hLsLt-DdhX7)~^1~hRj#AZM(!IwI}&+
zmvvKhmPUUL>?Nd*4e>55u$AOgBL0%r(%$&A)EmO2d&vVQF%?``Ng$=r$GD>fl}uLe
zZDoxZIqv*mRYl|s`jN?UqjBl^dkCc;?4{V-_}k0MI`=~MU0V2|qmJ-E$Gyj8gtyD9
zkna9wbD6ChhK(&V6Vq?QE(t)_*pJ#&<E_HbS5b>rxM5mv%;jw((nH-y?|gq0Qfc|t
zk#eq7NwD*sxeBTa;S)FoHw?5C$IzgIbM2M!UoWqKHxp1@?y3xV)PFTGXTa%1hzq;K
z*@_)K(SPKRQG#=VEa-jVat=)z6fw)}308(5w(7)FQMWgBRCAK*5j2g~j;QHyUHdRv
zNid|1=DlfF)mi91un)Sh+I^#)+{jpHqrDtQ%rzBQwMCf2npgKC(H1>Y&y|wyt!C?T
zaClw*<>JU0^fWwx=A(iY?S5p|Q<~?#&^3`GPGOXaBJ;kxzEMhRRsm#d@nyJqtUm(X
za()V)imEeP+GVhGDQ=bvzErs>1&cT5H}&7w%5Av+(EO)iY3ifY%dG5&Ip6Ibe_DZ+
zqn-?Q@f9MV_#T7abV;-&q+@h7ng1yuq`A8_{+8aXxqmoourMMxXya~BPu2)l0)JY}
zM0eb@jl5ZX3EEFI{+u~{3|W%O_lEP@^!AT?xXFy_n5_vLS-p*wca`24za(a*&e{GS
z(|mgXMWxkpn@CGI<_cQDHREf^TzJ`E(xH)|N}l2JvtH?QX_XM0bh1R@+PB~6x)ci3
zbaXcK&h_}gfcjBrsCY$|Yal5+0H)>MC(nBt482V9QLD4?U~aS1TIhdyr^RtStd^QI
z#Gk06soNQBE7~rSuPT#Go_ldi05chK^UEaGi`ohhKfT(74^rrLF=x%DZM2SiROo`g
z4MQj_4tLd890O^$HyGR=!PL{={pvh8D=rmXd(>akhmJ1T^&?!e@GiDcx7l0%y}DG%
zH1zDkllExJnsy(n+hoSD-|0!^#0DVrC$(>H%3cHoMqg|Q;a-%GELg!nHB&MBUsDZ4
z1E}vQWY>X4TEW#f67<zn=G8)+7tZS|&Z?#N;40AblYjItHjN*+Zap#>fF{+-zf=$~
z<FVn|C$z6L@md_V+g%1-Yp_7_iY$dvkpOnuhC8Fl)yOyE(nr4IcoAGj7)P4bm+!9}
zxqs1t@9TF=5F#L2T)vE6P7_3PWcTvnF}3Nnt^XuZ@hAM-2Stu|Nn6eNhT~dXkGjZT
z**c*~rL$>mE%+AO+l!e__hVxH48;}RZs7)R<<S@4@e?;Y7Vn7;fkbGg^WX*#`)+w#
zxh|WD)vnj2;`-T~+d2LtPH1I3Ywejj8$mfHY@yX*YyWV-b&h~EspL=eSI+`GGNZ{P
z`xI5PA+x`a8u0Ef{uqr{{&IzRXTZ&u&xNEMvV6Ee^d7IITpmxZnvM2p=jl+b?fW>o
zw*<Ia0AB*3if>+tC@Ikzk&tRgl~eFy5?@YSO`)A(+7TCZcnsWa69VTsK{XrQ$^={e
z0Q=a@;wb0ZfYXx<*$URZwu+EEE5kVjRi2`_<EW#Qwz#{Gfg-7`m4=enKHb0EA*HzJ
za74^gbuclJb24Ly8edIT#9N7p^*Gwz(!WUDN8N%HTt0o8H%pQ*f6LZ8`X^kSpYY4}
z5_4gZZH~9dL*W@E{P$prps^3*^5@{ku7|g^j$nNqZeHJ{6xByy%Y)He8s*zl1vsGF
ztoG*6>Qg10xxjd|QEc1gYO$zaP(SC`4KdjQ91Lq7#LEdw%6#q9+YBaSCJ3#u!w->+
zmzl8Col%d~>^<wIS7A#%ZUr^QXkOiNiqMZ3Y#Dvrn)35=TdmV1TEJ>8E*cS}Lv=d(
zdKz1MaU$Vn-x*EfhAZ2pqTQdp8}h$Q+YdK|@;Up&26pU}r%0*-9jUbnOcC=?R`R>}
zFZ$s<=DK?Kq~Qz-i$Pb}RaUlAE$z`Uk!O{ZkguAp+=XYK_;P0G605JjGng}&opk49
zH(`4=5gny8KbHxaMn3RfDVl$DNy-y&GkI2I-7hIg2BIgEgTIKYw=Gz<ly5UHaxW0A
zq>p=Tw-Vx1Ydh7fYcA?BbR!c#N|mGt#gnr`ufNv>zAqsYjGC;_-Wo*(5DF%3+4muI
zyhD<KLG#xguy0yE{)mqbt_`K3+Z8t$`a+B;ZztDlPp>%f9%keiXXVWuss0x#Wm*L~
zI7_m3Q6<aoOH#?kIVu)Mh7%)9owYO#QtVCdd&5gU+?t1<EWH!Cuc8fwb(=My#?jjF
zUdKV5-beGxqXjY4LqQ*xFfBCEHU`#?MN=rD8JQvZ_)aym$s$sVqdE-0LPfqIR@lrN
z)9tDYs0*oj8zKev9PD0V=YFgUiY}#rF=Sh_JUh(PHGEEiY@$7ba#ZQYp3=x@RO(|l
z<9oOscJVGt2fiGP@{e!p%+(;OfCzAOGlq@uRk>&|WWFbMow9tJD7zG88bqbR*D=!T
z_{u2Bc<Vn=#FqEK*c1NS^rlH0f^IMdS6J#5t0T>3FvmVJn6mw38=e+QvXoZfb^+h4
zIxcKz`Jr;8C_EfYR@=+Z0t3Yl4o~UrXG%ZF1d8``h(jSHfK$mDAzjrIirRJBe$KQo
z8pJ4%@GK}e<I81yO-~w}lKs&SX@-CNk9nojHboWpDR&Tu81d?~9gL5-!$qF$moF*%
z6WhLb92wo66TV>n)!f<k1pOb(ttqqb#g8XU<G55$69+=o^$)uc`wcc*$F@e};gh5@
z{dlS&jNGy(rOM@1KS>{V^ocZ=i=%XR=k9XccD4}nBrpaKr$p5wn_3$)!dP=kf|>bn
zP)E1||9{u1t|m7-h^J=iy~FkjgQ;_h@tgO@-|OYS(Ky2r%q-1SU-s^2Vp=${rvB!l
zPWB>HdNjvFd$FEegQ;9QyqZ%9XcwH6p9!dv<A3%H*!MBw594MINML9uPiuCfMhULa
z&<niLjMYh4Hs6s6I#!h9^0mQZ&88{wczVR4_nO{)?XUfYaIVEP<~b_;J=7ySmlV*p
zT}2rH6|-(-DE^jXg_b9&OJ0%(1}r6wH=&}U&k=Z!x%F=ye0kk20;8aI20hm*gdlgV
z@C1JGf->vIdL5`2Zf9k4ApHE;MLC^ott`@22uMpdV}&0cshr!aJ_-1ScwoXDgS^C>
zXOXB;|C_m^PmEZ1%lW=(gv?>Lq2<$R4=3ku$|m0*m}V#*p9&~VU`T1geP{L|pCm;3
zd%W0w)rMT~SOkO#hE;3IN$FOeTcK2W>CuvO{kL$$x?R<}WNNuqE<jmRMN#$?$<>>z
zkt-)STby3W7{Nu&$4^b63@J-qV>Fly=FIbAEwuOrdN#A>ChH6FxBfikNM9*5)NI;f
zo*!CYtd|bFX;o`0-oX~?Mrdp@A2&s58^XpOO1hI3g6c55!i!PTgH7}&8Z*NyuAG;C
ztj)pB54TQ$MyJ5ebGW=GkT@0W`Hcd$ZMyOUzT?H)Z+g6ZmX;(06fPonSZp%~FWF>_
zv8p64ecb3xnj6}~FutjiuJ;qk<w-ph{a>Q-axo$5`tWg%2>}9%VrOMMV&`dC?h{kU
zkg-jk`y=1^hnkuj>;pfu>@V3W<+n?Qi&OV=$SED6u$hxss!k~BWA>6`Ef^n32OK=-
zwW;9JlCpY7wLj+VFD3)yoTq%Y>K)IKQv0lp|Ch>_DnC@!)xa67QGKkIr-A&ZB^}E(
zQRUn3Lc!G4LMyY*d4;>?=0<m3HkURiTBn*CLt!Z9!v0JDxIR_8gx;P)a(iEv1MhUe
z8!w{e{g9Ak=}ZT+)iG?`88<xS+_EtGRf)6$;*c_(+2oO5V>CB7j-}3D6C`h+G%Pww
zeh&>--@tw7Tmm@%PZ^PWk)kd-;9ejFv6pILKI$S$Cg$7-W>oTr6vl5ZH1qJ)+`-W;
z#=LR&&%EHyI^X3{b$ddC2hW7oIL=WIJFr$ZS7@0_OS?K)mp^`BX&!)RwQo8TpPmL;
zwy}F4-+WNE!Y?CqD<Qg)3O7F9#}oSIMxSg@nscRfYuc7d-csPYqhRnteEp<5KLaT+
z>s^R9KYx|PE#_A!F|wO4Qx~>AjhlIiM!9TDP}%00%G};3)%97<V0v1r<?B4}>q-M+
zGQkTmrvrYDV|>fDsIO}EuWV1af`d4R^v>g`$Hsv&@^0t$Nu|<;iSmhX_|n7k-){Yb
zO3RQEgQ?K7NP@=JS#M+a`hwdYa{RlMr`o-H&>PMb4nap8T{3l&!(k_6+SvBAN?I_7
ze$@|^cR3tFh@081fwpV)baflcA=RdlVyX)z|7|rcj~zrO3m|QP&T%g!tLt%fyrOtE
z!I~&;UQTRrB3qIwc(0qDMLwOj4d#x8V1DgnIA#pv#&4Q%Ya`>r>L8n`2nt(5Qr<eq
zJJ;cW_r#n<dsOE7N*uj2VMDHlGp%rA_cD%@*a%Cm!okKpI9-fC#y)1KG#t3!4o<97
zlI?kWKzY&`C`z|2s*Ky-7q6(X_8RtfoS6T53>3D6!aL`qJnQY58vL(zI;#p!E4%fj
z1UGbmR^6z`|1v1$Sb50@ff6~-|BtWkk|fH<Uud5wjo$7PR<pSPR5S<t%hX#(yyq>>
z(fA7?ahF_)-07P;0Md#@Jb@Y!H65cP8b57zu!YBhXY6V!Q)wk@Z`oZxa=DF7g9$4R
zankp9;ZwrfCY!v0=D}AAb8bMl?r33tRI>Z-*K`yV>f3d?VLH5_6~Fc&HA$)sAw|&l
zrHcd2EG@@+LN+c=Y}COdeEmS?8Kv8e9x}p$@26+HHd6<t)ks>0qC^Aj={;)JYr3gG
zviLN?Y&I*ohTF5lWqG(Ov=~mRih$IbE~g)F{h)qakNt-PgvDjnixX@S`}3iYx#i(L
zD*#+FGRBy=9Ge_d0!uq%f$Cj;!!c7zprw?m?W$#t7!k%^BB1cz5kh9?!TeJYITcTU
z$F=G(H3pE999Ruoep&Vn<lcE^jD%mkQsOgt(jxd25LDmR2E}~b_*&<e)@~GxuYO~&
zIwX_Q0nMcG_is0GrL{C}2*%I<*$9oKtxTruG5+<lnahy1#tr@_@FryBufvYR99{$l
z7FydRM{6i>oDh2JhfWb`o|#)pEf|!hzGvA;J9@sQ5?AZ|-C^a~z$>iOdrJWT{nP)t
zAXyo5-o1DX$FnbH8YxX17$}B)(M}VXljb?kB#jEg02NyTVJkBT55|z{U?mR&3*%>K
zcyMJr1$DR^XH@izRzlVPbTmJ0Nx)K@SZ%%LF1_SQR_UX=SzWu=5ucGzfBeU(?p_5X
zeNT)P)Kf-H!<1K?dY_cFY&1Eksi|n!uVg6Pa<wY&U6%3^hClg`#+?NY4ltgQ)XP3d
zM%m@caWh%vOo<>-YJ-3$x$|?0G86O9j=qk@$ypp3W>Hy=U4*yyUuMbs<F9Hdz@0|<
zoW#8vVf=z7J!Iz%4e3h0F`SQ3iOE=O-c9SEee)7FPVA_5>dFLrMDBXGu>UaUST4RD
zve`;r<_NCIcAT)J_^%X76j)M-d1g(xmOVu0;(if*#T>k(D!!fhaCC3mb?zcCo)At8
ztVcK9y%*@~9Q|=R<2&Fy{^pwVG*a(DKHa4Khp@vk1FwF&)N4M+l7xr(C2d?_XXxXf
ziq`T}3HUG;r_F@D3FA$BUA*&9{@au`Vs2!dY&~Fl6%<0&<Sd)$PuZz}UbG%Z!)n6@
z({o)d?@${JDLg(ie{(y?gIhwFdBufg-n*o5z;zPzkyHclmrpfLUW<+RFM?=o084FF
zzefAIdOJgnVP()ppL1zMZrEW71N1hnwZ(70{mCwrgmo!2mKw?u-xusqTyq?+RI=6o
zsnxQcf&S8_jHdrsCCXEHlP?1*fFmS&mWnbfqd?yPrPJ2cI~v0_HDvTx2!2_0CTiw&
z{RbFD3!N>Rgo&pJM%*afS;*9-F~PM$Mfyi5s_ciPZv3<TRFX!+R6q|{S{~gmgZH!D
z<c?{$!9(CJkzIo`x70#@^#RI$v_39+Mt*S9+|cg_HrP`43^$KJMewMOem#r=jNQPp
z0hG0=BX@zIEg~YKKzqR0MB@qrMt!Gs5HN_i7`u<Tw^thG&#i|lUKM#;bMh>Mp`qKM
zhIxornZMeM-!|)k2PEZXRmbK%Ai=YE)%UjFL-Yevi6LI!QPNe0)mp@N8ypuz5B}53
zu~pZ5tpu*N2#u5(uk5<YPa~o`FSWA^?_iw<9NgYL)&IH4Ylkmk2#WSsZNLOZzq={p
z9ol2Gczh~B2<VCsPzmFWxa)NuJyjCEM{Ur5Z8|M^ak)2GVED><(kba{I}auYwLKsx
zQ7Xdt@VGiy<D$i;EdS=Yjx{G-eVH%dG7YIpd(b=kM}a*%29q}s*Z!;^7#rX4vVB9*
zz!M*zNdCuglITMJtBaq}5zQ3lf$1eBUU_KsX){q5E4UgMmgX#83+99^?ZW}$**qls
z4QbAwcV-+}%{sn91CmlvvAwtEty0RWpHt6e@brNc`O-dL55PwBR50^9CapnHqQ@2%
z8>C<wl`~9OFIHi%UgCcdB<5h#c}?9iyH}%Pld?AdvAs`hzQG>lH!W?ROF`1hl2tkB
z*Pe#MIkdID5;aTvq}5|ZrYDiuu@g)#qm0waFDkasje;N&f>zpl>L{Zu-R-fV)6pxI
z>k{<9<Za{r4c?nS<|6AQfq5xj%`3H%*fmvA`fE*w7j}OrlKIwBQ_j?B^Mu|2e37t1
zNYU2N?Xo`kG>edmve<k+y`|e)TU3nF;c9v;1H-#29kIRNq1fjQ%;>sgg{~j$rJ>i7
z@>z8>AKi5C*p~h$6?M7$TounjfjW=J(e2dU;oTt_oefGgFA<+s?+ji>Dw#LF(gk`X
zvP%#!N0_n9Dwalx0n}fYRG7pD&-UH=l!_KpkBnP2yo<4k{-o!GiYM<Po2qy}d-_$0
z{e#LA0SS5BYE~Ctpoc;`Z5yKvhbGL=fy(m9ud=nUwqC*7yNZ#OQKuJc$^g*N6yvt>
z^D_E1v*IuO!cv4_(kL)A%dW@uwXT~E{zy7k-Fr_EPlL#g30eJXjup#X$Gvs36G)j!
zYNizhRYdsCN$gx%zmn1<J{IgMX0FeZ@g1HK)mxSCR;xjS`p9$rP^S8P*n27pu9%-&
zNp}X((seIv7Rs5#EOmt^(-aiiFYL5%TG$!ew6Q<7%OwXg2BBM%g*i*greKz=4fYu$
zoKV`fV!2n%bWYaVeN*tt@cieE`8h2f55$K>pbwQC35ey!+4e3?oe3`TomnmR6CGD|
zNkXf87@Qy3coZ=kfC^sde8ePS8r6~+R9tvu5aeBph0CvV$h7#$pX*|b6`6MGWOc<}
zVY<}v{0>keM$=j98duK!DSK<c=|vnGq~8+6YVIt8;S=d%LOA7P;!U>W#^Qh%)~s@X
zMw=Up)!rH6M+JJ!ocFCLi0Stx;S(r7wzp$$uA#S>cGV8)b2d^ZPjW`6foP|^Z#ghG
zBUkC#W;>ShE<BOV;2O&3L^;L`^tOWaV;KdW<dVO!U<3Vnj6L%Hp#{`P{wXg`i;K}P
zFw-hXBQzl@-`oM|<CBD!{irn3I^G<(rVq9GeM~Df?a}stE&C2(67VV8v}@>yb-!I7
zw#J8T;}Va}^E7wPn2gVmr55613R35%C<65P$?jZxkTF)fMBV8acBwLd_NnVS%d!+u
z5s7eNn~W;kuHu0tM_*s;0ZjHzAR@3kI^bnWrGE<II61?%rkOk;XqqqVlgp%XI}n30
zf+barJsZh}giz{#gJ)S?%;f(|8U$*I|9g(tEL4~<H0k6{D~t@RZNrn^fhCF}q5)f~
zM}x^f===IEBi3fGeZraz24xnh=%|}}2arBu^;Pedswe?9i0g2gi_p=GP-&y%P3y-W
z(Fq=*#<2KmT#PKmIT4`S`iS2+9zZ)ppx;ZQCJT!OCzq5g=l5k~?a4}M@l;OWC~~k0
z=@=v_m`1SOxd2;gBmCq2K$n#F&%6}imWoP53}Z1k`C?!q6;Vu#!i;EKdo=PST*N#1
z8$~c(0t-(h7;E$*%Fr<|w2X|hhC1i(UWY3OC1+y==czlbP-Q^w_SMAnANN%MLXP`?
zkb{7CRpe@a#Eb?yd7eVvb@vkhcnmD%S;2!@Y9_kavCwVUzI9tiuZW<EazJG&(k@ag
zvrXf-lC7*J64j*MqLL#1+|loe(ZB5TYGT<<cGQl0$sV{F);|`SlwfgW1PA>O_@}Bd
zCj;z~dVKGJZ~x9@Q)5j&%O?>9m=~e`A$7{T%JM?b4+a$O-zTVAF4i41xDeMo%q;W~
z@l^Sx&55w_c7`IEJe`ft7oJDAv=^5MFw-?IS=s-Lp?}u9pHmWl$U=_FDd7(asLg!;
zMSmeXo+D)Bs#N#jBGu+9U&DdNLKl=`U@A`4@;w0A5-(7KMnSI}aV+1cbMxu$a}@o|
zJDl<Dv)0@anboj+2J5*w5veeofXwZKhKD16>W1RTU;ZY5Z@D9O&pxE;#BEGYKL^U#
zLEWrzN#=U-0*Nt4=PLZ$CW}wHarF}tuL_9P))b5Qfl6limYc%^6Z0pk<InXYPMc1t
zu*8m*nQxx~Gsa~qjG-Bx!7`WB8_M#g4B*TDkbgAq@vMrsEX-%8ZK=P2)DB4**i>Ps
zC@u}Kdm2esOT3x}+QR1-|5IP#kmvd@OMpFuZYnHnI_J?Zlc*K?r~=wwOGn4j-3lY!
z>?Q}Z+``B(QGVL(h*H8sP3QhU<NPh19ev&i4}*Vq;doe#njP^!srO}@N`Q5&GhZ_T
zVMPj7PJX5glwHXY7I0ZEC6pxF5E21#vYt9HDt*Mmov7WcLcvgO{J(o`$kni)J4bQY
z3-aN~-C$hTdR`RDe%poki&8G2ugi~ge=HIv>Uz=F*6ixlee-yxCj4NRyi6k8nAU#*
z<RCFKrx(t3_(JM>E~$$plc<jsBv+jFIe?1j=(0({qH4H;j<I|&gf;YNox>6VX(}Y!
zCMPJH9#(xe!2Jgm2&K#)AE>d&L1`5q6I@JR*#(=aB3^s{noCbM5T(I5ySw8$Cb-4k
zsY1zK_^X5O68y(?NkK>k0Au=Jucz474L%P(zk1224G5Xr>hJtSuOc<g44bwEAx+aY
z8<gz2T612Pgr7gP0pcVp<?CnTn~(is$@T4{>N<hkMwSF;w72FvbNX%@K+NY`HC7>T
zRe+3Mv*RE*fcmN=CU?*N4ejMyfd1N@Dsx?%YWF)bOk>p^^YD(ibxgEdYs}9Md*J)-
z=`a5EElKb70kZlIVZ&@TcbdEe`ilOKa0l<GqbdLD5)|~OAwqTCx7#SH+_9t3DeW~G
zr6wYbsZee5(Qiv26)sKph6oD>l9RH$0!q*3Td8RsFpL)Lr=z)?Pa1&URdhiB0shj^
zWR4vnF^e6o{39{^lgP3#FcDtU95BIvSiv7_bGL85e^}Daxj(G&<mZ13O;^{Xv&1ob
zo!Nwmmp)n$5*Rn^-8s)_ER-(`4s_PW<ZqMw)8D>KsNm#G<v*~zqN}aF;RRfr_#+S~
zxj*1re~c&URz=C{&b3(Nsf;|Gn^s-wIJ^k9>hzX6pEj#ci-X__M(~N*z>f}kS1&02
z_N+qWfq*#k|3^T=L{gK)BR_!n4&*nUVIe^<^MS9Mg=y#~19N%f>;!B*va8{26(_Fv
z%$2gkg(@-3X@aR#CX3Tb(V0=-$$aT_h22%+(YLE_z>n1-m(*V-9eReu-smj@FDx2B
zM$d+b$Ego!)KJ}30Y~X;wFlHDAFee&d_dWx$x3IIrkQLd(?}_Z?0`|>5kui1#H%Tj
zU1W+gD1&y=OSXv_<WAy7aJe}4+uStr^#gGJ=QfaRzA;w$2or!<bkZxQ`u*wGXzwrK
zIv9c^B$w3wmOp?AvADkI{_P{034!WKUUu2Cp8iYwC8mu`1Z-Svt*vz-PoX-47cY#f
z-+^5L`8z4Wf79|yTWS8pQ5YZv4R9*ROHUYSe#FWbX-jz{RO~py_ZY<l^ux5qOD819
zu-_<ke*T%2vn-?UfFRdYEnj>Vvf`D%9w@JJ8L|argOWX-K%jW1`<q`WXG!000H195
zzvdzkKV0GH{0`0s9CGLl1<P~9t4=okyAM)+?BxVsfz54ywxCSA<rdZF6HX+hD*Nm!
ze-pAI`<14ykvNDGfu8PH;}X%{r9W}FwI&vQLlEy2GMZcGOnlhMUKhm~ok6n(!vK<(
zg~AHmIcftDm_OoCg$GM@@NcbIeD74n^M9i`x{a1c#mKp~K*jgx)2pKV-~0zQw^Ui2
zG?5kLS;%eXv^m7j#nQPLvFv_^+hCCD$hb@$S1gSuHM;P`pS=9@NM!)TBH3)n?-lm$
zON}W;%I6z}YX}n@iS0657wadC_nKWRH=}4euCcU5gFKJ()~n}7)|SA)Ez?<DTAhOG
zr~lZtJe^SGt_vtuSwky0Qpof5-9KF<cwgl(D!uaez%UJ$s(pP}=*x2EFQ;d}q%95S
zP?;9#<k{;|7$WEA^kEoG8o$}&XH!1c=yCoDl0_i)!ECf^$PuXWx$Ysku(cpLM^!&$
z-|6%su={*vYhrt%B&QQ%8BL+V@>G0(^Z~5P5X1yX_23t0tv*+-(tQP!bDgWkvirsF
zipx_9j>Xp?Jt^UqaO+HbA9>{Siy{4&YLn7uF`LD_3%&s{lftuXsbzgKc@B9M<?aO{
zA_`JEbZb#O^7`uog0{Wck}xm`X_{^QDTww3)&P_{%mu(%r(rl3@Wi^ibHOB1J{mP0
zaq9NIl_i`dN;<lfB~L^{W-Gc3S^FSn)^eSG8cs{o*+{%Gx`OK=!?Rz?_iK6fo8})p
z`*G~Vw^SF{T99zEd)z}O8Pj<l<GpBk>+<YJI_s|i?EQhm<cyK>yG@nvP-Iig-t41a
zeMegirFNOpL3uq_17^q0fU+NLv6;(Jm8F?z6h{rGV1uYb5|Ep-(X&Uf2iGoafw5Q!
zZHbtvR)6M~dq^fcbN$z(_=XLQU}5VNtqDP}3CQ@xC!W1et{ME#3t&OI<Bl+cGNny@
z{y2?2ClOi^2eb0M(SFvW4!S6ilacq-`Gu5UgI#IM6jlIwDf_-$x5@N-OABWI!dDzC
z8FT^iU1cL0z(CqyxLpzfawHoXBK);)51$lzS+0^Q?BC`1Z4&$DJk0xk9DSXw?tA4z
zMFyv&_*3->Km!Ju425?-X;9Qe#c#%U<iO`jstJsJMGb5KN9P0N?fB`v7~c1wF@VNB
zSSb|rgDQxVx6_-Km+ZYJ4Jlz}Z7x!t>WkN;w6B1Q2mI<hYr;LV5VV?GBtYwvx(Ac0
zny3vP-i#FIGTuX{@~WWE7%@ttSL`Y>Wn1Ubd>k=)3K@t#2b5k&NR<+((AapfgGH|9
zvv;>J8aeDLnX7aS;otsb$Qs=<#B3ojBATrt4$_3wne{-~glf@>9NK#m5rM-`0D;a<
zzQ;2FXFa-4|7QCt^2a$VnO*-(*59&|vi$Jyz>ibKo3TusnWCblC1Iw0B}0P+z4IY=
z8qqUbL-?PvRh+uxv90l~%E@}ubZ_Wz#!yOn^V0fesy)x&|GD-*kk~b5J?;Hk+<Oop
z|1F6!<J~31B5(Y{y(rDnbkIoj(dV&bXVizv#rUU899z*|qw~65;wsQTl~GmnOEQqk
zlm$l8HWW9Fb2={?=x9<kdlZI_Wx0Bu_=IRyWO=1Hg{<xOs6m0XP1964EP-=KV?Mve
zuu+iu5oJIlNJTs$xb*D*P6|E+Ao6QaEn<SW^5E~bXB@}a!T!<SvuCn%Ai!<0xh^3=
zI4d$da+H$(<`*MFL+>9XAaZYvJEPN1on}GSR~5L}xLSs$l4g331Ip0D3HLs$JH3Sg
zglI%ALBeEP4W_-TAGE1l=OSB+5l#hhzbdB$G<a6ag<}$FG@bS?V8hSiD4>{6o?#26
z&n6d^l;!U*{f!O;{zd%Xp}>^Z<DCg@9^d+U7-Afd^jTS%&&mazmljI+lRsD2f|l>n
zS>UUlcgPixWZ^Joetp@4|8tSFOHDF3<n{Hm<-ITjCUpms%{I_k#L}Ig^V-LSYxELt
zZx2|Ff$^)7(9MUR)9nvXDG?;=d%hpY^vgCq@3@Lj8YzONMC73}<V<14l{|_TK~%e4
zY!$bwxYnMjgL7zDa({2Q9+SG3o8Fw4E{@}ql>o8C?2K)1sKv!g7_g^u3Wgkp<HJN>
zz4rOfksSHgxLUl~Qv~8cCj0qEtebM@42i-YS~<#Gp}%6`mv11POauXSo-U9oVPK4H
zT^G)0);oLwA{WNLrSDIXUkuru(0xPqqegr_-UxYYpa3fZU>^`n1e#8lEDXk9p^1JJ
z2{>#QsS@eFAn+nm7BQPdilB`X`rIHtd*fuEx<*Cay|#6hNTRux|Gx=$De)??#+Gz8
z5TBq7h?qXczSkE{#sM^O{<#|R0pD*!USIQTFxwGBP;1LPvG}kHhf<l=A*S&&1+=aS
zOV_IT%i~zPA_~b@x`X3qH_!%9@e9H?h;myRJLS{Y6p29UVt=JN1g!!hg#p@!!Sscc
zs*97h(N2K0GM~gG68J)kmG%%2Gg*ztf@4^SuH?Gz=CsvgF>g~Rt+d!A9+S9Ev-VUN
zh_K>z*wMH+wrstGuU!<q3WGFt&G>+u4LXLlQR);Xrc^p>5-nBI05Q}%ahIYXK0;;F
zCd}HW`a$@^tJN~X^JBLNIJc(zJWj)S9Ppq)QWDkH2zKl-q;;PdeR2N7x8C`cvHgw0
z(JJ0rvO(9ac2{?|*q+NnJzjpM3P_rEjj9loLg%pCZpVYnYyNm4pcz%;$D8uVO<@cZ
z+{^fhAwPb+f`dD>C%pS2C@2^f7PdogU}6#|lswm0({htrQ1Ft{8LOE$p0_g5KJCT-
zq|oDHY?9YPX@I-wG<A^L@>KQwNJiq05ewZL$kONQZHF|vT`l1e`?n<vUQc#gyGP^I
zC371~9YOvZp7QyRbT$i{l_9{>2hEr791M3uGGw!=drS2;s&mvhP+p8Mv>^JhyI&$(
zDp+g%k*oL-q96ldzU944oG$uZk@4ML9lIg*JE`LPcPVE>_0xsyS(M7JC9Ycw6BE+`
zKhQ!`UyNp}tMbK|QwN=;FT_a9(vneU4y9=jDzpr@R^=`WXyWAwmdx&bt7usxy*InF
zJ~jVzoCQMqkbOE(vxjICe{LJT1kC%1V>`s6P|~QHO<P=B(>#0ek%j5iC6j~$QZPaE
zwm|E#27zMcoIIfSlF}p)1N|B^c$BNy5ILfk)oJ9r2GCXx%CD~<M&Wl1IXjTF08-4_
z%`rj#+RDn0$Vm9Fsi|M$FOe?}+KAB5(VKmc@QkX_Nd%BG@kGh*;oVZBy{~U!1`7^E
zf<cSaZx6h&H`#p*%vA&6E8b{K2a4p9Y)><~sWWZmS?7YX{N6A#UUA7}C9xkLTMa{v
zasRJCbP-U)iF#XzjB|9h))Cu5Q%1>pOaK<qRm<z>{&Tad1r-=2MQ=6%r;Is!a@SJN
ze77)^srsDVL(P*l(1m6_k48g|ng!IJc5}|sJsZ%OK|M9SI3SdB)(1hdW1L%Tpl~C*
zu(=N=0i9T%-0Ri(lqd|>;kC09i+#3eAC`}fE@Z|x?rshuEY>3)7~V^~lJ~5IoLhM`
zdev<D;!|1O2{Z63>tsfU?2c<VS?wRaCBS1oFXaZ&$7k#v(K)UQ#v<QV8s6VS%P`SD
z1uM13S&LZNlnP8O8z;lm@8H5U;uk0M(Blb}pERV&C4N?xwV5hpdrenl3!Pwr0dDAS
zg*Hk_;nsqJMxW@iwgLfMLP@Mo5EoKjG;PDgSUqZL%=~i4P>fJgkaNrxWjG(NH?P}W
z=(>e~kaInXcJgPWU(?Q9HNZcW6b#rN{;4{tI;Py$>zrS>!Rj$I#3<M}li_G`YtHc^
z9|1wGB_Ur(Iw}_227yY}U*4*|^IP;VfqC87H>?;Wr)6f|kr6@&Ki^;EFAzVgnE5`p
z%&TEUrqash=_;~(U`=%LDAn`oSru_ZDMpas76YhBO_0UPkkp38yEvYKX1Vs7z4p`s
z3G;8#()=GQ3zwD+lRWoEgRzM}?om_dk!uEeCN%C`COBZEbbQ)86|7Re&5fRmdADG$
zwX&?X;)suDpQg08M%mr><tWR$oS+;GIbVD|Yzw`2?c@^RjbR<qw&ycP=6Z6Su3b$|
z?Dt^$?T($JrdTqpYK)KS9PFI~r*mw|a&L>Oy0VKzNj|>Esv!c&Y!g*i|N4>;+P%^x
zNO?Y=GN7z63Yb}ZWp!<*AyuKVm_fdG7P8bVcnWmC3W)az=RUM3zu0MLh+YB#SsftM
zzrP$BT271PeYO|TOyfII_zOPrl>X1~WSnn3gg_HnRoHes(uC!Y;NLJ>i$M!tr@Z12
z^cz8q!WH&^4P#UN5n=N7{ttLq3h}Yx-{1WaNX@w|GeJ?qN>qQ(x|<|}94-<x_nqy1
z#qlc%qB*&RC@}H>VyCTX5^P?NWt%q!!h3W(8jau#IJ!(WK8Khfo4%@|yTgLm-lP65
zLbsaxSmqmwX^>5pIa59J2i`K@V|yJFtq{q+berBKwqtRSy6Av_-hb3zyeF&S&hftL
z%pOJ=DcapWvg7{x4v27~?l-;h7^@BTony#RbzxvC{Y}j24~SW__wH}tUzSdOwL__q
z+|nwZxZ806`opqGzdFYWu3Yw+jp$f9yw27}bA%F2rHcc*LU3Mlr6EdOO|_2Rl22L1
z_wC_Ni*&mXIWq}94AvYR*ZB#$37)Lm7dvsp8B5f!@}By?vX&nO-0AI&5CY=qk2up5
z6slk_o@!>o{XhqQqx_vXlBCU8__?P7M`=Q;=RO1J>-nJ-b&VUiLje4Lhf-%4Skf63
zdZzLdpvDk%NTvH%TjBtJ>8?Wkdt?)oTKw_P<V%<A0+y!wOiGnUHB{~ZuwLL>ukkBx
zdLpJHCyhibF0j2Q6*}Yh&QXieTG?(2^81!T-lDy?tR5sM-;+zr%B2$02;!;0Gb83P
z;U!U73w2=s3WC>R-xwK~l7%CMLEgJpgQf?p^7ZDi5MS_}63A4-^uU)wE!o8=AM57?
z+kKKj+~zgO_2)`U37%aWSEHA=$<@CGhjWRp)HtnPg^P&8-Yqz-faFy`ML)MHr;F5I
zH6;V<FDv#F0U=)ahSu6}a5Y9Ox3J*s4S+I0{I>D$DxFxVc%H+ku-&XaDzG&dG#WPi
zxs=?=;vfTs>PVF+8yHB9ri^@SDiX3e?3vuEPC5VdYx`oe3!Gjh|43m8zpPTY{3OaZ
zErQy*MmHfl<UmPlDs|)qb`6L_2&g-~0K-F+BD>DUICf+jcbBQ0)mHSIkw5nLHSx$i
zF!^o^<)!m3M~A{vhj6cej_%jgP$WWAys=COqxvqfEG2Ba`9-%WH@Fr!Ml+V%M@dRo
z$|3R)sLQbZ4^+{yA_)TYlJu>gHjGavfG}a6Zz9#$d`|~8=eWomd2g;G2$*)USy?fe
z6aM-lhos?awwhs(aJfEOTz%qlV!37A+nZ6Vub?XXKTD}c5y2SdBU;+f!siXx+TG*c
z5e(`e+>kQyeS@i#GSSQukDMazpzNbNtivH|APC*JKB<4wc}Clz6;q`H$Jb>qN~@wl
zK)BmFa!CN2Z*1;P5~&0sCrT>16ylF}tpH1vkb#`LlBc8?PGBdm7<8nl9BKYsNpX%#
zmNq+OaUN>sd)DJKtW7!S5l`X}!)J~7`OdgwuLmEaM}}y|$*w$M?=GEE8F<*Cgw6@T
z1_0+945Jn%)0K^|M`o_Q5v-+W+xOnUJEqJ7`@Fo*CC@F_NQLWAMJ22-aru!z-*+f6
z^$7CiH;q&vLCY?ZI~=*diJpwviWddK3&_9I{k1csP}GpPJ7QAzP*E#l2Y)VFjUfdE
z8qVpw58+}%=_3!nimvqzr@e5T&}~+4s9hnBH>Lp>Vkl1FV1T*Ysb+6=@|c)=*Q3{Z
z`X{0Dt6z<tecJ>xQkf%^QWw8Imx+5m#{cAQYx(krf~B-EKlU;VO66VcB^yLtZ;}vO
zI%Y3H$HtpHUeBAnc>R`c`!?|5QeI#_LvQOyWCW~6OLZGs#>kgGI?~-yeOc!|iB#!@
z<xN2o+4q$jj8Rig>v3Bkm@NxG=k3wW6-+lJ*nUa6oTjJnzBMzNb0z^da4w_Q)(Azt
z)E!Qg0?+E_5@ok|)A+Omw5GbjHpgx}RyE~?bq~oI&;O#z$tO!G5?&XM@EN?w$C@U3
z-xR7z#w*I3$Gamure%RrkLIj}<$lFxwb(!|#B)g<B7TD%wA!nISanL&7omY=xwMU`
zjO--7hO0xjEie!PZmJ$^R<7n&>*$(@0EILRb0u<WsRSyr8qxtB*&755&F&^4tcK?9
z$8)1I2;A1#bYDDcU)4ZQ8@rI9VsL@@V>St3m;yQm_h~QSHU9*mo6fz10#Y1x;}_i9
z^9~acPfa;chBHU6#KCPQscaSyb-KTjdtPt@<d0~NZQX}>EfTm^6g5?mOiN2n0kPbp
z2?!!Du4n02R-B1m))qYa+U;d=?<i!$=RCT<VhJSLB>u8~z(YCK{CTQZ@?C6&&u3rW
zI#@q_8twQ?+C34Yq~xtiw0crlt|D*r2Wf}5v2*7?(0F%N{*cLR4hPJVbE}{hEl`4s
z(F-#>jPXYTz9ttmiJ+z+5&_r0eqT~TdtcXTZKN-%-6<t`L~scKH%cO*<Zo66YPPbH
zimwH*b1#cbc~sog1Z<iG^<2wv_(?;jRuj(`uS_mZtWg_Le&E0I(^;YesRXH#xVZ~P
zn4K!5T#G<(0~OrSnIE5>s|2y>`@7&PjxfD@mruU#!V*FPsafafrDm{%Z;ki!{CwJ7
z+TXT<K0(k1c^DtLeF-XnIz&TOx<?&Vkrm;N==pi!{peG`HKDs`j2cY&vXWh<@8Ir;
z<#HQ^-!tW)Bm+Y*n@t8qhk)x#&zFAY5m20;;k;4f?kq#$SX!PzqP)U4QYn9ga_^|B
zj@Lee?(e~#udSMaq@y24Edp>5uQ`!2+^KER&X)-7ppv|8qT7(#E98)*X~OrnGZC}S
zlEm(xVZEpc1YV{FNXr&@*3Mp4ueZ!rE40iuAX7g3v{&o(zsFA=yMOmOp6x1|BZ8Ik
zdJ}8%_)+hgql;?MUX8V$9bA!qUxIxYTlX^vZjU3FzV1n#7}=2pwT6U?_f(x*+}DhP
zZ<xle96`G+?sH|<1}9TqXVpmI%0uhyhcc%>+#HaC6X?87&Bh)}Jt%LgF1c&Y3ist>
zz*OAP3~z0=W-x8X#h;9Gruvq^hv&2YPm$!}<fC>998P_Dvs;3RdTW-%8YZ-2XhL<3
z$HZjYa4r=Eh7)pmrX&6(d5Nn^i|!)x3ZMtq#bJnTaIDpwHE^RBBkK7Kcm0L-DA*iP
z{9}2VMrVH@FyE@JRcw(@f2)q<Z4T!vOgQ!2)9D`KJlJ#Cc^X;`gS_khBI@Aj7*EZU
z++0$Vc392H94tJ7wUzDdgM@MDhLk!e;)8vkPP;4bX!P{ceS~@+yjyc-wfs9amhhH1
zo{H!26bh|kuG`w)YUvW(aIZ$zltrv1U)C?42$`ERZBAS?c}De+lh3eM_dY-Qw$r|;
zNLFc_?V~@{B(q>kVgD#eb^fN8p*ONK$;z$rj#Q-RIlX$ri68cK^sil0h)RsNdBK+G
zos)C&@0fUAo+l0SzGecsrY;_eU^`zXvHW~;uclc(RV+S7JT*}YD7Jr}oS>*<i6VuS
z_)Lhq5))nlV$D5I*fGW$=DhWm9Q!SA+%D<=rHjp8lWABHUq|`c%I@y+$uI5e9m(H?
znVcQysI%UMLxkyRMNzg8QBZQx`>w^CFkK@3rZzMNoRjI(2AT@T^HKKE)=@&qMB)Z9
zhx#dOh8yc|G;I7;A^Ph%8X#kwJD_gyr-DWa{7wPWma_)V`m%GHBkEEgb+&rJwgnVe
zJ=K|7OB*}mKc8sr`(%hTjn^7(Q|T?V$51fM^;vQ5cTJaY<`ZNw)jR$EWB@a8&}!u(
z_4**uJe;>X`0EGE)7+7)c~Jd>u|?CU5!}rAD;_2<!UMJv<9&qE$@CWPexO#5FN%u`
zaZlCN#!iP;Bdp^J9S{GZ#m)fxMbg|thH(G;8Wx!p<dk~pH7G~I`1NkRa%Gk<k*zFH
zbmwR09_<gML)LT^lv7HlrxILu2h^W4-(wVim)lbp6_${$J;Fm8X}-Cj-r~g#nb3X1
zltpCEbt&&@+2gQHTXj+!oD}#ZqINJgr&;{OT$42v^@_7geTQGrlDx?*?oX8AiGmMd
z(3FI9qdNYK=Q{7wH;!z(EmV9KABi!ZfDQxAKMM40M$lB(%;JT4a(?FP>zEBtWwirS
zXVI5DW{=|A#YDCDM!(R$)LQcPvR^2STm?Cs@xu`tZMWE<-&HlUrk#j`idPHR!KOFQ
z^%W#<uWqGHcJS$>n@^@rHxh@#6#gh`VE4BRcE~p9m^So9psAo(8;uQM`WF^HPHM!y
z+I-Qi2chBd#5cRA(rXVZj8uyosxxKJ;dOM(@GdOxbS_I03M7Y0*xna*h1DmI`wMdw
z-7O|+LQ-y+B;4flGij2ctIxJ=2;6?~gXO+<+w*a16FR~9GqgwdAyq!q)djD<Z==1;
zeD5`_8@pU8J1CUnkuT^eEwAbqZQbt7-#~@m)-+jTH%9fJn7b4=B}Ki%rR}RZgZIp@
z4?k=i%5ibVHl8IC91Ok)!%ER++GNDdqUz!Pu(;Z%bn67Mx-j|21iWsYyF~}#Sn>E;
z`#z3pY_;;jzNAt^W$C=BD0Tx={P7AGLY9#ra}`T=rs%*NwL>y66t5r9vlt?mLdv4Y
zPsph1R?GWhnYpr6nO#te=hQsnYti5p>N?E(N|qW1u{t<0JnM>{bn9hx9)9Aj@rmc<
zY5;#ldecw+n4i>fX^P++ZgeJT5v&hsTcd=)#;BsFDB$*JeV9RQ!fuXUC5Dk>OF+EF
zN0_Kvz&KmyZV>Byr+&x(>PQYclsQYzr*1J}m6D5T+VG?q=h;)g-xwwT*!`WC%`=FJ
ztu1{w(nKEjhgzWcCv4HvPydApLh^vj>(@0sODl4s1Bx_;$6rLyVQ#8jWV(BnR&>L%
zC&S7242LV2J|bSb*OkA$_kGL3&YY#omA!tNv<_=*FAkJHGGWoCW6qrRLj&wv>Doag
zL|k=e;C5|;yx}|c-OWMVwEH?)-{QVb57uLk@f~zLtz*T6w99V%m2*?Y^E-k9w)VE;
zYM!lg{T#j#eP3bkDR@;kO-3*)YWV@#Z3lj78Dn(W(Id%BKz~i_diF$CE}}f9uGAGS
zWT}=2(z&=@N}s4)U^}IEb-xkSfa$HTvt_jltY6|4tGdF$$5*41YF-naa8}Kn-C2So
ztV)x7*AF|(D?GUB8TF!7eyF_cD5&<p|Ap?&RK2wcKR!56FRLApE!@Z2XCvIt0>3nL
zORaC4(#y5qALEy_La6tbu$sT@4j(wcTtky8V={yER?lFlgyCs%_g3(_^+hFrh80Ab
z&p8LAkTR|tXEDTzgXS}Agha^W9$00F8M^CK)nS<_OZ|T|eRWh-{qr?Qsgy{!ba%Ii
zv~+iOcXLomTDa0FNOw1glyrA@ch}`TJm2@nTKom;!uiC^-h1}U30ZU$c#>TF3r#Pc
zZ8pQtkM31UN(5J>V`(7sRK9hW!;?tKuS~qOjuBSvN2;imlW@i_p-QKUGNclN2eRMC
z2A^br@s48h4xj6ki`w&)2}`>(F#TM1+WeD|B`@)--+o=LbsJi<@w}}4dA<+c-9Z~%
zK-5jMd?kJDpyqtiwy$<!B&;cIVl<NrJ9o-`!f|<myVhpobNX;c=>_t%xx6_rH}4z{
z0&S?5F<`$uJsMTT<T<$>!797cD97nZxoH8|^7_o4LAUSNY8YZEnl@ICT?;$^8%AQu
z7Po4b-QO=09v;r@h9{@4X68{SX6bh6Erb}jrW{JfOm?wC4}0Y;<n6cQC)FK)a(Jbv
zrUp3lPG@~!fvc4WO^3LJ;ri6Wn6)iNN)US}E|dvAoTYg8a?W@~?hAw$w7F^E0yC4l
zumi_Shmoo<L`!g`(F1FPjRSo$&5ULEHe2R&ePT%U+X5b1`1z;EVKv@u+uOvW32gPk
z-AOu`e{H$JGptlRa^yVyzT7K5-OUBZ0-F}Sg;9sILO(+W!O`Cv?YeDvdSqr$cfoVQ
zW83XA-Agasq^KzMj_ZEm1Z7AVsBbt-8ZsO`U?Yf1<i{HOM&SMI!2WbK^ff&EEj@#d
z!b0Bu{=UmP<|kb*%8rH4^zm<_^4xsIOX;9@J)73<++jJYseN9JDal{*T;f`sg>;vO
z+oWyw4n7*Fh`XsFL=PXMUPqKhNaS?evS$>g+1<Cxndkg+Z!2EApRtZN00O^A9EdD(
zq@vWia9=Irqk7Z#%R0KcS2w)9Qy#O74_cYqsOy=#D4&|Ixj#5QY{#UJFY3D~&Miqd
zDCKiuj@rIQ(q8{EER!=;bL*rtplOnomrWpmi=?`r5OH;4QCpjdId-lJX0|D3Cfzmg
z>7GY!FiQ7UYNoLw-esEsQs3&s4i66<-R=FbQ4*@GLX4vsg6C}Ws;b_Xs_Q(M8|bE6
zgI^+Z(|fLoSE0Ory{D$8U}iZl43h3s=<7=ElZ8?femEKuno-Hbr>k3t8Cp^bAuazK
zCCtNItb=sBczYr!>i!TDS3-;?>5z*qZTVQ(e|d^#dA(sD<@*ST!O*fu)q0mz?+g#{
z%&xmwa3{|;yWtDWJWGJF0&CkWqY!Os^@vo)_4xpw*O@-u8>NaLJO)Skj4e})J7tP|
z9>TC>rO+l0Je0}FNkbEpiS;ld9l`_e?j{6JdUWTaQgmvK<W7n|3;PWzseP8@MrJ6I
zGlw*03gjT*00QDx-O<l}Esd7B63^g1GyfWDTInz9pQm4*c9-w!*3i=WDv=|aUz@#B
zYk37OpMjE+sJJ*n)+Vy8ozBP#Du23sT+1ijHOkZ%jX;-DNBybm5&c=rf7l9Ul7LuV
z-w4b=1+@K?RC^!55Thk{A{VmE8z?~PLx_75vH&pG9E^u<njrn-t%`PsJGOTA#eZW1
z`+aw^xz9yceo^r0^c<C%<>?)EcQS9oJ&#w}CC5&gp2fhh-d|{`@6~k<C8+=mZN#U!
zW&}f%6A3phlDRz^t;SPKb>_z)d%yCHHUN+^)C{HWomrfBJBIrx@t41NdrTuf&c&Dp
zO&%zNhp)GZ9__-1RkHhuRN3w5qc}lBKP{$=AivDqaO=aV;5f8-o98**QWx;JX}g6)
zfbF>u5QwM}oAJ2Gcn+zay*kq7kB&5-OrtA3#(U98$Wj6+UwW~><Ct?2+-@(?XjZP8
zz<kT;88w;>y!F5fk+{)NxQshAq3mS@HcwML=D-9UYQl3d=p*|kYFGQ@5zGxHLnj7~
zz6<p=2SEP0-k1b?svObESvvd5=vmY<)cLn^;jWM=5d-cZmDPjC=cUn_;8hTn$NAZE
z$TO3;?s8{!oPj^wD)u1hG-Y4Mo{ku%$fPV7*mc42E~Xu(bbBTx(!RoR0-9&ePw@Rb
z&!;kXxh%gxjBV{Z>;yIlM?(%LpRj%t|DkB`fo$_N?pS;YA;^AE_*`tWG}WA#`lo=K
z*Sk^CzX3~L<I@kn32N<BBy*0V5I38#1f(cq?&`Z7Gh3)-^7#%eZqhE&SXqCzqlKA9
zXWc8!q`p|L%HPKV!<&uh?}0^*_3<vImt`CKMEPI)OwEVtxxqvS;lceCk>2S($S)3n
zI%^^r_m%URokBg~vuv*sdZpiTYufw`p^G`%R{F`!I^k^Hu%BzYIo8)-vA;=IWULpJ
zKQ-|Pi>B6?2nXk*kC(ZSk~iy5>>j0d`p=Xz-wQ7+ZiHILh6e%+E5Dzzd@5Y7trp?U
zIoR5VdQ3S-FQQ?)e0()dPW?n2wnysyKbDS-0r$G(NFyzVpO!nOGSS3N@}K@R&$VA7
ztZTFk+bBfsT%TlQ4RMv`#MSo-`DtA_v-{veQ)3lMByz16HAJrBsfaZ1$zhI8Y<#5k
zpL^KxP$yMq3{L>Vm~i#D!xFH=Gy5dcX6|&^_?8|!Bj)!QSt8zQqN^gEOD#^Sux?L6
zOybvcMCcR_qL=>dtgoF>oXFxXa%OtAuQ-~>;+?PuQ&N;cZ(e;h#!H&FwEO)g;$=<0
zaWg(7Otk~i>D4?Y&&|!lo5w42ybWxZPy|;zZxJ(q3ZVr$36FB0!ryDdsckNJPYtus
z))*0&_FJM`^9N3E87b+j)#%>krXrlMGwTD%%gv5SRh;TINGcotzVHaHv+0Jed^cm)
z*g&wq$ISD5UN2Skd3UCgO--b)FIB9VIhUqg>Gb^QMcz!q_O$|A>K8`uO%9pr1B31d
z<2^pW>z#E}N@S2*7n4t;(n5-y1UKNtfGv*C+aQ23*?Nr5`g!82^+G$%Z^=-S>@j1m
zAqv&pa|`b$-Y)kUfiF3n4)s3?{yH{&>En5u)mC}dhUk|kEL^LsV-<Ssr-s_T{LD6$
zTWaCv^PgyLBmEzWw<L?l&K-QV<OX0mNi;!!Uqkmok*f8*{KLBycQ|Qj{(tjSR~kIm
z`0N>M6C^oJ*+1eP(5QL5V#FW%1e;>jPEdLkRXS?4SjPFhzjw3g(@6-sAo~*s_6fJe
zPo2bM7crKvMGcyxmH9&SL`1^1i`Nojz%gWn8RdUchf2JTG`>hs_~o_406=e#nwUGd
z6T~6Q3T94x;8xEU*y;37k7yXNr+|||4h`;KJ^1?&RQJuUy5IM|pB_n<(tgYYJt_2W
z2l!f_H^sl-?&nn3o1HHrjm=uFU0~hP5f7p|NEX6w69vOj-PWL+Ub;+Z3#!W|p-4=(
zTo<2*0>NTIIa`FqiryH&%}o|>PC&Xrxm7IQf2Ij$1nK_<oiLdmt+UY44d&WqTHV%?
zF76&p5fJ<3`@{UncF+ft`IG>3xOc^9Hc6m>cG$F=5N`HrqZ1@4`i{2J^6wYgnnR~f
zqO+P>0toX3|1#YSTDeAQyh?9nQfc>u&;0kX>OD{8EjWLWNb4EnzG(cSCuNA5J;r`>
zga--ZV!^hij-T3(74k#>>gH0gRv@-!7PA4tHLr6NDFw^ovP+|~(qrpSwq0p<xW^2S
z%(s$<S&Ol`=6CJqs0pp}p%X|NhV1ymgboFTtFzPPT93LG<Le1)>zJYsqBpSSP+w?h
z1kuwJMTHJPXb)Xsbam%G)gut}m(ubaa{c<nAr~XTtkciwolc}jkjO-vCp@8K^F^=z
zt2YcLw7l_fOS*=Jm^iX3us5OK)VqfrrnAdS?63cC`S$)7#eV1!;=+wub>+xqbaK?t
zv>hSF`uQ^SqM!WooAGyxp0ca~E3x%JP1IpWOy=;~zn(^-IbPhDRr`iJkH6Dnfkjn<
zZzs@|f5yTD<Wv)S2t&~*dR8Vj@yg+u{f`%0SFO}rPASeJJgN?MB<HG64Q_*OdgmBW
zqK2cxv%fgYAO0@lM1nc_4ny7I#{<>-M40ycFXaiwl?cj8Uz_)J4A=Z%SJd+vz<=3*
zUhb|jm-*b{y5TQUxgdU$jCt1-<Cjsftn$LQLCLG~8sITL*!`-|1w*v>3KU7ejgAA^
zt%M&o%&##Xw$*4Bw*6wM#dE&K-ti98SGAKQH9`L{mKuH5Y;m3sY@DNicOFu#76r=7
zuUf%*W16aUt|{ko!HoiDwg&dTdv*5=6D(CZ7RMCS_grn6r&|7Xj!^>lr3h(@{@~}x
z>3mRpAurinvHz(-W3Mi`rih|ezNSNP_3iF8xil0fQB6r#`a6=MuBbg>!EqQSc|tu|
zN)xvmp&oOU_GEKbP9W<}=U`^#jDww#Ge0a<SlwK=R|9-C-V!3|P`9}8_?_T&442P1
z`l<+m<ny1*qYYTfC;zP>;JwKHqu}gG(TY)4D(_I8`b~7xSP@=e^YWkAE7I7OgIy}Q
zQxeS@y)c@6L{MP9IQcir=kp?@bTJvkr8gwkG??w!=2W|pH|$XI$N0X_JcZVZ+@S1<
z`tSQo$RF{^YqtN3@Fvp0TtVv?yx-tMV_q171}{i0B?fs}tnr!HrXPOC8;`d+#+Uv$
zQz46KDHP{Hj8OVEd#NO~AV$EJX>}6vHHH{4(^JciP!NDQcbQ;X?wRTg#gA$0yOdo1
zAn-)ls8#qg9&2@dJGkC}Rv6Yuz1x-Q;)Lk0+{sczpCRRgCO^VnBVI9T!)@c*0`qI1
z8Hvj$!5qrscTv+gk~bfUja|;v%7SH=vVYMs;oct1!{<7<rY@^kPWZe9RYEKaDrGGo
zElfM7b9suiK0w~Grfz3wYkeX5aq?-CNFfp40a?Xr^0Qn+NWq&KMKxVfXrZl)OZe#g
zuH&^^H^U-nRGC|w5Uc85AtODKfsWZ2rwu$k6Ah>p65`RmfCDEl%k=h+7dK&IWBB=d
z7*}^@^$|0Z@<Amn_3>RhHF|7+Ky>XEL96Xgs$xOyOJ;CFy?FJpc)~cR+UoJMImeV)
z{4Brs*P_A|2931VjG_Vh(v)6pi<|Q5_R}jnlauiJ;P?ZJn8NI=60dKVIF&ALA1ZWG
zWr;(&t?7SWlkjqvcK7iS{p7Rm&qeyr-&ckb_ut?%?SRx!7MOS{U`~BRO)LA-*X`mG
z1oQ)XW=i<dspzq8mz=qC;sspRX@0tACj=5*S;Iqhct3y1D4ds3{O+MtN>dB?r5#fp
z!_amVXvI^)Ff1H(8{Po+(Bva;dEF&_L7d#rd%kz8doG7qY>4pwQy8YU8X$YC>`Xht
z(lOmff-B<j_g?%xKt%j?{;xYGjQn!h#|zyRi<my4*@6AW&j-pXN<(3Gp?%*)#I58p
zr~{=NW5@%)<IQ6udC)}cRC&04&Q&*Jp41`LjAaiE%&)%sU&D6ItN(kzh*aF`Xl5)6
zVsQh#km2d`&pjer`CTx&Q)&$ebbPjthHu_57z3=SgSbRFV^({@Rxn~D#5$<~wk1Vz
zZ|i`{aG?vHtn!lPCKh8y+z=5ijCOf4oS_4ZeK9S2HiSZ?)3*mxzh`02<>_X=`FxKt
zBTO1fo6Y#sbCdAH>2wky)E4@VNJtNl^}ySEg&QSGuCtxw-u*ST(Uve6o3seL)6cfj
zlL3l?-YFkH1FQZEnOSxZzDnde;}SMqNAMPBZxm#p_!z&+up@w;%H%3-kQ~a1%Ajpd
z1!!Ov?TzrOqtV{vz99Fm_>cd|BcG8$!ua{7{Y6!F<cBKEUuKN|_8O$7D13N^0rjEX
zZIrVrjE4jDmmo|Xh>=!1x&pvur*$mgd@=u58Z!Gnvia@Y&44)?^HO)0$BEtLy0q0p
zoaX6i{?DEGtm*l8<9>o)^(58~rx0Hv_WBp+Xa^dzXXCg!B64tZLj?JwJJ6#<bIaPV
zK6U+I{@p`FW47X;PUWd$3}|c<&ewKShFA*?WIkowb_>8dUPVyW0KK0DJH3r}4g=eY
zg5pX{es@1ho3Dc@zb@bjxsGkCI+Wmx?^T-}9-+u>AgHm7YQ$$(+h|jn+MkoqznuVe
znOfRD8twLt)b+ckFqT5*;#UHwC}%;I8mE;AQ8LoQB#a#`IRQcOY(#pP0|k64dwvU}
zh&oB1QTF;X?hcFq9oR3YHrfH^hCZ+`M<JZPr2m<Jb<zGmtp7ol<6!YHkI=;g1XpD6
z<S6sBd6<q5pU@4-=BdPpdOn2Kef(Lv&~uYmaRm{H?;5@Q&@e}O$l^#MdH1mr)?ZL=
zD2~zJ&ybm?%~e1sAGOj_B<x^A^Y%0!1UB>6$$&(8>dK$7jmQ1PI={RzpZ;b_duW<p
zyGZ9YdZKbt%0N*4aS;#S)LqZA?Y$wLB=>OcML~Xri?y%7x5uBF%IJ*U9i7Q_PXf~d
z(~7w}4+1L`b_Va!%Al8^9&OZW&j)n0)9wnMOTQgopM4rIF{QHKE>6|oh$hYaTQBCA
z4qPlW@~`4-N-JrTgwC``-Z+i$+<U^HG5TNxoObA?)U;aap)t<)j9Y96&Y_t2#rsUl
zc|iYQ=%0SXd&~&i3kB8QGn(*;3;eUp)PwZ)gWEQlZyutEk=L*b)QojFD)3=uo^@y4
z?E_Oc9z`#wwd{*Jv*^pjf$vS11e`#~wcdkLlK=Mx>~b#ehpI;`W1J^an0k@P+anf_
zGg)`~K0QgMmN6x196j1bpM+u@5|luTg+TS8a9n`gt2Y+$1_DVMK2h)u>ofd)QuEBV
z*+$Bev?wH-Q=2<c8^gZ6m(JUSMCat+l)dcNfaEh_a{lpBE%z`<1i5-{(h$%EED8z@
zA6k51@g7I<T(7Tug_dm1_a;_i>sa25j9T^x6N-U3V_4xB8YSt-BOz299o<QqSM)p0
zz})YPxTsYtTa?d{Jj}XAo2b8xZB8`O2diFRc~adl?P8<H0}pO*pRuOjhkQNfd@Bqm
z%o))ElJmN};FF{i*rNv31N>hh3B(4a=Fflau8t{Ki=vd1Gd180esc={dk^RP6Xj0@
z4Gp)Z7t4S7r18i9ki;htE1(Nr@YT7a9U7by_15`Xoz-n2Sd~0DqNwFIzUZCA&FHor
zQaYL8-;}}Y4pGK_Sb4oW{aw5p(d&<iQBo8&H0g;rWFPRxF=J2Fx*im~jGuBPy%9_=
z^Q-A756j#lnT*GXVHUI5_>=eiBx<iL(Yg^T#0Fdckww`Uu()A~G2zB%nvZE~ij>Qg
zBNK5GA!{Yy<+F2a-H0g@`15Pa$I%siy-D)9tApuJ_EW)>_PyN0j|%;I#WbZq^E|jy
z0>UK*lxqN-fXSpr<#CP^mKjl<r3j3b(*UTZ3&Vs6bE6UFZ)*`L=n_b@e$B0J?=<D?
z=rm$2!NA_vOIvHa&Qh=3-^!#LU@G&t1uu4`hf9#z+^I^_xO2yQDnw6VjH;qN)+(fQ
z32_~g|ArcUW}x}6x|EF&TpO_tYUUqHr-2GT=$E>Y=9r@Kv?d<bRK8jVi~;>UIiq{B
zrfZVn)@SJHI>Mri?Sn5B*OnS<50ZBym;W~<;nA6>xWQM~Pc4nZMv^ncAB2SQZ?b4X
z{8={EeD?ZS{-$!oh&y@i5=wd%?dIrVyuX8Pva|kzy(+f@$3W;Y-7Gg8M*Ile-NTHp
zCvK3e4>6I7e3#Jh2xY478w~Nlg>xLbRH*c@GtE?v4~OMw`!n5<eoG}l_(gl=`YH!J
z$tp@ehpRLm;u^M)zvLma6d`1lGa9WpC{!PBoFzqaxwf>?z@4=Ja$*Ztw*XHUR^AqU
ztGk2kCzh>aX4Yj~KeH(1S_5kI`i~0KBYy8nOVQe712v*W<(H7jjU}uGPXCed>V_NS
zzXP#&41_hA7+3>YiPZ81@-HD8jL9wUw3dbJTBock-iS1a03UYs_rDWoFuDTA_U1!S
z_|oC<sdW}H^PI3-s^*WS-xj{52#-Y?&gy+$1V_@ex^y9grs)Ng^fv4q*cmBiM$ZWU
z?ND=wZ4izIlkcG}(LT`uCjLL6h>W#FEmBBqMMC!o;6}}Bjnc8yv%b9oONiuAB;!KI
zODrjKXbjRa*mv{!LiN6nVC7i;swd8A7t4!={p{_5xZ@0As;eM3n!PK@9c86##@6ci
z5(?mKyc7t|XVJcRdy(HRXZ_{9!1l=cYJWWl4cT9+AFDV4eFgc9BJq(yT$CrE=I$k2
zprd-(x}hyGlsx|%PyFEh%>&w22MWXSj1HWU7B^Dj5h$h=7I3+(#QgG`ci9<v?kz_j
z%yqSaJ!A~}N;29@{K_`Ijyc$yS3QsMpL-2T0NTuz;a`Cp!wtiQOGsV{BXHkYKJk>I
zR8aPpzB*EjPUd-r7WcFuKh#2uoTy_ap(IFEraHp$TB!d0Yypp>@qtkdHyXSf{?g_W
zs8+rXZ;HhfXy4hagXdowEnYs$u=7p6>#aUaZqKmGET_EsL;9Pf!SSj*B2z$&lu&ZA
z3wlM}GR)t0_5sF7$f2(26Tnt7Gt|ZGm9RmGKApLWFsaSu0evJj#<ewDB1C>E;5~-V
zgc+ROBlI>S_^_=FHNV+gD(gE7o6gD3`gwV)vT6oLoSrHiTa1>iTM*7f&BPrxx_op%
zOft$9iKO20FS=a4`k~ovmuIUp?-4or%64O*f`)o@?><Ah5ljdypT0B1{ZE68=o;!w
z^m%?FQ?Y23I*g8|JY116vPYa!!*1}p!@whuE*CX<4S7z*XNDFYUc7Gsq3Uk2_34|F
zu%4dfBE7qsqu#Le6$n7TkbTDQLx?y?Ah#kQ@uLMIkw#(-$nbeA+dXja@eVs*MRoNo
ztMQRk&t?DCyxbK1tU%HgRvJ3{@Fa0AZ?2@K`&A~VXL#ysmc8As$BmS)fTq$Gop$=8
z3SU3hc!zZ3PRUru8{-H%sbPE84rD{{hM&^Vuscp$%G=6<1ZOyn5onL^m|5{I6tjx&
zKbQT9G_87)vGuE_o_2(Z4-oO{A0b%EzkW45tdycMS#VV2&OeVDmAy1dTWk{09zC&d
zqFh+vah1c%5W8PRZ=24f5D<#K|LHc1dHTpem%D#tQL=$&<;nXRY!QpfAtsg-pHRY)
zfAQdTsmc5Y3V|R?HQU3H%ia=pEfK9;44CwdJ0YHt<6cEQ`z!lPHJKO%LZoJDJA<SI
z+3jCg{m$V|sF#$iMX2(^Wl<TDGikl5dWiJc{0}`h0o?eJSQc1{2|AJ5#T|3K*dT_#
z^Y&#ho+KP@k)X1)<>Y;6SX)1s-kp)!aooPQ?0KlIeSPz<l)0GlL^Uph5&-{$JoeD)
zm<2Bdz6bgTx;{RmD|?v>`HbX?$EV3$UZG60kwFTkYxw)3l#7gdmpU3_DUF^ML1IFh
zf%m<`{{2u{XZcn9`H6K+XrH2<yqObDG>JUOoFeLFmz{$A+U?i662w#!(;x2n9va5{
zJ$ddI*@#ZQ2T=D?ml$#bF1{CJXpD;z<I|nH48<c}^W7_f|7@DaZlW<fgYGmr&5l~*
z=c;ZFa?J~-tAP%6wDMV5?ZY>N)E9gs4<2uFC$5C{+!hpV(@Sjb{_8It)BGRX_+0#c
zoa0j=c_iH{4!M&9Eu$c@+*$sG4MKd+<%w&1c~43BAE~YoicDU-m`Ib+Kpt*h2g`Ez
zE@)NorC2c-b(|p(mrOPQ#dH$LpHwG`NYDVW-}u~0Y}HNDIh+octGl}PFTBAYdUAbz
zudbmLu}PUc6AyL2B5&JM$L6P3ygLLINkL{r|GqzK5Jq0k*T)F)nx08sd*PLDFLzJ%
z`W-1<AEToGYF;2W+G@61G+%AppctJgiHuRNMlt@~rD@aB<TEgi_@ZimFAo3j9=;wc
z;`QXk9Ow8pnY2++3WMmyX_u|WXUIBXZ>gHDRuDVmM#DoFNI@v_>tqr=r6#pS#ivhh
zInX8m36vENh;544PVa6L`{eo+-_Oi=bq+z_Z9CeNFPMe|vET9$Pb`yjw?z_VeEGl9
zcsmeu*<hOI4|Qz4kVH8G6-Y8ZbJ$jUe>ee)fg;xDy?-H%FrTA$mse*ekmSz~o=8<P
z4BkRP(G(A{A@E6L*ejMqea9He6*y~9ABO>i7a9m=*Vw;f+MhiBj3apaMvV&lohdrU
zw^afUs=f!0JFW_P+>5^`*Mlorx-OYQkw8wOY`RM^I2#rHqz5!nb!PIh=7vqpH6xSN
zOIRtaFWp{KGMPQ>Flap=8=H|ewjHxF!SM5E`n0Ef-F&N=B)SzM2)!MtHzkr;yXR9e
zbK-f?Z*{R=4$1f8#TOOd?RBZLTkZLjnT}2&YYP?_no21>+ogl`%XVmx0Yc@h(9Chs
zT;o532M0vc7Wa<sKy!3XsxX!FKgPvJ0d8v^YRFfic3~|-7qKVV_T>A#xJ9?Aj*FvT
z5f{NoMQZ&n<?aEfUZ=;>(RSZW_6{BJZ+i6Hh8<4!rADm8z_^BZ{RBmUJGOj$O2?^>
zj0~!M!NhwZ`d86RyzxM%!VNU0X#FkT5gd9b-BdH~dY(!gkkhi3Yhp?)v$69ayjf;o
zQGcINoLHS^viHb-L19f_WRy)hh=L$TJX4sDx`<QIBEdpJA;g?)aQvJk;Q#2+8sGAC
zLMXntPMib`1=zIyDUS6^EtWe~qlbYzCS7W<(@`8$NY#7>2kXg@KN?`Lc3u31Q~Ric
zbgKZbdjiKoYI1TzSm9P>atpy^G5^mI&>%B)M%eHl`OTfMm%{yv<Q{TK?|Be(&IQP-
zrLN|hC>m$=GQpi4y;I#$q;b$d6n2E87DEMHL6T}(8Cwr*sdE<x5Kw&!QF!>6O@m(K
zaSLKD0OEA|b3A12X&Cf?JzF*4w!D91jg92L&@oo~a6l}4`-^S1na|1gV1{VvvM?Aa
zhqhnvZ$fz<9INs5kfp2J%&VkGY8=7Gu(-w6K2j0~9S`T5lSS84b(-(aWDmPm>TS0w
zImTEC9MgE3r>itF&Np=*JwKEjgX+K6+#i8-ybt4H;0>5=Yb=1StwAr-xIbHIoPILi
zfK=y5iAL!Bz_n!wcqUQvp5Ug;C&&;)d6Nt8ifVRQJwdQ&x>_<;Ly98~>evGYN3v^s
zA(-qT`_6T~LV*+qB7cXwW9cA*zX<?=_|I3FBT&T?MM(t<`@38JNdi;>!1KqGY(}T4
z_3dIVseWfT52B)q)|Y1=$uTEW*Vc^<{{?w@<EC+&o3<Z4S2EK@t8>?@MHKdD@}M<=
z%h<%IFe#H;=|>#gSQ%>$^qOp^QmqsDs6#Yo(1kvKd5t17<BQ@=onr|Ri3k9%HFXt@
z^fTzRI;?_^6c<3J21m6As-5Bv+}HcRVEk6HcWNFH;mB`mLunRXRUDU!Cq+d`dH&Zu
z&H45J2PS&H1_3s|Wr%0+gP1MbqSr^O=F^d`SbV8c)wXp%4<aHK8eaZ1D+A$>=`tun
z_QyRk?+GN1FrqOaeS8t}c&3c*KA0))FH<t^ap6E8QO#QgrSSL3Bc^eG3aBm@nAj2e
z>hs;B^<nn0U4(T%O2*Bgo%%Wr+|(93P{3O2niji4Lp>nuV{#$=XyGvMb&KVq_d+Rc
z_6iPE7Iz{!l2JgWZT_?vR^kO<O>G5(kQaUidXR?*bk5yf5o~n~z8fg|^xUUbe96I;
zI(IiT-{ZitZ^8Y1#P!y0-Z(ge+ffz<M9)hX(^CSfp#5kcH-;AZBE2NQXMHsSOR7-y
zc&vS(EX07_um^TlC&(g+1^l1Ga`J=Kj+hu}*wQf<&;m1=B5!QwDFX$6K{y(FIi##t
z{jp!(9TZfd9&xqGMG$2>UlX=9kcdf0C~I!MZNMNtvvGU2!Gs<6a<iB@j>(o6t9PxY
z+bVxNA=ev0nNizh;S6-B?o8Rm?4zo9v9c=1#&6nUQG9#_2?z&(AyAHaDd~WZLr^#T
z1OcVwA0Rt9Olx&IM-d0Tq}j9(V0;m?cEJ_2%J1$t)LM!bP!yZQBjxy|Bv`^#D0RMe
zxaCU+g|~OY1+uN36y--fAO|q-z7UBc$f?m+w5j3sku*G67n3u4lj}t@rjY=%dU!DG
z_MH4VvC@wag-V1kq6MNQlWR=F5a&$Mx;CI1?GV{(%K2+r>PXb3?~4@hq7U=)(mffa
zTGyZt%2GJ_5#?m2syhUi1~y~XkNA8WB4sj};jv+3><<!8Ui}5<ptBVi@JK+_fm|?2
z-y}!y=br_&$&Rsj<Pf9N6YN%pltb=(3!N(M4)d>{T~|&ugalKzXaA{4k17$Ao-xD%
zJ)^(G`UVGW!}fw-W=A+-SGrYM3^H)MKuTj^>2D>fQ~ykN^>fkYRTv88`TX%^8b2N`
z5JP<=nX5kbysUKD;05#GKliut$|)^cHm``Y-aQh;@kgN^IS;>K;Xem!33@n5w4G<=
z>ws^nnHs?Mm#1MMo624jK7wn!$Oj-bSFA{h$JAC6qKx6y(aLFVSTMEfVhd2>Jy!OV
zV+f+A>KTfo_p|iY*V^8V-@$0|swB+0tEv$10!n~p{Grb1@WG1%EV#tFQOarEo}z&M
zBrruvT>caVk&Py<0Wxpv^B@x6kW>4sd=mvu)|bMEWy<8FDmghhIIJ>^j9x}qu|=E9
z=UaV4L(=Z9W&gl{4KQ78q2PUeNXHQS-`k-2TITQB&iDEveZAR3=!J3{#lg<^G}ab9
zLbcL`cRoBch|2&RNNHjH+)xqlz5Wyan>RM+Ypm6BI$pb5=Q4fe^8L92kt_TFU>2Fn
zx`h8k@*g^FPA_L7V$S8v*L>eHRKU6OMId_HJ4#+%F-6Dr42gEQ%^t<iS5S<UG)BoG
zi!mP$G8P1>X8>u#fr+(f?wm!Lg8{=xdMK);8nDCy-is3V&3_m;Oxj)ZwKjaNyYJ%T
z;~N?qF-S<{U0pfh5K*!W+Wmo`pr9;kD!(P2xH(&i@);gMMcuO$p$-9$<Hh++`-M8Y
zxi&ume}B0{z5lvYoM5#Dh>tt2Zd&cotV<Cs406L2ifz`*<)|U!fvPhoM@Y6m`M${M
z)kx=pk-THSypzyom~lSOCWJ@cVmRWL_Hr60jo47~#a!f!-qFD_AVrXlGxJg5V530_
z#Uf^+r~g0)VRsU*DAj!#L?n;7H8nr`AT-t1Q$-gU1O5Hv3=9!R3-uRclBA%)RzZZ#
zPP;SW=&e((<u&|Cc{^15!E-EIJZ|yDWY#q$S1M7eWJ1psu<UC8w^R|m`K=MawmJE+
z5mwiA)d9J6=_l|^#0f;PD0-tV@aoSyh)-P;Gh2rIK1?=ozOJl_{|M4ay7k$6Ua%$v
z+T8*`&Y#wz=Y)PuZTwvs-)_V3=|D+EUwk}4jM7f{vynD4HiFq~<Lt)ziD;@nu_a)~
z6LBw;sV2bPj<27?Ayk&$qP=4?xd?tnpFnJ3ng9*Ju*|GR@N4|L%kCCr5$DtQ>4Lnx
z5MjT2$IIO@Sz?)DOacPyF>j6a{+Yu!%O<_iQ^LpKcDgA1PiLE}baBecu0R37bSg3y
z_(lZRTMj~$=f~WZ_h_$doZ~l-H(`=C(d(>vRo@2K{K=KW^Ab<9%5`j*Yzjj$=T!~?
zK!QXu*?a#dMF}hxfc5br#^alq@V!Tf*6&04O|S?&+sMy)3MJg-@Z;6i8Sut?PX8q{
zQg$+3^%S9|ny9XC*!2eOxe@mYS>(Nt8LJoxegue3123kp>x6XQ)@zR4XV1g;4vft$
ztP_mZK&KN4PhJD^NBl);&593s3V`dj|5fylX^Ky;jiN?`A4<3zizoIxACBsiS&d%7
z!ophCY7d%oU}IyOT3BF`l8&5j4OCWFH+Lx^J2`>3l!un~`)^7!C8Q{ci9ePvAoe?h
z!7LZClqLoo;l<C-N629^SH<ZBWj9~jj^7!h%M;LY{|R%RReGIzvDVDF1w6BSrhi6i
ziazm>jsY|AD*+9pqIH>{8>g1O#+Zuqi32D>sJlT?CI?;=2~|6I$fQtJ>3#vrReq7(
z14Pm5)dHpWm%Rf<K4fY-$eK6X<nPrF&HD4vXX-91K08Ukf(D7os@KUf2x5Xp0Z5;D
z1B%i5qkgErX@1LAIzX|V56BT|wZ`;i3VwEU^=x&i-<yF=Z$-WL6IZl`*`bEcw->)F
zlKUKaxLDF!rc-_0*<5)O8dX7uZeOt;7|L;D^RTv*LdOmor}<lNmVIhtZjKf_^x1yN
z<o@mJdqYD*^S84znb&E(8%^oEY#DxvbNaz{szaHduC}$AS}2%20kSS^!`+;XpZ;mZ
z&TD<0GRm$ExPu3<F?C9Dq0l;+_ssv~r7wkuVGZ0===~uKtG@yr3$UW(#maR;^~X1q
zY$jdXtzhtmPCNJV<cuYx_2Go<NoIXJwDvZy;oCCL%HkG@kKitR{)ick4#;U$3~Ciq
zdWU*-q`9ApW4XGL0~>~K4OSjgx{<BpO>s`IhRW&smY9c6)vzAxiBYapw>_rm9}bQX
zV`{$MEv@l8^k2v(0}l%_!B#?`1kn_ERIIuRja#mP7;$#q^dBrdCBVw1Zm?rbv%ClO
zLSDsJPE>NKl%?;dw~P)<B}0|Qh6uAjq9cxi2uDXp2XhJRPYo0~grBUmlM)gV+S=Mm
zDQH0L1ide56@>37!FO=+^7D_E8;A%$o_5aH<}xZNDRsn%K>z)ghMeBq^e!)Fpb_(*
zt_Gr4d7S7+>=c%in6CW|{*uZaRZ+oeYiIZ7^=m`$Ci^~~_13aI>w#X!Bew3|-g?V4
z@79d&2wVaVyLry}I=hI{Ql_%9GVoUelQ}G^Z0D*ZAb#aEz_&FKjMe3jTQ)z4Tb3~-
zyv35M6jes(0>|pU$@G6kUkoUPq>q~$TNrK7BxNG@4xF73eaa{F<^lrb6hUI1Eh(ac
zQ|ZHsl3|I=jGA_!=;N7>iE{m$viJuAy6-n7gXe5*u+r5o%$iuWXiPQstWNXVJ_{Q>
zrBH0Nn_y#Lk`T<I7FfAEGtGTJBzx%XDEtSMi*gAM22`a=DSvCgABE@&mCkVTGj(PN
zsytm?zoE4@CEc*bY{`J9=72Mg?pdCyybHtvS!Iy+lg51i^mJ3kjd?YGtA31rd9Nw^
z*;b;#c%_NtGVtuDtgKv6P+)j-yd-!%YXT1s?+DzUt|9K|w)yfcFE20pKO8+mA5Kio
z%#_sB#Ep%~GyU)B{O&J}78{&Y)%6QdCkzsp^}-VqaUnDc+ldDLWK>kaDJcZ(92{D;
z*4QzE=dVRYMR%wapZ&iu;aixSQ<IUAExJy~J)RG+P}0z-sHliZOTR%vBU&9`X*UHw
z6^JC@D(Y~d)|RNI@eB+M)T%Ol1H)l4Ov%Gj-{|oYZeBahIi0y^kSwmksldgD>&sfI
zoqFngSH3UfHvd(cu(^8NHLG!=Nbh8z@NEf)+H2@fIsT~$KQ~%kByeD_YAvfEF`#`h
zU#=Rj`d1Fh`U?XY;b8I<SbMkk+1r;Lc<^6VJ~?e&argi>P`Ns*$JJ>g#~1{UD|n=_
z#CLC%__!#wcIrJYHm-Fq{6xpm=QsC?n?V^AL2b7%-r4`tFms};Eu|~o6n!L?Ki2vI
z<YIhpy|@zWG|dBcFv$wje&}Q9x}5w=Km}i9SmtP5n{vF=3|Haz;I6Eql4`aLfVUtB
zJdo3OzTm<Ld0&3u;HVyETk*X>1#?kub~ae6V$!LLv?*`Jf%Lg{?CcDu4}i#=FfG5Z
zBx=Hzo5Q&4byZcBHl^^GU`pER(BNRnMLU?@PWOw;l~q+gOG&*Z1s>TXqlw5UC~}sM
z_M4qv7L|-yqfr{4vILk%D$`EFxAt<rVJv-1;s9rjz0j^J?xyvm_5oKGJm^kG?e^d3
zDFhQ=XQmak&$HDkOZ)q!hfk~b{(c_n%J*9pkEj>MKg*ZWB}H21mI-z<nJZYU<c6_L
zZYC1D3LXJFFGB1xOI6Fh_AY2<EIu{+h;`^m{_V#D`Qns74FIB66z@}s+{}{+$+ZEj
zTp9;GdG$XWlB=N;pzXh5D+lTHwT&L~svZ_S_QpyJXsN>Zp^Ao15~Ri_vk;EV2B(Ak
zQ=PuD8>D|O1HI_kV|w@2pjWP8vGsCXwl;=>oIFX$hr6<}Ql0AaisuF@CjqJf@Wcg<
zB}w{1T~rJdhlPa|+QI#rIcPj6Yi4fn^Y-SyS#&{%vA4G$Oy!A*jEtP$Uj=>i*GCJz
z6-J#fvyCn^$hgdc4jG%9rX>EiYx5!sX}qxw4ZQRW3~b3?@K6xXFD~XPjNnxA<?Sw=
zxFyrHw6rKGDPcxyZ8VIHjcJ*fx=q=qhMo2&>Ds<LLw&$aCLqqSB$0zgQE+nBbi8*A
zC@P{S@!G_Q*x^b&xXqP*F%PY?$5)Ue4w@&emNU*zVToR|{tMh6x=Wk&)+}VUMF8ca
zD2RCtg$&5LUhfjP7j~Ard3Gl+yFeb}gHKb)|6<%XDOEOqS80A9hBv;|LPk?hqo}1@
z#+Zk*JzDcLTWFLm;tT0t!)Cj6FAyOUIt+UQnW&+7|5JhFFHMyj<thsT=zas4WjEw?
zEfrJ8bmP}LLhu|hvb!H@k^qlR!ndn%U~_>(+a2}GHw+XC)%93%d-;ZF*GdC9k}>8(
zDO`fjj~8E(*|*Z1sx3y!{2rHmPR_b871AzRByu`BI|Ge6L#C*W2y^(>AANMbQ^_Wb
z*4ZswY`;cDrQ+oLx!};=I>r{hJ$A@M@ZT}+Y6L9@$|RQ__4e7YBo^B?4K$10lV)E?
zNwl5`S{+D3<a1Q!xqOpf0S=UOLSJ@rJANsHUW|t86^N(h)j%nIBQgL)34%_f=ee0(
zMYxB`b8Wl=+svA~zYll$$2{npYi{Kr?s}&kkLZr|2bWI9C(1*fHlmQ2u5+$8O6fWM
zFC2q@OA(~m!pgQL((efE_}-1%u51q`7;#H1G6u3XwQ*|1&e#WhE3pRKwi>RAdz0?x
z#9jdOck^p7*j78N2unl@)?dki;qqc<#Nz&P4;~Sb-SgBiE-sEB?WC;vxX7i0J6*na
z#sJWJ*x|C723JeJ+yeS|2h;L&y|6<i_bgwkezE8>0{bH=$y2_nrl~0f0T~BowAIH0
z+?zAR4SOJRJDh`;7cWYJPH`4wTh%?ZDGCq<&w7WI<us?41NKR`GF*P=tM=4o_)zsc
z|09<H!v@$z29oowrn!1YYY%eJ97i|AB#|?<y15}?SkDbFQESEZ<PO$18FnW1;ryqW
zR>wzeR39!J8E}{1)*z)fDb0KX*{O+sa);=ZOCD%8d)+9V{BsK{r*drla6>{drta1Q
z%`Uxsj0%T{c=+;8K`BGyZ{N|I&5Zu+>?{-zE4F;{wVY_jEQ3{ttkp9E?a+JvofI|`
ziNFVb25#=Ux*=Fv>8)x%Yf{fuSQ5X>F_-)43_ss%!hMOH%In6K7C|E7_!U@oc6Mt3
z3ZAnp>uvv=A?$n*4XgmqLiNku3=?^B8%rK5Th|K?`5<BY{Q2|h-rih^Q$(JX^S64B
zljR}&7a#|kYx-<~Jdg{d)kdAPJ}8v&!_dE#DUB%(wzPBmjB@;`YxQCQvqn7xoZi|`
zZ~*qM+F!LxU<b<}!|RqD4i&x9tySGpGX!n1HjK~48PoET&A*gt2<iRK2?jv>aoNU$
zl9pQ7-0v1LcgKqJKtga!qSiMzyMB;~;jx>(e!ib+X9ZG+D5<E<z+-c<m!J8H)qD`o
z*~P`lqy3b}{L`mT`%9tPrPSo)-+uq5luzSz8g>B;L}bcmbSB#_OHxwWagGLuhs*pH
ze~BSQ>{O__biZK*(NUAf$*_(?Yxm>b#Y-^s8=jbe&}4Xi^Eqk7H83zZfWws|*Lx0r
zSDMfVm{QV_g``EAf&;c3o!xr}Sc2TsAF`^gLU`8J&yBbtE`otzn*nHH8l78MJuFs*
zpXb-hFbFJVF5Prplm=@`V;a+|yIbEh8c3JdphHdCtx68y+yN}A<U3WIc$1UkAW+(b
z)$1gPf7e8Sox6q6Zi{=@HwV`A9Ffn-Thsx`zeujH+$e_AWrEYI;o`Tj`#*h|ZfP;=
zU)OYJj#8jamNT!4{;iahBIe~!S10cdIs!3gOPN3t&+x6u6*L|H`#&2Qh9nEF?>y2v
z#(V?Bs0jy%vNmXTrts8iEN6$~j&^ru0i}qBBCt;kq=UUQnZBP44L`VS4|Z5`RvC7@
z0%Mul4tK(!4;fJITHCWSlVyFpuV>lDJF-7pNwMPpz+~Weg>7YJ1xEe~>nZxSyKTbl
z!Q{>~K9}3$=97kvXn_>B_P^{y8LNJ4!MN5eKsJb{8oTQ-KIt}|S50VM4%vWH(SMAu
zKfctz9J5DOofe%(h`!Ccw!mYAN8~ou#HFv4P}!fMg-?vM<4kI!f>GyBfL6|%146Gi
z<hCuF>5_A2-SlxDBps%l={GGpSu{)3s97s|6eTkHJI^9MH^kW8Q^Jmq&4+)-1*8CM
z3{_QlEFZCM5-#g(Y6(bEQ5ao93)@wqeGy$7fr)R<SeJed+x2oY?d$K^0pmU^_YWpm
zT;m=XlMj5ME{NB|=@$KE2OJ(5gqi)&MgX&VgEJR_%A?iZ-a>i+_SdEF7RI4qw~3mW
zEC2P-(qL@x0$pFTkNGwQavd2@`~>h@1%{%{heTE~)!NNwXHDuX_>&YfEc(TKZ=xcK
z6y&FxhLXB4lK)-k1tIu)i42dK10Kmm_e{=)j;U^@M16(JXbz$~JAttL^?nJZ<>AIc
zLOaQutj>mI+Mju9H*^rAonxu#Db9b&fhJtjg;Ll%|Bkm;FvB}~0|-_3=xKp5qaBw1
zSFgW)c>PxnB+9mPUB)0#Enw|P{qZB9yxbx9Y1fsXS-<7A<IHd4va))6%9Ew$#0(*y
zY5M5}$>yG}F1#mcFO%*&2j|8#YPpor-_n?f=p=*aB9DJcw5mN1O6xtZ=PdT-sx5MR
z)e8tE6i)rmx^c}q+lo_5N{0acZyy9b!3;{6;5#txMVh4^PtFm$K{85+RQ_SGG+haM
z>j#8L-1*Y}+D@w<YBzv?znLodbg%JmYyE2wlH6t8gstz1N4SeU)~$m%0_^mvB_fr!
z6tlPoeWL6GMPrRfT!6nF{`FRMEy0{83)T)!)tU%$*pi$2ePHdYPsG3<y*Dy^riH7Y
zVELSq51;l}qFLs;N0Ud!$psrE7RGGc4pVBQ#hw>J^o!wwQRWw9!zc+6HmyFieUJMt
zC||y)(A6Y%nI`!27!yh4(v4?gBTnmbQ=`6&PY0o<8ji<>6(Swh{?E7OOk9`$U7v?u
z-k8Pw>!w!Dg8nwY=2~)K>k~OqUr6k6E&90>H;n(1j#}~Eo*L5Z{hYI^c<%VWHFfUM
zMHzYu#e!g)ZcF(sjMgX4u>}Rpb4?yZ;1cVt3*7rOH(C*{P%ejInzf)yQszUsT$f2H
zI=zEr`tO4@8_sV*Xxlxu;=%4bjm&AkQu6YL>wdm(k*qdS90DX@R@Qp>tmjJ~T_$UE
zWQR-mXloCQK006bxcfGT6Q+X)<C`NyB8N_gnZD`4{7DXg-qC;r7Q2lE14F595yUA1
z`WNTt^g8-=wpyi_`577Y)ErOXBsQ=8!u&}7xEW0+_3o~&phA`W^JyIi*b4tA@4PR`
z$}|h_>8aTP@^2qFU-PFYWD0vHw5C3PEK?jeA6=-yKgE$bXqLn&l12W$%oX~5h~w^H
zL`%2PGr^;6A)qn%E9UfT)Siwzomo(8-%E6~@SN9?34*3dp!={JDUpLK1o1!P-@uWx
za->jpiDSRrm^;bcRHC5zavH))^CR&B;n@2PqkJ}Cse#BqJ@jj3A!cIh4(d_Q5SsUS
zs9$5}ef^1`X1`GlPoyJ+I{Lo;6>5pRL^E~r(n>iE7k^l6wf>*aO7aV#f4W<DG|$d~
zT&1au?Qy>SV_Lb$Ylk=8r#|HOPg;kM=0{N`H^#Y7^#pkRYX~|<es57+AaSu=@rB>?
z@c+pHTbu`K(2^0eOYb`W%D9{ecc}-~8jQ#*_<Uy>wgT4*F5(1AC-WiEabt+*emzfR
zMSYo9q#G?4f2>I6L+rOp-Mx@*m#AOWVs}(5%(xwYKc7dNS9&hiqom$3ic&PG&I1Eq
zf2HZx;Z9R`!X%!L_^HLou6@MG!|8tuIfV)V3t6l@;?*>m8E3C%w0{zf@8<G;O@Lca
z8`FFTFZzyzBolyo%{g0IS_U8aKV50$$){`EAL&HBWqr{5zp}MNCo)Tm;<e8+)WaW_
zhlj@klp9KP8{@n#w%I_K?0vbbw4oU<LzN>DWjs}+24?7Q2L5-!VVE@24M1i%9>Xxm
z8Ne}qlA4;D!EbgKAbX&OQX8#R(25wl+REoWGg)I4je`-?<UDUL>K8g{X!~r<70Tn#
z-QN*$+IQe|UU*AtqLRyHhuA6yxz3qo^TEUDbwA(RF%pDM3?NQ1s}-tQPL2~B<SK{h
zLLI4__7@u4qh<Q^a^E}xz3t~1_HXeo-9m=1SvjHyHwqauZw;tRlD;OJbrwvN&@&p%
z$yVoX+S4F;_8Q2vsdse}ncs@xBHdoU2QOA|X)0VPY2kV-Q)@=l$rCde`59LsIIC*X
zbTBpuf(NKV>8sDi-|u(er<v<e9#{M3k9PeS1=!NDF?Xc;mNEc&)+O{o2Jo%jv#*3L
z(H17w9tP4@JW(VwB`&>9G=qyh5g*oPa<~$&CI6n=unrght=k%hyD0lq7>$N`uck$?
ze(JE0yIB_XDV(dTZoS8xTKQE&Dupjm=aq|w*3~>qq|U8*p!6-S`DW!|_nfkGeI|0t
z8bh~UY+k~t74j^Q6mEZv&D~P{nve&{w^>#2tMmjjtY_#w#FHIuI`@wW7m-yd(zZTq
zR0cEVgNCMY_u_Dzyh5mpBNXTMj^1~+W!o)TgQm0geTpmEpL~KG2!iN4_%;9OLKPZ`
zC<5_(`Sce-tD^%JT>LN8V&yX`ii#h>Lct3?2cDk*k2&woQ1tSemDN?lm!bklHY=*C
zJ{3>B$Hwk<84>CQWqR1x=;*z@y*p)YAP}}uFVlVVaU?^?>tcup6d}pwQaD8me?m#K
zr>y%%Mt0V_BM#?lk%K>V)L2a%?^=NHqeQ#*ZGbd{x)ONye-VTppPtybxWvn6Aelhs
zNTbWn?NR;8EeI_us*0e0s7<<(!NYj6-1-I;bt@-EINN1sn3s=_4AKXZSSo7jkVJj&
zz32`4MV~8S3Nd$g_c`Zk^L6B+4}y`*q><~V-%69TX4xF?hh-qm_0>*R{aN+yCP$lM
zlL?SC?aKCO3haID<10q)wuk1QK*4i`*A=Njo%)|%rVsxTCU+^_eQTTS;Yz=j8kA5z
zt-I~9oyVkTWGuGoC1JMApqQS{2t@o-7%Hss>bj>$dpMxwP3@>JrB^s!E-R$+U3zPa
zO{CLup#CN-^xwMEO7?_h?(ch_q5poK8R3fgYL>?WGx@AAo+pz{hTadT;C*wwtf$iS
z$$J&uia1r``qI<PF?UFXnX+o*V9L<?l>xje_{Y<8=zQni`!Y2;Q>MVRKdL<89nAK4
z05mLAgx&(OZ*KeJn(GPYU+dq9mrqr058eI5!O~b%r#AWHWAY=le{BZ7P*u!r?<<1F
znk6^dGPA2!n?deSbE_aA<vk8=JfqCd>GHBbtqNAy#^i6%lW6Y!-;pn=XY?JhfO{lJ
zSRz2kvA(wUH8>c7nwt8bDf^c%Un**Ab3whxgVH0@=a^5swy5Fv)K5AU@ZA@Y!~!@r
zHa4d0_-}*1Wo3N;X#uQMtvN2UuIBN*u&^+rc1^+I4XOVv@HHR+W^iaIw$Ie;iI10e
z6x0mO2a`Ik4rWgFdq+lMPTC&I4sUX06UU3x7zPIi`>#iahoeC`>jGruQ<tOoO!5Js
zEXeM<tEj81+aKND*vQw>(Lp5SjRiwUm`aa%@M(#awu2BCw-?l9Pi`Ci655iIlA86~
z?=UejJAtooNXJQfCo7%H2%?^+dv4zDzDz>lD`t-)v)tDDIafn=y03k#@L_!X?~&aq
z`*+#i-VA4u@?_Deb<%q58X8o`W3zj<<7@WwH5!v@X^ihv|0DR<@N+PE3h6ierNB@>
zOP^x$FdFMPv0G($D_CQ#NLJ4^x<Nn4HCVnbzXRZvj%AYQ<wz<o7HXA`mP5N5Q6>Mi
zW+0&DjaL?1m6ar7qPQcpX4I<qm(~=51)>n(DeZD#2iHjYeL*qrbg*`k+nWT+U#vV^
zVP248#`KDdRIc$ZlMx@W2!aHyiXXBR{#)#5Lo1(@)cuF1%i%O)dfzt-e4=+?ueBN)
z83R6O>>hmqA#Uh%ed~B0iCqfoW_7-3eki($(Uw~w(I>NW&eLvj&G^XVcs_+k_Qw7<
z?}zM|yo{HNleCjMs%%c((p0%~*$G-FuOXVFofXL=>(k@S=Wbx~f?p%!H0eRv9?3MY
zN~KfDJMBalQN3-7ibADP$mm*VUI%P40k3U*_ruw*fq`PBCJ0IVZuA2QOvTmtpbN2D
zGKK`y!q8<0`gi!9+}yS<_a@ZpF!w`5sR*=dt<}`^e+h0)u@?>&m$z9<@qYWT7#cUM
zOHD-&D|zzHS0J)bjkL^P4;(4Q`zC+pg%^Ak6Ts(_S<|N-V%c5{p)jNhr4a<IXTn<P
z1dpTro!--E9Oi&H<p1&Xm0?wNZMPuOC8BhQv`9!d5`uI{BPk8iT}wegL_nlFH`3kR
z-AH$LcYhO~_naU8kPG%+G3OoQ9%D>mu1+G}D$9a!Q{q)Dr?MMb=Um#Gu=dF^|IPb;
z)85=YvKV6U1f`Um3jG(%gU4<vn1Y|^$-135=92{GohY6mch&R6rTfwf<Oo$?&9}QZ
zojLqYd#dm}LeE&^aAQ-|XX^OEl5l(eiaeGYk}PB-^{bKW^1!yIcg-4YiFnB!jY+KZ
zLf5A(kTGq&*4uqyHUwpkiH}126~bb6xcTjvUlb39-McdSi0`Ko8#f<U(-gO-4uzbg
z(U)t}hZ)P>G!dT}+Zk+<HPc`V(-c%DEre==?e$Qmm!dz_vyXKM=H>Ykr^G8kRHqEB
zpTEVsiFou*({h&d;jLD(=gh6ErNLPzUM@4L%EXG^^<!|yBD4@mmL+*I!iYuOUQ~(2
zDts12ql!vZh-KaBF+n96E^_J+d8k+R*bosB<AtEa<&F^hivxY8kCksdJosH*U87@S
z7T4D$6cnELAY&VM{-T(eRufPVgJ{AI+lj58;Ni6cAV8zRO8_*lV0eA75+;8%EX}O{
z>4U4}>t9mb>FMc(_Um$y%)0$4qW<4QLL{=3mzI_u+G9`+pKOf={`&O`?b>xIfYxOt
zR6@PZjbnSfU>z_Yz=UymVc{)+q<xSu4OTisIbIFu<#Mmu@h~^sK&%7&#~8$%mU?YJ
zp6^VS_|MFIjAb_y0f09ln=uCXi>ENPZAG8VCgLYg8od$GK{hSeS9mO3*<mG!wKs3X
z&#JLY{_0?#RY&SzkDuXb?Ri$E=Q@liGa&(nQ&9GN?@?gb_gUABIo#T!8mr!V;=Pa8
za+c2qo5*(OWSB^y@2RU_YG{>TG34lVIvKvz#sALq!f8Fm|7BCo$VU29Cu_RIy+uzK
z#?(|(5lJxP+`IFDh$sb94wh$t{aOxHNoB{}^r&n~Fojv!tXNnH6f=k6@E0?9W&}01
zmu=nWhHIni%az4-?q=ASOoU7I_OqSFE{tY#e4KLDnTkq;Mqf`Vk#SL%mZxldB$G5*
zKD1g_GdoVk(YE=#*mcINh68bHX6NypOI5AEr{PN3#TD4RU9-uJ&%O&;*yZ<*Jy^Or
zpX9I*Fc~KkYd9f#T}PuG#))S3xZ-RKWhrA;bI7`#jtKE@p<fBeCNE|>a&vPx{$}AO
z@H!eCt#pY)yL?ns-#`h#3YZC+=laIR?a9JE1nj22go`F7CRp^FAJfQ0E9^RYcz6tM
z=csLW4RZJO^*y|sPwjuL^dc21Fr*F=dF<En(dp^vP!i<aW)8r6d@iR%71mW)TH4yx
zm8I$Y?y;~73rvg#ZUJv_I@xOm;p+YEJ8A-wFi1hDyI=czf<9>#LrUoH-V5cwZLc<Q
zHry066=a+Ys@Y$Sj5eaqgQq_yqm=rzTlRoxcVaekG}ew{sL*Q<H)|+h7{vn0nF=Ya
zrx%ktpdo2M8I8V~fW;txx<5x6f4LMe=k}uqIjvK=D}%cNg#q7vr!M{g+d?!)IEDLC
z6@sX&>?X?;dO~No%1V)7l66qV8kG_F_D5eil`c_4zfCk(@q=&9mN|G-S!UjQG5_bc
z%N6U;v>Xn!@yL~aGJFO9)2exhPX&6PTUp9z@)<X~Q<kIp7?TrdK`>@@$;AE&(SDzH
z4&M2OfmU+#rlO>MtnX`vw$@XYWKtc7Le`>=6vXO>aNT5N6W|Ah^{g79&Vi&YT8$4v
z_^%`;SrVgFy$Mq)Hfi@~QVd5Uc0o9i3fSHK34u6Ii~Eg#X-WyZ<qHSkixPN3R{O{E
z(^7q*;&{9S7w`WDsiO~*2cIjt%><!R#Ru60o4fAt`*<8;98urobzp=*@N|_tgh?sm
z2>}=+vh=b{S>wdiMq%jH1TJBZ&7Fe{S!<^+BdEV|Xkun7J;#g0>oos}QE3+}`w<8x
zS!QWf1Rs6;ck#kc?)o;UiSl*o%y*bTD*w{;ljFKVhxr!eydW;h`;FYE$>m)nhm;Xw
zcAmOitg}DOE<eu%Ss^*D^T6T(@-!X$cG;^@b?u!n3hfyW^3k)(@ZFF%-fwCH-6ab~
zDA}^Ex4RbXsUCnD?-zgh(r8^~XJ><EVJ}dySu9NWlq6GG7;x<4rb0@UpRXq?K3`m!
z?EF52dj318C+O%AA<WM?GNpRRWRyeN`V}J`%~n!a>mI+T9c1Xu-YrklGBt^J{+3+x
z8}f*`!nGYO4?~FK<XxsY$(^<pNm#MqeOvozaB8kcr>ajHe45LMN`%$A@bZrHGM{uc
zPJY!%-1x^^GCqs(>3lw21Y?%5O$zCt;r3MVH0Qu#^zS#<+(kA0zljpW^i>6E>~M3n
zcMj0Zqyi809zj<XqA9s^j^@kez2`$g_R2w;?WYt1d%U60Jl&8DN5ayKNN;X$k-c+k
zg#{e^?etLkCwm{<B40w+`1@;aeCGt}-6cmfw&P!MuhQ-Rj)i=wrzqXP_ns*gmKmk^
z?`bLfT&&Tch}79xFSo!DLa8M7o4eNcrcy^8Z<`^xtcJ0O0TyQVk9%La`e!%!2^KW3
zgY%=_YSQ=1&MHK!C5R1gpOD;t!oR+&2}s@X)?%KJx))o68HPpumhBB@+qp1i)DQC?
zyvv$X3C{hXskN}N{7QByLK7_r!=?2`hD_n`ba!sbA5`|-Vd$P1(XV$LtYFiwgY0Mr
z;>OTRQF7hKRpy0kFL=Dfd7Xmz%2<>f%9e39L|dy{l=w+8-W!TP|03n@UbCf*u36?M
z+F-T1aKo61CY>(S4afJPlt`dzoVBYiQgUU!qv+Tsec&%nDdAXFw;>E{NhxA<sWkLt
zbqjUUIpXmOt?l-*{WPqVKWl}cxR0N9*@4F18~G}~6jjN52_E{U?Kn@l3x3cv*F)iM
z<*76F1)w2!N%w1=&AXX_)>OUL@svU0@^VT@Jd%W<Pkn}Ljo*BKOpu4<M`SJW(E15K
z`Nm86v=*$<oHb8xo`YlKB`oF)$NQ{KLz=QJ+P#H+C1|m!9Yem-|HiGf9TKCEaHi6V
zO_|Gxychc{kDP9i^9qL8mK%2_WZ;|-@szYKA*F!r@uc@rdzpH+Pl;=xKtgI=%*}BZ
z>k_sE+fCehHc8)ftvrhao|Qi-;X$~`?isBbf`t9AIC^jZ?B3KlSA{*mPbMODt{)eW
zP1r6^2vF@);`*Y>$r6RyLOh60#*Lktg1+G?zkmC)5|SpOv044%_zON!OHdoDhHa71
znk8q}WGnAnarBt}nekeQ`O(Lxa;=hO?#NYA<N6lF^opno>AoMT3aNu;K568}-3jHB
z1nD10$Ww6TT+~^m{#qu-d%u#gTuA$2tA-OSEkk%!IZhX(%dz|ri1(Mv)>8)E1ZjQo
zL<b-JN@53lICAMtHpvhNPtq<wM|!IG%1-Mv6IM_bqmoUQO-*;n_vyE#-`m~Js3=4?
z4vG)2YkxV-vsX?I?H^iWFP4XuT$Q;xcvNHG&Xf8C)JBenK7<VR7rRqsqRidkL~c3|
zJUT+iL&Vdvh9F+n7z{o>zMGpH(~%5JkQSTg>h%Q$X;HjwT@kOBhB8pc2NIT#LWK^u
zS?k+9({*O#4O(d|XGu^trpvKXpY0*S$`G3<YwX-sz0}<>b@JQjhF^2!$~K8n&^lPI
z25{zSwj4#>^^7hTE$CzIb0I)Ec8}P;2A>TN&|zP^h9Wn1|7Zx>VlCs<2Lz-p;rOP=
z``J1i%AMrow@~Km=?3DYLNvkwk>#J<^^G;|=oXCYruFtW#rm^k=!Hk}@r)jwnTS|Y
z1U;%f*c_}~+n)d9R(DC4qkI3ZdhM#_iz8z_(QP6MkwBtweYA#Z!8Zx*qI0_pje-|U
zdjVfCvmlmiJjv~E1{1Nt!&C7KINGm1X9LKm3+P|*+-2SFYo{@4yLpfz?Kbx4ZsWUr
zn3?R{M&SW6%YSwo7!ka(%Psg(>fkOE%%}P+LM?9?aZcQLf9~7MZ%wI699`zTc3tCR
zr#9cr@|6@EwyjRUFte*n3DPTM$dwhYdGQq!vSxmqsrtKus^a^6(v`SdY%+|ab_8Rl
z`|@>>>RWl*&2@DwkLMjlZ8Zwdvk1=GMpDHK9$qJk2!ip+<NY7c85D9xk<O1VmP3vK
zZ!sXY-uoaqtP2I&eiO-mOX`2`@tihEVMm~w9HySQH1q7N7kc{ZDaWWQm9A-}Tb&^E
z6U-G-b>dzJrS2{%+(wM@UQprvC@Y0{Ns#xuEqJZ(?IOZwKAWS?muAm>UtCi?nCD~>
z_`x8NI(c|Xbk)FD-Ml7OA7si^jZ7Q0>pQ^I9K~oI6qyr@piwA@$R<rKK#QGK7&D_R
zoMc9+&gTg4K}y@U@$t4g>m$Qg@+Ei(CFxGxNAk-odLB|)$DBJRPXHkc8<yZ_iXf4^
z{M}7}f5ynPcarPm*1WZN;c@kT=`2a2fdxlXOOzpQtKegL#GC}Lz$`6Rd-GiN@EX*o
zPDSnP^LcUP%NDh+sHMIfvCA=kC`8^NFzBDpjFDMb<;*Aa=&Fp`?b_^u25Szo)+s@K
z#+A`rFO#cnxyhfRi>aE+*H-158_g~uwI{fXDDSFLCvO)MUTAHJb&ubZagM({HB)kA
zhv_C@@#pRpwKLyji+^8JDbadnnm{ksH~P?c4NtfbPQg<0=^Pp`Hu{&bXGF!JxooWp
zD~Ii|AV8{^KXhheW1A|6K1%~K>>I8IdqW9RQ#wE#ijJL=>dn=uBO)g!Pfbe;B29gP
zg89=r3a(n7=w~%m663sYnAhd|vaufNpw%R6H^p$zpvII0l&q%ZY>^hjk<LVkhS(PZ
zkvEs`CK43uo*#1@cwvaGE|YDDUYvwRaLNbPwHHy>z;G;*N;5Se9+YX~ZS^jmt81*y
z!#k-H$mAK=OD|qx3;qlx^T`iR;CdR|cwn7aV7FA#NI{C~nj6Rd!`1ECN5=d)Cj<1w
zJUY6;L4cfJ_#>)X*CBVZNoPpoN|6rG*GMWb#*OHW8Ns>9lj^L>`ntE3`O+fHks=V0
z*5V`3+GST5_(?(4cu#*EZqLOsMNnzOgNTk>wn|m`YSCwngIHtRfTk|w8SB~INfdjQ
zKDD0*Cz-oE?L-nH8?zMbLv|`Mz_g(Pz0_aF7xH1RND!d{(hFBMP?HXuPHs^-TaS0a
z>zMvj%`@6wXRsRGWt8acFKON7!v;Wo=j_8T9I-(D_hB@S)2Fk%Qg(snPtyPGEpHoU
zLV>0l_j4<!y_tZL5@m!!@aDwcy;D?DY6f!*ZpW=>fJ)Qa+B%q{#tgT!yUXKo!2*N9
z#KgrTsr^UIIqKZc#bK<?y<H&in%2#e$W1HGf_g&#SWSSZ`bPGLeUE&6)(<=G66g^3
z?usgWCY<1=<q_do8{5+15VBZ~@GjxO9D%|%?D4t@K1`y$CvB`r$m~GraeRN<Luv2j
z_54F%w?x6-{iQWMJ?q#9DY?<`xTozdspFfILg)KZaj}|ou1itlHdb3S)KfY?zT}xH
z?W|U5VkEUUtJ!vzx*>U9P{H5kK#Fc+G|Qp#?L#m5)w8*HFXNSYUo?1svkNBjszdwm
zl0x)doCM#<d0t2MTt7l*|42X)x=vL0XOeEeD9ZH5H5uNmlQ5Bs7f4=}7@7hUA&#)?
z)~0jLjaLpN^Kn+p*xn*R;VTJ_8nHVC{M5YaVXt@3+e84*{cr-CSw+Ly5t@g%q|gUk
zqhzJjDZNx#(s8y7o=bR5>zI9zmLN{NIduf7CS9`@mr_rc#<Yh=!tF%{+*}phasErS
zXqKh8b+ehdhP5|iUqPK1%s5FsDv`MHp=4!GmU5<X+WQ+WewE$p@@8YB^$nO^w~(>~
z1M$Ybh@=iq$K8ZV;9Ept=wHR+E}X$-=WCXyDh>K>9^0kI+SLyF_jk9QtE*7T-%W>a
z{{DcsL+0)6&Fy+bh35MR9QPQ<`56%V;q^dT{YmhqIa+@#R;S>Ps6W=ty3oD;$;J=z
zp3fU&NxAzdL|An?^5E<wV`F2wXXj3HO(Ra|dz)uO{B?<+AZe$f<G8X+Y2_DTi&LfO
z!E?gVS~Uuq;j(f^h)SQ%Q%B5MLVGr^mf+kPF6me)W#H13at1SDhS5q{xd$cEm0SkS
z5`~Ae^RO;|xL9>5+($gIQm9!gPqQM|gN7pW$Tu5*4Ya3~mS>jg%4Wq{^U4V4eQ3!(
zE6Z6&9d;t;gT4+o$8SoQNK(FKP>j{>1(<G>IR7K{&XhdH${6W4hFOauUXz_#7Wi6z
zeenv*IAv}LiFGAhc6xKh4n%n;>dwiS7wYdRBaYdh=spfbOdd~Dh#lVUli0lSxc9%K
zksKRU#tXiTZSEx=kB=akni8gjh5zhP5;hd1ih$F532~+D|5SAFDz_F8*RJ}0>OXn3
z$trbZHA7`)Ono~MC$&nDKkGuC<@h2D=SoZ-ss3IJhMV-2y=V%IQos5<iu4_0l0;P+
zcD*qpA*wl;3LeO*MM7r-)6viHQP|qQ?)370%}E4>4{!XHc^*FyWn~8k;OCzgd#+#+
zj<PmFd}gIY%CJ|jUo-w1wfi&hnY+1q^WF!ZiKXvLB`2&|{?LJ7-d~tp&~?0rXPosi
zYaGLjYOnd%R8m?_+rg5?jHcW1urB$S;2g4yt>hcGgp+wXY3UxfVDU`rwzhY4GO}+}
zzANct7l+*HW}cj;roQ~~BmEC4U<50?B5wac3Bf%OCOU#iKLM85Dd~?FD2}ZglevKH
zWV6(Y3g>mX8YP8G1rYBCRvv(=fu2Mi&{mh*gIfOn4F|ZzQ!+BF#$ux$A{lQrP<m5^
z4IOUUXkwZVH0j<^*d3z|Gm*aB9ZunkjhWVXI#_`_ON-e)^e2)n8y!>Ucup9qEk8yN
zzO4@{*~SVbOhQxJ^AJ-{VQ*vR{{?n$w$r2at15=fKxvp;c$w^;cX%>pIqbxcscXGJ
zwaOC~gcvfA_<pI28>@f5dDTbJ6w7FS^nrr_3c}&T&4X7GFQX)kU#5){x1hM@G9a&g
zUTYrv@GPkh)bY-xC3fX72ml5Sz)z4fsI@f%E?6)q+G&L^_@+Do3>icfRqK_JFLj?u
zeE_Y(+E4+Fp4D_AE}w8q&0wuFi$9*_>P&muicashKwsz5Hzmr$mT#W`z~OZSZkAt;
z#HVz{g9H-uB<{!bZ?-KYojiDX#!J#!F(L}>%8h1H>05V0Ryv{ka@$o?jOttGB+81&
zxl+<uAE!@l%C5(&It<cs&e_62her-hUy_yrga}mjdJefNMkX8W|4uxb2~Yp{KAo-x
zPTIeFn-%?El1DjD9BMecyIfB)>H!)9mGblRhSSO!FhcFBv|WbVm?-)*Igd@kt=Ecz
zC~Ud`+F%0c;12f$a>mGdBe?=#7vKp|mGHcL5391`EbC}Uf3Zo=I4k`+y~!boJvx`c
z%uq)%<%Iw^vYmP?Xy*XMQfZmWROcwj@R!;urDsmt^dq@7sWSGQ8mUYn0G1Ov+R~2?
z=)zub;5$mvtM-?3Use<DX86`cv55-0$2O6NP(swKUuVK!OWC1;$#VN#Pk=c59QLT8
zImggtm6wB8J~^UURs6I=_)i9U(;uS~?tIJRu;)R1>^b{SELy=RP;f+i&}7SIc=>mi
zoXj5<HgZ@Q4`&d`0m@01UF%-MD$%j$iV@b~yA^5^wU`COw=^vcCNJ&;5a|1uBj+SG
zt02=+BJdn+4$m9=TZRrQx86^#`&NAI=%2*sm3?({Q+29gqO-R>N+bG%0~THzZW+4p
z`~fQxHvRH5abXrj<-Go!!^-PjN)Kz@@$T+yOCAo6QMw{jZyP8yrqjk^Ge}E%{|f7!
zDI7^$<<dm!!HUUb+SZu(1?_=2d07nP(fL;-K;D1NuR{S5w+fX^SoQ$LTxosUZBV^t
zYUnV=MoJnsRfhHf(Kv&HTxC|T)3B2LxDgiP-=rO}GF!Qp7o^a=)=uS5?dq$-UK=;u
z{%D>_1)I+FeRps|1Ws^5*&`KkdfvRfeBvbPOUDC^C|Vj#-2d+2JGcXggQHHxnU0wm
zo3EzZ<d*4m?Wxh-nWMD{m|n#%Jap6kiLm9MA(JZe7QUX)t3rq3H?_>dnV)=PE&+s^
znETBEfFBB)T6)`SZ^QZ-Z9cClrV;h3auv8z)`#TAo1b~g7mS7i#7@$LDKsh8M=mWw
z{DAW~*{nh}T7xqln)0tGv8&yk;32kYd5gW03Glhs=UI`(q2)Z{{1U-SC364ZtrQAY
zH}4AhB*A~PR4^So?qwTLp7H`TLg!rf54MMSgd<wzmS!t-00j8vAK_CGSo%cmc*xti
zO4%h%-lXOF={H_GfhDw>2HgRekD}7-j2$<pEez^Nmq6i{l=Jw1WCFKR`!brm(uPaH
zYQu%^a<_Sc*(!<fV0dBYFoYG%G$pL)BwY1*#Tf08zeD=?c1vN7al`qK+rEK7>30hz
zDz3}jRCqq(AwFuCNkUjbK@PzVZj~Ru$yg01OY8nuiu3n^=bpfAYoOtM-Ui|8rh8tW
zm0HbK&y)kcH2?KZiBUttQ8$B8k6QjmdMzw;bT+4Gu)Ui_yYpaNpmBdDj`u!oc*lZg
z*)anx6}`~5tfS2JlqTcpH+RWDqeY%;Qo(oiAFh1Tve*|@?JWw`-lpK?M>GzR6f?n!
z7I%H))vNu7>~*s2mg;B{Ak7zCD#-$g2ryoSv0|pHRgw3MZFP>Ha^rKzn%o&*CPozW
zQVVe)%NRZ}Ku<XvHvDWy2-r#<f>X;p@!Ia+kK=;hT->)0J2`|*wW%TBxFgvu4G7q<
zpkLMetxxk&3evtW<Xo+vDb->yq+vagC=pIWe#dBGCGq@ier!8w9Q4W9NKZibKL9pG
zjzFMky2jZW35%pNl2Hq^mJgg<uN#;4AYuuSbHEq}<f3ASO(i@qNiElLap49JkMa5z
zNaS8=k-ahaR@CwgELt)p)+*G8+5_~hdY6vT8y{+dBXyAlO1;v3qH|?TZA#5C8EE<`
ztmYxtEU&G{)&wh>UcdJ>tiN~qxT;k-aVF|Yc(k)Y^}b=&X=vA6Ahf^akIrPIT`YJ;
zpwd%aFI@mX==Q$j6;X4Wz`FY7*2mIUm=2C#@IAVd-tH)J(G50sZ=0&olpNNz8PXaJ
znwYm-2u_w-FuW$srRV*P85NkEWz@$<Kh_LQJ%iK4-czu`l}5I>sji63{|XK3nYOK$
zC~@^G>T&e{15XIToTlebNkxLXu7#B5DNwNK0fb`2O1M3i_i1A=WoM!Y;R_&b7aDf`
zMjPK8N<${*vTAKz8QTZ}^r&LHRVlcs3Tx{f7IL>gyLf+{Q84|;1lo+-qUrk}h0dV6
z$S03e;I{;AS^lVzzP(jLJ8}4?%8|?9DoD-=GXU{_y2odW=DjdWW2AhPbbtUe`b93H
zi}ITz)=oSxQ_OkY3_TO19^E_UfdVgN`{^HHA^O%B%gyB<HtuU_f-ug?Gc|r8m*_!w
z0F!rc$##axjxg((^bt1z0t$pr+_FrjFzNXmuvNwhrY2ixi%38d6)c$ljeji9_{|u+
zZxy|Oj_bF)q=GN|%q`A&jk!pa$)}CXY7}a++Z3c=PNSheqGQefN*g)wX^M|P<!B+(
zAsE_)`hYw`5i)*X^k2Mt(yi9BzWK_;%Bmkcl?kB#Oi4+Bo|mJeqAr3IsSgMj>gqF>
zN#5`XNG#@m=>UDiQDRB$A^cz%?HWj2oOGmjL<$A_UJ(7wtW~DVOB)@rZqL*TlSLIl
zu-$x11po*{z7Nra3XXQ?G;FvoJ=nYxVT4J|Kvk$RH3O_pte7#qUov_;L7<0?Vza))
zuCIO!H7<lCHvdgs>S{mX6(F9|`JSEmEb%B@99_+M(BnI!#~_A`q2=9%JeWFUlYb&L
z7b?dGP|}kRaJp?@rPS)4pEed-6!Kt#X65R3{ZJ~-{V_+%@`&dPT?BG5QNQdf;l$3b
z8c`q>xvYfQcFMNN+WZHS`epaC{$Y)Qe%5<l+ksN==;-KsaIo=wz30w!MF@~u{ECa~
z*&O*p$ZhkYn5cK^^+WJ#AGGzbo(kH8pEP%&d|D3FVyG%&IsScS2gXz@*=G03%pyt8
zIF@{o#0yN;xh!C-U9#y@r5yS%eeN&5!s0(LoP{R*(lI-pvlf&Jtf#Q4dkoPe`P^zK
zJuQA|PMTYbWv%<=DP!Fq4BBL9TB^cUf0{l&spiq0>OrY<-x+dp9_ddT(GRJ{^E<LC
z6%EKB3X9+0pp{!;4Zn%QqmHo1l*vFd*7^1>0FY0ACLW`e@2SYUjol{<vF}bj&Bv2o
z-+1q|x}K?w3{f??zeye&x`z^mE>7tnLCpU}ApJ_(ilxt%mZn3gh>V)0mH2TFi|1Q*
zmm&XAmy6}c-AO)o%HC>>oZl2P8bqh-mNeADtjabt*iIRuN~mz)(*Ja%H=PnUya=bH
z<_!qXe0)&$Em5v)yS4wBUrmaZThMn*2^9CwroOU9cDUkwP-17~K%_LTzb6p`wShaI
zq!D?1Z+hqv5VNwXxvdE0Jp)Anf)JVXbx~b5#(M7-!f^A5&M0}+wY9ukbGvf?t>d?^
z<osV}a_=P4rYfh7Z(^M_-k4#HQ|33Oqiig}&5|??NX5bxD<A|=D=9fSBZY5)jPa>=
zAnXv-8F|odG~EjS<@PpkH$(+9OmFN@$-ZR{!J5<W*Wc&)|5Y;kyVsA=sSLQnY$wMi
zC`3arxe%ZKPzt@ut9vJghjno(X8e+J=T!f~=`(OPRs~GK-^w|~kAkMJaqq^KT`G=i
z{1wqePPJ!9DHy-;j0zIE+26v(zt1|BL&J}m-=nrRkqf*#o@?5<al*_Rl6<13u77P@
zH#KN7^_~dPMTO9>TK(NkifLiEsQWe1zBnP^ww8%Hvd1xW7#wWRv52RGJ7?3MoPJ1c
zp7xxI9(2T5Mp4tMR<HiFeDeaeD}glfHqmDZtBi#q<|Np=`6o`E6~RBF4YMGAF~mn#
z6)-Th;J&UJH|6`UkQTXpf~o_S#A3>LWt|XiWP_}f!_$Wu($ko?I@#fdMW0u`xF=)x
zT|XYm^u_;PEFyvs>2+9LsxT6OaGw`It=`#H8IIxRS(fj9U-<T+djM@r>khxDW-kXC
zt6eC=ih9k&BH?g2gtyoPs$vzspF6D%AS*qd`r6!hu46&W9V^=o3r7lDK%~E%+MKK}
zB_0X8Xu+?nTqgk#<ul{ePgWKIfcz(0Q{LL37Dq(uv=3U;!2(9ctH)suYpf3!3mB+C
zr{CH4Y@%{vakM`KZ;=8FN8ik0q5XH4j&f9x%frwF9@NF4xUgb`5hcN3!`~F8j4&;t
z(L@AZb#{i~k;r0hzo$7(saN#|!HF$rOZG1fTj(O;PM}F)f&O$ulj*g+7CEp2_B&TW
zmSOBQouiEgvzKiLK9GgyQtkkoZ+z1UY@>&r<o>7sX{}~nq=5Tbq1gxxB$vSZ4xB#e
z=4r&n`yTMT)zv6a!PSZQq|z&fB_;Y)tNWIfFyft!b;|q{5>fw<`V!3B?DCD47FK^s
z+si-0Ohz)x+dKtR;19nN1AAK^6RXvj)4m{gKwAwu&K@S5n+y`%D$&Vn?mY=s$FNJo
z_8ms)GWc(|3!A=$wpNJUEiGTyxt+e`;i+N*BVTZl*Q&O1@U)S8)W#izAOP8QSVUrt
zU7VYXc6R2oZ<6Q@Iw`-SzBwLJV%p=8P&}J0<IIb<MBTt|{(_xYYyG><?9DhgphG>p
z19!&b;PH5mSyLP`nT-h+l3Mb)j-JmSt^a1p(nrwnK{<>OSjB(UdDSVo@*1{VGC~td
ze<_)}m}DuD8+E+@SNno`=85=ydtzLibdA$4-RoM8Dsh0-g6y^QM(y7sg^h3W98Ghc
z^-YSI2?v}=NPMv2LWZFlX3<O0Q5nqYzZ@gY00lSy(PpJ4HN*MM1wLqn#s3URK~Q11
zPX(fS4~0B}BsCZ=+1gzS0uHtqK+89dJRr)wJllvByrR{TRWVEMm!m$22du~@Y=iKd
zFvIpgHm1U85_r4GEg$Ltj81@d>~Ec(;hZT1%7`zxJcF}EXQT;T2zm8$<jWQ34((HI
z!QPc0I~9HE;MY%Fu;AuW41Pmm|GkRECIc`|05T0g<O9&w>OzxGWt9rV`Y$fK3}yzZ
z<R;w-F<J5<(y$Wjpta02jw@w5x_$VHgk{{4B>7R0pv@a6iOsy9D%QbVu?}<7;dRUb
z!l6IY`v(81CR1$FPwwlFRm@4_`+Zi&2NMTOuA_3^^V-KiiY2r}pOb>9*#v!!nEPn9
zrHCe8>HS1Lu|=XS;UlJrY93>Eu=v1zbuIT)UUvy_%!-_fa*`UiB};UKJ7o+v&W)PT
zgPN)%bx-f)5R>1z%#Qiw{pb}F4?K7x$25<ApWYOfShl?(0{@YzIi(04*s!-_D*f-Z
zZp8#ZMSLHN4Bky1ZH?x<vwJ8uTrvR(Z>+(QvTFK0s{M3hoNUTw!|$3G)XPYfjn)(_
zGeH1qjI29&pk9nljr~xAB7fIN;kpzM_FAHT-rbYgdlVh}R>J)vpF_q14UG`LJl=4*
zrh(YxPf$7wWeD|v2lKy~LFyBNRa<As4j(8G2HaY#o1W<PoqlI{?Rh8hw@uuo^j^?l
z)ll@H(`vZSn_9ALd&QEBr2m0B(MXZ-lUR=tH0R7<f9F<W#bn1iCy%z=<iI(&+rH@h
zsk^TEZ>#$;FbF02YE`gZjl_JDiT~LVZOaawA-c)B$6@0$F8*z+yuyMMpI{iDJhlda
z)TAANG=I+4JtUVIHGshcp=()FIh!C$1<YpC86HUC672g{i?#g-uvfrWOPt`Xa=?DW
z*Z(C8quv}<UrxnZ0tv&$S6=f=z|WB!>?d`n%8pJsdz+`Z30DlRT^hsn9y6(KX^IJb
z3)8AkOhvnW#W<l-sC4hsXFa#NH!`sfR@!Jg{h5>9S+Ii`_C;`l0VR=G3_7hY*OU+;
z(mTxyIy@;0mc>}I3l8{~$F`sMBs(+r9P$4AONN-zwpl;q@6@G64sl&<$3k)iY<?Xd
zs&#hut=ie4A@@rhwG}^ih`*shn7{n;z31?hwD`r}*dbh~R!2w2sK>5qW@7dM#A)_S
zqpSG>a+stMp`exG@m0!lZE~0Zfd0T>+@x*Sfwvs0)712Wp_UeBO*#QRNSf~H!TJ5I
zVu|DS`<*=$Fv5^7^Z9cv*-=>CE3<GP>ytna^7>2aS1EgsK7RpdHo@g>;PCrVX6B|T
zLwaaPBLW@qKpd2BsGcu>CQI;*yXKCL0@4z{k0(V2T*f;QaSrjLYzLNY)?C?kfVd3V
z>o$p5ofOpn%OUJ<12zdy;;-<MUTa86X8kz)%BXQZq41lg;ce52T30M<04{F(k%qcj
z?8Id8@?SIF22S6$X@<-GTez*UJYr+W*f>2SW6YV(`|Hm;#&aOsVAkp%7~pa`fB1Nt
zd2T>NWKWrT8?_ZQbIV0|P>dbj$$hdtrWKZP+fyWTp8V>yFHIy%XPHxz;2KiC6aXWL
zTj}O}jw{}y54cNH)0%hpKTG_*RYrE>7>GkzBgDQ>+Imnx^Fl4Uu{P(R9dZTazF5e}
z?X+?A<;C99T0&HfSKi_d`V0Q@?3qePXfS&Dn<F9}3R#NsCy9LMfnhK5;-8ZEYEHyr
z4J7&InW^(kvpO{qb6@>Es{6<HX!fM$;PX%XkYUfe_f(sImgoA0+lzE2O4v)*ro6OK
zjHYE+SvWZmcZARP>0vX(Er`DTLp*1-%DQ1as^LCsU%S!-sHM%zs_R>x{A}C^%dG=o
zfCs=~;VfsXBNWrb3Egr3JsaKcleXfT1jcU#+?df(ieMMYOsKTW-3d0kJF%j}&6N2Z
z#G7t>A8_e$>bL`zrO%(RBx~<};#AN#77X~qBw6NE!#!3Hr2(k<{*|qJe~OC5uQ;Mx
z3wa+r_EhxX`k;xEJA>iWf+pl4J^CoC#}jcPp-A}=Z>!mf#%9KT&wcP4S<KSGd{yNh
z77>OifBz;SVLpHLg1lzvHuNz>C7D)BOb<Jvig`t^?@1ZDJJ0ruA}Oe2Z1&6czpVD{
z+7I0IAyDGez7g=j>AO{DcYy1DmB_>u))O;FAO1qkL_B}8mDql5XzI_KSp=>5v`oY%
z8*%cc?cLqJtJ7_GmrYpIdwF@TGn$9eBBg00JhRGc4~;w4+nP~)e=a{w01lOwDe~G*
z+~Y^-VAEXT?u@z+N48s7^=L$Em~Ha7G>j4h>SELEr09$s_I^5SkWa{1e;uIq{)5ik
z0R^Q=-CGdj=}KvyL}2?DFUXLk=h$GMq@0H@*4A6lkhY&%{py;H1K==HEtj8Kg{Ray
z>aERz)7Mr5?ob5(60lzE8PKQrr(*uxNB2+M*HQegd3bLJPc&*hY5mifo(y;T&O?4D
z1*}!>HFx!bH-lUlL8?95_G>94YJb`zUa(_`lXw4)kCy@P{N>H@fa`wUx#>*hKbwNZ
z#l?xqNs62$A<VtKy>=|ghlsmZb#p>u*3>ssNJR|@u3Kyze`caMsLjoYgTnnjcg&K&
zso3&_)Xj@tnsJ?t%r@SB)q*;h7?GVlF!uKaslwu(aF^l&mT-!SS*LtY%Ky;R{4=`G
zuebPGiR=%YQtoTg01KE7tpttk{eW=Ju)(gme^4MPtD?ofMwyIPGa*?1{uj-}ZILeX
z>t`JDt^(vwJNMU9xxIK2Pm!_T6uWbM^4FlwQ|$p6PoomDp#XQ6R$BT@NJ!`b_W6d2
zYdTqM<hz{+ShQfi07eUVzr(@~kB>k7%~Y&Co3h~MovYsTyt%Ngb_7=XiA^ITxVT|j
z^GRcQS{;-j{7PpvItA<d4fn=#wXP(*4k-Uk)m(w-kC4Z1g>;*!m>C#CJk+0Vr{<6o
zP#Tw6jN3Rl(eU%fbIrN<EG#T&l8*R)9m1^zOnv9S`*ig5i)G_FIcHaHw*F!dDH;xN
zkz^hjS$z`=QzCLrX3Ektpsyo)MpMQ-pioXquL_QIQx`&dH7%b;!j~0Ne0=yTERMeL
z+ztTTRrR88uwpP6Xo&$h%3TB{|4Sd(@yl<Ds4`3jM|*ta$jTT(X%I4;&{n%n<235_
z8wKPK%*!;Jo13{^4i-ouA?H03FehA&v#zILV>=EuWPD#$3R-nGi?Lk3J2FB-Nkc;_
zU5`UVWNfkraM7q>HnC>Jyv($0itnmM5~lP<uMzHSXVP9`ei|zO{0R^MO0DM;!OYme
z%ys=kVA*7p*+|A4_p=?oDO#YTwOJGaa9d|@3|q{bPwm<rvY9?lG5<yXFJ*zCTs8aF
zY7qI?zpO;NiaCyMC3#FZWvw4y%F%|7_>LHxyuC5}{Y3*ov^!$(`j;);fEmk8=|mAT
z&9s9PO;QMzc4^{rS5sB?F6N!QSKlzc*F&{pZ`FeH!yet@Z&F!U-orPbk{727=Ei@<
z+?Re>=>UH(EKdlhK_gq*w@?tejb(3_VPf9i*O>ya&cki1<fwvqd5Tij;HE=(c=!__
z<^gh?ZckP(aO1tP9D%BNnzn$T4gClV9NKhPSzBX;?mU34J0Md(d-e<t@TQSZ+&nd!
z&q1i!IK3)Pq7d~9KfMM19^e>gvYD9~Zo3r}^D?3!@fKhta&UUO+8TfhWFKR6*Gt>m
zJ%AYVUQ$vxPc>?{1$;tLaq&`u^Be^y=UBA=%~$;soZ9>TUXbu1l;)Ps;W;#b{{480
zp7Iu94CyA!UHIzd+Dxh9<+NSlOCd%z^YJDfshxaTzliY^1+6{aTzPmtME*@wQDVuk
z+;^)8dfI)4nL2-H?T&4opoYYqZ$}n;Nji>jRax<awC+R85fa%P!^4yN;#C>wm!m(@
zmf)&(CkR=F5?-J{5AShsfjT_{esF5CE7>czPsYsBLiD2()@57Ug!)X4ohsmJA?tz7
z<Q&ed@G7Fza`Lxz!|hqVBN`f7G^-JsoZN%x{ZZibQdwEqY8$?;Sk5S`*)SFGtm&3!
zu2Vj{>KjIT9-;nTR@SJ=8_}o;u;C->@JL8{fdUV*adb3l@xyFwYg^pkw;rs2aH+^|
zn4FqA+DHpI8W6f41hxqe5{kWQ><88<pTmaU1Hl#a=694~i}N{{INMk)O_l@;N;O`8
zdCf##Hwxv-R>{UXS+q?$`s?e`xI{~*IpF?){l=UnO$kJ}Q1>n)IpTPtOV#G^niyKs
z_64jxTXZwe7XWj5)uQw~ve0Vkk%3cU#NFT&k{h=Ox`hpkuL0vnX5X|)({p1x%R{+-
zYu@$^o`S*3Itb=jgM6j8YALO6C}^13IGKxtyu(?9_)GPNd-6aY&gj2=g3^pY&!Agf
zv~Q|!WuqI6xAqG&#?|$ly~!Q2*UA3=9pyoznCA#$QIOo9WgrHtrvy2ngUtOWP>me|
z1B%(ug*?DF=7V<u{QkZ7dvhxr8=cz)4gKfy(7;+>oa^oBIObGTx5Lf^&KKZZ^Zl_k
zk5~ZG!$!%Zq@b*kLWgL+W<+wV-OKsB=D5@U`kaQ|y`J-#A;db72p={e*){9rRQLi#
z&=TwRV{%nQ_|0d1Oa$%45xfKYC#A(-D@CH8!TT2j7;)ojsZmN)>3Kx4=?<5Y-FOQL
zsJ`bbfPz>l75wKWH~rhbgV04C<B4+$9Ya#vas##1{H4@&R7@ud+1$Hq#v2Kx-C*Lz
z{o!h(b>Sz_zs>BT$1|ml@Jr>RfsHQLwM}tdg;BkDFiAjG7Cq2588?1`2ZW^9)>bix
z>L2T!l{`vL?LP^Dq9Wqjo^6uo`E!*tmj|OmK$}<a{d{${i{1!Sdzf=kac|_31QUR=
zFE=l*=d7u_8|}-+#s<-DON$8AKj0$;3I_fB7nAoN-qZVDs#$<z`V1}@;lF^fqUuC%
z(==kz*<JF8G&P|gr}qFtW6qBuBw5DAxX8f{!lU%7I>xKAw&AAgC2+ngXYJ_&5mSlP
z+37`To`FMdT2}amCk&N0D3ZhaM#R`d3YO#G?)m~SVlQYhgN@bZAXhNohXP1;1QJZS
zGEDl$AE1YaEY2WkrNmQmycvUZ8djiW%fW?>Y_aXm>xQ3>Q*?NE_>){H@$+=~8vs!g
zaany-g*X`*ck>;Ul+yP{c@wb8G`05z`cVH`CB4t%Wo3f_6XK7DkL$Q-C3`wMSzL4l
z?=C)RR#^2p&wEOUi9wHjeSMMkTUwS9saygB10`i-R)7~Bu%^1yvHaJkZYAl2WT;ci
zcGL(WMp8*S>gOy=;nd<|QV)~v{jN=wimPEi4#(r=X*AURH7{@}QY9mszGCr<owup1
z$^_QOY$?>|`}t3D=u<&+cutPYGxixwaaX|aZx$kF&MAemeCs$w1qqo+3u0_-SSzlk
zCnMnP^YvbGEe;hViCP(-{80XH3yC0%G$be+TYiYwcl`s7{LOUay{fP_`B`Y*3!F26
zw(Y~mkG&uZf@q8dhTjhrvx|$%OoW=fjSZaMYIlUQOt|XYz-SH!v4b6(t2^6}R$_cS
zL;@EdzXzP-!aM_lf*#bC%`-Fcj@x6Qk&&vLLx55v0eDm^qu7PK(g$0oZmI=3vOsbZ
z85zlLHA4Vu7C>CxGd*7p5n=^WFQD852cKpob3MGs%6l<uAyII)L1moJFpx%g5nu+i
zJugTfYpQ+*(DPxb6c~Ff<aD0v_&oo25;(_N<r?-`#_Sb!xP=IXPnN7O({iUZaBIkW
z5tBP!;y&hP=2~39E~zCedw}Fj9fd7!d#b+5wopzs{8h(V5bd3&3zsHQ-$#WmXW!Qx
zH#UUzKM_bVE-m>5N#n-(#s4j>$BXqN6v>D|+S=S6j4`5;;}+1Uzg)q34LBtIYlpiN
zm$3_nv6Yn^&<YjQ9-;k_PcjDS5_lB~g0Upul&_!PqX*C~tz$4Fh#k2zn3V68Wr@W&
z6WmmJo(oyX)=m#lNy1$KqZC3x4{jzVCc0;6hOC=HT)>OTZ?7LAPwr_gH8u4&U*Ct-
z`;wB9f$8)0{m!hji?Y#K<Kcd9x{MnoD#cqARK}(!+0PKOa1`k|$DTj8L>aqn^3Jaw
z3%j^|3F*IxA2?oFG+wJ)Pvm2xiecLquvS#W0kimrt0m#FOQ#=xIMC8^sa4)RBB*1k
z`M&y9iO%_2<FbVQ8ojzJs*|=K0#6}5!ksf74T)b!KYd`h00GLrqpzwa6Nwbnx<Ox3
zeA3nF3#0^i@hke>C%r>b=suow46iA?ri&mV|MbM_x5i7<D2`tR6W2&P<PfpxO1_1t
z@e4W<aG%Pxi^+?8D)_oee=w*I`%*(edbZk_Xn8@8IrrU10;|Q(fn+w`#pcjnY|dg@
zsQcgfiLIk)s#z|do9E$OyhfbR9IM4UE>;d~v0xdvgu{tLx7Qkc?*cK%_z4t@t05h;
z*6%oQuQiIl0S2IW+sTRYT?R89UFP8(P==|t%SlV;I6+ocOu*I%Sq~J=n1WWa>q=l8
z=GnJzGUs_di`3D8!y)<A`R3wqXQm2QA+Z#AkEW@Pdu0B=N59d?!v_y9-<c*Y;9J5Q
zDfDl1iBb*I@-<EA<x*NcqE(QT?i|-r_+*y-Q+-b&!|m`x^4Z70Skddl`3I&NsR^&z
z_D|H7%(|sJc-B3?ndXJOv=D+vCN!GWw)PQUz0X|bhMvDECm*o%h_hP0N@Ef=Yht;_
zsI?=~3G~J8Z_e+T$YP(l=~PIZy}?THsn+}joDG3fQYD!z0p~ZZYCp)HpE4|9n^P&(
z+NEI(GhjWs+>zFa2$P)unB5$lsCE#$4oWS3&_n?5J2jnED|RADDXFC{+N8&Nz;{pO
z?C|8oa4c5??tu#XoI&+hQj)GK1bpCtR@?ymd}pq%rn^2gG!)L`bX0A_obEgF<HrYs
zqS)|*f8@!7Ng+}txc~1xz^6LN3aYwDFhxrp+5G<FN9w1JY_)!lvyZ?~3ZW55HbKwv
ze$~Jd?10{Qo-iQPV!yDh#nGsvpKbC%)+jN?ghN2aGCW@E17^GIM6y8)tJ;^3Cdxk(
z18PHwlhlI+l8TxGPf~I7Q&?dR<NYsaL=kgYtH~4oNf<Wm5k-#7X34e@_UHKUm?9nM
z@^7_Iv4N$Nl6X)@b$qyvQT(LHHc$WM$*bkwH8bq`;1!thC%O@?Oo@DnWb8B(`IR%Z
zj<8BLztWMel|SpLN;x*ir^zZ}d{gJxEpirX7{(7CRVcQzBblSQE~nI2Lgx_nSbLKG
zMMchq<zE(FH}}zM`K+eKT2Fr$`*kvz##H=7n67$MPSnYBgHNAaS0mKS>x6*8T6^`R
z(EX)6@Pu+XpLKR!_=;>k3xRu7sfU1G5zHRf@+#(*z(0|+p=8w@jXF2S)t@sUu0PnL
z&egeZg8r^78uYyo@OBtpvL|HG`}gEP`^Zu5u3o4I4_qAbIOc--x7b{SfZtj7E--)?
z<+ZQy&e=|4a7^O1ZJyeEu-xKQj8?hJsxrYuln9d2_`B2e)`gQG!@qI)GeviWo9+1z
z0>l{CGV&f_OgSR3fQ97rRr@;Vt3P(|a9pWNDMA$oc3uJqId%??FD(t3ks7%DYVNO5
zoa>HRD(YBbgGH+)ZPor+do8SQb_JiW4d;A9KGT(gU7PeoS9(!eEDzXZB|oVfdT(V$
z=s&_5ecItw&7{*!%y+(^XFc$naKy#KFuc$19!NmdE38OyY2~Q?S%~@k`9ljzW$07a
zZ)&jf7nYPfi+zFB_TEetD4*YWUO8A*Z_)tp2$<+S2tt8Kun^ds`C$+$88rak&WBbP
zSZ)OYBS|D5_yjrsu9NldK`@J%t#JlU#IMAE5_}B|L<c{ZwyFs->WQ=kMjEfm1ndXi
z!IX+XACC80{jG}&#0}c##>Th6_e&9Efc6ja0!D2T)t5zA7;`_?x|5X;88XddrmRfP
zP0$DlIR*%xK7^CVTIr~O`&|M;6Q-|HzScI<ycofa0?D?!*dQ8)2Q~ZY;;H5w5okKE
z6YY;x^8hYr?(0Eg>n4+m)ec*SoJ(%3{O9Y+Jt8KAkdaN%V>^e1bR=?Q6em@*sWK92
ztwU~YpA#4_Q-ajXYXh1qR)M{{#K#VnKGNvUIb^_pJT&x|$Q;a4>IfX*W7k&Vh+aE=
z<&DXN`Rre4nJYUwG@h<0iD@v$v{&^QsCZN4L5!=leZhBivepnHAmvh%oD*e;S*ppH
zCJrpCl!c}l95)8P0wZlekteQ|Z}7T(RnS@xq?u<7$QYmvEN`;y<G}~Q7GO2g3_JnM
zYwx?8OESB+5a)oKgNhX&FE6M0{P3_6g9#<rVC-LpIwM_$>`X7QK--%$f4^_U?OHu}
z+%!#abTg|kFiH=XU%!{hZZ1Pl4e7Ceo$?l$Z8LzI(HiqGi*c7`w)?=u&^GGCfp0?W
z6#v^T1mUe6s>8WLxpYt==W|Yz=E>A8zN#sDfp7cPh{AHvUYMd^S`HpEe%|FJXXpy0
z$;I-BuH*!cPk!aqJk3{}Cb2oLFMH>6-c$WI8Y5NsB@<;A_;S|#V#4C58qkQ&XU|PH
z6Y?urqOW5u-s>t4i+u?x-O)yG?czGxW1Td2x+V(K{Tvs1<6F98ITIg`dT?m0C#KeZ
zvM(Wat*7;e;oG}horTkr6Dkgl=Wsqv3Ou3}<*JxQMn)0S!ouaZ>0%)41<{&8{PX;W
z^~22g;fRA)cF4NOtJgmvR_;g%TRDmb=8pq+)LIwUCIchY5~c~Z9A{3YyExsKx3)Ib
zaLqHl@^;o(aJyGd*@t^ho_zL;3aPC9zaQ<-S;Bc=JLVtmEqEnN)?MIFL}5B#w$zhE
zN1;UZ;2WC4eEajI(EXqKD6s^0`O6X(@*wP;_;(50P+@ofe2iTT$y%38g|Fcsk1$Ac
zS1o?A^GPFea}|-6fFX*j61J|`A5Lg6k@5AACs`Tx&UBcPPSz`Sydjj`Q7;Uh7t#z7
zHc>F%^aG@pXl;&PfBaiTiU$k}?HAq7pt=YgW}&w$0tm4Sx&O+7ST1=rd@>^5Z~lH*
zSy17Varzv57wGzQy90n{@UT^pw{PDjEEmKd^67m!ov#E+@8gM2A^NN_4_)=<Ficg*
z4>(6}k2J0FO&shOfQyjQcQ7C}3N2|Uf5s9WzO;^yPbZ+;iw^~^<vWv69L-fj){enj
zDbHNgV6nOjL5yFbJ!Z7rhO#YcVnf(ck@@iR{UJv)mVB7&XQCe%2hvEH$__+$8q2?&
zxSU0{_hh>2KVG~>RZahbTG>WSW}a92>0(=KRRqEEy^r;Wg+J`?Uq9kAQdnOc#g{LP
zm<iN)?Y;vU(&#8@O3RpHifYlln`4Oi5_WkoipWGRxt@cZMo20UknenEHfRgBx0lNz
zt^veFPOt00tl-*jzGV@w^r+_p-6e-l8CwIDz}ZVmooe|h&VSJtg(AD~>ZJvsjpfYD
z6&p#{EeqXUEO+l7E_Y~g;D>dI0T&pzIx?T8FHL&Bg5N<Pi1oF1H<&!obhSOVg4A8Y
zhNY4-yo!jPpPc)JywnR90u@ox-6`~35*c@kUp(a6TJfrEuT#l5k27-aG5P?!&Gui#
zOmnm2Cr{=g2ZgmM@}AX8o88#LMN*082gmJg`KW2l6ZxBx>}=eNJ&Ts<w%r*$utRpq
zqbw~*sYG8${d%q0M2j=n^o~B5tfwU>h*^8zTwSp=E8=<N!EP8iUE6FTSs^#{Xzw6L
zX9jVVq-9pmi9iOcIhpHdxpaWLqT(#kDnf>4tgcv@ey@DE*`KQ^owJNtb6g7j-W0op
z*C&V6y?&xKzpX*jA^#!pfyJ@b0YXyIr;gOoSGOmLz{UhP8%7|E4yNh@zv2h`2{>kE
zW<M-a^U;q$i~`z24Cqt)&m!}W;IGHI(&joV<m(IAH_1nyCZguYt7y{lJ?F>AaHfAZ
zXHC<GoK`fpcqSK&kz`wN)HiMW>*LG;m2Bg+$)YkmW6RX=?VQ#YiJ(Wj=Y-j=z8UL{
zSJh$%`~?T8s&?~GUqI7+ru*Wz{+1dB^gr2O)cBe|3u5w#qeIBi#Hy42T!mVN?`|#f
zHpiXoDFyakI@tSHgd3RGDK}Qaxd;-ZHFb8c3##H(=d=usLN+@cLg9qX`*)Y>e{*V(
zd-on;nV0u!zdH(L!<?4?Z@Jit7P@^MZ)@`fCvqT2xqllRRH(n&VsPNqUX2DB@_46Y
zK-H|P&;z<W5@+sKjq>(>1aPSVU>HOo=utVkga&HVyU#cKXbx>gPqrS(a}?j#pt)3&
zmO5wTW`+k13H{~aH;@s!;^!CJ%(?m~W0W+_ySlYFUvwbN+dpT;e=zjdEik2?RP&A7
zm>7)ZMzxl;WlUk<Rf!v}hb#P?fR9w(QsSlj5=5x(+;n#SPKd$x@LH>>FYz>~CnBEx
zz>T)>15V8<if@c@uX%a#&x>clH-*ylTlK?5`o4Dzvx~^0nhKzhoJ|@NtWnj?1~Yjt
z<$6~hW%?sdab$cLRMgcPxj`PT@%O)_fg+6i%mrRNgiL$V4dBdS66BtxQeT}Lp)c?r
z7mNN8^UON6(KwXi>Ky);$LpW0tp~#71+#kDAJTm{>E)uc(Q;8aM}=W#NCFy78REi$
zGv5{|+369&iM!?FXB&&7>d~%fpc7kS&_s^;!uC$8=`^&Zkd}D=g%H`TF)vwOEpfQ&
zMdsX3&vRS$Ga{J8URZ~o=v5LFn5@$I+NaEnWfR529tYV(Vo#UFckrwB_D|h#?Z(40
z_i%%ai$<hXW!YrBy#rCTkL`;Q4RQD_Kh79uvr9J#=F#|jKgQ}A5D#Wf+@D$#oN`q7
z@YWqZam|{vDTylnp2Q=5C2YaCe%=x4#yeR{@V}j6+_4pytKxD5b<u)B0u?aNU<IDW
zjzOOQIkf{Y7fk2tZFA(uA1pT>D0JgG?s>)955+V5FC6F4;$|h7(OYcnV}#SQ$IR|h
zCDNU|aenfJP@bp?5l4&3oW%AXg=J#h?;TM>8#=WZ%pIeG6SultAEi&L7i7!GZw#l@
z|BtD!fU4?!qD53tQY91w45UlxMnOb6q+3P0yAMc+Qqmm~qLR{`A`OCccb9bY=IZbN
z-g7ON0+)MFd}qFy*?abm=hq~S=tra-CIT!B(P}D-;?pkvd02B|?aswF#~_<a7~AGE
zN-EP^Ly;`{e9S)u{edZ84xxy}A7$s2ScX>o{?FG-=4eZ#svAzYBUxkoMqto@!gpZQ
zq4dcK`KLHOs(04nyw_wasuMDC-;Y@<4*BIza+4C_?E2&~Un`lr!c<ZCNGKBdTaH~z
z88XaAXL$^my=u8FU6R`#38<!NRytmXfLc={1Rbizmam>w3#6aY(!4=^@i(aAI`1yO
zQ4RoG5<Cii)fu<XTax#8odsZUO@G12#M!I*><O3OhlO54_7pLSsM|4rK4^b?mU^84
zav2-KZIi>U{dEcs76hwbj{a`XMfbMZ*<8%U{iljm6}GpCpBz7V8Gp9>&`Kl7dg?K{
zkD3#BL6OCi|5+Li{WW5K|0HMJ5%ciE{As0m_g5l(Cbtk9ml^KUm3y7*yrSl<Zvqaf
zATn<nF32`#zqBP83hNgTCUuRUO`U6eC5bhYT=>(nh;r_L#}?t$%Jy&KD0&;938c*M
z%hQ#{YKvK)oE>XFX@&+Tq}y%$GIq!y6Us2PE!q6OvW}-y*Z6v2)}ahn@g(ur%C5D~
zVBq=(PP#2b{_&y9RVOBPlijNGft6hja~*6P43W8G?R!tRd($qDW8vZ96`BkL*{==v
zIoA05OTNK=m!FTY`Ue#mcHftfkT_@(I(um!@H<0xdjmPI=zIjtmDz=bI5!*(28N<l
zEAyb8a+8&)yu56NkygKXsuj)H{Hc*<iZ2{=!6A0cRq{eAd^j7=-c606JtPgdW$V0J
zq_qA0XV;SQn3qwXfa>8-Jz*71->R^Sn)1=c?N(bot;d*%#?j!{@LN9%L}x!6%&g8C
ztMRkF(P>xU%sNnSmaftSZL^qg;rxQ1`7vE+?yb(34}qLiG3smkW{XewMEr~e3(afX
z<5MK3K#G%L`|<ho^fSgR7otVma{4ee?`s`hN_3n3Vi~{H(xa`9c8rS6$69;8zWB45
zXXzUN8<sxOqDOl^(@$FPtnIbpPj(vbXMyUkeb8UKDblYdFln?GZUuT0-Qf!;j+1_6
z{C4|MUw&Rfi(dY+RRD42JnDIbovWjbLw>Rd1Js@QHY!Er<UTfH7hlH2WP8&ZPP)Ms
zBjvi&Eym5wJ-NDS+HGapGCp1}DIa&8fEqZuq(~q}QMq_OIpM%nuWmUt_U=I=IT#in
z@M?Lw6;klSf=8mq(-Bt2FQSO)o_g{Pdv|Y~Wr>WJV^({`9RV#So(>x6yBA25qpc#+
ze{5oHnH+Q}g}ukdBzQk!?G695$aX!^(e(At#T2YN^Tu12XqpNMT(5HBC_Q$fyB$M}
z2-;Iq`GPOz*Z@JqPtNlJe()*AWvC_|Jryu<Why;#{^;^Za;ad|u<kHYXwTf8<Bk|R
zQ);(goo;RCI9`OPfcB4FncM*-d!vu%{=yl*KlOj<{=!r#A)@(v+2tIW`@tb5CdM9X
zvaqXbU_b+1F~j*-Y0rHW+{nchwTLVl`n#ln&e0>Io;1>pFm#LU7iyD5zjZT=IM1oP
zNu+pRyzv~6dF>#5D&K#NuBq$0QU8AY#7&i5JQofsgWnMMo}E}U4Y3|$9?6h(GKGn~
z+Rz&c=v<L+svA2vO~tk3m!)HmJ!WXQMAmUJ`S(Y2+@mt*DY^S``fjpYQ$G+P3&jie
z#uA-rp?1^}rA%T&H%>2}CY2_*2M5-6{uw@WBEL4SJj5M|S>habhxY3&Yr6vnBQLLh
z``jnFXm{ePH0gL>Tsd@pbDh^FqJ~)ArzG=F(XV{T+v&9~jIwIUUG+1Uyo(|x<~WnN
z79ZWh85|jr^xVhD;96uj-fTS=ML@Owmm&M$NQ=oQp3IF?1k5wwgh66ZtjT0y!xW0X
zC32RhYy+m>)ZXFW!T4e|Z`Mea4i^li3g-cSCb|8Iwa`<k;wf~+ysfSCnedB-$&?JT
zICGV$C|UkV$c0zfx?;rTfh4P%rHQ54Vk9+-B>d=ZrU_?CedR8Q%PQGgcSd;a68hq+
zdDHo5*XJya<1L;ZdBt|Y95I92bNZYeM-*Y*m!8NxY}KLLIpm}7lK(^@LsYxRD8Xg0
zHR2+41NMP^g#9v?5H{9Z<z6dGuKsO~)%LedBe6mRc9NfOSUl%q&>wK54~)h0qU0_&
z<}z)2TXE8X9iHlic$E$E@$}?aWRSkICF=<s><|mL;VP*$L3ht|T)p4!kIE~rSS9=l
zxb<d*(7V)}Jr3RCve&2epL@Gw;O{dK=O$Ua*{0DJaer#|?yWI}q3Te73};7L9R%BZ
z@o7&G&GzBN0>@Xo45dfos%~w3!Hb>mllJNs#IKR}AE;F#)ZuQPc_CHH<Y&hzNkhCA
z9D`q{hU_2q<T?u}tMH|74vY}|#H?Svget<KW>TG8``WmjhO4uE<7#LUeC;Xi+vHWl
zu=Y!gT|c(*s=8e*O7LwyHvjG(8rZvS;T4X)!FBUxDT`qTtvXA^l{e4s-&xjK30TPs
z5f~z?3Z0%nSR<OUt4LHXtTn&&7FM@6645Q$3!KnF#g%E^tqpHkvjpMPs_RFk*)Ohf
zebN8z{RTASv$QITf99g(;U}l36`bPQ$BBg5F&FmY=OKT6f%Wcu31<fV+f9e_)`{!-
z4xUd0XEWyVUbWWWHn00!alU<p)Fsfm?0CyR5FgpJV~uYXo6&9hQbOI!!|p!vD7|#$
zi*K1$cud7EQ*oa<_J;2h9hQ6+y}Yfm;E0%-0gsWyJK2OkYATc4xy^B^u8s$JDNrW(
zicUM>wo_P;4<&X|mNN+0PDm?$*bn>;OfT`qx}gxy7aLi2(W#jPB{C%GKM+0&terqU
z842X-$p#~uNXCR$D=BvIU!yft>7@T+GFwtKmt={D(%Vo*3WHLp<}h2~z@?hp*5U`F
zCLfU$wx`_9(BSx~=2k7~$c3V2nU&hbMRAEwTmIO2{OK=!1PEHCy3$*c-+tc^3`!7L
zyV(FclDqqi#$24yy=(LRn-9o)Hx+1+OFUM4!N222Et98?IX_{)YxPfdCaAP7&!;i1
z!ouC$P~{H6f7~C_Yx3DpV7cR}W8s$Xn%YCI>#d{*o=riF(Gh5<&n+#q%(Xe_z{jVu
z)m3@U>>DM+<@&vxP4wXmA>rZ@pB4AL5??M;p}&jy-VfX<U)oKqdrxg_w~#0=y&oC0
zq7B#G1Mw%n_!<mf$eU?MdN`(A;8-y}^iCs@BAw7lNcL@%%oE|4YkWV5>I1@B+m(oL
zoadSA%{SB*9tCBedBx94rE~ne%ch^nYc#oJk7nwENYA|$EPlm)Yv{Xf!dAye;?{aS
zHkrD5n~&L<Z)gZiJG@((bdey0m9MOdZOQ5;Pu1b3Qd<=>v!MCOTG&FkGF2SwWDG$b
zOU7ObqE6Q?uBbm9NlNmk=hXK7viydXA}PnrxXauAkhO@!LD@TCbQ=E3N%{j$-UE_E
zp$kUo-ecYVFWJ^P&Y%8mKEN)iP_R<R;C+qjLzqxh*J7nm@*e>`2E@DUi3-=2j2^1d
zq3(eHUEO`a>mEx`U5T4yfBVUenNOP0)|zT)Og$IXxoIy~w<~amVv0AS8Ln~!#BAT1
zN_8H%eLl;A8G&o#+^4)ab>S1U9#vWhMbulJjK|xr)fVeNT>_3g<fH94N@rS*c<xd@
zMMjok0NY(=aY^Tfk*!O63&Q43e@c>WO5uBolx^DcXSfSKkGj5l^I0xNjf^M%_ah=@
z!ShEq@u+DF!Zf%av&b4QUD-1()%?tub5jG2+H%WsYF#b3$4FJrD>&88VSXw%<Ii^7
zgtcR%%)un1Yh)T(Mj-L%>RmB>1FR`aip7909L5;c4`VfwC=jY_lK;|(K2^oZQy8~5
ziV$tHEmzk+RTYPkEVT-IvoHz~XnN9zy=SYCp%(>F>d40B1zegfo`RKGT$RborAD?o
zytsjS9g;ezSnRvqXjxr4%pF)op5QBHy^q|ft5egk(1a?sOyK)umZx7i%UG_85X*Dg
zS)@HDjFV|I@$hsu>Chuzb+kiU;?7KVj6{WRO|hw5X1MUCmnTwueQE8gJ^c+NxXihQ
z%L76mLCq;m1}BC7{<Kc1;|}9VsJ2^t5>3cHUh?^jJa;)Xx4G?u_C|L*8+OQqK9b+b
zp|mo<@g-O5`^i&kxMZz@k0EG7qiyDjbnw#rdn%r88_0WtNBN<&c04%K$!i%sxeDU*
zzkZ4v73sdeoBlW=&Uy$dMk|R-vuJd`P7m#d!_SX-r`$s=F=&n@f6^N!ue?2WFj8!<
zZ|BG16}VJED-g`oT|CY6P5V-1EWHlpTKsus_Xx%3LN92)m7_67J`R#GLRh-%FVv5P
zt9@0m?(s;XY^3v9{Lq%?U|njwAjcB%fmvQ)@kXB)r01b{)E^T<-+z({hSR^0FpS#d
z^MrNImAQ&upEPyJDXR+i)Lq(Qe)A$!#jl{`c4D^1ug_u-YRd*C{`Capxts>>!j|=E
zdUQ+sJHiwnlgf`gvyve{HL$n~0fNBVgd55iG_cc(CCb?PV_P(Cqo_Bt@TJp6Au~Mk
z4zAYooN2o#4M#H~$K1)~l_wcfArZNk`CRzOR<n~bz5JEcT&4eg)V+4y?uyQU034G!
zxVkCV<N3A11Nj*=Mdc|PTH5-yR0nv%x}Ubbw?WbprS8*EVJp(}U#z@(1^a%oiCpSY
z60egIznF6y0YmVV>CD1nmrZ{CN=_w9Zb3%n`hUx{{su9TqX0jDRs>jU_&u|<?M~UX
z{QRKk)(6~Hs*QPGjY}Q|$-xo%6=&hV?^y7cCR5$J4HG0BK~}rF$z9&}Mo$*yR*{6|
zq`sBunV^JJfuq?Qzepn7TjtB3B~s@0Gld5hG`6Pb&eQ)Ac}4<p3HmzOVe$WRZsE~n
zM$=JR?`!qciuA(fxR_N}_ZC!%O&mCr3#m)5CrsRFde@+>KXSrBs5j3b@cIsp_kveM
zF6PC1?SXIGHH$knI^gG1^+=t!^UOC6<Kg`cbSeyP1)J{DCsMWM>HB7Wfh;q*u7M{n
zU)IdKflc<%iR*u7p2G7HJr}%IGU_)LvrQRg<v)1LbH`$4U~d}FNIrH1>4ggIc-4~N
zbb@gu{k8FBTpe4l{rahHPDr2*oeOc>@!xgbpCY!{`fYQ7EPpCPZ9d}%FCBIiIIF7+
z&uj|0Zg<|C8KC<}Tz96A@ja;4PnTcaso}lrTpQUl0@o<?$Ait<%Iz}$-D-2G{N`BH
z;Xnv@aJP~v>Q$(&G<|^4P$^m?#=#~dT8RMSkb2a{i3t&Gb7rUOQLNOa>BCVa**}Fv
z2(k9s6_Xc_nEAVwmy)JHsZF<JQr%(6d7uOGMb9jPuX&HiQ%*>u^gJnu6EJl*v0p!a
zGqUhJU7l^RtU`@%TXGJXr3{9*57qvBQ{>ios{8aIxkbfS@vXmL$hDNW)lbEybHbo`
zAk$yQ4vcC`OZy+$=`>HO_#j|DKoW_vo4reAT5z<X<0bR8m;PDxnM1EVY5JAYitAAw
zQKVBse=Mt`p}h_$LDFZ*9a25o-v!I(f{kCeYG;#={40nQYZoIo^)#q{`ujgxG}d`4
zoh@(EbQ=*%mv1=ltg9RKrBO5W;F`Ic?j7VsxyW9q(XekOJoa;k@{5BQ!u{e>!F0rc
zg2-AWj(^xM?!iY}6ZpfGx+v>($V6BmXnMXGIhdG?apgJ`!DF82EEC9RdH#HA`g=kA
zxMQ@=#9`1rX3dXph^vH*_2=vt#Pz0}EWqwKqSt{pAkDGUb6IzFd22m%eA_CIz3lAm
z$k6=*F`U(oy3A~reWdz4T*0H#lk7nMHb5iEB^e5NW?5Djel>WlVv~>^nHwaCcpckO
zFVA5<w(v$0l+}$Fmz{!cU`>?~0ERGumwk(G<B!SbTi<%Vlx;M$v@*UM8+y3*Af6@1
zB}0J{)$99kG3<TCxKp(9^|c&{>c181^v@UZFKP#ZtD6wtfhhUt7)&mgx}9qWg||<B
zVOOu?+0;)~qiyCb0OdiAec5ihn4-vTxkExxkr0{)n&q}jzMWIXc~1Tq1_lPGmgYWn
zSeZ1I&$a2Fgic*(t3`NOF^89r(pxrdct_VU-T3c4#+ET|?yp-yKR9JwZAK?3Tq0>b
zy(1VHhGb-B%T9gEWNei&O>J*AV^lh<erL_IGbZCK#9rDt<wimA*2|xAMMvwy)b<0_
zq|f|fRs7)#&EHRWcw_MSm{mG|f^e*l%LIZYrb5FOYKrQ9N~u2zBYHs`>N<<V?}UTl
z%-|NlSN1lliuEE-I^o`%hAtAzG8~uei?8Pek!Mj~D2fFy|J{SSr6OjD{xn`JXyFgT
zT8n?Fe#O@ID{pfL-U%5b5d6KQ)iR$~k0lwCDm`501)dxvRGEdjvZelB`qq15-b<E)
z5qHxOvK>`sS4E2xz7qYmQKcW|g6KLl3_2XBBq1_mBe|<V<k#SYV+{sm)E`4WBIsuE
zcYD*%$c=w@6R8`C^SOb7MO*9iYdgP61GEcMPfTNuFJ2vytyQ@ho=aMKQ2NMnIbp|j
zle2eVan)Ew2cJTXGocDfpyL$WaEmG0k^e~#n0CEgQ!5JEU`@dk`5f3tGZ;Yf`|rI6
zc(BFrRiD3q#I!j+K62n7X4%tQ6&8k(c;)T2o6!liu`xQ6<g2oe9Y-?53a=^aZaeIm
z4Ht&=<AD9>WhP;C$H~;MdJodBRUBSf*l!F;N)%Chay6%#0wyyvW+tx)W7;*=iJGj~
zno$cdiw>2xbmTv#PtLrZ7XI;il7fZj&Dt6p6N-}^rydy^f(m5riumyis!IW_q0GNs
z3RD^R%X13KLiNZArDUtTR7fdMS%kM4NUz!lv=L#elp0&@@8ACtoNp=W{}BxZ0sehj
z%c60Q#Rsj>)%obK9}K9BR`q-&{;2JHw*A1}-HjwnxxG?<I87@Tx<JM&gyy&4IYmH?
zW|$XlzoD|Lh{n|I;SU<r6E9a3bHp!`@U;2{d(G9gLSxAE?s;eSpN-Mstr;CkGOdIw
zH3cP_t<--%)wD_P73?#udu(g9>DL>_u=X))`=y{Zx#xat#XCy;vX5{@K~>ccka9r7
z+oD)r8>rZ{m;*^;)T#*~mx_Mqi86H^KBbX~*9Butl+hwQWA8tHyap~qf{AW>YL!(c
ziQzrzk~eP<5VSNm-=(2>hhnpX`3SJt4?&`Zgx8i9!^<;*L1ha#vnZfmLQ?W7%%X!M
zmcjIo_qKrFsgHj7LJRhgT%AcmvX39{UcK|^(IdnaTsKWeOTSyUH-}K7Tu%J_zMKul
z0q~Qn)0HCXKUn3$1@8=$lr8YydT2h%-Mjo#=hRT9-IrfmC9VUz5?0lZKE<I(VI8q9
z2ZE2OFZ7q32M5M8O89>U4le&&bABc*>Cq3ElVnx;y%(omYF{hD`dRW;?dlICOUP0Q
zyAs%40TcqUaQyioR0h<2)tOSj0pQVc+-jl895rW+iRLsTg=>~+?E<h79Qj{ms(~CW
zASYz1OMz9UOvY=V=}0B;7PYzoCBdt%7Q~MS!LTplIOvX0Cm?kwHXl%D`mmadc5A>f
zy>hc2!FJuO!&EXCa@k{qIdzmVDU1+JZ*1r|>8PoZKsUy8q&T}(5Ht|<0Q^HSA%NaA
zv%0Fe=2-NIOW!Rxib*pr@Dx^pANBP7!dKI@2<0lu0;wep{$1|T?>8y9uk7DZEKcIV
zYhxBxR_k-?iMW+ht$%ib?(gD3yz@SIlhcF;o@qy3UY@y1+6R8HDF%s77#>jG_c|il
z+7uwY-PF+$I9`>hEwEpQdFP?UI6v_BK_yX$mYMmdA$urH*Q4C6;DAA6J0&?exqn7B
zfj0s`mT<`^=9lcGU}^}uUT<L-@;fl>P~?A*{sMsf783XjTED2T&i9e?ci*nBF#-wU
z5X@RFEiH9{8v<nZSL~#83=EXa%t3GgmE(B$?3NWl^xzqY%R?(3A84Xg+Fc=(Vps6+
z=4(AK@*hlk;da)MJ~DBsG=1`!42t^FPnT?Qh0~Q^^2TSj_x6zhJeMWo!<x$4jCA0+
zwrOzdg+I~!v{iI+KLFR`s%Nr%-IQi8DF=i<PQE+;%7aU@4-DvUuQ91UC#mhjJ0Yi&
zXc1G<bXFSpc?-BZ@L&P?Y7?G>nQ4VxO<?z6)^2wR>7=8n%gNRHvs8n|IX+lH)q;Ta
zmY{=!L!#sKRq(f!MTAb-!KqWaXsCwB7x(thg(O5fTeSkbHPFDoLouDmD>ao8C7=ep
z7+k@o(NRsX%b-F7(L=Ek^D+93f|8QIGHLrY=h^wP>guFmw`%l;Ia}j=XgvpX_Q<#`
z(SX<qOfV5)VRV?X-JO3K7+oZwU3`}fwO1gca?nFG`b|lgZ<j}P@pF>*M(a)O3KleC
z(b`t+)*2P_U~Du>i4cb)y21FXPR_%7>wf*IQ&AWGYn5c>(pI?Ye|twX&l=f8Td8?$
zIsgJ122H>cK|283*l%M(2AyPR;6xchUvdu70NewCSSp`h>XPoVUWSQyz`RPo^9j(x
ztA4(EBYHjx9uyptjLXGN+xlprTA`F)su2YwaI}Z+CkOOCgIHiP%Mb3FXr_aCQ{Ni1
z*8b#QF3{OUnGooGOGQ4#fSYnQ#S6$yOBE)>8vp{r-v-#xp0BJvle{+9S9U?^v5P@8
z*`?PS^|!4o5!O35ur4g_4{m10RsKuuQ%JL24%7V6tLEyqA_Tw5tg9gRn_dSC{H%1^
zj%G8Ec_UAJmx?Mq->z220780R0*?(f%304iz=WM%g&~pODZl@<11OpN9BBeD4X%Xa
ze2jr=g*}rj|DQ`-6RvEYo}Pi)gR6k8p&?40*@8qqbFh5-dG!bLURFX&m5kqU^M9v?
zT?ynp4o-<;#TSHuaKro<>?nxe`<LAIiQ2wbPd>hc6NG!v@MGfoIQF|JnY#C`nr;G%
z4;MdwmR0?)>Z1wv*t%RD`r%Vl#-XPK7+Zhv@+THk$+inM@n%M}%CCVIK2XxcWo5yl
z1K$HYI`BMzZKXIu19TR-52Zvtqr<&LSE9kyA+W^7#l>cCRSmTWQ6mw67GC&eTm*p(
zyYgV#Kf<Y>hLZAPax1#=ZcA$`da|fLaZ%7$)k!88?fLJFjJlx)o%RiL-N{lI-9?Ju
zsqk@Tm1@+<<L24{mI%!faUQGcr5h-BcnDGK&o@1`#`9eAnXqkXZ`wW+#_;yUWv)^A
zswkBtNbUg<8@*Mp`NJL#M~Ca~mI864#`jm3sl&c@=58;TAm<7y6`nxvieyw12GKz1
zVC%!whVA-j43HJO!Gr|qjOY2CyuJPB&xJF_9nPnspNdjwHNxWK<LUq8K5I8g7r3Y@
zHyIchNcHGZ_cAJ@e){70G<AzY0>9)@u2@ut0Dsz!(UF^cgERmCx`#4x6rW0u${RxX
z{aDyp{H`8ot0kqjtqEjBX;KE>5uyOvz9AY&1vG*Y3R1e;uVE5zH4K9xlU<^Nv7ySD
zRIIF_;ILr{6Fn$LR0tC=qGIp0x6v))q&bkK_IvX6xNVqqpPuOXcr46O5_y1w$_V3U
zrkJsmd)(ZEu}-Kfr{a@S8vbMAb*<V(O4km2!Z{fCd=9yZ+j~5|?5f?+fAtX}%>F?=
zb-q7C4An8Je{1uQ3ul#-r@`8UI=(qW^f(gwXcgjHrhXuo=e)Pt0yNL}!NCh)y?jY`
zxX_3r=+2{8m(V058Bs%Jln)-v8mG6yxcdb^e}9u-2U86Z98$nC?*N_8MaUnjj&aC&
z3FSIjczMg!T~PK62(S+DY5hw$p1$}|NcG>^u>P~ChV#aGalQLM=pbh{+4s4_4i&G0
zOnia5`@dCjWqaqA(8tJ^W)Y$x3rQDVQ}#RYhTv}`T!{EEqH}!PIM7T|pU8}K9-NNB
z=D|bAKNXo<T516gJ+wub)MZf4x+^49vvn4Vgx)~i3HSjd8V;#8PC&O+EjA@-X=z#9
zZM|>*EHwm-FcV$3T6k?T=?38GCGdcUWgFa2<iP8Y_SCa0NvI2~FO|}LL>@mz!xz}O
zh6Z-mQy@<Vp-zgL0Ccc+)<$k7$cRDt0{%Ouaum|ne@C__qMDJyh-UDXLYs?qQGWf~
zt8<UGUf3`OQ~lkUz4QEjjt5~1#<ph6jDNO#jiI+Pc`h7Y#AW=Nz0TKG)sbT7nQupF
z)ZWGV9POT>FVP%FzzRk~nJmESH6}mrLe9dV0|OjsEegj;PxU6qqQ7>R{rLG4W^i+R
zoe}UM%SH<PO}288Q-3&1!PA`=>tIn&+N4Mr#J~uIh=T*iz4b9k*lE56;^X7zuBwg7
z!NDu*VL<=E@3w@R{MY;`Js%y<IiOZc{=XxS7GtoWW=<0K-(z}-;PG4EB2G}zUtA7o
zy-V;>8*AnF<kHOMnVRA=f)b>LH}NbuFwf|WNK;EIlTauF-0T({BP1F}%WU5J`@aU}
zG0Oehkew99)E4_PZ?fsP>w>Tk4hd^ocNyFlnk<cyAQ-;7K}Hq;{=YCj@v_19yI#u;
z@JY6`vNE#dFH-LSA1fZqNi<-erolJ_ycyWJw?Y2WiJYAtXBJ683<IWg91Pk$0yAvz
z9)fR-+K-qqf^Rms4o@#H4}aM<{t#Vk-KW2~Y+K%HG-~&C<u~d%t5;waTyK3=+_|MY
z_efK{l*^ba9e-1tnM~>~3)5{kD-z~vibBJQP`nb+NrF`!<l4V{;%OjwA^aIcTBC-}
z>SR~Xof&+UqICPH=44d(Xdii;b?xB-buqy4(0zCZr%3)M$1w~d+^&*+her<d;vCK7
z?0eHQGb@!kFrQ}J`{mhwO15fR7AH)C&O+j0@%OA@4U+KLXZ2Y>r1_gDva5LxfSWsb
zEq!<V^9mAdopfYh%<UKzxTG8Tnq`Ol*lga6F*MRYaveV8Fmx0*zg~l!upLL5KhaUJ
z@pPI&lz)}GToK6#r2NuVG_ms6Ra)53BqTMCfKt0UsXENpW^3W%4-3A|IU6HFz1gYB
z<p5|`K5^sic2a#%cgpM4`gWJcJMTL+-RAtl0-Gr^_z-fL0m3O%m%iZ80qaLq`c0?r
zLm!o~&VuGImn&ku_i-#X|D<`Q6bv**wMTaMd}5q*bA3Gpe~nyJH-3`o%LqCc?-HWK
z1>>7lkR*|ft0I0!0~a}UCZGJ>K>SuBfKv+KtjxrFT#ZXoek&n;6VQU9PDx3j_eUx~
zEaesCnP2)4isg!pYSe;*^?|1CICU8nxgfjGP6&d7laG)P{Socnx|i=i-*m`J;^Eir
zwsKl1<UV2*$<HoEw~l9YK}WQnf&6YOetDS)jmqN2<*?y;gE3v_E2Tr<ktOBoxJtCt
zIz#uw|6<s^SdL8OrlZ5^U>z8^U4X%`i(spaPe3j8g>fEz6b#*-iiw3qNBb#fzUfX8
zg&N8SMpjk%$l-RS#GVC<eX?`#)%pEWM`q^YNeFHno8^}6CmjL*hrmIces@!_E}5-%
z_HOibEG9DJi6_}_r>Zxnq<^VPo)o_}jqN)a*MQ?d{>B<}9A%bGX0`k@YZu5fm9Q!h
zB0|C+#~U@WFi8OU$L!pk+P{w_&3?z`Q2EE<&F<*=kB07NCucnI*abIgFQtyVUJ)zd
zCC224QQdD<XvLD$b-f#a?crOzhh^zq!iS>k)r__`rW)K%O#PQ=K?Lzd`mihc9M-c&
z@gTfdwMSroRmxIO0cGf$I<7C*Y~kqqNbRZ$@GmBnVSm7RxE<b;^C>Mz`AQ>(YAg;*
zg72$~D4IG7K;?f{-+3z(?}x~38pkxdnK09IpQaJ?yxlRS|KLp2-d$j@E7}h}Msj6q
z%ggFz+2u%z(5Du?ZztnTO&MEza2sGdLB@yaP>%#h`PyLx)cJZ1w<mN@meI4seKacX
z^_+HbVv)Ro=9hcdTTKCDK3s?D?Q0?X;|dlhuZ7m-InOArkg}=F1#bDgX^B6bVFxBw
zX~_;N3u+Oeo_#<4mgURZH*emA!si<@Mgi#;l$sZ4>FK{Qo@uKcj1?FJ7yU-f{4Dpr
zK6(7se7)IX+kF;5rk7f=Te~VS*6XnIs@eZF%H7986Z+`&do#KjZ>xz3-*C|1n#2|Q
z9GZ7}Nr0lX??m<AM?xc!9LTmJA|ijYG8j<LR>pn|Hq~zs&wZGXi?}=v_Dij9ea4v%
zKudv7o+B6ybrFh|dh(3<&q$CB-8?^>2Abm~M*^UOqC#0ueO28fe;X{?7KXRF29`Nh
zcig;e72FN91yw|*)c&3*)aAGxK*)p;Ti~#vjh3IEe+38U(j-&?3mHl7=(@*yYY4O-
zE_i^@v(owq45XN3{3Vb_Lnmfzqrzu>e@xiOhz>0X#M~O`n;CUVJKDy#exN!c|I!yY
z)K)dOR#Ym@Py%AGa%FCyE1+yHo5&vng#~mnYQW$+x^*N8>aD+Tj;o-oDB-*++Ieaf
zJ1g=Y0YyXb(_WsC_fzLCrB%t3GtJ{y4f(|XIi_(wK$Dq5K(aoFAdQSq?4@$`9M69#
zo_8s=svE`zJh1d7H3!q!z$?V{NJ_esTYW@tQMF49dgc)&Dqys@;E73*XU_}KgCVT@
zM++_{_f`XQ=Q>ZIyiU=zh2<3)r0G-0P8>T~(qtrOtAb|ksH?!^^ta&(4=?EFe8;Gz
zN-wFZKpI>QBWH=e?`JZ_AI;qburzaQf;Ee2%3&vE-y1T%I2CT<|8De47kHALot^tj
zn@MYLTU0)Qd3Yx<ho)s@%&K(jNfSr48w}Zr_h<=){V*i>kc9SF5}l9Etm0*LeR#ai
z9>EvGN$(YTzdhi`W16U&+CVY}CNLhk&btlrs83nDScyQ3x|=s&pyeIGz3OYc<<WOc
zb@0$2z;hZNOcC)75mtqUj^3a1RaQ4#NUQAJ-mxt9>kShJnCDg3&A8Ijh??u}1{T}f
z1pf{gx$=??j#zl@6ibPoC|Kx>JB(WTzJNeNem*NWOLnYV0LC)!H)b8K#Nz%or&Nlr
zTwoOV6&_<cpH=6VbwzkT6Va`nA8{!JC81!j>FF8$xF3yn8QYuB(`zfJi9WxJzBwy(
zovf`P-G?uma%6y~SMT8#3LTw{GizN%rf79abBPM=j@lkinqFtDc|KLl=<fK3KtWbV
zckw(>Y46{G*8d#8;x^h(2g5^#q39NFy~Z3&8m)7k!hV@7hnDvll1{I9b)w;y_|}x<
zoc1{3yBuiT4cPP<m_ebF44)Y%r_z7Jm78tKDD8w^qMz2CbKyVpIHkz%JLRRE%pTpB
z{5w6bnmZXkpqC!Y-Y%)$Plf6z@7(}zt@JtvzT$<ogsC~ahod*g-x~}x3C+wd37<Em
zl1BX(N~I^8yox)M|NPu3w8NNw_6DMyHqbe>-FB7WZy)uT7(fA;S$W<>FWUSo!n`Iw
z@qVaK*|PFL%3fYuuJ;jnZ|dZr6t+Mgiz&uM15-fHVjdkoBTI|obTuH@6u&{%(m+#6
zC2-F9rE?0p!Fek)Np)(emD(l@>t{+9ve9Fg$$CwL&OBj@{(s^9N?d`n=RV9y;Y0nx
ziK=13*5UCE+-)+rP1oG)qfzzJ6nQ*jZ?D*JJE2y-qnb*iMZD8A_JL@@@0GeIPXpF4
zquM6R2(E?coXb!>iw;A}WS&}_|8WqU(LZ!Pr1vo2Xm7uji>3J8-3;}n{x4w+{N$IY
z3F1;!-nR6Xo}w}eRJ<JwfIS8{JWQ<1xT{+Em)YqZv&<Uw*UutyFODJv9`Y8t=L0or
z0%F}#Ulmb^@zZfOH0G7b##@dw6hU*IQ!GBc?X0SHS1EXK;@GUVw{`1P&TDB(*NdOM
zuaF77o|IDPacl9lc4Eo#6Ac>3`zELV*vvQn_vu%x`m_k7AK3AdqwVeG^HyC;F?v8Q
zmHeROxtdfq@Lx|JmmJq|-F!dh>OIA_9&!2WQ!m5SwbJ97U=(_*v5@`Q(uLTx3!Mh{
zfQOdt2g^!0x<wi6cYe-u7e-I`R|4;p%zeqXsz`FWqwTRZ)pSMYfc-tNn63m|;bi;x
zzbz-TW&ZN`>gsCi8Xh_C;6no7e%SwMCV7*J7@Rid^A_29KW2*R%6Jj4;TL8=d9&jN
z7jwji_5BHng9JD3qXy<zTzBp*>7pNPH%5q54j=Tqw7fL-bB_|hS}zeDe)Y4X^{JO?
zK~?-yzte=yYyOM=-lg+uij&gMpV#h02|`QslQkbFCvj2vJo7autDrQ2P9MTxaT}Cw
zsbH4vX*K|o;kXE9o7W*ved7)ID!rD0v;)XH;{?9HT<WhqjG^shKyf(zgKv|%Y`Zfp
zWGv9LIk+*ox*hYJko-q|zxTtgE(P5J{Ey$b_{2i3Unc(toR*En|At|&oRbC`K88Yy
zk{3#WblymrB0VsX0Uh~H8}Lagw@@{&@XhnXb9^ru3pFy-@<E-9N(aaHW_6G$IXvB;
z03+KXJ9^m0Tl-2*dUmt&%P~2hI^h8OjOE?UdI`-_{>K(?`LVGJwub+U%6=<BqHK=A
zWIPJlq9RY9UIH4h2UHa)-G;`<s7kuR)aq&g^ao+mEiw&0ZFg&h@QZ1D^E=?E;VUM;
zvuWgZDTBafo-pg3_fe<mG3)e;$)^-Jo18+w-t(imid@VNkNMx7{;Y51?_1Mj$PRlt
zXfa6EW8d@c^TxR%^4j9vOAAgy(%STS{G{cg8P;3UwztR-PYJWBxlLhJ@us@#Ke8t9
zYu>0SPo6SKzUL~8i(vgPN?+;j%^FXc0#iqDZu}S;`fBYLfH45*!XC8ed;g-V#e5f_
zeT<SA$;WOb>W}#DG5!W{wNpDMj}rR^RtL5<J3ijrG|8gK@dd4{xPLRSkO@)0M!Ipv
zV=rC0q{^2Dz*VDJS1`5ym>|wX`eO(D#eP3xZ0bqG;BIOFAg6{SlFq*VChxrbBR#uD
zTUacZ-knp|9&6p^k`xfbw7GGibj_#Ra5ejE!Smn6;C#tS`kwE5lvf4k$528s8D&y+
z0~Oa3^s5_!A1N+An_f1rQ$j3-H81A0@|8|czhY%$1N9U%kdHYtCR2uqn@yaZm$N4L
z`Rje@o6pE9ho8{x^f&3g6<v5~Lk)mafU}G6ZyJWDl)FdAy3-}(G(a&qSz$k^pmZD>
zsCn&7S4oV$l<pn4D*9uwS~@A~-%&}oLkZRt8NsM=fJG;cSaAdml`a#ku9Ewe8?f#R
z?P0BJSF7+y-pguOW+J7JmZAoWW^uq7tnBSS0}mbSG_T#bq3~w({A^!HzcaoClx(_J
zm^4o{AZzP}d=Z16LqUykE&ccAAE)Nb83S1<!4pc1{<P`Ysrm!auF3fYr`;RGDW1^Z
z-mlz89$e7!n$#1N;=zk!tu_4S1Vo1Z^txUG%g=tv<K!d)$MSOG7Dj&`1S>ZNI%d7j
zfa7SkZFhyx=}j|)t!g8#p&2nZ7#+zdJMOzUy4UOAy@y;<w9nEx6YLtl)A^OMW`x(S
z05YPf$;n#ic*Qn;H**pkNTrUc(@;=A>iw=$4wl&%RK529@F7~jg%b@Lpe#_78Gi_e
zGN<R4(afgNLX$67^^@}kDwsPjuMmm8?KmRx^=glVF2WqofQ|z6)^)fq=W=en&ndMx
z5{P%&i9_<&BdEx9DpZUrHq~y7uEM0?q&{YuGq=X2<NOG-<b{JLqi4g8h*1Z$BjF*>
zaiMDTsXXe{y&<#SR3yrGB&+&VchJw6fW-B!^NrDCcjqxS|BFKH4_@x8RPs7qVW~Q%
z6u=yPnY!AM@q39rH{JrTsd(;CGdOlOVf5h741<3Kc%#nm4(J$cPS)KD{Av6eN_X|L
zDzgzzIw0TWd}wyDxTlBcs@S`h2F;l4$rW>9?#(OSULaWQ@}RMxd1Z?9#!^=-KDm<l
z-N9OSwIcX8+}+}ubApAvpY>f!eLugl8w+FD&lN9-{iaVD>yuUd7k3#@l<$N22dEP%
zIqkr4*`p~mx0CI$+A?g(4y#TRm6P3bJto_zKH-{oMT(+o?Gthp`h`_HAytP%&e!iC
zMFRw^8I)M$Gy4qLV@a_qp>LHnI{yF7am!P33CTXII=og=!<x8#JoDUNmYRhF(<NJg
z;(}n|%J1ut&=$+QxKeOwcfGkm@;tac%N3P!k4mZ$E)Y(y&V3MDA-=S~f&_%U^UwbM
zFPpD7-rG2NuN9{}IXxEnvMVRM1W*EIk)(858lTrfUDIKU29B?%=Be9zj{rg)&I&SX
z+b*L}bK@OG!6sz_fJ^l*m*Y>NN;@}@f&QED?;at0>wJ|h`RB(6G}2+VEc!#?`39aT
z%^yEaKD&nOqhp>CKB-FhQZoI+Gv)qh@P|p^H$2yF(%ydhY{+FpSaM$~ZuXF}@U7V=
zGlibQjR7miTcInwT^2p;W`Y9S0)hHEqPvEv%q9Esx!nVc)7LNd85!c&Ybjp!<apb+
z7ptmeeNMD|K8w5TKJmDyECqXt84V5X8{xZ#Pq{QgEQomdK{5X9P0fn-*FQAR^M#GY
zOnxs56*;ZlcZhP7iHV6}jqAX=dX;B^LJ2wf^R8XL0M()^$PrTM8U~oD&6JaYgF^{a
zqCguVm5_*ts$eJ!2F43ExikB|;R_x-21}h6y1L2b_G{m+5>V4}ajCrcEEO_j+8;o|
z_8|b#=@~WXbXgsI6i6YU9(bEe8D0ybGno_a$JsxQ)nX|s0Y&Ot>`v{2Zu#NyF#|ol
z1f^I|?`P?_4`gvsQ5gUeE<%+f=TFQ;c<a^|Km=A`@~#{Hm5guDD0REM5p&fAst|0J
z+Y*C?jRX5hH{j2x79JXlep?$d5VCt5Zv8+$Mto@tA6L<^>I%o|s1!BgPkKW`;eTX;
zV`^(|!kW{(X$K1B;TFO}bt$Y|l^21$Vhnlue#i3{;EJLt`iZ!dEKVkNU|IZn63DBq
zlJm$lfZDFqI3N<S&y76e)mp3Av$Khy!p6br1!R`pWzRHQ6Dd;wsW9ffQiwXTpyfmg
z{XeyVq)Y=dclT<L5od=%^guG6t*jAK&bSA3l$0V`7504~UW9D{*Che&K!&W)sUoNz
zD1!Fo_oPl1qxH8=3+*wfFib|+NyQ`7SOdMbs<WdNQZ{|5uT9|(D#mgE7*&K`B>&10
za(cXP0^Yf?fEqzRPYGlj>PINxPTz03!)yB-Bn5th)ECIJa0`G*L}TXy?kL431-E7L
zW`3hWF&Z@&*FZ-f+UB#CcQ*=cd-%=H1!Vf$<L{Rot4n>TE=+J1a?M^YHdkGDvhK;>
z8OYc!dfADUo<uv0s9<=J{(QQGgdXS0wVjdsv(I7vy^44%$|v{5jYpmGBsa!L?aO5_
zykw^wZ}vdLI}6ZO6DT1YF)G2=`3S&HA~LcpF!p5)76M%e+}oT|08zgH+rv)v0H-d2
zeM88YjQhSt%){jX2ZYusOT*pezVh|*l^tLfL>{}3xgZ-=&QOtZ*srR>0+TE&D~si^
z`Mh3zY%TyQEvO$65fdvf{~32)VUmvLme(&(gXP-|_m2=h`{cxxoZIp-XeBW9$R~r&
zTkNCPzRXoS3VmPXMV_U~FD)&p7aE3`L+KGn!uCwW7l$V^B8$t)rS>*4@n`66izA&i
z=!99nn7%4`^%*l&j+R=zc8I!I>(9N4#}vmeD=8FDQ_;a3<C<pvQbEy!@Kq@5R=UVg
zt4`E6^6deWrjy-eXN^0!u4nor(r;XNMh=@WI^5>j-el-YV2EdXfjo`^ZdkZU)U-{y
z-B(_+5*0Rz_%!O9MZF)d?lXK*Hf%&joluwkaarS?AqMepL#6(bGY6OZ6y*YcO2~||
zn3xy06C)1cD)a)*JD}~61Uvp$08{86Zmt#^105mkfxuQX<$~FJ_4qdC3Mtf;*rGcI
z4AlrAU)C5Y@K%oy7CCpCaA~26G6Y*}xF08Rony``*|6lSr;wV{lM~wkEr%r#-x>fE
zHXCs4MG%H#?o<8x^{WqqyGpKhHYg(ykL$C>^j13AK#a&~b_y2QQw>=W78YhV{B`&A
zXeD=ftfB;^#smOYb(TY}f~5=oIKWajwV9>ceArY@_7=Mv#^%sAZ|u3o&GT=C5Z4H)
z*z1TR4xAtL?LxH8sK$OjCncrBHKDEgRqu2vBAs?kC;Qa-dv)c?*ejKz$k|-F8k3)1
zVewPzb46(^uhRBM#Lov*%fBbuid|fKuirC}kRGKj5qM%ZxseddI?F+|%A+ig|2H%w
z$18-%NGOy!Z=A*HJYOsLSt;l0=C%teDB_61_fsQ?b!Or{peCV}^9K|-G)%GkME&DW
zeZ_*voRmv7l0iMfV9xE$E)2~QyPU$>dms*af^@s8$jQ{`$vRvuL55K~QGiJv5S>T6
zV3%N4!cwEFbk$Ft4@FI1`9xWi%@Ly(mSp=Kt`{Tj8Z>?1pWi98!9Xf6{G7(;Jbpef
z5J6oMm2%O}aLwH9{D6>WHE6rDV0VE-eRJNlWt#G|#Aip3Z6Ut#(eq^rW@%%V-R!T0
z2YPON(;{y&^uo`B&{6AXl=<#2(6rH(iEVp5_lB=&6TfoPC7;kVQTf_DF}M9{G^goc
zGK6k=0N^xVb6Jcdm4hqr52JYbPLHG{|NaMt9NK~SP{myBs#m!H`9z^u8WGM}rKQ~B
zJ-A$>a`U`Ez4%O%=qL<1fg%i>!H(3fZ!;ya1Yd_cwYS!<UDyBwaBg9NMx0c@xlrN7
ziv$9q0$W#5t@07z2Y9}(20>8Q0((`v5>d;nQKh|nEMZX;5{nVO{$rSUDX?E5Ey@0@
zOn#+W&APLJ?S=w}L%Tue<X+jlLatv=$Ll{KiSHwsuc~ZKT{_+B8A#{DN1eOnlisYA
z(Il$fAIwodzt1hu&-6Y~J3%MBrFm4Hx8753XCnIS*)nKp_P2zcXWdPNQ1M_1S;|Oa
z-XGL2C>vY}3k`j6N>ECiVZXMO8^B5GuT!ZsYCOG7w@7R`k71nt2Q2KrLJIB-7$F2@
z>NooO#OAZgd50r0akWmj;jRO)e+1o4Dsg}hKLiE!Lt5hA-|4`hl7rfvJB?d5K@Km{
z=>;d+c-KpZ#jZa#Nc_1YyTk{^nU%+WxjR|gxlxkToe><FvmK;SQYYe0L&^)w6aJEC
zbxMq0X&EW$55s^uoI0a{7r6#5KM**4j0TEwA1pP0PSPT!abLJGr%})=Sgccq{#=Cw
z2Sea+HY_v^;<jWc8J^hGhw-f}O#z$LX~IhtbaaiTg6@dA(Quyj6%OOxkBY_4Fq@eU
z<MXXuU9uC%d6i5e|JTY&{-HH{9<rc<J<M%ob9_fC1N^|Z{n+o{zbA#oq;|m~m_lG@
z>fIgUvtt;U?nd=QQwY%_MyE=vc+0lG^xl_SR%<0HI`3zFzD)F~{l(JoBcYr@k*)mf
z;e%5NWWjiJgf!eD5iQkL!%e)W+@h;W{np>xIaL_Al%4$X3i;Spg+h=S!#=mhO{tF$
zv_>8BDkK;R-V%R?R%_$~Bj_UYx}TK7IBt3@he^3xH2_MIp~_jR^3#v7i5V@AE9bhQ
zz!2|&?xO0l4Ka&0qj>DYmv3`w&Yb%Y*mSC^t9?8j>ag@Jl<!^0eGqu-kuap{J>(9f
z@DA`GcLU&SH{rIgq)38BU#5kaTWT-rw3SH|R~`*m2bEa9_|CO)Zc?#jvxV9vzE~VC
zEWHq6jZbW{I!mlUor=eY1rt`2%`w@Qsl1g9i#s6=+IztxcZMITh#@5>pNB)FdyhYi
zFeJD3yblOajAGWZ-JE=~HeRIxgUB%tmfq7Oe))I^xD=(h^ZDt1M2Ng?R}4s@8Ufvm
zh}A(&3ykeT(_mq1kCa%j^zmHYZ_#5Vayf=+(3!+({x9|D+TzB8d7r9Jwj}4~=Z(RO
zQN7x=3^)-f<OrWVEdzt3&3wDOqN1W`BwBK)|AhOX)2CGbjuqLYJ2j`j$`+O;r}<w}
zf3@;py`d1{L_b#a9d%Y{<~F+>sfxMs*yJO7@A})F^QHW)LxW9wdgwYuKQ!EN5u+|J
z81W-6G`c0^UP&z-`(P=_^7j}iyCFnSj27dfZw?#d2zdD<k>R~twXXmKq8DV(uwqJ3
z?6Mu`-Ajgo=n{uPXMDZ%Fr;5FXZ=M*+i5O~7;miHE^O9SipzBHZp9e0IIrDuc%svM
z%>CQLPLwwzzegkvnJPmgxfe`bupg15v4n;!j{3=Se6ZNlHuikhGVvuK>}z9JnB@_2
z6wBO6DPGjx<H?v#lSZF6?=zu9ggQI9@J5!$T{6*`8SB;0Vdcr$4_~#bd7a;W_K7NK
z)^hRjsDrwoUG$bU8@t{hTY2fkM{E^a;GrHFi}y6$K_H@Tck;~vk?-ZU8_4%9_kIln
z#5GsHBR0WK1E^~pwu^Go@d&rPpdeAXa-kwhu7Vs=K2&-GaE8BtJ_w`vM7iuNeuDL9
z0tCtr@88o|ufi#@ThBIQW$p`GuBgmGX0Bf8_#K!InIJE<vN_eTPDb&VNO5p>tj``3
zPu%_;j$q@y6K#Cc`Y?OwgSeU1M}POk2;tM8OuMbfxyBdLqT7ZO6FMe)(fY#&+a?l$
z6HfhBPCQGzdFDax<P);~XIzR-WhCW)#ux0GL~@<JY!$%#X9AJ7;La}*B3>O!%RJPc
z00hTO*eO{_1C1xDt5EG{v%$P4fF6dP4b8TPxw^XU071bJ>Jd_qn1E$)H?{i<0u-;M
z>f`m_@aQU{lHVCz46n13ZCUA9c5#Tc5y8R1%LBQ&faybeC+!SMTl|g|qYX?9j&Spn
zHO$LI0X}53)=*RouRffQtDpOYX8Ac;{S)u^8NYB{yJ8{Q+#OfCft+OLmsD`0C7PMe
z$u8fB=QG6^5Ny)%ME+h3itByf3VtF*%1Jf*X^|&K5!2O(w$+3XOBmvJBC<>`M58ri
zL^Es2t%BBP(kt9RL7qD}KHyxsRRBhlnY~SbDCYeb0ZA&YZ;u#>{ULw15&(G?QFDet
zy=+i9Pft8{-PX1rdSe})H@Kp?pYa$Ta(27r-fe_}MjYag^Wvv>5;(1r-=JcueCq%l
zCDhI49F2$%jAkicynnjcfS&`F3|QX=s^4$dA5Cj8gby!muh<zXobX>B$d$D*OMIGX
z%dPQg*M#;ifAfD&DA9e!lpqSHL+mQ9*9GRc1fNOAI()24%qS$N2u~o+{pB}|Eb~n2
z3!EKs9qIP>OWa&nrmk?<=m*}d9%r0?p~d65$sR0d9Gl`&@Rx2*fE1-OsAZr^LAZ^Q
z@PsKim)j7ez%;cQ+&l%jYo1p;kiyjo@iyvnMyBT<eV<8C6{>RQ9(H0ZGacJ8?#)w=
zskf3X(yWvd8a&q+*G4L_<F^YJ2jo*kFQT40MdQGf;pL!^+!grSHf2f6$Dl=AgGLf~
zQ%{3FH{|2TWT1LLFnbA{hY(2OBErH_QR}{V#60W)q^2^@Qo9=>Qn;D3ClmM_a-gM0
z#p5u_2bs?}uq<)!2x>x!^1C}#tPfII@z6fRnCW+qHCmrY)m~uV-On+$4=Wy~F%{q%
z^~4hT^yhwxjIyKOyOuhnqGeH~*y`{+DWX8gQ6iCjI~qq2qj9{u#Pv1mu+jMIbsy5+
zedujja?8F#3-l&Gz=Ea@YkA;RcNakrEs&0rkdf(EQqf8mqj381`%TIv=F62f^GG{g
zCBNA)yN_?;STHL$_d%BSEeL}4Z+Ov^Lh6#+63gqHoO&)8=_n6%$`(pka33G-xU4#p
zbWoiqum3?*T7}qukvxdvkm_`Jp1U5&YWAV*Yn1V`KBkewzsIGh0ILjDjKdC3%V9CD
z2`b><<Wqey@F}t&?UhELhIqYePBpJkTLHFx^=%*y!?G6*MO}TH$v}=B>Npfrb2|DK
zLRQi=D|UA5jcn+SXDOOP&kB?zT3<BXi~4eEcQawz^8D9hT&E3p5?7p_isQ5-ndj+C
zsfmd!0yQGT;h%Z4New3pxFT`{{~lH4Q`lg4fJ$tLKxZ?5$W%f~suwg|GdwS1wm$Z+
zm*`0M*atqiPt-+p^bZ0*Uwv<GYU1et9bBy#yvIcSE1yi_&F~er(`ZU0H#VYVW{D9?
z<3#q|kL<?&gh`poQB8}nD(j3&vZML}BgVgOHzedr66{NMnMc{{6W-aGA5Pbik=VBg
zomN1U7<d&*T-ha{<6z=I{P*Z;r8SLtr!7yy{Nfo}xJU!@N7izrE+>CrJGf;#=U?(w
zrVRs<a>ltcSK*^WZ$fff4&0CRuIaXMwzL|<iT8NOSYzG#R=Q34R_ZbnGOZ=QHcYN#
znW#z%@gIIw=%<ocQAp<{{wkeEXdaq{*c+5q(sht*Nf^6GXXQA}$*8?OVOYNR_dw8S
z*~%p@eo^F|=+Vf=u<zG-;XcrxG(h{Ni_S=PUajd;KaTv$g@nEtPX;1&mK4ox_GQxS
zkZm3L(Gtm`sfq~R>N(vIq6Rp%6wJatD#{$W(;(?Kx1@h>PLFkg#RHL`xy%3jG^z5*
zY0|Bf7nm}%bIo*bsM6b5*yTQjT>sWrsU4%d8u`O;OKCOXl+eXMC_PC^{v?M|!e2JB
zr7J7XnL{zx>@{>`%uB`oy)RLN*@cbQ!2-&ICi@f6LXohGe^3Q`;X3_Qj}o?9im66g
zeTM2K^%+<><kX0yqsP|v78TF&y8;o~zi;qmb(DCs8u~HjJV72e|Gg)gM8<4-;eY@4
zfaO6{3DX7Z-~Pr7__o|Lc$$FS{C~d#dE!LlZz8mDFJj#Per+=SisRpCkA_w&d42P5
zy@ZCKc|=)%#bl`*z&^e&z<b^%-z2~*^zS$1i1Ob4@AE4OFR=XgK8|$g+W-A`Q(<TF
zvIlm{l$^eTFC<5^d3=P542qr{!g`yJ9Lrh%zHHJ5K6euR`Jj!INd`-Qrw|wYnq!Yz
z$>lyQ`E8_{yl=N(l4vOB`#btee_ucD2WL(Hk%LXFriN4G_I%dIhILErp+TK*W>(rV
zh4eFzWBa=_Kku?deO>GOUT(Z<wuO(qrP=-WmupR~ZK6NEEH2<IF%b~`C9EPy;z@On
z^lsXPqO>c&Yw+5`h!aAq1;1qr<#vCU#<WFNXi^mm)Y1&F{<eS6w7sPJ(UQ?Y48`O~
zV1uV&Ea+ZOp-d0c?g!tBP(tI-oLmOCe(~}Lnt=cB^le^km6EF1^!axia)oQL(TFB9
z7OLf@s~x|ZM6>4<5vqT{m%4N-{icS5n+cqE{yhKQ`&mmx*MBKwEn)(9!Xu76Xe86D
zCMUu;Hr*hVN=IZ#mh<_MY<TZ49Zco5Upb<2^jR|hF6Jp*%-W8#*sY4~`UKqgDhC)=
zq8uBM3oYCVn;2Ck`TMFb#;8{XLPlO_Fhp$de56MI5ObO0o@lf!w|Q=UXwR6+-G47c
z6#l+)y5x?6yG)|ZYmNeJmhTwiTJ>-%ZS@)iOZxs@mv52QuJ*S=^@`LZA|mFNSX(=2
zSh`ZT7n~f=*m!0~f3!wggf(>fD)+^VnX^>JTeJzSOr%FxuZC|6rzNdi?kjp8st`=0
z^lc+%xqjW!{nTkSv*hi9=PcqHIC&^nMRcOs+}0NM<jrDk(6@yEX*nk?j|W3$_H#Ur
z%!+RoK*(g?#Iw&hqC-D(F;s#{NJs~d(*olgJNmda6l*FnGC>(_v|KOTtHl<&x^i7>
zP6pRFht?t%-aX`RGcRb#$rrj~T_9V-96h%A^Lz(uax`mI;!xm))+RET>d|1X;?8rn
z>GiE?erh6Bu58l=9S?JZ>YLYg)wvMc<0AEJ)?4&lchJo^Z(m@MYrMW0`Qq{8|HIo`
zhei2){lZuvN~j152&j~TfONOg-3`*+F!Ze;AX3uO-6b(JjDU10Fbo~i-Cbv+&-eE{
z?{%*CoZq=l%mw_x%<Niguk~4<wf4T>gFDK3v_+~%&dzzc5D4hK=puiFFpiOlQI{cG
z7plGTjgxUt84Z8man^)oMY}0Wv}@3o8$98WO!Z8UPB`4QyK@f=00QSbB*Fn^@~8bk
zJz`nWwCMc{rD_~8(v062v14^yx}Y(?xY>7S(wv@~y4Dt-fj>PgF_BU?LYzuBB3n}M
z7T472WkEqflX8QgpkUMrOf0bFIT@f%uPYNy&7e(-u%@qYF<45*Xq}TES70rE?OCbe
zxc67~c|i2g_h6wFzkKU>5e1K;a@zE>8zg;E&c-h&n1q}(*lcw_{P}2##IxyJXzZ(w
z!nljR3?Y%w7X}>eHX{1~rGOLIADi;N(--X8@q#Y8dVCdxUpdv=ljS!f7;5$sC(QET
ziux6=O|-*DxBs-3hXo9XEi(WWs89%@$|pi*20Xl=(6X#fEJg4XEg=@s-h!>VH*f8;
z=YK~!4n~@Con>WI3)SFN=@7bKdB`O^_Qs-vI%7dR0$0ocCSL7~{n#O0V{g^vL*A`F
z-<s@14ctm`b9<+okuCXR?u+n<`UsQes~uJ%UTqwn_Baz|FP#R6q60YBrIp>(0c?#e
z^GIvr!>sT{YO%X`nZ2pKm*(mTZ3(3k!~+@dmkLGN*SrX%-GB4fq^jbYB?rQ88|~N+
z+R4WJ=}sUP>|}vkHFc6f&N8QOZ2A%kFVFV^nN+gH@)bWnS+rraIC8qNH*Qs#BjfTF
zgXZrbp}*En?TziX)(LJcgUuRYNp`DWH52ZRXM-<CNO0Z`eUfX-$y!0>Mx4<Z%f{UK
z(^(xBFvnU?2_nQ}U;)48we@+l|Gts)eAw_Gga7|$vv1b@X#))ncm(LbcL~5FJO|SL
z`(k(tY}((K&gXx5qkpI6f3IQas_lQH<%T4XHhSMV?dut4a?TBsIN^pRYCQ5dG??){
zDo0vdtjC>}Ozu_BwQe(rk{mi>f>s4`Oa7<s>n8MH*5cnA@xM9X|GLQkgNFai|NrnJ
z@9aBX?MVTnou-C%{o=mL5yPIIDCj)`i%DxbPcGJQajyMcrDXW^>(_Z8Eo_?6E2pmw
z@WA4aJOAm~|IG<pEyw@H`2WVA{}qw^KUF_~fBwJf7g?PBSKm`gf#I0rwCU^YjqWfE
z?}({HDQlF2H{XZ<8*24$YZdgrvv~g<n)_d0yFdB|)?iEF|C+qP*8QHSprBB0H!D)8
zSt$cZI#ql7QXyesWP>N<jt#0-!M_h21FlpFyr$WdsivhRB?q)&6UuWZA3xrt0|m?A
zaHtYUFt8YRKPC(o!BzU}O3=_Yl#Q<I_%x=F^19{NOf{Gf6~qGeY;Jaz9OQpN39LD|
z7hj{^4c>5mpg93f-zL%j18KEXjNG;$bPSADz=5lPww2OFutD2^HdcQ{u)p+67(B41
zj=~IN??8k`TJ$$KLPOJ1HM%~zyA1d&iVhH<^YTtKBma&Z?fq(=^mmKOvAYbKuLy8e
z{z7apMtRUvQXj}7QHWy`v3x-wE;&O1tmtfShh7}8`8UfST{43LrP(jd1C&Zc&Zqpk
zili|sJ$9`zQ_mV?^J<$=?nxNP(O6wif1gERC)KSQy-MymMXPA>X%82GlmZSPC{G?w
zUUK)kFw2kZcr#nvkYiO2e2+4RU(G)}0|flZhezPB*a{$1ITjf~UL90l+e>;vNqQZf
z2acFgGD*lhp8V{50i$<4J^PNBITWR?^*_ui&OD437u{>Hl&)8EawgGVi@MEE{Zt%e
ziaWcyb{z?G1gec(E_M688D%?qdW;yJr+m)evA_j%<`n@(Nlm>vYEcZ1l@#)5KlqPD
zs(Krx)3|EO!Qyr1(gL-#<cfxp)__xY5>7g%ra4@&4Xxknc6N4vM+5o04Vr!L-5Cd8
zlajG@rlX_w>1Clhfrg$e<uRD~ax@)Rw2Oq<$VeUuL6&it6YDoglece&_8kM}jbR*j
zPFa+G>^|eA)r1PBQmS~}@skr?f#h?|{h5`Y<+<pN0rD<#(3g0VkF;yq$#?nCbqK*q
zhpG`_@sQ<l`cwE%^lr*?tisTfaHs3v?AtaRUb4j(Zo$ygf2rl_5CcBXa2obFMIyiS
zT5<EtPjGf3A6zz@57Odp;8c#}xi<kGUOM31tA5xS7#IL1v>04M3yby;963AV(H6>_
zxk4n+2$<v_NFtLNWdL6R=V8+-jk^_u%*{*Mc#b=ioGg$E#n=c*Pj08W@Ax&nSFj~%
zcNS$FAuhtsOBO<v)So|q_+_&D=mi7fC{%>d)5Fmrd6LK-MWRCzu1Z;*J?8k$DGTez
zKeYsmCi})^BX^STHZ+LDN<+rDsyAgj|GJ2lzUMdOW#G|?5sJeD{h?F&`Sk^Bmd7V2
zopYK-Q#!)o+dP|>Wu>K&QBjoQcW&R#Q9=mE^(;u>bKI~+&>Z9CJRR3AH4L~awEzb-
zw)D}f^iW~xL`(<x>l_|$(8WE8*$1O~;$Cp^NIf%-kTEo}@Py99&%ap|t85PEb|rhs
zG-K!#;fnH71U@Qug!ykq7ysO<q<+CnE!t>YW^Fnk1zJ#Q{-Q@nLDJPe7%nWjH?ZM(
z;x?n-qf?_5Ard;c^&!T+#_^A$P2bAX`tELAMGT~{nB?ATmB00V#q>O>zUC~~Bq}^M
zmoJ#u)B{3HnycoEgw0x;5*Eg|!%zhzwflojMsWwH*VlQyv=DU%rDxqphnBIq>RlGM
z{!75Jt&EBr)U?qwA(m*8$HpoYG@yK{Gm`8|l|tnCW<xP~K)}i7dZo@%_+MlkldI6B
zL4xBy>AFZlr(XI7<Ov7wGSmrF-L0B_1KEE@M;v`tU}9#?v!Fir7}#&|tG(rcg+4J}
z<Et|1RK_(}l$#k}vDA3T|LW|vffEyCl|Y(!7WPUc-59Gu!sc*Q*|JLW=?}}Hz<T<j
zx_I=L!<Ivmx>Bg17D)!ZmOo4Jome^ceI*c1jbkpzHxz>u*Mp7g^lrb?UqW2rUvuI$
zD}QR9yLG5_@(u0#`0O?Rv`}V$A*Xv$B8Mna*KCYZp;xc2uGyS8G=y>NA!tv`<aUiS
z`6_K@{Z7tCKQM{8E&1R^MJDOk`TjL|#i~XcEKhjdlq6%gRBNq9O5cJL4s_eYSM@lc
zOrlOM^O}xX*Tsz!W9XMYa_bQF<AY~cU6fRP%F}w-NM_BvzwX}nV4stbCCa?aVfE7M
zM7>Rha=B(fzNXMBpD7tqE~6qEN-fN?8qKRT=^d(a_g58&`ZL!Qnj9utmp;H%8<veo
zoE39<X+3$7UwolOA?{Y786?y>A`FuYG+vRb&0^c+Pl2RQ4{FQmB-R6OK@#B?W4meg
zq1#%a)=Xk2E1_>ocjNv7VF@`Sl`S^Yp90{%|9wY7fvD}SDGLg5VGWTOcPyWbM!IzP
zvSHf7dA$Zn{M+i>n$Fefj)+mW>6AFpkGY<Oar30vg0^33koozD&=!l?Nzcy#^q%n`
zw7&Aca-K>Z_;qMn1~88D)g}42N2f1oUJYiJ(C4YDhL@kVGr}TOme=|0mBJqqtE1*i
z?vU1MXSe+&OODt<EcKQm+vLTRvEsycw3xO$2e+W3Im4K?9dzn&MT!pTQ0)|oWfz6;
zE`z`Nm>2N=k-+_H1c4MUnHiv|`4Xl=*YFGh{=hB+S06NlcN={l_}^|pcm`l@W7_^o
zR2dJDWzRtb;)`)>*+nF)bXWVX>tSYwc=!r3XC|bdrv4;*bFm>(DW+{y_h|TS70+LH
znG**>w}<(vr_QCLmKyd;pCo6`8aCfk<ih-&t#k5rpw;155`q_A8qP;wCG%aZaQqiG
z`6<#vLu)F{)?4rV8L@yeAXZY;eCVsy!uKwJv&Fav9R*2UnT=Oe5$KM~YeoGutbrXF
zv<3!I5y{02Y&Hys91hNFL%C~~`{q+JGlGKF1%G2+B}xEuK05&eX|7Zo?KK921a=R}
zUZEcMF`P?|93Q)FjppOKxix=Vna+*(mk8I<t=0%oA2H(N?=eRPcBzPmCsogjR1Ts4
z(X~7Ru=p)I%<=oJo`+)L8yCy4bs3451IL$vV;RL2ma2t6<#-*sr19nTGx6~zq+r^U
zx|r%CHOKzCkr6hu`_-<_Q0+!6yY8u?f|GKkn)VQ9)^*yy8aD~JKY;-2n(`PC+y<Er
z&gvS+lUH<HE=P5Xb7SjncFVeMYoU}($3(2BzKcrCEOXbDG^;_~qo9)|K_<Rv$nuWO
z!jh?5t81Tlcjpki%xkcY_Y_~7K_QuUKl!fQ%Rdv?GKRo<^ez{kZX(-Pi{urYLy8i%
zeXNgX9?|i8C>=YWs<&m%EtsDSk$jHj(zYqm$k&6Q9@lNxeN$AcaFn}BN^-K7*E;2u
zm$C>+maCO*7=&Vq28%N0p0{ntiX&HPQ8_)-x~AKW<mO`;C0-<ldC$Flc;z_%_D=;}
z1$F}S3l=OZ=V6!G!zy$v#+@uzvh^B<tELi^#eMXn8P-=3Qi(pG$|mq@s2{BynEOSJ
ztdSSSRozr~YHo&at=qZHlrikjgz_)1E3fVy<T>R%dYFO!Opr;8D>^K<*=#xdN~J`!
zJToWjE=g~lp1N1K{29PBgb(yHvXkWe)XX+ia&P2}e{Z%EQEs_QV8s(F=i(mxMNb<^
zr&77Un%~6*3vblh(t}6&Cyx|Vij3!np>*?KBF?g^q4UB*44f6UR(?8Sjltq#3;K;L
z_hO=)Xlx>vZyJAlvbyR>1FV;b4lMZU*V4K6fdZr$6&;(&HS_Kc=Iy70y#B3li`A%i
zux0T3lAS@eEh5~olKUay{bx|%$iVzxz<(8PrDz3#L<TWwQO(Ru5`D|6niLglu43DO
z(Vx!V)4T_<Iz1XFvfOt+Sjk3Q42@3II`O<z2g#YR`}bD{UrC^}#-$&@A!|x96~RYg
z(NUR)h*NooXgbs{#q7wa2wL5A`DY)leYW}~l#Z^O-9W<aQqwK*l95%K|FT{K`eF66
zF@icOJSp|l7d_Ck=#3iZp{R)(8lD&J14e&rhL*q;9ctMUZRO_HsZ==Hklt<;f3S|)
zv_LKQ5S9&=s7JO=K+IC$n+F0>QK78-y0}tD0=f)Zuf~?K&+g1E%)bPl-{Ddtvl=Q?
zXx4!5E#rme>1KzJ=2h5>_qP8ep6VX-=fB7pud91r$uL!_tb1N9_rc4{ZOw0W|0tOK
zTM60Ka{M8zbDrY(3m1Ex7wVko*HGjt2j89;s>nQn*sOCzo;3nRi3`ik^bc8zx8jDs
z+vL;UAHVD7*uL*J+c0wl&)hM1mZk$DYD@>yC8VSyD6maO)!55Ft5_OpBeAjMjCGO@
zvNM(s&oTDq7f84gzNCEU<0-xzf2{^ej6(hM4-10b56!lQRUMZMz0W*6Sqp@Twed(Y
z*&tT(hYKcG8?S;Qp?`ehl*QtdzMkt`vqz&;YYMTa3pbUU8kqk>Fn9hT#@opw`1=7E
zd8oI8tUYHyCa!CC9TgK`!EfAnsF)os($;Ev5Hdl;3WZIcZeWneO7yGt@u<_W7jY*=
zW>g%WC)*dOz(3taQLAca1+G9y-qo+SA%0dYOqETMeby{@_IkK^;iLv!WwZN2rh}01
z<f!BqY2IYnZ|6<<De5G7W0{a5n_p6!IzyWK7fA1CO?|SzX5Q5_yk)Ncx%)gg4k32;
zfJ`A&loVBZ*qX20j82+IdL;!pQobt=CofIb>sUHsu~!irg6JBca}n&-pIKs6?o>V*
z#G0(A9c~*1pXsMPYIRCB9^9)<_*&w_>&{NZ$2nl!`=)ZTlACoUVpXR$g2oX#8xz{P
zxHSqQK7s>Y@x{4_3$G6}MLaTD8K&*Ss!_2K>Dlxfc{W9eysIxr3kd^cv&!;3t3I5W
ziRGvzCMqr#a{WN_$Cy{$0E}9dy1sc_(#eq`AvStX)rEvpj&|ALqKLo0SuZxKeF~*o
z?6D8W0cDsUqEz@8yco5?{v4!Wc~ge2TlhLKb8r#IjIfaXbU$3UpZxMv?f7zKOcNP=
zu3r;ELA8=KM7O7~xk0l&YLniWf*GTo_dz}wf}bui*7G_z`wm@kR%o^iSKMeB%mm8b
zTeF-PC@FMfd*v=W5CXSY4jX=~zPr0Wx@SY1($v{uU1`yjut_OCdwQty5uqE~=91%p
z;Z&rlIZ|Cs#B3B-D_Kwu#+oV%oohcMD5KIEsvVrMz6Scn#n_7)F&#XeI#+{+PybXK
z$sJ$iA&5Dj;9RYI_}*ZPb{B7N`h>nA8`1v6PE==CA57IXYP2sO^`gyOc`T0A(aCYZ
zK6u99^+Uc5=9pJbCg5FImujbR<!VF7x>EBLLd$>!3v<ZIN|R>H99f4@t*sWinxeql
zKV<T#-O(qUB#0;XD;Iq5HAU6<>__!iD-+)u6quReZf*4JDr&CM=S~e2SFq1W=Yv~T
zS3FyI3o6GTJ3LQ1-fs$307_uiv>0<-nZ^5UEQV^d@UEJlZhCsfRde0jT#Fq`b*77H
zqgk4$y?j%51#i5)n%BF2mQ`_lpo&OqZkn@&ILbGN?XCb8KF_9>=h~fQhizp3A@fyB
zF3eZ5NdLiNve_SZ&1jSC)Q0s&0vTLZESL0LP7$Qm#aX*X^rxk6g-&dww`q}f5mRJk
zcJ!fg1R^>z?e2ghRv<lg)l%`y>hg88s=kiMhb`-qvuCC?Jz4FaZo@4`p*Pa2Lw2h?
zm@jkV8f?RyNjTHy_OMuUlqE~(G)={dNq6?Q8xX`JmTc1Jo*%rl>qY$W9^mV-(bfp^
z@^Tq|8)dgr$SP^_@um=e$vLEN6WZ54c!G!dxvx8NVn%M_yTFwS4?E+$<gD&x^zrFb
z@~OPTMY5JPw@|~Q(rAE!+OcVK>jl;HQgDooX{e{D_;n4c+g`eiH|2FW;9QM8@A`F}
z?_F1lp<W@h>3QgLF<GLB<DcHnXhp={|0%zC5d21tRGv*8B==nk_eq~>QHO^sI}&s(
z&8B}{GCujeOg)$;cy8b4kh;cAS`3bg79Wss>&#sg3|DAF;5&*9BIz4TQ?g{dC*Oiv
zcQH(`EjamqH_R$2%s4sw3o}=&_s>-YYdF*$g*L!)v=zy@;HU|b(RN2JiYZr`;{wL~
zniEzF^RY(ZQ9GKBOr27dcxiV<1qyLGVm4c#v>MH$aL27V@pUN)1<^-B>)wl2POiOI
zzEnUN0#3@M#D`jTWe7DnxI)5BDemNR9a(HA%z6hix;Rv@jfn}qwk2=H?IIQ`b0J3N
zahaYK>M8-G6KR&DXlAy;cfvmk8y~sL_R~<|k_sE2gfvSpP@S5&?BOFwAs?Xq^`Z%L
z_o>U}BwF;PbWzdjbH3Ux5Y;Z_^IyEef}FWQfET!eRGvr?n4eo*6_AL#;BiqCAK^T)
zytwXJZyQEXznevGj@(I@t*{t<0~ZqBNBx?W=<JnOkO|rNdK%A`Aohg<iHoB!)ryUL
zK_P}t=t#jUEQw2o%!&wbbGdl#`OVp<p=RUGmDNwi!0qU>R5@fTq&;ZNp|FOQ2Q=>v
zwJgWfWBkXS-yv^{eAlDc^YdQJMaqoyNsHs)4xQ$H*W?Y5V&>l*VGv&-8|cu^{DVbk
z)b?E#?<xsez}AQ0eWA)i=6P2Et13hNn;2Fz3NoM6_7b~o>iwpE#9rG6;g#K^fs?b`
zHr8@jpF&UBEO=!1m1QrrcTnbjEpWMR7q($Or4g;KQbsD>_mdbZ{FO~yxiX*&p%%_h
z;T}yc$4aFY*xI6eY1r-u65)Oo=0%U(bTSC;&%6yO;{>;LE|S()?J=xNiYxbtCjX4Q
z<jJeW*kiD>1n3X_u<i}n<yn4`MwQg=!J4Xsby1DT5cG?-zBXSL1hR*!?a$2iOI&yn
zhMCbMe0h8WCd)%?g<vfPrb%O7eAl;|tuL;#MieBIy4a+tknk2Mi-xZLJ_(yAGHF@?
z)xl9E^SA-^J=nbNOY;nfA<<QK3rCd9{7OrFWJG!&yG=}JW6ElcuGTf>Kk$e~XkJ1a
z4T|iX`S=FgZH5v&wR@#OX1)BZH=ZTMY*xQnTYN>Uvcjvz?91fZn4;_Pam{Mawg%j@
z&2_P(z2g^UI+>~pm;)+wux6%m`9fy>Gd7HJab+;z#>q-HF~-Y5{a-62wT`<MHdT(g
z6IM%I#b(V(^-RRUanAein~<d~j6_&9P>HfbA*pjdI;#y8K#Ql7p0e?WWE_i6WG8r(
zz`o3~8Ox0v*9zVk6%Vu_%WG))SQ~Em#ZkV#CJtJ?S3kU}Z!|w|v6597p0Q|yXn}LD
zz4ZVLlqP~NJfrPGEiJ)AGq&x>3#CIh42e;t%j1di&P`sYEvb+uvJmSd{R{znl~1b@
zbL`=tVe+V0&fL1lPnR)ebdUbZE5FYJIChH(Ah+fzFI5oEvH;v9{p6?Wg2J93tC$-F
zn~2T+Dy4;2gSG2Ew$UO1;=bC4)g-d~@`6`R1=^Ck)<hW^*Df;iN}ve8IQLm6CPL4>
z^PQ*@yLsJAuHrPhDX3JLzsPoI=NNv&$zsf*QuD<<L$>AB)#AbLIj!~_PgeGaBNllY
z?AAvC$4CQKNM4(JxzMX#0)Ctz*#C8-#H|F?0zP*K8%rs^HCRH~K$+2XPwdw5so8nJ
z&ETVb-1(99Og>o8HeSu);&FHewKY-RQ+~u}nlc5q^T~G*!2LfzBJEB*b}N_?PaEU*
zPb>%6kaw_=fv`y(##?_?<>obpX(KP)TA4t0YlYoUuj?V#WPR0-NnP#qBu-9_iX`Gv
zgBjKQQ7+>P=NRRdT`IAl*sy0`HTsx*>JcCvV%(f!D0uf8GniD=8T}g0;+}+)elja=
zq>F>1mchN`DSbtRIRFJ38Jesns!v6^tR)!HI<7@Jl;zubx<yv^n$)zd^ow9e#>M8!
zLh-r_yY;ak26{U?75S@8l(}-VpOqOJ{IwV4kOWAKz4_?%nzdEV&dx4WCdt@UJfoBH
znKpFEE$qa+J7S0N@!t1hXyuraHYyQHhI$>C3S8X`8})hXO{=9EM0(t2)RFo3?fwfe
zn>M2AJ68|i)9(wEw2ZQ)M;9;FaqCzvSpWq5L-%p!WEqDFilspr`FE-FdHbooB>036
zbMQJ|W}0KlhsjW6zW`y27pOAa2)CXmA|LNls5(XwGaE;{j6-wD3*zBAE1q__DjOGL
z2!seyTFR06e0zxFI9B?Kuo(3?^S#(@Qz+!Hm#Stn8AJEaFd!t1t1!!V^Ws!Iyxj4(
zw_?K%T2H%WQ@1d}JuRr}6o|xSK%{_3Ir!2&any4sbMly5hFg}h%lX88V!_r_vgu{;
zsYS4c=UFwxr-4vs74qI6t5QXlX=~z6{ek)d>LndS({F)cgxfZY+=`X1sc==QZZp#+
z7fM_mQqejSx0-|mNYdg6IF09r*26Q#Mk{CzdAu4#u&t0<cqr?i?o;O}Hp%W;;|A3$
z9ijYDR}qUCh!Hw@dkd{JR$||Ff#B^~Jk&!qix4`<*mEo#E|q4zapvSY&^n()%~l!$
zs4KH8p9<sAkq>j~;*DlsK{gZ~cri-*v^5m4ayy&ZkBdl$jJ@a}EH{~-f8<vm{!p)E
zY5pB-<<~Qmd)}PHY@T(=$!YW9UU#Hf+$oBNx6;rzNwZW4;Pq55g3P(OPA{mH%-MQl
zK(ly?TPoNBb?~>0=e<4o^sTz~-e%EA1Gu>W7?DQ3dlp|X(F^*EjG@gDO^Rvw*t$et
zr^1(!40<Q+yPhX3<|zRd?4|<^=k&p|dCFsnADog~v9*Y@W%S;U4hH?K_(5?F_Y@7o
z57j&%i|nyRWn!;(sWm}auuNpO{0+d{{PL@J&)5D=EAG_PT)uM-gM0Pq)F13)(M^=A
zo2kc2^marw>V0W3RB)^po9M3$O{#M>tyYm#Ri+~vM8OP_Q763a04n3hA*}0hl6i77
z_OOsRcNJ%AoLE3!V&Z4M|0vtbOS0!zpb9NX6yT09gKS6!FlZPU)c}k=2Ejz&xbxl~
z&%U?!+^oKs`fFMH*6Pz7mGa0SDe(>WwF*kA<;`{Oz?*g8qvKqm4&7<G=_?deHL=Es
zt%B^XEeXb7Jr4oVRE$;IHIVtIBbwV?iSO^WZ`ei-?^dv2WjCw%ck+vT@V@`K%f=In
z53){fQ$BXCwYZVckW9~W=G)8K%CAe#-mUmL>#hNayijDd{k`)c7?lrq)}2QfetPDa
z(+{HlWZ!IOKz&$_z?E5c=uqg<glml9koE`68OBOey%qVm`g^F*X@c^R*-y#|Bk;7g
zN`1qR4zSbo<)UIgy2t5y-DQiIF_2(?$(|IbbBIydCg{c^RjR-*oTAfBJ#07?BtBSE
ztUv;zD}PjSHWW;>(m;!`^1-5rX5Z!AhYu1GlBoOae9n*H*6MQ=fCdTAAU9jKj;Aga
z`Bf}04n|wg2f}EzQl-<-Hd+QyF(l51?&E!S?qnCQ;&xCGk1Q)(jfcztAWv79j4`{Y
z+2?ZnodH6_hXfCEL;w%7)5}OSTAQt}Sz+s%1QJe5@Z~A<`3j(Sz2OrZUj)O8i;Lq1
zbJJ7n-B%E>agCMD^p!av9~P65J9M8ki-oE2t_<c@N=g#}Hv7RwuFS6<9&M^Cv3$Q(
zM~qG*5CYC$us0;;f^e_6U|dVQSNSX0f>9=A7SI{ydMJOY4FR2slA(6JSKn*w^@-iO
z*$T^H0Wae+6;jTk1dsSLkFF_y*mh%cct&r_M{ZEtdI#sWoQ6vDP)o90!QFG+M~O+<
zie>#)cPj12oF@)W;<>`VsJSGLET!MRM9W;5qk%wZ2K1KI{<q=zA>hz{rKhu0EmAI$
zRJz9xU2Xk9T+*RtZLa9eUa`->S05;fgL5pCHB0D@>*;ar0OC{nhjuR5M%LmM3t#d$
zGiG8<Q*|HfFnAnfmKF{*t$Ozu-}>qi^);#x_~ag4;_PFK*})<+JWu`ug!#!OG*Y{|
zMO9q+VRnkb36@HSuTd`WfdB?WVGr9hePuG}e9Xk&s--84K4uF$Oat$3Y{ZF7^7rpt
zmmXlFE4AZ8!J&4~xstV&#NF|>mg-4|Knd~r!YMD4)7WXvOJj<N8U1%M;MvSGj3po2
ze(dbG_F%0(X^-jS6P64;LGeK{-Q9-c-nf{7A=U6Wb(*kgOKTBvUvHu6g!qmQkCQ{L
zhLQ5v<52$AF-2;md%|TJEhc8&@#^MUt?l*@BKEOqk-UuA%T)PaA4T5o9@hXz@GF?X
znD<<4$d%xx;JrUTM-2Bcbp%ZfUyg7}sqNigES~x<M>9WyV--6N^@q}17d=+7z8<AZ
zzo@5m9s2mz_62=dE*aD+c|ZNg{DArK(lAd?_wd&bBsYc$JwmM5(_{N35Xi7F^^uPm
zCco#`lxEQ-dCxV=PRDOu1XX3hyepMb^#<TC&~!-ni-wLiD=ZB*io4`<ksPt{+xFRL
zBB#ec$92lm)UYY$phw?mxuVJFKi-ds>6;7Xd)HgV=)vnA-P?fpy3b;POEM#%b=~Z3
zOO@K!RVOJ6^5PqM7G@u$y5OboZz-ypP>7#z89P_V{f0@T{>pDH38G)s$K~5QKf2Um
z4^%Q07*~AWu5lNOgPZKQcD4K~u(r3Gesw=Y%hZrIb)(WO5{0&RGi{&i&zhd^S+aS(
z4!y7Ky~0!M`G|gJm^wK^S4#-*?p+-m%f1)<WN{;-<252BuE%RZpXX2Srmk_qKH$ZW
zAMLKQ6Cw*G6WcglUPdyKy-KS~yRLZSM$+nV$TpqF^3rcdTC*f=jn0OzE?$_`-PMsS
z3x#SeAHLi>fKKj1{h}_-@WSlgtsjqc6}43Svf=m)PUd}Lx;Ax|1vc90ISD)!Q$aKS
zfvJ|fPJZdBGiSRO+*oVFGlH_Zf29ll=46W=GUy3bVb}+Le!R87oq+K2d8+z|zAvs3
z=5x|N(7di<|LFVZ@Fz?<j}cbaS7THmt-U`!l2xoj)76<%D2?>PmS5>;Y935qnS4vZ
zQ%7#>h9nK`u^R6c#+ATM*J883h*6x~C+O%&r~}V~-fH%J-Tyhz4qle-AUV;oqzl_v
z2-b~GRmgWOGT<)a<BnjjbxwP0WdDdIKZUoj&?VKtT`1Wr;Z}W(H&MexT`;?sc2*f=
zqDh<pSr)uheQ~@s$XOpkon0Qp(5Hi@cd^%I9^`J<Erv#rz#;5Q3fEUQXWb1kEBYmA
zAc7>~pPM*d<A&}V#0JebW7zAr8fKkV)<lip`XWx@Xm9sBoo}L=?G)3GeK$=gRd3w8
zB_SxNV4x1i_al?r#q*s&NtZ$g(dWsDjTVXA&~MGy@moEtJ0&H?GJ0q+$>3HD?Hey?
z=nT%%LWey!JQ7<up;>hD1y2Fjx3_%i;_a;t@H|E<F$9ZxVFjNm*jeXG*yuRfJRi`C
z;xrC^e-9+Imw4)&*m}$`>P7WcoXktePMvVm3HSL~>~h_c$M0cR%Ytg}>-A%=aioCk
zhop3Hu$~rIeGS`Wm@`#7@=0iTdfWT&PYOoZH2z1CG?o0=?Ge>+DVcYlLa8ybg4?=A
z^M;&-J$N`d9hhVmj0?u0FS^bfLxlX_4A5?H-Mp6=-xqj_gK}ps&RnD~VW+~8rKV0I
zQ+B`}8%mO19%!Ju6ZP3_%vq5FD$YT1P~%0YJ1Xqnxs)xg%V9pSw|v5Id34u&dZ2+<
zjq4lkf}bDP*y9evx-q&>{GzzGrfYk~hV2##dfDO$dPJ|EkrFi4<g>4Myh_Qn;`Th%
z+3D<LWM%q2Uy^PSM53v#%K4lbJo+~obc9Bh%+=QM!_ZLSr7R9%2D9@vz~6QhELdQ_
z;P9~}EngIxPek_o>W;*o72PqcWy={f_iJEI_6~oLX!U@JDfZ--$%{=H{Gb066DZR(
zkd24Ro*sNJkB+OPLT4g4+-S9FM!#CIi{+R%*;y-=ic~SoTRn$zc}hG(S(KVoK}{p$
zQYOThui;J(I1oSI{k!DCGNL;3iSx9M6~_!UVUgY4eF&jKl`{PwdVEVG_y#04H_^Tl
z5^Wh>>bxQxCiYe6n8W3#d`T;5Li`^V_TJ5zndcAcynl*0qbD0|9^YjtgB$PXr0(a1
zt{NEh_}P(vk)Y_-Er}aXi(|hYAx2T)vg@R^zNQ8dy9STA<0uprcv2P<9X&s~zDdIE
zTax;0bpa6=5Ej<UHui|_dux;Tr*2EX!Z9HYdAr0A6C0@~vux{cQWY38ZEQrGkmu$X
zv&mi`Br^5=AX=UE@e&yWrP$&A&_j1eLG1361<k|~yIix%2ft-ExxVLhsjjeye9-Eh
zqpRhBwAArX-NT>K`b&vsFS+cos)athc|H>(IaBAhrIx6&lQ(Eika&Axc3=_A;00Lk
zBNO4^ZAzngkCQ$(v}kNHwPFP#rht!|=!coL?s%PZ_DL6RxxGinw5*5oyCh?bwBHfE
zvuNFxW1mr6u#A+9kFHQbKOwjK^5MgnsQb{|N_U^|sexRm!d;RC5_L}d+h~&PHifKa
z=eoPSJNOTyqPLAvGEx?bo;L)vb(V&c)URF4JtQc#MaH?a2><8u>-=$X6l6u`#P9n~
z4t|WP!G0^n(@Sf89jU-~ZE5j9IMwzJQyn<39ob%^uc#<!@Q5m+y@cMi^*oN$=;Y4%
zQ1;Rc>)3>u><Pb5K<a&{{*diM&$)SGL~H<K<CLdQbpYm~;=pqGZw9Zb$G}AkS$Jr9
zUF+N6Y_AIDei7NVygcOx-)zWq2>Oy^TJ(cj@10@0$+_EL&52SxFo#V8qn``TuB(u7
zWUz`9s_gzE>R~lEvw;GIGW8FVgr!DH1A&MOt>z>sp|?8ZTcL<y5)%J6Y1J|#zHeWR
z)mOG2cOTD~pe31(-QiZm^QwL*6N+3Negra7ein|;8DqUH*mss=chw#fPD-SCDa3w9
zF6`^Z$fwHv>>Kb2ptC?E!I&c@)R|J)(Z*zwQ`ahbMywx2LE)Xc{-OFd=ACJi6Dzzp
z-%okDW-YB-TW6`O!@s1kili!R_k|gt0llvsLS~N!?d==qgg&~H86HDh+?+lQY{gjL
z&4C|H-iI}?AwW0Q*BTh;cdy*=|0F4EGe>ymKP=hP+lQZr4W;i933`a2$5@NZfL$TR
zPgW0|y0YfnplvSxHa};^_fvir|LONi!bX0N<~>=&oZ%aah>EqQ$-zjkqzxA}b^RAn
z;aX}x0)7}ixO3;4>Bd6q=;36Vk4+g@Uk9ou>Ij~{bi6{WB`Kx9kviMW?fltGn%C6W
zA~-7SJ|ShMmV_iV1uq_2?wh8&=^|U}fqlqP=rUYd6YcBF)E#%l0gW^`b99GGg8}y>
zr+xU8^ZL7mlxs-+A}T6--j}h?GB^pRQ!F~8u&-0i&qwT-{YcmiOyAi^qNs|hzQ)Vv
z$=yrVZJ#-MbB8t}sg;S4+E^+}D>nm`ygFjP!cr{#Sko#&p*rF0x4z*6%;dfGM(zpk
zuTY6615hi%GCOWy$M-B&LTaOKJ)od?gm1_5EOt()CvTk%=irRf?M07ep*o&juzq#I
z(H<60Vq#t}ysVU>g;wyITO#{=G!xMiS2tS<zObpowb{3HtWaelpN~f2LheCfNuCiB
zCf$vV9gF`nI#ednXf>C)+<LTJqdOS`!bw%iU0PVkJ8bnzaSb_W{ZUeBKTs{&7#$H8
z*VE^`RSb6&k+zf$I@TQ<dc{lDbTD8N%Ura;MDZ`dHZ==VxZFy;{{$0hDaBF{8>2%j
zmtX#^vU#jEuB*h(aA9%Q)T)qqa-#7Axvpnx@3h0KiMptW`(c)K&Sj8>7RHn@54Zl~
zOJbgT!o1=9Yy=B$US0a`x5;b`l}?HGb{gAAX~8*e#0JsRX?{Cduhzf|AAN$0!1}&r
zJh`8QS{Gs2T}ULS#tbdwV|XtdBeeN=(f`?bEGnJ3dB|$s&3U8nas%M-Z7Fvs8f&n?
zQ5gWyxLEszPUX*CozQD`#%l5l;L*^Q7A9|U<KeAINpbKQFvS;(ZqK&7_Kr}54_#v0
zJ1qdX>{=k*rAZusy<^=uMEuL6?tsk8-Lscun<nb>-nf<&HrCSEU^=NuUd>ET9mj*@
z_EtP<IF$>o|3~Fk)-@t*n}rt9>|XmNg%_LB@e1p^PNJdgP$vg3Pn@2a;TvbcYxcg^
z$$9aTE~)&5HtP`ktWBu+%tc|2hYx6gYxxHhTt2uwN4VXZ;S=i~EYPxgXEK)JMIqq8
zO{~AWBi`Gc$J%%|F0L-U#_63>B|Wf|4~X!EC1lv{WPYY{cD{nfJ&g^5yLn92vJ6y*
z%fQyiGp>e^&WhMGao+M}ak78@&6kXv3*%X#XphqSGkt08DS`uRZWv$1kQn8I&bfIF
ze%a1J%O_5!toHD4PxqFW(F$tq6YER1f6U|SSsQ*T4#sxyzP?pby-NhXaJFXIAtmvQ
zk{g4+k_zTQ`hY~+>IvrSR9^->jiYdF4M;*wgdq2A3vl#WbluyqZ*472b)yOKst0!D
zoYkWy+k~+>*JLg7-H`4>S>_4l2`iY^rgf*a<_U8|Y}*nYLy?`POZ9p$qJA3~NUMoZ
zq3|M1SJ#m#7Z=#!Oe~A#uH@R}CBNeSS-d8Mzu49tA@X)`(QEhUkE%w<b42AJ)gm1p
zH>|GFj}SgZ?@I-jBrbH5a@NV8gwwL@dlUu}J8yt3JJc#N7i}iHr9v5QQ6r+LQK=`f
zT4N=p>=c}LQl|~XhsH)VDoHpTAJ;!sW=(5tmAB#B)MTElOj)p>d+1S_H=AoVTa`Ck
zZa%HHAmaB}ER%I|40@ZHi8jo#rStxhABIU78~VnZy}0Tz-VJr1h;!vdIz$LD&41cZ
z+$1o1Tzqh$I`YSx5I?_&$k?}Dc8noxLq}CLOdQ|H=_{}`6P=4Wl#0rGXkHc`sJq!=
zPHEqsQP7}@rKEf;ky)kjmAN36UAEjJ_?I@V4m0$b)stgnc>FYARWMVbWZm&KM8+$k
zU`+Kk(}+VubD+Q3C#EepEmQQo^VZkLM9+%lz9=hb(5c4)Gu&8p^Y_l3_E&T*y+3Z^
z(n_AV*34SCFCK@{M$JV>k>|g=RpJB3PdO)=kLluk81o&zrqvG~(Z>4vJF+QZW1|RQ
zbrFcBx@=1X$(pya>ny#-nxDxG4^q@s_T;dr-t?T%y4Fr(b;Lm!HLGT!{uV3ll$(<_
zeyyJ=p{Hk$l)AX!iLZq6Y`vAiJ3%24s@Py#DKJU2a$8ujYVFSFJ0ZT>R1{os89`DM
z4=}Ama2)scGhy6N5H>b3Po;a~uhGvrk&yo9IefLp%lvJ5?lNQbbpIglr#uO7yj5LZ
z0KFU{;cacRxSY|@R|vK{=PqAtrZ)bWsoZtSQ9F99WPKYyI?Q!-N`^Xft9!%e$K1SX
zAg*|yA<Xef@e>XShyhn6_=$IYE+!2;t}9j@Yzdx(C!1qut3pL`Tfbotf9b;XJDSK&
z^JKESi0<NA=UF|g7WTeAV+{nK(+a9_^fLgzO-xK?#`)rrWU^w$GaY)MufgG>!t@)P
zhil!Y2xGDLU9oYfQ#$kEoSw1y1B<~s@0Z3s2okZ3yY^quDA5$IYrX1rrOUZd#$De0
zL%>ffcTU3@r`6A|Y7`n#4%|fcCd2(h?j`-&QXOxmVJ}>r*7Tg^Q|0E)c!0Tm>^m0~
z!)}VSKzfdq`?>eFwfKHpdJINHxVD~q$T1T&>r%~OaXvoQ;=A=#t!aQc_GgporX5(;
zH}ebRVPac$NWP(o(w)0UYsvesrphW?qq?KBvbu@*A}h0Hpw`Jwh0WuMyOTxqZ!$9;
zB!!8?4yoYmIf6)D)&07EiQ_gdKL~|LtM<lusGvQ}?B_*j)0?J!zy+{WmNj+`e=P`C
z5b9?hCdBWW`SKda|Ks*q&iz-e*3mk|eiL^=Nc1#H><UnUsgLmbtHI{E%)d9ae|ZP-
z={Uk{yaRnX?K8uV&aUIMQra>x`udb$x?CJ_clqN2+!o#)uHbf3+1M#{85JJ9eilrO
z<+02Axr-&B7`R=(W0x5V6Q4`~F<L0K^6Eqgk8gD!tx1A7H)?juLwyxpwV4}ZySakI
z5<%Dl23%LQ`5F)z*M1ImhKE57eC1B1Z2;k%?bsI?&wu0Q3IS>&HBuTcI4P~pYyM))
z#?*Yu*(|fZyXi}c4BA>1Ve{g-`}RP~>q&06JH8UW!6l5%ajw5I1MbojEJV^g{4xK?
z_9@#3X7QJGo^uokUmfG-sxS(C0)-bb%mof9L&yYK9HRE!DGymdfk4~Gw_H}LA23xG
zB;W;fh`m1~jn86j?HY;wTF!FtriHOkdZJTuTsn9m0gZ>%@Y~SWXM%~t0MCqB|6qC1
zSX*zOkSh2=gC8<Jx>u3N*L)Dq_-buJdK3HZVOuM?D$Ta7V}YXdlD0`7<L+*8I|F^A
z8kK<yGs6jsQ(I5}^`B4y1%KR^AE{MuqAf?8$lMRh7ADv&ctSAViRn?}Xkon@>-maK
z=~6i{$RjFDoBZfM0sqXQtzp|oQ58!KuJgqPX?<gYi;x445yYeO@*!%gX7(GW-%fUv
zC;7eNl3{DV-4q^h^JyK))iGJw7<Ht`kc_})o?FEfL(ljSXg6{rtF^hdH5NlORx=uq
ztY20~s_WeMC~DJZUF;VdG>tH__SQWSxTzwwy-6l#O}O%TN>@Zf*G0X#)KfZ=oVn0%
zVtgESes|fefAgM0<66hu#$4uj2u3p=e(R|YzkK%U@F-yy5dt6~kI`(52u*fTs@$Z7
z(<;aY_duj=dngR=$tUTQ(-kK}<%V#6SpUt!*(z?1pUntXpz5#P&e?F<sP$h(Z0qzD
zbz@@vJkdC<o$&tD-o3C+PR;Q!Mr@OHb-TQu!V-8Urr01VoC69Iqg~gnTV6!{h*krX
zQAc&4QjXH_*LVYMVfBx(J{aF@_ca=L+@aJ~2jzHAM!Sz$Ci;e^-;a_e;H^O$L%yoe
zi7)GYRo6=QdmB*!eJ5hHG{$dTC`8cFgVradExNzRkX!}%=FT`{HX_We#p?BKfZ_A-
zTDXp;xccq$8A{vTW%)E&E+5H*Y``YlTS{|ku2uEs`l6upBMr|)-RbyF`Xa+`Xi-(?
z_wqa872Y=YcOHGDb!8s65dLppS4vVp=_xNZL{ah~-XqA14U8?j+l#a}8eiS|M+H8*
zr+SS|+!Z3Hc|~Wm=G{VJY|IF7MO(IP(X;COcTb~slup?aVmW<>aGamuV9xq(dc7%B
znQBu<S#IMET$2v}(fZ>n#H-j^`b||eF<uY;t<q@Vqdceq!e+e}74Kgbj4g|eYzW`+
z3Y3;K`TaW-NL}jCre&s$dPR%G6d+2iHs^#m##_3Y#QhGtOWP~NTy0ynju|!r+>z$2
zGlc3uN1nSZJj{Zd$s05&T*tClXvvLLe7rLI)>4Phcjt~rd==EspRypmz~`Y)h2}qt
zXUD=kKW#AUTI~L%H!c$>^x9VzKUmEzd5Yp`2^BW!#Kf<ol?!t-(QkLd&_L8=%hIqA
zPR?5WHM+5KQ8)|@(dSv%T6LHu-Pkl3NvqE3LFkb>IXf2_wSCe!i&n>D>=>M{ooH$K
zmDw3<UV@Tg1L6)UP3lXKC|sDoA^I{`ZVQ;rD$Zex;0TG#I!@Q;Xb!lEKQ$Pt*`|_k
zpS>c-W5&bd^RUnA9rlS4aydRtehorOUh~`h*vrMN)tZ?BzPAuK#NR*P_X!iHsN0$d
zeS(RdfB0n6H%lhu%HltFz2$H~mfA4&84ys94oI>fq0HZK%vsW!2M_;+aD{rVT(2HM
zp^l`X7LO<IggU4BZDT(tO}Py-tCkRP16wwB1LTFz8WtZc)I@_FUxM<10990}*V=72
z*D7Rg;!T%Yi>D5?jzZWEBO~=*L`-z$4vn-QK9Wn7dtDd%2!T!f-i~B(-o&Ii1QJg#
zVP#wZVeFLUXc`DoCa?X}8e5MLS(PfC=<RI(glNMd|5n<YXGKRV*+nz=p6Y(RYm=C_
z)~bq$5>41QpgP7d(?Ftj+E+~iZPlh58#oWHTz&4Cv%^cebeUN+t*+j8{R@fhQ4(Lo
zZd-1s?$%`}jJmp$Un5Amz0e1J%69@trY_;7rt?(ZecmGf17*1KWD-}v!iM!)V89~s
zhN%3YT-lMe1UwFqXnKIIsMAh}Jv>rM+}`Bp!-t7tJh*!&_M0m;r_t!@jCt9{3ibjC
zcks=lnaoATpYlnmRRuxR2rX|Q;_PSVU&r%Y6Wd~kvTjXbZv-l2Lccp05UJ%_I*S#D
z{oaJ31A`r5bnI!ahwb7@PEtQ;m1uM(p-5r7fO7Md#_^6()yBL3k`pfVl}rBgF-Xd5
zv;mJ<_=ZZz<DhpWD~6N3J^kKf=!ZA-CO<1X$rz4H%6jf&n@=NC(A?MyHSG_b1KnWh
z!9Kj1_7EpofyTfQUP1QxBL+KOMod}*@@QPLkL5|lm+oU~WrpvKqgI=c7V^k1_g>M^
z&Tf*6UQ=VUe0soY8Lol{Xu~UX^+BoAmsoyWOrnU<`G9@u#+LPA3KuoZ<`58eE@Ntn
zatci-Vm8!c_SMzQ9He=WArqh|mn)sg5q^+XksUev&jR~%+8v_$i*W%7c}8@Fm*()e
zSa8b2ARa&+8Qp5{%#!;G0Uy<Cad*Hz0(fYZCxp2pckjaC2K9poGlf{GWZf=a6D#Xp
zurW*E-}KZ?ZyK;!Q*%|GpnF#ab}{IkWs#h@Fk)FaFjvr@pm9r1&<bZNyj+B!BLZdv
zgWjx+Uk>|jG897|(%0_H*<aBQ)Xr<dnJUcggSU-wh*HXWaL1g##6@%sFZ6ufn36Cl
zV3Vz|Xa;G5@Y2#Fv$kHhO2nfl1t5uT%LmZaRivIlS1b17#o)+-u?#6WyQKlJ-Ulkt
zmPM8f|3Xz~?$D>3#ajYro^vnkeZ5CWMv<=Yg$a?9<DVTZ0kqqgT2)@@VJ2$)oW5_z
zr6gJbzg?}pH|8Qv74{mDowYIF>4`z);O2Jhvd)jzjk)$dF*iWRQ_5^2f#9G5(0e7T
z<F18_Q#K<ZLb(~q+5S2KcJGZVhJY0f0MtY?9I}9d17!Q2GBV>4;~D5|JWN4xh%gcE
z)?i%Q+U?b5ZC&Te=Nvi@l}@QnqyzTq3Q>RGv^ZKCR0kONi=3R8zTh(x#r`YaRFL*!
zAy*p%FmZNe<<n;oaluHLw<1GdpXgR!K1JKDff=v<kXIun-X6$&t#MxS8HGz~0$D;l
z^3|>x`1p8nNgX>sZ;KR6Ng&|F)nO4`_sC>>`Hvq<7rx?rKqTVfw-MD!H$K66KJh@q
zt;zwU$PN(2GXW&0!?x>GAQhJJD8`nO)?XTx?qZMf8l^vt@MPRRs%D)TtdumJRj$My
zePUHm-C$1$xJM8qFfVQP0JiZigz(j(qTr1fOw&)Z-{rU$|9xDz&FpxUCOQdM@Ojwl
z&63=+t^QrEQ`LNA!Ei0vdyqpJGLsoqf@{M!QzkuGDh_%)IQG|3AOQSEOG9vJ1nmqK
z-LPyRZ9%R*=?|3r)ZJ-3fh6u8Zisn5g|cPg3ji4a&fN+2qv~aZU_A6sWbD1A_epx_
zBSsp{HMSC?p5kPb$aN{x<+he`EI?_eJ9H5Qv(%oP+Ku51e%G^+wREz4p%S0UrvrHL
z@T7Q2z#_1@Q6Y$%TjJ%>q|CpwhD%ULS<Zl``t)t@hG+-%m)Z@~7{n2Mi_l!=56SX$
zQ1R29`I#vP6(%!G)ZxFs-&5x_G+x;A0%81E0v9hi>4R44_x9=D;$MO6=`UU~fQ-jy
z`l8;C{LG1Osrz<}wr-_4>>e)-a`JIy))#zIdj3x3J^vKW%3T5KN7VFLM&I`L2wM-=
z<mtRn>eBppnkO^HW{oY`q$b-PdP8rtX0s*VojHoTdqQgLl6L{g?+a27Abk{MH5s%?
z|FM+z=m|p*Y3}aP`FPyDvvY-L>&Z@4VzyFXT!*xg7dHX{8*C%{xN8-!&A=$jGrcOV
zYINlOA&O?<Yzq%s<NaRpks^e0P^)hkgiHLBZox@bql2S|J#!D!6cn~$<=Ij7+Vu-g
zwG~hZ4xq~BC4{as=9(QXmC&5&n68gR?(%ze0h3kcxT|vUxvo_=3{~E$5gW@0+qtPj
z$d2xSckb1G|DW{kJ<?Zy3~~N8Cdkq_+JZ;i@lT9vv7S>>u(g)&InVe}H6oFP?=VU6
zZV-}aCve1mmD++^U}3KfK86PJ*};RgVzojF0~mYOt%wIzQJoeKqY~2gcQd*RxPR}F
zW(Q<>%A_8s_BW>)yuU=4>={{Ez9AL`ckeS-KOq*S=AYu`D;{;skB_HSeoIHNx7KvZ
zTe~&uE9ZrYNuBxacC|to4N8utWn?^vdKA8pxRF?0UuvVut_s9t3;nh?gx|D@)sm4i
zCn8meu2LQ%b*?O*u6%bx{bw01K;Rr<U!{OierwO!yI_1YItn~Edp)4c95u0LimZz~
zV9z2g0-VO~^)?NDJS0o*Ot4G8!uA;Se)p1-1XTpD0i(y_<Y*|D9fF0qHkd;85-UT&
zD{l9JE-!arG6Z6S<n%VN&d#))axCE+ghR#)jusQ)s%@b_=W9?u3r`wMmv}MaPB+|~
z=}`~x$l`7?m`a$xx|qV!N#;nZUu3YIR=5qA8$h-F)-&>r31s%+b~3cp6O|AD2dUni
z(UqZa@jWf7OY<I2d(KnvYE>JxZ%w}trwn~CtF8`aWUbGfpS!lCZT;K1D5}ko`Q`Bk
zN*9BHforym?Ah;BZ)?3`X#?5lg8kk8PPdZYg?-H#S9aiW3xa}Pe6g6?qG9Go*z;i{
zMCcb51ON&4AfM7@wx78?g#5<BP4w5)-!|r-vCInS8$CF4UUqeb{TRP;(IAxqb>G=4
z9^fd7a**!IopLZVGH%V7tp7S#t6D#bi~*?+8!4hb;)qsu2}pVB(P*!>=d%08#v|gp
z4p|4Q&4XT^q<c#L(!-DLZ?2}<`|~?mtdc~mki3CZSe<u!fu%y==WRGVwYr{8M1xJA
zR*~k~YeuGi3wntt=yT8+Pk29*?KgO1Ph;vR?V-~<h2u`BS|}IAW3B2WMc`JHM@vIT
z&Pt8ps}PW%NHVlwEVdFWR+%g1q1==CtdpiM;2rj&vWlxVvD?O=>4N4!2@x@oN^kbU
zX;D#UWE|fsI7Mbpa4@k|!cM21RB-;(6qTO{><7ymq?R-hBv<p>3j`^Kj~)Uu78o0V
zS>#;d#(PQ?b?9B>p!zf>LPIU+O`7<SCh~RfWit19gU`RFdowi^0M)8Zz>?5AS7aC(
zDN+<yDdV!%mIuWgkS&OLJa9Fdhn0t=sJ~=~JKY7TNk2g&h1Ve9d{4EXlb)-PKVlE<
zuv|XjY`qAKc|lKGyB{k>lbH<>w;rxEMhK(7GuEnftFgv8K>MNGoBb>=BsbV(f-U3v
zr@r=S6iMyKeUA!_>N~ii7$T0#-OSB?_ol}RWX?x4)t`ll(MV#x*mOxmG~|V}_4X^Q
zE)ryAMMlQpCF|N+ux9Y}2Yv=LUF8F)8s-2mBWK~t3M&fB!up7aQHA%Njk6NBu(op8
zMa>>nDM9;Fzgmd^|NikzNXgvEN2OVnL>AB8z5cW+1`sVJC*PkrCv)o--V~J!TmYeC
z)Cjph(j|1FaD`)bXQNI8Rpy`$QpCMEYJj2MJnC+1jSFj-?7m|fn4H)H6+);*Zn1w=
zGTb9l+@}olwk3140cVJ8U+`UfuZsI5eHpiF#GW^<uGHRATI#mFlkxi)jP=APj7(p`
z6qLHd9b#-x4pfI!tRI*2iwrb?(jh0J!?S&2N3U)CERVgHnlgzZ|E;O_fT#L>|HqF?
zN=XaZ4I_l?O-W?$y^`#ey}gtq33aS&viBxi$x6sL_LiN!_y0Qe`@TQ_9z2|Qo%6b1
z_qeX-^}Me8erBX6REOL-T_p%_Z~W2x!EhqNb-D@BH{JVv4=6spG_S0MB&xT8d9#sU
zl+6{{Abx93Ryq}Z&`YMTr}$N%RsGjOi^<-zBk^21nl9jOsFr2tx_e@0oJt|`Tx)F~
zIa9NX&mny9?$4<uF7uU{r&f>4m0JA8mP+gw&5>!o`RTlBod@1|gebgc*T{Im3CcPu
zAQDbX!*{2ED!URL5e#|XQI>X$Fm+!N`PFd^(e)fm>%Kw;rffOQ>?o^Nk2UPc++ZfH
z>-4nF6WKm@iro{7o@&{wvs>N$;8#cw+Xc_oas+evT~b!iks-%(-*<QW1-uYsmar8e
zdraz6JKwt@`vMHIu0`!j=EM8QhZf%J@o$xKQWaQceLel=SDUnIYA;JcfGW)0ySe@%
zDZ-B&UIAp;81Yp&<)vTWdJwEYla~O2+Y67|zy~;(kLJss`Y$X$^HBd8E?~;{%cH&o
zP4cVSrf1^ggt1%`YATAyP)-4lpU3v0)6l!`XnfoQ#Bd7gY3Q~ib@6z^Lcf)gvpu1?
zD(-MT)A1)zlB<7s#uW7wA4-tWk1i88-(;2ldGX|gO(7F=hCXe5WKW@V<o(Q=eC_yT
zV6+GKB10?_P1DYwpqDa@09#m6jDB@CNak&<OSsUWas;DqV?Z>;iPY5<ilEp`a&;Pd
z>J$(Yr<?y*OyFHsahAQvipj&uYWb}+@Eu1b(H>VkUe*Rv(C#j%mU@2i?v;&R_=CJi
z3(6L9cs-pMK{`I2*thmPBD`?V?K%o_d}&-Wdy04Zk%_BWEcz2|xi3H6=s`;I$ux+`
zaO!>wSC=#j@;TgRJz3>|z@a=m7I`Esq^@&Jm0B)NX|6#4<dknwEVCJ3`Jw$Z|A~cq
z88!8CDR6?DlUMJx{WP6VJUR?s9>r$Q?p+8>035)*E5*S04A|yWaq)|D3OoCCJA)s8
z-KRr#8Ok3&-tE`e(A&8Qfc~ldP^KH#<?y(cyju}i_Qf92Z6C-&`t}T$Ewoy-R3L0g
z0CpTpooY85h5LzLjtmPdDY%m7JpbJ=M|<4Ban^kN^(#?8Plog5(s28mynKNtB^Hmo
z%5^MN=>7{T1Qq^`|H{B=Ruhv;4Rn{I;`{^s*#zi=CsqpT=VxZzB%lO_+%yot7Shd;
zwrWe=W<DVD?RE<bvk}nVKzz+n*fd2J9j>&}s@%Qq(VIt4Zz5y-UcX2HU5SEX16X<*
zJl4EJ^ns0o{hw%~OY>h~$YPJAQ<NIF=g!s*NyvT9u4nvKNlQXp@cR!fDqN69+QibV
zJf?Gl#1;~N{1v=cj(<9UMMl(^%P&OjC)%Gn{bl`+jQHN(<>>JCU(TcSv;q>dH}r)L
zF0yBYLPX3LmOREd+Cbl-1an_inb2cI2Z*wTL0k!P3p9Sf?y~4xy(!BtZay0sW}j}5
zu5b59Y8k%2c=1Mm#%bUKoP*;&AFH3kg4VTXH?EEkk;G@QSsza!-KgEw9?kXE5}W;0
z=?u}Y2TgLPS10E_ay1HZJ-=T{Oq?B9CH`bOm}ZOXq@VQ1b^JJLu2c5cu2t`wTHaFZ
zd>>c)79a(-glKcVDROyiA?P=YsZ`k+UV-(EU&k>~e7Mfi7q}@zzQQgR<wdE{@@w>{
zRiGQijE%a}G6ay1MaJt+eRbw7y56a3;Z=P?Z<|d_x?1Z4frYNJ-#mk|k&F7pi^uZc
z@SbynN{fk?_qjQfJ*Ttw!o|fYTs>}|u6D?~i`!-~Z{slEOnv9<>8LXuyLv(nrT!hw
zPa9|Tp&&<-&F+5QS;utc>aMS&aTIQrr{KMa$NCPxQ>;)Lhu0o?-ZM+nW=QVOPz&hh
zoq1;a7#FhtpTB&5<!vmyNobnKaU7^0VLQ<%3l?Nzy!gD<H6`aFcJ^l@{Sz`4-LFFb
zP^5YtMz+&@N($r(CY-~oGaTCfprWA?eCJi|?YuRMW+QObSz~co9NTFT`eLW<@n1oH
z0`(kQ;2Z!{>Rz=Uj*HEpN^JCRX7?o`NE*3+p4#2q+;a;I(PuH4{7d3Q@$+@I<jq_q
zE0Va>)b=&*_qW@|mSa%Gk-5PO<2<`VH3_+u-kF(uUeDYD3$Urg_~RS!mlkcOkBT}<
zd4~JUzuAHRpLlB2Fr<vo_hl_rE2(HT?)p+IF-8<6sH@lf8OxI6N)X@`u(NA?KFrLl
zf&4i$W;2q=;iZsviFZIlIZ=hJrzs#dq-p%A+5w}E1c$PdELHu@l~k3xs$cpCy7Izj
zRsZKwmudfo2k1gOr+5ra|M3mBw`yA+x?Wn*#8pU7hLHLA2$ViGvS^gQ`P{{>N3c*A
zs@As16;zufR7*g8u1>r5C`eS(URL$OwQJlgpZvsb8?=;5S5=*yv!h&71@?(_wT}+0
z?Nc-39}{`P-K5l{Bprn#?m(;=AN5aQ)NU=_4tuie+Kd|H=YPS!jvp-_*GpP^xTOvX
zR$Z^bw7H#>#|4;?mINi9g}8Sjas&@&_i3XcS2#yHd#;JI@g*woP{zZ+p{QyamUH7x
zUA+Ofe8<E%hbbjk5#oMqa`<M79$;KE%~j&~i!(X(^b4ULfi10RV}k@bZS4?Ae~65i
zce9Rb=WdGd@c;M{=fH4^w8TIF;Y%^%pTfbPwoA1dsTWbVetI*^Zga82f*G^6-mfV0
zXC&nND^a(cfx79R*;;=oZo7R!`%4rJ59kc*@m37<{ix3```K{Y++yPhQ1lY6>dx8j
zVk)yQxNlOw`w`N8#3$w9Q1yj`@)CPRzOV8S%*oLZay$2?BxZJqv;}#Nt?gJ2Cr+OJ
zZO1q%RXxM-zm*m!32++k*>NTYtl;LBgGq9Hu%2htSkVP;x<;r5ymgOG+(Jf4jo5;a
z`fGd7Yv)SmUW=H5w&>38`oKlW7E6(9*&I$93O7ZnHU*V3z+qh}KJjF>xSgwXGw(Rc
z3&;~BK<-TU@eUW3{x>sjj9dwSYGj;h{~l~=f?Aeqk;P3&S!@YoBggnnOt#xGB=PtS
zvDB59&bzcC4u;1Kz=8>NtO&_GS92;M`T;Br1Z{I$vu;3QU4=~bMv}y;mr?nv!NcRm
zEP#hZ0E(0=FWvVTQ46A_K~$ydpu~(NIdx;VROr1<Z6M^q`1khg#slOXxAT;;_RNpy
zXzF$z?Hd?c{&Xf0o!@QJ&#rXyx$XM6gd2)+tDHBlQ=VwFhGkPg0nDny0XLxBjqdGO
z{-zBFr)MW5)P`VGKJJPQ-qUUapjez+*v7y-1qE)qtOp@h2u~7QXiNX<SHVFycZ|~@
zw$7DphpTaGPR_#%XGAS$3ggrCB`#Kbi7brqx@9G)Y?<mJ)XCle*6uiR=Th~c`50=&
zD}?F#y^&&ENF+h+S@O2yHpcOV({m`sTYndzS?X>RXse$+cCPMQoNfVoX1Sh?tmXSa
z&0kqeC?q~_D4|K_f9NUE_Ad?d3~F>xg5<x{@@OuZUcINR#Wmaaq1B_vjEQN?qgv}H
zRM7yFnOb7~vLr}0!r>#T<Ls_IJ6q&ZS_gYg+A{hwn5VRyU>-w~dvx5a@g`2x-(Rd4
z4D#`fn;XcLD@lqzyal-2rUaS)J=>I+C*1HkoGe``>|N|v6Sfahu7$ef)j@bq6P%T+
zvUAS)zoTR0xIowxWpP70GSb$5nmyIykW^bf)&Mv~rf+xF@(YtAbGX#R#h-_U+VR~9
zn8cSw_w+QsrJk*7{`v6q0#{z!g|XQUggbw=ql6+z{x=}UQaZX~yL_a+vnO<T9ai})
z8cyE~lT7(Z#`d0C&#k7nrUN#dX<4%f4M(r4RB~@_znB6G&G+%s&uaqP2KPcb+gor<
z>^Y=}@FzDe_-E>*SU`SEuI%RMN@pabzX8GIFUi~wa0r!0%?+dyC}tsPC&SKPB4fU}
zOUSdW%rOk|4o)G~X8vNQ7<?Oa+JB0wYO>UibiO=__mr`BF5{2qfkEzVLn89HEhj^2
zkFiyehjt7UtQJfjme~dO*=cgVyng%ho)qJCl!N(lQ(@oCulA<O_Nifurg~Ng*S{%H
zvqlYQEsk^)Q4SU^tP_>SUz*8jxt8Ur2b46RG)r0JePsU+`4!vLV)OPzykPsgG@+A)
zg?_1T3jRmTnO?y<+$WhHfgTCYOODQtka@?VI6mEarY_YyIl}F2%YI(rsBQW?w->LE
zRXZ}Vb04+=eYDXXq)YSsxeAjFBtbfRhXV5A-NW^)7Q?fv<32&b(XEjYB>JhvGaYvy
zvM0YNx^3VyiLSIjRPWFTSc>~65<J(VV$6CJSEQa7qJd>wCn!Dn2m0?r@lcJMHSyUk
zm+2!X8~NDp0frK5T{|KLi+}j+E(Gv;_R-k@4ZUeL7Ot-!>r)gSI9<%A=(?1do06A9
zaF7-_`@m8#i&ftU(U%c~`nMI+?&0=u2n(zFGlZ<eiKCFBvXqQ4+A9UNzez|8t#aLo
zN+9-EKiLhmO6$0fOvmfU%iM;#;<rElXtOE#pMmP4Dgi60>X6B@oweTI{)aZJ4<IWM
zl9T4+|2FUrOYOMY`^W+|!wS6hU2m`Vi=C8`l;(<K@<>^ndi)ztR$$G(LlVBIFZYi6
z-As&;FmiO32%q*S+O;o>vwj3p22kBnbYy&y#CkYczOs}cs_ef3x;Y00aJ~m}iAqpQ
z_u@VWW)U+t<T_-kwl_^|kyjEK6RS#Uf=RH2H4iS4j#hvi!mruy{Xna1AdSgup)J#}
zHN=+d77pOL!)hV3<ry^#PfR7Pi5dFUXfJ_4+|a?3mX-|rXye>mio?TesL5`QbHRbM
z96W>~&d!588@VM_V!E22J`HPqj8)bUGj-quHt~x9|F6*|a`no6jOE!UZM_s}!EGmb
z7sC=Hk733ni>BPpN&kctPd<afHQSQ=^`pi6soGaLcXYuvFocmOVj4azkKE|4CWrF2
z=G`;5Cnpsd!=QjHIVZSm^!!>S-#~x<ix~&v_82*9#VZ*DeAmN6$1P2yff|U7{v~AE
zGN8IscAVw>Fyzm6X~AuQlF-fJDs!eD8zBj90`l2yVWmRYXXJnVKR)HP`_#mdwEk`f
zBSW_(w^8po|I|!^#_RY?(UD=HVd{;f+%ET&Pd~D2{h@d_>_)HsGM%`%gY`g6JM&^$
z%g~6_kF20Hw)WVW7?s$h{dET@xgE&(x{)3jn}U0Edazq0!P95I+;Q3bi_;q=2H+Z9
zdJKw?g3JwlExi+a`Y<W8#U<6W{R#pj3xk(Az70auX$71!M^;*{vC==XY^>F}g3JsB
zB~+*8_9)K(mjn}i<bNjb#h7MfB#<soSXmwtxb=<WT4SD-+|R1%x?0!JPV5cj{xI{F
zyhBEhIlGk9i~Pb&XS*`Ca+UBqPmu0HMRm^<5)zC3Y(a?=$M&*zU0a`n%!-wm1q~|E
zbaIlR%-IsQ<rTEReh;^tmSX*iZ>GtbqHB=(XO)a|HFkkQvmN$#7<}JX3Eq>DwYASr
zbJj-QI9~eRUbsyy|G&P4Lm9BBorf(ky*M;~Q|43*<<>GiFTzKJlBZu{lGJu{$>-Rm
zm{sZmex%nuM%s+@t_ow6`<KNL3By%K`$xAwzF?sx2-6BI#^F-&3F$_eDh9Vz9k*$@
zxu9Nf(*~vu9W0K>C+f(F8FQ-{8*7F5y*k!X5!Q!K{4}U_f>|C&NdcN;!V-$@ESpb)
zL-dx4_W6<M&wH25mye&;l2cs|(`wsKYSPR%C+N(`&@%KID<Zse`m}K{MSEp8-PH3t
zUN`>ZyiH{Sl{XbHFJxq@e4SZTblXWh%dJ`(&z4`?`)MN9XS|fxQ?<=rEFuB85d~m?
z9jq{D0GjPI7V@b*kBuc6ApFKgyomOqS|D$|U*C9L5YRnNj~k$lcW$I3H=P$Mu^;*H
z8{7=xn!M<15PKS@--#m%KTK-Ht910OIHcx2OPl0u6>7VEnOd5Zf!agJB)J9KGYj}Z
z;+@mkzP=;JfAr{3n<zY--L|ty=H!p`XJuswe}9l1x!dj*oEQ5U^$=OTrmQJoe1qdq
z+u-KQl``wJqaxPPK^gh~;JYcs?XOINmp7=Q@#}&~>YMK|V6uXb*w$jCCF`5(yy^We
z^^$d0%Q8ikSJ?40_(G}{!WBK8Z>+Z_Zjd#98q5R8$eBuimhjBe&X%V=Zrf|~^BYhL
zA}E|Nb|(Z$E-os54wa(7VXVVesOo)%bQKS~eF_u@9)ao4`SaeDKBu94>I2mKy~}@6
zy?<AoaqrJh+=LkmnvYLDY3@;Yx~yk=h2Rpam&UH08@&K^gPZPCR;BUh8=?MpM7h!K
z_nJSHtR%qZ03cS*0W1=|pbz^7nlUEvPb(0FiG5>(ba%TFyP@qN^?&Q?8}bHQ>nIlQ
zWVN=ljST`J`}$d>k5x5NQRf?MZh<KIlj04G0!iNkx~Y14lT)jQqmQ0rI3lwZGF*&z
zaKI}I^4N(5rrcakqd!Yf*O7C+=^A=({+UMg5cPZM@rVyEe#*)qF3y9Y?)PB=OrEqo
zxczZ;qNV7y(evKQqr)kaFCRmsA8=5=Ipb7;$8lZQYvDW!TTcYlL0i@$JHKqPp-(U0
zhqz-a*skPws6mmFqv^bLmOOVPdxsa%FK>sL`K5o5-DG&qt?YzH8ERqnrmu=78A8GJ
zj||d~`IUJju_&;poV4`~;qOJ9>I{Xuk=tvtI`UE#@*Mx$zGw4y?GBXAod!aopZr?W
zA0a;&hs7S2lJrk(<Ygf$8n6A5?!=31B)9KMR=cyq`QT6>4@Kqfu{>&w>Wq;kvp0cs
zG@YMl{}vhCmoZXpZ2GkB9glt!0#gXe3dTF24#C<TO3s_gmL8|5Z5&B=j19W{QT(W{
z@?I+^6*lw@H^#m;c@EVTWY}!$SWM;XGmw|-bKfdX(0*z!G8!fL@s%B45|8z)mXloU
zAlN!6a4?2M_$4*9!`1Z*{8P-bvYOlbOhE9(jd*^DRV;9fn?g#XI|a*TFJJY4J#AC<
zs@z{PcIcSIQpocSbDG=~lzTL;-O1<38CsL}CJ`u^-D;;~xXUmRMvYXLT_gHG>*gYc
z&0^QN@Cy{Uc6z|(B8YI<SMy|HiDa?q!P9hvkvnK_5#E}Q)PTzGi8Py!QjTMda8D6r
zrbav>ocYVA@P}oXMGa)@EF)#sn-0jwBTM4PK~+HkIn3I`p57(p<)cl5{S~F=Erosh
zI8S?*eSVJGzg401Up!SF(?IV-TI0pEH$4p?Wc`ljvvD6)3)I}gjFgn(70*8F@u7`d
zp$z87wzv3s;-S)G?duxHryI?yMMcQ8xDq>gQ0y1|Zm-`SyT4^1Y55Q$kLo_lae$l6
zj)y1*uqXxO&Z`5kx#Lq*zU|j4{+v?kff5e=pKeh~zON*so!5IOuKf-T$J~gAeTL+I
zK^#V{)<43efc6T2>O42fsoF5exBG-4r}7=1QvZ>aT^woprl-4BT7b-qr(6#5li=LG
z{$C72#do!&B5tKoWU|Mc3r_W~+UnJoi4)v=C@*32<EWBq^SU*k^U~SVV{Q@XbvslA
zr>()4JRYx7?*30VLKJnROx?~3e17(}y$!Ff3FB)o;t>$|cqAk@W!f(8VB5WyivN^1
zfWV|f57U+3$lH)|21t_(H@6r$I}GQD*~47oW_F?G!8;HaUTCykRXW!#d>jKDmkf|;
ze?~%&<EZ}b1PMVaY5ZLK{(M`k39YRbl=$-fioV-RlQ#sezm#iNlnU)l_xojr?=gRY
zt$;QB?jIZIIUV1}{%pH3z4(ZShXhK98C37SMLJlYB9?*@lObUDzG!OY8w{W8dMACO
z)(r1XrIYdEXF{q?gCi3LRYqZrJ$eY%P~KU-d`U+hbHm?pA^HFg;%yt#Q$VpzSJA-K
zW7F`iUk`TO-7yx~So(WYU77*X*VtN9VLgc9u~m;HWN5(1Q1%!%hw#EDfLAtXh|XiH
zF$r!IS@qK~RpETeS|jRRbjKv_lhhXTpM@DHFx2aJ{-jy&x4yx|%C$g29<7v-T(y28
zfA-DU8>D1H<D};a(V`h*dU`<_kqlR><53f}3B}nL9jD(U7#kZCx^RE<QB%@-mza3#
z4OeL3wPW(<J`!_>NL6z3Yc%>Q>K-#Q8#i|ib$4b696KEY=ZqGB;1`|{(ZkEjS3ZvC
zK&?+T4|aFg$u{1xu-!%9pfEQ`Dc$?Db^}X$?`Zgmin<~V=I#2miiYpgLauF4#s7fV
zk7{4Gx$px1-m{UjUZVwzYUo32AXFJV-HQM_dgvg!%NrM|FEQ;<|BlRyV6S|nT|?PJ
z;tgA&VC!f7a(X|RG)w;t@(rokOI5=RDA=k8^E%3->|&Uh=yv!PS1&?L)|3*hSXFDs
zb-=2V$p-br=<*P{^{!8fpU4&?Jfz!y%p{c{Mdl|Df8+c1-Y|Uh=$Oal=8Am$o2=Ma
zJ!I-|$a8<kZ6;5r?UkOcYnk?k(WB?#p<SDP(_i>_FPAwPCAF<!H<wK~fge`EhGOvI
zz!baP>@t<{<jI3?kG_6yeoJIi(_YA<HyvLuu?I)0DHRsNCenuXuB+7QByr?iuY{Kd
zxm~hxC{Nw^b;}!~s_od?Co)=ONQUM>zouqqPm4be<zV;MGSaIpwNr#V?o86c@mtp~
zn8?jO(YP_2H*%8}n*f!(<}PqpU3KGXe}-OG>`B09WhK<y-1<r0@`1o$k=fxBc4?^r
za1dL6bRg%-#=%+F@GXYQT6S)t^GyBF5APeT&x@V%r##I0Gx+!j8yek_6C=takAb7L
zXG->V{r1bJctS!X+u9qx^_)?67jP@vN&|pIaD1x~sLptOSAW(x&nPJUJ6BFBtCVVd
zN0^YFo`09y@M+T}jh=$hKZq*EzUS*e!NViosK`lcBgtoK#M4AXH^|Aj8#n*YTwkV=
z#6lVs$it^1gnv3VUmmUOX|(C+x*BXg=}Bi`W@u>c-%Zrd$qL9QxB|t~Y*j_aSqsOl
z>>HfZLKG$0g-|AG^ajb|-<dWwcFU57sVM@RDP7Fx#@&U3QZg8(+#F1Ff|twGtb}E<
zSgDGC9-ew|V6UPtXPinyd>tJbA6)74$mihNnTN<foaLedC|<z?OcaDUPg>d^=BC$S
z+JlIxmSGcw$Xv#G(VT3?Yf8A&>t}aVVY6<M{6iNNm5~kKz;z0<eV0$%`TvD6yF6FV
z>m{l|EUsB-djXF{@3V^6EcF&R8`!_J`g1i=hWgive)TD}T^oD*9whs6Rb&Wk;pusx
zVoimT{X^Z}wsU4e&(^tTc!S5zZhmn-*IC2}pVlAuu|{O1ftmTBMmMzk|23ON(t_{`
zl`5aFFLq<U=a@_MlF6%&>8h~L{W%oGusWG{>=M`TMn_-q*dpujs1Ci<Q7vYwvgs37
zdeammk|2bj+@DR7^+0-V=A^`Vc0xAn{6W)7Na?&(VzJtf+m{rG2XHmdTzLQPou&1?
z+i$;hsK<6*8MP(tQ<T6PX2o?J_f-^&dB@Iw@6{_6CJTn>yVu`YGBXRLSXCsq>>a&q
zQ>a9Z$aQ;&^kEj?wb;iE^99~2cVDM4n)#gxXYvZyB2-Ks9SOOZjgwB;GZ?C`lKgO_
zLgf)7_K=EID~hiH>?u_3zka2~sV+=h<RY|5Fge)jO}}JC^jm-ZoNk$!x$K~UP{bEK
z3qq<cE@djrVXa{574@BhyL<|zauH=Ji=S>-!AYG_#q|q)Rc0wOuI}B$l3mSSo>?4J
z<3wj2kfm(##6+#NR-BX=-V~>gSHhxZmt?hRCGQ;Q>c*~Nn^8w^dxlktFDxW!qT^6C
z!;izkK6=b-quSYAboNrp4_(PZ(l@zgoibU=Rk9sTTdFlh3eGixy5!C&8XqX`#v9yl
zTW{|~Cri}Is(5`|{5-1>?WA$C&+z!;1cAc|^6M6c@=E%~sne%6XBEp@t)lkFRL5G4
zqDX|AYMoaw&?!CqbmE3*8!->0{&ouU^{Ttm!7Ei+BWCjz8C%o)o8i$!)@AX|<d3?h
zJl)Lq6D(B59TtK}^wg}qyxRP5S-$D^9qFl;SY)+DN^I4=QzcQAJF?@KPO9CNN=1@u
zN9=j%HGDeUcEcy$S@o)bDeh4|Z{fg3v=V-&Y4ar96J$@95IUy1c-E``4y{vM{rvj(
zi2L>U)oK<3f=;|gHXq^*bb7W4xBjFW6y$_@O}-x8duB_Ja)z;~!?@d|LpbGAW8hYc
zJVV8FQ3L;Sv%rcI+(zXjSQx>u`8^7;sdxzAiNsx}9F?CuO&~tnBK7rqJ(rWrTw&6J
zZ$xxm?Mono*+i~Th4t6j9}Er)^R|YZ*)}VK=PzRWWQziqRq8H_(Mi`1BiBX2laH0P
zMo;#&hcWClZHIOR!(qGoED9k}j~|<3vJmyknYOL1n$zYYi;skMkRyI85x1)EAKzXw
z`52zO-2JxWI4LpSD<i~pZ@+#u(QP)Eu=z`~hzDv*)~{8|bef4Z%T*=xcx%Jj@zYjf
zYL0Z*a!8ugdjVyQ^bR;hdFfRL*Re=z*rLNi`T42M;ayu|TC9SV_rBkX__w6y>8`>+
z$@Q`U&Qikzsg7<Igk7{cy1~@b?G4&_rBiijP|Z*(VPTNlCbh(-HrGUW_eoK>SIYb%
zkJ<jdEdKe`zLdnoA54s01v-8v=l#MvJKbEn!~%`073!vb^mA*8OvxTipB!iGWvMOZ
z`7dtuecv*FTE@iBsWIRyN7?G=a{QueqLA2at46Y7Ei*2B2Kjr;+5VeXBvc2%6qpw5
ziPeM2N)jt^i-Y0J#lk|5qVKvWN^o5yzziW@^n-`_l`hoC{z7k&huYnhJ&-eLxqnf&
zD?f4Hc_Dp5eW#bww(su!n{^A#4`RD%Zi}9HIAl^$I#vp6#wMhYqeG*-G~DHlT8wX)
zlM!WnX^-VXCKrUs2J-R_kEG(W|DK<zz;(T9wVBse6z!nt;=|a15fqd`hweq^EuM0e
z&H6;=<~DpoNXYs<eylHlNGy}Pcdh6n(ccoJxH=!0wDeL*=;eL|T%&L1l%1X-xy$>D
zMT8u-$ifIbkWCu*^=n-UPYw_+(9%4-IFx5!k%<VBC`60CI+YWIa^b4Kf1Q}(x#QnC
zwMq)jc<XO@Pg1J2J!4&Gp3h{~wCjs)*zCHo3B@I5y|RbWy*54ac$emyGn5Mt@b1#p
zF^6RQUoAml!M>Ae9`s0Og0-&ERI)C1o}y6cXYlo04s#{1`QqX+@NJ}1Anmt(&!`;F
zuN{4LkW}hZtIf)F(My48@}8AsaV7ry;`d0)jx8MS=y$``MlpgJZSG238zRy@C5}wh
z>nF>Yl95WyPk&EPMWIID!YDc+$jGlEKVf;DBv!B={Xt8NT=H5@L01V=vTS|MF1SDT
zGndociT&B$+aHl?fndUO*l9*Oh1HxLOtZY&410KaHCW2D%?9|i%2yI*tWPAPcaA@k
z$GK>Aj5?;p)mYq<*!X)W9BTZP?e?`$(4D-6PA2KKRCt+Xw>Vp~TJJ_wJ`&~7{WM;{
z&`nOSCfEIsWX@ors=63g?;C%^t<F_)F2T~U+de)<Pu9zj0SDLm7qZK}k`{FDdqw;%
zshK|T*85cZAh<2caO=^(Uh^hN9DjV?r@|4lB~)QCHh=HPw!u7-r2MGXYc$7IWLZT_
z;}aFRr$pfM>{hzkweZrDLiwX89@THU&gs$)D!x+$l7@MvX|F4-)x{JNMzuD5XY%$-
zoxZ_yV9&p=TLgm^dJ?_=HcCuvE{RfLYqc+H+M$xwv=UcjsmoIz4q?0zwX+*fbcI;0
zUUD&9Rl9VdXa!N&SUn5Y_ZEKH5Ua~jdxY5h)D<2{G}TcPh0*LZMn&dd>wE5i#ZJ*w
zNv`pj(0uIcXR!^ZnIFt4=Lo+@A4ZZ~5x&@zN|L-^)zY<!ABS>C35C+5-L|qZKOMhr
zP&0p^spcFve|#2Ce{;LmW<WGapcqq-eXqLA#muqkpYGoSq3@Et3aM09h9zE&(PT$u
zW>TZo5<Cq^VmLv)<H&TxeBm2{>vOYd?_+dwBn7f%NhN3`BRDP738@bdeS&4zl!kc=
z{r$CTS_j=7=2u1^I2Q-!Yzm$XRr*ajFZY^^og_4gkLU%S|9j*nN`V@>OD96#Eu**k
z=#8-Ku3jEe`b1l4$Gp9iGNq9#_+3}>EuOiL+7g)$X@hif2x~?vWYL_-ThoQ4;wiK9
z6qG5%69mPs@|kimEr{B47oUZ9|I~E6-QK<`L+h|a#K@&#*cjU!%2{0gP}&|jO6VXd
zlmfYd43%N7>&f^rat&hQne|4@J=4w+t-B>X4!Yl~93QWBT}vIx`H?E}!FBuY4AbY^
zS)mOddTtdKT#9Y!{dz~D=PqZm)W*c*r{xnCu}N;(mJ9J6NV;&U>-OGwV4g*~u=Pf5
z-i6em-+SXzKlMAL$L-2h_e5qt3*YZ>wHkNCK|hbOj=QiGb08Rh(s6XuS5i~A<P_mD
za!TPuVO32@PcN>purNDF(d;EIE=hG8E%^5hRUs)|eGLr_%8&QdK>{n-C^npi3m-q*
zE~xRV$!xPA;;OV1#g((}kxpSE<f@mnfX}$*=D*)EZMQZ-iB1;PBYAP@IM$nnh%Z1T
z`F56J!dW7n@3RP;N#-NTqGw=`I85TQ?tmKghmI2!6$P35J6VP*Q37|H7tT5QnPny<
zJOI()!o59J^LKM^&q;Q*R2?!l@h$TTNXV#!p4vl%Fqya0x|K)I^e+_-@Dir5GO_HP
z{^Jx7QYU3yHngF$y0H<0Jn(Mte@eTPC-EHj>{%&Q4fOfZJox1OEmPjnyhM0)eC)Np
zF*W7rN$e5Vak96!_oGAPK$C@rMr@mZld8+QtThiE?}+1qcZ{1!6WvCKAS)xDEbo{I
zk&V8`WEZ1ly~2iBYc0CO(8W{z{*1;Sz4*OE-x#D%r*ww3`~0vSxJZ=OZNPDHyEd3<
zzocryb$qIQk3Y#^%JFL-5mJqrF$wv4n?Shj{>6@smwmWT&;}~UMQ5sf#$1I@;@w}o
zjgQxlUAVye0yq6+x0WLPY@gE#UH$!H%N0bpJTt#k(XR1|(eK|2nNC*NHU9sf>8CK`
zU2nz7ex($#+u3cc1?h&x&0@A%(8-FYb2D;H|1I(Krn@dL&DRfKSJFs!c+{$q@|BZR
zAEA3Z_w73LxB@M+53Fl4DyIIep9HGUpbaBgE2loU{1xe&f0FyCj8)4{kzQxWg;)7n
zL^dq>t?`q)H23pUpZznn2~>&jmd>AkR4Vb8mcn9ORn$8I`YJZx<2R9%kL)*vQ#&5f
z2ZSom^>9ozF7zHBnKLl|Ygv`<2w%<}1;R4}OZy8etmczUd5f(ji(GS49h|$)a7;wz
zqwHMkQ580WM8+584)>X~$NmVU6C#J2mGW`uw)>ldCo;Y9)}#jie_M>3sp&KRl-XmO
zq3yspEH0i$Uw=g)A#aEBuJa*Rtbp16Z}*-o&q^(szekD#_VI>u`f#}Ge|`VjXZEv>
zH)ps?HK5PtR|noDWC6S<{>5_&A&r5UvPbR&wH3i`-%CFl+)$JFdoDW4n*_ScT;an=
zZlb4~nQ6S9n#8|;d>|zDbusDi?3ktFYySErr>ybck^|(NAM~FlU~He>QkM(DU*DjW
zd`Lk4gY{<Jq*%UEXCwzY`NQgqzmq#n8CO#iEktwY?h`Kq17iADVejzqRX)V#{Avs|
zDzeReravR-D8&*XjqbV0_+yeYUu(ku+B)I+kNPk2F^VXZ@H^+!vbAO$YMc;-(~pp2
zHM~_dk;7FVxi~>H$BQbbyE(%mBGT&W9>8kQq{Jg;ymwDUQSl=U!`-`2xXF>z8%7}F
zj!uRP{;f4_xDHJNTGazfP65}xV($zW;9t5F4t>$|(szAhW7?^bR`S-ZTf}KEF(B-F
z{qki{ND<aMU{C8t7}B~KBpsKFra|Q^I|!BVW{p*WR!mIHbhbTqu-rZkqo^P&)BW3r
z?6*^;QB*1oCWGgBzWAE`%f7Q?#!;zh-`Otuh#Y$FT>S9e{cI?FKcUqpvE<f{br&u+
z?fkZI4rTtqfVlP)j(4#8@!o8S%i1%^uHkZfOo}u~z%?AP{K1*FXvZ<n9Z9r!5Jg5>
zyw~8Q7;53wl{?}fQKswV<u#aZ;DfP6_2Agg{OojJsiv27oA4#!eBOK+-oHvq@#a(r
z4$cEp(p0l3H9wSh#Z6{rSr9cx)09|`Xknk4{?yZw;bEPl^;TANGDvD?X%-p7FBM3V
zlatf)^A*5eD5uG(HQ^}WVYPoGf+#-HU?sP&T)6_5jZ5GCpAKSy*y#}DG$XR7O^HHf
zqd$F0=NWZL`o;zl_sW${OXdt$yenqrI+L_YEhEurw8Q3{qOM&F?LuE(d~WTr+hYj{
zZA!ui%rE$3$w1f}H?SHgQZBE<T{5{d3Ay)Wl_e!5_h+JQirsgtr)Opu`T4bk+%|uG
zxb^7zwQGisVkqHPuSmQ**b!M}mWahFL)0!+H|N&gY`n#2l`CzoytK5khey@q)KrOS
z=k2|N4K@5#+*$Wpg>0*#Qm&XY4|FFS__ubaf|<CuO1^KInVCs}TwR2|SH{6gt^T7o
z7Ypv*Lg{}?yNVKCd5K5$OO|H@i)tO4zxiI^LXJK54h|MSN>FvGA|QvJkZ_ZTh-iLw
zqFxKMvoKe1deGk9sH0ht6Na7bB#B`2_1`V*7axHxd0xYFRQ6^Ief(W6E>##1tjriD
z@ev;O<9<eRVBZc+-a0;5S^UA2rvmEq*+CZ);tUD&P{JL?Qz@705K0j*osHR!2%74H
z<;r&fNPzN-7x@?K6+q7xx;jJTaDtGc`oaTr=!XwCAwp>+H!2&-Avm4Tdr!Mm2F>ZX
zpyD{+^%y)Fv<QTpdcN3B^%NLNNP@f{I*bix_na1pk9ynR1Hrzoff7!>$+Hh0%XIgA
zejGsKZ*v+)_&cX8oFp%)wAh=|O3_`3INL}n$;rv(OrILB{5w6oU~Q=7Xwgbm9PckO
z%biZhm6nwJSU;ig{CP68(9DqnRr3A*#jQUB%IIRy`(tmSp{JKd^Fzxl%2B&4HQMLu
zW^*(OMD;zl!n(S;unyVrPrqWHu5&D0L{oo1I|wg<*vQBz17<9O?n#!swlrF;4f29L
zrB*}N$;dKubt}#L^9`n{a!%KkdF)L`!0Ea+wXr)Fy@Mk;&CuwOkS&;63y~&qik6ny
zhlhutHMNOR*Ebs~wKDHc4K&F(i%OS#)50)`vzxwBd!m<Dy{%PV=Da+t*&(z;u{e(-
z{QTtj$ZdDR4}p<S0jV_8#`mN~L6z7^Y&-dVkN_`1&X$*gjKlan&HPlbh?>t?f_pHt
zOW_fufEmVsRJ{7+KFG<9xX#9su|Izct+*$JR~PZ(3@a+0=<3zaBFAgN3j;-Iusq<I
zp@pHLp~<8i#&b>7NY7$kZEo%1G^-jSfvX^mkB?6yNiQGQ^PZH0)9&|^5!)t8{0kS9
zA3l6pbFeJX7Q?3;K+LQJLiLzxcjtg>r}Ka8!=c-fFeB53aXnPDw4%JAmz4TuVWq*$
z<m2Om{!7GdLG6-(o~Uq^%Sg7p@0&N~<8?2w{X^xk`VgxF85tRg#F(C6z0x2t@MXk$
zYk5Q)M9=ePOJPNfyr?Jm(bk5+qxMKBx1U{zbsjBXWJCotG&M={yNijS#)_Ub;A5NC
z2_g7&n0ilzqXn#kRuE*ob?f2DZj(r{^$34k6qj@$8CM}ci?>@)NH)#`(kLFAWd?8O
zr9mTGGE@mdD=0W@@4CA>VH8BlY5IX?zV78|JO0=b^o0i_=(xDt_wO%)JCI8_vzu^u
z=z+Z-n$vUNp~PlP@82Q-&Gm}CnP^8C*}I@h4TDp{%Z$>0@bv;}`mV5lS4#*zv)rkK
z(wGCTA7C;g_~<w}d+f`1Ut6L)THhf=Of-`9ZwL%MW$W-Ld0#w0_hf5wJQf!Z=SxLS
zj?wLLJ3h+PIO%5Bt56C-o(J2KSM@Vac|3XgG$D%1f=-ej+7f9<aY@2|4doHgI^Bh)
zm8OyO!6Pi?B|tG<M|_&@zqf0;J~jgf{L)@Sjfq3EjGBM`?184xN-hl{Wo5hs-h(Pt
zamKbUH?;lz{n5$XR)fuB@!h1T8e^EyjEsy*MaFQbFqlmW6!BeQ5eY3Vt#ahDUXgJi
zO@>ixI1%wzlmaCz-reMEIN>!`K>c86MI9#3>^^iEcA$~{?sytWbv@UVC~hlrRT&iV
zjr&i%L8Y@8F>WJkix<&%b#+CzQ{@!7ZJD>l32CFHV!R2wS9T9KI%p-6!~+X?&!e)$
z9zV_iAxSt^o*L0{-&qc8oklxsZ+zz&^GM>epLsG_a_Yg-eHmgOoN*tyc;TG$*XUGu
zQ=!k?)*49m@VUo*E>R0-XMjYGM$m+g^N1q}bw7fl1}(#swW%g)G?-2z3JOcx+bA_Y
z&;4R5x4BzjCBg)Z>%4KI_#GJ~R!^nd!Z>(9aYbrxj&@ogIgj~RjVD%r987IbhB7^9
zdp=RuvBkA%L{WzW_(+kCSO8hb?!-ra*RNbj1c_49?{BY|cYgN8tf4fXu1$(IL<zZ<
z?(eMRHBq_$T;N6Bn{nU|18<1QQU~9n3J2fA{9YI?FS4#$6ECqE>Rot?@;HOb`VU-!
zvQclul@=U(BmsSiO7L#w2n*<2E)A7wQ#Q?l>R!Xf!4rA;GJdR%hXNDUzCQ4ES{}|I
z(f3dUg-!F}F9olBe273G@%ZRqal(&U4ZPkZ3f`ODoH=k5hfH^9_dI-MZ&C~l48$qf
z{Rm~id}=>PNltdzmX(tF;o1~`aIo4idj}iEdxX?t02n?P35|`-QA`py+gqPbv2OkO
zQw7^!*_uhtqplL@;-_Cv1yTx>mKye6XuvIKoi5~+98uEgC)UTJO$9RSquosb(_HPg
z)diL~JLoS0Pe=Xz!}QV4*iL9AEIqd^j5-$>D=w=+mE&XNm|Ye$Nge^@=-gf!DtX>=
z6|kFT3>mhcJP7~2H;`v^RR^4n8Xz+1Ve1=jU<O_%CdQ8CVz{<-#?o8t)hJsu7f|2i
zj~_qge=!b1btWBgDL+OsU+_}wgU5FqvMOVsqZ9x7^(zd1HP)8V($XTTUzWm|wFT*i
zrR3!!XqE;`qyY(GrX%!o!ANh-cT3pR9+leJ+jm6MGQl|p;mqAIUmW2Xm>3zQu%9(l
znnNcWl?-CHdD=TA6+1R=ZWVEE3k)#kYVh}k!ICUJw|UxFVbA?n>z`6H;KD#EAuUBk
zMXaqw-Vjpf?(VvDxYhFvzwiO|Rq<e|rGe$<k9P`(mKOR8l4dIam%6PJod|CAD9^Sj
zV4ASc+T%z876B!dKX81oGgenuXCCd`f~F-Cax2Cfl|^MGrBF2bGni$0uu@n|1oK7O
z3Rym1Hf&vH+!2=pFd`35@c=Af3Rk$zM4A*kEe&49ohSzvf=O3QVz(J8G6~-PGmr#!
zJkxn)bbo&_kBEv&FVasZE7SfYDm4R^iTT{GMA}Bnu^L@a2kvWSt<9E==FR}o$mtot
z>%XkZMpEIb@Go9#lMPx`E(I0NE4d)<`(xe#zB1F1^z`pvChs*sqlT@W{TJJvl`1lY
zu`{ZU9P1KMcR@ci1C&|O$y^q_bK_rqRO9TIhs(*hEj2RM*UE@l^|CGUs!LTq?r4J&
zt#yXoj*dxtjN;DnNZ7}ZiWyZ6+?M?bFHht3?%i?CazF6`-<|TV(;l!4$QS3UAPm|+
zIb{QHuzgl3ch6<kE;R)J0fr&(zO<3*78|y!M*(r7P@(JJF0i_=l$mvl%L}f(=C&F2
z7(G>RZ+?D03ZMl|2CSR?)p|n5-Bkm=Sh?~M>*?uf20lJb003wjtStqclKr9`GQ%w$
z9r9QpKi{1;4Jv}F3JM9(Wv=0&fQc(W@AOBkA+%wCr}iWVx{-mEwZs0ksgqMF_CYLi
z%f-6}Zre@&2qQWAVj(U2XAI&6t%-?=cfr9sfArJ3GvHo~+}vt^k+<i;Qkju;{D%(-
zL`-T=_}$Sp_K3W%ukZd^6E!AJUnEBdDaRr@rjnzS2t7tX(%teRuDJ*EY4OjajQ!)G
zwU~ZGY$sH`frwP$jRckJ_fVAudOEtsnEM8=_Bth~BcTgsV@)Q%`=`=y+`j!y5CC2I
z6i`uoE~`(K(F0%#l*;Jo=_8s8z@)R!GTghzS?Mr8N$R{h{`CTa{Hkr~F0;{BM!PQN
zl!fyd1Hg(q7$~*UPFbD>Y#hf2_B=6Qw1#dwZ^W91gM*_rW)@aQoGnO_K8-Ydx1;^|
zpkdlB63k+pt<lm{tbntd^;o&RNqKM%9Tk@68)xSIJA(6VJYy=LUwxL4N@sE(076?d
z4<bL+6s(T*lO!Z*Jv}|_voHz(HR5RI0T?nnk7&>T>AKt@e2|@7wAiTT@VVH1xjbz+
zq?U-9P3sT9F&5{zcm0;!LYu!Hp_4&HSlv7*N=^B|d9c{5VzW!SwTc-+7<lk78mp0t
zlAfjV^76POyQUz@EEe60&}hU27TC0LdS<5J-$@^>uzCvRZKvxx9i~<A%(t+pC=5nV
zUEfo+-j7fnRF+L-o;~~g%+Rp)-a`&{cEQDJ;nlj+_yh#15Dv`J-@mW$^5sjg9Psrn
zAYjAN9Ymz<>}5aQpQtJiuo|r@AGU>icSE$32k~XPDw3rJf%5kAf}y2~#oS7oSlzsm
z{QUfH2gRbYOJif<`?Xl;*l~2yJ5lc^6wwq+tv?}HP*8w>{(Qy@86(x{e@o>1*aCEP
z!t#X+7rMQ1PIEr3LmlmPAmL_FQnIp`Xq%c0tR+HoI?O46h^`F8m2pe$%pEauNLq~C
zEbf@eW8Vd%2wk;_QtoJ3tHEMLbMwq?fR!_!z~gn=i!J)Xu)s5s2-a}-^c)iq66U=a
z0<<ZKZk=Xy94c?H1<TA7({%r$<3f)lIgfSr3mqes|JS(JcffzvJ?6Oh2paCTGvd5e
zR(mAdHSqaQw>%=k%$A$_gQNa_Cg)5&p-5G_U0H6S-AQ;8XPZ_fE7J^C1#AI}hEWQ%
z*XZRob&kQ?N(~cJQZAJrjyR9CaJF4I_74nX-Vr9G64WrDgRo1;eftSotHfM&p>(pu
z(dX5x&98XWrZBO8w6})?_!(N-ey#!B0PfOhBnuuV=UsaG-Hz#vqc|Ef7#%J&cGj*E
z0c1fXNWFXaPB4NJZ2Axu0hb>_6sa=g>2<i)L^N9CdEnGBJz1g_fDI49)uWU5Hf9%>
zD_7K@KZuBl>7v`zCYw8=H24$vG1wcW%0|1fuxjKdd@BGXma~dHx0z+sv%uy-9!UPX
zA=X=UOatTz{lW~43!peOP4&s~fs~9)cKAUIz#i}usxwyLXQKpN3ukK)_re_ez6@qP
zzmj`?mngenRp>P&b7iG&+o-<w^~L&ObaJrp{u7atL;J`nF!jr|C&&1+{5xkX5s(ni
zsri7ty^!c%k2XJ2S*GW{#6?R-$9U(?Bj9%8N>pXLRlouQpMae~05pe+M!*J0w{1bV
z?84I2&|n4#M@mvs5sO~kB3e_r)8v^L7$o6GZNf9~H@W-2L*xY|yeo8AjsOfS43*_U
zI7lmr^y#@<WnA;bauji`5S^7>y><=gr&(7^dbyZiF?{wt(8&S(W}2@24s$pE;xxSp
z|2D?IPgixg-Wt~m;Z)x%+zW!W^f31D9}6&STIF^|z8hlV;%+SR@xr<gq-96e!BcvZ
zB#pzAO0gXNq6l4*j!aHY_7s~b<Kp4DhqLAnSmF~Bs^{rd7dy-=b#P;&s@yM=@09FI
zEC-aRJyu`N){eO?GGn6{C@+Ih{*6KN*qA=7rv*$_M4Zdyc}iiAFW)?rQ)Lu^;$E)W
zkOOn?Iy2Xm5&+==jU;ek-6AJ@9ZExkOHLghlMr}H1s$CjV88`~nE}#-(95d<*_J93
zsT4V)E($NP3pA`@rD`K6?V|9Qd_h!yt*YaQW1rctgoo(lZ{I`#PPcEk!zW`2pki|j
zV*^OT171MWMCWV{bpeiYga{g4?6GGD0n!g(nuT2!GSbX9ZU1JGMuFo*p;QL0$qOa|
zFBRr~0Lv-|6z>^t-~Oz%Yy%_f2Cm^pb@fA^vr>QpA*TINb%X_l*Zqvn<KuS$4%|8z
zJE{4h$*N{c>QNJKuex)b!-%D+2MU#14P&wln}W1zJj!RASK%LfQ>^=Imx4)8NK+u0
z^!H&{#v2ZEuvmK`E+&RzslVV6&`?@s*0EU34aR#O91&O_bFknQ4)dRHJ$hpbu*bY7
zz0-BJYJ1QkEHV-s;mqrw>__zG>Sn=P7{C-q=-BxKb&-s7`m8C0&<oH`UwhIwi6Tey
z@?;Q58Mj3tDXcZSqOb^5%dAH@o&IRRvwNCD&z2A2;o(`pD%Se)A`vc30ic$ip8mvh
zf3p*&K9^Y+jjx}d#mZ<kN%cy$QtUw){~26dEOZCM6E^C(qno2uA_aqwy%7YBX&@aA
zxTeEiw5q?qpM4g)%)r7TSwJnAHoPMmEHaTVxfhEih<ixhXnOR6O7#QEQr8XR4$0w5
zl>Es+QfkMXQ-!s};yCx~A-Wi#`iRFy8xVx>o+?6VPWX|~3%rOKV1c-k<?gg#M+5jB
ziWM8YqP)$^?53rTj#iL9D3iPASAVOU(<ll_wX?GW&MKijR=@&qC^{KSd$O~4O@iz_
z%G>!FP+%}QDqr7kszaf0Zc2(hRDmH9+WQ@XCDE8DW@l&9A~{TYm9T{`3C9!Xu-9wc
z2B!;n6zkak#UU=tLD+TopoARO2e<WbI=0)h)blud;l_?LKS_a@%!0rj%iO%VM41WX
za#j-d!Sveh%XSy@|I_S-i8A?*0F5NsI>CuJE%oZ<>yc5o3?=zkPApgS>8t+_w+U}&

delta 107495
zcmZ5|cRZHgAGcB@A=$f<5g}yntPrxZL&)BHT*}@>_TFUgO<A|S_qy%9_x2p>`+J_(
z^H=`hzOU<?&-r}b>r~Q&{IdzUpqmRKAW=AE`E(H#_1<wi(Bp}?Td&i0EsD?m-g@;{
zpYC7^VHYM+(|kh2Y<kPkpN6IKjylP+_w8H9ZJwxe*9R-vDBn9@ylA>q?I>+>I^sQ_
zj;ipAV_yF0$5D7@^PY}mmxw!V2{9x5LR9$lJU%x2IKKUf>me~wIMmrjKGD|E2GQ6|
zHP@2zc)$O&;K6on=y*Lc>B^AW>@}^kP&?(yr*M3(`rw(z<Leh_8sML|4|WrlN9r&v
zWJY|s%j3M%6Z*NN7b-UqO1{9$i^K;yHcc>}_n4+MR@_(PJwUvZO_@SVfA4pD)*X7Z
zYAw@w5ethu%3`xk8snlGicbXA(;*Aj5&cIF=MTg!Fr&oSGAA-AhRcG693-wkeEbgU
zKMftLJ{83l7D!=fPC0#=qfVI2Qu>^M{A=n?Xp<;X@e!<Y<KsCRL(L`K6cV@WR2tlF
z8cp?+fr)g(5IOcjs+MscTzq3L*;$)@OTfrjT!(KAQUdRjDVF+&_U3h<XIJ~2#{`Y>
zk`k2_%e2l_+`i%pSAsdF9a5ETfuTl>tc)p=5dsPybWbZQMp9fuSv@@C^?DkTgyP`$
zsqNnoy^HZ)np>Gdzbdu)npg3CCv=^?_Een-t8tXOn!vP<B5<v{R#d4`{U;qWPZUPR
zqs4YltjLd?y_PGl2#@X2A#yDLejTiQtQs5ok)Zqq|ES<6IIg~;3DETwo_gRLLUPU{
z!sa?UKFlHrGuNv&TPdFQO5$Z_PqI$FOk{MJ>+su~*WtyvUZA;opJzgJ1HT3k2wB(p
z4F@^iVmPbr-t%kl9m-FkrTKI+pFiZYiO|BM+TZHWS&tQ_A_--{@EB)G<JgByG=Ej1
zCJukeg6~?0)Iqr0tSNmsc$hA;(w(nz2=-dDkE<`3aPV-LptDA`O7puJjwJYpR~*rY
zP;TcE{m&fvt*e8FXSG%LbF*-a!aHYs0{JGFe`UiXuWJKEbecTX_uJl8D9)b54u1Jo
z?uu}6S@#4yE+o|T?0=P&A)95t)Ro`+CE$VDJ}2Wdr*^U&>ewqCR?G~xxs|nMLFpO=
z8&5_4OG=X~&QCcVP20Ph`{Gf7Q_w*4$9|`N&b1O^bT^TdKmApD#J@_lPN`z~91ysq
z`H-Zhyeo5^D(BWp@I~sC$`gn23l9dAe9+k-XTEV{-xwu9kpRfOq2~P&9if1Ro_3`U
zs^!4gahKqNE@l69(sCv|RT4+>c7<Kux!+U}SnEU!|88zJ7sjf&=SR*w9^YM1WvrX+
z0$;XDF7NDRuvc#izJK2pRXKz4D(7HgzNE;7!-(52`9po9&YQI>V^3`b8WtWX;Sz`0
zNU4ZPmZRb_g63DzMZesW(Sr`>K|(qgN)vt~ZRI5Uk(SyoEEha}{{AP26L!Bc+oO#x
zI>`FKKKE-g6dpDny_kMbqyOG;(4ov_@JH&-Q^Hf5xk>}nZNaqL>Yz5*d3gK879GF)
z7+xDJVPv{tWv3UMUWP+zyPD!sDxGk5Y|p_J)O%(z9cKh+6}IT=oXZFZPZ7j~1mrjL
zH;AWBX3j4MjRt>>8(v>Oe7N_3$LUd+ieOA1*=m@ek)CA7Y%#*eVrwbg@Qet=30F!6
zDt{54_{Q<HBn^&eG<!s5BfPsuW-mCNRfW*zjDfOF?}5Kt=GB2(#e+Ugh>~#ZT(Up!
zG0CWSdYo_3EPN`YIT?6eL^aBtL9uEPVKC+JblbHSF}KiwW%}z)nQsZ9D`VZf+~c`x
zg4bsbYpjNC(x@8CdzJIV>ZjZYB9yrJF%uyXQ9gN@2-IWaNh9BV-#goo9!V`u(9gJz
zRyz>*B{$|;Kx8?@>#K1Y0x%dyq2a^bJ~nC^n%}Lh6;HBrBKfcB5KgQeWVeY$T9yV)
zj$cO{F)S_oFnL6L`BpdVnEy3P@u%^c(u>MnWJLyt1aS5r@Kxq5M5mEw$}i|rd`(99
zq-6qYl&){wfoTfn4$JGW-aK*z>VNyO%K!yN+bx9ZigJ*ZB)B0Gap09CESykV%BB<+
zS^REJ;Wep3PglJb&6KM*u!&G@ul!w$P;q+1NF7{QNfI<|B36UhI)wW5q-;%*U{N75
znOd#h`XM-gK{gWvKRc$%8E&5VEKM#<e#%`qO_BM_#tTQZsMfw(Ba}UOd+GO3D3t=1
zMEww5YHQGB(C>)HYasWUoJFq#^QK-Vqu$ti-mJC<wYzv)vaoku$x1JtTXOPqODn5_
zl9IFaouX(O3PxN6)M`$oWv23d52CL^Qx)<iXw#D$j~?z>6<U~l6W1N3w_37|M@Xe?
zRunP7cqL~gaTa7BT}K{|#1HkYx`iMW+<Xb?tNt0B+bvCweG`qmIB(Uav_YGK>B8kL
z)khF3q3I`kdThm&LH#UH{bt|)940!yz5Rnsn>^2Z7S3>_?)#hMj9ZjrQ3f9;IK!PU
z>pPxS9XR3J<%<ld>s3-Ti|(D<Z9gI2p&~G7!^Xo&lKl#`yz@_yAORzwE2XDCxG<|y
zU>_TMp-``lIA|NrFS{Y>=lAdf_Wi&gxU@u0DmKKwQy4n`Gz(tW>XTX92ALv!&JZDp
zUAIN=ajZjH?vJQULsQX_D;h|aIPzvm>~lJI-7w=JzgbKl<hYAy#oLPI^?!t_wYBi<
zy;CxTNCaK|G<rtW_D(`1xAZB+%W#Q5q9>aZ$~8>G5yDpSzjie#7+O1K=SM_$E()rM
zM~|4Z#4NzWC$kj4HTf(qqjd)i<FOJ;-g&37z5S7agZTA`W@dc&fL9>x5~kBWji={n
zX5jnYXk|Gy`FARo+{ktp4XHA`opIbAHL*~Rc+CK^=)P?&f)5(crw{SpCapsnq?2T`
z3`TPVDdf}1!o-DZTXLy8yPlpd8Ms?cxl$eymgfcZ-%gufaXEkYtRfgJ*YA<ygr|HA
z$P&nA&gXUFakz~IwnqYK!-=%?FId*N*?NqZ`oA`Bz^lFuz-$@k{kz^6Leqs+9Sjuo
zXPg6+WmL4}IAP+F?owEpU<j!Z+K3uyqxiiANCC`(3QK!}!ld)Di;m@3j=Gp0MPi0y
z!|89|xb=9a!q$^0x_Cxfcnk(FmARwQeh>De$kX`k75moPdK6$~SGZo_Ns&#o6$`HI
ztoqH{51Pn^1t+x-urEQv;SBbn6&=WJQoc<M8uhjIX7_Rh`fz6FWH5Ae2E_A<v^fv(
z|Ag(|QR6Cmg1lYkxJXUkNq(`wi7;+MrWmz|a<jMcM9P2Rghc`|y#8y?G@*p6270bn
zkKTc3gaPl2q_G-NimVaB2cr^p`|_m2Obq;UF??S2_!kfzN06Q7isZ*kEm}NURY$|S
za-;?ViXt>+TV434SIUmu=cw_n?h?_}2sAk9ic*8yC+zO&TPfE+HzPleVp(hw%aNLV
zm9pD*QMchFG2<TgTTjYhWGznGriEVHuiYWEj&{om+t<__VPh)%oifH>#aH`HhD_~u
zck%r(WrG@sx6UR|^S=66K%gYpam1>*x7`<vs`<pir(^fbP7&M~uVZAQ7FW}HcCNLI
zwMABvw%guUB`F1cckyo(^G{Mg)?%-8RIv2CJ-9=Hgf+K69RKMm@OS1pM?rApz2?dl
z?H17aY&n9jW2IWfR~oBGKArNEFCrL@F9zv3?p;!st4%v5lizpXU_ZCk9SzVKb$C;+
zACLim?A7zIp|d5hB6Gbnf%hn$QL`tW-?~ZneQqoPFYV)eMe?mm<m}uS20?w|=28oW
zhx_}Htc5O>+u3FImyK}#w7IoD95sxSQtdx}4YLVXqEaevaKl*-pp{CRtitDDtU>sU
zS7_G4_lZxowEP}g(qJc;U$SSg^}Nj;em%PwoSi9tdv?#I|My~k`RU;~lO(&Ohg{r>
zSNvgq!;SD1jdP0^R6>PU8PR=_<8^wND220;6OO&di-BLdaym$i7UdJ_@{L8(Sm&Dl
zbP0KTPg@wP9oV2>S3-oK;NcN$+MBXhg3Be8;RjJAq^M2|Ay_G6C40q|uwkS=%p?lr
zkUOH`d|gB2d||YC6;JW5b)_G>8uBFG4U3M=IiSfYxoF`zr<YW*e#S9}?Nmqj{szO$
zX}>)xI$lp&@q1{07I@$O{6xXx77#o>4*u8Je?g;Rs8sA$#(3dDHU^AB!TD9$^49hX
z5&;2SqlLDN1X+tlh6-szd^O1Y6MbZ1#-Sesqz*ZID_~aCehtYS(o+_ba~SagGT4N+
zP7n1H@^Io|P5obOb(ZfGn2w3!1VP_(m6?#;C~GN`s5ygj?7c4GL6mfqshzui$@6+U
zWRP*xMELrhdTrBXfw=I-*6#PZuyM03{Z+eE7FMmZWumN+OdJcPi_`hbaJ3`NZ+V%C
zox2=aeZ<!`Ppaz)g<Pd*QsN_9zKd$q_9&p<o<%y2&yqK7HrbB^UQX(k?vgQSrRv4S
zSZ<Ty3fooVxFDwrHqHR`tf1|Q17VR^c;*Dicjl;n={5K^?GTL-Ewb`wyTWy~+X+>Z
zgX{y7?aZ6n<*bn_OsTvCDO>Ruvr6BxCNjgOD@n0^E!Zj@86VTa9IZ*O(QGCHo1xWv
z&!x<s&mB;1a9^75uKGpga#b}r7lkzX)>U}dblyu`cw8OdlmDEL`A2K9DhE^tA?hT=
zD|gu`Gz?FB21k(0RtIbThL+8URZs5<7-h^&m+%dD^MW#J>++%ExOjo0btjQJGD>xK
zqr`LQ%!>S`56dY0qcbwi-S@W;6Qw$|kQTEBEZLRNuUp4{*A9(mj@KHXd?-rT+H9OX
z&|3V6p?a*B+N`z!?VTbJp4*(k4t<Q4sb96i#>bF}Mp_(uQe?BT2DGV@kn8o2TPV1%
zkIEgF9Y0LE_BfOgBhlcPN}j!53N=z2FZZDp4gW!C6F-=_f9)faHCXXxBM6>oFlZq2
zMBjANFU_4zD=DMxy>v@^OA@DP2DThCsz$;&tstzn;cSr%!Xhq+@fjeJ2rRa;PJt!D
z2rd__b2ejtCCT0%ykAs**?|vDIKCZLsfd569HRT)Xw~W6@5E%v1iGtqTT2QE##1qb
zu^q!*5haoS9-1ihLo;ZoJaPv6qmz!d7JXVs%Z`s<G*8|8J-oTCum9ON2tKoVf)+33
zIH?QuYrk!k3>G%|{H-#V3zYUJ?bCduJaNy!7zX7w%oi}xp-7}{lfVm3Myo%#dzM8N
zmb3TTeESyk^gkJGe={N>TVpZy6{oB2zZnF{wxLM*H&{jpM;6;Nr&Ip@D`NgMChCYn
zPZETv)t6y1(bG)K;!A9-<t*$Yn@-|VkPp2L9R*{0{}Ll3LlhW*Lo;d^-*P%b&fiov
z78hHXsBJ0qK|Ix+(dQfQyXXx1lFV!(ZdT@j4BdZw?kP`$14#4dREm@pKeX}wrwqjl
z<}5LWSHDg-Pxp&L9(CIQ?mtx{p>9=m2;XCp=t-gG3?Y{a#mD7`q*NYNCEx8do+_|1
z^D>&(_)?Ot+nv%gaxkCL$PVoxpcG<$PS2Q|+3?C1MO3j_3r;I_@!X%n0AI*YQ7fYC
zd)=z6;U@(f^8c!ahVq<(5lP(w696N7He-R}VV>9hghXmFH%sF2{om4rSy|+ZI*Q*l
zD2yoBhrZ<5fzq);V6Ps@O>IAzSFV3aJSt?kz|r{RWxM9_V1a|TYy=Uj5yoVdned%3
zzqtl(dXf__|7yJNp}2B-D&$zG&YXM#9kbiort?Vr?YR!zHAG>>Z#=P(Sqp(YFj?_L
zZ3({=P?bD)2j_mZwV?v{3AM@fI}p6Q4qF%!1*}QlkR>QjW6`gzqf766&C^)*kj3ET
zu5J(`b9pj6Qy>3)@ckIkE$M6E|A$dCwR1T!766x@<SrgdL^5lN>Ov^l;?BI$KoS{4
zKh|%MC)01FR=mz1G#uL#53J+Y)5dS1)U!V?p|+XIl@fg4`Knba@HqrPmAoO^c|C!o
zUO(E0EOXOdf@3|z0_z=-Tt-SNIxi0+Q6Ka%!JC$3yP;pjy^DA_9!S16X^Y?$;B**o
zqnk>x+uSL7&6(Lbo`z|og5KDYhS{RW3W#fn^ZuJ3`sEgT=52iBR5fi)O3O)og8{)K
zbX-TGX|k9OwzQm(I;LT%s||(M{qHfD!DiW}k*XizRu^D5ojFY&wV}vBB;!ldC1kiq
z8ksC23#lSlcFhT+66d35=Smw>H34ng`H6A~eU3v^9u<7uqZc2iDCDbUxK*bp9z1Iz
z848e2yu|~hAIq<T3v5@BJvLO|4Y<nnMyn{Wi;7&}Kn>D)jhzkA!|K1v|E$k!e2q&X
z$`UK|Aiap!yVM+@?HMB>L+mZFIpzv77^`FWdZo%<c*X*XS5E%ByXd|^31ojo61_Ok
z+95tSc@+F$torda8I>iuIhupA*vAPDfaA9FSSVVi7N(Y+Fv+YCRoOJhgU^Y3mfSlg
z`6??Q>Vyt&*O^YT?Oj;D=3^Yv_JAI?B*uQl6%P6PL{9oDCEFdKfvRy;eG{2FG<qL9
zp|MA3L$uS%14BA9o84cXr+(E5%N}|+9(Y;O>Ee2cBu&M}6Eu@c_SkWUi`T{Ik#udL
z8)k5Qo>JiLN?hAmrsn)@5!rI}FCiWz9w`mI=b+OovTG=*wf2QEB(C^@E~vVwBBneP
zA6NF+ZH8_;^oMzF!|speGAnX(7MjxNmsX0?7Ly&t5lM<hKJ|`2lfMzG1jjbIz2w@e
zdNMxP{{ubPS**To0xUG9gDDY$bbZ}S9?RmQ-hLYqxo>dgSEY5CM=>})PzL-v7imxi
zBU6$AyzLQ0_1GvvB$l%%E^VJGVt#8o7^ZQ7;3h|;9}@ROdG-WQTp#Xo5>X|Fv*0Qx
zclL{RW!z5ZfQANVu&A+h3}#j9&rJ1L^Q_}{=v*|$eSQ=Ybl}wp%p~#AsHObBU7*Q@
zad*h!;{?QEF$SS5+ji5A$(0dXH+$ZUJ&Op8xy^Z40qj7PPIav6fJ6cjUWUDpUTimO
zUL=>&P7;HU`#Xsu`0i`#$?Q?iib;!{u)xOL^@pzR>3xPL8(gev*Iff<uWL0vqS4@h
z)e4V0gWFt0Q*2w!i@(kbbnHF%Wcecbs)HpMDwPmKs=J~}4-dMioOS9Zo>0iE_%&Q1
ztdA7n8eh{XbVN@mU@9_pJ{wN4wI_gneei-N%nvUeavwhFo-kjrZgysIdVcD_Q9^on
zrwh48d26J#x=YX+I@3qM?RN0X2^h2-p}yrx;{SxBIbL(?tj2>L*2JM`bI7pq(M1Fg
z(3~37b(rSzi_SZ~s_|428*x$Ty*fdZWA;Cn->r42*?0gNZYs>P$i|cl=p^+<^D3ZN
z(p>yUcMYT;?icNtjD_i^yV{I!+{L&Q5Z8WW8}C{|aw+C%bZ|iQc7QiOdBr;C(qfAQ
zMgefDOV{%^Dsqn8$3&1J`TWB!oz;&K-B|UAGk`gMXdnBXZ`dQLDpbtM^Un8?-n#du
zFUpwH$S#m&V&N<ae0T>LnNK577@+snTa8<U^Y!a?&-3D7S+1fgXmJX&gzIHakw6)K
z&P5(3Nz0dQUUReYO;FrF%()eso?E>?+b;-q&JyC8Q5vqFa+|NO6B;~?`%F2N_>?Lz
z;3V54L8T`tZGP-c^~=H}<Q>D_*dABjum0y#$HBF5Kvqt-g4Zh?SYEJ^$Je5I7rJP$
z(Qj%335;W}D5O;^n1XTR-k*u3qSdMNdpXeoc@=I!wIaK-ve!=Ch5A-)e|U36I{`CY
zffV#Rw2FK2qs+}G#Qh$kwi8=@SvJGQT^L9mdGTt4`!G7gG=_0>#GpBl_74zO#Pf;6
zy51ReR-@YWoBU4soFTNg{OxKZ7Vh8c4i5z|=ptt!q;WNmY`!!6c#J(8g=(bU^I_zh
zePogA?M?9IcKb{S;{nmCTVd7L!;TBkd8WMKk(qK(@&FLhZq<Q$@qf}*MKB@DcB@i$
zchM3IW3+Y_`7MU4C>R9o#726i&6!l`jM^7OEQX6)*ZQVfU2}&?7gZ-FWqD`O9_z4o
z6|NgW^K>~IKu~2^<=JR+TB2|5SyY?h1sub^3_sZ}5U|OLQ|Rc~+JzK-{BDM>w`*EU
zEVo*bv{@~B@EQ*W4bI_sFgXQNWMZT0uEDs}wOUwf(_kzc?^zr=kLb=sd9hEmvrk2b
z_ZEAdO@zRE=S8eF7DJn!No9g8wTHa2djT;Aebpj>%*47>J%$b+C<%W1GFX!`U3q&P
zu8eN*3j}JzB4KQlYyU#k-{EA{j@`kYNFVVy1e;6@uwYhOsR|&wOOE_0S!Kevl89NO
zKaL+XWRZS<6(Fj4;l{`2+>^9f|NSwMfQBinVc0{UE!7ad(f=KZx6*&CHMq9t;`@r8
zGoaxc_hj?|AfehH0_l0Dd^qbvGYh^m5!FCHwD4qz2<AVI6r|@!G@2p^d3xgi%qYr|
zr>Vkl4xo2|S<W6j`*u<O1PAx6q3kP>7Rphs%7M{*W)jDs-`RiSFbdAgBYiqRBVcD+
zNj`qKKB4AZg$dpkd~n)dQ9$O-f1rU903$#?InMYF)NCSTUj_t1k_1<yMvwAS-Dmnk
z<`7<&+c2|70M3B$Qmim74+nQ4H0+0;XDuF0m=;%L6kInxnqMb14eXUJYvtjOXf=oY
zzD@lI$8{UL+v&1M{tqr+KM<B_?dpEYLve>$!pPV-?n9Uri;2za{%~SzMK$@Ox=NfR
z+ID1q$VknRqM)VqL6}Nl-sp~lAL=%@w2AdOFYP=FO<jkPJgJ|QsH4jBN@apk_KHt>
zrk>(GJYR1`LDCohy;d`}@H6&LY*$>+{6S$Hz@e8m;EcQn{bO}f|4pqok8h!`GK->0
zbgf*QUw2Nbxx3X=x7PNG##%I_lKThnp1}onz`kEIv5r9T3#$9<VeuKPRhQhf`z%FP
zudWSeE^3IuA(8t1N75sf<04q_08oT46E+WkZ<Q!~Ue~{IqXP=%Qv7>+6&j{j9%M_B
zb8`j|0)SVnd4Y_jmg%n<+&i0IcgK%5j$t`{EZ-o%I)=ubtI%H0d@mGVn$y5gb;1MC
z;zvZ7M^+lB(_}$g#NG<wWu|;K)4nLTOCsOel3-3c?{&VOX4U<ZFdk`>dF38FW>llO
ze3GQse5kp;X|BH5o-6*)-QHB;2un}m@m*S4JV`15ixG08SdQ}QzDVY_cD;MbyvvDP
zcG>wAGU1ya$z!`Qut;rUR+JKMp#d7@e=kDa3eloYJH*yZz5*K3faMC;paBQ5j@3kc
z%ZeYWZ@r_(_gYOW<`ay%$WweJ2Bh7EpFJOl>^~{$e#jas*=CI8iGdc@ekAM=9fWH*
z+`JsI1)=_Nl^r*Kj$qG^d(Z~Fbx}~_z;p+9Ngorp;-T9S&w<$!zIOp7EhU?r9^}<Z
zR+ppV(Fd+#`P_hq0}scmJeaI|%VwLitfIna5q0|0d9*NseK~>TKwYSX=_pR=*>?T(
zf_{cMvpwpvt#jvTD01Y{XVG-mqG(oiq;-W`93UjAV8y16^%lBmd5)FlYflrTh$J;k
zap}Cfv<H##zVtfG@d44xFc7z#vUc^nT@R_I3=8N-Wfx^`ns_}q>Fp9kM<+&^gNEJP
zv7)9T&Ytr)!lE<dSqK5quMdMX2eyYHV)X8Kq4)_A;}zFhL!NpaQU?MQYoZ<MAbOjF
z_T8t19J6Q;!ng9Sw5L?14ZA@4`Zg>{+s0zC#skYgF{GIodDYO=5dYO|BdJG%$lPRd
zpR7wA0SncM(DsrH!^Ir4{?9;LSNF_3o5&N#qc5ID*f(b|v+X}#=AWAJVrNWM=OsOT
z^8Sh07iL4#tQIApPhx1VOo#%JPH&(c>UqFku*xg%WTtT#(`UB83>L2FdV&!fZ%Kjr
zYOp1$NuAfn+fD~`j7buhL7-&RPCE>NSZxm<uv469s=W=x?yVF2+`^Z^($T6(qsVd6
z14q_mC-{}C7Q`o4m;g~NVklzYRjQYgwSNqh+w|p`T{{xj&&Pyixo3n}kQJ9#H@lb9
zfa`JW+uXi>(@t(q?mdcTVb9JO&gjmAg}uM%MW#{gukDqLt6aVz$TSdYFnXBv5K^BW
zu6;R`aC+Y@p+5d77%}_T0JE9vjbbcTVf#}(;h2JrXDc39WN+`fI(hDM@ili&`jzv+
zJ$?+I<T3IfL(+-YpbjH`{62C}7kkL3oOhN-dg=>QqC3J`d*>Y%+4FLF72^xdQE+h7
zA|k)@Y+PABK2<(8O~l1pwEEo4tj4*6rpSu^KB0a#Pt_ev?%n-{5EITD-4PsnT6vvf
zS2-Z<aIa>e7=OGH^VGg=UWQD6Js9Zi;1gwnmmA|dJisz4AxY6h?8qK^Fd1`1PQS(p
zJIc#mczCFzZP=ZYU)J}#oDO9qxNuipvwKRb!wX<;(|dV(0~<>a3%b5bEeY@dj$;9{
z+6A+0mE-eWpi>#2S75;u>-Zff7iD#D)ENP^EnN3^C05>jM9mqWa0N*Qa;9#%GL;wT
z9W}(+vwV>P#md@g*y%Z?`8^^`-q2Xf6?Y^k_w&Fj%{)@|pj>y_ZTiWt{!X>k-Lv&Z
zIARuTy9Y$y029JtkWEXTij-z#V%33-H!pnFm*@8W+f810=iJKFgHJMow^A!Izk89u
zbo4fp?MUI!Gy@h^Q3RIvRU(Y1jtQy<(`~;`76Pic`&3MXC4i?=Cb7P?`Vy~5YWluN
z0?qf?%Zmlg3HCQe1EETAP1l~5*=zk&7QOb^u#=Or`QvSUD}d-DE+alI4k8eZh~USJ
zg~UYh8!n91M@QtvquH;<QJj=+%TXL(C7N48`+C!RIv4B^6KIF;d&je4)s6wz%q=X)
zcULiy-y;-`6;hHf;;d1~xKqRAL;z0oaVMF*IWt0?iv~a!R$Lf0DuK^mndAJqQ@z@6
zm-7!@7(V>xu?;@K>p4N3>2od+Z~(h1nl}8{a=f0Y{ns1H^raehXy3_;ht<<08}G?N
zz30!?PqNS5uL*SwO~t&obXNG*lvQj<?O}pLTOw()j2ne75AeC;H;r&H3m|f~d`-~C
z&op%g4bi$Lj*@|RwaR<}eF-QDvkb=IWVWK!a53X(B!l8Mv+*qVc3A&*-?eQ<+~JZ{
z9>1?CnwcLUxr;dhXi%i3f@Rg@Lm-g)QBc*Io8`R!HW42cfq%ktsgUK{HweSsXph<f
ziuoE9NNe!257<D%6-#OQJenG@yCgS5L~Qp5u-Rlkeh;n<<&w(NIG!{rugNhXaoF(W
z;67>~W|GD74Gw(D!-Aq;&Iy!zNK4OD;pP6Ihz(KM%%UA6)VKQ1FZcxoqu1PAa=k{1
zUtI>U@!GoIH@0W-66R(Jlqyf1oxT(E0~X|MaY_EO?|oFB^RW?88}t&21J62ynE}~K
zg2;Ve;eMF7XJ|GacoYTRh|)N$B}H3oZ0t{Zdg&wWGuO`{fIC8C*rZISkrvysza86Z
zNQM9T3lI(O{%;G*&?(NhmeOHb^W#m4YM`}!36x|kE&whRjk>ixbqsVG+p!!J8ubGJ
zeSB0}Rqi!bt(}Spg8}7W!3v>3U0su4mE31+Jl!*D9Oagy92JDI5;ub<nG*U>m7<&7
zn!<N$BKlgvx|L=Ze^Ai@2&Y`RmOo*+>>&_I>*9HSjJ6u3<pIG3lgJ@3)Vo{@pl&QV
zUY(g0ykAM;GGjnSGSxuyMaQ6EUPV$2TI6HjhCeZjMfg2cEu$Mobt~HC{x=*XO4eC_
zmP$Z;Pl1<ofh2{<6H`2!F&we5&YQ#3!93dTX;yd|(IbO>U&o4Tt&GTDwD`_PPWzN$
z{LoNb!HZII59C}5@>xPSbnzUyEN`YU&HFXeuUHoc;X{KF6Y9R%B#P~A0khQzA(;n!
z4(pdwxq7wr9hNOiur+#tpMrT7=prg~<AhL|W+m_@uhDI6@Sfh{avJYoGcNA+(-3Ue
zkk);!gLawkG2P8ms2{TA5AfgAh0qSdmA0im544|B`@P-2W45<b^SX++(4)!(Hutfd
z=g3B7Zj~*v48x^KkW3?aJ(Mq*&gqNVe^~;S(mAl08iYfS)n#OC;NPIvAM&4J-Kp4F
zUdCS1T?-?h>!g%T>7Jb`xDSFn@_G)puy#n~zf<5Lo~w5uL3YRdA_5Nx;`tx%3eS$V
zFSqqbC5`k_-Pq(f^2!zyiVO?A-I_94fh#-)Jex3=pu);c5+KuvfaJxa#%YePo-QO$
z=1upmAX>yNo#td3;C_dgB9c-tF#iBl^q!w=(n)LyKA2he-Fz7!wfQnK%$q0ajrd1m
z4SxVW*>}qTMGBYKpDm=q!fB;4XrF%?`fPeHx4Cb~pNa}nYz_;T7SKTFgII9fy65ra
z$~Pk|VdGV&{o6yi_yjQ)SJ1*8DKfZ%g0A2B!YUi?r|pIALpMznH8vb!CAGYY>+Ahq
zrxmP{RnAmD+FP%?0{X9l{4?cgG@@z)DCw%c6I(>*T1{Axf7w)5;z}2Z(?9-pM{T_R
z<}B#ZpCA%BEuVipjB`2O^~c7+nHx^U2z27soV7pagy;-7tQE!bFlb2`3~{+c+4P)u
zk@awg)h27*rZmx>(=E-Bz-aw~3VtgXUl#DZ{Go3!oQ@nieJY^1eUm9kR{SA2d|Nk$
z1XyxbOU~sHW_*tEf?TePGZ?|<OTouGru&c4r%f1(ZPTB!xdUbol2cNhX)cM>_QU(S
z&=MXrvoustWj1>w?jd|(nd7O&2s9cud$?@tDlY5%!k}#z2nF(bzH_|$2IE1ed<nOJ
z3k<EU?b6@127KI?UY`~>+zX>vDhYO9cqY66<_l0#5PI#J<Z5<~>_D&hb&CGCw*R+%
z&vdYu(>s;LAW(g$y1S1W-b!1U!l-nEO3SRYJLQ=1xow}Q=nR~zfoat&8WI{7loiHI
z&#Qjeebaq2;Ge)IYn`v<6I}1X!ak;Dq6754v-s<iZO-Mb@9tv3)`w>o!7-d+;6z!s
zheF4B01T?z6wV~sf4nJ8o(96n<wQ#a!%aw3ZE3KNq)Awv3UD?nR3l<F?p>pi7)rY8
z<i5e0U{9s)`v@~LLxf=}Iv?nP5k2`prLk+>?uxVCz5qWalF@KZ-iwdb^ZN|B8oSN+
zpB`T<i<k>a?f&EjVVG2-XUWj^-vXasZ@bQ4F#KJOIBz>h2gv10b_6(2h-}ah<cu1R
z5m}4r$HJena7@prCF}L!8-S-Es<*ApuExIc9efP`%WKJykBxHN`4AalV4l}d^9Z4n
zE|-QloBE-}j}f0>6e@B$0xBhGH6Ef*%PV^3>O&M<%z=?HO7A3t1dToi#D#L%teqw^
zLj8(%2DfKC`P;w3erA|j6{vFeJ0=9Y5CRm6?5}ZZwH+j(*ko->Vf!fT1-=U&nM~{J
zj4KvMd>2gVdbmg<ABU`~w9&U!ny2}~F)f!~bE@z1<;AFu_m-_c7Ll?KMXEE&&bqa7
zAIQl3TmHu^vX_*Q33x*>|H^lYmiNNZ7tOk_+>;*zjSa!Y>w0^-(S~#Qiai32mN%*3
zDH#H+YO}!Bz@sn<%^}M7g=p+^Ab&RG$Le;NX)+ls9b2z($kv~2x$k5zpBNqqJ<p3r
z1VIGM45T>5atlEvHGVx=c7Pu(PI?rGudHx*kb;++Ap*FTp2jXt&+Qt_U4{HLNjx%t
z0F~pAuyju4-Q#EHxcyp_c#ey{RRy5|9!6ldXD5D5sKam<miJBrL?7(I{X~rXfT{In
zlkL7$3}gaJq)rt{pt4_AYmm-e4rLQ%1YS15{@|lp!yIveE7jAvI9O{>S4)-#gsc)A
z?O$yMmqi}IS<<!KJQ43Ujq|_mhF4f{MB3qP{can|j;8siE+pk`-MuXxkBU$+W(Xn7
zPvRUm6xXV-VfQ94kQ;;a`@01L+$JZP`c?G)8}Lnqwnth`8v`Xo%=y+<c>HM%Aa_tp
zm78*$dI<X8{yN0QJXUOA;Axa24%q8Wws2utZ@_VZ*sJRih%30s0>%aMOm(%IOG%Z_
zK=g@%Y2!JW6oI&D(5%zbTf^7q*FVwYfsD?OS${O{<~azQeA-UNT+nNZ;8FxS>1;y$
z3p&bspc_)wovo_Fwx>&On5~Z1GSv2c)w|GUpKgTPW+EsFzL1+aNa%WUp!>Y>ZM(XT
zLVHKo(tv>u5cxAM8}|)ZoLMvFj{}$s3-o^*aakDk$Ha^<6m0CjJWf)%IUNs@82*74
zAHvOA&*}mzYYJw@fLyXNs+Zg=@r<BD-SC&$4A#Ux{N1*q6ex4Z?rJ3_BD8sJ8;+Pb
z;%~%QY?1pR5csRTdJD(^;O$M6Z_87c{W8|9{l4Hq4bX1lsT&(DcJ}a>)x$ZJ_pdg_
z>TT`=Fq)Z6^Ea%O^uL4jrt}A<W-f0qNH`VAJs&%h5xsz7`HffZ09puP$YaZ<Kn&)N
z1FwdAcAl|X^Qq`yEjd?5a301?+XOXMD}{zXk)A5ku`umb&gYCdgh;GZ?WBk&-#c4J
zYH-l_Cl3^w42=N&^2Ym6qqV#40=@&Tpk5;NLok4IzWse1A>Hr)CWKI+zR7wkhH9ho
z>wU(cP0gM|H#k&5x<f>mOR!Y7^`(SzA+`C#_fT8fx1QzFsMG2dh{*cEPD>bj@Vope
zTRki4DJ`P`XC!3H4G9TY3-w`I!3ywUBv3aY;ZK4k8n;F(So^qwr6Z_NRIYJB@yY0|
ze3;V3DY*fv9+{+jD7U$Qx7?4x*^*n%)mkl!fvh*#&}oT~t%V&vQ)*DrQF?rfee?c(
z=a=C!DzYncw|vC}IwdQ-nYHGp@-!@YfldVSDS#}O@7*W-3HwmKT(2Hn2hu3!JBE@f
zmc#&k1B>_X*L^Pf_BSWW{ed`^Cz8{wOB0fk2%=(OK-DjIw42gDmI8rJ=$*7LtaBVE
zaCk6r{MAM+!aK150Iicud!TAk!H=U(K(C1&O#K);S%cPn3n8D;{5Qm!5W9x|+&Ayz
z(&$?S5w)<Eo!}OrmrxrJDpAGd;Un14RZ-m*r3~xeqHK6N!jLeR(F30uPkRcoGBVgC
z5KY*_QWf{wVq`&5S5Fz~_bm4D$qjeAiLXO}?2A%w8UC{ge~=LENNNkPyA+om%#+gf
zK$Tp^FQmdU6_&hizXhSkPn<o#Z#_iaE3h$d&K6)Ew;>X#HWg%KDSvYkZx{o_S1FRI
zwzfu9-wNCK!BqA)A|j()RRft0qT|pp@r_MvILVtQqN0;yOx!9JEZB#W!ZgG4Gn5d`
ztv1w2H!K5B6<G<vvj35$nDco&NbNl8R=@$4!*bJCNi}vCpPR%a7<M-1f4T;|%ci_~
zla}9skNQ>p{&r%~{Ob=Z6oA%5_1IGJo*^Y(%tt3vW8*=1Kk$kE^q^+!PS|bl3>qac
zFZ@uV<ql4!CZ0azs(0b?_ZZ!g9l_EeuW;UCzfpv;1+3yDFJ6L~+P5B0pTO1am!E+N
z<ew}f;9E{sIC}x?(>&hoYtd8JkyseoE8p<24LJ^+6=>q2qRTZY@Kn?CsG2;m9JlaQ
zU7_ocGs*Ftg@v9kz}jb#chsNeb-fwjO$7)*HeZXQ+G>A>03_}Foc;sZaxGMZIeM}N
zvtCulWe@^?{&MaG0Bh`(gmqJbx_;;wx{D2~_og|YZeL4C1d)UB1X6i!S*O@HGf)OA
zL(5Q41}*Fsg~-#L68S#9(cfrZ^+Kro>AW7v14*(*Kq|QlI^mx}QzuZ@)HJQn0LhS9
zwlhENeNrT6T4a5NefP3_?~1LugGXAxl^*VVLJWjoL~-;0L-41-T+=pP4@zFSF}Eo{
zGXUvgX6X>inO&DUg;Gj)E_=VdHd;vryX@+Ug0?Q+3rZgN8bR>L*2}hR#_+BP%{w{|
z|0r|nr~q=`8&Lop;;6GW!dJPQ2aZ9d0vb}dn)TN&?wqsTompk`0IdL4Dr47aJo&-T
zzg4_a?0qZgje1njJ5X~h%h$rg79*PAl;4uGWCGV=$CDyiIP--cn(ok4X`9{)+dwmB
zNrilvS@{(8d~swavmfZ7%4suflCzPMG8<i%75K21VyesBIYcU%iEZo&Q%k#gYa9D{
z|6mb&5$BWEvRF!dj~vX58l-dZqdo$$Mo9KV?75W5jN|Y89AzeN_ba~Lm172-d94$}
zX|+%_Z1pyVvd*rqf9F{1L0B?%J5=lKJAo~n<Z^ygb2zLTK*W|;)zLSxBsZik&;^3I
zm6etFL_|8DU)1lPFwjy{CuU}Th6J#%u-*_h)-BPI8lRq?^0zq&TnwCso;#Sy@?i`A
zsT9BiDZjf$5ngu0nre|SPd>io1F~WPnJ@jxs0kPHj-JL@6?4ydW0`)=KyFi!6Ar!e
zQDH!Uuv$hCes*zy6qIKl4sxJpkFLq`JFif}#BF*iZvxIN5=(B1XrVP)KFpchPweQ3
zmek#2DnZiYPmQkVb*L&UP~Y~aB`zzlo$pD&st%lwMygpwfh;@f3+FiM@W)=%XZ}1%
z%Y8hYeExAaE-kHTADZ(y!^XvBI&1up6T7Cw4lC86<+vCKwJ*tmF|X_Kb4)ua7>rB~
z-OHO44V`A_bGTr`31Xmc*B_{sY<vGmXRe(sYot=_hV(77lIic4BmfHG%u0t;AJ{Sq
z9580AM)snMmveb1I;)=>tzz`F7PXB&pjA2m5D0a2bwPjLWeXdn=5EOf;Y82xZ>Z-l
zFD!fy4MlVd3sVR@dvv;*k?;f`ztszk=obLMLO59j5|lTCewV+st84Bu0trkza9(tK
zOD^q9w(f%XObIcoKXAfDfYJnxc>Sw*3$S|hu~O#0O^v6VL2hwn3xJw1KR}q7oFT=h
zORT0hkysL=ttswD>;+k=l}5k5p4PqYT)g=`GSrWw#L=pl7Kg<COWe)SuH6!!D{mAC
z2^ZuK_|0zLo3>ia7EenT$Uu1%V?hkmY1<t2PH<slD#LKQ8(1qEf(iTp0)5?v$*6r-
zO*%QK_;A-1y6-lfANcEN38Lpej_}ep))HU&YF6o>XVv@-$8)~^{d0g?kfx%CB9}}%
zPtD8f_EBlVqNhWL0W^!PX0^KmHj8;DkY{jv2VdEDHOFC<$Pab#H~US$9VGn=u!SnS
z9J2U0)fv$>a#wdd#ovf5YR1hC8QvoyRYv%o9+n`Z_Q4#26|yzb7ugIc0iCe-c!(}?
z;^(zF6$l|5I5=P#28L_aMTx11Uy$5_-T8|)3Mg_KAzCHOiL{etM?=S)b#}Dw@E-C{
zXx(>w27u*iHu^t<?Y^%0rr@8f$wJ_RAH<iChKYCW<)??lMJalQg==nf(INo+uHBOi
z%>u#wCoMg&Ir;D&CB9}Tej(l3MU!3l{1tHyKxm-|#44QeAfVQPlQDpUSzySd6ig3m
z=!>2$x0L4+)*3<1rlzM(of?(Sny`IeDL5q3V(qj>6}x8hbV-dpV}v-b;q5Id+xiC#
zlhqzRw5)WVy@wD-<?Pf0DOl-(ODqq;8p&95;?{6o1<eoBUm^%+q%`;1f%tUR4!6`H
zSAWGaDgegC4vQCDK@;b&q7cD9oMQq?=MQ__12MUkEO!QfZR;y>U)?ENJ3mJF!n23?
zYGYlVeL7dn4BTlFuoa=1P;^lrTT~cLn<0-y9S;V%yfV|eCP-r(7LGA)&NIV?Ev%&X
z9;nTKRQW!gQ>otsqKlh90MsK2jf!b)#X(;!_*h6LgRTZ(vhv>>ugqZP2g~|xQgkkl
zr%iU1Fp70<J@|mA!27{hI}eBRSagC{jl3D2;1GMoiCTe+LbVOMgt;%ld!xEtvrY8k
z;3(NbzPH3@=q)Spt%X;abC&lRLw!Vq$NU#uET5PrjIdt7%)-1)CziJ^yOZ($Y303B
zv*y;OyKr8Hb@hBedVtOmsdGNOzwq-^cdOaSx@SBDau`(Kc37E3;G0@lYvu<^&)mXV
z8kC1D1UzLR6f+iDC(0b_A9T+{Y$ql_SN~I3yKcX5dQ6kg&qF*_jqbMOo<2KZ@Z39s
zPq=9{@Z_5PYhtMx5Vr0a{Z+IhqWCSF3_!s41IQt%m<Yvq^QJC-kJ`94eFZKnJ+a+*
z5_n#gGwf8AyI+i0xNoV~5rCt{Q4^3Ysx;GIK-A5v4)Um?N|BJXo+UdYo?R%Z9RY$8
zoRj;a>&3>}Pe|8R0d>sci5s}QmqfFC$M1S+W(xN>{SqWG8oNT3vL>?hjCpRsoz{cD
z;4WFslufW!CGZV(us8gZydgOVE4J6TzC3vV*37=#;Q_hMB?d!L!G6=9YBXQW4BtwJ
z`xx6$%XLLOU>|A4qJKnC>u^S+XW{Zca}cJnMM1v;A)oVSuS)-T$|3iy`rQ}}Dk;}V
zZ-K8;Xlu{}BHkeMJU_D`<Ex<9xk<-bqY}gq?gAx5us$ZLN|?#D@~xli8A|DpXa-zw
zAaUTHN}&CAe%O*muE=8xNAeL2%}j_9O~Z{6QDuIr5EImy@QzmsaJ+;j7X{N=IN0_Q
z_n8sDFcx9aNOqkS!4JvtK)`g+fN?O9`{pySioDmo8N~k@Q>EYjN^>&wAb)@`>iC8Y
zT#|`G3rE-ZOzHI_1mC@OfKRE1vPlHt)3WI(H-ar`msk)xB&lsIl!atwsr|-J3fYEE
zBD%yYQ?=hmsz$-dB;qj<`cH06!7+z7ROD9=eJwDDhG!J8Jv}1L1G<c_3a30&LwUj6
z6a+UEfU)x<lu%eeiZFec{v*c_tW@h*RyAzWm2$Ojaiad%m8rNe5w5DUa=s)elu#`&
zSNgH&$v_{3Po%YI?61Fm#$wpKb)7*=L&?mGhRP-V{uRlicAIk+aeuM@d)eZ@H}v$R
zaPo^L996sXy3qV7Xw^g!JwFY1^mKo|X?bKaHyMxol)2WLoQn2R5wW}K2`~CoyUZ|m
zDToT#ou3lU9j(VkDZ2FJ6+8`w065QyZJ++tIM`n}iR=ijcT^ha=6}dXktbw{&n21}
zs}cZ71yctC4*ta}=ouT!#EjbQk!AUM`!qf(uc&C_{4*}a0{owX`x}5u#ZImF=h$F?
zxGL>WiENhi%-0DuuNsLgPRA%XYKY3HZMyoa77B_l#A9LYP`b5TYyZ-4k|BYxE;4Lc
z@Ezp~ytiM`H%e!8ktAOUT;@~Mms|d1xydAqO}W-xCnT@D3u2cvJaDianE#L!2oe(@
zc?4s^=&ZivR2@@<7Y`7|B(t#Q4j#uwnKZlBZBYqlJhtMEZOZQZOre+9XINla*j#Lb
zXv{;BU0nk)MX5yOgyav7w01ZA|8KUU(kga?duH6G01*$b%41<oGAU2G@lk3Be{LHc
zg%8^<EPW2ab-5F)hj@^e`<l%?XCVfGy=w5K&D2;Ik=tM-s$-9Zs^NjK8xsbvmH8fj
znqrF3!ne0=tY{C-pP~#yE&Ly)#>7lYUjN1<-92$SG!rsy<1RPM-;7l^0}3-#FXCBA
zpf-AWU+E$uB$QNDFBV*P=sK$pk}l8qkS!6fuF|<Vl6e*%C!W?Y))lkj(*bTIX=yEI
z+2u`aa}Z&ix6|bJFQhG-Td&f88zFnf#ZlbJ*d55Tve(IfAz}^X$T;o%3VidCV<IFB
zU(nU|Atmk^AAC?1{B8Ml3&ZCdtRr*&*Zj>C1|08ncfW08y5b%JmK9gc#G{7c6ava6
z+XY>!oAlEPYW0KB`X3;B%XqOn;-I~k@#T%(^~c`I?T<IceT;s2ukmjp#}RxREh@jf
z2Xe(wkx~zf!59vZW0M)S(kq^p50YjqhQ$#ITQ<EUR<AX`9=WPUNM8y+DccZljcPQ1
ziA5YR+n8-$xsZLRD229T%=-X9O#r)q0v}?&NxiESoS5V9N#eUB;<BQL0h1_u^<}8A
zUZa%vOLX}Sa|L-PbRd5;d2PY2VZIQ^$$k^}v~s7fc(WGdnTrgZ>Y@;GACxgcl?{rZ
zJm+|Sr}}hY|0BE$u6`ePCYgQc#^R<JhxC}lBC2(U?#=3D$g~5v@_Z*zpj<CZK_#uC
zhAonBT0=_15j|6WMjx#!M_e%-?<EE8uypzlbB^dPv&$ymgyX%PIWZPPVaUvVtLwqE
z`OntkiJjFw5Aa#;<riVnx9$|{Uvo-E0u<TapKNy&9w%%$S}G9q6ihMgNn~ZBXDO?A
z9o&lue^O&Yuj;DG|9Mnc&$*Bww{VqYKNva{EW0?OxamyzxNdTnQaE`sGS;#1)0YS)
zRgdVc0(`SLpv@|b`O#^c8yCF`St@sj8)4bb1&M#e6(dD9%*9V7!W9GP4H2Sm?denT
z=4Ph6j&%4MbHx%uL`LgD9N0>4ojC7=!Lw2+F_^q*X`n#a`?~X?hJ+E9itSoYlukAK
z-hiCP{k*p_);Cv|Ohmex+p}DwEPxJ*Rbonr0*wR}ZKec)dp~!?HN4k`)8LK9n#De-
z)p)&`=f%2LlamsTYSdt)YH1%r8&h`n*o#NMcel$9rk(Dhco067UzL2E=Hr?CjYOyr
zeqO!SJYB8SJpCJ;`oX*HS`Yra_wL@21wR>|Uot)xepeK8_fGe+gR@$Q-82U?7h<(O
z)Du4vUxvt8#{V<0cs-^x$HC?6o_F8$YaaZXb@sZGCnro{osC(}9J@HC*Jq4(lN;z&
zat>GDq~jsGPIM;?nPE+D+sYdT4p}W^Y>;2>5EApKKiHhgyPtd;2B|nv={e53IYWJ4
z^g?|xmL~u7RczXOi7;PPe&l=<rvT}Ph^G{%>}OKpv*|OKUXhH|Td^e6Fd-I?sV(yw
z8*doE(=ZUYAm;}<l4(O0q*JdWE(R|_>LuR(6vhdYth?p_>3@q3&7?QS9z*ACJf+@E
zk_lQy+pU2E*Xqj^MrJh_b_|dYt*186CNjKdG{bHVZqal<v>E|@>+7RP=;}VX9)p0<
z`U9dpgh<YP;q3XE*f^ePtZGOot&eFX-r-T@HH;JdI`8a=`Q^1q`nt_^dU{dh0J75T
z%tHI<%V3S1g7v{H*wf^*fW$Y)_jYMD${n3TWqP{FNSlmV8O@jQt&hs+9;Y1!<pe2&
zRF@grm+^2DXxLfeJ#D`aO>aM%cgcE1vmAlUoBSesXYRgQu+N`3l(p&bh66!jjpMIF
z6GUi38yCCJ5Rmw+5e;(>QM7Mmo#A&+ZZ4ZO;e6_vV;C*gec3%d8{q03*vuoO?))78
zvPp$l?Q*PS&Z#V4{loSN9<RDLIQ@JS;dJmx-?k4r76Ask9nje2`n;PTiXFFaOZwIe
zxp($>Es`z)(e2v9E^p%7=U(|A%Z0eaqLIOU+A^1ql=EknW>uFz5~(_IyL>}np1m~H
z|6S5UmpA!nmG^5BGjs>YX(CR4?$76a{+jfo6bFvDeKtOq-<dr~=s2~9OXA&w)LBvG
zG{3Sj{OyOz)-*)L_j|be>!pO|Oj91&tLbiIrlNgs<Tmaq%x&8{2T{Dq_ftyL99&;Q
z7t5-P7}CPOpQ4@*8!GI4Ai`giR5nHFh>sev6i^XNo2$~#NbWQ6RX%}Io9N5z<-it{
znQcEiUNF5|<%P+kYmSO^dx71;LhM$Vtmsrw2#5AN@WfdbzZ39jt$&_s5V$Fdnq@8o
z*E8$x`*^xs9n;gk%X}L5X>i&YQI!w39*A3pti{2<WKwSRC@Ej}-Rq<z{~>c(vHzo-
zFu>?6#jG!PyA)gZMZrec^k)jwUl3w()X!14@{u7KjUO|CZKHLIRBk=f!Iw$fkS9lT
zD-fyB^{VEea;a*>OlF}q;qC*<l&|+SYmJA>BC4DB;Ub!H!Y3>P54P>RhbE4r_IOS2
z#I?Sc(B2-_rF)!o__m>XXGLOPAWgeNX~5@n^W~$V#Y0m$jqfev7%)rYb*0g24#vg(
zv6YQnv+dp|WiDq_EQ<=SyYY}dR#Hbjv@e0sPR1;?*R?|INKAg<=i)rZX{k`k&|!@k
z-C6A-{&@CW#6_WY;^$JF_MuY>d*wM*Y>j%<L7V#*>;5dMn<lM<E_QplKB6)*w}ZO9
z=XokD(7t?iAd}-Rm%6{^!egd^5FdDT(sZGWi^$$VwFJoJeI|pqY+G>_ZBA4}V*Z*^
z3I^zU#V9T|1osUrn)2=A;MYWA8*?0yU3yiA1Z_m{jRI=QK#>mS?aMzi>!v9+J8x*=
zq%G0MqM;=^yo>j<A_E))ybQ&yq`z03T7}n7C!i%vzLyln#M-hC8S8=M^tY#qyXi$n
zOeTxJm=M3>Qoimr*&r@qMK@N1c!Q%2*C!O!g5Fb-8$~$#il<RJHb?|UgX`)~Tv9@?
zHTNf$y9hKw)tE2zD50~*tmwAHYNwR-JOe?95}Gd<IFMX)Q6jtK!xn$I+M{tnk<Dg3
zCFGaAG~5#>7CqlG%rTDY2E7s7WXwKYCvPUoHExD093+Tlcp0>pox6@f{Q-R}l7me~
z6{qe5UrcN^G|%qVUzYP&Ky~7-Ypw?hX^~x&zr?NBUCi%KEKk9<KV<Oq{TC*<4tzmt
zfmKKHtwMixpyH%bb4~D8d;g-i*H;RDvWS|f7S3{gv~A}!PbHJ}VB3jH**9jaTK6SS
z9)bGy4zD{5+HWrUyfn$@u@Ue&$m}|3exja>IgGG)%(~k?x7&-0%q6B!j<k^G6*%#v
zPN&Gp=fSfu>E-X>U01zF23tD$9FdUAH#_5kYs7lKvb|w1Z~Od0Zz3UcNM-yW`m=%>
zq`$l%%)cy?x~19hr<Ul|`0*nB>W+pVs^yLmA3hb7Um((hg>8e#$)!JcoR{&dh?vU>
zwuD-Q@?6wfr-iJi=qe7*Uc?y=I>e{zemuARe6!r#pgYHEu;0ofI_sYy-{-uV?bGw(
zho)p~JH0Hf$zMr#uEauE$(laDE{m=!jWb&bl>Yp59ex$3Lp0~3&h0(gY~{9}Kpt01
zhqw)qz}yH@R3*(m-}cm{7brLX&C489gfllOLj95C`pZ>3j(cuDQ{8a`{|`-99Tipk
zbrA$a8l+35yQJFyLAtxUyW!FyrG#`#cXyYxbTf3r&>i2!_xrIH|G+Gmd!FZ<v-dtR
zSEs03RXNn+60?JKPyzArpZMn-O5pK;3CO$#0X<z<6!IY^8S33QW(T%U2WbL`4`Ry$
zkEunoO-2O$ed{zsgP%*LSB^GiT=TTg3Ha0U1<3)~BF1mRI2*>Y>M2nsrZSeqNz-uC
zWe*tgSIqAycUP2tt_XMx-*BEJO_*ApAhiXZcUk<A`1Zcot8VL6nVM_BjqQG4)gP&W
zm7Syq3Mj9Ho$)&*N*r#trWMiysqP<4j`(wMV$AX1NRHepqkhw(H*&IadZ1RjWbtOk
zu7$$_i=WiR0@-|AKEo&orb}IpeTXa-5FR-CEK!MEz*N-PKL{U%LH0h#nbO|Sg?KTQ
z$*KMy`(YHtYK~A;Qxlhy8_Pk<9xE(7U~+m@5~*F*r^}+b$I?w{_^3gdp<O0cR@Q`G
zP=A`*LRwh32R)o6e6wMcF}+jnV6MK_vkWi_Co*$&6|SS1FT|M$X)8bzws@oDyu`4l
z>W=o-k_ZKK=US3ef4=s~_Rx)4x94^^vacLx3bV*9u&;{U$n@7ZOKmC+H@yG&KAB&~
zA;0o&pT`mHGUKX+e8&s=YKRKkm_2{8&|w}>RH|#;Zssl$c3YW3dGHa4d%bdhP7EA)
zr6vpNusc|-OF;-`#$9@AoFdIGw|8vH2-Mfln3<TW2vFPwM6*xigWs#meu6$p<+io*
z$jZry7#Z!GEhuVfc}aVI455#ra3vDYk<T2fjx5XP2vTP}j34Dq?v^VY-e}ewmZX~C
z5!Y@YEwwg*{H{57$|x(yYAl2mGXbs;?s}%bWRK}{5{xR2a6Uq)s>WpbAXByv7BvHd
zZ(}NU$&M1Gg64UMJdVN<1aLMfxp|w^Gy^>gOsge2p$B$cJ#%MeQ}fkH`=uiuawWZ%
zA2O8-;=cS*XRX;9YWR?}Pi7j*&O;xeJh9W3$AoJT;hXf`{6Ldfwm^evx(!I5n)a6d
zJg86rGPA)N+dzw*hbJM3SoJSpLU?)Rl^w^Amp0^cb91W)i??+63`P^YxGvT>L&ofJ
zadD<K^YdJHYkP_c3JQ1qwbFS_#X{YfQYu<ns6JBcPk{wx1^R=9>0p5y2??=QPB}|)
z&diDPWsvB&!wM=&i#K716~L#)pBJnG(*w_=_{L~u2Xk>}b^tlHP!b*LV=Qzu{ijdp
zrO&C~B-N<X<)YgSf`;c2c%~k8Oa|Mx5HPQ#fjeu!Hv{^g&68}_olal7oWjnefo#mW
zh4r%cK&QsA{n6_<hiz<(ar06>jb!92GY)*zNqmd)iW-rr0d}C-`bFA+f6892%&(%M
zRm>Hw?S=+LNl|PhdIRo8V<B1(EZR|K_}G_q9wF$5LC!h**R9tsNLJGi%F0U9$|tq8
z4)2*-2L@h$AVlT0!oJlCQ!dNAdcwiMQB+ifk!Wq7uD!Tl%9z-t=%??-k*-vGMr>ay
zjizLVo{oh!wt*A`;gYe{4V<9qUf=IolJe1L(b0}Sdd`*&cTgu9_Q&e%>}=`}KQ=LW
zaXTHr5Qt5WS+_4mL^6}>F~VS`x^uZl4u&S@k!iPfY~+w-WpQI8syvB~#(8fo*6dr(
zHGG?Giz{394p8VrgBi$@y|jKYJKH3u^-Ik8R$Q6sJ^+YA7RuCsSTsCA$guFIGsjLi
z>YP*WV1E28FBL!ADU#^tslZ00R)O_f;1Dxi3-`(A9PppjuP4tx!N%EtC%(&fum&CD
z@8F18PhL|yn}<D>F`^gxo8P{GK>Q)$`}s!t!22gf+@@BoZ+&uO;c4s*QzZV+N$Ru1
zkC))6Fz{-uGh7r7(tO8FkL!sdY&!|k0uPKQw64E9k9(oo(O%&StjIjgTUwk<>3yrS
z!~AE<@*l|?r}=>Nj&-%OJ3L~@CF9{OB*#-_Qo^_@be*qNVT*{jWABhz44hP3{?uD#
z-=P^@3y4i#L}<J!{cuGw*T=cOu_h914M#`}w1K~Eb>Ef))4#S}%MSk@l3!yuk4a|T
z_cclRtEY}kB^%5-55}jo<zQ;<hfr<*UQW8+E!@Rg_SP3xf25aWMCM=c`!#Wyr^R@9
z_ZsHEYkc}#-8e1R;*s+z+}uAq9(gnP+o~hw%Q-IVtQt36&IxPPI$SF^U7V`KVKPR*
z;7n{VR!FR&{j*Srvl4C~AAJ*HElzu<WMpwDr4z)CYS`<?N*ZQ7mG!>_h5P<rg35wV
z95!zCbl@xko`^N5Wr9Yo6Mdxdv|s({sS{YTzGfx~7U;foyZYG-<=Ps;AK-&Pd$`K2
zQtx)WhfQHWk&6qp#!c4e1ULe9f+EQviVDm_A1`$vd*7sOaB(@T*S4>M=ap`Cbn&U?
zYfnd<&A9j9mRX)VRW9E-$srpVJvPD}w{4g;fz~3I>(+FxvuzC_FsAPaSuy?eH_>Ey
z2VL2d3C=|S>%AiIjA%ekv|U3@ZQJJ(HL;a6E!V%WzT8^ue)Glyz<W%Viz*Y%sXQu5
zza79rWAMWCxxA`tp%{INlgv>bbGd!R*<6R;DiOF%UIHc3)c4UD7s(l#Qn$RdCD0Jm
zMsOy)R?)!Ua9cGvs|iegSVhpeW=K56;kmb+xm#>mp}ptjY)y^$cjYX|rBp#K8iHG`
zH&<8u_~3Am0<;iZL6)4TT%JaHJ0ddC88W*5BGS;)u3S*tbc0dap;<KB=$70Dw)`-;
ziMC(7lCIaUn%nu|8Uw~f@OE0M=4-f$hENtY>N6(<VW(fe?3G*V4|1aIXyPgiF=o8m
zC50!4WKx&!am<h5>e)t<2K1rCY*iiR7B$mBX<R@{FNcfg_c=rTv95HvB&{d9WxQ{r
z;{LK0lZF(<<koY8%2pPf1atSBR?qQq<><5QRijWlM#RijDJJ(_V@Kr2Yk`Tz2vlR&
zP5fB=9nKR%&-ZW|RA1>y7Vyg`Tn@n^5jDQ1nEe&>VDi{hBLO#{wJ=DB>^fm@e7dT!
z<$3<>-DidW7?J6_IGmiJryW|DpkeiZDwq16g+fL1Tj>W3h%=nHIPaf%$_o{)O9D3E
z3I(#9#;hN4HfU5=lcZsnF>a3JMau+{nYn%i>^i?4&N0Fan&rdYx0LUB5n{YN>5Cv_
zQc7ytT||k)&+xg)+xumWa#sl%Fx)v^kXF{7JY4K{s3Iv!?nTuW1&HkzzT-pj@<IDV
zmQh)sB1OhjW4`dK^5RS>AsmET9Cz`p?X8fsUQ0I4`qb9SfC~reXxt}iutfL0(_R`X
z<c9ga^>*jLD)GaW$OlgKYkqid2}Fmk1vzsc9<fznB*8;Cg0sa*+^8u4i^~5eN7+N+
z{i@C?#;WN}FI}v5_k#e<2#xOcjD<Rd=s&mnCi0JKe@0eoy8b5d+QDqWgwFdAQKO|r
zJv$p3iw>iDK3tR_mFwOIxtQfcA@ZQc=}9bhHGpaIPLXmG*@N%<wLwvF*yN%1aP+ls
z5V>COv}@5FN?ZgOG-Ex#(bJx;>(2<$arL1+qt0u8ZlnIH#HDdfTXOycEeo{rgd{-H
zK3Wv5Z=oAwj2AtorS&&qi-!sLY6zQvGRjmwcQ$qF_lOpHL|iKi9Hr)_Ryi!gr9OeM
zk>$YaIGgf=DZv!Mdf1qL+hx~JQt*AX-otC#yfL_S0196Et9kTBGM2OLGa=25oibA2
zrFq-IC*!QDDqraMO2wyvPK#V*LwEX!at5NKC@}apliw%a_H@m1&!-AqVtrL5_#|SF
z=IG$-8==Y|1<#N1S{;Y4O)adovT@GfKNaZBZ^Y-d5Mv*mtu#@D^L|Esa7P>cn|57G
z1bn_cDIpeW4!+d7o8nQ<9y>iggN$=lA_{r)27_K?Z@8`rhS$|C2$F^vdf~t+jik_5
zIZ;@c$Pw#B(A426wl>+I4MFzf0J1H$lg_Aa=H>p<bTYlsPGA*#nare&<pVDt-FG%H
zlf@K{6T>u<!`7c1dK^Qs=2z|kvFQWEjP>KW{K+H)QJsX6Y{b=vJO{%EG2#UbvC)F~
z(s^F?ci_D;Cl>g)BY5Q2=`Ldoa_v7dbA9tqOFrQ11u@}lZIRmzCqddRad~2ef1v~u
zAPVXB;cM&8FMDq{xV=I84KW63Qr_b))_XDM?X*CdifbL67zYZHG=LYjly_(2Wg2G&
zmxwY4F_V%yJB(KE#b!{vby*wR!+SpWEkXUP^oxgp(bK9xi^VY;#+8jr^3Rc$i?F``
zpm<3r#q*Z35L_@^ZJFMq9i%iLn@{A4JHUU@l6kgn_?)VZAJ}^+&+5FFjY238C*<uK
zPqL32or)_4h^Q~}tb-YW!1j=$cj=N_n(0XDJ5)k%U3W0TX6X}zPptB6pMx0kMn-v!
zsuJsoFk8Op)$UXzX&oAc`9~o|!a=NP+NL;hziew$`S1>~e*nC^+%@|<lv2l}s?}Pd
z)O+sWT8A^}pL!mTuaRnB*M2egOD{^QFIGC#j?HQCVlJK;^G3PDrXN4n2wkD<ih#5?
z`ihvkSZCvlKx9#<O|Y)<;|Vch{Gstvr`7MHZ=KCm$0@9}($1ezb_Y1FBHw3N%;q3f
zljb>-z?ZzvT1w)SBXgNm7e0-#j3NQ1stZ5kr8~oHZ0Bk;M-&C4PmsCAn4!0cfB?w8
zN}iUUw!LQ<*``g_6Sgzy%|zT#o5L%v78f!wEF2i@zWKeaU(C6(wjE^C^!7ljWarLJ
zR9yKTVF_d8k%-jy;U1L+TqdWmpsZ@ml`;$L*w!&CcbQx%X#b=lu0l%OS0liV$~*$i
zLj_ob<;@(MPveRYv*9~cxQcP$*TQOMl+2OCc3S87yyS|i)Pu7@ZK9B*5HKwoTvbM|
zz4~bCSaNpgn<*h<B*$|-PZyTU@g-y29#jd-eg8Z9_~7ZA*UvHxsAa{*!(>!Bn4M~J
zUJek)8Br?R6v-DH1IVxYnE#x@F9&=qnOean2(d`OgHcb#B}$FBz)l=ij91d%-+?de
z^)A!2yL2yPDzY5smvVA`;=M!GfpA`yx@xF4cwZ*ay&g!S=GMdHQ$d7ns%e<iQ!UQ@
zXa@B)SXWO;=@LoI&ee4PCj+ewy#0TGQtx&RE2{=XFKsl?eTz*ivrcQ3O53+rQBxEh
z-}+=oA|PevjA<T0XhNoP-xP$&n(t7>h+vdI?qEKHDQ^%>GeXxcThcezhXc9ns3w}m
zF(IRGeP1fSy0iPyO<RyE=0#2833J=X<&)EC^`pQ4y}}%Y70+i83AD`{%30pI^|ftj
z)wZvIUpBd{#;i4Ld;bECY1;nX0c!)I2ZtNeSmUIkqJ#c|9@|<B?J#YpAq(x5oFNA0
zr>3=)Bf}HT%hJgFNxGrH+gwED-$Sb_yNl=ljpeAn_)qX+e7-jxVtC)HBD2bvKABj!
zy<bm1RnUhi-J@IdizVjtE)NL60az2HwvZ#ocB&oQ++aSD5~w8mR4Xfq!!NX{_N3hq
z{Z*OPTEb(kNA5~b3DL>J2-~`Nb&<(Xm*6)E=Nfonu4oD4)<00QLb_%!hhWT3T}xel
z(n4uSEZMs9W0=e@I}aJoSX&Ib5S&?IcD<PiIt!Na5xJ02dQRpHfzEOwz~kqx*)nq#
zm&;!kz~~T<-K0z3OV|b92R<Bc<WYZlpG|j_R2P|c!3Xn2{}$S>{l3kF$L60o9@P2<
zUTn7h7r(K-{!>g%Ux2$Lb{=OcybfWd`<}#;mt%-HqUH=KoSw*FK3|bdq`cGS)CZOZ
zzT<g=>Wg0$?k(UU9Lw&y;Mb4RFomJvXQ%QG6upKza+(sZkGzzb(_(9xL*c80a`?+$
zf_Ui15%#MCHbRp!PE$u`#WyC=4xW6pcX%<?n;7C6XsjDQPf)x~8R(oIO7PwXIlklA
z;-<axwvhMZqP%o5YR%LlBDTn%uDYLthUzq9!>F`<{RntScC!~#JC6v0T!yBfS5*Bz
z-H^)KaM%KQwoR8PM#v+-qgw^EU&V>P&{v^m1TQ68-lkrqk6UT;pT6kva-!AZQr>*t
zckxu<Zxec?oHA0<&GUVihH>~>ur@=Ewtvo_GQs7|k8-x4Yzut?0h8A)He~a<$T1=B
zwC30W0l<>phv<P2FN0uK8%DV49V$fZU(k++`~O>H_$PsHw$=FsSi*Q$yv>+DIeir=
zIe#ON<CX{VSd8VzOhyc&{o0cUO3Q~08A{d<jIXRu5buyv`o;uhEq%yj6sjn~bWDc9
zgmILyJ^2pzklN0R7}*5)-mm-HDjg^|Z`?#B78ofTu>vtm3w7JKwXM;)t2=hfM_by!
zf`-NG>7@&*JYLA(WGOO+%|E>jj-(G1&eGF$<bPyBDKFaPiTP8^7TOG23{aKaTif;p
z#xE&XL3p1|3&oz^pFYF85Az$zPRn@97`fEOFQL3<+}gOyaddwo?#cIZ2H<H+ErPQa
zNqdvCY3DlZ;Yjw>Jk<wht2-A46M%j+o`3jd-(Gi&^Oqny*^n{*cF|&$b>kUg<hh@k
z-mwQ&E}3-IX-_(h^r*=OU|rzySc-j1hqMe&z|xCoNmMbbE{RF;Yjv&{?}<qT0x5Vl
zZ9N-@9n!yM^vRr=44|SKIv)jAQo2iAhxZN#PZdJc&H<RklOGogI!9~q2a3^6k?OA5
z+0hJ*vi4UDDvbR22&1)#B}q8UkKF|%VuXt_()zs5M-OI%R)wd&A9p~rF8Iob>0i?1
zq<RMve0GlL@@w~wUs~=!u{XExw=a0r=2!tL<EnxVp#E`0R-pr=VlP;(Gb?ru_HW!9
z=Z41f+Q_YRu+lQe!aX>HI9sUG(-$9zqE@WKz3oZBB1vxH?-U(lZ^<trW-Y7{z#$XN
zno6sPw?wg0AKVX+uH~VZelD3}S-of~Ue>N=qV<TZ@Eo}sXbQ_CUs>hElRio_$_rY6
zT`d8mSAHAysT9q%*=n@hfs8a;acg_39glVzQBB_`O8N!ML1-wd^76k)0!=20e;XUB
zG2$e8icJLj{#9nu)@Oe)GcMOQzm0Gri_9tf5I@E>{M+Elml5A=UqMdx+1up{^`D4d
zEETINTK+)fCk&c1U5m<V?UL03MzbW1*})=BaBF4K(RzB`M2HySebTksk%$(-%^$XB
z5&jlJwn1ey$>%-0&b5ZS^F`tSdhoc(Iz*NI<!($RzGAnke<SaR`nSqUfjqQQm-{g%
zMDA)&bT9*J-u+4Eu7_8xCaY2Z)>c4<HNCI=o&B}y-By-+U+_g3g(skE5`IhfT+7M|
z@4`Jqe9LG}0P5)I?SIAk4Av+4!`4H;b>(xFpBF$`2&7o}P=eItGM9S@(k#%;af-Ab
z6Dt?j+=HXi*x^06#8iKw3gyr<;xWkS`)p&sD@sDkcK>VUSiQ;mucf%!KKkD>P<2_=
zCYool92|@V7Pd!q@0#xl=!OeCz*^9TSD5Ohgw5*-^s$;N+oQXyU_9m(W&MS1`U2WA
zzY4*>gTuE-qIbeNToP=^vsY(Zyul<|E<Wei3iXIz#=I6Cl%c)k^?V?-51unLp11=h
z`KCLTLTS@Sdq;YF$@hiyhY<~EMkI<|l2fqxuFp6xUi|K1?i1)v))4Z>h$uZm(=d)n
zM1?XTtR$JG>qE_r(&gj6znExipYBM#rsN}%REIoXkjIoJkjf$@gmXAHmr}!#)h{>V
ze<`X0Ryki!*vDx%aec^&w|BuySOG!4Xu*70uto-T%$8utv@#+DT>8=XHtz@?lQp~A
z$n_e7n5m~s;T2b_S6K{kA_=fLJ6*<IvyJ!BV_8x*e&6=gIt*04z(S+F^brM*F=k-i
z`U`L;C8@FC1(LcJwsKe-N@4^R=%}zYudUI|?>|rY>$zasg}IDhmHxydws(F}NxGh1
zRu?GZ#T)6G;9s=-9*lpW1U}oqJ;^go&Gy3sPoRsLU5E-)%)-)veMwMa@@tCd8g7H*
zL_1Q9+mg5Cjf|b`0jqVeZ>wcItRGx)(T=1qdZX=V5}UCT!7Ti~ejhc0my`&R3k0fu
zvy`Luw!79rLaY@>9p)V9C2oN?6U$T_RG}niCib?Q8sKm<Qro4!-G8LuSGt5~C!zGA
z)>kJt!F&zH^A?Af+1rw4is4yKp?6-vr@HRgM>7ksdTN}{823DdHlM=@5|!yS?=Um^
z)K^xuCEr*B<BZ!WjJ~su_M%ooBmScRF}(t=A)h*eaIl4k?K<C;{<`kMbxd>;12BIu
zvkPiGiYj_>YQ33na-J+=kzFSzx~kq`My^<VQ~pSoGxvr2E=>c;a*~L<;(|q;2oY4g
z%+9SrSA6+psXa6m;$W`<n_3E0g1U}!whmpdt6dK1jMk2-M7yt`%cE=7-_G=mre;P(
zQk&~PxnpaakU(59e9F`6J`{X>!&*GU*v;hjYn=#vI(-FxQk}!)Ax18l0I*w6zQP!9
z(#l@PE)n}8Lc4p?zhVSOl1e1>vj{CqGygjy6(eWp-{r%764JjM;cE2ShR<>9VTt!V
z`BYEg5x&p8_QVJb5#zg!Y4Q!v+v`K<qyM8mVco-B@pq38?-Yz%zaB)!Q5--d^sEu4
zlyK_ZZZPo)#mhus4Ul5?!MEXUf9Zy_Jbv*0T#4g4;Oeu10pz#VP4!_P_}nq+DOO{y
zsV%5*#*vxORC1o0{a4g=WSJCIK84Y*!8~Y(eEMGhD;|g`86*zU;?j##_@K33L{7ve
z4Y#8D2GBy{s5whHc?S3Kx1c1n);;~{;K7^Rjk+qIu2eWj&R~0wOIWs!^SsIAO)E)~
z>?W;x>iq`sCgJYN??nfxu?{5tFk0@oC0#$fcF>zwWh4z(hFDW(e}P=4F@*tWis+bG
zaxVB}7B_JEEf!I#U|C@BlbMZNnuh(O_i1dp+EU&n?8t~*l7iusvpOHiVUXyUwW(r^
zflpgxRe#@ac|GW!n4R)_5X8*#YNWxSlT8L5i)ba)0(NsWLJnlB%br*Ja~>1&oRQ9s
z#3WK;YTW-OhE!<(ffWEFOR1tlXo9S2VDh}5E&^tUcof>;<}4JncS;2n)yX-(uC^<;
zn7j-$YcM30$YjXFpAnXU&7NP`SU11mZ`|U~-D8t2_&cWoftxpEkx!8mR<mX`FF^n)
zfad+;=Ke`Zh@1`g8F5@j#n{ThdQezd^3>?hk=3D#vG6Mki&xN}8d}QY+V(LAsT2T*
zSEZy8nh9CmnfK;F(Lo-?_U9NEg(R2Lp_FZC9#hZm4TFa<saXZ}UzMqpD-S*s4-hgM
z9SlyC+f_2y9USBxv{7uoOv@BtjND8I%jSx>CaXKE0qR8+u1yMjk>$`C#*|t6kiANW
zbRiQmu6cB*XgXi;Itk|Ao!QUW3BWKh#4`jAx$bB61<UQgo`d({8dPK>#2WN(3Q|2m
zTq!jV8>8BT2loPk!ixi)^0+Sg%V=B=*s6Zd5$Hox!1FaeZ}g(6m6o!&dH>uZqsRCk
zCd^Vo+9k+T#O!_Jo7j7GQOwerDQ}s2coKoa51qNRmUb0;`6D?7HJt|^J<8;-KUj3N
zF)efMeyj{U|5Ri;;US+Lqn*wQpc$S6)HBM&VZT%ok_N}~TFH%|tec3n<~s+))gM_C
z-Zdpf$G{&v=#jGu+2mdwej>O5>N09d+AjQD>V?1%s7MhjlK3Kp@aD1;<v>Y2rMedW
zI?^iKq~}aXK3Xl~zFi8@in#qt5wU|gWn!@WAO6_R$TDU2y<Ks~rh+E`+ywrwK>+7}
zPi1#(OZD62WdUoSp2iE5Q`5!IMfUBwxl4cZd5dJp%MPWj<q^a<y+1-|$v&8Kx%GR1
z-8_ghFzBF!IKwxsXs~jI#2k|xn@Lexc2PG<*!90Mc#y{ZyP?&^&$az1NJhsFytyr3
zEIk1LY1b|<;q>OfeLD(&9tTt#3liX;!SF!}V^QaQ9y!q#ECx@dhrfXg8Fzmfx~C-A
zWpcgHS*?q|f_C}Dz5T^qH8t#J1_dNE5~7}L3azfHW=n6?6;F>o<MLl4fhh*V5Piab
zh0ukv|IX!xqnr-!tP|yO2$G@ckUrcCj1A_DG$;{r90t3rx7#eNjuIy<{@FWN3jCpw
zT}^#b+bSUbZSL_#V!EwKRqbnO8J{GoPvpq<a+&dpB0R%ThW#hf=cX4nv>e-|6xz!A
ze_KSEUjMUZ{l;aJw^P9<ZW*VV(<vhjjRKbDHwZ(-*%YpA;(^+w+tkQ0VVSRi@cljF
zX_I+dXP_xEJ6Vv@{*UD1J)e3FQA~e(nofFtL__9d4n)#TLHww3Z{_ClZ_iC9+!WoW
zOPR9|s@w6Ac%)T`!L9;??J=ThKLxTp^)pQ&xcur4_2BM-0AbPWp(PHtE<8M-rXJB_
zV)s`2B9igz$ZFo*qnMH%eE13AJEtdcaTHj;8@mB9wQ=3WR|_7$zIjA%xSPpS_ZpoV
zR2!*q^iDj%<iMF?8ACS?@XL+dpPWFjMpsPB#*6w1>gGD6*9m;Uy-Q16^8=JjYH^JO
z^gSO;?G+;<$d);_XWZ*{AH-hOIxFPwH)h!TkQ*)WfG2v`AXjbI4B)ltKFm|7&%Wo%
zrU4j=;Xu_01<+sXK~ej6H4AvPeg8wgyoVs4^wc%^TIN?LO`~&Gr&xoe&5J;Z;J(O<
z0YMXSLUTHGHd-m(XC*LKeHwEGl}H9i)Z7qg#J6|XoSyTDSOR9UEnQsAO$p)t{Srh$
zS9~+6z~V#Rs`jx2MZOa8wH2b+%u$)k^-6HBK6I<QqD05RaUu_wG(3yu?3G4q5<>f*
zJZq8cq9)vS!w(g@m(&vWS5$y&vDH%lQk%m>heqn#fxv_8JhhuO8RZ@t29sje@|6ym
z*o~t@Ul#h}qm3ho+gv+e>f7<q^LmA1F9*2?)UK|FT^^^9@zRyR&-*dy*<XQRjURZH
z%sx+~O70V!6nTVc+$cynFC9^hhN={Z7N7GCxa{<cxHQ(F&f&M!-5_kXpgsFqPk~er
zTXv50{OnH`RD|jq1`)#XwP}~!uRgth`=|2Dg!zSSU{mbpi}cY)S*Wa-jEVKb^$T!r
zk`HkG@$fCJ%h3FKF^5dM3;523AgiDz`E(~bVW;TWvcB%UAT6!mIBs$Eh_UZZX`ys%
z>0vvCS?5YUCpJ+$tEx)3l_&M)_l9say<}ON*B>{_spF3xwts2PD7Gj0JSQh&UhOU8
zM#gm+1ivk6Tt4xM#_;CE<lJ6ggLVu*CC};N-1k&ImOwysZ>#g_GcdK3wdVKna5c0R
zrr5&*(>NOk$>yKLFZ!<@`RbiIC87&;?yv`JnJo~)7}SJDV}<va$jt^<5I3OG3=Q>P
z-+U<+%ziN_Ha0)iCj5p*;IqxGF2YapCWR!AXd&&rj>sGKewk3ETkA(DBRgENX`mZy
z30eo++iWkL0F(LVK0EZ2QWvu1rz=F!7P*3>{h2>$Ra`gzb*P|pl``v|znrP(Mc&|a
z!~ix9601J40sLe=p;|(q-8@IQWBP?bjy*k^`Uh-)Qqa==3TYB$Thrm<z#kbYo1p=H
zNL3l<R;E-vC%s^9XvAh(y{#obeeS+)o2J?>48U=pZ!dX~5JBZUs-Xgv1$${|b@+=J
zhFmIPk0Sh;<=^JWH;6qkH4%FE`|Kj>+F+*XwMr&my^Ku5Rj#QkPw94)w&X+cu4U;c
zvx`AF4Uit$ART>M;CrX^>`_?->cxm<W2S}^xl#-|ttqp!s%-4+PwKRdq<v9jHU&R`
z=R{|h2wGrk`HM`|WkleCPCe?Gi`lUjAt-V7Q^ye4Tg3Kc!bC*;YQ2impEbb#I}4hi
zuealuxC5EtHkZXL@0AQzou>#WB$qo}_V%;Ci5Q6lOXpz!*QE7?o#O|Cw6@_a_{`^a
zOVaQYGtDg2c2QLFmep;?%OL<Fe4W8`ffhwP76B52;qTi31&63v3;4l5c9)Uu%RRwK
z16UGg(phB;FWTp3QZ1ew*g-koiY4}5L2V;-NS+RdkK-LJ*8owg<mPp+E7&Pu9x+M!
zA$};}gGA-)O#hg1vf7H?>-OdkSTlXBTMY#l%HaAAKD)l=zPbA`Ho)1d3b!juQZ*Kd
zl=rw+N$}O5GC8bfABmEzg<&}Siy?{w%_Iy1_HUYiup7&OMggY{3eG^oHyTQi)qr@8
zL&Ni*e)Kf?9%<R2FOt=9M;5fOhQ9mrCD#06nR-BQH{f`?#QuF&r_R>$Q_88+$w3>P
zo7a1|P~x#vAWZ=Qbd9lmu!olJ7Q&T8T8E+wVb<WT6^xv)%gd)5E+KiAPVZKicD&DE
zz<ytbv?gwH9k;&|C2|#0dz-I?@?zqCn2*_}U-~XyOtgv5yC?X6bBU?mecPiZMmxg$
z+DE-8)9NYG5g$kDDj88#@dU-z{YiXNL?NnA9QhSVN<P1)gou5NaX!UMs`0zan8<v5
zU_%51H6IvTalw}+qT45}d&5ijrJ3qxJXejl3knK8fBp=EfsOr1G1fD~wZ{DNrH!2(
z>)6;>Y<&F8W3kC-+WEnpg7W+4Jq@3;-D)ljz`%fiaZZ7v<SyY5CR;QGc0{Qwl`LR*
za8V-&+6sW2mek7$6`sHizX-Md&2`iI#(cfGQg+8HXY(9J5ru1-vKXF%w*Yye@i&L>
zv54No$MYjIxM=b6NeT=|%f^aH4!I;^4gzA-^fbeawbyWOOmw({1UY>(qomsNxnzGS
zg!h|vf|_$`i`0aHxk%2ONvRElDvb}F=x^a;qYijjALf3Ni>X$dVp`0UZvM^uq^PXC
zwzVbd>dFm17YaU$kN3kZfQg0WR~UEhWFu}|MbM%RITp6-vABE*U2Jf$h-tD~s5QE~
zJiNZ)sC0<~P4%uD-t^G19&rY8Rm+oKLzv<fM<-ZP`ja_Xk}3zWpp;-9eU`={fw0Lm
zMj|jZ>{Cn|noTiJm1%M*!7_I$4|9;Yo5Oet3Wv#v3A?sS&1e%4_kowO)Hh8kYW+R9
zOHn$RN*h8aoYyP6RDTAJfvz#<<m5;ueF#-5Qck^UDlTTItgM`E^#a=l&oN<y9?r29
z@;IAkOEFZw8_OI&-tL#xMBy>2Q}jQ7fXzqErAhNiLXKQ{xE-l&s>RnA<wY2mc%P(E
zV~N_oAU@XqU62B9Mn%qMu3>X#$|Kfu9+VZWAgg}HUkB27s7%tSYOze;<?wtV)*$OE
z&;susO-A8wSZW&;P{<C&&r%cxm$W_H<kObDHyeBhHUws|FvfrPV>{41usgZRkZMD#
zCL0^h@*XvriX65H+TM|jvY;-Tz9Ht|0{_OTv?1O02vC4z6HgZ*Yu3$kekyK_{O+>_
z*zEKvUC@i${d^|}EoTY|6}85+h2C!S`BW+l`pn5(4(Dqc^8RfMsXqW!2w6E*1SF%s
z!|>lPjz96<C!z)ctu$o;{_RPclns2eFxHod8XUxQ@a-Hc<5c0z??<ZTxi#mj!Og?m
z@<5P!Q3v^C-mWi4cIBczr&qnA+9H$;XX1vs>g!zUAKxzv<ql4|D{wOtAj+$*{dK|e
z$oDBOO$$<Dbca`z&s5Uf9)b;STVdYA^QLe=)eYk3M+4gd5w=SXtvu@ASX+JE4>QLm
zW)?HC39}GhztzyO2FttR%T9_sE}54RK>E(m!yd`I5-~8g(+b`lRLyUcO6RQ~)<L#<
ziEJp|zJ)1K4x6*g&C2pSIIvt>Usr4O<Z*U(j^{Ri$5aX)Is!cW&)DA#dl_S<g!qjA
zqW>mxQ!X45y}qtNT6DhGJGlby8ig(=!M`&k9}haj;d&6b<<R(E>p2K|#sj(cPk~=I
zpM+<S9+-AdUW$q7Ztn>zX@=SplTK4}zR0m}7!scKk)GwQP2<vkmO*VV^CaY#Mc`&`
zMRd4cRU(Q-L}v(Tqb(p8%&WIIa^d|f*Jq9qO4V8z@51UKfWN<O?;Ch;NQZ-R67Yf)
zV-=>9mxk-3I^_XfcQsQ~0+7yWXLun<<#meDF}8X!vXrZX=y=JwbA=;P@FIoirV>zj
zr<fdqKh>b-p;c!LLVW`$I+30NsU{=<opfZe4E%{j?jO)o_P~h^=#=1coejjTCji_!
zK}}7al9KWY9o=NDI~0%a^E;*nW77erQVr{doJ%`1GwdSX01N>1=~9?aWCUgB>zZ-M
zGQ`T!SzRr<+UTyoBcl1w>y)BxMVuwFpu4ZG3zOwfJPKh+d0z!fjO~0lZ?xK>6d<h;
z1%ohx&lrkOiuEFeaqI@i-}PU<!+4t5sf%gtR}&C$BEi5@*BjlH2F1OB7Wo&CMcij7
z3v<E*T5E`3Bp?lznpXEUw5wc2N}`8gTU$3dC>s`f2wK1UJ=OYWrVZSWs2c$-OQ7n^
z(#EH`F9njQGpH-bM@oXHrN!dyyy9Lj>WmY)u!S36YrA<jL3lIl95jPEZ@++2FA}1}
z+tS~5F~5~M8~MT1wZp>O=mz)ZV@Od*hwfWSN_fCWyTzSjH+k*u3fkA(i$Fv~G(Hgr
z5Jiy+3}DOfi>RvNM@B}T@6SY`3*DDHThV?4+wP9nqlE0whPl?E(5RLc0TN!vl@3HQ
zL+Ek~?x=#AT7PFCaw4x2ot~cF?EE|y21a`VtHJg4dX$C1W7Fj1<h;T6s3@B;?hAK;
zFk(Dr<Gsm(YO4iY4Gj%VEv?<LY(#eBA>Fx3Bjp&AGHSqoRS0usS$Nap2TAiXrl{Me
z#7bF((SHB2T0e;a->TBS>3(T5_KK>%w@6!g9Vhnpe8GP~qBKmVOa3Nzt8(Y(XzZ5*
zQWXEb1;1}_+q{#c%eycA1!()}3_FTtSoZdH6bEHXwrA8FXa^aUu^w!k{8kXuU)1ld
z=&afTG@?Mu&%i37Y#M>KWhv0NrOk*=G=9l5T9q&>ysL@+p^aHhWPeeuJ+V`b;;|-q
zF|$CO@n=A=cZ$kY07=pqVF(-xEiyPmqy-!Kk!o(ss^dzmCft{khmFtO1>9RmWBR+}
zb-cy<B1zny9<Id2#WO)iUutqW+sTf4w%jVv6-<=kaC5rp>fzzgN8)mP2W^Xqi~FLa
zgcBMX>Uq1HcfLCje6rrBpyrmzDG8`GI?`%uYd_}<9q*gK*w{Db=H`Efh76|)mH6+E
zyI#V>r^X9E+_BaerX(gN1_T7yPDpW`3^KRb?oHByuce`-U2XS;^Ss%N4<{2!1IDt&
zxLij*51N9Iz}A-K+b@O}y1n598(UkPS4_|0<{c9sllRep|Gt;5M27sB1yadtQVV1`
z35b#f?!37;sB?gyBO#rGb#*~8Yo!-GgL0qrLMYb&?0*QTI}*vMJvXc_1{;JVl*B0)
zH_*A%D`d;1QJ?V8eoFuDOV<Y=qZ)f8tso;dB$W=79+4PVwpcwWlD*xHyvhMvmz->v
z3SH{M()niET$@fiI=}TX$ebqm-sG)bNgimD;6L3iNg#(7eVwuM+5v~EHWa(q#om?m
zR-z7~9DBZ}h{GPI_p>-6nyU1D4Nm$<mncuPc(}mS+CZd1TT~6*0sYkJEC#*dynv3p
z{729&KwB0&3~@m`yXx)M^T96{oxxv>2Hhc;36i<UkKRX%4U!2gnH_F~MFn<A!^WbK
z#Nx`z%3<BqNn+X9&7jSl9%u4o$h5sKl)iue{t5|c6?CVqKR?M04Gkxow7j??{{f9n
zIxbrO043nAT?v`}BcsewT8>hOft7HH{#`tncZ%9xZ|LtjdNGUM?^ETobvlF|i`zd;
zfm`GxPWSf}t800!A0a3pZPu=K-H*Xcmt4E$KCYFcD=sn04iZ~{DK|Q@N14JCM86#%
zto5puZ$KcmFmMCA*p86lT}Z&@+&_Xz%Fw?@02jJ0D1p&Dc%}=!*2$$2aND=9giYgt
z6-p5`u>l**hC9iNofv&w<nk3IUC)?4x_=Rt=izt7P~N1mqG#$xv1F80FX+Zy$2l4;
zArlG_b=fUC48&NozZ(C>JS7+z60%Z|m-e2MvsN5wjBD8yE<&_xMhB3Rlf$Hn3EkTR
z%=Pv4%@>4sF5Hh=?~ahI)6>#MTCSFneUg)j&bEfh$;q>8dFv}HqrnX)td_GZFwRGd
z&k<-ed3&-zVJJ<Y>2egCTAG51DRN*yLXrv_!Qc1(-jmaEwmp_sp;&NaD}<e%J}e;M
zB}}99p@gI)3JAHtqnzi7j)}Q>{B{Qo{r<g@S9FY@@Zmf8hY!^qShhZeg><Ce*Lq>w
z9Lf8xV-)|gZlZ)LCY&+*;!YXE{1m>(CCeZGz2O^>ZmJeAS;B+bk$;R_HMzw2q8g|M
zG)=C~@HVeWLK~7F;sTEe$vZJ`8|5mRLN>HQ!L}Rw<sUH%*!c`hD!2;B4yKFV2a_yg
zWUb<x$KUZi`6knUkX~L##I#nEV{OQgc2ZY5QLHe-gYy;>TLS$tXVsL9J<|=owM3wE
zGO$@Ma+7)-unJzUz=1KxruByE`5Wf(@ln&`u;%$Y=1Yw@0N!!E;Uo_Jr~5O>H2#|u
zI~jos#l!WhWsj&nGT=T1V)22HpPwI}(^Az6YnpHva~rbokwL98pw?<3jZ4hblqNGX
zbJ4DkM7fV9Nvx{Q^!XMCU~Egw+kbWY3u!UKMm^?c*Xa^iWy)nTxH~?+o!Rl~V{s%*
z9WPxSd;_?Wi|^|3GNYo!t?le~h5(SEPenS`O8r{bcMcn4)uFfPUXf@V<>^$~$Z&Co
zF`PQCLQj(qzkG+Zujs~N5Hpp`bj4nH{>-YLaorB6Oql=4mg2RD{{ZP9Qa@Q!N?jEA
zW{_>PXyIaiRVFjGm4qg1Dl6E1ef_4{omh|`{u@yFwkXLbpT+uaPcGN<@&|=WAVyYV
z%hTiC!9rb>T=s~SS4?7J|NYHbBBvFhudgo(DSyK6-@g&Io9>R9_BG3;@+M9$W^`sA
z-9RgD-)oo6umg{tFw5R|hiB2bK%R1u+fl<d%vVWCerh#!Ev;cccx0HJ<rcmuflF0?
zJ3u45VGRWhO*C8X^_$Csxrt)d?^_c)ua^%O>bOjZ|JhQdIBUW!Lz$yh|E#zc4H%vo
z07)0y*3ClwWr2qHa_ZNk3P+HPjhS+A-_&EymZ18|S*c64Ejc6%5Af8?_~=aF+M4a~
z6h64QV&g5pu0~9qFSO_YOHCDJk_g+4JPJHDF;R7Rl%r#mxWptQ)i*NoX)2xlhe#V#
zCUSDJL13eS_B+AqT?o>Q4^_-y9TU)W)X2M?B<SVtf5%G?`f%;=3mWmwDlc=B;iMk&
zD1Lis$b41QsQ0HO(+|^+Ow7#aVq%*d!lI(0hoj)7vQr#hgtj)URoB#Pf_ZJbcEQ>J
zL<7yBMW1idGwcWndBravkj!B{#vLJje0j9g#IyeIX4SJL4g-pBe)(2S-!3;jZ4|;~
zWMx$E+SOoyl1oU3RtUbQw};>KfUK3Uj@rce29uA*!>HVs%yHNk@l=<_N-s*)3s31l
z)heIk7r-iGk(a&nmu4O{PS9=I{@gxuKfqg2SxKLh)+U66IO)3Mxxu-Jn}>(h@DD~1
z3PESv;|22&V9XAt>}K$iJSSq9a1y>>i;J3zOU(BXi9de)aKFk_E(#3^c}q(h3WlJR
zv!=E(vYeTHMspo^WvOAEFlS>ElNrIIznT6$4yuI>YcCT(*aT+(?UK4>{m1*8?(Xi%
z$B~JN7|;V#K%=1R=s<E2nFzl$J|e$UXN&>*zsi3^;UUPQp&jVzpCe&z!<ctE@*l}1
zSPlpq44#u5IY9yE1r*X${6>}$OmuSI|DE1Fy(hGEGNa5}mzD=fbZaF7*hJmj`EZPO
zWivgs7vT=KYu6uZxTT<`7}?T<Z7re=hN~*Nca9u%5`*O4b4=hL%q}bJVEhwy3_Ow`
zk^M!QR8HZGd(CHpNH<6y8<>!Edj5+x%wVoTHC>i1dB~7NJe!3h`1BsEu^iZ*I8c{s
zXn+u3>eb$O0SyB=xnSaW%6mx!@<?r_>uU6F^2x?89o4n!U|VsxQEw0;*TxX6v2!V&
zccgiGlRddBhn<cQf+g|chBhuZ++e4qP@q_CF%u~xcoWdxF1$6A=ntx69e(gbDLhpp
zRt$+(A0M@M=idN2{)G)iE}1uL{6(Ow<9T;j=Xo)u@;fKz=Z_!lg-S)&Tgf&F>?UXq
zTSKc&(AM=4KB(pN!|TBh9)4GWc-r|bA==(Io45<2uk1Dl2Bi(gxXLO3gQbrWVP~Uv
zEvf$4Uf{aTX^(A;$=r3y5Lr!zshqfjdt*khk}jL#{aBMRZEoryg$U+>PpXhVPbSdT
z?5z5E#7H0>{@1cdVMxaN_ZhQ9beH*Z*c_}$-HK9@6&n!+ZH@|3CsX+AL=jY%gb@{2
z%U`|-zrp!^H8Kq$%gzUW)*igizj8p^Mmb338;#YPF*)L|BYi)q)0v5nHnJeunTc%5
zG2f)^a%@j^UQy~u2<gk%a8mwAc(-*gb#SQ=i}P#wqA{=qhw(0WVm(f*{0b{dDQ?1i
zj?V-;7|$~M!_!e>BKj!g`=X?DOr$T$(~(<oilxP5NlP9!CEj=7ZO$8hMsqSq4VG}y
z@8aC=RmihXgF-0_wgH{t_f6VuJZrTJ-6E)^OxA&py!y52TgNsFU-=1i1izm;Gf6F(
zu@~|k_FZ-=>M47rvk_d9N*n4(1zX~9PwA!(C^|$Q*gQU{BV7~P`qO<v%27Xn`>X~d
zg7H$Mes|6!oiSY%uzY%ah@)~X(`pJzPuIzLFs)H%(r$ibJM&xL+Pcn~0y<X}Oe)|$
zNjJS9*3{eEOZXt}X3%qEW8aWMC6)B|w>UNeDq%NBZQ6$t+1>U_>Rd19Oh5`E7T2eo
zPb4aH?0wRMx7pQdkyKnV%<K1--_?<qWFHS0jBX&ykpBg|1GhP2&LWFJF~CpHkj34O
zPR5dw7-G7#*6wUA8N@%`NalaIRr#~I<{N-0H8ibY=?Zm`u(x@0ZDWk;%ei8RmP7cJ
z_LB!&)VI442Q!Ep>|L$iPW<S!x#4<?Y_v-E?GwXW>urKvbk1CiqgSM_W&ga>K*D?7
zA2F&wDg)R7mM79dL$1GYe`dN{iRFz|I&7(<J=)OoM&L>s{T&gJpqNYg*h+}?oi<`%
zd-A*2=MsUhQR1I=IQ9O1(s5xY7Yq!P@pjrzs$vhUy<D+=vBgGv;?fEu^^rCHGJU&g
zL}6FJiCasDP}K_Jt>yo<<;SaF2i*X%Lx27hI3KW<2@rA!)z}MZ2x<gX)=Lud5E~*7
zm*lX)ej%*a_PaF{WL=QPtU7$77v*I1S4>9FTu<1q3JvG?X65a$n<gLALohNT`z|qu
z{MQ-xUsxk}KiqiYtaM5@{y_QgS<m1J>nxO?|FOCR(lR*@i)<TayGy-YWRi2qH8IEG
zeDp*Sfd2Kck?2QCO9e?A)0$}B+gE;X7k?`k4aSFsL0i>;iFDvm98@-Q%EZ2WiDRxa
zo7jq+X>Ms5jE`s1Qs3_`h@UCxmjv!l#M0dNXtN_XIFwVSSXo*124ZQSx8=bUYt%*d
z$57Ao*9HuPhKIa;h3cnVw3-){W;B{B^dlLdGZi?tQojO_1BaK$k(G=ByAM*NCUXhy
zFSaF1HTk>vASHdB<LyWm0SN2u&BF)R#x>{Pek1cAl2d+sk}C0EQY&ZqnP8Ror#@k5
zHZX>0F=TS_x8$Sir3bCs<kuJLUT-@IHgG(BzcB@V5q|?6GQ8=YJNLS?;}wEHkDPG;
z+4w@$lRe_Kt28)zp72xO)uZulxhLiPASGH;<-GF!JEz}06s-DN$O#F`e=pzJ+KsCH
zYP;7lQ>=*(bc}4|5y#GcSg66e7QkF;wRHBZ+Ul+Fs2F(kM+k5*eZ*CbfBk7G$MAP+
zWo$V+c0*!{u>gJAMMH{o-!EomTf<dnz>F;DgCDE*ilzuHLJn@NVfz||beW*AyC#;*
z!zI-(G|t;T)!=b`tedk2GTNa(G1jX5Gx)qb%9^KZ_N)wFDW1M^hV^<6{v{a2VrzDY
zd#EGE7zxes(3N_9?ox8Ss6w5Ca%o~<T#)|(%rn^>xwyvpE+!r9%PY!Z(Vxsyfj88D
z?S&W236cHYzUu7a<GB(4U(I^`Y*dI@2Mry)ta_=e?GZds1-fFdA`peG2I36>&>83m
zbPX35H=K~;I}W`vi_4KF$Z9`b9W7l%DrL&D*l&DQF4gP@i8V-sbrM+g{FWaSdsXey
z_}!vG`tbh!dpap$0FovtBPCVhH1zS1BjhD?c&y3T7}cRV<Sl=`^xS9}6l1{Z&KxJ0
z)2d^BEcuW4zU^hnJ(&S}AZiI>t5l3L`JA8yjhvXv@$NPz|Kr3U!W5HIVQjwAIEhZ+
zM_<iHTdKyrg}JuKul-tqNcYqkdzjX_Y&+|)sl>XX{AzvUfGX!|TeyxOi9=D4T%MYm
z;=p*kv3UPdWaP25PpoH2cpkhYy3a`D@gZfcR|6KGi<r8A{4lu&FA34|otDIryGAE9
zV|&wgc^J;z<6`u4+n}c~%G>ny+4_0b6@O}ry6mfplx~z?G6{^7E*{)nAAs`wtXwmL
zZ`T<-mV-V}fZ4&iOD@HeXzVN^Gp=+EXD=AIQr4VzA^oKlT(KNiTk-=f!2@TYx|$49
zH$RRM?7cum>0%|7quO;qXJcvop3<j@x1+CuBE4l`V8&bTz>kYava^kbJw1ncBs#ju
zOKTNXi1ujULyez|MtjhQYKPMD$JD^tM!vO~hEpU-Ajui6o0{LJ8~dc36Nhxd;dc#M
zf&=^{FR#pwidiTQvIjk{AW6TkTk`65(XpPx!#-P?gztXNhut?JHmIW1|E|qLF^C7k
ziFu+UN%%m?#V0$P{B<B~qGXVCUUqr;kAwt#!N+TZ$HzP9=>@o@QQ4>sBsvV5^*=H{
z0m1rKRuQ7v?|6BWw$nT!L68T7K|l~%RtDAD_kOsTDORn3^#K#mF$gZ$^g3TAeK7O;
z`0hdIbM~05tZaKv7~Wii19fh0u4F3jMWna!P(sH*9Q}&~ZoBtj*<t<R3PKGk!0hbo
zVA_F)hffjk_yzI4i%9prMY7mWLI*6_L{q>8o#%T~VaLa|g10;5yiR)oRaNY4iQ$!%
zZ2ATUZo4`0N8F{IEh<oUwvCOASY|E@s@9Cxon1SpzNToy5Wj{T0l3`5e6|J;K8=}N
z6i*90&XLy7Hi8}fn|FKM@|mTZlqOZHOOTw;55dD0kJptEw4RHw9}m8Ui39p~Dn)yo
zvI3Sj`tSsJF*$vIgv_;_?5~_LM~=1Ge{|?T8GgTItQ**p+edS_EF)jhM}r`MWUNI?
zf&QBJ)qx~Wqv>`D78d%?F9vO2il)DcUtA^q`@N3JF4fC&mrKz~tnor?L9gG6x4Lpz
z2MtZRjyaC)WaGTG5yU&d<3I*oa-Q~A!@HTH_Mw-puZnrW$Dn4vuB5zHv)i;qxdU$U
zwpgbtP@#{~-sWiM_CK+Xl_*_oxPD_c9|cQi8(k!y(a<v#G?z$<K3$2DbH#L(&HR=}
z<;a2C(g`Hz;l8aOzqx5u;)LFw3h#>h!_Aem$f>vGq8WDiS^NQprFO55XtJdTXu!<#
z3Uv%b4$mlJ2NW^W($ZAb)j{#XPm-#uyZa>{A0Mydjx9JS*sz^OPLqEctsFxwBU+%y
zW;}#%X=(YR-!LI52@4CW6O1`83Z@T0en_VYnVE?XM8^B|^w@<ign^GA5*P^2mRQ^@
zBqk=t>j><$gX*8Sv~+6`82|vFKGJjwiDyvzC@YJxwY9Z*XJTU`Q)j(&4VDwm1sp}g
zuvfZ6u~7)vUaF|foUYsG>4|{6T`ZFLMf>!0?R<BmEEkHzac^Ir!^v6?pSLyWhMeGx
z%8bopcW-Yc@bt)MJaczd?jnLg>d9LYc?jryi&}Fby<&HcNZPMy4J@I@WM+idy^;Yk
z>Es9nL|2#R9@gC_jp*=qUmTL0Ys@JkK)Jw7Mg0p(c_-n%^s*p$=o@W{c8!jNaN9)j
zS~QB%?#3%}e%UyhIrjZBiDa`_bLWS0>D$dim<Q^RQ`z=6W~U(}kkX&#SArBHrGSv6
zwmm0zl6apqL{m-11b+j8m&AhVN!>`HZc7{G?v`gaew4X$jY3w*p*ZXV%NZ7fbmwGe
zIQ5ovVXb3lq>b`7>?i>!9OF4dl`d_(=>32ZaY|j{+5zmtbQzpVR1anyy#$>NTTa2_
zwrK_tNDH}}d_FxbC>tq^PD|n619KA@Qysx}<?q$#f@R}8IXvw!m2-8itvayVBnzGW
zJ|pwdrzPCc8UNZUKj-x%VUCk3_^wV}^FV>-$l};rRbO{r;7;<4XkA}Pr3<FIooy-2
z8J}+SgF-EOT&uN(#UMx=k53BjA@e3_Zb6_Hu)edSy07{oS2D44JO}u;e^B83Y5f3{
z5gfKhdIk#j8l)E&7rCAG6l)Fx8uesDgEBHQPETLs6y*OBS9;bQKXqchTosIUyk#pG
zo3TAb>MDEKQNIA^{}Owxg6s|_dG+hk-@ASlBNwAxL^gt_Owj`Hi#Lx0Ob&VNSp540
z{xV9eb)U98qmXp~p&$?Y+tt_5+MlGlI)i|HpKFKEZi<>ih&MF6bMf}*XM<f+)^7?U
z55!j`tER)i{=rLMPQ(8r>MMY{+S;!{BqbD(kP?vwLAsIdl#uS0?mS3}(j5W<f^>J6
zbVzr1ch|S^zW@1V?%bIxT;#XU-p^jode&N}{jL3)sb)-ADZlvnz&sT_&zADZ(AWdp
z070cuIHo6>kuCMQN=#S^`m4#%PLu9W7vm4|46;52_>6s8lI!m3;@wgq=!MvAUm(Iy
zsVMN~pEdXJSZ`kP_#Ai%A3AYPF#TeBZiSLFZf)tJ(folunqcIsx^ukjpfc~PibA3r
zgSc|WaSfU|MS>MiT{)c2xLpeUVO_)Ho*QJMA@~I8j(`g<G3JL?pC#i46vV`ARK+*M
z3!=7f_Xh)-6w^NWCI@24^?AOWFlL2rZq2bb+d#;{pW~o1>d6RXF*`j|vYhj>)EEaL
zal~KIl32L~a{kdw(l3os?~c`c5S_E<O2G4e?sp@tr$@Rmki0TkVnE2{@bww)J3twH
z@bcmVC-?H^1`AYzp8~#kJT8pr=wlez5#5IMM;!zJ%`#fg^Nbg0h`G5zTwun+bF~^(
ze>yBRIjy3F`Ro@aI+*6~PL}u@^~TyBE((B(B2R@~U0aJuNGL5V{FK-I{4E__{;*ey
z%Q7;t>*@=Y0`)(&jyoJKhp#y~IXSqvFmZ4?zytvQy}qCJ8h-H0-<t0>00vr5FiFw7
zyFPoAodAv^DlSe8ilUI&9L>SX5DPP2Z2ppdYR+r^oYe&m4laQo;)8{SJtO+|8tG!P
zzq197D@=$Y?@kCFHn~#*d7j8b*yn=pWz)qECAf;H?<~s|m2b546fkYMWhdNy*sV4~
zrku&N%ol<RsLP!{lu>?{y2&t`de}#f>FK4xD<#7J+xb!-1qzz6y-pK9G#^vXVJk)5
z8hV(vV+&+Ew_%CP`0g}pDL6)s?StUAvVG7}$)GC98U@wR+;tfZqlG>>D-b(3ViFW{
zNV^zanygiRNR2tRAZ@L2!m63W-g*MI%$#klHdg(sm_#Y}L5Z8;i~w>Cn&m=ol}hZ?
ze#$NBCw!K7P~K-^rg~;8`ufT9HAzhLKjIaR8=5(Xgx#wY<*w<74fBKn74~NjFY1#`
z#IdwBFZj4k3Rn6P;WlG61IZ^cEr!RHUbPg|u=-S+O}xIAC6UPZz-z(r50-fLO;0BR
zNHSbvAO?C>!_mgJ*4F<K6giG0V3VK{LLA{=;1FM7e=jJYU+D}pxxGAM)$W{_z)dbR
zu1G-*8EI|x0ap^Uo5LQ0F|uRt&!-1lBfk%7!vzHe|C(oZq7DAw{J8a2ged`)mX`KM
za<YYks_=6^#E;uo7BJO_=S<loVa@Z3(9i^Gs6;l&+Qz%S`QlI>?7!jyfJ4Z|gfdT&
z9C*R~NlN)%=vB;?@uHu#<*uF_4k#@&(uc41Cg-xp65U9LioN!+%-#nLBkjQKOotUW
zQcFu7YI6g7BI8ww1+!eo?EzY$82R(EwYPB?!cnP-LK52V#|zOpT)45G4<-Kmvcz#S
z=JD1_@~T3<H)58(q8!#=q5;Z+)+vk>RX(A#R4QH}3HeR6OnE3~6e%Yy!rNe&_f;?a
z4tLpRdaofDX><7GDu*)~Ct7!J-8&<Uw%@Ii!c>==kf3gyi;h6mJz&5kz={FmEx^9*
z2fR`P71G~IpA{EZwG7NfWNE2mPBzic>&aL&{HxW^pDwi+*#reZQBd#N8uEO2_;aCJ
z+^0*V^5LX1ghqR&lcsO&X%VGgjstfqJ{esaadaxTD%o2@tv=F%;MQOsUrdGo#_Rjp
z`HiNkm{{5nItP}+h-`uq9Ah5BgN)!y0xa=ryd}c1<4ITl%cYmgVyhZq_{4DXF7`}(
zLPgwI_#6oev5M+N?C&IdC~+NF4=IpG_X)cVFU!vu<dE`M&7+MZ=?dr~wzN>I&1Gd~
zH`#2T&rzuGMtSrTL9&#0NEby(aX#;54X#%95FFCR@L4z;^3l(K54!}P^D)V?Uwl6v
zB+1aW?~u9PDtLF^9sbU$G0fauv*)4{uLW=ZPs6W~?eK)ryEDT>PE1hCz(3o`Ls<31
z#Kax@ieUf5hkWNXd#ezg{MFxbPfVRg%1(jyyY0ETVf!ZJ=iWVjChGd7kExBjS80B;
z>^d=%J?3TX^SeRhWv3q$YqzXs(W%k_A>WCJTN=y1qLQ2H)}c^x=6#j9x~Dore|X0p
zD&d;?pub}*XX^yJn%&&(tx%8OLsNC~)Ap$8D4DzsX+pE3-=fefEIDMxaw0humKo2&
zF}y%f&mw06+MHYdAvo~h8C(s{<add5syY~~8f3P@bC*EYOc6t>FL%@Wx!F^W<~zoH
z7egWTO1|nHXQY6oJb@OdhHO}UBkSMv70?%7*t8Z&y$0Vs{7cB*7pp`h;p)fD4IZv@
z3hA9blZe_?SHE<o{b)yOGoN1hCPBb#zJT-?DG~&5qR;D5@5^-xB*BJb)4krzyA6q;
zsi3lhxe5^rnyXgyoF*0JK}H4^P2v{zfw>%y%>$L2&`2rL2GqE${djeyRN-MWN5UHe
zO81Alh~*!x8Vr*X55nue*TccV**zcm{9ugsA_`$-KSz5kH2>q|r{&ch3Yp~?6@(B}
z7pVQ#flvPGY;UM1t~~xNM>1jqT0GsZl6DM8>1RPX6)=7y6eo}BN>j}Kn$t_2%PEAr
zl2Os2Vg-L!YIIsik!KH0+)xzPNAbJZOrq8^P#%4otd;<`LG@iSlfMVXoted)$aSKu
zt1jiwRa2PW!Itm=b(<M)xj{v@q!@`Rgk2dM{RdkxsH2#xp1sieLM6mvL5zq>yw?a|
zc#gWI&4P%YEX5#Hqln1hOG98D1}y?*$P8=7gL{Uxr}nmhR<8cSAY|{-!lp#dj=9c0
z5TRZ`#!ejth?4<J4j*v3BOY9Xf5I+x%qaZZn@`gQzCJB_jdYgldLR#-Khy_Z3Vk;n
zvqoV`kSMqg;y<h#)iKO|--fkM`mkTX_&z3s_Xk<OytPIzlwN5BGhCsZT-u5`E_+8}
z%RML_1bFJdo$!LYA8c3D=|)-;JSJlWj#B9EuMDY?1NHIF`|Za<*DdF{quZWUwwz15
z^yE3+MEwpRtLf^OF`I_=_NOn{tL)v*U?&q5Zk`sQPkOCP^1eMF!}^HzBZ=kBi5<;u
zm*0yD4;1Q;)Z`dOOw;@?^=4n6jn+F;CGl$^)RivclQ0&|nQRwaD4d;8V;j9Jwi_la
zBnpc&zd(10Dr{4!R7}VG`3a8lMAx3nRjf>ndk@m&#O(bUmLg~cX<xa|wJXVf5N6z9
zCk6zR;zC7)Cd$UM5~s>PwCzPso}*vdqv9S}rEg3pnFOe9vuSx`m+BI!5?sG_c5Kk|
zW!9#TzTdK$Blz;WC;ok*WnQ1`Q?3~+cgxUiQsk3^jB57Qa~^(rmP^dve2dUhl5)ck
zAx}F=D)6`a$iW;m6yIgxN`wrKfgs#)rKN^LdVcDmH<O!>^K7+k_LzCYI6yj+Gc!P3
zBDXVMur*mihK`PQa>Oa{Jh+#86axTnaX$Ch-oS_UrW1~a8}Yj8u=ebb$r^6V$=@<W
z!T`0R5J-WGeF+er#j9NPn%@<EP0@Uc`WBS)<M2JCOgYU$!`iET%>eNIm*4R93#d=>
zX_5M#a&d_BlkqN<ap+ej>reSj1#2BrP-Mk~QK3RiK4jmWerhD3hE;zNq8I&opYSyE
z)pYPyP>u4R4DX~f^AO)S<{HYfUr%Lb)T5Cn<KS&Yz**|3^0dF3Xp%egI8jh(xc~0Z
ze`XKCC#id|Uh}>f<*FCFWvt3oLk_E5UfdHbOCi;qC4%1>(KV!x9G{TXl|n!vcqSO=
zTv(b@njQT|4ia;lb=<l;v0^?fT*Uj-ot6}JYR^T|SlyKqbEAM0cQkpypu!nh)mAM|
z-Et!q8691oEmC?^CH!~Gl)vzTH^)K)i+TepEWJ!P0KDO^-#8T4lAT*lOuxF90E;JJ
zRZQ`+|JEWOa)<p85?P<=`sen3lzn*QDZS-lvhkvJi%C!j^{Gv)SLk}H>Jgz<IhIjL
zWSS1(|4&i=6*U%JK$ySNX(Wu1G_v6j<G<$s0>!B8Em|ZDu7u#ODq@49kp;+hb5Pax
z9IdmZY+2#kOX_YZql<b<SZaljQDVpb0eKrcLY!4gNV%-k!&K7)p{zP;nf`LiKZS+r
zev{dr=G11LJR-U7xh$1K8To!uRQ9TXhspr6Ax%!gw(}xYu)`E$*qqNTgl81yI7=Kb
zJe*!Mes;Ae*Ob{E3~<behcu`+-<#QF_OBXeou+X8VEwq>HgTjF{EoFKspm}bk`&gv
z%PfpF{^?}HbCzNvbiTZ9nLa=KqeHW2X4vL3mc8Ee_+5rfqPK_obMU!=U1nro7o5@O
ze1uJn-~=jJb!tA5HH4LO)ym-cW%|m*^gkpv%1iDCOt<|0!Yb7mFa@>jjvuabf(r;W
zOjoD2{%_OpIzW=hyE6fR2VP8aacQdY!lJSg_ia!EG~AGxhYqhwJ~z3m2wf_NhU7lq
z9-5r1YLlE@LUgh9XV?ea&W;P*rSaLkZ96p3`)1wJkUMj8H3#;$<Rffk`@;311axn3
zHkk9L6nL7kXZ?ow>i%&RfG8Idh30dCHr8QpI>u>ldhW>C-rjz?+)OBm&9b<-+TKtM
zNM%8T5uY$8*_W?gfuf_M1Ne+!qI5uY1nX>ksFfk~%PsfrUl573y|M`z@0I($l4|Xy
z_`^PVk(3B|WW`oFO|R5)wK+p4B#v0>{`Spo50pRk4c^>^A1D3UA<{Xy-Sk&L;B;zj
zY%XgJf31*H-&93!-S6NUw+p>()N-qDLb;pA2dLQkYfm7CL1yPq3;oUzKQhMMq>a4f
zOF9=D#8%=Rnh^aI95hua^UCDvH!ifVP{tg*8cASZw6JO%z#jK4n+^LZcG~q?V~%!Z
zK@Rb*_}<zd1C-rAE@|6ZEQvfgi}|WQJNVOjG@DoupA%i^Z$W<PY$38i6IloYD*D%X
zk0-O(hXVF@qT{uRQU_!o@!h?yr4mPrQGEFc0>R1O?G3=k%^ap48-le^Tx?PWSIEhw
z92PARJXEyz;EDyH=dkHi;#+6yi6Q8PwrQ|vre5zxul3ko!H3;YW`H;mwQ+{K<{rP}
zxZL|8io$W`YEE79<l7mN;v>2O6R)`7l*PFU3tVoe-R3_cxv-NJ7OLqP8PxRjYpdN+
zKL9JeysQuSzOAV;N*Hu>^v9MGFn|N8{5mOwD&W<s+hfl%hmn{eKzjcDdFE!1XDRBG
z@fI)=I%kVrrR>z*c^tM`bOURr0dki+Q8vel2FPELj5l!yzjlg|JQskQ8G2|K`Z+9t
zJTUl9GAvkCIMW{*Udp86hzSG;Ui9G>upG)^+H)5t3yO+{!Ew<)hU4m<h`OhZY)`Up
z9MB{ss?oSC$3YV|X4~(`r?tQN&~GX3zN}GylG@Rtdep_jh2VAh_RF2Qyqhq^N`zd;
z@EazjTz0Old?l`*KO_C@3Q@coP@)EfvLeUBoYDt1!jzkm5yKnem<`sO-o?yt#KSR&
zNgz8pk_VMZi5v>h-P+)@aDm++ScUAnZ~1p@ld%w2O9vuG5;3Rl<~}K9nbHxFtSIIZ
znLhX~;!<_AQ9`1fTe-hI|6mo4lX60x8ZFOoKh9YFA{Uc95PIZ%Xel*PZqFQdXNfY^
z>#@>xt#M3gIK;SKS3aMwfw3=o!_KOA$v!wzt>u0He*6o-iJ|Z2smH`p7V~uMF0z*P
z7Y({ApcUL2JC?#><F$>=Oq3q}QQ>g<Q~OMoruYnKH(%uEze<YK+VFY=zoxZZG$ZZf
zZ&^ZquO9=++-24ae5=Rbn&izZ)ZdViO*Oc)laiAD;IMrH;vqpsYJo--1sfZVH><@s
zqsOI<L4cYCS90k2-ponr!)>5K1`R}4l?G`zr#F(BasWb^oSZy9DapY6+NPh|rN_(5
zGth=IIW*(DbLmPFYqn&}>!_via%NdR(od7siYh;Zta{*wm(PEWWxtQ;-8$BX+#0T^
zn8>7IGL)zl@_re{3#MhCmPu8!FFT2pJo3fefiQ`mB(R=vCP*OXL@VaRZiEcQLIuQT
zgNuV5X~iY^r8<rY->j$QM1iuc$EhN319{kqggdl#*vCxFSd#2Rp1g{DYk)HpC(85W
zQ%_n5QMHEnd&9g%L4m6-4FXm0VtKrxi?8a*DxE261e5&ESn3s`Zu#7kI3Dzw5j|Jt
z;#2`A{y#_bav3u=lEQ}9uDmerAeqSBJTp4HJ5~`j_Cihz&hnB#Q50(YxB|Zu_;4i@
z@1R*fz7etEttq}UJjb#0aU%Vb-%`~5O+ZAOP;t7VOA%vG?X}|hYuHCx^b(5~@9-W=
zS=Iw#xZQlgF$E<wtC*$UNkTF_a$rSvX>u}Vq7u^HWBnAJFgTraQsZ$5`J5sWeBAOJ
z6(<dm?N@BAd9p=;xVX8}b9j)_Sr-D)!_`NrXNan0wGa1Ts?j@N?_Rg%C!h0g9-w?g
z`)XnE>Nfhrjco(|<x-S$!w+}f@2PIGaiYubkn-+0Yt^4Hq_i|-VldpD>Fup}Hb+Ot
z{v45f2mz*n+znH|YZXOBC^#np-#9B>Y0`Ih#Tz%*qoKMwep!FZ3=!yWvA+czjUqLx
z>UjPEw1HZKo1<so8ze!Z_ik(!XWL_79};>4RcU6=Q=mOX28<&)BV%}WHdQRWMhBp=
z-ZL{B4E*2}78QL`?65VA5ViR7X&QRBp@R!o<j*gA9EU&}04+HDIF&+_XmU6VHO4tt
zd3wGjUKQkPb6l_%ye;TEQLl<rw0G%~(R8rbH|=Xk{;MU1o8$$Xa{LkrzVt*K;yA~b
zD35cL+(@PzVk66_y4U2ULVHx)5n??xd1F>OdKTR)Wfb8qZ&Kh+*zB)9ux$HsI#<}6
zZOd+ceFuJzQfNA&>t#ej*XC7+JSYyW_cck?#cOZLO^%nO?}_&EwRVvJ>j=1j_R5I(
zN$lqCsuv$Ipw6X*%94jTh-%<n`1b3E?>VhN3Q^#4ns@YRl6ALTwySu6<=|yc%asP`
zV31sK%NF&Ah%EaGuvvCU1gqi@CDw$^7q=rmnPEd7Vx7~J=mL^`*J=?_;R056l}Z6=
zDC6&(g0(mjE<HCmZ<`7LTL12}Ft^bKo~5Oz<L|xA{jqtEnu(sM5u&r~_u3J%p@$!z
zAS0&JdT=hSJ8|K=guc@7sq-0q7KFWlt-tokQOo$h)@Ny+`I>#6|KYaGgOic5c={ZI
zg{>lvy8+Wtd~PQ?7@*yzdH}}$n-IJ-aH}VW<ni7uKt5gvp^sq!prw+uLn$e)>KJHy
z5vI~O#>Yj+g9kpALlVbVD=fqaPx9n^+8k`-RRgP>=Bx<blopv&S)cXav*HM@s=UKh
zq8@Zh!YN8pWm$QCh==VAQSwksbcDw~jlY>|wq$RyQ%T#@;G8KtJz>2T-tE)~>a`19
z&PGv0T%wGb)eA5`1`N;$)!&V@L}?k4<dhw$-y7G8@hP2CA!1$ik>6^C{e8op-6#8+
z&Nixf24>@efuEZkku2yqyS~Z+Qpz_y{|Po5&Lo>%JM(1K^ozO&DAuI!BBBcZEp}9D
z<VQ;%T$qGlZ{n7xr$|}(KZV^$j8@YWa!PC~I$4{s+~p|SR&vSKQgQhZh`Dj?iy3oH
zPd3I)l12QBCgM@4$48ltKiUiXsQ$a7(_kN)7%)Jt;woxc*{=#aZ1B{wu-G|XD=%ls
zXs1AGy%RBh1GRx?x=-)9&5fM~Yh%6~@DZymDs?R^cLr}MwVA+k)cBZ`Sqi>!FNuIy
zh*f%(!X<#Eh5466vh|LD!HTh5pvfC*+9>mh#KZ)BtE`9YZ3H2Twu%JG!hRmK{aluZ
z4pwUeKSefn3QoFyhTN>BInZxkIu?3;5D-%I+RYDvY<53OD;US7jhP<Q9;vCzBc?iD
z>8G>)+b(_qU7N_-%w-ymfyXB&B~?Xk;|J;K8anINIv2om;?#jG3qGpLrW)rQo@eZD
zrus6?91L0$jkX(1KQ5JM<>t(WjjqO60R<I4(1U?{(rh>f-m0JHEVt#e=Sd>_@=X^s
zt=4ql&J_~d{xMF1%QFbHj^*YdEIw5UlrnV)dRAjL<|Ok2(_57W8~V*?a75vGt*!I9
zt-sZK#HQC@9T9Q{4RZbe-*tYz!M)6GO%ew1;hoHbJRgm2LI~LBOZH{R$*YW|t*1&e
zBQmf7f+AK4<trQ(tiOjc;1BF_Hbqn41zeCLIli4RT9y7k%>d^%N?oH)XC~Z`;@m6>
z2dizyO}iM9Yx&0ziUNn~E52I8#jmn+1$^d8DX{W5akfAK5lS~JYb&egZ@Idj^%P_O
z!{9`qzD}Fh!OOs;RNT04Eeo~V9m*1PuodWH2jBQ1LIQ|ji_px`rUoxz_Lo3h<A3PR
zB_5x*g<G%H@A&D#cZNpsI;TBTpx5`eTv=XzR3!C6&{~XYPh62^jbbC=ix+ykjPPKD
z5E$()MUtOvNzJ806(VV$y%s#7Ci#`ReY_vGYMnJHUh?@qd4XteB|a_3@eTSGhOF_h
zAEeQ5+1!Htjm4$`J9IVDlREj#nql!WaxOUWPO<{)fOTu~NlgNEqCt&(7Q6`kPP)!J
zpK>N1b=HpQ%<C8R6B(GFj(P}DhV5E}5%ip%@4uEAr-&1%#2k3}lQXtAffbRn>WIqs
z-WnCuts*TC$Jal_7&{+k(GW}1QZ4qlxuDl$lW1Q}wU&Cp;|B7nH10?^bE5e1(sd^X
z($yT(*)-1eafI=cm;d7x{Q$4n;C80p><#;Qz8*hDQUsU~Ryslqz=(szY8Ed+T2x>E
z4R9ss0nOjqj)<t}@Nc=)<An#{g*#qb0`3WJX|P|GNXes;{%%K{LywJXEP310g=xof
zI(&!6KhrN|N^2R)9;HiVOuB5?_;sBF217qJ{zx>Mc)q3~S=U~>PI&4?bdNuHdxbOL
z6+*MOFe@qWy-PlwJ=MEOQO}Tm=lWd@`iHz0qFgg1T<27yKY8|lsmA{Pd7*t+;PRwI
zY%Wb6jW%*9UtAArbN}8Y(evvfI^f8&)GY3GKA|NQ-w(LB?`b(P_sP+wx-v8{1qc;I
z#X&I{X!mpK5);k*Ta3SsPt-CAHmIr*HpCzIOt3`R*}*lsfc3Fg*r+p7t56R(xL|ZO
zU15RGWQa6VWsRr{X1H|P_0_|Uway0vK)Jy9zJ~Q0%8*H7`zK2PEgwF2U*J6T&D9x^
zpl_OxSdc67I*PsYnkXcN<ntHCF^Tc(N7{>wYucA?90K7fj^u+cn>$z@%fF|q?%Y#q
zB!tP^Ehg%c4NK(I3YbB2QSE@q7L0<$n_lewF2kwbhqWu^16@5;*W>rgA!(YEZkTha
z&_rfL3t#Ltyz6U%b0dQlbGe(}Zpu?EEBZ`%3zs62`;C)$#20?K`+1Y`t@a^l1#lji
zo%gBiORGDglLlI6-%HTr)#HcmV#Fqouk0@GnLKcxuDb<$kHa=;`t!UjK&a|0@K?>0
zac?}zb|9jY_B%&kEHu$tCQ7j3!9}#jg<6S=9t<cS$Mrt1mGWZ;N0j@)lmuYsQp*{(
z*0wfUOI$Gj0ykz`3a@){dHo4L{N2rk$!b>w(1;v|wgV#3=QUnW$5;yB2_(YkAwooC
zx6$l+rMCqcLm*ALOCw4KSrE?F?7>x*ZVv9;n31gbdUaoBo#HT-zuXymO-mY+yiib3
zT4Ui&3AA-ce_ma_#!>7wCpJ{#^KeA=79ipEPcT}G3fU^>8gFWVF|x(yStAd@Dct&b
z#mIg66F<Id@^mN_%ofIZI=Duyq~!pS1bUFhZ-|+d#<bhsnG<>+xm08;Pwxg-l-Vv&
z%>)7UCiLYM-`6kFMo_;mPs%f91~TS|qb^E=h@_yT#s7K^H05L?StBNV&+F5r%cGUm
zjScV?*fvmLr|O&;aNlXZW?%>d&B%M*W*9@azG=G0CG7aCr^zrKx{uCD<Fdl@SvHOW
zD>qmaF==?I|0Mi(`rJoeWJq!JEAA`2<WQo8{D@deh&94|WV=lm;o%4o2m#^_$T!qQ
zWBT1eFF}9pCtyb{&Gqc2H1n;Yk&^nS2vlnwZ$-2Q3JkJhog|tnTfH5L$U*HdFcvP;
zVoi6jrUn>oa|&^}liU^tXL|>*H8s>3US9A(=d=|s6B66Qszfe@%>1skRT#`)L4iYY
zB8JA~z@?U~z_K@tSr{w6v1ubawvjFU40_w*Uh;Ez7<}o{xyXeN?LYjKrdLf_wQOl=
zX*QG&Pp@8Hjhlo8j-TIlw;|7QH^=8W!bxTexW0N++K;b?MP8k4=)WTuV2ZKfB5^YO
zphyh^esyq(2%Hf#y;gb+60Am{O^Zbnc-|it6G095sP<xYt6T~@7+6Qs;gU%`s}#XM
zSikhA(nnFC->}F^wx8R@_V95eNs+CM^ezggoOVHO_!M?n2p&8EP6lo&1Or>jH<|vw
z8!Y(jE*4TwebCH)-0d*J@wq&yalH!@t#Gtoyc#<w^;rHKnNir>PPMrUA!#KyFQ_NN
zl!~JTWhyE<sZvCnl{l7}oDN1EqM9jY?8V3TxE@bTOkK9<59D(qk#q<?19w(MY0N-`
zB%+zoaP}NgLU^XI6@^LH#^7@x+e;=!{BN;ZY|b9`R36LIc2J+NX#0-P(GF%iP+3H{
z97ihUm12g>z(dck##g|gswq$bMX+CM_9Z#JFvkQRzMK6W`9#o+YgW%w265zea0o<B
zw9|GkomkD@OFY+6GP^P6)fjQ#MtTMBrb-ZyD=g94f>#_V#EsbCobe->y1Ycp^=%-n
zn8DG(IHTyV=~_-kudNwEozkR+GIm8fMnv>=rAZYl3h4LUWg~#ENVH-E4V7f<>}x{|
zshJ|Fj80xvv3--+IXuGTPi`vGE~1yJFWomQ{x|VU^ZGYJ@HzZzG5gN@sP|tS+2aI>
z7Yey$q3+h7=z0P8yu=IiG=*o@&yD>35a?F)OUr*Aw10URZ%3h9Gt`!ZO)eCLCOiT4
zem|!X0lJtHZrLbe_vF?-p4BWnpkg>ev5*AU%8$EQ3qny@eS}vruYC4eMw>6RX*D<^
zh1N4iK~DuweKmbF#aB)==j$hgTU~uaR}++gZG@|<XX2k+M<!EO{`RMdMFUoyb9grr
zpcQc%gaFV=0KY6ynMy-?#(Z}6A3tmD9a-(jeeYrfGtgJohyYGJZfh<|%=*V_00Vme
z`C-J1SjCXRQYN!jDu2PjS6M#tVJOTenedd%ZCsHOp4o+Q>2&YfW^0YHD!Qc&UHq??
zUT&LAo?A0c-_MSV>4~N+d5*y{OWA`}$#gyV%QY7&Z3X$rxqw%F!8+Fd-z;@Lr|fBw
zXoem@&Yx$iBY-F>GBVOfR&^{GJ@Pv)!??2e<8So$XSS3ww&7|zje`D)eqg3A8WYw^
z*mIRQ8i}TO>HRZ0!UioH%YA<{xNHG+3ofEecyiE|kTSn-8SiiGK<+|J&ROnoRW^sv
z4t7OJ&eV#=M`w3KKcU)XV0`p{5&7r88tC)ad$gPVyt<mh?R-~4Qc_aOf#I>vT@*yl
z3WhRFJ&xCaSkbQ2LM4>#>FH;g$*=E#n9%gEQ}mzA+mj*8cl&%&wgYn&Yz4`CNV^20
z4;U%+){zZcB@>(_9t>c91cf{E?u1T&M{M>2U-B{@iO`dHzO&bI&T2GA2Aq5DJo-$W
zQsqya)34d{ns(YMBO1!X)s0<D%oIu1I{)4M;6V{T7k%hR0Sb(nN=s^*TGpD!RD-3Z
zrRDdkk6<S)b7Y(2c18uLYsA&`a!$o4nGcI${E-dwsGvp_VKB!7N^s$kLIIxRtx0uq
zTIYvL+?x$6K_I|`iX(l)$S1o<b<iX-v5nU3^{2Ffq*wePr)}%jgP&}_z&2+6W!Y7k
zk#GS&-Iy=@VI>T%{og0}Mxa->v+<${#I^c;_mMZ~$2ndW4#tAGOzI(wY{4u>r+r$|
z$31R{D;eXJRs(;>!Qz?^=r;Y2T*LF8Ai@o7m1x3vwElS;^xu<c_5HJnJU>-t5=ZF_
z_~;&>bFgtPx-xwA_~pZ&3Nch_>|m?~Fom=sV;5RLt#4|K(#~m?AS+?l7yhVfkJpJ;
zZTu#yyFn~llj+6;!{%x{_lTJ2kaoBiMpd9f%HQez0pwgIb&eQ}-ogw{)n+hPhy<!A
zG7}lbX+DC-AU`RcM3M6thhk>`p_IQf@IU$wKF2E}XbcY(N~{5mx;T$x5Kxk}c8oQ8
zJt?B>cMEt8@a#+y8UqVLlDgN}4L6!MPK2WdMkN;$7C!x(Hja|d7|?ALx}4isVa79m
z&;)dL6joom<FSkXS3m`_O(;6{ZGhW%tv64l0JB@`-_y(**#;V@kX0)(_&zf<Yc~I^
z2g!Y(=-{0LlqUZcMrX<6Z=bGP<$%<C+*G4i=T1#>pGJDI!I&|Fyu74|x#pR!@e?A#
z#X1$n3Ti06r@Hv+DZ%{R90#}MTQIez^ICVOu9+Ev#R+H;AFKm!;v+_J8AtSb=Kp~J
z+#1EJS0oBCcuPa~H(x8t=+nq*CE1_hP>vNwF<;qGA%eRWh?-ceyes~0)$nz#QWW#x
zmquSZ{~T(m&x3LBw1w#cqXPAR)g^$o?kD68uTlAadj#K4Cf;vjvxEgYhj)|pOv1o@
zx4!!4@)}*?IjZsLQ)^94%_}=F{|lLGcm!Ots^|}uG0!xt-a4DWO9jns$M-D8P%sV<
zDCb13VHs!8<ZzWZ>UUcK@frd%zQ-8mn;&@#p{Sb&2&v^TSoptH_9%^eyAyX+pt+H)
zPt0qo#~({GJ+G=o`b9Hp<J+nPo`06w_7`UJEdRA{V5#fY_+BZ|!OD_mw`bIPt(@sn
zI`Oa)2!+tn`~1sNx%X0nz4g&Nun~>iq^q;wyc!TTp4nLJi%fP|IpX|tAt0eiNhCzd
z-G34sv0ZVsT@TBjg1qgKRlA_R-W5AZnw!DojqjPgTZ~&W2=ynh3UOH5+=c?#tQnLp
z4UhhWHM^RUI^4&Sd?RWrOX_Z79o@s^F)R3Roo3LGmH}5YQN~egDz>&)u3qWcAgS(^
ziTreGyDHfUeRxxq==KhRHmj_Nf@WTd04Gq6(;*AAg*P5ugM16=*cY<{36k$GQJuWH
z4+O8l0~GamNM}(z>y;{|5Rr6H*$^bHw<SJS^5cIY(3O$A%`=|3&F%SCJI(942GYkt
zA!9VrW10%a6Q+ELme%=i3ijG9!d9n6jsFfV%ZDKRgwyzAoK!k(L9>KN+ZKAYo5|;q
z6b&Di%2m5lAfAYomCi#oVcf*YnUcFzYOV&h1d*BC%Yi$X?db|yplbr2tTAUQpD)4p
z^cP2+Bn&$5sHhm7E*^ip%>qY>Ll{fiG|WU|$m|_E{$VLYd>6;%POL^$)?M!qzjFp1
zjW2}?<KF_O7QS_KE0k}gD6iV7<^dZh^=$A)oE)M&HL^yKlQ>hyxUHS~8Zk$bc!8of
zsk#!8l-N9*ci6$n5&3KzGIDclUemwa$E<fHE{*X+J?QhsNHvwuL<K;S=cv~Lh(-Lt
ztZ5&WPSWQTMbZY_49shJAcV?a27CIQ@mJQ1s~Wu*0gqc3=_025ho+l!N=MG6lmm2r
zyx4bQaOEgZf^f~=@}~zA3y80nKl*4TCyc-Uk&WAN4Auq&dkuQt@HXAv$$1{M@@A;y
zDdZ@f5e4Wz6-_nVi&52H0*3vky1;l2nv&tkq<_cjMiAUi!>#gg+e7Xeg7HM;lqc0r
zZyH;XCkFEQfXZ??Ye9Shnn_3y584!wAk-*LYQDF$ASwF-)<~SkRU)@yW-8~L=*-#Q
zRW)eA6DLG6WVUZp$Ng1#_)@a%u|+$WG)55H6Y+h8rLA!wYi>dPHU?M(re8zFI8G|3
zi9EsQa%Z3AyXTSsC*gV>W(Lq3TIc<w|9RNkOEQU9YjfhHF%V_9q9pPTLC`K>-B9m8
z6FLziXo~&YI<xHHV~0V<oL~o`TJI0GlHMsOojBCgXbzNzF8Nj0FzTO)&d2k{sWgUc
z;mmFmi_jwlFkd5SlaP=A{s*yHilG5;&jBUuq>syvmQt%Eay&)5Xo%odf!E+(hmH=T
z$qjjxw8k-GDQ5bgqX$@SbX=Pau_lcc6uwRQikK`*L!>93GNg!3>I9r|k;LwErblyf
z^=*mA$JZG6z><`BhT5WJ@MH+yhLJ#aLnr_fU_Lh);*K3|lfPb{ix3iK8@iSB9+l)L
z+v9RuER76+b_)OWq|<&$9=-jgo}MKnu_MNPbV@RUchbZzZ1;~kB!2U~a&J-o6u$LU
z&GF2l*BAwu6b=r?kz5Od*en;%bItPF#L;R@zkrraAvHxs#m&EIaKPHcdC+h{4EQc!
zE@&iWV#0KHcQ+in{oT^m);3A{sIX||RSl1zrLWe&i~aQ!=vY0o)*zM3(pn{Q%Z!ND
z;yss7FhA*X7E3PNvtQqJc)8(K^V{r9W^OEhDO$Rl3pBLD@(FZ!3&g-IlC1DsH9%_e
zv$&)g&`bUWMu2W*2YR0C2JUY_?E8zVF1$5ZlSLAZ&FXxzcA5;U!s!uHE~RA*nO~h^
znEA|++?=-$D?ty@6DN8tw_a?C260%QoEKk{)jF~|#g7e}vVa-$H~42bHm92kJv}{#
z7Z)ag^?3}T0P_Wv+Wi{)jYs`s61z3UV{!=ufoO$pi=u{rsF8xCc@UwI@fh^9hF@P_
z52%i_J2Q|I#B<z5w$`-czy<sp*6Rbwy-<rU;*cqe!}e$pu8v2<3o_B!l0wp(bj6%W
zFi!)QuKq-!mS&R|5iTw+7_zypgs>x_p%sk1_r$~FZQ0q`aTmJ_9UfNH2EO@MMMYT7
z;*Uid(K~mw<E=c7s5lGtuL#84VM&VE;jQM;U*~K1Jym3Xa>q>3b~-<--Izhq29#$C
zUVOxM0EGkA+vd<I=RiIfkhs%RI2I<nlsJ!c<<q4v_H=Ba?ZMKuF7CraWlrS7hRxBT
zOFT7JSb69BO&iGCA)-P+&5MXmSiN!p;<?s#r^-A)CX1nw(Og76JartK<ur@&zz^5X
zg~3$*!K;&vtL5(u95&D*3|JOgXzzSBSu{f+V`z40K}Iq&5{v-K+%NXIujgwWX*H^>
zGPUZQz>h?v&Erm0&+l&p29+4KGUK_Sdq6SKeJb__4+`XBpk~RXdQI43ZT}SZ{a5UC
zSIimc4e?1!;_1)yR`hk2dMUE(<jvB=JvW{isdL0ba1S!x7zTRBu9r=pfLAAU&F5Gr
zXqIqGDkY~?y6Dj175zkLXW7aOYyOW@=4Yw5iR28oU0?{f0R~~wt1q|4QbWp;%fpTQ
zkV#52`M|se^+4*e{5=&gEv`%nr$8YlTr`pA_e^5x(0s>maG*V1X-NQ<=`|JA*Ue2+
zC|Z(Sy&x7mK`=cBDK+N)avg(-tltxuO)#;s+tY*sK>&(C-dJQ@TxY%Oi9AgVP+ZG@
z`h@fW7;=WT9M+H*R#y!jxU{YY7h8OLZ@hqs_;C`uzrRn&=Yba^>91a}4on%BS68;T
zm$txDW)SUpbOI{=@!GGftaL~*r{_X%uGaCBhQ<l-7(d6wHR%(&Z)Utknuko*|D72<
z!!_Ly!2bOS5|PhFAb`8?Zg}na!rp8;d)bmx{bP6vWrfATYjUTFU(MePsfLxR(RUPE
z=CMKlPa5(|Vm$pBOmU7LlhW@oTd$G#PM%GnKP+BO%sx6b?U*6XCAsroH?(1+roS@3
zKZMxH;G&pgi><-h&6F_frgJ|o<nyO#Ep@T6=pSftO?<CQ-m_<vB-0;p)?i0R>sq)T
z^r#VWJE3}m<2bL-_#sXHZ>f<~<K6j8-8eNJogr{9aBy?mgU?Q9{e5C!AQT+47a-IF
z7*Etn4Ur*+hEv6>9d;>NB^7;AdioHsGvhkf_x7~z)qxiUgcx@A_lKPAOkk0a$bI<W
zEpWr+dc0b951jIy8DV@qfS3~kDJ=2q*4q9%e7isuVY$%QP*V4JYW!}(q(wgrj)Ugr
zDWcuaCX2mm1+z@#5^wLX<1g9Frdhu-TR<2s^K^&h;i6mGv=NeQ4)#6+AhsKx&5lZ5
zNU#KZWQY(L)+$X}JL!L&eQEZx%tey|wmLAo4VtqRy%wU?*xY|g|G9YxRgxuK6G6=G
z?{4?<W;)T28JB2<GMzNrI&YerbENY>(|<zK!9Y3ZDXu*)`lRJD3XpGx=`Lu1z`OtT
z+K7ryUmOe;x`$gA8-LKzzeX%QJ~=S+xmd4PQUI?}2Wd*8JG;{r&k9s}XRMn%fpN$Q
zL`C2MkJ&0PG}e-Vyi)KSU=G`{<Yptlz0=(t1cFnN51JlQ0L6O*_-}%82IVbqRltCf
ziGw_k@)Z-!2Lh*5UaVErAp+-K`iUdGMc+d*`V`HRH$Jp7isvqtNL_}-^jH+hF_Cqa
zdwO0-9+QOEwSx8*)K@dD2nf|Jzv&k@o+Fj|U%rXQgGaOv2b0v}Wp5fGMVz=2n?As_
zOpH8fG)G>j@_9k5m=6Bqwjk5*fCaj2Km#%djylL&WTlbZ>!60_CN7~1iD+Ccm;P2v
zaW9mh`CzzqZR5ZFC*Y#XIW-LvuBB*hqr-k<Q(0GK)F>K^_az|#A9xPZffecTvS@D?
zE}!n%0B=50bh<v>0%``;OMB+2S42bwsytv3VNeM<iig<%qY3+Hd_p+r?T7Gw%Or_{
zoyNw-wsxayX^HkhS~^(!=CJrVvi9R&o=qmh1-}fm+lsxq^U8PJ)#^{0wBUYTjJ}9r
zvp=={Q>6uC>Zm9YKpL`m^5L-1xLg2O*|Nw_oxXVFcv*jMv@EtO&+#Bi4n4EJ6E#0&
zrB2|c#hX@Q<{IyomOO2QhMyLww{5d~k<yY@^eERh^1%2Pkb|aB%r1Ao=?%fiqRTsU
z?E%2<2tAzrZ>QXRE2>i(*wqw`6}ZEuiqY)v?HuY@PYIoeHxvFWDEQ=Y90FRUM-O91
z2tnLLs*IdmlST5V(w2LLN}fojqVlD^{ga9rO2xb&D4%ZYQ*E^ppTI%C#4t#<M4OKT
zQoVTK0y&y<UU>2HrHFz;L<?AtYD(n!$Vf~~EUX%jE2cOGokrZGf2ZvF_ek111<8@4
zLMi`05{g=riPegz2Y)`~I0V+QG*$Pk7xvfx@(&GV?8aDimTQzJWyYeudi5qm7PLsp
zkkQGr8QDi7*YuZnn7VfX@mNYgO`Et}4~o>)#!99#TZKch-XZ16uN77@V%gO(4{mN|
zz4Mn~63TOZX5ZBxq8%amzk08GT@lk0Y#Gc>cTmo{^oph+)@yJ26&YQFtE+2?>ChX%
z-vB2RcA3c_1<)xlG~Hs#$w7m4$L1i?3<k_VmP3PvX|8MBl$!)7*`uo1u^L`>_xHm$
zPhIU#cV6#GX=)Na<}3o2aK2Vu?CI(0anLx>%}8dxpeXY49#ivC{XJ4*K1So?>swM<
znr^$&0em|^w+e~XfpC)uV6|ZOymlBhcO-B4hsN$Gjb|XUVg0D)wkOyWiar-@c;a+9
z9^^4N<5{f)8=yj{OJ<lk_iJiw(ZHM4{7|ovo9<Z6QWR7p$zqBh#$IItO>oF?y<}uJ
z7g3#YVN@*Tl>`U|wWtT_a*vEho>*QAY}MG}`6*DGQjIs)y&pDRZ>4FZ1)`b2-a{iE
zQ$6U=WaV)0ujT{Z|7|YWn`RsqC;PR(XV~Zz<P*6V1+tyaV!ssO!Y($A`wmZyflCm?
zz!lXWqx_al)*sJTGi7=8m{iC;?H?EjS8Tf?n$h{Uwt@w*3uujKEz9gTWMQVk86XZz
z-0cN{EwGosJB!cu8LzPLVUT9tGf+&~7=!R(K)UA_d*t&ce&K&HQPU(9!e70JNvGQR
zhQjuA=l;Xrinu7m`!|rp>mP$Bt4qe~4I3%k%x@Ez4|t2%!P(7L>PbqXRW1DF#?`Ox
zi|z9L+c!KG)91hiLrFuE+ff4CvLMnSkkfu0z{#NiaWN1^WQs`%&Hw()GmUdUp`375
z%kYet8=*|LEw{J$28FQCr{C<b$)52D$-_fi?bj|W<kCb7l@Hq;+3={v<=(H}o>~Vo
zZlW>{i<t%LhN`IJ?TScvy(pOpQ{{Ynbt*Ev!XQ+Pb!y9gz`;gSt&>3Rk-mGB)6#6_
z%tT?w^hyIQU)%>Vj0%1>B&V4F*|~A+AoZa!Dv!rgwk-XF<z_gk6!P2oth)@&$Jiu6
z7UBxYKNeWf19G!{l5=>`$QXCWZ!uXL4$hdvpXb5*3xm_t>V}7lE#6EcXXa$0!7#d8
z!<p@!ofEWKb#=$=o*)D+<^qF=tDMIPoP*4pva&L@VqI964AGF>6WctlxKhH;mU&+M
zCy%j-ocAHw0x0!Z6w<l?{(2w(+M6tatEk-;f?^szEf$$Kfb<*N24;Mj_PuY4>aS0o
zRIlf}#?dIMl&4=24yaroH{xgX$tDwE5GcO>_&fOFk^<x>9w*hN+D$3XhGHob(^`pb
zr5jbbsK+{N-IvVl*!;L2z-&~9FysX^Sv#-lic9WX+(x(^2qs7zktLr6z?&?^FdwS3
zKprmWPC6~3nM><|fyaHZ&d9{0V?Iy8>-azx>zV0tG3o3uYc!VGmb7|P&>PRWn+mI7
z{=rkry5b?z%fqeT@_BSuz7sqLjW_Rl;}|I4zc0C^an2YL2USy!=jbU+TOdYJaadg)
zHz*PaKLPH|*Q|*I4$9*ju%S3SuOSxj1>?aK-k0ZS7#N6)g{!i7|0fsNDQXf`6%m*n
z91KGw@p$gd=Xq@d99YM($RJ<#bZ-VXSyl`{bRuq*db`5L4KSA`fq8Uz18G$+u0d~C
zmo7L<eaYNa17Ry6UE4BS=tNxY0A0i4aoD`psQfHYpppR%ndWLzA8g|`gR_;u=UH3v
z5fgm$S6VGGnv%!gqgx9rPlaUNS*1<kSMFZZ;L5HGiUtjvOtj%lb5JoG@ZO*2I3gPT
zne6i`rwMZ;C#`7ZD1)D{`|&L2Wk*8)98UdSx5PlZe0tHxFim!dL3A*oeXmINwU;V3
zwFzdoOg|)=sj{@ZCz@C6RDM9>6Efb1-v|DFkhfX;TOm*FYFp|QI@ihn&f9x<pmK7R
zt><uhn(dUT@B8KI2IFMBup!?2&mg~0Q~0}(xKm{Sxm|&fA-92_gQL2qcIr{<psxPv
zv5mc0fS`&S>-qYc)(eow!Wa<|Q4cUB^!P7OZ${05sxhBBHKNY28xzP}r>d+;0Eo`R
zW@cfD2I07B9`E1~P<4R`ad)<+tj7rh*eMNxmU*nBZb;1&;Q)Y8g+^ZJ{?^=NKi1ai
z-Xe1!3>v^IJitdumsmB4_A=ks7HLt)P{d(%Zzf4W_QMxA{G{;}IxN(jZJ-g=c9T8p
zEdA)~+fn*UDl95h;mj7LaMVWjj4xJkw7)4*qqq6nZZZ)P{?zvthTx_B2fjY71HN|4
z!LY2Aa4W^xm;V;+vbwJ9<j(bUd-4lhT-4V6l^uoUWP9yondU&Rm^DkJmS$^2G>h?r
zt9PMP?9qB*!z45{UFxvzKumx?WLN`%kg0r~rWA=NiWk5NE}GFYH<zr}7Et1_rI?<c
zZey3rGbIS#LJ9n#K&K5tHxbz99I#%6Ab?)h?QHuGD2P){5Wi3WYEH3^xHyvg`R>ww
z-9blN8=P;8x9smW!)ps8Q-C+Y<ogqd1qD?FYVPhv27h8%HP>;x*30R%`^kSgrDo%?
zUj~gLSmn1YEZ7z5Ax%yEVLZo30Kad5#KiJtvboz&o(R@z_D1r#mqtos(w0>vSG4}+
z<+&!P$Y9Y!$TyB(>O=gp(^I`a6_<!6Bhh`^Y|ip@Zs6Q;5sk>2_T&7O^H`vO@v)Mg
zx{7z`*c&)B&&sFN52q}*Y<IPC`9hQJr#O|%{;H`XYts185piU<^rAl@it%w`?HrEf
zm7RF)SrXd&8>_26I<!$ZEU$>$i)dzXv~?`E^F6-IuRr7w_MB1avrh<lf7n{jJsGT=
z|NAr!;opm<G!N8WT}$s8IO(8I4Y4(}&|+P-Y2){Ag-*9bo^B@#VkD&HS|3)()!@O>
zWGiHBj}4xi11a5WCZ@=)t}d>Fks_U@W7?xUK-@jb>_EUofxV{vz@LtefSixNnH;xA
zX%3od59w%1-M(CaOu4n0^Yinw>8k4L$)2gl<K?K!QtEgwM+j+`M78eT_FObI5`x%#
zH$EEzp_DoT^BOJyQ)XVqh=x$-AO%y%YshEUDIs}V^yV_q+gol4I1Rtdte!61Uj956
z`cY}e7Dk-TXWy#FspQ2td0VUlhpN0Qr3pc1;Gt*1>*l2*&tb`8=W+dwU#$EK2$3$Q
zbdc9Mfh~8&U+AUE{Y*PQUHIV<+@{w7eYRy^Dr>SQweOsi*IHjz+iIqA3hUm`cv7MB
zdD_D23)49|7wv!5+3JHjH@C42-e5HMf$m+S=jT~x?@1|%UxvoZ%Zp!A<0#?UYMHVi
z{?=(kHpg1QPwm_A+Ro0-AaoIy5f62Z*&)wjOkNRERBR-A{<l0e8mvn7(Obf=Ycrkv
z^4DuVr7+Vng^QNIJyKNwy++m7vk@jjz)|c#`*3^9seXY}_`a}XJfM=-r^e<r!oaG)
z*Y8&n`@K(oYAwhr&(9TFo4g|qRqaqyT%3zT{-E15pgd1BS;Xufdy8_zy!1|IG8O)0
zWl0j+@<A;q|2w5-wj^&TkJ6z~Q_P80`Ku+lk&zrGbp1n6M6-a*$V@)hpBn$$+a5((
zOP=gm-J$^P{lCN7W?v!{o+#94a(RVHbq*eFm?YsbA1J5?RuxUyf?RuU+`^5pFm~O)
z_|a;$r0*m#j7`v1H%a+Y`^wCA)yu!YEqpeB+Iugsyav4yI4r=f5d6B{KQ|SG7>K^q
zk|tzMLf7utvkgBFa?DGN`(J%9t6ALGP+hsqGr2{_)4gMV2f{{qt~WE{r73{7SFQ4M
z!dMviDa+RmrhjDF8t>64G9;*jWz93_?9MM@pCCjj-rgR+`Gbpcbxaa5d$!-^w*YCJ
z925+^;EjHAux)tqtbWM#dCYeUDGYT-OVu-5(No=tbhKhV_>KyeA{n7CuwF87G2dj{
zTJs2p_t#}jwTP&vQ_)f>n|t0e!dAJo`WD21*Mp>H|FHHBp+3~Yy#4Xy2wpb>b(^RD
zco|uS{6-O>m*kFQ>nJW}91}~U?Jo`_8F67wv8eRU!!v?b?Bmt-7jaQB*{q#p&us@n
zaK1@WAEGzu#iVfOPE=FMa<5-ezHIsS_M4ZK!CZckct~UW*SHR;8HC5R1Htw1DmsRi
zz-G2K18({P$6Lx&55-1akjs*{uFsV4H8@zfXu^OIj}b5bFQNkY{xTcL%bJBihn9YS
ze(%|LD@Ao=r7;GT44?ayLtDJy<3ENhM<-296if1&A)TAH3-ppr<H`6v`D53^NxDO>
zFjmQID9Wss#J7Som^|6Ezp&dbB+mHJn}>zF#%)KmJP%EGwAP-7B%|Gtguz5faE2DV
z(%y$Da`Zb_#&(AN!NS-Gz4Um6K=dPedGm;nXWC`ttg!Z_2!T4loeR82X^OLZTL@CE
zi|DC6f3)NW-jBks`!z?ypt_Oq%RKxzxye1)QW-GRzoYGqUbdp8_i(_Hw)%LN?pRQM
zSai-<$ILNPe9_y#t9_gV+%Er?$j{U(UwaYo3UulP!*T~xkVz^zH@60-{7d>*AeFKW
z)GFA+!$V!bK&$i{0jwx2BJv&tnB;YU8_w2~c{{}8q8(8(>i#|b(pK}aS|)3$LQB~f
zn>fYQJE!9(jIZ*Dtafw<4-+nE7V5F!g8UujhM#)apv!IyG5w;SJ7~!4SZXr~?G&|m
z){D6mMhiKD<+9ExzGgaqhmLx$+W(Bwfp}g^jYAo}e)@&gN$$*h^~sj{@MFIE`?)9d
z+d399O$|4^Iq&0sCehXxV1Km8eA02%V*dpXO;m;VCaCnU7FuY*Rn!XqT~6O~uY;n_
zbd|X|U0K}Xk7bVY=9pnA1v6g%?rOoW1U_TWpn6$Z*+;pY{5089M8p%2@R>2P)zb;6
z(Lj*MktHY<0MesI-4!YmSJ&1`bX%SPp<Tz}Qd<Fz>LWt69)VQTY3aV`Qo59;+$|_K
zf86`oKt>M#=kvGQl8z-kgX^J<G*U^OF_Vm*o7r(H4U8hOwxS#kw_2Nr&JlBZD8S9x
z^v~v0RCMd%Oi2%_h7f&T-?q%*n25^(9ZkhIcO0HCbx=S23_s38I(El_1w4D*gx5P&
z`@hWu$<3l7r{fu`zmwkQP4$4bljC~DDl_{=va|N(b&*pb2CA6=fRX1ODtEi;;4CdH
zJV;!vujyuXX^?;|cmIb13(!@NlCi?SSNI^2K1GvXgvaA`b8y}9KRtB-;Smui@bj>6
z2!}&rBr#AEfM$TB(HkSgq5I?wA_)M(_G20m2bG)3@wAgaIz((nWiB#}jg2OL7RQ@x
z;UM491=P2ZgP|AUzXMe>^v2C5J>gM`%!v%24B$|~PXvZR&GXVca2_tJGx*gN$N3-C
z(=yzc&l@`bPMSv(*T@kDhU7r>J}MLqPscNxS2IA+)h^*)jrEd}5291SdoKZFV#@J7
z<!@<Nw>!}v$?w&x+Yq(%ki}H)^t4fFr<*<}4QCk$(%LZFoB2pse#lsEW^XbWRh>iM
ze!Sb>nzTaw#654k*73IB3Mw=?o(ml(2%qRbUeI8VBogH@l&Y);QKskubzuGNaqlAm
zuN;ccZbb;Z5t*Jh*9Cl9Nub3jgC4vd2L{JZFTnrz^n-dIIRNi5v9P?eeHf!NL_D9V
z2k73cXrgF;X&ggWHXB7vWlb5M^POaZOkO4gjx5?#G7(I+scfCOi6H*zI6mT~K-KR$
zE#au<9HT8r+Hb@^(B=2?bcaVqG|&^+TDyJ6)}tD#P0bRAQp*|*vUKIX|4hUcf&U8j
zHg&-Ka?a^EtEwMHz7(nP&U>o_yTB`3IwK|!)fa!vgIZL3V3T|~EyS++4c!EavSEyM
z^5j;1h1qZxU;09?#g>HzZyQZV=dJOMg1VELffwEXifh@BF1)rk7&{byo?_yAxU~l|
zo@$3J%>T#PSB6#9b<rMBKtaHuB*g{<6hRskM5QGJX%y-15;ll{w35=@E!`y`Ee+D$
z-F@fL_q#vt&+GGm9z2}A*V=2%HDk;%uw5{qaS*cbUM_-E_8q-~K<x}9o3T)BfTpyT
z3^K@ELbqB<^O0C$$n(V=J5MjKw)VW#1^mN5ILOG2<|vu?&M(xq%0?d=((O{oeO`$D
zy7RkDs>e8yf2^$RO~w7UDydE;?EActGF4OwV^dcOR=eK4C0oI9HJL5C_0xd2q29%G
zT6<cZB`n80zcF23^pkO!$L}(`sq5qCzVB_73MI?&6fa}7uX#<h8oP5A9kC7y%RihY
ziOtPJYBD34pB1_N*8Cs|P6b5HSN(!Wycp@cUAoBY2Hg^_%|TJ1mx;Rn_h#3RCh;a|
zk}q2+TCsVhixb%HdBQx7zI_!g3?q>iljT1;M`|0h${0T4hxkT^jLnG0Ebo0*%?@Gs
zX=EW^ApGsj3sYoF^p}3DLi)-{m1=gj*^wfm{*bJBY!8<<EE$_|JvcWnFCGdrKzN$f
zd=wcG(X7M~7#R50A_JJjWMpN3Rskz*o<;<xxf`T{D;gdN2?+sFdzgcFlAN@`;1rZu
zEc^SI-*x6^88O@n3-R8fYj(5dwU3cYWPQ<a&|`O^yz;P|g38VMT(!=__rDi%kzTph
zMsogSB}>ADH+xTKC;~3R{wXA|GI{eZjjypum~Mr}U_m2U`aM|(@z=;Ok9hYy&5hPT
zTROqPoJoCJ?gNpT#-0z-xJ|y~En0l%_U|tdPbX<Lt!>^-VE8t6Ws-RlkwTGeV_)jc
zU%1jH=rDRlO46tCZsbKw_X53`7IH;^mET8tJkv)aIZx!#41_QSLxn!N-ANO(S`(M9
z!`f@}Z+d;$n8~K)6Wnm$PgmlC4|zcYE>^>z6xjhT#lr8p2NdTR_^imnjh=X<Vzf8P
zw$)nQ4QPTWKyf|n+CN**%H03lxt|=8d+1C75Z90}50H?N{pvwdWzx$#KCo;65J5v*
z+dV!$UZ_!5q_B`g)$@-x@VxggKs7O?7yrj<OtA!+x7%i%N5cyjMN6tmYO%X;oZPZn
zdUTbb!9U#JH7kZJnIy}maEn&wwLBO8l5f0rckB0Z5}U#|iZoxfT(8St7rEqziY`4<
zW(zI0xgYgC^CeOOe-HOO*mk`GT+b84^P=l9Sr(Mks%xVlR@vJqIQXh)J@+x)BUy5`
ztkKmJR`D&QZKOjKJ^3}g{T0pqpYn<*218f^`C4{KB8(R_)+S|kOrO`Wko%Bb!Zt84
ze67$kC8WHzZKq-zf_+eRALal=<9+_rrnCNJ%R~zI&g3PP^NNaI#+O*B5<=!}Q&s+p
zR@NYqP1VV+(vQ464{z`M;(LlupQ0e5&f4UoTYthogQZG=|0GOnOHJT=O~T_KQX50o
zC;dG%nwH-zCV9fO&#GNd9ucqQypgeBa`iT+`+4g4Dbw6BVJ%@|m__q;Awh84d^xox
z<uKyI&o_GEo>`zkd`T6HSamKXiVJ^1E;%}#l&~?+c9-xrwa3}g3u5*7q<I6D>)HVx
zgLdAT-}t-EHij1V<}Vm5XGcc&JH{FY`Al(|q$eLZU&W<WayJfnGN2p2$~HeS=Nj++
z7)yJLRzCZIfOb|_zd-tLt2dP-VA|q|Z2XY9`*_<5Hk4Ig3pdNkFV5rJUBW(BSXd}M
zUQq)E3fvA@zw<j>tmOEGPc{{G+V$+u)PUMgb-W-iuWt18&jI&PL&FQZM!e2ews5^Z
zP$WX7Mc67?3G1q=QbSybH4Y?y*@gFVK(5D_AbX&xfjyW3Yq3wUxSRB_SP&6q)RBFy
z?ljLc5bA6sLtaK=ctdOG(cK4*<m^axl0=nM!RjbAg}u_Yu#m}J!qV@Y;hglO(q6fi
zId#klhHV<+cXu5NT-)*x(7b!*bk2M84X#7wZnJq{wtKg2&MlWo#)ae4Q%SUybfA%}
zuW$VHDTWv(-Ut8VxiXe<B)4$3<M~Kf)=!GAHqXJ{Xp#7&y77m;nWN{19*;ArZDDFI
zjK6+TNU)o~G?dy*OXY5H%dSOgX1I-UJ<MsWg}UtFf9Iw0J=R+<*er1cl-a|?or}C{
zUYlzi{aGay__EY?ny3VfB*Y@$Ohi!LMyUpBZNlexhXb>beMImB58b`12a~6RJe9M=
z5f#Q6l{1kqE#{^>jwZa$f4P);_CWPobLG=tzf=nd&NMcd4Z^Fi4SLqaxXBh|N@&va
ze5Ci6RtT;gzmuQ@vpbc3@f%O!p(f>Ucx5K&pK?dKS>3W*dh|B@(q2It@|A(>t@-r=
zAGh_5qPP4_c8kb?-!l>Cy&=(u!xZ^iM42IKpRV4TAl%KiYOwtEE6Tt_S!iNyp?Rur
z)H~hDO6#dIA7fd5O_kEo;TkjANie0d<z3B7l87B0z2umYDs!1VMXUR>CiJQRwQN>%
zHgNl4j!#d^I)=B75(}{6%<V-@6j@iv<GP<MT$(zjM@X&D+WV<zEu5dJPA7fHHCv3M
zFsex^JKK6luH<7@a^xwxL%1$}Gon>UL%S}e=w2TOtJRTdVzGX<o2^P<XjuVcVJ~#I
zyFOJTWMNLqT^y7Q46W*S8j7OGB>GuZn$pUVMOblh)WiAxtw=Z5DM$QPYGbO=sP<c>
z47?U&UWmNs-cIfjZiaY6duNn$_iBdK_lmr8MG{nrKaqPRo7+3w64F<W*C<hf+QTl*
z<amYX<PO?7a@CR<&lFb3zP`YD>$8po5PVma??g)P+h<W(WH&xgrXUt0{G@&lOytab
z+6aE|T*0pw`i=9ynnYqZTfA~^rW(2OUTa3gIy60EKIN|Qy7{Q=kk*k|*g4v6anFP=
zJi9z8y?1-1XI=W?c@C4+j>i#yY;snQp2T>$i+aV_T`M<BD!*rtxrx7S#~RU!bz{n4
zlT9X{dZ+LCu%d0P@1^gggm)~Dt2MN-TYb-|TTFk5P%Q4_IG%NPyULTpaGRu7RC}@V
zEiy8Q`Q4U$b9+0IH6lhyaH%jFJvs=Y-An|DY(OJ`lWS**7M$zAxQPqVhd4R%b=@nd
zfrHMD??yQa<tl5-b&Qfe(w^mNx2r{yX_=}HsornCo2Ap7O~b|Jbs(iJ+RNA=YwSN3
zmIIMSa*eOo^WIRguFG<x3Xz<`kI1v=r1z|3gC!;vn{8i7nrlBimaL20`kI=qb?=&i
zs5@Qb#M59F?vjdI5b|*3&Af`AQN4DLne=pHAx|XTA1WH#!UNH796I?E$z~fE9-Dgk
zc-t)Sl|J82+2Nq~2{T>emsncg6_H#{mZ<Ka6Ui;7$HIOacA+EkbKrY{)p?{oCv-0i
ziTig<e$h%D+yM#pFBJ8uY9iVeqjU0S)IT%@lw;EIIPX0!u{fs8EW=S?tXiu1O4=|!
zQh7&T+rf3i<cSz>>;or~RcyxPKuk5eN0e>Op9N5?h2#yb4uQX=D)fDj#eTtNH{hvg
z**0-ezsE8u{mMtUHaP&PI>>!sv0*}Kt2_9+jaUnd;f#rkkpE6yDN%6~78j?Bd!}?m
zx$=3OAGBrp`$HY)w9eN|7}(qk?yjPYYx(jAFW#pnIe_0U+t&3>heK_>plS2+G>O&k
z7AnO*E}7(6f%m<Zgyx(tQVvHH+wb18ASY3sn}WQB?R@yc@n)YjLa9)K$H#*ceeZsw
zZafw##jmcq!H*pN=eTa^`Gh%l_|5huzYx~{UxMKd39*gZ#|@$M2Z~AEZyql-iYOKt
zwE?>x`8F8{?WKs_ZC~X}B#N{vRU%1`z8BFtv%D(9uIx-Yp492Ez{(m?!0aH@+NVLr
zL{#fqD>B836h5&ar*0oPb0Fh>X_#s1gMDjm?}bY|cCB8eMU~tM&kd&NyfR5PUFW@J
zn&i`FoWS}?qutksWX7WMf7p&IG+h)Ei9XS+pQZb}GPcf<x^nhXcIu8A4pZf-*x{;>
zgfUlb)SW7`g#pfewt`I#8td9Ms;6g-H*>5_Q}#>^mJ!2}r4jde&UZJs`4uIPv+iHw
zeoXI^XWrE{lSO^$<Dmp3CBiKw{#zci9dVdzz*|aQ&H?h_aX%)@Z+dQd#`w5&>D64t
z5OJpGUhpJlDdQXD)W3c<L4=yL7P?Y2bw1Q%0cBITfj(F;kE~PWTaULXFLCjn=c409
z{1`h6mpRjv&l`x%WPh{rMMmaz@nnW-0#O_RI3b=lM3eJFbl%~_GBs!Pq;u+2H%8Sa
ze6(=AGr%>#_J^}aBr#6(IUy|f8_bO1p9M`n+)dQD-#In8<g1qaB!9UJJL7kMRj>HU
z_w}akz-iKNCii_KT0)6trl;l?KG`pkAmEjLZjfUtL|0wFu=R+IvO!#LnewnKUw(NE
zPkZ*hkJR@!f@gyiuYPC7o*y9>0o_CtWOKFZ=@6dC%sonC)(vynDbCb4A}xNUOt<BC
zOIzBj3USMU1;;ABdwqSg|C3K9zExpgM{m1u`h!Z4{rh_@t(<ukZA4KqtY8y<AVmm5
zO3g2PiL>JF`%H{a7}#bkMKtN$T$r}?m%Q^eC=i35c-1j#-}U@>KU<9Ii6j+IbJVSh
z8IPa_6NAaQ`E2tawM$v~%-Qi}<*WZ)q4p{)*OB4jj39SNE!yQ+cg0znC)_W&jEmoV
zXKj{C9_56TRfPooC_4=#BFckQ5vB4ex2-Pc8(;Ljcl@LsKK}LEfJ!gAcWHXY)2D**
z@aN6n*Ml6JW=mg0-SX>W4D>N~ZH`l)ZSEFSyeOf{G@xUc84&Srd;-HU4-AIyqw6mz
zvM_PDcGoC>{dmw9CH7QT+~iiN$qhF@H>9F<7fZiMwBwCk1K&7d2Nt~=;lAJ(Jllmq
zIOyRG73e)((C|jTCx2}W%kz>+{$iYI<aZ}tk+q?iyOrKm$84L)t|wu8+RimS({s1}
z8+w~F+IUZL5!<~q*3=O4mp;$hm`lkK;Qo@r_*mdvM)1(i<n-am10;Muiba1<PICVT
zML7+wU%s<cWpn>6L-l9(KT;DE+4zTt)%O#&PxUmo6f_W)#E1OY7`d|*9q$lo5_>sX
z;xzxlHrwMbmiz8llYPzCBxdQ=W*>%)AKgX?yx9Hm-s<0?x?rl|FVqKT3RkbL4hJ}^
zC6qxZ{eX93Er$r!eb(xibEx8#lCq^<wL(sa3w)!sK00@P_@s?Fk)`OFoLR80_wsFs
zw`nI<U+X?4RPm0>u;j`_`^${9`r~{)M2NYGr(T|rsk*(jowJXLa*3unmzLz<3lW1$
zR7JRDUZu1uH(u$d!l-ECYza4WEV$XOjlyv9+mf`O5%jZap7*l&xyC0nynF3Kgd{CL
zFAn!hdu7GaYro6>It{9Qo?>5?0$b51!&0C~=KE=iv*(8KQl3usi>8AgbFZ9U!z0DL
zxZEEKRRZ1O=h@HPuJWdztv&X(J12HArJ`8r1(vtbMald(@5^(}yJQ^-_4JUlS(hJ3
z%7Vhc@fr_hbC$ioE4B>-w!h~*2Zy(WU|Fxj7<P}L7(b?+_2)dShum_ujJ>^C3^}(k
zAf0+k4T)*O>-#&qaR)5q-jvFfk!Sc(^?NdWlfONwBF609@{4aXQSmp1ioe`n8Cg3}
zUl+|d?brTrCyCv%S>StS6JEJpMP)4#0z0ErA3P%NvCtId-BYbg4@5I6#;vkYr^HxC
zKOlqKRzbs)N_4~c22Rz8#BSoH`IRgpgC+e4nr)L?Tkpx;tH$0;{jPS5^sqT_{7o@2
zrX3heUaA-sU;NQ#_cmo)O_?A=Lp!n@M(38ss@H(93Wd((+`E<({Rz*NZ4Hq%_?*pD
zJM>7M#k9YSu#-MRC4L%gD-6wu$5p}h<JPYugO3`V2Me4gc{r^XK6RAV_*IODuRdfy
z-*Ds{&IpKy#z`sT%XuqrWs*lT0L-WkNV=Jp5aW4Da>0Ni9TN+S=gA=XXUq~hl<*op
zR%~MLQnj3Z;~xdJI5i~rLRCWT3g7QWs^k3e7&kGhh$a{Qtm^hl51)lc)eP@-%5W5O
zv+vNcmTtx^Cp>A+7(SE^TD9j0uNzy7+jDWzryPX@=K$wm-bAk^S>-Dye7a`yo)&Ri
z-(}J0c$t!%Es+C*<?2(F;ysspxz<(>Z#B7r?X8^-%6rRcc?I!!*-?9~@m)-ES7UL&
zW19i`$6o_H-ywI8{judma(e$f(i)Ao@R;xHNxt!SR>D%_A}}k<5rPHnfY``e;A8qq
z5aW~#IQj8fn7iUA)v<d79-D*`%aE9L95-I9$g}RgJ=QvJQHsk}MZxCY@Ov6d6%2ha
z9#t>JSbHWi$YspdspMPB?$InyjP?>{Qn}~1eqUxSWFO6TZuQcILgF(DoC{WI4(^K<
z7H)Ei((TH(;hg)i{}#V??Yfhi4oYe6<khfm!1|8ROgb)D!AlWV_jqZF727^<{0~cj
zzr)5=_GfU&$33J&eXJ|<IM?Qu5I2iD^72K&RGIgD&DRBu4O~`@hWvlV(-M+dZ%=Bk
zCKeCX;j6A)AP5VyHv3&Yg>iZMy!qr??j=#4a}FDLhC>{}c<->pPunX26WA5@>;i2@
zjL=t_J%q;xHb7hv<xK4<nVC%9{U(Om!6Ez(t4a$>$`_Dh7~s?m=Nh?-1ds<L7z|Ip
zSGYm^yBEvf^Ocu)T0u45_)rNO;m2J{KnZw^+68_-mkU{)ScV5R@FCex)wl>2UaVKc
z(Xh<el6^&I*<k+c?RnmNmu?#!3BNC-nG+eTC2m(#cHq4*AOAqP^X9MY)b4*J@;2q3
zhI2nxcrIpA1!DQ@BMhsmhOh8mF+MvJ!mKLKjc8)fR(WjLl_d|IZ~BWwMr7iu<31oK
zxkv;DiNXefk*l={1Ee%x$|ReKH;cANKPg{+KqyBZ&!a_kU}fZEJl7XYb4LDdZTabZ
zWW|j16ZENmq0>U_ZXxsAbOZhof`~<_VMg<48UxXL8}o?5d+UgpI>UEz;<<N$Y2U`i
zCZ<=f#B2ul$&)uB%Y_DmwN%!_H2^L=INs`D7(q6tg-I9{f1)KPV3S?M#eENu5T32p
z$2N9yymX18AxwbL2?!2;Aty%y@^u<@ZuqzF-ra<Dj)xZWuOYbr5cElS^#PP^h)7A1
zYXC|H_V)G!;eh$BMA5|JD_5=@1D;B`*aU;Ws`(lp6#n<810)CXb!85|LwaEoUR))S
zJIG}Z6*CyfafZwwKv<~ZlR+SF2GUsCe|Z*nst!9NzFXRslPd*~!ci#jr9pNySU^p%
z!}{9SL>aAx(m}bg<RXN%!DnFc-!b(?ziB4re3tU7H+`SyC$v;A7vSE_|1@a&+>m1+
z1xpq5R$us0w}F(>K%O??IUIrvvtjTBe=k_@Z2Y`NYy|fs1{~*r_mwZ^rKk4<-k?)p
zrpmG3GXikJ8@eRb?3aM2`*>iMBpxm|!3U6p8g)edcg!=OH!rJo00AvDh!X%aA>Q8J
zB3RRlhhvZojzUgt`G7ASw<wz1+E>Vn1daTID)Pq84r5hzLc(WI5Cb_!)pyab19Ek{
z@1l_qz^e>)FT<jI@jFFcVPiU-a=6b4EQBxElao_f1_?y+<+m!HFDUUKE9ZxE30L{g
z?@>t?M)P5`Fh6@CT<;(qMAkm4b&3J9p6I#B@hpzK7vytF9gMX&ML^!Gwn5(6Ey<5x
z4LXGB<^1>Vx#Lqj8K0aKSQ{nIjyeKzcQijXkary)%taHCki2IlZ)$I!g<na88-C`3
z{VgO!eq{rYsz?K<sMW$sdPMF6Wn3t))XcVp|Bw2M%hD4Nn%*yNL7q1PUYs3Ooe942
zvmycX%*Xu+_kqDdG&Tg1bnxqtC<cM=x(tOp`nHHj5U~T2h-4rMUmhs~q>{U8_8iC-
zoX5sC1JZD6n{j8ZIdz}!NH4Fd__HKRi0X@D_bu_CCd$3z^3QDT>An6=;bw-g+tqlM
zU-j7KqDmYIK~KdhHwz}OdhfV+oW@Aid^3Pk+%XnI+=}nt0=n=^L`?q4AElhHEKYg&
z`+sD<26|=X6if9tkUdmK0V+R;L4LEIRBe1(c0;)@TaB>V9O>Ydk41$67$1+h+$ki4
z6o-)dXH8A)F`vA=JT)lD!MX;aLZUIpBg4eODou3t4UP3leQQ5nje#OzR#lEAb^(!E
zl~Rj$qvf`ie2pNZtZKJ#2erF4hSo-CUdt;eNECD@INqx|%gm7!5y6di*j2D)%2Yb*
zT}85>QcMYQmgpX%EEO~)w5Fz}VD_EQ_rEivwGM!h;#dwlj_#zF4?irb>yx)~V2y<^
zl|LCJByq1vQ*Ex2H}=~)cT41(b%QRTTsZCi?5=+4%Ulh?_%BtGk$of2LMtttc$Ou&
z;a8RPi9x@cx`sv_$X8+@`%rOY7~uzb8Wbo_PT&@S*xVYK&kJbnP>3zIFR))N!Ev%&
z>U(2<4C<Nj&F?i8&V~U6*+20>uTZmv1ogs){7r1JQUEYqU_Jzj8C=%^7SN{P0RSfo
z`mJ+2-^=aUuwmt??Et#7_Cs1M5*sQYD40nt0}HztQ5<Wk6Xqbp!)vF@v*PBgOQ)(%
z_LQ7^^VI(3_p2xkrP}aECiJ9}n;7$Z^Xm=bs(+)T3sN3^1r}XGQKOscB_cF$aQYjf
ze{F!IEel8@LKlvbY9Q7t5l~j3THhN?(0puwmJ6&pJ6WTjBKK+T^@KY~+Z_NvO$xHP
z3$WTppl1-}c&LkJ`2n%i_BQr0@2V=Kg7gXZ$Nh&-UCq<Q72erG5w6!4jWY!ZJfAxO
zm0teeXH>I{X$)pN$j6_t_p*=%*!xMnNO#w%=>>t)Amt0@uGZWx4pgTPHwxVM1Sb-^
zO;|{^joG;_m8MQyBSGs!!yfPX)s}7Hoc&o~e#Xc#=KZY%B8|C(`cVb_$mzvcu<6k4
zkqhj6(lRm|8ygtopk;`o4pMYMyP(a19eC47fwqW?jS&-?802SGVD=-xRE4P8dBFBV
zn<P}eB4Y>t84q*$x`1s|fv~9Pxs)6yMMD-AgaxJ387EYr(M)0^Dk*9FZ%EG+S<w$?
z;x;vRGLbc};Y6>mCBtfPu6&(7RZU%RdA@<agsM*Q^S?kG025v1h{<PP9u;fBq-w{G
z{1$^j3s9%AbXj_UBVwS%?@*d79~A7}(zGh=Z0_T=WGa_j4QP^`>UTx(daIDs)Db7Z
zcfmLlK@jKGOWz!m2>~wa)mEs2>-MG*L;F!?_W^*eF%V+ArMQgWigJOOA|QI}y6<yk
zOHsE2MM7i)l@kv?zoJ8Ms4sN>KHV?QckdE3Y+dS{F~TFHcL+Out6W4a`QJ(7hTcu>
zqgf^hE6)3^HoW(kj<5UX=FRL0m+r_*0!$SJP^+s>-N}+5jjU-@vC~Jd9Rc`+Yy0|F
z3f&TV^)7<!Hz-L`a+-YxG~s2JnriN~a+%~4`|a1%{RR5{eFxv!M-r6T6#A-)=)sjZ
z7X#%;agsc`(2+RWhYx>;*`lAjj{qY7R_vHYYCOgH%IGYP=epaSi{PAiRK9Ze>Fcu-
z{)-m7K)WcEDP_JS6Td%`xu!m0`<)c?$xg((5=JVC*{(PTQOKi>+_3F)jgF=P^@Zu*
zv4S9{ve_%IhVc+u6X+ZL$R1%jforh&M?d=t5mC?ER2G|cbrfLu77<p*eG^Um06;&Q
z9B%8Df;C^_17n31xaH;TozXK&%gr4Qtyrk$Nkhxd%a6kS^U=bqrh+9F%ajmhSy-o^
z(R`H8lMyUtIJ`+Ee6K@^k^isDL;Km*PK@kT%h$r@xL<hfpK_^}cp_U}_c}g_)URZp
zoGVDu71h#249VfVX@qmkxymyhnOj|Lhlh?a-<L@N;OpD$>_x|?ad9^#qWRk3=g_V%
zSz&IwH)4%esRxe0`JLvQRxjeefx0fpO?Z%-r_=WrjoRTk0j<my>}2WJuTl3;oSBu2
ztD(n5CDzpi=7^AQ{|4$m78(Rh9ccFcKY&`R3>j+tEAX&*T>xeAxtJPs19qqNG+nW-
ziD-aWsRFg=-)(#0-4#f!pq<^0h6=&cB-UR&HD_3K`o^-2I-dZBJ&!r>b1B@Gc?IV)
zt7=#8{ctuwSVm($G@X1-NlBSTbvB83G5ZGxH`uRLG(w$5ZKb#^oV}kr1RBXy0m-Vj
zMo2)AWo2dh=JaF@2>>BU?P!MvP$yQJ`TnEUH1v*W@MQ{$mP&^sm@uWp4G@l+;q_Wd
zqkaE`IJ5@BxJfr@>DL(s9B(J&)pflmk8ieU{C6lknC)v05N1HMI$;5Ee8nQ9ajqu~
zyk5!$=k56fk~%jcn?*EmCa4n)*o`@hzZJ%M#?s#2PV5;CEIpZ~<symi;lKk4QFNQ0
zQR%{6s8KO7F`>1hpxF)7d>lc?RC-OB?_^sB4%-g!FfUxX1j<z7@Q`!f)x8BJaVbPf
z3PewPIinOawVM3zz)AQO(5IcHmY@cVZ#YyqliNr=1!5sVig!VK5seU9>`e#dkn;eW
zfgc!^tvm^4R%_4IXf(01%Idded)}JVrSsUv5*p}OSHpU?b<rNwJGy(!@|w3~v~6>`
z$Rrq*mkPqWaQ2Bv3G0z>0sbD9Ip^;gvn%Es3%d{1uH$ZJ@_za2c3qAWiCXz+W6|5P
zyuYqZ@G%d+Zm}ij@P}D)9zN0ar)Tm<*k?IUUzC5OF|@;Gmuh%(G~rIOSRn~#17N)#
zsKsP50hCz3TrItqMXx8tZZD2eDItRky+;9=oMO&Xpop{quO&b@GkVP)-qSMllb6xS
z!OU*@FbQ~NZI}@7uww<*{Qm5s3(4f3+D?8yu4v3Qm&$v|y|*dYO6^{cTgE)ypC;X9
z9`dYAS;X8*I?G6~+`Hi>LG$uyw0>28z`vpN-vrnz`c@t85zh56<9LiXeqxxDL4Jw0
z)U*#?Ayu=+JhX`0$}zz_vcwL;S~9=T`I9`MIy^kQ1)kMF6@mzUCWU)wogHeT=?|u!
zey%Mb@@W50IO)9Eqn~qAqouB92l9*ndeolD7n)9<bIK&gwtHsGu9J$DNzP(na+lOf
zg?8f=kJ_0zQ^h?3+R3p!_5)0W@SjCi&Q?~k`)%E2aww5DbW>eG*)QYNxAOxFL?cqF
z!z;b%q;NV2Wu7j*9z)J}A8WNfjqgOQLr9bdz(jz#YphtH-D^)-`r3wVnZgh?7XEg2
z57cL$Ri&lmliVT6ob8$t5mW0|c>7$Gstc$qe30$u;uJ&>Lcj{~tzkd@R9)|OuBC9}
zB+wPGTw7a2|8DF;XNE#Wbs6-WoE~pUqOdQKRfBB_<uHA3`-pTL6tQjrCU<||22NNY
zlSzWR^uGSM;m-ovJ_Kze4=^*+Zyg8Y9_204jCY6)_`A83T|3cJ_)N=t%EHfghR|V0
zVkDifZdKQLCx1eqM|^3|&ClplpW^!>8ttv+!@be}62p+g%dvB|h8~QWA&BeJUk{v=
zIXR)ws@rIlcS1r4j~f5Gtj-8!$y6IcpFVebkHoikYOmv)!FMX0yJS0kE(2vfIUFfX
zKM&=DrTo>-BT%@Epy??n2#Mw`@4g;kII!1C!bHU_zeoAs;Vv-z5D<Gn4K06QrK6)0
zq>-iiD$^jS6s&UP@usMLGCJicn+k|Yvi)rpC?zW-2U+0@vE>UuX7R^!n}th6<i#ro
zou9pz-%LFM?%r=E521OWx-N4m`U3;P$kRvSWVn`@b$0sSgwuxc0`AEQ*yT12wYBFq
z4SV05Tok`6bFc?3Co@H7+uf2pd&A~^62dUYDP4vgBR+t=rWiXWnd&EfXUH-<d2u2O
zDUf=_;2h7n{u2AclGKn+@zw~=#j3qv#VMLI-RvJ5lF0AxCG^Ih5%vd?)5Fo)zo1K@
zAkp2{6m2fh=NGG}9P>ICC#RIGtn7jnKR-W{axv=ZvuDq2H})25p?T6iGtw8{D-}<D
z4g0Ez`Af&>vunwYyyur>0<)`RQZJOmlM1Smd1*s?^HiAkU639@Q4ntz#N1DpszFKa
z(}fmO-CHn4&N62k|M!SHm)}-YQ-F-pix=3a`uch>aH!ha+Gf0R`7C8R=5=Tk3tmf?
zQ>xHF-=!2oc6IRZ(F73ygV4LVMT!(galX2(<6714LW^R?OMnKk^i}vC^ioO!<2=_V
zjh&PErG$V@?ycqndfbk`<3#(PMmjK5h*qK@fXDdy?VB_BVyL1qM{ekaeT)J^aP`Tb
z9w_KOF8(+MJt%MgH8r$HRE*5?osyLLRk;E+IBI!irOh)aM77i;+&vRr7P#;CU-Wue
zAn?aw;2^X<2OHDy*yMaXf|el0u_zWz+$E$ha|)nDnyud-%@pyU?oaheKqu8=LPE^<
zo_W!Tv#m=W?o&--uj)Rq_EyGU?#(fm&Qg#4t!|rxDX1&<*sD!<P?^ih7s6v<F2|^2
zZ{-&A|IMbUc+TM^G<q=57y*q<Z)#q06hv8QjWQ5TH^jxoZ8l~`u91^B!^vxmwjXhd
zcjG-|C<x8A8%H9B?6Ko3-_y1a4h8uN)-ABxa;Q2phTj)8tuWiLIv0fN?^$#Hdr1R1
zDkNxhH>=WeZAK|jrn)k_$n3CRFrh&?%>n`6=ksW%*XT8O^*J5YiMRiJ2HNM{d3MhI
zBO?{fLVXBo^y8K|fY(}pF;~*j(eYpfP%g6|#`iovH|J$6ga%U0q0jr;kNu}*E;i%0
zKl$B>U%nhOaG^<Px_;Kgre4DLcpazW^wF1zyao*oW7n3Idi-Y^+JaDZ1?-CZz{mOd
z4~{t}3aInj3vZJrUdpY$H=m{l`Jq4xMcr5=(q)sYqqD9A6J6!`Nbg*sW;DAifBFoP
zS8W}V|GVRWTsvpdv^Dt@=~SEL+{W)bwk)6m1`+@G(vrRvB1rrB;mhqljSR;L8K{kk
zbgslc`CW%^aQdX4MG5GrA7zT6k&gj@&AG;Bi+ztdBuQ9V*_w-x5Ru0%hf2hYueQ(H
zNvby}p-A4vRqo&;DE_yePX#&etX68@b+?Von9RB{Y|J@a_4+%TMcV6Ac+vhJXKxCg
z8E5?XuVya3+&11A#)U<jMS+iobSY*gX^NH%D_rQ@zv}lhZ!y&G(fp<L*ZTX~ON-2r
z)_Jag|ERlQY;(&^w6Pta4+7MRPpEuU0*?2I`a`3xVdowNsL>7kV_P2zkj@=qJq%Od
zuzgs=DCzq=F}1|<v~UlR_AE0hTvx;?ZYo%GrU&G0`7nl?eO+1Z^64}w<!~gVC7+*|
z{X0tHFpe!!QXi*OJ>m`LI%gXH<M5`wF;c#Pon3iI4_|sNt6)lZ-uSEiY}QSzAECoo
zSI4g(Nw*!WDZhNdYI1VIrm(W8w8osc&85QYoQoP+eCv0I$Jo+dOO<&S_gOJ~XUB;L
z{G-muWT)1_xD1{&F^{K6sb=>QTc2eWaknmDTIm%&KmAo-w_(`<@N={gL1>=`z-rv%
zKRX~?E8il3FNuHmu`@Rp4$gs_67RbVwG*pC`MU7PYE^*J3_+H~`1@_4yzV9EXP<pm
zaO6eC2Lmn#DIG}&z4%Unlg<nnu<Fl}NxcJSgC8FLux%%7W@pZKN4e`Y%B2Q8<{?G_
z@b{*?bCYb<3G%;FhV^<nI!!pK_@RelpvZzjM|NFjsI+bPgxi?AxREe^wA$I&vB=Ih
zE>t?T7)r($s<ctcoi%t;>qZYKJ22|Ta8noV0E`>@9SaktCZ`&HW}msrtG5~YOj<|Z
zWPBp5i}C1pOF576o^v6AG7nK>;<dRpFIwb3dmq=Sa!vfUB^MQ{u<!?0#pSqnw<$g2
z@0UGQ5X%heJAPVpR9zJEWz;Ni;_+G#R_gOe9RtJSqbfp(BQFolX44`sv6p|LaPapB
z=pO&fo)5${PW}3=;)~6Uqw>dLeIJ@U0JoZ!`PtH_I&DrgUNGK&;0$dA8-xfR>*KHg
z9XVebdZw{PTU%Q<SD;gm_1#@)$5;k(j^-B~f4nr=^mGYN6(KvC_>h|hXU~o)?2eH#
zJ@Lp|S<r9pBa`LEk>|#oBDJv`+oZTH?l$><%iX0uEADFjPF(Jl78%<+BXw;*DD6w?
zv@aqj0DMHR0I>h2ZFwox6r`l?3&swIm#_s7tSl^8^(=c;Zb0HsS63G(Z-U33y#IMY
zwCjq%7rmJzEjo%yt6-e1Tz63+$jXa4%lCvw#`A(x7rXb9R-HxayNpv7xYW#~tc9~D
zitXZUa#|n-ynncxd&Cg+zrikj62cn&ihTXLxr_T|VYtM+ViFRlZT^4lFt*BqyW|r&
z9bTjOG%}eB=Bcs(MV=U+{UAjq=%K2ctFGJfwd;%qJ96{~$G`I$yyG^u++T{Icz6JT
z-vt5!qeWQ&ZwBUBFC;Eqzr#kD;nhXP#7Ck^!2CIg<EYqq^gJgC38FPq!1&)9=%@11
zS_d)Rd}%MYoScU+dK8ep2*!8HmSl@0)^>JYVP<9q-mnM4f){A@X1bfy2023UtPjFy
z?n@HR;hAomG&HsJay4CiM;q;XnoxhYriqelT-PxsfD>hdbEHJ~`0tj3Z}@GKB=|^j
zr%p41R}Km&!D!by{UZ70PYr$8b(Lbj*qvm0Y502e=7=AL%|GP06!fRIWx_Sb-`Az?
z%O4YzV@FO#^^v{XRbw4?2F?E!^y6X`Dun6)@dc@`n{;$de_~xP*BV$!C<?4C<6~j(
z*R)1U*1vcgb>ZpWMv>wpZrTUky2x7&U+VUp$LA(D+wjaD;1Fm}t*!|jVW?z;$NcX}
zVQX|=8cl-~lmd6cWr4T#h8;e4^-dfwjoo`@%|0*ZM}l{E9MPuD{0w)eGbC!~=B4i*
z84&C&Io6bIob7bhP=liagdDu0IObCrXjf3jAoFz2`R~Ugm&I#8uwFGcqj9$~zD-B#
zmf=m5@7r#3uz&53!wp_+@1KLcpVG%a7h|(Fa~Idv3UqaM10WieZ#-7{hyWs7;^HIw
zN1FvWZ_gH4Y#w8)uEY@8>1#Z!nZ(PvLj`%Z7Zs0lCjyA$qU=u0A3_{(xrzbz)Bzhx
zTGWLFYu^#YOW8t#dXH#t*<X25YFu~kU+wexxyPYm$0e4;Eu<9!(gI>Z8c#A$i71q~
zZ?1;cI`~z7N_V%urA_zQ@@<8X;Prr}pOE942ihBG{LeQW3XX}n@#)hidhesLv(rXM
zh<|6(C$Xz=O-Jx(pEE*M<d@UEm-JqV6&BbCUkQqn?ReRv*%cA0`SSrPfZucWb-`#*
zFu$W+d<A)*jw4BcT%aUoYJ09hbC`Ehn)UB+bq8EA?u*)Z7v@L4b(&LS1Z@YC{M`us
zMRB!|0;TL2R@nO%c-h7JD13p}k?EG&a80pwyUAWUPRUBs+Qrqghba(SP9o=`N|*z?
zU!LhK_Oy{)x#AodNd;64MkI|!&08hZY`-pzf}@Wv3uLL3=}{?xs5GFzQN??s_6xn~
z@`4GsKVP1KSkqmeo90R70plSJb6p4B?-61F<?3z{Pw}tP0G*onh=Hn%R5BlnOSQ@W
zCTUS$Ai6aB>(*H1uR}yW*zElXU*byb+(jrbpXYG$!vC-|p^UNIAzSDSXRzs!{<Q=)
zm%odpc$JuVc55-f#e>@tHAk{C&u{wZl=c?A_=k1ed13Xnqj}2?1A<cyuVP-ct;Imv
z^6gQ{ZAz1u*EEg!A|>P<XFi&9Sjj|n?4&#!Z4w7?b$Kzu4Vkb}n<;#3LTc|$oeWkh
z=scPkg3?6@X7=Scsn28fc+bw2{n(1%iFjKU;_sl4!<^p50VPOm9#igj8b|SctOG)X
z0&jgXU8pq1?7blx_%Y=Y%~{1v?kJqeDv>C4jWvg%h?hM~?%fJGztleykkJZ|?@pxv
z4+KWCw#)u_IW)Snw1-Eyu_p6Qk;JJd2?p=r(MVqT-X{O|qaD8EAxoo#4A-29Uc7S1
zF1!bdB{cWQS4O467`b#nh`%AR0Z099Wrf3$6;LJ_tptKQE5gLL>TS35%m~rMBqjLy
z^=kMl2{V#HMWx~6a-II<Tj0<l;auXhtyckPLCAQ-zxmjV^K9<=4`%Jx--G-BxNMZE
zbQ!u935+DfBa{Bw137c`{*L9*@~ihc3NP7fsbqqSVtM`W5|)B6<(cM_=ofr#;a-KT
z>PdZjXUikjHUXMvryGyO16D5*HM{-Tqi-g9UE_0O)fKPG^9G*;((}@E@~PC7sAx|P
zsZjb!Oih<x_wCnIbQYp?1CCEGWU4Faga>Xjt@WGQkog}~W=VFj8n&w*DC3blNRc+T
z-#R7x))*2m9?FuqzFD?b$v_jGCwYaEQrQNyj@zP`a8T_{20L%2s55&GcKH_BHpx|Z
z44k+wEj%msqNhz!M7}7hT}xjjRE`ymE`K(%N)sa|Z*g}R3x#?bG-<)L)0QxIN0@YM
z*ZEj5u9@dW<C2=3j_fr<<(=dwnF{j@S3Bq*KbDb8e*>8}?UDFJU^@BQtJmuQ8Onwo
z?IndFpe|C#luJxXT7)DPIVj$evl~4~?@fcyDsP=5ybVc$Y~)kJKlxIxUw?({YaEDk
zGqAJELl6)QAC#}#?dh%7HJsdG52R&2l$_G|<g5Vy{(88(k`6Vg%%4XJVbtZ&%-{Lz
z+pXR>Av~<}=To4fK@QQ{6Qkj`(*xz@JZ&J-1eb{kW7AIp)5NCp`zh?->?Z~TuaQa0
z7ZV%A<;nM!p@73wZb1L8n?!945qb6M)w%^I>QZY|NOs-QF1_@&Eyn-A1T!XaECF4_
zmq`+G{(MX8r+X2M;_ncjwxD+$qOD)!NXWlNx4DydKhDascdh67vtB`7c0v^Vi4mY2
z&Qrj#GM0WmVwE;_T59HlOlTCQjJ%02Sl&9ypYBi98%oS}Pk*2QL&PSaOo2qAUcD#r
zBIp6<HbUOhzPY8P8}d=@#is_m`deFDK_MZ@(DAIdzdoTotW`90NBH{n>x*59URKKk
zQZ_@tc2HooqUZo=SMrDIPMe#Xo+lQL!seu{-1>bFplFI@m@qZ%(49hEcL6G4{ZOJ5
z<xRTHNlD;C&L#qd0FJ3c2#a;Q5`@E8b^Rx&rrtotsdflTgBv#+A)0*+#k^2pD|ze#
zJPaK?TVlCe*PR>BjRL1Tsp{?)gAytrJbY<F4@D-mS#zE-ltX^V(Jz4Uaq_Q~N=Fhl
zq$I(|j^8X(uOLHy)z+-*$98}6R=#7OSt1SdA4L9cz@nu=26}nUh7>Y)1=66W@ik|L
z%Bk1w)Mv@hj4G_t`zN?U#rRyOA|;_g{tMLYQbE(3gkn7@F>zro8j*YT>I-BcTX;^P
zYmf~v0O%WqbU%;AquGi*WHcR`F;igpusDLl*o~QzyHkPhV4-v=s~h%{(&~}rn8V&e
ztUGlVH2llPzrdfFoz(*}q*4%r-&xM<ScLzQhWb}(qdh3ovqG~gEO_xK9tEh-%fhum
z=%WB~!RQ_O>ZLPO3w41up>*0`R5nwQEfi_fN=#1HpQv>$1tJDEyKNnOYMy;>xgRek
zH=Njcc$A^u<*y7VwYRZK(&^aYZE`jxP``$a?c~fsCBYcUDlf_x!QehZtMD-)YeIVJ
zDP7M^&1q4byr8NlZ-(YFyLWAudS(jw$k&hCSd^M~%?hXdcEYw==b9zj0>%6hMQl+<
z!2~@Lx_;XFIT3vs%lk%tNxZC5_{@4)HB&UNO?+}ZOSaI{?quf0U;ffxw&Iy~QNspd
zR@S&4eqcngWZ|%Z!U-milxivv!3dt7o`wO92I!~9#l$=wN^6VcmiOO=(qvTL9X(T~
zvSk+ZO#Dh-29mL~>yHtwU+;5uW}1U=@s8#kjDe%N000UFI#Q?Nb5M4Y6AxoeT_1xF
zh38SW4RyaI;9~3qrUXmig$vQk=;`UHC9VQYhxF0z3Z8M^+Xs3;jnm*ur8HgMNNu0-
z%MvUTIfqG%<H?Tv&gzH~5C`P~iIaR^nLM3MISyhBC2CxV$7gC67#zXKOe~n8$Ufg1
z%Tn-jOb}Y@_*N(16=|7xJu*>V>D69eh}&~f!6jj4R&$jz6G6HjdHqU-gzQzjqJf~w
zI|urkYhk;;pUW59+u|>@_;wu`CY<Tt7<sEy8lScO+<RO6-a_K6^a*O^%-;Pjp`-Z4
zGQ>}=M-V-`sQbx13D^8(vqsC#W@*SR8@|tZWoKtoJMO;|rZ41Qf$|NbFnZ0ku${s}
z^fZhuHHl3hTZcv6alsmo)p+#SyF5E1%3*i#eKrjY__i)gW+L^HAHQ<=n2N(rulDD%
zRq)Z$6DDcQ<JWDf;p88$MOu>C^_ZGo(m2T490V&FZJ=%5`_X}gKZiq#sY%h>&nV=p
zDRh`l)zkef92j+_l~RSj5@Zb~uF*t4NUpPOw=Fy_SKJzlpgTBtb!K>%_xe;=9>-nb
zk(&t~hKZHEK_t-`Yi<rH%YojTV^ZjGIyGWpm+VS5wFsNrXOtrb{bU%=<`Bu?Pi@_}
zwZoV%{iu5ncb0lnq;9c+)lt@9hr?lD3BKPJM(-LCo;^>7to!{D{<5Ir@;=+|%a^aW
zH8nMZ>@tdrIfT3L*@s`xauphlq}_|M5m*^2k~cBQ$&gFWkkg+KS%B9ewyn*zQy6P8
z&s?T~N{U&SoYO1;3d<_ji1REyg0Xew<+*wDro}>Hh+Qcq3Oh%?^KRy$nKRKOLq)HM
z9?{#<Yw^bk?2015)VMK+oq1xlp+MO2>Q!$<XLQ-8p}8=TikqVBe$js0V%JjbKl>Bj
znGXr6q#LpQJNT_QJI4Og2dRzLrEI>u{PM{4dxXC__+LMxO<coUq62?M*1{qmO{N*i
zvC)A$35|Hd{!p4<F!M7kQ1q(cqoWh<;&;|Jjf-opV)<jh<hUVcn(on&7^76ZK#>yE
z0359ePzzg%v>xq*bsiKNs!;q*sb<zV*%(usm737T<<qBaAR-g?2MkYd0wR3PVWXZ@
zd--G0`+8g-vPYBN!I!|`sf4m?o>TuQ;Vy&thR+-}rZh-fTvx*pa7ffvK`NAa-uW6R
z+%fJewyBkDZYeC0J5yEOB+PmrdYiZRIc||o%WyAO;lb0vb{zDmWxqdBITZ0HFg_m4
zG@Z15AXmvKYN&A(#NcFsgfxBV3XtTZ6At4vPrBW-U_p{K&)Cu2+ug_{@hm6YU%GX6
zDmgF-f9c>g*h^u}bZv-m$scQZAD^7;K;m)<AWLv42GmV_?Ox!~Kpw_#+Eg=?*y#}E
ztG*Hw>w#i$#|0bM*s)Nj&RHLe{<!IJxVNTuB@vY}O;C@lbb3&|Owf8J-65Toa#A!~
zBc1(9GfJ#~*VeS2+CRjw+igw0SZ-=muIS^|`LM*H#aI4)j@G-qwf`PxvSs5aE`d+a
zx3(&VnfX35zdX%T8{w{c2OQh9d=>_scZYpS#=`!EhC@lvjFu{&t(vA<QQA1CKUA1`
zTlZiSnYly#7(g3-9ZJ-ZT$cW~Pxhz$68W?Nli;5KzR(iT49de%JiWA}L}o*pq5k9t
z?nRqBx&t}yxNO#AR*R>6ulikwq%xES<H0;u#WkZ|H?16-S1i>gvMHy3Xj$qa&)Og{
zc(6O`fEda+N<{UrDdTQ2p7A)O2`9-QTO1uDTcqg0yx#c8D2HEqM8_FPaB?ydZ^qSp
zYI@|dG&lcic8rL4<aI@MaRp}DgGeaOTCNT&0O#j#2#U4=VkPR2u0G?2W=lr%`6oc>
zPVM@2p}hzYHpvcVQc<Ke59yr=QDYK51Z+(KR5hg|WlW+Ukcj3MAyq)q8AaFbMgL$t
zC;^~j3+dUF(ndyEK*xjzPw7G#<WHqT87S<rg2%#UG+caB6h>HY%spNDd$etKC%N5k
z%xZmkE8pBa%EYVZ#_Q`BQ=550S4ZEX2Zi!0Ih?(crcxE^Yjh%*uFa#6B*fX|&aT(=
zq5o#RJvCzgk|f=F=nF}{E*ZN~#9fJq`wP(~-QA0Qndv}@->j9FYP~l42L=)$kxWV8
zJ~=tr18_gw&>|ktve^M#p{PH27ijXNTJb;2R%HZ~95M1xyY?m_fJWk=X*;q0Y&B;0
zI$B{a+s%Y1+u6{Y6o0ONqi7ir>Cl6YIG`8O8_1bB_)g&D{NZ-#llQU(li?L1ieD9Q
zA2Lim*W1{-9Z_0Y8ou>P8b3XsUv%MODsGzNsr4Ih(NQR(nK_yi?Z%Ey8Edr!#;ryM
z?v>)wkw?OPPC5RPW?TjQMuoxR*VD>}FX2-kdNZxE?;jJZe1a)f^xp=kqNqQz3JFi}
znPmq)syXZgaaqjA0N9D{?%i4MA@GK>Xa(JLYCm7x+(`Xh5SO?)Y~HZ|w5(w~wm;Ct
z?d3-RuVg46j(<srK3hR0N(wSYUX<KW;K!%oE3aMTrQtbz!)Z4C1wfk#Ko*Av+rUM>
z-gC!!jYa&#7O^fGahCC~qYIGSaG=JWGc9KKx!ttNV-BKW5Q8MXPJU)yv3W+4qA07f
za&kn`MoATq&tH|)vrnZ>84?nZb)v7m&~JK@zm5n7hJ?pbtl6tt^BjkT(DTcfn5<u)
zn1ND)a_pAt=g&Go;Rw-31lq~V8_a+$xwzAhsP)5XLrg`*xDYMgUf$H$_!4a)prRTN
z$gh_$7eRvtawms7mS-n>XpP1+AnBdhxEmmhoE{{@a!CPW+XBA@ltZ0hYYKp^k9U9E
zMYWtgYT_Gl`BHy&KV#8w95|UoC7BSI9PSPo<An&Gt6un||9N2_*{zfQ)Yf|%Q{J_e
zsO<hp-#vuG5Ar}o$L2(@OjdHPU$qRb!aRqbjPf{BFA4I!9dd__(E!EcrGC$&Q`>rn
zv+@TRvA7p96OSV-G?S@9x)&5NWXl{M<5RIo<W^+wn9XR1J<@oQ>Zg*U{>-BYBIk5j
z4M=7!!1M2)Lj3qurwc!c_btLQT`cKfThk0txvPEAtD@Zs!L@*GE|Vn2IcMq*1}NQ~
z#<BQkOnCyAckK#s`PN*AJo;(lTi=ntHS5Jx`ilQ}-K9(hBjuA`k>Px4PS$=OZlcxV
z@t+lP5`M{=ou#hf@OzzishZ{9M(W{uU$8QMfI`qu+4X{7M(RtoQ|UGFp#+N6>zQui
zLrv4aWd3YKHcNQPzx@zsw_By)S9O}n_dPeU%e5@12@Fp@fTR>M`);mKyyh$}_{0u0
zDlXXg^vBlf>M=l_aiu_~Q-IrgwFeG}ps3Y9pXY(byRRwW9x<iZ4a{+xHs+O>!~n!G
zc=p~a!MMb$iVq^XdF0RS%%1!qwKuQJYSYBu3^6qHWJq8%W*$ArvJKnOXYP!HAP|M3
zFF5p3TJoL9Mt`*|SkQAbs5qWFvxIlFfZF7@uU}&>Uc0LUnUplOs>+WJ4nY9{@#qMv
zaL6d&Hah;$${TD6isLZKM8+c@Yx@FC=$#U)VQ#za6m&Uk#Jq!-954~Sl$>#Z)fMqK
zc%l><9DJ_$VUN+!Q%mlKS2#cL1plxoMc>Vokb=q7b5wADPxwg5p?9WFG~r|Nwk_w8
zOY*dEw_o$LhV(L(Lkl}L`o60f#O?-Qx(Hmf_L?x^*luRXM(*Fe`+~$v2o^MS<(F<Y
zGXzCMWTa=6iumyzrWCb2zc`F*1L(mwHV1hfd@^qbLQn1s3JA2%yFN0X6|+t{_nJ6V
ztaLe*LOJB#YRIlu(8}@i@NKvBY{{y|tdK#)O~OU$GwVdm+E2>Q%oOUT;ocUR=(@9f
z*>|LO8VZ^bD?ap%p@L-F!q^Jl)nS7RA^>^tAU{7p$SEd4tg`^6*lx&=D^w_Xln}4O
zlP_F11Fzka7u0wQ*Eb<>8xAFSO8lluRA$F^n+qZt-U{(Qvdw%YM|lZn2H*TQdmK%y
z3o2IQ@~EjHXs=o0y%FFiY-gRl-Hn*0vO4s1ZzawZiBfdS<_W3b+w3NF_5Sr+oSp#@
zgapmUb!V&9VY;%BEa1a!i{e!RCcLj6B(6bWN*HUPfQQG7gN;pM?;RZ-{qdu6HwXsT
zMlPUa-&IgoHuUb^V3V{n3*SEFTd*<Oz)7AJmLU+i^+a@|M9<Xrr3$ju9GST2LD8L8
zfL&?ow`8eX%eOFS*yZaz=`yP9zTI+h!P!SzIZcn%qmPls=j)y7jtXwPzf(3&Ys5xn
z%5^6Eyghh2yiE&YcIza}*Y%+v#Y#8phh+0?RoW19H#3Bf>Exrb-qqV|xPi3}i56uR
z50(A|({e_#z<NRSdkPkJe)nd*DM{=xdd=?IpB_sSbu3G#e=V^fRa%@s_01Qeci#*M
z<crH6wGnTM=sAdIWZBM09AM$d{ujBRUTawdVZTnbC`~L<yfmLsb@wcbkw06mc{QMG
zwe_*@#Wd=);1llDMi+Vd466?%v#Dc=9&R1>LO|$*bRG^eSQu<M-(+xO#BkK>swyZY
zEEBN&!%n~FfcbZvDfiEpqv#{6zo@GmPUDwud(?21oE1wGM9a!ammb%BA|jzmB^TEZ
z(XBTUZY5ExA03p;wceFE^dPEV<H^Y1|DbT3iJPq=UZBMOH9NZg+}?%19~<-D=E1lF
z0@Ykx`;}8Az*eOC1|(RI+BVAE!Fb3OAlysdD_!uSfc|6J*A$Zjq4V_o%xeDW2nYA)
zzxU<T+^&JK2EetsZ&Md+{+b2I@K}hbiu`*#C{#+%uwqU4*RSKLTz{_!-!!M?swqfF
zX8-R`5aF(!$O;t<b;^95|GmlO=J?;k9bW(c{&RE33Ik>G%#<rAG%(itGZ8111XlkA
znb!rcVRejVUHW?|vdb9E2r2^SVws;_TaOSrnn|P*<(euSCy&+pQ&dqpZ08X-#JVo+
z`uFYCUtpLqn1d|l+eiA{=o}4vYJU(uX2N~`{+Z|DEW63NBio6J><+1lYE<dz4Z}%J
z>*{O7_<w)<l8X~5ivHE&w}3;df%B0Aw$E{QgN)J;#+__hA;nWm^{D*ygzdH~h6lC<
zzDC2@O8nPj7aw(w(oz1qSM2Z%jHg(?#5?+W?8nSq%T@#vJGfpI^)QjIib#JKDy<CO
zw#E4`-deJUPV7f3$D{=RF71lGbhTuML+a0p4e1aQT?0|N3?iEE7g$9R<|iSCa`Jz2
zKeX8`Uu#;(*nj``Q98Z5vX1HGPUSfipIMdlElnv=JN#8;QdhKr|8{hPbOmZftwe>J
z)F>l|>)-n~{=WZ&?ZCR<m3yhwZY?+{+Rs9bSGdbl(s4^&^Y5L+q|w5oC2a0!>RwF@
zSD{@=M0#8JUl+Zi>xvUJwGRvmTrI);`*}};1gh~+)I?P)F`MR_$`t7eBQ_pv!i$tl
z$#to9OM7-1|Lo@Lg21V~y05px+mbRe9vFv6839aqlqt5$q`AN&h<A)B*#$q7=>vg#
z{JL2s@0m9eBG=p*kcHQIx=JeBBI%C=+CA`CQ(!jXKufnD;VIQ@JvFO%vHr0*hxl0e
z7X((qn)BJdX><AFuWePFZgVaZ%y8N=$!gBoogs(%&TU!&BSzWt{vwS0{Av<UjIq;L
zF-OgsJ38bgBz%in?iU?@bUfH+cduh|kKa7L>}JimY{e<NJ11zgo7#_+j)RQuM=N4P
zga@_lWsq(R#5;9QymsU~U3W=qjEY`cx%%2;-5gWm-r60e^XMzuIqBsMeqebM#ee?K
ziWR16jOVR!#^mbir=Uw}SLSVo_X>C0>SB<ajD)7PX3P76=Kj{ApRTRJ+A@w&OX6j{
zi}&&dm1iH_)ay4<;aG5Hw?~8vYagA%+>hcNGcz~8BMb->)vWWQABu+XqVcrU)D#%y
znB~tKHy6#d1nUPNPf-ju%qD}JgRbc|2iJX;mmeG`KKQY(SgwLg|GQ-LOk4B+5ck$$
zRd!#yC<ZDZpeP`%Qqs~5D$?BzD&5`jC<sVPcS$c|(XdDX>F)0C?%EUI-}`<0ob%Vd
z&e{8Tf!9)ao>^m#ao_itp@*Hnj}8{CD%1&_pPjX}_+TWwQ*3tNKys7;@lFG`EJJ+z
zI>gn)+oUsE%^U-||JN{J)OWRzc9VP6p2CJ;wa=4iwmRS3ylb&eC&EJw3{vP?w&PWP
z`S&-o<iITa-N-fB+h{k}(YQpi*ddT#_`o5GTK0zi?{CN;p4^(QU2S@K!-hhbQl<z$
z5yJ(NzqA-cLH&|PWNLe3=j>lAgWlhonSR;%`nlCfTgxV?<Rn9v^qZ?Tn$R>W7qT$X
zdo$I-?yv4(D^dP^`CC3v(JZ8+Dwd&A_{`S0&eaxgelJ^}(mqPIQ-9i3e{_OezQC36
zOqTxDcv7bEKa)fD%()85u?5W~wTA1aBQV&>&VAO^kiVkJ14?aRaT%m+PmFp=Clb!(
zg}Ba{Tj5$wlN|p^ktZ5rO;mJWg>^U2WoX^6dzy~;ZyjFTy~BW{kQ|+uchs5$H#!rp
zGwNdZ*Hw>rohL?R9S^S)B-*L?SmUFen!KK}QYZ8Gp-&<pipbOhAYwZv&z`lD^=-V)
z*e3`w<4Zn#mTk>Ni0X@~+!c8=SfMU{cCKEnN&QcIUx4;<NOvk}HyxVhfg)=(QUCrR
z>-DJaKb{Q<>9HShF#r7dU;TKCe{cW%uRizBQ~&QDgZ?kCXdjtCo^3d6m^QSCl4+&N
zEGjP2J30u1k8N&Q&kC0D#O)vK|D7XD2d)*#KnNGJ=Eb#~E1KBy>Ot1_hv(wtMA3M#
zFHLL}EiF^OeEW_frdkt&dh~Y4tAM3{mHUj$@Sm!`cn(JKFaAeD`sefCAN(Kw2>q?a
z|K5_r;pKlJ!PoX>ubMvc>BxjI(o7%-pgw-g18I||gtoVLM>c&4X^JWzQB-?o0c?p}
z=l|AE|GR4VkNy20?)3ku&;L*K|Nm6n|KsD|_W%Fa{O8c^B>4YoaBmH5H!VE;x!GvJ
zdvJu^5YRZn0i6Sjd4TGs1Zay4xY4Xe^Ywp<bd-pIP#g#P`)x-w9k3B#!{mFAR_+Cx
zN(aGm=Rjg!HNaERu>XPKNM8*UW4?~)GGzmaetn`G4z|xcCnlx`OG?K~K{B{l2wamf
z065#Ny1WKdjuU&tf6AFHqaN^NCnNhP&3{!=X}<+C9)h)*vUTV(wLv{S(!9=xy<qhx
zgXnGI2uNsAIsd}n6Hwxb2GEOMidu6eh)GMIShN27&RVeQ#*Tx~E9&a<vRVb@-zEt~
ztkt{F->zRE8_H8x2Bg-2cUZreij*{UbRu5LVD3g5Gcd53DzjWXi-j-<f@(8(=yBOr
zjI`Y|Px%nN*tOFKJQc|a*Fz3mPJ9}xs9Ly$7@4m0(Rx@MxLUX`MdxciMmXf3+;k62
zODL38&mN8=DM;de{B2S^2JTA-83>Sw?s2;G!=^a|H^;;uMR#^ekdu=;rSVzpcC;S@
zbRw15mBABNp7T!ca|hIQWRH?XR^(LkhwAv!Rc~d*rRuT5T?8lRH2s~3NBop8MFF4#
z0{4MEUe-H=p5u*nzXziQxJE;d;dQcK&b|pIP$jjXD?4&Z$_=o;NzosNu38i8-|qMr
zp;ZrV=K{JZ2Ut*jJ9L#`gQK{emWoO=7ufX&&2L#*Suw?R6{*lBQXf?x%-MJuSaJ|D
zot?HWZhd(Ca>jh5-Hf|L_duUWk|(=l(^SyQhlZju@`3GkT614%wdNNDXMA@9A3FA)
zt4L4#Fv-R+jrj?26D4KGoMo1h?)Pz4<FxApZtB+@qm9fBL_zXB$Ry9;k8vv-f>mNA
zARZEWODRz)AJ3Uo%jTWOJS4wee9;;Kl$zK&ZkWOA)VE1@Hrw6X3-O)#{FOSd{Xw-<
zraU-AFdy7mmI1VYb#NLw?Gc(9P@;~2q7}tqZfcr)_wHS=>vy>=w1=juuI_fE<E?>4
zqYi|QTK(ekj_eX^Q?{#}OzlHxxCBp3M3U|}nvy@}S3X_r2H5c_H-oX&W8<}5yfhY`
zv~qr0I`k!EOrXby)R*#TebD0Qtop3IgW)<=R-c4qYBICDdD6}&MPMIBcA^l6!M0po
zFb3+r;sJ-ND)|p}VAsd(Z`qGo+bpf@P(G9GMyAt`j(Yo_wFHm?H)Uo)W;%bJvi5FK
zQSs8o293W(gqsVx=(0`qO;3BfoRU%`8P(gj1#(vyMx|;rinEV@AHWu+lRYdQfH;1e
z>%ZL=?vFR?o!}E}xErlpa@G4ehU~~pMq0U<MIr2mrI#8VS7Jn--5%v(V>ho_shbZy
zbj~@&JZtR+am;@{!*fipqU_Wv(>ST-)r#TdL@iXa6UcdS`-Y<r0=l8LETr=On8d@<
zRmbab$o?!gZA_AcoEj&OD<&t>|7w#+_p+?-?5xe1-wp#1iJeTbi7j8LmAeGfL+h&P
zy819_&r19ec=`BXmeOJcK`eV;M@^Mc^D(69Y+Ntrc6{Nw#c3~5SAAS`nTNaVS1xL^
zY`pmAhEf#2FWpnf!Cz7LU7sRfdAWH?iMSd5jYIsj8bS!O`l1$tl48BCe8#y}{I9}=
z^gup6vXm=9ss`j0)T$i|$IC6?xLS_iEO9Lx_ZFX{JMUqG>KK2F!Z<(6*zkwY9nemr
zB`Ohm%y|_X`lbG`d~yQzjg%99+RH}^Ub@q<9D%WiA(x%{k_5I9)7v9Cqn73GvW8%|
zE}FRrc65awL$IjYc$krYWPFR?KWO(U@LJ_BCpC+7eZj)^97dhyC);<;o##{AxZ<O;
z;?&s&)UTY|G}`%w?LI&rM~y#C6lpj_w66-^)#Yo+*GDMPt0EAUH&Z8zmRv`5H0MU2
z2BPJXCd%LQx>;=@ffU<iy^olfG`!-&KjqCS#^9g}2NCv#lIIVZgS)7Pf4&kB<2m-k
z<ddm!KI&FB`ufUjx?FpPUL(!;&Yjq|$rdr?q1i-)xhM2c)yMM7UH`<meXsn0E$jNt
z$@97-_u;-U(aL-yovGnU!I+*p>WmccGLyU>HuJuT0R!Vrw-beT5q~{QGIbUcjidC6
zEtg*Tg!tvohsz2M871u-giPq6%196@odM?5x2`XP^sq{{tqFLb%ZXD_wb)k286*~Z
zJ#p~M8VcUi2+n5`JwFfjvfhpT$MK8F=qc?2U0oR*W4mXM_by_uV<v}O)H9Bq>f{B5
zs3)+V=4ZaK7bn!E;1bM~b!zxMN-4vR>wCudc==_luqc=K%2)O4YE_ZfeE4v5mt&Yt
zSbzE71S}E~ZV)xX1NW;7p;V?lF(>q^6$xU+z_c;R^*0^~CEMb19ekFq=v;+C5<=D_
zG-B<N>(OGqlFj(HtC6k)zv8}o+O`HR9h^|4dDFX9Xq<>o%EvjjRTwbqXzfSlv|5N3
zN=76uX6-`zw})g{fiUpqb#IS!Xn0K2&ap~N8IP#k&$^@u9lZC+5LB-dwF3UriA_|_
z^4-@7oqvWD=6hSt9d{+c`wur%gffZ_rqG;DGSWmtjg@nj!L3R)T$$S|)8Pu`E#u0k
zqk@&ZIDbpaiUsWOd5-b~ThZ8-n(J<X`od+cLmgSRLlqlw7Hr>yWI(0<&b6BJYL<GA
z<Cat=w7kXrrZ_mbvebC39`$eG?-YUbkgRgAN}E>n9I}dQ=M{1l%5Wb=2)pC>pF3<X
zX7^LG^7{)aH#bvL@@s3Gk1K6k>6^`@hmDP`50wSJ_VDCYdG$|q#BQszvCLPa$Pe>&
zpQzG4I3{{C@}h^~OnmI@%z1Sj%E7ibZY)5nmuJTeA*-qV5`k4*)ZywL+m(SEl^;&V
zYts=(oc9mWQlJAA&Tq-0bUpB5CxUG1YG7wPB3v%x;0Cu`R@-7a1OM}9=jF&a>R*4v
za&*F8Oxc~Ki?y~7cky5Tig$I%?&-4Ou-%z-8jlKv<y1DW7fBJ(Km~GtYx>C+yj#q6
zfV_sqy4OyuyKRVE-oS9~o4^VW<;+~KzxZ{(4n!N-G?E*`z>~TDIb|hfyj`z>6f@gR
zLWhpuRUr!E78mYk?sk&;qEwcju$Q{LJ)KEWntF65f5V9x@ehU~C7ngKRI=c;O!C%t
z2n;Q<v4xgONNhVW>kP+INZTDax;f~{*_GwXO_f`y<d2rySy53Q_2%LxDF?IVRVyGI
zkk2kQRz|esMsT`H&qu9F&lYK}9NJK~B*f^gH*3QcKnIqpx$URp5OL%VqZ6!-6vC;Q
z$o`;3{Ntrb_ilATYIf~n)g8<m{-!cRD5Ecoe*lyFWYs6Z<@vIKVq?KuuT#~CpAqT2
zZP{Ka7Vt7N7NbSW(p>GHjv*K2ydRI~99?p?PA_TPP8Xz|U6#~pYQXZXhYxei4GSbf
zgY<RRWz;Ik+BO$bKPzizvy)UPzq)hhf2QtV<q!+~?)ZbLReVV&kdM2@UR0#{crH36
z7ODjt2mjaY(3H8m*{Nn?FvUn{<(BR5ma82N7hJGho|~|qMa{6R!0Sr&Ztf6uoo4C@
z6m!|xb~S0*%_!a*tTxwVQ2IxH4e$mtC>!*b$*J9lA^2|HwtVx7RnaI$6^8dTzxVgM
z4^zd4Pay1cW$)>~-;>omVBUr{&MDaHrro{Fou>c0XzwpM{c6p4ahj#{@w>6$N5`t?
zjeEr^g9ZAM7E%AW1sKw;N`M(Jl{vgMotkugu5Mz#)nAyV3p9B_VIi1+gpQ_-m9pkw
zk(a9qvgSuMrO4@|t5khDO(=tuQkt5EQW@Zq51x0iCvszi9#=}aO_U$l$3)0|F^+Y=
zc8R$Dp?>S)-jMSOgrvMORw>RncUViKsj`>-XjxF+@mE8t_;#yuj_!A<9~~}IsUgf<
zLMGU@r!r?oa);su^s|7S2G%%afd?18sA)*l-D`;nZ@xjbss$p)@&u>Ytz8}R`cgs5
zm!)zx*J|0$tj4PYltS6sKF3<HGHcQ9)?R`csSzLk>$J)0>euBA6V-yi&%b+reeJyI
zy>WCJ$X-xLayys)vMGfpU>%)u-I)}F8>eFE;4if_bCW3zH-bU=tKKIuWJGPPgQ1>2
zM_bS^ky@PrJl}a6IS}8qX1d~r_$;<-y5^pi?d&OvCd-#_EcpR4(JKPhUUa(HH*A%W
z#>mLj%nZ^Oa{Y%FwO8lH1NKB>BE4S?5=S8O7iqxWMu44;c=Nij8g|lfIlg{>U0aJk
zo*$8K)UT2U6Q-V=kro}}9FMn@v$5YAf&$FA0jAqw(b;2iyXl99a?Ob>qW%v28`4{;
z>t(Rixk%V@r($xG@(wsrV%u~=ZW${5^aB-Dp(U1;g{dr6-+U(h=>g7q%^3fxid<-|
z`*Ita^+KhNsHm~q%?(wHzI<`(BRA_Iy~-^6Jmw(Z$Ggfky-$wvwS`lwAiE)TnvvDK
z2$RF^&M`62%PqlyhjK^L*ySeW=X`qsOg7fGZpGkwWmSuzYh^TV5eQ=SeKc~^_2uwG
zk3Y+o=?`5q4l=yNZW`h<1erKT>(1HEx1;dR>C~|b<RJX`&X(9EW0{qQ?%pj~Jtkl+
z(ad>0x4kA2YHr%6&;j!QqILovtY&>!HG5M52g;V@ATu~<clu^&)O208yex9kNhiee
zWX2;8r)IJ$t`=a$PKq$T3$w(G%kv8-Sj9d^Na(M69W6y*Jc*SR{<SsNFi=1p>tnPb
zOo9A?f*`Yq_>h$M47Is~ot*~?jJLJ>{p_!bi6svW91gZ5wVN^-jUOq)G!>2j&>?%R
zQEXlzbf>n~d1=F<<mPk_HMeBnX|E%jD>6LI`*TVx<EpXO#aYDpWPMlMYL|P>V))_e
z4WH6=-TASSoeqFwjlRB>*?P=1gH>^2RmT(H^_O~4hVK?0jN2_{O(*gdf7tjtX&fOO
zK)@YGtC$UFc)8V2$}I8{kJ_gPO57CCdGRM?%Pw7O*p8}DvkI)OISBFjH(W&fE5u!B
zn10uF>hMfvCvXk<$=RFd3rSyh4v;IUT?CViWY%5h%kbwkYY$fq&Y9mSZcD?wIEffC
z9lj`rXgA&d=3cTcZ`<z~F9l_0(5bx^bzYyRUm`Y87zrwh?TW22_1fP*;8oO0KR$pB
zwHJ;JRywn4PNm5rns*svP)SbK{G%KJjxswg!Y<+_tgLe*P(*tK;wnJvw(1+p?PX&u
zCp(J7`A=^9T1NJ7S!E<#xxj~H>2%@4aPj4~ts1T_2df}0qI7+I<DugO;@pD$*!>M2
z(SAEgf+#2C^)-461#S0QeIni%+xiltQ@0~E(fpQfqI+@8BBAc;HO*VGx35Q9d3f7}
zz1NP)J(Kpbtdqn!$qqZU3w!3ldiRm9yeA@446Lj?XmMCkFu~%4!Rgc%9$B??<=twV
zlR_HEwYy1Mr6%dtkN$4`TT!e5jT-TshmkeVW}PCYt=k~qtEGljpV-(KDp8nLH_p9G
z8pln0BF^<=6``Nekra@2F?f4<GnM0{RKMq6A@*@25HgW=<DXGc`Aw8g9QD~&Y4(ML
z7!PEn;19vp`kb)j<XG+MKWy<pt||b{!PHZ|zr?W*%xdv2Zs=6k5sYowOP;yax@(rI
z`CwsT8<xUEr^fMF9e<9q8LMOz0*FH{O2t|B#Ui+;b+B>)#jmF`ym3t-FZm^45L%|(
zKH>2iBhjzw^<ec`9@5BFR0xE5u;T9M(8+HxW70>MsQuqMC6*Q^mY^LI`V$W1VD(ev
z{rKA;{T250EACqk;5}8vqc(0`%FY(|qAiy{;Q;Kt8+z5b*zLEfTgQ#rO;a`h;iQ4~
zs-s+yU-#=6H`l>J(STala}?Gb^~}kGcm|i?rj2%ko17ZqU~B|`N4fa`?!k39d@Prn
zek(Iri1&3dbT(P9T7#b-Tw%c<qu`j(xHedl$IVJ*?x{pVbDg@g^$TXA;!{>xrAebI
z<+aEyMUlj)uc(~qY-h%6DKjBWffy1W)-N}a7V{&^cB-jNq=w7o;3^6{l#1^SN+T<E
zARVN4oYs2N!+C2eS*T!y?{Qnse>ZpAlFS>g6@Wq&0Qb4V{~H(1&ZBRsrZ^^1-|)SM
ztj7Jq=d%|itagX;erD6T5(n`T;_~GoH+|D!4}QUW9DjN<6=})W4cPMrL%yA~O-)I?
zu+%-ZVw)(nnH#9`#P9mtaPU#^jDa*_Wor*G41P4vDSO!=JJ~HY7&c*A{H5y_BrQ6S
z^bmuJid#6TCk}3bLp#{uFg++p7n@`^$rgwY6H<x3=CYOmAw#Uat&hmYec*fnB{Yo-
z@QaD$#C4ws9b{n`nFfY!DWv0$uVsTf)w3{%b4Nx-G*k3Lo}M54et)K(w>}KhYvOwm
z@rTaM+2neh<EqG@4){uOL<^*9f11T)>I5UkSsx~02}sj!IaVuk@^>FAp2_=qvNCl+
zCK_eGIh);o+kZ&dn+=XDXbSI5?6><%dxB*@7{kT|6-RQSY)ACaH#%B$sjW>lE>UU<
z78$MzZ<5b>u1jhi*R9akriv8~O!;j^G<|~Ql+5#gdE7mkA;w+Hhsx-dlsk=wN!Re4
z7vMwZwv9M$=NmiYB^F7Mfh3ekV#8%A9lOG%?kB{e{XcOek3i5Fq-FZmYGt1(U+l0s
ze2a_$4>=+{`c83U5Wc+)8TsnOxw*1r892CI9>8zueb-ctG?)SFSeo+i!|e<{^!2qF
z?N^y;ioy%f(d9fgpd&xdB@$B9)k|f(>7+fKZIzRgk{<53E!kb@T^-K4JS*96Wz6HT
zmcSkHr61u~P;L?p%@3f1L`+%O4>wby>du8o@+Q(kG`Yfaa4qDdf@()gB>5=9@;B_Q
zkQM%FgISLvXzcFDP=o2ry6Y7c{xcSK!Cle^6>(e0-3X!kfmKY|@_BErYzN-ejFo2s
z|7$lAw@_j|act8a77~`aO-iB+I<9iacX-^zeBd6@_;Z47vHYRzXXhtF((D)t(Md^4
zW;5Aw+AfWG7wbj>TJDk?4m%;h?;9weFlf-J6M{%Upw2P`ODQ-0H1L8mL2Bw|8*%|%
zeWSYKxW4>qpEZuvZUZBIaoJcm3Pa8I;>i&m#bg9oSQ?<F{nhFZH=49?X=oU<<=M<b
z$atX(y6vgY_x6tV`pmA*MM8-$bPBEaxAce?E)rZ=pqcloK(u(VS73ma`$vu4G`XJF
z<qaRHJe*!zarq!2G*PoeBO3uBDp&ljaV2)X+?v<l`RQrJl~B2|<zUCdDJU>e+W-C9
zO^;5{LM1FsSciK(&%H*sqi2K;UYoKxInREV#&YUFCp<DD{aBWj<s{=9?CA>`7(N4-
znKtItQaHuIwc2GL`14P56jl^V^y11_TbA<GYAfu&nwz;D+oM6q09d%$Tt(pbR(u2a
z*_rovQ8n1&33htc?px4Bfvf%REsi!6VO=W{p+E;3@~@$f99UCOk@=~{nxzfH&+!u_
z^`YVTt-TRMX*dejGxf~9IYQ(F6OFj>_k@x0S`9owzPM0Uh>7UmI6+r5{FK}b_Zz^C
zD~nv4#8iSq11kjOx076}`rz1)>Ysr53V12cGq)}^y72;8Nz%rn>m@fo&8N<6jojrU
zLz4JwmS28Rbf0DaS-2H7(&Nfo=ND2f;B`fvBxki*_H;FaIdNobbC?v7h3$Tfi%TN_
zpL}X5!yly0kXVV8q;PZZLK1v%N<+%64()VCYt9!IN79dgw0u{7^YNPOGFQ?XPLBBc
zh@!E62H%yLR!k2&6O^Nf)>1<SsCtIQM7BxBO%|B42D~kEw(B`v6SiaQ#o-ZAaPjf-
zAsT@l9%*c4Vo77+jg3C(_7>fK&${dE6<oNCz5BIlPg}v#THl%f;bDVb(J2(my5MD?
zSf#e;1OpeZA|cJ9+^_yxFA;lCOt#jPh#NIZJ!dKrs#TMAkzKgl4wh%TOfo0AFB^;Y
zrUc^q_(bFk`*4>k+gkn%qRb43k$)8AN{s*YXOQLW;(Q&kt=YjU^u11dZduzxB2H`G
zD$G5WM$|d`q$ynuaba<DjSp!Wa<TdLkyr#q)D8V=dWe}b)S|u457GARHws6XIFy=M
zigCq;qyX(nwg;ZE@lE?$CVwjbPGJJ;P2fni%fsqbSbpL8+!F3{tS8A<5L_`%y0)Ji
zYB6eQ(;bs`gHI$#nMrYE7eaYT{=s#<!W(SaYKA}D2`*sHWky>|kjrhOk>1d-sor}`
zM#HX*Hf$kNF!akLjt#_Rx*S0ma0<7PCByAvDBmjebxZDOkz^=9SvX6HEr46SqKh?d
zh&^()IL18RMi{j7Mph+Wp2~{?r*edi{aj_+%G_*7qlZ=_BmD3&fE6(}*hQ4-nI>n;
zc`)J^$xsIY>zfmv!amj^7A6gKOv1rwojqeh&74UcJXv^IUg=6l4~t=EhgywWCPkB3
zZ(0R<sK$F2B`x>Qqco!5+})j>Ruk}9Pn|6yG>kNcSd}2?G%+#eGUk@Y_0VbMui!95
zjp@oo*CokWTQ6~JVREydtmTdD5WpVD%1+Zj__xiMF$aVGZIi=Rg_AAroZVO^f2b;R
zL%uG5eqP#eYMWQ^u9i<@%3&@ovO3+mMn_F=ru0>1Y=M3aAebw|9NTJ$E^L((Bbkvr
zE8_>qFvTMgGf9bl?TIp6)x8R*nLA@d7@BhMpiXjAXJs|F^^u|cwC8y<#Va0YkRy4v
z{>KWX>YMyq>O4{#lL7QsDmohZHckYf|NDSwr`ea7uT2O`PxGW{<fP}o^zAkbkE2&g
zL7Jfph}Pg1+<}y5l;PotDOtOgtXYck_qZJT(QBNu>>BrPUj0>GX5DTZNXY9X7pg!>
zHo8@a+RPCduCZ7Kk!CIfC6tzgU7+2=DU`W;S{STz*3}i;pVTR5WfX-hCS=KGc_YIY
z`<yg%aIWw*MQ-i26F2e59?$z0Z{7YAOu2n~s@tCYirQdJIm=_skwAW2O@}^^T`-ir
zA1s^v>HGY?I!Rb?%Zin9ZR6ZDQs#Ch0Vb3J6$kG1jI`c6cpk9U`~t(L@v0@i#Kvj9
zp3ckFM<D3HHVGDcaK!qz5%K)Fv&huFK*g5SuKsAFoZk;n67bR>%OKbn^0cx-Ph4~4
z>NWc2w4~BZzZ;E^aHV}Yf9UqL@u&wQYBP^b;v>MkC3^r;wKc)EsONcFuLr63ak-3H
z*RWqa%$uoE9%GBU&5^b2s}I97FW`j958(*d;DXmfWGX5;O&{LVQJ`Tsg-ie$L?jd=
zGu_2Ban)F`PKdnCf#dKT&}g$gtSxRFc=e?#C;3^l1S-a(-ljaY)v*b4rJmt!i;P=Z
zt52&6b21rOo85aR26eZ}I9mQ(7ZzOGmm*?KbC7Ogvsganp(!^TF_Etu@XMPBPfQ^9
z0Sm;*z%6I2xvKI4&L@M=PDSb9j4*C!9r(T6^bS^j^8}1U5gFhx0X9cdXPF$W%H(44
z-LsFm9Tg<@UGo=NNWi$KPt7Z*6V7UitZm2<N}rV=%*Qf`Qz>qHEu{?C6}6fbB7vzU
zYez0V#v(h3D}?0$((ae5mCo-n-lFv#^=rOre+<G<GCOLDNw76fR^(H1e^F5h+jYg4
zsnwgx-9a9rFJk3Xlc3(V6B`!mmLI;%*(ntVzrJ@jA~r0?G){8Eey492CG2pfr><W(
z6%sO(>DdMaspmwFyU=QZm-mw<OhoN;i}(R*Q5Gi3B)UFawEDo-jDS6StVP%#lPF|r
zjn;ugEA!fElPi+xD*<4oHM?iZfc2rm(F)bD4fhm*1_`RgVj&s7oaZYaCxm`<6@9aQ
zQ|vG@C#hv|ucc0Oefxttb%`PL?JnK+c0NJD%y&k;=5R(`^8mze@5g0x+uMRPC5BT|
ztbx@~+7fbXhMAh~f@{9X&CiQgov|VA6-02kzfH{wPDE(Cdfip`F>`(Sb(aW5V^c%?
zWIt1b&KHzm-=&SfIJ{xbpA1QE-jT#daJTsAn6?UJ5-c1d8xz)Ih1d1a*knK`DzcX~
zCwo}f+2M|zYq1gY0SOdPo+$-8;kXf0#=>$u7?~M!n48aom>GLDQsevOKBWFa3G`b&
zup@XekxHQ5>3klNdqDfh<K^G4UG0}x)gHaAaz!@L1*@dp+I8Pj6xxf4#+eZ5B$|^N
zjJcEdsp~Yl>nrYGLy!DtWi(#P^Ir!<y%yRV+BA6ETj&%UQ^<Ia`#}NoDdB{<5OaEj
zdk!gSaS@7W&!3L5>}K>x=ks#?69n2No|WXzeT(c5^%*9zNYz~K8hLv&W@F8h+EJ-;
zIaW;U3rkXW0*((-5H<ROh4><^44;&3Y56rCky0=uwUM|r#6wm(r_RT;32qSMhS4xF
zSHri8`~7}?R)udzHll}xv>yH*?!OxoX`*NH`8~Ysnh+h$z`~Z0)U_t3X|M?q&hpFV
zz-VE~H>!A>EEgO|IQY<i)Ilq(6V29c8G&JjH^1PO8bgI7RvO!TC{d;nJij77_wi42
z-#NJ6;qntPX(e=j$VuyhUSqLPnvb%s;yVQnPBr0ZqO3a~OK~Z}F7V62{umy1#|iF-
zF`MnXl=Q;8D-#w>F!d;|V4GdJH#&SOADsk|GafwP87U-MP(SVOYg$w4=cy_QcVpbU
z#IlaDp)%@JFZ8?j>ec4WA!%ChuD6N0bi$l3Lv2Py?qg}lghp$9FfpWd+Hp)}rcy|j
z)5)|BuMrbhV9a^_v9Lhn*~*o}i_yYH&&1k)Mi=}oBOSc0{oBJ5a}fudakk5W1G;+f
zKA&DAk=1kO!BKjq&qUrL{nygMr_a<*i%j-a$|MN`6MRwlO(*JY#OvUtLOcGc!gmTz
zb>HeisCqRkSr3+pc=OgwI#<<Mb#{)&n5AEZh@6trhQ@I^9T7jA^+EfbhK;*mY#_W6
zSsB(>w>{ZKP)93}u5_T$Vz1$5Phnxd_0hk5k={`0y^EUq-iMas?|fGrXwG)Xtgcl~
z;Kqwe34@#NGB%De75RLb{kd!#XCi4SBRuUys0Qix<y!F2i<9~25F}a-D{oX@<0jak
z@iNhXMru@Z0F2eWwd!EZ;rewm6JCc8=Hj1#Hi;7pI~oMq%P^7E#rZ$NZk0A>$Il*H
zEDL&lsINfw=fuNP&2%C{MbSIF6kNoO+ErT#`pC!aIl|@er<@6a7HhJoqtg}t8;PEu
z8S*Tzd?2Tw)_f{RQg|_W#uaFD-Ha~OdGU4M5A}f)`_lL(<!OlpriVttj>=Or3*&Yj
zGc{||b26nMLXYfkcfbg(ai45)1s-UZTuHnQ<GcJmU&J)MLHzNQdBBPDEz-_yx8#Q;
zC7@f_u8Az>##9r&c&BB>NIIk4+E=g(J>+zI|K$FEzA#_c(>x0Nn5V?NPz<k#iEns9
zB8?$I5FAyin78G9>S%4ftQ+6irGmJ5S+_Zv$L#*81?hq9qnBLlNw_CO`6v2X;ys#Q
zEF=gMrKEWHVaxbsdAxD`nQrROsj&yjwlYMtWZ_m)6coj)Y0KExNm@>_kiq8CD2LYd
z9`4}pOgEPT+fO8&=;(H3jG=9g<?2tMAY#<t4sqp$3!SP7W^8n^!S_#jX+7+eEe6k2
z-pNncpC8vOMnvcs>;CD`5gvKLT~?HD{ZXIIf|w7Kgv7^Z7Zm2()|w`lNXR5&usP3v
zGcRKJ0!$Y|S5uyWEA=6$k}3|&P=e;*#Ws60ne3s*ZU%hF<m6lvWV*UcN2xtBFa=ls
z3X-Jbt#3wjtQL}wb!oS!0%yF@Z+q!#fWi0XTg&3V2NOt14vl3<&>PIg?W8DVJLXD?
zmr1<vDn`U~vtfNQ-dHAdcYTc`<Q*SW-Ok;iylQ%&4cU@X#l$2*Uf}%ErXTgLy~Q-m
zT+@$`-ZHbyc3jt}Fcq+)riZqXxQodtMRv3O$HTsS)X*qM=7^jUviX>=W%>ZuP~W6$
zT~=U2kz}N$ITn3E=w*aRbln$LgdnTMeTonfmA&PSit({Rwtf8K6T&nN)eA8#)?L9}
zWN4b(9#uy`V7LpyeMU-}U{=PU_BGH8`Dx79WA$>cl3(*&L_O2d9L;9qz8})Rh|Np~
ztivvbE?fAo?na0;J-$(@9!KaTG-ojtrjzWPta{)_A9a;;=`6?%n3MlWnzs9@7zH0R
z3b>1iZ&SCoswkq!k12dAn6Ag#d}Fux!W#0r>-({RA*#XYXtt0mZ&5LOh?eVVXG><!
zs|b<XzTfNd#$OO`Z$CRsIzpl);49EdXVJw%nL*zccQ~O4Z(HKII+9D+J5V;>?pWb2
z(^2{gCA8d$9xKC!+ii=y7ezU^R!L{JKknb5qNgEk9g0cvIc`ZNPRHTm;4ZRWgq|Qp
znop~l^q<!)CoIvtp{85t%1?+M&NzSYwX(Xmkhl2mWg8W(?i`Hu#L)Y{yT9gdh!f${
zxL`+0I@<CrE!_T-&88V!@wV8EX5Y{M%zeW?p-^&jYjkh!u7;_Zs@oa0rt)=TP#oi%
z(r^cvdj|6TFZO%%AwCZG2hRjZubp?y?#5jVaZxfq)A6XgcwqDFS$Af}fKik7r6nT>
z(O$!+6_<Atr)+*&5je=ai5eKFLpPam6JL4pO14kKSNeBD>(8F^qb$3lY&PHyzh{<W
ztDYg}@>z<Q1gPaLr?SSH4E9gsc!@P_^7=%V=;4A9R-0&rBE+u!s)c4rph9jg4BwA7
z2}NlBDrGJffd?@MjXg*_eP!7885Eb+RPn?)BZUkU#ER%6s~^8GT@9Nr4lu72zGbS8
z_;{7Z^ev}rqmDlHLrQW|=EG;|!|uF0`fnnNbl+JZl_l^AjG4t1PtjA8N7P}EG))hN
z3)WN(xe$Imfn02x0R#zbk+vp(_1+}lL;8FYX$umtdxyLn=~BG4jV>7yhUa4B49Z$Q
z!u7yclqI2L_<AAl*zM#hllZ$hv^4*T9(suJG%9hrmHGL<p6tmsc41rWOBRpQfXDPx
z0S#bQgt+rLpBxy=JXG@5{@l?v;F_G;Oas09YQgLQoqDe&xwp0b5TP&|^Qy_Fy5gDI
zD=7LSLrsn-`0>oQjqLgJyS<ZLfu%b&f)8i(?GnS_aWs1-%6$hz>=lvGe<lT5VjMP~
zU&T82b`ImykdiWqi6O!xr9_-5kmP*7;|c}NPT(fxl-`iXl_9-ttiwLe$SD12Vn7Rp
z?YXwtSKBO-)oqP8cD}y{YtoRESag<S+KMOSIH@zLF4=io=Ob9Op#PZLyt%hwIx=5U
z!-|nTRUzdiU)|5fJ>Ta_6ug0YOj4cE5(qg_IR8lT(x8!6+}pBoj&y6zv!<q2GAzEe
z2~USgWJa9}Enrp5`V02JjjxUj5<w0$Xk;QH^wjf?ADG&_VihOM2sh7c4rOA;)QTzQ
z@tdqU_tx+nxwsVxObqly{ggfK9&%YA-RTZGJUE;caeZC4GuGm-W_)eV?e?HltxJAO
zT}QD4wi6pofX_seOh=Ny)iDe*`BNoErP2JDgdeiy`ntLlcLJ~89T08>`a(JuOBpu7
z9(0tftZ_+Ov)U$W0kNa8FAnO;@>HGqzHeprFd1lc>52T;z<&4f5!prkK+1sy8;dqn
zVP>dNN>(!ZY~fQL$Qq@y(AwG4(l@SVP+jGDPq_9xAfbLmrP7IplJxbM<CcDD1x9U`
zqy3+^*X)ijy`9M*$dtCzSLGUQBRF9;Pp>X4KB8+ny29A%0TZ64TG^+g37b+1vQkcB
zcf7B2UDX@B@QFaf<t}bvDJ)`jH!UO|dar55ZcKr5X1K6{ciu$%(dMN3%@n}l3o*yn
zw0K#nYcK#rS(ukPBXXWcMSM2@Vl8|L0@|J4t{N133JPd6Cfd@qS(i;ac)rE|^L@%p
z<GdS!#q~}Am-kiTBFovZ2o|Tp^Vol#Id1V^9-qf^&nr@6uOSf8_%C|PvxRxP73>=0
zoUF+BLINqJj3sx8F;H>>Oh#Gpr#t<Fr4p8U(`BaP*wV*WHiC#P!mXSgS?c>+e{6c(
zT|`9|a)u~)ZX_|hrk$V-lkaeL@g5(`H)IMRwEPPSuF0=2pd-`<x1Z73s)+H1?y~n+
zB43y7+5Rjqg~i7xnt0%{usVT_x(!+8nypl{WGvL);qj)=#jqteHw^(aj`MW$TvpVx
zPSTRt9`LXJ^<|idw>fZ8*BTYoXnhg-_z8_)$QI*+T5xPO(pgh%Wi8z%f&I3tX<mpJ
zS$JDVse590_87<dvB{$G=l81v19wy$mbhGW7JNHseowYO(Iy~RzK=YsW3$5gQ;h6%
zY%DY;{DYW1hP%1s_9+h;SLHYd)~63>+7u<b<Ny^H1klR)`^jbxRj5TT?0gD_{Joq?
zIoaWx3qsb`r9Jb^OtSY9hn0?CC%KYC97dc&&n9~vPaK9g+E`8JBSQ12BKmq=>Sd^#
zfI}$=kjST`=o_BMd4F|9K&EHdZFC@1DfA_k_216>4q=AOVt*I<sf{>`F7>tYyeP`s
z6c`aQVJ$2!NrQZj(V;{EGhl)-yw;AXhvdqJ>t|{N?3<HqWLUX@fvt1{iL-fPWFxkL
z871cR;lp+l(Xe%V#UrOGFn&m;5+zXpAF!s^8B8fnjw?)|d3g=-<%p?l$nTt)E4IiF
zG+QaQ7!Ca{><oO&_*2ZJy4+t({7!+<Yz@*4w2j5YMB)qGw*X*};z7{^%8Q(Cb$23m
zh5z=U#5|z(#MH*6u{plJgy`s$RLuUD^Ir(fa`&w4^ejIKYnT$0acyZ?Woa12DaxsB
z6(7!H)mA-eNH39K5*K?LE@oBw)<A{XBGGUJKebA!TK~Dpxn)ry;^f(gCtynxYoDNN
zKIee6P6<D;4$G3VjC7xe48a6~;=$(;Xm1%=K4_)9%t<RzD`Em}*wYH-&l2`eRW(1r
zC5V+G4-afxMvl7=NR_GE=xBtM8d1MIG@*HS!P?FcW%Gvqo5>g=4Hs9(_|w-MFV3kc
z&9%k(00nD<xj>kS!qCcA?5K&7xB+C0mstxtw;Z;fxu`_k{i!&<YsEy}QaD`X_k!ib
z!CHxJEM1Qy4xC{prKQS=8Ubr|nP_hC_#Pm^L@*Z(8i`~$x2*-YU&6*FM5~ezv1Jj(
zixDBinj|DzUEL7hvA=#nfY0V=e&Uf@iY-gk^t_Fx;B7w?q<9iv=A!=OKJ*_`apds5
z58ssm05q4|IUsoCy0P?68veq$gJF-4bZ4@7PUMM|JveWD@(DF9>A`Saf|S@?%+y2Q
zwhlNBNoZo*_9}-Kr)BuH7I7{<=Y1FXX9~uwQOw0)N;q<GPM4pJcxj}}Hto3UcGWj3
zyong%<HvE2(RSE^n(N2?EE1EtO5EMn2jQNj*Z}-~*VFCqR0?~;pZK%0A*?=LN-Voj
zQ(gGs^v+mJpL^F&q$p-q_=);pfoxoP$JR&|Hc}trjFt<pGjc;#AxlAR0-+LppeFNB
z-@(ww_uFHHTwCcTVE{V1RcUb!&P+}MzX>MME1uLjc!dw7Xs$L=Uw+=~2Cf+pA;+3O
zKx?C6u;_)2z%I&+&`;un{n0-6F@hQ~jZVn#ST)eJ>{(k4pCiis(dHNnzh@dSrFg-o
z7eK%MXzvn!9WuKBbIYyLTf*HP<KIxNHLwVziJ_B}jA3<sV<DXoP!o|peNcs4t*QA#
zOq8656bjqkmNd4I@r1eLuRA^cuP8(84yX{1<r>0C;(GO&l%U3)-!pxQAOskh8+b+(
zGOCiNe@`<gV5W|2eD#?S17+b%mQ-bblD3BJaTPA0MDM6W3IL!2O)qxADBNzh8~o}O
z@~Y9bdUbpHJ}#Qgq8Rz!88Tr2BtEPA1@)oMv)#)*he%8kfZNDp$kPuGEj%M)u4%{#
zCe8@hUz{$xJdZNM(gtDo@xflhNcRJA05SZ;D-@gpnfB45_r{_C&!uUpJ1ZJS&-Zq~
znD;T`9(8Aq1r>I(n)wWbjC;x15#*$9ZiXAW!2sZ#EiGwuwSRViYzcvxL>K9?dsfF4
zIS<N|&GZh|ZuN_~#CW}WpMW%K#q8a4aWbFj|GC=X@HuJ(DuIrUH|xPK@n^Uiw4GFn
zZ+>(z%)T?dYY-~uc+o2Aj#ks@ZAB_dO`9e|Z=kaBE-!~OiIC;zO5Ir^kU+rj?he|J
z>zb=_QeuULAnicBY?3AcQN_%RMn;>yemZJ#l@U=Darj-FcN~pdKjkqBcyJX0JX1FH
z!$`)%!DWd<!1HdY`*r8;D^qSa?=5y6mC>rWX#$+>;6UMgh55NvmCV@Qs_sOc<rVD^
zZ4Hk?iRN8>twW>rptyA8WK~gjmRG(4R=%`jC0_i;+XhfqmBp|lA!ftR-%^AXFB;0?
z7}3Rc&8xDVg1nXZ82%gZH}1@}2B8G!j+WS-_qfIV>7>0R+Bz$O7|xCIq{qE29vpu@
zTFR-0E?M|@Mw^b($n0G!c2gYn#;4mu-zRVt?KB9E(`yU3y5D&qM`crGJ73yOR5~mQ
zx#gI=94Q8OimZy-3%BeocDt3gV@Rw`T)2~b2((Y&?K$-#`K6G3D>Ou1xeP_WQ{$3p
zBl#TSYB~QH*L?Di%lS<OV&9?Dc96ndr{B?ivNjA|*}|Hf#%DKx7L3qYi8Llfxp|O;
z9B<c&)eP}wV9-JyDJ92~xOXQ*8?6w`l^ohmLs4t7a<EMHflPp@e?B}nBMg%Mqe8Jz
zlc*W<>_zfX>9Io<^3#x*t_fi&%Qy5P(<Z|UM(LuX!?uCCOn8!no1)$A%LlV)mn>;e
zU`u{Ss10tmOm+bydrPHv(Ie)En4jj>OI@9gKLoD8*e<@zKa%YE)7txWj~A*k3B;Sq
zLxW!*qM5wT-#0&bIZ5*@=_o`=q0?_yB;fJm7ZWvc!_lu2xV`|AKeS0$HWQ#9pk3@7
z3>3bz%T8pZjN#ANXYi`y-To8)6r~&|m>U7#YM<bSw*&mX-OKc>AA*$KMeF<6LOt4l
zdf?T61{%EwU4l*;z@Ae_1LzFZ9n=4ld6X1KtdVwGzBm{lNLZ_-b2K_cvue^)74^SN
z;N-(qSrzVgAq5q+_EBVi(az2<09OP_Cus;+D>uHQUdlc;<6yQy%_G|6AjE1$o4NX-
zmRB2GoIf^CnQ<?c%s0@N(Iq25?=w7~i@o0XvqeV+N-Cr>SVds`lA0aUfPf$oALd5B
zbcGWa+5M-9Bvm0hdV9vNF<5#K!PN^8B_e2xgH;E$wu~)pWw~!*h}|}5Zf;d%=Bxdi
zA>#(-B48sQcRQ&hOa^{6e^0;vVz|>^o;maFV2l4~wJYip_9>fHI`pn9uG8)Gj|37%
z3+23_KGa^{(tMzG@-l%{{EBF6jM3=vjxE|R5v}@&T+>)CZpKoJuC_)x|7FrUfHut;
zx^|HZ8(TBdd0O=NyuRa}bfFe#wIdnl;gwCLTx5O9bo!yDrHzVa;mEV)j-ylOD!y_R
z*VyP=gUuK15n@`C4lne0lQ<*Px;h>P$;qY6o<XvtmNo?=_>>W1PB!}=k(P<MQcAU~
z&1!i`KBx#$=u=R*o-Rrx6h36|+TB0l{|qRl^WYPA{C;f{>vFMYXlU+4Y+tcSACQKL
zIL=j(aAM&r?Ql+`NhT29dVIls)a6b74{mkX*wokzo7@tB{+c{%=Cvh%^B+$S;_vs^
z+tB-)R;?zowZie-xw6=L@JZ+6jk4s)fpI^gu;y7)^Jdu9!!bP9ZQ`C1bLwz0vab@<
zdRL=Q5BDu|<hfCH?ML1%+pm4cfAfZ_UWTG7N<ObRWQQU7L&_&dW(IcxHlziv?v9>r
zZ=x_E9g{DefW_91l?VckQUl#fMwCrZ1`fyQRwYD)JcWXZgg)sdC{BPSs*#GVcO+js
zAAAqFj~LGD&Qf^&8Q4zeb}tU=Pi2+w--2xVCZ~wVut2DnFW*|0Qxi&PzEcsI(3sZ6
z9YPcFGUUnVBz)OPG`-~DSm-V-&(qGNr^*{VQd&-&@@1Q~ilojOn9x>^aZWX_9ah)W
z)F|$=o}7n4zp_&F_ol*)S%HXF5A+JhDh}qIy=<<98OW7-I|zde=Q-2o11anFl99$C
z{iz{>PHuaAl-)-09-ZV=9sw&=JZX2teK5w-0_Thf#RINj<~&KXT7pU3bg`x|ZFd19
zM0K~>PC7C(ab`9YMyhW9@nAb7ENeabhjNxeb@j~~pLr~!p+QeXSSj2C_rF83)NM$9
zf`S)wS>i^i@BJW`knu$$*I-+90&wm8kGCKGuEx$;k8iDrN<3dD82p{{{?4JfTKVGk
z7S7SO5Wzziu_bFhmC}qoN)_n~2CaGLFIY-Yvk)x_-y<Vl?VFRbLj5+RF6t4loRMF!
zSFcnQl$vLRgYFiXj=tM8870f30J`BVy1JJgj>i70A|zZxXLHk{_TlW*_Io?2?dfeG
z*zJ}TW+bNpCS{t>Xg)k5E}Jv-qD>b_lmix9Lhk4YjRhg|WYwi}<waKKE5x^O=+~FW
zX>x<ctjzp3ux2&@b>eHSk;QdL)38vA?MW8FJOX%V#EH-#RDSJh<$)9##XuTU3}0QO
zOZWI>JeZg|CZFqzdLK2LPD<IDAEDp3z>U)WeJaU)y7%S9@Up(C=HVUt$Vz2zv2?la
zn_W8n=H_@w(bSdoqj-_rvsmcv^VX$X`9XSf)JH5&|2XSuCn3u_gE*41t(UBnQH5P`
zj8k<H-702@;MP8nCIDys?F|Ga9GOE594hLCdULL{l9IPfT~ysZL`Ej7xT`CkN637%
z8F=r_JlZjhF|GfSmrqZgplHo~JH2R?1?01syCmlTGVZDhJUmDV^M9s1|E$9DN{F<l
zz}*_vzal0)0O;E`RjcKDp%Sk@yGu(>bwpfXuq+=Q?(dC5NYT$m?{LdMt2vMEujHiL
zGz0RU&IsZ_O{aFIh{#J$lwTUN25SMS%QP=$dj<<g6Fw$4!}!mV65oH(r(mOKuAZcN
z2e*O(beWdzFz?Q;Bvcs(OxP*e<X(`(y|dv~zx7uDQGcBo+3m|N1{nBNO7hP?9v`|h
z()NIG>Y;SmC0~UBl(+?bqxv&SR)U-azkpmu&RXl6_ejfjJDtP6sRe9o^O@qGOU~`E
z;UqlW*Q1WVtPezJs3}-@!<JH>2B_3Q5$-1J)})kCulN$<p?$_!HC1aemVnd8Q#66<
z6;(FpRpcdymMb7dwgqW7KI1+voEw*XU0fu}KtCs@gb|l7ytkNtRib%J38LzrjY&hc
zMO1HdjyXXI=eUZE+Y%Gd9sx-(=$(nf+wZ|~M?**eO)6Wk>EqI})l0T-5W**DV1GoB
zmY2li<Ll4UI`U_8=__`}aHSm8LI-{%kVDBEh%d=Dt}>o^Uvpt-x)pb3xjq301n{%6
z0_}Cj^<uW>!%<gWznn47G~B2sF%eyrIqcDeZVY$o;2>ULfee84a+Srd%>d-d_AfEM
zKUe=GE*Ay!BcyoO^h-yqHa|S@esphtu)h%=khDRbhYsJ!{^(?(j410{{dD5!`lw7U
zn}Az2(8bAk*`t7rV0tLw&hG8co{AJb_|=3A=-s0Y!NJl=3oArbFVgiEGVEGdSU*vg
z_d|q|7t6Vi>l#wTQ`Gpiff$5D8tb%?i%^e~b%(m%h7(ip@!msx_!d&1pq+|qAtWoO
z6%<v9t*McK0_k)4OC>IK%%Xi<Sw85?wsEx;b5VVo{jnhz392O;#QiEoFQp;T@o8g?
zIbdfjRqZmrIM$t_?69$`-@{*;w|hRUKG(ORSvsoWbS?DwyudsX2Wb3gAocJnOytEx
zHH0^#`AAQBms9HnuiVPs1%ma}=|wv2RCNS69$6PS*W4-)dT@XmJe~M9a)dIdrCs7e
z$chzEDbj7YV&~Z$<@Kn`R~-~lLh&p_Vj-YQQ1uZ$??g#<=}uf5)yMR-`5i-P8kZ{P
zt_^!LA|gY8J0iZ{{~|ZA+oz*=3f1(E7gMc3myy6_>8~y4U1B>0*1d)s1(q@~>ufKg
zR5O`BUZyy|`bqk4diN!d%3nvUyzuO8Xxr-S=4i5eB?4i5E-qpkV6?W?fG+8glcU~<
z#Y6FciK~*siit61)@0-b{%AI1hOz~9gdY=vQ2m)s?8F3H76i|a7<og@DY&sB8l&p)
z2k`7?k{9tGw%}<#f-OL%=vnO;AN|1=bo@QTz?zdSNXos|!BwktO$i)Q4f9i@T^;i}
zaj~CkK})&B!jcmE{M0&KUok&L!LJyBCtYIF#XZGs<Dxf{Z=Mf^Y4*=jt1-$n6Cyr(
zkcB6g)+z_e<=r$ls<&u-P?T(-3=p{0YsQ%HUtjzdt>e48c1ue?J2G21JF(QSmf<&O
zPHCKNe|R&3wNdse-YXkrho3Kkb3v#WV0#RW$FKN!J##(o0;gxRzS^2BMTCbnn*E+%
z?uAJDCCdtVt+PGOEj8d54h{*A5R^5-^M1gT3w3p_4Of!GgGhkQ+rd-RkKBO3gu!w$
zb+oCGPfo$df}+(MKB#h8hgZ#K#pBYWJ^MTT36O3;QsPg2J1PUN!2^f&*4f`dzsNlQ
zP0%*f#RG}-L=$9&5rTq_DyqX(`5RLjovb-`2dr%fbc~(FJKFB`R!#odf;4D?*YxG~
z?}_g=XJp=+{bDZFnE#|g#V`nDqg6Im(j$((By??*HyoIQh^vB!C;M`wc1}^>9UNwl
z?;fbp+4T*GU}Q?*@pJ=2BH;&}D33o$eVXbJFl_$zr=9F{tN&ns`v>ASM1xdH4EC&c
zrj-WGN^GQj-AZeFClnxB<HDHj0Cz6UVF0O*>1dWjc8WlY@M~BtLrFr_dH>1e6ayR1
zvc%SiPxjS0&uaF6+2Ny;AGgg+Ncnb$DLb6W>sNF1YbWFo)v+`j>;)4O0UH<ffD+R`
zN*M}wKYh_n8_@`-#>P8yB1_ks{bb@9=P6LCybWzMr0qxCSbNFAohI|wqiGK{6qOa!
zgp8#yP}8xLxw&(@x_%BahnUO9oWf$o1;P~*pEcrl=j85JvK>xl_jQ(M=&-V}<i<;P
zooH%CP_eqH7>kO<5fHGCb1X$HMB^*h)k!`M-2P$o1BQrYNw#ge0te=|(Z2y+443)Y
zGe~30^vL;Gl6nVOb0z;BJ!NUZ3*V1nx#h6W@i+Vq=WY*C{)>Qo5CA~caz>OKOJmbi
zw5=VaQ>}7{jqQ!_7`FI3Lr%%|+uL^xN#7UCnHz85fHWn#XM5TwKxmL8*JXv})0K*s
z7LBestpl!b8Duq->^GQW6d*C_8@7O*Ax}!5#|@no=0~<ugqwGM<itx0C=kgL^i-OO
zAU{b8sfTmh7n7cvmRe&I2Vn%P_Qx1_qrduHu9|ROTwe6yb%;VFmdx}X7v!c$6r7Nr
zKHNhmAxTBWK*?7(J8UfCmWGQBWI79S>z!qg#W%E-Ak`$5#CrrWKyc%vGlU77pGkn#
zJax;SRrd9IU1}gJ(BDzcys`$@ALMz%xe_UkkDX7f7>F=XbdJ6WZ1OqOs-XlW<8%S)
zNK-6~k+Vl!r9NcUyfoz$m~XIB#6#<mAX^9&>Y2Tm;6PSVx0)F2UL1bzIQ5(S9*3X*
z&J&N9EKtU}_-o`jzSZeUR^wS2*Skn?Vh*7JlE~sEXBH>DrP0{vCqRuyV!#~d=KP}j
zk~BRlntX+R?r;m9n>YE^5DLg}J7LS)VA(#~U|+qQr;#{Yl`o6u3Y3$5??Ng3z|jdj
z9uN`os-=P9pdBoL+;39_3SYI<mC}$f?-K$@yNB5-bMV&}z~6p~rr&2f$z!dp{^j8i
z2~Qph>)CHQ57dwpNdwvq?_^wWW;S=%mMLJK^HB*^D6V_h`mV>=W60wsgVS9=$Foik
zedarEvNAk$-EPz0q3weBilj|~gz<-h#S#R;*!AR~&2fZmvu}GBDH{U`Nc&bpn(;Yq
zeEGucWWNi)FRJH6{g0~N0<5ZT`yM`s@k)rIfRZZRA|(wXBHdjo0@B^Fv9JiGySuwP
zrID0Ix<R`6=JCD1d;cGgKAuO;IeYIFbB#IXSZnOYDP~xX%k>?pAyv8z-+PKwq_*1H
z%f5wY??2i+A4+wNP{mzN$rDRTe0UJ$jN@TMZhDO@Ikq&E>)PJV{!T3LLQq;lD?+x_
zvYR(Pr)a=}Yf@A`=z*zq@ZxTOHJ#v%Bu$w(?Jn%&Ykj{Fzalr}ULYi>Zf2QFbHAUm
z;&H<(fGi6Y*l`z6s=Y$?DdL{#ia_S}```)i0gL^WFCs4fFD-vt-G~LC^7eQYHi>V{
zR7uZ2f&%Vnt_ju5Y1teZd(F;b`t52l!Dng$tUbtehwsR&vR7QQVxa$E&ixYk_KeDV
z5np*F6gbJ>!<N+v(D=SAvX$e^-$ZnyL+_lO>&Gf*T$PobtJsb+tGLrS6XQ|WM8bV&
zHmJm__RM!<m<%(_8(@pJWP5DF*8tsxIls+kC{1TBvGgRWrnQG#TOfkc@6s;&0&v#S
z9`}{usOrJU$k>e#pg4{YQ|`n22Nw_u9?8vt471Yh_4)r36C7J%WJPcd44+TQl4-au
z{?jhX%3j+pnlsbN!fdd;ONW9QpR8%ecEZ!#^LlDko$C4s9ZWX1)N_*;oViRK7d%^G
zvxASb&Ej+2v)@LLIM_1teGGQVE<F*Dw)Nrd&m{(K3c)hs)9SUcIb3M7KO4%n5E2|S
zceEp2YNA#T5Bu3GdVEMRUWKlnX%G^NEYKX8Zw4tUHv-!%?c;sBCvJWB%|;Pr+zaWm
zO(mPJUpDjd=Bw64ymz+i=8IK!P3|%Y1n7yf6af1&e^wq@pq8`UGB#8gT$K6}^i)8O
zlOjN!mV9ex!t=TNkdeexpHg?D_j8M8No~V&*Ff#nOCQkasn<3{9WHNWLws&J<aJ_G
z_A6ETze$Bt?7#g#S#U+StBdrD%v}zFn<O{&9c0NHCiOq|b$0Ihf&_-Ehj_H|PKuH2
z?-{}Hfqdk>n7@u@TOs1%Zslmz!!zN_Zge;qB9$sncaW(kC)*X;al6LQk;)MR#0D_+
zaxKQJisguXit^Z`xXfRG9Dw&YL^8>H?I*5hiC^RsjgKnt9i>>9UbYTPp_rU4g1dh8
z+VCVSuuena&(6;J{z=#pMC2?@8zPQf(S?hbt>%WqytlTIdtA&>aT}F#k`BIIY}GD%
zcU~#dLq=@NAW@@I`Bg4D3C4a#GWkJynTgMxfg)^Qu0bjS!aw65u2(e=TkT$6Q0jZY
z_HcGBEugUZ_MO`ga+R(CAFx3#Aa<H_9SvG%njTQqmaqz{>5rYXq+o3RPAixVF`e9z
z2~qs;b|3MJx;)kNlhV}N73fwoW+rtKXP<9v6cWp(XL;;_^tbqhOfJsT(75dW{Pj3K
z{*C~010V%jL$uiE$39*&K-Q|$6KHYO%3*dP?v#hq7Hd72(Cm@NTiAt{JQ!4IODl_-
z0>1H6M_aO-lH+}&&e^r>Hv|3^D48rK=Nwn$`uFax#E_39z(PAMEM7%TL~_JgU(^0?
z_g>!trA0mF3$xeHj@#6>nSHz;Z5=$h?6Rb*fE7N}JGt2VUFa#!Wsa-Sf2v#-&2GY4
z<=tHa;+%q)v(I&Lm*rDjlHOGb@a+7$$mHXFFuP&D%Ihj?bh5>*+w`mF8v@1s4k3qf
z-K&k$U$CJ1FJ>)v38T|(k)lV}=!B$(Wv1RI0wPF#o$<{=n#6TOgIA5^c@1HM$(QJd
zd8AJd*rOyKh?AI-Q}p0@wiK`3OUq3q7r%&z>COy&Ylnh{YK?%J(W)G4&8Lg%cK1Vp
zdgj&NAOt3L+Z7LUa{?($<K=GE+8benqS4-ymOo2N+q=ZHppI<(OX6_Y+r$Uxby9N}
z`jPrZyc1~sAlC7#n&^Y1%7QiYbN7ZoBh*#^q-dNp?hxynLn%?{lns6pP_lYIp)^sq
zWk>6W0r39~b!}fj)0wR{>Y9{V==}#<C{dZkNCfb=+i;Cql!>q-bm6RbYpIrI?%A)a
z)sy9-MkfGbp<I{zIPGPs9}yp|UtL}1o=T~&Ve)D?zn92_OqzVURbD}o9&>lb|JKwo
zslVv~wZYt~V1`8QDcj5IS<8pzCxu+hPH9T6{k0V!ee%}HwZR1$5K>9p|Asy*e8`j(
z5qI?<5xnkM;&Dv&A%8wRosfpcO#1;I=!5)iL!6w?Z_M&eXQ0`qsT>)Iwza*m=TFZ)
znO*Q9OSD17{!3uij`A8ZvK|kLF?`?L{)SO#Z#QK(OTZPw4^p+mm7J;3#H$e5cFt@C
zi25Fykcr|lAp#Wj>=k$yHtW1wOqh)14Ya%0J4pGjCuw$;0*8{3-)v)kt(p1(T=JHh
za!-ai$%L!fm#zcc7CFFi-zBJ+1n+cceU$5`KO_zak*g^pDGiT$_vHRPcA=P^G5#oX
zU)Do>N=!^xdDu-F9Bd@p?V2%fwarP7mVZRD$7&JajTeVvhBVV>=JFCke^t<azVu;7
zv*1m8m`0H=n&O__ah@lA^yQ6ovlre(L-jQw{4u;qvx&zq4G9xY+?-Oh){@jL7GA#S
z^7=ywC3B`TTV_ywozqz}ojknbeCVX9mFKc^9mkZXyM*1p^#G(DW6aFY+FR$88-rXr
z5vBjlG~xh&(|C<1Dj8|=bc-uXUC)Fls*ap$jEKynz01NPAFE4y4~8aYG7XXozKw$P
z$6+IL=#hQ|cS6cXlA!?qVGkO0qt)D4VGpMz=S0P;?V^;g&yO_+pJ*ou>z;C+0(n9h
z^-uEYhAEo<@9pqrhx2Fa+B&2cZbM8J$x_`m91@0-#fm#Ra}-?4(AYhJBA$ISnz}kI
zy(MR4r1<3vuwX9HldhpZGFMF)1KWcnA!+-$(!Gl&v7l1DAnrGH+Gca9=<w7>4Gj+g
z6s0(jJwGVVB9(fU9_$GUGwo+G3xCXAZ<l$HfHUiy-PZP8Em6$!svamx2aaT(d2W2%
zX=wfxy20|k_m5wVy5sI;v8^vA<{-v(-dLpPJ<AJb)Zjr2V3zjG>F8M(Qr7tFeHWHi
z8_!q;n<1)D{`p*}Q5+6H(OAbILP@h31a5!y_Gm%OtWOt^_xqi5;qG4gTH(F#$g>0e
zbyM$^pHT-V7s|6VTQaWrSqt!$jBqlj(EV~?54Xq|%mbk;&~_*68)Pm;1JxK4r!;DG
z>gC4RP)P#qnZ(-4YR1ac>JJd(%@Pn5nC(vx&s3||-0=J$Q1Q}0*|usT{MT)gf><>*
zl%QUTh+o+Mz|&)7<}VM^2pS!f6IX1Ot^{chm3nfaNmtIjkVC^hRn_W4m!eQm(a2{E
zNSlsdFeQs%S-OuAjk~$<>gB^xxflaCxpC@y5T0bUC@h;<)TPaI1-n-TKmDF)D73Yy
z9jK3l9(e!uG2rqElfKq&KT$pRFJjD`@<9v)eD9F%mv@7<UR7|pkm1z|ZzMZuXx+dk
zkb%KX?#ns!p!gyC%zT@H^x*+}aWroOaEhw$#iw5wNU&>}W_o-7prfC&c}mzs5Kfhr
z_Wg!Xm-F|qcbEH3bt8zXyBE&Qp8t+9<spsv0m#uzM*8*K^N$H>^ks7}$-Ahl9!f9-
zNFrgJFK-J~?zm;S$J97P6^?1Ulyd)0b1+|A``rv7+7{>lT8GsJb((~z0L-p=!$5{x
zjSEEhDrJ&GP#=r35UHF@U`Lm~@qC_-hCL@-qKiTG6v&XotDeiAmu*<F^L*I%jea@t
zQ>oCpuw}d#|MNvftgmH}VIR`-wY@5k1k7Wh<Iwqa3wuFJ=;X|Za~3`GUyZcLX#)(J
z!#TxAx`V0*ztr_#yn3c-ef36|MxZhsYO#OJ&mgBWF3Qg_s-<km_oE&Sh{xjubwM>g
zSP^5yS7<OW$7m)*)}a+Zt9o1slr*3;&{eVL1;tFvNE;7C4&jk6$Uj5qnhXr^WN`KW
z$ISh_k6v)UD<2Xc3U7?8tS>|59gX5_%e~YRGjyZKSJ#Gft=%}*<eokmq_1{eQGI20
z5CQbjd<t2`vmZZFRVQlYX%ezZi1dW^nW{zxnKfJm*gmhmeOz9_T{qZRx_6oLTP#oV
z<YN|F2IRg!Wc7>QfTher&&hwLrFG~Y2^i(*(&nMr)_T4pyZpQtARcnt9b>{-{?m56
zHWAJHnW)8gI%xxWIW%lLOGs#alr1v>H1twmHFHfh?-e6vl6GU;_^lBY9VtC6l07++
z?l-@jRQ0OcszaSX|2FwExh#qqvO?!PM9FoGiJ9Zv&{9f?QR@E7nIEhy%_G|T3?`8K
zXP^Bc8I_HBan>eO79IKobj2W=EO`mkJgAGD97h6OXqzlIX0lpu9!^fZL02WT<gVPj
zK_ag0QI~a_U0=W14sZ4k#_8>$1U`S>{`l$`3xOO#++$rJ%|1a9-Y7>05zm%a&S-Z#
zJHlnSC3(cXovmi<RdJ{F6{uT`jyrF&jFn2nIGB@gSpK(zM0(Hyxc#WdNg$^C{oF8p
zDC1|zcJa^5t#;-?T~l@*zKn-m=^`^Q=b_jnE$P4o=~vh1JfKw;d?aS*1KMF{$c)M3
z3jpg55LegUvCgc)qjX29u3oB3Raqns{3ZL}{R5O1xH`3Ucn%I}QH?1^*U3qw3@yk#
z*4GOy=XFe-KbPk}U@I%S&zF+>sb}#u(1})#cCl4{+}RGh8N)c=!n1f}DI(7G?l6c1
zoKUQ;NqmqL(^OzHZfXME+O)|FuPWnRnG&kk$a)3D21N$$#~EMQ#w*y8UAdJR2)_Th
z`lPWbL74&eR!Bftsk%AsXba=ucgGS_Bu~^>{C&j&TRJBCy5X*1payuk<6L|E3o_O%
zPt`Ww(hRSe>x(&<(1UnmKs9yJ#n+6Ot82|Y@s-gKKK=j09yK={halCPF)E75!)9%n
zH}GXtNWAH9jc~Iw(lh3Gvzu+ICtvoO4Gv0>XJE@*7?bh#UL4EkiP9Rh3@E7x3REML
z(~soq<VoWd-<?}peV<aG^lm|cgii|l=t^M<B7A&#*LG!Wz-kRANfF4#KdB9d`YH=G
zg_UOuvJ64Jpc|9jMeIr12J0_qVSjZMl+`!xI36wB;tdH@n;gqFA?<5+@=b64xy^(B
z-y+z1FMKGWR2MvC7rp=eo38d-;+^a|?6LGFlNYM)Nz|TSqnJ^^?w4vU$1EzQ4ZH$}
z?B@n58?zSrHfc=a-;xayvX3NLSq67BoPEz$A}x`qis{SyD)y5+19|-K<+)jkRk}^}
z$0n&Achbxw3*)^eu~oXK6ShbPo}?+>aGN7`?TT4=s)%*R*;zd_e0u)5+(z*k&I;+L
zD1$Yj=>IkDF702`Zt?Q5f-(JzrLc>S_?E}ztZW!i(Xu0I{>3|+&y6;Dx{{E@8?(!q
z`;S6;AIhgHbL?9URrobet;~Bpx&K>Rl9-{0*cj6!^)hV=Mo&C>g!5@c(at7n%1n}2
zu4I3xBIa>f<fjg^%#Myi8oW!VMX7GpT}g>-MMY;B;{{Suz;iSV4?Z;={&r45TQy>|
zYtJRNP1ZklTIa#dPsdOczAKLS6+9mxj#pADYCfqpqIi1watAq2q+^OqbL-QS2lxN)
zfqFuTQx+5M_A9BRz3m)I*jX3FHA^=W(tnO*u{H2rsWZ1eN)KQ7#3BORNE|T0OC#yj
zfHWf|2BW3ZzIOO4Li!DbI*}5vK(Y41%c*-q_FW=r^~)Cobi6$kF**v?KY}E^F5A}&
z)0j5i!D`^Wav^ukkb{6BXyVkY>V^-L=AMLfX|5QQ>{zvoHid!q5K2iq<O`Y(*P!tJ
zdVSt2M(bwP@nH!A6SKi|ykFSaaLALC<lfHCAl9S(=}rBcpF%m`P0=RjJ9IoSJSeFb
zzA$MyE{hx+=~WhmMgJGR_dI&?R?X?|yZ{%0Ck3n5cMZjKHHu^XX`T>&ukUX#<gSrr
zawY=_#9?bcXQzk-rCLZ=s!HGMJ+5ct_{LVG2axf#v>XokKeh1zb*a0nGktvvpapSq
z6Rr`bkuWwgZbb`4S(xWwE0oJF5+|Q_FuTKpGQ39I6WVtZ<#^=^dPfQ9{Rm!sciC0Q
zR^A#8!xruwdUpT&gMuxdqs>{3QyOfhY%{)@8<j+$8(iFP9yJrX=?(hdiWKj;A2T;W
zvLXzd1At|@4zTDwVKwX<_?FQvSZ8;x`f7fGeRItby`Sx!;6)MSKk)bcISyLu_(^bW
zWO`!_odiPl^+{T`Ui!m?()X~r1+rwega_$yEZKm>KgxAooI0%f_#@rup+=nIpN@4*
zSe2bDW<L>2iHym~;ShU5Z}`$uZTB_hx|9gr4yk3m_d7HqbaTCse*ZZPfTr_^H!#E-
zkN~5r?LRUdj}KeAt)L9(X()(ub=3~fkk0L-|0W8pCjxcQ%DBgR+zdMP$+W#Ia5~z5
zdw=;L?~$ff!TK^Tmv)85hV@ysO%#kwtGptEuF{{o7U!19poQ6Lr+)vmr3VIq5<5-b
zWS^)JiIFXPFfPT9AxsygIQfUj+Dvy@bf8^~;s3VpP12y}Zh`sFD?ljZa?|(@yIzBS
z7);YKQDD>OvY<`N{RdB6To9R;i24)Yvj6IF4mk+q9~#c=@mgjoXDdH6?G^#jv8T5o
z>H|APj#7ndXHTssfn~KXj3BhC0Ifq9+XKb9AMEe_Dw#EZ91>kq1jo<cuSvaKq$vX%
z`bvHHu6F-OSK|A-O{JR0@#<B1McwTEWn+<&+If#kPUriV0Q*^vbrr2f@)jw<z7Y^O
zKp~toGxGqpquVuWg@@;_?y3R7$6s-LmoMSv3V(|^cY?D;_dP27tL>9cmuznaAiO1~
ztiN22aI|EjnnCX2Gb8@oNV8-n1jXuL(!CPJrzo#j&nR@}VoA&Jf5Bk$)88`oA3KMa
z+wq66xd<{GgDJg;M<M>I`jzlBF^t@c#mf|rd)f0q`CTtJc|T~m+J_qdD$?QZ!@{g(
z1^O8a*iCV^jEgAqs^Z2#LH>Ni=Y-6ej7UqXnCS>*YHM5NSCnUOKHzVVQ*)s<bNotJ
z)xz~OUi_g;HauVDF0;Fxs&2KnJ3Efdisv`hUdb3l^$c_d1;*b$${w#RSrDaD>OYxz
zY-b?|Qjh6(Ge!^CzWp{bYWGQe#Rc|IHOZ$x4mVrFypfIJcaV9w<P6sW-2A?BfUyV>
zMW1_pssO#YgY(7ATW5PsbD?}wqw3IIjwGy00bCoiSq-#5=$SLv1VK^FeT~ehEqFYT
zF$iccLQv=J@-`!kP;bA&cz*HuQmx>xi15J*zYkUA)8_h}byD|8@%&z|(*Gw1LEhfw
zx1lnz7nt-x8q)?$z_!{6Y&=8xowwda`%VNNI|?k)kJ+q`;9joYWucy%L)jTm*IVgX
zo|u{Eva!)5lm_<f;5a;Kx*ipQ=joS$z2I>`Nb<FuR!6FRj(>k`KU2W&A__p@)5F7_
ziPjsmwn|VXDf=GKzP?zh^?4qFv1F#XDQIzR&&RFbXJ#ttdb0Zo4{%(ew{JiARY89$
zAh3zv2|`1ilp8wpxpnokFLg;KsVL^1$(o%D-8dv$u_+#tNv{@qcJjhXH8ibYE1;YG
z-u<s~ua|H0g+;Kn4Slq*U;zoSQo8*07`<8$0r)o3LfEq^D16?4l-@`r2zlVqckQXe
zTF0Om1^;5hQMV#&e9GODgKP~0rG{<$^!Tu+>@zlGAHYFek5H=x80$+-1x!7<48Q$$
z|Ihwj`j9%B{$9!nR)q5Pf3&76OK8=j2`3d1p*83p1%6t?7d*FkS^tGwPv5PZ+y_CX
z@XC5pUtiF3`+zbM@8V|*TGskC&<O&=%G`|y1zvb_3u?M1{XAS;Ns1EAbLTNQIJgfU
za=6s8<C9Z;Q2bn3Nvg!IMCBrcQICiiYus6G!xic5=pf%Tdw(S>vGA6ts3?vp9SQ9T
z_m9hdJ%`Q`5?A3cW|X?7=1Vg(H=$%ig*FB`;{i`W{k~)wJQs$ycgHC2*J;V!uXD9T
z@9sKZ!#i=Gy^Tf2xXUChqw!#d4NZG*<X!vle(UVZKxB<aLi24oTH1L5!63-}_;(_{
z84%stYAVM)snj3JqB<N0LRIN#&k(TVL23Ee!W}_U)pemC?0T#Q4j+pVH@;L>6u^4)
z*3WT;JkDV4TsE%xpe{0}G9?u3)q`=(acFUe;q&`d<`vV-b0!X7DeeR(WX@zyuT)td
z1jW?0v@)|vusCoIvK|J*1`=`8bHjnNcRvWQve=rvyiP1Pg(qSJaUr3oYCbGE-r3mg
z&?^hS1UuU;N|3u%$E{5CNlSR`*0xr6E!NteMogkNm({PqI-H75#o)#MPxN-PyZjxE
z=WpJBeEa$P4aNz#C<DvNHldGxTl+ut6JHr9DT6RRiCQq5g`eBxqWegZ`KCH1@5KkP
zqA19!R{4<cp@kfZ1^Iae`SEE1H!*n&Q{Gv!BUJuK3x&m=YOtH&3O!+0WY_oH2VHs<
zB9iC=s6Wk~!-1<a8&$4Sl`7vkkyuJ1s^{n2x!(A`W1&&s!S`3ef%&jNtbo*>kU-z?
zwddClJOX3kJq-z0y-I>G*n(SMuRU)$?6*_2wIToFvwzmkdvut8JkyjoKbyNScF6*v
zp$!O+`U=JL>^+AaixoLELOK!k2C7r2W(!}B|Cj@$f7fl;!xN=DpSqFD%Gzz3H`n@-
z<)-Ra4WFl{FQmRJ(0_LXTWp%f+x#0dnc5f`9&mA(c`yC9aLVtGhBT>AhnJ@K6|;nS
zJh`10URJ(UzxX{BQBcxU{PI2Tk7$ji8j+GdC5WeAIvJg+4V;D;%p144J~GiTn0am3
z2IPwUdi&3g9n}~Znh0#B{L|l4x9>(OrKHrnyVzAqK5m&c;ufl^$NxI)@RH}=LTYtn
zhqMF}9TmF}h2xcvm-lJ^yMUwuh>z=%Y@d6AG&X(rk9w!A@FpXQjG2GTy0F-zN9pE!
zO$`;AM69+s-1Q9Dtt%e$b~81#d;yzej;D9$&ja)4|IK4IEvfJ;A7{xveUkrbhVYtp
z<&(6NE}>;uY+(P=RM>CPFu~Z#8kb9%o0IeNsT%j3*l9~dPE+5cj`5?ye+eOx5gS+Q
ztESge%N8tWXM2bGbT&NNZb{-}f60Bw-q71$n3w#&`v&h#^85VhHdxqUu(AKB`VZG=
z_eDyDjN8*QtssceUG0zl!|Z0w#Z7sz%o%;?LQfX$`ne}{@rr=CNKZN<?26iJY~)Oh
z>Df6q-RNqFV7Q6w`AyQA_XRyR&I%MGjb+Nm&i{w-HUYu#nEI3JA7XNN6DTWZ9%d)_
zUMtnZTCT-T`1FXtz)t<*r8Lz+MJ@%pTf>@~4w9pG5&_%Cm=SRf80AsPhaTA(Lj(Z}
z{AISpqBi@pJS~A?0U5e$qlYIhnVSOd%2m71#WV0o6+INbBbV)06miqgXH?XdSYKpq
z3v_uTDjMUWn)Z>spTF`=^xBYC0&k(pH>8!)Zag`S)A-}A_OwfufYY-fX=ml_^R3XQ
z+T$(pv%<r?IunXn`Y#K4C`C<suVv_MCkKTutjTCmNl+f|W_3Avs@?7>Dj@#gWzxiY
zCD3L%a765>GGY>4!fhTz<+JK&Z{Pg)l-QWY8<li(%Cf#*LfdLd<Xt_T<{l}X`x9wb
zqxb?D9I0CkZ<i;Fv1gTUa|uuksU3Db9zWc&Gan0jDPj}r=JjH+l}}vCpZt_d;ZQ95
zYqY-k$cxjJlv`q6+&!nFAC(=S#CBoh{L>R7KF(5Nj4UH+?yMokx;v!j6IE*HR#^$X
z?Q`e08YcbUGXYEFH{%0^s<*QpQj%7Z2SZl+v@BFh$2}&@aU}fqm(rHrWh=btOXJgZ
z(kMGT<fI@h{l`&f-&JQIJB!PRd;pVCRY#d;w|$Ap=WLoW%>T@xdpz<vuHedVEicAm
z4J9ON43U?k&ivV=T&d+t+IOzC=R>trbRLG*S3UcCMx=^Z-B)~n#|;Iy`u%bajwcj2
zRU#f<)(WFS{RI(TO~r~wpKrP>d~M<k#7`v^ijavBI1YSC%sogel(ZP!O}J7@xluWa
zzRao6k#Y;^HJR$A%kbAr^BH!(407+y8_k^|qg&+o@g@Gni_z4KfhzLqzqyhVKUC~3
zmS2itOzBBWY4Q1)TU}*J3&+NzQ2n;ID(T`%15c$UGx{HFo`|oX1>ak;v{q~X#<AVr
zWTQ5{QXwp3LP(INl5?iB7S261bbKxKG!t)kY=5Qjbny?q_<i0;F-`w+IZ8zE*w4%3
zC`opHZUwL5<8qXc;N6+~=bIS)++-}KaX~qinez$?M?>8e11|<Yh8MF)RLDg$u&qnK
zh!d2PfO??d(7`#AT6SOB0*!Z{?M#bC;9tY8Wr}K&tA6)Gq}sV?S|<(re)+C33dmLb
z_-EIGFkyG;j^sp@xLT2^7UDzyCf#(jGHkH=v^@H7in3GlC@48+GI42zo1UsxFT~4>
zAt&`&iw#*cHL2!P@s&A=o1Edhk1_lh^+#WydUNlFoFAvm=V)ft&}0wk-5knzGoHQJ
z^W1=l*PdzU_hEy?$9B!dfn%Db<_s6ja1wO;Et6pBI;{%XDAN`cAOji30g_}@uGywq
z#XUzmIOg$FVq^mkIFv2Ve|rentUO~r;1ZF7L!sK@IM(>0s2d*GNi=vZPo^b!>5XeS
zHtHnQdZIEYG3{joKK*kF5<M5Y{<Wi5_HTnNo++9fy;u_nqK<fxYOI_%&FeV$DdYI{
ztD5SqeMa|uS@gXi!UOV~^^`M>S0#E4e&eR|t8H&SkWiertj@;n=gfK;zPMzqzq3=y
z#$ZRdEbWlJigAX8d-Xeo!7VG1?AOj=x{;4WSY;#c_HhJ7=*3P6SDct-Ui^ECI;YQD
zB65}zEq|7yMiYZS>}EAxOX*wM!sb|4u?l%FOby^P$g5i+-v<6Ykktr%8#MWORP@Yz
zDQ&h%3z1O?5;M&z(d`jY){hFE#xFB#)F=p%J&L7Fk!@Y4hz<SGygV9L1+5_duG~Jz
zCjd^CM*qa((>eYAQ$Zucy-<Jfe_DcDgrteH9SJs%zTU!@KTK)aztTr7p`&%0jg(J2
z2f;95uh!12LGW{K8s!TNUHN6kU{eufy_LwdAu-T#&ZICjn6=Y(UZFPRuwnS(?@@Iq
zloKU%<FC_Gd(m$4BYtb2>ai*{eZJz1etu&(tdT!3yBU`_b%TH`oyTC@@gbj<zFv5?
zidJ#kH#_lT_3@u^Sx20H3dqdV^Rh9EB!h+5bJg#{D^6+6Cyu{bI87T1xDG|R9A)Q`
zcK&;yv?q4+u$+87gs^A+y6OcYm$KWkVvg(aR#avkOnYVCxvzUK@i-rrujkPgnY63w
z+!#<Nq46CMZ?Np&{H6K@=UVVR^hB7$ZQv}^qRk3>pm|QptB%BBV7R<EyCA@2>N|k<
zZwsPy-@hfI##tUz8QNdWc26nTX(7lk(4E~oKXNXaKP^?d0q2~X?T~T$=dnbbG;58V
zwF~Z=PAFe>9FJr9sX*Y%P32P+ABDeOu5&}=jf?3|qVSsK(UGhFJ_jmCyzZ;wndLI2
z3I2^Z#Lr>1zd<i2xNogGc8&0)P-@@3NYU%X>xmge^@+MVE^U^Us`+qwclEhQX9LM%
zpOzQhm_^N$IiDVWlArh(m4wR<g>7I|h?>%Wj>wnyLq}{wh3`jOAB+eH$g_vF9O>2*
z<cf}J<81uqPv`7{*L5o&@3Wjf6$AWf$iEY7u{=<q%06_d8z~g_-|6+tAh<*7_H*>S
z+5p>Uy+^erKYwRpI%$CH-}AUpyVq$oM8jMPeFl#1SeooiCeu6blaJpz|Def*Gh`p4
zm)uUwB@YKkK7<{$G;6tB2BhDqbb52+1l?8<&eDY-!jv5Mr53r@YjQ#vh?o=R_)o#&
z-t8014e~FOWm+-r=kLf#>|Hr&{ziY4u=|s`uQoa!OG%phdn893zg?@&hVPq8jA!tS
zIs_6A@iRyUc#Xx1Tacw)4tiu$BzWMk!Ubn#4u-TF-VTaRv9$}-e9zr8FObG-O{4W4
z<~Za1zY_(ud=F~^k+dn(25%eHmQ;?ux$>zz$Hv_uGV9YIht*Q9<W<%A)&K@8UV@u<
z(yoOR$%jnpu*?i8_SxUq9wcazk_)vcVTd>ro*zyK&H3~tQ;|?q@#BwXUOO7CD}JO3
z2|eE!IzN8ZFF&ui{&|#xn$=A(;V5)|c;Ov)b#~lQY3F%03leG(u<*{09T8xC(>|v<
zcOLomsFr`B=*`fjXa2_J&Mf`5>)UQ;Li_t*lzgFNRA<^ATliG8PV28NO<P+&@6)79
zgtxSJxI*9|JJojS?A>3ycwzq}Ut8r2LonE0)4!jNF1L+`_)1I^<@T&Zk?CMz+jg8y
z<nFAqPdX!#+PFf!OdHbP^wIH|KbEsjEzg1D1%KsPCdKbxGJW3=D_h-&IVZa&j!RSK
z^x?w?8@;L4J`5*XtO-f1t5+2gp890^Dj2WF-p!!wCreIy`}P8uS^t<{P)nnar@T)k
zLA$+`LAp>nvr5XQpnx^IublBEQdZ9S=JS$v!x{d-Y9zCC7B6>{S@r^B*|<fwZ5{X#
zonc@Q(!LEEt0DhRaQil$$_oh^ryBU?rJ;o1PBmWe>hSm&3W<J0yBeF$g69hh3yebN
zt43wtzFi+)>2ZBlG0svW=`na|itK<W70i1hPEM;F_si(A`8)K8j31bB9je&WoMcPf
zlz^SM`n2hxVn<5{SMPYm1&2IU<HciWF<@7&l>ONG;Mcie_Ttc%*TNe=zJ(QUFQ1`H
zdTJ-Tsu3|Vm%nlxa5OyN@4kXICGt`@?8e``=CN>if~?ow0~bU4bk?a_g4nZ}bDLOV
zZP>$x)6kP&bv&D1Kk~Hoj(FhrfQ*chA{7MY`2Rc7v6bIRQ4$*qhjg|5XdP(bb5L=K
zRvV&~@_&k3np__LpAb);Fm)-tb^GSJKb64tsiwi~tnSX9l;ZwUyMXHeM}FRV#bNTY
z?_UpHl&8z5w?wHlr@T6G6<739D9f&mSHEr*i?TAG`7JGjXy-rn-x+*&p75=<HvG%K
zHcp-{5CBu3vQxOoyYm*R*~}~G2(ap|^s+OM>diW=z}fZvyD?0PvT;h?-Xs}X$NL#c
zh5t^eO$YmhLSJEl>#s^yGHPm4Kj&0tvkqks#rg^LO-&Z{r(Vk|FLzdIdjBE4cCNQB
z(+qJh*oYN8`D3&u=q$<A&|`Yeu7Q*ry#N2j;#5Q2x9^HJhf_E<B2-*FCfwYsHax2*
z9An4+41o>OQox2n&fb~{8})zYH71sB?B!oLqBr4r*)ZV{*=<#3-5L=gibtuic+x+(
zD=b6ItqLYXoF~Y#;PQ>pr(Q4A$?p97X<c-4iHJB<c<n7E#ksp^ko{|l0%7k`j^7rE
zTfxAMWt@`HthWKQ29`a-9&g^Lw!HbE&BMd+qdSD^5~mqmjVJ2s%{oajqpoNEkLw%$
z4(=#ff>#C-{@$^74?ld+i2O&Dfq_n5#DS^!F$s5YXLH#X_pt4L`Z8@63B`A*bYJ^w
zO+L`d{t%*i;!#(pDvp6VpqmH_$^DIC9#Dh%Dx~{2hAg}gMGm#dYPC)`8}QKX_5*D4
zxYB*fy1LO9Ffr#A7L4pk)l+kG<;BFr{1ani@4}g5xE&z`w{AW2hd2CZ&cXYn{7+6#
z!9a#t_97$P0lb0k(Ia(qn_M9TE76eAwO}eRa95AKtlHk*t|Tn%d2$NJQM99vPlq#O
zsRqirgZUg*b9pa#5;xZf2VK8$BQA>1(F9l>RdEqcHkklClW*`P#^L_TDHGr7h?}}E
z>5{lb7>7sWw>-c(+r*q;7slVXBM*P?2D$0v4x8DpANItHVP{aHQboKpPOvVb1lJE2
zVuq%hzEVrrE_A7cu;?(!UVKt(g7({S-ya93A|9K>XpbY~uQxGLv7dV1j@#L>J3Krz
zZt%iYZNZSg(h<fkzEQOwO||raNVn(PJyH04ZHU!vtdJY67{ad&LX(X($u6Ll-xd~f
zMDf~7`uCUHoBn<5e>?}koOHLHwdR>WIM`<zc7{U_3hR9H_m8iye++>`Hi}n+?20g%
z)y|><8IR2q|Ncs6Td^lk)Fc1#WR3f;oiJ+?K5j@}9|i+5XK9Z@<x0MM89(7P<Ac>A
z9l<4owuVo7f?5tw5ce|~&Nq;hl<fVPAa37E9g?b7<57Ngv_e2eW^T`qDym(wGwDxN
zAmZ2?)Uq5n_%mbr=!R!`J%<dCYfBN9)j!wAD}o{-G8PwK?=Jon74^bVke2oyxPfu-
z50)-UupQx$Dbvp0_QjZTmG0|bh(WTQx95XO1(}&;;8evg5L)BKzRqrRYth#dMQ)rv
z&I2dAK`VTe{RcQfe#TEIyT895e<&p>i5@<o=W&>{UUTk&d*@+e{Ckw!)fh?CLFE;4
zCt>b!y8ytq{r9kmSo%*6x92uCa>U7<pQs@^_CY}PDl01|Hhjc0ZS|*C6~8Di1-@ZB
zG3O^KW21iL0_ort-~Ir{eq*AFFaRUBmo~;NzQ11?dSnqi$CN)_Bb$Y<V>jx04If9}
z^6^GZ4r5NGqD(kP9xschi!Y}W#sPnXPsn4`x&6t3b!Uu_p8w8z#RdZl!XjH&SN9M{
z6#VI+sguq&KMGmxu%ILqA3uIewbV>v88~iH!iD3hSztE&P@=a#O=%!WiU_^7J;32K
z`y~%vdrIK7MJ)k8s?8UtC|{$E@w)P?em|Bl8tR?>M1mq<lTiXQIDe{Y+=~a-`#9#T
z9#KhTX_c_pN^c+&Rn!tZJUm6g*q$5l=<#t7WE`$pWD-QRbhMIF@(9<noQ0ZN?Ak}s
z?{IED+~I&Vsvp1ML4SdfoJ4Q6`*9m-Kx8D7#4_Aj*4%}uiUW&#sF;$MO<3-`18R~|
zQcAEi_vh)0>o|_v!<zzd8ghBbJv9S^f|6j!A`*%LG%|&*`xeaFrShZ2=1Q+$r=pz!
zUo?0HxfWhSO>WJF=t=rp)SNm@u%iSGdlMxjWn~9pXr&oNJ3r0K%iG%C4rut@qo(Jc
zWIkG)2A=WJNyMmFnmaTppRP<LRG*)@R_%?L&P_~|*%}G#kJ)CXTsio5dc42#`y&OS
zs;nH??0qNU4!dC>&pLV#H-5fp%1!FY+w|M&bw?<hf1;nznK~{pM=p%)5r^e|`pzgm
zwMqAVozE*kD%&qaIhy?Vb`S2NfQ}5MSAU}t<)-@V*{8FE7NK;b?r4}E-*FngHT;%W
zRP;CO2#upaP84!#&dSfWLTc0zG6G>{dOEsqR8joSitL6RJ>VK>-$Ld{$HXLk>C&Ym
z7$qFG^U}<kh0k;<98~t#$C+huY8h6BazdzJL`K;VZ7LWW8y}%tNhbzo8jQ@j+1XTR
zKSC<ZDN{NadYRMq1ziNh#OZG^ucSl!xNO#4DY6*nH83<pY~rbF#cxp<%s5mGR5)6W
zgO7RiD9i}E23&4B*xH0Py`S%j%0T;}r>CNr2q`F7msW3_qTIT{*${neV&dW}L29}R
zD}z}utytlF;P>w@p3YF5_mtbOaF`A<ahml1ST02^^T_63Mj5=Gs4PoCO%$1iG#R7%
zk>JfutM0kq4Ie{z40~cl{1Y*+T>G(PHz>)1vX^pkDS!CzA$TzJyuqJ(2TjZg*OTbo
zS?U|8c6a@E5`6ITTAq)v9E6@tI5}~H&Ka66M7d#rZ)#w|0hJ`JTy){1tDYk^>qB%T
z+!osNp*=nF-E-47^D#D=84=aYZ&bJ>T;98!7!-C>wHOkL1%{os;NZmh(6iHHDPGqJ
zJX!aPvB?UX7le8C)<$V1+>)?1fvkCBGME`gwFLKK;A{Pel)4#nlbIz%N-CJ?I~6WD
z|F3BdRHpn1E_uU)HZmo*h|#s|AZe5v3Hr9~v9l||tprjPSt64Y@J17*KQ7@-m%1#e
zX=o(<M+%K85k%Gqbr6mo(5b1<I>R}Y&|%XWr5<~5LJIvocP8Ql|LdO!?uV-C7cqtH
zW2Y`n1?0n7ua4{5Ym+s4FnFR@wzs#@9euhahxk+JIL`%{j1;D+uwke_ul2;h!p81{
zW4}!~kTbWrxj6}5*TSylfecR9Ju?Xg>kC|{>KSn3459KN=4_j$LS^C;5G1K)X-fE$
zBPZ5t;}vS~xV^*OHz(b;edasDC_fKjp^6$FXy9D-#N%~8Dx{GPmH!<IZMM=xq&Lg>
z3;x&bAU9byO7kU;m+iRykdE7P1k+6(d<~oZC|D&RbgEfd<u-jrsn0>Px-7<hgeal1
z9*Ti^`89+N(mL$eX6T1ajj*sV%@T_!E{m~L_v1Z!a2$tFtB@ia#qql5u8wsRLFFAx
zH+6s?!7MuI;1MYcF<SA-llLw=i*Yw7pQeg^B~)#}eEz2Ct0<A_-s<r3WK9jZ>ndBu
z3@SGi4wZ-1P&G$43#QUv-zM=4Sm?|DG?Rdjs#XT&h0PncD*wf3sHr6NcOgH2V@LxC
zqCq?*d7U<6!MPOv6kASO*ptST7_<|1Z%_&dj4wZe&j@EhKY@M2NSSSluG=3mv=^uK
zOf%NCYpT$}NBbK&A1OT4Vd>+wn(~zHE95oFVwX_@r%TzCf?nHKMn<lZt@KLC%KE$;
zA^!f;!`$@r^b2Ij`bGm$j33&=c2nj|6^*y;l^h(3pZX9d&9#SwlrzJKLpNH1q0fZ#
z0+~cFSi5pKto|DD)nL6v)W{OjK22pWW&J|S4GlhpjnU5+8wl}vcX>dn!eK3+_pjM?
z?1g`Su|J7SDy?6>cyVz+O>Y1q)<B`L(%~_3+|D5bErtdjV7xlyPc3!p)~#38Ch(o8
zO@V(u%RS-LU<)gb1^_b_>xvh*VHhfb2`O}FmX<|XQZeP<Gp-vY27w8;yhT8;RQm?g
zKQUUsHJHIMqyD$mA}srZ<|X6Fp`o`gf!!F40CYeL3JgS)e-~>NoBfHjViXY2@h>zT
zVv&jBRSD7aD0?g_I*=kqQFXdE+Sb`AU>Tg5$;7NxEUlCxI{?p+m6}S2{w|nr;T*>P
z`1tsh7UM(96E#aqOD@Z4DgDs#=--2vuFFzVh~3f75_X$@crv_^!)50+;Mz0@xCpVJ
zTj})7jKB)+q*VhhxM33uxD*;yP>TEhuqesUZ}E*@;dNMDY*jni?pKnF;uck*Kr`*~
zLqPb$+4FUV#~YIfCMITv`2@98@avt$pI|mboh>IJp&%*wWx{ns078{Cj1CwtEJdef
zv|~0E<KY6X#jr|L<Le__X1`+L9gAT`(LYCuOr>tUUtnNjQic_{t<t<ZT41EW$PXRN
zC#qamR&r`oot&Hu$ZfLHyG~8RlhJb}PV{5&?E?2>2lL5l9oCwYd@l2m#A{@{fe|UO
zvBD+e4iWG!&FQ12uSBM3&(Vphiot9h#*{vK@TrMM;^XmWz7*WDtr&;<qR!@<lvqkX
z`s*fOSqtkTD)wSTXR^h3xrF~<rbfcMD|o8TU1hclCckoDB(=3nqHQ{}VXamao3N}p
zUav@M8@G+(v3)vT;i&fX>C>bw!qMCp?=D{>RbTL%&00S_h6OyB;b}Jj8V;+eT16zu
zjIED@^JU{F#kuveVzc3R=oE+!HT6tEDg!O(QRsvXuoETUhRs?zRW_C^Ln82}@#aXS
z^R_c~V}4>viYed=|3sKKOAUBJ{je;}+40UfR0M^Dq|DB~0u=DfzYqc{fR)^Oo)nn~
z8X+N}PQddiaxuZ9$eOB%$Qx*~Sq;VdCA;Nl*L8YFU$}>*$@8-kXvW0KO8)K}@x#Nz
zBk&;EW%I-mMqhsfF{<hA%w#ij<{>eVHhlQ_kx@`^q5z4KO)0h*Pyh8xT>=fich^cL
zgUj3j7pi8e-&<OhIO;kTARr`MZlOSwYdlU1ZIC%hfOXmZdm*|mX#l6eWqG!2JtfA&
z-#DDwBUvUwRo@StxNj2Ez$Cvz?mT-#Nii)3=8<jv`>XTf6z;E*Q&M6aqd(uI4fpT(
z^y!~%Ky`Dzwq@G$Q+4lTBn%-GNRCHGMrz-0fw<ltnUb5^vkJzuMq_pFuCO8}BIh)6
zlhy7r67yLl6Poj(fU;cIjR4jKmB#=;cA9sBpw8lQ9R91up<=wFql3vd8fJ%6v%taB
zI|E<(2}^=CqL6lj=B6g+=Aa@!$MvzLGIvPd<^+qwozRY1o+VTNpnEWxd(us*wY8O)
z-^uFp3j4!{9F`LZzYm3wE_$)l){azEQfiNc1MW>GDxLik0UZ{N*;^(RNQv?vuNL@s
zWs_+A{M@@<wh(P2cc$vr_&N`b$;O-+UKR}R^oZg~_y@D<>8A0RXv(Zj8uukH^-8fa
zZx^HUn6*(g|3dqfLB*1V`FR)LHZnek%yPRWMGGW)xWFhRA}&3B>)%2DxDKe+Hv73C
zHIt?<1Z`bis?dA7?z<|u6oT4uadFXXpFe;8;^V_yaIy2pkIl^eqgqTnIt~tH$O)oj
zYHNirUB29AQPOTv!pOqXRj~GFrlq};^F{sp?Rl1BNR(_^;M&p~HRsjnG*cDH(xiX%
z=vP3KMafEb#k#+pL)Bn?eLdDj9~Ng3f@m$Ds{(!$*gx0$h)1!fO)li&_F=mb+&ihL
z66@K5RD%AUA~%r?h)SO2tN=m5D!j)N19N9V=v*V&n1Pd1#j^S^86AlvdVAtT?N@)X
zL3Zg>%vZv1%Zo$-fl1^RLqL!oqT`_PVRXcdtH`+TX<Ay^2w=-Fee{ageIttqCI>v0
z-sZuGywWi@KYwMt`MbAosU%oJGWyLI2wNaBXIcSrg3(87K31AnZyluYe>dlgUO$OC
zAK)}KH!p+3_`u}Z9yd&*d^sWHw6L)>oS_@1A+<LYPv~SbzHPj`t?g8cypO|ltvWLl
z?Fp8OhQui0AnC|QKWz_TwWYJxva>5(*m9h7r|silJ)q^{DjqxLciMdZA#BW|1mFRs
z1o)Jzq>X+V8pOrI>b*kY77GKLk&jOc5RgU5Xuyk~@nS@Ucu1zA(s2nxe0)5ANcX<0
ze<Jiw|Fdgw(ONb6%?l_`26_gDEo}#g=@zB{gBP|Sl~gp}KLo#OHK${kQ34@fhcuZT
zAV)MXGLrJ2?~ciV$gTzLp$to;8JE^)N+HviOP<Th=nI8&nJ<@BKCgR=)dFdDNj(x!
z6_w$X^#U%4q6VM@7j%Zzia*2^*nsI`AYK5<N&4@c9d9tQvE^y5+rq=aI-wR;LO?)Z
zyfxbzeSWsqTHK|RQtU)Fm`}=U_Y7u*3qinOX81^r$_+xoWLSqK{RNK}DNj$fyA6CH
z)T4dG86#K5ZssCnu-qLzMz7zy{{YZM(Iq)<Ot`ES4F!gVrh0qh0L9}vsLCm$1Q8HT
zD!>Rrqs?vyhyhHTLy$91myr??r2xXAgQW8>gw<gQk&O||N=&4Zi4i<~`(*&_^&KE5
z_=gl@*`{m<z(Bg?WHsc3!BmKj>snEg><*f!aAJWJZCp}X8qGp2<V8KQehenU3ed=`
z%ZUbaOFeP#{^B&(542nvf1mEpRy(UJ3*4{{HX+mINi_7m!^T7w1UA#TwjhG8n|JOk
zmGQeD*+Hh3Z+HQ{aO--%g&`G~Z$0x=;Ak-c`SSIv5~zm=NJ&>Us%yb)o}XVEjF)?R
zlcWSj#~`&TpZ>xq**TDJAmt8_CHmY0Bxe2SROZs9)(1FuyzU_g0`n<VEl1a7`ZKNF
z!RBwQZJDr#V!(KpgON3a`}wd0sSp;(5dx@MfulY@SrC#kH!T?z5cfr*;VHqz9SHG1
z{KWuiWTN>s0GotAG>|lK(7OsR(Gygeum-*=`r(0_P?O2abJYPr;wd8UE=%|W6hj-t
zcW<D-8Camcbd`)-&>w&or2Nf_do86P;~juBTB3JruASh{9eN<1kEha7qSJr2V4whh
zIsxSg(b+(b|Id41r&QM0U|N-^W^1Q?`t-@9tOj~^5tasq{k_Y$#B_iHA+e>~-a~^z
zCDLGOI=W|CB^Co4$oVmYQI&d=!RDbym&*8YNjjRfHfWb*_Os6RCF3(W+?tD~Y=ED3
zQpfch(@kIYp=yDoN7q8HcoJ=w@WQ{Tt~NF{h59YmfQvF7FSkcyZkS^!un?i)7({Ti
zXSx^8op^vf=}<ygh}s@8%~3HlOolWk2{~QS)2wjFLVpB8v5rkEHJZP;jB`5{okzoc
z5_B2d_Bz{15i*8s3KA(KxcXok8zzA=T&?d{p1wVd!8B;uISUF>(oC4euYcEnfHwxC
zfm(e0{oSQLaTxmji4p;~$jP&-T=#W%stY9T*C&^+-@26u(A^XQkIVkJW0A=K19ZL#
zTnLE2$Yx(MMnE3jSO2`!*N0^oMlD2GBnQne5HPB~2CO%bZ-+KQm)E;(_~*q<cxC@<
zzB59A)|$_5@mn~zrG@o<W%OL?j}yHj(F;)BZo89PEAk(B=cXXjPx!}EQw?Y_zydC3
z6>>3xgCEe?q4G9*Vx!W>+kzN(_iN5|gw9V2NUwdA?h7<V!rNL~6<~?ubzIK`9Qw>3
zy)6Cs@$BnO^lX{<3_KXz(HeBi&TN3(GO%ZvR*^9=F@{}{8d7NOi|5ftN+*oF!8BDU
z6b4G-iHI@^_53p|x@bHK3wy!eJY=fq5qjRC1}tEHjFeg@pqo8Qw~9lP&vCs!jNRxK
z8Ce!&@1{WC5cArl0>_-?i%wviI2$-*`j-Eid!euPwdG$LlsBLoQyI-=V-xkCUkTD+
WRk3m7nSytqB*bK&WIulX_WuJBaxEPI

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 50f90764d43837f013db3d000fda18dfa35c76ad..0c991fee2efa85784280e4788c597e65c999d394 100644
GIT binary patch
delta 107353
zcmZTwWn5R!(*;ox5m4zCkr1RizMvo=-61L6Al+OP>F#c60qG8Dq(i#9yYpGl-~Y|?
zPW`~WcXwxJ&YW|`ie|);X2hQ(Tu{v6G(HkpI2=;#2^Hb}X1={KfoE@CeD%JE^;AH>
z<k43pN|BHpYd&P~GvE67_#Ag&F&7u4ZoJ}dI-$W&PxSN3hYzNTPsUDqmhvpDmg8j@
zD28i23oBtf?)n4!DQjdmX{nbkl7TEvlDbIsmVp7eJd2AVgo{f$5mnoZ<B*#=KRLZ+
zBYt_@^bJe-v`5KkpOb%0xyp{Lc>N6DBa&~Y6o$KxNxnVh^S<FuLUr-+s}fwSBxK$`
z4xe^-*}DbTP|jbULnTgEUMAqpU7ZlP5czqn&+fD!!8gC+DIBtwAUn%;V0Hv@OUnZt
zG7oeo3u1?l!uPLP-47twu3)G<Fc||oB*r@7TwGbx&{QL={1DCKfrE3sJ(W}WkGIko
z%*xu6%BYP4*$m3s$oq`XKRo6YcGh`{ukhy6if4iSTwcK%&u)J#?N!}@$);arIt`Ly
zpj2>}XG*p!b)lc%z7Nv0`*2@$22asV{!_k3(s3W$6~VHe@?EajtZ_k}u()~(``#6M
z8<*!1IhS5aQ=ti#?i&<rpA2;PCi2~LQhx7SS(?6W(lbRuhIJ>V^ZRh#P$bU|&y$Sv
z)WA9Q9`6?9AzQb2d`r3v7*v@iVPf#7u~M0xR~yLM9?-W&nG_gf91L@S#-(UA*uR8S
z66SgKq!{+D(9<(g!&sp`ymu2=pN~%QcoME~r01|5HqTDBp?;PRX$;A-JJOpI=LjSh
z_1}+v+8zG38ryMtd{DeP<b=^mo}N?+mP2AB(NbS|cRr7#+e)^%gIj4hL-A|^`w%n~
zxxb`-)|xV2^?i(U-JDyh#pPzv8_B$%+Oxx#t}uA!Lx+0`t<!F9*irtOd?009!n}ES
zpLB-HPk*9+)z{CvZWo(eFY<}QVZ?=+YmBeo<B!}2&u$mN?;hyyxI;kmutr(W3R^^h
z>crlW<w|4BcX;&cQRm9uNXE{#)gKn~uA6DTG=|y6B)^Ui70Pp`@k0h*%eoP-T?@eW
zKazF=CjtFCiflQmM27_{W;}#@!}Wq<?TPz{)Q>Z7qD7n6#7DWDc6YzC<refkHW>Y}
zwS~}HC2@(hQz!e|Wb-<1xILaH3`$eaor5!A#_OUnZc9<dS9EDU3q$;AaWa>(c0pfi
zOQWa?o6$Ukbg7L3;AiM!`K;}iPSRY0D<FGjz2?)A*Ik`DGf|}Nonmk7Ox~GsUeq~8
z*L1yq$bCtG5?7knD&he?I0fzBO?U*)e6-bQ#T1o7O{>+5&Hreu9790Sc_jMlq%~@R
zqi^xVOk4YC_?YN*Ll<8?Od@w&->Nl@@-Ye*k3<+zuL7k)y3F9lSQ0~_HAGHL8sX06
znv+4*Ne3hQ(#=aFnAbmGw=Y;CxQyv*XEq);+%%J3QQ@6b?^+$3^P{%4{t`Zq3+)xn
z@ym(nowxHxZceS=1rziy5EDK;M*Hp~sh=2Ec*rA1b$exfLmw90M#;CQY%u2Y2jsoi
zsF<I%wwTLmMtnP6XTriDtb>ym0wgTw_ufrl?_P*h$qc~3y?_&b|4#0QpepQ`l=Yal
z>&BKBCAK8LaQvqJ>sM~RULyYV;Xd0#M8~DRSfv4@Qv$eQh_`^MbFlNv7H?=&m_S%!
zujbm2QUM5&x<H?F^y>?@koc!+Ettb*cmeQC*ygFCGYmRpLO-t{cL{=1TY``kTPf9G
zCnr@4X`cRH)q5%&J~Ri)Aw+gQtZGE_33RrpEX?`8xNeuyoX}2Ue77XNOI`#aUdb!+
zyi?5%V^vPO_%!20#D<3UC;YO;`;6%Qn3!;6DqM#|j3GV+Z<>cwoF^VkqJgaPUfQ~X
zdCSkk1)&x?U|v7GPcbY(%5%;sAn*VO2S-6O0uz0<922e}UvfR6si80Ye2aK){#rMl
zyC}g+uwF`j3HH=qyeqo!GXi9K9%$1!LwdPr4(@;6Vr(TWOKwq@>`N-#Cq2U^Qpx)6
zT?%!9yx}38wcBs5K>ja;_UOcwZRgXV2q6yQ3gR1tICkucq{UN8OPRF7pG8d_c_gOQ
zC>a-SLfIof9mJzlepkBH!d2Lx<jE27%i?dsCPM6CZq(TKcFm@0seb3u3rJVk?qPrw
zxw%nZAV;gX<ElDY?{ADPDKBk<h83aQom?2U2Ge;dbYk|R&m3QoUm`S;z$+D^9ObBx
zsTVfpJS31ffe#vvr4hQloEG{6vzT|goce?+;_*!2)$Z0Pnp<k>OG_&&dbaWk7yFN<
z5urH9ZZYDuA}frg7G4|h@73y+v{9xfH=jJ-w<@$W^AwS<R9hA^orM1&?wjX3K%~@~
zv3!WX$9yHojpCL%2v3FreY#D>+)CenQQy9n<UNNWGurQwmC&S7MFG-t1=NJpTx(jF
z*CKyrdh_B_zcJf_o9K-=;=Yd}e;d}RJ(YOR+xJ}XIj}2&nk(>i^fPX&$o|1izWeI<
z%`;V4?2vWLQX~d(N5M=DfAEcl&UZ3dRk?2>5AYCZjwPU@)d4%^AlMC&XKO}8Y-aL{
zAdtZ<oNoj}walgD<lIsz4W5)JDUA(O4ZbCE(j<~FJKup_Bo`Oj!N1^Q$r-3LTl#E1
ztG1?k&d5^EXKWy5t8CqI{WHSgaJq_kBChwt?N`pfqkzQk7HE+q;OX)6)B00AvGw*0
zVr<BQW$_iRVMkV}-uG86mJF+p>YZU;ueEE>ET?5luQL}PIB+<hjz3ByR!BmafL^h@
z4xCTrD<Q|qLh*l1_2vcbI$qq!8$D-pW7kV9&uOcx5)-Tir;mk(M|@i5swqF|iA%pc
zQ$`mB*;`mNaAme9DVqDY-ve2%;HWTp54&?{)?loT4X!W+kt!nd1zW=RUD+%-BfRoe
zr3VpfLg_O7$%e*{t(ZR4SyahXn!&;ziciQ17~Ta2i-f{oU@$4vc4EHhR~gpc+xcs<
z2Ytu2K&sacCo)hyYX}ciMv$FWT4QKbxqtd!O(<BMk0y1UTO)(;&sp1tM1B0~TK#<p
z+kbx$09i5X;69$M<GoCBSRY=Dduni0#Cx%JB3fc~VUtzYvdcpx{lJ~&g+FbbwXx;k
zti~km7L*Nyt$HgB>0%G<vC6R!7+CG#B79RXP38-TNH&|5^9gT!2HR}+HiI4CD7z4p
zpLgdRFbr{`1vrGri%vr?wg=u7$-ZPL{$oxzz42v3+FF3UlBJ=qufjL?M|2=e=<`_(
zuVPT%ub$!rbq-(TgGan)_;6cJPjY|n-$ksE|Bt^nr9v`rXQsW)Jdli+&TF`7h4I{7
zw1nXFp+b+ibT$iI7K`#G_r!qSkaMdmhNt2P1vrm|1PJ^5DVB0&gr7VOjOB}A(0OP?
zh;OMevS6KEWMc32x9Z;6&}CXT5syL-1QagVujI809C}kkLA|o~s<5__Hs}7e?b?D|
ziOhwgHWe$HLuOBMg~PagME6YMWa$+o92NHI>&m(>A-HBVM5%3qxwtGh_gJ^<boA5r
z#!p#GT_H?sc_wD&3=6-&RbOO|gErM#ON=Fb^b#8$IPGh;T#(d7XXb92`r5zs@y~?D
zSgREfkX+QgqM-3ShI!7-Zc4Xc-l317C;u|rBqCem*E6l{S8$H78){R*9)9gl)+0C|
zB|CDL`VNy$r@!EhLen4?)^psyqJ+v-r{jlwzSa;B)K)Q=3^SNaDo-=DPU$~tTDh;K
zwZx}7#DdjJ;M5yUw=D63bN+cNR^XE)bT^@mXD5^Hf5uzlQi=oTCrj8o&K$a;M^h$G
z<P~d<uj!IyES6S{^3QHm3qJ|W?V{a~QqbTrSy6+Eu39>z^bC|Wf1Df6Dk$|Hs`J4k
zfd?FzzLFKT@q;SM$qL5NyZc^$<kIiT%@qZT`fZN#zW8*sGAOFB9cUlrNk~iwC5s5e
z&Fu^%H?fDJr=DJUjrT{77ha1#*J@%YwN0Dpj043nl54mzh2$#3<Xt-JZ1XRg_g-Gp
z7AC5PET}i7s1YLK^-vU>=9=6!{|YGghY6QldSyhfC<Kcgy4#il^N>?X`9x7)Aq4(#
zjHMQGrBqb4ct`6>P4Uci-R|uJDi)59Yk!ms&jOr|31iP}$vEaQoOp=E)b+F~=e78b
zi%$Gf0Q9pzKO(btJd?9Zzq_?3LCp5IV1^`TYzdiy9kSY#?AqE`;CdgFRFx}d?R?1e
z<caubr6w+J_L8xYe5TDzE%Mvye&TSGFsXN_D4YW&D^`@gjZp`QRSOlm2TAXwG4Sl5
z$3g{Wn2BxelYz(Dpe9ZHMx!FQd7)D8Oa#~MC@CmyP4MyR$wfVpxU$%Xm<3FMwt!^b
z(W+#^5$F<X9cm45%<~LMi22=vqA}Gwu=By#pt(zuwLe+dc%JrRugxj1VzHb+P5NET
zf|X1f$Mu1ZEyUw}880U&9{Vd;NTae>9{Ki&<mfnb#P6y?EaA)v7hLy+c=bxz6FV)V
zEs-XPg-U1p2NXe@$ve9@@|qr)=!6zAKOlcOu{Q|@Lk?f39h6aH->v>K2X(Z^30Z?@
zM}zS=1=q26Fu5r${?dhf>~+XD4b)WEuwlm)N2yn>Q?nwU;joTTRvE9U*UO3hW=_z(
z*=u9G2jdDYPJVUd;*$^OuRcEL5J9Ejv`2G@L3-OkVq2T2A?m94GDN@pw#KybK(K1h
zot$r_dFaIX+>LC+b#62H8ti4u5ALyvwUwU1y&<dxVH6toOw#gH_<7^yJ861`(%CUw
zQRBnjfV+Ze6Y>}0y=URV{)y%Wc2b^7vR0w#cD4El$HN9`MkmWlZ#!J8%2B-ry#VO>
zp{5dyh2E8ZaLs%Ktz8CfJX7qIRtgSEPs0Ye>L?MUec4_+$`r(U-II(Z?wjF}+XYWL
zu1u<+AVfTl(sRK*=$EJ6z6k9UQUJY1;!xJXjkk365a{&2^d{pD9nknNm0Gt>_JsZH
z8`R05(ALydL@$^1E%d=HpgE06S;*ME$U#HepAm0{*2L3W$40CBtoqRXT`kEYSy5S}
zvhNml6!zc3ME$G3D5tdyq)4alD^=Sg1QQT{#ME7**>1&z`=J)6$Qf1Y$7aaz3mgyC
zpIH=Sn5^!B(v-8UMZac5&(VtyzSZz<rJXaqJq>*rNxzu(aJj<{D@4Dyd<%|)i2TMI
z9?2Rj65v@kw_j?Ys75@d#x^esCH)W>l*58DtrB}`t#=%n2O4IJvqq#)^t6udo`&&Z
zq!mYXW`oJc5z4md=(oUSiu_-mF#1mU_+v~qPR!UB5&3LYPd!#Ros3LSH61&<C}U>&
zhN=K)n3A>Rv(7et4f!z0vj&A2c5r{h&A$EE_TO(RCDXXLxu{n%mees_F2;0+r}gzs
zi~`NuTv^mS@3tl&_`j`@$(B=6o(Kzx^iQSEABueF5JTpb`|;1NWVd;DQdET8nriuW
zM}JH45Q|sbjf%S7ngfP_kY%&s^QOxVVZV2Zker)ODKcw5&1bW5{D9$#E{DUj7*}U&
zoE1r;7bL7-yAI_x3DiE@8o%G@L7*=$$@^S)@oQk5C@Zy>l6sGrqf@kht>QUBLqzaf
zndjPlnAn=yJ74v+HIz%%e7+3-94JM)y_k>w)do2(ijjxkGhYQ`IX%NKk91rSh5|Sd
z-O`M&sB*-3w^8DnzI1;VOw3h%DNyMm8K7eO#-`H354EBDLIFc6o07I`Vg5_Oqv_gj
z@g>oWgLpVGbANQ7=Xut<1AYPSHQS2Cfr(7d0EPs?XqP7@a)m2LRu_X&sb|^L!-M9<
z2h3Tzh~*rWtZ>(L<}R;&13}nkK|@BH39#RoVNr42fi|LJb;rm9GV`4T+FC~MZlrtZ
zkF&>$*lu|PlJ{i7e~E+C*wXF&r%`nAvjEgY80`wikKKD4N9&%G!(96Wk11wG(e>?+
zhZ+X<a)dICih6cGcDvXUP(bfX(#^<>sI9ZKwm1844pD}h#|x%I!#{6w`*q0G5F;M9
z395ZIWq4}EB{_u0Z#kg#p~CUoYB1~@(WPf@o9p+D?>h#4t_0F6ws7=1lhh|as7Yx+
zx5sOHk<2KXNj_xE@`O5*94A&OBr4{^e!$@S#HpT+CbAn`jyV+ax}v868Me-~W`7Lf
zs3LsZizoWle5G=apj>K(uB1tS`x+IywxO>5#D#+Uh(8TlaX?KoBOYHKui*tHDU0Rd
z;&(xGyRo2Azl+i1jg`%*qR{M}ybP<Ee#bu)cR1)Df|uT^LbDG?ndzm_nURS-N%9;X
zO>N|0oTo!~-DeZ)dSf_Ps)t`=8)!B<S{)pV0f~Z)YsfGwU%MYabw&c$J#kpE5UoIX
z{%p290)0ar6g6g0oGVHR-g^mQ&7^$faj*IDs%PueEx1;6PlX|51Ae2UQ>1aA6`rE}
zp<Ui<hV6$t+%s{f(HkIB3~EkNRFo5_E?;-F2%M>M9un$b+eh)bUWaGxO6j>2;RZlo
z%J_OyIF-gY=O<E#LB6y+S~UVlQlWr*#Wvhoa0-#fv9}Q3Rpm@)nf)>H!JYvu6qLA#
z%HD$B*_g$Qwr3@#JRnnQ_Qz@4uOE^5+t0Jq8e!a)+S7tCQr*9$_G7JQ_Zr745@<@1
zCh2P%_p0Xk0`nc<1r(vKRf?H6BO<@ok2n3(;qIHd1o?l<#S9GME_Vk@lgj=h7FWI;
zxS*a`{k&9}|1u&C19P!YZ;|_<#Dv30L|^Oj@omjMmz)X&<V(BN1ibxuDH+%YI<~kV
zbz))XE*tJ~f34AYxJeh!$-U3q0bibhvZ#AH0k6c#GB*T;5CW>rh-3Tu+WxZT3@(10
zIc_t&{c)QYptIVoxPICle`?68X7g^MOVaJwY<vBtE3>IWa`j|?xteHRPisjm3Q1$|
zBwZW_C-$dN$Hskv^Rz~mzkJkxfDjv9<D++a07BGdq?Or7V<=3w^PQq<dm#(mQ#)ww
zMWQ;-6DQUBKSlC+#`T5F`({FbfN_dKr2y^KWTB7dHG%Rq4jg}5?bp(TLBcj=rLn1+
z{8=W0hk0JTwc1TbGm|b42QNHql4Vk4at;4DG^n)|Xj~JG&>_xEey4CzR|*<#uTeb!
z%Uw<xxo#YtOsr>h?TpceBB}GMAQr&nH1zj|Ma8sF(+LH~Ngumb==xM!2RHF@`-r5^
zd&Fp@u+$u|5^$<9B=shIa8*Y&73|JjYGC_)h3in}5=|s|^Qd9KK{6yFM0p$U?!4b4
z{kL?)V~L~avpvBQWl{;#c#oOi!i&1(lPu(<*N>}toxG`qAVp2Xg@zww&?gfu+a#*G
zax_IEkY5uIE2PS33?u>5YG<mt;fq{Zu_J0ryDp%qjy0WEwbymnyTd8ekiz>x*#B4S
zl$UK9Z_ej{Z61Du=@U$hgg_K(tVs)1OFj&)VaP|&Sgd!&^_}S_r4$E%^#*n4+2Afg
zpD4nCe+vECs>Ib8NHvv|dFeiS?1~70<Ft)4pkY>SRJ$}pA4<sucYcfF{q7Q^X&10}
zaiI7zKd~)ElR_CSL3R#rqa6CyI+gg6i!`F!m?!^p&26S^h7$f^%<ZscQ^(!w6R)3T
zR?D$By_eStq=7e=&!t!FZYkfd@yrn8e8E!yXHi{v6T7YA(D5DNf&^*KPTSDgE2mv<
zMlZVi`l8LRD?I&#v@MzD$EvMyxUg(j02>3kN>Vy%pZG(i=S>n%OI+aNW()&w#<M{i
zb4!r!Ks$E59LFXf;ortRJCMp|ar`r8G)lnyP>^7<&NH>~D`OkM_z^PX@!*Vk1BT0l
zZ&$mVJ!AW%>`2+_%JS(5ZQ=m6kUUc#-hR)3nl<q$4uF4`r8;ocSSRF%v*qCJ6ui`x
zi=V5E+t`bbV65GStsOHn?;nUIUJ<6<oD9QOMyqwn;7o;RC{KP*LB>q`FgYrgi}HNF
z{<Xv3cSUhegUk#1aaSNGg>KoPgHrvE^;$gDY_38eD?WmqipOY+i_ze)7(OIztyj-g
z*uI7`^djF}z`~QFYFo-mHG)5H;UJ9!PFd8%c%byw?BTk2!W?=hADmUjBaeCww-eaS
zkT{MQn9RBP*=YdY_vNWSaUSO}9n13KT&Rm9zBS}5w&Ng`ZP2Gl@H*t(eWGbk#e?JX
zl^j07){L6#Uj^2iEMGP168R_7`J2@^t=slYt5CJ(!-EX^z8}G?^mV@j%Hl%@li9H1
z4qmqj?DOHe$YlhnDiYFn;Z6z1mFIsG;XjDl?4qZ;(SBe~yAYB&1HND@9t9c|m2rLr
zu#eIh`|RSl8z|L-WgDNwy7P>xCY@w*|A@t&GX&MLC*9uJ6hhvRr}*#G%~*{h9&5{5
zapRTh%tm9cj_*7hTHn9&enZRY-}seVHTFnA`;%NYUP4V3x{WKgv?hrzt37GcS8044
zEsc6h@sF9UQ}Eu8k+bCzp_~kv?A5)#1?w%gNaRekXRGA_SwEl3glRF*GdB+Zipbzm
zTOBpNPZY52>^4>3@?MoMle)D~TCrA9X=Z-$Pr@m+koObh$bWe3!9Q^!)f1FA`q{Y^
zoZVe37B!IwdS+(5(}e=Y8ei0}qN2VC4MXGU1C?SwAT{_8!oD;%c|Tty{qUFTGJ#2D
zQ<b~Ntf?($qcA^IR<_yH&<j%`&hC&u!JTrn&1U1Zy9*v^o__@0X^<3#2cX;rWS`SB
zlsEq#hnMG292~xnVA}CczePZICr;&ARjM;wiD$avJ>w@&KuKBkrRJ5im`hm6_Jho(
z^yQreTkOvKfAjaReUKD=#YT-l%|#(5MkBRWN%8aULc71mtGPYNr#8{4l=sheTFpo3
z;(5ya{7b%1j65mvEHRkyqjeD))E^DQ!HXICl%77u8UACnOXP8Zb-oIR$(EX^nrtqM
z(gbHnEP;^pJJg4s8RxxW&Q?T_*TBJ$KzX9RysIGV@sb8gd{R`?bzbf+Z$(Tnlh=)f
zcOs}9TU972Bf&_gwyPA~cg^_F$A7EON_*MFcWO4L^HuT!K3T^2r=^*1p$Dml1FjCa
zr^RtIQBUG%lQ}i-EL1?7anT?`z^Kp?uZsW4bmP5c^XvU0D3qSHe1uEUy4?&e_T2XM
zfurmU?di_U`^^>?fT&b&9KM?OEXW>n+XOfvm;?OcQ}6h#@$wC5*U!mhFjC!b?j^^V
zk63yTjGr+X*ToBqXgz@ge49x^Z9kq$a9qe+y?GW?lbaAg-Zf{s5Ubm77*DRV4Hod+
zo}#xhB=+CVN+r**M%G;2cc}cR#KD&Sr>{JF#E9K9W4OExq%@y;%WuK88lOp5(ST6I
zuJ{$f!E(GUi`wCnf;>1;>s5|;n2N&ZrFOK_CyufO+#Sy{KLoM|0A$TY^kTSh<iq8O
z`TRi2k|Ucv@<H(<WF_RcUWw;yJ`nM%hsylpHu#MpQyUewCMezKHQyBzD>_|uUa7VW
zb|MOmoZAtqsH!&SkQEebUq!*$ml28|xXN0XOrn&Y??AO3y%KYReF{BWm%#@VX!Hvm
z5@I(7gr?P~!?}WYYaGuxHmf4P!-$8Lb3^a2x^cwS&9L%3yETU{A!O!t#y#S0HMmki
zKha{I!?+@kF9q9;eM7N%^>5}@bc>C%_6Iy+dsI}Vu?N*ypcB@Zvlr-p_EVD+E{$BQ
zexdiU8V$!QA&o0!r3@XvN~1uzcSrSuy5Kzk-8^e$VYQar-hvYLgl@@O$dv8Ed7o?Y
zD}ozl^uiZo*{>3z1x1J_fpDX~xxFjWOMQlaCW;l%Nh7?m)o_4h12QKsQ6jPbE!NZ1
z-<{haMS?k%`~x^cdjAC>Ke&9Sr`?4Z*x3ELmx?8XrO9#&xo^y95E%~h(f}kc0));f
z@u)L#PAyxIC(SDFKJI?8?AXl_gaar=Dh$YI&}5#Lu(`YVH3E_AcHlyZXRBEBJ_#IR
z6t9GW<$I51AMWaaVlLkI#s*#JVc2q%$M07D)F8l!9OEYs*-@P@D68fhZ#6Uc4RzAj
z+`)jS?K9pztw!Z7a)M2axPA%AM&Z{dY$|oQags<GHZbJXBOxTd@`nJMj3q%VuIFfZ
zW_$-*vzGot@?Mu^`z}DcG@RVZete@kJzmNIOFTnzBd(9$3qr^>8elSamT=E71$|05
zFYlU=nd2l(8}PlSH;f+Os5G^VeyJpxw7xaAABk3Xr?^dr?VTS!wKbZ(@<@84=%N^S
z*#KP^w7I3yAJxiN3>{D@s;ceJZ<<?}&x!-q*A`;+zWm|DCIwVnj-qp1TB6g($kbeV
zwA1z@dG8)0@1I&2cCZrGCf;CzW*IZ5P~gUpd7V4)aOgg3{&^N(;a$oqjrj*g*1doD
zV_Z+3(mWK@E5T3GK<G?6dIYJ)olpQ0z3uo~p#^jOR<`3vGS3XL<9eP?s6EIv9berP
zi2^zhRWzlx<w_cJHq;g8LWGlt(<kwOC)vA<>s)2gxyO;+Z*udKscpFCt>a|6j5ekk
zgGChT-b&`x?T1TQbMqeLD%*URy3#Sy>zGWh-ZdBs9qqd^iH8Qmk*j&h=%}-YMnPK1
zCsJuz=9pIHsCsP7r9b_#3LP}eb_q>8y``^OrWyb(3SvsW=0k+<u{M_GCxgZwl+N~!
zo-BLRzv<)y>c?9$;#(G}1o4!Bg)~V3S3q0yfbb6Yw2L}9xegr#>eahmr%Rn{*Fu}m
zOYS?+l~U=Vjjj1vqAb3?1&BhEDHTn#6@VGvODI1Po}(}~U+7CUrBin>{{zGKx=H#n
z9Bg<nJi9cNs8OHLNmjq4x~~%G!B6@lKJJr&mbjVTz3dQub2hq!Dg3>GiO#*XzH;U5
zt9Ns=^Bs-yZAyjqRET9aA7(w*;xD(vO<>UYCp{Ik((&X)CSP`+{45<Cdq6bAFxrvX
z25z5p5r6)S3^(nPxvG|y9e_`iC-e;KXXJ~UDNwByc2xhKugCw+cb3T&4s?pD+1&Zv
zh@coe;@DEe)Ro62)s|I{zoKa(N@_i}5k)$j5E0RN3gAw88x9l~J6PC;^U&Wha#)S1
zY9YnQE=?vI=lGP_5tXPn@JM669zti!S9`&<H@<btB}8sH4g#*~knoz9`QbQKWgB9G
zPxga?nrh!U$(mL`A#ga@bN$V}M<+N;qT|Q7E&OR3SyHmx0DBoB@B?&`EeD{wI});z
z=q{Aa|2K)ePk8Z<KjM7JL4pS-+9nwBTl<rar%wcv;p=`3Y|P=R<#bSk)Z>p(=xUBw
zg!e+elefqK-Y(Cb8nuV$WYW02=1ty~!=;A<_jMyTqYS!FP{0cq3zATjBRjj(CoLms
zGEi%e@GtiFxu~VAIYZ}3g&|@8l!}VgV|}@PyBt8wZ+0=iA;3%R$5YH<^^1deWJ0y5
zB82WZ)w2IcC&G}VIFxW^18PxY9raKqIz(BkT#=PrRG<?g>Z1|F!*X0`Mez~EXiO+6
zDI1!b-xoyux)Go%#Y4oFBv9q1j-V*Oo8GF4{~@yS9J-5El>btWkkUlNW_4T~kynIO
zvB=VI@gRO<1Am_wUHNc@rPL<=vHd}cTcNkhU&p$oCHkY%f<is6dt0s81g#@MBbgBm
zvZ%_C5lZEnMMO`86_Zy$f>pZJ9l{=Z_+PRnRfWaf3$};Btz^_5K|NtWWCptZI@@uc
z4YfR7!pVwXCAv>Qjq<v(pz`y=5XIXA1*lsr_Lu=Jz%}Jz4u-sVy28SIfAs3E6k}Mv
zwx-6l`@uEM-Z>H+c(k0n!(Cjmi2PUC-(PI=kL_MLH%eZ6_VP2m?OirG&7Kk6Kf+On
zfR^OPW7ExxmNF%!FdhJ9_m)Xk@%s`z;gC9v6B>rafBA>>8=~0@(mARb*VC7XWqHe2
zJ7}7pq*yNYePajfE~LP=h)(akGhdH-Fa3zu;oxdpy?WYQx{OJI+NkQA({GlpUvF~-
zR638TvE_|1dWWr%fV?ZT9!vWc94iD^a}>Ir;|jK2^4FKILRRe7Bhr8lI6EaOl`VHj
zz+C4`Jv6hjBc^i@+d}Ef*ZIYKx4n_RN?cju%2vCuU<Q7|G4JyUu2{7f@9f+O-N04{
zra=Hrh4pk2VI4X5e_nzpK^8$pIk~~|7gX6N!9H-RryJ;x4;YVrL(eSqn6jl;c7b<C
zUwbC!jpSXdwf={r0|B`4ht?3_0&fmqR^skqiY8MMs2p&gjlDi3;`DJuiBsf{z$|x}
zl0n@%$Bh_w1746@f`Xp8v~Iz^bli;`mZ&*8K|x=fC!{rrG9Aab%cJlnhaZ>*R(W<A
z_8~@epeKNbV*2(9`~A9!dJQM$eAc8~e&v0Oe81KG7nxqx>oDJK1Y%My7GIdg6A}to
znW_b++OuopOiS2jgyS9}ZjPe`qx=%*$NQG2dk1T9*0}Zl=Uc7eT*E-6x-KvQPJ(X$
zA$>kSP5L9BQ6ZV=>Vi}Tq2{X|+{C!eWTdV7c4D5~+;AV_a{81_dHz4aX8mb|NK1Gn
zwc+wa#G?10iArDM#uxnUs&NLEb?)k@CySsd1~-Dqse>5MS<`JfJDYK68sD?Bo&v_+
zaQxuxaofbOu|RH$m561n#-6=&HRw?w`i((edWPax_9xG8MU|51S*}1u3_?wK<1xSn
z$h_>m<p#*>uM&Pt-}(XGC%p9f+p-QZfEKaqLXS=Z+U54Lz3%$8L$uUIyfZ&`tO<lp
zO|7*L@5~<J?kRe6*1JdhUA&bBI(he|<V{0MnuYjGqbVTO;0++yyjm?<fblehw_;24
z@-<Ir$)f#_t@xax_ZCSGEatuQ#FDqL%Li$!Gt$n*!gQM0E{Q~m-BHs(OV~`18&oJC
zAiDUUcESN;0R~8nD9Ze4cy$PO^)QF*znt0*1WZu3`kdO|yDGE#0bL?~;!GNlKs2TV
zI0r*-9tI`&5BHt&C_3tc#JZ@Hu5B{s9j3@udw>I2=53~EAPtWdZX*NtjV7kTYoo6!
zS9;E6tlkOuby|b#9`6x%$T{upVzg^x?LL9_7?>q?cobjXf?f;gf3Xb6`df$Aa023U
zLJJ>RA-m_;kWIj0_<gbNH-%P~ZJ<AhEnW8%l)b3KQs-Ggb4w7ymopMiw;XdQTX39p
zG*hiU=sjz3Zvl&uM$crr+%x=kpKz`*+x`2i_hDfPUx186dsJgEV)`LZIfVB1?b3us
zGi(mGoAF3`X_7l`etE1UqcbabG%FVlM@7LE;{Al)8ET3v6ShE|AUZ?~Oc6X=7qf$P
zlHyS>?|zH)-whx!BBRDfgafyhIw6;y4gpTq%I|R-l`o&_qWrveW5);4)?Z_U`X_=Q
zs^qHjV?z^S1iJz{zmvD6u+TTR{)%u!sp-ctFs@?$`!(<G@-YPdQYr&(BV4<8z08^_
zE)rSp7s7a0Ct(<{Y(%p7?eYGMS^Sp8xUm8F(}%qKjIHE)5&4BxYt}6U*H$!6H%h|o
zF-@!1EjJZrMrH!lX8bsS@5XZ|&rPBFOU4QO<L&u|M1YrzUptwMhCKSm;-jvu-!t*h
z6`}yi`XnmcIqJ<!t=ERNiBh3kKrV0U&lkjS7w21r)<$1bd@vnid|nB~zr5W4*a&#g
z4Z0)-hvRa<jmXrWZ@ceOubet>zkit@kI*nWN#rG{Mqg?P7|yqPGVFkyu{bR#CAc=G
zfguL1w0xZTq@}!v*?U(a=C4Dczr?|}tcW2a7GSbkHu~^_<RWswVlvZl$*1b4A7Fz)
zOFcVPb)nXVwYak1@k0mz7*9dt+n$7ZhZjXQ7cy;2ru;6LL}4&-uTk=<N66e?tFe=q
zaXI^=;~%Ajdsk(9YZF0XNpS(dxL1N(3J1(V>8SbCU7d^T&>l<tR2f{oTK(rMnWsB%
zrhMFsg;3=0Vu$26zuMq~+k6b7m?BIcLZrgPKA|C{A+KoM6F3_KCQbx5k_Gj;4J85T
zQ_JbW`{$-j1O0x_tRx3ryBY^TQy1|V-%jIP>^cCX<u>4?N~(|TPVAmhtY}@gCS!MN
zVS<QBc-i7uNmm$5%6QFa9!45%-vjkycl&@5jqcH}mTKY<&0jUVrj(T&5i<xM?><J3
z_#eIMt(>b_%nDcmzx)cfT8+RLY@~IQg@9W}z1e+b$Ut(N4fi4h><H_f4lDyV)xBZ@
zxcK#mTE|<eRS$;D**vdhq4E?N<5RQ#&i9@k;nl6{;XnsU#nq;~rMXG{3}Mc+S#l@-
zOpL|4v$@ru4Trt(M`(ZUhZ&xvDBtZ%3f=#{Wos%O($d}$)6>GaHRj3K&%*8~%d4>$
zjgqGCfe|t-LP$Dy=kgC59fhCGOdayu9c+m0lh%@I3Z7c*QTkT<8Gw2Fog?Oolx0<$
zO6B2;-zd42gOUN+Oofp<Y_N7U|Iok7WUStl1whiQjDG>H`@aA;GD1M$(Xb<O%5>Dp
zd|$NXE^(R{{nlg^9&}|8njD5q8r^5d0|M8iB>`ZkDe)BSf-W5Cj__FPHwe7!3P@?l
z5<$h9AI<Vpw<*Kf#mgAZk7uv&I9;W$wUomY$ke_Vr<2MaDj}}UQl+-L=?2k0LsAU*
zr>%`h1^$S;piZUw&!3k%TwGGNQf)f-CgzI0?f9mL<wm1o`34y;4+%WjphR%-dkX}t
zy0%Pw9+^>ex51Z?HUT?8JU4xiEF7eNm@lnuz}F-p`FXBD^3mZ$>(4f0Gh=t^U+~}4
zn-_*E5EXj54|^8=90}ipps=^~rj*6jS=&XAmG7ct=*gMqU)F)>kx^;v8uG}&)iJ*H
z1}`e;A*}@*bjZ~~Btzc-3GMGMX3(FHy#v{~^^PRaC|Y3yY5xUP1Rg+$)beS@2iI%|
z59FvMb*Ex)6kz9!WuF^*(ohULEMBz;Dg1LorTTU}zN4ld?Y%9uverItJSxB8py5cY
zsd+=7hBs-GG{yCrmi8r#j0jy+pILc9<d2@-20ZjuL`y7vUv4<>%>0im?ydbpD4v|<
zH5MXvugJ<>FZWbl^dFL(|DtEQa-V&Tuwd1i()X)~>{&spQ2-XDA>PQVgnS%08^S7z
zzsZ&LZ&aqr<g0PEpe5Dx{PFvOvAmK-$tg#LX;k`9MTpIlo6$o$49ZY+OiXRO*F$jQ
z*U&Tp+-H&D%{W7NC}pVm_73Hl`~}ZWOLp9q^RkDyeN%6}Q1D*YFJ7%adhNr2fwib<
zy5cXChE=k|<9|HmrDbBIzUGu;3|ez`8*6L|G;`AVbqATS`92(+PsTeUVUY=|ea81`
z#t(nkTvXs0pcMXuib%O&r~-=qInQ^bB#fBhz<cH86V(9(fNvMhy%d>OInmMH0hQqo
z33%46vg<ME1+@8>6kG_#TokI4wFV<Dj|iOtnj3%TXLSChFj)U#eRZHoM_-ws&8HP0
z{N~Mn?aOCj@yw_2*=2?G!+PffR>@lE9EqRECB5+gcv|Lop{1yTmEfoN>g$eLCikOy
z2=&{}>GnXcE-Z`PpFi=mc6)80a^c6d*25J2pvbDoSd7jiQGNE31N}DuMx1PNJ;CF`
z703J$?A-U7lklRvKyRyjCp7-4{JJBb7<<Y}VByh7FYqwaNJ?Dc08gb+GBm(-aJYY4
z!e4owEv)_#-{H21LW7^}&(f8T$cQ=*aDn%(wtkcZ;C(_SJ+I$)@383TkMpytXIn%$
zGS>E##*sa*W|!d|ZZwoF3Fe}DErG4XUp$imSo-~A*%7qSi}AALZZAZtV@G|+-5(n7
zU6vJnK5wGrB&32$J%N%TRX>FTY~T%}jbBo~5->i`e{SPS5rs-a!L{^ohAvUiZH9IF
z!pN?2<#l`SQtnAo6X1W}c6^Zz;FN5r6s8l)Q<KIZ9Ltt&{<gWaUifFS!uIP)CW`J*
z^O~5mu4bj;0hMoKnFg=+5jGR(J9p{7{0i)w#)}VVcndwq_IMhP<Q3Y%9@_5PA`5C;
z>tWEawwYnth}93|eaoWL`C5v-nR+BRfctI@>OJC)AZ31)QGKuQ8wIFUx18rUkHqDv
zsB@<6%1Z9z$?8%)7IUw}rpnLjFGeITbL`v|`!TIxpRcx@S+X7C5~HJ)pMo(2&`m@u
zLF?rn0?patz$j%v-xp<N%d21!agIFZ5K6<OC>^fo21<swBTN#Ol%mAxsQF1pP+j5q
z9q23#F}~(4c0J=pJ-=$uo~iSXP+3wuSjJmQT@|_3>Kqy>892MGoFr8I>6E^fJDT%Y
z0@AB}DG26uXXFd50l90Ihh<BUoVcQt6KJf=%xFD5J#lc_XB;l0FflNI_q3=!IV9v6
zp7@gMG7XW*+1c6Kj_McJYnL5YHC&avzNxdLS=6vZ^?-EQ1+G9kV2)_iNcwypRGCc4
ze6aOfh&NoFM1?B9!W$d~Sv$m*!D@EstcOC|eLl_c7IccWA!)fuP&#P#>Y8ra!Yg@`
zWwN(9Myi%}3c;%Sax{x5o><&JQk&1$Yii8>Op4vxTK7iP;gGjdmtXT8&raP?lHWTa
z`tkzF0)42?*h)^?`dh+jgf?&O2WeMU{#Nn1O1Ao%Ak3MM$|Br!fd(HmHA&KD&{`T(
zG(D_TVP&2`VHJizPEe{?x38S8T`!prGrX;OdM@gzou>s<1RAR0y~}p7VKF$Z%`E19
z$)d$ZSw-_brW`$yxV&(Ej;E^5i1i+=eT#t$Oz2JQMeDDDT}*TPQ;SvH%BMLUwQM}g
z1AtzyBje)SzbN<(zf~Db_cmhg-59G@@9pRyp`bv7;E<91l#d84`%?w(-@pIUVxcvb
zR^XAexKUf^9dE$>)Y7Whekc74I3v*U;xcRBsKr&o;pH&^0y|}<Qw2zi-x9f6B|G(|
ztnyhz^`9@+;~Qg}d-fvLI9Xx8O;2^LF1&Rt0*lH;d7j?Nxf2`O6`BdH;WnP}r$UJB
ziB%Q&xrWXsk!9Z^vEHk4v;lA3eEq52GBr=k@0MiJ(P886-ExHO%6<9223gN;&cKiu
z8ZNds+B3t%+p?v0`asUuNeKh?8_U@V3k>W<1xg+mEyj&TCfx3+-+^&WCGu|jp!uz-
zW%WxzrG=ho%U4W~1zvG-2#l);PN@vVE}M0oeJ-oLLZ^y|z3rj9+a2sox9=&jGcTr$
zob>%%SkNKMD85cFprj2)Sh(v9?EW=2ORPKT@Bo32I+UcWFd6sZghMvIz@TaEY>fmL
zH{vX2@$MI4+w>!3y@~N&2Ij)l8EO@VQmUIdiAV`ZeDz{qZwQZu*M%jnB>a>skfyy?
zw`D$??1sMh8FM-133s&7>{Q0AL35!?J^G@?bsoj~*Ibe>o6{u(+Q^9K^+lN4$X=Fh
z;}sWu`ujV;ysY5Dld#!UV)Ucr5?G)gzPPnT))7d!+KN=p1CjgGxZ^EoV&ETDTW5?L
zK|dZ*s|aOusUbC_?6jd<ILcPeTUWij9FbW1-7g<KrowocMv+_GwJ(ct9bX?g|CMh%
zNbNlcgx`cYtn}Y6WO--%l1Uj>8aZmOKeJ%n4qbqb?;jvrl93ATOh=}GTV<TvaLSLR
z_OqLL4R+~W|C-W%4_NEOE_CILk$^grVy?ZmDnI8yfM=U(Sb*K$+pX6b+hRvhei9Tx
z=JnO4?2B0$W+Earf_?ttY<B5PPoDiHf(a-t_kb6~xyxijo_#v~BXLcY#H|#+Wn7=N
zzE8EO)%UFD!FC(4WP^azqTANqxGV!~!vlf-2wd6dB!RxNGidWP3m-_^6&I+$eq9V~
z<NF7e+XsZ<dB!(F_Lk#XzK<<OZ;o&RuTS5MWv8PxR+C--*0iPF>}k2`q%UM#DBQ{T
zw5G~+99`$g9-Wm;5B$3WS`EgSfg=V1Ym9ToaF=!Cf%#b#Iii0{^RuG4tlO;kTX&0%
z%xM+kWLxv$Z|ZSe0=N=LS7Gj*#4Cq;yqyHH!AYV(5_{s`wYz;$G<|gbB(Vd$lf7)m
zE-}8WqARJ3c&a#Mv7UQ5=Y;#P#GTmeGw3M52ms(ZSCcm2fgeL8_{Pk?Ys+9@0k`(x
z^ogq<wUDBf$&NYq`DL}dVlq+S_7#NU7U`}x`^Vv>mm*963>)17=2=fRKxutVw?cm+
z{-dK)41^Ff1BnwQRr|X<*5s|Gok62^lVgY}5_mn?@qb1Em-x^sY>Lrn-a38?ztH4*
zP?e`96ub<8RIWhvDKLN}pcpNwXUWd+t6Dgy`o3GS9gOJAh>`UO`d!ahUFEUKjr6wD
zOD~yf1>W~YZ)FU1WEsI5X{;$AOfhSw($)m_7uMna<BBUMjXWJoh(2a>oAJ0yXQQ-t
zc2FIhgIs~JxN^(CJb7}&ThrdvO32k)_DV3Ge7RK@^#0cij_MLv*u6v#sA=pKKN)*M
zveHn|J$oyHch}Xp^y}MRKZhP*OAx*X3O7(1i~_E`9)ILuYZ=H!0+w!J3GddP{=CP)
z6|bUVVqb#yWw<%Cj?VNIt2HkeVS5UV1%VYu_yXNL9XQWRi1xO(B9;MV|J1?}{axD)
zFjK0#^|%ma=vJ$nDA#(sOej3FTZ5hgodf{w{w%UIa*4L}LT9<y@gwbeW1+9WP&#x#
zsGQAx`wa9VFJmwF^t&k$Z_amr0WoD1kj&wJOerf+1%W1g*_H^1z9FGF?s-kTioN8`
zf^eS}lq((JKum@bSj0Auddr|pFE$&18@C=8#cW8O<_Zaua_;GE*{JcEuuJVbQ+KxJ
zk>Tz|3H$qH6+SFTrF_-=P1`{0C}@iu@b)d~A2s%iLIEK2jvu0YY&k$*X59$<-8}(L
z;o~`r!%V_i9y?#b2x3?UtyW_=TBRk4Tl)qHpy^Zp;Aw8miU4gH3>Vy=_16<XQM{0>
zO4WiG7Y6BzEr4@6AIw*{2pTeU$V<oGUXb?z|NO7N(|`V4lww{GVO50l@7e@r!iDp6
zrgXfTkGN!rD)MDMR>nkbU55h!@jnjZ5Dh*m{2ED#<M<t$Nn3TaX$G75U2(*rpYdn1
znEFk7+Rqng$X-R|>p=CUJNcq+2RW@6Ap>I>_%bllCNr2GuA(m3k=QDKc7t4*Jr>bJ
zc9-h$|EBt^Cd;A$B%UoHE!F$Co(-k3xmD#PEK`7%jnq?AmGejhJkAz%<?AdWlGH+*
ziVMxNjh7?Iyb_ia=eV&H(&PA+71+?^J1MES*m;Qg7)dIjr;n{a<vu_|H_$^}hiJ}f
zW72*&tf#8<ZVw+^-Td{c%B1y%0YyN>)GfP|8Wn{r!r{|&Uhi0A)pBCOa34-yibZL?
zRjEhnpat2h;P#)NK*K@bhj(53b)@v2c$6Z!c4w&kzWf<(%xJVcXVl%^n*J4AQc<BQ
zJj>AA2`&JTS_C}O^NmpgKtn+pR8Nl<B)D`DQ(KviQppkXT|lEDujOs6F?Cn>3+l6n
zaKP9lqGo5kcmD;;^4Dpr$xuAPCsND5^leeivWASpy;Y?s$1>i*qrUp(iSwxQHCJ{7
z*2D3pG&88+I-s2<W(CAtLfSz$)Ju83)0;)l<_=`QrJ|XW#qVvAxcXc3;(qW2bAPU0
z53PslM`2lQFy&P0XUb%Mw%&UMrXT>;jvlMnNZ1(WT2Vrt@c{RLwcsq;MFn~RBTMOG
z0G>Mkcy_{EA)8VoH1*Oi50Qn8i=Jz@_e%U~?*htMHrosbb=3I8HwgQ-Y0vs+(6#vh
zfZZ+Z*cR`Er)MnvRlM~AT(3_$M*QlwjDV*1soSLh!+aL=4SJrudw3wu`O+#cM!wQs
zwH6&kkxjB!M_?gnj8wN~Ay-RT3*f|s*E;5Z%ryfP)>FcoxcRS>GEVW4inc}=Kq3R)
z!UpI_<6`F?PF(E5R1Q9gK5A2o*iWW!Cj|}Zp-nSDiJF_Xe7bym)Nc^bAsj?Dw=O+j
zK=AI22FsSbaRVi#nudUGtSH$g6W;O9d9vYH9U4&RvJSZVF|nWHR}t4^qJ!*x|IS%!
zVKjiiWXuML5HNRmn@&nb{Rq?!AnspI^~pP;`0yWss^0^frtzU0i`zejwmmtIdE+4T
zcFx(XHlctOiTw~!bOk10D@c_kH>-_A^F~A=1N$`@^{`~dGm9f`(HWQLQ7A;*MJ9~H
z+s7<w=cWOUB`*vXq`~jE<!-#HJ$V1nv^L?d#6f!}r$LXTg%+1}7;hqeeM^J^>k0yx
zm`M{p(U-!<iNsS;jTQoc4Qag24Y(ml7~e#HOfq|CY~b8>xuD^ilvnR82#i^`onzlm
zD-UBi48j#VntREnBfoS585mj9foYFm;sPHV%~_z+D?`zq$PEl#KY3ez9h58+;o_@2
zat#HAGutZ)@Y&OT$JSaIi$vAtZB73;6Dl}b4;;T%dK28V7cOYXwEP-Q-rbpCO6VpA
zvt~JxwPc;!jj8Q=u^mkkNU~T0{rx`&(Sh#)419a)wvwrGNz9rv-5LY7h^-s@bE&2F
zIfhE|uZT54LtXjP2}#_MTV**Z&Iu9@LMEF()B}buz?s+M<qY)>YeZfWJDm^`3H*V5
zzO1IHJxP4ax0515_xPn+c<NU|OzKlCIg<j+3(4l`w}n#g;n!-Tf}e`Kc>2QQaJsW>
zh8Tg8FSMKd3*5>9H;+Wo$utKO7rt%pG)L49kF&y=W;B}^%EKL_z#XY)&JmE<dB08_
z-U+82bnpb8uqj{Z2a6$B4L)Mj_G0_E9ny{s8&zpWkM<%t;fteHP6C+Pqpj)GXQ_8!
zreZ&~$3^k>0{x5~)0s~jJNe9QOOcoo;_v!8b{K&;msL2uf2`kZRZOsZIw%gp*Rm~G
z8XU>gWR1(6@(Kum(M&)maGy>Z+ZYrgMmTyF;+8|=2;4>T^p=g0b4}Uy=AwWI*PQR1
zFuy5lGc9=><I-$?Mxf#()GBdhSGr|}d+~Hz{raiO6Awq_V<<YLcdNC&JGhhF4xV5V
zVU6iX)RTLUOIbx|G#sH}QAj<{Dt@Hspv}o>hffZV=87n~*A4Q*R*0K?u!~ghGjUxU
zfm(hYm$T2;L!|ER9!5<{0Rz-XzQP3U)#|ABqn5Eu_LB8H&J#A=qs?mzPQ7BEy_4=7
zThH@|GN(ZqvP$DzPi&RO;<-N*cJQ244PINHO6Zmzq+jl@s`@hS7pOL{c(2X`CiirA
zNWvQXQHcLk)<gYL9ors32z@tu&$3EwB%j_5@-Pd8-p$nIBWy&vbkv!4S{$S5kNVqS
zjXbs4^Z7K|b;MqAUab2EZBw;AtM$0RxeOlEFf$qEiMm?su$15PII^gDIy(hSaG2WL
z4p;%=N5UDwyT|^qmrvxEJWQgwV)q-L_?^gxJ3yi0N&YQ5Lslo+1q;h_DyAy<Do)%k
zU*VYMYPU0b!|C26R^kpgJV78n0}S90-j*)5e|6#m&$24|h3Xxb871e=8j-w|3BsK|
zy<K!STtz!`f`lnN2ANLS>88-=ILFiCOH7Hav;KKky|9oeUI~h;Q@yWo<=V~PGw~Qi
zIw8zuU51m0iwI&pfw#QK8^cEOyY@^t>*B(;&!`iVX6)Y$dx*5;(MbrV4L&bD<$WbU
zS6q?Xk)_{ehZgMONC!{Y2(U=~GlaWg-HM`DcL+&!f$F#D*^;O(>yn!xujYnaxrh60
zL|inSz+E1WZFbWBwoR<9MW=(HqK9%+&(pAqm#;`LbjgO;>-wdHH<QeNT5RKVk&^<i
z==^l%bT&(Vwq!kjW@5QiFjkxINOjn=`t)T~JtpD)`oj4DqHxR%CWB@<L5X|^z16+Z
zgjda5ZT&FsY1#T8w!>yNU%3LmZv|)4<D$GdKQHAxsvV|L*`+<tdE96Y#q!w~IIU0C
zw|z&mKsFF{6uAdzGlOrBHfUefz$9<X;lZ?msfHwp(b}6Q5$P<)ue(Jxui8jC3a^ag
z)nE>DO-o>tRa3l;%NDEIUei%~O4s#e2Kt3&yJg0GNutRtBPO%x-4paF%)iI!$*>-f
z9o<dN>cX1n-l@@&RgeW0N?P>b4!hEa4?01O9TO2qNfJV7VInrSx4Odh!`IvYE25%R
zz*OTMP%}x?TIZ*Jc*$$uFROh}uYmADQh2uSTAZb9Yji;`(douc&WI(3!ftnBlXdfS
z+f@cN?|Ujca`D)=x}tA2;Yz4^j(#)h1_FmgW6|AuoJSQg<xu^UuCy3lK;#PVsrz;L
zw1DCDM}>MWZyRD%Qq7TjA_t)ixJNiB5g7VOFI_jVe>mr|#qkhI%726hfgMSqeqA;`
zUcHi_;h#U4h21}{6cxU~Y{{yOkH}gOFDb7()v2KBCq_=X0o^<X<A#gTHy>F%mGj6f
ze}rOalpp$eLQe|R&G5JrKknQFCe)&PDT2ZM0~+><OL@!ZX?;=HU&h1)gi*7FMhY+5
z+LX?nk!6B@kkFT4nnUHtGY#vaa;m})T`=))PO+?wDkh3>ucG?0HGKz88t5kQ=&R1x
z*tf}BA!d}kyuj$Cf}e*Tk<Hg<4=XzuR;hg%;>6`JAbpGDT`Joi$oqC<1TV3iRS@?Y
z=2t%NS=Enwy|@LR{mF~D#$??uz5X*3fL>@GuN;bFykMvqnHok*m3cOOV8%FKUIa~8
zR!uVhXkCV#2Q7DE`gFP4!^If-#7~~VDC}GzGfEb{;<eOc<ph)yo-10?4f_qKDj*Nm
zTXRerQtMT`TRiriNKI(H-v(Q2nV7}hJ+6FVI&-<NbjMq{jWx#q{njaZsj_IX8TvZL
z$jlgr!?Qj!$_|ElEBhOAzhjp?QTD2&#+o;m%ExUT)LrWqUpWKc>aTRAT;XR^(wQ>a
z7&{#jcOxRFoy~=<jMf^yX6V#_Nc7IA8`4M0ds+h-POhW&(71!#(>mpSPi2I(%U&1T
zIaPe|>N)>%?%3d!DLm?k6ZUIVXMMN~zn?Hk;fsm$37dq8FPY<*NorL0L)rs^v5A3`
zU($y$tWOeZxBfi76U<Wk{|jTXMDF`;8lC5V!VisE;M~4{Yl5&!Pxeuk50)~8y(fC@
zh`B`@3n>rfF~KeQs}!Imz4e^BezdRy^{iKbe?l;;_L^p!5$@V%*_<i++8kE6xbi(s
z!T9sZK1Gcwo8uv|cpDMFR4$F^+a=vhY1L>yeFH@kf^|iN;esW&P;FO=*K@llA+!I-
z(^rQ@**$-wAV{MKNUM}|cPawXCEeZKdD9@N(%l_Pcc*lBce8ZId*k!{y?=0BpbNWu
zcFxQvX3pUS25+X$5`b}IixaT;Ol0o={Vw{U$RIc1HRQ{6VC4!wl;-26J%L*d%{sWn
zHuQ63RMov2a*3P02Y((gC#AslOet^Qn2PeeDKxjwGa8%Kgr!Z+3#HCsphdsJfZifZ
z3pA3vj63hb)F+Z&y}tD&Jq*kJ%$(Zy2Pvsh{rL6CM_p|MRbX)e7r`mFknq15ugu?Y
zkm$bte#+0EYv**7i!<7fOan(Yb}n2|;5#PK1l5JMS_a`J_AhoWCZ|7QO}Cs)3knGh
zRuLxpG$yf3;1N1XU5Gc%G`#w}Zjd{<TkD(Pw9+L#H93=$KBU)^(y=pBaXZ_kpy1kC
zK5c2J$^-+1A9-Jd_pdJ0_4j?RBQJ@+?Hl_3qP4Afbx1>4^^@nv;yqqX!80`Dp}!3&
z1{)f~x@?JziGz8QI7*ym!NV?6TPg+}{pCWiW*hkfFhi1q?Mu^jkL#=q(}qjuNsh?8
zxr=fZiNSsFp5c#|Dd}xJ*wX|RxG1i?!kH(gfuHaIS&P&s$()uWg3r{{nT?IRQ5FS7
zMU9{NK6YM+ylqqjTNaF2B-gQ7&F;^9d&3Ug--^UrkZ*4a8#2YyiJkUcSnF{=eJn16
zz<MaWipDbF;A)swrLX3~3&XDHb++6px_DxX4+}tEUwl2?qFC67%7nZBw3#xiU(1kF
zwLN4EjL0jh`ejxmc2)d>V23G+UMKS(-EshrRjP^Pz+&=~)7NY{l?T@*IhF)FtFy4J
z1bo#s^&zp$NtT;+HbvY#s>05pt{2huC%O-JDi(X&-&pvfk$#v~sUuI<&G9`wo!Nbp
zzhy6`?E2(qOz>x=>Q8bq;n0wz6})e@Ii5BX&~wpl&L*#^8Ln8^bgZMb5LZx8K<8dj
zJic49OK<b-K|xXRrS^C$z;K~}Wv+I{d{=!hwa`Ww^=|WpSlC?ah@$-=w~u#?<i}bU
zwUD0kW27_&`jB!T76M+M;_zY?>G<f~Rl<>#p2c@J*MTrH39;(XvY}LN`ep5q;pAC;
zK=<hq>tLxiO7M>TPimy6rDGnv)sxI8w`H;A^ho3RY^|nsGvg)CzIHWK+oRKHyKPL2
zu?+~fZ<~OE0u~z^n<U@w^+Ts4t1pTM4+#B=-N&zwyQgwYCQDgyB|D45*RJc||GDgv
zPf$bgQqAn<?F__*<sQ+om%g75WZUuV2dMk#2i7c1x4)^oY;GEWC+OJExlS}-OKO5~
z(cjkxMg?q?yLYk0_|1IU_%3NLb*F<=xe8bvJF&_w2Bm4PuUj?@DQlz(S-FxL+ehX&
zWE6i%$+heJd5czI{zw2H9@pm~-4YVw1|lcbN@!a=CW0t;wKulMQ4LpfWi4<4^R4pG
z^5XsUTXt#wMt9#{QQogAI$UTakBSk|u15=1#zwPsAtAFv(@?%H@Rwg0$bNRKfiXW=
z%%kc3$b7+4hy(QPb)}BwZ`1PJLn9DT%q$$-QHzBdh+wk1Eb^u~Zi98pr4GB>dZWEo
zuu<P6aBSb(eY5|#fT{l=;19Td|J>0#{|X0)2tX|HPFk=Ha<HQNbS05<iX4X2Xwmw`
zH(x;dC?3UjSif|;;i3{3ueWp-GjV|t#03~%S>Qzfoq-WHV9auSc*m?<wSx)s40kTl
zcCTxQ{DiNRB+oMjFq3|&4vacLD_l0+!-cIjZ_r~f@kZ0TPuu{oBGfN8Z)}7^jYseJ
zHZ@1Nm=YzE7A~5YmRcDdjsRM42ig-eBus{E{VqLo<61IY{Avv%@m`)XED9WSRi{09
zvpeuVPcGb~^_Xx^koJWe(wPxW`R_CQm->*#MZ<4=Nz9f$Izn9^d8-hZm#=?XJ_xH~
z=?NlRE}<<0UWWSQJ-=dYF(Oum6vOOXmxtp}6uw;&1u9Ng8bLM9R4ezr69egAiBznZ
zbHdH4v5(vmFeI8V=E-Nq@Z@1u#24Esl{y?DZElCtF$|%9(ufvX9qKRgBXlX{NY>lW
z5aqM}gLsyMxxl)%*Gc=Yo`3%jq_TXfm>mF=>{7H3t+cfFH+ox5_%TN=Pmy2LT~}v~
z2jLhq@7O#egHBlScwUPS(1%ju<<_mqqs%kOj@Kv_LVLZ3q&;*LS$2z<m52@4zko{y
z+PRA~g}}tUg(p>3-FD0(l~Y_0(xe--Ae7l%=Fh_zGRuk^=`i8FXWZ>=*-KH>0}jWA
zZ%#o6)Xe@og(*VuAy<R#BXng?dfz)@tQ-p4Ndx!`7A~y~8ueHYUojgy%8$qO&TqD*
z6uBIt+~GVWrZZ1v$W0zjNqe}wL5OWZ0%hDp%=KFbJ-^vB6k%9}XoVT>naqrzq}*m0
zE?QP>9;GPFQUS4QvfIBxP>#g`P)RsX;>wcMt?Z78P%1r)#L~|723P|<Q)2nqwA9{1
z(MjA_&%TK%8Jw(p5ZgXws9aZ-H7ETy`!7!-qzo=;kZ`3*ms8QhA@V^|6YfR(l4)cE
zV6?u{-711J>bjB%CaaUF3O&Z1!U7A(sB!%51OberyBoid3L@y#Fw&F7@FEi<tSsd&
zo19h&;mc_&n+1e>#DPSR&BGOeT<~K1#x~p{MMQT(cY((p*W+pcZ$5>iJQlL#^0rF2
zf5b>}+_aUXYN%S#%wos&E^)zr*eFhtJcj3nvn!qzvgVO(pFDzIB`m4B0I#dmsF>Rf
za-m|aMIn_h03O<r6r4lJBQlCkQ_2-;+w3%|<7S7|uEV}LL&rWv+FyZO8}`+4s!|)D
zqD;(0bjn^0H&pi#>|3SCg$HhKR-Q$I9Jq8y-rBSXe|mZCZjmMZbCAhu{l_5jL|s%E
zk++6kYLDoNTVJBp)I!9vbGA1kPAo=apYc!8Q`6RL7%zON&c;S=K25*TMbe{w>a5YI
z=U9Z5;h>u@<Bw>@o8L+%IGn9(?;$u>gU+djwA_&{w%lg?1j0h1QisU{hG>wKj+E5Y
zubGo4sZKE~D?4(6cFfBj449dt0ueun*OqfED?X~ByGMblKz3tk6fH45M1Hxu`~e7I
zjSB^#7AuGaS6j!n84^RQF2`eW%50W6<j@|#3Dg{#mZ%EGT!=s?J5h2IE#{V_h+d{Y
z{hyWG9}Lq=5xTGgd}s(E$0oRck7LQ%U5jEEa6^mS&XhI-!N~Q?+)P&ZTovTzCKDIa
zb7VEeDi-?8Y%r93NuQ4a2?B)6J!h4`6JN#~d-Orz3b6yKghhr$iC>g%%(F8jO0ZV9
zK;y68vz%&z$wcMCk@%9v-Dj+EqraNAPfWY#3cdXDAu*q3P_MI-qVBA1)*<}FRoe*q
z7n!*D14{-h?unZ6rLJ1IMN&v}#`T|SP%}7RPrR{{;?IZ~YVYm_!i|>i<Jpq$rfMTd
z5UQ>3N9Yz_&VfOzr`Uzdd4y_yput(IFA^Eo)v&Eb_GQXtB4p=4gQN1+La}kww<1tU
zhY6R?#43+!z4Ova`BIKsPFHDV4f9LCCpxU*qx(${HNis|NvS;Z(RGW+$n0$E73J~3
z8-rEWaRa8NO<`r=HzIn^SNo4kZyqHcPesiM4(Tl1u)xCkiJlRet4=_R8e3fcMMQv)
zV%L$?bS6E-{P_L9w!#wl{C`xN<qvOS?2D1ht=uA_4(3YgX4>1|U~EuC9Az2=YED*9
z8dvh%oI0^sP5S$%4sOkCxe(0rphv7o)!x)}O^pJ+!G8!&<<aGTVT0!&RN!Cno)NJ}
zxQy(*hajZzeQA6+q-#z(<4Al`Zw2E^cX<t1#azNSf-tafarYsO^>S4u2Cvb;fQrKA
zt`ZK$QjS^iI&}6NAI1Wn-&?FWKl@^Hmdh1m$&IRqZD_MIN4Z|A*%d@L>woe=<TOeG
z_zh)hk{8{5tM=gaU&w+jdTMzy!{xeyIf{T#10x~Ko9)u8pR(XBtMx*_c;oQA?ut6}
z7HO@<<ICmG9}MvVw07utf`y6PXRT{0G<VbN`h?>CDm|?V2o}k$<M8L?fAjn@yTnZ3
z(;o6tLa&z>XSh`AIMhfun(2gR-)}2j0j}-aH?!@MM>{p!Hm=)@s*C1}FTMYTsN~Q8
zLR2^nQm}EIt}(;uvN~BL($tN#=ZqguQDetpY8Qs62?KUT|Lv6?x0=y^ygu=8v{05X
z6r*vWEW@cn28%?h8acyW2`)!n?KE}szJIbWrgRfm>hH+Ztq9^}3R9pbRln2fBgrF#
z4ID75*0^QVKZivjV0=~O-#Z5eBn-xa&9^39lp|~W=LH-{Syy6hbF=Q1B1;Lc@fPz3
zMLkg4d(MpDng|)*D}wQ-J|SYX|IK>(q!OPCGxc~0#}kS>=D)D<{_RyVX^~PPwJQPp
z+vTY?p3@$2fIzb>Hm-~NJKrxa7pF%d3-jcIp!Xx*i9`fq0)eH4n8v6>2b^fCSl{X=
zt<s2&N_itXd3(tI^TGCg{<lIhzY@J?_1kAD2%-x-E8c3jtHS(BOg6P4RzfGv;bRS?
zlx~(Z+%qX>{zFOG*U>eIY}F#;{<3`!YaF;Ja@qfG*w~Db`9*pxKT|)4EFDtNI!@s=
za_q=aH}NoayiinOTv^Rgk1$EihNHEqT02LEha^|--BKbcvnR~H&-dGKkwQ*E^Sh2P
z&!+!w9xK;@#sjG3gb16t^(xaf)BcqV|G2MTKni|pVwpMXGl@GKm*iAP{eL9~vH*@k
znBp)rBQ2%I_J`4(gsmP}LI0FfAGzZy1#HQ}2!ZU4Z?eLkw5b1WI#kIs!Xz-g+k{b5
z)4miiQhI1tizzQt>i$emSzxCXE#|K!!{PZQhW?oR9bR$xND$ifs~R|g!bx2{J^5{$
z0YhekY#}hdefs9^`mnYkxj^5aQ`azgt?esFMa%t#g&s7x#-Dk5m{?v#Q?vt)f<||H
zPe41@P>CBQL?N1Tlh?Ykq_bCuAd0a-Ec?IS-4gWwm6~Uq1hXj=e?e5(_&xt-Z);Z#
zT2|uF<zM#=sxG<Kt9)8ME>p<r?2Lf1RE@7;i5Jl13)uM{<Q9}Cb=<MPak8f1jHlcw
zq_Vj>C85cH)wm09?ci=*D+6h_B~kOc4MSZfT$#dR;Lf4>5fW*%*uwFbH&V2{nq-5_
zKsG_muQ-yI&hyCc!c@pwAI>yvv*J^7qlR}tUGp$;q+6*{Q_ovTaRn8!Tt@=3Lzz!q
zDzD;D0P-Xc^%D!MZ=J1Z9JSmkJTXm$Bve&McFoO?-dLaYIGouAmj7w8L4}fUpND|I
zkyzWDwJ(kOw}-CpP>YYus;2zC&Rxj*?1k@&Z3*gk{2rH|z}}R0b0&@IFh!$|U$R%s
zHgo;8ZZ~RGsbgy|Jrnr@`GI6{0~eF3L!kb=Iogc~h}<f!S7s>ztl@4rcQ?q0JE@Bj
z70O<0kWx+ixCKQR5Dpkh9fs<iA+$s~a-5V)R(qm^r_i0dY$(<*^S%y~JhVdhP(#41
zxrM0{MW8`#n!C0g3VSoL+1=C(VbA6iGEc1evm-RmsEA89FPyD7YN)IPKlnbx2&kBA
zX_PYon@QawW20ViDy%46&7R{V$!N|KrDHdTKVylH2^{}J`2yLBn6r0^jwM>1vx2}F
zm^KT$852WEd+uKd=tEagao1S&9?Butx<l*lQC+R4zFN`m?4T@FYcmNtl6}L`nTZ^p
zTW;3gPauvznfvOB@?IXueISTf>`ek9azD^xR!s%`8q#}h&W~aQF*LS!1a;L&2U`kJ
z#{Qt$s(?5RvpAMREOMUnPgD+r6<Q>2=oEB@+@c6EQ`BvV?;h}B-oSEC%uUJU8*PU0
zTN}?>!l*x}zL_&1NYT*%@$VNc*>AGlh#;&Gb-g#7YN`v069Ty7u5Z?>$eY1s3>qYU
z_yUSwpf)!;tHO@SbT`1)cPd|IQ(E=8mm;A#QVs27J`|EilMlm<igN7tUFo>&RhXFQ
z5D{H_BXvaWq%46J&S2?m<c{Z6tEzb1H<X>$=2!SSB1EAV6?6j3rOfDF*s!Jm;lIWz
zb>wFn0L~w={CUv<)!DukdNw*rOPtc=rkfn~+mDDp4{-{D*2D27x`p58#A^T4A;aDp
ztIt2vDZf|;^3P~(@t~b~e}$#t5$}THHeMVk6L}ml2ZR9=84m~Z)k#i|b_l}0DO<ec
z<A>ojHs`PFvNX<!B6s;tJl1MF+q{=kwqxl5t&i&mlqs$bd<A%a<UInVHnxYg*1#%V
ziGNAsDCO{cB*;6-)3!|xXO8hkr$oGYJgW11>t%#2g72+ulc_|Je56J5$+2|T({JWb
zlSMVfXxspogwX3^cMlK<Hyo>gf~FF8b8;c9<(HIqSsT{&iwO2a54On>2rDbWO2CWP
z(B-9q#|#q?<{<Q^YJn6b^i)B!!m?f*f9c>=gnygWZk#7i|DF<%!N3}u_~a%Mw(j(P
zTwe5;PRw_ePur|Z*G%>mhL(>Jv)29mvNIt6oGNWR3bx-?KG)GasTxlbli75fn|zTV
zjSIu%yVwo#H&q+_FbikSi@=@Mes$697ttdw)73*ZvGN_;-o-QZDh6H@O>;z5R;?iy
zAHh4esrpn9EW01KYk+a>D)q<@&(nBZ4nC4aRc#K<CHUES0tJ8CLQm~oCsuxbP^7DK
zIS>RRX|`(R&A?WFY<ymLdO>LN8S%@sr*)Fi*=cLH`q_a~V7r-OQ{d&ZXXeTn3>Fg2
zgoOuu5FVkpIl_V%=8KRlWed+KqJc{nYd6d7;2q2}_Op%5*th3_0ZE_r<+166DyrU6
zJ!I%XX#7tl6pq^HDb2m!neJtZx#P{ok&B~gEj7lH39^)`qeoK=SZhreM6`PhE9>hj
zn2>?%{;YZg6@9NdBbfIrb_7%o57@l_%speO|IyU9Ai&TY2E2=tVHsOmI)7SdsVvm6
z(EFnjM}X>nqD~F}*Tf+Ye8#by+|T@p^vj;=bq06-uXla~3U%GOw2te<NeFK>yYk-_
z1e+!%3>t1VU1CIUW`jY%WDyAPDC}FTZPy_<tZd~frd8Rj<PK|n+O$^>JA}F(u!2SU
zd+=jeQwxvptbV02|83vfYInZhAeQ)AFErv{`-C@7ylnl8g!*cAhp3ZQ`r)T<q5*Py
z{-mm_<YQc^rIQ>8`)@hLP0fj?3b9r_=5;2NN_CEmk;cMRD~Y|LW&o^%q$dU#*i@ph
zu*l>t5210r34CR~eK!eVOrAIJzIzOUO-;;=6K^M|rW!Aq$!7go*@)TP-d<XUmwv+Z
z6jKnRQ_lDcL*FAPOjaD|n0{FLy<hLY^;`Sj`el@26S{0w5z7{`9;h>ffgzBS!BKb%
zx<@||Vt~@Mi=&NPqIAKlHlA~61ZWaUc8)=pa~g}@3*QQ`*VLmk3?R7H|Bj4rPasQG
znm1pb;^%O&CqDW@xXZJb%?0fysbx+UY4QUNSa9iz%8El0afX3X0udv2kahIeM+C^n
zz0uN3Es}4zJ}c(SB2fM57R2u4$wU}da*G708AirSLg`^wawt=EpsGDw*;H5_Ii5d@
zeYpcYmn9v)s)h|hAGatMe48k#zZ-Pn^Zs1EBqJZ+#F=~v`Xe^lyLRT+`Y7ajSQeIt
zV#^{{g=(Q0y+2hNeea{pv!{+;Q=3vcIeIfT@AMbWJnp8lvN<51>$f-V8e2FVdmsUg
zo-fO(3ZdG<l>NsWZ#Yy#wdOIpoA1)Pv^M@JOzOy4|6s=GUsqJT9_q8HlOX`Q(Rm5s
ztbpQtZ3w5xdno2`VHYu57jV&mSIKpXd|%~vUl9oNp6T`lQTzp^ss^uONDF+tgfO_z
zUrbQda{s;c&Vlb(dRcv6=LN_*@!HLjCn^mgr1kH_n$hUUM6wB{Y#b*z&9pdhW_P{f
z;8IHcg%-Bc%6$eN`V=~v)m>~fD}MO;h~v+SRHpTrRalrWH&)l|MsgS#r7JE-_@X>{
zRc*H^B34(9e(2Gdh`jCPj1E^)X7}L^E!}fACs;6Mw<jNMY5BXl=?38P*r{KHO_~yE
zU2!4jG>Y5)c-er>)4n<h9Vxnld$v;AXkj8{r9~~7Ol<Z!Dgegl?7B3$(X*wGD2cqi
z1@DE%-`>sB8R(hvYC@b|wFgU}L1y~FA|zv00PqG}8uxneKLu(LCyJb0^2{Wod+??O
za>LC~8EBkdSZ90#RDK;gBf=D0?Y7Tc5rJcBQ@0%dpk~5{|7yOHK;V&Yd;@7X*>YoB
zLUw9osTm()!<JUHMMX?<{As6}eY1&j7_yTF<n3C<=f2^9aJm&)LDb>7ZyY`l+T3-$
zzk_03;}pjkfvC4sXh$l6Q{CQammS4RVb9-$10)<(cL?CWi-1IBVutTS78#U2bZc0P
z&v5t3D$@NJ{SW%1&1^M@>rmWXSqOVBN*GrimU;&RU%cLT6=^d39N^I4yR)pdv9Z;`
zO5yyXN^uZ;!?7UNnLWPL4X%*3F`iT-P`#Kkm~VLH#^^W%>t}DJ<KUjL`?;T*`~r~q
z<UA{YZ?unKJNZQfhbjYE^rzH^O1FNzL<IZS>tza%^8Nj)g9lI8TiUu(va50n-ozUw
z<8I}-r8xz|qvwOZ*6K+SpC@0|kZjJ5H`4)NAWKA(%yrsyO}9ESHCfm|2-i~66E2`S
zL#^evU)7GiAi3kPxFnZSb(+H<P@!O!{c?`uiy_%AB?Z0c@s?q(oQiCK^tGh1`N4h_
zb?R2r^9qgqkyf5qni}$+hbBJLy=Ti!_sZe}1&M#?X(m{l*ReN*I=wF8M1+;2W_an@
z8DU|dTfD_w;b-a|YT%)SVwSq(af8@oU7sk<No>FJOIh%@KN7P6k#=0Z!R<^x@K_y7
z(BK?Qg&_z5YxDI)6+A+bk%@8H>LkU<QPb)km5x3;Ur&ehrxI3PmD<^89}n1^d+2lI
zI(Gi9WlvcfaHkTdvOc$OBbj(s*fB;vN?`wZ9^dJ7_zyOBVr^}#Nj<4SGoFf_G!A7F
zvf39U3^D<}3a8;SnLafT-T!Dv<(kfK3O&;UEw$<VRM43ODL|T;R(8_Y(EaI);7pc=
zoebxq96256o(RacqRFskxYSZ$(4o^Q$DW7aVO%zYp9f`2I2Z7bE>vwjC6H%1=0K=R
zYN5aSZ|ZU(?Z2EHMBX*EBs-hOFbM#qwj?l)m$5_ed@aO+1}05TnUK%<j{Gryzw>tO
z)*7S5T6VIIM7D+xTA$h!pK)-xWmg;f6381M=Ka_?*w=)e(`NvSjEJb?N7G_>BVr}>
z`-RQRYx2%K_9VCP;E@?)>r0!CuIb{T;on263m2nakm(#vlA7|Yq@+&!KHyXgT+P-R
z6&P^&mAGl1MfHW&|NkmGxCLgv7v?=_?_%LZxWqraGe|DDvJAG9Qff%u)6b<~gkA)4
zcv@3LM{maRSZ<q3cY2|Px?3QzNjYMN?H=UP94kgSAPtX{VE3F#2lUDHAv0T_TYoK{
zR9=7%l;{y$0}lAP)r9Hg8*AIcvpr!kzTBhhOKr{~%AP6TV3`s*>n>g+;6Oj<+{vDA
zM&s}vtU<pNeoWqRfet$R-6*qURSee&y&4CX+f`?Pyi+Eh!f@0vykt&V?3{9d{7fj~
zt_D5(uvlZ)@k+ovc~9w7=sBrpl>FZt%myTL`SN~H4amOI?!K*0F%agE3k5rG(%%r$
zxy;N?My_mGh;kB&$ZjF3yPH97RFWq;3SSuG&n{<NK3$Fds4jVYQ*opI=kGMO&taP^
ztPnhhhOm$hr-<SaiKXN_*c1ZEmR%B4LiS2E<P2a|c<(nW%f9BBt`#e2!n~;HH3nn}
ze|WW5@+k7~r;c;4M``U$j7;J3IB2s78e(<jB6A$~D`wzWWcDX};9j<OA~S~zt@le0
z8Qv5@MdyGLsmfx!IZo<q85G06MOi!cM0db~#Kf@LuIH72<>N4+v4t63(+c5Hgl_|;
zfeo3AE(`@Vzh<2G!z&m`%NNa<i2!=p-lqS%>}|>x>+vtl2-c%iJxS_Px4%ipB1p@?
z)Pv$-`7+rO>fzvv{k)q{h5nIHKVvy?^2=nrj^2iG9wZp&XQh^DtSZ_J<HgX|EOY%7
z`L1XL;(fau;TDfkDUp2TrtF+Q?|S(v(5+f9X6Cl={;!t+NT@nc*HiON4~}<o0R85q
zKef$W_(#Wq41pn-v*gU3(XGSzahEuyHyz8OVZ0q#^eK5Z_d(u-GG?rxsq?Ldd#QV9
zwp=Cd$=-aN2S#;K{p<uLiPt$ni}=BBY(e+9#-Flnq=)d}iSevP0Nt#D@hL(095OG;
z9rZVhKJ^D6#vUXb4Q3FbasKR#JQSY;9aD^jjS^CqC%PJ>DSbW7r0(WdiBV=pXB5`w
zw)u|c(ZZ7bZOI(9_7wN=uP}>~lAG9x%L}7UNtYbxrN@*VJ<Afpp!D+CfA)9LB4N|Q
z&blB<%+@(5*txRJl!5$@cZ4P)e`*~rNmUlWvyuT&3ikV_BE6>5SSTo$t8<v0QZ#(8
ziJF|_^zmr*@YsUM-mZy7MLCg&_dW`V3d@4XO7?Kg;3K+A+FftJx*A{Rq?ZT?1LSiT
z<0RlOAGk2Rjv_AB)8^S(tDkN!*}roqzb@f&t|O~-YIYZ_5_35Sn<}M5SmGc2DOpAX
zq+SqaE;qN#og8Wq?sB=n+yOJ4AT`S!ZgSA7q$P8F1y+3<^vj$>i$zoOL{KV|Mt>m2
z@;6{uKuCElZwS>4Pm~gUL_G_5O8&48-VMWl&v^C3RoFYN-V|ey{CzXl#12$C{cJoL
zXy36+k`nS-tQ-k!a=Ko?<(C;<ZvvyzwN9%oW=XyCcdI&`bo2f<(=6}TUTHcM&8wUE
zKq6o4ob4RfUs!yT<+spTcWa^ic?<}(!VLGb|9CM-xYF{*x)8n5(dU8RynU)Syp6d>
zqxaW80_Iz=MOgc+9&}EQ>k7EF-&OlgAUSVG*(SDRQ%N4TV9MyH^(O-!Cl-W1EaWMy
zly3PP@yP|T@VCQrGDv$)AL0fN?i|R$cA}MWreA*%A56`gcOMM02&I1XF}Tl^+-@B4
zqh9(<t9h$-wPiPOH1p+jZ#@tb|7=Rcq^JfuPH0nAoEd_B&h^aiT#;bdhwEWMz*`4H
zp7~Qq+P#3($?7W5C;?DP1#J{|!p3oIuKHgt_cVoC5BXAzNd35lC^QL=`F|FgxG{*3
z3;F8qoHW^vHPwtaznLE3P7SNb*_~Mf^-@G>QPj?4&6da|@dL+8l;7znij1gL)fmj~
z#r|@{v?n}06eU5}RiB?GSBmb~liM1cC8|&Wjn-Nluxai2{uhD&ZRy}2RNWrO7?xi&
zL8U;GqhundSW=N}fCtH@3(o$v!F5Qa+2PpL@6}3ziQgGTaEi^Nk@po<6#8IanE9RI
zv%uXo)b;ykDzE0cu6_U&Y(tAm;r(&y>(}Mxqm{_@we22a8MnhYKv7wKW-han=jmtD
zFd*Q$CvWR2mLcAh-|4x}Y44ElM>NjZUAhkF^{VLI&17@;21_fL9`X^$)u{HWVl~dZ
zRd)a*MRMyTipw55CivC_*+o<ZV;4!@cYkVD5fxGihY-djrK=b=7|6%&4fZ>(exu~E
zgJYsHGe`f7xQer`V@5~!NEG_iA_53ZvWVfCyBkrGXCe|Id52HN=G7L-xW&y9c%edR
z+RTdt+tr?HMmy4fD&x1vjg_O$s2k=Jq^rQ4ueH!LImgx1pCrt`)JqoD9t3ETXvjF2
zl6tjx!=_5MUS$a6e{{VQ88>#YGXlZjS<h2_saYGz2DuQm1N&1>Bf#LRhreBZeuJGf
zcSbVQd?XRi4-2->;9ZqnJUzDNu~SvtxvHqxT{;5C)#)rXLGtqv6EiH}XMB`&^`6U3
z4XwmIj9>HUZ$2+Sr~y06a!2+G2;VER4dM6HqX){WBuEe6SsMOrv4nsJ4J;SQv*q|J
zvr(U?s#l5e%2&h#i_0YtH$axT;>K!NHqRjh%XO|b`9)k9`Q^~EjGjE$I3}vamrpa`
zRF6Uw>Q5i}Az$4k>`}vz?EVkDJbL@%OP&t*cqV{6JOz8m)Jk9FBtL1+=H0!yPH3uw
z@A;Wu^^kt51W9_#bm8+}{N}-`imI78`ZtDFO}WmNY!nzB&C1Pvs*<xul}h26RsQ&_
zO|9Xr6Ri#Uil^v3%!hzf4)zRmjXaR9=csxqB5}cF+m#Lz78a9&j&3l*tT!^F{K(eR
z-4+Ad+!c584(6#>6esh|q@MG!Gn{Vg8@R#3Mj|R-xIojM?%af#=Fckqvcag2OwW`s
zH9pYcQ6ygZVkU3Z(CV5q+5?&G%@V{TY)}n#xAXZ0O!gy-hRcc3gSST{HIbA$oun5u
z2CXHH@AQYjSYJA9$2^7Sg4Rk^z#lhq-=uNeK8ddCi|>7?7XgqzoOL%CU{qBJZxOLX
zo!7ntdvMdMBISsuR69JhJ&{iE^>817u?;+<8LnDxo8fz0^W+8}-sO^>Qc}X0EZ;!1
zMvG|rm$rpbaNu<cg*z<G=z^)KUdiITy5`y^_~=ylf=$u0TT?RWNe#8SYwX=)y4pSP
zA^*DD*HZDw+T7MU5#iAYzPq}*>Y+TlvO)MkphQigl_~~;?6(j5Y*OYxu!1Gd@Vyy1
zM7mHw6?Io)TJ*u@<f_z^pQ<W&2^}@FpU?xuAmAfIH_Z=lse&u$D{W5fayM6`?1`2G
zy~1^{BQ)Y-z!orv3aV=4`nvYveEm)`wTXpXgAZ2^=T>j4i8qf=n@G($C3L!NkeI_^
z)22*~ml{xu1BJ(2+yUYcLmt?qFwSE0i7V%%%8|LBr<q+FWQJ4~RYJZT_{nMJ_Qem#
zL3l1ZX?3tdxo4mI{z5J;vPVro;!3axCF=A}t+CaWr@`%!i}FVwwn0D10bkoUse?eL
zZvqc%W^hJ&Lmt1!F(67S=I8mUc+ZS!N>oaU{L`nGUU_-IXD-Jm`gc}8P*YZDX=q59
znEs57jUl(}&s9-tG&n{_|9vhdI;@Z_&u1hgn40L7{#Dw>@r7v>9);C3*Wgrf7?|)q
zh?~NiBbEZ1%Yz#$2|}ac)$bQ=Y^yA*3r4l!xU1x1()iq?uH-<%5OXD)4|ZXv1U(+M
zndf8xwMigK0Zk_m77H7670s>{y)@{6yjYsQnQI>f`6Cip+q_qB#ZdIRao+ZhJ#N>m
zKQ`dKcCg>wYShhe9MAav0$!{fi4j+l!=)jkq9z}Z2+QL*Tevt>8@AilfmNN<g5ZGf
z`NHvFhB1alrTz9|pVM~Z_2%YgZC%|PfQU#+U!T<D_B>q)qUq=7muC9OVG%!a)?64e
zJ-sBOeuc-=#%;CKh}^rlHJDgow=H#lUsR;=Z!F0MSCyOasEu{H$>Q}h`Oq@JPE|D8
zoF|VI&Hqwh=5-3o*7u&=497w)A0!!hSBJ;l1$H6TyLc<U<8^1cc~R&T0BWy{!;w4N
zii6Xt3xp(0D#R?!=wZF~E6R4ncuvZ64bn(H;}2D*B`|#N!E3nU5HZP14F?m%iTeNk
z{XjzEo6POR<aDGC<9WMFK`ve5QKs>+1PA;uA1<b~=*{U>%D??-027a2ZCU<g9;>ND
z7yL*fomB1SzK%3s=ew_fB32Y$s}RCt^~j>+&G45qr8}PNsnW3Z$tLQ(uZydV@ZbAV
zxXF!DU1Q!FlY*8_(#-q*c-|gLiT+FwE`qQD6sCT@Pa6oWw#j=slT}dt0{TIu=Dc4$
z#n=p7$Y$zh)<Movp;>IgnH4E?kuuyc!dCMlCDV?zB)KtR3$U9QNov#p6`GPH?p#-Q
zbt?9+b#Jyu%qZ%>W*tdcu*r<q^?c>}WL=3nikyOlWg_FEr1JvOhS^9Vl~h~0m@XI?
zK&1Ixq1F=gw$}tyJP|vi1t@SnPjp1&9~+Usf4s?QBtV1RFOS|h@`C~US*lGr<WRiK
zS|=sny9fbE)Y;aRwVtrJ1J*@wRf@hMdD0KCW&)tlXnx~dS6;i!Zz!>yGW5a3iU%R~
zpmk`{8;o<|2uk14IaS4uYbXkq>!o{3O0rf7*!^+OL@Q-;KKziUrSg5F^AW|__FcFp
zT~)YSdiCL`ZfyloHa(T3b@@B6kD}zo)MQQb6yPj!9a+0wYkTzdu$XvfVSkeQt60Qy
zp>Th^*~a<Pz?B^CwCoXErq}Zp1H((PunX)KR8$l;G$ap1Bh=mOi=LUC)!5AQuPp}y
z760A4Z(n`N^L&Lf^Rxe5&)0cDe~>fPSsd&$$SIa_8SQLvS(uDxdEuRMgrSQrg$I0Q
zY0a0LiAGnmTfKX?`2?wydqlT?KfYf;Bn_H)ASwRQiA~wFDtpyQo;ILP$A~eNOUo86
zx$Ubw){IHxxyDRJ2~G1>yHCr-hnw}y6c!yxrSVU``C&e+*O5flSv)t~Ih;}}GuH9e
z-^Atqc88+xlRgy|=-%)VVXVTG09>D0zlw1l^!8MK49rWdwl+U}LGFIb(e`_lHGHv1
z$;|$eZ3q3GK-r6=V)ywL3SuhBZzvYiX;meO<sexud+Hf1B}1?*?7>0b)`ZX^IGLvX
z1w*0xybXa^eTL6^rIjoaQGp6eN=k}@lk*J`(NL|;dWn6|S9J$enW%kTz-hsz$Nz9`
z&2?z+xj_5XqM^8(R?0=#NwN*_6p#ubxV7EW9sgh^bJ?*fBwh76i2Al{_vYAv!k`-(
z&*Xg1KtKtOMvNSmcq?5vPvEg8TR~xkneUya>c0$p?lnsT>jyqvtt(d(b2xEMGlbMB
z@UGQ|?%4Ugy`4NZjU6yW^>gWqCPTw3_t`gd9KRhoep($lkFUL|3LZPV-U?@r-mr8t
zx~?Z_9C-=vO0~{y7X9Gh<V%s*<}+j+!k%5%MN_-2VGv+~TIt9&7qOQ=(E30&Ss$sx
z(t+N0Hh(k3Hx*`sinKq8R3jkGp$u0hG==&{4h`S7)S*1{qVz8ifGYF)qvZ#9U+=6e
zQkZ&&eMJa&m!7U}X<1oYcXz<VkIm)o^_~c3yDf?K_I7J~`z$a|y5FDl;7rAoe^XJx
zMaH4izrQ}YzPr=2vXb7PDvsy#;JUkl);K4fp4wv5YYuGmMj|33cT7#iczMADpc4;K
zq`KRF+-eb*l4=I5CBQf~o#41|<?WB5)jQqjWwBZ!P*YP=*U;D<%SIm0k<{x7#&Y1Y
z%FtnT_HnvB-^Hn(F434D>G)?0tJ{IbJ>nh#8hM3W%*vzowG~|LG}9UDU$iCgu*;_O
zZz_5D`<;)*iLDZS%@K<M#gWN!hPqCkiwMfUUQ&nyCssIL=jZEL-Mtk5g1SPld@6Ow
zj3l}*?qFMXj2)pVI<(6FG3A}XM#WB<G~=F`=8t}v{O#$UE!jUBd8D_?Cdh47Q7^40
z+gjRJS7kz7NkUSulYAx?@EX9~A0@}8YZAbuR$tOA7O-La?3w7chRM51W{xu%Z`I9e
zL4gz>AFe6J5|%bKL~W*VqdDsX4MKdr?_eN)$(sbSi_I<emClyT*)pg1NB1w{B*>d@
z_o(M8O*#UJJ-^Gzy)`f}I9P0m%}w=809QAhh*xz#j#{atbACS2?Witr+4C+eG!%u3
ziYh1~V!L9PFH<CfD3pjd;q<gWTIaSOINj_&IX%^F_d~ot?WZRtBLh>ekgF?~*;LVL
zFGVWp=g&KT;P#dy@)_49q@^iHN&V)nT9UOqZ~U2!hGf9U%F4b$KquNLgf#1(?~J10
zvlUI*k>iQ@QaJni`rbTTEx(9kv%s63okeZ>&&M|l4@qq(w@E~oK>A?_oU48<^f5{h
z6v;j^)xe%Jl>qzkLS5H24akqw5KE`qNg=q`tf<&Iy6f8gaZW@6+12OzEF}(Bk77cy
zoC`<jdi$j^rBbL*h=GbySwvhNARL;UaeqQj{LY_%Vv!u3RXo+cu9QA8PIxHpqC<(l
zMU!7v7m~_d)8txn+SLv`Nhzmnu)-n_wOj4Z)NwJ~7T-m`earp}Ez@<x(V6{kf!}t7
zpj5O0p*YI1Ee~nb+c8?7L?XPoT5lv!Ta_wS9Ivx8f^j@r5Vf_n^$oq-%}r&pn5#4;
zrl+FH5Rat;(Y@0A7Pw{x6M3>x(b1E&5G-m%2K~Pn66E6s#6T*$6_H+dXubo1*(AFZ
z*pWKOzAPauES%X%ffq1|xO9GgZu+!8nu%c5cmn_G)hiI99D|>WHHc<Fpk{iYG>-i5
zc+@r3=;U9*3kk3*1^y=KF%(Y2;{_}6hG~jrXN9akVkUTyH^fZb)Yt>=_Z{;X+Qm<f
zO>_q<2!Y7{=#6RiUeFB!>7dDVB}$|2$rLHro&m!1wt3-r9StQmQ|EeInE9>ZP8pY*
zvov~$ZqIgMv>Eu<!KZmTN|&*oepAC0Jg`p5%67=pjQ)@lA+vYG=S0vfCTDu8xNVQi
z1E6^FgbXbOaBuSkh$xS%!a##%^4~LW^8w(V5co<d!H;5A9#xJNs|R1bO_}{80~w^3
z+e6YhnWGvHvzMTsTzt3u^ys9+{ppjBN`>JYLP8CQ-wwE}VZxcKiB`?&X=yN#z26KB
z$Yf<@ary+niHW0@r^o(FX-UaVuol3z4i1LhoNY(O#5n3WTD!P#BA^jykE8)#U=Eg=
z61zfhw3vqCnT?#+ep1lg^H?r$fE_!O^z=RYOgaqj{|*isfQd*VfhB*aXBNZ8P+uPd
z2M0j_OkF|2!5Ou+?<0Hdbwa_TW<9T05x5)=l(*jtb(G!`iWnOk7t6Q40QaBsX)gsW
z(X@HVmSp5VS%#I$QsEsCH^?xhrBd0C_vLbA*tf5^E@{s5N4&-i+_{TBM~Qbzk+Ub?
z?J~2r0=B)4f+2z1UX-7*A{N1CugorOcpLzoe84T#3r$Cf`yM&GgiFcIvazKcBZc)7
zL3%E}fgAGo1x!$rzKmtz1Pjn<yPN@I6$8TlrCx4;%_AtSf%kt1t<KkP`JJcX>VYzs
z?j$w8j&c%OKXZCvqTJoIJc28;zT(#(F9@h#wDNW){!HZymoCit?bU*3e_1UN|3p)$
zsq%ebjqte5B-_%S-pTkV&s|5}T&B5pCK4St+N(Tot`_vQrs`tK@xk&+!(=+~ImKlQ
zm+y9m<lL3l4%iyd8A{@W0n3M7KW&XQ_{HFSto`!kOBy0}a`LT=gl2Ha!2WP98WsU<
zq*#p|jgYHzuF6bQRP-aq^%`JiG@OF!)zHA5E1lXsc(^8{czAFS6dJ1E5rB$9%$MY_
zKXn4;HD;T&&XBpm1lIlbO9W!~)=*q#I#>0~(G_5IwJq(7FCD4Qe#XbOdpDJ!W#B0p
zMEY7x!p6p*)6>%z?1p(A!ev=P;ps>J&4YcUizz(}Dj%J1@`SxO;b7N1WL+)Ex+HNs
z-sooE^&I|SkC!Z0vRX*$%>5-_hl~GJ_`CI;-S%HV#P&%Qgf)MI?*4)EG;dj;X_#LG
z5HX#V=7sn-i#_=j$1V@^J2Yp~OPf3U{{8tdQovG5jovIF*Jsw~B&H&9sEjzoXNlsb
z^C_a(^&&v(C?&USlxc6!t?p@Bg{2{%bpt<&o{^6e)+)&_Vz?EG`%6|-poih;_SS^l
z<FS_1$sZ%b+v9pQkbuM5D_xMD&Lc7fNEz~Zg&fCZfaLA%eGXkt1@nt)Ps7vWU7<#)
zWbXJ0n2yx1T^8#daCl3m_f<-@U%)_{?~ces$6MqJCmYYl#9=yndbFZwii?W$!9$E-
z?2i`WjZ$24rfT;y($`SX(1bH($v+>=*Fe7Ly+iHU!)@etIx_#u0TP<mA4>qWS4MUM
z#pRyeTP}lyk9d%uHm8mT*Zod5{YBiO3;5Dtn`dwZ@~AN@$4%{<nS5Mt!bH}1*9IKM
z#ArKH$cvU++7-EZJNwnE)eCiJ>JcLQJuVGo(5(!0_n@j7N*MsT%3rDPxuk^{VPV)t
z#OyxxEIln6#^dhcCFLuBt$`dG=a{jiPg2lO5^hG{^2WCZ>V(HgC+RBC>RR}Tv;G$#
zvtSJd1L|<DDk>@p1e7l>q07UUhx4-_lJxi60KYRC&3IL)1lgKWvnqBh%9qcbs)w9q
zWM<Mwal(-0yEjW6>P7H&%S-b*$S%&E&2K|x*zv+}i28eb+kjD77xZ6XQd+%YHXIO`
zY`r+br;-BU&vPO<s^6z0mrfbDgSHIwUM&bhG@F8gf?CJNE558_PnB`C%{RGS1%wLz
zV?Ve0Z1OQ$FNfypr<Wcmln7zKmX(W$!rx$K|8{OA{q0G-ACpBPKn(WwwBQyTkX|tL
z4DUMyqTjx*ov2_dM6#Zg`z;^PYy7pi?7o|)I5`}&5yy1XCVC60Md)$3LU)B+gg6Z&
zPrYZ_(o?h+z~{`vq@bXHM|t1s^#b;Ky5-S-)|~v~M{jig`(ML+cPL)JNkqVDr?7^*
zO=uh<B4WM`F?bGQP|)kQZ{Kn_>`e$oJa{|+pPMwTWN*gK3JM-oE18*@&)OY93`tAW
zA%}`Z(F#Jx-Gw@4XyX|gGB(WzDyqG9y_A%cfjCCrKqA-IT3XAu2y{iC6rTJsBx8E@
zNk7sB7gaKYS5&QR9E8Vey$0to1dPUEh}*kZJIvW4ZErv4i=I3;GMa6B<w}2`odd2q
z+eSA{1C65l)@+BmQ*yvS@o>QcS(~%}-74YHa$4ae{sp|89ww#OaCT$bJ2xka1)s*(
zI~a<&ej{OdAIJg)sV{%Ikz;xYVS&oA-<74Hb82;3Nk^SLNctVq8Y?AKg7W#Kt$vns
z#3YL^Vv2zcAEA1b8x%%h%Va1}97za9E1aWg_|Q@g(H9L~gB!9eLa{8cFEz3)C^HtB
zj~4dib@6<LwyGK}9zJdNTfM_DGds%{%-qxc{319h4RR#TlXhBU@L=!wMimW6egeQN
zWsJd)O!d5_1b6j73NJpWIA#ciIB)zGDLB;Yt&KgqiZ8i1xnj)&@LVtHSk#lYt4ziN
zz>OUaX?aYDih8%Uwx(QRhzi0-+6tp#!^?wN_os)eg}bMNMJIR3tEbkJ>A|Mn!KI%P
zZxBCyLM&6ls0z*yDy*CoE$p~e*sf8_EumMqpv}%^%Ql5Jz54y_HDdWy^VwC;$GibI
zE7EHo;s_ohin>(bj1$Uy<j^#{-N-M^Z!va=#Hx1+2L*GqjOQB<kY+ekhktl|1sSN+
zSG>Q)y{n-E`xN)Xr)i5Vc1QGHlcc!e5I@CSV5D2ufEV7CT}H$oC_mxjQpQ4^HA)yL
zDm|<M_?eS`1*JIg3&4^R1sOaogpm?)1WFY)tUHKU;Q;MKhhn)l9x`)ynjMVPjA6VL
zPvn;Sh<!-nQ0K+5wuG36Pdg{49URUfJq3Nc_6I3V-o9rwG6}%;zyQLwg|H=Eu4vDP
z0FN_*sUb(!y{WeT_&;_+zvPfKV*+|&f4ux288!I>)XUs}$3W(NG}*Y{K+itA@PcC^
zLuuP&7f}35Q~F>klHlF}xwi+iPegX9uPPtGLfA3)Ax`si)Yjw`W#sD0uEgc0{a;02
zJJhLs3D)hutKG|yVy(kLT*bIlpC?x_KKH$R?epPvJL)jcnZ!+}uG`g7F-u!wRFq#!
z3y}S=W6w2|%8zc{lcjrf#BN0jovR8V;&pxhZXYPl)L36zlX%Q(%wM}_uTz(bB$qTE
z%R&Up?M4H%S#N6~ULuh#49w^O;BWkJG54?60Y4z4R1ZbIgNB@5Rl%t@aI23XJyB`c
zF~c5qL9Z%6iH|23PSfPL@zXWg4=&({U5I)wr|VSewyt|={utRN-Pk4%@$gosBl-^1
zE)y)2CCsKaS~K@955W7Fx6Qs3tNJAuaOq4$fo8te=`Cr0z!|0V#7vCe&_{uuCr#!e
zRP@kt(`m&|_RyyI;*HcwK3q0Ixq@}cuLVCE$A@Da5o`;i<4_~xH&}{=hdWhV)x+w8
zk53lDkV)^a@P`E1-~J@OJZ1f*EMZ~~@KMAZessO?bd?~Gs5yZpzOGNiMWm#0UPBhM
zg-BEBK)kBaR&)@tnJ7=3s$nd28F4OCHWP>BcX$onYiJS3SS&wk_=@tzq9E95*XFK%
zAsTq`Ax+IXEFo37IaV_D{%1k01Rf`WIl`bt7Sl^1)ml}bEoZ(mH7SgSO}Nzvz|mj@
zK{}(9<x`tVAkvtv%1yWWpK!6K)%H<!E|IXPv8t48pNd!dni2;6+x?v9^K1db?kU}e
z2@3NyKx~Yh&y@@sO5Wn@m3ei35y=Kc_wWQqi{9SoyDZ1Wu<t{f%i#cyO{P2s+!qYX
z5o4Ck{;n>krJ|PA2>!<pOibltz;tG2W`FE_rHS3V86O|tX1{n04I_um-l2yb(>S!{
zDFr0!V_76Fl2}F#0|NtbJT6Qi5{A_VvD7mhmZ~QEA*KHf?u#*oJr^y%`!Pxb+kLot
zde(&(yUC&zp4dFM2zEg~($7V{kR4R!>-~s$buJ*drdcH<JNh@~6!03gy8vuS9*r2!
z*2+*{rPQK)b8CgidobYhs+VA{E?^^s(VfAJ*)qk`iG9U%L$i0|%BVT|54bc8LpYx_
z1T@=lYR$4skCD3OnN_%Su+f`t-Y2K-yBReg8H-KQxEZh^HVk`J5eVP5GoYKS<pI<a
zt-<dqspV%*4)Mlx0SJTp)>#}Sl`+`a*|N_fu5}AxDVUp=;}Tg+%Kka}NShDmDK6YX
zk!Yb+T<>|HnOFX_^7w`zt_E=`s~rQK5?4h{g=!|{=YqOaMGDsNt-2J%Xxyluqf)!D
z65SWS&DhF1Kkc2b7(ybG_9&KA$%2UMJyy#71!@t2g&DCPKqwX>xfNm5lFL9MU9WRu
zuD7%C!=^VRq>F(+{$6mFhTx1<(2z%Pr*uf+a=9FU>xnt=QJ@?dE?%?KN{(WhS;3S;
z{=a5|#IU;P@-b3(?TD$*G=rny8s>VmE_(eD#A|v-F#?`WV2X$QoF#nM$EEpcwZemf
z@_@=k`j-24u+CB!c@&NsJA<lYKLDM-x`eL9#KeRWaQ-pqk6|<#A`p%slDs-vG$SmO
z109Z_H#Az!NuZc;dN$0@?0J8!dL@r5-{o?DZF72d_A@lJh~<Tc851un>*Pw@s}9y6
zIB3{t!`2q6U1iXR*ACl$9Xv4A?qxmzEIymg`i9^{Cu{As!ec~P-B`>N>{6C!!yEGl
zo7ba(H76&zzj-b@#Dyp66!H~}C&mc1vgLlNKWdZL@6GSFpq1VBxTLZ8J|P#m`Xzpa
z_fSudBpLX=GWXkgDG}7Z#B=p@78&+f3WLQ|PS+yjY-=(x8?io<X;H5mAdbTV=@ppn
z9^SO6^gT-*KNZoH&i=+|4%a(*M*eBwQ#_VMKF-g0=T-bN<;0(e?l_f0_mMNWIFa{Z
z*{!{luqkkp)Tq?{o`$k+zT;I6yq5tMt0cZ!o7<p{IJ~>tZ!q7lA4~_OK+yB;v7ijV
z%lewN_LPd_ZH4#i@+R|0yWX5YY`TQVoejEz8s1tfL@jHhB>8JFYh$qLB(|a4%i2O_
z_xC+j<FP7_d&8mePr|;U*Ej}S6&wD<*&9cKmdEui@0<)0R6BZTwKeeSXJeJ`rd}TK
zBppN5Z{Wx!xe-?J-+VtUgHujMikPy^sv?T>KxA>{ih!z_{KOJRFyrC_u)3P-@K<yD
zk3G-krVTWIzf=-88KZpNf)^@<4JHjgH<g}nLR>~&I6`hmQU(SL5b{Ndkuk-iG-*nM
z2db8pmAQkA;vUp_Huq5pxhnfSJKEcoL7E^KfC37L@WqPa5)vpN9;5;P4lIb^koa4;
z5hC1gHe+ejs!;$hu&Nw)<6AL!-a99-nE1$1KJb4loRpK1X$9%}Y_kV9lD`k7e9pqu
zFi6^zo838KFv%sfhf{bP?~WSa;Nf*GEyFb-xoK%$<GIpj`;eB?4s?EPJ^3?l$($f&
ze*RR=Cf7G_-hdi_yFji~vbBxP7xGA1P(oK%C!lWzuGcJj6l(AGJcI+1l9En04F@}M
zetdDp&{)9f@ODWoEwxR1ny^qHcd|0ro{D-g?x?-LCNj{iN1#=o8%~0?r@B;SKGpP)
z7?lUKxC|0&RV%ax(temoshpuUf6x_LgjYpQ<Wm#wzA+oq+aP9vaMigLe#ddn#r?Yi
zslk>3e#NL6ob9k8a6U|bw65}k_!deXEfwUmy5pi;BB9YQcvkpz!s@TaK}Jm~<>Vfc
z4HYPytS){f+GD<);Wgt``DvoSp33)?$^%S+2lW-r&e`X!&@nE@FoNC3(^jg)+!S%c
z{aEVO1r$Ey0`SSHjqI))$*$GxcD2Dtfkz;Bi*c&V!Rkra(-iI8Hu~+c0-2K)YwerH
ztq9FRy-j>ON@|+?#$pNK7AN74?6Dp6qH_5G31hHlbRFpgN$!g~_X~mwY|z^=nEcC&
zW^?6VG{su-cjY>sZ*7rZc`Or|jEUmD`ho$M{*lZ5ngJ{w&jO^gi;Iqdfq#5FE&xUn
zi24r>4|hk>J1lw^L0|#Sul9+nN+z%%f}QU>qnTw@RYK&E=`x>RqoQttKYRT!b9}tN
z!771nZ}))gwGVhYYJk2*LTVoydmAsIxRe?hiAl(1uMKL&>J5&E;hvzJaJ`ji)dy%b
zIy1JlwJ9qrM?q74ePI`YdWZFCEui=?omnLb1`JWX<CQkh7>p(c#Kpxy#hb%^2Lr~@
z(Q%S|3Pe6XeAq#TbGALad}+7;|7iN^u&SQ#Yfz975J9>@Iwhn*K%_&Yq*1!N;Yh1=
zcL_*$cY}0ycX#(Ye7?W;dEgJY7w$P{X7=p8*4k?r+^wyx9qT<-shOC9K_#~EzHXt(
zzu~2q)#K>ync2M!B}`J+gO7XqMupnq$Ux9PoEVtmQ~xIUm~3g{bgRp~I}b)Eo|Pl;
z$mgOWKQk}1rLMnOZJ;ExuDdZ(i&JPqc*FONY~d6=W$L1dvbBU?uAPn7a&oqm;@-`1
zkz|6g62605)u)^F{^l7u^z|H{?vb_EvyEDm;|J~^617}I>rJPZ^q7e1W;S|UYMVbe
z4()#B*p8t~sO7vg*oqmHY;tpfh_N+c?K!$?#b2^9H~*mr`-ZdIK)Z<)wb`vnCj0S}
zZR60NXnpMbwUM9NHG%iGoC=AW@^$CC&Fd7$?T0_3dlIBA)NXCrn~7X%z)Q`ZkLEe0
zIV!tUC^u`gL)**uPO1(vbpG;DZm4b!#L-zgy5GB31?BP`ygo{~ni4}Pr^p-1*u`sf
zZgHehGC99LF<0Ev5TCK=b+S2?pB1<efw?2(1S69okW^#4sXT2kk|E)CIqw+NXklpB
z1!%FO^Mbp@nJ4_aK)q^ZZB5TAA0~oY&Y(M-(*8oRuEg5ma3P7?c0<S9TvMOd(a}*h
zk?XTLZ<+<J(jPJ|h^xB3URabaS6?AYMa#J()(TFj8bOoWIczNUZK8YIv87;7s=(ZH
zp3&-l`Ec2MGEf@`zi?!?j%=NmSC((xu0Uw{gcjn8H}NZ-n{VTn$1+wh<9xq}rGUW<
zsp<AqxTeA%b02Ch+$*^QBg-r5q%V1BLze|%s#9~^tl*qgs5~;m><|#_gusA5DT(Y~
zbl{Y!1`nb3divgF#C+L=)Jr+1NjJOmrl50AM!bynw`p`t9j&z|!=o+FXO5>rSSO@!
zN#aFSbZlRsoQH;AYeg%ZDCGG|brdrrO#Bu0e4)<i`f0!gZ!k1>S8DUEzPLW=T$DnA
zw#b!gD~zKev^)A!%QjSYP?;aUqLAp?-M^frQ;cr@+fVDpV)}se07ndNYjjrU^?M)G
zLoaEn@AJG(X-P3~1S6TYW5Y;%kt5tATbVpfu4T(biZm174!8SOZM9kozV-_=;lp2_
zoE?s;f~*XFmUUU43B~yBxW`3Rv<2d}XQYC$a79H91TuYL<AKn@6f!+|gtlN`Qc%at
zeMf%z5)2uNSTNyov*+_fUdLd7t%Ag&q_~)rl=R2YkUS7}`~$$9m0i9^>Q8B^RX}%B
z1MvcwBc%rY_#>mENUvUP{-RD^>cG)%A1K@jS0#<I5|@$DpRF{%yWULMTWm@J{}Te6
znT3S~K)ulM@Fc*l0{*P1Jf({SN(2N;EP(0(+l~(qA30fBG(aNy`1`ZzG(WLeT6rfZ
ze}8{ria8yTVm>|_l`C`DSBqxS>={Vn?e6VOmx!RU+Zp@yIsMZ+@CTn=Du_9L;@1la
z3H2p#hBP-f-=K-2`_w^uovTZpNnPX-cn|VII-4a5UXHW0VM`5SlzRdc;~)0d=z^R$
z6e9^3>V@wrBhHhTf)+j8FXm}f+IUO8d~g*juPd!C=JbocKS-3%sn7R%HyZ})K1W&^
zdSy;c@)NE|Ha)_GqBoTIo-;B4zTEmpTMeOBb&|zu9#b0noZvI43P!Hdtx+R4?(S57
zTZ}2f!s?BYg+M8D#OwF<FB`H@+6EeK9`hr$UxqRH$*ehuMAGNWQCQ4&yT}}wjN>zb
zi`<tA!H+h}`RR|LHvQR+Cxt!><Jg=)UjKARH_6&^xj%X^Cc=qSIk$lKR>5~uoQgBO
zf0cjhyi?vxBli&sSt_U(PB?0fcbpcmMa@#)7iqepV`!w)jBr3UhO4^RKwSB&>FsQE
zgNP45Ug}r&-~4DZGc((oDxm;5iQY>4FRFcBKEA~1GD9}k3p4NA;^JZ*fJvd?vNUI_
z?(Oa2a#|2R8QvfBczAehopa=!>zO4!CAat-zoF0RC<L*TV~pt-_0)37^{c|jsHmb@
z5yYH)_B)E15pTNnSn~@CK%mbV)_g65ejt})7y)yHxTCKqm@L$g`S2|~Fjyj+WPOuq
zuBAGdFKS3!Lirp14o1`&q8&dCJp&mvw+MdJw#Ax{r8%LF3U<yr2`nln8Hc^XNfR0V
z7|2XR@bYK*l%BDVeQMA$MZP6s3`V_#*nkKsm7s&^M#t7;!@&5~$8VfOn=6gDCtqx7
zy6HPR`(KNZ;5?&H$K)e@hVvL7(k?^WgpESj?JgpvE+0RBYiM#1B@HkPvUXWI65T*6
zSFR*E6KRDC(5@b`)ERhh&e~1Qgi&|R1^v1m3`>a@F&094e-O-dT|1m_XzGPf-j-96
z(4sqC%MAkYsGX}Z%)3WIO5ICin{Zaz9;W7z7HrV`8Sh{$A>UMa<-^J0#0v_Gr`qDa
zcMAjckt(LQ7qcFq=bd)Uh^^04Nk!thdOEwShsCicf@vb(zscB)BWFk>j)YQpIWgbr
zX)=-)1hxk9rDHI6vRpsR&1^PRMMcvGG1#%5gk|rRU>oq^9;OHS>|%;n$1f4$IqlVp
z+#jQzQ=81BcXq+NS%?iBoZ20belf5SnHf2;8x(%h4L=`PFY(MTi6RbVu@ZJ2NjFH@
zJ6GB_ssgK24d;1bye+0<tqg=KUWKF-0)qo`-KTUHVsrlG*Bir}P%8xn`#bBG{)d7|
zDZ$o;-Ofy%-=nGzPIa#}x6VGv$TH`7-{-IsM>K>;kM=kGxjXI3m`pT|@TgG6xz&@#
z;am{&BaS(_e<#Y)>1fTlZSSIIU=Q)*xd;E>C%G7&z=8hntgM{6H*m`W2D>{S@KtDf
zWIIO&j>4jI!!_7*ziHGH#If)!Ta?<?X=#y`^<53HDox(zHyUiC>ibf*-`NI&RW8Au
z@HXfqqPy6>qKmCP1pBit+$$rM<+p}dC5($CN(F6AZRg(=7B-uvYIx-nHDS0-tlocn
zW9gUMvp;7EX*@5447*ucIAW|<<MS538}54a+HYF6qjd7{>*al(`8sZLW??N$X80%|
ze0T9BqeCl=rVfU!2Cvcr4Gesys3NYA`|Kb2i9MbQTh8^{tsCxKA;#64&@ii@GDw?w
z*oxabyq=uFjKTWIuCqxB9lU!aL&P>6vqUuRIuW|QMck}PA-})l_+5GumO!6jJ4W>*
zRbaW?iH5Mo**i8@<Y`>sgtB3e`(@C5T-K*&LSDKKhGWzb;WY7@226}q*v@7=s44Nc
z)el)Y`)LR|QlF%y)GZTf()*sVY9zR%7Uv86%9z~dKDl*pcy#3N;P&OzE{=Kl)hE|a
z1_^TbR~JxE-;C3dQ=|Qvk7&px`kGZ<(K<b-8Q)G{fP#;A{PVioE%jt~wk(RoxI|@1
zkNE8Ux6)!vXC;eGk5QU`IB{_(?%&?6<+QZ?T)UFkp5i!|YW+Bj*Gb<VvNev7;WR@K
z_AihYE(&e1lA{j0+ywO&nr;VG*JrFKNeGIBMcXjqa5*V72y^5JD1%5^0ED;@;Xmvm
z-enRICEI90v!72B!03ZfU~TVIq`gR7fKRX>Mnogts}B-;u7-o7{;(e8oR5CF;-a5g
zIl$>D)=tX~_*JG)Df|3geLlQQPCUQheSgcfx_pd=T&%XB7=BRl9DN)>Q=qlJS`o19
zU!OxE7;}QHcD!^?%h-!_3@>a63ocEI<B+IxA?1Cjl@3|KjqE2XHn!{(ANMASdr{vw
zR=#B><gEo{HebT|7D8h`7vATOzh=T=#V3{RLGV%KUx%JhI!Ydw8etL@#0rG<bJG%O
zgsc-#9(UB0UTvmRLQk0Nmdc<qZ+w;WAq(2T;xl!&WeLl+3u?LN{x)IiG)6P~12dPL
zm9{qP0#k{)PF^an%o;rQh(j;K;#yH+sDl30$0E<N>Q;Nok?H4VtffhRx}~O&A<$!d
zy7bPFM<1fEN9D|_xEFCQr|ydPZpflv;Tcl65;D?}6!xJ7*n<)KsUBHL#b<9!mz{{s
zq^S?B-sfdKgh=_X;9j*v{9~kpiV|qo;6C|TIo!XVilaO?|7n_ORf$f3!waGi>TWvP
zTau>cQWb+Qk#WrPI=DNqnn}%rYB$m&6q?!q7PP065F0$*{fGCI61Vy}<{UdbA$>f{
zO&YwAJKh@umg=xGXB<7(Q$j404BK+Dq{LOcd?$M}qU0T!#VRpfD$XMwjHl@IBEx*P
za!d59;Ie=|0AZ&}z7HF*SEz|m0f>0%!-o$rAO-&Wy(^fD=Y3z-CY%rpV+spYug1AN
zvpuT8cZC3()HP8XQhw?x9;_A8;EyTf4$X`SfmAlNyn@`!2}hP;_IW^2tJI4?@_yyE
zT*baOHsqgFl8i=kIqt4kpc=Qg#Ptlz^9JZfUxaiFhFU(~;4Q<av||D&6ADi0h`qMd
z`J3(bQMVm4j8Lr__5Q^4sF;@vpGEjAgkC;3vwtgDRaJ!#x%z9-1eZ%)s4JfoBso;h
zkt*Gs%<z->&*#ur9{UNC9`~+_gPg>=9j1a(lN#Eb#T-j{!<OBt!>dF)A-?wgl1^h-
zVk9c@=P2uU14gO|FDOvN)u^MI|Dr@zM*E=dYUqDa`b-;;eB*a^l{a#`xo@!arR805
zv<Ro+_NJbIn?o{Y@8uslrqy9fZ7AF3t)+j-X8@#L(km#8jQr4BE$U949Q80etW%h*
zq*55rFc=<POr=$ll4F!UiU=Mp&=ZenL$!c=KCi1pH{&dk?`|vbAt<~))4AjBVohaa
z6pqn+xxirQbBp1xf@dz(;)4FR%K>npmJiNV?Fb!A#^7*ys)&Wq@mV4uYNr?`&EIMT
z?`u=Kh+P9oJg$Vs&aixLlcYb5m5Y5n8{;IrTE)`1Bey!sb&C8ZDtV$X&3%dVL2_(t
zEzB)#dFOfteITFL$3^Kk`|rxKFa0XYyG!X#G{YkYIW&oGd4=2c>!<4g_5zRsmPXm@
zwtn-arUG3Plo&K%ncfRUow>W}W%g^{5ez%%IR-OM!>1(r52`N^GOA#@mg&mSVNXZ`
zY+TWpt)|9n;fr61$qC;TtI+S^iVgcOHEPI<l+VyP&WB$_z|;{u<qgA*lr`OPEEKQ;
z7Gy!f1v!j(#k9}kq&trewX*9zn39?7SYccw3JD70sZ9$PYeM}hOcRoa6iLYzs|erL
zjQ|+Z^p`m_yiWJx$XdQXylDA?C6A-t+V3eag58a2@G`Hw#&Q<Mi`YbDFB<bm=~?}3
zy#ppqp_KRh*Qwg+QR6w03e}Q>$Th4IyH<eA-A2CR+x~Jd(5<h~+E_w)N1c3r$kg6Y
z<2b&Kf93)y%RGom9yr+fTqbxsqYivV5W9K@N0qd`_PfEU;v#*1E4gA?^uN)9TV`Ut
zK2-*)=uFee!gABegUTj#Rn?)tG6X=6<(;{{PXW{^<cP0cb<N!wd@}&5wX5sv*8~Kn
zJLvF2Mml%A2*c*BAI37_&F-jv2L>RRmTP%HYJ87w;}I)~c|Iry;!cdjDp>ebb)SOy
zPCB@@OV;ED+k*Sz3sh4*9al;J!VIDd{`remRO;+<r-a5lLl&fZn=QKEBQ&M0-Suf8
zZLv}^<WeQ=P@Tf)&D!(f>m!fEGh90vv$|v}I6sebjS9>ZUzD8>SLvFWpZmUiU$T+t
zhbl3(;YPou%8P7n-eBP3DU_WbP{=XI)5*}-zpc8X5(+XC_m9JY-QsXIs^G*rz`n#@
zICFJ<V`O(EOGwyXF#S7@Zd*;9jx1H3o<}9j4qluOT?)WsONBm{@sk#M1qr@q0t)}8
z&fBy{GW?Wq&$?9!8#Hy~-uH1aN6F>{L&U04&mJ3m;)Dmj3rc6(y}6sO1m{JqWo?M3
z>zjh^A0;trr^a_`711&&ObuVFs-9WT%7zu+qByBbQaC=np`{n&fkI}d@;+faasz9U
zT>SqA%YxO}%Mk!NX0w&3Dk?cIbbx5^*RNk`<>iq;&u9yP>fZqYzo3ACbRZ#t;o{-3
z0n~m^PY)NUA)IYgfv_u2g;|jfvoDbWQP>-#lxgL7!(f>!zIPk;7J(XtB(yqyu_;+I
zQBS(JYm<nE#^8(T=pl*-xT^69xTGKw4+VQ120p9MlC<)$(Q?tNCpRIwi$G)k)b@*S
zF5dawPL>KwRW9Umwv(Ll5FJjQ-jKPbXzom-8_DUClniLd!qS|YsPHbx18edxvBIDI
z4x_0^rVg#JUdkhDKDV$R(s_o>w{qxXRWCBd^jL`>j8GW=v}U-ZW6DWoK1ZgLpxMk_
zj_GCDX3)zux)-Ouk#;S=*`M<#tk=ZlKjg@n>wF+|qJ`8Xi%sby5;bYOE?1W70#e5@
zAB=EngGihly$e72mo#hvWRGS<Tw);(C$sy3P9fJjigv^!g}F6L1}4Pg@>_+yisPeW
zQu$H7<#mQn1%}RZZ~Qc-&(=qij7sk-94c3TX=}ADv`OT4@Cx8N7<yKE*=8WDATJl#
zvTU;2_(FJXjV{5R`_BLuLIg^im7_oV3hryi9u-mt3oL!l-3r8cT#}{ySLw}_J%Y9J
zQ}0A%;AVxiIVDd6xbJQqo<GC-Up{^CB5t7_>u-e`V-k=33P1^;bSV6;7o;&NcfJZ)
zqpz{BIzVNP1RzAPZPL6!2IrlE-4dc=VnX|&l&Ad8Y>MTJpD_!7wQn!>ubUok1wRGn
z-<8byS@f6O1g9<!?VbbuhmNkUr!d(DQT?neS6|(AfY#tL=ybmrryI+p3+=YDoWV0(
zlxHxXr=cGp<6I{+KCT-w=7{ZHw{uP-X;^TAi$kLN@%f3yP+*nLQB=NxrYYx$j(Pul
zy}Ym-SrdVg$<IXcOzh9}QIyP4>^;3P>Q#PKWjI9r{qu~P(sa>Z>3)@c>a>X*U1#35
z9cQodd9jc`<)wN%X>lWnLv)c?k;Mo5iy5y&V3ci)zP}=lRzR6H;Zl(oIGLE}PbOpQ
z9C!Ixpp|7f8L~UYYxtZe2vT;*=WR#3%g+ewxyBb~@GJ98olKXHA3Y5c+d9!$M0lW*
zxL9Z32)POWlfOHe7(_V3^?@QYGO$+}5Vh3`$Uoxd0tH6JhPLAb&T;L>|4b6<taeG!
z^f!4#WW)uXtr{^!eXn-8ch8vcl>XEX3jfE0RV7)s7bEQ^i+0J+AX#Kp3|-Z5jEs!a
zDNHU7ht=_d6lGUbY`}83^zbnq2vP%*m#iTKpEh$i=PMo87t;AP7=SzDUi8~ul_RmR
zG|w8B<16aZ<>wcK;#H4kO%m&!k39+ra>jo9sIfb`<o|IdWWWE%WlHtx`dr@EyCK_J
zb__?4>gCNX8>D*33OOXYvHoiE*00Mog_f34s3_urKp$X*gTsjD1sNH*zBSw%$e-!<
zJT7#XPL&$9G&H2GOTc+gz{giozs<=6ZYhAdap`EeEnlOiv{B?WTv~7}UqXe=x+D<e
zA~z2TK5FQDwfFS}0zeBL==1ID*rA7OYM}XSnVU<50BV7Wg$3UnPAP>i?WKb5!}B&`
zcsYj#?4{oq2Oq5*+CGmk?@`~kKFr2{<x#~P%c7X4LTR5KY{nb+eNp6GB(F7pf#(m+
zcIJ%Vc1ucoOH~Py=gY@qqGJh9#=IEg^0eJQqa35uSKyZviKru`G-}(U<&75WiR&G=
z&^Cq)JH87BQ(vn-yWOx@0|*2At?hx)=FFIP@v`RDk)mP+MHoFfTg33`Ihl{(jBkVO
zdn=1kYf0Jlww$}!nd#~oOukg2c`f`OniTI!Xnz1#sK11Fe%?(Mx`%_0pxh)K7ka?)
z9ZCFgP~fx{V68gCxz{ecH8j9}h}HSvoB)CP1>JeGY{_ld+?*jlihUPHvMT0<B<S*7
zI!WU)e&Z%a-G=E;N@;Lk&@ItNdZ|{a=$D54zVIUOE6Xn*enal7rhW~cH~tobNt^%D
zGnt>16Z_c*;|p#zvpo@`M5dp=*OMnDf@y$p*`au4vl6rqFa<)6%4e@Ab2h^v+5y`3
znVg;4zW>87<_%9Q>!G`2v1=Q7dHIU{=g%z#m@WYx$YRik3&Uf#CI4^<u>N=!VUWM^
zsB<FRG~99lo<me^4$G!UPa*uv&rEOBi)2}+%9Ff{Vcq2i9o>SASszck1eVY}sck(?
z$xIdsl9%+9l|h6@$dIwZ?pq^|o^!oWxp4}2)AmW^;B|=#cGFtPkE9`we6Q^#@FoW|
zo=nB<jq$*3S6{it?)~*m525!KaQ^K{5=?e(hT)tu9l3$6yKWSa9`Z8C%Z((vd(0$#
zwCvUJ_Ow5Z^x*2*jfkF>r0ADFox_t+KL%RZEf(_|Me?zvAOuqvIlNDY-6GxM;sC4i
z{*(U<WYJR`Hm&sSMKiDK9-|F=F2h17-(p9kgnu#j!hwkkY$hgyyFg6KPyOgZqQ91=
zWLj)*+sruh5`o!pbv2h_Jq4%n6EUyK%VPGtv(wED!z6v*qRChk+KDk%0APItv-~M$
zr@?|98gkqmNaExzYFRy~3O;JE*CZfFI9V?*XOC#7LTqgiHK1w(gG$WkzRUY24c5wH
zCD;~oH6mp_BzHD`DW!SW9hdbwv?dCp>7qX<Bn{udh7F*JOIR}*16B4-CH9V~ZKFVS
zEm2{LwSmmgXDXOm+XBM0>4vX58*_RcT|d<1(LWoX8rgp?fhKhtkZbHjotcF7kd#Lv
zKNK2C!b^Zj$g2*Y+sA6BHc!MUAoX%*sf<Zmx}562DtQP!j<3bk!lUd!-orxx4NYnA
zMpM!5wD+!HccEutVS&8`2~4Y3e5RYb>4PU@osQA;=eqmf4BfOr`{nOa<mzJR7G+uj
z?~VR?I=~zN15K5rwzSZfr_2#+ZlQ5X<wi1HJ$7)Ltd&P`-7Z>~SpPvtjfp2n8NX$=
zF96`8CB+t86|XGySKWF`Y&ll^Jc<7)ec#s6T>iqiPGN2<;}o3OVVLwEGAAsecLHa5
zxX|$Ri9>yo_XDYVAyj$~8saR~Fa5&96WFTdzmNTYF~Q#|v+0rc&&W?VKu4|3X-t-O
zkq&=omA2)QQrLM(iD<v3c1G#oR_FKf_FG164Hg(2*-NBXm^b2C_%dqSPOI*AE&pC8
z-6Xh=W%E5x1@+dDR{^0~anwp;EAs3=#y3777lj5V{g&k0JPxtD2vTrxev>u--RV@4
zV1f-IrcU`S2qoetY!PmsS(V_%&E*TlH2x0^atZ~{ofoJMKoR_sx&8wn`HaSMm3+-t
zS67`k;tYEr2=L1Q#?!1-Y$U|PGu~kV6_*b}q?nYpnZ_Osk2+0|q%FHu=Pty6)sFpb
z^WdlJX}pXyC8A@5Y~J_YKz+g+^Mo@5LtEq4f{n_h>>uQe8jQXoHG10OhWiPp+_g^B
z6ow5N`|YkWJe40T3O~5M#p$t2WiMvB=^z)UZ!1&!Zl;k((&Ea&6Q<8R_?MIsr%w;v
z;??Wm5j}GoJiWXnY1kQxnnX3IB86@D72A<ux-na6o$m3tCs<sB_ur}adi@$L{2a^W
za!Uwsr)yhd5*$!J8BmEpGzjc2^0=zDynoS8#@G*r2wxIV>|Q?&6Zo=kT|-&`g!gb0
z1iiPI`>9qy1v)kWOU4SK7e1#&AUKPIn>!B3-PxdryOZ-j(4RkBfH%KBgznE(u1|_}
z+oXy9f`4`NBfPuc=-kQ`l&xrxmiMuTpZ(b;$B;;#|7<{akIxISyoo2}1BOFN*iKgP
zsW0pS7WFkyNo={r+Q#K&VO;A^QS;q*v1{BBD9tqaR)hmwvfYgD1(ex}_<Z+<sS)`@
zo)A$kWQ@4!m~1djUpAb}KGcDfe3|F;tVE12RsMNzVVY)iYpo-8xGiPGLIgqQW%plA
zuM&sS%qtx<4F<oz6BUOUA7ADvz#4HByZg%ICz&B_O-R<hc#!aKiAouy(-agV<KPl|
zdNMkaGyTw{BVq;B;jNXsKPg|Tbn;SEw_ZM^)ezs}jZg;qo?x&U)!Kndl1?)V$mjCa
ztI*iW64C{qdTmPedtr<w3zKr%+?JB8VpSH&*l}IXB2^=G(#k)OabBwn18gRH_R2F+
zwGroBJ7=g3&&%|y%84{=zFb?Yn#9vaEEDRb;}b;kNK1L)mcJB%y`<FtuU8&&y(+{4
zQju2kxV!5S$l0PpU3Eq!*$v2-5jWqHh&^^G8S51nDeZgB#h|*Dme|{--UE!%EZ+J7
zpI-a%tK>oB5H9hlSMEXH6y4Hv=g`D!f!~IQas*Zvw$yjSg%trMXn@lNT7WjQ8#`Jz
z*Vm0mF9Os8e@r|V)b``!fC^qzbT#-oX3DrWKK!&JVi5Pc2wwW!TiXs8J0qrGfkGca
z<;O1B|E;_1SkJ*xrT&qT@csnOGW}j`Q18;2<AAO@!N<F{Le0A3(%O4Lco0&)1BFl^
z?z@%B_M)eWiQ7BT@0pzwA{oL6;lBD+k^R{@O<YYZZxc$}rm&EGPl<@y?%QC<U8#q|
z726h=x)V{ysT$i-4r6pEuKbN|w&|;;zHQY9=GsOsnBNsgT7((V<9z?N+^$xh!8{jq
z!m!Qjp(X&N`x6$EyTa-;qK?m~F)DPrd~u}3#u8Do>@7$mK3L!!XrB!#dME!I5e+i8
zFc><3KPZnqmnv-av&&D~ciMU+&MU1^F~a2G8z(?!KhgrVlL_(Kutu!&>$tc;e?ffj
zIgL5E|B(eo**grjbP?TP{k#I221+0`nVM1sY;*s?d>yyr0im<Ab0U|u*TelSMP5gK
z|9|rknW*#P56DkK)BPp~TRZI_W7f1Dqw``iq6My(&L%Vb(-A+ZKL$Cac@FPBl@-R$
z-!OCDo14dWFFig3;(%iy4v^@{E_E*ABu*1ev8q-^#*}MIHls!QidTUat?A7<D3lYe
z72$_5RwWZP%T;yFH7Tp@^?|^tr^?~>`A{dUpEJO+%5J;P$E%uNu0hzypB~OwFNkGF
z-X&mAM44twMk5;}`onhv?P*}*IqOC8xSY!bV655qM=+RT#5cEeWXCtNr4gF6Jsnp*
z6r=;R0clI?*bV=G5lFbrO2wjMc6u800=!f!H?rn@PyG~`#v{6&uF^{Y2T{7)WzTX-
z&LHPWqhbt=rR4cW!gVXQbh}MO!>ULEq`{bGo*E^8@d`feSx$+Hc#y<1$aHJ@n<_Qh
z2Fuv<`IWqnMf~DxdEWHqbP);fR=SSO?$+dtSBgVcGe4R!<K*ml2l%op2sEI8Oe*9q
zD+b`F>5u;D#+<c=70ZblB`Ru8Jy{zRj*t6DmXK;pClHa1JT`QNsgFGxDe^jsC97Y*
zm!j@l0oz+FI%-l<Y91ThlsM)GY-7?IMaiCne?F{44hLv~H^$^zX)-3|L~Oo^)E>@3
z=ZB0dW4ZT79)#t^Q3IdpdAA93&O?={f@bEUk-pNcZP>hkZrIb>cqX^4|F@pd10j{i
zMT`IRG0Fe>^gcGehkImd0n!Wj?k;BW3Of4XYE9D5qr`1h=FOUc3g{*Wcfm%+WFs9%
zdRsx~INFpRw>fl}^`w;O2p&go`(1rY_IININ8{=9M{<^PcRC&*c7USDnqe8Eq{F!E
z@~R-5;hu2e?3t!d4SJp{!VLkff!K#Czjx&01M^|lQVVNaD@l9F70KcE+{J=ft(7*m
z-1}YpAFGg_gUDMj?y!c9OniNZr*?*9c*~jB9VivnBce#2ZAsuFN9<?%!?2)|;=k00
zbnMj=fgprV>=ZN$&w4P(DG}80AA^)ELoO&ISvtjO)5_Kkg={p47>~6geEdq52HVR8
z$*5)Tw|hQLwBzTm>1m0>HZ*>kbO*Q!?7X@)C}dVDliu<l(D}wW|HHB|n<qR%vy=?T
zX3@CgHzE8aYx6>_;)O8b2i-N-q9`K59<ozm6K{zluSmE3D`}FM;2ohAPus*r=lHob
z>EER<i_5E^PLOo{)l|cnDzh)Ztg8y^Ms&$WWpwekUSAVk0mRhm9S1VoDc~WBZEc&g
z%wYh=Vr{Z#P)qdYf4IX=qWIC%=KJi2v>(VW_|zmh1bTZikkTW2)FxQ4d|VV`<Cqz_
z5(p;4q<2KzE8Zws<6*z&NXV7}sV#3GL;`eytCdhv_r+$k05aQ2;9(wGtoqlXdPm%T
zSO4*C2R$1$H;QM+L5TUI`~Q2gh>mheluiKY0F5+yAl4EJdL#nL{@G2gfn8{#Ak<hW
z{6Drp^h@qvQVJaVU&!hzu0U@Z78mE?Q~Xy^^yb=hO{>4Py9*1d+E}2$|A4CCy?q{u
z9Vs6h=HJ(c!t3adPfdF0h`fG&nmdBGIDIa#16uf7Tkk4W18o{!`HsMKfBPTKCS>$s
zXN|zs&5aL$Eueigld1M8Jo>mQOIVPJ;T)1%>`D|SHQLRZPRLfmU6{#S9Wk+QztJ0W
zQ0&OOJB3=l4IJ{xSPvvsyf@d_Lvo=>^f=_V9(?OMx>{&swr%k=28SjS=0!DEu0!N>
zfj@`CxUL1NLt<uYw2QtZs8Pu{7|=wam5!_OPLp#MHSM+)hIyBVsu{Q$8Y_^jcZvVs
zN;|-$dhAS&i4g<(4_dlf?wU^@Kf=9!4ZU7@rvmO|*)tO`gjc%;4ufiPUv+cN#0+hZ
z%Hc9WszxhINA7YC`$vT&)p$~e_`xYqsJka=@v&opn_$See1s%_eaRHQ(&pb}bCd!g
zF}W``Er8Bf?vS=^>rO)(^Ps>uY6Gk6I70B3maBX&!b@;+XCZd#bJm`JlG2d=W4$68
z)0E^bwJQe4`R6y$uNK5*7Jq(_WmtSEW}B8!woH`NhfvVCK(gw1=FkDQ(N-V6e@foz
zuQ$)cnP3`Ue+&Qke~9P9H;@kJe=X~YsFO)}ufN#n?&LJ)#)|s%a`hsRjpfJP2o9JF
zpc$~@ZlQ_rH7}~=LiR2qX)i9;n@LovtC{_r+XVF6^O4xMr~ViajBKm2AMj#af@6U~
ztbT-P4b41UAmxtnaD}~sv_xJ1{s5}k&7SUnz$@$e?u%)~Qj9N85?eiONd#B;z!R^9
z!liH4u5Cc192sn(MCb1@`|0G@OFjWZa=;y*V><C0Pq0C60DF3WU<NSWDf}y?!$Wo$
zwU6WG{8Kw19G3q?-keIQTS~EWYMm_(9>8kt_?C!Yw~W1HJVW|F3I#4LIWDfYEmJBc
znz=}|MEI@ae~ErPz_|L$cf035Nyg6OFBl=qJVwpI)8b7IqI#hE1Kj~&2iC4h!~-Ci
z!r{v3#*>K1Ii>g(-hQ1=Qbwy*t~NM8Fv-K!%2}Wz`9=O8SJg4@?!rSqJFhu8V67Qk
zZq$L~clb57Pa9$pZ$uqTvZPZdjt8nYsd}NP_P1l(#dM~3E$R1{PaGj&{Gu$d8EU}>
zmUXiH0S)Z6pz<-Afk7SGo`wYetr--b;p^p?EP5)xvSofgbng2=%I2PA`roMH7w&mI
z+pwHz{TE`0F5p3~mEJjY7q_|1y)Lt$UHo%(6LW|6qYz7cH)j3u@$q^ufHdFRUp{Tc
zR&r0%dBD6jdG7w*N69^9F*5I%gl1I|fK)~Yr})-p?6gmmKpnilW?eA)^p8ZPe%j@D
z@Uo_+@VbpX>(fhAEs2?u=`t-<x0fq29ADG^C)XW)vEGKwK@hZK0;`Q6@{<SvDAPwR
zE43jNp4-<q=*A;8<2l6?We)UNQ7zOXK2<<JWQwlO1fsLi<;S;xU-fpP$?Nbjby<pe
z%ccIq)z%Vf6|>7L<N9wfjZ`Ij6y45lzr_a=b-iS3G8|uUPn4+tXrM)m^F?9Wb%P;=
z@vEf&FUz6Fg3F(cv8sz?FTGkoB7c5-OsgH;yZ-k;>uI$g-~oys>|t7Ty+^tMGn21~
zR-Y3QbvML!Ec?#VS!RuvtoLG#ob<|&P8C89v&Ne=K_266i;ve0p!F>5if9;c(orS|
za;Txfx-ws5TcCJvZ*HJWqUNB=Ie2FNV~;-(_hJ9#i1_#*b93$++TpKBw!TL<U;ll<
zJa&4U7Gjrss%p9e65AuRQ~zq=SQ~6sQ<iX*Oz$`xT_xmbw28pPdsYLR2<RLO4g#Z`
z1}i!6FQaSC*sa{?U8X)=r(e0=dAfHE>#;L(R-^Dm1`;6=XS^JF-B5c=azU5e(Dt|l
z!<<5g<H4;wAT^#eag2-w0D276JF%<@;lEJ5wh<EY%&GfdAnq)otjj<ezo{EHTRUj#
z{t<K6EZzB_{eC`?eP0_wdo00J`i#e3;s<-+2Y;K5P)$$ULuZ7Z?sVBdPs7N!NgtJ%
z&4^&@q>wvVdl(Z$6}Y!BRIfrK(uC`;q5Hy?scqMANuJ5*A!>X4IZnWG!po!1um6-^
z^W?MLFGBS~QLOSb{myrrD#u8c$j72bd$g)f@FVmzEuS-dW$=!mIa&}Mg7Wy&HyeV>
zQalg;i&p3dw8q>uXR+*2m`FHVd;8rvo1I6!%>pQhMqYUpSO%<k6C23Fa~iQT@4U7)
zyN)FEaM;xt2R-f?1t0I_+z(qd(}#`Z^TyAK{B)jCrReRi$7pYWMg;^NAV3CXN$Xg`
z(?s!c6^7;}B1T%#DRzc$RfVegDZ5U^d7|Z)eN3~I4`Q&19g60~C)MK;Bz|rAs&JJ$
zQ48JA;1idH0qjOxz(Xjv0wlUe!8bK6bG>i(_=1!Rge)vhiuh60K)(jpE$6rUx|b3m
z4akQbUKJX(ua}I_I`FI2knu*0RDt-_MFw9I0&)L+_SymJdnO0I^IOGA_lDmTO@|*Y
zZ?UfKf3x@0g(=X<g=dCiOiW@<d-4lE5omk-LDm0L5mg%+m8%ioZhZ-71CNC72fX)f
zTK>ki&C8m^93l!tdq@4Np7VMEQ?!CR{87e?aX>Vq2O6Hdu5WB?1RKSRA2^V<TNv<%
z_+lyi^g!_WfK<dZi`lb``-*_Y<x-i55Ek$+jjb_km|rq|FM>PGH$gY?JC200g2J~X
zQZNl#bVRz+$-|OpX%~@L(c#37<9f&QQf{pYCnGbLcmk)+fTb5r<M57sHkE=(eM%<c
zYxt9-feLlKVxP@O@UIHwEq$<%3iyV@q|f!G*WY=9ToD1-B+lfSUSmayuo6&QUh?Oh
zBFSmzGAHn1ahJS?{wWE173!;AX-&F8xEwThWiP)74xFmm**)xXbSED+s;GZ&ipJwE
zW1@j1+@^Dq#z0F;YqvGzDJOUz>A2wR-x)-3I<BG(7$IM8$oqMz$;0T<_O|g(D0l*J
z%Z2p+{zYN)F$75|JaeMS6d1Lkcl-J!H@b3I+<4A`j}cZrO7UpwR#P$5!(EU+XpW@X
zd2ht_xP9>I;XsJB29L~}7lXB$LM_VA>r4a86ZsjnCwysSAQu3x9N^w=8u`N;ErU{N
z>OxxsA&qi6#>xY<sTyul%nFC6IJoSYKE8c*<!J0Re@L?PE9%b$ZP8|{$ko0biWn1B
zsllR$pX08{4ohC<xXIU&Z}32kz6t0D;4SeQ9sR@O!#y6*J2I+Q=G)B!iw+h*X96Th
zTv8Gl78VxFeJ>SlcPzur1kQ#KLT+IQ1d!fdHYZot?*<O1x1&4bdCs7pqZgXoJ%4KI
z=!;|XzrVdqA2-<^%fV!>KYG^Y|EkGc^ZSKpf%20ZmU6yoOE3{{waYnOVPWA%QPJa@
z-6BkEY=fb9?&z_6ExVhWj_UWpBO?l0?&st9MMXG{;!jKO_DAj7&0BdWno69VHZu%i
zXGD@JdSq8f*TBiy?_^)RWT%{G_2!+zwMYEgl!SX^N(e&YoUkXry#Xe1PW0ROpWamU
zl<ulR?=SfOc8}^q!e0T@+!PT^0$O%wtoa|~QDW!bc~MW8>2w@4IP$mTjhleJBt4K8
zRo&m8ZI31jxW^YkjW@sZ^YdrJD@fNz{>moxfE%{Fy#=fcx9bH%iu(l~U<1<u=f7Sw
zA73|WQejjvGka4fl&LUWq%8<+Kb+MU6F4mmPB;32A|M#NQP<r4_Dg`@UT-Arm#gD0
z5=+yR$jJ7)>(e+vNC5PBKXoNI5^|=YNwR~P3?X{KfNW?)I2s8fU|NBsvJefc(d=)i
zw;F<I{H!+3Xv+iPl_hyS2|~m`E3~3>LCF>VG8dw>vCUw1M$U9DXdN}Re@i5zGdr25
zDsW7V&38;0kn)%eIyyRmHl4PcK355<WV(3z04leC(Kp4`6`YOC|K%2NX)QC5n{+&l
zlYDB^=X<`fjCa^J&d!kIC~zm7s<H%1w3hbv_Ed$Op>4a!=x7W~%r<aSbZ)E<fDh3F
zouP0D2)2MVmz0zQ?^s)3e+Qh%z+o|AVPS8uv7Ze!y1SmaxI4FA0>g%_^P7_tA@3Ja
zm-jvNMa5v|@bU3+IYCFHz)3FXQ(Jcg_3`E9CLnHsJI&17+m*w^!I4ZU2^kqM(P9qU
z;yFx#CDC$0)dIPIz(S<|U863DV~+!U|LbV`H#7Xglv@3FL;yezD?i;uxXRv;Z_Jc`
z?wGYHO(W%(GnriYEV)-G>8U@NDz3iFlP>{}0MEA!p?VWYMC5n8fzeTK)63S%+T8P_
z*t>E-<(9yjoD(c@V+IBAYq_(G+um1`Mj4r=&F3~Uo5&__#MWVLdP~^8{o(!p&r(WK
z3L!NrFW~e2lv?cQ)IzXNq6V_0WN}%o>w11LA@{T9L~dIPdhJw2I^*e5N^O9+6753Z
zL9rT-yw%Xq;Bq;mQLbt?wX_@rm^QHb(g&tI3|jTEINEL{cjp9|SakFyz$&1#-ublW
z`zgqIAmG>+2>NMO4h{yMJTb(@Nm*Eivfbm@jYo)}vZ|`hFfrog6zCxh;8T4`yj89C
zBz9DCN0pX~3uBs3RckBMAVx(|Y-kFqn;-7nLxcUr9%DSl!VIJ=rS`ru@eZgPV>{1_
zd0*BuWPYxdpj7DgvltkfM!L&yQAUFYd~x(Qp2X)M37C++D)ucG0BQu83f&ygYsaro
z43ruYkdZ#Zd{UY(BUhZrFYLmomNY`3Sm&y@A~2jFF<8roVvOgQihCOQMee}N%-S-8
z)yr)j|AURNhkfiDs^w?+bFu&R_)|U$F4mp4?n7<Kp_#t1o%Voobm;T1*Vm5VIIQnl
zxm#OCwT_2c>Gr_p&r>#Q71&nPGlEttF%tJkV3~IWdVJ=imkI_d&1YepKwG&bkQXfN
z?e#&XMGc)*iypmxs02>?6_bH8F}Yx`fc+6f!1)`_68___7f<7xO|jfoC;k%BfDQ8G
z*m#unH{<XOBa5P<b*<;Y)nlCsM*^yg?)geI7Unk#mT0ePglq|DuV-7aZ|XpA-cpu4
zVx_kaIY5(cZA0U_w@!o~=|mK;VoG$NGZh7!$zF^`BP4K|@p7`lAMnF7vkafZF_^0Y
z)HC~ZuDTz--<zm(FXQNLW8%6I31+_LGTz-{wFMJ2TaMJOZ~kxo2~}Hi&rUxJ)l@LC
z1AB~BNmHGEBWP0yuBqY1q?XOlZt}=iE%5^VF?O{*oByicDL8;Yqzb93Dqh)AOygr>
ze#vtJo)xgTEIJ=D&B1HWn*nbSxDZ3|PBw`rs<@cN%-p>9@~Nlq$JEr6+J2k2@LQ*+
z(3=(v3VI($r!V4+-Ze3%i!JLIR~$$`0Q96X)7L$T+T`GwIx0np_kz<)&bv7)e)xq<
z!E#>o&@9$TCUn!lNNnO=5n;Z*yLFzQbU_aU@(Qe%T|i>kzolwuOFMjPgK3A)2*B>R
z{Pr<AmF*h^DAGasNREfFC#UxDcfOo=dJRw$4ml`>%oH*I4<ecSa>tipV_?Wy8n~17
zA#=q>OF*`=($W5mNu1&vCMG5xuLHwV*lve@zGk<WtFC`Ab8)#L+RgYYxtM3qLu67>
z+nPdvnLW=FrY57OFVQtJUU7otVksRS9T#VO*7(%(SZ*@@+p_W64@|w^zkdfV_#<>^
zN*qN;MusLP_5p^a#_4!vek$W>>W&FE5X=0Mj1P8{p<}2eANMWzCDm~x%Md9EIaUMO
zG0*H^qEB6IYYPtZUdUr}?Iji~XGC~W+Q=5%<;H^2d%>!8IKT$@f0DdPg`XP#Gczjx
z3U^ubt1j)MTXjayU*+IRlpCXbyLyAS%_a5mmWHlWeYHM7nIYEbhF#;^O|S68LxL|<
z>BUNgxRP_E&h@1C)KqB`(Dv4BVF7JxE{CP1B{q|WxP$}}Fjmx=$o~ol4}T0AWmebM
z?ZveEyStHofKK4~_n*~71!I^eErz;-*7k!wT8i4(+QM2{*OXmYn_o~02qeD}xPk$_
zbIL_pGQfHD`lG2!194OjT!hME*A7EV9xw_3NedYnnX0<_Xf<U09W>`5A|h57%wH}-
zgJ55N@{hw0`WwFRUplq`#HDYItyP*wF(M6sO7mV-oGHhRvI0#xt993qtN}$$GTqp*
zPeXH?4&Jy1@GoLBof&@>glpcvVU3-8v&M?{Q0qHVFB#tL!q00^7#2-^BNb6TKEb7Z
zq-bNU4WY!mQsIAYktejjeceeRpYL1sJ#)CvSIJG~pK_Zc&hjbTP)n*@=gMT7^Zxh$
z!~71o9|srQ{l%0FUJ*VHNcUuLhND082_3>fOq2W^ivK!_oM!ho@_V@5hYv#)F`9@!
z^AZ233cUgCBl539$;A(0M`*}=-Uz!SUED&eRAI3KY!al2gQsiD;B21FWL~znacqbD
zmPNrtM*FM%DKSiC)bF~%aWoOXE-z>cUUZ^XkH>G1$0_Ggg02_fHe<8)?=={@pDj8|
zNsbG&oP{9T!<bDKx9o7+ZU4!K1?*KcUDuNi?uW&Eu2xt2`^{4KTL+j0=+PYlZQVgt
zRh+<NDO<58ApzLfsJZaSs5QKCdinAt9+wp<5X({1(dEvRTwh;5WrIM*Vg-!eZp5E#
z$Bf=kLo-=sd^7kDlV9V`Ynd^Ld0&*tw(ay5-_jBG`7DksL7O@&udlZ1OU0+3KOA`&
z8l|nTEIY9>D?M~yZ)-!ZbcgVCd_qRhS_m1-S^fpA)C+&BA}O*HKiLk@_|=5cB0DrY
zU%xt)z2u=Eee`RjD+~Ym-}#QMf89oA9L20i#C(6=SaD{yK1fd`zzSwIC%ij4dOa2S
zyCLBDD+c%IBkQ1Bu^RLzXGbG0lc?X5R1cUxHa%vAl4VskaZ?B(?v~{>#a8bb)?96E
z(=4MfkfOl$NKcxgk_Zq!P59R&B-&ZA2M0ZiXEmTLbU{$>UufwRs3@=w277wGwRpk9
zfF}9UiLQb@4N2SLih{fQKG(wrmE9}ObQjtq*;p1&kQBo9L{xt7LL1xZd8Kso&9f=@
z99&U*(10Gq{<?}@(%5sTS99C()>tKWPR-cVic4y{_tqegmDk>a&rI4AC5iPRo#{xN
zex1`^d%ZNTnuwZ?UwJM#T(sBXyqsY>{Qi5@<O-L45~W*#2XU%bHC^zjqwe>cld}!S
zEuKn}oo|uYD8S*4-f*eHNYx3=LA&Eq=5n;Z|4|<l?TAD?HSGoWWb+HI$Y^D#Wp0K$
z5))iXxh|XU&CI3g{2M5Fd)7fWeYp5&J4-%~;o5tW4|v5e(JSYZzI#`)N9UM6EG{hk
zOpfmu4h9%+(N~4k)rAU1u>F560Wa{7d65cKv-xTjzd@UI=`EOC-0nB*We_lz0bO_N
z7pSjZAuItCL17gMp}E1q5YVoT{$ruR6$mwtOQVs7hlfFEKuD4m1F-G>Y)q3)Vd3UT
zoueg;L0@dkaJpD8_ID{cIYSV)o)Vpjkk$TwJ2Kl(ZehUF5(aE)FzS_NLIuhI7@*Un
zJ)FU52F6Ws0+4%c#TN{4>-ZWE<AlFYk)v7*E7wG1T{)zUU{~)+=@4W$KDd;O7*4if
zt8vq?edm9;$g$_u@0beoDVGef=OC+?<0^xnv`T!Lv)&okKaX9%-z_`PET3Pb6r#Zi
zF@NfhYbPp_CG}8wLu-i9EzvKWsl2?h|0$sOefd6)7Zfznhu8NsJ%Mje*5|yh=Q~no
zt{2+*A^g3?mXk-*|I*4Y&$4Y&$h|$UZ(p5G6gI^Epd1twSqzm$jXC20JGS$BluB;B
zAQxBlNbNMxqKT`i;U6|!5`%CGMcqDL%v!F4Huq(JQ2eazS1oRU?@#H&smgg@>m1CX
z>A*Pvo?;LX^ftQRZh<V)Hx{&Q<!d#}tk`ZZHVHm|fuaK>I#HFbs^vzn0eq6jy3jwc
z!{!7S-_z24(4()yZCe#tcB)NppTg%nGPC`3FG1wY4QpR)FC~$K@uNz!pH{v!X&<6j
zmzs!=W5>z?$2*{vQgLVCg*dp37^G^~xSY46NT?r8I5rhGztf7&*XMC_pfC(BOnZjD
z=Epm0H9ELHY@5435Se?pqnUYiK-ZxOK1LYPGo+4}cQHi_7;<r0amQJt6zGg}_CGzn
z^*O$?PZVE8I|wBtzF(vuY}MOwTOP~=Bo#3B8pv*zynY=NeM=3cxV3`RFP7aH16U=2
z<LBq*=ka*#Xn+?f1@J1M&iWm}2r3#--ce9d`Qfn{3}s5;u;ZgLF)^L(O-BI7z*6YZ
z<Hajt4U4AGP;}sv)CQcoy#4(VUU<VM$@+JnP4*0_6+{Da<f)PLDIo*R>EIW@x-bcJ
zmqNhEvBai=-*!+mQ=tT?5P%5bBQTx=MG*u+#LjE4gGmGjBIE>&JH%?YOKwmWMmBm<
z6_2HIa${#KzMrZf8yiOr;{G>-B`?<V+8;a;%gbVfGFy3zx%XfAF7K#RT_9XAq4j*y
z+Ghr02?>;fyziNgrkaE_J+$)JkWpF1QrT?y>QtSg2O;m~h;8n-w6gMV)ryF5JXZ4F
zF+^}YbnyFh!I_NiiG?CJ<U}!Y)S=ef$T#2k3mXrvDlC1!u&Lw`cj>W*VaI&DSVuv&
zJ+VHs>TJ+w<5PDPq0qpmc^W6{bkpVa?u*NfVx4}Tjl!E`%k)3j@9Aoxo@2+0gARzq
z-Dl`MENYA%Ww{z7g;;R@7r|N63Q$r1`zKCLL1Ee;2kQ3T!uBH2DgjpfJKzQJcfz`#
z8;5{^;HhjM%lSgz@Ho?2Ml-krzW^YS81vcL*-7KYKYz&S4Jp7SjCl?Jo46LVaaxx}
zE<K@Jc)UR-oJ)eXj+^&UJPwGnmv`R47+R`;gP8ti2(!e0Fc(TYP=JLg?5A%k@2jXm
z_$r|iwPKRc6Hr#cE2|sS1xs3jURy&uEBZ+_1x3BEx#sRSFZV1Bzt5p5EG!zdAE6xB
zP=?J*U2uIp5z(G*czrsm{c9G_w=i{Ok?KL$(<kImuF4v!`cwH1ny{P5v~ys-_hn0X
z6_KE9Jy9-ZYRR)Wp3oTKpKf2a<c#$cA*zv#_xEBsboss<)JCA*UX&Ow!hU0>3=0qQ
z4DhD#&`L%@-0j=)a7QvFn3~!~)_t9a<1{}%56s?QH0!WL_74^0Dv5wj)@g?iA67jS
z$U(^2fFi6&qb-RaKWxXm?0IISsnIekdo56a8*KO`G_H#zfYU~2TKVC6+O~X;Si@W+
zr~tK~4#9T$14?4IFFMX?#9R#0hXes4x`sNLlwb`$!lJ;kjOIG#=RcYoiRc@6i!H@2
z81CUO44da0iv=+^y4uFW;qlar($dRDTGNJ7-#~%sHcM6PQ{gWhuUjKwxZ+VeK0QNt
zFKbzr`)2*k(I|Z{SH*U!jdugSORb2lYbN|Bmv<l6t+ubYnTQ0tQc^4o&g#XC@fIp&
zyMP7!o<_N+puo&eAwzkCZNF+7{@5gr&;tJ}KGuQ1C8o!p(A*k=!088aR}J^O%lS`R
z9z+)Z-qDbnn2<FV!5^j2xfq?320S(Z`t*WW2%EX?m6D=ev09XLJloWCzF4iNr+2iK
z9`cPf&dR~T;QC};S<vt4V1d!JDBkI_n@Mn(dL|1`PE_f?=|y#9TxrSU(|~|CnlwuR
z&_|ogx>uM2g*S`pL=?63@CzK>8#s_tx-A(U!sG^V`PSUulU0<%fx->~T@mC!a|^8m
zKey-Lm<c*8B0oPBo{WA*qM=$Ig=)c<0#ZMeBy^rxockcC)9vFv2z=%qgh^0Xntgjk
z3%_#O@(FLuTD7AC6JH9<D`mbL!NPWJWi=^jGbE9CTG0j5xF27*kH$p{iDjVmJOpw)
zEbh1g$MeE*>oZXv52pvrKUIW`U+o69od@W4O%oLn=_6?FID*d08sBo<V6<uDGn*4U
zEou;f*HzuaDnOkWZEiVVThVz;0^M0pU-%$l=Enzg>yfjvVl-%lwCFVJ6bqAl1n-w=
zy>1vbb1BHEb`4=>IQRpVV`S%R=YN8~=UYx^!QBq%Hd@JoAko&eYLk%R`$qLBj{@G&
zsIPhUnVc;vr!aHx&EMwI&G23MR#G|X7>d(+9ft3CfK2UIXl8ove8aKPzUDT1P!j4r
zaR^L=1b1)&zLo_-QH6hH)Eh_GM0S=&csGm}`-zb&TgPU6YR=H;^}!65=&ri87TKae
zi9GGWg_U=vu2q&?#|jh8ChpENTVWkZm=_IUBw0}BwY$=#9sPIKlI5s%N$V<KY=OOJ
z5$Z|}Mw%}f@A4S?4iE?4@7=F(f~7qE$M(NP3(lA2S7ts6#$y9kINMp5du4;+QBj$M
z+2Pozt`~cdE^sn|adL8cas<xXC^emA`t#=xF275;(keKrp}GX`{8OX*%JWQ#$E}Sc
zJDutL5tHfcrRfjV>a-^wY|gN_K}u*<_Qi%%ebzmkhwUG}3GGL)!;&O|69(8{8Ou~z
zS1I$lP4yj58JX)kliyiaTy()+abQhobS`vw3Oz!(9?L$#ZMC#7vvPl^%(m@oRq%dd
z<Bi2pnDC!$GEEh|H!awnbZnyF*ub!}_pOThW)h^>tSaclgr45=Z^uwj8M`*Gv!nvD
zx$6>)!fQReKUW>gP9W&p29}1E1L`6x!(|n-|DDMJHi0k2E}oO!-M^yfRh~(@`YbK!
z0Q_}_-i}9eIu`7ZM}2IAk*3|nHtRh+%st9Ub8{Nl464mTX(D*<&0@z_o*u}t`NJWd
z9efA+9-Q1n$zvU|Q!Z+`eFj(4&;Bq$MP5ELJ8c(1cGc~?8#BePDYGW}+so%hSCm*{
zaySQ%MruJF@`nzLxE!wG3b-D;%5jjcX|`_E*9N4<dc?yd(M)CQLoaHp%c4e|+ph4G
zsMafY#{WmzTSsNJe$m2&fP#QZ3J5AKt(2sqpn!mcbg6VnO0yLNR0NciE@=trP66re
z?(Xhyy?TCkjPL$;&yeH6gZtf2t!Kqtb54-%hquqF6HquLU=E~w9CHdPV=-tkDagqi
zpz}M+)?i+$=&E?@d1gAgi0KjW4dL+?j!_maRo3}2y}yqrRXE;PW!~x&7i3ICel`B?
z`$H3?*7Nu5h+TXrj1W7ZOZZZ}=yRHE3T6@7`a?U5TLuaDc_1-jJ$3^IbYQU^NxRRg
znvDjmpwSLfm-Lgq!P%3aS|CK2DE$jE%jh4;m??1<^PoMc7*?C!_!zpLyq-{dONB5?
zqV{Ju1Y*RR-2#6uM7FBnW)jl8FTd`9eX#L|@1CXUq#BZ*C-!9TRg1HK+^=3HVe1%v
zf(V6k^$Qt2!LL+fFarH+_(HR7^5!eV6)#bD6b~I7UWXvNQia(|b|dSBEXjfCq$~|A
z2bJ2Xw<Fh2qw_6usR+MIY*no-iI*7iF?>H`>YV$@>o^LgpzgM)Nc2JnCZB__*B=P;
zdaNfOAeS2rueJL8_2nQ$v;JA}M?IA>+QI!Qt;7Zp_ih8Phl+p!I7#5vhZdMG7aZBj
z5YULU0IJRr@^wu^e--V}jJZOY+E^6Sv2E?|i9YAB-^fm6O8B~C!|Z@{q<mi~QU0D;
zpM&~ayjRVUy?Ru#9<^#sX1atLiBRWwafA4JbG>~G=hy1)hKY$erR2Pl5gWX*YPXZE
zJ*jI1^)>DR&4*fML0M+y3P$W#Z8ey#o>#;eJnZH8jC-Z<^n=3X@?OQo4AabuG^#OT
zEq$*?1ZHKor&XKuZW|19%{r1RJC0APZ{~YvW)f4k-JaBCH9xZb$w}w5%%zXe*0@g_
zDRKU$tSEWf(wD%!-qUoV`fq!DWJnmpEknyt&wm9MjmzGkA>eDQ7Q5&#V`D-AFVA9>
z0>@i@FFM_$rbZ01@sA2EMi&NilhwEa0t4?ur$XX>m=yb`p5<L)5|Xee!`I!EJ3NFX
zl0hQL5(!s@w~9tuz99WO3Op3P-@i)7M#QJ+NV`?o+oZe5XAdc$Pg-molgOSreXa0+
z(o-Z|3W8KO$ZNuM{IBHhl9LhEe+!NFZ8WNB>!i+VYrRxaju&UYUvsYthjV@2E|-AK
z!SWC<a-6I-Oo!0MV&6C?AoEoV^W)cs9vYN7?iP4)-5;sJr$CgJr|GC!m`ODh$Z-;R
zymIj+8QT)N7t=I_GAm}2^uGlLq_SR_m#-5u%<mn=Coufxa6-lAjHOZ}xcJ*qKOw@?
zUyD!W02jOgj>=}rhl}f%%C$LEK8~(r8VBA!C&LOV3;o(J7O32PjFfBZEbi^1T%q!H
z5;1O>UG)hzMBmWxNd&LC5kxkL`-1oH2j^Qdv9lK}r~(gSTz2+X<aAlbpn)IAXn+~j
zDIdj5)7aR^bxiJ?QS>LLp_sXAIWtSACf-VGSXqLbZE;>P2!CB*??J{h$uB$Gsdh)j
zLl&wkL-@<M?>cpV$9BI?XN&RYag!}vAbsj%@g8a7V<0MVe`R)KWg@&rNy`u93A#iF
zx*B8?e{)7RGqUhtJ|Q%0o(f~0Rw&7NS=TS7yw&^JKeh1(alj3pkX9G$4f#HnKa#CM
z<Fmd`Rah%&f9uSg$D?sB;FG5rb<>7>m6V-xFs0Pil;A%|?hcM<x&6V&iQ-3<22qr(
zVel)Y2|}2G0&_|je{E}P&Goym_rmD{<7WIdOj{_}KFO1IM_!@eG4UnA{lX)knr)Y7
z#?_s}hpp_g!pS>kc1=+@(m(aH=jt<tz!(0n^Coy0|9J*R>X|+mSv<fzI_A2sh2rzp
z&SvA}{0jLhWv7@%y{QC$k(#zPgy6}OCq91N-_6X~%82tXalI7F>i%22U)CReXyw64
zj;}9H<m-!_zl7}~{QVgnweXXbzSneAU+5X>iL~6V%C>55noU3IsA{t%AHsjqULVHg
zD7;~oktcGq7583dS1>s>x=8+cvB_Pl&F4Z~8vfSs+x^tA1-#F|G+m3Zbc6!`j<G0f
zSzEb1rF*)+caBwEO+(v{-_MPEikwOly13iUdF(|w^lLw7VX5fsOX=XXjafh&kIE72
zu!dy?y??on+-$?u-DV^1SDfV8zw#gI%liZAX7{q4z7_>62Gqttd57BsFRSZHe}4Gb
zxsIS!>)05z@7$3iA`H3oKh9dW!iyLbCr*K+>`zk@xJ_MC(MC|OeHJZL+;#k;YIrAz
z#KD|<z$m;0x%3h{uuVyMBENG$Qw0g?mFW7keZqU}%)a8!jCf~+*Vr)sh>^avw@+ji
zwT<b@fIr>GOq*?&Z6-1An{pbPd9vBb!2d0Sn4<MG_$7kjr;pk3(z0LuRI}WHKYH8%
zk(>H<P-&rr&TE3(&JgIv8}>b;VSwTg^OU;Mqr#t{q5kaX&~cS)NPUn$8o$snoQ5e5
zEywoYt%+w;!KStcwVO7-6MrDWOK82*blZ0Bjw4&%MRUbP(>TXuWvTc34h1<6Vmuo*
z@r>Evz#{L*VjFZhbxlpdA&!U~i{)6ZWjQDF)XsX8(5Y)mo8_7GGim=Ea*z60+Jb_D
zE2~*CmAU3-W<Tu9uE=NxrW)M1Z+#)QE37avG-z&a4%O#?N^G9qamK)deAe7XQ<Rz0
zYfpt(i&W22m$*Vf?tXU(tV8UNZJtq!3%p1&DlT|caN~7NK(f{LzRAiCrBbtn`LADv
z;xS4W4Xxt>!*atCtZ(LD)W8uu5mQ$Y?bCa;9;$I*zPAs~{zU-BRY=9P_iH=)ZS7bJ
zkiGq(m^X`CS$R$id(CS6iK5l!uh^SL;;T4z`m{Q=nYIo;^AGRI7i!B-;jfw0AuCI=
ziSpeR&&eD5R^G2JS{7zGzQ1c{$%LnY<2<yp_<Zr}o=54m^4S#YwK-A;nX2_f&>M5)
z`g_PMu7pq0QVL?-vLKg9s?uoEO{w_hDzlFox50T%2I~4se9BLcCc&dpkTv_`{hCEn
zzv#)WyU{^JpIqF|&8T&b5!|*hK9%t;zuHo(t>n`UM~uf_J+7o#m-KaR>OF3!@IL=F
zV_49tg9joIN%&tjQV0Bd2PK$KcCEcHopCbyNjJ1_&VG1Gt+9QVV9j@28I*RT2p*MO
zE8ivmDeZNv-Xr|y#f_ikUxp+e9UN$6#OU4FJ!p84SasY=Pd<mQeh>NF$HM;b*S8Iv
z;}`d*1@s;h7Cg>On%HM)n(piSlK$>((j!gkR>Az5O4H$zb(DzPMJr54;kE91*|@f`
z%Wgk^DPG}AW$6B(51H7XI+Yb&S>isV=o2$DN^c`t#=y*D&F@D~f(NUJ>wUg>@%8B)
zQv36^e(E_}_xIW}KM=hflclK$i|Qojvu#ca4PWcx<0ff4!VSCIku48j8`l0QWK<U9
zetT?{SY(>z<rozdR?5Rx(5ryA?)iX*@Gg7OucJZ#;IPX6_ro!iiNwDv%TikSGx5qV
zjeUB5n-&7MX}23K)Nihjg}@Xu3)_zSsC?eH`*nQhr*wT=XEcJ|y_RPEy*wAQQ07MJ
zZ-kN5de=@+R^iskIu+U@gAt!53cLc8CLqcze*hDaA#0p<eepi+2P3Ix*tDXRyBP`x
z&YAeO>9wqylqAmx-E|qS`A5$#ERF7l%<uk@Fl|Wr(<l|U<>QlmH}$QjFj3R9_SsLd
z?p*7_UcOqr2w5$;t&^MEw#ID4;=?Ob3deM-JsS#~7x*pL+V4dU*=Fw@_a*pvOMi}a
zA}X^^`pIbKyLEZPi8~T_j!(_D>=g5;5BnaEs5{j7*?uP@tjj;CdTl7s;)kiLG@}<8
zUaTy565{Q3MJV<j9ch)cp-H9c=%CAYM+)oRoiOe%vB<EPX<^L2SaK@KQjypJL<*P%
zcAIZ?y$TG9pR|lA!O2#s?KVdnIGmLGZjqx_7PY?Anjz<_^tmjCu}V7m7JJsxjaOT`
zbIkg4X}H)Tu6(!bcT)Dq8~V>A4W>vEdFmhBGe{ROnn>kEGdtJS8gv)&tux}xvFXH~
zji^RigFG7zqP^<oZ#`0}lVuoqi<Ly>jTzRoGE?C9qh52WC4?3~#Mv&gO!J_DvHebl
zl$*3ivRutsOPbFavp6T8cGxEkJ-q|z{Z8h1XWs^S<DM_B=T_=og)f|5iP~YfNy&en
z4Q}MC%6aUtM)!mGs<ok<R~wb#*(a0`-{1Vh4~BjZ70_TJO?G1s>vW9K8Fl^>f@}4@
zja@IMWuj{nQZJ->`6m_^Jaou??T6=nM(onp*Q#;ue#k*nFQVkmRm62s`8;97OGk_r
z`>U7JX)WP{-^z>ECyJF+3cM|gFEr9hDK7|qqYf1_d3H_Y#!0Kcew>f@^dFgK_e>;{
zlSTE=e$(-Q>v~`E12VFOiObXlQdV2mx2XQO{|uOVqeJ-p>ILuIi6~XsbC-{@wWjP}
zcrU~doC#IzktmDqbHSQ3=LSX$K9sXJvH)?lBK4oFnPgFT43?HQQ_+L|9WRX>GW-+l
z>}KM!lDD@t#;x^w*`Wv_BK_R2TMEHr=lz=Mo$F{xM`<eo`dOIKG<D{3zCnGL|8s(I
zS?_FA98CErtK~gQPJh+sADm}47_htC-@go2F#ciKv&d0bDcCZWg{;ata{OVngQLk7
zB^VqJLgif-IO!toI9w&cDJN$quJUxpo#rm!wIwHQ6Fzq(7kpxl_*(g5)6VZpcqCxg
zq*dBhO1KK&WcSZ4mE$pazrRA0eAa{$R5xX@`TU6dO=*0d>49@IcD@wz1<{Ff70Peg
zYq|-y&-_!oq$@}@T}H!pR(&GJTs{a(Ba>afY&N7#>a0Z8KKL)oRW{ttwzo>zwlZ5W
zFFqdiUf_C(CMY`c06EEIx-8hu@+H@%t7~y>?XtrWJESEd>}3BNeo}ip#yXIkE|v4M
zz;NT^N5no`<HZGgbeU9NLzjAghh;I1<Zda~;C<8Um6o_@o2Bs|hppVd{ANLe4&<cK
zSWb*%yEMg?GL}I&=of7joTxS8?G|m6eq9{++N6!pnp2&Zbas;_6})Fgw`3q=ClO1M
zHRkASArR9FQcLZ(?s(@fj3w!~fcyEEUAAeqP0BYZ@#K$dwx1VdE!0Ubc~AX$TH+q3
z>hF^!=ypTBvrLgCTTJJiRBMQHayps0jfHOVIn`k3J*Jkn+RkpQ{g76pUg3+r8JcE)
zB@_{mk%%rcMWxW5KU9{?8mJ^|YmvW}a5FdjESK+F?6BCS61E^n;-UdXZ0l#mJqH1e
z_4`e4X693dY!uRiLTK`}o@VaYg$CQZk~m~duB=LUzns+1!lKlmTrtatNskItR&xQN
z>Ax1VMYprG4VsHhXW{#J0#yDt$k&WTQ(XR9ZFTM7mMR~AVCUnm#%;P7shjthZz$av
zIThN(yHu5B^C0RdM!+FR<cU)=5zD8^fj{k?I`ev-{aTVb+E45qNB^_YALwDbmJ}DO
zu@J&X*y*nS5MQ2q`qCPmm^?^&tksU42F)JfV9~dqW^UtnwN8dqT3Obu&m`%q2-FV_
z`;HOe*fi5V%C<{!RL%+vW#Qc`SRlfCXsC2VAF(Xp=qIGktH|3%!sO!~#-c)KT*vQN
ztecRZFKhj8w_LOe%*nP&fZ3FYQ$Pi(3o@2ByK_{X@1iC0RJShFyeAi-I=VYd_LFvB
z%CGC*9y)?Jus=U^s-h9o{8JuKS8!;0Xt}_mDMYODS~FilX4ow%wgc@63kOC2=Su?#
zI~DjQGtCFh+O&xJK$auIK>p0vP2|7N?j(Nt9%aX#$Jy^G@G3Jfi$AcTYWBKU?2xxr
z&%s?~Qas8eYcbHi^=X|s_wAeQ4;&>m$4DHMHOUv#U4*-ENk7yWPFkijf_Lt+J0T=%
zE};7Q<g1|b9m;X-yXW_ZLea@^Hx?)4av$j2^HB9hRBu|a2~FzxZRryvT&j@90Eb16
z#o)jA;|d?6njWmX#q%3pAEwEM!kgtUWwT$G;LFl@hk@zhK=5ujrZia=In|GF_wk-O
zu&SRYmtk;VxU!ZR=P;yzmsQ}s+el8Pqr3X0-Ew;${14sWQrn7S(zk45kB=Jl1|o5<
zir7itRUwpShiL1tGuC8e{wd@cHU^zlMVF;#^K)bG87&{EZHDXdhx)_IuD)4U0L<Rs
zohUD*lS1pZ&pd?`#M}f*GDnr1Up9FSiN%8G<wM;sjri9RGyZ@CT;pYJloDKO&mD=f
zO}QRkio}aW$poRTWY;G}e_6(%vh0g4D(gqmI|6!4eYtbzB6TTzRYC8<eM%FRs#4}Q
z9-GqF<4JQZq-j5v$&k01k2rsvC|`TtRvY(oh*3@V9a8w}W-7RtH4B_SOjq|ao@hPU
zr;k-}=8quEeI4AdV7^*{D^1ifFvvNzi$hVa3!<E`+~Mqy{HrDzJ{*r#bK#qPM36C3
zeOPYz)JJ*>?yJZ_mk&DQr+$Rk<-fxbS1E(3;CflLi@7{b?^Uw;Lyw_YCGX5uJ8a!A
zNQ4A8kBxtMKl_P?Y@}_rwlZE<hk-hy{Q7=%pcMKRgY`zlDui0z+G|qP1)23n5AFHD
z<+3^`)#@d{Um5-exh|aYt7onf#;I}XllMP2dl<^qbefe7=j*ndG-O7+N|8!1IPq2t
zQX_D_t9!&;^!)j$thmI%*6p$zm<&YrmB{fGwwfXCrxkxU@zfktJ(~Od%i5Fpl)u{3
zQ(A1kyZ&&)Smhv&(Pb2W&a28n)G~K=M($Yk?iM~o4W=tS5kVkn11zIOE{9t(QI8%O
z?U3U8`S~r_mgKC<FxW&8gHBv4^U2uRfXS*V-!k4#GtuO{Qc*3AQ>jM`DMHorHO}K+
z@kYaofiMCFqh8wY$09DcIQH}PBA(bP%0W>*QenXwDro_(godbb?>JdL&#!eB<t7&d
z<ak91RqgzxY%Xy}Kc^Dr_HV9@Hnw+g-fFW&_}&=EBAOO)xaMq#b1&ehs9!YuGi@H9
z7m8$o4m+$uR#eyycU->+BK#iS8Fw-7r(T2_O}nvWKDkE5jf8n&DKPQAuDE|^CH~W$
zmbQ7H<9W+jv%7g#9~K{WjHRC3B%|E;1u3l!UFidXZ&T|Wxz<g$*P_n(<Lq}$B4;AV
zqhwVw=j?OmI<8QZSHay?V;x89JURS}USmGvmQ%C*fq?vB8uGy-f~Zi$j<KAE;!FMB
z+%eBrD=y!>B08g4v}#$N_2<ERZ5?9Ll1sMn{{s^=h56{wYrP<)Ntb{+&L$qd?s6Fo
z+<Dh3N>W^2)HF7Gf=?aafLWMOA#122=v=zF@%j3SMnSM>h*`Yl6M>bdS=KJelcU##
zTrX=Q@1dGV`fr$|&m@RUwN=xI?Mq9mmshJFDxS(h)>jD4JyKVeYT3Z`y<seeTU-Vu
zgYm7Ov7O_Cg(kgH>i6JoA>qc`nm+SSLX|*wLJ~9wXdcO$Fw_-$@;jGK4%Ual`MO9w
za9xnm{B`oMk;3bMAYXR&5G)1sC`1d6U*5nE3cND3iW?eRRmtt&tQo#7WIU1Nzak9}
zf!t7J`RE4jPU7iA<^{rwb$!)iIVTkF>f5~B{KY$7*Pu-m#LyWWTV;l@Yfk>G>+Sb=
zH`Y``(&~OfA^5LG-uYYHT;Rb-TVO}mL$#a5NmFb2*nA#6uXbV>H;`>gmhWXIeRkap
zs}lLZE?6%+nbf&}HFj`tP%rxsjym!B1gQf~;Y1&^J@4B){fQ%a&-81X+qQJ+xtd<}
z1-xkpIe2Ske7KoKwAV)PVXRl0U7zUheEK%%y*hTcYuf8JuSzAT8&%jptS(@I1djBI
zA5mVZa7!AqG9$vlK?wgq(HCSz>*|xX0!9IRT(sRL7{+gP8}Qd^p2u?0LXA^(KEUXp
z1BdJ`P$++)5sd~#8f>}tCdNb*%ksg21E?YXX=%|LD|0j(Df)(z;yXJu^tycAy;@gy
zcd6H9vyKeK`3$avySsbQw4WfV<{yO!*ri?J15rVg>>Eg#PRy%V)jUyBB8Ea17a#=L
zDet}deKFc)OR}csj>sNE;3oiv=IAwEgG#?gwQj6jjEK<L!St6Kf~5H+Ybv7QM7g~F
zDvwNeFtWf<c7AOdQdme|JhZ^LYuCi8#W$RGyc6=f=(>j1Kaa!;5iX_*uIj~0Z3@3K
zI@*F4uU!3slvB_z#TSP%uWET_=68yGf@--F57dfY0;zGZp9W|vfG9BJ$m2&=;e8gn
zlTlm_)`d^R#Ze?z$k{dxG@;7?s)pq^<bELohr;OR8Fx308gi0zl3%_|#s77*)IPLy
z{r4w?2-Ttvtl0t4(b{qEG+&%+fL0W=%Cy&b;x&)Dt&QAANw~uFLm9cgzrS#$VruFY
zwEYhNxVzYu_sv)fh~+H~<}wPtU8(Bq{8h7=_R69qoI0!C(v^1)wG);;`S_RSd{a<5
z^JA%#8lN&|oM>~kJik@6{x#<m*}0@@b*amzc|9=gx8)x1HsRBMuG)tezgLA)&33l4
zS)aeL??^fO-2X1x?Z_C#(x9|>3VNcj%vZ_C$e{56*&TtmMd9lKR)`|^CMG7<<)&zC
zYMKJ%L>#DQzsJN0h=|ab-AtB=sHgQ8s+nyL<$|>LGlT&k8EtPWPwSy+A4JL~0Z2eh
zM)sPUf|-?d2EI8NKR=~O6UfU&i3(2Ec!J#WFSEhBIXO9;6u$7204A-4R+>94kx?LX
zI$ZcB*>QUg6j{jt{GpwN!n1=$h4se7Lu!YK_v-VyzTaO6Yx_P-q5_ApDD?obBY2Rg
z{8Atwv8AW?I#lR>%6Pp>xcjFLyG&Z0{|CvVXVF{vTi3p9dVD&Im9PHB3L%nP9J=q@
z|9M@1o|VO$<>l+%(nm9-+{kAXWyaBZDDRQ!i=qI5W0kI)F$K%n(3pBs446Rlck|KC
znVRU_e0-r~+79S&{x}Gzg+6+CJQ{R`1`gC_Cclgw?X<*PaX1%&n~O+;sz(s8Se1$W
z+5;|jG*-2oADS%K&d$yPc1ynji*DlTr>3U%R|ybxz<>oJWYo~qR2oQG4%U%#4GZXh
zAQm~@>CVwqFE~jb)q<9j`wkm7L2c-#22$u(RK)M@;bGcROTm@yR7T;^G698CD@6d2
zB)!<z{nJt1$Nk+3w}M5@!Mu!y^^!-ga4?cJO1+FD5p;X~OLz7xuHLfoXu7a^;gAkC
z-CzIYLw3c<mw>QT3+3<4s@G`OO?q>+mOXrZ@fDM0u_i)w+)^#iM~ZFG9S#uabs0Sa
zfjwC9u=_R!#b6xQ%J;R=`m%I%W;e$jrhTP+f)h_|{AonBeR5OAz&fFk-?!E)p0cnj
zuo%Uyf?9VI$t!6`C#Nl_?P6pbcPC-{sP#_wlDD|^#t7Kq2BwCec^Uw%c)rs!*xOU;
zt_*~<Py}M-Pi`4PRwpiJ6=23liA1TOo8NdTh4#gm;7xs1D%Hnk%vdPnQSdtcI}net
z)cXS@$%>ttJ!Y9)m2PNKy4`MoBzhC7wM5X~LQS6ZCNp#DPrIsd!@!WX>?j^n((|+Z
zN)}&J?mH?h(O~GBUKaueuWJ}s5u0IaJ02}2BWjgy!XO64UpQiocXoDWr#-L&%|vLk
zZ5a%Sp0iMeQ@U@k&=FS#ijY(Swkg%Gly@C?uygUsCMPF@d8G=JT#0=FNb%9sj*7xB
zFd5HL#pY%U$=K?}z5IjP-~H5Mgr1?YwT)D8i5c-_OP?}=LLKUQNC}D1X~fV$E@5?a
zkDfdbmw3ut(<bRtp4%tT*!vKKS~j+~UjydnM^e(%^fYD_h@j(C0sfR_B``AbGL!g7
zmzc$iG>r=_ZEY|jVPRnu;|53h&NX2DF$B04no+)bG;-#s?J((?BV-y75`z8Q0zE$*
z^ms8lPzSalh<)SjEYcMG-)j93_kdwU$WHyTk+SOVCe=3ADV?%lcIGMTqZTr2lxe=b
zIOBOFt@O`u%d@;cm3FFKYt3fum}Kp^erbsY!`Je#Tq^NZ{g>-#fS;py&hRLtz%a#Z
zF5TNZCwnCAMLK8LvtBo?qUXEEp-V&fAbFZ=h78`t#l_{fp1J5l$`%jQy`a_?0Vd88
z?icl_n9xwN4VAb;UMakcMDf8UJp+RfpYbbPS)qHEbsxUx6A%c8n=c(f1)saC{{B)<
zOu{`)Wz<@01skzl$&3E<`OBOC9f=gtj1rvbl~vp)xwm-R%Wg7EIP1ImA+q+ih1%A3
z+%Ib`paIYdogYw{p&UAM1;ZBUw)MvT)WYv7NnfVR0c34s!x9Cx0$E4ZBFkH#7!J*J
z$;!v}>zP8CpxKzd%T;PCpBc`&I<650o&-9`wH8YF41RC1=)6KsZqzCSEB_HYj)e&_
zCQOz+RhDnIF%fZx+O7*zZ=)aUq|{27nwHc31)@Ue9N(8xyqfy$<fN}KdqJoL4?RMl
zEqmlGmh4=|PjefwwUrkfj(uJ+F*L$XI}-pA_+4MW1s(zhr|Z#<R0H9E@ZHTJ)2yq+
z#68bb?;3U7M0@q>l?jjA0H;2FAb`uPk9H7auwqKnBA`};9<=QPx>X%xw^&$wu<%IH
znx@S2<!QJm1M`ev%UUTG+pbV~y%M&yNk67>>CaJoRK#*5c0-b}U4u>Zze~%pM2`Hy
z_Se3q_&<1U?!O3@g`!&C@3*^4)!oQC#mrB#k~T;*r+^!?0f>g)RenyuK<0z1(QhMB
zHjW@_f<ndvN9n?D)9tqs3Ew_^xBvzJPbN0$to?;!PI&Kt*=ZAi0bgTc2$@dbb7&S-
z0gO;9)6)Z{0butx(SVt5?qGA8+*jfm1JVD5nJvtELA`r5mn%#dm2Uh{bXA`*bZ5ro
zr6(4V6!L0@&QP6N`rqb$c@=P8`v|wMTtmg9u{mzA!RJ`arf-z$jND~Xvkx#9d@o1c
z_bWJkM8UpzC#a3$bS6*^?K?PZ1i*pjhk&akDJ?x+zCSh%rZvlY1nyVmdCmVHR<&uc
zF8FZ2@>~XR!!d5i5i?4m!S<Q)WKj>aeSPxvMe!Ga?r(n0K}SbNPIO)V3s?%D|1Jke
zo+$8A{oG*5rjwA@j+_1U&!}7>wK1J+{1KdM)K_qy>dyyn|Ed7IrpsDU^4Ik+6nu1F
z;|kefi12$i8BO`AFm|kWPWpx;%!Y>Zh;>~%sxm34ov6U|W=sh7$#<^T*jgpF4?!e&
zYDx^eDYzaOK-b=IW1`xwZ5AAhmjHDJ4ZKfPRSy6Wx(3NRxG7PPacC?J3{=dRPVF^8
z34#E>6m(==gO<$437s7!cBa(MD_jIr{Lka`-J#119{n@4789KexeRY25T)Y!+l&>T
z*Fy!$wKCJ0vUV{NsAUPU-m-qdd)GDu2hvYHE%~yg%&za`pr}bT$b9`@H8VkNcn%!v
z%UKV`!#(~j<*hrjgZ17|R4QBiJVmW0md(lgG`)C2g9W3^A+d8*ImCD1jdk~A_81~h
zr0Cn-%n^VM>&&|X`0WW`-3Bqt^MS!wWAzlyKOFqNVlwl^+xjmSL698!${hcW5e!}}
z%+}8kkNaGvaDzW8CvlKrAOiVqK@dRKhlQPWLVEI0S$=sPAxxJbU-Ek@edWq|*J{&!
z<ubg;3}&|*8bot}v`u-(Gnni<-?<?PTPYTo-jOPXlg@d0^uN_Dq8=>e=Y)4%xaUZo
z>?ZPbWsgWxI7ZAuo%d~<U(rDcQAa7}4R0@^fkN68R@InCA0?DIu*M=Isk{hoGyz~_
znYIkx;U(`l!C4XS{!KvvOoh)?{bOYhLVa$E5bh%(twGI|CYGC)Lg`6t4C_+718Fbc
zk?mM}%&B*z-O|D_wASr^an1ktcCYAsm+6%5?K4*-fn(#JD)evPCP0Zoxiz8v#}FF^
z!*Aa|OY4wQA_DO-{kqX;Vdot18+(>233f;A$GKkDUxrPC29A{w+{yZ|(fbftR7(Yu
zT;gp2G3h(k(Qi+?L|uF^kRrn4Hbhf;dZC#*GM(5ShtDIcW-|Egl}Su3Eyx}0Y`<&F
zn9R-Oq`jzV#-Hs%IWD92p;MPzRPH~cbVVk-7q1M+VrJmeF2o||XW*ZPej;KFA+bRR
z<t!1f`=9*%pCIYIsK^H8)`8m!3c#oZdkC3pKpoa(rDlufuA!f`ykj%wFB0r5^b#ir
zRN<WWyl8doy{aIz@7FK4uB44U%y54>w(*k8D3u<HW^NH$`(qk^!$J6$1py%8*r7Ao
z%o)P8S=>^D?8=oYlrS48$xTiolX`}R8iF_Awc{}|B53U^4(0TBN6vrO{1P}hyB~m+
zJ6&~wcHHTTbEpX^4o}#?Mw2xl_FuY-e%bXHD%jh~^|lvq&&5Bk7^z8oJ#P{NiO|2N
z;A4arOaS<!cZLb@@iS~&^6^rGGZj=+RsTS5<;zuD)Wsxahg78VVMbR_ZsBt-=XGVq
zWrr(Y#KKmMzQ_^v9s{#bRP-}o(bI(`MlzWI6(q)j<4~kO8fX^h*`SV@y;{y=H?H&F
zp?e%?xGUbru3e^D0CMD-1sNqtqH&FScaF!Jo6}eJAh!GlZg|@eGwR|FE;%iBw%oNd
zM7kr!bx)RVL*Z0I_}0j77B8-Goh3Kr*;8*;VG~Mx`|{(PlhYO%TxXj!`*$eMcW`{e
z(zhfWYCk`2J9YG0nYpYP;ryKpwTVyze;lZ{i;)Frb5BoC5{v!q?P&;-&}Qc6LrIV<
zwiAeV`u~CTdzVCBqM;Y&A@loyk6sFWlzmJ5@JKNDGvr?H9(Zy&6l|t)Wi}e!wrIGN
zE&q@P?{f|X$<F=<n!r!PRsn1K;|6Z5|7WZD!AwKnFv`t(uB7xqIUVaJE9=jnRn8ED
z=U}{n;C-;dB)?biuTA6G!Zm9v@>GOU)a|#&$R+kCS~msnW1{hi%AJU5bg%7kLItr;
zz~G?)yXWBHm4%CbFZL6%f(e?|ThixQn1%5E``DsXweqTAXvl`14x9&0P6FV`qK#NJ
z+&BSUCNuz#dBASRusqn9%y9*7Kl6Vpu5PpUHevWcW0Chv89MmTkZfp#Vo_cT4)7cn
zGGmkbNhhI(^4R}wqWBNJ$t?ue@BT15o~ZWF9|o(&G8_%M>C63D^AJ(Igq|q<LQx2x
z`cTK%BoPpaJqyy2=i8S*c~1kmYONl3Z&|{X7uhxnnYwXRx{lc{k0hVn^Jz6!(-e5?
zR}{sFND8<`m+eqB+dzi&@8?=+Wdq_LELW&G>QV*4B3v{Y8XD-Ne=+8EYz~50%@!Oa
z6ckO+ss9EwsCX%cgineQ1lBv=qmjeT*a;Obx7r4WgZCcQEDAc3Yrk)$WR|u{u-<%&
z8H^k7x$46A;(7|S(BQGX%!+(ueOBXpRC`5Mf%Q@7BPf2RSt8)ed>9)H)F#4tJ#VBv
zsrtXYHJuAc&e75GF7ZCJF_&!_^+JnQ2s7ohwY3LVQSxgLDjRW=?lBRIeN>maANZ#v
zOy{f1orlHMgV;l*??VA|cdPj8$bcC^)=1;G_4C+yQEcQcVD!vZGAxVBi<pg8%#5ID
z{+o95N(9boR10B&DABujW2>e-Cx6nkPDqQxYks=SdXcIdO~;nC<?dRt$IK{Rg1U|F
zI$5Pv1Izi%olNby`Ttb|;yYE#tMFdzvnZu13GEHM24GECPmdB%z;EvE@|0G_T@m|^
z&;lZ_uWaKI<lGV$_MWpWtYPya5wF4$J!De9E7QaEIW8$+6@7_D)FUuddYX>ig8zoM
zCjeLM8aZkIl-)Gko^;>^KoSa%J`~5`@hxWjD8ysjOvQlki9_uHX~)py8_f4MhuK@#
z9EZCPCq3n!ir9R+b<QU-h(gFBV1*?13^rc-M!8?%`vTo#eUE{LzN}btjrD_(!>1=x
zYNstv$ZfG5Qs_@|^?O6xN(XC1PhJ-Bm0hg`gQv=4ca@MW#;-n8-p%DR<um_Fn1iRA
z9K>dl9)Nrav>C-9m&u+H`aaRe_$0PVX<PsqJ{E;W8a@q&j2pIA8Hh_PuF>*EjzSNx
z;1;Z<qG#7}>=}=QN#zTv{(0poIHJ@hE0>onZ2jxo9*q11wB``yo3d)?+iHR@^{its
zJ6a_3BsnJPl2-OoXhNzoR{QGOZ#g+GxJMNMz9?foBXwU>yjypDsw$RDg5{n8WdD2O
zht|z7L=ee$34;z;yvYrs*vf)J77dbbk&}D;CPDqCFX~1smuCrV1l~u5vwahVE81n%
zhWNiz(z)<cd*ZYmIja?1FGGDP{(Fzp!$AY<?wko;Ir3?h!l>2W4FkXu<RXK-cZsrk
z<O1Rw7Z0l*)YR{Irq^w(NPa|`QXP}UXj^<ectU^KTj|y`bf>MJo82WpG;iB);^WBk
zMDxDWpq%D$PlyDKfk@-FibtbT;pu+?ix@WrG!A%PnR3$Wp&TeQV=+?NFj^jJ9XS<z
zBUsc=m@roLJ_{*y@{9kZkXi&K;ER=oqME(cc=9mURPETc;|Ol5%IBA@`VI#d&5GT9
z{3(elkI1y&_ZISe9(Vl=XyZhz-)J1$TV=I4>VlTw_=U<w@^L`>!QVsq5!+si`&-YV
zE!x0P)6$OE$D?nSOnz`k^ttj$)weN6Co#u0Z$sYQtCfloPvMh#=fIF<J3<3VmYV?Z
zksk9yYzMv(5KwIVQZ)Fe)60xCcI_6e)3wc})1rf-esWO*fU8au&}@s8q0QF%PmXOE
zV{~fxP0^4V8QlDzmBdf&#4Edg{kr}zI5eiep8?>1FfqX}*q54ztU0ZBg!&|3v;3>n
zLuG1$H9Jji>5(#}J0+v`lu$F68$bw4KaE)jeRyQGz(E{@ziBipQ~L0w?Tv1-$~N?T
z0@p>=#^ut|2>ttew%&3?XGiDI!SZ+PQzeG-@A-hpzkB>>=lW%@oK@(_hF;6k2M`EB
zUKmn$tfl5){X>30C(w%~$yN|?SY;)gdrcUM3oUs87+ERz1fL+zPBK)_vOalK>&4#`
zq|tiFC9BPDWY&6|9~l^)*fO$Pu(7-PdS6B8f1~BQ;DcKRDM$TPk~`hd1R+odNYZe6
z=zx11e7PimFvjd>htr6}3z4KNa|)?|hp4G;ZH(vidR5-_s=6!W7X7*CWr1>itN(_9
zh;e^ke)TOM=v20VVB-oE)yP09WZ;<$*8`;cyM5AWR4M7XG+kKBsI)J#SnXM1mS4(Y
zV{K_N)%edWTkw;%PCz_t<VlME^5shyJsLdBOCllFRwr1>Ga-Q%>~m33k!k?!nw~$&
z9+f#(cX8JCctoFPQnw*w6AP1H0z=d70*GqFL{?8FePrAN2Rp{QPaZJlqHhrF6)G6~
zcjj&_MbxoToXIA10x%I2e_1a9HcOePNAv=Oj$W?UzZ#+?H;KBWb+)DJa(4o}?M}XD
zXf57L`c)jytNEK~C(#ya@Avm@>BqiiDmOj*`~C>bn`^F<j~C?OzXk<8R!|7Ye*{<j
z;7Vj36c+Hd@HfiCL}NaG$-aH{oyGK<+*fbKRFi5uEUAPeUm6Pho(&_LQRSs-s~H{l
zDL55W<9qnOh1#u;zg`x+xxJnH_UWw;*`EQbCEnw8N?sxDZeh`pPryd{tm?cC5mV=a
z*Xt}v-nmTHPbkR92Hv2L&Y1S4b6*cafqwH4rzC>8dQ8jX?Szg2_0x9Df6u4zY&4>q
zn-L*c9eb=Gb~$X3+2}^p=GG=--Y{CpKF#~wbS#zXOIA0GzS>)2VPR<&TVqAZ;yeH%
zLdvOZaIyVLE6HfJo!=q&S?u0?uP1unO2rj2U=y(xqiJMLtgP1mv1jN*a>RP%+Ytir
zlP*o!i&s~+S3!xYwqWjtq7QWSK4rTN)$baK-PCiv|7^7(6bD2LGY#sdRGvzGOto#9
z<B8FIZTjSQxnj|Jr;6q7hQ$P{XEs54p%Ih*-7wX1#EB<L(7wL2^IKnr1`$XV00Dln
z)%o%G`FF@A<{Gq$WffZPzJc7`!~IK`3+FQz^?&7swpBHd-7F1{5thHdk;SAM`UMJl
zL?J@HKK{<N^sJBBQh7|)T+c6^2;e|c^Scv<n>1!Z>eh`NhJmjb-Lf@<E*MVvhCPu_
z9?&D@iGN@Hw{J%FP5<J14fpK(T<Pe4PmJ!7I{C&le<ZW#+>5$5vc2@s0@)IDKNh}x
zhmtq*<elLf-t11ND21S1uK44}x}XaS>HUb`zbLm|EFSOP^(X`!_hhYiE(BT>4RiDJ
zK#N3$1}sodAoO0QrH7d79)y}@$k0LdVk@xASzdROdT)_=WeOv@;JKcAkQo10tod#9
zjn#pGV^X=ggkYC=mP{RWz9-a&bK<;ZIbVpD_nFPfpesv5=w=nVrrAwt!uhS2h*u8&
z&UenQg4z)`Gp_$V<iGK=U}!2aRF6keG5PsSA60&m-XOfL{qnYl$^(QD!+G>cyE8CP
zdaM0jJ^ZYae0eT$-|fefb3saG=j}x(DYMK+)ciNDEm;d7h=I@&TGlsF3fhq1`Xce#
z^B3f@n1{zITuV`4I{e!%BN-R)7h2i~`7B?JYJc9>uc>Ikoi4+0T|EDN<Ad*_x4PmR
zGGUOQMG6SsesNsqf*e+rAEDdSNrlnD{rnpKU<sLycWonQBTVWa@{JE3|Jz^LcLN)L
zd%uxkSnEMA8y>6aFmw}JRyd`*fm1j)!h0wkj<ut^Uf8}h)_ktJrgFUTru*0X9MT$V
z2mz@DY(Ls9_n~uwFlvF>pu2hoG$bpjBNOJ3^BLUx+nD{YFZ-iZ@j=nOtqh#v>EKJ6
z{8t`HncW_23>z>pKaVLtCU}#xGA`<tWZWejrcc7;>_IH<bs3A#y@-tc<c|9RzdTLd
zCroj12e&&@kM+|&;(f)xov>1_!klN@o+s&sd{EBD8e*U?Z$7r8qV}4pK(dwtxn&}9
zRYXE-;?q)Jh8m>xr4$r4UudEiC<t3SJG!JF^{3;kvVdiE;nM6617^<KsxNob(&hU}
zaWCR+y$wS{LwncoQ~M}4q&*@j_78T6%fX6`S81}zpnXw};&lFIA>VG}u>aERr)`#O
zg?6aqJq!*Xj@WPbC{#gZzJ<&*iieInuL~a@M7n;Sn3#xfh6?1K?R_*$S%dT}3VYK;
zpS6WVYTC?R`bWp<wG$Q}N>i@6r+KFA=Km_$sb=ZjPo?bLeSc<RH!J?36y7F9AyPKE
zBI>}MZZTk3GzfXxde^^m(@TX|J@*rD=8<AVAM1|oC&S?2VEXjlGz1cDxoh$~$eD{;
zhVm$YsFSsjg6MomTm|)-yiivUhon_39Hexs>G~Y?j3aRPef)9<&5kx|341{X3yfQ;
z&3qg1%tvl*Z2<`p9Xv-SN+5sWz%cx{#}U~1d7zpq{^SV@6Vp@3Rrak{p1ai_tA$gS
zL3cjV`SB`k570(qUAT|}A3*_4UqD@^x-xKb;sS)@a)iP=Ux5cIDK_>dJY%`yc&!_H
zx7(QOvMK%t4&e8ecW}X4Prc*}8EGH(h_-BFpR9FJdO8LMrxGeOY=d@=j!8S`U?+Y_
zwZ`kPYznGTPkYyiX{F!AAoC2?Ky<VE_1ar1g`-WS!y`3f`Q#D`Ta)<MrMTz_9GFqG
z=eG#f27h9f@-fG23-TBtuEq5mTpB%VOY-V_b*JiE!^HWkBg3^XqSc;1|B8K?JQjM!
zjX(-)2`I%rHIzdcrK!EWAM(|5rDtZsrg*rxK`}8|AaHLqfMmZ+9yKT&rk4y03tJi~
z&IK+bdD~%dr1C*p*cJMJmc}b9A?TAnu{{f$V~@0lrXLCcTcb|RfMvxKC?Ir!Gad(u
z!@*9^B(3?WDDGSTsQ0HqhIXN!wfpxcA|9)$$Mke`Dask@;)TMSzlrMnL>)F_Mb1xb
zT+a@Xd5U`mU!viIuFc=W@W&~GY$r#_%DhSe8UqJMdu=ZoXMB*yC-u88A@zHeQA63{
z*Rbm;tBJK5<*v()K(B>b<;eC-lHqbz`<%&X%WM<Ba@l$2Glt&v0o70uhs5xGbDf{e
zqNrJnwN@=XQIS-zO@41tZl69r=@j-%1QDN(lJgfiN&!el9FR5zk#8g=Coe;AE&ufC
z$5*dkHw&GCKmj*&O8Q>ECy{T`D`3tw=ej?dd_(YU3c$X@@7pAWL(M+9I~5e27J-Pa
zA4SEnc5HWky1zW>P16T?trrOr__MRK#=z@e0!f&i5u5f!_+QEn4v0f)gY)rfA&=XM
z18U=jb4vpI!E^X*DAD9Wo4*WnW2ioP5B++^OFgM2D>=<d8R|S?tp@sfdPWfNmH?`N
z+oXpHpMrZKPdP_oYV#eRkdP*nl>;;pEGEr3=Okr(-iV9^Yg^mA<Ysi;o3h@D(u{4L
z0seD;UfLs1Noi80i+cE<<K{mN!|JN0BKaxPy;r=P=ej0fS5Y2%RB7qc5|UFMx%$0Q
zJ2&Q3?7p#DfTTjRZmRxArPkZC_kVm4GByt<s9Wvq@n%e!tZU^l8u7r*ysFG)PyKiE
zT3qtpe;SbXL}#Xt|E#)~M3YUOS@Kw~x%pc!M-@Ue8!QDqWHamQSy06^-kNE|#T6*e
z7y{lhI!WLa9k{O%A2I`7dHpkufj+KHP{Gv(@rcm6SDnzjNJ6^Temo}4QLY3fI29k#
z*>T%7s3M{EozK}SJhODYLIad_yem%kG}v@1H2XDeWK!#;q@`KZH%m==?;!ba#w!+h
zed^+qWFkzU$-l168K}B^w8$tZ($F!n7l5Zr5w7grM#dN;Nb2a9I|sWrIB&d#SYyH3
zSe-qxWU5f3&?1^0=iNho+?DXdfd=)RYP0(;^rJ82!ri{@sN*eGV0TS089#MhtTV`5
zrZHynJThr5^x4@RT@OZzy7r%)h3|MLi3nAbtZB<XI!a5}c}Tf<9-kPo^whpDF#5~4
zidDMu5lxdTv8Tv-$B&0uELQd|;=&hDv-{3kN?L?na(1cw_{~`yj~uukkX+~mxWR2n
z`n_b=`6zoj$^8)<x#D^#TG@VLQzB!x-KC`FH5)LVw1}HSS|egY_JezE+J#l7c46y%
zrCS{<FYS5V8nZrhvKDTdd7o{>?g{8c(gmM<r}S*V9D}Q$L`x)Jeml$sW=cHtP5G{}
zTqtJM%{jlYq>Rd}V3R{hngT1VviEZVgat=W`FcVa3WCEgEKY=L9Zi0cv=)e*k2GxN
zNatD4xQ+zG`0tDxbSZLW-j{wA(8D7Z-l4ptrL}m8`N(bI34~3o=QprWBX!W!{c3t1
zbTVRF*wEB&<$Lvt?b*H_^N3IeXA2uK4NVrnv(v#2nRNY-EUveJqxPx)%}}0XfbF1u
zXw>EiOIBHqrj3*WNVQ3oiQK*|77!E_rG+g0nwpxTljIw7cpbJpP-2&k!rj28(1wH}
z+&Az`oF%akzMtb_!2|n*i9--nSy@SE3}O?ojcjC;YPvRoRD*bT<KxGV&%v1x)*?W|
z9LP5cRCdjOfV&wyW<k$`%R2gr=ky1khLrr-(B?dL#;L`o8e#jhXGr&oyiSq(@O*@n
z@NH>6f0zw3qQ$O@iBwWOSsaK)$u?^2X!GK`Lz$;)U4E;!rc}-kqwmU5eT_k4-NmHF
z$@&!43(QheQkYI5BPZZ?5<~JEAP2#8!rLwPW~qlw@<SY4t*vu@AQ=4tJwMgXf;$at
z8hr|-)MZ@QJ&{oi+0h3X+}du3`uP^4>OirR3HM(L@KeV7j^Oekf_6mw37Za8l*Gh8
z;O5j02nq;@XOIbBXy&LCGM_Jnp&LPzo1CwA0Cl(KGG<D&X&ZsaBb$|H{h|`QA@NNI
z0#}w;gy)ZzXQwtnze8l(z}KhY{O1W(gz4@&LT`9h{o;1ukK~A_lceThx#~~&Xqbi%
zQG<Sm+nr?dMzU+tx3Y*ww#KjSPPTr!xmk=g*4MSWVb7eFQj_T<f!*>e-TkPxc#`s2
zwPL5G0ea@kZxt1#2V72mbmhGirhytbL#o;C%1E&y&=!Wk+PNJ};@l8*(}V^uYlNT5
zS|BQUmC!(mn$Uq>h0wRA4^L83lH#u3A7k`4V3~G0pU{w;9(T)$zNmCNy%SH27CaJH
zBrH%d{jrDj>5_I6N&b1Ckm}Oa)SpDxheY@?e?Qu?F&o;b#qBp48D0*46a669)>%uk
zYiPZ_LY*8nFw3*NwE@NIxNibE$RS~3{TsG@evLWbKsv21=h$Q^mfRZBBq(1N&>2I#
zCg8M_;<U362P~=1E)~EfC>0RGq2pHn8Y;T;n~u#1u-E}VV%&~Wd*)fy^PUC(w;cbd
zWi5W1e1eyYtE&&jdJRd7gz_!)EINSIDB%!N_W)&=WBqGve2O0kw7@UC$DvM1xr?S`
zEqJcEUeZzMWg2YexSKIa@us@|+MBWceqNH9Iccq$M9%T8_U`m>J%RiF+atmVI%+T^
z?ftKMtDK`lx)(df$3+>E6IN&nMT^`D{LId@coHbm=7a5(9t0<X;#{H+d&yHA0;&$c
z>+IB#={yaHMqIWUal^6zh!^k!s>u(+!s_-T2TSd<popn{%O<p!L`CzK<Plt!c~IV0
zh*ETjVI`IRR9JbuRvOLR=5vR2BKQZWNi8pDRVr(1M}r@q4CPSc&8fN}h(>a9b9rp$
zp78+P-fd~L=Z*Acf-@x@W36+$jO}s*Z#Pzu?K~TsXiFdB8!0Wnxxv@Ae8Y+w)w>2w
zn`X;i<Bn&W8uO;RJIPIXOpLP|5^P~J>OUOTr`MZa?OITF*6`v3exeufICQ5(7EbEr
zPP^O?U<{h^?yd|fLfSvLdoieX^0}sFB#J8X)^^Af2ZthU$5|3d1v9gs25m^BTApOY
zs6{`Z#dem`v!eTI;-G*Tpy|H|qnfqkr;`L~ypG&<5e>6Pg0$AJuWKxVpTJ`_@Eo6#
zSBRz?o%3r8UBASaG-d1DeFeO`(&z^I)N(A2(Z2)xRqSo3CDUF8?-{bta>+dGZ9mUU
z`#ZR)lM~F{k)lDTBP7}<Uw@2$_6Ln(zaV21#g(^nfswMS{t=1#f!<7_9Eu)ZLg~^+
zmjw{<m};JWO<Kt#sJqii+K<|WLWAucvkGWYR=*)&3pM6tVj?1&A#?DW`=MNc@@NhZ
zlor%G6&%kF=j{NFhVpJaa0uBqzsEiDCe>Q&%ZNan_Eu8qr9)C+8teR`IA6Mhcdh(#
zH=2<8G>z)M^<$KWxiXlWZ`64kKIKnfEoO@Q=kT)9;k7E}#>t?@<QWk{@ooN*2#Eka
z`)Cmc)c9f*QYc3?v(759(v)+f_s&i;48DBs^wQB*QLpjq_G@l$?KsQ7w>vx_qK+I?
zRyVab_sHy5#tPCnj!Fl|u=DdD<ZWyLjcIDfp_rIh9xx>N-VjuVafJ?=DZ`Q6g~-<k
zL@=O$6R^|HU^q52J)Hnd&q|<h)(HrNbf?{=Ml;@YxG9|`-s9vs?V5m6So(rPVB-&<
zuu>#qc=jAZg<1Gp2uz#FpRc!uimVc-ofd9!j8V34{J|ZA%WSz%<4rH5(ms#&6<NIo
zm(yOnISze7shX$8*6jEtbLX51@?85mmO}KSYP*_tet$Y<uB;OqVkz6z?CT2KtOyo1
zYBC(er%9u@_ab<Xwoa4t4hr88@7LmeKAyx<!w@66f~ad?6<=?Xgmc@SkC7tlVL)-$
z`sXWW?fKazIX-~3DA1JZR=eLDD}Y3Xg74(~d@`Uw#ej9~@vmWAJ~h*h*oUd}BK>9H
zF&9S~s;<}-94juw`>=O`yVmX)Nq2a3v;<Rm|MhdXB!MqAt0ZLrlu`sVnDk|PdKFq^
z4%ov~p@fxCs)(NAzLaOaP?DWF(}$b*cI!l7l_Eq)ZgO|IYcztfk>N&@yy~b4?w5;i
z|DDkXo)`UpCN^jvO0CO{{-9jHk?18e+&J?`<>*&blcd~pRXrKIb$g#^hiO4qepl&4
zJbRWv;KefruNkiOx*@(;YfHow>y50eA2tD%>drcVDuj;~5==VdJ%MF86~X}~+S>d}
zoot08@xW)w#Q*)lg(}hT{ngk+BIMf~y_847k9&El{K>_c*h{U)c^sQ_URvSJ6f+#d
zLrN-7HO92dwl%l8m%Fv&#W7K9p=5$jQP>w%JJtbTU%MBTWRbUCa2EA>V?xbSFVk{y
z$piIa|JPfOdSV}9!qna2;{&gjtqb_m*d<YsQj0-uD2uq{h<ddkkZkMg>q}a%7LkGX
z16UpD{W0e(07fJLmsJWl(*bl5t;$J>vuNy>?S>@MwHwG6AIQo-)i7A@PV7y^P1lfV
z<v`Y-Pu+q*Uts)gn9@X01(PAN(c@yQUiM6cp>Axj)a5nmz0t~Q^zJfqG6h){ULf#(
zlvQz*251$%U#fa{!-%h3iA~Qekpg<clB1>c-GSHsj-}5c-)o?+19Z-Z9fU{caq$Gp
zY4o2y{bO7Xl9t~m+icO&ZU;(+iUO8ldF#Vz*Vsx<RE%EWnC<TF;`RalU5)k=kCNeY
zt;BcPH&woB_Hh<hF_#xeH|3;Hw+btjTdn3q5u)Df0X+m=Y~};m4}l_F1^5I|X9^CR
z@kG`8BXEa!AnT^QuKdaZs|@xpTgeC@u&((7(ub|245-oUlMu}3;yeo@O6An0+D$f3
z+sy)bgLCB8&de4?HGa&v<kQ%u4rwFcs}Wwr>x$i{32JPhD=$<zI{dB_F0ynTs|#;9
z(RVSeH$M7E)ZS)s>F+XE-Gul#nNalPIRGGnMlDnzt8CC3o(M(N=EpuYQmq;9_5tuH
z9*T*HU`h+sx3|lQh-4ZA>)IM>X0oaMHJ44CzlLxK=gS=F#@MoE&3JTgB{t$DPDUht
z5Xvu#PsnuL%~aU^?YlUpjyYD=zgxoX7PB+LYy8BZ@!^~3M@PFwI};Mj>@qyzj~Dy=
zy_>I1%#TI=^`p@?t`vs*@kgE0-8PioLRt-SsR+CU;qXntl56dDt)}&sMbDM<XgO9=
z1-jj5;JFk)fh9Frq2Q$@@SWDYIVe&#ETdwHYfE6);eet)ZA8I@FFU<vxxd3-&NZng
z9~>R-)R35-JC8qSKjrmi(H#4}hU-Y5U%{)~U*Qyb;*NP|=@x43(9oFfl^{}>DM|vP
zLyBxi_FX!GXB{Cr$}D#KmyGTEj2kYzAbip67S*#FmrktAm;*;?VdQb`Tt!;ZdZ^Lq
zPZEKW!w0`eYD3UOc^3QCLkY?i?1uU7?7L$8eR>Mhg)+1|QB@yxcU_kxc3rK%%C3ke
z_0%#3MMd5^&%Ws?E}kZVbPMRK|8iz9@X30$zLYRfdc-vCkaq1YwX&1gXVx=i&;HaT
zfsBONWy8%X;s2gXPL1aPaEi@%ZBQx&s;;zYx<!E3>?%p3eQDEE$d|}}ks-f|8*}G}
zLD9)=j2B;ZZZ`IcM&1zn>n}afpa1o4ANJ<{AFB&xp&JV5mk>j3DZIb$4C%8TINK55
zzb~W;{P!nl`ZLN^l%%9{|L0F6yYR^zOEIeWjXMALGuYx)|9is!`+p*TpMpKRyo!M~
z;XY-1hvmLv*r!wo)F6&eiZzM5d-%Of>-yiH;=GBWjmA%S&A!33HAoy4%Oujsh^!V&
zP{f%Gm6oIINc_-a=Tx!3Kdt*1*5LVjGqvKQ>D^RyQJdmOg5>3<1+>(pm}Tn#nXu#T
ze2XDOAtRzdDI@DsZ*RgIl?M*l`=9?_U1(kpVN@h^tA8%J^4XU8tqa}8I+o(=|8~^-
z^)}nN0H=<+gV@pD$AXg_^-aqC<pMeHbf;%Q_PBquW@uj345nz$(YNkebI(@nYfV}l
zKVRG?=*qDB>|Gi_Z1Bw@h0V3|VF|Yw!t8x*x3zpaC^*RN4DY|cEQb2%iTU|Tg|4yx
zkG8iAiz@2EhA|K&RRkmigOCuEZd9bZh6d^G&I2MKE!`#Eox_mQ9mCKa!q6~ud<T7=
z=ly=Y*Z1e+Vy=rgGiRT*SKRAfd#`;z&#sm=JL?%F5Kr=cLRM^o1gwDm$~e%F|2l8Q
z9^zj1_hFk$A3EP@Zo8#I#^=XH;YA~MQk2>X8#3%-=Q4VEo?gc#GsVeR91xY{f7hS>
zz5Y4@IPyofYN9k?PQ5*YBK1v;lqT&tO>U*ke=n1TKi<3Rg#kivvy+kj;-A%4r~|w)
zHu=EHKe99A1~f*_#UKA&nDzHU5aZ^4n{h{Um_j(TbfqE~F=Iy;#}-j;d=QGi*X649
z>`&^5M)VshI68t{^IM<>Bs;BQWlcC&r+)WneT^M$@q-XC`?ty2*EMTdPc=!<vt55j
zevWFoK!+bkQf9r7T5}u6B9lfSQai6(A*Z}QHA<fFQZSnP3J`%@p@BT{c|pb|mjS+_
z7ST&FT&UogbPATzWZm3N>bct2ib?&NY5Mw1d61F)8QGuc4bX*n)g*z0S5#sma+|rQ
zD9uZo|GIy%w0AL=SK#}iCN5MHH~Q-AXRO_UL(8T&$2E;I+PcB|9v@PO_xSCKJ*T2<
zsK8~~W*N~lXXt&!*Oy6Nqup)+SS`1ABuSexXOEYRUdQuxgeH3t<t{yR?xlVIUKTtu
zV2&|k=57RV52pD!YcWcJYm%#6rRad|@SV-8iq&bwpS!N3mR08<mWy=>{=v<rK_u<G
z(63cqWZdV4)@U?X)6l>x7Ll5o8cXlZF}Z#GYyI0KLKzYS?<HtS#N0wJU!7c`c-)LR
zhovEHDLqW5<thAzhDN=SFo>a%&?@#DrhW0r&EzC7DOBL@<KsgQ-~H+7>3Oi9E=VIA
zz5fCijU0@176{)W;s_`}IiIjW!jfH`c<-+t@yZYMB0}SRTj}`Sw8r!$)ruGoORvDt
zFt3S7_UDY(N(C-_g9H!%PI&$^&}R2aaXB3AG*36J&oIt5f0}_O+loi$0z}{yp#A4+
z{MALq`YN!+H6#*}BQ8Kee}NVxBMMN{zT_3v+}&M2yZg_9Mc-N_xncGBreE1fZ|Bw<
zsR^cj35Kg~`iQJDH_Aw{dvo>5p0Dl@KxqEn{yjf<`&={}ZVtGE)y(pg-c>SR@eq5r
z;~M_RK~}&;7BI|{#p{0_QZLqjUMV00|L12j4p5GF_CkqHgk7NbOWd;d<*Z`8=!&n>
zBnUKEBrAkH^ONcVM!r$zDa0Y0NIZ*uq$e_QHbxQ@X$N0{y$bKmHOdqfH*`VU|HfoA
z!Iul@UYO5DVO8W;hYs&<PMwO*9K+2E<g=74^V&lXQ8l^C$+3CsrDg<*d`Fi=n19EP
z9RLcMPUWM@Pe$Ne7)8$Y@j`ifXt*CE%7#};+F$S$TbojWfd|{plzihgIHmu2mmm19
z0sIQM@`qV<`fN8dtJr_fl>d-j^<L~9wL1f`dun&?170-^wH)FREB{pQ{Hz;cer&6>
zN3Gr{g?@$}K`9_05bwgTF$9)2DC<(NYOVh%SxNgpCDVMaW#T?IO?5j|zZOibze7et
zd;b;`_`k3J`;WJd|Nnl)|GyUi|KEoBe;I_f)vW@fC2h9D<mMRDzo0up&0SnDA7cL$
zqqd5@$JEo^A3eJGlM?;G1J^ndXLU#EP=;IUYk(DxBsf0+3s#&gkQzDlJNNI#%BtUC
z0M1;YXfonbQgAR2tprOTAnPCnLR+u}mjami!NCnqRp1%b(^UWF6XXdL<Gnce_%y-2
zB2HyUaA#K4*Ot5QR7yXB>$r!(eRTTli7?s!;2kt)Nc1CcnQ|nU^ksv6rs=7vg~2T?
z2mAZs{f47pm5J36*ir#|uZe>@17SP0%eYpt|BdRF1<ZWTf}cMhf&B_|L53h7Qj`ED
z17HiR%gf8LB6SYfUMp5aCJ6hxEo<EB)NZ8$epzdScO1+T^v8{$1WOd)u}e(K|Lg$`
zZ3p`GuGX!SpBx+<42u4HJ*W|0MvC#|4S(CUi^+q81_<!2$Mr1O6Pa09BpUiR><O=K
z#)c-0lWbj~-^JS&PuI<QmTh(HQVv#~0(T>^G>4IVgFaYOdD<R1k?U=~#d{7di`G5s
z-&Co+?G_#GL{QF2D7vh>-t6+O_J1+x#gZ<<`3SfFbiaW;Q3?QgfN07P@D$&q&fBx~
zWcbfkRKvhsb@CvEDYAL^plOQ?j8q#d)8;)jE>KKG!s-TI&@YI;_D4u0=`+2RmHy;i
z)BBO?XDk>>p2`O7+1-t}t#<v5aNd|v%6=y|Mr!ZZ-hn6OZVBC2v{CZh_HJ7ox~*b?
zd&U4wklDCR0Ba=vGs9?Mxg+$>$<jF;p)U%+ePdt|SqhJL?_=Ho*iRhf-jWBL-9UN=
zoC{E~a>V&|-A^?ZO+ZX7$<NR10)2e%ZRvEC?KJMY$p+PZF`B59$c%)=#lV34qH7`t
z`Q_ZHImWUXGHD~ZRKGxHen4M;-T1m56*vtOc|=PK7S!@z5T`qakmSt^y&!l(wzWvS
zh}+q{{w*v(*$))82!j-L3LX(>4rNYT%dQ;fw$p<Yyam{A-$jkxn{7w8;azOQiaL21
zJGi&D>-X;akI@i=ds?6-fVrX!*l=0_WcH876l9{$(ThOQp93JfGr(yHa#J9?1D+GJ
z)*UfO4{d6yk;r9DvZbTp5g`l`YBa2<>CLOOwP3&6UDrk(k8OxcO3?4e<?tsLGQLYd
zqTF^NySUqfcD_6RqUiZ)6<jxbQgG1!jTp_*RMhTK`;1Eu)7!HQIYSDH*^z=A;dzgo
zbit7jAcXF}Wz~ax%3b|LT=unw(*Jn964n2OR>#SMgO&AN8hp6{^mjBIHMu!i0!_-=
zd*ybvD@Y`LkX5vY8;;nTUBk`bzyJh<Ih0_;FO|Q-HK|ggS6aYsKFnW{No%!s0*S)g
zvL{$yYY!q`@B@<l!;FXiC|6w#>BLi_LW@fZbV>5XzSMp$B7D(1FkJtA`P^omylARX
zP?`6zVP_lg<tMrRaf#h9GK+8cPQBP|o;as)M#ER7sJ=`o*i{6fd^Pkqp0eG@&}6@*
z?c4>S<pKQ~ZpfgT{pgL~#@w~CGt~aC<$Xm{VqD+@*k-mT$}(<uk%bkt-edfNble>#
zkP*lkD`Y?g7Uz_CxaLcRn6`&P%UtkbBNYYNR`XdoL=5nCpl>uKLj+_(OS<M=b&8!6
zIo(2qB~O8f%*A9LV?~6D(1T~%t9BwtLh)U*(b}fvqvIYn*F6}XJtwW$4Q#pkP!q(P
ze<MO1T~#4`!+sM?YT)25TsXMx76x|1`406W4b93lU63fDW22X7<8c~d;+U7!tyS;p
z)E{~GF43fYwkB7ze^ag8rh@!>A>&urdo9tm;(*bv0E*s0^YJlnWn?lxpE$M?makfo
zYhEW+g54n+C@R;p0@{Rn<8GHi;4eXr_FN8hTUc$gx6B~ufl2sWSr-`>vS|cTGTX$&
z>az@3V6!>?eOX(2njDlW0=SLRRX#pPs7v&73seGTaiwkdK=0_iT^N)9Xx@$Hq{zL5
zj;QmQxnUG))#1t@;;74?f|lz|#uDbgl-N(%K#ld5ZIl3FY9UaIxdv5`r{rHQqwqXU
z-$Z~Xc8TA^HaTp(I=5VWt%UV(==2)a+Yl|8yzmtUoEg-kI<--kD@+bb=IdC7>*F_9
zmqAN1l*se!5C=CTp#lHRW}F|%NO2Yae67%~WMcY<lX>+!N`D=USAJSg;R4?+PdsZW
zunYi1OP%fB<>zaIz7d(7P<usVF1J6p#)KJ;Yz1bcjaO5(PN_7#1!Ii~CpnZGO|wjl
zC(j3+kYEMHYK<w%2l1D^MzemPWQcBc%SMANkyY#+MdGGwm|6sDu1lnCm`((1sSCaZ
zZ%QT?P$GX3*Fm0Ne86peSzEm&0HkGyFPf@nxYl*%5j`zrbTCEkp5C{Zs8D{97}ePS
zFLcyeD|Qn!pv^2>b5|X9JrR0vT43WGr>>`|kptE&5R1+-QIL&oZuc({&q$S~x63do
zpvg3yf9sWKryL+QY0uOLk6f};-cI{i)0yvJ8~i=$SxqPNH$7S)FJdLVlTkFno|S8I
z`(Ix55&oqaD=qG$fPk1h#M66P&tUR^O6=3Mr)8MIeeY|q8)qZ<x7BG%v6ee`EW`dy
zX+HJseeR#Cf=;tABVM(K)|zQ$HZg%wEMK80fkwRGXfvfD)lFe^hTKehSdVDqSmo9X
z0kkyJ*9+*fzsN17y9{g0K%2D;IqaOfipWgao^nph?<w(+)I-NOHQPvcu7T`NB}@0u
zI>~}w%)tUa;zRJCIsNWM^fBOphHC)j#!b5g-O2Mu@v66e<4l!d(@Bm-CK{3l{#Z}P
z4yi*&KQ|q%whd#kyI@T20=nhGsx6*@j=lx+AKF#|6VbFUHAkCy57^B9oWL&FoaVi4
zL-t!EnxYhB)BH=I!a6h@*wDwk9=%`1Mo8%)w6r`qed{^X=B_3i;$kBeHFGP&Q-6KT
zNBG|lyM`3Zxi9dGjqnNFq-%EP8F1eKHl(QjHH&=U31PO86|C;Hcwr?sd%*Cl@{y9R
zW2oiXzVa&JgoKA>V_W*gS|7Y|j8$`%3KHwIsp1{|FH&Z_SwOcI=FlD&eDgr7Bx;6N
zlii@6ket(gzz&lP7go8##K#YmF*F+XJMX(F;bHON!6A&x6do$udL(9Ni_V(+vyv}>
z?k?VSF0o0jddqW}gbKIRgt34TBAJ_YUd4aVIyl&is96<uwQxn^cvU4n&<$5o0LF?0
zuJaB2Lvx!KIk#)a8hE89%Qaun$PBB~Bi>TJ{K%s1`e}Bm@<4udj$mcCL|Y3`)l_3e
zY-V(ht&NtT=Bel_S-7$$$}Z%A(W8z)xG8EwS<QZ-RE;Be+)7VNL0|L-5#2vtKzQkv
zB~i2lV<7+YV<6CCSTcj!qnN^JXE==K<Za(zzMqZ`=6(g;2Of1d`H{fS^PnafFa6Ll
zolI+tEjTHjdh$eVZ{RuXv>kE&Cnps(&{O4Vw5kSwgbRTUeFgr8iVay;Kg#Z{<&1QM
zMpg}ZgUqWtcjn&xT~2}a{O%nlFZsF(D+-(~$GpZk9&Lf5Qwa$Ui*di@0HGRXESzDu
zrmllq>Rf$;0!F+fH4{YL1#E-p>h=c|S(`(`5pp--3Nb%be&e;oa$_<%RA{NG=_W=r
z3pfC5K#B5T(0>!KSD+!7e!9NCmGteaoQDhy^a`$!YEz-_^sMiH%APR%G8-Q&5ZoCD
zUl1+N#!zaS-bHIVI%fGho)D|pe=@@cGu4lEaUSl9Jg+}jy?NcB8Mt_=%3!Ke5L46p
zS259)(C#pyC1FSdISRKCZxFaDVuLzv7~2MdNb1+?H}F^4WyP?}AV$l{&-{ho7^(Sq
z)G7MT#aLn)N5k<!N-vQ{yFxYnlXsxSxMUlKR>|D<v<tKkWToF}HeJHR50}T0&HKyZ
z{auRs#y?2L^RqmYc{cJHlW#HAHuGOovjj6lzzeh5{Sd;(Vp0yVoYvZ^4sko#2xBut
zjgM`ZQjDtT?U4Qg-|3Rx)6)q7t*^r)*NTWC;%eoAr|-eRVRUq2!6Z;Lw$%j5_Y5I9
z-|7zE98HJC<`-$mGqa%9oJNm^Wd!J%1ezwcd)cjqIRHl4pG{5VP5Te{Y(_)Ow}8JA
z`%B>w{02jCl~og>S6eEavMQG_Z(5y0dH2t`>0=Vm-M@)fRkPhIZJe$bM4oKH;e$=N
z;Xal_iFh;g+7)00wK)ivjHfurKTJ_xt)4PZJHvQ(;#YufhG~JLY+CujVHFdbCa^VK
zQo><@NdNr>1stPat#<(4uE>^)29S!n_4)Tin(}=*@mF_|1+TW%Hm0aU;Q6kZtyfo3
zNs$Hp0W_>`2ra4kYK0Q1Gp$h~rcpem8|D5@ILh%~{hk^0q+XZ@6TidQY>vs$)!b29
z^S$=U(=p691?YnPX&)U5l?31D@|boUYm)FgrO&0JfDBNBkcmdA^)ES!c@mNe$MG&x
zL7>SN6g-rptX&+09OK`P>p}zIcsO!{4;<L?&4^$y+mJb}I^o@cxU8r?NsoimhfGSl
znl@0!f|JNWI(iOe4%VPyx7ZWKWT5KnkE0Xi$RF0TLkz-9j!i0U>x^R!9vJZLGMdVx
zt1`e;t|=up|L-hB`G?hDp?T&18Px_5zbEY!-@W`8v{Nf4I8{YQxRcUDxsgpMRkc-|
zsKi96qCoU+IB{M^06<rM3YJJT{X!Ir`?D=ko5tOm90XWSH}57)hw1sCo6~ocyUm$5
z4(HcR53GQsoIv`p<-9x+)^j$4)wNAvLxLL#B`szo+__e7E0w+CJaxs{o`BHC$CI3n
z(2Mu>aLW+<jl#U%xAYGKsNXWcL3|%|WcHkO^c1{MYtt_A%XhQ25CmN_Lb=5Au;`eA
z2jU$K;b-z-6B3DxguiD{9L3Peuoh^NPvyx4VI(}qKPiIzVxETX6i}X@%eOpIR=7ol
z95v?Iv2Kn^CJ4_&De4gDB{Ry3tVmO22{lkjOrS<{Yr4%9e~F{H#19h0TwL`f9%c?D
z+Q%u`+h*fKdSx#R=-&<@6&g#tFzIRh+tO`VfS;5h+tUi+cdiY{m&2vE$MPJdR##R{
z#xe)_G<CCps+)-q9>>!Zn5YWVnN>cC7dwQ^iIV5J+E%Llh92*cIltyQZ}02X6L$Oj
zT^shMo4=?cGIN+VG<Ix+9XH+<Octv4Lf$r|H7l$tnKl)w6=@_&6%{CtvPqFdn_*K#
z=KZlrt#%H$MNuGPe+6wz$C#>-tMc-)4Uq!`SijWNSQe8dO4(g&tuFox!UVpz_cFdn
zZ@Yy_+~fJ_=jpFj$!TT=W%!T?B{fJ4#A1G<1(Uv{yQx2#Qggo2CSEx$WiwD1XJy?C
zav89K1-81TC_sqY@;UB2A7T}bIJ($mvqb#GM4SS*BJ???sdZfgstZ)xK*R(t9pH&s
zI93ssov^C4Aa*5;rR>s`i`=v3N|xRZ5!ehEp@X%qCqu(DhwT^_TQ{$;Z-pW(e9O#5
zMeS^As_~U>UFqEvxuzjAjsx<DgayVb3mL@ZQMT33Ih7XmYLKE0(|W|8<X!78J^C@r
zRc1(X_H1~7Pk4VhnfKVY%u2IeXiherCBhORGP><3BqW?KmEmWq2#pb07n774jY7w&
zI^5IA0>PYC2uYyDoTQJdePl`4{<S5@@UoqSxtv?TO<#yh%BR8esX<d%N%J<H&ee(U
z2z-beLz%TUCEmdIibVn$N=5gIUII5qp$hQ%HdbtidOFROxwcl<f&GY40Y{j?oq0m4
zzt_&Ku(h+;8v`Ow`pH@LC$7E$ZA~*Hplw&zN$Wn-%Jg%c>{bYh==U4ET0lAVYt(D(
z4xY;aR`aiNMpd`sAj3GUM+pVDFMZUzo0_XwyLIH*VAELH$}@sgA?AzITtF}Kxd*B*
z(iBo+S+~xmeM~Mjq%?S??6w?62o>n9u^f%HyNN29EabU39`1ZY5k2X?vry=XgAkdi
zNfl+RA6NwSaXdNUdCjMSu$ar?lUvws>|b4<qe-tS?YF`soteaLsH9xu>R`cVEjukk
z3583HgFsWpG@mlh<qB|3p(~Lu(5~>qB@-xq(cNvjE1au&+5nSHvAcaZSptybf8e-q
zYBPI$uC7+);ZK?8X#Mm23qsjv=<(Tk+_Tj9tStXfW2Cna3D>@!$7!2P0`sOxJG&Jm
z+eJDvrvUlm#&}mwnQ7u>kdO!I_^mJWJymt!rB}Ck)L8}4@qaRN+>u9OT+5tjPRdY(
zk?rhXY|)-HvPjSO-Bi4tuu!h3XDn7w2NSL+b=kD5yp8k2=2ukfE1N-ehezAreAP~O
z)az@74l6>Zv*iTg53xqIMMCMgXq3C_>gw%|chvv}xd=rtZG&YX3qBc_N-}R)3uniW
zasLYGj0BGU%=n&L_q%%7T}7J`5~mB`kwrDZ^NWkrAX`3iVlz_(Icr5Bqa|*~NYGWr
zx+NL-FGrj7Dx*qeYp&CDC_T8V5#{~rN63^bu;|;gHAf@1IbG(HBmUQEDwbj`Co{>S
z7fselAYkv+0!&Oy!ROI}ZpmxTm3q5N0@=Eq9aLn9Ac^ad%PR~xGQj9)byUXKdT=Kz
z!RR7gaFf|Qv;6Qtk>#x{$gh;`SWe4(QWYBVMTF;Z+^7`|vYQVZHC?4kT^yTC)|MUQ
z>06L;<GD#9op!@1aV&=wZ(~mX>o<WfjGVaH%sMS^aq+g4l)_24jOhp<3~Z)&dJJ+_
zPj-sjgKFo?c%DhgFsbBoK~HGem}HVOOgG>unsEC1#_KBrTX8jl)HEW(lA#enkukF#
z``Ph)UuKIll8J43bg*Cg-kh?Qlg69~9!x-sK8y_B^G3u$Y&O?6gvC-RiE<-<;=up6
zu$|tCrRyrv3bkJ^MKW9m2-dmoQ?IXW_DW5!=rM1c7srB&w{A7hD^Rvho(L_=_U#9;
zds^<Qpy*uF1WDgPk%lO<XPUaT^kT`Th_OKqydiZMl)fqLqf-+bgucc?V4XJ}oO%1-
z_&o!?3W$0+HMiYu*HXd6r)`?uRoo8lO>^gG*>^Ax55cXE;CjPeq*c9xHlM{FeTmuC
z4qtgf|M_Ktr%-``gf4r$y+Tm78rupxjxos_kj287bJI>d-WEWT$;>mWfS5end^R1#
z{~>>7J_YSOFPx+L4vo-{u(wH`N^qNg2^(8}YOqdZe9-YOB9myR77pVvf7!^-Uk{UE
z5*t}QB%%jun!%Nxo)dK<hsQ6X+RevL7sH&Zo5c$q+Kh2F=cK?q&eh^9pE-3m>S*TI
zs(5ejC`|bn`qh=_R=++feQd^h9z41+2HS1O(WA`+lm6(A6XZKHTGzG&N<PAAIjfV6
zSL&a<Ej_luuovC%L?s=m{k-UC%e<;YRnX27AWtLlzLS+-X4Q6{;4z9SkMlTcD<?QN
zY1hu;F8TFtx<N?bofG^3q%j=Skrg%SZ=ac0{h1~$mcQv323Im}?w=|c@$uIiO2M)5
zBEf;|4ZJvVOeN2navz{H0a>715rDqY7u+_XWbY#nS5cBA`yknJ=UUB)K~4|=5=-3u
zGsf17`CTPXZkUxLj#m182)1^$<qRO=Rx<)$(!IS&YUrf3EvMA7?F2Mn(%R-T`6N0z
zt{cTXblD$oHRYo(2&A#sK-64i+$PKxf{nmXKBBV``ZI|Q&4m@4jEMS2MzHD}4;t}W
zf{cb*y-N@SKmYlTUqR0$5WWjSyzySXdhrPIGwaX91%=Ojv|Ti%#qb==3%vVvd+duR
z^+$*h$IAL%P0V(-M>|KVM!b$|BddV|i&I;sk|mY~#fOu{W%Y%h__7QPHY(=VW6s>)
z9M-5+S#PO>KHhk$I=S9)N`^Zk+-2O34BV_T*B})O@F+~4q1OY29_ldemEuE1o7Le=
z>8M6xRFu0UHQ&*6s;kaE=IXO&NvI@A^+xi<8qq-p4eFp0)lqGCX@`8Fl$^l7SwzaM
z3iSP+Y!Z#o+Vt!uR;Nk|5gRd!)pWHj<N{}iiu`2DC>5&q)ZRFwV^L={9|fWy0s^v3
zP+hW2fk$%TAjd`r!n*6mM%U{R##NFxt@9<uhb0_2(mcyH_&~}VG1t0vsw}m-3#*%J
z5>4aVJpHn!vuoaTc*(~fz8P6^lzJvnd1>8nuvJ`*BQ3TcAAjz?Oz3dqnm~T((0H)r
zhzf&=9?rRg9r>Drig%tCP-wfzfjK701#qHPb#`P1yKqW!OcD*hS~;no-56W@jZboj
zFs(XWk??4+jLi4-`*ZT4B+o5zT&wXfRQ4*@jbv;72s*~fuU8My)!)?@n=IRgjh3rr
za>I-BYQHZ}Ra!Mou~{l^#Sy4ii$z#k&HkD>JiSO@s})iR(XP<C4b<}jH-LFvr!7Ju
z)ty>w6Orp}`T#w~EooYf4N-v=OUw2~!4ayt2Ka7u6BDa7x)kJNJopnRJiMAh!=4|L
zAna!bM!hLF)=wcBrm=1eZ#IXktB&1Kbq_*8_ffMu_PZA6HZVuf8xIaJy~T8+q8%EP
zDZ#Fjc!M#!P!vHQQUC>hZ|;W8Au1yI5nv9tdDh;Gx?EZX5h&*2iGrzR3R%vj&U%a%
zAFAU@@R?>s_J39?M3LTM`udUw_XPw5{hlGedaGPd6k#M|3L338Mboexm1d+hXH<A{
z(bobYBDf>d(FQy%rK3c(1WYgM*g~NYOFQ&&*e_%qz$RUk_OU9TJg-Qs``m2$SeKrh
zW0&V82vU1D5z%MZiF47Ak_V_tdM+g}?W<e3t0k6C@T8c4q9Mltdh!Qlm1El7AtgRI
zATMs2a$6l@AY{wK&r)|g8BwUgW<;27<maox>J!*N!XK<LRK&bP%v2%m|D4)T8R_R6
z0?xI+%n<T7LEIZ1<g;V-I%hYJ|El#KSj27qxFnPGlM;^e$Pt8MA3~{fo<X;}x?MK!
zWAt|R|17soiR!C3kt;IJi&u@{J(;kO5+9$Rc6^5A97av!W>1cYYi|-JiU^g`WnFGx
zZ_xC6<zW6a1N`-9^CuLhc}wts{M)zEnMa&qWYX)zZgVky8~PKQif5f~6*)hPofU<3
zcP<~+7BnwfTX{vEjE49(bxn4F1*jEo7tBt3J#7EB#cMs~sXE^Mg_zIYZLd1P=2%Mo
zXXhk;pJ9LkvUiu@kBWGH_B*X#T=dE&puTX&M7%C&|CQVIP;;a_3n*?n(#5H*EGTTu
zx_|@++HYF|lx<h=QeVAip)TjVPng952fHHLHtfv^tQL8nv>2!lbDK@t!UT{aMe{Ia
zF80JEX`|uJ19kS?5{%-_68~aY{n)ZE20ASlpP)bA>4fG?xl7r!smK@62;3YaXY-U?
zQ&1{K@(b2~LTKj!v&|){DJRxXXXxm*;s+n%pOHOIlTfP~va6Z@wNo@ZFG&x=aVhO`
zMLR_CXp3HFG}v-*Z{a0duBw8v{bp_pAC$h@g{5X)wax;nm3jb!n9pVhCz*4RNyp!=
z#mv_!QX`qEfL^<;TV!4H*Z;hPSwIaLHH%$rkoJ4dzkreQa*%!082h!DoiYng7YhdW
zBHlXnT5qSekLX=hlSgYOqepYW6j8~sG~HF06|P^s4z_W(B{;J^9(5kgwhEb&DXCk{
zaD4G`a}-#w?MEUaLHtJGA4Ycz*5?mS7B%e`eu7o|M~k6hVPT)|-iuC2E~<Le`3xz1
zg8+PK);At2)JrYi4t#pvS;8MbGO)NJ!Q&b6<J`&_-qi_nwHmiDzmU}28_)flNhorU
zx}?T?d7je2Dg7ORTIEM?LMJB#&V<$7zUZ^(%p4u;*`<rq)=nqCz<=)SBnZz5j2zsR
z8fAZ1zU^LYcEfrS-dP^RvtBny;^~0`s5($vT4qhNYBjH^*+TU80~#jsV4fKFG6@Lm
z-tk~O8Y)sk;ry98cYYNq&Xp1uEMGJZ#Vtv=5pa638RvE3cnIOwsPmJryF8v}2nl}A
z!uqlxGwNH!Q(TRPKJTE|{gY!|<Tn>*C?~?rQjRvS0X4n5#YyUZVI|B=%t*Kfti;4T
zA;*_+mSmVWpK2`~@OEdrAXal7YdW^kHKbl%1HP!(_ie-b<OA*-*|CVzJ$u%(IwLN<
zR?#$mxV`qS&+YQVTncyI-0~%<rj+PHzcyfW(=JhSW2l=}C(M<V^<7v9pM;>F2+6@*
zEyB^5ggOGs2P*y^)DE82-U>8%i|%25ek0L6pwrqKR%g9T?I%xL!Bc5qD<M!{Tc@j1
z_ec?3=__dS=<*3+%VO=9Y|jHOCgI>EB-h~AG`&Ms7CpTOj@Vz7_gt#y0}aLLuOK4A
zh6eZ%DFLw+3L&Hns3WWXltUs_d>hT(`&N0MTh@?rXbs$+8b$l_R<ppsjCO2>e1Q!!
z$8whRo#5lc^g5WKP#KA67t>cKdj<(6zbBEggDYVItE)D8r-lOz$sPa~(q#Mve@{#8
z{gJ>xo3i%^WZbyW99r}E$ldrDGXpc7PbJmYWLQ{6+V*5`++hXnW9?^AtiS9IO_o=|
zQSDF`f-Mxs{IMp1Pe5cJmc7H;B(4SV(z0(xJT01dRl?AbWcA#SrS;y$FTXp6&o{Wh
z-KQSH_K)%p8;-Rbe(R-re)6<=B*nnsOk2vAap`VsXl8do(!RD@K~EP$!RS|haw#OO
z74Wrx+KM?}J(fGn?vL$TJ$?;Q7a{cQ2M>6s%U&$2pN<T-Z#V)YywJ+1s}K8^Pi=h<
zX-yiMbO|s7KXdL)c8cD#krrF<c*M{&UivMSGxHV|IhQH837@+z8|ZmlkO-}=lr2NB
zC7686w~tgqef<tkFm~UU>iPTgd3<_&d$2wDot|<Hy}3*H{o-X!?vFcGjcvZ99FJ@r
z7JCYb-34-)_yE2cOB$y;g&`qvL#?s1k)7opOU7H0k}CBcG=q$1>Ek3YreJdP=3fde
zQM6I0lnfH*n+akmZ9U=nVq&+F!BiK_fG@2SI^i=t-32F{XyAN>Lo|X5l3ygNZ%PZ_
z)SKa1U6x}<tMymBn6|MvmYmy~3LuQ`)*LGQc3fG#1e}F#Kk_)YSv%RpG(ENKb(}kO
zh#$C4wy>}+FU!_QU&$C<M{%bljW9QIJ3aAo8Ko%ru;csT&7C;lfPuhuEKn&A-T-LW
zg=mqn<d2`e+8xll@s)VqtIt5e|1*+>nq{^LzP-ZeCHXo%G6<jIWMcruCG)-^o`?Vw
zeR+*(2~gD6PhNB2q>LIqB#hA3Xf;dbdCKMXpu}P4hn0aX5i6prvm*gqxLbSBhdsAR
zA}VJ>yiDZT2Ws>t>ws?}2l4*3f0VM~`z|OgVx70e$iwaN%BDLa(AZ5uf!u5~VtJIN
z%y#9mBn~zwA&K;SofX0)3CMV1<P(21obxOeh~C#adD)s8KOo_K$61-$>62Z^<(9c~
zhJ?s#qyCkwX0(IAAmX!@yP>|m0(~1;pw0gKc76Ov>yt61X;lJoD$rlF^33nl+17f>
zxuLMg)Pm!8XQToP&9c5_N6$U~Qv(Ef^8P)Z2k+Aa(T3~J8cJ_OVH}5Qw%5!TfbP<s
zz3oz+92=t4VtX=*e|!3uY+hWo2cFPDplgU7-G_MJ9M+P(o1-2qZoQJ<9s46;O=%DK
zk<&g@_b$F#8s}DogiCfnXTPbixV}Z3x0)Ons93kJo@}rtf4Xo|`fywus=BUsN6KP)
zgXcJ084{Wo{)8#yg{I!tL{ELs6Ci%F!DU4s)^X=_<;X>viuz5YxQlJq%jB}4@yqVv
zq4&3fJDPCLzNpD}P+%8fQD6_H=#Jc;N|P@jJ|E!46c@Lz7VGNL-|XxU{4V_8wFIdd
zig~9x)&k%-?~SUN&LuqWQW=AiHsO%|0MB+lH`0ERoRD836|VsdG;_=g=E?7?B*vI0
zzsCf9Y_bJq^VoPl*TLx*hG1+ie3uZXn3AE%wfR4WAgAASN>ZiPANa)Qtkg(3*rDMG
z+E+v4X5}llGUu;W&%~)B1<r#r3mK9v{$qfj9)A}J`s-pR3EWdWh!{uu#uMNfe&YrF
z#bF_A0BL8)N4oI$R_&&?L8fyLC0~b^F7+d9sw(s}R~**1_`cMYA4fNsF7z%0H8@?e
zcKG=frk*fT3k!2kZ>exnkYZuIX09Nv_M|rvYIopH@%Ug9Jl%lhE8ZT}2qT6oofMpf
z0moEy(SgL*Mb1@ro!lXE2D%ag{C|eg+1?|kfnEEKj;!oi=^p}M=eL4ljiS-#=(b#u
zH9Kpf22vU<dT#>7drgCH4*%G!V4^<__^!?9f4G^x?4Z_NAYY*B%p4~(kR(DEo(9VM
zU^j%VBQ`4faB`j$O@-b4D@@#^i}GnB#Z2tZj?~-8UYGMkD(>a=qOqCSK?n2BLbMxh
zAne}z3JZEI<V-jiH0N&gq6a3pBs|k<3xi)O8+npXQTVt<sr^~l>gov>L{lj@P%jcg
z0*T;^q~%ryb3h$+)e<*J(KZj<JCFM?_h@oTf1($R&*czf{?AfQeEy36%y?EC&3&$h
zX1t>wKWEC`WT~~Ux(Mi^XI%5Q;IZ#uFRRg87df;kt1>`Eo7rogaqw8Rr;W+aeZonS
z5Yy|LS3lkRttyEfa(W1>d1St$j7;j241RQdt==akS^G0VI)?x#Kazh=MSxkfZ*zv;
zH|O@lCecy;ZS3f16Ap6DLt_2;E4g8f5SixIuAhS!s9XnB9y?&=xcxn+ia_-Yy^!Bo
z#VlB@Xl=PF!E9_~ju4#0B$$7Er31&4_(`?tu4wV=;0Eah3ZR^Z8#myKU2-{AkV>_Y
zcF3WN%f=uY?{GDNV>b1rEmsnG0ir^Hnp_D7@kd{m`D)x^siK0Y^gVM!@|UYhX77T2
zWD{e?=k&DH8~_7fecY38*Whh*x${RWwCYqYiOCm<6|=N-RM7%qWNmX}QNA!J+>Oi&
z&(*$Tawb@cM%sq$)hDyy&sj^m8J%}jcuSp%3S_E?H<10Z6ilEAFez#I-+ZGlvjKsQ
zBt90_9CA*zgr5+#6B-imp#YDZnvo9{ZuIuABLi;z*$fkyO@1=IzfVl8(MaG6n#G$d
z_oJP4cp7dYc2m!|g3+n0NQkl*+PvX?rmL8ZRFC$&AEMM2vmvlN+AH4K{nGN^vUF5+
zr9UDAx+$MMyE`<~AL6hJ6T+A`bV!LTNu=LLLWU3F9JMKNzh~NLoNrD|&e>3D@kML^
zL|9lqD$17YUGmQ(NwmzTj|Rg?TAY}8)y-~#IIPgW`T_J?Qc`9C;{z(x%XOjK&g*|@
zJ*Q!@@jed}gX&`9x<WGbr?|MRuYpl#o0}(|ctQH>Mb91Q0@HnA?=%x@mvk+K^FDua
z_k1O?gil*2=}OY8pOo|=-Zm>j%Z$AgKphclCBCno;>@z;I%{tS?FDqfK2G)MAM`r;
z88fq43`cR3`uhJM#$GtlV^NBVGJ=gpJ~MawCMZo?kZq}gUbo-8NR`wkS+_3f^az_b
zIXOS1s-T>*o&M7u*JKVaG$WC77guXm{_y#eoptngY(PyaR}iRXdB;U%`G5i3NvaPA
zC!Sy^CQK$(Z+fGm65LCz)|cSpFPE=qZ)XR0(-Rot6}iw|nozFWm;qYG6%rEqD|k*3
zYTOyr)zlrAB^G8e8V_|k<Lw%Fi`SSb|7Urv8G@9~=;=?$aNy>WtcCHhTJ*97m8CUp
z(;W$*dg!mMF9XIFMJ#i(E#4X0z(DcBA;V;Oh{BU_^Loz$H4Sk3`p7>7-{6q{NGI<G
zeVyIwn44C=Wf~W=$z4h#g9;|<Vznf1N#WR-cPdxM8by!9ETq3{-}&myvg<mClzx-o
zayT%o@Vyg~O4PjAo9+}+VJ`~I?GDH5AZu*i@SGe2zzK6iw{td3APB=0QlCF&O4cvQ
z<%S|@H!z4xjGH02xr*6Fv$wPQmK^ZNk(xRw?nC@Lj$fzy^3Pkk^VE<x59+vCpn~dP
zq(3+o*7a#fEZ;U{O)b1goYw-*ik+jyg<LHeLCkk#u_+hEqBOf_s6`ecqd(fMpoZTd
z^0^hDTPAS~@#8pLS{_~JbmxjX$(G7HK0DW=Au&iICzmQ^Wqu7!L#zAj9iH-N5Qu0?
zcbcC+Cs{=HM_|1o!+k^mqJaNkz3;#e;HJqHeesTH-&moBg^tm5nkVH)J04Qq!vnE_
zzI?V8tb~MyjCuzh#Tv$yXM|5l2m}FfX?D!4Kx#+F%Kd}i{Vk234)dAoWSFQ=HiWd~
zJ|M$htQ*=gbD{gPI@rE0^`&^jjq$2Tq+ikNQddfIp70ol$BDmsT$Ji*&&sNrfK1Pr
zIhn&cn{8F;%cG4Aw8AD=ACI#A->U>)E%kqtgy1k?pfBjzov{5Br#zjUjEIi{K8riz
zdRj{Dobyt0*H3Xi{rVY8o4Rt33Pi=F!3^?2fpU59TD7?4-LGLmzAjZ<93@-JhPJj<
zgG(&Ta`#fkA*lS5LMb?>2^SoHX2|)(3C`KgX1)|1QA8U(Jml6QOV<tpxYA%rOG@hD
z@#%t3S6BFy2F3#>hr;#3x)}i5{}CT<_yWOq;h)-dXK_^lHa-{SMO%UhVKcU}iposD
z{}?Al(88Pn+~a2JoN+{@Y`l4<M#`}@(@pslM9*FCMpG7w#3>Q>A=#CdEm7kR({cHm
zBuXfkoULzn(aa?)W0^}i*s_|8W+5|EinCb0zHt5maTNfmuzRMZ)TuDqVWZS(0&gzs
z3&H;IQ{1efCP-Z3PN~TP4DANkdCJUuga2r*VM$jKTOUXISUI`olG`%(pAWrtp}g*|
zTj+V^$4uD}Pw}}xq$zJm2)Yx_DV62r&7Otk(Ao)hbBC#`CE2<W_JYjHDWi>cDtF)m
z@V(b0UOy)4>)YrhQ=NfQ!#UpJVm9HELg)Us<4247dUiK7#Te1z?Z+%FmkwK;VB45y
z(J_n07E%&;cv>VJfidJLdZMqo#@`!0gJ-LmROuVo>OApZditZ6{$)kwGah_=-a>|r
z^^MSy?b&saV#Nq`NNAxS!EfAgh}W@M6rjzxZmn?_QDG;>#Mm*`jHZlbR}Y1bbhi`d
zt?aatDj{<72YJ}a5QQ-VW6aVXN~py`aC47?9s0|H%Ee`U#ZbK>1A-66JG(1|=H(K-
z*6bXkZb=R0w$JF4LQbkfe9<2f$Lq7VelUoMLK|wGg=uZ~(9>~}JNo|oV;sTE41|kP
z)^c$V4I^Ubwl6=hvzqVC>|ltP(f^VUq>WJis*`y`R<OfmaZ>$X;QejS_~|39#c&XS
zssdY_x3-Q95M%2=?GB}crNn)Tg}kF!*fL#LsC%WTBD}WUlq@E`7|-Ybqr10+kRl?b
zJLM0j7MFAIwHA3H2^VnRO#xrgluenr0<;NdPA>k^GlZ{3`hxE+pue5|M)7SloL_+4
z(^uPRx20t&&?+^tt<uwDbF9RtiU6I8>brsdNFOBftw8wCzHbUGNn+x8Wt!?D82o=G
z<A*)_f1)Wfvz45vkCn<L*7PPLatUq|00v%yzHw4H#au--V7g8v?odq@W6j>!KfoKi
zPQDwpMHY;M10v7@!uf(^0VFO36>r93Nv*$<r5d88=DfqB*wGQR_@zUvZh9uGp?pOC
zMYRdeNm5u4md@}8a74@#(+aybt@NGywl?GEbu~d)iyz9q=NK`k`<BctKkKId(bF&T
zI(%U{-vcOgGgu|sJDVp}X>`&Kp^twjB^A%+{?<w+IT#k5HFpT5XwcOBAudM6`z8`Z
zk)~F%KKX9Nn=ViOhp)*!ZdKoYqs~S0`cy$$L<7BfemG^Zzrn=P$S1n&gDO?l_e`Tw
zFg1VkN{8&xqvbleH!7D~3@~<VC<XAq{vn9JOF={hKKd5SUll+-Fug@+I{aGpPFCN7
ze7V3-eR_Epx1GwG?t`yj*(2xcQ=VRbyw=gFh`tDU*Ylc|9?6^Nh>-2p{zPo|L%zUX
z)_@W)?E0NKFPa(oWQ9K=Zl}^HZV)C0QQ_$&edBa_2>!gM67qp^8^D86W`7~=I1^~-
z%wnZA0ep7|N8B+sHRu-ArkLWM)Uz@b_jKVxl-o|=zvztaV*m?YZ7+AT3(47{UU0Bg
zSwSmDv<!7bWoghr%&{1B4LjYCf4=*IC**ThYsW;KPJhFnk>BV<Sa#wBm`ev(oX-=R
zCo`YT@w*1je=aGt0cJ!aFtKQ(dWpxmWRa!=G#EH5bymPjR=!B^nR!tfkK)xK1Kq#%
zDVyEGq((`WS3z<>jl?ZuDQ-O5a+PW)OV~~tEro<x#jNjVnzT8ehu=1b9!b=~tQ^<C
zP};|$Q~?(Ehtq>e2AxIE36FNNPyBo_beF^Fqn$aq$MdLxqaC~T__KDz3Q_D=hR6?K
zX6}(tzieo7#~2bpAtAbCt??SfVXI*t7%CS#iEQMZZSN^d@iCZuORpf3K2IB_txQrf
z(xV51{*QhBabqplYXu<YYTQ@rjZ(UXgx8_R?(im}tIBa2i?*737p3PtZ2X)yeyL8t
z&YH0NQyu}eJ<$ec3oC=}H0c+U!2VN<*b*xKCbWBbQFWT!`)cc<YRJq+INMLs$+`x&
zt!Hk(mYr=8*V6_V*+<(~X?W?v&_T_wm*2u<LCj=pvs7>mGnIV#jyQKEtW=br*Chsg
zhDE+BU`FkX$B^3;zv+VsufWbkdk0`vQ=1Si=qyYKl7a5yIT(@Z93s?tDQg#*R(Quk
zo%`d%kz~PAX&0L)eWk_-0jH7sMa#83oI`C*Y&ma=nO)l)Lx0ogGDPlc4Hh?$Jx@bb
z9XmnMpM=NvPm9Rdax#X`nQ<(bWQk3T+lS~g6HAe8i4F9u1uX<!vStDyoj`H#(qU|_
zY+mUHj?Q|&@`o%KcwbvK9o!vubVJtj*)Q~#9!d@V?i%vi=YyJs5YHB2m0TZTnY}6A
zw>)_{Lyw<|3Wq561@4IkV`D#`h9!>2y-McRHP!!JwMBNT)xp{ou3;d-ceb=8gqA(7
zLvU6CJvmsy6G(>?xPbOg%D*)%g!FSNx$p2YEBA*l&0yJPkCO19de>k{L)U1V-<VtI
zN!w`(pR>^!CWx-TTE<FqCYWe<7fZN%-T{Q>n)@c=!Zo2Y{;L14luz)g8cOtCSW!i-
zXW}eKw69MWgcZ_~sTzW|%55L%R`ZT6IC<<ai(Y^p{OQw<!}%MK7x46|D4zU9lYK9N
zG62j}+;V~qK9v-6bF`T1SoN+w%21hrlA6Ax<G?c_B~2m8_n=z!Axw-J_$`zQX1{Se
z^MP-|WX9^aheWBDCquV6`SdUwtJyQx*M^tj9QL7$i+@Dty&T`dO<P&Y|Ew&1+3TZ~
zGy_I|b$rjd|9reJXp|-A{a9zvM1wo#D#0ncO&0L6Ke5l_=(i*qIK@lGhCS(>9fu%W
zl*}e^^<sOH%LIGZ9_yE=Rtvn)Jb|0%LzPv3cUuwG`l~w-KkZc&_kJo7Q(M}1KD&(H
zU*GXey-*9W*_BHC=9@>WTw!}kcd85P>;`D*mr;T=XwD{me@K*}+>;Y~2NNHn=%YcB
z=wXj5za3pwLM$$MM&{=hyA0ulwXY>?(sa=WYbyb0eSGfnX0VkhcPFsjX|QCUq~z(J
z?s#rJJWbriEDFz7Svmt$?kiBun#NZwqp1lAFS*Up=!LO`sl)d!PXugPSqtavfCh>L
zZZ2lCQ`x{3f(G}6x2P)!7?*gPZXOcQ5tK+hKk152`!AXK>9HyG(}q^T3TqVi>*b{l
z<i<dcoZx$w!s&0eGG_FS(j-0=b;Jd@WV*_GXBWJq^?9;%;b_cxQ8;=cL%nT9(%Y2H
z)0+3<r?c=))6z2U3&w8&p3+qV!k+I_F@3**A>-6y0xcNH$t1u-3VvEhQNRaf-Q`n|
z<WJt;ImY!`o7#zERWpj%mIl+tOE$;n-q8*a3o}(!HOMOr)__0n@P(hQsMT<1$On8!
zht)IZ2&#{L)Y5tKfO|m5&dKSgsB9iwa<<XG*M!)C1Y^O~5f_N8-uhgu`0dN*uXard
z>r~Mr1Jk6BN*DCso4qcEyCgLn*n%G44k~<9OFpFxhq$<HQYJv5WdhVG%6R00a)SL!
zZnSXIIbPd{<Qm;QN1L^)&YaEQn}z+nS|0A7DzaVQ-$A=*N44;j6Hy4ugX?`3QLii}
zet+-K18gwK0eH{gx*+}#z}e_rf%GGJ2P)a$oqAepN6`_;c%t5AtsZdKz*5%+os+Vq
zL(O5t{ZJ8^^nAV7TIurpBi?#sp%R%&3V3VM<BgbF|C=Xg)`0sZd$N@J`Uj(-nsLM=
zrMADnVk2c~4y7$)2pL|?;CY=W7v|f&UIsi}H<4Cy_evh14?#_4BcM)dpp}7%Ep2q~
znKgW>BlRoai(315(f~0PV$w4A)0dej_p5m@fo#9WTA)ExsO6vU_x-D!Dlzba??H39
z&b_h{oyf|BIFje3rYn!JGvyiQf|m^CZ9~fJ8n5nnIYBYtTy<_aCFdut;D>wzBAzx*
z)@F6os?IZfDuU=yMuHrrbdy8kDcg#82x*<SVKg!|9I;Vl_=Z-f{kdpXF5J}6bKfDv
zqV<;^ryYterb8ISLWbL1zmfjZKb5tL4><>@b4J|yg{7&fM>Cn@`AcHJQy6BFK*Q8A
z(`hLmde~t;^~v56g>!!w<bSoBPy2b!z<hFi&Cp!^^p#^wy|SNpmOLZF6)sp>8OX{U
zwnIITrXF{fx`Vl5JICc3NaMI8k_BNg8hdDnOOjU6KeEju*57SlerPCpVwJ@LNu-&s
z(#b605K~pXth7^xP0!<jXxm)#J)VSu!fmELGCm8kB%9`T^fW+NIB8AAYu=}B^b*yI
z)kOb8Ro7f!z9~Duz2<hsuKV1QguA`jpmwMQ@-m?7`VBw5&p1~**~L<rn$-H*l8Q>t
z^lS4B<GDoxWu#~vLP6ogcJ<mPgo_X6UcO_6riHuZm7Eu#<soS@n?!hA{VX=R?BWu=
z=!1YcFzfv}FEcgAUn{`I%dmPJe1ZUwo5*kgOhN)=248+)!6(Jn)7pHNhTsG%X1h*!
zcNf+*p=QC;9@2J&803jyeyiRSEq;rrz2*@I`x|y327Z;E_Vf3*&!F)34fce-K`z@B
zpm-n?2jYqrUm7;jg5=QP!Vg@H2hG4Mw7K26cnf?H9Sif{GO=yOC?uVhfc>|7W&2-N
z*TC2&Y3xefEBNMlnGWeKk!C2(DCo{xx3{9lT2;PcZ2WwmxL%p0#@cDU?<B}%V{rg1
zkxkt+#(<M#w2V^!VrUSlrO^s6Bx9fdvD7__VGK3L0fXwkomuO~F5xK9QD8}0$u*^7
zr?<+iVQvx7-D@<Yt^WMr1mcNR_gU1HI!~w_9Ei<(cJDfWnV|!O5<}hNZ(m+7UUg=d
zAGNRBAD2K1RLgeJ&<@=kibj(v^Gu<pH?Lq9#^&3JXV&LXt34YB9VbPG>)!Fbu^k`t
zx+m>mXdVfY_|ey6cm-4?_Dz(9fugfz;13v2k>a|zsyVIOX!?}+e*l&B52^9*{=ew%
zu$LM1m!u+u?@AhPbY@jzJEEuN*!2>IutLu-Y%Zk7n!j4fm^+xig(Rl)Yk_7Tk&-9@
zvIy+1)J>#qS?Y3VlBUgU5#p4R=8K&)Mgxb31Hf$XNV(1E`EmkDX{7pwKxnpy-EH*R
zTa_(JWtuF<@p{tV;4nT@)c>~0@VR){<J1&yjr8WOg`JXNiJD(;X_-2Erve#^2fhoi
zB>UzJ>?`dFfrY-a(<2vt{vuf655?Cy5HEpwoK38cG(<FvSq7!seX1QNTXOF_Pq?H6
z0DSfHMNliUa>$K_^S+J?Ypmz=>Rj5Zx_PK@NMP}AP5N$J7yT5Y47QjH-evMD=Q-sa
z3?t7wJiVGpczoIRG{$_g13}7Bc0DCEu048vFQ_iyk(JEzUL<~}o<X)pMr}gv)Z~+s
zbviArb9`Wnr)&Q^Whl*<#^6L}yRX0;;2M%*F*P=I0#<N-$;<CMRpvy)G+}EFIn|-K
zAXT<%`uXmMJd@KuWptOG#$bU4OTe>}gS4hpzn2|kg9kw0RMnvtAOFGW;Jy|y#{)b3
zE6=X~S8l`zavJ1;X8VtxckA0*JDST1%+6@AT~Sf8cT^rZ?q>tkEpT{|&LeXCd&Dm^
zlQ^T}B==BAY@DsJ=c5H2Abl0D5ei1Z{%s4=F=a%L*|X@RU#P>(zC!5I5qVBPwuf~3
zvGPdUFFmiDCZi(*3-d3eB5t2YS?ZpWiqHtmL-|Xl><bbT>6L`ut(|POV(>xtmfK}L
zA3dVUDt%ZdUjfh|K!ygSm1(A`){=LU>nh5vlsS~ad9pqb*RQ&76|W&JX-Yy4iK$f@
zCwHu=T&U5x7e8M?4<c~R_&7;%xer3NTmx%{r&Cj4S&sw8oAA|M1ZT0>vKVAHqL2X&
z(~k#Q$G+o{r}p$N3s>AMQQ>M~zT)TeDg1VKtB3(G+5FQ5Vt(SM6M3Hm<Ut32d2|ru
z&gVubushQ~-x(Si6)gnT{QMp;7xwpUjMr0@gi?sC$2gRvtD`(Xf(e)PWHw|=qnL)8
zpZSs2P}G>pQ4?_kzYXsNqsGd@>SK_(>854)-O_`}L}c{9X|oIDVt!Hj{MWdER)Hk2
zJb|YHmWJ!VHqt7p<IrLxOrwvj;O?lc9jTtFt3+@2y&>q#?_>@7unma(fuO`*M|RG=
z1zk(&w&KUND#oESO|>e>^{i-=ZgM}eru9(E(C>i?F|piBMB}28p{M7nJBeqAMqj@-
zNP<y9z_(`^8NU#D0HzDgK(Dt06J9zo##8#p1t;;rUMT(AZHxwMDMk=4Ai2Y)D9KDs
z`YPoOJqrspYNtJO4z>v|ZUjpoE_x%*b8)PVa_0r-jYEz*n-cSOCutKZaV%BZ6xqkf
z!_u7B|0!026hF;vl5Xwy;6YcWe9yB$5<L*_fzA<#R8tHtDZ&92?88X;6ZLrcM_vKp
z;aYYe`K-UP9v&6r?NfwfiHi%@sw=^Y>`}h%B@l<L+2rKwl@Hk|X@XOe$HnzFN?c=m
ztCXEXP=4@By_CUHfhi-79K=N$+l;fMyazcoWS1k%GM^9%pBO5&<!76Ba!|@1JJQLc
z0%$EP&1a04qm)DSOf{~q**~z>(`UcRUo#!?8X>>~+lEs^(|#V!Xn^q6x~c+>oaD%B
zAhJ(}-mOa-Yt^NEDpfw0?`Ly;GqZ8>g7SZl`w)i;1giIFqD8ICDyXRF6JNWuZ@cDB
zW*Ia%FFX`kj_jhRZz>46-w%k_<rS%y<~ng;(?x#UKSM`j%@cVc1=h=Ns~#7BY;ZN|
zC2dqEw+Hp*Y5A`HpR&FJtje^D_MjpvDq|oG21tW+NeCj+NH+q~CEXmv0i>m+MWnks
z6amRYhqT0@yYuej{PX|!K6jqs83)hde7v#u+H0@(U87IxRPYi>S+B6oR1Ngm+ZN#5
z&$AJ@@ycGx;^}hP1AHO37nkeT#u~e(YzNwupu~RiMbapdi1Tk~o!Bp-CfcA|FRvyc
z<UF)|va4kO)Bzb4r7l?hgd_E9P5i-$*J$>G-B4}CK+VjWgrBw1(8f>hVF|JeVmG*d
zQd^VdK99~Dp~+)u_RA)1X-PfGJ`d&`B>ee?a+ctWd^Sl5aov(r?C|LFc+06BC1v0M
z4qh?(#M_jL+e@9@$gVuEgNP@m*>wf@PNHbXeU7To1Z4YG!4(J5rrUAR$RQ&E&cVQ)
zT|;Fd4d{Veo@*to-~X4Mz4XjjY9RXgweJ_ZYwgYY8>SB~A+-4cl}bhz@4jGnHkccZ
z{RFM|4g@paFFqc<H}@Z<W_)TwGtcad=jUXPJy{S6OmpAo%w1+*0}aID$#;6G!vi+B
zdtBhX{HtfK)MA&NoZ<a}k)aP-8uK^WLh8G>TGes{9lpq%Di4V96O!nl*sRO@zvw~z
z_E9|b66by{M_tVqFHA3TAiVg{rtxte6=iRKTNwG8^g>^M!aSTpH#FRA&*Q&Tw{eY4
z;5>Ha!YYU9`rK;^Q6QUIpRbnMBA1th=6Dw8@@qAPJoE7oIQn;pxgiNk+slbLtd7yl
zK<>X47adZV{9uBqYe0ZvD23$ql#}q*>OlJuR1@Xv=_aSy_1eslFBg7gQ8~}eX-Y{C
zxso27oAuX9XJ6j!8-}z(bkcQv-2?4DNG>4!%$fz@_2T;Cv*^c6+{l<X>*;3c`CQGQ
z$vd*ekX8Ck?Q>5rmu)l(J)U{%_7LPS!j<vpeTw8pisgf8J(@!2!-HYfzV^l-ZE{Qa
z(;blK>=TGwJ#c4aCgYX}t?{}R>HlcckC2G4JJY$5^XlFIV#-VIe)v$nu~fMda3*rM
zS|#6dLpsb~NM4LV1e#3*M!)CnuJpq#;!~!KQl#Pjiv)!>;wVI|2>e|9dbeyc@Y%yq
zSk7{}`gweA5hIfxvJf*NXE{}C+<#y5l!fi&yhSKe`A&R%+dhHthzEUai{NpXB$w6Z
zVNSYT-bd2OV!q3)MA5yh74cEe#0aQsyV@y`)1ioxEpvZGNpXKZgV{M2-D&FhT9+HV
za|K2JiU|}hUIgXGPv{nJl$KV~Gv#J{x-n#hL6(4y1E<V}hR=H3)VVCOiqF3#@D{CY
zyo&PG*7c0`)ouybQg4eQ0lgEaZv5aU^oyXWhNkwMDw{~I>dSh1a_0lc@20QSj4vU=
zk9P)%w`aS*)s*ZQLFHh}&*vNEl2>uUFJ&9V+h0KZ3-+!NpGL2Z&EfIVjkBRVOF@A#
zGg&)Q;PF*sV|sc;jt?ouYw6P;K2*C${0#m0$YLW>UhxO8%~C;ugx$$&yYJS^Q<8P1
z&KB5g0{k}$3+HRso*@Hww!h5>EA5)zXW$Rl5y>k6_GSL8D&BU;a-(Hzs3fX1Lk|R1
z>kela>a_eLJp*3QeO7uBGyO-}>%D$0>g5fMOKSrSSFycwf-1Vgwxcm0|JH{1+-%6_
z#HMmML+XEJVaAVZ+SAE|lf^jNU0tNX()Zbs{1|-iAWhz=qwm|-*}0oH%Vf;iO*~pd
zWMn8a9WD?%P=x$jGz`WOXz#a<Z@YOWKwxq(sS=PWM{`i4BP-MSse`cI(2>d!>&@+}
z{9|zi9%FAEq=<t`3t515Gx(NO%6l9wks7$#bHlSdBvDw+SLyG42_^=;&B}HJlEK3v
zs%yB{h9?EhIJ+`nW@l&p@FZrDVQD;99ppc|B04v(Pc5y)1a56V;ABilUUw6ca0u>V
zsY^V)4$ML>R|-$U*Gakw(e$oYl=+{b7IqugENu>4`h_h<VitBS;NVQ}gZ7MhjYBwY
zrpYHlqOHAtadmwRTQ;0|5$+?8bqq;JJ5#xWL9I{*-kz!^NX%uioS5lD94zecY_*qK
z9i>TvZVsmfh1mGM5X#Za&++C-*}Z!?0W~r&=RzoVIo1!<+U{OUB;mMVd*@E=k+C@l
z4FVI(KaPEQsD14^wBlM~)+U!QLn~2!iJNb0yZbCYkq6sR5*LZth<%f88p{$Bm3OES
z8^AY0TJB5F<O?k}+;SLdFm2WHOjC}k8RGx&bA)YiZn_y*=m{Q&+n!w1(#G)@|5q|z
z;{SlD`Uy#hd+W9a+$0}83KkFueWzWe65%ypy2DjvHyd*6J(jeWky}1F@8}PsH&14e
z0r_&-1382|pH$xS_GfA@LIrVCkP{2uEO2VZUyVu_BC0<LQi0CQEJzu)gOf5<E)-UL
z@e<c?RM33&Oo)x*2&ZI-=)5_<y~1hxUP4^OTjO7sa<68(su2+tdaS%nD>ZorJw^q=
zLrtUG9X-vlEjT+DrwUhH#pxKbT9CKC0fau}n7sy+&43kZfathSlKDQ?y~OMDZ#BPk
z4ZDiIG29-jdxWTg4c;Za6;-=;4P-F2J=Sgx^TM^wt<2GE*6mEQ5>(O&@e8vmwpW9A
z9{JA9zm|nQ7@Nb2NV5EZGb!YA+vA(Csd3C~ZSV23)HbU^H+x<hiNr|bW^=1sX_zvH
zSWfTRcmw}$T~PFUH8Xs9V9$e=j!_?IWtzzLn8|E<o!xb%bmDHZcrqRieb%+8d*<98
zVH9Pn{X~Kp3OMoRIhk@<giA9`d9C_ypJcn^=YgQHnmR=HqoV>{1fAxJ`JzhuHr@YF
z%6>Q!13pE?)p%UHOm3Sa0|}pJyVNlW=1Svguk@31Y^-hJ^XjAXH6;OP<zKdcEs;IH
zOWXqel}R7GOH;X%9%9Kkm1$Ytepln>>qDMD)ZIL6ud$n=u_4=3x{Z5{XajA?ly30a
z#IP+iUvP+sf`|ZB#D84C!D84rJUyWN0J;1y^5uc^PsA@?-nWmp=rm*!dSJqLRW>Yo
z|K9n1+a-UbA+;n>y+!V#B-7s;9f=|cH}082;&viMqM>3wV5^^JDzjN`%bexp6l%tS
zO!fMKEn5>l{*LLW&pvtF7dDR>zi=RJAT|WZU~dnGhTBWjHP6oVS&hHQGe9HnhrayT
zfuiQ;k~NCE6Hx#CXM-n{{(-+L-h<{aVH4qP!_!x`I=7z91zlO5y-tGCs3j+p6`IF;
zYvOv(9@Gw6&Po_6=gwsGn+ovTzKg~qhs3Er<GZe_<4bXs{BNsrb1D|LfL9&@HGR2z
zz4VSYg-?uT4nl!6x|gdahO~UuDJVgTS1<fRTL8M*IpqA88r8eD69V6`4hE9@SEpi|
z@KnYJSE%PZ|Bv4m*2lCUlRuT4a)6G}oqwEY5uKt!L9w&q1;llEn8K9+zaC;JU7D&M
zi50SOOpmW6>=GLRL$tH=cIJgVD__^;<v_A~cYh*#36ezUA`70fd+pWd>s$QbxHdbW
z(<jS7^JcpBHHOFlF)Tx82lw%*UQS1cU2vh(wlZ{?tsHLrX!>3=(W0++f2OZ4Rtc4K
zUi-`_N0DysBt&w<&fzhjuL(2G4cIymPgFFLHS-ifw-gMbmkrQlz>Fl4@O$AQ>Dux8
z=l8j^4FDd3!YT2W`^RYzXFPq7seVB@RoRw>BcY{y-C)*dbv8P>uN4*jUdJP?7zw*(
zh=lVImIq-k9*MqwT|J>hi{ROt)6zI-FJ_hcb+)5x_0ty*T6pse=ekjS@6$)ilL6x<
zIjr>ZB&HYN!KpjBq%R_Z3~r|GH5@U$msFo)A88?DLR7DLvjsUq3}WVHxjz%x2daUx
zJ|Cc5LMroF^28jW@wrfD$`3KDD`;iRbSOhH`So$GaaFE53S_dMh)A~gkJr1ibN6m+
zkh24M7S^c&Yv-`WO4d}zNCzx_#S9P1ALdtlGH~5P4iN6o`A8!e7Dn*tQ{XD4!5`4I
zJFKm`u+elB<kF$|-{kohvG4F3X(yVp1OeQJ70)skDr=$e*D<|WJoP@B)1#fEs)zH-
z1A0Koh~+A?<=W%s*qc{<4Y>4)Yzme6<BdZ+Ps>4Km839W@c1%;iNJ*V5725rCXU!%
zX>=DeM$dX?d-KdmP(dOe``J}Ho)(@&;ah)pJwV;`Go2D0{9H`qq}9P<hnL**-GuaJ
zQQ1REy4W8b3{OeT@^sA&44ZQ!+dpl!qRm0Gj6WTkb0y?QV_O)IdQgjozLIZ?3}o=9
zXBU@Oc=X7_?GPKtOSyccvxV9Qmd;6SD7M+4HKjl(AV&c5-EH4)iYX}Aw6|A;CP=y%
zgeg9bWHtNoB&zN}GuK_G!sC6(n5Ihw*L0?X#lq@zv+Z&4TUM4BAl+J2AEzXQUj#-V
zg6{4QG{J49N{JzmZY5YgbL&oF!*sv#f?gkwyuP<YlZ5C-7E6<SQ9hSBuUvmi&*Qxx
zoG^c?pKdZ^98#T1b^GeqyZ3CV*l?8$Dz7_=;oNjPyV^NEzT`I15g4Q<xUN+<Q4u41
zeiDZfMYhkq1??@ss%B^&m;AmBvP_yxIYf76B}KcoYm9@yggrUhgYTAy_sQ?uK1WVH
z9fRuY9!Nt+<Mu3il@aU5pUNN0=Il8?6Td1Z(uj$Wdh~*|Etm<;r{MiI{ZY=py^TDE
zxcPtbDkMf&iYF>RXEuAW)-1m|yB)Hp$BoMGF~r)Tc`o)Y1?5JK&6_<Xg7~-8(ht^h
zH9`wFnSt;2PI6d#{NPcFx-;^kNK9bagtpuEin&W^u0|24m~M|?dt+MPxqjBGH%#w8
zvz4m%)||F`PN{Ib+e5@0DKWym&i_S0A*Og1r%S8)8H7PUY;VWi#W(tmvjVjTM<Z&4
z{WFpKe6{@cx0wUUngRq*en6YIm673h2Y+0RT5N1@y1NFPzA%V~+w~aIJ6qe#RO}hu
zEiAa7SUknipHXw4BL<SR@?)%rN*t6?P!!I4%dA=#mvo<WPDBA50bzcqj%HwK<(g7I
z$PqM;XShJR2zNA@RfK+e{@jogB$MU(Am;;Zk(1-E2p8(6^_7|2);oujQ}0en^a)10
zF5S6BB8JrT__j(2M0>Lx-s~pUDdA8GUl?!y7mtsYg82eBj<rF0`T<mi=)6gxJUPYe
zb~mV&=<Z4!ac$?RnE1F7wZ7Tewljj}Hq%6nXp(~k3A@$*O4e>)%ml>?rwV7vPro+I
z8p`f*+$!tI*=lDT)HY+|;mby{c4a-A873srH`9}naNsm8t?KeC<a!*04icZC4R601
z;y0r%d}03a;{#XM-m%Wy!K18wiLPEs?Alxsho_Evp<&?)XwFVeEuNjGbi|Kzqrb>W
zBn>UeJ=WGrtmd`Moc*ebp0HHbKIBU;_|UyztS8_1*3s^BZ3tJM!!F!ia{Xg$E{_aF
z+_?@B-Lvo$ij`GyZwXQLS1cw?P0&OEQIJoq$*y!crSV$z-(JG<bX>yO0W9OyENQM>
zigfs)n^&LzXi8C}gQFGV8WdKwcYg2N6cy@z>sVy?#FGF3RKD1jhJp6mNLSS3y**y8
zWY^)a;_Nl6Q<ZhE$0Msq3w==sQ(6#j3@B$zdIXyjb9N=Te|=*#bes17!YM5&$IR~2
zI@w>Pibo>U9#O_-aS|%flvo)m%PVUW84XVrf=;}ZkzOoWQBM_B8AnLeQ)yt!Tb?ud
z*lM4SHlH`KncFeeR^C2%sl9LHs;p7^WmF2ML)1Ow4xX{>x#!-ECG=ip_Vh0ziB#a=
ze!ueVY|iRJnmC`L5)|oDbIVl{mX<jqa=ozuckjE$H}`(pFE^R9XQIuotJE6cP7B&3
zRSHT$+x~wPkdc8Qq>cx*6_+>`wB8d`Xax*>&pDEiL=f<4Amjs({y-@R_N2;Zx6;sl
z`_`5<(K_BdE3X!OU%LH;)^4P>X3k>q7U}6I&!m83(~Y(4r8?UzedjC5mYU|-i9^j5
z^^$8#3ZvX6qBrUZvJ#ul+iVr@q)Lcx3@oqIzI_)bPWxtdPFL5|Kkhw$?dTC(j_rMi
zsM2$%f0;R?*Zuw^%p&y$-dYrn3#A&}o?6Lp6a;xA!im>JNBmcB<9n@dXr8377ay~W
zCuTHd+J%zkd&U~I)&+>)=f{x%8is=_fPcQ;$VeuRkH0Jw;9@?%D?8ZRc+u0A*5v8<
zgwDFhwK*Gzh=r_deX&pP?hJj|I^<0XYwkx59_2frm>8)rZFLQjTmz9aWc-e4(!fyV
z-A+e2+kM~T@r?rb;hdy=-%gaw{A8pDh_{a?PJv!{Vl?}9B7u{Lnq;%_!@q>8u)b;h
zmk=*GS*6r#Wn&+DtMNDrDvM47-<_5=UtcG^<oIJT+cfb})g_LMB%h*kWYJf%?5M8<
z(Qi4`Tv=0@SS2NRVm<}JM8k<N(;q|vRhMxbo<EK(;-d?e_ZD9Fa1d*Hph8DP+<*0A
zOwKJMn(ND?R~q&-dZ20TvR@41PiUCm(&9DCjUMl9B(owg`p(Mg*VYN5wB->V&Ca&u
z$~Y2;NuX{{$|@*{K>Fbd>K>sn=Mf~}?*F=#$$&>8()aZSuX`nZYO7*t{4_mz?c&vw
z0t=i~qx!e5Z2bR?F3Hw^%B`C7VMT;T&D|G<;+fj+q#yjyZW6>V;FJ-MUBTd|H4F8g
zj)6!vaa!oq_M626qz3Z6rCX@E-@|=pDFw;G&*F3+fh&uhGFTjf>eBoSvTlfjXKdmU
ze3Ftc?3aZdOlSplZVqMRfgU~GsO#br+iuV1_(TM#P%sBdTD?Vct_1XFhk#59h0?xc
zTK580L`)>AM=ZT-kxT9MD}zy@^KY)npx5`Xs_6_a{^m5QT5Qy&im48faI|J&ZQ_!S
z2FZnZvKTmnWYCCXW3hJ!8iAsQ9s(C9J)%zunE6*k{tJW-hT!PSbn-%99^=`B_FXu8
zV}^PVv2>J5$L25Tq<ek~p-sqkErodLH}TImZacG5h{RFHPdP4+!$DDiTH$#q?(Jm+
zjtPZ#f4tv8{`ib_P{}b3l4?P%YuS#+n1gP&ZwJF}SeJ>jp)fpHPzOT)CCIXQ?Sl&f
zJ6y_>9Dg%#+1gw+o^SlvAhWOQ^PO6jAFBa}y6f#DzVeAxG`RpKB{uTn^!C+D2U$bI
zARwibV>y5P4KxII$<g`QT*Eext`TTxXp<wkswP&?TLDqadX#?ZULdThth8}yeskpN
zvS2=AZrNNkeeuvx$O=+Un)?<uS6CQ%cOdq^Yv61g)1);nV?O!o(Z!2<O+Q$S)^*z=
z+jvVQ8R-l=(#R<B?}6^*CjshdCYy)+dZN_Az<Y?|(1Cu14vK`nDbjG2|HrrCOX@Ba
z#jodQ>`fi(iJ)~3M^IQ;pG<W=qK0#Oxm^~rl$6^Cd+x9gJ)AZxM@d3XRb~nBQ<Sr5
zm1qddO$erm)a*^uFV|hK`&o`=euO2_*1oIF&$w`Jcq{~9&DYPGT?^8D6^RK7=?<_#
zu&LhzOZ4u2BtBA7$3@$yCN-0`T^orFN6A+BeTVV!HPAek_4u8|u6U0^U#gxp(}Nhh
zzjC$yTN#@XxY%5WO39p9@B&|76<Pfyr(^|J?e#wgNVE?(hmQGP{X0H?pI^cBzlddx
zsLL4QeWmCOpuo9&Rv7~Ui=#Qa-}DSw=Mrw2efF4=|BfV^0@K+JnSAEjsOrN|4EeRt
zTZ9dC;X;`UKS|#B^HlwTjmil2qx{hyI-y)|;h@O!cf3MQuAyU95{)5cMzy^odPK)B
z+Ydahc;Q)(thMe(7cVG2Y2-40R9_SCm@a)L8%22Vc0e6bRaF@#ulRh@o&<TgCVa)f
zKNhL^p{vuVZE)Pp={LjU%dEhESH?tiG|x~|^p~953b@!jo^jo~FHr9Zdaf<NnzOtf
zh^oYd#$jo`!;=+k1}W)b_VZKT+%k%hQZ}6~msojk;7A-cq2Qg3h#DYF0tSu5SN4g)
zGC559Jwd(zxIE2;3^^kh3H|oQNHc^gYXRO~TH5ZKKB%5xi^8GBE7ac~5xe^T7jd5R
z+%Bz3TyH*a$s92C^73+8&Mp7+)Oo8hF?b>Z$3bv~cFf`INcwW!E)&(<90GI3=B<v^
ziMd4q3(Mn_iiqx=>kcoP{z{0$^9;$xMMr!1B7;Q7!}C(-Y!Xk`aNtnqHBV>;VUAR;
z&wdE6qy<`ktfSRH)^2R|x$SFPK~B^;f(z&Pg}_p?H9TIQz@ZmF3@ISfR>mXF63HQV
z>(PXV$*Xqw{dlwYq6MQiPJS0iq3xDv<a5J$SOGaDUu0{1G%Pj7z*r%GzKX09aP$+U
z=CBV;=;$PzTNdHsB8czogk}8!#XKcs{uFvOav6#hYU%~nV^@0pB?;@DsUKB47><6&
zquA3uHKSFc!yEu+9<;jDFz`LT$3wu*^0ybrkbr~uF^^vWLY<kb;;?btG_{6V&o7?T
z)-G`jHs7IkK3b-Vh(Jc`>%RH<go63V1*3X=r+x$CDXHX3Ju5p>_zAG=F?$ghQ?VI}
zgGWs4bg>Nt+7B8=LW^akQjhLA*l&xeQli!czvoPg93Dmzkk|(YNoOj`YagCdUlh3*
zcs+=GEAVe$@npq^%*-wQ9e4dEWIZ>m<kXdYs%%8$*pO_+zE4S}b?Ipto4ULIY9wxJ
zzIuMM=-VSZ4{dpQVR7-Rh^VM2LJ|_e4<-Yx1Uhct4!6;8WN*7M(f|IHeZQ9*gW)eX
zCMXPScu?-;5%t03V#)fciJ&ptHhS7@cBgrfe-KI+0oB4-TrSH&!kNX{AIBlG2J@e>
zLQWpP89`!uyH$ETf@Z5bpSTBejA9#{TT^>t%PWw_gy-ZF;uDE;bF+g>@>5?j<DVuz
zK>KakL<3nx^B^h$^S~hIR9hU7hj@C2K7is?x=zPO;^70=Wgp^YLENC3ty~%b)ymk5
z1k6K(`k?L{6TjnZ-9aN2v@2OhWn{3uiL9zuwV(RS##`7d5PB@<Bu_u$rO7r9bjjpf
zpDhX;n_b*i&NKAL&#M__e7Wuc@W7<XjtBaYKXNrjj6t0Y0k{H2;F{8Z{&2<xS^55z
z5Y|oP{prxxdT)OCbLkk3kmPUehw5V#^t5UtprF??GVy~gfI<cnKf_cFPX|43VqIVO
zOUTZY?wDEg;bX?2{v7j_oE+n&*&m)aR~M&7ulvm=ZQ^l(P^a$G?p<6%8MLt~Awk|G
zB^5P}x4ifrS3YRb5kenZJ3HY#=!=3pVqszHJb&$&?^m(&$xC?j5rR7{(5r?+6)bG*
z4Nc$U$t{u63)4N98u{8h;mmA_L&1!&DK9isLw+>7o*R;#x{YkNK9sk$8FoE;SRNKG
z-r3dkz5kMmE1ygGZo0;@@{O|x&7I#8lXRRznT-THliMHsJ?8(IR-yP-l5hAOGPgLE
z0(~5YB~`1v$*d)wrKy$S&mM5X#=5+Sl7&8C%l{e@q~;|0F8_-#;(yLDfNm=dZw0mU
z?sWZF6^a{YkDZtO8MKVFO+SYZLUnSp12PM5HS*!G)D)j(FP(L=ZnMwvla}TbS+lbl
z_#m@+_vefaU8-e6r`(@FXV}SfVd_{Q7_!Wai=B~X$XdCIg@8mhGkF=Z<K%+Sv8|$x
zjA0t(y}MyCNg>tmMcyCZ`L}6xkPlQ_a+gH=9Mn0XVF~XwU%i92DC5&H>yk2$VTC(?
ztE(wJ{(7x#4#2sPwzW!<*M`n3dlUg@o9)-dnODCW<o`u3(@8HWoxIf({6D<eOv@#$
zv5S#EM2??lM8Bl8Ga2zR^1(AW^7J@gz%Te-9~YNm=slGiKX!RH<?<JsZG^Gjxv^y(
z^_BgA>XiA!`j~LZ$#b*yx|EcxJDkCSj$u-C?>-V<ZK@EW>BXk?!crC$jSYI}MN?C^
zltsUyK3Fw~ef&y}KwDMxM-WZ6{3UF^hY8)!AN+;<N=0cW93~#qF}9D_!N7w3vO1<?
z-&UrJ$jD=#3pEzl*b&ZGPcmBaeR~$a%5KB1?7N7tA~SbI)^E%e8=kU739eV%2K;gu
z#<#`-miD7tb_QZgS*6PJ%*eCP7ss@bOTXY=LCeel5s?yy78z-$Q%3vBeJl&{5q*2}
z&x%Oo%9wcA&TjrO9S16^Q7kevAq6n&@rg^$2LCCaQs=oC{T_v)wr=u~<C-dVCgy62
zhH~pm%o33{G+OTy-Cof8_71SuBqlD7i!_2`;78=<BfIgmOr=cUot3syb@!>yd9eo{
z=1*2%-KZ@}7j1plTUyVq%&4bVLd&m}jg-gBbnI)!k43Kr#Ziejy6k7QIeDrqjAp(k
z{@cfNG5<Icj3w2@UMLzTmUCIconFiC21?C;E*DGsajS-A`i(=G&(l*!Fz_4-Vm$vW
z?KqiNNcJm=#*wPkkgzIC?G9u8={<geNlE9f3*%>7cXKCW%%9nOHuuq8@Ea9FzJzO!
zEK|3xX%ReaS?U)#+DEneU(Zi>#krBQ^g$?>fbg%+^*8I4Ec^O(qR#I}n`m#4o{yCN
z8I*uegh49=IyY!BGtp2kxAEm~oC#-@YWY4pkpRivMQJX&F-t)-r_a|G7l*!xGZos&
z#eNG^eNKE$%@Q82rYB$CAu=V;cw8cmHF%*ewVgR5{%BV*?S<QBsgyYLggrI2>U)&S
zLfQj=vtHHqJ+kg_zihh)`0MkHV!MI?^^ugdd-ZqjPN4|=Fw^q6h{oxaRBzt6#LJUs
z1in5C@1G5|qMk4rALCTp4^CCgOYnJUuDca}j+MWFKVww9SZtK2h?HvdIhsYk8Q&SX
zKxR9eXj^S?d@}<jmb8@7n0%E>GabHEE}FdiMN9caZu|VK1eV{<pigI&*|faDJV8Wt
z5Ut5X*LaoT8oG_??NVmXBaRA{kl3q#u1B;d#b)-Wr877mT)sAd_jEtBy58ue3(J`V
zQCOOLrO?_nH+LjKhK%%=hovV2$PD|wv?IfHkJ6a4<iu0mpfWNZ<&~GueT(IA74}RQ
zjC4%8({+}fTPB{dzxg=fJS9LxaNvvgInTsZoQ=_S&VVDxzqcMal3La$*gwlbQnLJ`
zRc}Uy;O@fq>|HJXecPZzQ$L${8Gb&8l0-7LUteZ1gNBsb-xz=Cbhk=Vy~gAHi3Mkn
z=}_AD?ULX?xsb;!>TywdU$v_3MVFP8KHbi!QI?J~-W|<fUc?jV-4WF>Yx@-xTjY=;
zeBs32+Bm}o7uTM>Wm@g~15RFb<`9bXTR8;RoHWUY<jvCSGV4z#>)Am=f9gJ`j`<oW
z9>pgkM&{&PNDrRpev0JA{2sL~C?_qXD3_$|pX<%EM84sMxci1IJXemY;<8jsPH{vz
z*qgKEN=pvCxTf7NzbhiLyO*>!(8z_?H@1Ge^zAAy6@}EmghCAKbrFJr{CqgfYVtFZ
z-%jI1>Du)x1-z_s%k2Iiy6eefzKZ!wX2~LzDNTE-Zb*ppedVWSGhn;U5$vz%=6jRa
z#X$DP^0?%g2bJvLRm@0bkY3Yv`br_lBwj^Xqn(NdQP^YgwuA8@`**+6oTVP|bGUp>
zB<7dL?z)iB)fUR%pp*RI!B$G(g|2pk!srEWYyI8t6@{$Q3Ovq-*KU@yqnah;5QylY
z1}r&-!JNMDrOUh9XQ#YwTVGesu>Ad4ITz3W6#$rLXv3Y(@M{dz+fqkg{f0$@%siyX
zqDxevwX0;`ZO1-pp}xxN!P{=@?%A0pT&}T5Eo0obc31k?G?};5I8WsaZK*Y$TI}uj
z=UO7fR1l5RwC?`?ko|bo(spF_&RiZp@j#xv_q=##+hE&!6I0dvL>H~WGJl0Z59*>K
z_N7!>xc(~Ox_Ei1U!R4AMkdm+kC5Z~&8YTOWmDy{s8#;YYNH;5Y5es|RV_)qY>s1j
z>1%{dJvEqt1IrVu_|r>GV|>P+DHjiwmLt})hL0*y=M~D{mgUWpPa4o&pAr+|(#^=-
zh1<ETdJJCOD774;ac7@(x#4G)_jBqsQ5dXDEp<W7JG|vXS%r^wkonIp-X;F)GaB77
zkkV(Qe)`uxpEM<ki2>(b33e7kOut;|6z~|d9&BGIKi_ArW<O3R^M~6ZQyy7nM+KIS
z{LMAdTc)$01>sksW`@%~FqqD<=7CASL2jf|ORoQwYyBGA_1*DvlLJzJu4)HsOa5!x
z5gW*Ga%J(94_Dkv+<nQ9487wm_nj@c_R$Gd=w$;~cEuFO*$=;e&*CvzEppqw=f2k%
zVoIznn>4{LH|=ojY&AOlg~t8q&Sml5N9XaG(fi-A$dJQDTALK^S@_3zQO@k^`RT0o
zXs^^N*Pe1ZyK0WzZ^!l``<5KBm7Gp%DV1SgqmQ&Re@;!|PMn1lcCvr5peO6>=y%1v
z2ZE}9hN1si{F@u8OTXx3Fgx!Svy3u?#i|F>xJ`q2LmbiCUnG}?6h0><^%mQf(CvQY
zM5a+hQEDmH*9o}oFcWvL+643s)s*PBlmau<&rp+SX?90t+WD+ECHpq?uZ#pwGQ@l*
z)xEF&vz+7zUTu7liq>{39IQF@5s*H4mXGAriBs4!o*!JLYR1H+gLgxV({a9;+StLG
zxwvouO$j^!no^61UU<XxOJvAY!|B};6G#8xq%Qe4PPJM7{}@Sm@_z6t_tIpFj+aRv
zV4>@<tL$=JvDkFC({Q6mCFf7uZLN;5{{8n>f=tu%7g?)Ro&B45(QOC0{dQUnHKn3G
zy?4?^bN$mMldR`y=G(tKq>pU+*w0y%d^54D_g$m!z<u_w^4rtjKOxu7nV$dVly1dM
zLS^u#InN(X1?8J%@LF!y=i{c0&K^$9bWC<hPT5o{Z3?2lFg-!JSWJPl%$r!tL|ps$
zW74E^)Y<7!SzW`b1A*)K1-X5>4P`k?%RI2K$qjy``|j|Npr%uFde?3SUrSO`M@~)-
zKa@cw2I(1Mm>5RMfxL1p$1d3Co6-DFZfoJ(IQG09w+uofGH30Stu6Wu7r(03Ih)OC
zvaqigv<cqcJ4^7S#OK9T$`H=h|4clm(||^Ttxk|QljgH$@5a#1TXu*EKNvV+VPTA~
zXSTlLCq9~=OIJ#LP3Qy#b+E9oiVhEzjQtlrToLc}8>^8(w(!hzvx!M5MqD`LAIK0+
z?^Bq#G`w0g!i|^8Oiy=sam67Zyg|aUd~{oLV|zOsDdgGY_nK<oO6(079Z5v0hFzUt
zhM#=2YRo;6{{a4UcIL6QJu_pkb<Zsrbq?M)ZS8#9YD|=r&-c`K$a^iz+w;AW?6I71
z&akuY$00i(Ba!Rl?o9slc-!To#Mk4bJ@Chvk;eTm!%9vUHYE&0Jp>#yeROH|elJ`@
zKWnNm_wgTGEFijkJlTSOraZCnbi>-nJ)x&yvubMDziz^GySVjr=;b6sLLc;D!}0~n
z)+cmlY8NKHtz)K2fcpmum9M;uO<HlG-~B}ACQ|WTcC}rUE}?pjQNCuR>L?BEX#s2w
zJ4k#GGP1w_{m;J22%vF#NB7qH`NiGU;b)5@JV#E)UDr_COTQ?^dpGJ-l$6;2vjKJQ
z)(GJ{Tdu&Ar6X10zRh^t2@l;iZTCvo5MnXTetxbAxq;1pUMz=$e6?{w>8s;$yB*ro
z%#OPySp&^n^Z8C*u3$v6HEy&-*hJMi68|=#sus(Cww$jVyu<j{`~7UAQBH|A?l1I4
z)W*{4E|oL5<Cdqk+p<-A;fEMRve=ViXPY#^5pnOI51Yu_iyyp|anWnV!TmIZfPF1;
zXm<{|RJwG&X+*r?t9rwVB%MBvQ&~ziV@r#P>}N-Bxhvj6W0FTt-W_Vlj~K|cgO2n0
ze?sm%I%Q^V)$K%Q`Qm!5JhUOIhs$+;Ij;7&ofN6$yBEtk&17%YS9Mw8ALoY8OrNK)
zv)*++VYh7HARb}XMz(WW^mvsn@3fy?#ALt=7ccZ@=!)!`j=m7dTgl9`W;7u583_GP
z_VR{FFunVx+)NI&M%-UqJ@D$^aeTFLL{pE&rB0SS__lhyP0(6+&t-R(fY8{URP|?M
zLhcFJ=gG^R9&O=XDd4`9qv&*0LP<;cbnNAC%7U=bMMV!N!dkCf3)4~k1nWEX`-$6}
zkpXRquE>aaeD-iOk8EN@`&=01K!AS{_l+E7xSRV#M0l2f)ZG3q!ZYM|0&R3bE7tpP
zxr=?#8$Ypdk#LhRR?R?&;8_OP`Kpp#RBB+Co|~Vi12aJI-BWIP=M5%=imIwK;_%<F
z&gRJmH#HBBI*fNv_n${v#dwTUwoXp!I8=}SeqIpPJ|_<^@Ou9qGa%Nllfe+Ma@<l}
z{^iRLO(w5P$Hp|VFJ1aMnG`Zq*RhDSYg@FFkk36fuXI0l$bN*#Fg(Fk#JY9s)~`Z+
zUx$@pUHiL=xBA!*KVOC)5|c(NUb=LHDCgCV=2Ypo@|d(mTI=hLQ+R??{1+)V6S11$
zOMY_*8(e;u6Y)A@If{8Nc$&WXB$_XTn3_p+UE{&{kcF|E4KmoT4EOwgE{e!0GwFLg
zZj0g%`>NaMvzVElSj2tH^VjQz?qmr`$#8Ho9L&`)vL{89Ho(OOyA!VK;^5JK=6r*c
zyZiTZgxgi$WJL6ME0&P4k4bFWzyri*&X?eo)*4PtCC;c`B8U04{Sc~*)YM}2XGhtT
zneZv5y<h#{oBfwtTU+VC@llu1bh55?z#7qiK`fZGXS&gsN?baQQ|bBhzu{#ExBqKn
zw;|8ag9yjMqC=i3!!4uPIhvL7RyCV1D)k@BCg#9{4#_=4ly-(b5-V}q)WetnV}70}
zffi`eU6+5E<ty3QmB6O%N5D<N@G(h4!*N!1YWmHAne)#Prnr6VrMium-H%tx!LB#j
z#>VC@8Cfn^)0={Mu)m*QA@4&3#j|g?4Ms?qis*LxE*6Aah)L|x$u^3^b#Fy)C_OEW
z?!kj+SFc`;6-3sZoI2BEA9&f;>d^Fp+qbFAh0kSgc@tb#3R@z=!sJ(u`ui1A)6$mO
znd|#8CgI@tt{(B1C-bzzRYVfl$L%lrkXRP*UiMUfGg0Fl%VrdOa(cSE(TKN<^rz2*
z6!q~M0Q!aQ2aK8(-Af-3rZ0trJnO)6nbk_rX7<BlsZaMCpY*3nMiWb2ZjeZF+wzl_
zm%n@a_7{v{H<&0z*4Nj|Bnjkt;}T$=dcFR<j;a2{hQqY~=TZe?iF<GgdurR9Q~Mg$
z$xxww3p-dRi}zNq*JU3d51g=ZaN3_-hqX+@$*GJy>PzBwA@Bslah6wMcBi{T?Uv;l
zB$tJP)M}gxDNTC6iZ6kQ6*!2!A@l+Jr2L)~v21YO4*Lq8^{Pa7JVQPYK#1UO2i}k}
zK}|CJktHIcg<ZNcNl+*3fb*{@5vjh2MsV!|*Y_{|Ec{59osX0TMn*%bOo+MwFL3;u
z;z~0yJ=~fL`)Zb7JNTO&VOl*q`-X?~rs#_obex=nOWX+454wa);4n{VvOZCp8Z0g$
z5k^^RHCfkegSDaMR{obKN~6jyv+nH33??NDtlCY7@)GhY=ZWD(+CI9fWOzb1+B!NE
zA%!C;Ghd~=z$}va&bahkjdD76cEto<`(CiQZ-1(e<7suc)y_;M4j$a}>qH0&Hjfi0
zDk`eQPOi!RK3TzBaE&`WJZznQyRe*JKL`tVpvXWHC4U1UmvB8?;m5^|SAr)S@3mc5
zOIfYfyOZB1y6>qC_FZ)QS?7Md(=CFBq@|;i08{V&{!9f)GRN&sF0-);YYvi|Lb<=+
za<kjaz7+3mp+M$<?Yvt26(MOp=dG_8ziG@A5AP-?Cs&>BjdHx55`w2ti?|6(?us@-
zfF6=R7}<GZ%m^>vApZH}I*cD?7lEnW0JzzI9nwK+A&*BVbCAK;KKsm~(UM^@@!ZLo
z3hAc91-ezMWussamj=dq?&qibv78pM^QfeJnFQWUuqV7neff1Hm~BghRoN{HSK=OI
z!0JCeTF$4+L$$WPl92E(DlS&4_i)Dz!`k4uTZ3>@?x|8U!o6_=zIoaY<1>ytKUFU_
z3}S^#Ik<anPd)MvF0!s3U+PKmU8_5`L|Ht0_N;2HYB?f2JOxkSK&aTTvzKqZMqf!0
zaZmE)%a3;RA3VULb+I=U^JrgEQg++u^D{&P?rrVwM>kH(>FIqrJ47K(A#TPBxK|Og
z8^5;KMA(y3t6lVzXNNOb3W)Ww%0A56LP|I}RiL-PmR*9Z4;&!Hd%?+j&W_iNlECv9
z6(ghc;^Jb_=xV&s)s1HWtuUWQY0?%!Yuc9<^v@;`77?+Z3F2DHt6r0b3<8s5V6Bd*
z^MtkBCTDj$6_bwQFA2eEJF^JbBMkOZ#>N?Fo1Uv9h=Zt9q$EI&%Xs7+N3ouGzv_|o
zl$qk;kNpD!;dMy@?vx2U>p1BG7v#oioUlG$2kVGP$~t-MgOA{6&tcO0Jj`yXKU2Pi
zn>1K&2J7HnbbBg<0QiB|313?iBa~^5E{Et68U)i;b*?%Ly?l(Q3xdaz0C(Kv-Q^+4
zuwNxe6Mc9gN2W1<=a&oUedRn&T5;2nqSTX9o&>BZvog!^tnc52F(1Xk!lK-Q_4?hH
zZQHMDX;Ftb)8Nl89;T<S-_GcT^_dP((gs`^69ZnCQG2VS>9BZEyD(&!i9S0)3rdFl
z2}zb?!&#pY4Bh2cI^N#iCdh#-B}_cDv9+a2;88bf^2d*5(f6Um=ffgFOl}^clO)4@
z@F*;;ZzGib{Qc#(=R2vyL-_YT?X8SFif*S`>_Gcj)t|x<*{+BmX4sywx}O*>Uu?jE
zEi3`1hm4L=ahw)A-*K5Y@XZ@=$aw8uhCi03D<f?j!FX^A6d<t#XxP{kAaR6JQc+UU
zaB~ksGR(1A>Zj#Dn0gOUg27g5vd$ewfHD|7lU89K4syV2Ggqzn#qv=85agrC=;(AX
z8O(G)-dmciuV<=WVVwxX5@3KPf51J)-u+2rK3LAtJ$xt(fpVZbS!5tbjV*fytDwOP
zTLhO~zxn#^VxL%uz)=QdPSp;uJ(+8ZR4>%?h9E!tmK&40Zj$jXzz$Q%R?f`{zS1BD
zgZ&Q>^NiNJ5v|*bWs9U+Xn1!KmMZ-FtIQ&9z}jsBLK=U4NJ$BM_6+9L9A23C8e!HF
zOy)A}O>mcjA|ISk>A1O7VQ!cD(nH8y79PML7@wb=vP$<YWpH${OT&foJP822Ln9;S
zXsM`W;#IvDYjnV|(`b^CnmWn$WTYNX$4%e@xC=IL9cqxl=k&{{si_eQxK%_ksLB5>
zgTXR*^Trq^+6A59NGWcM?iv91R8wBV1UbdOsx_Gc!uy#-y)AkSFb_!u_#EMR8$eor
zt*hfIY||wg%=bNQFu$|s?Fr<$yKE)NUw?g8(Q#L5^tquh(;PInzK&$>W1oA=ZFjJt
z3kgvgmw-tT(rQyC2FNx*jrixiPBS-9b6>wwLaI$3UWFQmfk8?aBr{+NYFV|I790Gh
z7XAz31Oid7psJd2db}?g22oV`%UCYlK%ETuQnh@{CO}V5?=C564yVPhg|p-JAs9wv
z+qeTP&cr*Gsi~>rVGe6!n2o&HE6J<```u-4C9{0eLxYEh2i*}C6O-8#Kwt*Bv}&u3
z9v>e+b!zH$sp){kOiPFg1h1joWA%Y!E{Hzhk@{7=)D-jg_DlWZ%(XjV5QXivou&E|
z;e{^(36nE2Xa@3ibdVxG*Ix+>dl}VEFXvUuz-8)>UIcEKJHmL3tU5rr?y`l|xBB$Z
z-y_j&tJT&pV?`e`uBLpjtPzN&Ve|9TW2@J%Uq^OUznvBan^xhQzA3}Ifb1k>WCmT=
zYtz5I$Ds#w@$K6;sxrajzcm0Duo-rA^Y4!-n0Fvylu9hedHGLw`qXP(><hg%%(1bt
zO}jro%Pe1=sMV-;Tp!F)EACGb!*ho;cMT6O|EEaaE2T=LbYK4$(ExVi?$3JqWfr4p
zFn^14(K>_8K_mn4|E$}PpO&LEEPLQp{#y0_mkICEKsLl`_*Y34sj6zO`6}umJd|a5
z^{_pfmsZPw0z++1-QdQo`@8zZBmI@OM&cL*X9qY?0>uOH`3CvZisy$1K<O}a52)y9
zeffD@3y7`!d8KPvU%G5IBs>XO*?hT)hs%Sx3Xs4|PeW^zYLSrE8*eu)jDUyP(eAR!
z7i0vZUTm7rO)~W_WBjAA^EjL~UeC?VMaHFmp*#L#eQ5=V6M>L$Iy*%3M{AUsXF_7v
zgz&^`^iDMvc4-iWpmrFBMQ}fu_T{x(_y|<Ym}NPH#>j|>uaK7{!(aqcoVVvA`Ovei
z;d5wodJLNpWB>)5YUFqu3N-{n#xE(sogg!AKcr>Q#RAW8iWWTYfvJlG19l!^c6xeA
z_(l4ZRJ5oTrEXQ4j}+m0jB%v$?yW1x@~3WWyoCVZw3iR|5>bpAcL@ln#d!gN)*Y`^
z%a_YaN%i-pN<yF(pB$-A_wu>`lP$Y6UR@r=h}1~qFdK{*KKmWO90f=kA~*lgDd1^V
z>EdA&JZ<(W3ue7(H^QCcG);|og60x*n|JQq(GMo$8$4M^3F#;)DY5NhvmE1A2qU_6
zOE>YUb5VyiQ*=A&lh>ZKt>)uZ_I{(`VPXDr!)3sWZr{Co_Y)bP!%iUI+o^^_q~OBj
z+~PoXG;M3m;e4DVS+Uim0O}OpRVuu1d$ePMA_XWhg>9unc06QR9!Eq(l-gwry8~5m
zyk1}5&mwx9TXXu`rn2MUzzJ2+ySE9?^|X8)z-~v2GFb>Fj%?f`{WF8R^Kn*mw6x+E
zFJGr|aCdR3GDIaJb6ND8GcdV3Bt%6*LL$=^W}!2YUn6X1DLX%^axK@+(By+rqVtR1
z7<>J8zE{7%j#z2Fb<#hWkZ}v(WCV|_vpHuRx!cAw@!l{hNji2h`|P|^0DNz>+%}t1
zC+<NW(5`h64=Xbtk^j9HA5V&URhjZ4NFL&Odk2yYau_R{qOt+`pG7YXi3%YUgUPv!
zT2+b=ktJaEvh(Y*9*FqRu(S8G8A6uv4+x+yy(BFwyPvb~0(A;4pkLVEiAjJjfNp97
z^2oBBUO-@?coph~whj)cWNqh}iolDSwS|n|MG=ETVbN#JBRk{hA$f*3S(Yyc^R0cf
zGprp1*@(@0-?f2KBvNPTTpRp|pYET#`rahl(j;dyl&akR(bPmsN%?h*1z#!VkM)<+
z9b`5<zUMDyh|G{rl_2%*4%MG2rJ2xCQHfgB9i~Cy$}isg`PqBBorPDT&!0P$@l`*y
z<wYid0SDSWX%0UtTGK|na`cxuC)jz*gTeQ&Vk6m1eQl>{Y}Bn>G~DkGzgE)zL9CqC
zG1t?lHDea<y}T&JnWD3QSQrts6c!fFb?FoO{3^G~2apuam-mov`=9;${T8t7Q(i<q
z$pHInVN1AdHW5%I(6$Mj?jZa0X0o!MvRzM+*B5};fJ^6SMo+(Qr6$#vQ+QmYhq2!J
zT$YT9wUMl_#!+(=$dkfhi=LjI#CD7)xK!WM^`wf53fA>)NaS;dHr3<5YF*30Icj^J
z%937PgNugdC579@I}EtW$ziC}WCA8Z)(1oyNkaf+B<xow+dlaFZ)pk)L1|wGr4Azx
zX2<FL2|IMD1qI!<Bhzp1R_DU(CALEqDIPhF#>AktRq#LG)?`vQEBy)691$5wC5{1Q
zYc<<4P0h`8Fc+M0xx=AJ1BG@%N?CcQ8bm=PTzi^{e2$it`ui&*3Tv&%i7w@m!6)!}
zp1=Prz0{hWUdhCENIHxT8jy+s#W9JyKsixrGuK8}i6UY%G)8wNq)sS9p1peKvD4Y{
zKJ_TCYr?ND_NtP$8&l6}&yJQOrPOP)%+8fxsH%Q~4D(q02qMJ})Y%cdU8@euuoxH}
z^NM8~9FVg=L$h!(?W6t&P;($Y|4q*CvQRpt-PaaLADNh#2lNf<R~D=zh=3%V=8%`(
zW1ttx0WbtT%Mg8j&F@N5fJA<GsOflk)S#Y9T-IWWhIE?>u-zCiKUyir0g^DL{6=4q
z1FW6oaU>2#ya3a}D3Ob>&6%vf<OeWS{j@#o;#=>>d@uxEXY-a2iXq@!2GXSCaCuGC
z)BvB+SC2DQk87{_b<$;3?O7SroNTqjV1cpxSfwq0J$3nLE{vQ>gC~}HnYmi!_VX7n
zT$p5ISVaMWWh6FWhM6D*vzVxIbn3N3o^ByO{GOQ^sIUeDTO8LY=7JG($gnE=YvWL8
zKSXh2QW^l$qT9AGp-b=a1mIpK4XfHzKrl40u6%T%Mm#(|rU#HZUhOzE6C#)g#`2MI
zR=<Dv0khM+T0WUqQ1Apr3K6#-vJ90tuxdw`ovEi$J-)_cgWNgToWaCC;EB+MqN1WE
z-~YY}E2sa5-yQm*V}Du(hKRn!r};$(362w+5#P{9wDM%f<wcmXe0P0PCk)`)XCM<?
zJ!Ii0n?l%*V%EunSiHN?BNFzeIMPjnn3y;Vfbh#Oz?ENq{(MQ~Gd+R>2<c6!z)g(1
zbpy#Cm?z<Svf0Aneze1CRlnWE2SI=p!h~Y$sdPN|E34XF<(TrJY!z8?adA^fr<vvT
zo6RJNzYEx*==@b=5O8+TBB;!}>`h1a9<^&*mNiiy2?XFhW^GM9=fGkBQZkZJ44b4x
z<mBYqOBKZ-e=*CK;At`;RR%6yhH0t!1VkhlC9Q@hB;>A+mYc!y!jwP|IFoB#_m_4S
z`v9S15GSw0N(K-`7=#30-+0h9K=?#LX*Sm$y*zn-R<4|{oeOWlP?!W-cA=Pn1gt@p
zPQ3@%n%h>|LL3r2KSV8!mglF*B9ro%>kfp0CVK_D3YZ@n3>TQClCK>zii3k=2G}%~
z%PKKVHYpEm?lTJt<oI3p`U`X$*@4mU_VqPGCGr#0tmnu_9+%Ty!p1gf4Wm-8bE}+i
z-`8T)snY~3hxwovA>_G0n6#o$C|ltg2v_oGpuK(fR*)kp81?{aa)?@vkqil#gxQbp
zqh>b@315g2!%m{a$8M58&46v79)BgW)D|sMYB8!DRswi)+-WB09yO+@slk)PF#DgI
z@$cO;1ui7<d~KvS-Ch^IhhTnz$-ka-<uIqGGZg3D%=IS^eH$P2u^S*`(|~q}{PZaU
zmWVN6g|M&5BEEF2ti8W~z&kx1^=@E7u;SB+TWSad)_w8k&lD?HR?PV~f825T?MK~4
zh6XqVk#Rr{Z0Zr8*Dyp10-^elmscI4qXLwTIl-9cb7K4hn1v!qSl@gY%-6}s3_HCt
zM@}~Y20H*6qGDp%vPlBOJT_l}d-)2PG&=?JM(xfg(tRboFdBdUnfhM%{NleP_Pu+%
Z=WJe_1E*e@<(R(}7nOOQC;Zy${{V-6Y9Igr

delta 107225
zcmX_IWk6M3*91{e5Ku`;6=^Byek7%(8>G8S@}P8GK%`3=q`N~vy1To(JH8G2e!s3i
zkbCw%d(WCRGmDaDq>5&wf>BOrz1HD{=;MbU*J@8S9yUK%D-L+{?$tNnH`thhf~K#&
zDNzihijkLueBwiSLi#rKbSA8_p7+7U;KPslFJ3I%<Q$JU)j6N=T+US1Fq`+p$<nSg
z$0#xsS~gg$iL0O{Xq@m;b<CbF$yJ}qg|<<zjukBMOA?Jse$bGt-0U$AT7$SvU!;)N
z-Y5=U)~}3Ihdd{?j|_Ei*mV81eX9ESwdb?jx2PK6gYWTv;_7H2@e_toU!L;#^y$Pt
zcB!MvZMZXsElu^b^`()r$NdK7{YH&%8qr;)<~Fhn#RA|9Y)!w}3=%~tSIelV2uO;)
z-H$wx4hiGi+ht^{-^B2)b%Y+Km3VIb*kn2~3JMS6H=7yY#l*oXFrAk>F~5Nql<vcg
znVT_wtk5IDz*Ly}irVi-Ti9D6pMo8q^DU)Q=HA+KRBl9WxyiIy8x4u;!h#}_^^WKM
z2k+_M51NtM8;QKO!a4nHsLRL5I}Ry=^<;{r{-(ZL&g+}QsuTQHC_$pLqGA=LrkWcj
zDu2$OW8N-N{rfm<lm0zRntJ(b=1+Rd?A6T4R)_Wip7Hv<3-ir!vyUh$$_d1)kyaMx
zhT(5Y?Y@?kM74$0+c+Ig>XHi=IGT&8xpv|Wt6#!$y{a>n?DeU3rSN-}QON1b3NnuO
z*qJtupCY!bEVq(rLL|DIE_p<ne+-I=$;b{~pz~ht)Y4U98!;i;?l5}0s=B?ky;UrE
zxnCyfNQ?9?rGnn4$F=3Nw#tn%3V1EaCvw-Z*>}tFe0D4zOY31cQG~J1u=<#DPhuE@
z0-1cI#=k7pH#E3ThP(_RYR^`ggl^?ZQqGuJo#@*CL;e=joTSxi-K`E++7i+dPpo%T
zZLn|5ifv}}q!8z(d1%UC3%%AQzcc@nkq85d0Z$uPgr(__AH!}!%5sAumFW)Oxae|%
zRGw}_rz$;Dq3-s8<5lJUrL&sdHvVZnYro;siUO0@$|8!{1c{MEVr?U*?)9_OcL!>S
za9|FL54t~Nj#T0P85*|_f?C{vmgc{}smeI`?ccc0yYj-$i3E){fopFJpIOlmi^+Uu
zCm*4uO5z4>){RbnZ0wqqUkBwv@;n|fSNOrte#3%-U-&-8W=i|dSEYy=Mw)_u51wl2
zFDhhJAl{JR?j2wc;UPr0UcD#`z1S6Y=CwM8Ogl3D@}<>6y6f_Vq(kD>*2j{s#7|tW
z;U0~*KXsXViN|W8=n;;5_ry4N4}AB$4-8jk!wvLlKd7oSCeZmjud-n9FFS%U2%bk1
zXzV{UD6b7J^m|#1bN#h4&D=Rop}cohTtFHV&%x<oVuF++r9zs_;N`j|eW~?ph@3j!
zV-m_sqN1i_r16vBr4z_yNS*zGii@WzM>&(GOgs_j;xjPt!tHz|7w=cmPR!W?Qpx&z
z-@r35Q`dTu_EC==yf7l01T$h=GSc_B#PzgXj5_<=nR0gzDo^_F^1e35<2qqa@46wv
zveZ6iVyT~u>4fiW>-2>#okNmCh@8KMm>nwfX0tsS{sjm33Qk1egWMCv;A_uZ59(dl
z@<bsXGvnNo+x*N-_k!ZqPg%?*_?KTL*`G?}8gg5s!I?yMXZIY~T7RCD!dzOO8{B-E
z6BoPjBW3yLC|D$!<*krRn09t$W0gb-!{?R8Alq$LSlkdrnQI>ujW8?m>SD@XG~nPC
zPf<nY<JO<)hwq@k?n9NZ=x36Xc^;0JorJh}(L)P_cS~BZUkYKqdmG>eCw|aLOiPk7
zaZ@{@nqP(4*bW!N;o`>X=<4%-uzx@C{moBP6Ca)?GRE`eM`P>?=GE@bTDvRD6#m$w
z`qu@vhN(6Q5LrC&a`T4K+ZK(C)YQia2r0%nMaJJAl)=qSZ(GIN;NnL27c_aOP1Jja
z370W^*|HgypK<-&JQVj5Wn_P~WmmcQNd1k}jpS1MY-gS_o4css2T5cKWVjWDGOn{F
z)j*dctRZ6wvj|%7$iAE*i1~jn>+QneFGgY-UZ~@I?)o9oSX<@Xbt+j)NF-4}S}AwW
zPLR>4D+M<qkAO!DjdAsqb<d;iM-9W}i5iZ9U8Yuh#!?5zukzF$H<n8TZ&2iSU<%Gp
zq{-sOC@u6AcxY5kxm?;9-M+|E=a82ep4J4{oxepYskfk=V7eMM=bo#E$}J*f{J=l?
zGqk%V#GlAwyT$yS2!Z$)qblrvlTZGKk^+z9sIYG0FBWIBc;QG%NtaevwY0R(N9XnI
z<W;P>;8w~|k9zd<H{nM;d@J3uM(}FRE0r;^R&=%p2*O3{)mOSqx#2!a`Q^g~c2vpq
zv5$z4-(CxIKXFeUexQf?`85Q#WlHn^zm3$0>9linwJ32+QJCaH0WCLqm@_gDGwot`
zRY6i(2(KEv_U7wNw{vf*yCWdItA>iYa#n{LhcY8<-H2*k%vexlMR9o!mA|&d-+N`A
zz*DNOpO*Gk=y0;!0^`LQx$?`b9}z!(%6x5Uc!so!XbWw#M))MU;*el4)~--eQGEsg
zMlf6d^~V|hj~|#x6v%i8{O@wN4F?^cq4Tn$8#qj#Ce+T3b5uMCi6u&L$WvwK!aLuH
z7gN;79s!3mB=)&#8E+0z(2VM-j>F;0CwC`xYIiB$$m{OC?-Z{2ep-<tYzb`TDBRBo
zuEc|!zUKtW4SBDmrZwA8E<LVyWAuq^xjY!IA>X)?Y7am@INKYR=Hg6#jZ}6m;3WDW
zjr0$0j7l;KeNt+LuV<8rsY+>xpR4KFdgn`y-s8jX%HmsQ-7VoLzV%jK3k!Thh%Mm9
zrB!6ag^b@IIb3!9V3_Gj<Ouv$6K7mAcqt5Jqq)NQqI3RKOaQni^F}I3Cd+7|NS0YH
z9Y0)DFh$>tqNn%Gaf?x{^|TXNP)unqY~XI^UmM5g{6)hGAK30}L!=FHO2I316lXe@
z%X)Ey+viH`jwv!GeG%Aq;Dm3$Vzb`x;>1F|eO;$I>Udn=Ma3>}`=C<<@~v$NY$t4Q
z7ZnDenA-Uk&%q7TATa^JpL_BrKo!*1h-is5BwfbbSZ=Q<4Pd(~Vd1URwK3%YipHhv
zS5=ghZF<v`EnJ*<f}A@)L@BX-6BQKEqV-b@erBc3JPNz*TG!!pbW%d#(QqU<_3kz~
zU>IWeXyxwp7t9ZP2Dybyl^Awy+fD|YI_;@onmqw`kN*AptQD&4bXiasc!=AEb+QXi
zjq_z;0}BtpRY&{JCxPTwG1uLz^;IFE{_^+LhAv*R?zQ7DQod_T>29kGynYp@;Gxkv
zk@k|R6cLoyXU_TdC4`4Wi@g2~)su_5sbqU6Hm;E0<3WpMZM#l!DeVdVn<d>eoyC#w
z{ksWiql3JQ^Q<K~l_)=Dw6xbhF7^b!*S+9Dl$?YWRuW1p_A+d1K_wH_yPLm;jUBf`
zS=3NCgF>^IuQe|`k_OTHz+Q^Ki@&+3tovHXzC(jFc-S5u=(zi!jOb=*MR#wbxy;TD
z%NDX@eE&Yls8bTqHTJ_+wRr0vm@6-embu|tt}PVoA~M3<NN#;X3n{gH>PR@3t0dX^
zPF)2xga`<oU=96$iesoT!C~;syf5b99~DUE0%}X0mBH|Rm*caBoK8e|2#cI8xY6T%
zhkjTkFOSg#z0O_ECJc)hWp@QD!=Zx~op?&Bwx;%KP7-~>rjgoVbserNZw6~g`t*^!
zx6Nug^IiK6LFd*xZ?#hz847K+m*QS=O$JtN66LVu)jdtL!;I8-rJ#NHxn&6!UYCEd
zFnkI<3iqcLR<x$si~RM78rcWBB6fUP7^SMjxcAk-IJM<hft{t^Q1fVC1SVuPH;F_^
z*^w>dGEll0H^YTgs?wB-!<&;t_4l=M8|~c}|0x9O`Y4SuYlk6Drk$gY%LwvR6EGJa
zAwu$y7|f;%;!PnP<I9QskN&~UT`lo9bmq-{L%#+JBVa-6w*&fe#t4!GQ{tw&V`i;n
z%^Hh)i6)=2hK?XBh(w_$oY%IeZ|s|!?1+x}s)(`myI2KRnf0-YS1eRHTmQYEYx5_s
zv|efz{S%J8j8SmK@J1>ZN$#giSY(*8N4Ua_XU1%LCB!O|B3ZZ^kc6pAu1G~oYfI-`
zPXGhdk6csLGqPM0Rq-A$t#&_p+|l4?$=>R9mfsj#A$zU)zGt^u99Ki?X~~29i8`Nk
zJ79KVZKC;Vve{(0=Qo6K6T#O%O=CT&tO4=UuTA(Mjae6S+HBTJ<G4$SDVSswOm2Cw
zqp|E5NVBz0?*<Q7Pj~yXWB;_cRBRQ#ucj9hQ?N5MT&nOkwotd(OMy*8x@79{xd+YR
zqz%m;gwtg@ZPe%VuyTAI5c=cVch_Z4g94+^*M;!TOJ2`g!$CEZF?;T5hGPCycNDU7
zUo*YnGT=4lN*d!z;mfDaYs=1RrFV!V(EXi%<S#aj?YnNm8}>sJ>g9SBgpByCcn%3|
z%T2tN2W|EjL06j03QEj{(vbjlpw^*|Xfk!O^|*}Tke8l>R}m(W=Jn-!%ZFd}bdY-c
z920~Hi5HeGqL<PIF&x=Fy?M-Rd+h2zN>=<15BQ+O!IZq&oNqLy#RcC<_T0`1Lpp<1
zYjfVG*ug>EY^o0%=X)5Q$krcj@K)gUe8!I5?3umB+XWJ*JNB>fac`HlAeFl8_bm23
zE)`dgXWh;S;5nh?v0tl?HCPGDu@MTb4_f+${I7BZWJsjGV?KZE|1B$;RH|1=EgLfb
zyHtOszxZQ3R{7Hvj>(XlFP{rlC3xvzo|p~4q+9`CzM7ThQOC((tsPq&?K?s|Er2h9
z6#<@!=&8|~kr1R7T;&8?Nao9lt0}ZMN<ZYH3Xeg&X?ntWMp(^CyFAWX-_JgJy)eSL
z+VAu*Q?7z#x3wZT&)R5KQH`f4?kMUowKeWmI8ZdLrP4_1%De05S4cT7IvfT2L<4k8
zWSopxVkVap74g>M;@ysRH*`-E_g-u&x_taNXPzu+0lj1G8TlQq!B3>Owa8dlWS8S5
zdVleZ6aKd_B~aLh@c45GV%Nf3TZVDmhnm+mD8vk&u;#&iUL4`uRRcJn+pPZf!1`k)
zqJ_X%v~g_f#Y(Z5Z%`lS==Ceoc|<stJcySQfrRnOyQdj+#_us2+kH_;c^QeI(yt8q
zEaq%18=i%K>hUV5utxIgmQsa6V{l~+@M_A>&g`^Kl4$^|wLEWpA`_<5-rL>S((^JB
zafZow0xw*}J`LmU>|K-nW!k=YsT9cU<LkJwlOCdJinJs)s&GY&gTGRk;`I!|yDfC}
z??}Vxj~GnV9TgR8HK?>($5equN>ZU}rg8^~eO&L<$R*amV0$2EAUpZi$$s4KbUZpr
zd2TifGK;+Dz0%bG=#sPt;ARReXu5i-$%bOb69YeARc~1`Z>rp6o#&pTSj!ys*li}n
zsnvF<+ti%bW9ddF3QL!y3dNJLzX_=cWGf*RjDjXAv^Pgy00;%0wj5h9E$^UIV9?xE
zJHp#P!lCieu-Y(c+8qhQLA_Vl3ik3n4s=T6Y;eOzFPGomk{IYws?aFXB3h9$MU^bE
zm86l5aa1e}4<$yJIct43Om#53s}0Zi@M`WqnR~`_pGO;s=r(IMj?{+tI1cFaK=5-*
zBLy*3gF(WJ*p{C$*85kF#8N4sY1u)A_zrdRi6Roq!#XU$LPY~mDs1OWXm>OOG=$W=
zj8Fr+_jfLEbA{`IqD!gaj96E#P7ktljgZOFOtq)c4=df+QyUqKOTF!<eGb-soHPB>
zK`IBW{DYf13w28W2yk*Uhm0WtH7;r_*^tD}6Xt;NvJ0{I123ovbd2@e-5I19Zv4iJ
zSo7|{#WUf5o8B~GOV|b4;0i1KVhz-pOvYGv!%4f3cH!w^uNTt`+|H4j)y71OtUgo@
z7lns|&T4DvabTdt{=o^I!*uBf*+7Zzc8TEEfKtgBqh3PwL}K<GcAv5=jR&yGBRmQU
zPWf^f-q4Z2QnQ79QK$LGep^&JZIM@TpKu3ph`(ByvWN4Q_;Q|Sr>B>?H@@X_%aPgD
zG42EU-_P4y9-#h%zBP4*P2y<WERIX<B(XnC!{DG3rO$AqWpr~S9w}KS%a^AbE4S=n
zsY-d(cZlRcd#`A7xdeJwSMCnSO~;>CJP8c2q132)G&36$1~?l|DbO<?4Cshd5d3$Q
z=4yJqjdEhH(KF<rIFL506u*JNcf}f4c#~8n_W7yQsQnYZ$5yI#i};!Mkjy=zGsmB!
zdP0<Ri9M*@!1Pq&U3MG^_5S-%a)7*3{$<d#artkP-6a@kG}kHj0RY1y5H$NKd&dch
zAM_=g^-jW!rVDvLz^k$qu3&*0CoA*A$2t48t&8QnS<EscMVrX{)Cv0I>FOwVVlo?U
zLz+kIluF|q3tKa5FlJKnskv=~%Me@#LVQI{4#_jjU<H2aDc;JG)Q~;?S(}w$$l+`;
zve(z|sm@47Y#%PJ!aDoVLlaK&E6$Tu(>{$nB6UA0scAfn)v`!eA)qwbjuwJvIkR1P
z=pTTxZ^{^hw#b`jnW$Nx#Ms^|{z`Y#`L2cp&*HXX<kM&mrQ~nOrQGdAGZaric`rOk
zH{mgvf5?9wEE5thz6VvU%fm(^utD!?Lop%K!gC{(rXVv?l40-;@s)08wJxc8uC)sg
z+SJfhJVbN#Cu-!&2~QWMmNSR(UgQ&`rBVc!rK~a-PQW<xJXs1Ye}WRu_<4f`Y_uOw
zIjTFkrn+rw4Dx}^`C4iJ+ZOfK;%!`^t|yI67Gq}Ut%JA_?qKq*oKPK>XLvDs2F%o8
zyfG`h;?j9BbafVCZm4A()ICLZo`a=bfy7CeM*ulO>r~|j0>|@r0s6drR#vYG$z4Qm
zaagDKpRvlC;8aOoc)QV=HaE2WzzWbHS?eQ~&y&8#fPcJWAtCDO;6aWl;S+SFj>>ox
zsN>{E?n5&phE@e`VZO5uH8t0``@ZJcdfBSwH;YCKlXq>&2`!O`xs!OBP8i7p_L3tl
zIB%&hFY%pMCt;-}W%Z8gzb)FHP6Wm|PkL|GJD#DY_1c*Hm&TW-Fj&>uzzNf=K2p!q
zMEm_G1IIN{H6W=_Fm0vK+Pq^<@wT}cYW&K}>e3pmQ%!}XI2d#8pw~BMK-ng#zpI$i
z*8A%V@09;rPvWJ$;NTRQtS{y(qqum}Zun@qWk2XvB-0CC1()f}qzwNYrM`aYSnBMa
zAa(PwVZl)<BrIHG9q+!73B3G&28q^#8g<@|_!KpWy;KVuzLPlR6~ws?V^H=}9J@aM
zoJXMkH9VK382kFGU)DMIud^K<HMfV<_(-5!jd9c?^sko97Fy-f(5wv9<&W)KS@@$^
z@0m@<XQV?`t?b{>u0N<)6O<9Tl@MP_hnpPj;R^-0(WMxc=3Hvsn6;*n{V8zWRy2Gn
zv3A^*pNSgyiz(QPAL6f)yuo%)jO^me(nY9G=Vn}_Rw>&QRIz=mI=lN(ZEc1VMn^-r
zbd~3IRcZK&RPbEfX`i3th`_2f%3Zzwx!o}rEQoVZ|16GbbPNb3OxrgPE0xzxRgOi%
z7w^}9v-uA>t%6GoC&Nx72^(8xyiC5<7u<A{5kR-gkF<Ms-*PT<2s*yhCDnL+Fyw?r
z6Wf+vNdwwacY`nmmxIA4ann0B&{nO!u5M#Fq}DWCOnI*Cx24YYZ5zeO5-1&@lHBRa
z8v0!A&&eN;vm{Dbl)th(mMcjUypvH+BOguK!MI~P(MF6ojftpAg)%r)gQGlJ8lWhT
zNd;*cHM#jHeoiQ@A~R&gHT|>sX3`)R>SRz;-Uobebndhjj^#{qL~o&*+Arh)2E@%(
zm>^obwjM9n$ABmS#g(#4goZP2XM&-})G;#1CdS748b8yWvbf%0>vpDQlfapk4(aD~
z)Ool@<)ww0aD9Min*MDB7DS<vE6ILO<!-Y7DO<kIZ5i?gbxN)grMxw*JE*f5_Is+0
zXl|<WO0tY%EnF-f(iocAveXW@yo@%~5jp6Kk)#$ycE^!#)j5R*N_EN!EzFHNW%D7)
z+2_Ht%ns68JgeVMv<ywZ`rOjTn=lpHHV>B*GpssT?~6llxtypJtSd|NMXpZxc1}vG
zj-Rp&hpZkL^}l{dJmc>WfL68oOQ?lRzvkq(h3y8nzh-?jKi}Ew7meuC_AZm={83i}
z!b}H7HltJQ%Ij*#>@p{Tv8RKTLZEQO)l`RmS$?d@=do`zth}mr&jN0l|H-HotvTCq
z(3g`S(mSCnj;ki9_=B^ENFW(GHKB>BaNIOrf8OP%2|IlMXfHdnNM}Xk(wNqCx5lR4
z`qo7Jy2ZYeL=pes<X5~nrJ`Pb(T@@&R2U4WY-~4gUm4piy9q=*!~b9<4WHM2nZ!Up
zk}YJgO9Zrg5pe-SFVvYe-vDq0EDfPdP1%u3BXl_JH+;t(veST3aU=2a0qm)?ym<B-
zUe(X<F)24G5dw+M7QwXP%4oP`t?>N#(h1VowW^vNn~E8wq(s$vP2&YN`2BA+*!gUm
z1^-=mxv&4DuzsMrArOD@{))~eJDDbjLPCNe#B+d5lD!VSstEk?5U$Hq?YKwJcoFeG
z6;YQ;lRtu9cS?Zys-I}uESkQ)LsSMWC{6=agYL+&Q&k_Oa_f56MvBGi&nxd-06yQ2
z(7(cBGvt{}v9zim8Xe)8<%KoE0?VYTj45Nm#f>1eRZx-Hud9?#qM>})(3g27`Z?$;
zJVaHBXO?wy_yc>^E#YEq#8siyK+eghH}gGaZbVn0?tx8FL779#P}hranDc}CFPoD3
z9~BCw^ydRK{(=2X2hdpBJ3gLo#(YhBPBVF&Zc<-(M7Yl;)RJDRWaMZM)1&6Fk$H36
zV@K^1@1a_0VqA)d&uK6q3t>=Db~H7#iJv;kSoAC^-9N-(PET*!NYX*}R}_?kv3k8n
zUy)d!JFV>TqV*RS*`>OkY8?@=Jh_Ekr;Xp0R9`2te+=_j&Q^e8mM*IC&mcTp<`7<v
zdJK?G1gt(_kGDUM<1X|mCO_G`JU~C2rAAq2_!!LJ0{x0ERwXgriW3RqxKnP)u+cbH
z{ZXe;|6{z(!_u$-LVv?GzF&PwFko04JiUKn56|Seq+A#K+i+xUs9)q0B~{QP0TW{S
z1toGJ&h5jWl~B2;7s13nN_Jh02moX_?{tN+$}??4Z4ozx7vt@tYz5nZqFz3!M{{-D
zfx0a$EymQ!FYX~=gZ|&q8R7>=Px)vh1#@W?NRojgO|!`#Hba4oa>Xp9sv^D+PU5+9
z{T=@`1UR5@g*8J|YliDLm0pD6Uw|c~>*R-Sjb9-s!mfO0Ls)G)!tm_;hcXA*Dw%7x
z;kDzPg6|~xMs%9tO6a$$S`@!PpD`TA$v~}$3oipbQ!?0Kg7I2!lU9ZPnq3r*<vZ%9
zFdz?Vv!%7km)63xG}0WUBfQB`w@2=Z)rBQ%yjU6O)6&_uH>0tQQ%vTbrt?!Ubav<f
zj*)1$>nM1nFK_ZwyzyV9D0^ck-Q*q#_{ec2A;P7_mpYVfy-CNu?052%1gJ;CHaP#~
zGBlSuYF^ZrL<5(qxvk5ark-{olC;<)>)=3$>jIgoH0F>KCnPakY59v_|BDiH4mce$
ziTdr{Bj=}a?zLe^pjtjBF4N3V>D^WO?}Ne%fZBl`HVE<7Q^7(~PN?p{@SyAvgYvCw
zfO|JLYw%+^hU+?SQPGVde%qwJacm0qJAXTA76sqZXylR@IAl6DoH6#d#OO5+&+pk@
z_d?UC5QhDUVYtcPm399Az0gf3-M^CpX=sw}<N;QFx5^ep!6{W#t+QBRhWDk~gTJI6
za}}%;$wAHz2|=pwKOf0ruA~>=##;Tr#*w4E0mKW?j^^p?mjE_a$`+pniANm`_`ngU
zT|8#bl9rDlzt~df**Sa0Pl7=d|Gr+d@RV~Gn{b`|?D+vB2HV~=KhX22iE}6(<~oM|
zX#0%EY$CRsgt5zZ)wWY=s(MY@qF)B2v!~?{Ho^{vmx|6uA1jo#qrTN25tzlFFO+~f
zN3D5JTm=Zp|8=%bOQ*}bVe!$TgF8}k=c@-S<}_>PL%ncUN6Uc_fs%Y|g3~q1-4_iE
zr)*u6@SaCMaR};<z{WV+6pNcZXGqtNs5xIJd-^nJ2Y1kU?JXkD-qS~rvWTf>Up#Pi
z!+-5|gc6V6cC&rLOs}hcFp4-6?w>7{ka$f<6y)NEGTpS(>+>nf<`mHw*t)_BJb%&<
zG<A&3db_;vcPf}a0;uPup`A=mIW4Z&0`0RXa~T1B5Ri7hM<+CUV-d?!D%=ZSv|)dt
zQh+-&EW8EEL!H(7Y$|zaRmEHYWW~nZKX5O!^Q=sFcZb!I%mXF5q7MJC#GI9puTL<r
z)NOZGPMVH?lKIhk{FB)<W^yCHWj>Cs9H7Bz-eIO3Rf4O|lUg5%H+0+)X2kaMK*?zn
ze@w@!&>!*rnVYYABI;aQN5+5s2U}hFn?tYdPQOFZl~^`7yUoQS)C)}~#afb#v|BfU
z-{O8)>ja?KR5+Pp#jc!LFRvk+ZmEB9-bVvfaoP&zuAVI-Qt{PimL8m`ET>Q^EW#Go
zz^|c{oFg=^=~%Nj{xXV<SM0hH_lr(obG0Q0)&bKt*K=8FgZRU<6W7p+zS=m5?FgxM
z=OFCd`A<)O5PHO;0_}d2t(lrYS$usf<3-agE)^ZEU9XnDC?FUBd*G%th`-am*71BA
z%fXD`1_G2wJhmL!VixV{WLm8eA>sv50j5|hnNs!HXI*1{H~Sas{RX$j8dYDI=Do@z
z7Dfv-gQhO88~cX?9#O;go1n%lg0T~f-<#<+2dAtBho%9Q>Ig%|9!0$r8*eAR7Y)yw
zRXtrutL!XvIH)kW`RNf0zqft*y@xqwylQb=|LOA&*-xHKNq~B7*F$=Tq0)-~Q#cbU
zzL3GeohtAjcPx%*zoIJj?i++3mR2ZbU}0E1AD?Ad2sJW=REwkL+-Xt;eL}Da4@Pk@
z7*RNm30fN%qz2PB=O&4M*^>ir`gyQ;%4naTspcit*GpFG#fe8nVfok0Lb)S-zg&r(
zP#u5VspKsU1u8rHZU#U+=-kcuyp)n(<l3z*Ni-v5JVXgv*K6*UDSvy{R(jwf${YJ@
za$&enKn;&aS6q*O^*32ifVEV9*$DYd8l>s|e;pqs{2w)RWcg`0J`L<<D7Al0LK89Y
zeSu@N=E!iV^RT#XxstX1H7P^P(T^|w0my|mwoUzi4)dJ16(Yie5Q4c_gM|}fg%zCM
z`|V5Xrf1A8O~47`NQrm9%hpSPE;6-bAq|&HBv9{vG#d_lU!<U+Ip1XbwZSPb7bRtG
zj|e2}e%zQf`Pk)dUScCkE2rh>ckrG0|1y*m=E^EC^Hf}4;;H^ve%IF2)u^)T!u%mk
zW`1qKFD1Yrs8>&K(>FSeMo>^NDkf&5<o#Lw$25h-wGA&QB|B2X5!rZdnzpU8pEL4a
zHBf)`KgelpsWY9OzUq1iD*0wOH7_GP06g*T++Qgr#yI#dTUHWzNw21?ii&0WBO<iT
zjXa#2R-VT1z1gnXuzVbuS^O?{>r*!Kk^$LTJoT<VnyNESt<AkBH}vQ)3xzy0$){>!
z;6q%v9|IB0w&%`tmUbgUjlCGH4F5I1iU6Dk-G!9=i0X*&P(GI<dGJ2H!KDqBxb*z3
zlbG#AC1@v`HC*AuLdx<!5F}ZYIOJ~TY9O&Xw+C>nbcq{OQujAdLFAwCp0L)Z!?qbw
zD-0ateSC-u!Pc}#C_zOtB7fC)<DHa6Hg^4Quq;7oOyKn1Da*jS(E?XcMF;3!1b4R9
z%NkftYPQSi{2{?u)%UxGs^O>8$ud@H27mB;vygXq?JWOHAB|t~GslCK%jp#J+F85d
zmz%`K)t(%5vlQqE{QOTh$J~~|TfBidhf3Am?{?iYlW3-naL;f_Oms;kxp%jZQW)$5
zjsqQEnoPe*TeETACkM?M?r4c@>|_e$T*TTQ50p1+H)Aq*WyHb!O|UU$!yup8PZPX*
z@aQ#=-%jvM)W_EUWeINGll(}*W;$w#%a)Lmmb#+wNx6DUAkIOJiHS05A-S?JAf$Z^
zX7^E)g=%6h6b&5jDMG2x|CpU85I+`Bh~X@+R#$;ym80@N5WuSeWtT)s?_R56Uxgs5
zoU+~I$h(!k9?ef#sB4^h+(4V~4~_0NYt0@;JUepne{K=qFqH@2{)F|*Y1)U=0&Qsw
z2=AgKz4fOLX!%#p=;+*&<tTZ8_|~AQ`QY<G0P#rWw(vbOv;CK(@O4VHNn|*}J)*B%
zXV^No#&wL|5BDE57JID59lrT{)6#N(X+JA*$TBTlRxYGfN>d!^=Vmi0lPPcBK;fNV
z^~I4IQ7Ao**U^2j)nQl;ie4acJSY_$W3RHJ*~9tfmIzCx$(5VBy*=!6IMXTF<jL`-
zzE7Q0IQ&1o0EI>)raErrX8L0gNB77%VFgjm4lJ8+P^YKZx&IaIw&u6WG}OENNoo8b
zch1F0-SwWMsHne{*-S7naB{6{)ICN@h<GyIjY605!8mpSpg$?*xkLbr7U-+ZJ9OS5
z&mH#57O@1y05vSDn++{lD=sk|4WP7erOS4>>+7ZS+g|^ramRrm>~dH)-N{PK)tN+C
zZ@G8r!*<xp`K8nu&N=@iN`+neN1~v@0+x1<1^LbZ&nHY^nZLg*=CwRFRe>hU_ac9~
z)4&015Z2?<Qi@9-WW&iz2GA^UNWgR}KD$`JD%i1nt)Veda&O)w1rUMG6^AP;AWfpz
zYbkL!>gXB^x+%P&3?=vDsP)Tf?t}vwgNGm(M&m8iQtE<qe~yO#ze5NCUgqFRZZZf%
zT-E4eL(jD`*vl7FkXu#8Io|L1DL*CTeZ)@^=B&294PpbpWApP%!|0irw(HUPYzhMh
zJP#>f?=pri`fR~xSSc+Nbq6)}@O<N$ea>Y}5H-#-XDil190x_Vk!nF<vihUvAUiR;
z_|Hp3{U5uR1vQ059kD;X6M6+)O?Eq5BoM4?6BldjnnHFqK)>QRN^{)YF=VpaHJM+w
z`3g7YTqc+{vKHPslg&NqF$yZsHq=A6LkIe+z-X@c&|q;ArlrR?l7w$*C?jg7-hqb3
z%4|;Ja;Iry$lvz9MU44$TG#%!TQpwX8Gh=JuV1Pz1cf)Zsy%9G7QMZ8m&olX)B7kT
zF~k{}smg@teP*{7py^Xrs_uV}%DMjwuVjYhm=VHL(6)_mDuUJuP9Hl$9wv|J<++yq
zwngxJc&L9$01_>c1aMi0tMewl$Hsq{3hGSFAp30i*)ul<9~JLBnSnlM`z}ZT@?Qd2
z|Dcg?NCeeY1tBE-i{P%BvWQ;xS3>m~D#EbuxZNxAC{MTw{D3Ttyt&D|B<;;}_J%Bd
z#c5P&uw+68ZC9mxhBpGkX@y?JY@IhRR#|DV?vftR{CKjuZIdj~zrKn5|MFXQGK(wQ
z{kXW_pF2RIAcypBi+VQh8cFc=ow^SE!P#jalRMRn2VcQWz)Jcx92zR;=jH!b*M;KW
z92_XWp-a^3%ui(0J8=NAh)Vgz&Q}nyJk#FMP0ou$XlQPPZzWS_D7F_=RS~ckeJe5w
zvpVY?Cb+R-uJx8&I#-_6HdZ*sf_m@I=fceQf61WCupwjSAeRYCW~Qh<*MN1VdB*sI
zJ4XRDV0Six+2A5fCjj~I+;)WjA(E+}j^Vf2zUUHMdQ`?u$gK4zMQUpTImi?OlXx{{
zw3UAS5Bkf~=6CVNIU=B?Jm1O{rs(SZQz%V70gDcfwSaD|kw&p_bZw5$sCvxfr5buP
zD~X62Cj%Vj))th-!vDz&pGNKB?;n(Js^`b}tcU2IG*_`U-)g1hv+s>+-=IYJhJq(_
zGgdjt6fVrWBHs|ow3nB<Gb=OJ>CHc!uh6=Sdgc1?x@KHS&QfkN(7#`d%r_`D7%ZZK
zMwP<y)Shb3cOg-B^`>pFIR-`%`hp<Ga!fUxH(A6TPeoesS=g9N1d>e+uNm99<HhdR
zclV%d4Erar2*eLpII058o?lg^c_v>d6tsowN<dZifn4GOD?WA6m;u?j)q$Yic~Azn
zrQuPTi`1k~)Y38xmCGCgaE8IlmM6=YoZT$q_`x3M+6FOaz2c1wHV90qs0e|f)24z%
z$4{AQ<{r=%-u0uPH8oeOZa9$2nD#KfmpMK2{=xI>jS>g!-NupMUcmUdgU%mqbJZ<N
z2Bbxohpz<^CftQ|8iYH8lNDxfvMoM98fYeU&PeV&z4=W;6KSH%RV=&0)&095vD1}5
zTO*$~7IQ};L40n{l+hmqO7dlLK(_|6xLT&>tv9*Ul~&t)sPdHe`*HNYxSw=J7V6V8
z?QTbcIQ)=U?CueC4vf8~w+l^ZKWS)goPXXY{sv=L3f!Ow)_tu@2B!D=Gpx==eHDKQ
zSJ`TuQC}MQfn6cWAXB&dRSbfj48>N9&A+?fwg|2LNU5Swp_x>GfHsxhbQVlqYJEB@
z{*GXem<2^pdaO*}3RUj!<NWd<C?dOmunjJtV%N>$wI`1UTbkZF<vqITC~eRLv=Y+R
zU-bUlU4YcCxlrcb!O80OAI<^?>w1$Lz|7%5;TB!RV-`wh+PR{@xw-lQ@@s)#QcucQ
zfb8KLSRxP?d(u3SR@vhSGQ5&|7p#xb#{A!l1#F;Y|NJY!Z--bI{=H>&KNYg0>ps(T
zmQDY>bbq3xG6|*7P`=_k$9rj&gt@Hu+0<0D^Q6}C_7cIRaK$tYoIzl30_ESzXhb+4
zZ(8MPMKs;Ve6EWs=G$B+TRrV{-&zCJ_c9nCMFR8w1)~I*!?6DPoeOo^yGuPsk4$ba
zRzWdCMz*%JHnVMYwYSJ9De>$o5fIN+gMjS{sC8vLIh;{Dpw|7Px#t$cDd<2w1sy|J
z!i!))kY~>CcTwQ@e;`qjKR(8Z(8JM6twQ6U{BvOKc33Hw;bBKiOQ}dU&7=R+gLg#k
zSSO#H_+S3i*IBwJdv&&MqE?G3Zl(3wKUpwMCP~4h0J5<*81tZGSSeh0>XnsIYmQ!o
zUbuXr4vp;n9zo!vlJdK!9~E?Wwp-lDK(La|W_IdMp+Ju8MS5YohR(+i3yB~Q2CX;C
z3GEjaBM~1W;Q8pVfuNt$tgNf7np{Ix3Maec7iy4K3QPFz-qHZaiNp%RU+Fr#(j}&o
z*-+2+qVZ=#PY`;njD7+fFgd3he#iAfI6#Qx6&*PW3mxEJ(edAgH<$0UkNlS3Rbzg=
zvEe95FxKof=<`t`5VX4}_|67u+%-nCI=DdEY8jZQE9?BL$T0jvd`ZyOL|xlTz0k==
z=-|x*vZayHllFb-s}kr7V!hiXwp7glWfPH8Wyc|ff8)G*{r)Da4T$o*bKqv@@!zv)
z9j`l~=87#@D9j?ht1R(5l%VTSaAWr<C>HKi7nAZ^bvk>Awte=9T2+{g>zW0`EkDSj
zQor4G@K@-(F{4(#p5{y?`}3nZ0TV}LtWwAHeA@;ji(3~VS>E6EvbQRkM<+mDtFYX7
z)&Be|o1?=hyHw*cKOERod62G81w9%ZP~7KPmr&ymF>zKxUf8_=WQ{=mXm%%w<ByPc
zq4B=>H*=zp(_6h%vS+}aYdy7SXg0+I+5VoWc1C&z=KTK!D<ZD&{q73-uXhxmQ6Q9b
zUXaBE*k?GP_X2e?xG70XSo|F!h(wW4uTue&;F+5tWu+0TQkf+la6TPRH>blBGBW8C
z^JgJ%_xSn;dPqXd>+hf7B+f*QS0quWSP$y>0**GUa|cZi?}G19?Ss)*2j4{VR{%-K
zt#|F|pfmVsSUP8f%pVW^^gyk^_4tZB5HPRJLLf@I`SCsPyEswH28DE}1RI4~H9kSd
z^;1Y>Q{*m><lIdV+?O-d6suKXtNnGqtWs)GqbKY}tl$fq!bx~vJhzW=(7<i*G*~&<
zW858z9<CvD4q9r=@=)X4kL^L#5LE3ro-2ix2U|fqbmULBEem}Mt$YHe5<tc9%?dYH
zRT7Aw@CC6cA&)C|gWE6-X8DXhLh^Agn@6ON1fwRM<_0g7@y<|NJKFIPbZn5VWj^|G
zi40|-s-awYfxMFy{hTn-&5mHs^s#1ltVLXKGR<WG4I?9C?csgfxyGMH;ctDawSX-s
zy*5@zVgc3cBVv$=XKq{DmYvI)DSmRG@aJ=L;}H@P!o_Wyb-aqi!orf4ktwVHSy;$`
z7v#o!h<#e#*vR*DOa0lc?ah+gO!Rv=L1_>l2m&$u=8npBv(`v#9LxD$Dfa7DwNIEa
zmkx_5#GUyeG&qVXjhBPo7%wXD8C4yexJ!{oSD<Ti;-3@(2{R4-;Zm|c5=H7mZx_=l
zc|lBIhhsD*ZMVpNor_+s36l<Y<>3ZrvwFAJBgJ_k9x!a7TyV7pS$BGa?9hy(7F;AB
z?DKqI$c)sjb?ZuIvY{W_K7Om4TLRz*Ah6(PFmtv+q~PKVDhn6yFNQyV{sg-6J{5yn
zTjYDJj>o9bZqL|wayg1W|JfI<buLe_@WA;zdd2)OcS88Ej;_ycVO0Lkh}0dNq-m3c
zeLna#YBW?@Im0aPV-p)(?AQ_vItX^0+a!x*hWEAGXQ1}Na=v`??1WchQ#tHPXIJ8w
zT$AT_1D4F6*@MGFdU~ejzJ`J_NbEVYVZ!O4Lrp?tq_Bnt5nMz>gxv<`QY$ee{^`>*
zd=}HpZ5hvJ%&F$@qVH}^v+HYKULOZdTwt_-5}ELt@np6TWHcF=?i6g0=wk#zQ-MwE
znPYoSbx$d{mV6H!bQU){+t`0QnD0Eul)L+1p*G*>i|zGpN1*U6CFj!>t+}rfdfn<6
zsPB%vp7-4S+?Uv<1%<wj@0oy88qZ){;oSCI`AB70gN(2}>ID5k?$7Ex7x}CqruFxJ
z&x>I?fDM4y-dOo^tFzp8{P-CApuw5WU8r+!`XR{HXJaNtKNFg>po8CQJKS9#p30@=
z*2IN{LX}Unkm}}JqPyi$6fj}8Tz9M`=EzXlAg0=Cm1O*4q&HK}_(D=yYO?=uXKd;b
z80qPn8$Ic(7$jMGYfyCJl}||%fv|X2proOHl2tg%uGd$?(lpSuq>ajE_u<O9+4&p9
zO4~XaK8FM0Mm{6yezVpY--@;oXjl{Igc7%r<)8J{&zs_N;#}8G7sx=~%k9RA1ss-g
zLRGn_UUZa<i0h^<vcATmfy2+~Umaq;hB36B<dVNFI8c=uLQ>@S7_YUWB#VH71{(!@
zgu_aY_fmF$9D}D<aAAm_+6u(WDLGiMP(yshyn*RlhbN+Cu9CI@4J7lQ+4R5iM=65$
zeatJX!>r&{SxcJ2<3!-j-$@4MPju22N+w=BP?TEk=*NoURAM;LhpA*xY&lslrwRVC
z#l2+!sF`h>Inm%tBz5JhlOow1AFy;Z(ZsBDpMhzv{m07WM5U#9=5Zw9`s@Z+9n_f>
zpMx0|NNaZ-TQFO%e-guj(|o$<^izq-l~27_c$h=@s+H+fw0>=33Oz5FufU$Mp8`!z
zBq8$1AMC10c}Jog)Vb&)j|>HfwR~g$v~YN~7U93i{#`BeQ4qfbw&%k1HiLEX07x};
zZqVj;(0UQftL?eH+o~<FIwrjt;3=yag+aiGPo|ha*6~NNc|3kWG|9Um3A}Bhn<nY3
zAyc(V_r`BE&Mrt;CJZGA2nMhWkT<!W4P0E4k_wB}xOr8h3PLFQt>1{ifW$*t6C0wN
z8`2!LjTWLoMRd)#WN#m+o+^|OoEe8NGJBKR#vh@=!E0LZtIDWMlisIlhalu}%llE4
z@S=;!B5qrC7pM8LiU3J=IVxs1{+YSC0bw>*7ey}}f8@g_fY^a7bKSMZP^AX2U@A+X
zC*TaM?i-;{U$tNLb8mLsbT*lD^zzd=pY0<8<`x=fi2$U|X`5+s>!JZ56bBW0ar1B|
z4zLH$_OVu4+mb@9CpSI(y>+s#k2dDkW@h0HnRO9o)X8kV{J{{2pZVB;9`;z{l!M5v
zsla;wciWKnx2?B;m|@^l`Fv028Zt(h;NIiO_375XEnfusLnyGY*N05xa%To<)DwvU
zb2Bhh^>aZOb6JhO>cwh0H%Q9DqJ!`!7Wj%}N&KoSpx7=?OE@(b{JptR*#fgq=i%eb
zG`6iGaA0`v>#y%f6gRQ8iq2CiPv6qoqO(eBBxyxf6&~NyK~J8)9fz-I<?z#EQ%041
zt{pdi=`2)=w$7}*i?k3DFo&@?yaWF<(3ZERZE6GoB+%aq$CK@}=uOqeTrWC(&_IJC
zdx##H@bBd|jy%0hPd0tDm;Mj{7NccXJi1mEI&B<>uY#&SFdByNrqQvA?2WuZPKfml
z?e`eWfZ8x;W(8&D^0Ki#Q*LB_S?Y`0g0E)<*Pf>FQVYv)AkHF9+tSb$2t1g6{UJ^$
zm+2JrFqwn+PuOn^;*ItmFcA+y*!ev?1lMjjukm4b7K!H`|8fiadBxyy)`zp;oo}MD
z<!)OJ+iim2JfL3rY=*ZMcT+6zomy`1{j=GVHU8-mRHPs{0Ed67;Is@ZwwA`3FaTwM
z_!Zb(t!|8`o>9P=P9Mc0HH_3^Wl3j<z=Ekp=OiErm|(wgQj(x@m;y0sCtiO$t4aRu
z_XU!#MXT9#;Piq>3K{$BUtqAsKRE;w$1gPNn71s}(rgBObmbbP&tAX2dgi#!aMwI&
zF9)XS=%~e<yepS)&o%fihcx>L+zG@Q2G~tL?|)mBn{PFP$jR9;f<oaVBw4<cn5ktz
z3R^eO#CMq@3`Y9$<A$+DjeNCoo|M3fp$&?^*3HO2btGIyP0gMn#xJ283s{lPMPVcr
zL~iM79<1cqTc4?2W7*1A;M;oM&9=3-*EqUI$TPbTRz;^IJ8q<?_U?=v4!7p^vU8@K
zB&Y+n{zveU5{E*KypJE`#h-?LR5i${sF=xQ%r(LCJH;$)f;Mdcxv1lWDX-^NvrVcO
z%;4NBK>u|4WTMW__N9e&xbGOrZSA710~18#<=%kJqhK%wP%Xb_VoFY}2>C{Ejj!lc
zsmBiF+{kKuEqVJfWi93&9V7KRN?k42!%w9--~Bxx43yVQ_%Z=WA5rYR0VC|D`R_WP
zw4T*Qj!DYZc5{5%sqQv0*$}w%5>8u=z}p1UE`708Q=!!=GJ03_F}v!-&`dES)BE3j
zZK*A@l~sKGRL$|hZ_h>Fv9~K5K4k`Vf3Cmm2@4Tgsk(-jcut2xyW*;WuCYm;11;!~
zAU@x!|L%Jv0d8C7qqimzbz&gWAS&kHsm{vuKHW8Vp6g(Wiz`-c$KV88p=@bYFYLK9
z{Aq-TZC)-LZ$R7-J{%hxD}}>!`!I_8Q3VP|ua%h7mY_Q_I#02Kt#Fz`iomi5-#7~@
zvgNBsBY!>!X#5e$Mb!#H;99yl-I#?kuZkSaM>J;HE2pjkwQp|%{SS*P3N9JJuqO!}
zV_hQ~2rjCixnS)QN&Qra<gN~^P)(HH=H?6+^$N?zDU_aHN^iZ(E7|0!?#9Uyy5Jzc
zKV+*i#d*CM91<I&K98HBzhmocOz)P$F`C8wWf&3fUA$?V2rPx36%{zzP`pjrtjlVS
z`ldOC#=TKN^%qr!NI(eXYHwZ6-a-x8y)c?aW83@l;>-&jMpa~MS&?9NsVy({hdZ-v
zeh9I+=WE$qEXO)bF`N+1+62U<>ZaZtRIF;qzQ!DH5?EP)A*lMz&*)F~YE1bK3<$75
zA(d76p=lmG2z7LTFf6FZV9K-N+T>Op4mfZwst)8wU0M}CP)~&RZm=HHab9%~6h0x!
z^#%;{Sy0jwAiBR*=N1Ji2#WgFCYbJ*T!y0pk3~II?2>h^PZ=t^thpRVg1IBlY}}A0
z2DE{*3gi!6<f5<-X^mMTtZ}YR+UAQe;QPlxLw9Hr@*MI@$xc{an4w!gZA63vfjgju
z2<8$MAoitq-bu86Xybf<hpxo|<2|DOV3>5?ng49D8w*o+<*Vo;H-)}|B2zslKbbP^
zYh6T<`9f2-MPbRSl1QXT@hn+tr`RJ?kv8uX5d8OcC~vjx!5}PJ&UXQH63o9H|BVUx
zI=$hgi#^pK5*-^)iBjSius`B*apLltXvmxz1CvLMXfFJ1W#^{ncF+5t<3xiQDi8`V
zkA)Tjc8ixsdqp1e3Rnxgf(2E}RSF6ATf42c!j){Ylr%Dc?3V2s`!^kDXm#Lgz7g{m
zr|?zyyYXwdD)cYVv+!RFvO`s<UkY-5SHHn~bNIR#7t{$`g4-v6LWDCn-jD(_n^ASu
ziv9D;uKK6f1h48s(rGxa4HWiNl*R|DI{)h-nEVq;ocQXEA+vvS8tBbH;DHCYhSY&#
zRrYZpS`lJ<Ofku`78vHh`g6-B(`*`%tL>yz(@N{~4Fu-4)aMgptJ}pNlWZK3b-l~k
z-(~uV_-jaFF4%eOy1%Lj!reN9wx%7vVCs!F%+Hn`X^5b|wSD~Tm#x5u)BqW|f)wpX
zv)}x;OSNz1Ovm5~KMBCE6Y`-xNqYDo>Fw*<)1vK59XP>{MeStI=B7rMmqWO0R=T#{
zZ6&y#aM44gk885o>pcYNS)-3g+PerZ5l>w%-`zY<QhYZWQ-VJdgdmo}R&&DL=rj$I
zHCZ-+Q>oXfM9MxX87Qx(Ray8V?v*7)jluJo@H>#)=r(=7zn|mGj*a7G@bPVm1}v$&
zp*5RE#>m#bg!sKjXz|JX%4o!xy2<%fP*xPwy<!}!rK2O2jm8Z(R=W2jP=492=Wr~Y
zeTl3!StdJr|Dw^jRpQH%9!REHQP2KlyVQ*j4s&&)d!ky`JTeo?AbX>}U7^~_6<8ff
z+>FJal%Q};tkry2y=TICR`f?54(Sx`Li{}Dx6}=6WaK%^G93MS|L8{CYU$S1dBPM_
z{488ZtR88VVJ_5b7~28frO|xh<kmE8(d;(;S4~K<&&b@|Qirc!uUb|~_HdpU?;EG5
z>1>y8krZpySFa_6xU1dbyvI$or>+^Z<CEOd2-WdZFdfW>_hhKQ(B@n$(EjI%;xa@e
zbS|Uuse>-qRa9E&CS0rnasGM|W32KKbe`qIUNpbRv1`F3S7b7e0e1vxvl*LBqcq=d
ztVAg0-F%DhW<!;pX0gMhbzE%|E{ALUA8rA4m&TVII^1Y^Zl=nm?wQ8{#ss$;IQVyd
zFNnHsS6|Y<{PnJfVR=sU<cfpv6T`*KYTuZgrY<ttW<#4uuZ+A#o3|7sEUgixwQuGB
zUaDZdm?0|AX~*#jb&AU+o6tnZBsfFr)n~GtQSPJYtMEUJtR9@f`g1AuK+RT3sonZ!
zHCpT*GJ>P;a$0+<8c_c}U&dt3ezm%!_FSYPFuR%+2L8fARMs+)(~gemHi57D4Hdb=
zt>M0#epskFuO!L!nEp3NrNrQ@IRl?TtTTKC`>++NSG=KTB)Cu>D_6+=Mjq4j%tEv%
zKxx0RrbVoXR*uTt+2{pce8k$H+t-x((SNFA0-fw#ZeNXOoXpKK{AlyL#%GYew|l6?
zkndE;X6C%)nje~V3i`5WQ5P6xl6I<WZNItQ>(FoD8l*J4psKB8)79}w&ixJp5Sm^P
zKhhp?p^r+=X~tW`M-6YjJ}1NHB@7zZO|=uC+i|{_RymA0=&H8eaHM`@w4d`Oh4bYT
zRpXi8reu^WtO_;b9z~6|h^UE;z9P)MPr0#i1hubt2&NI%xtBMM4Yn1?DYPGMBk~5d
z8-DI*%^x+Ig-oW4&UcgYf4$mDZ}&W|D(|^GjXHa7ezW~iN<??|1S=>+Yx<Fo-CXo-
zH<%DD+>UJ*po{lQDk^Dl6+M(ux@~_9v$#bxP?XZRGGs5aM^mI&jM+Y0koqM`{W+EH
zC~GlUmufD0rLQX@Y%<tV|Flr3y)f^Jgzx3~3TBr+1{7O|+xm4mUu#Hls6xuB$7sZj
z%~RIdt+9PFd_Pf&KP61k?(QH~xPAP3H#k?(B50xc;e$EiZ}pbB=^Ei$?T01R;rh8q
zAqn5xe!IpR6X5qRVkg|rIe?+-Xmb0#$(*vSvm1CBw7k@0Hk9IV&$=SdnjdUvc}}sj
z8m1CZSCMj5w?4;lg_lF!q>;2FUQh%M&y~kb`Lu-5jj}>LtFIj~Dyh|~i|FBZhL^{<
zsNrBd_N~^YVvb8SYb+13u)H#S%C0MnT=P6{Vy1dEUNbOnur0E0Vkn<t)TsGg2@9I3
z27B<Yj;iw93hZccp^;l2oN!TzC7>Bod#gaXq}G8ln40BB+1}#4RIJlnuxR2dSbQyp
zk3z5hItmSY!4<Q0k3b3?-Gkz2|L+7c!8tmk!NJ8xUJ<E88vOr#G?_VfBd>!WF3+s^
z9_WNnD3<(RhPU&(*na3qJRTYK<<701DuTX2yl){Z!_H!sp_yGXEfk(cSN3M7m$m(l
ze``tyj+-VYvjr6UC_HE0)pBz#WCo{d&kUo@BOtLgE&6hpXZPX!#{a?ukM0jRA}O)@
zojWyJ;X3(aJXouCCkqra|AK_mkkEo~*U5;%<@6+GwB3FPQ!}5fDMuJFB--8^Lz>s}
z&4!On(oj@)dO(W_WXCdf+T|RWxu`g5X1eWoa!l;E#U1FS+4bhQjRP#UI)w3d?Ipob
zGWRJrcWj=k&nVuc?)w+Di&8_zG~Lxf`LY_2!RnG1bou&i>rGFRYwSCYlTzCKHXh`|
zBEykiyr5@0hGF0`5X8qbBAuG0CfAg-ALVX?$0DW;)PjB+^cI^8%4kQ{i*6k#eihrb
zhz}x6*}bG&ByxdA5{}INv7h$+B>t2B?ujJ=y81Q3@v(cbWN5|lGcu1t@QciNYdCpw
z5ysADEC^;Xg~4%Zei!>lZH1wzJZ+bsWS!BDg2Q<-JG|(g!KZ{ixcdVHRzaLL3$Nh1
zRg|Go*&^jlc(}KuL`8-*krvN1X?gGNY~Af1`-@<ql>A=}rYvA49H8o#vxoFYFyWV{
z4Tr})VreoAb0^O0uAS=5J1R|eX4GspZ(vt(wnN6KI~s1DrJF**Hz~gVN7GftRn<LR
z5Cx<~Iz^;Kx?4(Gx=Xsd<A6v=s&q?tcXxO9rMtOwz6YQG`+=W)xWIAmJu_>qS$ob7
z6JFJP`3of-XM!&OI@bWluI{#alF}g!)OTE$E`Ko5fhCMr5a?A>N%U2cm_>-(tPEIO
z#l<AwrxkItiMXS-Gt^Q%xw*Wi$yuALFv15H@I4iaKM02w`4U$NN5A*-ckFQ4N{4x+
zl*9~X(DaLm(}fIQ5}0m^7<n@Jo4M5@cxa|-!JVRAnXIYG{c-!~Z?KR4tCGq$ebAFu
ziq;l5vu`7x4LtEBHuI7)AzDl>cCP<dc*4Yx8ZVS|bYz8u+NNYPAhfgx#m-1F;W{Nf
zdW>3!tewh+j|<X*O*#=ex?ZFTb`Ly&fRfVW;zAis?|_7G(U|_Wq4sJQWJvdCY_duY
z6wPw%cPziA>Uvy2efS%`K+Lg<+qo`fK-s;ueB0VsgBd3L2v%i#)Anc{yxFDo_A0ch
zoF!^4%}=vKyCJlag%#P_%BeA6>u~w|8yA6|poY{AOLhbHMEL@SBl~H*2si&^CnjW;
zvnW)C0UXrA-#N^1oE&V(9=ay`>uiiOTafc4cRIcth{j=Da6i0v=;LKVTAM)#U=cot
z0^dZEG3HvO_$Dm3`s#^tLU);nnvO1`v2nN3D!;I>ky?PP^Fkb4k4%NUQPYHO)yci}
zR2yk)!|f2xeoL0RPaMpVeU9?KRgVG*7j8Jr!n{a#3TSf+sXNBznx?(Wrp8Pw<7B!W
zSWNbxY$E4OYcY<+YM-h!HjRD(eWnkA9>$TYp9gPscrt|K)Xm<2WfN!QbF&7kM%IeU
zV`Gus<E=#%yUb%V&(rY_$*lY<=#;{Ax2nakwp2#lmTwHFYQ?xITbMUwnpp)um$smD
zxC=JA{?z{APT6X2Mue3=8cE5#Y7u#+ZXN>qYP>`gZ#jx+HsAj+B~k!Zil}&ac#@J6
zAt>-tbF8g~;rsq`mf4MsL`5ndt3%DTOv=j2+Si7fITP9wq-%)#$|@=^^d?#X{k;O(
z+3GEmOY58DYd9Wi8U^DyGbi?!KPs~=mRC_4y_kjQ=hcZOItEhP8$b3VjoY3h=DS#D
zx%C%Hqk|<&;5fUthXKIz8E-T!Fi383)@c^P(dDV=>uf%xo2{~=0nhpe0jh_^gRiI%
zZ!(d~=CEdA%5e~Z6+kUaAwyCAH3S>;n{+e7Jp#3T%P#PKzWMpImX?+TO}`8*475ws
zH~={#@`CYO!}4f6l1GTj_SC4f*xS>B46f==F_}%TqqcD107>5kzS+PSIG%AkCAoX<
zAFMle19bfigKJjiGa_1Uo0~?atOSl4bhp|elqKNlFfuZVijIbrY-*jZ-o0P?JF%PD
zOV{Zno3H+iHn7XBhU5mHHd;1}lV1rIjje8A2TynVmVb8yWBqrOYmgDtn8o%Q>j>E5
z(^|B!X#hsch0$e61qA2EP%96hf=y7V5M6XZ*M_5ua65WCsAT|{leqE%Y3QVT+NMe%
zWeOwO4KauG4bG=!3pZO$Ib2@%Fn>QH869s$gf$UHQ*aq(r+m9u8&Pk%cHq8fj23@<
zcnzCowU?|bY%$l20AqGaSQwm9LQT29@rfu&3phnLpTo%e>2Ob;Ku5^rZ)>CX0EfLt
zR)#`lgR6DVOF&r%Z~tvTXu7E?XmKLGADxx<q#70hHTwl$T@FhxUvk^g@yvwtmk+`w
zGN$(1@1B*pF^tUC=X(jP%o$^?E~L^XQ|Tixh{1S5i9!=67A%NiM<ppcqVYXk#7%Jk
z`98lMwBL>B=UhKB=_jfWjRs+-9_OT8^~=p$o#Dr3b9dgG4e3sX?1`LZPy`CFPvZo6
z(f|W-hhGn?0C$LtX%FGOb;<H)+pf8sUfyPGDqQq=$34aU>#2~flcBp2J!bq<q(iYg
zG8V)G#)EW!NI%kqMEHYWhXsVfdIWHP<jX=}f!zPJeh{0a>j}UCD_!&BP@i--1;Zu`
ziBR%B_Vy$sl?gvT=ztHWsMu@-(lq^2gtcdEz<->)%eI>6#o<1#sTW$@P$ki1p1dZ>
z<j=XmX*t!>iwW%hRB|mnOO_x<WgLGVh6+sQjkYz%@Zu0Mx2M<LBk^KiXMmG}$eMqp
ziTuCPG!`^y%;I1o&^N!_+&<kG>8l8Iw=8YqzOX;F63tm<7<V3hD>CE}$8s0TY_TT@
z(Je(^_Q9DV$$-iTzeY5Z+536;p@*KT)?o>`@+Z%BByfpAzi{=}sV8x-@u|xE8XN&r
zv4M<Mlp<5$jjq2994OYn7_*E?nS5g=z~euqVNFQT)hN=wrO<7;`(Pdx4vE|zt$&&v
zz}X<ydN5@<VbZwp`Zc>*VsAUO(0H+Edi0>rJ!E4}z~a`xI~mDVl4`uX2Cx|3YAEH`
zdtE5%>U4tW5_WX&H%@20_XM69JGI}Kpt2G6CE8`K>H98V*gSWV)?F5VFdIF#jic{G
z(qb>k{x__*#H(n`g1p(YKBQ`|>K6yBCnY$eBha($EequJd%ao)Pq%<^W%Ya4Cvo0V
zQ_7+F>6YZMf9t47E~5q>z96wmv#zVcheNu9q$d2EwG{J^2Eb%}rMtBUXUurTAPQ^q
z^wPlqVB9IlCoUN?jXyiLgmLz~f)Q2!>D14}KoP!#oC>qjYyxRY!}<9d(W>JYQ^Z%0
zGi8q)cN*qCa_bI{4LNYWZNlR~*A-i0$>=WYK|j_%YLH8tHpM?XFxNnLt(=~lHZMZe
zK6VXBJbHKmrnWyh&6%7PTgr_kQN`<bY16rYM98w>V(`p<WyzeO;r?O+4c(WHwxyq~
z!&86{*<Q)^gP+}{lV9@9iKhFKGke@>*!ow)zZ0{GCuYQ>oD$Pxr$Y$QYsF3Q{FVq0
z1vE#G$;kvMX0VZPW2#E0`v$4<>WhvlHMq|uy)@=ODk-L*qcV6eXYzt^=ad5A^axaM
zECH=7+nI7FR&%kLox+7rO<QweeDI;hotY-N9|ue>4jv6tAr@nH;}KTILw6hoxwHo3
z86{)<_C8%)It)jtkFZ0UPL$A2=V@=&sNg?I^DfMZT$S$Z4S+X6m`V=-(f;n2A%)s8
z<oUFU5-JAAN9p-PccFeLz!CsY)I5r3=vli$wP$&w>>Yc|p>9!|AuEWveu9MYE|x!H
z&alu)z{~9o;dUs+=1MBq(ds{w;v)3eLo)PSADdfX_8w+sW!Jc-eeEsz%Hi-%-qJ_C
zXGQQ0f6)uP%d<?ekts|{zu!(a5Bc}2y80MLrZ>a1Q97MXeALndfWb`7J^_XJKS*65
z^p7)&kgK;?eiFTztK4DwX<T$W;|5fTiI4mwg)7}?;9yFWW(#k5n!CbhUE>FHiTZ^s
zsWXG;pYg_xfpGeMNy8p=^s-Ao&D@Ve2G>7wYWLIBorX`{^-PX?Lr6>BeP|>HvAx=*
z0#CU0J(d?BdP-q1Wc+rwhaI|o;Qh?)!orA0nl4zZc&>_0@dB@GR%hFiYj0|DKpAB}
z+GJyv1N4M!w*DcO`G#|RbR6|Uefu{<T7}E%%ASkbsWJL6iH3ilVXa|fT7KyILW|Md
zz@fxXP2UhX&Z1ks7jZYxXwnkfOO9|K*`Ye<6F?$rVge_S*;)}8?LQ+On|B^Vs4q&%
zi;2MG?;3c=w7Z{EYFF9dc%Y2Y!ky5_qCVR9>UP?oj1~Jt4bfEj;MSV&_N*U%<liOd
zHS=F2&!k&~e6Gn2ALh{7tscU}D_p;}P)-_8Rqe3WtE1s`pnRwnCI(*5^pp#k)&|tB
zyFP|nr*we9bb`3|BT<*fl9(@-x{plY0s~~&Fr3^Cc25<@4g))Zd&%n;trJwgK%&?;
zQ<16%O@+=qKqn5g?-@MX-PaGWgT+k5)rGj#sa)jclU@ma4RPj$&U?1cI~TD%9?7)8
zj$Uq2$>*_Eg*NkQileu?%dPGJ-Ev&bG~vaO#SqK=^J~NWep<V8#B6A<D&ir(uhp_N
zC5M>+h}t--!%41xT(z|gkw(hPndvZt+*WT=6$S&-4Z-UuF+^ThsJ*(Rmp=uG@@=ls
zk&uBY2&YN3cq5ITZet2F@))7iwR?Y)-@4BGKwjcU+h(en4Z%3#%>j|CPYno~Iqsi^
zrxC-8LX_FN*$X@bR-p-V?J`F@HQRQyd(=fXdCM<+|G{#ZpZ^Quza{w_;K7*Xa$B8h
z3{)<AL2gY21|XWKO>tQ(I5H^@k7*bG7B#t(RKUA+7&W9%`X@O9#ZkH)ELqX6!@jE+
z1hj|ax&Q@d4LY-2oTuj&(C43cygr3Xce$4kYsX>6d?d(FC@`DJBJ~sJ>up_)l-E=%
z<ZA?wVT9gv6!kOZt{)ODUDM>akT&GMg1||oM|r-tMu<k@XsHfOs)}V93z3**j~)C`
zl-XHSYf742^j1i_XUw}71>7y&>pBx#T>D7CfUDx_Wm=Pl?ijT8_jUE_wJ`ymt<hk<
z&QB1bvqzD^hBF2NgMCBsYoGAN^HGd@o&cR>5er-QE2h!>cjH6#f_#hm9wa8tq>=xa
z9a%B5!oRs(1kWOcapY^0R7NG*ePfFH>&ln!IC!SkytMc{<G^H3DR6-np9kdh4u8qc
zq(;Z(U}#MAUs)$LSC*Zrz_EHC|JeOG6rN9lD&iM-nY20og&ZmHxqW_pL*c3<VB0uZ
zG^46(jBhl5QOh&hW7WLIl=@2_{9`?b&@SOp<K4>ymsCMzH!^?x*HO6H8PhwZE*Bj`
zEyM&xhZJR6o|=CWiKM1~D$xKZ?99<(GM#aWL3j>tvHguUQu;DE3t)=E&`fky8{6+k
zcM`T1;Dr5C&b}+0R4L=gREYVE+ow~Lx5WH+B4z@zZ;)jFNi)&WIIw>!3cOe8rRHYo
znEIuH_nuRGy1E|T&<lFq6&Y$)4PvrYVl$jcB$y(VkgzZrAbA~>U-~6ON}bMMDtoo4
z7IOju$}v%z*gYEC5hWEZ_YYPE$H6uJEHfjiit0KN9UrJZe%sy?6wKLD<v|Tmj;1-}
zvn|vOi45VSX3Ce$Lijh}uNeJ*yWpD`$N;k`8n4nr%b?)iT<><PLA<Y_sud?CJ!w9i
z6a<hn@Ss*{_yyJo=pWXPfs(~64<|>jT|VsqSEbF{Qv|TXCM~3^{Ifxo<#Pwmc-65j
zZ{q`fc$Pw*E%YSo9mm_qJ|7M3r&!(XyC-h9=?6bQ5T<^x=8@J>z{#o;o#mceA0JE}
z)hT)%j3QoC=BbhySwlC<cy)E-V2^%Q4sgeqYJfq{Rkx04hA{FqN!h1QuYIludwJUT
z2l_#fUqfV9*?~5(e4cU~^kSMy<!`6$iDg_1!@m}EF+0t_=~!W{T|>wo;<!jG;b|ow
zg-k5m$ktQsBhz86chqZlX@1_HcS(s@sr{HUtUtgj8V+b9C)I18RJFZ20}5vBz{xJx
zRuId_`wK<d0jUDwnXD_T#k)?;O4FsT^jdGB)s5}U$XHL9l2Pl6e$;o{q4o^52zP4L
zeshC;xvuQMHIG@~u$FW0v|Q`hgg<`-MC{4~KCLY8VKdDp@!gv0F%^edOIDt0qV8}V
zp0Bp`;@zFQVuE64H5c7aK)sUKAbPdjK*=22D$v<z^qk3+z2u>TE?O+i;}?gUg!(VV
zS2|9qGC-T%X5HbNZ>!KR;90{+|M_3MjUj?M-|r?Ha`T7x!zA~STT4`vbm4P`AGj{j
zVcvV7Tl8X;;P1x{Z5GVQ>9Ov#sUtTvy=%vBv}!X9IFJmY0oqG>Qmb0Lx~AC#m>(|u
z&+!=RWqZKpLDvBgk@1e<ZPj$Z%5cW(9Y!pZk}Xq5=P2y#NT>I~nX^WnM_R%M!)($?
zx3ue|-mjzUx8_QIHwC^SZAi|1kSd}*q3pVc7rbr_uL@^V+Ec$7e}VRC;gyd?#q<wH
z{16jU)q^B({-!FHe3V^MlppXXr8M5jhW@?t7#gMLjYyGy$<Y<cH&BuO<!(rP8~n}D
z`6BUkJlJ?Jy2+d6oYhd;cq_B~lOEy5>Q)O>Qd(u<;CRW#u{|w?G8BR<0)rWk2g$vi
zSbL_s7#*iY>X%tcr*SA(;ke6P9>r#VBN!x1F$l3>0MQ771F%&B5&xa1SN{u}J|F!n
z%qs6UV8`QK<$iMRq|-5Re60m+7gx@6J_h-4*e|1r>JzB?$D1RWEJju?jH7e7y8*49
zH2$xqKK>?o*L;38se~N{BS71zU4*)0Qqw{V^awcWG##Bt@A*)_bb}U7(iRJZT8tTB
zE>U?3wojMLJ}n^MpIsCfCh_%6lne<g3)|%%|9X9Yzd1m(#2a`ioZ2y*cXWtoANlPI
ze<=|Hx{1cQwdT4Sn4Ml-0blwbMToQ>c9>4dKQ}Bn?S2Q|jDpGjOkr&nI>2K1!E!T^
z2eQ`eKeD<&nLB^&T#Q!x${kSn^{3;v6Y03yB%KpKsSg~T4&1v!Akh7gAt44%%DMF3
z|3;6!<BRSl*M4?`;|jgbi7WMeO!#h3MWOd06_TPu-DZo+G+xX4PKu<ji#!1e6n4S=
zV>OO8%xel)_N%Uvj<742{AXrRET{5G=-++$J;!~H?<&L|zf;|TxLReKDYvITiyzWs
z?-6`FQgiPfXX@z-aHzYq#pxR-RA_Lti)%en3UIRh+og&G3%2&><m_#ayEP&rK|YVJ
z+eiqiHX{;GJwD<%I@RDO#Vn*6+T>W)ZnQb;5z39GUc`9oLf6+pu}-s_H?%oDE*VPN
zW!dbR_;U|0ALtKCxMGBPNX7FVZVo8PNgVnJJR&GuQmx@*_Vvc5n?bzVv$~dAP7(Jy
zN4MFZ5lkL6?PjdFK*q<C+U_K=CMN+H1l1N{YX#}#VT5;}P!0mXFpUEJ^MxxAxQ<nR
zIOH4dS4WO=!5s<gP#j{=UE~{)MRh@FZ8&U|7yyQCrZF2_#lw0KgG^=jHM!|3d|G7l
zJ^qM-H^x|D{ES0Iv#PDfg#Y2=?9czh$MShfypQdwv-4aScXtCs%j`32UQ$qn*^@p*
z+TaR&_nwT(!^N66ELQHpXB9WM$y8A#mMwT_$$O=_Y69r4lvt5{y*VXd7bYJ_pHDNc
zi03L1kT!CoNfvMV46fbGE!!z{5st@xV1jORp7UfPmLbxfwz>P%Mc|Q}|1s@D*L`=2
zZc3LWvxXy)ml4<rOcJ!V|Jb(FzRR5lgsn$dakj2Td)PUmZ%pSrfxi9nSF4~}-9+j2
z*|eqsMEwR22NZ1K=!a7YKxy+K(;nUI$u2e0c(|G)kxx`DHps7E`uL87ws-dpuX%}1
z7)%nX@k0?n@rV~6$#0KQxXF*DaS>^fcDnJAg<{U14ULrB;{By{c<KAM-jd21pVZQO
zVM;y-icwevIHoO)ei?B1=f^f(`=2n9J{t3e_wF_#cM2d#-;$#+hmb2_nJBk_89n|%
zO6^q4<xCNMqU>&n_x@O}SgT#8bDR{C)>w!qb}h(JQ;&eR2gkPr7DoM{KMV!L)h_#_
zNu1@lSFMQL5uSC!f#6}021$VddptN6HP)PHe{HJ^6jyr2=E5=v1&o$@I;aP~<l3bc
zM)CkR&_n)ABCVeur5r9^?5JVIw`iKik?}I5AK_NA`*ih=HG257XtBMrJ;RFpco2YP
zQRi<OVI9Xemu8E8jUZb!>PqC3n7w49oY=&jdNCS{TL$bxEw3$EDGji#Z1yFWd#p<|
z!_)hIYB&1bM_Fdg9G$qID;yoX$nYAARRzi(paK>|dpIk7*IW~$^Ls0SSgz}whRP}}
ze)w7R9g2agUzmpGU$Ja81%7;dC;nV})->Mn{cDPfO<Rnr=_FT#yF7W|_S66!yG<<^
z<4#1r#Ma<E%~u+P2>r+`z1&qHpuW*7;DO=OmO-*H7N43irCz3S&(>?mTn>Qo`L$pr
zrS%pe0rRB?((Ku`rx@hnz$g&x8ln4q%FxLx^B0ntNAnIAwtyHdPj7+)3RZTsyo9(x
zCU-o%(4jdY{+pmB#fYKixw$8lEVi~a15b$H0tP}K523iiATu*&HTB(vX_s~?U$|d`
zVpv@!F~9I|pJU_ahBv^Zoomx@Pj_3`rLZvu^iGOMqd73qB(%u8RAcDvK1A<v<LXPe
ztl{>(I9KJZc7hA3T!qSmoh`J-a8mJ4qO`_6x9qW|w2|4TT&43y8%9uuD^v|nH~V>K
zl~c~t6}-K_1Y7z(Qkd@Yxe84t?4?GK@oCsSjXbT+^lX1^1o+UMH-&Z4cUI9Jt0gsK
zJ^6?j<Z%pnR9*-sYBn%#Jp`88l@<Xh|B;Vh|B(;6zXn%{H;xj#Qq5kGY*H;B?6!F^
zU4|PDX19vCS#J7jc0DIX32=|<btt;UK~pvAn@=n9!ssLOKX|^E(C1)&>l>2nnxI-q
z*Chnbm5FAX(2By~e2D|5?ij&6dqE!HU_bK$?LbF7RkLS?6CzNHP5Lo(bwrn6e;57u
zm*)!lAM|^>xw20~i}=@r)bG15i_2D=AU}r!VGs}9bCg-Igd3J529Fu+zCqIxkQ;q8
z3br0opQHxU4=r6yn7ZbuxN7&Tzz*bH`8u%ly%CS)PUw3>O`(f-`Y!bV9p%Le{z<uS
z_3#DyM!p{+F6}oYiJuZ5GCc|i0`FW<HcypH%Ako=2RHg~xAb+T6j$X2e2I6=rnh+W
zOVi56N6*I2wfYaQo?jcVhGYjl*<1yH_D*qKa^r3DHG`VS)MPOuQG9EiAw)zoK&Hl<
z9MNiSHvlwb#3ncvTL~p5m=sJ)zN&QkJ}5b^D66HgGBhNcRi2KLTADaIJFDqLor=@M
zS!V%j*4IN%U}Le}N_;h92ud}aAYiuh;P@5!`?J{18zSB$JLL2Co`MM9-*?c{g1_B^
z5)S4Vhi~7BWV`ORy1N?^y~qN@>#=JpVTBMaNA#XPaMnFA-r-nMZY#lqR0K4{9cU3F
z9!w_bprf_bgke|kibh6yP)gN_&vVj@a9LE^eUJNmCTlXSvGS_a!ER*>-(dG{hf($@
zXnQc_WiSgg?_(YaNnb@^kZPbE1OYu?)xZ8T_06#ThnrR7fXEP@2b+dOu;1%drD*mC
z$$=#r)x!8VWjn31hPrns_mb1;NYMj-+W-W?GiCyWnjZC+#IiHZl05!uGd6|&8LSOE
z>7IKsS?!Q2e2{G^#VG0RTe`vW7V09j-#V2ZU9Tm`CuA%-)N(9Bt2R8Pk>@z)$DB<+
zSQ)LdlKlX{@zj6+%gRBNUDFWxxm-qd(!xbVpX7(E3f~b@@R`jo^v|&=rbv1)@$;y-
zut_`8#=~miT`@7nh^R<sxWTka*Ca}~kwhy6PF*w6y&@R}<)-#ZUdeJ~c?&kqyV|w!
zLG5Oe<sWz#0S*!JZwLrchxx5*uj{97n1FGvqy_<klbiWTj}4WWs7MFO<nl($11+Kx
ztD}hIf{rO!spp)>!T8^brAwTiT&l8<(-xw~=-Al}4iM!~?`KX0(hPGcAKK+dCZ}{J
zagc1y`cl)<4KUm^C0Fo)n3H+!qHmUSP91BaF21}~Q+`?kRAn@h1_bZ{S}7Wa!Pc>k
z*HtP8r`|a(Bt}mq3!HU_jL{Uh0^Z<rR@T?O<vVJ?Keinxi8K{7W}&Jk*=k)Q{$MTe
zy}9u$UAN)dZPeYC${pf@Op?Zr<L`LMRg<gd-odBK&}sP@CEEQPuGMyLzE@^vQl}-e
z?zu+`8d+sS5#7=eeozq*=|r#l5Bu&({sR|3Ocmyx_rTeq?L^uz#(T>%6@!R@A$GR*
z6dxMbDO#MJFFxdo@+io5DMq;w;V#d6c&kudfa81SMK5bi@1L4@{fi2KPD_5(4`)^W
z?>uZL1%JcJM$5~q4TFf_r|#P!pB%9<oB0!s^#Df;i9vv2;R_$wR$Q{sIVhqu6X(!l
z$EV-(L;4yQ(-VxlWk*Y^bK@(EdYIIvh#<yds3Px~@AJ@NNnt$3UcWdMo5mnp%ebz_
z+I+uZ88dS4o9VTEE3%NE)>JW$)TI|NC>bofS$5nrB=Xc44bq8?n5V^`)74wk$Pj&R
z&RQOM96<L41-xzk_5w0XX$j2VKKfP{PpFVeQ~9s-bKXq^Yi`pcm`@s;Ns$8*0!bpb
zs~?=bD`Ln-B622(n@X(5pVM+}+JDvz44L4N1`bd*q~LevIq3ECCb>Yaq(nwBuD>7-
z7xTDzuRP*w@P!L|WI7(O_<t_}FgBDr;jq?o!7V);?**#7Aubhq2{n6K>uP>!!T2sN
zWSdHj3boBsgh%IoUjsvs=O|e^V_Qe^;x4gEZ!(r8!gz){OHvA%yg{OsTSXu%&XB&4
ze5|)^Irni8;I21Ag8|j68&8;F63(uY2rn@+Yo_bD_KkNt>u8rxaW{bG-@vN5I6#Jc
zYvwSMn75;|BvnC!BS_52SnfqgwwJ7^JpRl1jwr#}K5d$_M{R5D!oKcC&afsZ2hG)D
ztlZkprh=6vsGt-i^4RH_Vco~Sdb`KP;~~dys7iAv(YOyu$t)A`s!9kW{iulZXMYzh
z@;<Ey+BKreVq1@blLN>%(*z1U-tn0QM%G$Qi>NRDOHism{#81}Puh$5<q9`@<wZq<
zjEw}`GW(zJ$rn}|40pEN&B{vI-Q2Hfu$5@%`NzuVy9U0KUJ71CqV`mK3hl#2fH<I9
zHjpVE<>0{y9%hO3xCgez#)9Z+h}h<j^Bzj=-b%E}DA0Cu1~)~!>0(puE5mitMiXum
zY;ZGpjDIl<H`QI^;*`0c345Sf`tzZsNig+puh~>nauI2KZDyThL#AP&VTzOeSQ<xw
z?;cRXifLmcb!AqHZ}Q->=FYn2m?V63Neb<txxeqF;yu|Pq+E8m(1>y3jn>TuQK~vc
zs1>{-AS|dTGd4B>fAsn*NybcgB!NqUF3DZ!9nNDzcYI?LDZb=lc`~>^er-A{%010P
z(UjMe_&)p*s09nnVbr%d^p>7a99bBTCAN4gZa2QPfBoU^If$&GzB3UOHf@xa$p^J=
z-LrJ@|NIx$U#A*h{A)(kx@(~A2&ydum%lV%k5iPIw`}qx-`*c{V&U)DtLl0#yj0<t
z1U*R25zFCyU%R^H#O3Ax7~I$m8ecwg`whP3IWfZ)+_?5;Lt8#?V;)CKmY-+zoNT^b
zI}%trcx6)wUlj}|<zjE?Wf|6SC&2W=gM$k6NyQjwHkY0lV!hfGpSv6~gt3kX+^7h&
zFC1V&Y#)+xuSIhNSw$GB&TX!}bFt=jtsxev{pEJ>u|W4OPnHX50KWj`W5Q*f!eSE@
z3<-MC^QUy!PrP|d4@Jxu&8A0m6rd-J0`;;4e=F%++0v|$i{l4R9wokfqN+5pm{()8
zxR>nZOlVJddMMN?;aE`8T?8I-WIvq<dDxDYu7@Zm&D!VMz5O@tLhygtFj$GU+xwc)
z+E)i78i9dC(S2f?vH;VV$oKjisD&~fWJy#VjbDxFR}x?e$*a7lTRxC@oSR2x4(L~?
z_Tupe@qX8E*YDT5yz=Yc3<GH48k)5W?pFm?F-+&u&2Sw7JO8`r&ljl-4{GaQ#pO(n
zZV#9y2~a^(qu9YQg8}oD_u;eSX^O`ymzE*(szca}=R<k-shw0tujChH_d|Y6ts0#^
z4eZ8ABrT^-`UTOgvMBC@j~}rgCUDOFl)oFt&VD&%WunQ}IFCfcDw)Kh4w9v_d)F5{
zH9qTECkKFjZ*HL-?_aDRx{^Lcb+y-Sj7Lu-uaqZD%}P&~vf@H1T{-dI=iA8JsgqqY
zX1szS1bazyYJuPXYnt-?f9m_=*YGxSd#IIe5d9m{>LZ-wf<J&0J}78kcZPk8u)ZQf
zBlKHeI^hmQqagt|=tWlG4l$`*)&)I~Of$^iMf$YgK714LClM>-eCa;%tQ`ex&O~J~
zRBmh>k%Ij<le5NM&`C4n8d~&pg&@`>AHOtHkiutjd;(`fDCrOa;wIZLVPTKztzxc-
zjEb7g*8Z8B^=L%dqsQO*XYC=~@zU@h3n;f6R-5)>jp`rwbcX=Qeos!~K+*2Ot#fiY
z6WeULdfL-QzVdB+!VuV2@mjjVWn)Qvy2SIJv?=>6M_4;RS~1p{;znprlwUQ)bh@7t
z<$og^-6m{z;)MAjnr>vQWuWkf?zmn*?5wJfNiM_la*R#1s~?7aq$7)t<OvsK(ZpLS
z^cK2x`^;lfE|>XRZ2$@=KckCwjj(CI8B^ABqX8V1Vl4@2X_VdFU7!d{+wU!vEUSv-
zzdF+mE&+7(&$jy0@V9Tc$F29mNcX>9G*?TW_JLv^t&ttIpg~5~Hn}j~`GF}bdrZq<
znO&*TeUcgY-yvSuzLx+yB+pFIc^Ja!7Vvj$PG02~AZvHohkRE_Owd=VdQ@#Hl$&8;
z5EZ7+e?eCXJDa<TV{^?F4JP|H(<Q=T9h&Gzox)NnCR23C65%Qu47qF?pdYr(<uon7
zF=#DGac3X|#`e-_d-^FnqoX&aD*KNIw_lPXe!p~=vkKN`fNRI1p{yl;P@n1NqTh80
zs1$qvzkhnDr|Ej;d0<hAK0hOa|Bo;`6c0CrFcwSV#pGHSh-0;P{r9EiqMp(Z#4(wE
zfoP3Z(daMi3!+{y{8cE}VJ$`%*1?NN4#DPPFZ~gQ5EVuS>9>U5eofPAxc-L8-aTfo
zC*rqB^^9>&<-!zJ_`L5q;SkV1@bNV<wIc)1Bmw^0zTrp^NA~D`8(HUleb{>9T2x7h
z$s=3jOpX&w5rRwhe&-_?8k75+v{k_nbo8u&WUtw$0rqg4&~?j2GmvUBgxZ;H$$~|2
z%uS0Fw0F{nYsbYDSUNfts9NJXI{2q5_65mRo7Tz|z#Kj8E`#8$J#EwD%OYSO)E;Vp
zWlRJoSTr@7?xH!5@-q`!%a1eU3eQRCAvsri>6uhH+Wqs4ylW%5Ps!LK1m2FFQhshn
z;|>9X?QB#C83DvN@=?Ea_n}f56ZX*1GRGDZj5`3*e=1{VNF*bV!_V8pH=C%!;qu=8
zvp{3VL(xFVo1`$lD<N57UO?Bkf{DrkpBh}BNjZCa=GE0z7!h&tw{OTusUf)pjjft$
zYF`WuKV)TP;d)P3S*X@KAJWqOes(^lLmD+$4SJ#L_+Uao!FCU>K>k`V#p%{CZB}h<
zb+DAQb2oc<?sedN9<98QwW(Lgy0W$%9%=DlTtI*-R;oy6^|}yP&1j6t|Cm?mNg5q|
z@<@z$ybL>NOdC8vGeqA#-G?TQZlZJ%$EKO2{MVAz!e_?0x!N`}ZmsY$%?bP8wL*tO
zPJ%rqZe=k$=d><T(41b+bTT)F;_1vP*5v!A@H1FeWFf^n2Udk6d2g^yQW{#kiW;3s
z*#9a&TzG<G#Porx4DpC^!#>RE65Wy2jzAg)hMu9Ju+mb-_vGZe6FKNj?icSlIq{U=
z?(Re4Kx>G5Q}d<oTK&w*D+rgvmKHS_Gh^R#bEy;93QUe^!BzPjG%z~9Ym1&=cWfp6
z^_hRrR$N!syEj`h$48N5*29|$=w);=PPvRxMkDb6xCjL;Xf}o1HKvKhun^AVhED;K
zZVuzWP?WceP2eCdxFlEH<2#ig-_c0z!T!bxq_C?%mT3iZy1%No?1SSi))=1eGLs6V
zuvtuNoi&7pqA)Wv2L%VOO~?tnTFFWE{x+UDkGp`q<~{~Kvo7#((SG@dLMpTNpA-ua
zCW2~ZMz@kBOw;b#cg<!J=WSt2GDBsdJXkHq6%*v8rKl$&Rv>Id(H(VxQ)kIWRC|Xf
zX$D4g&S87wz`%&kS4<&Ex%D*6E(zil3bHQR;C|aP=Pth9#1q*5K~qA6ebR{haU{$d
zVoJMx{WHaKngP1D;Vu$f0rZ3`b3nMDE&1c#l_$i%tPb@AFL~eYaAS&<G$UYZ<4_V5
zX*TD%)K8uB77;hal7x$s*Q!BZlPGA3JKq`eGw2D|K2%Cb`t=JVWV!)!rUjnighasm
zzTLSeoS2Zi@mY25Wa}C3@-4r|3}Wz={q!|AkW_JXM4Is)D(c2X^kNz?qU%~ZePkR)
zx16%A#e-3OD$ry+LlT&gT>Vn!<cng=sqtN-ppCq<=eCpJtAsO|Cck>|$D^wV{E508
z<}dS+t-cgIr4CMij}2-j49<p3t?2jl?_C0r_!<B;xyc!e+>f7H27@ArZX8JWRB<u6
z(gfgA2!3Y7N)_2LoLBP!3Mek>W4Gz!FR3I!E5+KLZp-8J48gIj!0-WowpIhi-7Nz5
zmFERo>@+!eT*DYz?$18kspd?aLOhO{j0f@VA0EcctDbv^{r$azgBvSteuY|fm^?fU
z1xZyHsw#fCu5fU064FXIPNSx*RiQZlO39-?WHR7dd8`!R5~rDLPWU_MU8UGz#XmYh
z{Qa;^0?r?vQyqy1s&Db>{#W|_)6{hTlW%y2hxQZ{Y~%F|Px%JdzTf2K-P4(U85KW3
z0sr}~vZK3fxABeCyhl=(oHsbkgqYRVf@NnH-uH4PS03q{N|KB=HTclYGOp>zwenru
zNmnGUhPxV&&PX4jOL`}*{zE$JJ8a5m4_rHOt1?2C*cY0>rJFeR4Ga*;gcmQ3T#ux_
zqh>Dbo*nY25;C#z^9EqC2wUq?817uQg|mL1kR1|M&NwMf?`_E*+}W{M_W3RpU9YTy
zqQlfsG66LUNYt3Tu2%zbtFMlifj+~}&o|g2fw!`<vOoO&nVb)1>5|hiRJGw)H|uAU
zUBHn74YjlV>cr3Mg)!snD;J?~YggPCfG-}EzB8o5Sc1&Eq(wt=J4Y*8E69^IdgaOK
zIUx~$>QRr9!uc{VWNe4kvB6_SnUAzeOlF(s)@KYwAf-qbgGKU`V8`BMIY4SyGAA8a
zQ)k?S%c|uWMA&lirTUzfm{_hCU!g*Cj&}B4oqly6(R5#AKJ7zu6O=v#cam0eWPO2W
zSs}QAIB7Kz(Y1^M<y>iv$0B;&$;B@xq2qqowtyq6u4}O6iq9y(47Jjkn>+L%Z?N@&
zqRJb$!!+Pk|N6pBG5<w~9V*h{6hRFDBggsmrJd*n@*f2>Li<vu@{FO;-T;}eh-B99
zKuc$4<_DO1r+woojxGZOgR-)+w(jnLNu|x@ZZP<m47yQ13k%yhI%cYt=y;y>QQ=O<
zl#8gV<0IpKGF+^+yS};6x3Q5+WVecAw<5T^Jgnf9b#vni!DTQW&y@HP5bzDKun1{s
z5$Fg&U9TA7Z}<gWrl6wI<;?LzKxa2kac@d78i-*s8~>nDY3f^4L|<H7?6^DbpQo5p
z{;eyRTbpC76f99ZR`Zp}3l=lQRy!yEZ5L4qqreNEK;OQ?N($F^h>wJ5M^%HKsFi*-
z3+)&-)bcpyd)#RP^OTH0Wcl%W4QO!4sf?eE*1dv(DuP(m`_RwK#iJ%bu9V~~vO@jr
zJhuP6^TcO;cVyeVO^v|Wj=RIZEmWfTHy_>a9YSvtaHp7+_HHgtxuOhlY5gBv7KFai
z7r|&bos=-Tet4My>0P60$$ZCF)2$89%cfvPSjH+GPp_@&5J?8qC9PPmRub8+SL#gc
zRH7u<Szpe0P^JA1_X?$xIdOg;H+UY;jpCu&6CkGm?g@tZg{~2urslpoByq_2+?~(8
zi6pYGbFt3BY@wQ-)a(3>goFfWkS#MAAxUxD!Uxj>m;J_<A|};ht)DeD99IwzhL(p*
zLVSFDB_*X_zkpw@xVoM~5hQ$gZ01;-o3R!h=dqguvFn?gUDv1UH=D7#UnC{d^Yeqx
z&mE@=)jJgB_(b*eRw!OgwxlW=Y{jRf$lBU6x-8lufSCz7M=muR{A_yqdo&`>R`99t
z!QUzzh*orw@U#25{`~#>&im=X1twcIS<>0rSr{Pww_$4upi4k|u67B55}umHojy&u
zt%CmD=iLU4SpRB*NRV8RFE`4?|Gaa4e^6H1YXR>q<LCd}24=opLXgj+Z^8?+Ski1!
zkXJ%;xo(T&{)knm2+nr>TlF>%L9XIy+Y02VHQ?J_#Q&rvS143YX}CVpMtRp~cmgyG
zAB{S|ZWNk}+1Bu3$x?=A=}F6a46fTy>wU$ECO6n74RybR&GPR=&0Oy#0dM4?mea2K
z8~jo<nhU3Zccp;ElzM%HGc1FSZnh}mkFmy`qf)TDSjPe*FE9Urii+dQ;)uXg5_n=&
z?(V>KsdwD{?C$PfWzc+g#3L3C5J{-3tB3Vylzn>zp80M=X6?%>@P42^Sl_C;^B63k
ztcgR5_5<{WhK8(?-s$?24XueODLRJ$pZodB5XW-n-@iU9D|-BQ2Q1+ve3BX(8ew-+
z=bisXq}Hk!QDwhrF&3yb*EbK^>{^0W(z;f2<tmE|X$u=k3{?mg15wU3E(RwpM?Og8
zjY%-6sd;YI2KIFd_qXi$MMdFuPaN2(DzK^_mlHs7NVvV5^Q2ndSZyXc40Ji7wSo=?
zCMBIm?wx^x+0#w)O%m7YI`Gf5cyPX4bm}Y#r-Z6Wh6(dT><hP;bwqnNkC6J+Nl3~;
z_L6A*aKqapOvZq@1>2dNi1jz~l{l>p@0Do9Mw(N>i1?+3G=bKZ@G;A(YctOG&`pvq
zI_@)Q$En?~Ff(;j1*)4cK5OkiUbQ@F&!VBDn`{lncXoEFDPwqTC%b@!&hzfD1`7*I
zk2yZKu<!#n_w8pw#3r2)ICNgyxd%l>MYo-_Kp0yG2fzT*=$4t4r5#7CrK2<a_;82M
zYKj5V;C%S)bge74lgsOtRgV8=!(SkRm?x&Tma~6gfGJ)Y4+Y-d!C|h_>@CR83ZZW7
zA}L@*b@ueYjg}dRr)Ol)eERgfmqB)0gZMZf)dWREM3Nn_P%|JUoS#2`Hcx(m_ElcK
zO;HZuGqTgFs7t_=`7d$hf<+nZx8n&IjTPS_Rt9>LC|M6}Zd{W0AGA~c9m(jhMLjX>
zLWYh$rN9=H{|zZ%UY}LW!Rjo}*8^aQXy93%8NB!K`n3y6)BLWkt>IK1#9{qD%Kv1K
zeDRHYy?k|!4@<*aMAFgKKV&)SYtQFj8AlP>DR<fKV3Khfy;*9jH{aBAVN&ZC39A>%
zk&sTI`vnzB$b`NdrK6Lg`;+^gpbaRj27U6$pnw4xDa$I9>cnh`73oCPZx|BO2Td`U
zDnJIijd1+DS?6I#n`rbz(dd@JsQn1I4N$TS%&$1ggm;g^v)nV|VElvCRa7jTs1MBk
z1K`Y?P@NK!%Y(VW6h1;20RaK)rVEvSU(GkhU@D^zCnO~FkeAa1{{eYOKe%1H(?x^`
z=p=@R^HrIdnHX4(r!7?##zQY!mzI(hb7WVyoKN2ufSE=_OsrhL13mziup6vruA9-S
zag6%yqQJWGaFUz;T?e{=;qCb@?t@Oo*vjf^+g~BSPgwf<>11d39-2YRz*8~~y@nPh
z`G*fbK(u+(bOqtwrX~>ct1*2BnLjXar52sl=5`LPwg2*yY_Cm0-X2NrF{UHPS$T#W
zU7>x>8KXHQ*rIIX`6-D$;YI^OO>sF~_$Lp5NUqSUNdntt^*G{ppp&nADiDe`V^2hP
z?BCq~_*0VsNCNgeO>|QhPbh!Cy-ia#(=#FQOwO#Faec}xkDDLEO!8X9yIp%pqI)rq
zs=df3XOLmEp?P)MTEb%Htw`9*K~4J2)0qIm3!9o0u*~u<1M?~9>(S#DKBf1~4=fkJ
zA0yq@>wHZ7V7VogO!$+vS7b`c@cj)mp4W{T78dsWcsUgeERA)(lUBH?PqWz*yVsWs
zHVpS}OZ86RYL(2e0FRz9AQ)|u3eE<=+~)??4A?JUzSQQbX{f9BrHh8Z?11TTg#XU6
z$ZbYccoQEB>xWo|^w-%27aKodi<&^T$>c;#RFut-<i8jw`d-2HS7i$5-g2WSdJLZF
z1Suu+L3Bf%)0TG9T%_RTsS-#v$4uEcZ))F+=5fKk8A_CHSqh08>SwE(5oFoNvCTVV
zsy~4;(&0joN)eLlWv{+VvABpeGWjnU%dDLrWf)RrEB;Xq>J|e+Dk2DdwKY~|_%qG;
z#Z=Y58sbH63y7P^z5%17k3UjKUFS7e16*I-jW)XQXgw6&+X{jX+c1zGsy8^(gRi80
zUOBJ3%QCM!XLjpF*R*$a;CC9;mVJ9u1$Bqj3;AJeM*UG55l%YN(u4l&2!FAnN3Qa`
z?oWH%Zg+tkhCNLW@4I`B$uO~X|K-h%jg_6;M))sMQVl0ikbq{GuXtafM(2Rw2N~z%
z%2C7dGj#GiX+u~9*Zh2MD&*wEmY0u@kkv+B;NsJX^D(4}4Epa-*7C;>MQz6VR8$`|
zxjgkLBM7SNlwG*7VS<VMAJqoTJb>@n{a?)lb#;uq>~LF)wt8|u>YDlbghV%~EBh40
z1Sc_r6xuv&Q+IG;e%F+X@p}whmYj|@mwVl)og+g(T|ZM|6h|i-nu@nAIj9K8CM6}|
zG3vp<!K19`KHcJi){@Rn5&qj9^5*L`WEinfJRxUi_M@ZNr*3ImTidW608|Wq1O`UN
z#r5_@QZd{vEzd-Z?_c0lG4Pe-@0E!1fV0aOENjdry4Ubx%yW5RDP<Bwwuh75u9jTl
z1pxlBu`yil?cLqJ^W6zI5Z#|%TvYMf$uo>!=pv38g4azk*E9Mgd1T~;&}4^?T-lJb
zjqXTxED2U*YHN)_)&wZX5|gpf_<jdpj(JFQa{l_SQD6!uXT8?p`Ce9eZBm6@4s{=`
zh^~%*Ml7<9{?G{;)M9jMFU%b-&(4HeawMzqFP4W3)#8h$lVd!MobRJyx$h_f{pg${
zJ-FX`iDHAg@sA5@DYaDRzr`I@ig~3D&VL&#rPYJ_5&~;P)QNzB;jE)q6cv5a-D6lF
z)};PV_J}n;dg*BHI}GU#&MZy=+3Pr3N*ZWOjcf}7p93`3pCINfr0Z#D(i0HUcVikH
z`dK(w!?N(q3<!wfk~AogyQ;etV#2Y9JD+KRjL%M3F;A(|a_-MF6!GzaF&;|rA2T%^
zOBV%rZ+9}LjtnFg2X5}9RW0tlQ1b&)nvhXrx81B3syFYBTecHzn#m|A{I9QFp8=rD
zvR6-`dPN+Y1&+Y|2^_c&x_%^APjd~&*9k{YzUw849+8PnZ{y!%iilu-D|u&*EgGU?
zRuHKYcu_EHQzNUbo4=<iDP<)&1@S<B5B$WyG(7g&K8_In8h-`(a^l8M;>Ig$FR<+Z
zF<IPqPaSasek3rteZTTU=cjH7;(xqe=hhU*_;IHS-dUvceh7Wgj8*&<>(sL!GhOYt
z!}R0LNK;4zFyJ2awN^728LD>IBkf%H6c!pa;L`X>HSX<vGe<{iyzCc|kp|MgumC1{
zXE|^rhKZgeoHMm_TAxvatRLZ=HBrk?ZG~F08W$bcoiYc!t+{w<<e%lzXuYc^GvNK^
z-1SDEsIkjHdDD^Gc?8nhcB4r>)Z$>g<p*Ua=ZKxiG{ER+c0;o;X&`XgU)hxIt9$>2
z<`sif$GQ|PY^FrNc|K=@Pm{n?2N3TRNIlrwvVK8GZrvb*H@mX8rt+nP%6VS`?Xli}
zWBv7J2DQ!R{G=Z{dBcRaGy|3-bnh;z&-TQT^3I;yu%N9DeqSj&!C}L8{aAux#Y7}|
z8U;HFoIa0b1p875<K=3ANs3#x=eM#Zk1Lgg6&Xp%_T}Z4ac@U%;_Z=C1fbSHfyw16
zk>iic;k+nF15ko5rpJoy?CdB31&ET}W1`ll9EnIuR<rRRzHkU<OD@Yy4qMXMGKt+R
zBfP@k59IS!e_mn)HwdiYbflNs6NW`(`-tCdpT_nt3boJKLbe+NHIxGerNd3WG9-C^
zE~|7BRZfz!mPDZ`<Cz7T?3g^*pZZp)AS-z@&4QnB(LXHOpdG0V9mwg_#&6gd!wFTd
zSi)v$e4!Ec;Ei~Tz;3tJO38gm)L-xv9}%X#8Hq5N^QZCs^y9MUT21PoN4)|cM5+^|
z?+R#hw4AD6tUu#4cwdK*hg7q;cC%ClAr(wHpKA~=m=qg7J+Ws1Gkbz$e$L?v@B3t7
zZwS78%x2|IeLHZ8(c~_`M~RlbaP5okxil2|0}F@u^c6+j3{`Zgs-PyqN$E@E_^4|h
zO&|8MIAo8JGns9g^AzULVg|I_YMZCxw#*c%06#9q#fg)5=vy!tqx7|6LvkON2eRcJ
zMU=HD!-&wT{kGT@0DRx~^jh>n4!o(e^i)Ah_TNv$8<<c0sN~fMey<b?zcQ;U^fdS+
zA)fVQyK4F=IYUfxS7St>ZSDm#Lw5Nlbd%`Tim=0kUvAxbDF1MwMM{dM&4U^Ne$x8o
ziB)lGMz%H`zqm|8^~OfOmqNpl*awpmo1T`C+A$i$@9z*fOaaz}v(2?N9-Haq6A*r3
z;Nz?B?t@8na{%Pjb%$)Eq^W0{6jCvGNt_p#&0O)jSIfXiq#VE5`0vK-;0kVIV`J$Q
z-gvW#Y}iucAp$TvD^V?A(o!+M|Nf2@kvlntl;1T#o6uXl`0KcPYhvXQ2Wr+~GCA^;
zk$<Oks^~cofPa6{BmF5^pi6VPuxq@(_qD|jc!Mb$-O`<&-Eju_;*+S7D8vVjFcN3u
zw(o{$hg!mCU3mO)RrnloEe(XWZ1yFFXLl^il6~OP;B6!S{`9V71$nVbLCw{F?J)jh
zGGj;}#4~hw<f5rU`xR@Jc}x>?;*TYhRt+?AFa0<mkkxLo1>0Ot6lp%1=HS{H)vQbv
zk00tbh=pD01(USwl#s>nph}>8XhoBpVLd$FY5EkS^rD3#V>xoDx<o}=z?ng54zPVf
zL>pXqX`!GW6c`m+So9Iwazw=XQ;zybu!+~FoSBN<3L_$oQsw!;wlKX)j#{O!!9rPn
zGWeyy4ic5>5jo!@lHB7hyh?a=RYVzDRu?6-r(#(#?l;v^scSWpqkVquMpxIjxK!!K
zG)tWDn*=hJ6f#G(ZQ47VUNonb|2GrN34Aw@a&F7V+Syh)tt*k0#6KA7UX)Zf!{jd&
zx4??oG6mOgf`7$Yu#YTEC-|-DTnk{u7WV2nO|+fv=A_0jh>M6Q%^$_;dh4HU^kGjo
zPnojvKHWn%Tdod0?p*Aoq@<p~I@qU*V=+MkrSoi$tD~xL<vb<kZ=HxS4B(}_4Dp+r
z?NodLZ_lE;ywJQgUa;SCb8`dYZ<Pd}X14g0gamWj!SAaCBHypNcK~e%XaJXn2*!_k
zhjx7u7|mw2Heb}lQoA!U^ECpkd)lX-SuHcEeOR0E%q7I?N6sHR9Shgj6JJQ~Rw**f
zWAi6zS<QFDOBC`%J^_qxwJ#3ObpzEN)_K1sv^~92;cJ&6^W!fb?R#VV!S?VA@3G8&
zcWu^0S*bOiQ-ZQ+)<?kYRGg-6VeB;oDe6y2^lABzUEk5U*UinXF(a`C$EzX*fdy`c
zQY9u!JS`G^+N5;Xxv#F3<dIrcU`MV$qvKx75ivaxbxjYYJRbAu&xS)f?EkQ+aeK!^
zEgr7EP=sqO4^GGqD_UvVDy4vHZt4Zoz1^jS$jD5fQhl&XoB#m3q-m!=qJ9mf4L!}P
zwDLc1O&a)`o$*i`m`{&tw=zSoU_T>E)<vKrT4avQLIb-vh0h<lKEFHg5x6kv9t1Ml
zKHP^eL)FjPBNXhf$T(b=bb}k1Wr|kUaY{d-E*(e}dRf3algM7X6nnmi%ue<@2}AsF
zqxPau)?eI$YaGx(A8dX}z?a-}RaDs-oaM(xoErRXc`@u{!pOD7?QoyD<cn`X7V9q9
zAVw9VLI#sYds88({e|K&eGez*!NkK0ii~_KnS~iI9iUT^Q(F3Kc2<r5{zUlk@$UHS
zA^?q8!>C1HU;m?8sSlZOfS!SYpGpA*C1rp~s#lCb4-gKEfsO5-x}5AM-h6*16+^EH
zF9=o@*Yz+CJOOWBnFLm$yr@Ug^eJ;1YU&k`m{=}16Z?N|i=t7q-Z$wE#s8?=Oyu+W
zEoCXFctO0bv|qr%F>!E+m&_=B{n`ow=$nff-J8`w($ccLn^sE2&iK?+xe}eGprD`?
z5K;htL6TToTl+4u4;(o_{^?UVD2IV>o}gXyy5J`l2F2kU`wH;IANlWigFYRFthP4F
zsHt&{0(;f8kqtHF4i^1BW;~@%r{u=#OyLYmiao!Poam1=*!LliGLn-m%Z*zR3=6a?
z9XJ@~avBHN)!t?4)^J7sGO6_B>-TKCOu!j>f@<<&lX50<;{=M~$&4m4^_PL7a<}sY
zzof@~j9QJ>nKorA7n`X`f@TzWk5_@X)-CFv*I>(S2WNIzHE8l4Wh?T*M{)0!secW7
zkW!UW*jbsMC`U>b&9CPA%6Um$w^6};#l?)cK`h)6aMOz(QXHYExER3JvxZN71t^1m
zJi3C*6M}hPROiST1A-DoDBk{TeGNb^z=q&nC;9q1|4%~=!6Kp#RJ>%W>@2J<3Ohy#
zqC%Om%#-meabPv}L!Oeb%&xCn?CMuE*=B1;@1r!@L>DhJAEi5YUG|S(Vx~*-yyh4M
zUf215Z?=hG-OUUI&(Zf$3}Sx3GY+gBVA4QuY-}toFK-_lL;+7yd_4Z;<)!1!=xX)G
z5eO`#Y;89wioogu&R_TjN@KJ0^Y0`3zEWd>8nXTQj=W$#8xQb!k2G@yg&MSd|38|p
zGODVsTZ0nP3eqjzNH+pXm$ZlO?v&cJqEb@Qpnx<;N|$sa-QC^IUHE=?4EP}&fwTA8
zbItkGLR@WxKx2x4h-heO87fc9v_H_*6^cnJ!0vP9Qlwk!CM61HTkqpOUTC$EO|R}7
zK+u?(nY+&inw$OYW@`?&iyOzz1{0VS{T^U$j{}35nVIg0MsV;PSjr><E>GY9!D5B8
z4LTpj=A_8Trpx1X`|E1gxCXG61oavZaq#gw=IXudlXUmzTjvnPe&27pdRF>YWBc_R
zLoE$&k70%Omg|x$&}TWgiOW{~962@Gh_}DJ+<Y7H6cUIeJ;bkJW5ep53$tbhEb9})
zvnKko#Yq&_l*i&=!`O+%9<=Aq|4l8SF2LXHpfp0m=yrRdnblX$h#}aZO?7)@X0S{5
z@kp)3#PaB)t$y)Wt8)<w#bROUxe6Nxj`(n0TW3>YU4wK%XyddX@kB0GVK&K&^Q6Z-
zOJ777>vNG`O`@@GW^e+&!4NLkH-tpYjViKmDt#(c5<PbK^)24^aG+^8S^g>3X}_x`
zX6R?lnZ0sdIEwOEa(^wiszA$Ic*{?;fLX5tzO~SFU#T`ZlJAOx<a6|V<K)0sh`E+F
zRw75X0Z9ZB#OhHkDfzW>m`0}|DE2+8o;zaTJ)$!cnwAr8gM4Inw%VH?@l(Ffu{;Sl
zmI;MmR~q%<VC_%{`DH7m^#g=C--Z+L0{~5}mDR5*D_frUWqfg^ac92^_|LwvLTyJt
zYV&FMf$6f$r0c7ISKjOdU#<Ioe=@&wNo6HeRGLy-D@do9oKu;i%=?9L=qU*aKms#f
z;Er<Sb~6>%8OienP|O4tCD%z?$z%67EM8QRn<|s<!KYZxn9<A7jXedk0cD_IBA@J{
z9X%8M;>lE?0o8|&di3o-LnLI!uNx0gJL2}%GI+^3Z6hmOY`Rz(hIm?VNYHQS+{+=}
zo9O;rb>xlfF==L-J(2|CMk<)BN~L>H2~qgiO;vyYQxiGK9d1_A@ZzJlXV+bO@3ymo
z>FTUSYOdznubSe}RgAtW<Hj2q>RxVaF2L`4>_`z>XWwG=eP=4+e;nc|-Q~?D$M8OP
zv^C3UBL1<pYR0R;!X#lT^ZGC5(xazPmh)GROpd2}vbkf$l-|sSeu8<YmamKXrGJ{|
zlZC;?!*3cSU<Aulv34*2@aa#)whgrjh9>hpNvJ#WZ3%G~qmBNxmoh)tFBgEZIn2Mc
zfj}m^Nn&}Tk}_XEd7#Po-1^tK%i1A-@xoYvS(<)WfcN{WbIXUWEWkFwfgT?U;rePN
zdaW(T8+7tYHkT`7YDx`s`Wam2o$CWnIdTFGJH(%YIcD!=aBwiVWg%2j;SLK;@ck)*
zQbtBEKx3h$rS<gb(^nQ2^gz{NGMFyLC!is!{F*lU@M^mlhmf!ftdVzRW*9(F1AvC(
zKR8MUOP|?{%r#wQm?LE&zRc|GQrl^syPIRBv;BD`@I%37#>>kKVA_vgzGRvYq=NsZ
zIW@2xA=(d6!ro7~Oj_W_U7zpM(a|A;&z9C!LI8QRe!wbLe~+Zv#t`50elsiH<7)Hg
z141Mh&uR()i<Q;Yp-dSZ0oScRId6XejKq(=9u`3ay~cRpMArI}qznuUd_KH*(xU6O
z?0C@BZyeQy4&_jk9BVN&*Y3thAI$y0c1w%0N_TULm8>IT?8d;;qx7Bb@>Du#w{H}=
zwu{MMOqpOvucw;QeWx=)EYaiZu9%~yr_uavx6@C=?y0u)O=UFJ>j2H4LoyXnrqU62
z#NBPsqbcCk(ZOV!QLH*tk5_)ONA9ofN2e{K&pIl9cUgKu6)(ffhswoEJ{jNjgm^A^
zPBp}EUGEms=q?VH#&4x{s-^749sBvLYS*Nm)}z7-10Vd6XIWAha+PX-Bv1->=tWWn
zVpT&gDlu|O-*1%k|E@YP>sjxS5qiorI+g<Y<hD3dM<)%f9g>}R8q@Qzj^0ZhXy|1q
zco_E8+N4#7sOFz#=zHVh>Bkny2nKTnXnQ#$IQ}ha7UJ?j2n(=!*ZS_?`Y`0-;c?lS
zhypzchvU-g=rd9>vi_aPvIJqT%4Q$1NZ1XU;E>P>4Ahtb8vKQrH{5Z#ebfUY1t{~$
zRpO{82an=IcN<33mh(0QqZfNC?k&n(G@a6%BV}+%$oIc85yYtiu6e^UpG-Ii$HvA$
zp&!{O`bhEQ{O3rS44enTDMudpK>C~t^ioE$gL)+5$QggnKu0Q7)*_Rd=_Q71sH`VK
z4H?)wBcqck!@6B~om#4RAj}e`&m-S5QE}>M1>MH2E3#ANc#6`zUc+okxmpAU+Yg?N
z@dnAE%w<xoXCOtB*K!{lo!slKTe-aWl@#vmsE>Y`%OA6XYrMSn@ka!je^em~*;D_|
zP@mLNc8oq6Qus<kMivWw-_|WB_qi-Bz%ZT~(ZxovOa2y4?x*9Xg>t75*!-j%jN9*c
z5>{~VLub8{|J<&3D)nWQry~%%OC@{HT8x!(H$R?Q9GU1_D6yF(0fpGIrk>018x;Pv
z<eH0LCgHg$zQLx!=lTyg3bp0bIr`e(ElEUmq+qUi7%lEhT|z^k=$_S8l)1q{W#bxr
zghhDeNJ<3~(`CqH#)2BnTI#thG3xNK{IfV^tnPu7Boyq@BD_cqNkhrhx1spP-@GBa
z-Q)Rb*`8QQJORQ>>5)?{G)}TF=WiujPEn|J6dc0Qh1}*c?mRwR;c)N1T{eb8wU=)<
z-*GjkeQfUBts=X9Dk*ch5@07eM-AWFA0;1RKkauER3ZuGi>YU~bWfqaEBp+v*;gGZ
zI7^G3IUSTI<@eC*KD4`y33abq0VS$1n+0z6%Y<CB`?2jT^2IsVWfk@|OJ=BU_9Lr{
z=MUPS@uSUmYD5uTb5hxxYkF_*E{%t+`^#k-VfmkZII=$z?^8)c$X(ujdqJ>V`#b4G
zK$xQps%cF9A|mzQuM)54d>=6W_wU~kUDN;v@-kOfiZG_<rgX-Z<`>d(>N2cE>MvR6
z!sH239C#<Ib=lY=-<f%vC**DWXy})ngp{^m8{P3sf~Pz)S>dbXWtYiB&D1jClGNuM
z_5kGae21@PpC;n=Bl5-;=NGSDGVCAg??5x8^!==UB;0(&FF)~0sIR*1@fN|4H<p=v
zTS)Br?btWE|EtWkakb`{*9gftg{*Xxez(Ai+p~)AXJi>fPU(BiFl<z;9HZ-(FoUkm
ziwti3@C(K-=?%7T>8;&TrmB;Ud|YlxM$_XQeLAIow@)?}$6NeP7L(C?hQ|}f=K%>U
z&(nMaFSi`ds2&QP0I_KOH$Fkf(@IgkcSLXAU3T{24)a^#yi!hcfC)VhpQ#g*AEkV_
zF98%Ieiv)BPsHnwY<4?>>&<0CXTv|wr&PSL-kPemRe=igQVsPiehWzyrPbnG4i3y7
zgcrq%dlD8Y2rFx8q`e;AJ|Ve;d<1LnYg3n!Tawr&<nCXu{alMoOO%)yi?+G7!i~Z~
z9@^{D8x1c|oq{dytg|XC#cCjheew8nV<Mt_dAd8<9Z^2P<iLUDN&NENL}(pjm2;!z
z^OeuN@*yI>2#!V;3|>3kWV-ASx!V|i-#}l+U6MN4daNR}8zybi{yaeT17tp(sEY=x
z2mz*shp{W?lWdk~HypBaIg#*t&9ctg#J0)Chy?&4?sv6X2xl5sbji}`^TxPo#I)gw
zhmQRgs2eY~kI)Fi*#Cj~Jae@aH1^4Q?iXv)ron12B-ArsP%RP>qKxy^IS7VLb3e4~
zcD{OJ0oiJOscCm>3(yTqj$9F4Z*dw`n55EPtUQ2IoXln^!DIjNs!<AMkKC|^6!+mV
z7JIt&Mfw0WK)E>a9Mc$_t0K);5~<-sM-`nwIHJ_jFrRt%y+xid)d3&T#XdMeJbcd}
zoJ*oU#sB`bZ5Ldob-sXUFik<~k>~?ThfP|qlM%rbGzA47T~%mh2dZ9{&;%~&zPmXd
z!Um8m$4S0<K55@{ojU>Yy1jp4!a-)0;dS&+e~Ic&mgii%Zsxn!#MAuMPUJTI>_W}?
z$d!7cZdjeyo$`hua?z3|dh)@x3G6dXKO;HRRgRQtf80h#Wn+}aJ^bm*9Z*X+v&R%Z
zGy1hWRDUUw=L<xr1&`vCk`;!B5OUr2;oLU`Y)t8B7~wxvI0#|!WJkO6C5@`ycVtO8
zqTejYtl-5qj`uxxdPF6`S9|wlr}yg>&z}kvKJAZTpGiR#3ORj?{53+~V3Mhfp}RNU
zjvf8fuSjvhoW+NBGpeXKkiSGAHuJ~&rRxeh^w1L<w$_e#2m+{&IpsjVD5U`%m6Kpb
z`l=4CwElfKD)%t|RHS16A-YDLzz0mF)9n4}AMDY@9;7JXDG+}=TxqL)^1UF<A&ujG
zfcwtGFAJ_jSRqw(7vNVYC@B0u3;Y*w(j=tBYs=076yM^MzKQA;COPb$ShAAcK!IQ&
zv7VKhx)i3CD(;wT4NVuZ-J$@ktnw7>8BZM=UEHpJo8}_TW63D9d_#3cGgSU(Kk^mJ
zEk}v&ybl|7M6j{+;tuRMpOoiPuybxK(BFPJ2vAxI1xTi7$AD4q4KKT@n#gzdrBG>1
zmYERWOKy(|QW6XT%Bv1s`q5#?%teX0fQbxora;r0q+YjEIF*LRVv0za)<6mG5B?B+
zCvxLipLOG%aNas`M9ES`T9$QDG4TY^!RSVn$>7Ez=4nSujTw#5)~6p4OnsgqPQK<`
z1qy-Bkf;lzWerr3WVPj5LR^bVbMtay-YEG<c;2XQ-X6Pg9^@_%X!T@3No6O_9ONQ`
z1EH&4Z*k)fH>%jw_`g=QPS7(nUti-gFj!i?Dm@%G8QiuMtvuJwwbsUAFlQ~>*t8#v
z$;<kQM|&aDv8ln4*|mto8{ofJ!ic+DM?KcSPo6F*GdEmk<m+|Bv}qX-S9zepd62VE
z_D2I=SSK^Bwff-ABQVQDQtAapo#slRMD_pe0T|!9aOyzM_@+>Ucy_=zLbx4E^oGLf
z3bo}{k(SD8>Ibvu7BS|{Au6lyGd@psA=zIqXx``UBwr&?&{$a=NjD5SoO;_nF&`rp
zuV;Ki^k#Z8ca16E_28{gjD_X7+H=JQ^>LTEqe~}%dI3lQLM@v7xVXl1Fh9o59U_Np
zJbAS$d(|LZI=seu7iJ+?D%cUX6QI=Jy3S%B9+w|lKUqSHi)Kv&u<^X|gc%*NeTG&o
z?)PNg2^4J`u;ZFs9hTsIZ1|6fR~CUXezO=4eZxlW*O?XhZ&7gMRgsbzV<U0s6EXhs
zw8j_iY$K3vdc{+2S3VT4{`9mo^g<iTiJ~<oeCw?9C`F~J!^gjz$_q;`_5b2acg?bU
zUGU|Yb&WfCeH$g3<7<5=j9_<*U%ytLnyIpd6T_;Ybn8L6V1%Eu)vHB$KAzRQ$80mZ
zv&prW0V~jQmp%HVC141UIiGIi)RSD_v=t5p$Z&oF`wnN|${l}cZl-q23Byx3pZ5L*
zjk};g%XMFK9nyMEAk_`Qqv(Dor#8Jh*!7T|%fQsm_kX=XfCGNQ&HH?SC=3H-^i+kl
z`}8~uGqc6tYzzROHxD13MUg&x_6!UV#&*7(FUySDP;YK-K8A&HyM6@tIQv}+6u7}-
zNU=5ih{qS_O?z7)H;he`B^7Gel5y-;7e{iZuEUGSN5zQNrII4MG)g|vI{LY8C}q~C
z!93Vh>-bdF&-^3#Th$MmlS0CGj34lKBP9!bM!vhPQtT8L8&j<h+G}TZT;~s-7e3=c
zJ@Vi1_7GL{M6aQ()K%hpB+=+u1Wknwh$FdD+zb^B)34>ebg0fWn+|0QuzabUlX2|q
zgYpGEchd7b>L<QJENzUyS8d<ubV;7f*51!>Cy7GkQ%R`qpS{rRuj++Sxdc4uFvHiz
zY}({wU0A4NpAHt{rsPZ4JfbH1NCV(82hA3*K5O36u|9E6at4!~t3@_T6{wXF{mO(n
ze2%q*j@T?YUiPP&6jp((|Gi#|JmvZ?a#}S3)H}QB>%T}xs?M21rQ9-mb9p~-gysnf
zVSr1CpDUQw_Wy1BQ+mP&sg5Nps`lXp3#KUdQM9hpcYRi}$Dgd9W&ZDG@g@j0*BW<V
zR@zR7<manBF0memih#O783yY<T<+ipNHx$>|7mL*Dl=DtBOxW_cUXV}fs@H#h7jT}
z0J^Txe0;`O6f;FUgbcCm({|;lFB_6FdW^&k8_gM_^Xb&ZOMepb_~N6tMV5*qUUbTZ
zQL%FbQKkvAkdfCbAz)cZHM&ds!`Nu}METi-IrN@YK$v$jAMk%Hz9zHlt0Or5%k)iK
zUD(rgKnaqQi5i+yamuR{iec|2u2&oYG^FO_$c${)vS{NcYgH_Up)(^c8`*fH;?7m<
z3qG<2=~vmKoC~*fPqwrJhp9LSu^_Gxytjrb^+TaQvxeajW2s}DM-O^Tzq#OXsF(X@
z;`n3>L$F)7JqgS?)-S2&iE4Jn5N6$T7Mcm}Sn7V(=gnS@2eWSmw@{#RwAps7x>_;K
zAFUI5E<ToU&n+{l_j^ANn?Se>qs`sVEgJ{$7yr`wNi}-?nySeMJ(VKNW30+sr3Hpk
z;x87a9G}B?u^q>l1QVP-{KHmkxW7m}x<pD2Vbqqr(r)G-e(V^0FvkDQnXtO$i=n=-
z)UuoZ0~G>B8%K{Jw_K4UkH#q)cYA_0_y!GXQsIHD*27pHCQ&Ii4duHG+JN2kV&bgJ
zHj+D^T6lld|4ONQpuGJO>R(u4oMMWABR~lS-1mU6$Sbmk=}tWN?<Nq*8$n<+f4~rT
zCSE>AMYO6&1luLR?vR`E)bj~=Y`!tHa(x4^Hqdw6%-`RVQ%PyuY1v6st!ep44LWR`
zddkYmmXwsd%6NK=IcG~pvNn73rg;4T7~QX}aJN)&J57G`?tUq=r;n?&i@{xwOmYe@
zztG(%C+b~ss;L`{(1YxS0(uc}nm`C6mlXqHwi+YT?GE-{_yy@uf!8&!<q@SH?bvY8
ztNt|=amFmQyw&$U<)--`ra3pFf6?0(=Iv;zX0PR+MJsQQl+>jdo(hRg?fsB?3wy_L
zIw)_6V1e6>DywyVW<UH<3XX}t{1-Bo97Bi=cQO+_`FrbavjHfQaM?=uyGHTfvh5Dn
zos=^86iFuGF;XLBUyWhuRc{Pxyw*wa!R@#7eVdo<R4(s}_8&vh)Ync#$5K8rZ^urp
z`z0}LNz$WKWZO;i0a2^NfY>B&Cz-ZMwQ(Y!=J}<@YRDS1<ZziLW=+3QX1G`K%8@I+
z`=w6(oruczD_V$Z2@L+w`5_~n$YV>R3VQ)1csfMpP<Cb+6qVukJ1EISwTD|{O8NCj
z4=jhUDH;BGS~MVeBj=>TVum+Hm~oRCBJG^u0Pc)_y~gi7g+M^Cu*(I*$m77(7}MxP
zpV|CtjaBNob)TL_EdTOF(W&Ku#EadIGqa{^rI|y>X>XL-cY|M-C{$2%=D^vd8bOaQ
zSV~S`?`GSlb-8{ZCMGU@T;@Kkw6d~j-6H5pLq+9F%M6KgFg>oD=r#MnC8ItYD{J;|
zEzhau{(j7?QbQr`M+a^A`VJQdhQQ19Za(8lfU=Z)Y9FlHiAE}d3~^yS4JzUgGcq%4
zPv(c5pP-_?5D-Wdy}yYDdTIkZy9d-F^wXzD23=oiFld{Zo`{ov?b<j{dk*%}U;FD2
zU+p6OCjMEpui?FIA2VqI%GTc~I~=5?9rJ}RTjGJ^b(-|WcGq888*P_Iwr)*mvYWG}
z$)9P^_n~}5&DfTEY?a(p?RJH|^_3%+EDOr$(v8F}H$00W?V(!jUA3wzE-zd9McxtZ
z&Ft^v%<BXYhQwQb_i|_3EwPCMRu>P(ys4;8+^B@EjmxX2E#$i`bNs}7RhxV?KaUpG
z?re8$baqk52E*;xFq(>@zD+!Dnk;^%is{?htY1}fz7BzDGjfaK{<FBH;;VBFA?T=$
zk9n8a&4|v$!aVUD)%d<n0n9HS>yF<!G;CDz;El;9o}y(<%RSC=dco~*shrG(+VCTK
zF;pTFyX<akh1x9Ai$$BURT+_3)2p5N+v}FpGLoBlGgi`15>@N{&MQO1@ej%gH??ys
z`-U0Y)vnF(ikz2vQHqLS8H8R#MvUJ4hwW|xgzv+NwEaJP9dYh@zW&^3SIt$|?0@*h
z9_J<Roao)f4{v^Y`kbjVe}7)`_-lX%5^$QK!;uQO(BEGJtY3j3gW?HBN@3FD+c}?P
zz;nD{*+Z??<Y4Gx_%qy$<fTL>q<yJ=y;0%-m6T`<+qS%_Q6PY%&$3)fS+i8WMMVV1
zVoytAdIWRjx?*#Qu6lBG{@jXc;9jyu>^RDVC35T1yol_8Qa{#Yez7)ZRs9zCp{5)i
zyltHuVR^T|Ekhl#VrfEcO;{a~o;;y-BR0SoS>jaQHq;R;Db|2BVq#S>v^n`<<77wT
zdG@-a=Pe}%TNd=4;qCIqw(P5Q1EMN>jd{wk2M)0QC;E@`l?)T=On_AZTPr;1qIZb+
z7jc&~cR$WTJ)Eo8ogLWNpSzU0a(SB3{NyDKBA!0-3R2ckeo-I6aWt66JyFBIfVUAG
z7+-eiI2v^{fTm9c;uH|I64kFRosY}<OIyaR<wKrB)>ec7!FmCf`9skjgnQVco!9%5
zQpOh?&$Xp4=4y&SOx~pv43;QkgoCg+_kuHeacCGnBCg?-^wS_5AFpFab?9N&z>(Cb
zt-^uQ!l?*)f_LYPY^eVAT6>2w24^)PKobv8ZB^E%Yvw0v1P2#4>ASl#rz%jL5%Hk4
z3Uc*X15l^Ak6Q{8;)ir&L7iH&RZ#9;yQ2P(b@^tLt1f!DAN9U}6pP+Mx;r8z3?)3J
z@R7P1p+MnYIB{AQqU*uCcs5!X&c7+y;-h@(0yl5tJ9m0fQF7!DdFHpQj5U|0cVn*h
zruO#si3{LmLCC2U$EQQ)@2#L^!JgQw61QKR4QygL#&6ys>P+~A>|~;UDm{8T1q0jU
z>u5z52m3Yke4<_+-lF$74G}gvR_8uS#iLQ)OD{AP_0lO=@IlcITd-Yw18`9-E#Bm*
z4^^d(4XYE?M2@YmK#IyQmy)7`+_<Bt%4%96ss5!h#ec|L1|<jdKA_5df9@ZN%59B~
z*Y)iG#ROvqCx4YRCw*_PPOA#NBxJh!2>)tV+E^~p1HC{Ha%^;QQhB8E^-zHOZ)KNy
zCC%k)>>vRrmMd(*2Hn<RpO-4LGpOL^{JakOfZ~&>8b({_Qg<j)W3cj3@wQLIN-Y&0
z&=!u6MY9Gb%Y>|>qG7P@3LYxxMbp=<B^+`L8{I!IhYfs2YkO*->bjgq9q%U@5v|mg
zCaM^w;1cfQ`CIFYyxmy3oc@1ckWeSM?xa9<0F+3<_<HGp<SVxt&1<PVJUkRS?&Vx<
z1pb5`0OPUhGU|ta`NHM;4J^18=yCUtEH)lix92G=F_KyQM10Gxai~UUzntCw)W7ee
z&BhL@mH9dP%e0*X^R`mqDzL1YYf-~l+d=JB6kbtWONLG<TV~l=-zAA|U365LUXj~U
zJuRs=1+Qkh@K-cb0c9dB{9B`EnX!wLc`td2Mk5yLiBdXC<IPng<Iv1Fo<LO}o=91i
z@>HXjsyk%E?MCS+7ETQ$oPHp@NE<3AoSZGW*m?yXuJF5GIA^93`n;`zQ*@&f7+-93
z7kd<Hk*GMMGelz^6A9C!7gUp~K6!6N<Zclj92Rxz{UGZAcS^wU2*Dl@-W&6cqQty*
zuM0JBu(4YJOGXH)R~SUhD@hw2m7LtW)b<$&Dt#|Kfo}nbDSzIaH{M<4d%vG3G0etP
z1wYE;Ro9AHx#w?TFt&aWEpGM@Z56mWCfdjtZQX4%F%=K|p>7S74|piG1Jb9e9hWU|
z=bxgzfl9(fq_?_vR5kR5V+SYo&X@=cmW;deEif5?OF~9GWPyz;hC%+WKKS~G>7(hG
zqg+A9f=XpOv6B|+L>o&81J~+8{u5MW$~pd~<AbbnhO>fbUDaP@_Pa2auTB24Hfsq^
zs(7;_v*JmpZcn+Zd=FO7g)1kTXlS7wQ>;XAv8K^|e4p96&^=%t3lve2M9EYSSB9Nt
zcdc<75)6VxpZol7ka$X_OsjOzZNi~Ztk(xzbdxo%OmLs^Sd4&5GPdY%y_Y*^4L>AI
z_6?DP+0JmX)D*7Vdc1$cP~<?7uQz}HDG|D`r!2E<i83rbGU<ke5@0hUwl}X7ndd)W
z6{~G5qBmC*Ful))8~bT!SeOYm;w@TfF^IjTJ>otF+e<{_KnCp~&uqmsN7#EDwV_^m
zu~Podd$UofYKet;mq|%PvOy#WoHhI7emReeV`4ndO>D0xM0gSu+QGXnJ$0TwJ}4}<
z_@0@@UEcyd;=P#qxw;MRfsL`n_7n29vh$x9eAm@DcNXJQZCc2H({+I^4S9T^n&%Dm
zPv#K<->^u&+V)RllSoa}2%K7;YpEU17S@Me*I?pNOL$2g<e2I%0mhCCUy59_g*^Y>
znC`#lPOZ0TeKXI@(y|M1Sm2_dfq6H!5)BIrbUAm1<3+k!?`H4P0zesg2^2y=0PLu}
zr1`KgznE^Xy&8c=q_R5VM8ACa0~^VyQ>HHbF?u|B@{E@*2|EaBsb#K=M7tD?q}NbV
z`Lt`6n7P*g23PJ5=#<#q`GTU-YVCA<U40U`Hkbz+rb>?$nTd;tw<EltKe<q7wfbsw
z7$Dsruo!hmLZMBTFubD8^h*aXk7B}aT@we*<j*tLp)sL$fe||wr8<dQ%A%(Y(r~dw
zW=vcDG@6DI0%QGdLMKp!`s55#ISOg&76u}b2OF53xUr5bSbOU}_R6=nl81@evDyXv
zM;36YxyJFwcA0?Z=K~N&eFahz8yjZ8Hk+OA%_a-Fhu75!DJJuY-QV3tjT()C1_J^M
zF&NX>YsgK_z4LABwz;8k4OfD)X}RY_B&L0Gun_ld>W<7GobQr#gH0Q!V62EO7y1>d
zmQ=~pYjOa=198B`@azh;soG@SIB~?_cO^a@rr^+@hDwz&K?Pr6$UQ~PN$crqa>X(d
zC|lFn-L}co-pSTK*~HQ@gjTg{p<8&1Ww=%$0A@|zE?;d;&m%gaAyPq}gkD&5o8J^d
zqq@1(Xygz%`&$D{p36E*_Ji*XY|#mC1HwAd%@QSEF4NhC{pXgmT6W-kag(59`q<EU
zVHAQ(v4;r8e+^WI;cup@rQMw!i=pp`EU@z5#7Pf(Wzbfai4b%%q_s0cI(fP*@a_Gx
zxSy|L^0~GKM#W_VO!5{?L~MHen6d!Uz!kHHgEVG;6QJxrGQpJDPVL{wed6%400-Hb
zV9Vd`Q2qNj8THX<b1YA6S0BOQvImdv(N8YNiH<{+7nSj!)y|sR#EBTv^pHR&C3W|=
zhaE7E_pL+a#L#Tnl!Ic?1QUzUs_Jo6?fqF0Z-LzMi9Yh?{oHaV%NE3ymr_eLQ0SeM
zg=6Ljwm12YA^nPjqlAxaGEpc#b4BVf(wMrNh$mnQx+6>8ay=^^q~)FpPZ00NX{>9%
zI>&Rg8uj%M#e_}BuII$2xP*^*N;5@E?%MSPG2=QpL;QiFK8b^dDcdJ}|2<Go+Vk)2
zs~(ymWhE0cju-mG0Z_aFkX{6=EaPXwVq-04SR->RBNnH57FZ=2o^ZSSN|l#C-8Ayx
zIBCE7T#Vg#s}_r*7m5Axv8cz_wfdHcvr}2X(Yz9Wo3)#CYNG%UJ3vvW+7fJG<Ay@l
z#?3IyT@veh;?*78!FKr33=*I<pi0+nC5W<G-|PHMWAEg0sOUBTQ|h`SnMhML)UK{h
zK3gXHJ{{x_D({$;AXqRF-goFMS3zaBNuwt1S<<VNjL$}@FzV^BTDhCbG^U?{+YLUp
zK0FWz6i=|~+IJrjp06imwnKo!zGhGvpp8nSrFpkQ%)bZ_z2mtIE~~}I-wPrK{Y;Y^
zX<0_xmYN`IBMc2E4z^4`J^dfsA^=zM+bjGH1{ud`#jj-Uu`RtqE2#XHc9<R?c(dN9
z5wPBoabr@^@jotvJ*LD=XFsYR@5aPjjeJj8(M;_jCfnfVosDG+Nu5$0Hu3lG3n|z+
z(Y~2P)SA+Mkp2B-WkK1QtQiZPo$fS25FQD^t8e}7boAm75L1qKM8OG{fQOi<=3n^B
z<A?Wlk&oMra~AV|xPz-Awa3He3;&9?6*L!QELlWprl!L|Z)HtctiPd(3p4ii#^WXT
zg>6gILZ=xGuuplv(b19;6-H=5Z%eus;r^}Bak{bJ<l0$|Jecr(`?)%DZy~eAoZy$W
z`qnKML5^y|hG+g4k_cm)ZvS1`pxU4P*mVHuND*`|^}Y2BCl^(el|}XRQRftp?oSid
z*QET%76g80P*;tN`dsrG#B>P*xhd4yD=#XprJDER7M!zHY__<v{7Ku{{y9cEQf~8!
zr!1N*P^U|nnT4d(N`&D=o(l#aUA<S|{Ib8}Nq+jSSi{iJSDRU~dhUa`Wq`@A|8O=6
zE(+J9FkumqQ~<Vo{rXh}fA*hlP&l8uD)Pr?_h(h7)Z>c!HM`u_bddHVuS!~P5Bx+I
zR1Pb_yd{;pOM{o|?y1>)>;14K9@aAi;pd8t=c&KVzt(R$YVudQ@IKrFckb=VaWME;
zJj0dzvmu@AO+M4nvpZ8fo~ytTO)JP5Ba1v~Rghv6l{7woYB=6029svtY$)T(jW{x<
z{@;UkjuH()#sfV)KY-KWIqqz-4%Len0Utj;IsBFn-pN7J14DS@%2{h(__Lry?dnGv
zzXvQ}7`=4#IUVk^`YX7he5}E8Yj!M1s6E|ewSWKKXg?VPi1IeE#+(i+9%=KhSTD#j
z01}hr+L;%~e1D$D7`pft6{DPwl>=aM@}esZdR;6`KN^U5@C!i`Z4jQ&6wK?TkRBs4
zzD?F`v~XzR6+LPXvVCT@{c8%kxc%TC)q{PL2jPlqfg1Yz5r=}FwMJkYH54aX$!i|^
z=<G?24>$khrOfyL?t}vboD=TAwXxbY3ncrc_VXe_LRKOPArIf4u?Mm-XiWZ=e4U3J
znUfIR<pU+zf<4QqKz)Yt<7=md2qv8w{MI_dG!CONghUscw`CZcF3d#dq@mXXGL2Gv
zP10-@w$L?#x)o`1C+G;j3#&e;W}R18-bmi?&6!-=Fun`TWd;h{QnswX`bU!Q#q2d5
ze*LaDlqIy0m8xQll>qu<s>?r!WNDnww0S>Vq12^vZ4-V1?5VFi55RbBFmCeB-HG5>
zzHe2M*|>so2zwpuyfdJCpd=u)Xsac)j=E^{QZa1S0quaxHu9lpf)C|^UisO@;VikH
zc+CPW%CEKHi!F~80K{cBb-H3lMbm4CVb2wueJ{Hg0L6USUPo#SX;sF0pa+g_urk|d
z$1Sww^e<&(_Za&YQq0(Y(3Cs=7OqAEf=N=6#Uu^(ECza$<Y|vDuJ-q4i77l$fV9RF
zzAXfj-;F0|ErvnkDP6j7N;E>tXpJFaku2li*?wJA@4t=zwlH+(^}rDV!cUeL4Nv!F
z0Tiuz=|F)DHd`?HaM}7HZJgMUn>cZ5d192RRqVgY|4<zdzi}lFMJ6=#SNwZ_7@BTk
z@W;HihASPMPsl}|6~8sMxa&PWNj{?8XVdy~c#LvN`kjKHZv|!U;^N|H6|}8MHC_(|
z`1j9;<-E?LtF(7V@PfuI|3@U$+@oMnfCZpRxsCyGT@_cO?TMw7KeoasYaAEfX<pZ;
zHIfnz4v>mi!JBa?7EH+PWG{117S;esyVVif%1;xy>m$R@iD8ni6@JBmlz+kjeqwWB
z;5gec)pX+R^O0+FW=m}%>RoN@-|ht*%a&=NAL9PVfe)&4xz`KqPf*2ym{829Q46>5
zfgE-+OH+L{?;=&r=)2^9H<OE*#*|Uy%IBp*NBDAZ!hD(4zDPf{-V08)h|dhUJHpO!
z198k*t$%no?#H}d>&qAHM<O;GfIj}};v$BvXVpp7o$X;woPSwL76*G6Tde6t$(!Bn
z58;RY;o&QKebCZDi*0Px-_41mRaDE%RobUEnzg&-&M}gWTZ{YmH+r?M+tN4T7J!qM
ztqA;yB;8Sx&D^^n+SlBi43tUHMA2&Q{yeYF$Q04<&k!v$x7P6u)IY^+6!ce%d)zXw
z<c4|(eI1k%7=2&BeF_>|j}Mm;I+XJKXbIc|UHQY!Q19mwMnpMeYsr3JLeA>0#;bpu
z$RHezxyi8c5tj(8{M#@GuCWBH^lGSCC*eXUSXH0WYZxIsoVbrJO4nk^U|?*SZsTN&
z8qY3wN$}h!2jn0Cxl?3ZoCbg%B{?puW;)2c3cRb+)0;;ASBRG<zUmGKS>4WBK|D_9
z^CnGlz6P|$_iRaaiiu}zAMPhY^K#G=wOZlaTLXMX9?!%+H^t@p__pqP%`OHn^*B~m
zaU?E~Ux;DUdiK0W5=YX_-eQg^CPV*=L=k%+fBfPpCy@4W-FKgC`G0OZ=tJ~aW;V%x
zLOPakw=X)EE=x!svnF(RoPQC-3k6@i@Hl(UT=!@bH)er;xFs9iWAC|*Y)AWO>d>)X
z$K*iM`M+xQ<UFQ4S;td8u{$&f&E@QDWyIO|!qk8q34yD5bPAO-0o~dek<yM;ewRS(
z<k<Lvx)hD5FwVU3E-vl<?q`GhV$qNVSI+3x6{Z9`e6&*O*-;PM@#Q*{_PE6IFD;l$
z47U^>R!WO7rQNoSrQyeCryD*66f>z`m)}#9%H=q;RH!nY%b{V_j@xB!7xwmLzCLMT
zB6(LAo$hOk5iuzZpc`uPMvBp?-5nbbsoNQYJyUGARLT4?3SQpJ@l+%h7cKMTTjF1L
zv%#4idco&L%`n#4Cy)cz2O|tXnxb2B6#(&TJuSvV-6vLaI!7YL7fikKVVrKzYm1t9
zPrtrJz43iRv^p!z5KGskB7?HM9lFy*M)|;>oux|Z{3jxs;pBp`MEIOAbP<9}Le><}
ze5<T+>{1=9(?1gaEU2cI!+X_JFOB<yhHuy=-Y~Sq`iA96JLlySJ$-$Buv0v4`pniK
zEh!n2%ll7egOLiPA{Bq*n<hzaFbL|s@-f5V0RN)D*pm>NJ1p)cwOr}!b6fr`QDI9%
z^J@f+pB`ImI6Lit1<eO#uWnR1nTT4e(u+MBkwt~~mb=&JX_rEPrT;#dC?*fTGXP{^
zmysDQoPQ$pqC<|R&IrrN|K^L)y4x!QzTv!|UT^Z|WlFnnP;@adb$rv14qxQs!#7$}
zT0>V+ibAK?YLfT+u2C}&`jjB#&K8zwzNgON2a<L-0y*{-U9;M!r{~MHB2gCQFuf#h
z<dg<l?jnQ?L-tGMxR{vd0xla(KcO_=Zo&OJiRJdN%dPxkKnb-pH^aZ<zPCI$IpK1Z
zeyBnY<G3a(vS?&+?KBMVLNO|7mesfo1q|wP(-xUrc7myKa5=J!9y<W%lWuHCOO@Qi
znh|o_toB$k_lC|X5@x<cHj|<d&O9rWl@Rk}1?z;~mIdHTuPx2h=ON0mMDXT(=E-E{
zt0aMrO4?A86&8!9?T)-NLlPMZd)T}{D>Ag@;(b%g)x39(b^1s4?RCt!^CIPqn?MK^
zKF+%`d(qZfS8msDqr+Ae1zNuXfEwKy&<`LI=*g2O=@2;89q1i_wL+t2FM-qio3IxT
zK!T`&r_&GedoblT=NC78Dm(5=nv)7A$p8gpb(OHLzWya=?d7fI>DH*w%|YvG<Myd1
zFm>32sD)Nwya%a6E?Xm^_<A1xAUA8ivgqaYca8kXYG*nojRFG@z@$^}iL0rp`ThHM
zkK5C66f`u>jc+1P07`dyd|X@TD`jcPP}~4rS%r;{e-@;Ec=UW;={osf)63ydC&mnS
zK6bUF!I9fMO=st<6Oy=e%hYm(r8wRa<a&!LIPGMLbPseqR<z4#8tTB6A|JoPY~wQF
zu+iBmam-!)6Yy}=C}!Nl|FlIXN6Fxmvk7<-YP?WDlHVg}#<AupF*={CO<fqZf+`+@
z0qo}+rtf^OPAnDEMEW%KecotjjM;X<BAqOOJH!;c;i}tPz{&s~I1c#OzW_FH2yp&K
z<92<smKCsa*sksEEI9IKW4xGL(0#A2aKA5!x9t6~8BhdtfESijHr(dEkvIc3aE6<U
z<%p^ZC0SXcyPM12G)OcJ;Bdf}a3UC<q>M~M^sk?X14=qe${vK!Hk({5ZOsp;`aG(g
zVugV(`gYl3HVq)m8{TJg5rH!1i=vL~*VrUoh@*|z)%?TpLczmz8jG710$;;vOLn)_
znbR(C&{8kHwhcf7z|pY`wCR9R9nBNf#`k*o1vrq1iJZ^G?<wN>_<zksz@UMx2(7|%
zA^O4j9A7f4VFI-3y1K?%%bqxvo#`r|L^CilGE&Y}+i($(m6Hn%4K)OB3Ri^i9`GR?
zz!@4482G@N17AFN#nI8xH$it+#@J7tozkdiXz&*E4Ir_mL1@uycZ&PO)8`Tz07(no
z;Ecgo%?Yq{$jHb@ggl;Bz``VdEUt7#UtC{11L9UPg3R`-&*AyG`Q~sI5Nv~m)^l#y
zm&iRaF>x@aZU2m%+`e1Fcb@(7$sA<9_Rr8JX0mG$SP{+A*7Zg?2VR@%u1lEC=b5wZ
zRL+tu|D<wcDqV&3K0QR^rkT}LYOS1E=s-G_84&m=^&q4<EK_F82o<iR(2zr|_^-2D
z>oYg++K}YTPu6>BS`GMM0RBklIefeN!J2D$n0vvqI@=kM`_qpjxa!pj!k0rS|GSh6
zG+9}X2fIhxa(r!AdM$12rNk9k5ou^9-p$rKX@Ex!-T(qVgsNCJWyaVCdv-B^T*F<#
zg-YPE#Aan>B@y;~p0B-BQB}1LFzr|Qqa21U2zYFIy`V;G-CczZqsGQ+fvA9o5R=tk
zm!Riq#KXgD2X-1qcXx9Ti9t<G{gi;fVx&Pnk!v%&adLY4xRWa40NB6>f#qhcKLyfW
ztick9rt_GtvbW!2{deW&sG^@ulyQUUrm2BsW8F+k)GCFG67MO&WK9b#z?_|}PT?m~
z5cT4Q>1Q*sb4?_8%{O)se8&u!K0E?^vD{12et-UWLu*>^32_+t6M#BRm8iYRDoEvs
z07{Ka(el#3z5c-#yyV`D*AUux_kuDekgSvF{SM>Y#^Q7u>*2D)*D{??rSt_0Q`Mys
zjzfLH>h0&pZl5p2rP*3J|Lm`QwE7QPT5O2XrI?*?uPo)NVKZlv>+~<Uxv5QDsDH_(
z?&88VTkBptBp~Rv^9lr}ttHvd#m&@@{)iCi4no1X03IJZIp{#ce7e#WP6+tQtAM`1
z!Od;0Vq9-|+P={ggi_U=O<7Cs%Bh3B0<Nh&jQCAJ)uZpP10U}1NYue1u*irO`L#C@
z9Ya4QDifPJBW0swg!M?#IS6gKf`IB`#ZDXf8$Rkj?}raF>D6H|H@1dox3j>Tcc4m-
zFfAtj3<7Y{Eq`aPiOc&8D3&t!Q;!z=6%#}Cwke8!d62y}F!wMFfq%gBsI<6S!QG4@
zUDdoz)@^F}r5{~#EgyT^(`yohvCBM)cX#{AumfX(*K+pg_`mf>F;kViYXjfO%D`O%
zwwT`&^RpcKz^2eKGm{*OLpNNkFRB3*Xo&%XjDYT{v&jqrgCOun0V5RSO?8w_SZHXj
zx)AUJ!=V$frw{Xh8U?=qcmv7(o-l%+AiqgVi=eU+L<9g&-{V){;Mc;lMKMb3x`)!M
zECuHI7Qs4(%(&*6o{IeiCzP8+gjWE1VzOOZ*$&aKeKok5^%Xr3B5wXEwxHT)5tu!8
zpf0zui+ZVYJBNo=(LB{vdE|QHIi`3WJMxX1A>rB>NDR$f3d*WuHhij41j6C~*e$PN
zWEGq5;*%>XE68*~M~b<sHVeHO``J9C1H^<Df_YNA<M{tUBzp|LWGeh+oP_TrPgT=Z
z9QoPE>9nVzS^-b|%xAwqA^QRXen8#k1TZxbubqw_#3L+x8}6#2nzcVFkb+qO(_I`0
zLm9D4kuIbyEzLCE+=5*q)brBH^z`;}dFnrS9$^($ZM^f|5@6}QdGkhczcp~Fh}uI=
zR#sY3aSaeGI&~h0b~Y;iu&^|}i0s!ikU~c-EiRsJ)K<&bpDZH}Iy~7d!5N@rQRi{D
zUSnZ#aR~^$qiA$K3Pw#xlIa>(wsZ;ba^7c5N}F~H0BnE+6^-}cqaCZayOue~Uk<((
zl*B+HIx42BOj6oNK3p0CdYsq`T6%p|Wx_aSx-If{?Ns_ApJ!TDaC84%7U*dZQntNz
zZtuhu!t#+~5)M$3xSZIIk3BaWO$2#iAjl2G>I8w5BX;XC=74~Ji<{$t!=s}bs^Ybk
z6@(_>1hz|}FZ`a?6K~DCu_AeNCi%VUMRj#G-0$Br@2-FITw{}y4<M6!`vLD<zGg8k
zh+Xym&JB*OmNgCN%HM~KUE>x;K~ezdT0pQ)L1Cfg%<q@LoQHsbFf+CX;wQ{OERTGe
z2x*w4s*L@^+H%+j;?l3>znK;+QDixRO4FB_I$o{eD-1bbb@8+;QV)qGYp7h%3`E7n
z29)an{)NicSVff~Lx&iZP~HyJi4Yl@6}Qx*=`yKLg-AKarF(F0vt*{MtdrS17^{D?
zh4g&LPlc}<O+P#FUVcVD)-wHT*kY}Pv4P1Xf3f!S!TU!W27`3Qhued7XD|PU`JE@9
ziLu<J)n8*q<n@yA-K8YN$k<a;B0QnYpm{1D`<;UO1>+m#F<e}mCu6t%>DFwuPoKVf
z_WBnTKo6Yq=`SeWM9w|Vd@KHhA?QS4_q;>l#Y-Qvudn#h*1KB!dc6jO-OZ#q%;&uH
zVAmeU>sdWJGV&ZuAwwoh^**xs{*I6BD$G1xvq!B^Ko-{4NG=J#Q&2D(z7act2gt+f
z4%;63fec5?V^iXZsar#D?PoJtn$`Uh6h7#2x^`9G_+d_8^@Fk257_(D!1l4cyBii3
zhQSznw6+FfY~oTCqczwN=UQ6CkuXU*K`<N;sN}iX`1*=GWR-zE4D3|pBEt`<V?=FT
zX=e<?5tSkLt{P_bdp$uN;mrkqORhL}8c>lr7pu4~=J~&QAczPQGpn3Iexh~rgfNS0
z55Cl1xmd`__i9{%zy*n5@khA?yA&D=`q6j3;>d~J4kpsC-d8U)m)Imqa8BHK_xX4j
zf9ogEcz-`R%x(9xbbl$}#6hHuQ6L|H96^sbqmWx7iUUMGN(={P=tABIJXuEjOHGgE
z(rCPOmhaxO{yX!4C>(-iEmefxBP>+UyG8!|JIdcqC#C4X#~*u;UioEfe0+)l_TVY=
z<M?|D1Emw$08m^Y=4+fGySDa-*T>GzF6$hNkRy}X4Jg5auggkHwMq=&;ebo6;M9>x
zFn*ZC#+C^ZJq7_uqXKt*IeMsdIJEKxfEV*S%G{m!1wF2)Ig*HN&{`<A0+D7<9Di^U
zD*V~~qynQjanih+^_{|{t<snxBI#GTI%**N(W{t3j76+UA@TFeeGZz?u<WJ6tY-BN
zqg_X+Yy@uA50jx%ZDAI<8k75p-3mTMU2_!u{50eoilR@;H@Xww8ieXENo&G)KhzLp
zA+`pT9YEcL9xOz7se!YazusE@UR|UVDqGbJKz6I75PJx-ed0)!$mhad<%YTX8vH4?
z6Y?1iXSXokwlwprO5F(yE@tm5h7m@&y9Mrn+Wm$e;Et1)muIf+o((%FeDnWW0t>FM
zz$1&JBn017V%XZfyS%sepmKuWn2Lv(;0d8>pVrn^_zN(A4XUd=1s>kzW@K`NbRbM1
z5%sMrFck*{6>L9ZKmP)DQSAC`wKzF{yZU5rme>z;&kn#4T-r7<{P%AH2&Dnyk;2nu
z56)6g)f$jI_{(lq;KBFR1jZkbH3Y=*vYOqsXNV;rqzO1k?RH~=o<iauk>Vll4@-68
zSta=_myv(<FmPZ^9-<$+{iFFK^$JcVPfLO|FuQxTA{u?HW)<JP<e{Q095&WfkY?fS
zw`t}kSr@meb~JOYLbUfPjpk&3zHL29HQqP>`K>dpV@E#D$9dd0-3hOqRLYcd)qi``
z4!tPAu)Pj8_`&UrR0e)Oh!Fd|XgOk~QGbPfn|h1CZ9KrSGtKSI<3ztJESxv<v(v!f
z?yj}2!kUfNpzC;GkT+AcmBuG*W5Yu4{c0@ZJ{BfsQMuh2a5jF8i)#nTP#w4D^ZKiI
zAnQ@PdY|0)!Vm;jy$7Lq51;>#5Ua5wU2PsJchFHof7;t0uwf9X2BS0sPD)UyD=&1e
z&i4oW;2(p&+&pWS4CJ5vDYyKa_r<-e><WDjw2*Z+USNh%@1J2=-MJZ;ouyMs1$m?3
zy}bU>DaUp>GE@IG;D=-uv(=fOs1VK*&VYqqV^0@#fBp7S_%g#ZsZQRbh=EDe^OY6i
zSJd-d?dWQR_3@8hYRSIk-M3Q2c-&Lw?nYtqS_!G1J<;p#n%%we`e~J@=%=&5T3JWU
zhtxIkh2_2e{`2*rG9N)q9mVWvbEa?Qj-24?Iq~yi)elK)ynQL9GW_j!9bSM<w?C=&
z%*fojr!?X3dz;s<f2`hh(WQa)h@|%rDrI0mK@^7M?(PmE3%bF!_5cNR<t~|^#mctT
zy+8g4CLNGA0MbVh!z2Sge?}AbIx_*2;Y0dYb0Bj3G#iM917C9(5IzS1a{kQ*pyxE?
zxUJwQj}K#kCxZ%fYzJnX!O3klvo*1AKI1(jBKia?Y@eS`lOlT`wA>MKyb`ShJB+gP
zED#B-DzE0CSHDroW6Z9cp16q>hrw1kLmtAVL%#Z4GX82Lr`z2zRn9%Yv{9@TGyHuu
zT6pVU|2eQUO7ps>B~J}oVtY$#a~6h={K*M5264)U1-jU&T;V%ojV15suRcl6)7oGz
zBw)oA&&DJfSujQR50@3I{bXz@NB(m=vO&6I^)HHvPEA0ARz!ZWpzp5s;z;jkUdKy9
zh0HdZC?Ed7mSk@d;bceqN%VoSxxpv#NDrqjiO+1nJl>by*7GxKm4#R5H>^+RSTuxv
zDss9>f$F^u=cULg%N1U=(!4_{)adDqqS_uAID>&Pg^GjYYiDOCZ2xbS{d_6YVLlk}
z9^@&;?O$?Tb)5U#+S@}wf+mOS)`;wWW90#JcbTXEIS7L=odz)hu2WSxIi1~;4=bX=
zfUDH?E`mg_Iy#`B?$dc=!y{yOI!_@7NH8$17TM~z0AkbRG%h$@lOgx%7g}X)<(yhI
zI6v#-ZZ;NY+dVOjVfC!vQ`tvLHoM`fTFZ`b(#%12cHM4K&&KFl<Cq*+8<!OOd1SE`
zuVJtJTJ^6pagFy2+WKXzeH%`x4>#W3-JEDHxk+X)<y=&SLRu7y%h!KxX9{y{q24s^
z6jz3nSvhF2cdd0LTRoY3cW5f9-yIlwRu(O0nLYG^A7b0=#9@MhW(ouYF|~Uptqndl
zQ|^8fax$4FCJRdoAJgLL9=8Z^*68cEOd;{O)rf2T*+SF<tM+SjG(0;2x+$kizSXF*
zj*d<x>C?gAX)zC#<&v7@Bea^-A*y_}8c+%c!`D2hRBcdV3R#|fZnAdRu@pw4RT&HU
zo{_NBB5M+B8e2<j?=|c2J$-EMMER%fhWgy}(O|aH2ibkQzdxz)u{xNRnHdiDdf+HZ
z>^o%lF?Xh;gykUrTSA8_ww-~X%R`3>6R{9H*v3XdxdS~HJ>qf>33S1NQJLAS+rUl(
z7j7o;E2_yq?C*2|;~Rd&o0D!1MO%TK8H3`G)V;wOJxl*6J@mKyIs(4ofR#90|CvN6
zy#E#!QdRiZ5ya66;@d7nMDQ3562`o%p=6cCGBd|8*`^eU_)=!M1FMUAtZ#38(Q}UD
zYk%*_XZ^4)KW6tqvy-97AJEsDDFdy^nEzmOZz0Y#_*|EQlK0B8dSQKCbN9m9{N`-m
z{+5>&WO+X1qJ5L6O-@O9*Wg_}8UcRHQoD=Asv=MQ9XIU{8HO*4R$0tWXXkN`kfK#?
zZkArW!AHGVB9EH6-D?es$iF++==it?j>J?by+oYb(1AX7H&;|6vj|k>-te{7mRTQ4
zPDDO3PD&cjrksas%=DLi^C4p~7SqyM4!OK2{Q2HMzyP*^I`%g|ZQ;e{I9b7vMkTd=
z&Zg|=7d+scb^i<RlD9~%BpQx~LG53`y26dPi(op8j~&Ct(Qb(}lZ`ru(JrWAdb>vD
zOQ~LN`9so@ZxpkSb?dENNqnNh$y4uxHdv(O@Me!SQ_H%qi0E)zx~Wki1>@TSHkq)7
zKL_#ca??l;7Z22PehqySTVO3Soz||hD|Ea8Nl`kL!T^iNUo+xJ05<v8#pA~81ndL_
zgUH|{^4HN0zV=*6dQ9Lrmu0liwe*DujL<fI{#!9or3-p=$txp{#EZ>&{U`_eqA6pS
z#(j+}{q24i(=jXEO@)#RA-F09;?TTLJXpRNz|K5DsPbu{HnWBt+2Gv^lg$k$cY7h2
zm!`mLV{?hd4yq;-m2l(ykof6l>)i}nnl@cHI*c#&@M%F}f^=DYo<hVC8gniUwUFDJ
zp(~AK&CV(%X0=bX!ZQYz1T&kezkZ{2^M}+lB|rU!?I&ix(y^K5=&L^}Jlp+E-p0ce
z95rK`5sX#aw4$NT1{)VgySG9s6AUQB@M^pW1CMgWhR(l-Ht4BB#3))Fs8?e7>Q|e>
z*E?k<rlt(Yd6{oo0SOLrejpJz0B)fZa{S4w-kGZS42}$VK($7HKRog_lJ!ycX&0|7
zhed1ng-Eq{^2=d7j2=8s_cT7$&CBb-NgBtt+LG=N@;E0S^$EU=)Kb2tR%k+AUvOdV
zlYl)`m{Ck}9%msQn6_5x6G*4<))CHEA(%V<m%F3EU}K&>AC$3NV%Oc0%eXLl=1{A-
z8$Q&FxfUFAx_G)zWbyVMW_@)!yDuG9ji9T$Xq)v7MuiK)U{lflSg@B~O-959@8vn#
z?52_2A1axK9C$&(c-&iSNX`QODT#&0W04*wHoIpy;H^At0?!Nd7F?a2j8?m2{f5aX
z`S`vSH(WueDT<mrW&^3h|J0Z;4v<n`+W)y!osO3`9uW<X-DN`=gtMAe{JS9DN%ys=
zx?(R81e>E>ckYWGrCL3<C1vPmGjd+R8vQ@2-U2GCwTl);Jctrv5t7HGL1`pKMMCM2
z2I=mOjUa+3prmwncb9Z`Nq0$i-}UPG{{LRbz%z~t@7_<X^~^QboY(P4#zQ9XaOIBE
zcu8aj7mE}X9PD_AU4$FJfBA$jOiuG@f4K^_U`JHKolld#vO2@W2FN>YIa4=A#-EP5
z;3Yj)K6P#EES+)qJa<kv@rjEXc1c6!f@Nv2`<00$G&Y*JV)92D-gooEGU@2qKF9O@
zmh&#lF347iy->70u9?!AQ|r*^N|x=)8X#OgJUgQZGF80}Q)S)LDZH#^#FWdrjj2sg
zh}t_aAQE^Os;-D+ykJnFYKg@>mLK=FP${}8kid>M>+ssD)XJ==;ciiEa2*srYX%2{
zDWmKJLiHcQw+3<?*HN@K#%I!+wn*G2!GB|dTl0OxW{`t>X8r`c{>+*O>e73jtg^t}
z3S+3?I@2ZHoglZ>y4UKaSE4TQa&LFwj6oj`O=dUpy^?5WLQ~I(Rm5+M(zh<Dznb7V
zt#u;NU~<h|{AS0a-=u4$radxlm{wVAE-r!$l!F_x-m$k&4TGdH&J|1gFkw1g%Ux)w
z;?Vx8b_(xJ(iFfpW46?H6VBw1pQ#zD*6f#`xV{3hO^Z^uhMv6jzRLp9^kw=&h$Any
zLy?y|@BfU()q7#2>5O@DE1xN&)a$)&Rsb#62?_*3zIVB7=pcJfG#)L><W&iPA<0KB
zTvQ#`3iE7GH_Lzc726+q?|26WUm728$Lnt+<2IshrFwX#jHUXaO7^;}em_n)YgF*K
z+c1+Z>s`;jDYIZElHoeJ<vj;k1oKs6-qtuJXe{kM<tX7Cpw47C7<04Dp{TB?^#ACd
zn-eo<;X+6<Ppg!DCrtF8aRQP0>WyRtiM=mz==H{Yv3FjW?#Vc;hSC1S98A8EzeiLo
zR@!&U5e(<wh|L>&@=vl8wk2{-Yq9CH9-jF^97LkwykxXEpC=sNjCkctAsoc`I9s2x
z8E)n8QFU+?GVulfT^?~QDaGiF{ONL;@2Z?Qqn-Eff`YqpnbkG-TG#As#!RK+#@g86
z%Egr!(AEOWk?u?-c8C4n`f^|A=CnZ4NM_kN$zSyTo&yg+PCAPt!@F0UJKRlfC3dCu
z=)LsgW?*NYWv!G!d;}jnp1ZY_M71*HAH6X?Mx%BoqezCogyhu4L%NzgVPp0t>w2$=
z9+5HTC$oj22lYlg(2r<7m#C}B6qaLI(3Eaz>1Mhiuryw|iL-gB{ltDe;{`Uk@;UnA
zK?8h4UBsw!SVlrRktJ5o@W|9)V@S-Xz!FOwCgc<Y#~jG#CK2z3;5&`a^?jy|Ui5z6
zS+XjlA@oZ(E!9_;OVFahuD6%T?N%@bBZE4Hhva;(+f-}u*TtThtBx$`OluBx;8(dp
zHqf@QEhus9D3CqWhX0<u-0)XI^Tn%6k2W676?P{K?4GM-56PD~^Pp0p4M&d~Z>1{<
z>_kUKMv$JQw<Ny`^re^2K-5c{p49ux7y4&|;Dka&M>hvrwIL%<0mrBVR9JQG2LK%w
z78dQif||4pdB4p~ym!l1S3`32l!uCfZoU;gKDwh_SMJLrK`VXh!I_gmq)S#2$zYI8
zT^YJM3(?c1-C={aMNOaRKKe3@-Z-V0ZJ!q8V^$225++wk(i7Q`S*_&2UA&^*Txsg3
zUl+GqM8GMkZY0l6r?n(7%jn%!;kO{C!Wm5iCTb6@*FM_J(U+V8_pmIanCETnAN1n4
zZCu?VJgTNb!*RI$iRRa?S5hp)#sxOkN=}Z+<?dyx`N#RN%c8DOZw|gp?o&woVX@GT
z+=Z^kM!3M4S>~I4`U*MG;}$KdquD{?d_mFofekE_9|e6BH9gQZ-^$P!k<lx)7yS15
zfKS2o)rUR2UIuSbGwr<wS3rx_z$_8HUB|V*nL$|aYFv#W?z2c=kcABaW!W!;M%Xww
zS^z->P3SU(9xlcE<>cqr_4FV;akDIsm6Wg<Ul|L$`O&?)tH(mzzv7@^D>3#jj^GQ0
zo|l^O&oj#|so_1o{rPp*&GJ2>YnYlX!YwaD)ILfx%X5z&-eAdLCi`Sl*OPCq^F|eq
zTZuLkPbmNA*L15#pI*C)uOV4sZ<M+3h_?^Mqy8gi@c7v6eMH6hCf4pF#7<rJ7an%D
z<;M7@sg>DSBjw!9TxRO0GQ_M|6YD7~K08Rqc$er%&WvI0z))rHkXwE+i9$$k5O3Q)
zL4;|G=EjWlk@?FrCSo5De>XBRmQd)QeKoeU>!6|#as?&>1XT@uh<$$jOlPUL&P+PE
z7!p&?FD`x^U+SWY$Hs=RZL7+AJ<e*mV^@8&K$4f22WeA538f-GKN_gaWbN{TqY;lb
zhAIgzTiD|rHGv<s3CuzGcE&7hLjx4rh2O1axT`zPqp(jjTPHQrlQxyHmKUYG&d&Z(
zT0G^C7OpF?ZoO(mMi<^=|G<DRg%)utxF)7W>mg`~X#a5496`{mh{g)#_~aSMeHV*I
zCIjpqo@bNDy849u%x_7Ee*0Kw^=XmdIe&H2M$AThU2bl*i;L=sN0A(xuWp@j)B>gm
zl`9=tRivUuvT?;wjAt~VFL~!&bcZ`>W!}}uQV~^bdVx`yV@17*)F3d~ipr$x)}1y#
z6;MQN*ckRD<S7vjsJ4lGICz}QzSMTS*?RFRc4!r(`iFlt8^%(nFP|YUE-nfmhM<ei
z@nm&z!w5v)m&q7k-X79QQ$Q8VQmht?mmehpxcv1l^A!|SS=|~{gTl->D=a9(C*H;0
zNl`MGW>8yNyWUY4Bl*1QQk<oR?z6HExE3H4-+vf4{m7ZqA3Loi<~qAFkf)?%^xL=x
z65mkeC9MUcg%naD`k#Zf*Q>t`4*0Q#4xh{uxP+haJdVGHPO6@eEw!{$C?6Hi4Z=Ad
zWXpGqOMh-`TnM!fYK|TB$S6C}QC(G(JbA7%(ma$yOGa*JeEz0CQ$Enny<G8Wd6_h#
z-(Ns-r!kk+QQh&`*qWhg%o{S&R`|FmHs@zD&l1k2vR;~UrbU#OVyNEUbX?XX)KSGj
zZ89&Ck2;6VjbhKeNCX3@H-njo(oBui`p!`AJyLB%3*k$O`j*%=SILf`HA7DQnDX%L
zJv_wgE9Lyl`_Yqp;Y~wTZutjD$3St#?50V?sXm`m6x#QD*3k`y0waW1@Al+3P9t<j
zp~gEyZYW!a%MYAr_BE}GB(adnF%37q_Rq7uTb53o_xzg%E55or&K3`tZCNMReS-()
zxw7g%T}5DxXCaGe$QWPLbA$I>Zk1BuHP87|3F=K3%WI$P-3PzM?Rg`j5=IYCYh7eJ
zORnGTai{;DPJMXhy<JksI!hB1xqKt!^^REP=Ekq0^%Io0n7|mns^3XLu5?O1yE}#9
z?>qHx(^^<G6aZXZ9m>w$HH5AVG!9PApz+7~ws6eX9#!fdLIxZ7a~gT}^If$!xKQjA
zdU8gp#}IN-hh~H!Wrs;hIm=NO++Q2z;+5B{wxz}|J4IflwCo-pd9uCsdEmOwxC<Ci
z{z}(SrPN+9dHcn<u>Ysyz3wLWuV3^tT=q4!7dEV?_7?@IF}4^PX_$Ri1x(#+-^Vly
z!Nsja4Ku$<=~Hnc46z8__`yujjI#<PWvl+VmIb;s1?5N@sm{jl8e{3#G+UFOJ(Qit
zeFJ2!UT|U7e7k!Pm)GcK{5M(jdK?Q!M7lK89UVI16u%Wv+uUX9oPOs#%JA~|$`kA%
zL5x24e}_NU$Ln&IIoMXt+w<i0{CY}bktdY`x6q6?X(}3BADQ|r?eC|kw>$c2Xv*(3
zJh_5IMnzAa${p!A-1u11&nwKuT5$*G()Eaw9h_t~vW&B;y``CH|NAKu|71r{uU=D%
z=e}G)j!S8YCd3!+G=3-ZbrtK)VuGlg2V@HkFU;xrHitv6=g*=M;d9<i*2x;UUOgXs
z?;pG+oX;4Wy`jGt<!sp~bfP{f^Ef}oeXvy2{R>(9W5EY#1ohY*lsC7Yx7eA5SHbYZ
zL_LnN<n>Fx2R5xp&}>`|ZzqKt5tuHPJ6NttKM1<Nv3PB>SKq5G-NEw*_YZrK(9$%w
z58{(=o5^_6cpA3~5&uxTSNc9ryT6X4T_?6n!$lkAdnaEMBd+<B`0H=wg6Bs%X3W|e
zNM2~N#+H3^w`^{}wIY=d)J@a;NlaX&nGaH*hH&K13%Sp$l2I}eo^Q3}i=_Kc_D(Ga
z<nZOM#33R#+kND9W(e-yJ@0k4@?KJ%HHV;~If+CBPa6kC4U0#*MoR_0x7Xgr&7r#)
z#dVGP)yJ`sfLP|HP$ceulbVXO>`mbdc!f8lMvLlLO)SbEkmn@|D97Yuk5Dp~Se*j&
zf}ucPO;S5WC^$P_#JQ^D{CUfa;0;eKoh!lmm5$$m=#xSBIl^$SVW2@0|267_D)xcn
z0MSK1qh}zlrNcn*%KDXBv4p0?Z^Qu~#W9f`O$C@y4vW6Lm~`59EJW^H#fjYI&8I!;
z?gor!9PXZ(QV|u(S26ScO;0!+9lEUZPwh0F9c6GIA4Oc>W*BxnZf&wVpI^4K$3T~J
zJIs0O?>QYCFbE49b<ZTuBB1@U`fAbb8tGU>iSGUbD<Wdm)mc3iGW(^~<<rx_Vp4@d
zY$z*q-3pv7=ym&oi)gJ(8<CT?T#39{DA?h;_}Y9_`avSA`+rF=`WLJU#;U_OSpxdL
z<+LArnvdv}inO5jaL)lwK|P7bw{>l!rM$&>#^}Ec4BZx6Fta&G4bQ5<?$(x}c!qOL
zcU)8rGHaqw*cfK9F+VwcGI7a1d02KK6pfvdN}9^?NNw{`WzST~hnyA0@oY&oJg>Id
z-|Nyoic>r%O{vw}m$Y+Bl(++V8%MLK)`-t1@}6f(Vb$UtUJM8(gm4X~cwIb5oA%Gf
zAXKw`?0j-RnVmwTt5>bW>Nn}BzoL1BAJ_Ekao<Mn)!!<lD^8>&j!Nq;2lpJ_5Y=?z
zvEQhs#X{;jcd=%FM-*6<m2+K=s=@|XLMJldzt>;RPpK(auXS@^>;xbcuz})labp5{
z^DzaAIWJwMVp>JY3_RLR-q>_1HZn$_^Pgr>&Dccz`(O(TEh@7@pPlM!jZ5Up9p()Y
zwkwh3=#ha@O~HC9$E31!=xhmjLqa4?NVUmsfk&#bx(HfCi-)Mkw^Zq0qAdCD5`@}v
zD}|&iQ<~z>rO3<N-W!qqCjL!okRa7V9mf@Uzbqg;_|?x?4krIBb*jPHnN?pkPdS#2
zeyoh{q3Ty5Y;O(^^}TnnggNfK-k-<8%HJ{DwW_VuW>&Pu6VRk3RQfMbOhYI)h*4IX
zGT&r-Y4<lgm@NbAshTsAE`Gm#xKt$JW`doVl5!Qz(a|w~CXgqpbI{Ruo8*WyZ9k%)
zCUO3?F`+-L^SJiBiW<g*A%|mmkH$N@5yRTX!7E%!tBVGU39<xg|KSf;ncwYZG1Jp`
zDB8n?GvzYizuFH*5iEJhz<RSyvm)+6+)gc)?^fH~Ey=f<--uuaVcGJDp9-LFu98Y5
zpC?}@Pwe_Wv0X3no`Y@Ma(9SRVcJ^0H)N<sMb-B2cD|}{4VEj^{svoL0NO=1%%$H`
z#wqnf84csgyuYnwWE*D9_+*>N$he*N&X?hn22)aHWQ(g*e<aA)eBUPipuTo&yZMF@
znL*UR>fF4yPbK|vy+HTfAeWYfGO?%!e!(4qK9(4*aq0`L11}Vxim5UT>ltSTMEnOR
zj3v++je%oloMeg&TvV|fZj&|_&<#f6@aw-dV=goM_Y<CHRr@~1YICK3{O(uh8j2`v
zM5c;-)AI%)=zHZFpDXf`9%B*WH4G5RXG-=JsZmm~)bzA|X)Jt?zE4$Y>P~#bbe=cV
z98IuqI1zpAigbLm-+!OYf+2-E$4F$BA}a%Zf`mnGc9S;cO}}9v&#c(O#?4VQVfPLo
z+0S>Hacs>PALX9@lC|kV#odcw2_Dm^`4zZf#BM5ZjI9VmcCN<5;U^t_eUON6t~08&
zu{fk1@!i+1@j!QacGySPJ=Fg1;4}q+mlC!(M`*lLXq}$hhCdo4EPdW1|1GQf4q*a?
z+o;F!yf=O5QM6zogPtM!si$L=Af^V5BI+{4>-AJ=WF!fhnu+CbiP+*447e6`?PIwi
zRa8;`&-677XQsvm!gpM=%#asuZV<nlXNcUW@=*E0{@sP?E?cs|NAYq0_JKNoY`AT?
zu%r+la7qEr;IvOZ*mW#rzecx&^Xr+5T*c$No%6F41VSgH<^7^jf7eN1FX4*Uq_cO)
z8l?~r{RAe0i~waZ*IyUo&)$a<)7Md-mg$~}1d}^dvrHtwTlS{u>Fq)vL2C5%)H>~*
zSAvJ)bjnnKvcG;dIZ%FntMd9?@4bwpYVoHlxKAF%EX<ft<3CB`<nB@Yy!BHGr!0+7
zQrfV8>}Y{cQonz*ddG|^eP5V>XeyIP2%$+hi+S=4XIb&Yc~x?#$NAdQ0D0#^-S2nv
zg)&OzSwDOKVX<NRYIpWQkn!E049qL3P49_?S%b|vyB@*h+oNwUXl>10HtzP$_@m;L
zE*r)4kZtv>`*X(J1+{31!il-k1V4>@R8f46`6|fa`c>ODi^ldkU(V+pAncmz_U}6|
ziY<>ev#EwdKEF0JZFJsDkaDf~OnR!r1y`-ukaDS<%=BEB?r|STxk%4V<O^S1KCyWS
zSl4sYk4UtnqQR1Zo$t){)JT9C3(1BJJrVOo?rIWIT5rnTi|ISs!F5qMKU}aF?#NV2
zf4^B(an-cRwWWT1gZJeZB!>;xKl{%jE;)j#$M)WXjr?UF{5h4DPfFbCf;XIg;kztd
zQ?qq1z4v4SZ_J4EsE;@JtCKt28i={}9zVK&ADxSfi$Sf*NqVvS265#p_xSh+&myya
zI|j6jJ@$Zf+!Cw0n6F5YYp~-O{cQW?At(O&=7)*(sbmaHRxM;{>J7uGIlfJj`!&VA
zs|!_l(LZa)OLN6zYlHB$DGQ}^<_Ck)v?l8m;Jpbe$w_6_HUZ*n>}-kh$!2H*(uvU<
z$*A}!lU7SDwQbVK)(?^LZ91Bm`>XXOcv1BeBeh>2uKL9%V+Y80ydUUqcu3BJEl%or
zDN2N5_Il@(29n;+FGk~bc6ZZ~Sed?2&ai38-mjNB&QUYk?a!#_#lhiyG;_|W%o&bj
zOse5<qrjGiuesrh({F&ZxQB05i6tDKv}Hu(c{wifi7lyA5>p;M7EJC7N<SLU;AHtI
z$hZlyD@S>$S@qDysAbUJ&}+itPL-HU70vY20#ctTJ}$iBc_lu#^2h~OOL<08`P2XS
z0dF4AGh9>T<j;}r;F3wKWG?9Tw^+EZAS@r$(3E!5#S}RexIFx90~2E2$7}`~wIH)?
z3k^uBm$quwQnLF-20AB9xGUaYISDqN_j?phHTL~FS8qD$q`sZ)6>$o8`Ix)RNYY=R
zJ_RG7r#_4@n5JQT-DJ}NjNfId1Kh8W<GKDz`Qvs|nS$EX-K8&TdaIv=6|d6j?b_~V
zDi-<=;(+-oMu*#J$wj<Alt6D*&QlTYH+>5T6E-)w2cK?Nw3P_^3y2eqox-ET%WPpl
zTGNjI@<scD48Ja+5DoH>o-MTzozc~_DVTe6*)kzQ_@Tq9<gM%V{sCd10UOdaWv@VY
z*5@mst-UOeYn@X}SA8WWH$Qe#^XBf4C$4aYd>(O$NKeF;=P)$i+xs`BIM3`={=e0Y
zVt-5FEasBo*KZW>JS?!1AP~w*Epg1O?fAEl=xp)Y>1iVt@;E!pdb~imToRhyGIOO%
z^jRD}u=Da(<w~Qlpa!lIjEP!$01+a6B3rGg+3BT6HuLhi-wA)s`%ySIn)|4VOi9t@
zboFCp{^eNPJ%7;=*x(TSW}k)*SE)zlomr8XrM)mO?ss;{S;GPV=T`abKWip<`3Bwt
zEp2UeNL}4!DEbbh%jf2%R%jK2)A@SEj@V<dh-PpCkefaS5G2oT>%Gg-JRxkdHxd#z
zpeDW;svA|?m%lAc9r?`qK6Dfo6tFNcGl#O9lA{>xPoI8=F7r^7ufP+cca;vTpxuuF
z%FM|rDPEweyk0VU9aY)~vLX#q?*oeefEg_Q^uRtml8xBBiI3j^g{*eyTW@Pb*%?qq
zJNf72c>7hhXuXGi06K*{R6!qp@)Gz!`|+J~fEN6#%&e0EzlHjYj@f~3;>sG7i84z<
zUeTRn1Ba?`*Ef+&qC*`;E4-DDo|-hu5&n$>%Hj;Kkw<2G0%>`VU)r+oxTVNGiN$F5
zePA$Kdxd`U87JpcgMKQM?f}$^KpL7qaTYuSUB!}T0VAifEr;g;5?p&UU-j4;s588L
z^=d2n;?xt^t$q2rU06>6`vn9JpH}V(C=*y5+{L+qbLBEh4I8@7=hwN+Xa1oSLn74I
z$4V2DljW)Yl^?hm)MMEOCSShU^iPZ8Xy{CD_(GTu!w+R-u+X9}&n_wX0iBRVrjx2B
z(f_g80d;Nxd1W&zs|>S>v>zn-Nlocbm7{&mU<|w*W@cuyj1ukyRnLm#_7>ayx>jOI
zF@{`S=Snc*b4Hg#l-+h`woejho<>x5oL&YmQ`Gv=z&Ga-ry)7xro-+ZC1Xjeo?(pv
z%W7g*AV|nMAm9}S(+q-a9;%hCYCOee(*g@!(c(fvSCR80N(>APA|F9O(hh++LB+tJ
zGgb3pZec;r{uC?)wt(kscC@Yi6=3gx&(n2?=MHdsa~%&%K-inZdgTr%y}@?_*P0Xl
z=i&gl@JzeDMu)O>D>1tV647)@MPD;A=-~TNnsD{?z<<oZV%ALt=cE0#QT>SuPRE1w
zSIK<<VsSd`QhoUF0hsIGFv;y=207n7(2@0%*ZfBP8Ww;FjZ5TZ2_#kahp!ejv$eG#
zROk!?Vuw?3EOIYx_!C^$Z@3YV?37vYkLIa*9G{+y*+c9?K_4|e{`koMvMzL{+u2kk
z#`N!m?$(T4aawB;Z+$K=(H~=|3I2Q(u&qF?3`NW|x1d27faZrbRk|2OE05OE(J>Fi
z@tU@_F9^7c&P+{7*hHtN)1nqLnENJ2Ie42^9VPwzJt+lLsrj8QzHEsM9-!}|i3ZbA
z4|&&4Ohkcp+Edes3dLwmIXMEx%H3ONs0{`^BH9Ndzkc;*sc@a2OkM<!QbIe4INw<T
zT}Uwfg@Nr52qIo5=l1gR!&@E9e87M(_)bkt5%1&(wx^M2H~!0P2j}sa43J3l2Y85>
zta4a!bt2q0L<m$#HZVnhs#}ZCPZ}0C9e=o)*vvJ{VpNJg;-Gk7;5V4LA9Crx>@S0n
zFJ4P0EgXM=DY(A5Tyf86=orYzb^WQ*mkz<XmS&94c3l~b+~siE0x7-#LlT&A@OxCq
zg2NFwQiQI1t3&ekr|c7U@xAZ0^)JTY$QhcUOGBMbXkc^$|4=CsDlq6^75UM^V^}IG
zDvBFwyJ$wg3Q}v|%k9~*UddP6gY)I5R$3Ag8!GVPMJBm4tnCuH=$1Mt0%2o!Qs>)n
zanvE{Yc<LK;)}BUO0R$1LemVZ3x8@xgE_N}wgi$;pWdqe-4K!=1`D-%Uu8{ft8<D_
z$YKtCif-Nl5S<BJ7J;s>Bo~MQe!Mh>WSa*g35m=`x*cH+)fZ<QBD2IktpnchBx&2j
zz^F;z-Otr*0>X?;l*_R`nA;<N1nlXH$K0x`kw`)|E~mpM{NSDdt#0;Eg_HHY3p!RC
z)fc>;*U1cWe#r8u`qchLH?^~~OZ4%lm!){62^)4sk<9bc?A)@z!~p44OAlKvH)s0u
zCtHP{2QQ`*`^=ayA&C9L{VjEI?N0`I^<LWif_z?$yuhVh#EP9y;`>c0?;OkC_Zu07
zViwljsnV1nhdEqqN|BnH3cadmn_w??=?PFuu};7`_p->%;(~8VK`R88^7Hf4cWuPQ
z#X<MKNtCg&etW)cq{5yV4O-qoXe0`{kYee&!obM&QlJul=3@~1M`M8v+xj#SyF#VZ
zBC!_vJ6rla|H5k|1=#(LkQ&=7@Q@!Jf;nzg#mz3n#8U#WMCh4)?D;MHzXx}}i28;9
zM2;gf9kvw*ww0jUu_ylnWfyQ8`telr3L4Plt^~GxOI*846MnQZ9ZN@7Y%oCc=CO34
z8X<$s*?D4_&E@m+bJR)VhPa{ol5+k8JItlw+y`)MC&3Yi5|)%(T%4cBIF((>^yw8a
z)R!z`3JndNpJ9L5zM9B$`JKd+kFW1zUfzinHB`8DKADa8Cl|phWXgUWQclmEva5fr
zDZJ*u{oj^Y5l>&Qh5PZ(9wV;~W=5}ThpGzE@qJ|DNM?vK+1npaF1bQ04jQV@3AB(<
z_LMY{=D(m9WGegB@Jaw!^Kf&c%!)1Mnr{*?D0|haPk|SD>cIJUIj!F+IHwvow=<{&
zrau-vIgryqbxv5p_U?Jl_BdaO<)1586q)g0oy0v~!tw^h1dEjQ$=K*A|ILXUgylm?
zBxMibN>0x&MMmL2r~kc|ktwVOCXNOV`KtHm>7L>32-GC`GSd~S0PT{+a5Ul@m*aqF
zRN?gEA_(X4a9Azf00(j>Fc?JxMtEiK*8do}s5T8SdtsGH!y#t5>iXly4@Dmewl@lO
zC|lSFUgt2eNaWc-N!Pb~FVN;s;f$v(ux)N_eGg7Cu6cG!#{I9`8z0gGJ^?2Z0~wcR
z>^r>iv6XYo$->|LH`cwYBwqKn?8vo={};5PK#ZiDJE1U-YTGSeJ)`cKIHl#(Sw1us
zG#Os-1B2aPK|lPH{mw$JUeEoh?3ty>(sHDn*>T?p4dq%3hjnvH3w3!l6m<KN#Zi9_
zmbcQ2$|(CHIcUfL4+@9MJg>+ey@&VieE>E<+^;d3w!O^a3G^5+T8Cylyt&22%Oc1W
z(DOIJjy1$^8W>DX;AeD|AN)U2LZ*W8Hv?K;{{OsL<PULn%`oV-bG^|Uoi4b<Luh(A
z((;I>RPmig?MHti&X&;Nrdxkk{=1i&hIMBYZ)f{*H@+2f;xbunBF|fYk;py`Q<tWp
zt3q>l@Iu!&-uU9%C($yj#C$=%qq*B?kj;6yyT1sUg*@z4Z*Q;NM&&^XoV)3y2M9#p
z0nU*Bhp=j4W3GLEwt)lyIT)gTYl-<R4Vkne2ScINvebJ`%@H^mK*uZd2`>vvL58lm
z@c_)z{r@5dAS;2mn6#%@be|;y<@N08{w}GH4M2Ec(x0abD>S~2$uvFvAZN~v>4iL^
zD|02(wg3L`4|D)*@j7Ff!b)ZbuI)Nn*~KrYwT_IGDsNdr*>b>FK2@eYoZauo4<eoR
zAJ0KA3~E$h_RtNQun;?1P!`eLc4m>5N04=ZIM(HW$8YaHd2$0-p<pBfVN@F^;*g?E
z&&|oK+JT~IqG*W4<`nWAm5D+Z9guGeOvb~YC;Ka;Xfk@CF#`r8`MP*;T4@|^PD4#2
zcEF(YY=2Y%EWx3TvJM_iY*CitDT%0i$JD^Gm3CiAy7YK5iV^WK5g^<kADqbfT0pB<
zEi?N}L@1`Dq4@dQM&!!h;-}tV5hvu~NF8l=@xs4<XV*)_zG!Q@XMizZb6TAA(1GL=
zYludN_Pm%b2^EX$dG<c;xyA<^<cJq;swN3vX%B!*JY*kjv-a!e-Mb1!{7S~a7?yLa
zmP^;3YyNJ%{z|D#NI}+Ncc~T*W7R)z4O*5@blQp1=hem9!yw=z5~dic;8YJGrW2$Y
zMoP>zOD8B;|5OF~3A@lJv>!CK2U%z1XVz~T%u1(Edng>8Zf$k_R;$8y)|kgw=(tdL
z)n@V>CzLnXbz%s~ru;kKS+DmQuA?6H_zihuYBb)h^0R;ByOO_!2m3_kKVYa@|7Uld
zzbSSa`DcIaDh>AF;*GfP1EL?;DJeZ;UwGz0Z|w0^M%eg2I*Q%-vO0L0eLGzP?{5~e
zGj`A4_v)$xD06;0n^>8fv@1>jOG~*wZjk}e6h<Yqa}g962o2SSk`7x3)j!yB+UqLa
z4deLXg-lK^VRQdID$L_8fK&?A*(HQI)#i)qHS-5~$6SK2GvSwlc`>-lKy<(<JkEaM
z42D(hKb~Ip5xiUjiC7o(j)HwmW?>&{qFbd?uab<*kEoiroN}m9qTgUk6ZeVUZ!+Pl
z&BI)TyKHU8RJxg6BH7DBetG$&DV#HIbqmA84BGaq=!7KK=_VTyTWJCf7NU+WRCe;`
z<`$r{hrR*v^awuf5=tEo@>%Vf`p?=8%C6w1<OgwLW_A|Zd{Gx6R2Jy4t7n)DF24dY
z-2ik0NpaJV=66?yqbnb?x+wAWb+O${jhml<gm^T4B>rRx2bi8ZS6(oer`bxRt0O<y
zorW<K4L*;%L`>=oRUZ6De}R&SJ22;9W6+L9v74%v{#>8L_~#$+H>Qwy_fE*pjswUW
zgJQ4YsZ;n@7n&nY&!rW34)}7-GOkx6zFM`vIiA!u7n9ddB!)FSfBTrNA}<1X{6D7f
zG|ys+GAmDiPe=Mr8Py?GjKyZl!==9#0XY4$3ucluU;v;!x3%cSix;$HmTLkAiRu0L
z1O(e4-};V=48|!=Dsp3`!!>#1lhft)XPE_tR2fP50v|2~Q~ngyjj5?w>+-v=^|H&r
z?`AQrH;{<BTQ8BDC>%Ug(woj5L^E<UeLH^e_Yw#k|DHJo=kJAI2pNNo0Z9rmPS08v
z*JBE8tvcA=rYS#zyXYLG?fEt$@OUp1O^|hRs1lCsF0d0-mtBQAtDN{%+p}~Z0mY}J
z9iRAm?}_^T60B3%;1xP7Q8N6BoHkXqHu9Z4*IY=efPj1dyCo8JkBjs2&;t2F!_7ZE
z2JqrN4;3nFR_Eqkf{~fW=-KIMiP<zd8u%z_H%uC0SHl`8{LA4lEE6Z<uc2Q%az6Kq
zIG(>@BqUjNE|z#7pjsRk4V)-rCxfqsOkrN5=IBN;-A!(B8G;yRJkw2D0MaC!oE3*s
zi2i>oO>++&F*P%Lf=bKo;N!yqcj+oNHZAR@YNbPAMvan$ghWxcqVb+DYOTNYsXz~B
zHkrNy*_hY3i(8u{VY65#T1`jS?dtBqf^r@lC^#1SDtrqS)v)3H=b4-4z}b>vqWd=2
z&dQMpR@a}&h2KmJ<QO;iSBLzdUHj5EBr3|cx|$EocFY`;?`-!r8gzxK!)iw(<#j4C
zngsn;rN1*i7*RRC#C9%M=2IOM7k3}+_u5*AcaUSE)-)TQQrm6X&yocH+d?0yH1??T
zUwp!$9_0m(xRHGbV68>TZ$f__@2x_XqDI8yU^TJ^x2=*w1op9)A}|>5vmW1Cyx@K@
zGi}FotM|#>x2zjfFY$jz1k%2vdh(8(Zc^e_U;BGYMS>9D$Ok5i!v4O|l#D!0+Y$dQ
zTLEQ4RD%E<T-u`$W-;Ogeb8%~nmnOxJ*arnB^)y9yirlGEV8&bJNWZXDgG^bGf01u
zJL#DQ`{I67Z3?5Bm*u`q;@LN|V=?ki^6EQ+tx{tt$`#}MiEat^pIp+(%DqyfomgC%
zZo1}pM4+>nzmc99aXx<oYNa$|liUfcv9|jOL-n=8oMG}DU@-mXNa;C55kGBhv#q|f
zx<VUd04UGwz^ybOBt+f@Ie~U=)YY2U$Abq8T<3$OIv?hlED%177LhP9EBdf+0&NJd
z1cewLva^NP(b`(fo$xIk`F#LNA32F5t+END2^KLI39!H)Z3%JLFFbdo+;Ly)`sd#m
z+Y?1xzB~iM>Q`AwPiWOjV?79daZqLN81Y23Q69;lLz!pz1ZglG+iK;zTA3blbo5`c
z{85humWaT>8I~*!z8}v%5nxb{Sy_W-X5KqFJ9n&`!cno{HR14;mC?y9E9M<_8Bae}
zS^0;K$)tBW%a8{p$qJJm9^3khMd<!)>YAFH@8$UXg+#1LV;YFxDopKHnHZZ#oH(L<
zq?r+U6o1_&rA2*ivAAk}ihVR_r0(X#lPZh?o5NB5DP`V_ikB1nRxWnaxBa`6wVE${
zB>!X7#(l56d-_f@`QLinc<3Kvl}sOCF=Yth!oEkWI`iF5j&JAYmB>rv2G=f(6!sE#
zOo}$~5TbQl(|S1XIOX1|)W2v;jz@A011X|suJmdC<hMaiu`!f-r+_Ilr(BFCc$<zB
z?{Udq%gq#)f#S}|{c}GN^Pvs`RYOb_Wm<|9^k2_@wymlEV6Px)TlYg6YsWvk_`T|<
zfBiR*tkn8>5_7qik&MUZGT&gQ1#;(pBZ=uIbaQRP*WfCzz*R|kq`}@JBTP<eEb<l_
zo}BbpWpD^tJN5^4N(ORY-Bm}!6(0E>mf>=aCrchZtzSpjHUBM4+2;y(6~8s5$ENzB
zV|mN(!@zaFQFH%siVGZH;`u#AS@VMphtj$TG4*5}(K$Q_(P}58|Ix7pSe~U5+Fk5K
zUu<@AZNL|@(f{9rxD<ts70{J*+-^J^FxGWstr!9B!Ed!9pZ0p%y?_Kg3kTjO-*TQh
zPa8btcDxm5VK9-}ebsVSy+a#nyFWI!3Jgy8!f^y$!gngSJFt!K)x@^u2CYoth+{yl
zA-H)!ooT=r`8zZpZ}>Dqdj`M(<jCE<EBo}LSj70;zshWwZ7(`UudnabrLdb#txje}
z#p}h%(4b1jKR4F!<gGUa?o^OqH|5uE77MHdy?<wUmz)<}DQjwGwy{3@+(TY{&)8?i
zHu^q2o3Q?+r!{RA+#8!82CqZn*OYjBvE$4CK0dBih4Ae|>t|$WNKsMVv&tKBE)Pk)
z<6)O5yvg(!Jmt4NtYHrUoIX&4xv9lXCQgo}h;3|V(i1=#FCYZ@7e64gNpN?EB0%8X
z8F_1w55+H=y`o-LSnx6MuCq4!jR3o~q^_v~kCFdqQ?rEKDU5$NIR4NRBkZow|Bn9`
z2D|8{h@hb0h)uc88u8QH_g+fdEH5c&$G`4q$TMIq#zg4l855+w1$nwVk`Z>?VrLBU
zHaQp<gc<|qew|5NG3lde3DRS14DA=(iuQdb5=1IZyA9cv{FyJ*p2=+=udnMU`0|WT
z_kFBBum0~-KbEX3$V~Zs`BJxA%<Fu`9f>vKdiIPMx-A)Q2*!U092u^GZ6+je`afER
zpc4e_${<`dt~U3h;krZt=+sN7p<QY3?1@snjoD{7qy8k6s=4#Eyu{al5Nu+l9T)n=
z1RZK-+e&&T=ab<x{QvHT_v&$L<IqiogoG3vJzBw#E>~;|l-G2G5onB)eB_=rS4T-g
zTR>hc8RSR(u~dcz_!DMkmU&cPW8Y;Lyvr%j7HYY`a~CsJApPGQ;x3fE)5-+w`E;v^
z;vY;*b{iT1>|005=X&>cXI3x=1%+`4DX<wm34e-2IFIq0x*1_(=WqY|?^yOC@qNW?
z=ZBE<D8{~e0X_S0Geo(X#`fsZjLRoVO)o8Zg@h1-+C*>yd?f4k@>I4_MEi$w=MKEU
zbk7|O4hN3exp{-gxsN&D#2t~|_W@H{Y`X_6eBCWZXcg|DdlvdPp%U;q)-nI`sk%>(
zMH^!xC+CUiPk@02{@KHOu4#z4Yp7kByIYgGjTdcDutP)i7U*!FM2x5BHwE4apFG!i
zY0QmQovb7`b;hIenLQyri0Oaze96g-iMVNi?Lo2Tdtji}S~md4Y+RJ|P=|1Hy6j}S
z(3O1uL59#jF{6j089yR#1TQ~fA$^Uf3lt@D`vnZnU|i39#O&Z8&iF!X@U6uE)@5a`
z-W5}Ah?f~;07qxc0JZC8p$6Zl4gZ*M606J?UodWOR_A|{djn5buujcW*RWUDDclCx
z;7mxd{7+<d%DvM%JTWTOy>tt0?EI99rAY2#hyTxy_Y%}#BL2)->i43Y{d?ez7dnUo
zmf~yO3a8U+YizFTXRoJz8`R302`FO{&yguVe|||mOBvc*FLk55NjNfq`9Hh9(cm>i
zu3nekHr+vFR~RX^l=->#A5n&yU|Ilpo^D5e-iyw-VcowI?T$*54pk&KH<+QKwV9J?
z-_(B}+IpqFK(Ra^N78wPPvO54X(hBpYo;qRT_N`n-~ZNo8nS%*4$9%-UpLJ+?yNiR
z;w$r1kZ_3J%cxtSCl(2P^B9T}Z^0tN+{!A!dUX&mmYa9($e4^>oFDM%0Q8ulkl(#Z
zuY9I-^`1eu^G~#6K28Azy0t{+%`J@PtOZ?hnd_&tSc8vdr;vsEkB`J>XJ#kNu3<}x
z8u)}cP9o^lotFRvO^R}z?5)vDbHc2=#+Fc15Mag@?W($Vmv!-`gs`dppN(LyNlVwP
zx!%6j9%)S>eC?@Saixdq>>|@l1KC&SMVx~Qhi{|Nr@m7zc<!yAd}#TmLQe!Zop=2j
z#m0@GX?PeJ88JWFo_}oqyOBGZM6$Ad3XFQR;KB{fIO&bKkN~%bGGn@eDI}P`bR7!`
z!=(%EngrQwyhaW?B)fddpQ5E=!f8L2xF*xz3H>+m#&xaQBPTQ`uA;fB2I(!mN6|~(
zt099O-(OeRf5*+Ho2@@~78&*Ex-~_Bchw9HAtY(;tt)jJj*d0D%<~v$&Ufgskm273
zEAu*dx7h(o%$v3=Xq_Fux=m+poxb=a|D$;dzuUb=^`K{G0_3+e2+SR;W(KI;?UySl
zmU}Kb1zRVaP!;Ibqa?HcUaKM?{E3Eug1Id;tygR{lA!Rg_?~wfzY)N`T8p5)k^$n;
zOZTf3wzSVLKg-{#!;N@X95Uk^n!}j?lOvm7?U>ITQ$zC@e40jI;R(F+$#f^voN)XK
zkXO#Edo)KiBe9EE<`c0*$=fV3%*A}{-UJ%P#JrM!Ppgb}EvBt$rGU>|)@U0=#ZPua
zqr1yw_=KBkkxb6wx99#kmw1Lx*77S3w)uXX4Dg<gS{cP;c;rOD6%c^u_v2;OD76t+
zAb-Wkr2|wFM1WGJQ7fwazuL9qq#*Wuby|JCZl{e-c)kAHt#Bb5i^oNZN@J&gBp6Kd
zeoykbB<c6^jO6o=kR(t^e{w@eo{wP@y5K)|ib(OBssC_$`Y6@wc2Mel27WfLuXzU(
ziS*;l#{QEOvG_*V#sU5l^JmYl<Grw$9NH5OOX?V*jZP0Ewl>Z(@G*2Oe-on|_daYj
zE1N=op~Z^q(9l!LFIvV%Bnkb28@^n18L_Qls%y@_rXknR(9rDK!?gJfk%rS{EN`Qf
zHo;C#-4e$#xwF?awDUjM&h|g?+OOWMWYK?)A-7@=fL?}_rQ^;y(f6j1__v`<iJRNd
zS5He3oiBb9oQdu1;ieqLfQi=GVzeh4K%Qycv*z4y*w0#~b;xf``peynP*y;q@J3<j
zW5IEX5`nYmis$3&6e)v<yz<KuY&0}inucYol~Lg?QGbTgwHHU8{d7A-ky(gBENi$y
z-wI!${n~x;5IZ>lq4N|ov2byDR>@_dFX`KW!2k-PXx!6XRTxc^rB%q3OH4{y0i|?e
zPK$&v7XAC_peuE8n!m{r-jM{RWfD+mmX(nBR%W#vcQAoCiUV;wjN3jBe3T#}%4wbY
zFYr1~KqS)dPm%QY_LhQwd%VAvK8ajZ)}JVZkyk*o;Q*NB_UCDf0wj<k9PpU<>HCjt
z#=m6dhCxZY19WZicn@R3*h@*=%l^kDFwKI>FSNoeqa+tVvPU$cj*7d=4LR8TSot=b
z%_HRPBcoq4t0&^u;1`Uf>xYq@ZELMF-b7Mi86BxI^q+2b<k+QfpS^CoHMLm!IYO+#
zm@lIw(ldx4`o0?9m;0b|SJ(L#UxugOatEdt4UMKgwo=K;tV`!rziIX-vMkymB3_*@
zztmU-+p0^%MUn6{8<k_kp`_IxDwg`NF~c}5lfW7#6)v5Gj$@#4k_~TzOTl~kb+v0S
zpN)@C&EDQV49sRxa_No6Hsvb*f#1K!GN_d4!CCB43>7IcK;8SGaGvzxGA0cpWA?E4
zVhD?EZAHFLNAKoz9rmKXa`cQ^g<Y;Ha%SywmpKXP*A(%uo9<rfiD!jk7IW5QHBt?b
zWCDyTH3NfKAyh6Z)*!=Rv^WJ|Z1-p}dKG}7A-K}+wljC#;flU#$2|qLLH=oNzUn-T
zCor4jSyx<dKB5rm?)_|8c;@$Rgvru-Y<*Q%vQKn0?cf1O)TdvWkD{cc7e$;AHgOW2
zOo`*qac8#LOsUV4AFrI(;2ZQ*zxbYXiW)`NR*i9QwMW8|k(lLs6PeGOR<)Rp?)cjB
z@@zLhK#drH%kPY8!h?g8)Lf2U-Me?s`}1e&ox=x$U%R1+pT%U1V<1g-^=reAg9*oM
zoAt3&P<$`|MMHz0IFAT1BpE_k=4w{C*hbZfN$JKAEaqfSat9$;fU-WstbyU->EL>>
zx?a992!`|NsK1(?o=zm+0NA}3^rEy16<O`KwQz}9gy6GVySj#8=vhG0?LHX>=mlj!
z&3U<3lu;H2Qj0A-C@Lxo3OvgI&JRJYh(!x&{>H4>Lj1Yu^2Fw*8I+WTEza;`)ICgu
zGl`AZl&0r_^eG;4xrRUDdUEjFu=GTVj-a~rZ*;s;LuRSMF?S5UGi@in7->}tNh8}C
zG;FHaTd`20z;!y;38mL|<Qy$kcVFP%@yL^AeiV=(S1Xl%BQ@+*tF@4MX>XYfneRA~
zu8;RtCjN_(Ub7<}Rf*uchB158u=AHTkd4p^r;%HKSDHP<l$SRvq<TcT9zxHqBv*|l
z9w<-4<FNa!y1E*~GP0<plc=bc+KUXJO)uyn541=x8huubXk#dUBc1eydYk$LwVgpG
zA@3a>Xdwmt{CRFaISSSZneUKX0eYoiq_0%Tv{DuzbDGm)_@V1%s|;tb=#F8W3c(#u
z#d&Wak(id1wa~0#1B7P=U@86_r6c$f#%h=Zu^FFpBe#j1)<Q*3FXaj@al)j_$=gcD
zgF&Ef#DRlb$lkgv7Wqc)DD04*ZHfA&X2;%><U9^`Wuyq&mFebU!@+n8!!1n9xy>Le
zU1$BqSH0=<iH;;9FII)>h=mwgY*jAI9#Hqo8&<1iWN$bW4+m8}Ix^g*2;1*|DPQ7f
zkGq^M){AMJc=5&GB!hNbQqAGM;&D;_t`v5vhU@jyO6dU4ilmP`O*AX_P=kvWC~Y8B
z@h~X;U>&JG@*L#WY~N`Z8j^?8G3O!vX!QPC@nrDBaafA;#0cCK8j2)T3YrH@?Vuv7
z`>`BDu|UsUa6(=VL<fy>hwxzO`4<3WEag|I{=smk98?>RznTq^L5(Cx=2X+<4mqBk
zC(d0dyQ5vfjz>qJG>FYOY0-v?7p}<jb`o}#Q5QAJf$acZqg5=s%N-3AMtk5#O8?e?
zk`=c%wy=D!ZZ=@vild9OYfx=a?>wCO5g)~Jbw?s)yox13)5BDyk1V;7{l#TK0o*&=
z^b3simptdfLcQo_NzK1Q=hXI``X+usA#21B{#`oVXpzdXSA70M38fl_{L0aAN9_|h
z`-2UCL-L64cm~v)N5Ubh4*H)LRm~gl=fkA^R{%8ny$2bae?mq^hLj&Z2??!m+4Lup
zR^=whjd#JnJq3y^Sl6!y4;Tau7|<(ISHhxW@>5WN^NEbxF6Z>@?2#~_Vp0YMX@+s{
zgMG81(>Mq5eEss}N02nxShheTs#$LcFFa27kn+mjl`~8=S6>#(Sf;G(5lNfPjzmoQ
z%Qk`|_BOwMl<By;<3U^EoE6$nn`gCicqG)t5>)n@>d4H^#d&$OHxd)|@*Iq5LpEz&
z{toX$lxBx_EQ8FqV@^*m<KK}wuF8q(g<v_5ib%gkF3yT)W-k<;a~uC+5$^31=TlkF
z&6G3D9Rluz@G+$sy;w)syxLu3Ho|JV^pb)4MX7H3sa@QDk=eu$RjF_`<6Xcqz+Cce
zYkNE0SaYWOPH`Xm2WDbF%!=pFpC1Fi;gN7lON#-J6RM!Amf5CC2}<9<W3U$#MLt30
zrK&wV@6r?A<hoSwX0<rew7Q)tc2^Bkz&TM@1*ueK((pA>cw{<QqV=<@wXLXB?m|^+
z8!zj9=tG_X8LVQx)iKs*RrBg2oqyh1=0}mrv53E~(Q!kj`HT&NWEH*W+WX@mT34%(
z7Oma@ab9X+)VckzW!Z`#3t6J?Y8~inVi0?t6Yek7zA&2{n1s7}B$27eDy*HZ3yMg@
zTs9dnd3$wud{n%=st^M$I7ls?Me6>>(~*It{YW0q$88WBq}qttjE3)CGKHG(ZqI89
z{PPnS%K?z8p^3YL7Bn_JT<JY&FJnOdaF1@@`xjz(lMeciWFS*p`$4Gm`=0#>oAIee
zq9N<L@m?Fdc#J|>X4FId@dI5odYnJgiTtu!$(-sIA{*cKl=3yEJ>XDUtfmw$OW05s
z50cZtpno&zuu3o73p^m}(?fF*NEHQM+rgm|RGR}HX1>dftJegc(#Dp3TRfbU{t3}8
z$4F}bBAsf*JO5qCOO}sUv*jV^%&o2}5!n%DsI%o^U9)?n57!Y;8vF*lHQ!r)cbT-j
z_SeVLAwQ^&Yekc2Sv$6LUlq6@&XUcWAE^FhSm6=wU1@)zHX0NiKxvH|^3$n}PdwVd
zf4QH#)AHaH1NAO4+-hHB-8WoB5V7PJ2y&4U;hTqyop*oYd{#DZJCM;QRGD2;iJnFW
z{JNE;xGxO7AZAktEGg%$215S?RP9+{VYM&W@8L33mn8%Rdw6u=wUf2mf@Pb>w5<W<
z7WD^y4W{vARi(`0oiWNU4N||?HY#>}b-j1uu(0hOr9OARa(s-eW(PrtJv%utglm{$
zp4MlZ-@~y$umTlN(YIkNsm-_0WGu`_@yZ#j%<}EeQg>#HD4v~guZ>4gj@^HSLG$3A
zY$#>lI0@viXlPSU<tMqNtV!60a*eSIbQ%M+6$1mDv;u;~Z0Myb7ISU2avh3&^qz-*
zf~a5!Fuy-?evO?c6KrURS@gacv4uMvtY<=D0M=&fv>6o}n;cO6GQQcjJ>ce6W~@HV
zdr1BGu^?nqG-C-JCdI3e<~o2zX{X+FHxQeIvzn|x9TJwOe*$2}E5K5u&Wnw_S(^Jb
za~JYkVpf9$K_B8szhyKcA;CDq2ZFK)ZPkclDn@7`T0YZ^^p|ML>HfWe@rc*{2JaFt
zD~N~-$x7_)RHxIR-Y$lU?}DG_WP+5(`k>+0vkM;W<VcBPuE#CJH{`a3#gebrcMfh-
z*9C@z$KI=p>UnhbDWW<~)NF4+Ueuo#>hQw7D9(nft1DLc2J|z_K^3_W0au7uuU|)-
zD}kpG3y3WM7vwoFFMgR^flQR+Iw#oJ3BjujcE`T>!lMHV(IdohcPSQ9R6|gRg2K4;
zcTGyMh?JWT1ZCb;F1o+2rU6dI0Eim0>UNUS!^-x=qu8I@L`#eL`FLZ!h|5gGf<t?o
zx~l8NH23v`B;@<Ukw{%%)P)Ew<<82h!fx6N)7Z3ujBIBFNe|VGE8Geh3M4Mx`8vC8
z(QB<fPItyu6-{tvnzC)dY&GlLwd(7swj(vK3nxGCg9@j^X-`_KDv%6>TgF?UZOOxH
zLEbe0e6-5*<5g6&froyu9R0SlqNS<nH3&T)rw(SRq&u#cz6RP?$cT{~OrbRtNep2L
z0w+Umdh2n3Fc1rqA=6BOTShZ{8|>{q$S$F8`Z3Spl)GvLd(`xI{0iXMbpw|w4(v7v
zN;4saK8CyzE9AxHnq@b`kIVJG52V~W2F_zDdUq0)J>6D6g&-V$52^1JpIY7#GF!{J
zE44$1U|c~BhTW@x;7HbU2uIszl!bG}rgu*2cD&D~?G?}k@NXz89dRzVigDdb?khx3
zF{xW1W^<@`@1V9i-`d3jawAd{<v`)QWr|9&QX4mScKSevHw#LM70t#N;Tfz(!#Vwy
z(S{Rk2;^v;w+I{~*Us;hz(z-9u9oDK!c|pO4ltIYIIW^-byQ=jwOqh71D&PRVKaFP
z*7sG1Ghg=LN0xEnyV9hXlnZ8IuYNJjlrTu{%pR?h+}gRLT+Q22X@ASOciPBfAU&zZ
zOZ}h`9ra=|UYpH|4!iT&yXOTRWyo|CkX$n=-}rJTMh++bF5vKZH7L_{iG6izsJhRS
zxk#?ete9z}(y^F;Uh!f2fMSC4Wg5(4xlJ8iUBk_(T91qCI5_<e{J7&#X+1=XFb?y$
z-aeJ^8{<wp9nb8yLA&?bOSIGYR~NpDY|DiDE^aIZzA^$|JS@W&4PqvcI5p4rrY8M`
z9vRQi^D;%vJqqU68n`A|#dloD9xAe+o|r#zFh`hV{OOD}>Wec{Oj|dxt&2}WXUBdd
zxb?}cTenz1^9GlM?G1bdW@l%EMnD0OX2)$tjr|iKJ#8^!b1ap*)0Bx{0VFm_9i60y
zL%xo%{rbWY)2dX|(gR{~9|$ZCWGFBT`*-Nj_Vrqz5q>ILi-VSh0;?zGPZH?!arJw{
z?w|&fAjFz(|3?0RyoyPsOslPkT3|k(ok$eR*^f88j~m8%ik&o3&y7~`Q;UiE300t3
zG0}C;7uglt-f{8q+PUwvm_qN|xDi8}T_J@)Vo<h{mc9#QLWlA<fi;<~C}R+h5V*2i
zdU!;R_%S_vIQLbRh*`JOO7|I>?Dv+0wy3gNYx|i?`f73`o#F+;rX|P6yN<sdLNgy-
z#H?>2@-+X_HRC@FVSUqSE0Vn)%(S**CCw;nDphW@cp3Fv62fpL69o!ye>3v^0qUJW
zqc{a_2E>r3xm=vtc0_W843r1`P`b1M+%haiTe#K^Yl6Zw7Gt(b`C#NSnv5iFIg51k
z>+ApvvfjSY#E!(@0gnHQ9(T!6Q;U;*vXU=fW#C<%zNGP-c>DY^)6tIv+XG>~=x-u@
zuX%XX99%pLsNWVU4h|tTx~4QL&ZwtCBYJtVzqU8+Nr75Nd82D0Z{GBS93d-QZbS7X
z-=&t~Z;}J1v1y^MuAbX;cyt5@2<uxXrwXpVODR7cc$|FF*Pa_h?atXOyPx%b9-F}v
z>j+39^%pexK4USky)9O@5X2=qaGSkMmju^gmV9Wn_ak!Fn?ITudA9+xvh@1$klpf!
zw^BxF1}vHJO1$LMqOQ@7wgi73uI9y<!Lu2w^8l|1`WE)^4hS;qZ!luKnU8*0&3`FQ
zmHToWcCmmne{LagJpZwuMaY1Z%Hu?2*=AXysDS5%#WU@q<FuhA!-3V<bd_`q(bTgk
z6D(5+<OLt^QK;15Bav09L`4^=)f?15?~jmXnb}K+(ar_+FN6`P?6gL1V_#26@%-~3
zXff?eG`KNR%TA)Eg`~p@$lLQ5N{B*yAP;3CPpB^GZ5bY!%qv!Mewf)Z@5Y-gI{I=%
zrG0)Jc)zb!OT*2VUpWkRe<g>bgnrs@#ZQ7*Q@D7nSnDaTpZxRZ$4f|FwYwzq?8D94
zm5Vvtm5cftk=$ki)UU;8XsK1sHJ)Ro-%cg&SP9l|FcEGiP-_?)5r1mCFYVyuJ!3;A
zsi>SscP{5Kl$lg!95=VByH&&U=g>h*AtJ}Nqj3D4`%|_2A{Z*qX|oTPAsmu!$q1u;
z?bMPgY%G+`G+)Dtrb*B#IlYe_6dl4C(j;~1KG!40#6LL`Jf4h!N#@jBA)!qfvcF@2
z=E`J2JB5plz4O06aecUa{G~$Wk~(RD9$JXG!X@fpg-iF4?LC`IXlBpNp9O^m#(w>H
zhm%YUW9X{1M4?2e)0>H`KM5up(Tz)tXc3s#9DViI`Wt>7Eea-*i9eeynj(%h7%X-#
z8*@lqDe&2r^7!-NHC`}MG3bD<>^2$p6=)1%!4Dk@lDs7KmTzeOWjNFI1pTvJZACuk
z&|io$S$ulqM3U>K`UXDkpKr%;cO^ul5nAa9I5s)_r(xq64aSKG`#{i@{%`l`FP=RS
zSG1Bk-cKny&l7B7BkRtL_LbRNdNx5#`hRG9>!>K(_H7hfKtM%6TBHO71f)wr>5>{6
zk?!ue0R<kUq@_hVX6SAy$)Sd3Xpru%@51N(z2E-!+H3E>wo6^JX3flf#TmzOoLBt)
z_S3rw`8VF&`7Mha9&(>{_NyTm$w2Z;RSYqcY)UD9e_vf+zW-i4$NftL9{ryLcL=Ry
z^lAKr^*>+60{v9>PDweF3Ot)jXPF(SR8>@C5PJ2<Bu2-A>Hv-?o$rW07|n_pRNra+
zcLJNA&)mVl<0q?uO?YWCsZ>ZIxuQK}Gla)zR)l5K@7~}mGz)NI9#__L{P*+o)YqRs
z10)@R?K})QAE23;0Bfzk8<8HJa=F>m`L`QyR74xGF({%^BC`a$$}2_2FHv2>3KbFz
z13#OK+@qtTHmh#lxc=muXp!p=KB>6)wrjtuMVhBR<r2s#S^YvwE<j0~Ykg1c-x%`4
zjG^5)fJ%}pK~`N|9nF$92@boj)rLeeW~0pUncVBO<?fMcX#Xa|6`x)7aK9wJVu45L
z(`ys09@pq9SPDxjuI|v(8`m|?I(S~VuCrR)LEh>(AZgm0{iV8XPsPeAPZyoqz?Yjm
z$n|#7@kYc#A%E2JLX|>v(%Qw<O2HlQT&J44pVEwF=jR^0W3|ND1NU9J7$6`X^78Wj
zZO`7x;`}rjZVg#z=bq?O^kU(}qQY1TUe#EZ4pBaZ%RFRu$Hq#PYq~_IML+DOT!84w
z;&C%xE`_^SeTDQJ+FwXao}}8?%7=;9PrQ-aV-(q;?0Wgoq+$Zo#fLG}yCU)Cty=`d
z#48)@oy^Y|`X*?VNgz4i=%i?{N1IWXWBM@>yJoWNi7SG8Lp3j@+sHny{hw^jX2;p|
zd+2rAF|76n_h;M>J-T{&qG><@#<l9POXQRls|U+>Pfx=#wK}awwG7kAteObf%NsZH
zXX+Xamko#zbBo5Wi%Uut%!_`{neR9;WWQN<yGn_UR;ldd=7D%B6)$jCmY3-%Dft|3
zst`}lV$8fnB@8)fkH)Z6or&*S5+m5s%xicb{<VwjT;N-cpGC~Yvz^xb6}1qhOljx0
z$dS=bMlVX`(#^`UxBgGV7~b)xKw#^VGh98>C3gzpOs`Q6JaztSI^Lb&J`Exaf%Tu7
zZcMd0oTjp0_xA#e@d_8x_l<wv{PlTgqy9!R>%R|HCca&ZF{5XcSAd{|Qarka23l?n
zoPQZesb!ZWX@IfEz5dhVzCiR?;s0vq-dKHw(j3}nr&9>Q;c4_5mL<L=ZOXTJdRe|_
zAyc%Ae6OL8-63Y^f@as;f3wNc=z<Dt4Da0B=5L!=2;$7KG+~v)+S{gjKZkg`4|H*a
zN8Zv~4z#^Cwfq{j_KmhN_pS9*-Y`sSVQTK*K2bf~zWL_P*Ud4pu=A66Rd5N(uvd**
zIf`LmHR^f$$*JbhN~ko-y@eAim%{B58~Qhb`o9BCMDX(2ONm%P$Iqvs6zR9}oUapt
zRG579;H-i~j}53NypgRg19u*@PGtP+%Acub?s)IDfWn3Yk_$YtZ6f3EBzZIX$`1d1
zg@Mud15DVzum7K3LjRwi`2Y6<(Em%@{%=qAsME%%o%i!}tku2_8cJj8W5uT@9U~Tp
zPQ2UC|2ob|1dXt;aGa}<R-c`{J#AGocBReO#^JAJ--1?{Y7nZ6pebQ@!)eUUp5Wor
zXOS{!7Tl~(_o@zC>sT;EAMfYK-cY|02lA5I>%Vykyt!@7x}G<`@mMM<DUv#?nMxAt
z!GnD}&h7I2bqn>Egz6ry?%Elg*GWu)z2ceX4epy#l9K(Pj#35=QU)Yn8{kN=X-WyK
z<aiD65>UH4CuskjJ`C@6<v4F{W@cIEDMwiN8n|b;6ztey)~Zwh3MC52;IXONHwYlG
zPB;}Dy~)VOrvVi9-JPBA9-~39+?CVo+y$;vr*T~XvTC?+vX1@B^+BVp^4JAkL`6jQ
z!Iji`VCzw?Dm=d1s6>V?5>)Sp!0rXXtMk2xbfDjNSyi}KY5Z#+7&f}mm@dc+AP+{+
zWDVx*?>n%<Zeaiux&E_R^YNxmK=TKeF(eWh6#eg;7~aOpaTK-7#ro~Il8WH?34(_Y
zZHDzNxf0mf*}vBH043}4YQJyPB+<^T=8IJG?D4XB=e(Vsee&U=^Vh9NZ0*@KLF8xb
z@fBUqtjOhNKg5h<)2vPBvadRwkNqr&<1}bUTG<sutSy0!o_B^_*m4CpZ%__zZ`Xo|
z4tQ)V4bX*F6Wf7ksVFb7)1n^M(cP`As2Cx>x_7rB-@-apVP#>=qO-)c21}W+d{WCg
z)?4a%!iaR@CnOT)1VVS$8`^IPkzk4y0=MmNMnuKNW`XjOb*v9ol&1gD6)nX4bT4Ld
zQtR;f&D=`HhQr2Ln?DC5x-dJtVnp4cMw!Lfzy$4o%-CC8d4-@>E=%c}VOv<_Ur*h&
z9;LkO=jWFLv@KdR1rTUU#$E=s&gCc~`6-29E_VZ(sUpL|fHk&Y0PBR0*DFa^So9aQ
zL`rq_s`fzJHt$Dr;66l-s*!wUb=se*coTbn3E>dzct{{ShkXx7pVSx^%tb^l+VCjz
zvx{tyau0h4@i9E-8ExkHMINezyhDj!@PMfLSD0G7i~uEzjhv)hz~|{So3^b70+fyJ
zf!NJe=&K3fC)30r>14QBZUY+&oYwPZgm>Pp1xlC_r~x5jSalI#_3tV;1p4Ixwg%Wb
zwGUoK@mR^o1mC=Q^9dp0Qbz=msivW!LOT0(q8%eYlO|!%`+B32imseeI}5Iht!3SV
z;ZkBL30daD@ASe+KM~BDf6Cl96{kbVd+q%;OBTnXQkGAT%TS-f$3&3<X@T?yt3%M%
zerxSXXBX>b+6N;_%Bjh$iZ=8mBFFfDDwvL7W%_O|&5?2fF4t0R;eQ;yK>h!4_{HTd
zdi)Q)cbNI<vkJ2^gd1SG+r`Bti)(94LDtcp?l_W5_O(~Y&Q3*D)fgIvSFZ{|UIdpF
zF{wZ^4;uG!7vz8V+BkPa(5?<-1c8c8a}e>2zqWX&+3!(U>HPMiIGTeHvmZ=pq&cSV
zDn;OgNxwUXPb*z6D$NU2w?=ciQoQ7j9@J`4hIpuX-PKvy#4L#|=o7SV$m&ic^HYOY
zr#hCNig7A@?t2(VL*Dg!C_+?xe{a+C)M*y_)T33cg)bh`CwPnX$Z6%T=QU#FYBjpM
zRZf6CVS-1K_7s|a_q6^Q-Wtp6N^w(Vh>KgrSJ9n6J5Ze<L475277faDt8(V<87<P3
zXV%I@UVSg9YY$r+8PSPpIQXTt!)4m56%v1ZMI%XF<1qByitX!z7uB81hg{6j!_ai*
z(Pbz~GjDj+$62^9yye}+(V-^Nx#Qo4e9RpxRt3?*8SN?#c0qms%M$i~VY5EmylKS~
zp7^tgwf<*Rw@k?Tz=wuO2hXBes~eP&lH3($h-l}j{6o$>^Oa0yw?Ci5u!MiPP{mbn
zsix6;z7#G|AH1uZQE_&_$OTO@KQt>_ZPkf2*_8eK<Wt#dMnR_8_*}+v(Wk#n`1nqR
zZie~3GuV2P3$|RCg2K?38t;VUW0&Gk8oJdb)t&>@1|9a;;xNa$Zyzss_BBwrxC0~D
zgsT;t=1i3XowQ8cu%m{vdNR4JOXv5(IA(;}tM+GNoPDt)xKWbOKr^LRdiwtUSEeU_
zpZ8QJ+O!`WqW^(edsSnj9PE9>A?GV?(H90cnsA!*&3HuLeL4_xac_U>XKOCu!ylNp
z&SG4d;;NgP+^&kH9lOFBR;w6vS3AG65Ow#$ug*@0$ob>a*Wx+*EiE--PR{dwlA5ZW
zah-Y$s@4g5W<qUK&hmGZ|0;nPxxS1gzDEwgB>|dkxngMv3%X@^(328GBN^tnk=|!_
z*hfAjps``DI<j<euhewfOFjmc<H>m?FF63Y>u7DVC9n0X&`D0eRHM09p#HZ%KW?iS
z-LVG2`&yZ_j=iBw+;CYZ7V$d${h(c#UIg^I^1YJfSNWmI23Ve`13sP#%#}PiY-IS>
zJ=9Bftqn6})+3$+QD0P<mF0!!OF7b`O{J{N4C{Z3{TM7{Dt32RF^+f=<u>@-l*4aa
znmwji8df73fip{vdoFyZmjYWADR~(szvd`SY!qQp)M#h5{mXfexo_%VK<{a?j9*M!
z!8U%psc6r2Gz+d=FsNu}_r>pEh*(H%XOW0-5Mb!oDEix(Hz)w5e*E)B7%&DSo%#uV
z!|`1j$+wqYR$+#>eJ<~WX(xmCmd0o#n{M2&3i}s+<<f!YyuW>{PVk5<f1Wt5v)7q*
zVJbF)PBK>$4#i)$a5bqq<xEj=iozV4)hAjxRQq@6%mDjR$k*8z?=-3~??|P#3fXI~
zzZH>=-k5Ovk=t1S#_P_#WAm|DeJl7}8GllI#nrcNA)(db>gan+|8`OZ^9HN;C!AHU
zK$?}SRx6gHr?$j#r?w+HC+aUrY(5x6poWqIURSfziSeGCoSN+&+J$kLw_7E3_o@~+
zS%NmNH(~uvKyOq48T+BeeY67i=QRIw<azJ4^;MHy$j<pdLsU{qK>(r-oJYv@pyr`~
z<DP=#z#0>biH)hIdpT>&!p(M!QF4HzV)9S__<u(7?~h$Vs(I+jD;>#!DdDTsDM<!H
zzALCGVZSG4Bli(urm<C3&U2~!Qa;X*VeQx>n22h*)#VOsabZH*(<-?+)qS~l@G?qS
z8^tLW>$s`x6a6ptNdqiY;b#KNM#hKF1xj{u2$m#1(I8w3p=I6^+yEV|(^v}4!gqwa
zB}Vnoc4BUTUvhvMQ!HkfKss)}Z0X!@Htox)m*lcYo@L%Z^|5NFHu?A(<A(TQcp7W+
zDbh9jKb}eqhPK3Q&FAd(T(x;rR8ksK)<w3a*SSg;9;Hfsvu&F#16X8J?HIr(8<)!s
zgR4lq@iwYEu5~N5+{oyxWBNU<+W^%bMpTzuq;j^)XK!Tbt~E%*%wURk;rd5e>8l6P
z=0m501*OL`kIGD+WK6f!<^KNmkX0G??+^yJj7;3Y$e#4*G4gyhclBhjxT;O&QcD;d
znn&rpq4GUNCFIwyCv(OsBC0WmbXkP{^>yQAsPqriL%DjlSYcb*Wz^40q3sP@x7nbA
zgMNA8wGEZE{li?xTvD?9)6^z*NxqoyoMzLN2%3_xFn?Xo47A6Lj^FLuE)_0+8}RWy
zU`<ksp_do0E5)5X*{prMG)`h;EYUwhb};=aVAweQD%$!Enho{U`S#*q{UWQXrt7gL
z+|B&hlf?F-EdG3L3yR+xnI3nW`4MV(K3JUUTIW{16oIJT%Jnh(&`?j5_pb|y0lnbI
zkni~gCm8AL(Kc}AiTW0wTiX2+_4Pi?<m0o{P4#!E{2ZA7RL1qc$yxz&`RgS%%1OT^
z?i{PvX;b`d@~aoX7Z*Pr?tIwz=RPX*pKZvA!GGSkDEk9k1I_YRCdE*`0lH;F!ILBh
zpDuw3Uma8oZ=&Jzk{v+&%f%<o3GJ74=%Lkiw9Sjz)gy3tWU8)y`teXczRaR3G+7`g
zKT+!>z6T2jOIc=biG=r<?_i@3Jz=OS6_{gsy10eXcv$oH-u!%YL~`oyzK0Mfc1Yi<
zb{LM<u6Y4K)r8?+6P|o?P1!73V=Ud=cAo&ymiF5+;c#Pw(tKp4$CP7;syeW3ieCnN
z@lYf>I*b{ty{bP-t=FzpJ5(Y)CivacrDP-*x^!M9Wq+xWahKEVPYJzPo^Cs?PHwrK
zWbf}j(rMWdap8-6XtK8Uc?IiqvC8y$mBQQ0OSg6ZwS(gj?!qF<>k)hU<XSX(2vRVt
z;X;SilT#&Ps8}Q2%6w47GmRu$&*UGOh)^Q~vz1ipQ>zI%%h4KnnBbAG7+#f*^&}Jq
zY;y~JgO$(axeien8N6i(l-j-9K2)a~Dzy&fz{lT0KAqiNz=vO76kU1OqiaP2@E^b4
zj$1xRA-%iSb8qv&?f_k8yCsA}arPCrmHzki2&{R$Vwh7lA=~jl!td;ifq`FK^Rb*+
z@1|#bk!P$<(rKO1OLYXL&S^pXNWt>P?%&vD_?H4N^h;8j4<uG$l^(B7UB<5VI>8}>
zOBl9){9B(@<vf3}R)>LZ@)}M%^z%C5GjU)+1Kdaf*SA($&H@aQmYY|?i=!516vV0W
z1=TVj#dx-{U1yw!BtOwOYo;<42sd+-MqQa193982*+EQvx={2%Cl_7w$$9I;G=1_1
zFeM0U3b-@t8oc_-tm)y(RsodxXtuiEwN&SiYmr#5X8Oq)UQT;jnEBY*JgWGuC#wqH
z^(?%=qqHW%z)<PWYPdv5+ArmT`z{Elz0&;4thH$-1(PHPLYNawPm;KEyd(-UnRRlx
z>r-qmu@E$5xB=CnPvK+<vVyB4r~!WKmyb6uH49NcW8IwxD?THlpfQ7i<p$(n)bu5f
zJOPYihk=2F{ZPelY2i<fboHN_wF1WLZ=@9}0BFO9vo$0K`bBa+*Co}{Eg(<u*JA{H
zfTk(%l8sYV-g7sun=Ud_8_F0T$=6x6PZ0ZJK8!Q+qHzRP;IRkCP2~CD6Rj%5>cy$j
z9vLz8;w7ZaaaR{^Bix{v!k_+Lscx(wHZg+YVw%453O0?-wOWTMInuCw;4U*m#JV{v
znT~iR$kNq0k7baI(n!^WQ4CHD8mpQlT@Mll@~>Fw?!07?VQjpH#<d(Oc{}DF@06`B
z8rJ3z0i2{j_M-kKV}2qS7IebFXBtxv=nDPy!8YsOQ#@y@g5EN7Mt*LXJ{`2V@|X#J
zi(c+e$@HZ{c8bcD_HfM*K$qZj-mts-(I|`IdnFL#y60a7Is5#tY?UaN1%{qc(rq<@
zWs$<zVMMMtz9+uUBV%`WPf$fC<FIR$v$JSyDASEod#cp7s%@Js4hy0@T9J-%E3TEY
zJlu^g<|nIe5Y<${zW-w6DUOJ!gJuF@XZhIcqb(E#;;&ToSEz6N>_jH?^YhNmwb7D%
z@sDy(v=J6Quu_oEiHp;y(~rH!lwX^9&Ltn=pj|>|sN>^S-9^eBl~SPIE=Q;R%97+#
ztwb(jt;oLp2V~SAkv=_=?^r{3J%o)XZ$aRpaCp{>;;8WKA+ZWm>8o+R3go$lM800l
z8=ZY6l{|035-|UOt`!Hf>7x1M7w`cl7j)IxD1p%O97{eLdHq8l4Xzcnw}GYzbO!2H
z7h3pmI`>|+kXgz%JH35T*`Dqw8WmRKARlOuK)?yiqrLk}Y6%(X$c6%S@avlMPXw&!
zSDOAx4*s)i8#ggr<UFLl0!Om-d_83a15v%qa#AA1d}Q~J0476De6>xu6FKkCh5ZMp
zf)Ao>%<<+D1r)mnZ0PBbb*Gxlb2i-uxK3I#W*}Cn2CCFcJA{H=ZG(iH5gJr`)YKG_
z9z~{ERMBo|V0ICByVcOBbJii(N#WbI5nJ#FL?vpCe5YzM%x%i{mA_L1vN{--?NQGC
zMkzDu`*>IbP2Dm8lPqqTvmEi<3DZ2jvzvJB8Q!_x-bAl>K=4%WQvGu=CcL^u&cEou
z7eqG5mz#Q>QT-wCUw-}KRr!@2`ljH~Ni1J@RQ}Hr3BW}i%}C${$zCEoYa>iY0?S?X
z^S0HyV^`N>{(ckr)pFusu~vA|oNEmv;QgIH;s#tP{5<9g%?FQ;$mt`o3ha53;>#^#
z=~}IZ;^$kT{W{N{RT6Yj(x%n4MvM~GNpEC_aWJHgLgndbPvBAvghonQ9YmCkZHy;G
zJeM@<|52aSnKwi-l%EPLZ9Fb29oOp1_#NZ~_9E}!B}XbZZm@p*2vV8jnz^;vL1<D$
z`gqgZRU*_I98uAQHT{92U>Rn959RCM{v%1b28hnU5Su(h=KDSN9(#6X*(tFLap)|x
z#;xhD`pebnY@c|sFH9}S+~e+-t!=I52Sq&asWQ#$ush}6is32XmfpTqk~O(^=zeO{
zl;YQ9brz3uw4EByB_~TyJog}%kQ_|8hsOjGf^K(0fi)4!z)xrNP%E=$qT^(F2x*R(
z+V@L7H!aCAUa8KGesX)S5r2FLGa}uMs%e$<y49ZzBP7KKI7;`oKU`B1>OI1h<}%X#
z9!FTVtHGB`0T+o&)x7F5E0rQVF<M*c6m)z8i)zEB;Eybp6~#^47)svE2b#$MPgWn?
z!=YXo$@v^(J(jC`L}9};b76?|l2p8sBNtR5xpt5rzS>W>JbWNVlMTdF_2Eeo5H3_e
zhYnUVrX1DN{CC;nzLJrW5)}}|TxQ(4hns_(mtX73q8RBz3rce$Lz@Fojx8pJi%Ypi
z3O_C<=08oL5Ygvf)qC7t3yNBwmx-lVPUa%L&O{TAwnlc3+PbnL$}In|&wa2QErCV6
zE;Zz+>)ZRp355l}uuGg;8Cp+r-=7<3C}pKp5fm3r?YJU_*WBA@wY31)@9xn$Lf-2{
z_L|*SxgRpoa>KM&hX$DHYI5Y2Z(W=*Eku=`swF1t7Atn)QmUlna5$K+#dH=fn%abI
zV(stk&D0gtS~cXSP<TMTs^o$Q74fjO2sg)0&R)npPoGv2))pm9!xc~JcG%aY3+084
zSPy5jkawYixEbihb*<1WfYpyU)}-Ie{mW{UzjQ`dW4wq4?o6`(OB#$E>Vka24CFu4
zH}lz_nzbcYwQ4fG>P^3)*C&)*Pfn`WdUiC*JllDfJAQY3IQkIA1W5VRDyqL4a<>}1
zMLt1Dx+#S~<>24lM<Ka!6~CvX09_>`LD#(*ezs~Vs_P|7dboZa_LpIJFSM97$v`3?
z)#Msn)vGX5C@}%66Q-^<&&F%78r7FwR+mzDITJV?C*(2~mduo6Lp40k6b_i_CZlO0
z(}UKAhif$h{s6_{1rw_-q{L+j7Vko5%$rksw{{M8`YkU`B_gi*3Gc}vd)Q7hqAtGS
z>f1}B61rfI(ULo=pDan<n@S)lz5*R5k4ZK5#3L)+n5aO(%nGqt^{MgBV9nEO`^vp}
zQ>XDG$Vr>7XWB^;a{Yyea8A89y6V#Xq{B2T{om!>@^o3ppxr&kt~~1({1wL`jaXVo
z=V}>-hB=}=B=PQp&rtUtyif_bAm!{!BrS++T~DKoDHxJxx?)7R5?G$Zz)wy6@?uyE
z$g?#kz&>pepO1O(r*)u=yIg@0gDc7f?a2abJ0CdBbB$cSgNBGXtz=i69x$So?CB)+
zt6Q&!;5W45!1(fo1U!u-Cbup3(?IqkEZpspQ+!O4%ZZzdykr?d&OJN$)myQyVEvmF
zN$yAJ>MLW3qT5Bi9v04nAz(cRi%XV<d`HckYX_~Fme;J2XsfpF6V|_$a{qc1F4bz}
zg1D^|OMcsU+PJ2Cf_+Ku`6r0)&uV4}bd4_Xm(R84Y!i+s`-+)04R?0>okw)lm%==B
zwJOSWR#CDM``ZEdU22A{PfvHv?WEiamH2V2D6})}d*~wTD~*0UzjYoRo&<L`>Soy)
zy$<!Gk|Rd2`P2Ou_DM*Gp#BV;H6~6DrIZVyg6?I~xCT`2`}ygpv3Bvn3TSJ9a}XNn
z-w!THp10vUIEdM(UY}6!?CcuJARk|e)^kqI`pOiR>A3bMfla=%-C*Ez!$r<=Y`(IS
z*QGkLqi}Jh|0L*O|L0=yaXja|pRr1X#)50^UlW+Yyj|<vBon_EM6=SG92+&tIBOvj
zp;K3SmQ%FU3Dk4<Ne(E*bJI++uLvIhB_KL)IDp?-A!+6ruongu>3_aDJm-B6Xv^-=
z0m^==CdNGYKGg85rvd$)tFC39?m_?-v9Gu#EmdZBG<|u&S>5LVQq~>okjMtqPuhUc
z>;rn;6=Cccv(|9^&a#?uE*@GSKK3mql;<vw3?4KVU`Lrl73Lqi-^X9InWO7&={Ts2
zJiR#H#m2(C-*?<Fl5@?7lN$8P_6&;Gv+74S<g&_X5{D`qhO-yxbyw#P5>%L9ziS9{
zR}q{j-`N$swtI{y0#GBr8xft5=SRxLIBM2k))Nl^k}B#ss>(_BI<f3s-ZGm`w++`<
zwVHxhB&*@FU3SR%E%V{Nn4Z^q8+Wg<zvtB?$09|Sw!GJF91Vyn{<*|Y)HZFcET=?}
zFEtQR`wzx8oZ_wsis>`5%mGcyrFi<Ii&Q+?J^4Fd*$_~!uCpG<*k+1ZPF3yfKY@s2
zvIR>q-yV%EGiqm%{W_H`Swu=(1^FK4*Pj0>ghkxEIN#qb<mYtzeSl`H8mP=Q55Bzo
z<0;5-V&8LO9-XXNEP`ZKrvicJ6(pRou+V8cPU&Q|IjC7M%`F4G%&oxWy8Cac9&O*e
zTfDdZ2yhjWKAv2yJZJflEIrTbo?j%_y}z!rr0(axyraonom~#Or^=`JA8>+hTaH&u
z!)+(a%xahlH}O@#YqL9daskV-l$IILop64lzEf#<^>&OLS9|<NayPAo8%)EKWO!sV
z{RyCbN50pqOI6=hs(NiBuBGCU`lQT^1TH3KCZ!JED)u;;-4eB-QY-SQfr_x?T33@Q
zRcV+?541KixL$<3ZW(~eeoJu8g65qaDJi5v52#F_5E-B{F^p5^mMxpMdqwq^8_#$G
zv-Tn5cU6o~W~CVgqFx(Ma+KCWd0N&|4k_77N^a3kU)KnZUbJ}G&kO?tsohEa3T8G*
z$a-3x23&>_wb1XXrB8T4qAkJ4355)I6D1!w5&IA2nEut7n3^|tofxi%(i`iS9sP`e
z9ZwSNVJ$PxU``<_XE*G_kB?vGb~Py@!`hc^VfD?kXpF4QEK)ouwIpN*Ir96~-jm>!
zHn8-o`Jx?BYVSJN#O&u0AR4;okdhm}cfk3#4~fG0Gj|yK0vDhsvbx2}_AvNx`R4%O
zZlJF4-tjrLgJhm+HR7z5WT-A7zdqwkH6MRveCpDq`YcZmyY0o(y>8d%1`!S~1(k5K
z2H`5TclO@hj<tNhw!X-;f4NRsriT<VoUj1Pwf!i&tWkwNn+<k9C~B8-6~l-qo}{Oj
zI*{+o=sVly5kl{+^2i@iN%BR+Y!<n$Dl?sOGD_Arr;cQK6_`rLm|gAgEG{m0F?6&W
zDB8U7W-*;oM53Ka+{$es4*2Ml`)mCb_8!d&odCHOCSGdM*`~K<Y9*O0Ta%Ka?nlZp
zK}%N160<YhI<5nE_++{E%9y$xPa?bI?!kddnnu~{Qpj2U;wV&8`ohzJ>cssFw@0H~
zWKVx&O~J!gk%I%Mk6NQYXWP(%s3OzSimtygz0$|I8LWXR8ryF>8a&wSri_YCVCpEt
zDslVhy}ORP5yqc-VM~9lj|1l`*J*J6?ti%H+H0%%I-3>BQ(+1tgUQuvU)C>fHP;+N
zGkTJW%b$LcQ&Lj;-M-7n%p|=wC-<T)<gy@0u4j;~+9BcBb;B#_E;B=EXb5$TInQ3O
zKU2uztpdcX)C;K-^8J<M)y16oH9iGE$*b44rpzkC^D3pT2G@Gmjg;q7R?(SgTru@M
z_7f9ZXXh#J3%W+FxG*YveG7doh_oKZ*_jpW24CAmm@ZsUGqd)+wj6vu(YsuIC~r^c
z`m(0dKKU#?)AprMI1zu@$ks>i(x0CYhRe@xRF~z}-_X{iWHnG^XvdHIkc)B~nN``l
zT1JSH<J(MIp*sVZMidxhr!NyS!XFv=-!+JiQdCg<<r5UkYqP_CiPo{uhF;-W6D_+T
znHg!LE4<E<$c;Z+U-d}PHMjmsoE%{PrT+;tEO0SFR`cmF@3#UYVNvO@(-9%L?rww0
z^RW9tp;dM9>Wtwt7FJ@Ce!e1AiJBc99)}}t^;5Neu3^G$6H4?b_e3Zy4->QYWNmA`
z!DZ!eTIRv)kr!E|QGMAaB2e1TlTR07ZHIr_VoQ=Q<%eAf|Ggpp_S<9bhnPBVO$->u
zcxBX%*@&zrTND@#odU{btX9N{xM2^NxP3ZVW9?tC_z3@c&O;Xadq`erx%Hn99)ion
z4;(C#jf^b18$)lbV+kP-HTz{uGp1L6|IW7%8Q|ElbIP^-oRNG9hsNwLpPN5NtR>Ab
ztGpQX5XAC^wTI<#UY@}x;?CCWl+8HhmW*nL7`*$vgkD#S<iAk8g?VykJv7vCrS;XP
z-QBqJovc~`s|VBFaTz6Btj2;p)y?9(Ph;O`d|+cCTyi5Hto}`kF{+{QY|svF!Tj<M
zYB)t-Z+1N>RR2qXa~br#nkOYf%)v^?vQ}WnZaC?SN{({ysSVW`t4V~b529CKB9&Rx
zqt@m%$DKDhs`&BA9mAv_z#^l>e@1yUgard+DeIB{u)K9l9=QyPby*)|{msgf8t8Y2
z^-j5QR0>{?g{75l+GwqZ)BD!l=2nHn<c8lJOpo$i&t%0M*A$l_TU>2i!|niVhJx(I
zp+;|G{&IZ}Y+b%;o+-I_<Jg#DVY*~p5rmBpZO9QUf9|8i{d+=ZPgfw9<}%AH<Zess
zw1SqZe^wa*tCJa0Lf5+1BHvm!Ejv-bY4WSyN`^vtl2BJspUvgv4EbHB_3gqe*8A7j
z#$W#(8ibB?p+;XD$^}9KwwWI1N}GAQw+I%kIL^;o_YN|$0wy*VSiH-M9^q>o7rXDO
zmCKWbBn4thS8Oee(FzY~q#gP7r(vi$Fn(wED@NNZ-`b+&Pg^nX{xiX;zjbiJAx{-9
zaZLR(LX+3^fa2aqFkb^-gKW!6((I+DI{xT7Z_I}1LNTd|d9LO9{D_|9js$J~{-5ks
z*zqcT$-}Aw(?x=*^1{96I@)h>rY*;MNgG%O#~MoDIC5OWQA|-D4yUy`?U70vZLnpQ
zo3E@-^*P(>UK!C~?8zr*h=gerF_y(2g_$GzgY?BvB*F_*uImjpnEG+Si_Q1!?Ai=6
z&nl{upvgPmBxxM%Z0F{MrmBRNG5xvseu|`ioIHP@_%%3xuogGq3%eUyWH(6K1#U0?
z+!#4i)=Rzvm)Z1PA~_jV+q3(K0ZWc+($v1iz{sqB{woX^v`A80mCYU<HTvq??pLh4
zuH1{KiS)<$)_A^JNn09^EUnEt^({5*JWgp;p-?j49G@HQBF!uOM;m-0d)B4%7Ox_O
zPNU|E+0knh?~XYJT^U|sY;`WMm>|Mal5ngYRwNek<7%HgeWPQ`MlGz?SX8(j!0Y+;
z;r}=;-t#7=`yub(svPrs1y#|BzaCP`<I0hS#e%h-8~(>Gc4kWkiQPSFt}_oCp!M-I
z4zG{x7<V1+KjY(0Av&bYIWp3b?$tHYl_5)(lM~?1T_P!m3nmO?d1^jkBp58;N|n(0
zP+%)ZOINZTwM1~4qT~A3cs+WQN9S~hXsFTH)4kB(BV{)Zj$=7nM2Aa-=0kHaygM^z
z%oAg-hl#yh2}Hf3H_s^un@-KS*C3CVKXNSjJ==wLJ3HAp+2=Bf(#(U&HPvCf&pE)s
zgNgYgJQyj7LdcvJ0}y8Aaf#D8ociO1c}Yn8!MFubkRlzfRLiV7mrRU0+jgj3i?vBX
z%n+T96Mrii*vp0aaf%zPQGA9W`R=VUah%M>v3(RHf$)RPcF@j`8q!dYCjKqrAUEgv
z3cXw4EFqbzb*7(6N{;0WkzK^cOYSdJ7%9vLBjEUAvldpFB_X%Jk5}`GQ>6Uk7fxmT
zykMtymw<2LoODV?%4ms5H%&E9o$b>r%ER@qMWNg`F}fe%9Opu;gm~*AA~^=T;wj-#
z(HSKn|8c<kzS}0yqw5$;Q{#&)+4}W9eoUjpZK%%;)qYS^WHev-NP_n95?oEz>O&8y
zk0i}w)!Gw2gp*T#P;GBwUBC<2aD4;}$Aj1-{3tY~ZM5&(!5+G`<+p@!0y*}wzP{eu
zcSc;z!`8I$>`J-z5>~oU+XUxSg`DpL72Ijnt6rQ}J4mShU4|lCa{~d`B&?RjmMdei
z(tb~{`o~=(&$Fc7(aNFH3Cx`=XG#N9be<zg+8Y+*-Qg*+ocus@Ba3$*Ow~hnz*#Dx
zTB&cIhi`W++UN*6)XFirs+e(N%irDWh}#1l0OPHDSy$9{98qj1bmo2wqst#JAI5+a
z_wLuxXXtETst(i8j#nNkd1riH8y$bPTNS$TonS_H*%rg7+_Kd?iT^rjo<aEo#Qeq`
zcu!)ZA1$nQBSp{j6~Yv;$rH)ddF}3FNS9mWahhIO?|~V45*@Q+G$JQsuH<>^jdt?Z
zXtMf^%Y`R|#UD}QXRk&7>nXEd523%qPYFN6_C}r^wv4Me&nYLc$ZCBbD<^PmZ9Olv
z4Co!DJ9Jt<HxXhfFVC;H3qdel87;$~Mrm~NXZzSQm_9gh$EHl?2?|Vo1l1pL`Se_v
zB}K*svNcUZK~;g6M2dqerE~HzI#_AxBB`80EDhdc?s#lggK}7lc5cvVh2B(+^$Hm^
z4P$)X4yM{u;$7DUiZX-{YN-d2v|U>MoT<SZbq~)x1X!m>t*U*9#`wffFo`8cuOiDg
z>7H|Fnv+&3cBF)lV4ly*AHHj+3O7M0Iq5y>I$Y1Q6=G$*AoivoFaOm$^hK8?U{L?S
z?XbNiEq9hANk%4f8mgA{a5}t6DCS=gnVq6aXH({SF=D9v;>rHr{>M*U;|0rg4>A-9
z@&4d69&#8n)acun^S{qX-(3#ZA5DlaE_AMN3%j-bHmsK?A)#ocg#kfFV-f;xVmbHR
z=fMcNcZhI2$l;>)tRr5X_Y=uk$mm?#*0rJK3I#aF^pjqbJ2b_=hHC$bo+@HvGdgH$
zZpp~v-hFWAf2Qcm*?Ws(y89u|gs4AvRs7rrUXh*>hQ6pJ8S;sqg^jtbKA>F9-obh}
zGZ8$KG0lEaQouv-s3RR-Q^d`gk&?wvp&Ij4DuafI(3YJo?u)MoMtn_G(Pmx6pP7mL
zUi)1~$ez|~iqWn9^zI0L<ahVu+gp^vvJIIOuS?3>I)DEY9HGdPm4P3TjOnfXeD>r6
z_3@^kvH1qHvjoTJ-nRCu%Q0o{A1CfI5Ev;U-y#!iwXQj0x9{^dV;gylUlkrD!%u&&
z_sW`tbyN7)mR|UJEkSdyCR=G?MTR*(Osl)+1*G!gA0+F7gvWEgXi0gHoFX~sln}}T
z<~?`IlZBei@!M2Mvy@rr8!Ddyf64N6-<GR*X0$%q?l<JnW-;y)Sw(yiUp;SS=yMyh
z<Nf<Y2?@F?g%Y!xkk5i3>Sn~jInAjNN-tyaQ<8nMN|efqoK#}cU1(ptDcHOFaAqCc
z#3GLu+D5;72FCP9#?ls;WMF3ksCrJ-viws-U{xwr$<6WIofUfve$SX;a8yN*y8ZSm
zDR3s06~fJW{a*aC2}l%YjND0QtfOvHN+qsN_pUUgjfYx>TfSeFS@uL=Q?2vZhsL&}
zwEW-wH%GZR{ByhD*=t`YEj2zQ(4TW}!2FI`G~uC>%Y>Reu#fI@4>CC1v8{%m$7X*G
zIY!`-OsiRL%lJG(cEnmGXbsHw7`6XSL(?s`A?gzl>*QK?bw_Q^X{?#gk{0qU4qIDP
zf;hf){23^|lQ*55$4Qmb6S&QU@xXdXd&_mo^c{yan%C{?HZ;1rcaz71cOPLJgS&z}
z6XfHtxZtufSVmuN>4>otBTJ8rci!|!!J;)n(&8A53v|*d^B>sbN7l?WOc#@TTW!h~
za<Db^wtT_opetDvdm^lZMa{#jVtz6G3}$TPY|;fXJsK@x@y%JOd@6V;(h=I5SYw?g
z0LNxaAbVONfcfe{anLq1mlWL{l(rLS!RIPv?t3I5p=PMnjOgDs((&*(B@18;*hQBz
zpoDax^>=B(gUxA}F=Ao{L{FZ<Uc4IM0B4Na@$7G&Y3fpY)T!H1eQBmjy+#55!Ft<x
zJ6l(<f@cC4fpPKCdCcn;q(fo1O{|PXAqyR$jyUx)I+wRfk<Q8$EWxY8eu=+CMAAM<
zbaaTND)+bahH~j7SMrW>o9L*}9j&x32-W>&Vl=uanxB-Bqs#j7EmKbJcKNpcyX9x;
z2ib90-*oeH)+T<{>n-=T50m7)6N#(UT3eQkiw{ZLtQA8L5`Ga9B=<gmSr@W$B}0|9
zH@ia~?35MFF8;vqu-L4>s~C7&m|v4UQc<F#w+RXF&`t?>Ck`VOzbRuf+x3O8-XJ#S
zaZLDovkz>HWh5|c`*ZSj+oD$mT+2(*o}fjDLv;@^I<v;X@m!sQ<gy$Rdgd%QUs95-
zO9TmhmA`DQVlih1JBb@BhzrM~kQ#nWct=PJ&A6L8XLbYRoBp7O*Agu;B4JTV_ut_|
zQa!Hq?YqCr#7$Yk#b#xrTORHCXWRr%=M)pOztvaYWS<=p;>T+*hHD*FzknqxsB1p&
zpTr;c&KIEPD#LO7T9v?sctn5HF!T3;YAaGD745WKp{*!w1qj1CB=Au2?qO!;{!#wK
zSIvZ6_te6|abp!sbI(=5>AY1JvvD58Di555C#EW#>aL<`$dS_rQw9t!oLZupE2OrF
z*b?--DFLPx9O#qZ)Nz@+7o*#bL-4;tXn;J8i=X&c4fvAAr{@4@N_KWX2jN7V)Z9Z$
z8zLqfeMB_j50|UaC1Zq*8%0N{p1rKA{IbRBZ^O$R*&Xp1#D6X3ZsOtY1p4u^wm-=P
z1;nlKLWvd@n^!FdKNNFDz39F-gp$)Ua76wbU<Y1nx-#1&JulWjkmj{`Q|3X)H`t5c
zj12IYd>NLPa|&8FZC&pe#Ke3u{^T1F4bff*Ozo7_ot^)Vma^;W3`Z6Ejy8>B^?8Nt
zb}f>HO3%1Cv8=SVekXZ4W2PtP><r_p;-QY7)OJZ*B~dwW&7wPuM-9YjwB;1!3MH0J
z9@SeRFbtx(JQ9ODZS%|WpWp4dNJ@;BbuuOHT{3bDi-f6$ie%*Opv#L2pHWdqLfv9K
znV^#x8kIF8dD^b;@3c1~9zG_QRTQIp9C({s{2t%w=|D}AaGSqBPV&UXjEUuYWhu7z
zyI3!S_fzyyf>|5r|1#!>6tckS1Retbx)w&h41C>QkS(XJ@{)XfuOFY|&p__ttwik&
z4@T}USY5wsFDf`BAqIhx<%evpC#X5m0gf?z@^+?sVg1uxpH-*>K3bdkO2L<;j0k~C
z;L*3=<GQN;#`L6i`ywn~(BTVKM)CM^dvn3|+NL0nn&!@iz5Vy$0{Ns<$HMZTKhnPx
z2Ft{cnLS>le;lE1!}y&;l7X|_eC<!;Jwbks@dDz<Gg6sL2as0~376TPQ`H_x+0APL
z;Fsk)=0^P$D_Y8&p9tGt2Ya<K5qIMi+czp0V^wm}=b}_9fIv}m?qz$%Q?qGbj9;G<
z92IuTuJIkKPFEsPy|}>RJ-nz~7Z-6f4m2F6x!|+Od$zVzFZujIi|)F+uL{b!Eb{2E
z^lAtnIYG8m0(3Fk`!aN;+d!#!?t14#)rkbnV-KyeFvo)Y^P56f?zg&*HuG$0OEPj8
zO8oH_H*zA2DvF2Py^cOt<Y9kc$oXPy06?VV29BA6BT_|>^?wNN{a<O<rXRo}$Q{-}
z33aaglZSE4Kh9Ab$KA7GQOM3kmzNGjecOKkxVaq-O)xc;o&dMm?VyyTj1hmmqh@d1
z8vwpuBPtkUT;0*DwF2}T901VRyW4{{H)U;#UWp@=AE)RxrdHmTqs;^x=Wie>A>)(n
zw=No*5$cLk@>g7HaVrgq#YNhcu8}_0_2Ra1bUOqy>%U)7vU-3fyk>`<r&WIGiXM4E
z-QZ1}0<}7WBsVuvU^X>l=Cb@O#%j7FRI?u>tsd;NutJtyJ`7p-OX8I^QB8XHby?XC
zR@P+=Rb2O<^OX^Ctx<6Irwdb2s5JlL`OM5i%UIR%u9pumS8q20%;C}BKqRb<HD9Nn
z1>o7}Q&NunI3n$MsB-3$o?MQ|TP?2}_w@035~6-hh|BM2{vqHBC6!dvqw=%kZRWdY
zrrmS0@WV&oXany{rZMQVcUCeJm3>zyZ`}9b+b&y0rW*RSiAjNy()!lQWP{^F@Y<HT
zA(LLrMkUd^<{`w7hXUbQ$ifRFna_`F`5D`ZNuFdM>>2yxL3(F#CF_+0&A#ZVeRSdW
z8w&xujK_H5?R=;&XwOXnj)vp-^SQ^thV0kGTwcj}@+C7ykJPR{rX&as1^|bh7j0pf
zPkWZM&AUAHzwI<Plku>8`m3peuk#smkNxY?UDrqdLyR9%c!r5N$s8lU%1UQlO;$*D
zcSFCX`;!Y!kA9{MpQc+LJAI^7)vao~N8h`>*E=;SA@yll9GfB#mbn3j9qgb#W7YY&
zv0K)%O!V<tG<$_Kz0TU?2lMMwVD8#c^T#@eC4G{2Ad}x5hicXvqw4D@_o;?Vj;gSO
zb+;b@%p}Jm=YeW~g<z)B-ikhdMA#7k(If44mY8d^uEI2#pmrH?%Q<ekkmYvrno*Yx
zLRg92edO<7Knp~k?tiocCw|kQ20c>tH=64I@SyqJB0d*kI_TCV=Xb((m*wJrIFdQv
zyXSUXvE<OAB4$0>8-Dx_j2wDxEIRhqt<w78ORI`!&)G{{&;Cf_w$l!{Z`_3h;IA<!
z;t}C8bZLQ~1v}KoJRW(3r}*y|?2J09dw|P77zwzRmefWR2=$g`ZoE{h3oYo?&fn2O
z89_2+sVi|$iAi478#}|<(rO%MQ7K+}mMxs<4T>-BhI1uC17E+YKcWZ>_qm}Du4f|Q
zNLLb<2??%@NoodG9lT?%x&_&SV5yr}cfWCnFaP<K<B^We`7HgA8K-=ib#s7VwaL`>
zX7mr-^d|&G+Y7v^+Etx@T(M71qKRDHM!&fscf6iXm)zTWsoS)r_o#bg08^Y}gK5k$
z5Xm}I@SF5IF)2r}dF<IM>Hpyqw`aYwr=5;``~Ja_mL}$h9zlU|>esj;a3#`QV$<xY
zL<YdVIL5|usb!zZ#YKL(T%J(^K#XTI%&EEG;_xPt=9;K%MHK-dQDvD<jd|U2OhE4m
zw!5<^&b!5>SquGUCDF2ibdSDIg9CAufA9hO9lX%AUDyfagWV(PFc6q1BF{3GX&M>!
z-=7q4gKAIxd)WlY$X7`Fa5FSyUBn2Jv=Z+p^ZJ=uY7BiZjKXM+ceQm}dg6cT@+cuz
zKDGAdG#z3eFWWV;DlGXsW+RwlvK|@U{ZSX77NwUM<us%T^3#XRCRx3G(!o?2A1$k7
zOrLoO=XI@goE4yH?A*>H3o58ukdS4`Fx;Mk{1j*!x9oNE*f#4-l?q>RE}E(0>fX`K
zKjPsvG_3Xda~p#-gjjpt3`S2%szbr=Z<i4PG@`faNMpGS)xCix{V3%ZE>=VLC&|{!
z+9Wd+(TM_&ysHA_&A=5a30o{*9>#<R#sy+h3i2u$ucvlV)$TXT1LT!Y!TAMP*$5Or
z2_I<r<*OTj^Re}xrr$J=ldmqtL;d|x?x9DbO7!AJIFwT1wZ5@<WEu6-AT51eKB7z&
z$(Sp{oTsRI{N%r}UZm%pFU-Y+5TPS(pe5ibIHk^OcHhX~QBz^l)Uvf#Qs0J)TOZ{k
z3e;#Q#aM%zK#KP98@gzySa1Eln`<R1C+W6JjrlYC8V3=aCs9!!UqntJbB4!$<1Hzr
zD!i(VBWl1Q^|mEnUNkla=bok&ak4E*S&#vdKbHeNg}%|JH9;OJwkA_N)d6T@LmMu2
zVR3J+B`9ZfF)n!1J*qSOCE=^JrfRIr7+&}T23vA42_pH;^b-2@roGB!a})P5CXg$0
zCXN8;N|T$%&_ed=4a8^zQ1ngWzQrbNuQpwA+|`}>24Ug@BiWWRK7eF&dO^gu8u^D}
zMD#3ooY7i7>7BhZH(>XBt4;bZ%(B%zSTi(w62#?-_f#LCS$b=)j+K*kC<TNFoW<`w
z#>R>(G$ev}4ae7J%}OrEWzoW3cQFh1TbEz;eMnBN%nzn-&<F<jBa7<d`(&<bQhVHR
z=ACK0%^<~$xsHPRj8ZL~<?^G!bK<bUp^k7kcbe<b9|mPdnHDBxMjbwz;AY#vQnS^j
z$&PVY6ZU`SDU2?BhJfPhmtYyM@gi&vNc5I!=+j_5@>K(LN6&z7#^<ve+m~Ne9OVoq
z<>fsun9XKTKQP?5;hJ_wa=^>d0I|)m<E2DsOdR9qXWfcrv8QIkWYVXO!K3_Anp|+@
zKA~3P;%yYY=8IZWKz)1ol96e?AgvukjmzTMA*V%zD!$UKTdE;szCOyGtqC8XNf{{w
zn^SLvFvhmk(GNa)=dFE}D;>Z>m4N8PO<FPZlzT0KHXCIYY`_jA%9RQ${04um99wGc
zzI?k2j$OXGVGEj!?5DSKo|I|ybhFkYie%Ls9vCBZS_ek&=e+xsFBsr*(TBO)+2u&l
zKDMCjRxtLvqs1UznQm|yWd|YE6B0^*XM)m%8X-y3cUk^_dg1jxGf8v1*iRc==>eZ?
z9-j=h#1?MpRXplc$WiX}%b3;!rIo6v*KP=OV@FGT0mB;~tzLeRw_AEIvWLu;f=QaX
z-`;X4SFiS@2bHp5OLXwcCETISY|o&ED|r8aj)5sC#c4Cw`3yR1Y8w9v$6nGiHR}Xa
zU1q#G0mvjtmR5@kY<BZbs$axESsQl5bRPb927d_$NNi)_wd#t8QNdn(A57Iwf|;Yu
z(^K$3<f<e|X>29sCUx~ch*VN*fu|eD;5_&~@xCD>*5EPbo}1aK4q_52WXsBml^?#q
zX@t6b$O}0*`}4vW;0c4j&@n*vc1mgZ!MPLw=T~?U4P>Z3CN|u@?&0{AOXh0{taOtV
z&bG5cOk-s9N&i!~bkfazl6DEY`HRiB80WR-&LOrnNYl`RMtyw1Z>51U1_1S=?K2G_
zA<s~NfoYyRiBMDc4F<3NLA^vD^tHjo?C33!;sCIy+)g*rFB=RG=cH?gQSjVuJe%S#
zv7CF69};rBV)qp8rQNy~AZPQ8D>-{o0OU4IS)_rh$!%&bi?r!#*~#t>!+*l)iIP2n
zGgV{AaB!%FM31EA2{$(rPrTTe1t)pQVkOR|M|Qy^S=4$3fl}B&5F`ykbn_DF2Z5?H
z49i%!9wZ~+WhKjVG4*NO1!eh}3(z#Hd@cVVI5Qv#4^GXoMK95v)>N?jl1Q4e?C+o<
zxNUT=j<zv1E~~%N5{Z11psBqYRgS4YPVBuznQ<~5l4E@IDnK~KM{=oH!Zw7{ohd`$
zo*a~N&=HOh0%BPY{n`qRktC0Meq-kT339Wjn=NLz`dEI0QzS9};q}mkw$k^HmoOMN
zToDR$UvQh?|8GErThtN1t)NWXNLQYJa<Dfp;f=|BQ5hA}gqYcLup0XrZUlpbt)>G>
zo~~~M!b5Rd%{YeIu{lIV2kh-Tm>An2G30`?ex(BwmbWBMrI=#imq6z8<ArWlFLYJH
z2MsPO4M<DrmXjURMSFj#b?<fdswc}m*aKqFjaNV8DthB<%7_<79*C|ynj7Q*?C2@+
z<OQeOis}>E#YGo&&Hh$Y+_rIM$@UO$JyyL0sPj>#2qD|E%3=w^rD(uS@Y~BZZ#{4P
zS(t0dD1A_##M^b8ESTH6C6u*d{7G8)+9>vi=3nv=bdUeDcJ9C%oWf#WfVXC3+7uvI
z^J=PrfIaX_QBT6oMXQ3-y_DsXeriXGh7?VFUr6m&1oOnP+dSQ{ZcWFj`8f2Ejhk(D
z15@ACbIv8dyVUv#3?If1@y}gqwRTBEb77wMc$2@i?0kt|BR;FF+-}z_=r55?&^W7)
zP83m=V!RAJEG@3c8>!tK8t(TYJS^9em3T(Si9;!<>NkxC3N&`J?OVr2^Mo-LV{JeU
zKH=D{9#&)G;8>DNZe*Ho`)KuZ8jtsXdvP=E<Jw=MkS(*&4eGdH+}==})huu(S!TnR
zmbULPXcqJkO=Wd8=gTJjPtVfQg7C8*M`G4{C8hc#k?6Y}WGd#oi0lGcFte)8^7HAc
zhHJ>r2&Gc5<fu4DSR#7<b$=<6VdzrI4aCCjtE1q0;6o<!Ch{h3bg)nJe-@VAN0wC*
ztN=r{&&bVgJx|s?{$WhyRr|D)m1;Kq2_Y5COm&e&Dg(&MJMU_SYX|E%oK+7uR+l=#
zG|A}%=Eux1M|K5D$~HH5KCb8-Z{Ca{JO?3u&)vxoAaHv8NJ$39ObM3uvB?Lznxud{
zWX6i)2XAImO`>|Dfw*BOSY$=|OC9<izuh#i<mx(Ua;RI|(j-mQl4bl&MZvo+npT#a
z{2KF%-sI=}BaU?|Cn`n9E4;uMIk%itb8<Fn8&O>Rf|0qo?~Q+19PgAy#R6(){?BCb
zCZ|JQJu8VD(r2l@COoO@pru8et_eFnf+XmY`|}8WsPVL45Lqpmh=>>+FXZRPqozK)
zIM+QRM-H@@4^amH)ku#6@G*5}eV7WepH*bv$7AYp8*!|Q4XQFy`KV!Kr7t6EZOZG4
z;TD@v8EL-AHYFwHWrB$w-FLtM!m{Zo!A&CUo}C3EBJ+#d(qf%_aEzR)3%aYj4}*!C
z8f#&kySB3%SpX%<Hy4BlRq=l72>H@thKa7+e*mBV?G5;X3vc1(BGv2pH=hh2pUbVd
zx;nQ`UZ-eKOX<eiE*RuNMBB@D<xJZf7#g*uqwBs8)xhe;QL&)(VJ$;~B#mt2mV#8K
ziN<@iJy+Z}DV~sG+h-oGH4k}tQXDG(Cxk$6AAUJ|2Uz;O-o8eRIAy6gkq`PZ0Pun6
zs2KEhr>rcss*YJqgX<HM661|mZ0rN(EPzM_p9@3`u-g=tHT5*5jx!xO>M5RdLRBI;
zBLbTMf)KhPa+w&JsX6KI`6&kGC6Ny=8A+_hNmS&>c&PMbP<1jQ1NCs0Ra9LeZBn<*
zA(JJ|!6Fh;64l<Ur85MgGb&!_Wf@IIPe=%<W#Vq9txQPX^faBnm~-pvcPK4c5{2ry
zw+eYEAiqW=Cj<d_i*PD;6Fj4fKJrD_!=A-PYN!Rj`zblBiF(z0mBN2f|M|ZW^LA?L
zHBg;Ne~iiERGtnQ7%I^eRBGdK)|3Vl;gQXncszDBTy*|lWxWMdRb3Z03MvK?iZ2bK
zl7fIp2@)y@(%mJ}EpcerfFjb}Ee!%v(j@{S-O?Z>UD9>u@%_K|zjs^)4j6dO*?X@w
z*PPFM=3MJ33r$hG&s~PPPPDuJ{+WUNLxFk>54zp#v>f?7bkzf);_vd5h`!IlRDBvN
z$(WIakhknFH--yc)-}{DcdoL;+r#*#)RlENHz+62Xo4&K!iVnf)5D1_KS^B+)GM*^
zMX!k1Ep@OXP4A(0@=5x1L_-bA^^6kN7&e>}>+5oZTD$t>R~CsfGb19PdZ6>hg)?2C
z&)+YUDYE=l<-Sc>2{i|AW@tfZCeFM1XqB?=lW|V+X3mz!){(OZ6^h92RBsE}#n^y%
zC(jj4QE!!-R48R`xwzCGS0o3$dslQct-F^{FPo2C5|#B|2v{GA<e~N%bPS%zTYfyd
zy<R86U2LlcHR6J7)z>@Q8%G_jEpefBlO5Q`K+klp)vxSO()^E}y>m4%=P2yuGe?Z7
za?3!n#}vsJ4S83B`|Xx&zD$0gEBr*EaqAY8-bttDK_^Ev07oCv$&!N;gM_+cBoY-#
zewp$^Z|TNlCMvO|7J48ObsW9#Qz0!w{YJ8kjK&*t-!WLxrTd~skLDZ?2(=Hlh}8vn
z{C8WW2XdxzkFu}m&8|4ntw*UVN<#YfC?}#}p{}eYT19aaYLTVRHe5QEKfjf|6OL0?
z>uajL<`!4gKi(Dft>c0FNEzbOK&-9XvZxcTl~OCRR!i|IpFxF@CpNp*-cpM)n~%Qs
zq%8XS)y-YZ=~w2@FG3J3%)wj!UXVy`kN~kC#)gfH<J(v>NO{dV+X-Euka<~G>#Rn_
zSa!5`%8`%k4zFYoT+`AhqbaOO@XuL-YTY>!SXFK^nBq>Rp(u_pzMXPPZrrHp9&Hk3
zU|`7ve0|E#|GV!S7QHPy%XmdwO=RAd%P$j%D(}_v3(BS%h}=oO>r({D2<Ip(FhuX1
zoKf2y=urx4;?Q!jQc;ncK{nOf-h>OkL3_^O*=q*O`oj5)i=y|#RAf`-t!V^qQeV8t
z%W3tK&eQ7-^1n$;VJ(7X!64$W@wMd^{UI}lS9h1@9S;6m9@Y?EZDg<Ta$7&MZ}s~Y
zhK+UQ_3YuvC%V1Zi7Pyb5nuP>*?=(=vAdRw>OlDfKRO=UfNq);h*?$+*Y!i9=?&kX
zpOY{QntFvR9G*RA7_)k@b6})nv$Vhrx!`U@7h%|}i+$@;x><Sk%hCOhe*gZJKJ0Wz
z%}3y&oFo-$dSA>sE1-st?!gHSW46uznhD;Os1)j&zG{HZk60)c*L~q~XwD!cESWiC
z24wji(D6;zHL@|mIGL49Db$`|g@Wi1rNEEv9wwa4?Vc6KVhM>68)^jGW0)?WpKND5
z`zA$&W2}-&!M5S>Q%RiVdUaLY2iF5JUfCp1s_P|D@Ww<(opnIwP++v!Bi~IJUGHO3
zAm>A-EHtK}He+rip)SpW=+rizj}ET2`nat522s~mT#{gCS9dlRJWs23m{%00X@5@r
zdtA64w@%*A@8>N_pByCg=0)CtT7k_|dk&6(HYzs}OSzi@TthX({2xNy1%8ef<!a-;
z%P`tD`%Z{vU&eK!ZP!O}&ti__hKh-l&<#}+UCF?lo)Fo1&!^rLHZ;()F#ooq#k-}t
zLgc7AQA_qXH6^6%=CGR0*(138RcYy2WNkD=hlQ9t@|qX&zg`IZNL8TD|Cx2!YiJ*d
zZ7RA3JR=(47qsaen6194s;r!h9LMUstgaW&J=@W7-?dM)M13##^Gg`V*r7uFbv!)Y
z)-k6&it}SrHbQKy<o%Vl9MW*o+2Pd!=wgz*m+IqkYbcn!D+rQzE-pi04j-tNDI-Sb
z@;MPj{8gBnd%X$_E1&%~wcEU3JP>do^=R(y>Lb-qXEFLy<DF3J92mqR{Z5cQ@kMR5
zlEk`Kru*-9PfX23{%|>Wfp3`^?0*vRp<>m`(=uF={ERNLc-N34EBuc}bM@~59=|z4
z$b~%%<MW1M<g}`%jrFADb#G4T;2^zb34z0wHg`F3Z&3?PCh>}y1pr-k@d@fF&@1aj
z{R8LN&j7VT8YpzZb^mJ#XG^DxD;Wk=qJXG-<(|}(-F+p7gXsfO3~dtr;+Bf6_c$#>
zU%gP!4UlsZnqId+DJT*9ywNsI<dx;?k1AAw%+P1hyF(PFq4-V06Z0J+aQYbdsR<Zc
zsii0kBT@I;qsjg3S<{v#jj#@#j^eG%wA!9H{JMmJ<{Kd@A90}QZH6Q#*U40wE3fwq
z{0Q5xx^k1pg|3!w9L+*MX_daPxsZM>(02Miu<4a@1)cR{>j?>ss?^0<ndpD(ZKZ$I
zBe9S8$@kPmory#Toj~JEir8kr#-NZ<{6PSYq3aeQ2!;|;U)eF0_2C_4qFh~Tn`xzc
z1#iwm@{RT@cX(%`>$e!8GLs-FX*jvXzAJu*NJa6X$%>^P6XR9m1)+0~J>J==0~9H-
z)#OLpTf}%$@hO~lQLKP~hJq=(O1YR3m@?gE`s5W!FP>4=odXYd48AXJtYY3Tb|2q6
zfb<@c{bxYg9IaLI>LJccSTqe^!yAEm9w2oo{Lo>ZH0E&0V?R@Buu^QYcv$<{z0l9x
zyg}f)vWFiZ&fCygj(BV4r00a+N<1-TNo6yddjmw77|Vm#EkSM!in+W<a2oW#Sx)J~
zs(xi*+4Ye=bC|?TPq(XPDHuBAVrx*`q|LOuyqQza3()*Op0aWcKgy<dbdvQ&h5?#Z
z1~z2J1vQ^{R=cctvFdFJb(W!I%S4|SZmZ08)0afv*(<1JkMKZFLUBSNx0;`vPB?gm
zWrbf+9P9@Bb8HCCwxzhYytjWM^Qozm_J37FAW8`Ue5NwoXEN&Mk@8v#zWLTzse5=n
ztIVX)KXEke@}ySXbEDOhNHgern;lHSlQbEu(Vm6+%FfuPCq~Wg*akxc=dfbgLpI(1
zll{<_G!F!j*l0Y87X{RH;;*0Gy?8d>F<*+{JOK7`dv}p{mU{M=+7~0LFLW*DBu{Ea
z8TGXQBMIV@f=2k2E@5qh_8fNfgr8y8NyUX|Yd+uLB!@&(ZiU43Rh##Xq!HVz9{5Il
z%6dF+8E2jCjTWXo!R5DDOtvLQaz>e<+f*?T%gl<8UfLF8o0ED9b}d$X)(AnFdZ6Sz
z-QJv6le{*?(rv>rHLgebW8ceGDQF0;#U37}FD@*Z+rHjXAt5C}r~Y!}4j~Hane4Yy
zWLb<xzto<%?b2O_2H?8Gk3;g5tbvrx-zNFR9x$X2j|OVCb>yNbBGo&yj950e8}%S?
znxAa)udZxTH@@9mi^``#3=P?Sbo)3|4Qe~jOGCq|IfWnrutpy_SH3gn9*X+T&ZvC<
z-rm|j>Nl9?^GdbdsV+QJvL&up$hy4r4l{jp|4o$@PMQDs?Ptc4=n~{VJIeW)&r?$h
zvz}0&y8c<7eVe59nT>5h*#=vejZIw-QWsT+IR^Vogsx_+wnTdtO%WyCwuFj@Ve81N
z#G)pz3u;HdY+q&&MEQvHob97N1y+c7)*P7~`C@rL;3Jg;B7-0<$u}dG)Tmx3ATl<#
zi;c)?5?y8PEby#QjTt_6l*G|@9aSk=KHRD2{v8l0o{?~FeJ!c03(!};OXp=+I`6(C
zpH@woBHAOxq0H6`O)t=tfek#GZJOxjGlNR6(UY^5JV1wpz~d-X<3Dme<qD#wQIcrL
z9WZ6d%-lKHs|^_coOrTr!+UsWv)SHlw^t~icj(+?NYj8Xbfl|i-my<Cv~<v#jHWa0
zIlU<3DsnokaaiS`JXdWU{6v8qIXQ3I9diN5-RTd~;%(irv;Wl)o&3t&_<Qex@aO}L
z&R;vxWd+5Z_nBv-+h-0~HJYHAndAq0BHf^AV9<UPD3=qqea@EM-_ow`m=ntRi-SU~
zm7$3mYh;m(|CrYkN|h_EsM#@+u<Plaxt~)ymkrioPQIm6VNfkboS49PlK0od)>c;Y
zR-iP9bFMr`NL*sxT}L)sH~#sW2Nr+zYXe2w+Q~R{)vjJe1*1;2UM@Hm`#<;keD+5C
zr)&7lU;2$OV<>x+3u*9-2_7h@b1n{tFu7Jckuy!X)~a^^B?CNWW|^gDS&&q;T_{G$
z#S?@lGNvPleUW5J!9nZ;VDSE-*7%|wVyHEAv%BEwR%~2dOQ3jD_ApB`hXw*U`AB8?
znBDfy?B;N?uABs}VupBg!L%*9ZZc93sNH(;VNdma?IGoN!Y{&OV!eIzIynRKJ31=L
zY)BOZ4ihZ0x{L4eg%Qq#I;yJmbqtXq;EErKi&ut)zOeSD6Eii{{}#+z-vI38GG`(4
zb%AOBdhX&^oxDNc=xqQw0+3hd`8i{3DsJgS3MqnEp=8*Pba}%-l9oQ=<h-(t>%&$d
zQSlP^`*eJR1-P~R<Bpb7H>K(5c7SvvY1c2cwP_d`50LlvIyia$kx^v{A#2yK>1Y0c
z?`%g0)o(yB=7QuM&nf3`VK{#=5gJ&YS`yr+w4}}2oGj|EK~^~|DAr)*3E%UK`1-m)
z-!L`zLG;Y}?*`9q!PLrXjl}Nw*dn%-#oG2^a_*g_WMl`LBHb$~f}+a^0aZcqI80gd
zxv(J-Eg?1rFXz4<GGfNs-_=^(H9Kxa|FtkgNAs&J6)W_Cu1s{V;#sJn<{6)KUD_F%
z^BojmWw<|MY)k?UQqp`rYyUBNf_H`JA5##BTdR#wD>SM8U+TaL`buX~U*@_7x~4d+
zp&U@85Iqa;sytR`V(L>Wl*ZrNGynNmAP(8fx_58opQ&0D(<834W+1#b`@1G+^z{{)
zte^(cJ2d8Rm6Rw|_yr{~H2G5UUUf<cl)gW6vK7=m!=*6}S}G4=N(lO(Z(1+FJDVYY
zpM0@a)XtRAHtM9I(v17%0|uI+l+6NxM04|)@hoB;!6(<>I@fHwXO(A3|MD(Mb40d$
zl7B*-x6=DXLZwkSw}gViQK?mXVPRcBNmSr`DUX`dT5dr`VLtAa%)rH0=6rdDntH&I
zLRaO6DLRdUq_CzZhu<=aD6W2FtPkJ_H0E#n`xkKy*ZPePb_JOsCoRy`7F%wE%S{me
z;d=+POdi_@DBgqKqH7z7biDjV`az;w4X4w+b8^0rWc$_EW3jL9y`KHh&RRjK2Sj_!
zx+kAN3Kis6qVhiK1NePwi-TFZQ|Q)H=rWui-zEP}=x2(GWgV{&^L{RNZaBHi*!hMP
zST*;gb#!gJIllaNe}To@1?K-{J9n-g!RCq?PX#1f5n7BXIwW_UIgg|#x#E=EE1O5n
zEa;4klv>m{jVZ|W^vtCK#M-(<@|A~q*!Eh|w`$l-_@HBUhRs@2;$_45w@Zi>9FS8v
zsIIfC-KG|oo>Go#3R>(Ekhif}9?y);WAdpx_{Zbm56)LTy|#|drtXQvz0MjlW>z^B
zB7}~vTt$EJAEKLNlv0wWWa~kArTbegg9YxK@tE=@oUKcGJLgP7w8mlQZ)^E;WOH6J
zG1BE%kSJUff~KI)-1&7&QI|U1qGe@YiqDG}qu4TCxLfQt-iv44dnDlbp~5jjoECOg
zn5@j^J3K$pmgeGm5W|y*#hw+ep(t!>kB4~E(moy=>H9XRy+~(=uVG_9c$$B{tkpwP
zU2E~_yEHQ`KX>6M_U>4Bs8mx+Wl3<v>Z<y^{})U-bF@3pU_CL5j4*+73C-RKD-2%N
z!}i|m>=6I)gJ~9roG&WJ^KPZVJVS&y&I*siEoy4sa&zg$sNJ!-HEmjCUa7U8WKr5y
zI=4&C&58HP%+l(>w9`OKSD)7z=GqcfPu;WAaRCQT%_l-R|KKx8@GXBHQ5`&FWiC9g
z3>VKR-QCMoaGQ8vT=uzP=<Bb#;C@xIoincdnl9S;LGJQAver_zq_kY?KdwRlKMLqm
zM@I<qk|v!vI^}Y$+2)0Evm*5vB^{G%)31%LKZX9b^{6BIDyWZ>!WgNl4L+$Hi_v6k
zznN|_lATxA=kL$yrks?9JXsuDEU0L7Q<`v)Mm@m4I}qq^F}O2IvhUEQ<z`AwpPxOm
z|9r6IK*uS_^rK?V)jH?IT4I(r?DxGPsYM=0N?F@}&T`N|-lITDp?Hr<=D!<#2m64*
zi#+6JOU*8g`20umLIUOPI{jlFLJS%DH6}3)XUES>2YibPeOGNzLoNyJCZr6CjBzd%
zD-)u6n<!hMd!ev#1s7T;OEEbW--h>N7|R8B)||DeJ3C^$VWADo8bgj(CS+1IrJ++@
z+0alu+{f!oT~R;>J~C!h{d{nJBqQT*S=}tH%>4U#aJcqb$q{e#$3r~R^|Lqi<hO|c
zT)o$$kEu6h?#;<j*K?n$rUH6lA%LQ*x}R=VpW2<|zxEGf$v}laayOQ9aupk+$mN{(
zQ!jHNDqoV>KGzS$%p|^@4tngS)M+aW)3MtxyQdswXE7M|=NCa2otaGLzzajM`#>uf
zm3>apSWymq+$1_Y-4OIWUEW@_SuR)h_eA%4_TxEmTA{bRx_81kf8luRCSJt+>CO}f
zy;K|S(EKTqq0IhK!YF^YSi`wucI}s>st6q-SLx&&Y|H<|Cr^!s!@;(xJKuNm^hAL$
zD(aE-487#}33@xi(!#=z=u;Q#xuR>r9dXY#Zz-s88<Mb}Xz1SYr24X!a=J=vX_ZO;
zpLgNF`&(q)r;5jJ2#)^x{lWdgyx@y$TM05^P2U?l=zZ{p$ogxgnBpptDjT}{-cYJ_
z4h+}U{0ihd3EB6U(ZXPDH;zpD%lMaG&E&iqaNgS*iM_6vtq6W_3ZWA-F*0e(w$%F0
z!VG&8XPkFoQ^CQX4krvlJhGmc{#zK(g@WEsQu+A<p>C-2u3UlQ_Th;C)6U_S_d*|*
zZ1WsJN_<Lp{Xwod@?=)fkrf*Ddp82hN36L>tjNvwBuU#LJ=}(^H4_u@(`IHu9wpT1
zJxJyUPYXKTy`Yr&y`$S`dTgWk{8xsL+kcOJ^_=W4_;njCYn~eO|K*2mxW@BNUaMZw
zK*ahQ>#E>|hU<DFx{5+tn!rp3Cna2KKx~ik!j-!9i=m|;mjBBA9NEp)?>9N4yc|Q4
zNH{PA%juw{dLc_fch?eu3_|DmS9BF;!bD-c35QAx2>pw`!e2$Rx^6~=s%`A;NGHr2
zO0!^gx{?t5ToUMzJ($a($2ge&MhYFpL)E9_P4o~F^;H+2Iv<tiu&obnA$r!OUkp#n
z;-Mm#H#vKP+P$n{DLnJqti63!WVrfNFcuJXTOrXcGD;zCZWZb77)eA1KhYwgvs0DJ
z_0AN|Zz=ERZKX1G(<1VH7vqZlN2h$Pk4G12JNLY)UhBVVsEc{Lu*xr4DX*4gpYx(Q
zV5IbI`qz8Vc7h!Kils-j$ShBUU#oQvk(K=p73k%fB6$GZpK5&`lf*xjz;__(aIKME
zTjv&n(m8BF5xIQ1#%YcQw(Sj8uKB*xzdcol59<)0Rlvbzw&>jN<l}4T;K{(on&n!%
z$48wfy*AJg-AH`6cVa;ERLrSzGA|2j;zIoK*3cZ{`%gqWX|2hX;S$^B0ca^Hnbv;@
zG|`*S#jlqnx>o|qIXDeQRtoQgzKp3|yixt}^$K6>TKPvO`wd;l&)TOPv7i~7xBQg*
zr~SeM4VmOd$dsgJMPC`Im(NCHu-ch+ul&1#gh}ws8KGzVpRsIGK~FE#zh;+KzeRcP
zgotH8hSSm@$Zab?o8RIHk29rW)n_p{^)B7$wTa=6=Hxmd#z8l72^s<*zsR)Izf1#L
zIH6o<)KnIqZTE>xAUaIs92I}Sw0#*iikNF1+#On2aHcZTTN%xzZP!MyJk(t9*o-!G
z<xw;R*XL}A60r>r-Xj)s(a}5WrETYJ1Q(&J;rde*ZWSQsepg#kPib9Em6Y2FzTR7T
zC^=FUk1vzZ{O#%c<V0rf@URiblexEXdq|h4OqafE_iboN>v35VL1RtLR{W<hoFvsI
z^vTJ!9^N_E;uq?5vIh64tL+!s)npWMWgnWfvp~KZmVopXlg*Y$NI1=IUduF}5Q1s{
zWn{$0!5~|xcakL!_4W!ITnvfJdySI=rIT*Z^+X6O6!t962s;Yv$;kZ=an1;eepDXb
znK(FqT&6B1D|^GLc9#ewbliXT_r`BGto!hLk9FzPw>6YcW~A3pc)%7LlJ%9qWF#M#
zi>^igpJ+%X8@sYeKHBbJZd#*ePWm=J7iGKjm6d9Cl}w#BR5giWZkCnl!j{UlL<;Li
z+_Z(C65@F(SAB!36OHlA*IO43siqfRO??-gcnr6vJ4&w;qM`WC*Aw$?ZG7KU|3!%1
zI0o|a*#11BF&&e?RiW9KjUxovYVz;i>G*sX-(?wJQ`{a;iYneWJ_Bobk*{%hk)`N&
zdS*}&Y_dpmb^E=r;8jw|17^K-h(+-sa1ly1*pb-W64#+s@|n8zwOQ^Bef{AsbB%+U
zCV@`~Go(ha0eG|dC=jDUJ%)X>+owxP6WWl{fP?s1a{(H4Z26G8*C0Fw=UvxWKs@Tp
zRcc#C9e>ftAhqwpwrw5+;E*oNzR;0DcRkAH-35ONeUGbwLGTF|Y;WCywMYhpsqx80
z<eRt1i(3a@nJztkVzcL)qd4`o!1C;>|J>oJXbh(-gJAvk>$I+-qWh8JKbX$X^<No#
zgr@ciC50lJ5AR2a&&|)r#Kqq0J-&A13|GAUhpdX_RZ~O5y$~a6K6BcP3=%T3*Q!Dz
zk7uTyaXG&0>Xr}}XSjbKv!Sn#Ku>S>3R^z!;V*yI$H*C9JhzqmPlcP8W(00tn{+0)
zHhG*T$+B3v5w$(sJIi4AX`tjGr@>i3tE0u&$F;4z6bQkOD@@9^GcBEDdVGlqj^|rB
z9TXhXp;PZ3K&rtyY&8ePRxA3p4cc(=SlG9A-B7x<b#NZDd$afRcEjDIL&IyfyruK#
z8a_aK!?U)MRbRjSqB*a?BDO-i^#*7U`TX!vI&g4p`|Yw~H2PK(vU;aRsn(ixgw0o&
zR1n+gU-M*otN9ONpP1ZZ4I=f<DROTshaDsW+WW+LcofeIWQ_x%#xjSzy|Uf-$E0F(
zWQns&aKZ?&cs2QBuw!E|(xDn_X+j3zfmAe0YiFZp1&WE>0=kpI0XqPRmg@KKFO%n&
zAu3;*k)u1qr{N=z3<yx&T0mAFxXdb@3D^XnTj*(-;y?DuuRdVjf+a8_NBuWzXeH~p
z>w!sm;e)i4!MV4@xActAxX8f7NJ6Q^1d^UqeoEK)yP+rSXETP-s6$+m^t!QQi30cf
z?AP7+^trhdSw@rBRzCI7#8Af`G%MSwm$MVc#-P+XJ0EmIL=hLRU@9@V_v6>UDnb(_
z0CmDfb>MIMoAR&m^ToY#6eYc(7aK5yXRrBEik_2xG4!Y7#?+^@WL6geHDOa}JeO{r
zIc1wiIU|*-F>%wTR-zkU&g{%4avKG|!KRU&9VXW8`PD2QJDztT8oe|Ird+>v-q?J6
zYsC)xkv(1@lq;QQ8gS$hBvY^_1-96%!wJoX26EbRo<*s;dXh5#*TabY-NyiZTYo%H
zV+B5+yHVvqa?>4r+n0<pgEoVNlw5E-4~1gm<a!Og)9*q5LTRF2%`hpm#wRUGi7Uu3
z6SizAE!IK5q}Bd?f*Snyd5H+63@)YXRBI808O`AJfR*0Uz1&VwlilW~s8}5gV<c#~
zy{#kSo7eM4J3H16Z;am{|IsezN`N*7)J!T9K2$uGFSZR~VtfO;o4tLW1Pu-0yO?^(
zGUfevr}(V0rqPY?JJb5I!i_Vg-1Wn+B2GaZZhYe8s@VV4KTdA>UMiH$Q@9!WV6rsw
zZ8VDlWBoDJE`jem7G{mjj?Gj_s-6w1+6&4D4z{)fpKjDdhL5GV4?I=0yp18XO4Zr9
z4~(s~!@&3sx2?_6%2I)YpaC|mFV+jy*w`&(ZNkA$5%slYYf(B(jP?crb?S%f@9t@Z
z+#^9=DJiCpFx>m_btSK$>T7{PzM0l%DlF8Dx03MqhhoMTvLOueLdG9&T<s$u;Jw?H
zE@`-ae8n&-=0f^}<iarz{bB%n{c*gJWsB}YTjEqcF@>hm7W(jx2KtdM{!4Ys!pc#W
zPU|mr&)fBmOxJixGc|_Gj0?_}`ko=nLu%Yexk-k<kWi^*tF)BsDKn~eAJal?L=R=2
zqKK^asfGR7Ej(uA?2c*@g%f_#Zd;5UoH&2l<iE(P(s?e*v`?j|t($t;Zb=a<=kG<!
ztuL{PN7^uGy-ju%x$o0A$WfD&wA8OaJzAy=zMk4{A~R<&$LG&#E=t{YI;9{E`SW4n
zYz<_Uilbk&dG?_yN-Nb{z8NbxS1D)RmQT(MWa8)^Bj=U~d%wS=6}g2`*i&dL%6S!c
zuvllG>F_|T%r_vrhtJAt{GJ%u<l>LqfWJqx3HVYw%6)BHHOF;3+7Rc7C?MB3eHOo7
zr!~v_g-LE*JgoDfol*&%y7e)_?95V7zMX(obvmCj`k8&iYwqa?PE*gV^56&>{_P3I
zdBQ3d#~@C)=Z)lWLQ-v+eZ1R>gEu|ZSs&hcicnhTjc4Z6e>!-^(qA*pb5>#&th_Fm
z9mU2*HYiv?CXYVs#KS6e`o%T;nw<jsjLfC?QPIo&@s^4e6YgCWIATcf?s7rS#RB=d
z{+zKSM2*MEX^@Jj0>eRbM^vjfvy97tqSvimSyPs0cO^$M@Z^_SY?yY<inR3S%gMVv
zQbL4d=|xc+Pg3Y_DX%<dob2=oUPhu^nZ%se0BbcQvl5_>|9om7xA`$9U8|ZfdGjp`
z9&YcIw}s)!x|%(E_=x?+$Gb($5%wn)lZUT$IWsOXwsafz8+G&L#JtDdZR>8US}O78
zUHHzwXy3k1DLWM`Fev9Ebjzwj8vYer?z>Q?@bXm((aEWax6ixZ`QJLL%z7UgED%Rj
z2i`Y1>zb%CglS=Uvz)`0#*yyO9e3^Tk&!40ELOk!j0We(&@uKPKHq+5b-mF`XSQ{5
z(y&aVQg$8N)r~k?TI%DyvR{Kpwp4%Sk>Q4WS>je|(y4C#7fK%G)tACPC9?kVd!|2)
zbWAq0be4rJlg>EoK2JJL3y=~Y_~54$m^g^BGb8f`oPmCQ^~gwuJ-fg_M49g>>X&|z
zb=Gz4KTbA?UqRA-Ws7^6B9C-d5+%1tcMzqX92A*B7|`Xwl7<gki-mb4gaL_{N*gQB
z-{iBnM?Fv?7BG~N!DKRQ2~&h3%)G={i0Y?P-`QJnK6)F>p|>?0S7n)&SumCzf`MpJ
zkmZhleNNeF?HKjO_%>yJG=0GE%c1L7d@R8h83?L>a$xFgLnp5Kb;enpX(IaH^2nMW
z9<H*or-G<0MUr4bX?1$6cCC?DLH)<-j7s#3rWB-3d$-m#iQ3&`lHe!f&h6=~Hma_2
zV$YY57#6|S9+2M^658EMULS1Y#viWPASgASaHge_9Gp~$eS{-KJO~6I<*3>3NQ57y
zF|LG-bM-UtBiR)WzYjh2w_`tvij3#VDAO5l9SLWoQ}!i=#}<>mClh_aSfD40UaGJd
znUVT1`jvkV8N<V3<?9Qkil)&b&@Sw=<;2Agr!hELU#WTuC%I&XCXum=%LI~UTrl#K
zHk$u_d%CbE<i??#2<GB^G4?|AZ||+yKVJ{5qn#QsUp*Ku82*SN|Fd&;%ImtFzDjPn
zuk^o){~4yAuX*<5+NkO8$XbpR38QGNMob2zvY9HEF-hgVT_wj_yZ2Em{T8n~Z@aCV
zM`z~qa*ZW=Y2*I&f27V$LxFcnRXJO<)S4*s_xAmJE-uD&AcNzK?*9HTYNC2+M=5W2
zzJQ->qE5<tL9Fw~Q0`k3Q`N#GXPuz}KZPN8`r=}aql_H&A}4<wvGTIOL5nr@JVd6S
zgcAoZs(nq_RQXHP27k)PnEOyBfBkY*YjPjE!+1f~I!R}54ceqYo!2Vi^iuQq6UPvm
zrNhCM$c@~eN0r^@71G|8<#gi@dltK5k`m*y%x=3qXX3KzHITzCb3ZY5<M`u@8)%XG
zefkw?*og&~mb#$kt!lfTWoznPq_FWKA=!<TMK}vQjonb~^j}FIjOiB)j&dCyEQu(K
z@`6zpMg{x_8uxcDl%MajR&y+7QTkoI-k(Nol@}FIHu8;ovhRz|ey&V^i+IH>7s@Y)
z(V^z9$sYrYh&QeIepj&vG_Ke6j1x``N~-?*9FF|h+7X*bIF&Mz_Tj2qshiL3Bg2dY
z%Y7#c?v2I7s>PLguru4+$9ea@e$C}ESu1|ML*%yC6l_YSEt5RS@p8ug*vV>a>m!33
z-|A(tKI-#?j8Eg^%YOf!#P7lT&Wo6q!`E*hx?3<Ud-F}_c(Il>m0ND)F@BU2$3|h+
zBh=zogDRJ;a)i;Ynscr37$p?E&KXz9<+z>(XXQm7Y2|#Mp1wYL7Tg(cqF+748yB0i
zPsCEN?D+2kWWRmfMkYR(kNPwu<)T2XlkgGzBTDFm*i>6kb6C^amnBC>#?#$hyt}J+
zc|d41#$7ecPVV&RXtTH8_|<+pv$W1mUl8SR9FmO4m25g!<y)A0FCI#CTl!5ENhF_c
zrroBrHY23_yFxIYY6&;`KhkpEYF0nqBR}Gpxl(&Ja=}7+Bu!1DJep*fgHim}Wvwh9
ztJ?+6gpX~?SPm@PmBso}qN0Bok`N|k>DfiZc@OtzPTU9EmI$5M&1Jmz?{h38cd#e1
zZ7Mb4uq)-(?1%mHyc<R3&AWpZ)^3r`e`2_3Hso7;Z~buKC4{rP?V~vqvII35XUpaL
zQJ6bJL@y*{Vl2xU<J^&!*t8f{f4@(PL_+e_-{12(@=9F6qf&m`6C&Q}b6v0My!wXk
z7zZt@t3cXv%tGwqTkg7o)a8m!<h+h0#xi{IZ7@6Xgw?ZsUb*=^!N013Z>7w080%%2
zNtF}Jfc3hq>zUy3zRuXuh4L}2Rn+yT6GDv|&nhb`K|Z>aSHv4@VNqw3ix?g<P{oeP
z+6^?Mm4u5m4bCtH;+(|82JSF??&Bf5oE%F%`|VnpNnb_ZPU`Cf)SMF%_@lkBVTE>?
zF1o$h$Khodwxd>q+JVD`FvZ^w6jQPDgFH<m5X|IrN<2|cy2>|LSXeg6i*5awj`T2a
z*w?Pfy}gY4DM(ItJ>^aw^}y}S9MCYsx^(IL!lFv*{g2f5XvFAkb{HuaV`e)j*yQA}
z=k^uTuoI<aoNm1;Z#SGp_#)~)WzXS%9%Giiz*IS5*=^g{-7NzmVMO#^k>mB0p@bwP
z3`(!X=p5_elb6cVW*qC^!*7Sj$8Fug!NVxEO&)>s#rgUBtQ#Fhl_4QFM$CF$URF=A
z)ML93VNFvW5NCUVNbt#Nt;2pLLjhmM6p_h8bFM?BH}xl}atJkOKc}@a#lAQ?{Ao@4
zK6QauROK-~V~YX1VyU!G=fj`p0tK7p+wT`WgP*Ld1iTL%#?n#wv#T7REOptR<A9^-
zAz$|uZQ?s(@8!0-2{!TE7Y|O5y~exl?(Su|tkVsIDeJT6HYwPe;5Gpnh4H)0=YAU%
z5nJ+!OoB82k<n6RVO7=tx20cK^z^4D%ht&XLa(0IF$ZPGB$Z)Wu`rQ8)w#JJ|Mk2n
zcxlsEx!-MP*F$LdPn*axj21OF_aLAzw@<hXJT6XMuiSc{8}t{?VR-O<@FLE>=f<m<
z`Q!?mW6Xm$QT3`9EAfp*6OxJ(s8cE#_GjqyJTIi7;s4`8vY+=;)#$*v|4Q;LH??@^
zwaxnb+9`E@+pi?YBYkDb4`l5<6=Z@*^W%^Aa#BnGT~eF&Ws9XjPbqKcm#?fKH7Q$#
zH3h80pZYz51^XKWIfg&j7gZIT=rzCdW}KSuLp(bkxK;0a2%Ibtukl*Tb2astoj;L7
z>P%t=@&A2P3~lgc*{^rK-g2JD(sM435jtCTX}Xm-)Ypwv*LQ5GpjK0IZf(d1f3Yxq
zNc(@*^#yG06tbT<LbLF^a;LMUuKh&Jn&sN{tAZ&_^fHWqJ_E_P6gTV<3(h}3w3ms-
z#NX?xQj-4rZf*2%i9EBfzP!Jj+F0wXO8Kvl00!y?R$=fqxntVkO;g<RvHX);^!kM=
zGVY$9%B`N>nj<4Wy4%BPum%fR>pd|1Z#9a0XURx}J^$MD*SHg<=Zy@-!va$79C~}J
zC0tcz?C54yaaWFd4idDTEtP(5u{(2Om6|MK^0nCv{{0Q6SgpbBf;7VJjg88pm>35P
zld;i7Gnf|9^=+6|;jS0oBZnH44H~Bhv%$R9zupb9$zn<NE5gwhAm%r}xM+l;P|Z$B
zc?$YYVd*I;cSg)la66*mKP1B71^;Kg&Ckz^g@IJNK{>3zOH6QecmdPn$ExV@xI*-!
zy*+$vEsDnb-8<EU%gUhdr}Xrx$I0oT<63)yG?E1;v(gM!bw`brTb1)(@Gxzue@b}c
z=FL>fC-%l5ma8oKjF3$#9N#nqKExz0OfhdVq>=1JH<mWJYrIKZ5<%-joU_e~lk86k
zHb`YDHc0OfB3hqE%*)mfdq83~j~XLe<h|M-IF40dy>c<B|Eu?PTCozNZe@^odLVsC
zxWP1@%c2idhN#8J1Y9eUoR{zW-@?qseUD3?Sz5|*cz9?6;xWptn6lS9K%9J|c0UP_
z!(#Gq%Cood{)Xr`B-r7|#izywF_|Xnlov2e-wq5YCi21QkORdA5`Ulis>dLR1OHhm
ztQQXhwfby2sW{byt4}RL5)wXnd0haXfk(*;WA&8z&D{5!lmf1*VFRTm3NM6(;bk7I
zsi0~3?-y$`0dDyEMMh-a3Wb63`3DG9|7jt&ig;r9BjUbkf2I^J?FHE}3LYC_v{=#W
z*B^r@ctFOz>ZRM!pWd=z?Y!T=jq_$PvoS$phgu}yabXenlxs;#5T!<^FU9~-rU@e(
zo7|N9aXu&wWh5j}fJS(pM!6+gB^_jVKLuTF5S87$u&IV*n*5lzNwmemn9_^O6adlS
z&RF)WpSw9ZIex*xDj<$a^g#6vf=r5I)=g_}5c`!{PE^H(V`MYKY~OZSWy?rU|EF4j
zlT!)aakSmeih^-9?MZn?M0;sk;z032!{UU)lv~xzd-6{~m?=cDf=Jy$S0WRFAp9qr
zO;yM*kSqG<pY*SQ>p`j<)V_;(?|G0BfV^>G-BE7GSTH=d?=AWt4{=ayqh*)FF+caw
zC%b-(i1-I4<hL(<k}OEp&24NH_GihM^?rN-cRdVJW`*sn^_n2LR?N%d!5hTsh<U&l
z9~k`{;~cg6lMUpF;k;Rp_GV;3Sfm>p8v$TpVq%(3)>NMT$r7~w!-63u|NOZZNWrcY
zMo(ClbK!W9mFTw-i1p32g%5(36*{UtgahJrbU>_I4rGUE#e98zm0K=NQyX)_40k;O
zxot6{p}x5f<QVg8QVQ3J9)!@;;a)np(a!+_Xf$HX8ebluB)gF$ep<1SkrB?Q>mC~)
z(dKawXcDJh^xB(dY1FK+>@lv#C<_md8t6ZVhK8c=aL^iK;6~HK0Z;}P>jNEk3Ci$@
z2pX|vP<RiVzc5{WV0jlKx!io@k=yC+u(*_z5(MReIyV<B+fGh+(OaCR&%C!i$Rc86
zbHHK|G5Iu+z*4hemd<2B9k383W8-XIhxJc)h2BRvmEvJcZO=#Rh=*C$?kZ2RV+a_2
zeRoY<T)ZzHbg#3L-PMbVine!l!kcD#hINqqe2sFn;(AgB8Ab{(DP7iwE7NK22LRFb
zM$HT+CyVq%`HxpiK4o1w2yr{!TT1o0jmRr1Mu-HGWtjA5M)IyN#9Iv=Y|WZ+QFy2~
z&%7}n$d(IB4;DOA1;O@`$2UFV?6>E{I+OS{KuplBXhld!2z9(xZU*)ScTvJXiJ|lL
zW(|g`^2?VWZaYq6qBbVAaLG7AbV)#Q1uX{+g72U;D7XZ|c*9d?d)#?Q+J)@9;2Ofx
zZ$Wx9T55uJ9hTK=T;M40gXk(+v@la9xd7`rQ8wt}EsvJvHTz<!8cbB7!qP!zPJaD@
z03%p`{n?S3?~1LBO+;AO$JJGHFpViMeGFGz(3DD6DzzLMc`hV0SdSo}tSG)eh18=p
z>jMIC5b4rO2OmDBnU7nRJMXRLRBzOc*=oqjh7?q<^UlxB{eFut&;ugYyIoH<R0#<^
znt#pC(kAkJr1IEM`(hr8zBTX!wc9_KOh<|{z;}ajMTCT;AZI(t>ecr7h8;1fmvPBd
zDT$v_wu%j5r2Hcl&$atc!xM+~Qn0N!gy8n=^-4aw-#zlJI3y3BhLs}c?mUTIWNla}
z_F@te3eL_|tM*UmWiPnOYHKIK^;pE>cu~2dUo7%GJTSoF_4LS2IL-!EtrS$FH3>&b
zj5HF6JXlkvu{<9AlPJ9O&G!}&J*KQe*BO~X(|cSA%3EY)%axjT&$%$_3`US*gRpe4
zwa;~Q({J}SrdY*H`_g^yQBzA$_Lp1fh#7SqH|7LmC~{+KV=y#<xzqX<Dq;%IJbWlC
z6~_@ulL+@Rz=r3O_R|rgu6f?Z1YAH^y643!>Is)U2>t4yQ&A*V&(G4fdV(|Tnuir2
zB@cOheSKrNm9X?L7RVU9@TnA|x=cxS!(lBLuUHrenbRVOv=PQZ9NM5ie@0(Q4&(!2
z>8YmtKA`|tXK?n+5o|IneW5cxTI15ir%Ld^)8UWU**RVI%-6=sWx%MCR(7_xOYPUx
zPY=71C>kyI!+F#3iXtU8Ow|;fdUx#W*ZYy>!8}ffb<O#O1#w>6(ykwoOq{k0l46YZ
z7hrs5ce({ZIg*6!>n|FNPXvU78AQzLaHc!>-L;VtQI=?lzS!<tXg6=z8At0ojE%wB
z^iU1wvIjmdYB>VbJyq6YzF3ScdlPF<GyR0W+WiiplS$Gv=XqZ;VckmU`U<(|Dhqn{
z5IC5`&<_92o6pU@7^?FFISS63jaQ%>Qf(o|{q)FWd#;V5@p{Ur96D-{l9Hm0i_83H
zO0CO2Bd8jKtCa^yJ&som7+Urt1~sG?T~z^m5V2~1+NeL*zJh;SQCRpLh{0#-HoebL
zD=}1Vz4*%0`??UZ=|V^B@>qG{ZMQ#id9xUwWBQi{z=@RF%!@;8`9AZ80R%+>LJYu~
zEgZ_%(4f5O5e$R<UmzumUTedA1`y~GqSym7uU|d}0H(}hj4N;Upr$jP+r^#A#KZ&w
zKfJ~C5-dmzhT$N}CusRU1{M}+Tyic&cx(uZ17LCde|m(_G34y@*dQ)*P|hsDBn}cn
z_$ExEW@2hr5SNbQ$ieP+s-K(F@2(S*mKIn%C_$*|4qPvR1gVGBb~!IxCp)niD2~_P
zt}<&?m435;0sHa&dk;)DdLh3I1FOmEO%olq>)=hzzGGjhNkpHc<%d1?u^s(b;(r4H
zEyTra(cs&om9|-fa_GNf%>R9E+M4sT<0KpIJ9nO<W8`~$ssfnm_NggE2ckL1es0b*
z!TO8`TR}rTe}%saA{&3f2W`z(3wctzn3&k|&mz5|G2A9uRn?fUU%#Ru&V*gR6es0P
z^p);N7hlG}aL)qcKi=PnrU6e{D_p5T3UP=EIVrNNoF{7j_Kg9sp~cvj!PEVz!f)Tc
zAvWa`HBKdq?=)b%!_pI<IGBRjRQ>6bAS5Hp)vk43S}UJYmX_|~DbY7HGzFE)I6kN1
zf|_md7cXAihtGoLHuUb=EpTAdsFgb9-7_f(iT-dIfgG53G!BJN9FkzIO;-TOr*gZW
zBN9AaD77dfBcu2(s6-+V0|}|9Hq@*Rz#W0+^zipkYW^`a470D+f<?KnISi3drtV`h
zV2JGmQpRKS@%A30LDV2V{FIGr3^_j|Cm~Ta*kMsGeMv1E6wx%pVDAPJs$t>byOS;(
zCf`Hu{<}=x_^x#e%tVc?4Ryq_qq_O`{~<3y+)thJb?UWc1W!wcKB*~uy2GdO?3pKk
zF4<*^YKM)X^@(b<d)QqaQJ8Yw3H;jB#Izk^rl_>8)6Y@nxb+(MHebch7#LBLu`d?R
zhx5^8*0UmIpp4tW2kxY5K1!q1aqHLXVOG%Iy-v)cI9#ArX4pvvq7m>-NQ4j(k*h>x
zd;oksw#H&VQ7J91GSZ5kPRENu=oQoa)0_*n)b}E+)OpvG^6{ILKHDw`rb6#=?#l_B
zO#-4u?GCC$Gi<wW0BU2FTTfd1v(%3cdVT%<OcCYZk9Joeh@Gn=G*(I*Q}}`>pM}|+
zx!@RhST4~Fh=G4hW)3XyNCt=QFHSp#xTKGSot&KJ4Rf-82!K6dG-v?o#PA?ULe8AJ
z=auu-m$q7|hBKw&yVUMP(SgM9Ejq~%8c?TvO67il#?m10dmoZ3(6yzdr8Q~rzzj==
z@i09*Ih^m}G5GNh8r8#4qut%OB~lJ306k>gVFZAS(Q?JHwm5JP=iq<sHm9F~mvs)h
z(fQ*?A3O+ZP)S9MS3#IXr|!gy%0Vh8==69%LFPUmpXS!qR?_K;9d8+kW;VC9h=-xk
zH(FeFU=6K)4B$%4NV?BqZzBg%D#6n~;YWX#(&pQdXfwd5v|s8Ee))s*a6daV>P88k
zpOi@UKiFe8_z_|1g68Z!h980}65rZv!jBIVKY@i4Gusl#E_n2tnw^<>Enz2sN^mkl
zUqo0qhF&J=fo($j{G%&)6vpC@#vqsYwU?fr{+b8ey7Dp~P%wtD7WiiIC?3B#L-3Cr
zoSgU9EyN#L)g3O#x2}$rciRdAs>|;=O%QM!Wp4#I;n+B)(t1`lwq__IB9gO1A>jIx
zef{isgB2y&>&|3VU!lpu&cf2gUQp^*1A58bYftP~8Fq@4TJ_MDRD&goo<aHA!-aS>
z5Nz<g=&zNWlvKcDyMP#jrYSmp|N8X{aFF~m1<Zo&LPsb~?a^|cL4gZXq8iz+Sfiv+
zJHv^e+tLyC!&(lJPTS=lsYU!<cRINZx)g55cV=Fpa+M7$wOi`L?v$IzH9MD%ee{A@
z$JIKg9G%8EY)pow1ALCC;HRf=wN+gI-J1rOJnPX8(viSZ(t#7H6#K8y-vZEpA)AR#
z1=^#S6+khUfPmo7a$ce7@19g_T-@CNwKCJl*Lr#q%+LD!`}tRDCm~mX|4@Q>ut*OW
z!_(8Vcx(;6{rvrR%w6E|9@z$xbFU{<`kRlI=5$zte;!1q8|bL^0`Vup#>3mwFB)5`
znhknFlc-i$mXeY(eq8z-wKg>seN;bOW~S18dOn#d<ii&YIkCn%A0TL;&h9%%zEE0V
zLyz#}x2a|Lk+OFk^tkKYPX`Kh1W$Jhkw}v`v}QLNdGAlU$b@k?ZoZ=AcUHW4^Jdao
z7fbiQDfpt(l#!G7GHtY`{Q4Krp3YY<lNT5F+XI)g97C+_*k~)fa`=Wf5uRz0zm86E
z_zV{xpVeUu0)<#u`#hud*hsVW&O&E<BBbM}-#yRBxy%_HG+!wGpPPG!c9jiS>^MSj
z<IejN4!h7VA<nB1;dZoyWHq;d>Z&1$NJfk6!Hf^F;`)%X^7>S9Os1ONzCkmZxqoUQ
zgcjpsG&nifN@^E8JAfecF)fWhy8iskB{tJ(QV>4ZU~&-x`H#*y`|2{6eM>QU$P2~C
zl}o=^0x}jjn7)VxSZZo&XtK6?A_)hwtH*_IuaL;)(fY`ngoN{zt_4%!G@xr<tNynF
zRp;des&|Uwy1J>0#b}?d0Vy_rwBX@fIAe7;Bp4OrW91fO3zeFsL#CmiEPkJfNg|8`
zfZ2r8Vj?>2f)J@+GU7*hs75JqMKJ??q23B>z3zFZorTCJ_T5SR^GV3+n)wSeTy(f)
z2>N@$9}unPD#szNy)K8fYVz&F&9HQU|2lcf`MF`o2|Tu$_Uq$n%NCGs#67nDxCj|S
zaem2g4zs#X83&-I0iYVR+tLT~s`ji5-hnu`%l<lKL;^=I>rb{Na%LfISg+lmv{~dm
zzGRMw22tcxg`h{7@;Q2x#ea{KWNEvsae@_yFD)$rgR}D~-!v{0z)3_*Om^ejD`o({
zlGXr2XumE{(ROOK<pD?(PA3H|*TvvX`G>1DLGu(P4(To6{$(e^JaiC-@cE19!oa-E
z5+s50fDIziIVe&E%UmMi_X;o-+nRMdT~8p}VMIIK+hbd^t(;~<4*-_!c}c4)M*%O<
zW?2rp?$Iu<OZ;*RPBlnb>raXoZ6Ky`T21nGZ?vu1L*&yUgD?e}-rDPYkN_vOL+Bp_
zJ|LI=Aw1@EGXpwsUjrNMp$_IlK2}sb=CmAFg_J&N{yH$d5Mc*FN1j=`=4VMKgtYv2
zmg-O6GyxP2gOiy)+$f)PmVkJx1a^hf1bO22{=NzTRkR>_D;1&%kd~2qDqSpGQC2o(
z%&J~XS64TQ2gV^CynuKZ$khKSL{9U*eX*~eu$@DG!&4Fx6K4VQ+^y&D=hp|xkCLit
zwpciwlKwuSf~qP*R=<<B!Pb+Fn%VnryLT1Rd0}4BsX5uhR~|IzZ~}zF)=3h4@<65(
zdA@4Vb8BnsWtc{qHbbVvFB}{kiU5WqXaH9#!hk%&X)3`_a@}q-b)&#Q<4^9t5HWHO
zctA7I|2CL(Urcl`J=~u6`|u$<>^PpoBy7@oCI69Cp-N#=SXkJ=C$$oJD`Ym1o3B~d
z#LLU86k^`#Rxt9(Zs@v_;^pL(MRTQEgsJ@@4zpTu`nPQoBQsc5Q7l@^JdgPJ_-Y2M
zji~NwryEmMXd(r<e~Z$NY)Bp~0*_JpCr>nh4O9XC3Z3twi9oOwUbmxCkkV8CVxA+N
z$ZG=0I~kwjXFy4Bx_AK4ni@cQ*6c@kv|oP?hyD(s--3;UGw^~~D;*F)He^oAQ}y*#
zcA%7|`Cj*TH*!`2;XI3!9Wngv&sx_5h;Sf7f;jLtzdA^c2LtF<S=a-U7zJ#)(4AaJ
z%5G>3Od#Wf2a1K-wH&SoZ=_-#KAV}D83i^GYvW0t;`~`YHk3`&tL$==KdGidP*7n~
z{v-=n4t*guy%ys?zk6YI>?GV^<>LA>;Dk=y`CX7bG=l)o6>AT~1FeJ+m?#FyTbGEE
z3j)5Zzr|>2n=L{4+bj6FKz9()KQ~w#uT%iY5KhAkc$0h5F%6P0_}XY9z=a>#*5DS%
z1l?;E>s@x24Ybf;wZR`UFo+I6n04wj&ktG!|3G+yEa^LkR4kI21Jna84NW$9F!g%(
zS{jp|pLP8L0!kmFBg=H6>RV#P#2^4-m^A29<l%2hEHFhfd|U659Dn)Rw4&h@U>VK`
zZ1m+_9;2NYM%0I(pdr8;1L<P7^3vcB7~1~mD1ae5{~nJr7xJgS!~P?Zlc@{<L!yA2
zdT(zpI;TSuJK$rO9zN_@&W87SsIW`+bL8WpGcoj6#hVAc7~z<h64AInFz-SN$4qo8
Y7(no3hN?k(9{mEb7t+E7&tG}{KUZs*i2wiq

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 010aabc527779ab819fffa8f5ef5ce75a207f9cf..168693cc5f14ccb63b924d47b104ea37278f830f 100644
GIT binary patch
delta 107539
zcmZ5oRX~<Y*F`ykNGaVYNJvUIh;&N`NF&nS%_AiZDgx3aCEeXE-QC^IOY;wS&Uf+O
zd7(JZ%$~K^T6+yiZHV1%hy`OjP(}4KjF;k{|8aDTdYI-l?E6G132SojAvQHVy@Jq_
zG%u_%94b}uuR1SO#r>;YnPJyY=Woj-o<~z}+~kgwK7nn!pVqSIqQ<hUVFhs2-k1k>
zZ|@WEdgH)nVBe&tT<s+g{e~q3V%jW`#i~HP(M#2xz0!;}GXrV-e16h8%aD)b9!Kr1
z>259V_f#(u=EH^E+Qi?N*YrM5-}*z}5%QnDef+}btnbKV&ToBR-A8j;CN-~bBaUi&
z(E_e!3YMEWXlG-o?o)tVz5d$b*;wd}&tBClBjiK4*X5S}dZ`vaRk;=4eU)-?dn~Vl
z{ym-~>^5<z@^~x?#rc*~Qji}qJ}A!`tDwM9b{KJ%y8Rjz+Y{)k$XQ@c5tx!Rz>y*w
zB&f>C{&ub-ZhKgDQK<JRCQd_J#L;;10gsmRt!ovkQ%aRozRhk(4O4N!i_<o18<udU
z*8}Dh4uT$N-&9UhEp>!387CkW@D-U7zZz-5cRhY_bHok9Z@$T;t$>Pd)p@IUy(QmY
z2<7Z{iq|ynun;o6W=)rg5{!7Kzrj^CoFO>G{tsfj-tG5fAMFd|cbP+O&&^3<{o-AD
z&c;R`O?vUZ?Y-)(G$;LrgWFv;(PMo|XQt%1%U4CPo8n~r3dxPzo_dTiv8}&5GEzw3
zmqb`pKMN~8H~TKm1D-X=`e<0KF5%?m!V4AW!xg#Tb?YX-2G$1($EeYE&!QWIq>S0N
z$>R?-U>^jvWVwl5Lz+4&){Ak#M+l$1x};KhasNRk(aP}%{h7{)-<udQ?9MKWsNP}8
z=0k+$9t_Qi9@(O1h_82Bn?$wF(rPn@O}P1Le>||<Y0cBRgvfd{;Yvjl5r`8FmZ@;9
zELjN51^J&v-L{Pnd%cjW`7pAAoBdoM%$J2^?P6tUd~B~FQ;uP<=2pFx-7(oJ%?7qM
z*1y+7yiOQhtb0y$Gkd6rt)h9|#(T+65JD1lt0r!;-NW2&NGbBPz?NaZ9u{%8|0Zb0
z`%msx?hoDaMo6|I?|B4?ruxl&DGxxI*1pgYS(rYZ@JX2RiVr6YlX1yEd_wPyATu>B
zpp(<pLVB=$Bw83U0}XMoroxs5vZib};Rr$ySb}1#Ge=iPw5>L}b#pXp@i=#$Qz|p0
z1!iyY1$$WTzR4ZXYxUyrfBfp!qa+N&uv|14?-oLHo-iiT;P%=ssZeNNHAu7CuWs*<
z4NAS$V;o2_j+M+&0+sW$>g&HZ;prR9z1|vDJ^V>8w^K%vxATn@)RSI_ZOh<anw{8c
zjGw&vB1+k<Rzj?3J9EmJYc%J^Ng^U|)W7hGz!55xE~JTwNnpJBQs|4hktIPi0bQF?
zy#L_xDxlEp+Wv(}2){d}rRZFLj`V@7jLU$@SvwZi^3fEGR7oL%ZR~7JAoYF5gO_5O
z`fC|G0o6OO{-3}0rM53}$PZrJu%2$VNDW0P38vj$S!&vUxA?L3Si?a6Yv)n9%+Etx
z=|wx?k$1&pV6%LS@ahR9@hl2yoGna-h0x|`nUVaV*$o`rb2tfMK_%5GE*4n3(6Ht7
z?@5!}TN~SAq$(?n2sNSDfy9j;LUZF@k<9}C=rjj%VSh-FMccdZ-R`y{fH@%{w~KPG
z81c^&2h$TB@Rj)bjQc;yD=-)`I{sN`g&cnw{DSYr56wjT<^8ZF=y7#Ltk-nT<<gTw
zTmOn8^u+h!<Lns}uLj19W}Kgs($9L=dCIfRdft_dB$;0h(tW`n+@j!JGoKKYFRzJM
z+U_jAlFf0)hvO4cM3K$`rzaGfh>T^i({I`#Oj*Z8N_aVFj{6Q9%}m`l#mL%X5Ye{!
zf_u$){J>Zi%8!+lZZSVRvK<&=$!wtMYU?vZgd@>R7~t|t@I*S9<m(*~>f*F@!-W^C
zSCC)f!w3$RiROpMA<OfClP-Z5%iW1Sap%SRG10`CN}n>?y;I<HOmuse`k;pWjJdbD
zqe_Mwcjw^*_uqLOFovkxEeE})s#t~|LvzjQ73+|{&rd7naI)9sC9Q?B@RN4L(&aaD
z@5%*>HyO(PgFF!7tWLKT^LP;`N5?D6DR#?9IT$V%CVs1n{7^9nOtIS9{BRu7U6?(V
zr%|cFduemh&upQ|r!F|cr>11W=G`WnKK3;rj<etylo&sa!YgF+mNhU%$8_&)Ez|}|
zRF8A!LO@=xjOZCysc`3iC@IXOM1ws`e6cu_%m{aWe*Q^c|D%C{$L)Sm3=I>*6F9}K
zs+C^-k2_DwuOHXDmrfGYe8A6IUj3xIF^m=Q+pW4gTDAJ&S1i)^&(ipcf~C9Hc@~{7
zLY+w9M6w?}A#Awc6{tkq`wp7{6etWDkwk>``Dz#D`->je5~+!(?`l*j{0OX1bmbCx
zVZ42Z7-4y0E119Uia|KnmZd*Y5s|}o;P$!ofMbWaR9n-Zy#1|PO@x88xuuE<w!B?t
zDSCd&U;&Raf4I4N2^ra>=K`b=Pgq?p#<@n*Y!e4~q4Ps3JiO>1rU}#2X6$r{%~?XR
z0bx}_9v%#L4QQx+dhm-fG2;F*t*uo~>g8;47`G{pAsqAfrmpxcEwr(Vgq7YjT1*zH
zW7Oad@lfh%(lVluSyz~zEe!v@RCza-6ODcE7W`XP)#~|CI*RVaFEq1r{RbTIeP3gs
zxX)^HKU&VH^AL#EV0!|~A%zKa@kbgVw=~Wc4wQ-Yd|pw0BTM^v+Twv>>CXkw-SupT
zT|}4HRxpS?f^to9nd_pk>**5mjx_2SVmr?gBF#RqBxdq(baZtI<{J9f6+U`&?);Lf
z#p+XEzBI{7W6dj}j$Y~f1|E025~DF>=yPUL^U9y_*oe1o9Uqiv*g|3)Ge7^aOI66q
zoco3C{2}Y*G?z-A1}s71nlhI*h4ucZOsHs`{(e(G*sl+~t{7Ys(LX1umL^~_n=a_h
zO^xsoYTjkf!vH!>bJd4r`$btBtr7IbUj;>LTc%nEg^p-?JR5jc{uAOa<O4vMsboUn
z5c9@6RBma!g6HCkK@v5%OPKf-iCgW;&#>+M&Fei{&lNAvY<jX3@>hF$dJIO7G%K-{
z#;0vf2E*HCw13GSJ-~~e@Rk(ukVRnP2*Y%9AugA_{!<z5>TE5piIc-667M2Id&D%%
z3Ey7xXH#Sk!+qy5J`|w)b8@{5chCG5m3q>RXluB=p&`>Nn=TfOPt@xAJ0pi&Uh8OW
zt}NG+!y+WEBPzI^D)j7*!~3rLJ0xcw`>u>L$uiK)SG!<s-^1f2vKNZ+uP4~lP@Hh3
zGAJ{L3E+ka+|6dD`dx-K+B}{Gi$xiV%lfaTs<>bhDM57+LD=*ZuOhSLoWc`gd?KW5
zMLldhNf>Vev&ZI}RvA7^`hKhAL*?f0Nis&<5U}U?lFsfkThlR2RXbopT9==#oV^mC
zp=M3HPK3}FdF-yo<DEgAuA6~`7tJ$$TUD2k{K`HC{nzD@DIOb)c{FeK(=V!{)3Z@F
z4{SRTme<hh6!{}Gz}GnI-OFUT4sDqqk2B$ZX6ozB={U3-G(0QD!y~}7I~E&UwA}hZ
zzCW+P98PXL=8ktNYk{qie18#7{%wxJ*cF&=zOBVn5BiT*1dRA>&<OaN5!>*7st#ON
zj+77_FjaDWaJ%UhA5_5joieT9Y^fjy*%Y%_@2~$Uh^8vc8(K_e%F`0UyGa&ki|iHl
zoBOyBT!<b5>g(aN<gwSxjyh+49u=ho2mO4ixp92>+(-X_F8M7gyyrMzN6KBhv~?=+
z28{-rLpEn2a?es9Ss5$z`id^P^V!ds4}-UbQ8a07MFnNH1}$ff{i|y>XMXKaXI~GU
zqT6<^V%*f}3zO0>NcoVK#QS@4qQv(vWGi{H&B@YgRPrq95!gBXKK7)+&h>+&uTPi6
z{ZS2v`q)yLN?n^3KBm5XUHDsQZmc}X(oB9+)%FqroChIIDM?0{89CX@%$7n>@|>qI
z7jVzqwdcHi^)?{%_&XZYw=DB>vf|rQ+{V#T?MUu2cK^qyG`wF?IO{H)Er@O_m8>6C
zemKjC5$>3f`TRu5gn3Ajrf_hO5Vk1Bdb0cW?fQpPi09`WhiMf-AenDiOg1{Qb-sxg
zsc?0AZ*xxwp1v0&sJScFrxDwL$;Anr;~F@eK>H*q%_=kRX}Eb0P+bt(wkIlrhkp&S
zmC_7dB=RF9?X34bDVLxozq}we8eF2l64Bj=TazRzH`_&~4t$$x(YG)^P}+Tj7c((v
zvF!$TqjB;fdWlq@FtYDKl5u_~bztO@sn&cyEL9>Dh0dWS#i$aqAfEVW(xwZ>A)T}J
z4k}-H%{=%W@cHWBbhW}42w%F-Ops3q%gHXP(5D29JJ}Lfz^4QeCq`p6X3bF+?@b9V
zf1+eE8MOPtdtiXAawk+C%U3RcWxGcvNr%mOO1PMl_=|pXe94a=SKfMuE>(J%Q5c5q
zrT6U{(iSuo1w5{mbuqzc!wF^GhFAsHc<B*jsMb5sr>oQsXPd?t&v8PDIW^auH_IU}
zmRN(Y-|$2`PT)4#EYfWA#6&cXi7hk^ny2&l<J%#zS_i#@4@{o@w(3ZH<=Cbwj6{4@
zAI(<0*cttfeV&r#ZBy~k<52Uq<gImWbw~t}Z#T)x5>IxIuNl5S(!?$*f~=R1-nXAG
z9I5P2T=Gzj)p^~v8CS4wh533TUbNwSN^AT2)aJ{$j`lmOV3}@b>x=Iu*Ohl|+9#84
z@L%xxgnNzl@Ll;zfz8St7aj}CUG|#uG%;7&nW`e9o|c2nE?8B)F3%1F;PRb3-)iSO
zqf8C=^^(haPY|@cvZ}K%@0q5E8NVi<H9UMmBH90qbfR3Ji?RN;wzrZ58#Zg3NAKq;
z>1#X0AAG(@se;b1a&hqfrLQ*GtE)kQkD!=CL#me8jJWM<k(Z?D;$9cqQ*o{n0qOHo
zeNwr72TS*p*`KK+554sYj`5NU+lOwjXD${}->gR+dZA;yW7gEfXiwi#cQz%O`@=l{
zmw?E3E~e}yjjrqI&J}1l=fv?n5cfyQ_Aj!-abH7#Q)cM-6W{e`B7C`n)<*-KdT<^U
zjaysyNOjiTVc=QgO797^e^?#-=ZK@*gm^&Zw7d=H*_;pnu@nzQc}Yw0q%`3Lr82|-
zB+cjKiF$$x8tlm4gkMxX-=AV?=@Hbx?)Y?$2oLC%cgp8wi^$iAZ-EgZ<g)gBnRA&G
zG^N{;RCl~(Qs%#VB`%LOIQ0EBYw4qi-x`%D(zy@Br6h9v2kjyw)i~=t1KwBK&Qxar
zI!rc(u|!5|Mj{jrH8~F|hX0A*lK+~Z?3DP0f|mz0jWjK6_uGlIk!eH2PanTH`qhnS
zez}L*(*K~ALLo~@RdwQ9Ox))bdg`%Agb8C`zw8gnM=CvgJ&Exo+PfOC-Ok2|+77nM
z>iVd7Wjhj)f$vXULMwE%0fG;#u7N5IE|r38<6@5Ardld(^H#aZ2vCy!lrluaB*d0f
zlf)=P>*X9V7SEQw2HN_SZQXAjU42BQe$yW$u&}taG^5h#ipBPjHxT%i0i0z$^R-}C
za?6<VUC)F%!Aeuid#w4dM(siwP9!9PDY!OiM(x$s<EUO;k-?!coIo_0%rh12MhJf8
z-?1DD-2$*yZur<t29i5gM0=*-9rx7b-=w8HXN=AEt6%z4C-R9>3@28(uOBO{H{6Kp
zid>p5{(*Q6j~p?qEsaiQ;dmj}OoJ<IY79P*I91uyXvqGxoLqv~HGzXwY^s#|JOs<7
z1^a;IffS(@5+AOf1$`_kW*)EgK&l?4r{rQvA7^T#Ts<Z{d8n5dOX0e7-v@q#v;Ji}
zG#|2i;VV7PSsi;VX;r*Aj<49^)Er5a<kL1j_5bWNQ#-+5h=M76jne#7YGzC!qCrW|
zfCn$6H_cZQq~Vx1^;k$x5im!I3R|=YwdUkK(Qq6HB2=OwB5-I+JBRBwn~^s~>5vY6
zdAe#%_7H|?w~T0!aOKMo?vdhAZSA8yjoR?l+LFH{Z+O`Wcm~VY-hb$b2o(W@(h(Et
z>6l44s1YGK_C9ma#Wf7YzTvQGs+M-M<tLEoIdhsR(t`k<*Tr>68CgwL-cnZ4C}8}M
z1u;35ZGH*JY87gq1E=Y=p$A%8J!BLsRqO#eQu65fw`eTEV|&t+Jd#gqq;obpQ#9#r
zMPFpyMYog^?`qQLzmiPb8ofng&-*C_G%@QI2ev!;5%ZdONTqB?v;Y%(X_9@^ruqIP
z_i*E6Z}hJ-YOl~M@uM5aa^uHZQ$pVZqtPnY06n#`y`aQCG+<o>`SjZ^uNf$0v_78{
zs4MU>IJYYrTB@wLWO;~A*J!vRY;kG+oQIDD+0+$~Hb_SXNPx5T6|ENG*){+Qr%QEB
zrLFTHVR8%|eG8KViJbM<dS$n#Yzr?0yRS@W3HYv=`TYLqK*md-9WU!N_IVn5Finl7
zuSI~@GOc+=U{AQ;BGi6yzCE9fzXZs1WA2FfWQgXwZT@<IBwr&iWEy%a{~<p2e6$y`
zQ0a)nw75}j|6Bg9zb@9xhW_}=3qRQtvZl5tWpv60HA71!%<ZOtX%(JgW0^llO?1vz
zqJ>`3P6Z7`!96Wl!c@?+jO5}6#+t=tu2`Bfi+%4TMtd#4D>k4dKMAQi`g(qyba0{k
ztXlJ=O%CbaUf*xwIF2nsb|<%OV%U8+e_oEuSCGD)gHq>Y_>$>cxwR=@pf2gf!Y13V
zmgWOimoKbaJiZI0XSxgHKa1NPi$5;>P=Kc0N1QfCD&gX0n1g~_3Ygpd;?Bbhf~pH`
zZBlPD5jp&Y^0ajZax%acBwttLuU}q_B9bf`m~9BJA<Ye~mC~={;}yFM@SG9PXsr_5
zivpj2ugmY27LGMNUGqNB!!@RP>aFA^6#8+Nt!;K_g|&ISs{L@A6QDu!QZtdQK}R}N
zzNzoA5lx;H&7I;;812nKd5T2|AT{(`TYl8vQe9NFm71y^!>&P)$r#z6pkUXy?HMtB
zU90s0od%nxG$u0sLcr?k%Sxivr3;*Fh-?341nN3rOoRyuup?01E-isEdggi*26&H3
z)*_8MuyO7z1Zcw8nn;z~D`$8S)=V#ZTuM%Kd5#Y*N`Chln!ES+@`ZjA<THm;VVfcr
zV?GlGdW0JF&9RQ!9z8qg1{W@}+d1@j_?N;M^}Qf?BJo$uraS@qQok)Kc3_NSd3!bb
zj=4FgZNuv+=vfr(#d=P277~EMHDH*8LcbLnW52($O3ddD^^G8YH6XYhHx>#GfONcK
z6l~VXyuvBnx`dVkcLII&VN;1EDW~7uz|PAyk&io@B2xp}KPp94@?WrS9Z?1<cL8fu
zk|^paVec-g)WAC7z*5?r{1*NXV-l8X7rTTrcsL*u0~G!sc<8$tUns<sZ$N?{RpEF9
z%;i4Kdm@H1g1a)PjePDZ4;|VT|J=V=E7{`T_GIFRt_9b7q{cbc#j}%1BP-inc1SYh
z$JRQ;Wj=0AGIkL$-hKXoHWjMYJqx25%&~Xm*R;l>Rj0BzziCT{5!!5)k3Ft0i7Gjp
zS~j2zC$RPYS{%z+1%2dIH{>9v0p>BP`u+-IhP9pZ{!QWgo`wd0(z3axiQH!&vY3ce
znS*DJs;z;q0Sr#yBFh;_Vp%t4+&Nddq-@QOGgu{$D($w}0_9ZBc<n;3r<5P|yR;A~
znS{vNml2)5Am9@U-By$G(6T7h3*EhrubC!_IgJU<DuuFTkA_WI5MrxxE8O`$tE=>F
z{zy&P*dW{=_+D#Axc?GAK=-OPQ@|DXlx)2t?2hzM@g?tDRjE^P)h}U|xsCr|iJcXC
zB%N@AnVRT+n`}!oUiRcOd+KV0%gSzjDakn!aQuXLlv9mLxmWJ-@+|+m(CB@+{C>9t
z;SirY%sgVxiR-HC5qJg=AK>D<ZB;!?TTz*qUjKu=;mnBY+r)sZM_ufX4yKA=6O|TB
zh2^psu3Mi1mSlQzofl>-*ZH=i{Z@N>0AZRFJVx!YIw4814G~>E<#auLaW$xw7Y9q$
z_wx{Gi{nzRkh8o-liTD4*XCp@3?gTugS*w^IWQWPg=t`S^mq6;@?UzR@rmtU!bpKc
zrXCmzjpVGej}nx2TSDICusG?T27+A!l~e01IQDd)QdE(UgNV24sqGRc$HQym>6Fyo
z%KNXoG5wFnwbGdrIi1()^WIz`#&MMeA$uVDhar*ku^~x80KPRhhtjunChhQp1ETfq
zqb}c>l*ULX)-89ctR2EuNx7;nRe-eW?EUj5S3wu>A@yb)6uhG+niu?`YMNDy-$caz
z36F{m0rb@97^UldYN{AV6XOmAFK59xf~D#{_OJ`W_z11G;%K2f2V9dkd{;z9N|q8$
z6}np>1LWRrb3QG9<?et%LL{!>hUJ3rWqiE4za^W6*d??fSGN;!{966fI$O!)L(rxN
zoPAIDLP=))E3M-S=4t<BH`MU|2}Mj`4qXVGVqBBDW>raLJw-qeMS(Wyzo_&JT6G=_
z!2ML^isFeXj6jA9IFJ{~TUmitzXH{t6+%e!{@md;i*o}no2RH;VkGee_SEvIh{D^}
zT2WYn=HSej=3Q%f`y=#UaJ?bN5_v&O9`U0^HMobaprlCb{hyJN5ka~7{iI^At3_B3
zJ+iq68r%hfh{{9L<KuT(BSy%71mpy@I**1$VY|g-TB%-$HIQgpvQg=;m+jvo5(3k5
z$pQPxw426Mz3*f4WdGi0HEDS)z}gpzVUE4=(^4ijSJ$R47YMADL*c#dqC>9ZEWjmD
z(MuDV7{|?9mkvXJ<(TKnWwxymsG*q6raHm-{l`-o4+}~c1j?$ua4ScW%aprCBp8Z7
zb)usVY1*i9;qjfx#MuXYMG*~0?Kjr`3BIL#=GAD)YhRI^%J*n{?tx`k?LR)5*5n0$
zZhZ48FYf5Uo3v4#jp2XDZ4nQP6eWU?SDQAjH8^V%1ytPIiY@7XO?|c@VpeVhb*87k
zBo4GPAQDW>#Y0>w!t)S6T#S<=!>J4C2WKRcZ-+;`ht0V(V9t5d@$K1X&L)Jw;=_wo
zdun^c>CCkOI?6_*xV(SGAqSjHZS@bD<@JO))oMG00PAmX>xAk$OUnUu-?%<x75-|S
zSp=>fXU!IF&E&~y1gt(owl=+L<W4d?miPXzQ9w9{u!+tT^aM&CT4KQbONcHW*sG(p
zH?(Zt;F$}OFmZ%pI_WSuJ}f&r$;qg&ZcQ{y_XwzpLo|5*PlTGxB}wx{y*UL(=xy*=
z63p&T_*0iiK%6)$hG`Gr3c)R2bTvXA3uPFqY>bTxmo1p%hMR=-*GgPb4-9c6nUB4I
zh~~-@)Tzb>OJJtlD*<)8LMGcXfUjCvWx*Qmyu%!=W3*}xx<?JQJw`xQdkLlHWscll
zZpp}sk~uzfu(Ix3)8MxKxzT6$X%ctKcu2oMmhGN34=b@b4&96L$I0PchG7Ld7e+<J
zgT`FjLZUsj(#^e?w*%9Upm(On))teob(Us&4eG8;uM0o6?**NiPvn^VuF2GutM`@6
zT^*b~2SWG<mzI3VnRy&SLnhLY<W-DE`OQ*+y%ojwB14K7+J@3r8sAZ;Y)IPFXR1Qm
zoNUEVv4IYxJ;1C?Nh+fy^h3vRmKSc<j+-<vGx+STVk|QXfoB7u$e81u^?F>?l;LMk
z^4%>7=(l9~F+*pwT|hSao`XBqH-|6L?~AFQcTew9D8v5_gM{h7f5+>J+6zfbdY;(3
zRG-?D=`2E^-I_J@@u|?9fTw5SBlR(lfW3v>_wRO^E7^=>>%TF64gc1Xyx0B?h=laB
zVs*&i8L%Pkr(^qpJtaw$VX4#N!@-X@-*K2BF87#zh!+rc8{NLObwa#@<7H?4K=)y1
zR#Q2cr`)#)jGhJ3#5d|+RlvZk(eiq##qUh9Uc%G`jpVjrp&e$fGgj4#7o*Bzw<@6#
zbK<&=1;$G|ZQU8om~-#GCbnN0r%Fgc64dS?_ezZ|?q6gv`35cy(0>m@s`yn_^Sjh5
zSyL~O>mTlSEpFU$7X0R}-QO)Awj7KZkEJ@>%C7sACM>`zs{Yk)Z_m=oc<<qzPjQ=_
zyS_ol`#ODX9_bj`rYKU>o8YUbJ$PX#`F<{BNx*jfk|~K=`ur2b)@#oviLDY_!@VV@
zvE_;ZEMNCI6f&yo1=L}$E27YRXHOvgnx_XN@sT+MtW3YwOq+RpP^gd*>x;T=7|fE0
zHZhyEq(S8gp#m;yPj)@EV$NNg6_%m=54_0u*uK@X-}(b9)VzItw86%d?O~L`wLe;Y
zYbOK`DO>qgF|jf)G+R0j^cT~E-deOmF2@jApf3a<tNePj*xn)NGNc(DcdxCqZcPZ-
zBuX_`v&=4N(QgbC{;B5QYw0Y_|BB(B+o39={^qze_3Ff404)3g7CmF%S)M4df=Yz6
zH2u%oPmF-%JOyrc4=MY?VO@TwY#7AI%+5#^A4@e>bRr~Y5vLjwk&8)6KjED!b&c4&
za5Tofn7TK=$>zgNj}TES9uerEhm&Ne)145~)zK|Z?}~QMw+O`1KHTf=Op4z4aVbF$
zk`V6D_VFh9HQt0;Z7Sw{xQsB1jCP?qX6^T8+nQ%)zUbMwwas<rv=C>tGo@j9cT-y_
zg+bY_(J^W#(4YLv-Ph(sfF5?nCgyk;E$!b94RPpf#+5obl`Vyl8rt?4a*eq~>RHS^
zmC9(t!&Cnjfg`0b?t9rIMK$)N=CskKW)hb5!xd5*=c#6aP9&5W;+hsT>*wpO2@G%C
z&<V>&e@+gI<c`upZXs{`l6n|5h-ZHf!Oi}xeWH#y=C=ytDME$o<F{PbNDZ(jo>mJX
z8@D8WP<OdL=$P{_y)b$5Z_^#dQ0cvddzaBVZ?j@;JW2{Cl=EDPfz;Vecm&b=n>nQT
zr^8APx!PX!eP|fX55-ZCD+Jm)T7uDL+WR4BIH+Wkx1dF6ghV(^foVG{>llwWVTLgB
z4mcexG%4wKN3WTH0K+kN9nk9^I>x`D1KROPWxH5>VE~t<M@`IvTz${(i4>=;87@+?
zp~Pdp%X+4CoRST`K+ja|SBi>Pf1F7H1>xk_(aa;ztVO`wE)S(O43oT<3eeM1umA2{
zfY112I65MNNZK$^s%??7FR4VqMO9gfiOM_SbGBbauQ|MdwY2+u!y*;YvOr@$Wf<qF
z<g?4487wL&TFO#c-5;j46_*Jlr=<Du<Char+}6jv<u8Q+&c5A>gd?WQJzJ!?!UHA&
zfO|19eE0IYy3B{<Kg=ltTmI883UHcmA$9VVCM@-rN8j*BKjiWL1qR>$Eif(zC8<DA
zT?-6oe7TBQn3#&`9P9ohDP$Em+BDC0<_@j_d%eA7I%axWBRC|)UJd_B`<Fs?p`%0l
z>_n}Ga}n)AZG3#R)~?3ipX|vB{dHDMkZ7lx+#QP*QVQ)%)bcbwyaMV4PuI*u{#n}l
z3jeRq%dRi5#~2dG=-D2a3qD9+RUXb7`_Z_hl_6tDjJANm7Fp?w3Qj5K$GXkhpPJa_
z$<eIozcke3;s5G{VOIvuzk0_{F<gs#O(LL9)Iu40azfxdRuS=gjN!b<=~OtS$pWV`
zGKNS_X<Vk4Xv!Dx?(lSvPB1}N5{gJk$r?6^n<04i3`1D$%Dj6=60(}DZ=K;s{T$@F
z6i6d?$C^^q{_E|LMq}pGgeu(>U`FhO8i&TsZYb1iLDN0vt|}6#VqqFp(_RK?*T0p=
z0#oa(F$+2{sA=dPB`$6I+`TOR(Sn4k1tQNX<TyV#QsJtxMNQ1(3{22!-R`3OEmv(J
z6z~{{#tlMukt=^0__iaudT-ByZ3_f<s3)AeB1SYHraBc?NdbQnBP+Wa?hYm%%QNxv
zPA#a-g`a(8@zI=2RScbma<KBY)qNz>QreRaz(8lW?%Ie<eT+d^9t#YxqT1IGhwKM!
zHcN?vG;2@Chs^`B$8;w7Pw~UtT_W=Oa)d!J$27{#*<6Xd<otZZ9XTb=1vUC{!%BTU
znXp*p`F~jgMjn|AqU1H;@p(LhiOyDV9OnI<p+PTxhx}Jwz?17a8{>zcAI{JJ9zmd3
z_D^6_XDMTN_&4$V<nYX?qz@EI9~`Pj85h=P=iX(Vtn)cz?zFr60%J8`X~cvN1S~vt
zN#ojv(;4o}pyiMU?^O#kQ9w!nP(UhR!7L41=6k06c*W1cg)uib2dA^R1tpGE?1<g#
z6+4Rm8pbcZeEECrC$X@t&Fj<sBU7Q(xb*xiFSg12He{Tu@}T<%BW>m2U}sP7a+GDp
zgMQ~LX0S(!o2s;iau>Cgn0)Yh4Dnpum5R`N<>-_^-rHU>fV&kdmxLaWtk3|?51QTE
zTi#<5ZYY3V(@*LDPiRX0CDRr|CbEq-XM6#jyK@AlZU{*$CYfGcVO-*%&Tl5`|BO**
zG97RL(&~4?=J)IU^X%odwT0;tbQzZ#ja8@EDtcqC<K`xf#7I7^VLD*7cVD^iw58){
zOqBb(&!LfV9ZCi)VM8KPfD}4-OeVAxAOk+{-AQ`&GTZUa$C1zOcwu^fPfLT7zaiOr
za5lgaYr6;d>aCYEsP6o<=_%eXxvuIU=F!n3Nx`;?un^^d>3s5>`U$|R=#vu%a97^L
z=%?4#is3&ilk<<B_g5HPr`uTXJzUgl0z=ToKhS6)nH1Lh?$i%FI5##icTDTk{DU(0
z?GRu<R=A)xPB+5C>me5@==q~_39VP={%X@eOxp8bb6UA3?dN!%v&wVV?R_#;|5FCO
zC@3M(+=leEU!<4E^jVP5xZ7L#w6S<bn<f*O`g%U?k9PuSd2gLY5d?4$V&QuPgaeq`
zY{rQJTUjN)ar$(wVfQ2m|Mp@HTQw~c&3=<@CD#7I_QV;n`e--R;var6)}eLC5i35u
zlf*SV$GM9%f2ZMJ{ww8I5t1U=vhYSD#&BSA$?5tLePE$nTq3x&W%!}Gm<mt?>Be1o
zMyzbmp5DSqP#xTORjFv&=nvW`u%8x}mh-*2-|KrBN$3&2v<1g-hAha*P`sY!jOrtK
z!Nl&>9jugNMr570B{-EQWHKCRcTWQTygTCWM85yy6P5N_BtR;a9P#xrusty0$b$~|
z5Y5cwE9iw3-bz>62E>S@W`wC!&q7Y6TbeUz_u5hk`Cp{;C+HlUB>7ykT>`+gOZHJi
zOLt9oVPMKrVu$db9Zwa)F#+DUc9y;XRkf0rT?Lv4rt=!N1k%9z7QT-8-rJdr1f%2Q
z?CSfpz+-vQf6wMAn?h|H{R7|$%kOGr75Zj-?m$3J5@V8Lww0A4AMn*g_>Av=+}dJV
zH%1V+;TCO6U&$*hGA?~YbZJVASe8$=f5lWX=>=(UFctp->LoaCp8@!W%8C*9b()g%
zUv$Of7l@0c3=C0m9X{DWO8z+!S;uvW9{BdGbp{T!@a<+l)6%;^K@VhF_neUf5w6;0
zr@+cA*t3uV7=kZ-5-6509=!$5XZzk9h)#vst#c`qU;KX45fbFa<4Tsy!SGP$lYyUm
zb1%p@rqXv%(DVC@154->#vOWH8%pBceze(5`kMXKWuW0-=q>%<(q5LT$y~SPR^YG8
z%{M|6c6hK-S?4CkJqor{N^so3k3VfFyqlD>9U%c<B4vQ1$#fqj<|H;SV5PoH?;X~6
zGKOabwj?hMISyK@Q}0~5U}uYTSvF-#5btrjZd6?Uq_3l051%X6)7)BY`OSOluxO~l
z*@Pa~p|{y^Z*RH|r$J2WRA+BlSq6&uTQP(0R2(1NZ{ph@pAK^p-=r;WHtN3=(|Tp=
z1_Bi~kSwB3N5vQ^r?Na&WUn&`_@7iconC)D<(dqM)-I_;PEyd_*_7Gqga%y?DUVU4
z&vvRmY|mAtbh9a^6RKRPaxPih6fxh00-6oZ6*A60-vtld$st=_*ZN+)qlea~+fu-*
z;g9G3k<)#+=Sg>u$uwdA-YDUaktAbN5I7Z7|Jvo5M&2PA^2$C3Buw-K1cYopJqL^c
zj5-?O+xMob;+8H)^-OmDe8q1susNk0;XRoQO`iJ^526?zYWv)}jh<<Fh0OYf`aqIG
zc!!5JMz?2g)ld^rX-0&W?)2dPs8(F;>1uCoSI&ir3AWPT(CB~t;jw(Gd(ARq&Q6?o
z5VFBt5D)zTGLlv{LAk8M2Lt*ZBUTr;!p(n7>^n)_i&=&~+%C1l9lO|*14gI+d!W*o
zbTT;Zj-%H@w}MdgN7^pH9>LV5J$K|W=(*+6g+Qm(YR&j3XQBEz<Sl?@L@1@*qZ~Ek
z|8A&EZbp0g7NhQ(YE*E`YX^_fXNeJ*{!h4r%bxE+@Yue;{JY|&E?&eMxB$WN*d1PI
z)eZVwg&M<W6{>eR8XkuyR9mKJZ3K;f$l#8JL4=sJxU!@ZqSC?zvRpKQDI-OuGk25S
zoi+yQkv}sK3v+R^Ylk$L|BqczM7;aEoXp2I9Wj65_kb>E`^?zzagYIDYo|~w=yW{g
zZje~2WpF?QGJ(4gm^pU#NY@c2H_bH1hDB~SjwbgC9bH~R+i*$NFkHJlc|9Fz5dX0~
z-^<?zjz~eqZ%b2eP&neU!HtFTj&$)cn{n-_T`XsoF6f~6s98!gNeZ5*{RJ-cb05n4
zdLcmq-l6dkN5RK2;;K_ZzsfZF$t$zLMT>7l_~I4@%wN8=v06@r%R$`TVDo<bu7r@S
zmI4K6A_y`lYyZ4Y1x?5C=sbc}&Uq&GR*SESnYp}uG?%RtX#mjf_rk`1VaBWT$-N!*
z_&~hSKC`*xoN)J+o9o4w?QJ{2xdp5eXTF2jX83nJZxHh&a8I=vF)&#|c=J)?Q{~~0
zvD)*!5k-=8b_)PA<j^brC$5Gto3+2=Qf#SzXUU7-4D`1ib8Zsl8<kQAdPyqYgtdXP
zcsI_PC#z=u0B%#wPVVnwoc@BVaOHHBLu6vTi(jnDb!S(npnOTD63ws)Dm$7z-r*Tx
za13ki^byF7+1w1zosmvSbetfAAp(>#YE)3@*w|c5yyF_$kgaJsY$8*+Dv#y673Fro
zQx)Ivc?tZr@4#OL!qfsab|`P^^`t#v;J1Jem3vubon1Yv+X1z})eY~DjRF}p(bNpT
zka%Peas1`p_l(tYCbey6>34?ZXK)nwuLh_mA1bXlq3ctM!0G7iDab=B8!dlEkpSU6
z%zs(uD0kOUdf(%}wbWV!vs?{+#~*jeNEU^{BKww!5%02@Fa61D;@$M-+11kbtxl5s
zFKb(W2_(%2Mh7a&MfRktrI-|Ia0HmR*nb$M(vL3DN%yDKwV9R!IJ$VlQmBb8x6nOq
zZ-hRkLT>{#!h)WxOtva+&lX^nf1&@$s{a5Oj{_SCd78YfF|9)&^_tRA3MQ5o+B|)L
z{pf;;Rt2uW$Nuq7BN`pw<y6@L6Rr|d)Gd#zpP+q_jp>mKJ+5;4P7JvTd8eB@8T^TI
za$HwOjfm`?X$}HT*i+o?Aa601_#<I$0qDtRU}s~$kCEMEJ4ASXF)3ezP@7?!P!ngj
zDAcE8X5v|^cwX+BDgNw*V2&m;e0??I@-+nN(MO}c0c7u(n4|vymy6kC(GCPtvy&D6
zQUMBx{65PWeSW^Jb5qj(t^@=@QBiqhlC-9n#z#!UK&I@xGWu5=H2d;ukR*>-Q&x15
zSeY&^FtJaT|J*`5YpoMZ6RPF3n?S>Ov*0NY{%u2+wVZU0#yc!}e2E2N5HnisWSg$E
zkCDqj(;sdz2KeidrE)|Q*wiafr`xl*L9Iq#f#0lyXk_?E_DfQR*Dv0R?+9X;;(R3S
ze|#11Zg~kK#I9qYn>%EigIWZT@~V>sn6M8&XOBH4uyZ5GXq#awvlD&QLB)4~_RaIM
z()!BK%e(jB!1NjqHfWaMwy{J|98>NdwxGf9>Z6`<+h5t3ZG>F0Q@`L*w>g|+j}E-3
zfGU@-Im39t^mbECWXIb$CFhkun?DIj@OY$>RSjWYauh9x8-lNG8lH@3<I_&h8+%7b
z#0xU|6g+$hcS<Q?rkj3lZoI$^4dl4=M#rY~Nf*Sr??qOVkUsvWl#Ki^57Y~Mu;T}F
zDGFF3&S$T!L6wzQ&LzNRUBj%G+TJ^IaZfpvRnca0-P!iE$nfy+$H&WaywFXTjje5G
zALF!ke}DhekaRcgBb>0ZxHzJu{o3a@8-vk8ZoHMGfhqI-1;{gr+TOl;sTl3J%ldaf
z$yx5sPW=S<(EF`@A}Sp?0kj=!Ls>#d`!jyPHF2paz*$RomqdGq^|sf1wIv3(vqv%+
zWT2e&PWv*Ie*5*7dYnU4W~%&%$0M&Nh`(@tzIb5G!J=~`QxWb9q8zLF+y{x%MK2gR
z{KB-i+DASw{hSy{T??w`E8zM~t>z_n+t{hCO(8(#{bFf$J`EPBtlLT<B^$gj@(dM>
zi}j8(1b<F(P}_zAcgtpb&CqB%h5Se*k!Hj-pE>u+gr<6oF&Q%68(ZhSsb&V>v{rDE
zYVfP|5ev-E+o=?4=+nmLE?QH;=}Y+M4G2{pkX-Q-z%OC6XS6=Dbvv;Y)`s0h34Bej
z0SGJR5fl{{515p~{rve8&}gRuv6RMG3e?$KFD?>B{(Tn%1Ip;=Xt<xTv16HiUy*U>
z8v6RedSW3C6~|X^{UsGxBBk$o$>O)K5fQyq`KWK?v4EoMO4{%v8J~XC4`!)0<9^5P
z?jX+hB`JvN+<u}PY@kCY3Yd%Nu55SiqhFIwPd|E>l{w}>*DP?{+yah{XRN@mxjELa
z&aWJY-7$o<pda%K7k@=1^1SX7DYZAqwr?RzjZYv^cgA=Vd;TXs-HL6VligmO=?cEy
zFVF3L)-dZftEVoDoUV~cGqC(PE0)%1exEi1_P@bqhXJ0RpxttYg7`dhYlbl5!eu)l
z9Pi2Iw29^J!@F4<UhGdyyEE6_xH;uA@TOuoKO9U)(!mP?w)9F@mVK1DK351SLK3jH
zSU}8r;bW`eR!>py>~Iis3<QX`D(4z}RWnWV`+gpja`>;ak;cAOdSg`j2#A*Sc4X6m
z^#~99K=tfrc1|Ck!2!!#LeQcUHIk=MWxR4%9VM84<8l>EnobG=Wu>QgQB#Kv9-L^`
z$BNi|l!gNIY;B8FE3G6K7!6o`3|7+;5HV5RrvD*A;c~XZd{CD*=aRw>xN=_|eHRmR
zdB$8U&uiv#M*Eyn;9yc&zD-jozL+uw3Pj~&S{knX8aEs;;U?<sIJxix(gLMWkmvVA
z;5)ix;h&#Po=avI(%sD>^-gKvLi6zC)+-_Uw7?;gqXSJ2UqUKnFduRf5VKdr^2J7;
zd&$V?=b(rCE`R5zLi_I%l=ai;zPtA?J#U~>OHKKJh`%9G`oEMcMF~8RsBk!Eg^Q<h
zyX}8(+X|kkAe33;%9p6&xU<%^SOX_WWWeNr<l+Gi8i{^Fd!3+I%rWHV3rw}LTus%*
z3iv4%Y?3FVodi{rKb%SEPUh2^yfa3xIjgr1!V2sKJW$?jeY0gM%~D@=T1&4zSZ59h
z9)9VNtFAGy5-|a8r^c~>PjoJia&LmLCSONrrTHBT3rwMBrvwi3=Xl$RYY`P@jAk$#
z4wOtY=!!XC{$vO#9}r{ZOVzhK8*IKkv}NHszQ&WiIwZrF5J7LMrdn-dI_=(T15Y4c
z7Z>P`a`1qGOKwx36pq!VWuEq7Yn3c=3=FF7J|J)>=axGo`5Z#KCJ!vj0gAtygFIn%
z|9&Q|XxFca&n}(-)=(!%6>kIb-k{I5e_Mg%tapdCTSM>gcg~Q?<x9HLsk7#qOD#g*
zoyif$TM=NG&Omocdqj0=Nq3b<rD>PoM^d(bbQi)TXShHEIpnG}7vWzSl+Wg?(S=yl
zfeQp8om$>=Yi$8+alHFfay3`ht=O5~n20z3B()}OIN8N?6pr+P2qH2+Ekre6f1s&R
z`$cDA7!O{LwrJ=haB#&w&3k~&kk39bIvUR?Az}^^5(uU^p2lG~BTf<udp6=C2LrMk
z6`0+0v6s9cak&WoGfwu+FQBi37TLeL@-_15+vN!NU&7k(Nghke-KrJ%?cDlvK@~-D
zobS$#CO^*GDW&ngfez+CG<<O|M*#FUBoxk2pvkd6q}-8e;s3&Zk}Fy)!?`0JAl|ve
zh8Lk0w%vXFF<%>%Q`SNjXV;uWsP7G|IY>HJv(VmBNMoIgY)>O2)HNK^=mXKkO<zo%
zEc68Z@tB1<C9FKSSpu2Qd;Aw+F8eR&N*h<8VlfbKu{t5}QcJx|ej6z(17vV_{5<W;
z7SSJ<^Vx?L<*i@TyPkpBX8&4gkMJX6GMzQvk9n6?OD&sZOkkjUhfu)VsjDBH1_I`7
zpZSaV;W@VSM&$l1m|ls|R5-mUj+ND#NlHF3#l_?R+0ryYnGT=}$3)BzN<$%;;XgLB
zj-((YQ8g>n6>%qm$lzg^puYaSSIYn9%Ooj4UH~qeoBmHkd&4*ei#o^PqL)N)P)plQ
zVz!}P>e~ke4zK2T;Xu6b>s`(<v!4ug)~DhzRxZwFG&Z|&SE`mSIrpx`PN}Au(&Kk$
zZe3lVdLae9JWO0-HFH!g5qbX*l*WefL((Z3VAgwsJi*-?1Q3CUmRJCp6}?Xj{*~_V
zaRl~^MejRn4KdW==a3C>*rj8NJ|kJW^@enL_{^Za1cct*T?n{+MvF6*qPgSOw@(g_
z#DhZ`h9|LwSf*@Nes|`7yBAFiz!HFfZ1Gp<tajb3k>`q~h5M3T20Eba<>Wb7{qb;<
zckZ95mP!uvH)@h^znN~lazyVh%KRKA<Y<e<A!ly4H!<T{w_?3#FDsQ-XBZU~E2Cw>
zzIxhfIimZ?KxGrYmgG1g`xQ_DGBrFyE!DR6Q3@t7w8iXg0o=cO2@?4$A503nkFmfo
z{RIQ}kK^H&fM8o&Wr-b)^1WteqE%#YcZWYbvgwdzgcM&=rM~uaCEp_(>!y2x&eN6E
z;M`3g#O~Wow@30jGDh3YN*e8-biubI%O;JYq|?F4S@@Z^rOPmj(8blPkX^oN*FM;-
zA{IRo;PHVKkg&2OD@e$Lhk77gq`K|t4AU~S&E9l%xIhrS^XV>J^U?u(wfN@(6UWW>
z*BS7YP;bh;#?n0Oqg3b=ZE2GZZ#C|>(+TTud&=~+KuFNNK9qaBKcliOaR4&qB|Gdt
zGPI(-)td+1duF?vVsg4!zmqd4W}HKC@upo}ptcs!;2c8jK}^%i0rB~(!=hf>>2NIT
zM@i$}eP%mbNebEWZ;Rg~Tg4(g;v_;RB!9tKeD*8y;&lUBLgrEZNPY4L-=pc=+A>?K
zs!IZqt39?qU6Lzc#8<gn>;?Sek4qrFY&?5w&RfOjr>1$f{hfTTR&)fmeV55Z>)jRh
z(IO{*MWkq|5^p=>+Q&8A0DB7}>Y$f)HWCW}C8@rR=ZMZIDjO%-*SKUlOj1L@=MHKN
z2(Fh@<j;f@3FB=kL?W-&yzaK5B~|=hMuGu#RBSV#6$O@EHgG61Io6gekF&XZm{<!b
zdB{D%;~Zf~;|6(7NS)K&i55H*jg?TuNgCii(-T~))0;TBB-_;(7d(2fx2~agwo(h;
z#{i8HR6ARuuL4ZgfQq6pKt0G@<06$s6*xiL1)0I-746Hthj`IH7bc2mJ=ek-M_3Ek
zKFe7)lr}Zbb?*=^PJVm}o*e)_^(`7hWV;8#Y?<Z^Jv7AH4)As)c*8-+`MyG5Fgx_x
z{`ZOeP|BSgMO}zGO``7r8L_AQv{c{YaS*|lp;pg>heGMf_8?IBE#avo0POFnc{U$+
zx-_^7e)8v=BRr?L03lA+>{0kpI5m5MH^S>|O%EOV?{W@nNYw>P9(6G-d+U(HU4#$x
z(RW!5T@Pboar316RGz5hBuY>ouTn(1U7>(r@M<k1$4Dl2Z7vV`@_guskLI|qSd;|#
zG05ZA0{0<UVe{wrGBhQ!<Bzo7Zmc&DjStCfUDd+kgqAEAlSWynnae<mIjMOL6XX$i
zJhX*@*=68S3!%Ehr(zt1LDz~7S2w62Y=r^Tp}V&%2ph<kXiQeaeP-80z~yNJ!7PUv
zKkLJWL=C<xVtNEw+9O`LsGAFn&w1A*VreutSfvctB#dVJd1&5W>BKvmM{7}c+C2gQ
ziur~VK%LTo^H;qf*z)1QQJxSfkN=-!@b2d}8p5tIz!@#s9!@->x?Mc$>wC6cjwUiN
zn6xxhg3pF-d%1|d8CYxo<W?X9)Aa}pySMzk#c-4M?K3SjFZk|ahLtsz&qAu}Rk!<k
zntur+g0@OaX%Gn%uL{!pO#=|-P`&P{_N~V=LJ!~caz}zWGGQxEzX0z8hN#N)PNBj2
zeWTI3%|Q3hVEJoccQtNYU);9Oy0>$@_dqAQ{baXSYn-gV_x@|`Ki}wSl0K(YAU693
zsw-LDD7|?g|L)y{R}qXPEp7WPzWK(i>n|A(NMkbpv3WQewbg8X6ywlp<ywNPt>X7^
zsx_qTE9&1g-LGXsxFxr<dN?_&hoBsDm|BOzJ@sEzoZI~84i@Fz5o!}G)5BKG&iFO1
zrkIXr-o+fd<V}-Bv~_=9z0P3%bn%R+HU`5%2h*0{gTTA--7bH3Pp;808Qy1jDXpYm
z)N{+ceHMu_CU$rrnwmvQ|H^SXGh<dHP2t%Mto;*}_G@Q>&L}j@!XpTgF2@Xy|EAQY
zeTK)4Og2yJX##3AH^GThV522TL$>i8A&%pNoOri&ZVwW7Z4K|yCWNvKxYw)E=+eDM
z4$3YyxM=i(X#<1T0V2n4CK8pcnsBfN*NWQ{&61iuw!lM{N}G@5qh)9)BzzkNX_|Qk
z=G8+q6^s*|xSVx2is4ZCSkyY_{?rj3ii;>ef9^E?CdZWL4sFC@o-e_y@jv<hL~2hz
zGSWxa=I~1<`DB>Rr1$5YLU`VEjSp@AX~Lz^a1qN!Ar)8Y31N6S<12>chVQs=PELzF
zMkNMp@%KEYh5p5(JG*QVt-$$&7i~`~C|><Je7YHXm0xBp<oXmo-J)@_F(`fu(w~de
z#}0*{U1f0Vv*eSvwa)M^7Rgd=gSl=v{BGhNX5z7*VzaKU;6`sgD%0us*@?AcSaedI
zylhB`SDTpAa%6Vv!Fho}9L5i03hgCX31urUXlk9oef@O&T)rywa$r3=_GA(zb&Lf9
z@9NZ2YZo9dYe#0%+ctfcx@x|g@B!^;l`J+pCo3G8%bLe{R#rB&>TkA`T@NEH3kpgo
z&2}y|IL~4Srtps`h|YGL7yEV_@&#=XLOd#QidnS`)G+311q9D^)t=gM6j|mfKVlsA
zyRrE6i}LklYCvh!FC>8Wa<!-QYAm*IBkN_A)=i+)@{y?Z@yjF3m)g9-I>l(mqz88N
z2KNtPtZgMc@7~5%^k6Tq9k%o8v+MU|M19mXo&2*uD$8%@x(EWuz40E+A`_AyK1>|W
z!UzS^7Hv9g*gYnqa(;C@S&G8K#klE1DRbvQ@vie}%jG^A?<%J0F}?crp=8&N5#Krk
z`T(;Q!yt1|uJLSqqqI&l+Vhx_ktJc)yN$(3x|Kz!@5>(f;qf!GIJXUE53R~#hpweD
zlh-{Lb-{JNe(EFhGaFA$&1|+Nrqa?p-%qgq$AQH3@U$_tzHOiy|A6Tm(L`|`4l}D}
zZ3B+<g%uM`MBof^+p{7Yf>z&3Y>V(%9H=?c64BP8@8LF)hSTL%sj8~l+^=BTvc)yI
zj=ruhFG%`FT0}aP8h%~O8w-wm%EUdMkf-H}p~?IpmXOF#>MhuJT1n1Tb_y?P_XzyJ
zc$)wj?US{|<TTDomxEQ@?17l*Sudfx>z$JK{o&njv#^duo+t*0{|}J&q3Y1}JWcDq
zu>)g5h${vuX6FI<6fMydO$Di?M<oIX*e#yd=EThYs1=ZVzIrx?^@&RN_4)wz>Q@}h
z-h-3GqRpK)pS*<7Mx7*kD*=A>in+`<RswoU_Y+*cU00v!V_LJb8jjd{k~hmYt;^{m
zsf+{i3L4@LhmXh{t~=%u9dx0~xNlf1`mfElX=~1An`~V7B55>(^G?1x_Q$B?QjfXp
z+m)=2g5>%$5#RLk+zR526AJLcmlUY>YF$$0S6|d}(`KyFsOy|p!q}p<N~ebmUk=r2
zc&A0sQ?Yk{T2V*6+Z+W3zof3dhu)rEaabSa%vT#M`w9<pzqBMwL;hiq9{;7Y015v}
zd1gg)puLFMrRj^GSqt(@2FKd%(x^*6%sKLZ^MuLo2`)LWgbTdpD1Qm=L5^GXvxVF4
z=WN_4nu6>I-U(jYFZ1G5`|3n6Kb$z|fs$IBE5J^qpN8y5>+VTOyb#WPNXf!h%z?I;
z6rM*w?c_N`gY6klLh<4wu9!o9uATk)Jc&?r`7ZNN;-<vMY5m^4OW&hQk4ggFKmV-#
z7bcJlX-uj&s2S~!Yo~PY)xoT<varx?T@Ld|)r|6_i{^<RSBwqdE3Kq6ycuE{2ltc2
z)tPdq8ud48nU?n3j#=nkcIoimeu0GZbu5&YOFP?hiAh}c1yihyuyk)qb2*FrQ#!@M
zL?&Wc-h=m{OH(FK4I&+g1d2kuuBN$%%GoR0-_@yM{K=oI?*vGRq<+b-YdVLRS6|oD
zO^<&%P8jUANS|ZAZ7JV2#)Dm^9$ibg{kV5zoo%WIzpk!jz@3o#)#ylcUFpn4SQ8|W
zNm1`15JFx(_N>Z<J2^W8eoR0VN0}Xk@RE|etzOG8kLljhjd}SOMA;qBudLyC2~RSW
ziTc|39{30H>JB@ppbL1Yx_;{z5<5H3R`KV>^vcVG^A-$~Oas4xV19eI`7>gI?y?U@
zc3c#QdaPaRZ4B<cXhJhvhB|ibHTX$oT_kx;N)f137$3&kKL;PE_6P>zS$NnS^jx9a
z-AbSf>RD*kk`yF_xL|<2%Iv^&6!9!SjycL)Wy=^)n<sHXFU(*Me~KybCUunUn^wsF
z@3F6AhPlsI7md+&G<v|j0?)r*aPYuW;mYy45%lM6z2QC~(1LUm*VI=IE3H;UyG(H6
zb6NX4zFZA`cu6I^v_YtPy=>yI?QKwVGNQ(ih`iR{3j}>t`0%po|Iu`nVO4eAmJm={
zx<L_;?k*7lY3XhW>2BDBfW!fnZUhPG?oR3M1BdSJ=5GAn%k%JWpIR~3m}87N*NIcP
zz4LMFIP13yy+gX%O}0I9Jv*2Be~L(zBx<8tOnSPaq-1xILRKFtDwCZHs!OCoZdPsW
zm3r<g+gBYeW(YgZ0?zd(!0q=iL4)n{PoF+H*!9#?G1FjHb)(Db)L3?vIi9Jn(DWoa
zLz^+QPlhg&8Oc;sHp@Wzta3;7VQ(e#HMt?wiI<&)1dfz0-Ez~^=Nw8y(+MS*mUE$@
z>x47h*L2!sd3C^%c21%ps4wqyYTLRbrEhdZU1G>a#72akH#zMJ@E3$fdC08buLoOU
zoF%y^neEedODu0`-}f};cuZLhlpfo$S>KVaNB-IF*@Bf2@2so>bo8Z#x0}<3e#WVh
zKY#wzNe#=Bh;4eD-hI3r^lxkwIyv1{n9pf%Z?`(ZeGm2c-KrzO5Q_@wt|fl&G77*F
zpo2078H^Q!YmHD~@XYg`+iqsfv$i3&S-N0<_nn67J0md)9#@6fi*RO8&*{owtrKvn
z-g)F9mI>=$S-_ZXQ7>pjAB$Ig><|uYu4OV)8`hI9i)@54S?`fS(t^ATVtnOW&T0#?
z&hD(LHe4R+8SmXdqE7Al+OxufOT%Czo``|MBBFqsu?R@iN(CaE>))fEIF?p9y;lpb
zduaHsF+JS7UtBCY{cf8G^pDwv>q<<jT|7Lt98&L$!Gh~!!m%@5Kl;~%F)=fjRa7)g
zeo;-VV<OY-He~yro*t!M!0c7QdY~X7Az_+sot&7sFh#lYZHJMG=~-)I@T1{Q80t*Z
zu^Z;0ZU?Y!;?>(LgBU|mCGr6I0<*zxI-m>z*PLXJv+Fr6Cv#|H1fO-veSV6LZ4w9L
z`0QXqcT@w5<7Ga|J6mn%DwEhWL2&bB4X28gZ2nDi0jocLu1Zz^NW%eKT=S55S%Ep1
zx+vRwsc{f!T5f8LLxo}`@-5e(*^HY*3k~`}YRf@bcsM-+Lpv1OI7H|O6)JgP{>V`1
zTl$$MxWRnED?&ylY_`RPAH}V3Lv@RVWBeyvcI*B&>BrmR%W;#XtY<tn23fbvZ{NOU
z+?c3xa?CAbtEru&S7i_t#ezr%YLRa%F<VNR9zLF)oPex6VvMJMki<fr>!58Q*49@5
zdy;yR=?R_D&NeyFhqw3yPP|z!7`KEFgVE1ou&B5H8y&;RGo>GNnoSn8O~{bVW9f90
zLsgIYpkU^JI_Co-2lMK&(%r8Nn}lwjXmDbK0_yzT`<<g*xZt=b8!P^dCyuHL)QC0d
z(ie8aNG5eVPzc#KYTIOW+J>^f1GMn^`bJ5e7^P61ls`r^{S97sdQw`71Q+^MN_@f<
z-$gTkqejAmg-msDHf7m0ivKy<c^M8wwC=aug`pFjYS~dQe<iX_)$I6lxL_x0TzLPH
z1xtKFf(+zoSZ-2I&>*PLfi`V%>EOtjyywPQuQeOatOPDS{OAm>bA|x&z^B9MgZ%e@
z;6SM#9Br+ub-XFoTMD=q^}sJ<w;qj2XObotEooa?p_{dM<tZx3EuVW};e7b@;Wf?L
z;lQVU)K!#AqP`8aQ@)l<a}rk%-Rt7KP;_MB&CAEk&Ry<-wS7A&5c@<5hm;%5TED)$
ze_wE0P>rqYdICt`4DYv3Ehf?yHK+^tc9hRDT^Zwe`n)KHC%0Gx7@C=p^(#hn^v#4u
zjQ|J7Ri<_<o8_wGHo4TY7|aC}4f>v>jp+=2ip6v32uy0`4Wc1mn|!>@m1Wo{A>n9o
z|MN;S4zzMwZflB&tO^w3+@DJ_B3IgK%p-AlY(!f#K7f5uwGdlSo8i*%HRaJ}BUX$-
zWqHWUw2>h(Exck?<5zp~f$PU)(B?4Co(s19(@9BT#eZHWTRP=`<>!_^+6)aJ1_);>
zcchrC^|sE>4&8YhE`v?>g-J#|N9MWq?-6bZ<kwC{W95K)LW4WcXISaT>jznb!iY!5
zc7WX8weW|Kl5Bm_akR=Gw}jEY>(4NfQO0Z6g=y`829mR*aBE%mcwADmvAo-8&AGZ{
zf>uMS&+A-gG+pQ<_hXMYX>lgweUu5jyAl&5ob?gO@~aezuu>|!{UwQL{o%2^dEBo>
zKIyBwqU+-t!KV*<gOz(TJ(Xlu{xaFde*x-ft)s3tkxT&OL51g+a-#a(_`11+b&+xg
zDTkS(&2JA4l|Pf<2XV-5G-9A-E;n@>?wE^ebsb4D<*7bXIHMUb5-VD>%uju+iCZ|G
z^3O)rw}D&Gv-P&7AYdCqL-uIJitp7;?`&W_GiopY;?7gFD7WYx?v3u5EfH8%1-Mih
z|LYT@s<CiIr2mBW#cd2OIg{1zzh$SAiZ6<&+B#gq28ux7x5oiyp>ZjNTp~WEd|maK
zjy+af+TATexPDD&bl+p~z9>h2VP7WGJwbqrl220G^a`JPlkP`*M)`mzah+(LeoA$-
zax#1hNttC&?tj(q;Cx}Y1!So(l6f7hz1kS|d{HGS``Alr0X<EWG1n=m+NcDUa1V$D
z37eX~YS{^Wuv9LT=AVl5?MMef1-6?f<MJ+u|8>C3LhqFF6)QESmy`xc)t;`b*5U$k
zRN~i)2CO~B5VhkR8(nL^s9`a6_97{0pqEMdob=d-O;<AW%gbFLCb8atJCI@>&+Tzg
z86%pcmCwK#N#gE$?-I{!I(2gYf5tjl)Th=JADDZG`^}Es#^JW^p{RRM1jwR#w(|Iw
zff{$NJxSHI7uAJ-+&6UXhmFuFI78bTop}cK#7?q1B=_ka_Aq&TM?+RJg}Q}K;Rol>
zxdeIyyH-_z=HI?qJa5uq9XTZ^pT0*cUrQn*g2;Jo>5w&RH0k)J5MOhCMuQCss*&>9
zzkeqzDH*IzR+dQf^K746YqdQNdmJWL+Sq8F_nU8>(}220`&9GrXsn&_NW5x&DvM~i
zO@$P;(;X$~r2)GmeMsSR$K`0g)zqFn>Mo467zCJ?dfsZ1pau8UP&W|eqz9R3f#z7F
zr3MpAVGWb5aM_u=-t<1K66E*aA;+WzpFCf>InZ*lKoCW3iet(Cw-b@*|20#q`rN+*
z*2&O~Q@NvqFfF06N!vs+cN7L!YP^h&?&%Y>N|}C#)Py1`2}YA$(cQi_8rGjcGJ#J=
znU?gAz58`(9-rB$b|@Dz#4|J`EydO`(hWhDk#zCFjQl~2S^C>d?~YrNgW0^>7HWA3
zsTJE+sAxkVVCG5(?#6*ZW-vK`eVNPA3A*L;aHcblI_w#3;FtFl*2or`ha~vd#Ug^f
z;zHaCchG7LR!E93whi>90js1UCJPaUVCoR1Z{WRf1<ugW{&nc&ArBp`H~GbZ7}w4N
zp5ao6VtP~WEKHcBYOB*K{UNk28~+XmF<OxUtz_Ho)Ak^%zf0(mD&JSL6@s;{1P@jt
zTN97_8J(rGVC*p#Ix#D+tqp5WIQ*kRQ42LP%~~WqUn9sSShNd(2_fT0tS5dVGmZ>e
zM&f67uV3tlPWX5Mt|^6;YvB}8m)iD{a&W8ZdmfsTaM||%tQ+UAFyS$G9q67d;f~ZF
zEG>1&p`r^qo5B6unTp)nJ@li84iIZGQdH;f5%dplU<K2l;l@PP($(}Xvb%dzOjZE@
zvXy0~4mks0uXw{so%S0I<?UYu1dvSNr1dFHXwBJYgaG!p3~)&o9xr3HRY1#XMX3XA
zX}|lB^=;&L;awuOlPP=&)-;SPDp3^DP0vc=6AE~8sWPuNKIN>)^7Th#Bcs&_uTL6<
z<V~aTTYu%lX^-#p!6X^(s>aK&IBGEWgT}kg14<yY*&e4!KdL^GX~iY?#Q}drA{y4w
zV8iM+0Vjl<RCp9l3o|s9jf`>e_peg={EgQGe-nrcuy1upm6aB)tXNkq?QO8FcRyDg
zzkLyZQt9WFGO{ndy3)+ab)Zp&gz#^H?%@2lKtX8vcXsu*tkR{c%GbWi_`13tNbfxL
z;(J-1HKv~po{()Ow4*+)L=QLnvY^JQU@XT6MzraO(_=fiX)j8%ztk$FRc`s}zHX?p
zRyz6*hA#tq<W@>x&2K&*ioW@<_vY0KU;=hL8=P~^JshFi;<g~Oqz<p5i#mALS~8ho
zxxO`h;`=m$xUU3lzxcb&*8@Q^B1H!%v8YldIp~SL31D0doe@s&{HE9WwWPs}loL(o
zZZ7J#$K#(`6j%D&CfTOy-;yB?49<kif@jTtdK`FPSWG09#08gmttwX+Ux$LnWTxnM
z-$5cqQm?s7;=iS3B|m+J%XDJCT(gi1p=sk+Iu+_?7}Cu|cdrMUX9KNI=E84+0eDY^
z5Z|=F|48x^E9n^$#*WSRtiah{AVv6|WPLY%LY}ff{CFbQ-E>AAT1F_X@Qs#^5d#dk
zP@#K=OUyiV1-Ei!`8oFWRhk&Q-EZ~wVV$*aoh6WJYrR$}XUKHC##>0zlAMtsQwXGV
zFw1qbLx*$i+Ad~*1jmFG{;`2aY@1ZCW%ccj8<~WXFKyzBmuu$TSyKr{>q|jdZk$}&
z;k(AOUR-au1tPZ4=(1aN`ay~ggvwGSw>>|iV-<AI_{UUM*g0!jXD3NvH}W9wf`~1-
zz;HxBxb=7VFA^4->1gH7k&?E;XPVy{9H0IXt`}5Dz`vbGu(oanZ)m3V(5}<XQKM|U
zw)r`+u}CJneVi9t(O{L(9Tycch5nck=YGSD@UCFe*w)t8ku0$7oxUvr0S`+?{?26|
ze)hDAZ}Il@d-=Zp-e;`?JKfVYV9c?V9MWYK&M9=JCaA&X?w(o<m1GuJ*x`y$kIAjP
zYz1;GnH7qSiaOu+#Zo+-F<)yrKmKFg@{m9V5pK<#MqN<dH*V~7a<nUJuFTeLHu2Bb
zgZb9=)8*eZJ-^hg>|IZ3OSMqu7CGS<tr|#X6_t!c=pwB*&&qs$zv<3kWa#XjjP}cj
zt32d+Zq93$WdfESlcfrSrSl(3SS5j?;+y`Gv4@uI>3I5;%;A^yc8DNv_Bn^n$i^fJ
z3?0k@XF1ZlEEZhdo9k;kND~9+dY*D=%^3OQvJ=b_9znV@USYMX@x2Ot?+glih8EzX
zx|^S!TZUFe7mbz`l39QHqee95GO5TU9%{6&%nAIf)PH5m{oj@ZS{Fz0!KKQ9JCw;d
z&^FM1^}ZX(mJEV}PFC2jn!MH$xFo-@(F+zUX;~&JO62zeo5oIzAI9QS$;T$Aw&hE3
zUT+Q`z8x55cGI0_xn#cxQnVPUEmCr5x(oQI7)r-{Lj)q({20=AeWv5ZNZ^!R{bIi3
z*|cFv!K^#Z@Bj{1m-dDbqZCb+0*6h^e0@lR9#Ji3!tQnAqe)yQwxIpe-(w2cNvaxi
zJ<C3%{z(4WbubIy)>~uINObX%Whoa}2xJI1<)Zgw`>s=J=5bl3+(AEv%LK%{w&(WY
z@kw;*bDE!pHSMtf(>r&t|34IEtB)7=`|JVp<q?cMkH36-l+?mZM&hx_F9^XFZ|H@+
zcy^>Z>cg8oiL9TL3lRmGogISIU6;(@JrUm!zPBLgLK%O8yskEP5PC}HS0Duuec{C8
zP049MDlMt~+OmirB$_*NyWsY8ce9>?JZIYP)7x*TI`T6Qkgxl&zgxMm&E-(wFhf}b
zidzXXS1FkD;uE8>lMF|aLB+}|E)6YXW%Zc|8$<5F<d00qt!)+T4>nRuI3pCYf^N=8
zTAyiBs5h=eJEiuPHcN8KT?P>r?@m}7?*bSigpYo0f-|QFa($8--yw;nlV<N}4Fu%$
zJ(=RvtFNwj^Oez+YEobxDS(>FhDw48TpY>PHo`tx>jDB7LpGn+YqX9BE(N>i%X=H2
zkmV@ld!)S7=Gx!nnag6bE+Tu48`hdn;&8#;j=@>XRG3a0KW*pt6YT$`1k%I)CSV6C
z;$;Akdv}YX^hf@XoY-SYqg;h$w9v*4+IVO6s=wZ*H)XpJ(Bu=RM~9pSp)|GbAa6SO
z34SK%_{oS@D82NA*7uT~h}llGvM!w5FSX5@9;j%#1W2Am{ptuzH(J#_tarc&=oxXC
zi2ty6^CPQZaV5&*Fxq|{ffXoyu7=e02h6*VZ?O&?&3depu-uVkbu_Yu#r{m>eaeg~
z>RRlDOtk^`3}ia03Bz8)u>RR2A5B5`mqIsI2m(UR_W?jr1{n8$cYSr*ijwa9u(tx{
zUEX)Am<S|QEWFJ{O`H@W@Nc^HS0*+}EI7}EqoYukX6muvOeHI8@5yz_;;ZbKkptm_
z-QNytNmcBkTsQ@7A{eR(-N1@Vmjkmucjt^9LI4!T2$-WL7*QLTFp=v6CGZJ!(0cv8
zk?HPuElsW*+$KLs0pA2wxZ^XPTFQLz5OAiH%Zf{tjuMu>x>1x{6rtk|`;K3N$aaA=
zUemI$ulrb_46Pb%H<yn!9)*N8WRq1MdBZi1Kd-dh81(+J8d9THpAPbO-scz8F?hi4
z7T3n(*7d|sP-l|;Qc1wId&?0B6Y_&ZvkFc>_VuMR`g%X{b#_-DoRMbt4DQg8J0SCN
zF}byGqkO_?5sr>J-9TxC#jOy`A#btL*Z$MY)>5&-v&6;6;nE=GUYm0%wuo#0^mX0r
z_CNyY@IPBab@E(JO@{;cnwL41v);=KQ27154+q66L=s3GA?&_)ZC_#m$K{F)P<&ML
za62}`Zd85j?pjcevvlu7G2=$mVZJ4_er<idh<dp^o-YEqtx__B9}edb?kzOFI>3Qz
z%8h)NUA&kX@)XpuU4s5QU^A8fxE~ZzmOZICgAFObyfAFb#%bJm=4RgzwLF^msFjn;
zX1s`iv&gG_EIRRW7OhXC8ipA#b<u4HT_C1K%9xHnT0p@a!N-pgoG@g2Qg-k*Hpk4i
zyuhJz%r?MfU0<93>C0LoZe=?-M!86{v9gWYBYFars`Wy_#;*?(2jsdlSlKN_>)Xj(
zfd{s*i7mFI>g{fjqWxokx2y-0nU5dN@Lf9|_CFf#TR;QsEvw>&OqyvjI>dj!`B1Sn
zKg~!6D;Kw$b3RZEE%qydVbD+?#3<due|urj(=4F~&$cm(RaVWue2nj+5L%p2HSQ~B
zkQqXVm{IeBb3$jPtpW0`$aqz%B|16<&{{r%AD1#&w?N#*G5S(C=GaCACb#;}-r$Y>
z26U1zEX=(l|2Kju|BWEtNGX1_=nY1dc1LrUVujtP3Mh-jw)nTTu4gf7f^yQBiQ-tf
z2_<dw#r~jdZKURPzxmpxX@{ZP){=#%AN;{c?~j;HBA()_Ur-|;#O)qz><J3nHWJCN
z4wD*5&0BatFv|ieu-T%GD$AN6##y|vWae>=OqsnPPi`UpX!ia?(EJ;{nPJg%R~k^2
zalb)6`Ik)f8R+(9OrOs0)xg!7Lv;(u<n~nZl}BCfgqR9bXO4$gWWuKoy&K>`0%Fay
zv)opa?ls^#EMC%boT&|0f5#MA>>XVM!xA@8?iBW+{0#Et^8HE^{6_tkKPjRM&*sQg
zuLHhJe=))85E@x0EHfVx#m96xb68%U9nPv^3p6lZT6Oxas`-G798-<Xn)+2`)}5!I
z=V&;c>n4!wDMEgm_doj|YofKeNbL?>m;q)@$Jm<}OiN?SqDOY3I52`squnXEqB`#D
z#20V$gj(0X<Nr3`$9xC-rFb+@$&XKPcJKkDhrqpwYKbqjqjUal^?vCSvf!iEz;`KH
znEl_#7DD(D*DWvnYPP(O<ISuN#;Z(Ey>Dz?k$GI#34h-QYI9hPz7k2~h?yDJ0W|Mz
zo$j<Tx=YTnK}p-tcK?HzE5{d&_E1zk`S+a)a7i`0y}K3ahKHKR1AqT=zS6C8MxOS=
zys9Eo&0b}%aeacuUyD6L7D(fdwoW<w$iarBi4`0H+rt+d17%H50Rv2@GktqPj;$vU
zt)%SEXapRUvqA-%5@n_0RDgtJR0jM&DaDmujg~*kk(stH45|$3*qcd)r=`3N^<L9W
zksG(3jplK$i&EYPk{(|jmYi-nRrD@KBcqJ{O6Ghf=wA~wHbtX{vgsIuG&{5bH?p=^
zjsCbIJy67)pO*WN-6Q``JfXFiEyIEp@ePKho0U4OV|_RP3_h$qNbr*Bv?!5wH(k-&
z91wq6Zr5|m|2zIl=5bp6Vo6sFYq;;(tX~KcA$2=gT;0{Ru$j~l?jk`ZSW67f6Ufl1
zd+=&IS$XSVthD<Z@VnY`>K*#jwjVE^lN_)A{Pb|Ws-ymJLOj~@@cL%PfR}7{<)A;o
zTr^`k<Jbb=ifyq*n=KgdMYnh~tTKCx?Ck!1g!>S?{4yOCtr5<$C{B@J&uBLZ>F#9E
z3Um%EfkB_Y`C*(i^781<u<YUE`_pRS{1mX|@5xwssIV`w?%=OKdXtF!31W<1sP8#L
zUqCurOtMfzylT*w>dq-kfi_yq`*Qq+A=Q~g&kmrB<y?46*pzSZz>1agU)`sJ)PF*0
zG@pPM+i{g0@%;eG*8^DVcB$}hx*jFX?;%+QbRb*4J>tU#kD&vL4O&nxQ_`3x;3_N9
zr@|kZAizNjjp7IN{CT<g<Jtp+pooWJCbnutdUd>813`!_ID|e!dB+5CQ$k0;6&FYq
zb%s*UzYuMVjrmICSU&Dx@7S@>AOfPW-yrgp#CMhLlAynJvl{s$)1%bamc_oIj@c=Z
z+5xI4F2Wb2B0qdP=x{s@sY&AoH~O+U*UZc_z@nBar%a=81$4?w)CaMm{Mh6?9c5NJ
z|F#SvgMwT0f>7qKW<Oo&%YZQiR=?~fZl9dUjCoGPy3$Locr1zUsR;^33h+EE7ksx3
z<E0w1jXKDG11z87U~SfaqI|FQe(IJ3jQXth@P|&SOlnMGQXVvKXm3GzNX9EXV&bdg
zrHNzu9O+vKC<fS#xo!}sgDJqbi-r6th2<8RsEWCs-^xb7`jWKS`)Aj6TAK3$L``2B
zybiJ!gfG?X9#^VCu$v9F&#&9GQurcAoj9*K=KM}DsjvNC6;iG2|Go#dOIT1W#4+og
z!M))%4V<R7o+xRpvnZ&qPcA}=sdAzBMAWY?5C|1Re#VhRHN=a8<m&3$j;Uh`iEvIk
z*^~baoCsPw9pyr5{#Qw$E<QoRJY>6ATYI}W$vM88WUQw&Yx0|$vuAdY)SQ<xnv$Ax
zw?G&yNxasc_F5y@Sc`V#Piw3sW7;n|&I5a)1a1%G{)=@<um35Q?<eTNx9fw{2JE1O
zv)1x#U|dAUI{%t$=vAoGJOF(AqRA+`_M+$<FJ?qg(N0SmU~^Oc7kVfsdpe@VHYRqe
z17kdRTGE8}$98tkERSls39`Os2%zoPqOyB<K*-mfG+ASq+WArP#Hl1TXnSu%zUM%k
zi0j7mj)beg(Kd4j9U1Z0EpO?0N&0ia($MxhE79ln=SNv4!@tmpg@G5^B!uHb_dA&Z
z$K03R%;8!gc!WF8b*e*DcB8OyJ>dfv?FaAmneV>18h8l^&7q;q?^JpJYrx7D`|m(}
zu9VAVvw$Bf6ZGm0BD!i*iMjrZl!5NKDfq{X-2ydNOqMHd3(0}5@y|4A9gA5fY3b{s
z!e#J&X8RF3e0#v-BKdGJMvD=fE;EjYre8Y#j0IHz5zEGp_{82cd7u>7MU@o3r@#{C
zZyKu|oO3om&G!Ix`;YgL_^cP|KWaI3*tcXO^q(_>lQ&B#|K@IlV4cv+(!vad#q5)@
zU+3#FmhQ4IlTp_Coaw$DX=|Z>3{12HbaW=_zcLbDJpdhappKz)9TY^8G*4gjDGcMg
z=EuG+A|9sX!5n<i5hH7`p8bJJJ3I{S&emUFQ*idDn*fSqSX*_N9V>e44|KR47i-Nb
zqN+7$<Z2;c_d@IVOCKxQU`8|dj8&{j<PhD%XS`#&teF-=+<TaJ!W5QsU6u4x&FM<<
z)B1SwzfRDDvk8JO$heW0p&f~{Td{TqMEa9;hIozG`q2CR2`fH!LPaOD9x?&QLyk;?
zbe;FtHQ5k{(?Y3xNgEKq>*{t+NX_9m9ak#XHcw+voCUz`TA|H)BBl)a!%XP;f3}U)
z9oj#v87<y;+5{b6_D??kzn1`*cWa=8Ohw4P>fbtAG6(eqjqQXX>vW8FAd4IuDeSgF
zmEOn^F;MyDsNB!NcVzNKA=fnKz(`xpXLO*u%=Fy*wIccQzTMRF%7IQcS<tZp#X_%n
z351H55z)KIu#pwI`Ww&Vk-6$Q4{eqKMbC2Ad?-ds=N6@YXLr>d@fyGYNs{!@cCua<
zxuSa^Nh}(EPZBcrs_uEZ`gxfPXY;0$&HUvma<xyLw${huXxxb7;2hU!5d)0YY#~hw
zsfP=^2hUaW6$MEL*x6`#ZSvD`XN&s3BsX)&W9j&djK7-^m$Q9sv!-Su7daC|9Bgy4
zQpVdy&SNnlTL|dfV3@lNWjhR;+n-rlj@h2w7-K&2S21n&-NZNZ{{nLf7q-Al5;3#}
zQY_Vim|HV0|9$$Sb&1G5tsXZ@kv7>w7N&p65F&+p@Ogjp=2m{q*3Aq;bR-HDNZ>l;
z)-0`U1`h`Ee@%>~q~3SLL8BecpyI?ruI~tk@7n_wNFBIT!T?0*QiqeSfJa14?qbJR
z(cH=Fv;wevBaEJbsl9IwRV_9*4yG%;X{^P!t`c^FYV?&(Xdf%d(b0`PM9L?oXDiUD
zGuU7y%DiQ>jHX15FS@2*1@XB-6J<xb!^kwg?x4JE`6n#2O!%#vy?5i)MG~)M-OBgF
z)ntHIcav0G+m5UL3>p61jW3^F)^$j2@}2CuGFdZkbGmd6%4t324{m=P_5J^qN<#&n
z4kr1{2W%a<3YND9r@ZRFC`5!%v;=N2yngxx7xj_vY_0VEePNf$%3?h4CQkoXn%!3v
z88PHBMFzAJO`ut~HFeOb^Lh8~)qDela5F27uIfFmyA~dIR#7Q=k*mV?t$+>C>wu35
zfJV9*{#tw6Va1dM{^^-4cq-}#^ajER#0*#^%4kBYDB(MUO!L`rr<j!kM18h8G4O2w
zC$2!P!HihpQ1g<fM3Z*?ZG?uh>OlGC=G(3XDW9b}zmxEW#|=wN1Fk@zyBZug<Ng7B
zd@QW9%=?5inTCK%YB?Qyc{@K=mUqq~@Tnx@zYV%7mh9rba<W_B9l5Q04_YYuI$XAB
z&v(QRwrO`P*6s4QoPOA+{byVA|1T_}fJ9f0cjGg+!<$l8j75=xvz#J1<yUgTNVyd+
ztf%}#OMG;;0a|LhM(L1)z+RcKwL{|6k1fg`VmY;J@e87m&4KE}(lV!&LDp&2S|Tka
zC3q{}TjlHJbE>1N!aL1}q9R10;!2K$+ujNghugbuo@+6VSX4e4^V+8x!}SfW>%xVr
z63^3Jv)_tFv*4nPR00jCJs~AfDdi7^V^)`cJad3^nW_AxRLLbn$*Ud2H)#-dBNL;-
z@7aRVAcy3N$+kFL31OZ%@}(}Fdh5VKr^eU#n2fbLp97!LAJj|t>R0KB&l+Bb;LB1*
zd@ojW$p;wcr(UM3WYhV=*e%&Jwla!J4%;6c91rSjRq${+=0&ZajNiBEJgulq)ZZZY
z3xG+1K=0ycEY^l~TaB;E^TDq@QEg*0yJ%9#R00OqVVAhe_9F4LMfFERypM*qO4pVb
zV4oWevpWSaK!JF0tpsbn9609uEg8h%!}~}h%IVKJ-)sUJzv}FdcvBX<uMCF>gnSSg
zp*@tM&5C8k9}^d*XL~19ez~`zK?gu%u<V<&)^600Y71ESw5UO7P%K8z!^5|hMXRo$
zz!s&LbdN#4rsjmN-}J1Lg#!i+urk<KQx&2vvknWeRFj7M(|H(q(~8ZZg(HuSBMkR)
z7|57RP~)46Ok?DZ77FUed}j46V@xa$leeQz7Nb+BA(qSoM8bzOcY-_t;U2)zXTE{(
zfIRPY{u{f{pd%@=2vKAy4*M5JhS1DEGqptg1x;}@5}l4lhxA94enx3(Xk!aVNwYt^
zMI!$S(^>I|{lglMc~umRG&D4O(7NcB$A{#RkxwhDtB*Cx&PYS)q6RwjPa5kA59lfS
zJ79o$($8B`4&ni}I~)M9rPD7YNZEUgl&bEd78&YDk`CMFyEotkMz72^x-VU-DyvQB
zHVdwHYgn}|#(O^OrAc)jc!1}G(e6qVXa{bLa4(N$jj?f3O#dNte&YuE4%J8=TdNBU
z94NQ+tKO6P)sfuPwH=lfKAw8bEjSqjbp(2$n@+K!Tpw3#@&n}6A)qA^<dMsqu-f|;
zQ#0?FK$<KQbnCUw4u<s54z}B>>SzD?tuI5hZXCd&Ql`Zn=6Jtg=*zX~MkZyDDOL2N
z4uzZM!5Lla#r8<2`Zku18sdVH68g*ecH*mp1XS}GZNXU%*HJuIpPI-(w3<(iuPM|!
z11pGYwcb)b1AY_GBk8}(Rv-2JtZ~zF&K7qu`L`@s!dfr~x^<W{z*B;X-&ZG?D&JXc
zPjbP~q89Nlp>mm&@xO%DIi2lHhF@J>%{_mjmA^H7tAYp*w+KfWMJbq`?p%ZJ&&aE-
zQ&Wl8a3FW@yvH5`Po3U2+0j{&X^^d4KrHgW4cuUQ1q>E!IDq`#P`VJ9d^*k&5xN-&
zToAp7{la~2IW@X9L;pfU2fNa(P}$a0fUotiOjn51a^lF-pZUstktLFRrn49#%UACv
zocO}LSB?+OPtr<qcgfdzM>wEaoX*hJNP;Mqb-<NoFViurzQ?2>h_-qyy^nU^@pZ93
z5KGt}<WT%ol7V+lP?hz21u<7C?XEEiIG8hD{2F|^COO%|6EB^;D|w{=SXbTt@|%=U
zB?6JgzCUTCkekY!g}j4WBo7VN7HzN-;SLHv_pvup@mKx4&l%uADRGaEjy^*|g2%?e
zd0(J@;=i(x-;LG?7NS;KO-DkZ&~gA)1y%`P?9ZDRId~O*f9iVit-gmY;{|Sxckfqv
z17+0&#s7KpqzDzG&34thhuVUE;4VIq6X4lMMgIy}PYIsoKx9TxRTs*j=n#s@G%2GK
zuw$<g$tlV^za<X>e47TS6iCh+u@U|4?|s85{KAUVYcCc0seqdn$o-KT^9y=Z=zn@(
z1fN?D>Ig9ib{KsM^>dL~ITTsp9-ryV)wP$HO*fZ6mh^UPaeENa)Ehhg<N#I0G8d=T
zfxgBY{>jX}&VuP0DemFa3VsW<N3Uiu!sS(Snh%v4ccE5W&$jnQ)AbGy_l%DIXl;Er
zS7jwNU97hatpj_*?M#ZRN!`>>(+0NrDdZM-XVk7vy8F9b?9G~XEg)l%l-j{WgoF!b
z8f2cL<_huvbK%j^mruL5qZWgp_5%N|plGh2wn@L_?1Y=M>ANSn)A?iqjoO@Uwz84Q
z<~NAmxfyL7o}OOiU?K7%Um|^vs_t7m1xsaz(aF;<>Hwk*Yq7qYX;ecfmzg%1w{CuV
zkG!(3XUfK8W3!rHV$w=`ekw2c!H?0Yse>I@q66KK=@Rkp-@m_m{~qD?Vt=*auo=;C
zdn`MGm~+D{m%fu(Wff^r^zp7@?;;GJ@ptV%eXk)&yGXKVG-H?rIrU*KK&KuPGe=yL
z_6<HDX3IX4>>rQh?5z4w@t|SEC%?^Y{l(NNI%-9`+E%0GB3&I(Vz!{?^QQiwIzrVO
zLHa<a`Sv>JBc?=$FL<BVjdnNuqSHymgmthdvXC~WCevwjVszfE9YlXP5Rmqi<M8;x
zuQ{h<bhs{eKO}E4QFHz{)bO&eJ7=oZ#JQ>ms4B8r@Q$!dDoRGACtQhjGS`9B$X?TY
zHwZ|5+*g4TfXgv2FYnvZlHpLAuvdY~AE&$fdx-X-4D{NvvDNDUT30^LtW#aMee{nd
ztRdGCq$}r8UUJ5)tPYVQzav0<eDnSoFo@agcwy*Nt_<~=A(aFzXe0^@Ma0lX;Lh%b
z{Vv&RWKtb9H^L6zLA(o@FFH`WpbZplZbBzBQr6^LeH-FR*S*+dwo2+)&y5k3LSFM#
z8?(k?PC0vB_y+Wxi&+_&o%n{KIg{O)qTO!T{ALq;M5bo3-$|#=gBiSUP!is{9>ccQ
z1x0n8srd_49wAl86k(7jUwSx@7>k6tsAi?)*x-P6vU1W^(N}j`lkb@^wrTaflH%q2
zZ#!4ZA(VY#4Q6I$-VfKC=#(O4c6N4^+C^gl46IR|ogX48MTcO(BP2U}oq0(6TXF%o
z!<(C%KS(l!3xFM-j?_N<JEG=w`!RiHYFvo-J}RMB9ZJ2eFkqgv2hs(d9Robo4Ga7S
zjBCH6-Hsb+Fx=r}-96rS_dWWjT}Cx!_pDB{JeNhWLT#^zS+(bBDt+;mGu|oe#nv3U
zQ*!;OB-73je_RV+eB&8e_Fvx;I%mgc28^YnZ;@_J>q=Qu1BAa=r`83G??$m%9)!<$
z5WyZ^gANkW^m~1+uYg>3KpM-)Bym7tTN}EgEEO#!W=<9?>HXia)RbB0n2^mzKZAx=
zRFmmL=rc}6Cub>6ss^IH{HyXuLjk;<7dtRAh(}T3u<j<;&Ug}a%f4SKy*<D!;$C~>
zZ8MSUf?*0aW^v0cX}-zdyBN)qDoK{h1KR~wRdr2G#ety!059Qls8GRMi7W8W(FN`9
zZ7BV==H?ns>LE`DO((hGKn3Mnq((?D+j)#K-u^0T_*{GcdF;xs_#ox4r&X$`K6c1$
z0aJYuFj<eI_7{AK5Ou3=zaPVkW2;J56O|qrj4?z@HH>O##_#C$5p5{<)SY_O#=CrS
zuU+oqWgGyv3_&JaG}dSDP^3)TI0TE7xAyl@$4T28)kh}cYH^kB&X(vTS@UA?)4LV1
zd^7RfK>j>H{m4~(Ydy2GW}M-6ck4OpWJs6Il0L7aN`;otg_CLU7o^HKvr3{d=|ulq
zDeFlE$xwJ-H3}FHmP?U##Qdjso85djwXJ1L&xe70?e7WRGf4qKMKtorl?AiUg0NRu
zly!CGR8{fd$OWA9rg*zL+S}Q9d3&JHlo`!+Sa*f_Fo(;IDgpvRX<6AXr_CYn+x;3-
zD_LS{R@NvQ**G@GwU3RBjiu$~nbOg;LXUT+Wc-euTYTUZZcGNvDeNZQ*x1;9K|x51
zi;Dm{8Q=HAmd9qMO&~lnl0`6QwI^!YGEE>ga<V{!(`<l<gv;XTSxijKN}Kt(+;rb0
zm))sE=fw|$`Ns%{r^BKKu3qCg^4#Q4<~8h<>Q&s4uieZ7fr}ZHGzm*6cF+7(XgCdg
zn3u<MMqn6@F+oW;qDSZ(8#TVwRyCWzK~MXX@)u4As)aMSaFZo(7X889(Y*&u?sP-^
z-%@lHE9;aa@mtH+a0z2@dyI)yJ9bE5+Csqs+6XH4>iA=}81z*OYCrX)4-eAP@vG3E
z4O=4n##SQOvju7(F4F@t&NNKGF$#yV3M2IQrmT2O^-D(FnAx$MBub3Y>j_{N`$I>X
zo8m4vBr$GLrDW*g<__U=!c0%?C#-bIfW>#yxhTf#_F%7mv{#+`azzgYvwiKEp2CM1
z$cO@|n5haO`X%NMch_7tb1@|?4@r4>Z%<B6_?)+}qiN;a=jW4Eauw7_PQelG9vp;}
zne>!fPBP%&;DB~rs@Jar>Hw!r95S+jkzeAy!ypBn^QhI}%3NGrEOd9&ad+H9`9VTL
zNmccOx3>@}ukBiPyng%i^kqH5PD|7up3{h^s9gOfPobk$#HVL<>zzY0GhdgNm-D&q
z5uI)hn{SQ$LMLL&7;~bvl!6et-{0M7R+wSHDT>~6_4f91H=+IGU%t&p7UY&VuKKzW
z?ISW!^G@U&IwD$RpXs?c{~<;D2OmFooAk*+6{(lM|B%o}JTfO^3hjc<1sGMNz02mg
z_&iZOn0_ZMrY<71gw!F=H*99<E{shwkK=WR1EMCq<s<4ATe=5TF0Klduxf4UD1)Fe
zMYlYZ>%=+$3{9Es%NJ=1(8so{J`^;{G|uL~R&WFc{%XgX0-Bs><1hNq&l*1VziQ(m
zoq_*}x?EK8UHLOKnEy$o`mmu~B9472w`b50iV8*xg(7uuEt0~4fJ(~Q*%{KA%4v=X
z{#@s}OLJ4x(P|qVFfHgOudrW!AB;}?USA*354zQ^u_gDpy^tuGh8;c3R+w{I&y>on
zs;ar1auK1Q?M_F1{rXkAS8*bOfzW*weEB^qA7277CwMW`53wfC^Q(v3Lw=_XbO#3q
zd|QoCT&aKiz!l%=nCrHt_>-BD$iAJoLrsi2QuJ=1cGpf<#ozv-mFaFP&5Bz=VB~TV
zB_re6BE*kBo-u8oD76Hy#WzgS%BcJAqZi4gqzErFEUc5fMGgeu_M94Kn#p~-7|}b3
zrCt{>Ji^5*p{<kfncEKJb&>>K*wxBGi~Jc19VaN=zdE*-3lg7?*ps>3M4`X;h6$ma
zX2qm6uGb}?*rocziVe+7hr|JxaEQ~p1#P&)oKJs<cupO5T%HH~)pdV(W-X^Yrw;uF
zKNFo(sr+KFlY1mQq#2-q%r1=9fbBpm$3rN6HC+YYzWw%PJm)0|iFQ%ImU+5Quw>RU
z7$ja(QzJI%D;gQS0{={HsGJ4JacSjhmoLi7%7m`hqT%=i1j;Q&Ujr*ED}81to12?X
zXIo01mj@WF{)i}=&ED584lH{0o#Cv_nir?vUQ$2D!lDxur3By8!GZnHpFh^H=2Y-J
zN%l_R{$UP6v0TM;<P6jPIB>4Mr=~*mOIX7lu1`#T@6R=~t1L|&BQwoZukk}#T7Ybo
zR(MoYRQK)d_=rfriY?|eQ~KYO@l@IEa;$R?*(;ik7Uv&T1sn{m9m=eORcFrIm!Ae{
zwIuY?iN6I#q3teef3=lIC{zrn#_WqQA>_=aic@_i#kr))L)5z1pBQ2;8sgUPr<3ZN
zsIpsfkw&_m&Gc0RUyKJ-J{!e}bw!B8J`EF0W7O3i0$kuG=rgv2pHiET$|jL?Z)>+u
zD8g#XlI}v<UQ`KQT?2`@%XsBRA&M#5_g+%jmnGrETj<}x86dW&MS;upw2eqQML2-$
zO2UIALq$25Hfv};KH#*+9R`)KF!*E%3W_0;+K(`FomiN@62d`+H&v>7X&@9Dq(@`J
z`@}yGng0F$N-j+((XRO#IaQ%&AdxeX--#AZUtb^Ub=dUuf3`i20)n2b)!?8Yz47sB
z@FW6v*wc&&IB&HculCGVe!*pDXTQ4EyS3W~pN6-0bX-~K3g;8hZ}^*^olUbpS7ow0
zRrGY)3?}oQE_xlZTTbLzbp!IrJVkpK8U{XHG4v|Mbp~0ZOM82Ahl@@9VWl~Fp+b)f
z1jd6&+*}qTOZ7w#4-Z$jx16qfMS9P#Y}HqpNx*ywL<v>x7j~+)&JOwd8O1|zU*oz9
zsI}2CH?nQgM;xr}2Dr>*8iKEtj<kM{bvP@bgY0~FS6XukWGzS@pq%`(!tw-{`yebs
zlW5K^x8?5Z2f0bp?|Dsc|LXBB4g*ylrKGI3sbURE3#0j5rJ@)c67>7r&9xa!Ij*!Q
zf^JgCjlyA@Q-Hk&A!wl(s6xM9W2e32humCp`JWlDi$@%?wndyC?bB#R!@US{uR&e<
zffP-2-+`tK#RuMhBe^)AHh3D#o6DB!b9pe-hZC}zL2Dhr#Cjx43Jcsyj1XR<_OFhF
z)|{N2tKjZpxb-;Q7@UTdjBjf;y0gQj0}t0O661hM!DQp@K?B@}4<B?@n3$QEKF^k$
z!c~454+irujiefSO9n+zQBf`u(}4D6um>iW6rkjkFBIqoYxdzz{10(4vLE2&si~!x
zY32Ku`zS-RN&e=W&DQg?9W_bSy3L8}ir|Tz)M=JSbB<bY&(sU9jMsyO!#mT7z75l0
zhFSX)L7n8Na1nteJ4&fLHimRj@8Y)AlBIFo1j%g2@zRMJ4EE>`?n0OHlpWbXp!vth
z2s%1?Tu2=OIF#P=Jgo(F0GN{4vbVVUN9f$4*3oO}Lfw+q`d_SNuN!U~Lac<{cHF)M
z2fA^Kl07Bazg+j{%D#-#IyyS?d7izMiX<}{P7}7CuP!R2sxu#^AmOq88;n6(d)R#2
z2jTX;yOOz7W8mVdEF6S?3rMVwiBt^>Uy`x5+?{aySjpPO`-(UH{{4HeIVJ|r+B#+$
z4h-z3E$JMlsQDxZZU!nI+M1e2yCp3rf2pI`K!DVIcOu&zNiMrNl==b{wY{q=Tk@1Z
zO=TUyt|yY*iUaE(`WO7!W(Zy-fV5YcbGdlWioyxUF}{27x*E_%;|%buq{{0#-+!@_
zztV2(d@V7_dVs*k5T&HAeSvW}B&(rKWoR(CSjf;SCMzaGaf2gl2{|?x7}ri3E!AKo
zx%esKkyGwamAhU#v_`mQGQw#0t*EHrx(xyjPV~&&+*pOVqUXh4Z$U|O82BwXIDS4M
zl{%+Q1U7fz$na-$w1Z>)F|XMUi1?O8CM>zd#Kb;GN;)pL2U#85Ufw7Pqz>g>WM_wk
zJ=9p38yg$X?$gRA5s`Oja>vrOCa{}WjOSofm<<{quXKUoH)dgTWu+b5%?W%CuPG=f
zCKYv6L!Ivl-%n_P#)Tib;vd5rKB!BgR=X<HdID2fMyG}Y16E7&wY4@x=*Y%klW!h|
zfQX3?OiV=%4wt_`5f^me!*<T{->}j}JJDhNh$E8%3)CW6D6WvXx@vsKN$$^@H8nax
z7s-FNSV0s*p~hX@nL^M@)QFz<(Rh!>)kgQd?u17e>Z!x&Nv=z7bLI8~W!2mjV4MWn
zOuMIkq(GfbU*+rwHS+hZXjWh308Oxr>8BDUoMK%nPUaVkI$pX^pDz=cxCC)E=Kc(<
z!Xf`}yVMPU6bK(cyzMV|ZIsxGf(RgWaX(Xj_MvEO>>!xMVe*RP#$9Lt={g9-lTZx>
z_6?pmTKp%e{hJ^J0Lw9rI)cHNi$Y0BnWLDlAT9l(KaP1d(L9abjnrgi@o$THy?Rf}
zbVmO6Y{Mrfk2uS@N_voA-~@Y^QPI){kLM~@IIKz%u;~5;1Mg#c(`{sA<{Tp5!@^$9
zr3}NXH}9TfzdwBTTn!!`2~P~X)cEaKnkJ=s;TEpghu<<lGy(RH#ouo4<^A6wY&V(n
zHrzfWV82s#<9*Y5pAzjvE9iD`m-A-!?qcZhqz#3#)OlN3GK#`x<Q_HF<mhqFY%od9
z5!dHtJ9n}gLiIeV1iX)CiuEA?tY`g^nt0Rd$%)1|=Su!}Bwn<2Jl6r^!&@OH5B@}J
zef|(1o%kPQ0C4Jm?ij@GZ{x}Kc|a}+Y<_!$IMEgq?KNAw$H<Zb4cES3lDl9zINurt
z&=K`hGBr*FSZD)W!bMiE{0G4s;)F{V$GX!Zh0{#8`YAy-deNyb1uqFC_&i+Hyua_T
zcTfQ^ypMymNc>{U`LcV^c)yKmz0Vl`(b=^~R^bCc(n;vokaz&!y@t4?ntLZ*br2hy
zHV~X`;0W69Ih1iJH0`X7CHo_NJZz!2S4fpp!qp)X1?xKpPGmZb;WL~{G+A#C*HvM$
z1ReDwW}n^ga=P&^>7zkf=H1-jn~y5~;xyItvd7%z4vhB>KJ$VlJ!^qA6fWA)v7a*9
zrWAn~B<yEm*k1*^BZyTv%4U9vhxExEb$50yt*)w=iGsi5@86GpD75%a4t2K=hafm|
z-k(#;)jTBMC!wIQJ)&F3z~5}?U3UaGuM`MI-FN=vJ`olPAnpaTH^YGhcJKvNP$x!J
ztUx;VmaCQUpGL4)`Kg`qr|F^uiq!iz@L}baZ6SFvL4PCGYOmdT7CEsw8R=p?h(C84
z&_3*lVA>jVi3(1$Ed|whY!{Vue(dZSDf3BAc$~dzkbXdpTe3sKr2Ui;eD!|9oP*!X
z8O1#6AVaT4tnnH~qp2M)?D0;%uUBTZN6UoI<4AEEi4m#N@foiQTda5zfK$Bo9nU3@
z8_oQ9no^W~I}2tp?_wBxku{p9#eeAYUc)<FG7CBAWo30Vj`S4jUxt7!A;%7m8T|Th
zJ^-`%3?t<UbmsEAnvv?x>f4JO(?v?y`GK%`mS0@;k|i=N2mP@deRAfO&=C22Z>@-J
z<5f2yI`O?ikPoljboACYz{#G8Jh*n7Vy*0K<-t=e|N3CC<#^uBPUU+%HG~YaG&nd;
znb=>K8#*NALL0t%E6(368uQ)X<jbW8DV~|<8@>W1S7T(-Mg5Nh^xKpydlA`Isy%pk
z93R!%v#phj=}LF|UC0coLtZ)Q|FYdh#=qDfFDr{#5>CUs&*wq70z|r|^L%stlr^Rg
zClZ`UF=hAbw|=BBFGeX!9gF2sJrAIQDycfV!AXE{EOB4GSXbKK-ac^`qy%K*bSf<z
zU;>JY!#~%-(7l_&cYahWn7IA;a4D+qeJl|exyG%Ug3r#uVZ1Z(2V4n=exNRrBbTVm
ze2ddZ%kesZY63tdfh19yLzY{sVLuq<zDymGC!Hty|CtO#npX?r*If{dDz&MYEqBD^
zeLZ}7;FiqJIqlR=@@r}a+@9(Ksq{X$E6-Rxq(esl#<BJXRZWuq<NIHo`Yi*69)ea2
z-zGY1Ekt~?c#q4JlwVsG5<U|QXpG^tcR(t=BbfD4xdI&Ls3a91g3Tor3MPDGg08*!
zO`e-aS)-uwqn`$}6_0f)J;td#+g+uf!ye907hRM5Or#>J6RN8ui0d`EKd@}$EpkrG
zwe$LjExK6A9Oa8~B@o^lVp?YLAZ^kGO@u}W4aW?t$Ib&G-#4G@9W#HWZGERfPpqZ(
zM`k8DpdFa`I+dEvQu9Z>z_j;L<=SZyYm6N|50P4^b#$$*TLxqAV=#UV=y86bk$HKt
zvQoodQ<3i(G5Pk`l8LeBJ}Tu8CxiC?n+ewBwh_Y}Tz)nAY}2q7Z?^r^ERW1B^Ioe#
zPZ17oZ6oj7Z*hRBr|A2@@pOw*ZjP1sNB}En%b!TPH*&6AS%lkmr>lrjIZeVYa@#Ew
z?MHvDefpsuhHC+>F-3wPiD5rm?mfQVOpj-(v7F2waBX?GVFc*~=Bf)DC8d`ojVCg_
zmrY)m&2~+RiNqbj7^_jD4~A_4ND)Nr0bnkr%CD}Xq7sZr-dFFk%jdF#2Lc{60L;qr
z97RppRa4A|Qozon;Y&+P9JnKwwBxuq6yC;idA?Pz+q}%rip!3puNAYYo|hUQ$YEOF
zlFNE2zNZaZ+0wh9G%-<O1TPp0@hl1{_d0mjuCQlxP{=O#coJp#NAHc0F3O`mJ&l|E
z`Vl4UM5>nl^T_y%x%sxex1enSIJ%zj_r9Jt8ep=Na%2!?O&*3(P|>)}&vrEUDxXv_
z9hk9=V+qV;-CvN%0tt~>;dH59SgdZM+C_05>5%xJ4QU%k($q&xxp;oD2;J^t8d6tw
zd|)q$tiQss&6Ld<t3O2Q$4sIVdH2P+EUTPFa&7~(__GpslR82$j5Ig^Zy+^rUL()|
zbAdLVbFLou%JoZo6~NM#G|x6xu~(fvihySO{#WpnNnoA2om1Ul)I)W&Ucr=fR<)@0
zbFuZ9TQ>HeIIXE<ORM%PEYQyj$t`S-z4Ju$eRUd?na~&}^{8}48aFTgj?a|z2jvqr
zeu|`R>7f=IFvHUZ*)mt4swY=`7B^wya4AZIQPfybsx9Z4dvzFfQB~_5>8Ib>9bXp{
zv}5&s+XkirTv>zc-Iraju4%m|N_Simr+Yf~5q7FBdm3=EqW`TArKw^)L2!y?8eI2}
zPER`)7b(Cw|I9%Q4j0%{wphQZuZI#CFaYk=$`>-k0{ejd8Zeb$(yNaF({AH0UqZEu
zHcr<2BS?88L12Lh>Q(_Ymw|y&%?~$L4_B*LaP_YHee-rLgI{0Z=4q5r<m8Nn)POy=
z_x7y8JY}}QjTJR0;LG>{Wi7kch4oaC4oYihC@#2YfvcrJtf#LuGecCQF6-*bL&&0Q
zcnqv`-CeClX=<zc8b-fb8R+ZF^twE#w44mnsxTXE_VFSkC;!p=8IOPfhk{}VM6jMP
z4}6Md5`chH_}<p10U4u{R8`C!CsnjLd;}k6Y-}qF+xF^{KUnOas}s35+w15XO{Vsm
zs<0X_nI!x3Y2hcX@zwUY#yi@;C=%6&W5ARV$s76{!KRj&xVxP4QKiXWz3nHzN@$a>
zv}c63ddEq0!zHTg*Gp1A^bKE*$cf+c>pZX)8zgW|zfhn(vUzviyWllh<5z;%+MR&g
ze~D9e>Aha<6~RP#&7Aa_+qpD;$ioAC<lHg%@e1DFeo<XH3%B1+#3@hp#_*aM;5#z(
zIp+A{56iYG+?^R+KhN#6NwcSkTQDd=F`>R{Z@F7LXd;VP?(beQX@i((@SpX=Vx!8E
z?yna)%5lbiHAKhwv9i{AmB70!F;iArVmuv1Mhd>OHl`dmjiO|AWn{jPA2QskaRi#q
zxRJRgpb!)i-_QDMbi2xT#uMOcU*^&zVYD&O3K=()Mt8(TYJ!o{-Z$etp^~1NP+U3?
z;3MlbxOOcpkUKd!MMOk=@BK_fOl)LgB0gVjb0~Wd2`&~tzp?Zj_yjOXlF9!0l3LQx
z)HFzrR{Yl|Y$Bq+e^hee{!H!tq>+_bSOhuIY}xoY??+HfTRH>|qrhOrky!X!)XU3j
zuG*#_gu<cW;a(Y(1_lPrU^4sjP@_m&;@vwqT3T9~9&;<Jr$kGu(cM8!F|}>w8K2J$
z7a}rx>+EdZ+8x;HVYJc12~=b-{i>>5f<j$>d2YonLBhgbb8vv&_`J-w(Xg;UJeXQ?
zbYzbS((3?h&-0Pu|7g0(sH~o^i<C49B1)<>NJuvV(%s$N-Ej#KQR(jP?(UM7?h>R^
zy5GUy|8?O9<yo-q+?g}y?6dbi=X6LNYchqSWqTs|-bR6Pd<F`+deJp(mCodslanw-
zYV6-T{3>?eBUdjc=-hu7SA!8;Oq0T~*rql-;)g4fEpu2C_-*j6km4tG9gbJPqQ*-J
z<SiHcDD8-J?)`neulW6b+*9}0cJV`s{Q}B&x9O)tT;58lZ~BLz7dhhCZ}+a8Nz|)z
z%;)(!=q!1KjWzLR@Fg}&3hFDYLyn?QO{*=F<nIF|!K{l{r)TWji6bO>-L$ewwj~ae
z44ZgGbpaiNahtf31!ayMDK@O4*rCmP2*CyH5sR91!I#kk>=m|LZf3Je^DTG_+^_k&
zS{U-1`X;WGT7~Z*&9p{r@I8V%E0ubw#l*?X2NQZ<VN_zFHex$@j?aG$aoTai%jk%(
zC74nkjO61pT@D#4%U7-tY;-NHVGC_Kr{&~lqyD}y%dScN<nt17n8@E^tzV{w0%R^o
zGT-8vOe$<wJM8Br;~4{py)V-T&!(oPl6bxF=u0aera~Yho`l6l7XurcQimcKaGM!{
zhVbOMVq=S)^Y&N(0_sa$Jv}*VE?it((0|m>;{q>KDdX*%H)&Z-0%qccM)E0aORj>c
zw($I}@Bup9!j6|}1WkTNW(xNtTSj)r`yUry`hVr1k;0m&K16~Oq2fjZ<p;scIhQK+
zikr(K#97YtWm9wXeQWoy6zXf9EBsDGH6bq6dq3q&-*h6~!nM?e=R{?+L7-as;N0&L
zr=iOCYkeZUIE%DPRP1;onxOc~hLE6b`P+J--2(e_xr5BT&`AH28gW#5S4D5fv)16h
zpSAkBbVB&!ReXYkW;1i)uB$id>lp=1ZiPp`K^kRDuMz9T`#fEksNC|WI)b^`lQ0so
z-!vl8nBadI@*u3u`7<3W9V!;n|FJGkyxQAo)eP}YQ`TCF^!@3p{az(&f!ZDWzZYI(
z9ilP}lXjLJs`Tt({rO5a$SD)V@lAW)37FeoCCS%DE@IQQU%<^rWLZoM6KP6;JT)tN
z2f8H^_oqb;%kAkI^4-L3@^$d<zq>D3JM7+CHYs!d&0TSk)pm^bN>!xbZqMMY*PASa
z->6y0ldKdUJ>Q}z7$ACce95_+-*Kh`@dAGJd1Pc{XZLqp4!6IgBB4Zo`lD$+iHJl-
zM+XE2A%mtUJR;)X2?d$aRtVbLqb8z!2SxV9al0|Lx3|C7tP2C_De$vFLP7Z*fUCVS
zsiM<8cin!m`0x_YCjiFlalWSqut3nw`n<JeDk&vp=j@yV&}aCkPsfWj*!=wb`u8@$
zU9-8Jn`(S_efIgw7yI>}&&S5c%WG>X08;IJcm2WeXBc&~34>yieSOljHslcuP~=Lt
z6Jv054&2|}^bHPXfu72vcRN7H#!Geh<lL(fC%@q}0r2s?dKD5MUuyz}(6i5hd+Y6d
zPmXpOP~s&o0#R-`f4tkUq@=*KevZ5(G7(QQFIV_ItvQnUx01@+VUX=oU|8lK!yhK!
zBtD~9<SJMGsKNCOBMY)Wn33i@b**tfCxLV4?3^Tccb??&4y!5Q9l9A?WR}5UNpHOa
zv+v6q+jsupdKq&QxFUN2dtK8pI~}a0Q$n({B6RsD=aekgH=Cg1`vGqSedMU8sEE5H
z{*P*j?uVkNs2L>Z39EOtTmnhD7VlO)h9^z^REnv81IxX!a!07xurFq;E<pg>;xC<{
z-L5CIHm%oy6%l(K99=N?%-J8#j8N@p^WEWAxgC~>zF^a1PnCvN9wdq0Zx7F#Vm)+u
zdIG)MmN?RrmR(=%7tELK)Huqh(_}qg*l8sBy6b|HHa(a=aa^j8ZUXiPzw2i;b@efj
zqW}s>Z<bgj8Z_r3EBoxRR|T4AoELif`U0)qcU#k?soLeL8X5rVRR!3Z#<o>cbMvS<
zjmoarGW*cr;45BF_RhdvRtAYYZFe%)kz}rmw%`}9UVRlN;jwnTzja&xJ0{Zk(~$A~
z`}ekvjztr>pxEI~DbeA9^C^T}ny>9n4&@e>)tVs#EOTK;*W+r`ipf$EHibL`aZO41
z00g(Vgvmh?0k?WJ(xn!<;C{yslckm5CA^8=RT+s%SYer|rzx3Cuq9J&^(J~A1ZLmL
zBYMX)gqF3Ch>9QeB>ng-F6q6u(3bqXRcgaom086lY4v(_0CX)QQ{w}(D+!%bLM{>5
zNT4(w8UinBJ}kC-dr3QP&XDz%%yOJKyl~sxR=fy1nt+ZZSW|P_=+IvxVd$Da->uq9
z1|Vz0O#3waP-l&vshqBFI6)f~DG7@Xlje8kjiGY}w59Ph4JO#@>YkR8%3(b<j(JAg
zb5qXNsn7u^_1(@2!kB^lPpN^P!6_7c706@9#qtM~8smyMIqk*=VmA2dSd%!!LVqBS
z9gx6k^7Zw-cBDt@z<$F+Mg$|_aVJa?uR4>?Q2l;e#`QFd3>Wvrg>|%*i2_^nJt_Ez
z%E&itj>)%+zk1<}mX^)>?8bfyc<>MhQHQXA{_Nvw^~rTl|LlXA7VUM|gr0Bti=<(`
zWg?PX*nLOwVXej0(U=tOsR7Cws_Y+{Z1r-MH;wDAr?+A7j*Y^Mufgt=!0Hg%c^4HR
zK1e$lu2lcZ_KIMq{+HWt1;-TCdnDV(Z3ezB+5BTo>)cLq(i>mEuYMqBn%G-bFFqyX
zt`I*!5Y+793YFfqgUlcnygTDuD6ZF_ES03w@gy%V#A8@@T0mvePV(9@eN_DIw!5}?
zNbj@ut>x~{tVz@G*Gdm}Sr;r=|9;Ce^Q3m{CZI<Dav%@VrMTownonQC6JLIhuRaK^
zqdsAxl;2qWJ@?CJ6d5zJS>bp&G)mx(OG0jG*@O&e-t7#b3r=aGnU3m06CenOiAUoK
z<h*3H7Mfr2b=>2qnh+ZncerqkQo?w6(!H5gv<9RxBs4s)$COsmdMTF-sg!9<cqDkZ
z-z9l|A2F$Ta!~)<!9=TbbGSd-^?^#<EQ3zVtST%AOtxqe&vZE%bErF=7JGM|LUFu9
zYd>D`Ap#%L`(ydtMAi#0fVsETMdC$HbfZl1ArFxk&V^jOdJoH64(bpI2nQ@>yrj1=
z9eIrLOasTt_$cwi5^er;P5M&M3@CEw?wuljHgAf3`$59Sf?8iL&>&3fd3^Z#Q>Hfm
z&tQvgMZ*6!mH^pT7i)kzP`3I^C2`e6UKT<gcBj_)4t#bh=P)btb<ybp*hCc(7A~x)
z*L7P<a+*&_aem8<5MwKxpd?zb;MLq6V_V>Qi8~ATFXFJGe3URBqm_@ew)@>?95a&K
zcXUiVdRRXl{yLhEDlH%z-CEnbt5=v`|Lw%++Vc|!HJFJD`a6@ZT00u+PhZ;WCAZR{
zUPC>TY?0~31lQfAUXm=Jo?|3=J-5aEQ_;7WsPEWetiEu*B-}S+JlI3Qrh}C*`nOr8
z$h#AJ3`&-|>Y|0_G-VvS^=U!^?K(muYv0ohv@#Z9tt_$=XX^M`g?g~umWza2ADkG5
z5Niq<)XWb5fT8mo0V)n}bc%HdmwrPx^VMQni#K_Qvq7)-?%jEC@44+<vML0q!_x<q
zA_#Cq>=g+pVh*zbc690nv0F<O-ZQByru|M$efT{Sp_TXdJN5>Ekn%(&dWv5#T4&oH
zkF0?Zfc-bs1bK_*J_qIB%5$T9@>6;C8E-OY0Ok>%&C47x8}p!MRD-U>Dl0w_YLlN$
zJNdlr#3sr}!|)2D`5{Lv6>FSVJ9R8GQDT^8e<1*z7Wh2bBO;$ZL}3vwCa2|M!_WDc
z9XG4y`JSki(Zm<Q_1#{dC4Nqx>0FnIQDIN0m5P00sdmP*C+&;N<H(gV_>Mirw>*f7
zf8RZNF3Twv4AiTDcZSG!Q&uKn8?zmz8M5lFy}0#%ybxxeDirKLgP)0n6?F7Op%S7Z
z!!%-?YkFll^QJIQFLcB5!4Z8=B4AWoH)H;Af{RZC-9wQ(RlwSub(egIVnZ=Xv9UJw
zyCX4cc??!GjGH-oB=UUZPww!Zob6V-HCJ|-A@}F?VzYd^!YB}jce_-5)fwrYO!W1X
zkmeK3{`Rp^ywH!ti`RLG<t8P^z4#}6QhgD;y06wzlK0x@1+J5`gc@2yvO9C&^`(U>
zTxTq=p>wREK*yp}?+R=tRjy+SfCRt?JV(nTxobJ8ZSV-T|CVJ@MH_U=)K>&KJh9T*
z;XecpS)|j)ywGDD`K+D=I@hK`YJ967jE>+B^W-uqsi_ObSQ+EdG&D3ky}Zn(ie3X4
zw{y6}nU@!&)_N?aa~R+debVCXJ~N;>-e)N2_aYU(7qpyFCeY$Go3@f=`2=#^mocoT
zTYL#x>ECrHtz4x7<+9AEXX1P`^(4Wtg{b)|Q7C);I~81=<Hno1#P+A?Z6kl%zvKx2
znv%Z8+rdZ`vh{>0D|2}}A!yE5Q|>%(gm*+rcQ=7}wCKBku2!3L;PVU`8-Cw}uubjm
z!H$9n@?tK&;O`nd5yJ~%Sj@Cel-Nfy)SuSndzY~fS!Ja2Bdu-tI<ITJ!w>40hIO+$
zHK0tRa@nkIl3!1;QJFfQsv$V1NHcG}6kWe(m`OqsbP`qXUGEGrlq?=J<*gH}mx``O
zD01EAig;%&!N+&V0y82%>>EJ9$t9|@S+Vk6bhw1;Gh=C(Nh`6=f#_0Ara%__;bAT4
z=EY<5^>mk~O^z8Wh6c({*D(I3Uo!08I)o^xWyNG@B~N39$D0jBBRkRa;AHCCdf+F%
z>(AT{WBG`axuG(isGk33%UYE5>rp<khO?#p69h|x;OG?9B5l~FBA2Avf@_o#;j7a`
zYJU3Pd7IEjlsPKVYJ3?jp6$Z-*k0tjzs$_rra=njh?ne<u<`B|!4_l({*&kg%^G@}
zM6%oaCrsEgtw#tj+Hpr>*+hM~=B2Bp!@t|kl_=SIBr;98W>mZlq7iw6d^}maQXzj%
zqO$?3meas(dHdQH1W#<)`sbbKw5z$s40nvU()$Kip@$64J4uX*jnh`5rR{68<$6y%
zUn1+Uws5JJFk)o$o5C8ujDgTR{&X@XQvHx}Jq{kZy7$Lj-L*JUP_QlVTvW8iZNnUe
z`78<5Op&$`eev-wE<c(3C~S!`7~&^`FP14wW1+&hCn`)H`C&cqcoNqX=d&cSzyu%e
zC|ythx$-mF`i4;-=CSigq+h1n+^bYzVD5wLY6vY;ta#tz2J2Bi^aqM4<!5Cmecj|0
zE&(NX73Z@N>TD|*JCw$>Twa*BPimR0CQ@FerAJ*B74c!O>QXBYvsMvnTvJ{K((*CI
zR%369^)1G7uL_NP3l*b_#5#8xT4{Z5d8f+(isgR#>9olTyPZ<7M9GAn6VHCo*agqc
zwJ+o_{K49sBJOC8huDfm91Hv3{`cCDFvHtZYXMk}oy_T!0eEf~&DJ0z{r-~HbTk9O
z*!UuJ&vdc?2lPi@;V%=;R(apKgS@of{j4Be@CAH&jM-Gq(~)2gOeKES`Bv1oSJ&m@
zR<Y5q+LxCdN$BrbwQfSvdVM}j;^)ckgHsyEZIKV1s{Y*!FQd=SkRqKGI^t*79PR<v
zldx2Rmi0-T1@7#5uguujY8gDGVuH_7$E-0Z_nIes*2|O+pj}YgnyuZr@hk>`I6xwl
zU!^DI_h%A@t2G=P=~vU)BdDLz2vE($<Gf#6Z>U)FJ+Eh2V_bhRy834cB%Z|!&Y|I>
zG&zt2ws`U&<yBakd<|K9(k_f+DkqW${UXITX3|(;qEj~;_7`R=ced=&a`os6QDe=|
zB13I9$sH*vwo#sWol?ivrXu1hT$brg3Qut+it>q|yngDvARujTTt9RJv7&Hcid<5l
z8;gi%D^O=3u=aGyurbL0bUWw`ic{T^L!F0wFz%Xyi#|J=RaqKEG+vmSV3vZx;PRDg
zC)F;y=P~a*N8B<nwBa$C%+{^J2V6jyg0jUH&%JAd{05t-;i(=IR*>6Ef&+IjUmMEf
zalW!nItjA}UScf0PEJ~y51^sjZVbK#ygWc6%GIc~hMAe2wF86;6jW5G9yBz?>{YYn
zN1HudqQv0_lini<6tn7g<eqwN82trj@?1#;40BJs9sDdmPkqA3ez8emd6k6|kGu+u
z1Vh<)H9t#o^IofN%{Wl5<#>lmW=CTsr3#6M{)iRM_HB(^nPgy33{trxZ@ST<%rSoX
z=fGgmCRO6BR@havFCQwfX9-w6FlE7;qP=fg$T8lCX<{=$B|5v!{xn~xS4JZCt(lqa
zWOW?rk{Tr*ujj#pZod(;gn<-95ep_L$7<SFzIjv(H+$^KtPyDxlj6~?RGdAsd=xw%
zOGFA&O0g|pD2WE9i12%NRFVLKkgQG%J$b+8ydPb&)tZ+K1dw+fhB+<&h;!D-=UgJ|
zUpS^*1zgG6yHD$d<Re^$oy;m)$6lD~3N|?|gx9;)9Q~Ok)-EnP*lAC_k9yhTh9e`c
zD%2A5s}-ZbldSjUA$4tjfIkcU?e$C6l5=_5C1gEY2Aqj(O%mE?DR76{z6fmLG(TS5
z4h>9L?W>%<D&&ViQc0n_7rjFQ;-$L%MM!(eijnkSzNmH5SISDYJ{I@tsK1!e&vnt8
zlc|38uXP=II3X$im*Ts=GE|3CiBm3|e5}?1b~KtU_!LmIi9zwVENjk;2QY(~R)`N&
zFpq#!fB<lO%mA?R=fFU?GSD(d#4NdlQs)Et3n#|_Y4H9e^??CU{}M1nOww~LIQ`}X
z3nwFhoh*PZPy*sy%E-;C)oY5A2fn2)WtY9dG-TA=acOI(0liWal#{W(udM?5mrs}z
z-mTjs_-iM$e+^32H_YlC-06-IHxfg)Q&1RZgj|p@d-I{1zvH>3Vev7P&5-PfK*F=1
z?W*FU-MFIF+}}ksoDX=Cf)44sI>vGqyp{T4g#*~-^7;auik%8Kk;lAKd4dDtSq7ZR
z(i!MlNd>f}$xnE_0szD<K_54e9)Mn#Aod|Ee?{E6@23VA@XokB)K_=jQ^_u?!b7+4
zzKcVbFwUft7YlV_t46eOfXuS&m7jmwY|ke6W&VNngF)A48CT;8$Klz#obh~DO>pTM
z_8pb>NexA4mqiJpYUCMEldX+K*;3(K<G|>WvSn-zrdJV<S0ZN8>3|bm12rW075&B(
z!&4R4aqf}l_%h$kCqj+>)3D0Dg~p4|wPhh_<YR*Bej~<soA#4JNMk!D3!ZA5d*}VR
zQ?i@G#U{>_+mKYB{&<y;CZD~Kd%O7CJA%m%(Zabo)T{b6r*EI?9{w3zIn?f{3@8nG
zvQ#Rn^|?Yes6K#$kr4B`W5d8Q;U53g?u(gwS2fFX?;klsc|@l-TLDj%O%|m8^jb4z
zhu2B?IkpP3o#@@VX+roO$2&2hj-eMHB6Y$-l}YaKjr8R11HW~w(E|hBfbtJLe<uFm
z5_u8dG|L}cKbkB5&B`S#DlY!OQcNfl4C8q=sglHQ=ZB8ZI^js^_eK%h-=vb&a;EJO
z^uGKPtJ5zP2GH_y-F8?YoIplJH2?_K4+0J>0|SGNkyJ=VR1^uOP^Gkl1zDT_SwSd3
z>b5tRGZaeoELE?`fIABRSTzZ$7MHya*Q2AH6Mk2v!E)zNq4STQSZJkAvG?zhg7ra0
zHVc#mc5LrfdhSAz@Dt-kGj_+(UNc1s`~HzZbP`$5jszw(;%AgjQU`ufX;#)!OWTqq
zS{z!?>OF0dW&H6ZXGsdswixNy9n8=vii{DPZpw|M65WZMoSxVyv18RUGm?;KyY3sf
zqrK|<LRQI9<3cjKsj%uf<+Z;@u=A^YQ})Y`_@K=1BGDHo?Dj_`omB?Qc9zF<7<Yf!
z0KP0wxDLt1;jPVQBae8pyktt8OP)a{nDB+ZEG9Y_j$+W=x+M+%RQx*FTpAn{K!tt{
zKXz4}o40nTmhWvtgc&OiS5-O9V}~TKKh^7Zx@E}tdn}6)?N6Cwh2HuY)u%g$AcNNZ
z^`2u}<_B)qU`>3FQGHchb&SJz(aU*4;t}Kf1+oTB%=`PRy`~wMI&@A>r^$C!(<gY)
zB+FA31+0>;h}abW0Iw^j_{TmPNXY|7Rt{ZNN;K6f_6`qq^gQ@l8~fs;X;c^ixgG}e
zDK!3G1Fi<ztMoTW)ybgjr<b|$wNu!zC87S2RflhAl%;EA5UM~(zjIUjiLO-Zp$v7h
zbFkyvW>0gk!`c!FFL^{=9MYHqgw4QI+qKk32$FdvX}=uO9&ywZ#JI4L0Rm~0hy=&N
zvBXQZ!!8ep8riVafohsuF`1xD5PSemQ0KwMI1~8NX~E0&?k=E-)q!_()kI9}D;XCJ
zx`l>_=ECmkcp3*IiNA`u`}->5>F?)j?%|P9eF4S2V)fx4<yug<=qVsiE_m5mDgOkF
za`8b++N}>(3_LktI6QUA>5L)JGkmCK>}&|(Igo0mAt=AGi9hZZQxd!-a7*BKBvL!x
z#vylo64rOLJNKfsM2c!~nBw5Klzjvkoc68EzNUM2giyof>D^iZxf*ge@t*6p+d-0K
z+u)Wpa6OReV_~Fm@`AGa!t|vmNon3{k-Gl3^yY5?C-D4m8LB-I=RWX0L|XQ!t*)wi
zuO=-i$t-FK?s|1YVc|<=w2oXoS*gHx6R_4!_$$gP4_)fk&c~k>lo2nnRTo60mA8>&
z{>H^?jAa?A+!^HgXqL3mr=fd4u`2M=3B{+u)tC#xNQcT~3$zLuly&9#sY2AH$UEzO
zxrWgf&JGIR6uL4uM?VpteIaA78Wyd=-}y9NJ#lS3{r3}uYO%ihO6IouO8GQNs!y3y
z+_FsdZ@5~sg5RFQU3Ag?my%3SgOmghf4P_|3gF+x#a8wE4i1Mc@vScN9<wu1Ix=!#
zqe0K82d0+P)PE{&FROjAdR9&K|73!Rz+(UJ_g=^XECW(Q8vw1%>L!g$ObE!2<^Xr=
zgvByaEaCyqRBLl=c4l>zZd)peL~lX!@^}GDS4#76vBUIF>A{Z*Suw&#thTw{xiFIN
z&1LpGJ=8cC91fCBFWWw%8s;%nLe-Q`0!<O&HLnqygkAJ;|A%Vf0Irb7=~g>n3cw(t
zqZ^%WjsB@w2LJM!UkXo!TzXD>g7NX$#lcxw6;tZXY^i#tbT;aTSx0^**Sje4&;p;H
z&=s(#yL^$Z$>ph5kf?zY=uYPtf644SQYE4+UU9U%*Yu5#vL4FW{3^Y_$V;u1@<et6
z$KG?>I)r{vO50sJ-)fy4CE?sgqP7-MU!%{P_$An`W~=H%m+CRgb=>7wpln<k&!3n)
zcS3gzW>&9D36!)}ER@jsUqaZ2ma>w}peG>a=W=^_TxGqO07#2bbF4sh0TeJ%o)6c%
zrLpg!<vP1eLk@QKnO}O3HGm9dsxbcyXI+>Zl^R~OaqilW>3s3J1b_cZt|jjwH2kbL
zuY81y^``gu*jp?^PyPJMnOBtsY04Bnmf(SM=@@OSgsl83nJh`YI#usPySE`ktKMtc
zVj8Kw9QLdajdH>#KjbW_bW=e8t+mvS{x+m|jI=bbH+p+~?FqnxHc>>t67QEzD3YS{
zXC~7e6!v2DKfw>RwcABbxjqE*!FMw>JSW>qEfl1S7cc|I(`qXZ2nl1%_pT9Q?{6O8
zv!O6!RquGZC@Kn}iF#uaTs-wojn&tn?36Rl|AjVly(4&=i-X1O`C58tzSc$rB)?T=
z3c|s7-9N)g@-=Gd7~_A;%&00Ct9L*4Y->=F!EmpU6tZQT*@|pKx1*KzFBl47^STPW
ze%-gdV8IR*C!<>Y*7XBIisB$RTE_zxO-W_)2~KmR=N1%$k3EWf?;Tb8txb9N2%8_c
zpN-o65*K)hhf4h+K7-T#*X-95RFx1mssex^H~*fGmytdBvZFffGHC^qaB8E$FZgu<
z^5t8K3sqNxC){7fIUJ*S=Fx*=|7|6deFEw_*}NGLYmF8N&97d&GrS9VqB^4y?(&|n
zWRw4{K<WkZ#}+(Ra^MS<KelV3F{80MUmxU%d7BIm*WUuA{Go0UmI8I*<IZmno8WpV
z*nU<ZxpcFgfbsZ%-0d`xeKUaG4w}rmV9{WTRLlKvXq2PsrDEx{E`k2(Lj%Y-U(%`{
z0cHivuLdUz!9Y-y$XX;-YblPu<+j6<Iyc^YiLI!Cl~xD`uUKBP-fcOP@b&SQTlx!O
zP0Qvvfvibbe9tYR(<Hd9*)KfSm-R>x`|_TX!$V&6WKO23>0WElvR|4PQHJPebh62>
zk|x?<v~$K%^{~M4l~m@UBrCHM{R{g<a*>{Ly}$JG_AZ`UuET`=_+jSRK7f#vP3Vc$
zo}OB}y0S0ut1`LZr2&g<-0N{bHh^Z2uhhYB^2CK9SGo)nMZLAhW>jL{UPc7fpAqXm
z>(%RRUf@q{wbxet%_#^p1M|`LxRa6GOEyN5|1}HzWE;$1+Z!4i_Y5a<%~o6B7OPiN
zS89Q-4)k!x_7sDlZQ=GFLb*QO?gT2Q*FZ4f6{Pcc$Nq|quB3(Dj;BTwY235JW)#Y1
z{#7rj3RagWjTbzWW+g+a?y44yWpBQQnZfI{eE1gG=o0s;9Zkp7K*Zb;)}wV^(b~a;
z&xNyj4)gECksdqg3uWG)W$!O{@L(PhK7*4A3jA^NDobw@<o+fB*HJ+{TFqH=IDXb6
zJvfGla;quvib#l&n%IEFoF@aH$d?z?C^|yE@(T6n)onZTWwgksPc7W<HtX#Z8=p>P
z?o051t}Jg83RQpqQ#We34NVba(zR_F;dgATmvztnr48`y9}8$5s3ZHfW-BRjrIX(Q
zLKR5v?2lIXF^IT-g8bp<;Nbe+Y~|_xJgyp!u@&_4ao-6(R$a7#F20U9k}dG_ky(A8
z`!l9GYDisgunY@K@ATQ!o8Opsp;rG;9smt~@qx%4v1+rz4pGF1V+2$VlPMxn9`gEy
zjm2f}FTy<jZ%SauwS1tlEK)-?MZ8NAjHP;51oH-&*Po$7P=BuV4>NUX)jjc!If{T(
zCPSX41<;~7XXkzFZm-v%v3_{=Q|blU&6N`n%w{{V?pG+5q;uzs#}Oh+7=RRx$#Q9T
zrcQ}_TkX&#P_EU$pe~+XE7M^6k4~ll`mHzW66VwBFH46QowX9%CuaWzW2EX%!7Iw8
zxw!-oWJUqJ8CcHsxL$d2zzcCa^+(5CcL<!r!)K0ZJuU9f>|&YsNs?KlqTjHFjoMQT
z>y%(VUQqdOF&^h91nWE#54Sygi5zRt)(hLfxib}1`~l5gsB(u@FERA_?S>CyV*k1@
zS~8dKerj<zImQjMfOwa)OH@BEdx9v*pGcHd$DGrjko;R=dy;E!Zpd$nq(SM`aQ#_e
zRzNH_4WIqRvi6Y=A9j7&pI2Udo<#LxL!rXe&gE-=&=1Q&Km2!xbz2w1mM&&8LEE%^
z0t{m<l0NFATHpocgzmonz)>p2Dv@fb{LKddbF5C+&Rnme4BK$S5(fy9&}`$el*g2W
zi=zewjO8uzP=yshpRC`Tv~6Z06D*{wkNE$0qS9IuAQzg)rDOSIOiFTD?0e~mraQv6
z8^q%vRiLEyrJ?u^DgKxMG{cDXqx@vu!f!gV0(3<dBhy`Kbes~uHg66{9r;Cl^faIP
z&b+sj#ItPu2DE&>_S05+`CAWxEDy=)mWKvSFd8Z)z5c0G7A9!XNKGt^jE_)NCgjGQ
zj8D;Vysyb6oTc74^TGLy>+w1Rq=pEHlrSVKc23#)eeTeVwqwfWi<jrhN^_wLpW2yp
zLek|%-aG;`QAe)%+@D_Gdeo-%PPDUMH7J?=ud^sH^z5<BfnN<$sjVh9m~MguJ0>+g
z?eWNTWWTDw$|n6TMd-1R;67d^^2Q0z&5x3XbV9!`+-^)ji2~!c*8Qa}9mOAYZqw@M
zhQm9*#NVKtu*0=}*0M<;wp!aK*7vC-%m6gBME+}NjrLZL>1FkjXDZ|HD`?1WT~$tZ
z;f3Tz+S8^_k-GAWx76T6<B>6)jaXi%DB(BsVb2o<tkqeKSzpf7%_)GbecQlNXs>_4
z%lP`fS^Za@h=X-<qjcBO38_d=cw|Jt!YAMAO4<}!FS8d|pS`q9ItkPM7ZUd$+b!T^
z!vK%~-(z)7wcuDyl~U@Jk-2&|UXGuvzrB1#7Q$15JBTwv%q!0tH*JuntE<8uySnd(
zp)2!6duDy8_YJ*q*;;*WD3HM4-4r19Y=Bw2qe)+Q5>$4*4ed#q)UQ@2`%Bd?NdA}c
za8JgpC>2ibEw`o?Yt-5U$P0)aN6iNkNBOZ{zgE+Fiwt&@K(V}H+B+@|IHeXW8VY=9
zaK9+*^=t3XwsO7^SwGnst#kDcw?8`wV!o(OL(CtpQb(yc=@T3NWNi4k`ny#&c33D^
zfZ4H=lKJUjQC(<4kxC)MhZg#m%`Yy}8CX76s6IY5_%aVSck3m3{+z$odhu&$Xlcm7
zKYn6(bZU5KzS2{qWv`GrRJH>R$VAU3K&RFi9(`N1`TX8}UnE|oZMjnUdHn;d+cxIZ
zT$N*0d#lPbFJ8~paItgpfaQ(TIseBEv#~#MGsSUTH%6JoQ7jB^aCo-Pt8Kk6ubIo5
zE}O6?sUnV^m0wm?D`2hZa1rQ4sA2XtUG`xZE1Oq*IQ+lY^oE-pLc0&)VPcAmq1A{;
z^l-5c21(CoffCJeTp{?(QP-Gx72x?l5_;t06y7v%>7rR1>zc!xQUBGqN1k_#Dw7h)
zU#9ZO<s-~jsIY%{sIfI|XJr;L-v(jXHQqm(pyJz#@{{^}wIpqaFbDjQUj}o!oJ|Y!
zJ|Y&X2=Q)~(1Rs92wY-kx=*-=+WO#`Kia>|A4Mimi)t&+ZUnbFn?MfV_l~_j{QT|7
z{?N)@jP`uAt-Ul4k<E&a#l^?Hm;2mFpw=RzpvGoY)XnW>;fa6ZqeAmREaSgtJaYZu
z>gTf5<VM!Fczw3(aCxMssYSyV|K;)9+}4odmAF9eLl4L*K<)`TA0JhY%EQc)m%<g`
z1IP9L(3TLl3}&>aG6BK&PPK(%J)s$M(W@IcKsgXyuB`~rJNPxYagSV}h{v9YU}2rv
z5;s&rKW(?W;PhoUS+~r9HELoDUb&SswhCWNH`8bJ;8!Oyfz*M`D;*@Rh+meH(8^^6
z4=7lAPEg4~D_VoecRCDl5jOmF-!CFbf0Z4-=n$K;dvW+QInlEz|KhKMqQaSBZ=_2(
zU?7TiRH6naN3y3Sn|eCauT6B=IuZYe1|zWy{`^G<`Y$zxZh!~$uKqEYa*tNw(WWzj
zG{Q+z&zs4$k$2)j1+vG@GSnnCQ6+%)*X$b^G&-Z<i*RMBk-t&D=^2r?pPoQkvVp0@
z7et-EiEM=F0j)8N77N(%9e;tI=>6QQlV+nY9HwxfL#PkBN!jbxTwJPn@PdoPk_WJL
z@g1+|RTl_ui@%~awim=c1X3@<gr4&WyeBL+<IRdZ)oTa*uLdW3Ueu5rgEe!R@%w!=
z{4|WY(A^+OYj}&bH%e+}7!&HMuGl`(Yd3y_M@}XGQuw^)v5HEy3Ip6u-ejp?pT+7h
ziN!iUl>V}s<MOO>L@;@mY0mkH|5qZQCOm)soNNFfPE3v$;PCq$iE`AqQm8n;pm5n_
zjmB!kf#iO!oVhdfV=MVaRAX<X_tW#0JGoDKcetd&8NBwKeC0<-vlNQ9wFaBY3`_Fb
zWSaQ*P6$&pN6}91zs`zbcdl6PU%SlyU|}alUmLizd=_LC{aIEErL$i30mnpc*#t*x
z1=u}_{J9=r?-b9xY87hcZy^+qDnhKH-ueteb2E>_c@m%$XDfwziw755VQ03*|LfQA
z<ja5J2c&8V@AL>6ZyljKIfK`Hn~){9W8C*qHqpMpb$3tLKsxQ;-zP21xVEjXTW;R!
zO=ntg8uI^&2Tz`ol0r6MU%&8c5fwan2MGyKY(4Vwj9&9vE1q4alklV`6TXIO6*$U0
zWoGx8@tnPywbi|wiQrXLtoVxevqnqRFLzs8RjlUCvC8RxY-Ors&$W+xWjm;<kqaL!
zf6=uVyb2Z(ts_wI@|_X@H96XqcN4Sx*?>ja2QX<vd0iX>By;i9>O91fD@^-^JX8HU
zEDn@JLtbyVpXO-!MT!4dSAxuShn;3#g~k^nbZ)eUgfFbvc>VW|1N0*mwm6*-Q;SrE
z_|qfgr0^@ZXP+!2oT6}yFTb(s|0Equ*lyBJm~UdydB9|1_OS6H=8$@m=fWTlK;dz5
zaba<Bs^jTM?u?g>3t$9pi6I9U*Zlvl#@gu)+}LMxFDRLY4WvAVZ5B(qq-#G*xelZg
z&gm(X4@BZMkj~zxrGB?==iNo2%URDcp8h7+6o_jb`%?QnLR?xgy7)sA<)j7A?XEMN
z@VTk!3{M!pPOwl3gC7E$-d`d1lqW(D{io)>yKYB|EFh81q5?^dZwoPxPNDc~O37fi
z`&hPP2(17Ubfylt-?t)&Wk;?2R07>z*DLNl(T=gcFVDY|udNiXjr<E%NTE?(r)Uk*
z@x!xu7LG@@YqtX=S7-StrzgdMkpUcl9Q3jWC4rNRv000zL;H0fSQ{!}GIg<kar<|A
znZ*kUQwqo=sG|#VI-jAntwa_z-(T-So>zZjVfJRq{Q;VX=Z^c|ApzHuwEC+&5gvCL
z2c(L(50c~`ax=w~ZTD}#kc@SV(<$L^yVDzxP(5!3ygrrv-c}ijMT02?KtKSD0kLXl
zE_T3O0YSwzGVcyow(wgRKK`whOz}&=s>9%Qnz?Sbk0@FCf?4Vp0BK&oX&Zj^_uGo|
z6H<vxy&=o@Tf9ckthD!(3qiL_l~es>>~MVcWM5eNz6EpWuy*A_V~p~UGe_PnLNPn4
zj859{#>q<jVMJJOEvMIu^f49aAjuRh-l&)1rC{X<Xh-)_<4nWM!Qs}jeizFG)&^E9
z=~aauP2?tRT96B*#mQ#%H!R~N{M~C$_|n)|-1PKxn5OHULQ)C}{QuTNb{7XH?`xwi
z9Knxvk-$;Vq_cSXaWMa<yh`pT3Q<rRp3ci0p9jU#u;ptw1@RCfs23}83Z!yp))T;5
zQ4UhQp}MVU|LqY4C}x}0L&P5phiO2I|Equ%d~7AODvFmh;Aco?E!DgY9^7U>St>Dn
z(Zb070wW;*gVz%AFg_Pw_T7v&Ahifu@+S5BWNtqT{~n*sF_dC9BGD>i(q$7VWdASQ
zN_jKjvcJ0ZhX#Jq&e-l@35VR}aB{vkk=PwkAxrsWstLigDmvc|6qf$&ffZf7z2h~I
zl?Ly_t-})Z;I%VVJixZ>DF^`hnwpxQ)R(#KLqkKq{$))ag0aa-Z)xiiwqvv4ZCRqP
zB?l>;oaB7P=R^n<?#|bY5q%h2qk5N!GxoB6uQ`sNxUL-3OZ2y)+uYrF!1t{SGZX#_
z$$^_oSAGfnw)nwD$x${|ZJg1A%&oM?-zvv9gIBYRFAPemh{PB{X?sNTY&-x_7A0GS
z2dOvy+m&5%7R;X4jm+%_5AIH#nsR4R(*yicYR;>S6s02A?w3e{oGy%xaX9jYlSMkM
zDZmKE_IDRD=y4vOon_lBxBMXFiUFuyZ2dkU=*IIwya4*4fOvU{0MY(Y6_jYx66D{x
zh}kTY!Yu}BSMh!(S65eEL&u9d;Hi=53N)S)Ez{=Jw?MA`UbW1JL?k3l84Rdy;Q^L0
zCcs1B!m4njJWr7tV7DV6)tIfdA+EJq?&<3b3JZHiE|Ve&H0w^BxiYW-Tx@P`J}g|8
zlZ9m4!0%Pt-mm_RjrCX<d`Ro3Pl4|rgb5;7FkAO6Nc$0#U8OGjYFj%c-H=bd2(MbW
z)%#S@J_(V-MV^>!dJUO^O-^@n`*U)xf>faXE5_F`AycZH%69tv{}NFmE{aZ)uTs+K
znYmp13r;lQ<ZwAP7Tu`n^QzF%FF{D=s9@6E&)=WJ^YX~6RHx3?7ZsQ8Xozh&EFr;N
zr{ZOn<{XgK0vh`;_%1M5k%*-`?#9)z1$h+k4biGNSXxMa^!8F-o?a#d+6P{*>oVx!
z{^okA-hl#$XR`o{fPmc=2KD<*s`|tqRj`shz&P*Ff*2dSGlC3~P^~aT297?)8-wwf
zNo?tO;>?d{*bhc|xMobC;0e}uGkIcV>6+{kR{5X&*}e-IH!3g8S425)U!W5QbVf-W
zE!)OHaMH#;e|3Tu9Ji%XIwtDGSIs8z06D3Ab3V8z@svqI=Q2U5^fr`lS)(UlsQ2f{
zV6i-;9Kt##02MqT_Y{r1291=_R{DP{2A@7P=cb-OS=X8xs(p7_pn{8C-;~uFXa+3X
z2zWi?e0*T&bXtDM=YPr^1!Gg&{tU#%vs$U=%k%&mZY*#@0q#O=fK(0q-b6A-!{X!n
zIs#DTsG|W(8?eLSJ|?nS{W7m6Ltd_{W3PhLt@>0R?(Zb7z1M=-9=o-`{Dy$b87*2|
z2t;cj_2}&F4Fnh32mLN2e}eC!p_mraC1h{kK8Y4rvPT2KN>yF`??K}s*7N6jKNaq0
z7%q{PAd`)M)rJ`0_5}qr<G7XAhbh1-yWjRTziWQ(P@t20>NTqoksL);4OtjakU6vU
zbQ0Tz71BB33t+b?KcRPR^WhvLwTP{JQdVHds8~2^Z#FM>?nNx_F1gKeF#<?N3#!tE
zxDAhT9U7LgvA)_m&GG0x@oqV19{{hP#%)ReR|n;t>0?F0nEqbRQ{hu4@0qABC#UA|
z-@ElJ8Y0%_D2$|lWMIpHM(?H{H!dMztj^A$<?ej`X$^4a(d&<*0wyW;I}=oche6Xl
zJzs#&MkX-V2der)<>KFf9Z#zU<yHx(k-fyi##XB`l@qgkOnP<`*ldPD_tDnQP7j<o
z;2^ZRxtTj%4ip%Fe5WlfEnPR_O)hT^Tes(`X&|uM-#g7yB!i5}YOd;oW9+|JvlR}S
zI#WMDb+c;Ad1^OROdKto@C@|`)mqs+@;Vuoy>*+CKoY@)$t=oRvPH2Xy%Nr`?5PcE
zS%jc7=q8JONEQtQS%3?Q=Yr@Q(b3^zMn_pxJ%VLI&DSHI&1~ERRYBaRriKXTP)}It
zmpTFVl#uI>)PU!>`g;l+E%AcWN#iDZzE(rQ=6$l|&*ms}DsfxJ{0`(skxv+a5jr(-
zbMV#tzumf89u-6DRd%Rro#Uc|v0)sBk*}tP$pSQs+68503=M$ak#7auN&<l~O|r{U
z{i9Y2_|(JzWmARG02FhB)r12~e=uQ2fT-T9Z;^|DfWURPxc1T3XST))-=_I2M!Q!&
zrN(L=w#@6sIedjSfO7O-aWzS5Dk@?IiX_Uh@}65AF~GE?uQQ7m>o2jv09;<@?6xda
zKt0N)eUI`v6i<NO#MG3E48m9A*TKE4dVWiQ^kfQp7LAJC363*hof7t5(6W&lX}^^a
z@gyO0SsYp^>^btne)XH2%zbePpiSbJ(iZ_nv}v_WpriVp_>5NWDSga*E2xQ1_zAP-
zI$D><_M4u(kF24s$9Q~=AQdneXrDvzsk()2zG)Z_%pB^ZJz+8Y`Rqq<a6Gdame)0O
zss{?BuAZKZ(I_5~p5*Wf5b3(Qy0Rupx%#FCV(C?=Qot(0fOxKAoCq8MVQsHZ&6-31
zNdEx(z+9jmR8>=pCh!LqFEA>l+QnlZxT*e!eH4g+g5DK=GVrs2s%zyA;lf~G_Xn8M
zIM%E|<8$p{*V=QqU}cE9k0A}i^8hmi=Lc(^&68#3FGSK|WRr(h6NLx>TQaiAoUCx`
z(Z{0e742K7mr55x$TT4Qk^Sj#_w3h;{8@%8T%f1xU(6b}@cK~A8VL9{uI^WNvBnnf
zlKR_*JHr9GqwMAJzpWDfd!IU*sWC38K!tBsoQ#dB5-O{$aL>zh;m3Q9Pe?cfpvf5!
zw*><+vaJ^z$j(e6BO-`q$lvn2>lBZg%H}I(doRX+++PZspeA8a;*6D+Hj??}VG>aV
zyh6aD0L5(%uw3)JSakUu5D=c2Sa<=}15G<4of(*~0UBb|;4c-RZwM^c;Lkf7o}(*(
zwx50Wpi~n7rYIjVFfs9>UQ@gCIdUj>aZHA1HZ!4(DXc1uZWf4FLe}f;PgiB%cy(xn
zMXOjlY$6f?Ef@0so?e8^pYtd3`-|Hx@>HM8-fQByN6p{%msD!bx*_hwbrcDYeNdWu
zqavy-Uq=_x%BMZ_GaGYirlE6T_}hId^h3WhDu96th1U|Pm|D$P&jw|%(KmeIt4Rw{
z;KF)()8h000XnjWF~D#2O1wr`3-py@I4>%hI1!_T1rMhhF48f<pBQ?bVG!xB_&hEM
ziIl2YY%kfEs3Mm{1%UsAfCOc_xLg?C^Qa7@X&UO8_WPiWX?I#&Sm@oiH{Wo&{E+;7
z#OeJ`iQWWbMTCjFC!gQhMTn9*8Ap6nJv>=oH8f6jnu?DP8%*$;;sst?k4m7v2(rNh
zmhX=j4k4jdddbG>>XSBLxf#cx=cnXAMY9xx<4yylM+iQ#)054iyY_UzjJ4@%Bhj>4
zz2)KF0}<_&PaqWd6fYPNnD<NPb`B2@PFb<}xF@z-FvhFa+mj`8x%2?XQlMhRe(~ZH
zxa$J!fQ0QWI3b4Mw%$6srp|cu6`oFIN(_krOk;2LGL_B>c-K2L&G$v_6v@w(DLR?L
z!_?7|qh_n-H;PN_V*yz{<hDwJqsPI@LT7>20Lt{+oc9y%UZQ7{i_#MMFl(qk-=q_@
zPXQM#G`C|ePIFYbxp$zTIiqujQkS0T;jmk2kk$rMnj~k;w5fDb+hIgF$cI{~-;R{I
z954AX_1tMh?$OLjK-0zDW$&W}vCX}BQ&-Ao(n|eTQshg^xm9>|9xFprlBEFJqt(-c
z^du8mgFz|>ZXsAODdDi`L^{9@km%rb7aN?EN75y5jnE%0n^ZWAfqxT><DvE$Cs8#=
z?EZW$958DnW&jEuz<~s|Ui{heOADGUp0i`)W@9^GM4+N#7myf(*8v+I9zGcYMpWwp
zSJ%N5p2SR%&?ym4;N0{H%4EOV<uzJMpE}YrF`C+KdGC7pYLFY9hQ9q^z?K1(HM>1<
zQ}^%l%k`!T(b?wvBYqRp^gq8>*(<PU9@cM9i31t7&=|%<%mQ^oUaG&^6B6@&{$V~;
zg_HE^RA_9Cj=v21)Q;_tos~K+ZxD-D?)*~GVC8i5M{wM?9%VQ_QmEZaBntN6i{eZw
zf(O&;^b1!SJR){jeED)0yOkv}8Ej{lwR<_be>PKiA?n$`&-xw^SK_6(e$qF8Yog=j
zviIxk%m0i}Z6x{NJtaN*&i^SUVpM#I3f>2Oco>i@lF`wXNPk#djBecq1GuT1A>hgP
zr(8i^zNCSViz^YjI@yF{09IJ)#I+R;R0Hi(HGeOdfL)+>@!IlNu!>w}W2C^}2!T>&
zE`*RvwvdKTarQgHeaY0L8T~X@<mjip647h~0UMX><^9b?U2}`<;`UX`Z2A)QkADBp
zB{<X=?cbZ=uq)$5V07xglxb-5FOFKCxQiNqO7vZl@=jIDxx{g48)wHWvLr4-XSKht
zO*q9IMPGaPpy(H}5A!(@PKQ#kwFuT8-^s<MRq4h-FSU8#N=4fJ9LjDOS*g+4co?E`
zSz9HJ(uQA>g1M^T-Ij<%2BNQS45ufFN|^LF1i#p}ST1!ETyWE#X+Xsf{(>OG>#GhK
z^$|C1xPgGF*5b*5jEww8r5-<67y#F9r=ucIfX>S`z-wV~F`mWZB@Eb+2HoFq);Bg{
zmUY1H+8)n+{C^M(^2=;gRbRm@d!9(SX-v|9U~9MRb@Aj^t_(2{1BZ;7b{R50$H5T-
zK3AjLj(`9--s06wv$^9A(Rp2;KG6m4D+E05Rh}Pee*8c)9!}~@;&3p(V_*7Ov=ns!
z+LlFXm54q-AzuN6g1{gNFF0o``frKnlSdaiTy59(xqk-Z=p`%mbYxdT;(96?*Kc#1
z@yM%vkiDP9qf$7$&HS`oxvPg-Ob``cPV{g&(U4rBZ)xlasb}{#5TvT>CC&O-unz}J
zpsWpUSfehTualbSQ;tp-h?FF@;u;vstkz7EKd+^Uh+;wpPz;!RXnX9ODKyPmwNLC9
zJT>@w5Ue<`-)zX0D^xq48#slo-KKhyj4m$<FZCxF<~-aL>t)K@+nVkIuMSI^<9xbD
zc!gTSq#^X!Q=E*93|By7D{SIpXRjM;m<6VLqU!2+U>EXYQ%D!tKLC?9rOSKYr$mpQ
zm9@*>iQq4IOaAJ`#z3z7hsi$ny_$feBpUET&IJ5gm2Nb^83E$Ym#G6jPvBm$+r9yn
zP*K%lC%7NtdA;1DkF6b*Q`>L;#6mhTHY15_rTy`ZMoGyJ+W9x@eUVerj~<GqFRor^
zrzSM2$lFfjr4<&A`K{)EjC&4`S^LHIu<(1MO4m<`k^E~_XLH*!TfMfiwRICk?<Bv0
z$s~p3(CimZC-j?B3}x>;Or8n|pX(nv7}TFYn6|x*7mqR2lrJms1@K@3^MBvM^Bx@R
z-Rjac&(*HAWg9j?<w;b#Y!$2<?_>Y|9mNke#ErEjB@0%A2-8j?jGC$~=$&h7-g|8$
z5D~pzGOdTsXHAMUYD1_La)8+Vt8ixf;$kxJbSej#C-6i7-dgSz0btfRuu2A6SSEuW
zG?0_}fsb`}2WUgh?#E*i^ySclH!!4>*J}l(G6<5LAX5zr415aa2Lg-HK0|8a%t`ad
zEEQz_`dwe2vvY8yPny@CO=&zH8{jimC_B$|p$WXcHEJyNfN>X)zGf&F9$#E+f*8Gi
zdA!DbK4YNlnUzRXII==(QGzCp)B2;}TRybU$_nL%#VB9>cnkJVM*6yDgXCb%T2^v1
zJT=ae?~oZ=9ZiY1lLe%E_&d#Vefqwlj5@mKR|BWAd+k)W@9zZxBVSexRNh8fo{(mV
zKfUD+bX>XAFF7S>e7oNL9P{<ZYubDBOODD^#qPb;?3yF`818PJBWKEBm*Xj!yC&qd
zcx<tZl51R|%oEaQ`i_=0C`lsz#h;feTdO}j28>;6&c6uPG5<F&)_uBTAaeh3#Q45|
zWl-%Hds#&I*vn8cdd=o=Jn@0&^i8@IEFN8-IJn?JpIIr_yGF%)x`Ys5;J^@X!aWtp
zf1`DRba1NGhwALsUa?tg7DH<Z4s@)ntdF-KP!cP-KST$Z$c0a$Hf@fi(l6K4)|QU9
zba!_v4-11WnTL=WPJ80LdW6XZ$ZoCI_iQn1D+ixOJZy}9GPww#tK`D4do*T@<$WN-
z@g0RXM@4Z83V^9&ExjFMoZcWZq|j3SR%k^{KRk87F2q9g)aNtwZtEb&5*J_3iqTR!
z&;dc^aNRQV^jS+T0zCU|vy3|O`LOz4Qq{($h749n%AeAahd0h9qzxT~53I+Bc%n0(
zpE$PO7*DWnj3f+859(e-EOS&IGq{<lj3k_!?n`ZsdTp^e!K<^Orl>q-1z+*QriMe}
zBgv00x8~IVKm*~Q6TTt;5EL~isYyU^eDc>&cY658{&HxE6&ya7*4y{`XRxJ>ja;B0
zg?WYc<?rhTrD}6ZDk`dF3?$P98FBD47hHd1b{#E}5L4aR^Q*xWf2VPsr&`X9ZaXGG
zsr-B8FzSx+ezBvq)%8mZk6&SMVR?1BQ>ZQ-!PhTLZqK0K`rZ*Y6)>jbwSw6$wcGRI
z-vig8>|$QHUKqhXn<+KOc6oZf`!j{Cwew!Gt`qm%q`ir~PP5^SAS$h+qp~{ti`P+p
zc7}A8;{HQs40E_dm_C^tvihX9vTLWTh2nPBZ(Ca$q7?8ksM%%;(_^`qYgN>B4Wm99
z(gpR7guC8C1dvxtc)0^{h7lORMc8Y|T<);Sd4FrjgP6=8pN;=J<5bsq>SGAe@4tQA
zi9IDkyBtUR%EC>I;bo186P;@RM#3;#9Sln^usI@KpcBfU=)NscP^LdR2gb3$l?hIK
z00BtG&3kn?)?C_Pc3bU$!hAhFkf_6uadmGG8U#gXkvH1oWu7a@w`bUizpT%g5Y@32
zrQH9{<iV0kmw~E#q?AtHGd-;I9;cn~koV)v;^f29&Z{Lss-qHd*48%s=%?ZJb*vX@
zLZGQbZ1M@ynN}ThS6<iC;be~a#jBMIX4(8LhWABdZ>_9%e5NMDGwx=Ogt&`DVX-0G
zN{WSPc{4DjFsT0uaV89QnWDXv>X`}*YkyiuXQ?)KdbuHmaoKG;rdViWR#i;g0@fE+
zg>OwNj8AgyR};T$@+S1NT?jLf>r*F-T;xyMopy72d!I99uoBa1IgaSK4O8!0Bq_r4
zk;<Kl;8~b>OPMRaBhgNIV}<{iB|zZ8aa6s+s{yzXYxBr2>p4#KzBao$2kP*mgztTZ
z@7}$G1p~_4b=!2y1c)EK)TKJEn&m8Ao2%+Pa_QgiWK?1073mZSBi!+<MwT%2<3^5l
zl|Ed3xfCp)B+BgNVWYWIcQK}{RNfW0@e|s7QfG8csi;)tc7$prbp)jihsmSykx$;S
z!$@Gi(GZE{x?UNbYaDxv$Ay2*hlN2NX-}ZR(lmylZG&k=Lj{Z(&2$A&FN<iZ9OS=c
z_BtGy{*l_`GFN#~Yy)n7&xbA8?7I1J^ccs0d9M<>13&wabUFqucp}WUG`7D3s5-XZ
zI+Hw${qd4T9!~M{3y@WMKxkvzgt8L7`m_M*NNs=`Vf*ZtW*sa2iwM5=oc~TO$dktn
z&R`Eg(E4Q_?d|^P-#j}z+wV(lHd+P@E;oh}pZb8?0+g-Q=5)aN2e_R?0Iz9vmJ)E}
zcN67o)^9}k<P%OqtuSN|V@9ZpV}emqFg|ldy@L*26gK@wRsYy(JoWg}>R9`Ch-)-V
zR`rHds!e2b#gS-8P<{OFAhD2;_cj}O#?(qr4ffWp{=E0eB?*qQR>xAeZ~KT!tF1uq
z>SA<&f>N8s{;G269ARcW;UGHIVdw5L?XQgc>Z#M-M!pEVWTZvSRjcecbrKwgCCn7$
zE>no1;_tRhbihG@o#mMh!NVVVqc2A{5Z_Cdodd*5F5lz?S_hSI+f&Q^bL@W^j=;jR
zBJC9?U@r^&hhd`5k$`oC=gAP;mSe+TU^%q6nxC6e@GAmX;CbhrWLQr24Gb6nYXLNT
z)*rynL4)?+iY)g$6n|7vh5bLKzB``l_WK`|ifG7QiHsy$HYr3#_8ujB@9jlaWRw}Q
zLiXM}gzUYt3E4ZF-+ABn{rNtAf2c>Lx7T&O&UMbY&Uv2avq1*gQYAmB&-SGIC9EMX
zi4{0Tv~643R2{M|o_OL<@2->>@@UgmBiBrz=Ju8P^tnIFjXz|+ji}z(w-|ZHTKtXP
zceY$Fg1F3&L^)<CZf5nv*14LZ3BDunx;S2}V3#3}&!riHrUx)AZ1RsZ^L{6JJ>l~V
z4$NQ@Z<BqZr(7I~q&3r6?=4MA${h^-EoI!(S$3c#>JNI?QMq<hvocnC{Dyp7@KE$<
zF;X|}rv~Brmx%LbT0D*W9?AxC|M(tAU-n<kTp?mNkaAx2-P^OHaNBCYx-x}+33CfL
z!K0D-w{3TTWdHRL8H^i*=H@bm3La;$4ZMJGMsL4|ED1#h`>U%=g(cI%9qAW|C$-Q<
zj<wSvBi`~s@3o%;O{Q`zyBuqEH+)q}aM{oeu_NBCUJ}=}q)k8<<UgFeF`@1>{4lpy
zu9o=VyvfLnQ!;cyJhaJgu%7TKW4Jbkse<l`+~MQfeB!PfYTUsyUh&z%Iv-<N14>;Z
z-~6GuK-3rP@DZko4ez`HA8N)64rV|6b)7OL;{n#pD-|)791Ri5NpjXR<K{&Bk>%dJ
z5Zrj$Vywdbi{2;lh#d>RWFo|U9_~ilP($>udAz4>UHOm`iMl-y^VrY;D;n0c$BB8G
zUP~xt4`lU5<7F1vvLnB@W}8qz8DvBUsjbmMxgzwB1o1B<JPP~SlKkV2yj-fi{h69$
zMBjdS*}&La-kuCfZL{ojy+y9K;tca?cCARrx|IzsxsE=p)n;Bz8dVjnYfXK6!}&EC
zu7!-fWOhgVlUee$)=kR`;nN19&pIwV^&f(FB2!eby3*cRTz_yXgl9~kh@5@RKr0-V
zrEl(y6NsFuTEsf=DrI<X$<xvmI&I-W_A*B5xuw82{+msVn|zv#b$@%iBEL3`{h$m=
zmYxn#!3p?fBQ<ude1@9@|Hi(b<>NcG3-n!lrEEe`s4EyGbZr^s7|g<uHma7H5$m{Z
zUy+cIprNIG!$PzLZ6SMPb(oubu;@UO{4PE|in6Q9aZ@H=yS6j%)0Z!u&=Y-F+oR|r
zk*b}aUQsc+#!nijjhOEV)I<>li|x6%cU{VTVQDlvAFx84|0fMbu2`v}g`bFiLL$yD
zrY3WDL>o+AUh%1HD6^9}v;OE2>N8UDvih}_F(17TJ3>a~(t)r(rU-f<M13{9&nG@b
zKs!DT&&`xjWb;q3q32EAsSCU=-jbO*{T^!%9S*~D?#{}|^fsBQAlCA0!SOg2=5E)k
z9{9*myY?4}zsd~_x5_C|HIsNv7BNiJ%|#Q>9SVXnHro})QPEM-7Lrj#!e7MkvEzr7
zZ$9DvBHg%cTe60byA4fB&~-DCPBOWzRaG}W`RH_7bBb$xS-+lc&!twA$|0rtt90P$
zBP@D%W~zK&+T3pk7b<npW`7PMIa!HrER@~=@2r@AWs1$g!2w0J%w_&d(FL@X6`ZmC
z!@}--2q|^ieiIlN=meu_T_M&0%bIeU4>oCq4gVB}#F_qOCDD)w(9=t|>M{<}X?)Ne
zRXZhmR#}uAWTej4olZ|WWxwX79@9@myri(Yw>x3s^yC-LCZ=bf?vJDaz3kl2p`7Q4
zQJ=s^g2_vfmQu8Kz7IxuM;q{cGpoqx=+a7%g0#?h$$;1VD#8vXv>_DTQO)cM25<a|
zUN?`3APzeV*^~<55)3qaCXeEdZp@_QsC<lzeNg|5<Wt^r5ApZ)Z9QMnSpMJ;Wf=`n
z$NE*(TrfY%X=<#(5l$aa`rLWn$H;{w^}7aMyo_Pwn|5fFjaJz+LT$LUtE-Wi4U90B
zCOmmBRKv8Bgeh2#V&Kay5>As~e5@c&xy%B4q{NJMVCM;@vfCO9*Ob{E1%WT2nSs9R
zQncYpp?W1VoOGz8370|jya0@R&Odg%Q|<>^Abs&jyMT#><>Nm&dt%$$xo}y*hVE`$
z@zzyG0+`P7jsD6-MH_w{G6%7gr@492_E)oMdYVd`bvYskr8&4r2DAL!1yp^Jw&#s~
zwm-~x#Nd$HC=~VL-<==q#P8t}yo|FV<^Oupde7_&arg*E&#>g`rT!e2`G?N}W^d0Z
zoFp2C2Ym3Kdp~6p0l)o}?h?^aeHUI%s_j*G`R0n$$?R+UjoR^Bn(q7&!FJe`eBpPV
zrOC~&tcOSra0{7kcxbo#;p<={d376+bp&5(Ct+OU)$yZo4Xe7e(Ap2O3$3^JJB&Es
zoE8+8KGl~CCA6*`ShZi!BKd772iq`6!bwtE(qP(d;&Op3{NM!kG>%x)rF~sJy@8oc
zd*53)*=={7>Q6TKii(Q^-oG!Xpi>7DHRz<}?$v!$m@!|wkw>mVfH;|(k^B`GnYyK^
zA;qiU>@^~$XDfXsr%VcmqS$y?R&$h6Ya3UCQ*=rGCwH_+XBnPZ%o)3T`i-T<R`EBh
zx?@4zt8ym|lZU(EI9B(?88~837@IDWb)Y9TeY7Y&cCK^c4v{YQxIg=1V){uI8Y!o!
zVR##N91Ps+nFSoiAm^bIV)g!Fx_Zd^sn9IpFZD6*giGa4v6S?6w~Fi!9ZY=V%eJ+>
zDLk~fUAPsjxF~J*+wa>gK6Ga)xn{1gY?|zxt_)gaP9a6YiN3Ad*v3p?xn5IAt-36_
zrlyQnrbxd`B{6R{F$#SU$sbAXGb$+kG$Nv)JaBRWd{H|HMt*!e`QN`I-P2=m4eaOB
zeovbR8y8B(gYjy)bG+!KB$Vyosz`=L>9+MWQosxJSzK%n{o~D?Y%ArHEHZuj>Uvwr
zV@;o?yu-Nke4(NHh8^j^XsMAvpGGLF@pgE{kt?UeF;15LLD}f&_*dD7@7SGt+d9WO
zKUIEI4?=Ri$vvoYvnfs43J+Gh%wqTPXj86Zkm*L`%m3LeRY^#36Zq2YO-_8*QpRxn
zYW=O+1)>8tHXK4qY^-<6gC15-hIg6P+CBX>;)L+*Usm0*eDQ?QK*bS)m4kP+joA>z
zXac(9VJsp2%d7rHh4uyM($lz+xW+=Mq;heg50?-N3$KyOW{j?DRk4K6P!VH}H<_3<
zt55-INqujzYV>dTt&on7*wsTx+43**WFrK$q`TKCZ+fzJMXodQ7(=k!*=I)Hzo%Y!
zj-ey-rI|Wgr+Mt8WkqAk`njXbc@<?{kpKISRXuZ@$!pj?L2F9$TS9~vEyK>)4PrQm
zY^m2LL6rkN`!YE!q;~Rkm{4cPgY83}x`ew8P1)KF4Yk|jCKZQO1Fwyqd@UpM=4`W)
z{ZMtaGRJ!;`qjA6wK%Kd{qUTWgbR{XF!z@}vr|c&DMq#Yw<UvoCP(VqiASiYU5j|$
zJn$Xwk8I7&C00RS@AMpXi*+>dMas(%`)QqjSG^;)wiFyM5?_mdYNeNc&*GKzx9c6J
zN4p@%VxFc0ZPLr%+uAZuxngP=eT=ZWHwBs~H_tQf;!dw#n?Hk0|2#ZjpmXE0cg;@}
z7O+J0$;3Np$GQVF5{+P+%9EP`_BXf8F}`$)>KQg=l+&s3aTuIfr&S=P`F@q49bL`I
zR5m1!z3DHkbCZqvTOlebj;SB<)JP;wx%;EOE~Aq>54-mIRDj<bYH~l{Sr42}3QOYQ
zs1WKZMghC&P=&w)`@$2p9Eq0hz67rUaTT7<sWC5&Z^80&^>~+@kE_V(GOvM3>Gxh@
z$K;}xcjo_PO9*XpTh~mklv+OVGS6-~wZ+IZ-Ru2kp8%0+k+rq;g1(7cLU+Cfun-5G
z{+-^8Y@D|B>aE2swlY`T61hb~0GatCag~j4tv@tYn8!?mf`>y`Z!4(npWZoKxu2oP
zQYhNR?kxTNAfVt8x+2R4G`<S6k1ue%WqK?v-SgdcMUY!Q6@!8je4I68H)IokPv{tO
zm(JdUM#qT9g^SI$mGNyw%Xcb0-&fa@<4n8w7D;DniMV61g<HnnN{J0{%4Lx4Pq%6E
zI(w~=6SvQwOC{}CN;?>QCreA*`TRbUn=|{|x?=e}`ODev-9kRu#ty@o;J>|{d5iO3
zm^`0+y(f#vPBSz?5>djH{@~w8Q}L7|hPX!Lnso}#{s~q8o3Mzlv)2-<u0m@k+#GH6
zXgmM2MS)rkTy|#{>BJaBCz<DG7Y1&{>Fne79c}e9=r3krVG1~3r7GXg`AgKQav`cL
zQq@nQ_mQ|*iJW8WIVoBhQu591X=2M84x%xn(kgzZ>!U$?<6Gu<KhY&B1(jwQ+R<3t
zP)h$xYVB*k%R*Ac=xhvjKVf#%^2%sGK3y$Nk5qmT*FV!1ULcgcGFHRTEcyB%o`{}{
zZEnH!&)un%miY`0lNMy_8dKA+hXRF(-01b+=$m)i?-J=6WANMFkq^9HSxeZYT{V7+
zhv5Hf2pke@TF^MkQ%LvV%P^amS2>_4(ihQ8D#qraW0&Cf2#K2X&WW1d*Zpa$w{0>0
zg_FI$#fh8rTl^~mVQgCk1LauHy*J#km$rS;;1UoNzk4e=#Uw%6b>-k$yB|-jNyna*
zw=Ny1Fue$tX-xO7KPr4=w{7*v88IGf51Nc=Xs`d!-Nf8hz;&gnLinoTR)JDrcy!>5
zFxT5?0_4Q`d(b`ZFF2b&td7f{*^$WEzAm9UwA)(xG?ZVo6C+#cM6VnZKhcTnQu)l*
zwLqz9;f}s_Z3aJ!mv<bVW4Fsk+@?koce31U>I38iqyFr`H6&;B3U+q`;WpB4t!~`V
zj9Ih?p(W(EXZL&c6oZ*!y^4hZxse}1hTsrkgMO4Yf5q5@vba&CXop?ZLxVF_2_Wf5
zxrSN_@R}qMZ9m2)(<8nr$U1hs+28dEulzFUzT(HYg<(fL5RLn+Np*K;1hRU+SrIBt
z`WlljEyI4eO-&WM2y#<*DI=FPuTS^)=@h-ypzw%#2}_2I(9QQ<sK#Hg`i#KIQfB=w
zx_)6r9Djshi^f1?LQAX7!;MsB2HzRivkMFA{&8|M#UE$HV^(b=?>^-7DbyjK21r8-
z4BIwoDEO|{r)vvzx5gJ#U+uk^wOZ36TX7>Jb=BB#I`D8oWJrE?hmJsJ;Z9R`P<G20
zkMXjd_)Sbf9+FVyDTDa%|E>|s#TnoEkn>X)=58!1p9*zr?c2@HqP^`;jZ1X{(9(n>
z<L$<<Vzd(1zLZL-ZgHboSKkkqw0_s=V~57C;XGhrrGjzdn5EEpx<__{SKu-)gFq-#
zU)359QdM<r^kcgYwK>^EUEct~w^FaxOEXdwGz3qhWQ>{nI&np`BGfWfY<fLI1$-rr
z{$RZ<u+I>s{N?vmzK@~iYJ8*XM@a5ZlU3Qm`#~YldfUD={~05A_q09pvR5A-n}6P0
z4v+XvcpIUQ`lqt|d;ZVzJE$y|tlai#QAQq@-TC(fD9TQ8!kf0wV5u%qSrfuGT%-?{
z`>Z5=F-H2Lmnr%3<Q+*+9zy&3_pd=S1HvDldq9TIv16;a&XcJm*%A8dF^$UKs*djZ
zGOQXEtG^B@O<M&!BkzK9aU9Bqdxp9tUT%HwRt`m}K|;K#U{&4AV`pOOLe!n%j~SP)
zI@0M=w_>TtMYLFLS(Wh*+jqa^_++=k-<CetCfSY_iRCCHdn$slwMzCX{X89)LcizJ
z)Xon{J#MZY+r0xGwOM<WU(q9dTH22PF6Y|o>mRUC%yh%;V4%;*NBNBVs(9ohG_Uz$
zE`&MS#xtUQpPcSMB*(N6RNdobh!*$pw3xQKOqw2Kqkq_)uRRgqzN_IMYq-zrT=_=N
z@`GUVH9N61oubOTPfI^j^x5_2HIs|zKUePTeN=uI9*xbVuCYq<U%r0p2|YIH6EGE{
zyASc7mixS$x_q_h(T7qxyjSXY6Yrt=yEy+kzrY<iC)w$@jZ7#!IBsJ1Y9FaRzcnYK
z?ZTz8l#BWXo9$O-!PY}0oaQHv@9tkB%dIieWTe#%?LXnl_{2QVj0B(Rzph_AQ`2SP
zJU&7Na#;Oyg?peZ?^7Y;J2j;-c1R_`fDt>diu<LlDJAlWM>4u{$GVHP&7?<^w1*B7
zQ|kJG&TR-s@S5*cjRtz1h1CXS7NNU!*SzBdfc0W7?qt!%q^W^q-hX21<=CNM-|3Q(
z4y`8rk92RLE)4gs8*j+ZwJuqk&R(apjbyeZlHSt0g9a&w8Y|zwPXq6|?Y=3*llT=D
zjvMv8!<vCX{+w+rWzLfrcS(?$d0VuJ<IZ^d73T4~n*4V=nAz)o7_l`dcv*?#G!I0n
zJLL-GNVin{!N@jr^MYRB`6Nd6MA|O-lS0!zOg0(XrWVC(2bmgS&MvBM=$;KdAy<^<
z1;ru%dn!BK_A5q*F+ct&{;5W5PoT<!rb>*{YRvxg;m7-7>Hioaa!Y?tly}Y{@xH^!
zb+~PQOib{Gd-&t7$?Jb66Toj;F`J6g{;-!c*aCeud&x9q)}1aSrL2!$^t*t4g<0L`
zwaw=5z95BapSt@;a6kGF=g(JP4Tmn=L*y(yTQ8YB-iVSz*NbgR$&DoHVQNUqcDc=2
zD%Q1bc}5y3hC_)Y_&*yfl`ZN;y9oA6z5naxhNOS;q)EM{zzSCxC3@|6(-_8Q{FIs~
z6VnlW)jOdr-F}}eC&vF=G5u$E`|OBu#K3O5{k;u$=I1GG|M-)CnRfLn8Kt=snmFIt
zLVdcu5$eWqsIM$=!2_bc!SO}63{3Yh^0o8xw;<vpU;nE6mFn;Dg-3QwO$YkvU*{y8
z=*IuP3!F^Yp~F)XXMK8-=sTD;Th;4O2N#AmyRL@H{8^GeX$;{2_vw6|i9hDw)fZ4N
zdur)(@kLJE<d2C{$eZ2S<!~#GtD6o?L{BznO}`Ok^=HIvZO^wq@)wBNYGkN*rFtgQ
zEOv+-->)IZp~Z8oaTQ{!Dl`7*2D&idhJw>s7(*>7Pd#xgAJYj5T(#KoB!0T({N%8b
z&~2Q2(dYYRA*<qr-}xu1{43I5%PCCP2IC@t(|T=ft-@}(A1D6VGow8MoRE+ZRAF>e
zn$|Y%F1)~ChP6EOQImC5_Fcxkjyp5uh+;+8)#D29L;|wGkD1==&Ra|RqcP~TMnmkM
zT?IWbt~xHY2>N1P<kk=T4@{t<9TDFz8560YlGX2i)6n*?Rco)#(yFxSWJ2gNAD!!P
zec6qs=<NbQQRfhsN`{_K?#s4z4+$N#A14NrkJjTkB6~!CyKxO@)DSWOMC)_a8(PM1
zO!sI1s5$L%-`G<;CUUgGys4p&X(P|N*5fnXKVx&fOr5W#+x@-FtN_W$-G2``o>GLu
zcNILta1SoxU)+t@PkW~_=k5`bZWa1l<XP#wJmm5UOMJZDG>=_rIz{ahvN6fs_Jox;
zK2M};Q>-hQio?ZA<r?T8E{KTlx0LI>EbG(~9b8P=GJm(Dbp7fRIt!0Ehhtb#@Otop
z>Al@IB*ujjz3u4l;!UQKAFjtI`$k2u6GAO2=aHatFf@Z^2uZ$2RSMx%&(pZs${l6-
zfJeNP9{O#mw)|_In4t=vYxVZ`9pW{ODqkT&%OoRl%(b5xHi%pLZ2_X4M(!GnuFtGY
z|1OQ`a}cevK>2fqjlWjjPkwoOe>;vR<JfwEcA2MPO?9~Q6st#A=Tt=FiB!Lc>eH0^
zV#}ed$PxoE97ZF8->p!g4kw6v4vT@yyR?M~!2@SeXiLtHwM75*uD-*c!R{dQt3{@~
z4>C)HfmBQL5q2sDZ?&-eZDUpmP68)%H@>ppQ$2}P3sM+z8X<R(dwvsxkD#oC`Vjm-
zFHfi@uTc;M=6cuaA912V9qx=0H+?D9>xYGq#NOX|&Bq!QjcOmV^RJ@2NBA<vUOD1)
zsQM=;eGT|9VdXg*PnVqLT_kKH-8{*N;22|OKKN4~v{e2in%%y+)AXD4uYD7M8U@EU
z;-vIp*MRy~aQwSlA{-1;4NP0O77b7sof^&54BzQ5iojv!>+p_jp$}DiSx2rhH(r>w
z&{{7x%Exz$x2jyT9kgfR2a|-h{rG!i3ZSZSPX+S8YW<CE@7!e?GJ%(!AqZYcb!oPe
zLXM1whaiYmLF*U=ZVr}Scu)<fCtFZfWq00rTL-#1*&9N@Y7Ja)a|hGNQh(Y@5cXz$
z-FP{{WrcxXfH5(mKU0AY<P2ZDd`S&s-=MEm1pIKI^?A5nK25}HM~~_41q0wa0Nh2j
z`vX6G0J$(^u58>w51_twD}&2f6|=Lmpj#FUx6PR)$9<nr5dRT8+f(4Om{1io`2F(}
zs*7*8?y|3E2?(`%4*`@?9B7?^Zr1VN^$zM#{zho^%Vn>6?5FUsn(jJrmWVNf<Mz3Q
zlUc*$TNj(KNqg^QdS}0Vac-5YrEw4`QaJRF!-4i=*vn7bnspzua<2C^b=BN6!Iv^Y
zFMboCi4;7=u91<E0fB2jc>W@2IAP+FsV-LlbSY(E*$FD`j1v}YF;26-grA>SYWt2;
zQ&R^71^uo)n9>8r1ki|I&HC(8O99$~AT5EyfB2pN)C{)csW*aI2zn1#2k}6m1XvRm
z!)`J*HnzNn_f&>1hK)FO){1+I03VH2p2+7IBKm=#5Huq}una~TnCkOWQ+>eYm03<}
zzfKG$eNqIRUM~A9&47~v&gYsHCneS=2GXG_dKFxxCn%ua{*UHXt+5vjc~0Uk0sZAp
z$JZ+mcS<S#%LiogWr_Ug9e7kJsdIvvk~a)3*)Kfr{$sQvs-kP%;q79`%?^xx>#M;l
zn@%`?+UOE5>jKOoaw-uB&jB0@R+{X%8K~031d|bfioJd3P9vynh=Ks7Sjs0*#}GVO
zW0#baB;qi-<?QUNM1cSy0{`I*;ti^;*Hil5x4lK-I?R5R1OOdA9{g`Wa6_NH!H7d1
zkWTQ2!}^O8Aw_#5Lx3}Ys0(&Maj_oAj=wcD)Z-w5hCCH7Z!A<_wK;7a;G<E%Bxnm3
zJ8o)&^rlN~0zmHWfD}W6_tnaSDOb~YZ_YQ$`%at)W{eFD6n*^WMtIO7)qR&=l(8wO
z82ad6FW@DX3~EO?)W_ztAC)osj@n0V2C#JBq&@K#yCPl$>dJ?{m>j?Umv!9(3ZF3}
z(*}`tv~I2$Q`x8TkE<HzFtteazI&p>u~N=aZHU(K2nbhbwp9b75VRitT`6A#!$@13
z0!5Gs`bK2@PPeXJ_1`(s)Y5YP{YjjK=m!7x4?Gl{R;BBl33ORmKy0rgJFpaJCD3$1
zZVq3ab##1u01n<_cf7``ZD)7)B5-bD>cU8Y9XOFwLEukCagBo}Qa#oQAaJ)(0y<0m
zMGMpG8^C9Sw7`yRZ*QY`<pBi^TD+Qi6vOY;hUSNe-eTsX5EFqZ{_MFfm!EyZElX_(
zQrOa4ZZyR^)wNbp9ea6n(7wLyq5GU+r<)rmZh%T$c6DmKuWIO=?cRUc-#-4^sBS6y
z;zrGuexjr3?1Xi7O2FWM<}xQitG_r$PLvLMK>W?~Duq*0<2ndxG=bFvB1nmTLwh-|
zdILijcwdz$E?^=Bnj31vLI-fdOSwO+m`~LZd&;GV!8a-!!etaqDwZQtgsLnkJ)I8K
zAVwqQGBYaptR$cv+_0u{1{TB;leMg4^bJV%vs+QNoj2&Xf>_PnZ25j_+nzIt+r6dE
zx0U&Oz!|@H_%;A3th;7&jkqJR3y;wDT#cNXXqo?Ur#=;d!=-lpq6t$wx|NcYN>@kx
z5`S*@KD)_kImHjSSOx8fD=#D^A6k-c^^B$JvzvjiP#Xg5dXn=b!R<qVFj|zgGYYH9
z-ut_a%(t?e|Kdn++mS#K@S{AK^pi;+kZFx4<l=KXWD)kY&6ZGm?F`OODTJ8Ij0noG
z=>E9mCjVO+1|a`b(}Kg8lqf~@VAJO$$^Ct2e=m!iFKGT*!P<1Ve4>8(dd~DFrB7o=
z`s~}s-BT7Nq=RYacm%Kq2rNmlqbuLeEB{dB&w_<V(i`g3yLwB&Eu>}||9d3~3g(AL
zM}?Ntg7OLqU?}J_K=X2W29%hqOcH<pek;r{7lm(=EkAi*7y3VuQya_6ZQ%NyV60FL
zI&Prad|5bq2<69iCTx2mm(;>bN;o{Rurr2eKqA)@K?&&bd^7VZ-`m^6{P2(6rHXo^
z>`tb!&{ooVWPm{1RHu5*g4mU_W{QH}tnFn7?ecHm<Clv6UU%dey)H?*Hf?m3JnEe_
z>m|NUatkBa>ZleFXMk!%%h9a7^NRLaRH<Qx*<!Z8e^J2%V5C`Mq=r1SDqig)C($dv
za6nYG7|?1kE5>8<`x-=_RK3;^6u25_ZL9!}rXHUNlYv`~$>m(_Sh@P5W9@JC^_AXF
z-(t;={Ch+9sn?T-4+CMGCmZ?4&93*zO1jcdPPf-kXl;~PCHAK;_cWKTZiAyixzr3H
zI8O6x>saEp_Ym6YntQY}bo%ZgGUfXP+BWta0C+_M&<7ey*##(e=Atw#GVNN7A<TKz
z6TZ%t1t}w#NI=a|eWDguEzm`WK;k$BZGe-5Kgs-gro%az`>fS=a(S_jR;M*$fC>Sl
z7i6)-K8$_twCDqfo>3P+Jji>@WX4Cd$VSf#tHJNrEO#&Yx;BM5y>U!<u9Mm%W@=hX
z^Iu@J|2mve3FRPE0rd5(^Jh{FF8eR9D>=z!7W;F|ZAC0LwV0iU{Sp%?0XF>`z@jid
zu>~`7m$n+QnH70Gz^$X91eT!F9G{ew<R=*kF6R;ov!1=(g!nU>e$LUZjfIXGxUqDS
z17OIZPx_+s{P_>1b#mz4peFT_9!)8e?%ytEmH2=OCjKPihu#uP#GmA9jme9d>aci{
zM$JCL<rlopQzl2KHvqfbIiS}s9P3}AOdy(o5Cqw)E-k-(>}Exv_|U@jOecdgSc8qx
zbbtkM+|UH+$ZMxsWM_1aj*eiCSA>DHLACJBmCDM>g#$u-d~2BafWL2s>5r8!9o3t4
z9T3JPUc3;_S0N~5Ckci@%H2OHp~`U@`jy~jFRmmxZcd$^?)KchdlyX-nE`v+`a5o^
zVl+FmwAAW;FnLutQP7>+<CI%01-TvG_8g?{QTp5I1GgzCWB@~I8Y%An!%WJcX9iWB
z%oFEnOB1M6SAE^!(@#{{rWV!uU&%)Gxt?hgWV^dZhmJp|KV!!59@gK}Uk}bAk}jaa
zC(uHx{q!2h#KzI7aH8$-+YN#}0IG=wQpdy)WgfPo#1N6xw6txQ^kKK2`|+?9CZICA
zD`8@4dnG$x*%<+;y~yGZj6e)}>VJ6XgqjrH08Jvv)(HHNjz?-mKA%4mp`^xP!}3Td
zK?8fLxT0)@+Cc9Us)9^qLI2=NbGx56zQ>iQ7W!`HBL4rn@{MmSnD}FMnxtZpQghs^
zMKfO11&8n0o~alrV#at5zEMrYkp1Er+MUc6Td`g$v4<EvPg&Gy0gWS(ZQEZZ#zP@0
znTnk-Z3i+a2A!Xu!ZGPJ;;5*o=*?DP)^=HbfEvO4Nc9X}7mPZ!K%7G<GXW2(5eEr?
zsZnke1x;sTyMxftrK`7sZZK*!fL&+PZ@;TvW)>ipk_>~3C>44bUYyB_x(=rf%)0>8
z>j6yS6&{WUf1zPn8wt~YwriuTU*h9e?X<x}3pqw_XvfQ_2nJX44!M({j8V0{twU+-
z=b#!z4u_ut`v~v)w>@;h=H_RGzf+O7Uzr~@w2cQFn8yul4CgsrIUXi^{-0-KfHo|D
z$qv55qr=Ec(-zGID|d3LQYV(pE{S6HUQE_C)SKR;)~@td7Vsuzm~TJZ8<4f1N(Zef
zfO|$T_bQWuB=yi#{*VbVr-@>`YhM(TymBO1@eZU2!@LQAUY<ZItU>Fn+l1#cUh%iq
zY}P>yuaUQGh34a2QIFIbA+clJJI^o{f@OTIbHY5NC+#!H%W+Mkf8eJ=ny8M$-2Ql`
zM;6h$C}R5a;UP1NrJLmyrs}J#d%oM3_bUIo3jBK-t(l4(t#yAe9`rh{Sjn6@eEhsl
zSK%h;TT_CNt$iVfq9t9D>{VV|Z0Z6X@mFdEg8sLHbF{H)-~^(8OJ^qsBjQf*42@=W
z<fqo=8&72vX;bfd=hr%WJ;nz?<_)GehG=AH2IeA^rF5QqxZsXvy-p(BbMRhN@6Dom
zcGb}4^@#qRkRRJ!^-I03JGUAy*bDmzoo<OBsI8uJS@<>KU8mqT;3&CHfQ~ok@^;@b
zh&DW6Yiny~&H~^ds#SIjCi~O=#{emc0X;C7nejvo(E-4**k#|uG@TmWvtp`zpS_uo
zzN_=sBkMY8w4~L~b(|i*D2vZ7zub&Vx$Ef9;ceMCGvsh<<I<#NvN!wQfhUrgJNv<c
zh5DMN8FzshqmW?Tt&|2Mj|5blK)JD4j%-3F$Bl_LNT#nqB6(>PB<8~DWFIpxVn93<
zfH!@MkkB908zCEnJ5&G)_Qpq!TGT_hmWr>gTtiYD&vUA-LWPDaeHPHV^&|wZUxZfw
z=;`tBdx{uBk!mJQ*T_I+AJ+u`#(`gGHGqm>grwM5dD7MLKW~aE*w%iphRVv$Iz)G6
z`n7Wf=%!P;@7)#^6-CW@g@@~{+rcBh&o-JQt31m^7_*n^2rsp+Y@IB<;OY##f){kW
z={{PmSj?T#_M=KAS}siIa`-9ox`uMFtTZ}Npl0Cq7xU4BeS<rp@!9Sn|JwiyBEXVv
zK(ge<2a?tOtSXOFmCGcxU^Z6AeLu2v;5K~Y!y#uoN=!lLtI*+DG)~icCdZk33kwH&
zZrXyEd@(KD9d1BHQpq)FRO=HUhrE<=w}d;Y=*C2_ZzAQGx~nJID|OJ{{b$SH)Aj&<
zN~^}H&}|(2jhlwOC?XQxrg&Z(9j$V-j?#>T#674~0p)`q+OM)Z&aJF=`<2%<9d?PT
zhTnMND)M=C+Wc<6<ctySX$<<9%ODqtSiLLdYsYC#z0>*nH6^a|^_wK@DhoIVehbL;
zFDI)v0MZ5URTk!u|9gnGl`GnGbaVil$*Y>|=?R4xf(8Rvue48Vi?*uaR`Z|9Ww)k!
zU7i?BFb7oj;HJN>R^ZOQ>_dF2@`E=&ZRfYl<`wibQ8O|jc)BU?>Im-Yb3P=Rxn56$
zD$@wiPQ5`q9{p=*04ogv)Hk3B-@bi|_DWMTNICmT63iMeT$zF~$Y??*Y;5d~1EuWT
zDOS`@WqeZ6wcXI(;)YSoA7(-gX+N=HWCLSnaVbE@$n@9bFMPnXmxic)jue-ZLBGuV
zPH*V!k7=6s4&Uz1kt9~nzq<|iR|_K)Sa@?f5kesP$4{TU>g$Ej$|oI2QSzQ>d=6{6
zJHOk|VDP%aZ5oElRsS*WufwPsz2K}tm25@K+}s<qwe|I`z(`0+t>XN17XnZv5r%&b
z5d0ERbR01TP_8ITN`9vTH_+7Ed_3)Oy4?!>D})IyA$*YSG1Z@a5~nD1fT=7wMhIJ}
zVh?Q2^Vw<0La5mVal-o)cSA%J8@eSb-}#DoFEn7?NMr4K#Fefe?zBb*oz*bAQ~o33
z%BZFPa=cq}vAX-jH9(q3KYb3t6!3{R&;$hqYySQ+t$`6|NDU&UwP`sJjtCgU{pZAH
zgtZyX4f=VLojE^Q`kT;@k+1sR_OC<QSue}6G<k@M?ygw7yk@>;&7^SJyVN%<=mvR(
znBO0ymDgrluDc!)EZi;FjAQz8es(LfFF<8-nm?5@x%}Yk!EdTjo>&!bRJM=Wk-kF|
ze$>%1-|iEqXE`$E@&V@ib|F6ps&5^Slfg^{HOMJRf_(+Bd~bb?lJBM|SWL@*A&8$p
zF=KXVee!z;ZeZ|7&F9)uxc<m!+o#0y)RB5jR=35=W^1`tm9-UAMr&q9(3<z5w!Rg9
zb3MNEW|<JSr>Ee$Dd*|WKrORCRS7u<FB<@`msr0-w3m2XxK)IAGM&O?n~M*!r6>Nb
z)L$c?96vq%-}~br0UwyW;^F%+G7w;`pCL^Qi;b0(kU%M_85}%;yPWNFbDoy0CBbJZ
zs$cI;ZdhIsnE1X(sqm=S9ulE}KpI@P>fltorI@{C1#ag`4w<>RssU@K&+;>fR`?C}
zcb(WaWoRKwGlQLxQLoLj$xYaZ;HXdALeaZ@ZPv>2b1#NE->Kd`t2{K17c__~rjXD9
z5Uf0A+q-CC_@}f2`}%VV%cJ5S+=v|Wm8sZ_A(PlA(H-%CL}H$OLB^PPeA&Aj?V#KT
zDbG3XcYdOA7%ddEh4?*5k>Ga-!5<*7xhNv+9p!9C66bgaZ2RvyAjf>ZaRa&~%yu?;
z8fYok2@<d3O`MLtP>fyCSh75uAxx`SHNeg<_kNrgc~cSHoP0LYfn|bpD){w5dOgb1
zU-&nzMy34gQ}$<~p9RN1oeW+d8zp{E8aYI$<_fT3#Cg4R#0kDu>?z?;XNR5MSC(RC
z+7P5bJ5W-Xs11t-9J7}$-D8Rny$<k`0xEqRqt)Cjhn0oxL1X1#sY)yU^2K@f%*`6%
zYI}7pALTzC!VR6Y4x3Co#|$A^*jHDwK1g(_Zd#MkO*gRp(WA*vrgB2|-_=5d1L4h<
zed@Ydd(gXVhlV;B`(>fGL<Wa+bj)a&nMf_ReXKp~$H%mBEbTor#mDOH7~N@aHq`1D
zA4@25Fw}04KW{87Cet}1<afWLZF)1Av*d`Ltx??YwJq=8$G9M2<x+Z<D2+)SXK<0~
zwyUXe%__sjERC1D2@id1mWN9m$$8s;z$}@Kx>lgoc$9@8?312YSlET7rOm@)d0Qs^
z?@0O3VP|583{f4R(w`Ym3M`iidNXE{hfCusPh_VYtfDE!YP}#qD|QL_8X%upF+X{7
zz1kE_bGX6w)0V{zvL1Br%(>CMJF52AVYV`Ez<S$SZ|6zRJOKX@jg*SX^cNo_<kf|5
z?6EQTjJqY3ULw5N9Og7v<IMVhN1(!Vj=H62th%hMh`09ESC1c*fob&B!fH~vzQd=&
z<wxHg1D=};wr@pmM5%>*Rpuz;nmDSzYBWUsK6h6+yq@}!!&^^pOt-ZN;<;#hNw0Fn
zpFg#I>UnP?gp>y)^Wv6p1)o^doAM`2ayw*gBoz#$Guf4cURf+L3Mh>-q36#`_Ao3B
z3=EDO0AWwUiU)nhD5t%gqQ2tAH)^G^<KFdA9fa~Q7u4EN8}h**fHYcC^uX1E%Fd25
z<g6wxqf0zM=<bJek=IAJptW8?IuU6xe~iqI@FWc<<Ap}n)wx!vMd>j+9`c`WCKCQH
zDvj_%8yBW)KTF{=GjjsobX-nOE@LRt(Tjy77%neR4V3d3e}9|$kpv9QWktlrrkNf7
zQlsc>rMU9}P>E3(`5EkhzIKtN5ZP>d_WkWwmUJ|&*Vh3`Vq^vq9i?C5Vskqu_RF^R
z*Yyq@`Ty_dUV)!WMMbr+j{k4~k}BTjI6?QS2g|S#kvFRPZzh`M^tw!lDY1ULwJ4uy
zWo6sg(AD;p#qUmS|C@&XD5^9eO3e7FBJTgZE)_{9XW>tEm{dCg%G~R?QThzNxAvRH
zkm|ku{h+XwtHWZ(HvrVe>`&Mz`Ne%g*zz42Hg@sOum9Zk`_&LpcBfDZ0T-nYen9N@
zhJys1+hL6eHHqo2^bj8!T~5x<vk(d;NFahyC*q#Bx3p<iG>x1JJG*e}yoVy5jCS6z
zgv>SUqmlzUCfe&Vi{H^?k=;x2o#@~BMyDlX{+H`zSQ3;Q9nuAhDQM52%2W?#eQ!0I
zyu<WKDlBa%V8+|QnNd9Qm*xEBOdLe}3uhxx(tLJdOGH)Tq(KYH<!EL4s>(ByQOQ3@
ze|7!;pQ>!x2U4h4K;aU~V?_=|R6Z<mNPKuyZ(nWlcQm1xqYH(?+fj6Az3z=n%AGRT
zI_!uD{NTCQP<XZ7cVr!2O=A@&u6+)|bgQ)g&e+1WWJ;C8yPVC#!($j1BEGWqD!bl$
z&2q}7P>J42oT{PZ=6BZ*w;rWN_Ig;Us2_wSfE?ZHrBwZy@RXyobxLFgDcq7<s_K&O
z>2LVLjfTn<Ni?!AobUZ>^eHUF!Os&XT9YS9Uj&BIGE!g3O<~s-Joj;+d=ZLHZS7Dx
z^N^b=4;^O*H?N@JBg`K_xf#tI>X`4Fu&AD2**Mn@ay!!=t}VMU73eqei@|p)mPo4L
z5W`HSb!<FX4%w8-yLjuHk)nM;A=UX8@5g!|bWy#a>E&<U#rKV%X;to{x)dB%JMw|@
z^YDMsN?#LVq-f876@@K)(0;%G`rfRhD(-&4x#Y8s0^I^PYXh_W9WNO&_}Ht~iasL_
zZ~Fl<q{|bP>|i7)9`K~3djS1>7<97LbQ6i;v89E9{^$ASR{NIV=bna2Tq52YeW71i
z65aDXo$y|I?1k%n+u5TRs^LKZCkiJJ3iuG}+VtSjbEZsAla1pUxl`V&-!nBoJ7937
zdG#qC;kw}Ey!2Y8n>9m+{x!S7U+<)_ss4Y?!Q8R3;^&?((X=dvivl#w;`5WpUs@O0
z&rLrG9a+xf*(Z*EJ+*LO6MuIig!Yhh_+sm}>?B_Matgxbuwnqp#jV}lA#ic$4A6VB
zc%&J+qK|0_7XfOqAYI_*;r!o(J=z4+SfJ~YEhv9S<h{)^9imVbQcA9xee&Z(W*F}3
zox~T>{60VsDE(0$0a(VY8#)?v&un_W8atJQ$=Y2c)}G*67MV?Rdweqvli~F&!qSg;
zQ2fXjtLU<I!EdFBiY>ZpGK!#jG4tF>hp-}YL0HrC(b&tsQ>0krc68D^Cn+WGn0mPw
zmKEv}=eJmBdTggv`Z>(Irz$NXw9sEaCE4qmkKOJhR2Onb)GWq*We+5x#a^&_$!F9@
zgi)_|e~hAA(2g2(nZd7Gsq`u_Ktn_GoYmL=wmNoFWz_l&A1hN)gWTdI(L}R9<-3Sq
zAZxBt`Z|t@?U|`M&$`6A>S<-dS$!VK(Qv2<IZ_&$#mwX1^y#AQXosboc{B8OWIJCa
z%yKXKS8G)vb|^aH!L*lzm6~}Ntq!ZhOcy&f&b!x~&^2!{UDeEUiH|jDxXNB~o}O03
z{4y%)0U>wZoA`%;lx~r+2dnplY%|^epl^}9Oj0_VsuIt|bm%`lK!y|aHPWU-@VUF+
z{#B@zksdl~5X_pY`RmJk7uK@HRz9A4Q@iFRA6G8&YN0AR>!ctN|CY;!D6iLRrJ?P-
z9z}+4rBNPynwA3g;Zvt>F(!i+c3xSS{7d_ujzd{Wz{HhEBLNas{lFx>u(U*1Jf>;%
zeH_RZPCJ0{XFxAhU~xSRE$LOV+-%iivxfZ;`=wIHW-jtc4`I4cnvOfVa;l|)S(ABQ
z50T-_l~p#cI|0Vy!sW<OM!#&TNlzG)B#)a&OxrS6;59i_y9YzY(=WVUH`G1GLp^;g
zK5~{LT1?BxT@+PyGtOhm>cB3z8b?8Fhbs4Eozs1<<&w+m(>3{=!4p(BAaC*S!WrxR
zd}i@DVB<euD9<wY%K0CbU>fnz@bFklxKBevbFt%xLt3MSi}odYY$d;HJ33>ZDSpWM
zu(Zm|O16~jmTbib8>)&o8L^t<%H$ihl9OVZ`F{`g1|O)D(oS4KLldr?bEw|$Iy9t_
zh@L#H!w^Xc6cE##CRCz&sJt3=k-5*f<Wl{^R#H|rddNh{7Yo0;QhW%39n>#Xw}%!}
zqV-v&%P^|ds+_F^4Ygm@j`{FJ-=KQdZc@4e^gx|L-4^AmX9v?NK!}D<MwSDk9)Dm`
zTh76uOz`M8o?5Q5G&Etp9~#D;>^a{g=S_u?Ge%lkF}PK(^rwCBxHvl~p6dC!H!}&U
z1AaFt^Td!~5(XBQS5Rvkg356zkj`MV@{-{~H|izokw`d<l4fRfn}7WH1HGVr=w$Wq
z`pY@FU)fx~?ix-jy~7;WU1a>B{zB*h?4VOM85-uI-S0o&r>i5}W-9ve6!3`%atc;H
zWsPMeJo4h&l>Y6o`g8G&x5&O;hT>&ZeHZfe{Ys=_wo=YJMy0JOet5rVp7Mp3OPK4E
zh1$3Kx|Fg-H}3?`8kS`o4N;eF6<3$el;;oSNE0%vI6I)0H2@F?Xy^c?TsE{yd?^m#
zcG*5_WdtGthMGuTLMaaqj}153F{mwY+WqNex3J}n4@-p}NU8CFm_|8b(N%8_Caqp-
z5{c?`@2`z1;QFWZ$}T0EZ`3%aOUHeKYWY&ng9VR31XyfcfTw^HM4j)4GDtj*etA=1
z?<42~ySD5@O<)*0W1JqRPUJi`&y}(i+v?Z*#3`AqPV3xHIIQQMOMl^(LlKX_q=?!d
zH?bKwOFapQz)`e$+1wQtn5Wdv!}PrUVEtw9J8SlmqtKzEbVK}#$!P<*HGJuzWv^Am
zQtPfeJ>m9j&t3|~XL1e@Fzg#OoL95OD6;3}@lLo#k~o;kB{2x0HZjRewe*jQguGO?
z2Su%8_E=H56m9+K_TyCO^9jR74Z$oY7LJoFlE9y-<8dSkL}n%sGz(iPdF?VVaY#~u
zKrLG`@}c3QmYUVP>Js}^Wlod97hrb^O@25F4%MY@M>hK#lNy%wXAS|IBa@EPxj^Zq
zlOh%@55x<R@$tEB;dHAZd>g$AvI1$Se*$cQD8***ld0a(LOrngeQ085AhL;qSU5OJ
zGBP31=SjJaM|<}V{i}2taxi2G6-{g3hsrth#bY^xgNj}8d@AJ=HXp=7$TC4FJF!o4
z_(cuHmY=`>D#%8J*<4_KEEhR2%OzQJa6Mmu%D83prSHNMkIT<Vgq~pwex`~Zc%XKg
z>M$!mc`x&wsAR{dzxodmL#4p7&QlEygY#UjCyvTJh87WH@-@-Lxy2(!$Ki<#><_fq
zBuvQebK*umR8KXFwxiKdkNkR;?P$_}Hm`Wn@3(T%U5k}Trp)U|f!g?2Z^UueB50HX
zjo16Q(WxTu%Cm||v^sVq#4}_w^^z{&OSrk!!2A#gpgN>Kvmj3!8Vt*n;#l?kXqJI0
zf#34wV58mUlolSX6#W*%35?Vl13IHf^^tm&U1nqBx%2cmsEjeGm%IojiF6yNjI*ec
z0RF?C3rAsLPYZY%XHX(NUW~IW0i=SKZ5O8~3FCtY50K_fXqpUt75#8J{)phi%1C}b
zykSCaWD_P;A6sdGI~oMC61V_bg8X2(;PD^oU}E-f6^{8*TP`=JpvkbL^H{ROcR=N|
zrM1-+b<k(qtxbM;Q<_Y<4lL|M+i8?=q?PX3P%r1H_j{W3C&l*L;@IR<%4=lJ<VQ|-
z4n&vNYyFX&E>_!Vql~q?sF<)iXm9UF2(9~FrPD4xd^%Uqs^V04$w}CgW?c65lg|}i
zS1VIW+(SgK^WjDgKSOOU8htXfp28HrPE}tJyrNnOVr9OIAP5dI+w(B2b=ebGwq)2w
z#eB$}U?JeN5z)x#j*lC+bOO#BL!WAA5-u|*u_+ZzS%H1U<dK8r0r4=T1Zq$BLpfa?
zHWnR6G;O<xXY=U5&Cz9UwNB83c|m1N=wfF%IJRf8>84O;J-^kAFv_s*8H6O{NZ03w
zWsRmD#4TdhJY8(OS#~Vpo-H{aCsEk&tK6pA!)Xt%Qfp(kyLdYqKX(6Ev2Bq~U|g6)
zj@hxh@iR?n=*Jd>GrUT%zl~br*kbEyyOa#$5jlhsG&CozS5_P63?Vt;6rl@0JISe8
z4jVVy!pfc1wLix#fvZ+8HVB3{l`=XydM8*5IK4#jb?QB8X_f;S^SD<lRkZAt0v01u
zAqyjPaew~YhlYWgI_0Ncu6C_@nBCd_2w`;yu&OL0g{>3{&}O}R=g$4?p@-Q+vGwvU
z(9LhAD9+h(`92Q%^H(lkei4QW77-B<69HIiyme07bEgzQu6`RROXJEr&~?$qmvTpT
zm=|9ppZAkSzp#H7=<p^&+9kJ8kx5wNLNi-wx?5;z*7!O@tl&=Z#LVC2WBJyfU%T{n
z!;yrbNZ#l4#>hJl_u06w@mEocuekS6scSOv!NMP|&=%{(w~XsQw8t>s-6d1st1Fm7
zu@ow0D@(&u!o~0+=UxJbaTv{V09^w8_Na1pY40i?9^Qj9yb1!^hgrWFQ@R!#2`%x<
zVh`PRs;+#g^HP_qfzGH&e`>&#%PPVI{aKmi<>e#N$_4XthhthRvH5)pR2}vhBFSL6
zr4F1P1}-jTy_O&zy`^dhG^^v5^+Q)F-994V_1sm~q{q;R62}*Y6ZO6=oX;m$$(O#V
z!&TghP6&+jGw(eu92lz6aaHriqSQMdO<ed{^71T8GCKb3s!Fm{7(x*miJ9BiLd-Zg
zyOEGJ98ovB&}Lhm6~Db%6D)KzW@ioY>Zs=m@}WK_su5vmX{Ksjju7C}@OtXS77+Y|
z*&s$u%_i}eC_jD1d*m}FT13`2^D3hi9~FDuNoJD?eWQR@o+A4?n>AvSvF%@Xu6#^j
zi;)6yC?5*v-?ydX5niQdem6diuXpS8X@o^)r-y2NuF-wMQ~9y0RhB`LbQnANO><xv
z0E>Wski%q<9<1RHtX~Pd7&OV>a>ePh#r(G7?k4dd^&L^!lB^Wq*imOD6Y|jX#>C;J
zo(T>O6{4OvJDd+4g5cLre?^$M9NCv4A9s0Wa_f;6qWuHE^kSIh`3h0y9$w}zHe35-
ze{QyQ^^>N#iJ^)o;?ZUa_loUoaU}Y)HoI%piBR@oHOmpOn)(sz6K}=r%?Fgf6z8nD
zx?RM>?9`$?rXJ>w?x=sR+ioEZO9ceoDhyh$j25OmRLc6je-z7UN-Lobu2C%%n46+T
zmJp@@VBD7o^MIW8ej+;{NF@6CHfvHKW@p(fwC@}o^xUl3Rc0Yo%e-6@#^JQ33&dlx
z<zP8#q07_t^$r^onXVg^=_T!GuUgDcY^67REg~2C-eg!r+P20O=k>-pGLPkadE`I%
zSQBe^rU^TqQSzI@Zbi@ZE+J|u$gAU1CguI=dF#r_$4n<1Pb;;WFDlOJI0x*lnl=No
zlr`e>s?zReJ&H9M%F2W^4V4uF;w-O6@I)$XZ&WWdvfT}M;-Y8(?dB@^lCff`6Uq~N
z(Cm+FpJ=vUh*Xgb^EjNn3UHmu7fZh<={_{e|5-<oN;(GaCd(t(P7gAnxqiIU!Hf#p
z0RA!sTv@CTQded+_7&Kuz)MKO&t>YmgBD>e&b8T7+uPW*xS!rFROCEexn#UC%RfuN
zK>Gqtx1`Xgl`&P+ve5D?7h7IuX=i)kn#bniD~R1Sm6PfBvRB`d*tp-!N9pmBAlp)-
zG?)ls6$#@o=p+L8oJ_ya?6-1Jaq(=(35v$oQE>>E#~zfAGevi#L8osOo{>4y`Q?3n
zeof}O!>=@%aL<ng>eq8xbD4An!0`>eWUjRn!}f>&CeiLLHIk!7J(9`wl1}DJFDmQO
zea1(<1JSHeHCK5kPfMfJbOfopHGt0Yv1GoIlr`(C|Cz<RF8L%i<5qbr&`}xOkF4oc
zNiS0R6`nlkbN7l$)mC*D25Na&MOQ8_XF7gc-R65d)S|XYzx!LRwBEiZ$LqDBP}}-&
z!qBwotj|-Mx9OP|Np$vVBj<#9e~K#>V@+u!plSDsW*K0Y3c9+wNM_dobc3@%fBP7c
zBqiu|vRjyr79;~ZU=ozQ+I#X9M*<1j24sbDp+U0@%jknHB$j{yB)CV>^YimV06qiI
zpD-W*uY{04vH0SL%j7)0SoA$xClVU!CXlqoLgfs=c<Io`b+vUuTX^LW?kGD>Uh$xE
z0%1GCpIV|qD54DqUW9$wU~j#QT3^I<&Nn10X%}5`%^61P>VJd3h2*4{*4KQEpw{T&
z?*0^+C6N7Xxtvvt^Pv9H@1HUQ$cBelIj(cFcw{024bAVMEgK+YPUF=vGRlBUN-8Lx
zWdJ1=TU?jPa88W2`@T}HTCvzr)&gQ78DHKO!wigR%fMba43!f5rFh4Oyu9iL#cHJ{
zL;V5=6Dgo~l>y1<QEQwO?HK^*NMM<<9LD#BgoK7N<eq`2>4#gA9Z;?vZB04(t0-^f
z!N*^^iaQ9s73>LRAWC)zB<)d0GWv`iRb-Q$T?aZtL&ErZ5|S!!3t62=mVcr>g@n9_
zL)+5NaNTGcU$OZ4{O_o(KugRR&xmYsE0s^I=M&jQP9(DtCo$Al)OReboL5RbYMJf)
z&WaX}8(2vzr6fkZ&-4b%2J8j%PNfvSNipmc+AenTteOMcHFG}^SDZA8|ANAOuMeYq
z!Mq4j25N-+J??eRaDjMyphn;tzk2@Lc(SULR_e3W)I3ew^t29mXdcs79%Z~ijq~_5
z#RgsSnwpx)vw0EY?~eegHxw##{%2Vctvz|A%VbIWuzG74OVnZIbfwFbP*8SiA4%LZ
zOPF^sM!q$gC@Q{s{@zdz9knn?QKz>EzQExep^Ia`ix#r<VkmD~lQ#MpU8se^VjDHq
zhoJri0gJ(ue2?sdbhAgcvlePaz(9*i#39ZKpI}DdP~^^Zb#^L3{;_jGe)sM_*Nb2h
z&VE$1u$bUZ`?hEyVh7C1D?bA8u&z_xLf;DYn^a%P->!dAd?yus2uUq2QRU8$sl*v1
zbugFv(^Wdo8Ntad{8y>1lFD4db-TMgKK<zk!5t<N)FUTJWh<2jSdbrQp}r<8%GucE
z^|7yojk!H7OJfW2@wpsK&Zi9aAP*{Fzc)AYb8;R*5Re64^<o-hP!R+R6wJHpyKQu3
z6KJ&~6Oe?*434d}kY+CfWNHL1)0vPLug)}hXX(_}(X?5V6Pg<Hcz$@cZ@?NDl1;bX
z)p^tSgL#*Pv1k!$H=8z?umygY?W@Ga<$bi7=r??@b~9n^BQ0fETTg0PxdX%4KM@Np
z5fE=!Sb4$~8rQ&?MQgiA0C8P-^Bd$-+go{E-6X&br=_OSXu0nXxd5%DyB2MCV~VFE
zU~P_~!F^>E6tJJj`T6-J^kM6PV@!~`n#9i*GkCVWYReotPQBhvKXuf)-fvpz?p`H0
zp<=K)hP-OM&|$vfcV!8)tr;iWcA~#yJk2hB&f-soq<uEQrW<NM(UA5Gh>S*1W`^YW
zGh7Qmbyi3zM^z43c?W=iFOQhrpuna4#xDJgR;h{7`2ySo2wkM$s<_LXuUXk2zKj;3
zQ2J|7W$CKx%dyAH0bTva_y4v{hEoX?$w}H}h+g2D44?GdQcyr4;Cyq6F+5Si#Wi|w
zS1yy@eNe#`Ni{5cu_}`#nPRl>w&M47=${18^C^erlqUT6I<O<!^npyc;jDstQ~?m4
z{#Vi0F^h8FU1z>~opyi}x-W%Z@!0RBXudA+m;5of!aT#)KT9jkZ_1PPzwhdItxiiB
z^QHVU_mOrj+|T@D$lL6`Nb9P<Y2$%|c+h`ckRDtr<+Ig^ZrO-R`1QzKnvwrcmxDw-
zMfipmU9ODW!l((!;Vg?9hIzuI+CO`S#=M=e{C@mQ_P$LX&9oU;$AXPB8@{Ja_isb_
z{3T1}NWVTv_V4On*eD(^WZWfvG;}?z@hnX-y@Xle*B9E#Nj8nMq3z_=!R>QX#71R-
zKyKG#9bdb+$b)P<Gfag%S~=~7l3D+QWO*OaqH3F_1$#Q%XWA0ddFI)sI#tqTAHVmz
zT>JM)0)zAV(b^-FFY#r{^(d>u)#}$_2iAj4ZmRVcLi3+zB=H9;#2K5XM_#@jqJGIC
zo6A*5!n^w*_uRs3*sWgG{|U0jRPbi>j*)^{DD4W%;lm*#x%3|g*Z=wSq>@`(81%O9
z?u!hX|JYyotBS;r+A6kPm7%9*RYHH=fR#;<*}k7mXxh9n_L=G2#8pG3v^F<eKJ>xr
z&$3Ir&NLEHXxWX~DvJBGtUKju|BtG-0IM?CzK20XR8&-?%OIqsr4<#VyHmRB(DkAS
zNJ}?}bRN1nbax!OyFt472EF(G{-2NfAb8%1clO?Et-WWa+A`>0r$xL+ZcT$k^4r>x
zZK#%$IPfJ2mGBKBDHFM+WkVTCGn|>_Ynv5uCI|Gx!cVuUTZ3OwmqBm2zyFQB?u9)j
zs5_txa3g#H$fW<C`OB~e5r+HkYZR$8$Kr1sFJ9EE+WzM|D8B8<?i_S<UH|(|ND3m?
zRUH-5A}%)k?|<@mIs2D<fq(zMU+!gKF#2y)s3Tp}1)}F7RHF%^3&&0OIAUGZAGx$`
z=hK4z2^Wr;GVlMhDHJ_o+W+>e?$`gi`2YKIwH^}%@8`r8q#z*T8%R#szGR(g-2C~I
zr*~(s?w+mD-!9d;gAdQ%8ZEX+ygJ^8#~^ewACVLDFO<?ec~W^)Xji!tqhT!1KP-gv
zcVbXpf&RP8gvCT>3SA!;@SW>KsDY>y4utJ6`r|#Ts=#gGf%MwTBuVJ}1BY6_v<?|Q
z^xGFlz*Uj;Z+mYZ+|odKP2n6wBD-Ut+1JABocb7A^IDe)>X1oU>Ozwx9+<=);J{#W
zywX%DFsn+dJ*g%f$y$qT`1=;zzvQ<QqJ(`5JFro|9D-HMnfu)zXo?u9TSsqH7rtNg
z8F1sok}$51hR1WS)(?k@xMIQ}mpRP8CXeDJP=|?<I*YJxHQu;FlZmx7veBW=xl0yx
zmszxzBMh+QuiqCc^$v)vS4#f#T3_ErWkhip3lWcHv#l4%43T%DQP|_axzr6!5n2~v
zLu0ep;<_v}{gjFPx1jxl-0@JVO}oO3D)7wxdbVm*)gX5Mwe88PDhr9|_2!Phv+&OG
z1L%oHShU$lmwYzt!?=042%psXXX=696hs`fFDROw(f=((YtNhK8E)mNcXXAmnCIvp
z&Qc7MXWEvlnU_VFO|O@n{d?f4%(vc0A3`?ms=(e77j;3OQ%+(8LsFL|tv!TIYakIg
z$q*To{gtsI?%L%N7AH|5vg<Mm9Vl&k8lxuQ$ss9)xqkk9jxke&7q>BV$DZd1k7LA$
zh@>)(fwO6UPCMclRzGLimZE*2C{zJEqs+L7b0@Eeli(aBE>r8|>pj07)Wigh$4M9%
zMat;{LDY2mK^W|IHAPq`6q*b|q-LbFc0?SpW~9{eTXhD`dk$8<gKKiaZWN-u&g%J2
za{<?xICVzR(bl`X8!S9WZ41^_OY^5F!knCXE8^TPL5bKuC%T<|{M;9KcCIC{q_*s;
z#L^1)y}=PZ!gZK`n*s$lDPT9zO-ns>??6lX<x2|sf>2tHLiR<szGfsESt0jkht(oN
zxJWsdIdb+?sz^j`+Y?(usXNp#k{p>O^Xa`yuZATMN<+wFyO$S4W<@2^S*|N)cPt{1
z+J{!3c2kMdWsf#Z(eK>J|91Q`GCDk*EMk|1kRUo5=x2o|`$v^LEQu?Zkf5O8Rl3Uu
z0fCMoXy+$@WNvOYtG7E?5RIsHeuTsRJvI5%@_PqCi?IXb!U20Ne<3l-2_CSH-QTMs
z;X}-&2|x!Nk5JhA`p7_3bx^Lznu0A`c0zUOpCMR#{at8MLE9C0h__OE?-8Sn!j;KF
z$IQrD%)e_QjQ(~t%B<coq~RK_o%K<eHY7DM2p4icrApxsqAe82AVnY&k1_wXymaSg
zokw?T_pb!;$5a{3l=BYceE*$T{Z=}Si5n1b*13Lf99&{GW7~g@VByt`Bd~NFq&!=l
zo7;nlI>FLe_0w%8-Hxz9R6OAw$S|`us5)4lg{o|~^tKq}D*pDW?$?E@Iza!5WtwB+
z$d|8d_g6>4Sy|r4z|@R!BF;7SRJY$#Z4Qiw4s_^JPo(}riKXN_`jNMTzcVraZ{#mP
z$aqIxS!f$qDsiD?A(4=EeRYvN5J4r{8H!1aXIQk}GtGN&(2XWf@p2L){FeCNcK<mp
zTmr(Ze*2kB>9UzD^bK}H<Piy%%lvm>Oj+{&a(<PzLnV5m=;JvE2Asd2!hZm!MeRAe
z-zUY=z`^j`1(z}o71U?9p_vhM2U%<DLa?JF8z@CaASVU3KFKI4MFIYw1`2SqtQJ}j
z3{U<(1qG!#ZSr<Cm1M#Y*e9nkdj>#ccQlI-K<(0?(1I1*w;tWCPbT-Dh>J2UZ+=T7
zv^@pXW81C`d;%~|*1@6t;^IQLY@)<^NhE*(Vi5!tdm}&_N>??nXdYSm_fk;YTRX(D
zror|XUa+YM91I)(Pur#oZ3pW)4Lpx`zl$X#1?c9TYc^-)D*x%;u3E)Chu(C+5JSj8
z;tnX`WQu72`B_4h;Vq5G&Twc<zvwjo>A&wnD7%5ygC55&fWf~M_6-bF(9)Xa3@Vw}
z5)l=pJzy`Jh`+ub{GgL;)w#9twBteKMk1U#sI$4-bv$y09wEtKcM3n|!D28@hYqLA
z6R_~Asz=ZxJLzlv^rLV0c#oie4sl?V-j3h7tUWL0z}BN@d|ShgvNZVV$?o`;-QV$t
zu!8Gs_DO5O$z!}=%iszRB`>eKkB^T=W-p#8LG+FDvl`v99(7em@3`-5R|*$>M-be;
zK_Oway!e`hs&z!~*Mqj6jy>CZuRok{q6&iy4)~2=Km$cL_&e9VrCMEzU~j^h-`&?M
z9hc*108=Y!y2^vI<tn!;p^c#o=~tSO|BMF(MMFAIakp~H90=Ni;4)kVAYE-&fSg}E
z9xiSgP=DeNkAS++34ok-fkXBXW=~KrPD1^Xb^GE9RVW;j$sM>AQ4Z3k56=Hh8(JX0
zJtVSOJ-HVTxR0De`MP{|Rzj|7rg=#gf}sFv0%6xJ(QCkgFGXM1_-MwY*r`>5L|7!G
zb4do-c9#t<E5xhPO8t_&sgtDvJ5RP3ogW+i0I%vINGV2)hm676T-ZF)ZzkEYW#<78
zSsie!ji=Ms-0?jZGX!1wuxKZ<-TnPgpP4W2*13~+s?E!8vM{A!8$|{XK8%{6A5T8H
zz4!C;8x$gQwHdYoU_HC8?wzNnr(@9~W(UFJ*_*dsHrdhD7S^mCTsyY8Vcn5KSEYV^
zd8$~$WaRr)Mg|BBO|95NdD0(htXH<y2MU|f^ZH<R-s+rzlR=uQ^tjPA2QF_NF1~AA
zovFIA!y<?|Jt5YP&6RW_ZpeNhW5R;X=CekEZldvEeM=l3NBUs&wc746N2hVYS^3Ko
znZQA6@Y+pL*`t40GZ+I6^OTn!?WE@HcVuuhHIVd`FFLm-Og=OJbJt!VZjuWFtgp9j
z|2Z3}l6K1gWm=H|y0zq4wZGD<2(I+Gu~UsM6f9G2OGxTcs3wK}9IK5yQ&dd+FiTOj
z;K^K;R>`WNVA>U*wHUwoMCOue>3RoM!f-q<d<dh_t4)74e0zHLvf)$Zez$#CLU^iF
zw)}{ujHJ4@ahATPR5Uq>i0H_LE5^b8k@}f_50AKdzSM#j77~2^z;^Ag;nksZsMk3=
zl|w{$c<JehZsR7|yY7DdGu(FT9j%En)=rgNtshOqMkI_?Dzv{Y<XPZn<8haagwanN
zp2XF%{s~eZcRa~=b6xs`uEf}bW3iCl+qY|V9ZLStWKibSM47y<R=qr58LLrB=+vf@
z!I*JPflcbR>p3sUTQ{}dY3o5x27;T8Kq&ZP6)MB)AR9fbYGDLUA2sg(?U~@b&Do${
zhb9yng4{Wc42w(1`(mcD9!YeDOSrxKhb7uZXW@>D%Xxq?sk|`euvy>bj_c3sCoO*M
zUE$e>eJ5CYEU|2vGf1Z*lU9MoDIT001>TuI%-!WlK?1o7HM`9tF{~E#=sEbHeeerk
z8hEY!hADJR+zjO+eQ@U9vpI&@q*lI*j~IbuBLM+Ks{wf=dJNiR8-z(O&aCQevk0H&
zFdFpsThJ;*Sf}cV3fh%!JD#goOqyRFXXUNS)@lUzZatNiP1GEbr4;@!ef`cSGc#53
z^-XIiZ>p^af6oEhu9E*8vM-qciE~!1I`<;dbFjH48wi|uS*&$G**B+?EmnGDo9o_z
zI?U8j)pj|od;L{Y@V+`S2swz&rwA`KV~B}Yy*}S(<g$4wU9&NQ>=q})Ilsu>Ko0*}
z7H74Z60G5L%o3b&HK3OJ2P-=~tc$)K+W<3^NZ-%7R)_q_@GyKgviVm;LP*W=((hAj
zCQYQj=Vc4ZN;#}LHk`_vnx=z`oof##hNKuU))qJE@Q}<%Pzq~yC(z=-|B=jDx{mJq
z(_nv%cp~4_h#tyP3%pa|bI2KoKm8wkO5sbw7UQXcg_r`<3YDSkjnwsyyqeZPqlJW;
zEHa27Pd}VC+#znY?{P`ChJ5EElK%1ds%Lnr2kcmXXKQv|nn7{*)~<i&)^Mo@1`YxK
zugbYEqye@@hgl&hTNsZVlZ%-x&cAwuq?97FoOl$bb;`DeqW*ps<yx=~^Ot+d5bM(0
z_$)%=#Xm~&5&>4^)a?re{<DTMh2)UqS^;JRIf;bWfdk#&&q(n-0Zj6Fn|>&HDz`Nc
zDectuPjYmOQ6AFHyWei@k#c*t>ZEdShNS$gdl%IDFI+BRqcXbRx@D&D>HGIpOO;Ay
z7!RhSJw+nOt-_W1iBiKI*(9cV)!V5_gG#us3;h1CE$F5(xmJB~J$o)TXa+T?<y2<`
zh<oYF;x$t8V<W#c)ID$CS%qJE^Qbg$*i^B>j`^Z5FNUr{S~PMV|Lt%X6@X{=k?D<k
zAt~3n8k+M=fhnmQt!CVMMM^`-wzL-E3Tb0>xQE6Q3d)?EXWEy`PlH&@I_=`yy5w3R
zeQ}&8Pv@KO|Aj>F909TUCB<R4ouVQget4e88&lzI;nU)vU;k8Zd2#h>BTsrX^I;>y
zouHYw??MjN=a;w=f|WP-CC9g%uHdO5DO4C&t=T32@YNom9Pg5D<tWtH^{R)YaJyZ7
zn7|J8v%P{8;e(1q?Ag{zQBVT0!v~`<CH>~+;lV9MJ6Qq2H1mr;ih4_L{ao@Imj={X
z^o{MUee=vm6sA<0!?jcYj=63Xu+2E+Lq(gRUALO>u<%@3M8fY1=i|iVIO0grUPh(g
z-Uhvq%`-Wfy}Y?1UvL=SmFRVjVr(n=IoX!1*1b?a-C`vX6X%9@I9pkmZKuK#S#?^?
zn-$<`y`<3MExG%DoALK|$E)M4vD%p;96wi~(AUo#Ng-p~ZPhHXyneq_u4X_S85f@m
z=P;L^{0=`n8wkJu@WD{=N}{yOps0&ng#|js?bS#{4XG-@hb8m7(rlL}qP$gF1*H^}
zIB)I1s5I1}YJo_X3(w^(-d~g@ItLT0kvIf|>mD8dW-$2Vyb=XPFSd3p=h*8iah%=w
zS8$@(P|$?kI(WZ0`bAlGN>Ku43KY7!h6XYsZmSao%CZ4ud~yLM$OEfGggo0z0oXKg
z?h4<z4B5p&(X?8nhy{as>iUZpzIl&dhNgUgjMl~>X891^qy52I6Q{L9=N%bCtzQ;?
z{~@~Xz<T&>zBMd=InTXX6kT*UI8SkAa(O40H{vJc+<ZT&vQk}jYO=(yvdow|w0*JB
zxHAwXTi)UULiNMBM!KZirTSnb-5{Xnwy(DmBOX399_lNJEE$|qJeF%*H6()+{&5U{
zz(U>ygZVNv+%aJ*J}$AYR1^6u+~0oiK;GxBqUW#Y0#mRbKE4B@9_Az9k78+KQlpO#
zxU)73mmu?9EOO)@o%t(<NWIht!eGD;Dddm4nvvRX0qQQvmiuG-H{ziDB7P1S2WH)t
z-Np_H#C-_vsZgnKt$IZDSJ+U|O;H)n&Ng8*Nf#yJ2qDY<h50r*IO!O1Ej)OolHoxG
z6eU{KEBzvx+ZdT<9(M?b;boP@hT9v=ohb{{REM2~OEEQ00Y{|VX^TUwpA_!OZ_&sP
zC5zJ@8jlD*dGX?3VHnCR>WzG%=LfNy4qzvW^$m2Z-k&TrA2GiyunyNqWc@_Hjzq3j
z-N_V+Rp84g+$>mGTt(PVk8r@2+Z~P3YI(*Uc@Gp7vbwg*h-((gvC%3Hrw8$-<k731
zWQ-S5P}7UUUAdKS-gzNO;H@&o$$V+eOHQGBR8*&vGlbIt|42~*8Cc^W{yqcDB<v|b
z>yfpm5{xc?O2{*%n?FxYrX*E14aw!+PhuA4ii*fXKYz43A+B1Yzrvt?c>r_DfHZ+!
z$KOax&KNF+JN-J9HlrW760BuHhLR=x^DLfa1Am-lZ25D5<k6M{UnalZPJiS^nIR<b
z&57^SpR@vzs8I_YGmt6nsG8Vs%hXCU8qE~kMn5?>k=hrvF*8+Hm!X!KLuChcXY`@r
zp7;zxRF>Uhuu~ICr{2d~_r^<*_gE!FFx-p-MM?~KK`9YZ;_IXq$;71ReznuvcAn-Y
z!%E9giPn$%k7U%sI|!%QC1lG?NAja#hug>e>z3NkZ-kR|)29lnskx?)Y!xflVR~iR
z?{&ZPAu6jFay1V9bzY*O$rTA?a3Qmjg108t0#a!YTL!|NOxBM@L_`OLu&|D$4-q9!
z=Z|b1H-%FT^fE0xoSdqhspPK}F}>Q{QJMw6XWx8ldry?jbXl!?yhid57GGBFG)b}`
zTQcM-$9vQLU*s1qH*8jj?E0+gW&pllzsA^iLaZ656zC08gVE%d>C&$B)mwKyd3zZS
zRHHj>RZ-*R%`!ItF>b2Q8+4nNud)(%@+k4v+;**}61`J_th4K2dSn-6DCHHOEVkSE
z!Yz3<J7Q4@g==X>T3%YwAkASxXy&M<@?H74MM)#gYAb3D+uOY8kb!}rom0H<kOa4{
zG<R-#=YaNCy!sOs>U5DTsmC78nrXoZAcEW?Oy^Yt+t!&0&gu?45^*Nz97syvEYg{<
zQhTD}jg2kOo%1d2ph|F#T@?VLnKue>`>oeyDHzx1!PZyEva)K3=&*T&%krY!+;H+8
zqm42Acits?te8{!InKH=hu{Pj=|F5~dW3QOO*c?>ZbVN;XK+%YNO#weI(^LQUQF|j
z(fG@(LQ+#ADrbp6jnlI#nT&i>>k2g|gK?~o6s9RIYUVe;Ms*$-A=PrT1gTlLRbTN(
zuWmi<%;6OZg4PFTuB6ao<#%j$<Q6SCg2r?QFnBB5jM+D8cA$<O*9SN4zCH)xrp|BD
zyo|D2^|`RqMIk!{=}MBFRbfo`ncJK;r?qvpyhij<=N?W+;RG^8=94<MYWhc=lW4Ki
zp!3C3%xY!u)60ft;MFSAKgr5JctF>W&qn)w9a^S*7;aQ%(o(pjk@q*AOR4;At8|n9
z7eGnWc_MnqoX0BjRt;wCwpq#_sMpq15PGqAv~*RJ7wobjLsf25+VFS0Q{lG8?aLNm
zdxQ_3c&rz{LQY87`KaS>$xEk@lo`dGgL2B7k<@v<zYz?JS3aBfPA*G_L%gK4aQmI@
zWT6{;vv>B8gI2J2zTvhio$o`e$ce4tybs<r(kjuB$JP4>1)We2NkzF6{|%*g4+hZ1
z9gl=RdZMaWM(WRb>y>KiugRcsDsoNYODWlbK?>n;kx}hg8^@rjBM<MJ=LaV8WTHSt
z-{^6JTH2i_CHg#|zm)|K8i5RgvYVjO-mX&PDzq&$7%w63=Px(NRU)5mj05ip{TcTg
zzS4y(MZ*XSi%DDU2uGHFK|9>RXN33{E~nboY>XSoiVi0BaK;lY)!r*)YbR%KBwbw&
z<w$I42LOcgQdp#?5YOp6xY*{>mG!;Qz~GW~l|0*9NU1xS24}qI1t%@HeOKR4ZKRgm
zi}Y_(i<_oFtsTyr#*SYoN}hbjz%a3Jsw}jQyT8j3=@3`lN`LXvvF9d8zjFj35<d>i
zRB<it+SCp-R7kH1Y7Q}pVZvhc4-sjRcQgk@6|O^|35X9?nvif(5lBgyIZGEZ)+=dM
z039AQV+Uovw+VRK3rpJMa`{-3;G|IXh=jos!ZjPu={2~bhKiKS>^8En$eiBk@1J@U
z#e~E7=MhUEwXbX$%2Y$y%?3}9CMENhb7Za-fDd`7vJaAUGX^ioeMTI993aq%uxF!@
zA4sb?FJAiiTc}2NOCsHll1elLsE0?8f}@^FcSgs!^JY{jC1sW;Mv~q1{*m_oJGjwl
zomQJcTDZw+L~ck*){FXeLRr}on(rMmy(u4z*3L|>-<yqBNn!<DOYL0mh9XE$%g5PG
zrE(l4;&?KT8p-<e8Y^nFJm_(4v(>?_jk)_J%C>a8AF_jp2lt>4LdDh-b~m2aW<;L%
zVD3X>wOeA%QV+*WzhcpK7nWVDko*9ssK_$@oL^{dT}*YFay40{=@YGdX1Yx0%3_eW
zj5w@hFdN#9TVT(XRSj=!C|;cX0%8U=6Rl{O#I?FfAm!%}<L_P>fL|lC0O}Fr`Wcgk
zevWlRN+2h#Wwn}G>j1>MYdGOY!PVZH^JiKmS(g+6qQ~{GU)=#vSUFo^V<g`B;6%?%
zgr}zg%{xrrK0`(cJ?QG!7jpFl6~j5NfHT__(x6+@ns%EJ<hFh$Yq<kA)`pV0Iq$H7
z8TNs;%xuKA*Ix!CP-4J%Iu&iCj>k(D4VWPS1P_CCoXqQ=c25^c#*?`UB=TQn+_9;w
z{=9EGR;#(?#h2p0Y5EBrV{Pn2)YiGEr4Xku=1|KxMnknwATe($<0eD00V;YM>|5I4
z{v~FHEC=s%#|mUaSbs^o^4g!K8b2nGo_(7>k-Jm2-M5b!VW?>B%K)u1x$d=nfPm6l
zRupJ<v@cuR$-SDa={C3kyBhE(sR{?kmK~Y3nkun{<V;Juy2P;-RdHMWkq}+01a0wW
z^k_=<i$}7`#7rf*{ULd#Tjcm>I#RvNW+&@>nj||F1IuKszOgb#u;gjLe$O5<Bt^qa
zl=LZ4_hDq~Ml)k=zc&OQ*=2|5MTh)2z?({h&(QIQ3fD|%D%_X<%c?{^jEShDJZr!D
zzOcSOiFtztdFf+~=gc@VgyN$FK$^x_o=t>*rcVfNU!fKlM?7_PhEG3Ly;QxQqNkVH
zxGNMM9G`o{mSu#D=h&Ez5XXFIpE`-m)N!`y@4a42eiJL+Sq7O72EePA`&|!9Q5g}x
z*|_aGjM+`**?9L9J8@e~nb-IO)=f@-#SW#r@m5TkPwgL7snk06a;`XBw4eg)b9k7R
zE7H?5yvt6{wiAK6F7vsZR+v&aQURV7&2c!gPr@kc^-2hP*er~k7uNZ%cRT;<$@%Hp
za$Y=zFjj8@9Lga!C}=(TG9hsT3~-B-b-W+`G5@Nk6gtkd4mnoeP`RZ~oqnze)dAUk
zTxtq#ReKHO!tXWpzkz896_ruz4Pf*3h||(QN{Sog#9AtZr|j29Gz^M#ORb!eO@zi^
z)~vfrBVp0%CxTjzPW&t)R4mLxAZUU!9r}dAJ#P8Q>6DGLv0d`l=rNtId3*ZZ!2C~K
z@M3y^8~N!wa+mfuoYvDQSIo<PQZ*U(bGSNZ8cgAX39)r?XyObC^6{0CbBF@*(W}HL
zlcUm(ZeSTkDGYQVB$$EFi7f>s%X|i(O!*YTPQO2m-ZbQI-a%yIeX}!=e&?Ju&gmqM
zchLH_rallMs=l}j+`1)_-Nn3ytis3?koq~Tj8Fl~LoNq4qxq+dPqj=K82Q14Sy`CG
z*Bc{qmjN*izl)4k(P(<yHj9iP?>KK$ZPm9AMlfFKndg(6jMf@tGI@~L>oj>b<$N&W
z6;OC`x;BBR@x6qCcSN%lk(eYCq9I<@{CQFG5rJSKyevyOpO4AZ^l)nC8Ud@I$i$h3
z`OQ0!R&p}wI5yz58`t!+S^c=Jub9l;idWS`1xAWf@*6;=JV_`K5aIy3I%DJ{6TB0s
zjP0>yYvb+wYg)$RoV}oS?##MT-}YKRhMAbN68im#rG|C26jRP>?1^ek41$nTz>P9=
z`+Q51mjx<dhFSMGN&pkXFz6n{hT(P0TMUP-B*+~42IFOD{~bIB$YR`KaXVuSNm*g6
z-OR~-+S5$FwnWPnM`+?TX>Om?%6q!jG#PQ5J~qqTLCMFMWFX5k%k%bzp^SB~3M3Gg
zmMq+U1KUxruAC}AR6PxN@<gk1=3=8nI226SNut5$5RFwE<zR*FtBKlLLA==VQKy*8
z7DIF;`k2}scZSndOkQT^uec;{Q%nuaC+?JipkNeJZl26=ZC$v1X<J$tFL7u(p;Dyb
z++1c~=Y2!v`4J~^0%&M#`p_874J)oGF)b-g_#Co7bX+;OHSF~ro3KXLlhE~`g73n)
zhDPBoawl1Hcu73=<m7MVI*PBK_%I4udt1+N9GeM_cFAhxnXM>~#DZ_|jZ?S@E#OTL
zRh!4(YNeYlR}Qg(g$hz6P}R-kpaO&5>|k>`J|{bz8iXVx9OxtiPF7`wWrDa$p5YLK
z0!{PEl%#~XDyjBNBWtdB=4)DANf0`IKa7>5el4Xq>_h#B$1Gc3`m)XPChk=|B9opn
zmKWuTcw%ch{E@phD#I;VW5}0&g4OL%B|x)ALB_l<VXmzm%b9|@O@GkO!+h|=<@s6G
zBJ;vJWM@{g>0y7kNQKP?xEwDp6r7My`&mC!_zNID26~;g<79T`o0G!uY2MFF($WZ8
z!kf1)rG2~F{j2E|^iJ8X%4x`lx_^CY&^8Yz7bz6DwnM^=dAl0y_{bz6xL=?$RGDRv
zB)Ou}PgrFgFuXMmYW)*bDmgjww~4X0p#Hb=-#l!DQ$L$=C1g;<mOWGYQ~u>xgHxSB
zUAoAt%V6vvFN15uclfADH+QUJ|MyJ|rpU)~+}h*PtLXw+xC;vqU~@;>{cR}p^g4B!
z+w}`xmpZXM+#FDTWlk*h8#%Ca!_v;&hFH#=TP0j;Era#*&Q#Aq-d@`{<~P*Ph$CC#
z>r`r<y20I&ExZ+S{o=3ppLM<&Au*SgK#1Hc)~1Dmy_UCeQF61T9_h=Oq)%V+V}0nT
zb_^NlnGz}~rbtf}-2=Mcn!|69%T{?l%NrN9igdJYr8{A(jA{bX($Gr6o-KhF&=mpb
zIy@NoHQ}m$qwzo}zryk-@0Gi&nI9`VPvd&7$&Icnw70f}a)8apxryz)8~YA4txJ2R
z?y|>&9^U*j6)h(#CX1eqdo2fA4<b94PIVNADysj?;=T{;q!>Z3lw9ieJR|(8-!Kp>
z?A}UU1v(>TujVfT(MXfQX;)B4K%U+F+c@}1Wsa2&bhvGqd1hnCm<{zKaI!xwZ<|AJ
zko(AdN1eq0$7P-;fP9C=FaRXpPXnqWUkdn^{XRUuHrUB)jIEx&69|P4O?j5e8TQ{o
zR{^ILVTch-SRk0l(3U%*FilUqxF6V4t0aBe*iH~3bi6j~#AFqoDE4nZRU5flgONFs
zk3r4x`;#j}MiJ&y5)NX-diT)@1D20WPR(MR|1|yh(nZ8uIDI*$tP-LqCnkceQDttX
zg5+A(x4?QQB5f5Ku&}LkPP+D6F8kDQ%88wr|GkS7xFDK>@EuLCmY%u9acQoj$ug@`
znaPE|N~A?|{#ujpBenals9F3jLY{8s&>2D7^9?_H;-e<ZI9}wcusg-|nT(uTPC50c
z)d*TCQyZ7uxu>P8`#mU|HuU^RzdncH@QU~{GOo%lo}Nk)wrh<qtUUItV`)@Xd5WR`
zO{#^(w+mICRipd=u5<@~2Ry%Gqq}?jzCxg$3S3mGyU#8{OXHp;A0>giESEI|H5uqW
zO$=@)6AGGJ*=`XE;%!(&)NE{)g-259Q&PqW3#Y4PHgvV^a#goJ4Mknep{RHjE>cjM
z|0|<7Oq3w{JL~VOjsA73Zz@9;vgIs_aHjoaN#XXMic1dc=w}!(%=?|RuVr7~_48SZ
zwt%$iJJ>ER=3!#1Q$;;)&c$XQAT_5u;K7ciqC$-lciLVD!{|A%?)4>$Z#`TcKR<mQ
zD#Do%=PQ!TRQ?F=bzN&0zaH&=ZrsT}tzO~9>^XlllOryc|Azf#da?iS0hvc?)m>lx
zBJUj>ewp}<eo|S%?qJSjv{rr82?y6vIh|WjG8339L@Y*zV-w&>>Z<W*Lgv#QKl18b
zt`@P2kBl9*769>~q;pC9F<L&69i960US0%lm;w#8!t3(zn2uJ+bT*q_Re85hOe2vm
zwSM7(NL^fH9+?Q~T{nxCUF+-1*#_-(bleT{q2Qs6kmNAcc7c<+upj@q){Qs#cd`8Y
z$KN5e;8~pSlo1iJtx!rcndMxIxPPil!sbvyW-i+5@?4<RA*GZj&izLe<bPB@KJ&&7
ziN2bn;vr)5VqE6^8l2EQNGB#vFv<C2DrsfRiZ;T??13r?L42sFl0G|M9s?JL(mKg#
zM9m6gA?12y{k>`C)DZd+9TW$v`R`l{wPU3or~HgUx&5=+M)eR({Tn{ftdMAH7EUwa
z^UVAoyaj(;?3uome-lV^+kMh2iQgKzzr4P6UQ`iOw7o0+h>|>5`ekc($G|U`3Z5$(
zin9r=uF7_JO*oG0IEFRd%MjOsd;WXJppcwwWsvztH)BeC5C>o0S!lj-Net5tgZ{N=
zmaayA6|JPLaqk<5pDd2P3suU9^}B(7(U#x?evLqK8osZkLA^TrRTeB+s-J6CO0MTA
z^xY1qJ@;f{l$HOeSqin>KIw0m-ag{>Wqhq`FUB*Y@nEHQACJ?K!j}zR+Yv=Nwa(j<
z)CsPIwP(T!504@u_#?OSNX&Fe(tsm!LU@0&kol;UVo!Dy!s$uZ+dcA?Cw_prP}fTK
z{1VHey!7%wne=K?ba83PVPbsOG&0ct=(xGH=LoddPOG(&`xa64tmkOZfD@Or{mgz^
z%W_2m%jxWYHYb2~X@8A=V|8xA_dGOQJuo@UpU!W$gjW=u>VJluoE;}6wr`Eb#JtIb
z!ug%k0u5Rh3*JUW|CtI@=P5au;4(SjW3hXHvJ-zlRV~wamzb??qP0ivkj;E^D+Zi?
zzv$EY^E3Gc?vw7^>Th>~&~~Sq?=ml1Pm&O?j;lb-XDd_Aj!Wr!8Ek|J>yuwiBzxwN
z+v<k$@sple@_o~#>=$sDPy-qC!_&1A^0<STuJllJ(?~0W6rH)%Hcd07vK+1t96frd
z;5l)<z3R8M>T{yG77_?J2;DuYdS#5><tmj3Hcqr=Z}%51v<wmOZ+4?^)>__^jrL!9
zg#oiZlG8^ot%#pQ|0zrk_QcG20P4WE^F^?9H%TT_x1r(qEe?q~+>zuFrokrkt@&Ap
zsAHyy!SBSlZc%ma1<jjoIxiX+bne%`S98>6hudPMrR2o<`0;D)oMIjlt~7MV`a7Mg
z)6mco%umE^?2|4pT-DsQvad1uRGeJF$ud#0e3C<*L(KaC_Ucujr_ca&wE;dhW4JOV
zl~#TIWN*)UdYwRE5xdBN0eX9PW_^)CZYmiCgIi;5IC`}L=D_gw$E}LE{vEn`rH_jq
zw{_el|65EDhj@PXKd)H}<*yO{?l+9OY{#1v5!Z=I<iIcxYib=%>#=LRsODrdpkau>
z&Ec{9k!ZJrB;qfJJ@(*2e-f+N#&RCS>c9d|U15ZZ7#UD4@bD5yldHVURCksSHbzNP
zRl5C2oT;Vd*bU=k+cW9-<PYy(V=?k1a~z4d-0F3`#wMHvLssG{CyQulahix~)2$wR
z^_@+}8*R6Hm}Ha7S;8Wg-Vg6Ns$0Mh?Z%SWLq5EWEHiGC6~6l&v~M_=<|_W^*tP`1
z^Rgl0@c7PbAqjRvVs|Xb#1%Pf8s^L}A@7I%MCa2nGn|B^Un_#2c<W2(Drm)Wap=KH
z-xv%w_4AA;kr)Sf%~{ZaI3Q;co04Oi2JPOXDT%T0est5fH}K@;+~{Ec<FljL$R{5C
zhOAv@Cot{7!NXE&#n<>3>#7+;#`>iuP<N@Yj%7)4--O^TF^h?iS6JdL9ZgXz{UZ)K
zr}t~jt^&-{QkYi;8Nr!_Irwa0_RjXLa=k>TD7Y<xE`0e|qIZrzWy&xeUIq!bxA#v7
zKEC}QFMHzr<etWR!<SVbdW&`%KB<gO6gqY7-a)n=78U6YewY4C@f0^lNgQ(X*$${9
ze|nEr!<E1u17rA;ronWgcZL+QCPWd<1=lIKEHJTUXf)~}=5$k6(?^Tgct1RC1HDgj
zb_}oBO<yH{)PV1g;YHScr>~OELW*t7L^4es*2Y}-wI(JhWJi;Xdf`>s<&EI8QI<WL
zjn|m!LxI(vsaTi_IJInjKUsjB9~G{5Ae{oFQ>5P5PLKx5jWm_KVM|KMQ8T#M5eiC{
zq6r43Zo?UZheZM(5zzpaHy@hh;;_sk<8wP)rQ2YGW6htD9zMHpNQA1yZf)ryV$??d
z6hSban2^x@TJg2RMQ5th0$xsz6GP(m<kyP#Uk^z!v?R{itA|QhK0&Fe)t3`AH3j<-
zZhqC(agXg4-H7=fC*KKESzTQ3>gaDoQL40cju+m5VmTt<zknesR@r|MtQsdmS>Lbi
zA&Dk>_F2P<$xk7-Mv~>(e`o*kHpXA;*t@FaM>$D>EKJ1@sU%9)hPwhdw<e^XNA7PO
z#XIT5azb4B8<(EtV-rPZ9c-0wwcb2cD}XHQ`%qQ}nDI;9Xm=9c`b8JqtF1lcs;5U=
zj|=sTrcRo1@l1^-z#Cv$E1b6nzWBXSs*H(?z?UbISB^jK#o9~fNcC~&lWa`8i0|Nk
zVJc7)7i$2kO7hM@7VI#S9d5Vttt|H;r{|{}h|fXE#9uIZ&Z9*BIHM$)SEMAy5hB&^
zDCPaAw_<FTER}h}d&X5?d)YHRdYe2X>pV<Vt)Z~y+@@#cbjR)X?A(!=<r|KL^v>V|
zm71@YRz4ZFlgXj&nGcl8qJw%h>N=WE?4BMbuiwsNXS}ZV7fB64oz@UDS3kb2jx3Pt
zfI{B8qK;ZE(=FMX_SmII72CM+TR#wj!Elea`e&_Brq_@-ilUh8iEo>cDdf3KpnHcx
zL1_89mNQ(6iN-ABGm&#{e6Qtc60#9k*XWA#*rSf@VfqRR)ySgk_pa3=WmI<hX*qlj
zQQy0|C=-(Y=e{g!e~1s?PJ-@{>+4yaDm-wrEA@Z0%Tk_*_kpW`d5>m7B7RPtmrVl|
z0LsT_Q;Ij{EAi<Qi{1n&ad$Vt?3Pbt09%U;b!Ia~oR|bfW(K}{1Sx5b^9CE&X<5c`
z+y(EXuv(e#G$W$naPiqP8DITGt0c>>3OxEvvjQfj7um)Bod7;UdirS3!#|(QL`l0`
zcFEcv)#`S#qv&Il*7r|V2O4gP=`kmORm%~WFdxYaF&nh~Qehei39SeY5y8Nt>fvYh
zFgPZg8~I^qH|47{lKWPLb9}HSn%f+pHtKM4TwEAAQ}q0LN_wVPOZ%8?Cb-+WNdMJ@
zMB-ek@PbKzY#Jmb^*ZuZ=;xCRh;e%LNyBFRdHK(T_VR9StAb*!cQ8pFvz(vLUzDVL
z#}>{^!Bxy1H(yz7iz#R{`eM;Jp0Uw}bN?wz(XO3sCfo4#ddM@z@3#qTrcl1A#e*qo
z?r+j{yM*!g8fKnWEp#cf5<U#FHT4rG$wt@~u(Io$OF2V)Z(Reo(t8I=9rsn3OdgR6
zrYa0JHcrh3T>e^z6*Dq?bw$SZT;iE>%gYy?wo81qjp*rXp6&`08ZXmZ^L%5Mz}dQw
zZ4>$8$^T66itpn$#m6yq{kvHKK7z%&Oor&j0SbYSPR7TqDfqr_HVDL-nH{z`b8%W7
z_(77RBu)vbs%0=5nwahp3#9MH;><THp@7-oBo8L`=chUbdcs3{iK7k|<Z$})@S!-h
zqh^XBUzU*e)!|B<Y|TVC+l$O@pEg<)f77{%S;r|OIZVyzw@$uoC*%9K-nwvJ<XItz
z*ngHe+v+uWZB4dy*Iv%li0LvXtm3zY)I;*L>PW(eJ-EwpHBJv59XTWP_3fLXB3ld;
zUY^)FEqyPKXaXvINxh8#d_?WD_lCxm!)LWGw)1$#GI>FyN7Mw5KAu@NttII6-WcBR
z?ChPvfh%OUclEYTw*9#3t|7EcH6|Ks@J{w=H|28_G5L1<S?Qy#+QN2!8!n%VC3HsY
z&FH;9>~B;OXz|wWw{Xka67qowHnz2&uhi@CDEz413o1@xx8r@Y4CyNbUCrZ`xqM6D
z+x|P<C*CK=*V&4XuWq%E$<r#Ta*lFjk=^SXs1H2(CLmJ;TNm3EkQbk3;;m##6XMwO
z^}843^NOAkFI7aG1p{LoI(1l>TO7@BL`IFnv1HE=lYV{d@zQCnh>j7_kY?&uSG#MD
zQ<stV;xmyH3(hYo4f&C>aGiy`_{17|8K3XgpFD>bsK?}qgr*#WCUbM{LN2%{M0wL%
znl3Nzp?^|}!C$a2Xx)GD{$y;*6)WZGJ=8}>y6Gvw5wSw<LPb@9-cWPPFrm;bG2^A)
z6osqBi0~)D;UAoQh)V=Ls=_J0N(7cGD-XxOI@kHk)6#~yBBH_r?(CM97)jL_m>IXU
zw`IrAdZjrdOj?AI1wM)@=JJ{TEqs{3F93h?QCAQ4iqrE5!3QovNPJF_y&h9)`rDbD
zT?PApvj?p_z6sN}Jdv?MV=>X}m5k@>d<SO5X?`9hY-rrJrU%69Al$ZB_rHowVe~A_
zN`U7Yf#Hk2tREiweqp_JZ#w2<VV*N<t{H7wN$yOG9NTe5S}*q!E)J@eL0V1NSAYo7
zYBtK|hx3+|Uo+&{=1_hy1DwJc85finumBnN45kI^D97OIkcY3kl!|tApyZn@7NJ~2
zTOoyo*Vy-_Z&eVpyi!gMy<{87%}*er`9E%rmGF7@K153x;APU@PK+5uPQmRhP(f_K
zt$HyzML;-7z;WJ^D^aBxm!&wq92SA(ilAm(iuL=V*3(8=Rs2jR^O-$FD+)|>Y3`93
zQ-wjZNknxP79MxiIlmPeI1|w>?%q<6PAj0(000vksH&<sJN}vc&IFjMQUwZ&4)(Kf
z5o?i7V+vYXKhd{gKR*h*e=?@02bjd1o=!K4h2+K74ubD~xbMZ@eO)ev$J}G%DC0ZQ
zUp&7x-V;{q?WU4tfPk6YwWk$za;h9`?#9Vn+-@NHH-l`j92NTbT)jB2thGk@PT0=Q
z+@p6A$_uUSS){EBoiy4RP6%0;bvr$6D{`P=)(Rj?sVpQ{q8T?WEtt7>^KY;Y272^g
z;pqk*<MMI});#m~zZ4Xt=<2-+{=92x_q(W;-<2D>t!->Nxmq!W)l<sel+q4BueP4{
z39(56r&(|5|Dx6IG5OQ#hM8|F$zK`^7F^2Bc}GHJZCV~~4h0{RC5V@?=Uz9AA-myD
zA2NRAY)(Mjp&>2U&R9<rrIcv-=vZQF3hq@~?&71lRx7bN;Ri1318$V3yYf67`n+px
zb=A47Np?Rq!~k)cP~sgZl_`Vda^u;%&%ZO~7n=J}lb1J$0PPH!tx2Uy<SQcT)7FvZ
z=_>#G;-d_WpM<A1C55;_)lrW`^<WjP4w}ky)I43qnrVwvon&^+GVUE|A{6a?Ol23E
zFf$JA47u!*Qdm*pV(L$cu8kp7h~*UlctIN#j!}d}3){0&v<OM>cf%xnBDx9-npQy>
z)5UqA&{n`XAa_4S$U#Jtm6GnykcYGNn!!kV;}T<uo)lOCOJf-tnV7YDnce!w3Wju=
z9O({dD|dLwQWQPZF*9o=@+`b*5$`3Kc-RQvaE<kSzZ2a{!HDY$NODC)-vkC*4kYf`
z7)+lx)G^z=*v8Pb=g=M)@agJEOE)O0BwCf)??Sq0{MTw9f3M{SE|9dm>*CgpNRXbK
zckA)$W?LIBqSz+Cv1KSMZh*OgxoDIGyEi!hZs7M%y9ZIybdJV(?}@youmA~q5&n6U
zDArvd6`F77$10jzf5_?{?!)xz!ew=B>;VU@D7XpE$MX}h$me-EYeMh&(A}#m#fTtc
z0NdEcQuPiVnR*9!&e2lDj(f&(5CAsO)|vfO^L<{A|L%VMO#LI^z`+!&<spQWJwJ6_
zEy4iFjmOq>)mXWs_^Qfl7ZB~uYMjnCTnNEabbWW@Hyo{RvL&Yf%B|Sms;Pf6uEF<y
z-qK<L2m4prF;1|BGV%qYTbWn?Ew_aN60tQb=p60&<%K%Rs4Z=vB*xL4x9F2<+f7!H
z5<+fSSzypfsw^CnIds&rG{Udx=Cxd&8psYfT&_~P-K%X#yO%6X&5)*)r1fz`3(C#r
zBqL#tO0PYO*zGcXis`q@Zb_l*=7>%eDjSSq(H)n>*3ya$FZX|N<u9xySalX2Ssv+?
zWpZm5Y&=)d@$$6!GO@<8SxzvLaeJp{-O9`Rwh8Z5Q`?u#x3Gx#vU#PUGVgjKT+$zs
zACqL}mNwPkF@4TG@diEVno_SBn;|($9bHVaGq&TV#_{_ZD#C7b)Siv^+Udp$17?m$
zw6<d5>pVdiUz9BI9%<ayY0WJcq;RmG%sL}}2*0ZSFBg7(7nFeqZAR@T;s7nN?i3yx
z7ysc`(W^wP%|?{mL*a=($XT}>gmr+uMBH4fx^sDONJHK+-X*;T6m3^rp#M3tXS3G9
z=csceL`ms;uw2K5xw-?;1afU<G&e}Q0w!IidF*_F5%W|!UvCwn4W1oeg0oQ5c~=<T
zSoNO+Dp^@xv3pv_$A$e_6eIhD{$-54WWv&Xx9Ulc(|1)`s98YLY+f-<Z1Ft4YRqN}
zR}r@NqpPF78{qGt^FwXpiHJ7xl0g@;qn&2N=xq&drx1@n!Ut+KU*pyl_f{mMQ*4PP
zsi>ULOP6eh<Xg-q>w}wg0W_UnpNe1c@F+$MU+D)u{#^)jyj&H2L`6Pmza?00mi4r>
zaywXzGC4Z*RVHVJ3!zgmnb*T1$L%vZPRQqGt%!<3rnX_9qF8Jm`zgQ|_F;Io8rhYb
zL88xjSctH7UQr%xXR_&=2ANl%G)m?KRJO$Oe%Q^CaAE%_=o5z%+#iKcz=SZs%4KS-
zv}u`a`fgI;c(bg_-Ekk10)cOLZOjiv4Bkq!mw0^9UNm#%4v0=c;Fd#~B1bOKbj{-d
zAE-FwDQ+T9YZk2{L6sI0KVp|IC<uRemU9{l+xd7tV2SpG?WOPOROPH`k}E|OA27;`
zC$4CiV3phN*pcZ||8EIQVm+ML)_5EvWu;88fQJyWwtEz+athNPmLqAOq7js#2GF;4
zjC0U%W3di6g{$N4V?)*$sMd10vgUY~?iY=)wt;>};$TZrhNOK>tX7G2u|a1F6W#U%
z+QY}8n(o>0BtZ{H5FvYTT(HI#kNG~PDzMPzKNKR3$+5Tn<s;T28_#X&bi{hGwbj<e
z_-n)uzDm;AITn5Oo&?B?AN7~|n!7#wI0vKp^TK<+r{2czZ}1+3U};Q~z1{$;`Qn71
zaKH?-er}5|PeTrmP=ua_T9v+8>Fe^TMzVag=pJ}yAY>;_&`QKB@LF2VnvE)QIgdeX
z2hQzSZuaLEd0j_aHnHfCy7k9MQ37Qb*B8Y1b~i2}ZCYyBbO;reLFHTf!hWCMLsBYu
zbzCG?Jj!f^m5wcxXJ=bbv~c=oAww$DLI3)R+{25mOd#BGJNOTXP)G(5czTB>55LWn
z>&TX+LVI()E1Y{6vaxsQ`Q~L|VZj@gg-fB8)c{@_PX86eC8|zl7*faUPh<X9mC!1-
z8?ki%$9tfgJd6I=a1Ekc^4yb$2@98flg-f;ue@b{XF!XmXhKG-Wxr)|m!b0@sZ53a
zfs}ht)zfGHQ!~GrzpSU_dE=)Df5r78_-Z;KhnNM3rDyc7k4sCwjD%U0tEl9!*S)aw
zPN?l=$xRdqzkd(t_G53|p1D@@tRp3ERZsIqfhF_{6$DSFi1zIjxec=eZp=(M8UhGF
zb4|eq4aEuj_bu~vM#2h4W20|q3cS>GkZ5OZ^|sv9M7(Uvx_IGKdrQ-I{+6KU<&`~I
zt|YXgowPXb<X<{%r1XC;AA$E|$i4g`owyC6D(OQhDYb9G2`2pmgV^LxhX@}I+HWpd
z3o6vKpq3}Zwua?aNoUAC4}Lt!481$9PLbg{IeUqU8~;F0e5k}J2t?TWVl!upksk?t
zqUTxTgmB430Z~b;8^w6OUgx##5FH{?BzX?`ifYW3_K_xs;<TG#<@AX?X-r^x5qA5h
zX0h?g*lO#oL(7se%PmR^GJcQSC);D`OarnB?=>`fk`o!<1M6(eM-_w21`DSbx(r8r
zExe9|QWf@BS2U#+k}~Wk&W+|ay$!eLyGmmVsvQrWtvgZ9ha0o?lDuu1`>s3AdaQjV
zx0OAM{=ck3E!9m-{sE*ox6ukBOr2K>frP$Ryz9Et?TzEUJ6?}aX18#%*W=sD6G_kK
zX<6(1Uc5hm%rt5jx8s4@@SdAWsN>oFY4wuQHLB5f7!2ymg4Xxh(<tf4IBy%FX<i?f
z6zR49%976muIfYXs~1W#8a*Iwgqm;Kq(y9N8j|J*l)`L5tj5nY!1Lk*-EQ4K{uo(?
zQa#3!GTlDK?F#fdgk9;g@QA&ov|O%*Ww(lFP=KI^kC@<lM|8Adv*ne1S}qQ2570)f
zkY;*sjCe|T8&inLje2-`fEjQ0QhVVo?2u+SjS|Rr%iFw7o!$~sQqoDsJUwY?gZAB&
z5|rr!Ib6|*Ux5Za!vi5^et23Idd-EquUpKfg-zYI9^T#wlTvDGqOUr+lDS<PfeV>x
zUNAfwK2!7(NtI!6IcCF6ON$fj{B^t@5GOY5Dv9a~rUbBAZ~HCaP~TW|y<UDmkW$ud
z&utf+5ZQYYK_7)PDrk|z9%N_OY6xk{SX-Zb8QH#5R!bIVYBpMGh&sHye8U>U%$z^d
z&<sy3D`=o5b8KmS(q#=~A2x+@%^CmaLN8!HZo_xqU05C912pOw&o$&+vRosGancVw
zSGJF^TD4V?kHP30$7YI25#eS)a8IWmOgIQr2Iv2>m!IkkKgVKUI(V=*S9yrSLpFI2
z1IFTU7ByT-kxRT492r@(a)F-9G-E#DBa^D_UKaV@$I2bTJu2<GZyEps!=)``7$Dr$
z^^%jDh?_+=VdJT8-&>F{vp9YLF0z_(wrSboRJ6^*(3kj~>cNXcgTqM7BiIM3;${yx
zds0$@<>bVyrI~<icS3jY@&)bHYymF@$eZovm(DjN%M)eC2L=vD6KguoK1Q)P-t0JP
z*U4vNgXX+DpA?@u;xLkN)qg$tsJ({?l7QF=(cOEyuLaE}d)#efE21NN-|$;7@Nk7L
zvkwzL40_HadM&b`kdP2}s(&xHhKDomr&z1z$lh+=rD>&`-)fv1^HM7Cw>q5x_@q-X
z_oL3?e13*tlpy`ME2n!~F6(9QV~gVHITRQgq3-Dfkm~ZfC<y<uQT9i46bIl|Va8zW
zP|dpf6LO$iz28VwEI1Irf9mtS%s^tL${=xgqE;~ZOEUrug>G}OMgumGmbR63>HT0k
zIn&m3u5}Gw+tfQ1z{~XKHiK%T4iI<`1s{x9@3)0TK8+9U^fKXyu7Q!gugzOh(!LLw
z)x^S%V&Qjp-?(_N|JRmp&(%MU&AEs5QT&#vwaFAbapV)A=p=Qi{&c|0rkiXih_hCC
zox8O%u+~XwT5KMgzE-%=0hREHLZcHU-}?}AqlM0h&kf|_qolFUlL}Kjr7XKY+sLwV
zYBfUIVmMZA)c7ZhFkf~8!snBaKaa@qukM)ZoKp#6oT?;om!@|)9G=YWjB>}t7Y+Iu
z740dI2~HK=hWCq>X}`|RsB0beGT+Yo$j9W@oZj=tarF}*IA9q#^1aQi$@u~Nu({N8
zO{_JncX%9l40^f+mtD#L%E;5Y^;TrL0U^8&I3&>^-X`aCNI;p$HlB<@Xtz;s>+lP2
zPOd|lJqO4iy{8O*gV4i7RN(g{^zC0*Y#P?Akj?h{_gO?1+HBY%MvJ2#j+htD#T=0-
z(RFt@xDGgIB=9^T!rUTiUYp;Zeb=y)?Qm5^c49X3K9DNal1JE{SBIa2!_9e%0_5Wr
z4Q9$@27d@cqEvlj!R&#L9>$Esk8^YPcl0M@3kx@;8q6BhW$m6P<=Vp-T&~$P&Br?i
zawkrK77Sd|2%(A2m->s$P=_xKZ|S-poWBtIU*f?bAINVG&+{lKzq}0XZ%0;0UmomQ
z!q*w8E2z`-^8GlnpQ=sq5=PE4v4r%GLEWAp!}DFEwr6*iL7b`P4!B+^jP@A_f&&Bn
z2x(VaI6{ai(?}&KU3X;?$6)tuch>j{jtNA;`;z_9$+ba?RB0MzG0cvfhb5Nw&pn%a
zyR{n*-D#@4Jd>H!#s&HI9Mfx40XvLd`?5;pM$Htj<>a%i15#zw=;<BGKd4Sat7GBZ
zwC8rmmMb!XxK!|uBO<#Iw9jtFucs3rWV`i6(uCHHfGNterm6_AA)ssnpG+<N8su#1
zW~Kv!>0Q<qkPkw+jqlonXT<u=sY&L|cv&he>sRmh*(4;siMYgLwAz)y2NkP}&v51p
zb;Mx?23|I6UHP*N0cTKf0|*!1D~U^fR4aQ6dAvFNc5z{G#3CIXOza_FOuh&Qff@>{
z`P-GHdC<eDm>s_sKA6<u<WyVtij(!;3}rp0JO$gdvo1TMWI`JP6OYq@(;tLDAhcN(
z7Sy+PhbO6Xv9OjGy>9<MF#(GThzXi`zY9Yyppd84ra44vBZnI15%Sr$SoZc~wT<s|
z0zLCh`D9c4WOnbGg2+0W=x&RF{;gSkBjJVZ$K}fM@sdeud4~?lwlzwMonE=S)sdbB
zPuM9bpHUZEIE1S!De)$@%)H_d6@4B{B-~!|wEAAOvPx?(YA*)H9~aBZ&JS_{XzfVh
zU5Hx`_t_*4j0($ycKHrEGso+EjfHZ>Uba{8FNv>+Ffr2(wi?3_u7pE*NpxLej(Qy*
zHKLgdgwL{15q;olfsQ_tW5t4^LftcWH!-rlo%fQT{<jj#CBaHO=_DW`co8Z>tw>|r
z684(Lr>C9b|7+?iz^dA!b+-WqB8r6cQ9w#cx-CFJx?7~AyJMjUN=u1!cS<)X(k0y`
zN_RJJY|p#rUO$kn?6ucibB;OkpX0B29#~gH#z?lx+Jh9{R<97p@QC#Bd151*IiQD6
z@F^tNE3}4;TuQ1$t0Et#Xi52uA&lO0&MH01rtp!aqNbQC&!UqP)EpouQBYF+xVQd!
zrKa7lYy+kMSuHn7e>)9j)m=BUm6ARg4sFeOV`Ebd0SY+&=N6uwbyP4I5?0F2VOD5G
zpViFDh2);PMiWJk<M-X=u;DOyzUW)`P9*tWS;m)+6W50w9hMFSTnj}ON~aGOs_ccY
zV=OjWLlH)=CLqx2?0l4HkBsU+>v=VR8ojJi^ug0}ucXzan5cc*#&%_F+S{D=>Td@=
zNCQS7D`k>viVT0!%FG&F++1EM#3FbvA3nU8pn7TPO%?}nX)JHUQF1en<>I5Qt*x<U
zJvq9XOJl5GrJ2deI4@)M9S!vDT3TQJC(&6jK-NZTwR*Yi9{%IkN(4gdqwR3!x};6f
zk%x~pOI<ncF021P=LEY;-91g1KF{Se#UI-ZEuUa3+6dkp98?QZkBpXAlYR0U(dk(o
z_~hO4Fc>OL<}}vpe?WCPnrhscNX~k6>#d+*Ng#%!k{R9*-Yk_u(Coa@@c;t(MSs0V
zA=mw6P1pm}u(fRjJ4wl~*Jj;$aS6UHd43piIeVImOA2_RN2Q_;a{=Uu+eB5$olXz%
z#@&L<f6o>1-}jwh-TXm_H))4-ba#F3Ijil!HkS7AoLH?R+4{dzDw(+#^5V$>){f~I
z0k$PK>opHT9P;!-hxQSB%Fpk%uCH`<K;d2gQxGAUW(+H9NU+j<!`8$0a@SL~<t(ln
zytkAGY7MJOcZv7E2SnSt&0(auZLmW5<w<HzO0plDrz=to*Ecr(_%a);>^O5A|K@Z-
zgOij`Hs9T3DVf|~+H^X+T;16xrX`0I7ULV^>9T)*M$T`DOW$lSgu0emu?>ZU7@6xz
zGIty7n5>COTyY60vN<r?*LoN4^mZiP^@(}#9&yRObzw>3)6`#jg?pnFyLJgjQ9IZA
zwV3{CL8fp#Ak+^aP0y;y<IPB}axe?A0w?4nGwqN5;k(*gLu$=5J4<gbbH{}_yX+R;
z>nHA9oi}Q#AI%?<pvJ8%Gm?BFj6vAPhE0Cu?3{epbSO>sl)VJ%c;2;}oX@5lqf0&n
z$xub~b+rb+oz&ts#l81GSOQy<Id4BC6@$=xT<=@kz-9guZizB;(d3@~5ev&~6f7|u
z$xH5iL#sr4Fm*<0dt^oiFc+HIgj+&<>PL=FKbzYx!K+`_uOH66=0^Jy6_-RIe=3)$
z1XVAhJCu}_#irvA-u`bO_lou_+B7xotu&?Gp*f`H_YFe2HPE>Rk86zT8?G;CDmiVQ
z**E+CfY9^GuRDjoO%+$}xZJenr4AtDngz^>{}O@R{FmjM6n|RtYZF0%c9}C(Pf~ul
zCpf8c_x!y1aca-4sHiS3!tZEY(g(pae+e#&&DQbeoxD#m!&NjQkBA~36_&>Y3mb9~
zdgirTAi+JT;!X2F1qm@<K7D*V7NgA%=t>YEH?$!06SHTwQC3#Pz?A&Ft7*9m?O6g5
zyvRa{$%-R#&n%j2XzctaQ>?1P0d0FjK1)k;d%`wkq;;WMgwl#)e_{gEmLgWuzx0(f
zE84OrJXzU{4~%ty?Tr7rN9O5~aRB0Edms~cye0~fgS7xx&>&`B`3XnT61iRPSK?9F
zyUJHwKo_VYyO=??8B;+x`=cb$kJGC0Ot7%jT#V|@bY1M{6%^Rpzb7w@b`)UO_KJ(6
zGn*AY*F1R#>I64V?b-R?Wv(MH0SW{)5lAJx@44VUJn=P4>(X;Q17zD~7#}^jllTam
zwNY&M=<|@lLIRGJ+?<86d!HzEZlB=Vd;OV_eLnyDwTbcf_K8ZCA~S|+(={#Z7zKq6
zC_z3HKqB>LWZ$P5Y`pkjkEcZYLqz|-HUEBzT501t&W|6HH5*oTe=J>bXCg<2p~;TR
z68G6%A?RsKzUDyfkZ61<cau@w(x8i7a7DR9ZLjR(8|(JAFRTd^aZ*yw&0}_Zq(QE=
z@1cARMfAD;l}vxEH&)8lgpVHt|7<TJB6;AiHxIFUR8%3<ZwYeMvda)iSsZ8ypwMYg
zg<Tb4$Nc_ZxSwOHZq$dQ&n*QoX3MbHe|wE|Y1N2;EZmR(g4W3W;%iS8Pk%yAt^DXW
zDWJS=L#}6oYw{i=m-PBwZT`c{fKD;v(&dSLJxo8jFBO~oTvjZ4PV2y>+$4aguZE<{
zzBC0dVxzN>#j^<#eGdR5z{18lx4S4`+=R}Sl}pe-I>3UQPqJbC){i+tWa~8bDpmbZ
zH-9Rd`L}0mD1bikTxTe@I7*X_lxyW{?nZ5N4EONPl9O=axiByQK7=OcNHoM-&GO{o
zpMMiHsgaRS+}KMBzNj=Ysi4=n8{^9Dvx8?4SJ}W6Wr>$vuHfCC-FO=l$A%5=6lW}V
z>#VlB;=K6eAreY+VPmu>6qdI)zk;U6W`4tYe_E|o(*r^sw6HAD%)>_age-Km)e03V
zUn;VnordXujUFb{&t~4TaB!yS?q(kjK6pUt_U4hN+fF{MSJqIkHXxd&M6fZkaQW{X
z@AQs6U}f+diFXe~$1e+s)pvW5Os!q<4@b^6j&oE1r94;&!d~HV(6Tt$Qq}?m_yzpE
zhd}M|<l1v5eQNpA)hAl2Vlr4ln3#6`B<F42?d6W!1e(zmR&Sc9a<65&jC}c0pmXxV
zpx8uM&78bX=;XNtZpTR1=MR|Js8@4=LXh}2sweuk-ZON_vvM5@n8-kW_Ylc(gH888
z7YO2S{IPsmI63ShtUtfCQ%i}e=FqL2l-&AUvvU!`FCpDK2L~Nqm;v&TJKq|dpjDuc
zsw)3U{n+7JfK#&QubJO3jIFK}IjlX)R2p)D5@ti+@l6{6E*y0UiAfj7&s{2f3=#w#
zQ9e*bS}GOLYn33eMWpK9YwBq3*u|KdI=UV8W0urtgYOwxzTgJQ5Q4qMuK7#H{n?nd
zG|GljopHYDCHN#k@xRAbt9F`j7JIJ|@~P7+8half7QlvzdAOwIwQQ=>6_+`C7och4
z;lx~5Trm4Lo&GBq-+F&6M&;;rC8@ajUCn$JE@E02K>RcsvzAFHS(71dwf8pm#@rUZ
zvo2wth6FfT%-*)I)uKOR6QjcvZIPRp(<R#O!kN9A@&j!h{+(D;BKLy}exK;a@mb<|
z#T^hs^6AEr8&(8cVFBbWTL|M90)nJo=#5OeK}b<(b<(1i{b@H?oa%(z^ww%@EH`q;
zEj%~9Yau}UYzymv%S0rtI;_@&)0&PlyNi{*z86b}|LguO)H*y;QOuKoYc(#93>RK0
z5M$pdUHP@YMBHx26ZYvx!c^Hm`8g@l$M)46Hc^uYk~C6)iv##ywF7<FP+&}s@)k#4
z!lw(%oK^tu^?Cl(UguuqA%^o<45G5d^FTy|^rN(1M_TOf@bKk;D`4=xd_9#O)UNS?
zQovVcq1+;cW}De&mI@1t5I^6=BKteag*Hp1=nQvHBo;riaTn}L^rRbBv-8P|4dnyl
zWlnGc?%tV(T<sE5jWrDvlsF|x0~GGh)Nm}fF!!7$|JZC{(8I_3g_xN%EPZWY$(!Y!
zYJL7dw{v*LQB6HKIHG$oBO5y^{lB2$A9m`$={1aOCHYK+-l?%rP~(`pn^NoMpiZqB
zI08PFbLX<s@$lsESp)EyZE|H~Ai=7s%=C!`Gmd#mO5wI?#^mtrd*595SNdM0;(q=h
z+Os)?*g;xfBsgi$<Or7Tp&*esBr<U0A_GdiFbwNd-vxMiO|pJA9r1<)&A{3>sK#?1
zxOVvKx%Sr?4{QUXdnq7XJT>Oroo+HO4<fV0Dg#RCb-t&zQs%k%ZIGmK(Y=F)yjYM5
zc)>?3eob~vuhe*=iSAs}w6_q5fKvo{8lnmvzUELlW9nA`#sF`|=}mLI|75u*l))-5
zAJ}XZ1?}$T5575gJxf4UU)F&6%Fwuti?CtwWNh8E+&&anSU;C@6T=H#1B*q}n+Ia}
zF&$kt9h>dLloU1^x+a{Px9#m_>J2HN+S>uBm{CUc&XDbhl02lalYwC#&{ODKNuc`I
zyX|_W&iiJg$4C8kBWRE-=Q;I*oh7l?I=2Os%%&TkHCF^jLro~0E_|gNUm{46#Bp7L
z7Ls;3$T9X&@lfrBNo~nQg+kkQPX81+!NsZQ%At{=!PbGuk+){FpW0PR9pbynyE)A<
zU&*eo4Szuv@`kegVl(Nmj%}|@_hCoh_>%d5)*Sn5Yt+VLQz1734;s%=lCU3dvo5z6
zb26f}m$unRSDUhS9)1zu4nJT3h|JV=vJLyF-KgqApS+C`m(rh)w0XIbTSxT+&<jTE
z^X&&->$kOKTA(nkQ@wtbOe4m!3@J=`ntDGo;|oc}VcD4NwX4$`gcQ1pqSy+6?OXe4
zb5|@Cu$Hk;<}a}fm+=Uea=vf;5fT&Y8Bw%G+{8&hFs@eNZ$#n#9l*${w~0D^Ti!r8
z3)m#T0H15BsrnkNZJ@EksW<@hc|0Gl)Ai{IBsvC#*EbfABp<331tOF_t)yZvn<hK_
zIi+rLM@NTh%XUG%?NN4U3Ygl8WRZ=DBI%)#eXnJ%mX_}H2tzjV<)4AITIBpS>sO&h
z>Et-%4wW$8a_$*pqbHGauu?#zoNX%!)iJ<R@zqq~0^k*rdG;~b`d{O=uAyqiN+=C>
z^KZnQW0uJweU8h+Mw5c}QTAPdu=6N`AFF+h0c)3$-LCJ*%V#wmiet36D7(IFpkv(T
zjbX9Xn8xV3bL{M1N!ltMLAePuYu#N040m)N{Z0kMo{U=N<pha#6qpQtvJ=;3fv!{g
z`&-umISxAd=l99MX?HZvUQs!(aQh`~z9bT>vFs>oL?_xnS-{1nZyk;DY>pgEqeiDA
z0q6q=a3q?z>3`iR#19}JIM2m*O@4d20oh7+H{{>kZ&rOLBFld<RLv{UKU8Hkd!Se%
zgpaS8e-LLj_NinMr*^zJoWg$f9vwrrlG7#KNPob@xE}V0?ieB3b$Y7-zwrQWv<Q`N
z7q&jAw*Qm36ytua>ic5Gn>3t-bS^FW@Is?+>Oa!b*(?^z;P8j%@1)nD3D=kU8oL6)
zkKJgLw_!D|XIOV|c00n0BNFm9tUG-xFy9rbt?@n5h102slPg`jc@NHBY?m94s7OVJ
z4(#X6X&sFrV^*)4p5JVuDFpl_6gaH!32f(aGLo$jt**DTY4ZNkF;m>-5_O3hQ@Mj5
z|2R=2FgNu~v48F6O^xGKLi-q=_IKeX<-sve0S_xXQc<Rh#u$-zI@&Z9W70IPugsQY
zhC1mjyn>QM^s3=gA7ANz5(^gW{k<~jW}Pv&e<BGNUM>vW5fd<$Jr6fWv`&flYksy4
z5CH<ROSA6n)65*jh(m}S73dXj0KP1t_s0IC;MTj-tF2w?RvOh#e(q$WT*j0CEzrAH
zS3ymmSwnR-TrD8ftKvz%$W-f4o^iN#t%gf(Xa9K25_hynfUcpjZ%rhf5mHeBRY+%=
zCdjG;LxDI&Bk`X{Om9uZ#8jdpZ!auMO5g<XE@w|NaHP}IEtrgzZ&wrr6<1d~ml@5h
zbtOs*UHpfY&9CLav2Yvf>c&orF4h!pyF`CPc;v_`{s<)M-#0hmM#Q~^x)9l^tChnW
zl}-bXo$;*goucFO@I+iBE9Cr{TTP}yb}Mag=)DMCb3Om6aE!pY60V65$a7KS<A*BM
z+*Wh*Umo$a5^zaS%Jz|Zd7u8??Qn5nJ_-vsBSRc>t1iMXmdtOi<8@Dr>hFlKPjS(r
zRz`GrH}}|WZ2POOO?ZBeQr~tZsdXmI3Yn<2=lq=pt=qI^OuUYe)$U-;*(UTujgRx*
z|F4>KdFJfk_)J3kV2GYEf7G-s=^Cyi2Z=vj_$U_}17m(hAe;G)%A82k?+E1tg-jOn
z>G95&FBe8DsWsKPOhe0D!j<}Cq~$fLvP|S`0*Kqj1~4p?BSgBBq1K{w<Hk};eL`Ts
z()?Zt(r2{}XJ0^RZDuta=yD9z<s{7vNzsS*P7bAO?b?yFPeRi{hU{ZODS4x|`LXIF
z!3d@P$?-e`((SYZ(gA<n>G>D`WxQ??OMr7#7Zc!x0si7kCpu!5a!h8+DuV2A;nncZ
zDD#_psFaA7&0<dep47oNE%b~?V{22|x7LZY-%6T+|2%6K(Zmj?i%{NGbqhFK)v(a!
zb4otWa$3_^QYC+yLBVw#)omirKb7jRn`su~AJQa*#S!0l-ZuY&FkDS?VPN^hY33rA
zd~XXqZGGD~DGrV3Q7Midz&uEEmd`ey|2NJz+-P6osstMw0?y{gmW~H410&P!Pv!eA
zY}ohIh7^}X5CU-BMv&1H0NI|+6*0kHa=B@#5$WID*f9_s9007t;Tjne_7^0w&p6mQ
zE_o<n4|gob=fH(%<g{b8Zo;=yJO(^aiLZ5a5;>e`@8&7)&M&*FVfKv?*NzI5<@kR&
zxu}7xt`>jBK0LR2WI|?=VPm8?^@LG?V#T4sGimV6u;a>QEahBkP617mj(lXT`A5q~
zaJ0AP{4>jkeJDi!!vjUh_2IeUy~oU=Q!LmO@&D%b#)ENS)ewg~gTp{?=vFypmkh@?
z=>TI2<VHRx2XIV`6+e*IY8z2lMOwp-T9_DT^{nall$3VfjHO*OJlN-8FJ}YmBfxqM
zP`sG?MKljCc!@V|d}6xWswc_)Ob2EZWk2qlDwoSy9>2<^3yCoQr9nlLnw3GVI5_r&
z8yqhA(*5nXSpd1s8cUHg3h-aucw=BGbgs~-um4i(<X(~BVa)K3tux|SzCBVeNE*rG
zlTlH0&CQW_xhde4{VNiGsttgs+q!N=jzuqC`%k|}K0|x&zg8prE}$!jfJMUGwfcuC
zc-x2w!pqHt4Z!3$mzq4oa&Ul}m=*WEkpwF!MxEdh_<RlaU%Yfg)GsRU?6~j>MG${o
zSoak|K@nf^7!G^@F~W&qNEjps{TF)z;CzyMHK~yd1>n7dx?--p0>drIBXD8E*7~(f
z`ag1h5-PWEo5@gsgBKrDb9JFs6y%*}AU=6tI+o3}u-;|B%<%k5yjMt}HD>h|FJHI!
zFB8M|Nuybwqn~<lI5(5h4cP!Ean%~JGmv9IJ_nF?l+aKZD7_*n#KIyvIf@22&f=5!
z{~r5$o%dh6S9fz24%dN8(K=r9LB@#qb5=)4#8$@{C5oTT{Kf;`rC-FYde!F(q^P(_
zhc7Xh4?>}2ExjV}!YDN)T_mVi_4=>;osCaVE<3W%0(wqJ1Mz)hdHn2}7<z&L(r?lA
z1wbVS`@55rMR4w+8d)N}tvZ5jUN`+i)qL*Ho&N7ZI^Q(c8g6*uP`KRnK*Nbd%-uc5
zx`kZdflU;##ZGg0#TRSEEJYerl*>qxH=)<JtBy@RAC%2C>`aNt<>CS*iPCSY=dJ7j
z#&)ko-fs+ZJ3E%!IPWQ7I=d{I%^DkhD-8BLw3gAP52rV9J)q-~lRHZy!up@NIH~f1
zB5nPO$_Wz$E-QW#ui(500LzyKm};JP0dO2*+;0Fok&bs%us8{;t%>YUClnKI^dGA(
znj~RR%uwh!-M)2YHuTNvpVlOV2WKFA;?%%Y?}x9lb`xwc<)9cnGz_6kAULe;8Q)$a
zxZQgr=@u6Jd)l%p_aQ$hlcw-UXWRUKku`uT{+?0E@6r~z2(I?sYG<{d%;oy1e973C
zvR?1LMms=KD_?6BavpwC-i46%q9Gxvby-84XLCC{Rrj`@VZETjjT?9Q2npfcxLNSP
z`FH1aK)tkUJ6(`qD{?M7<r;RMdk8t(Z)1brop(fMUTM{BN1(cG7hBpj`Q#PU+9I?|
zE)XKCOWy|@u|A2VqH#*oQ)QNdjL!_KA=Ew{h~7~ti7V(Z&0*}QIA|Dpk$Oa->+}4(
z#kO{m`m&NnTT4>&61Hiz{hg9H#Rl}bk8*Q1<oQ%R)Mm~6rWyx(d#7#8j6rD_jI=YO
zSRB^x&7C@Zu@&%HzVG#JU&`x|^{hQ|oq-;_j{b8!pSKRmxflsmANY)@sdu!?eDay!
z%jWf;&OSADzt>l~Ix}&{5NdC?r(bn<hAD882^}g+9t+I_j)vu{12Z7-pwe(hNHg$x
z;{25a0A~76Rr^VI!AXvBbiAE6WPIE+QMENP-VErz$U-}=d@>#}EmfqV>i~AG7LV&Y
zqp>}?M7giOGb?W3T;p6IJ<kMLThG(DRNwpU)RLXc!ji)8$hLW?>+`dS2JYKL1O4*<
z9h^Dn4t~Rq?aFl~B$eEA3m5mhNq$We@XS-}*xpFvEtr2IJyW_pWbx>rgo`dW1C;<c
zLy@Cn+*cOJhuBCu)vZV0*@}ko?});qqTJWaW6sx1R&ti&Z6V#Z(LXd*2q!pEu<;=_
z78NBl)OgrQN1H!wpd|=lqB%{Jr#fbV{tTZZHxK8gDpvpu@5|;)I(QVX4D^1;OwHZc
z+56t!M64SbR{WZWCPe;_jWZ6g#r^?$uE_Dqx#Q~}1y=-i>og$u(`nfC4XV88(|LY2
z5svG6ie$Qk5?D<K$2|=V*vb+TZDHat*W9<_TY7%rg&A{@4=R~PvH0F_qPtVJ^4fMk
z50Dvao`SguBSb<~3_}}>-jb+tb=60l^+9=^<4}Km47n)RC%2Dhae3mx5Woh1gR*-r
z*3`Y9L+MX+&3LUsn(I`=jf63j9y9?iG_kIX25$0({gax;_pd2<Xa17#323bYXTN-W
zYFs!@ea<_*nt$M)bYMZo&Sr0Bv48S=p1Zr>M>IBy#o@Q5yX(xzgIHa8NC7A)oS(P0
z3}3yY8+wJGat3)Km!&{UPtQ`uBKmrR=J5d+Ed#Tte9gTKFR#W+aRD0%;h)2GOilg2
z1{PX<uk$orpsi_^kdUBbWW=Z+9fgiUeRZ$ZcDI18)%xtj3EDw#C8XcK`rYk5ZZ$T5
zpZ7)K1b3e{(hQKB<o+VL?zZt5uOh=OKx<(J*>OKaVjPrm=D<_t)veala6C?2>>zry
zU{QBX)m)Y}-fZ|sP}SbJRLGRXgYZYWh&DlIq*r{^m^Oby-G}fiH#dJoNJI{03j8hF
zgDa0E9j9;cvbrk-%RViw`e(_VZo6yf6I1|IHkf+&j4}9Am05DvQTNlna#sd#2c^6@
zq`tA8SC`Z*CsIiNQFkDTXGNw?YZ)9D5mec~l6{l(X`+I004oRQ9Gok0G90`CwZj|4
z^YLr5LY#%K{urY(ZP5ZxZ>-T&75mRkF*Y}}`n<Peh&FZK_XNx1fO}0V5p9Ct#pF3A
zcu$vS)?c1byf{#Y^sY{0HXNi}2`5xVe&0n%`J~XrKfuAOsL#Tgc6nWqp8RLWB0(G(
z^8X&djO#kXiul;x@{a%w;S~}RNFuUi$cw_crN{RV4Nn?6dzzx8=LMURc#61*Z-0Zx
zSFFALS*Bc1N6*{G6cpYV4n{wX=aJ;$Vs;d7)#-Wm@!PII#59sSYNhuPoQ|=i<ci<H
z<$)85kgn-%PbR9Vj?!@zCAtL%WJ{%PYP{&C5@24p__{e+H*#L$Ff}8qcu3JK<Ph@m
zk<q88RCrczWm2$uu@gS}vXgwolRY?;AHuBqT(X*#&9Eo>MDyUA+AXf69P!(DmxXyR
zJP#vCp{6Q+jJ)X9n>#zJ@4%J!D2{He*6t3Sujmli@0g!_=^eQq!!uu4{#othq)1#+
zdI*}28cB=$%1Os4WKh2NDjV#({!^=l$xM9_=QXjY=wj@Z^*#v)$^ODl^<w>QEoutB
zqFdU1U+&r&jXb@?yqxrNJI10kkN${n$y(M^e7EnP;Y8%Yj1I5uN%>Z+6b9yH1>wK+
zup44{j<VSYW2VG>8{z?Ej`}52%6`Kt6MUv0Uq-fdwuTF^6buFpG6qrp%=<RMJT$$A
zPobzBLB?li;2eaR>R)JdYvpJ8o>_NQKIXTZ((!nXCL*SKDQngCG9^<$g`G%pv@?k9
z=IkP3h76|l5GwI~wo;Q@L_`Vc9}FixOJ}otNcl`MDVbs7Z;!+S`s>x)78QpWqqRug
z&G;9}PZH(71_v8Y6#EeOXB*nFPse;m7+%D7F=yCnr(?x59$j`7?$H%25<d&<+L4VW
zDEQl^+~WCL!ug$sz4+bH^z$ArpXhLPnG-~^M(6HQPg=<X>{=z=eB$W=%2%>w9Ojk=
z@2BH-Lc^Du!@_L0!l1XL!m5TVP6m@Cag0;Px|C4nS!a|&c|7z%y&2mj=V)2*ois4Z
z_iy`A!B+dV2OEluk&%^)k#*VU@#F&$CPS?(6P0OBY^T!mdq=uMQk7rrBsB4GksLcd
zd*j}0i;VV(eIt`zt!&*(sg;E`G-AEVb!+(wLh3=J@@h(@P_bOy-v0W1|Hgs>2O`RF
zlT=PiClS+jO$8$p-C3o+jP)HHPU4Npx+l$_o6?YJj7y%De!TOur8F1sx3=s1yfO9?
z-ot{kO5@l;QvMW%2SPbKh|_bV!DYXgmPe*?FM?ej*4PY~te1W_EcK<xdL230^bv^l
zeiW?9dPF{re`L3maCGmU+R^fbRbDH(+~oW8{oT>grQ#tXJBym(n}3Hv1H+(>>1gXM
zNIy%1VN}-sp(sA=8e^}0Hip}%DOO12o%d!`rQM=Rtoz5lTpUvLxU7ank;Hzj@E;x-
z&IXp91G8i*F&v~y+{r5sYfV>jOXu`>UeotFPT#~cOl7<034>TSuKtV{oqSkkv?SeL
z7tzknTpo13GMSBk<z}X^X^#0Sg*G-Fx|2zLnas5Cwf&miQhS>8g-U{mg;0__<?;#l
zF5MeqL3=A{nU|0}ncBWg^*FaDBTh~O6a?ic-78v|?doN{bH&M(hKpNR^irqJ^v{w7
z&&AVH>ZYvtuyVc`3RpI<G;)}y=pW-O`GjZN-J_mlgbp?c0YWL}{0E_-tB+^KMUUXI
zzU;Ixt54NFt<InQN?IzFb?gcecW<jF%~MmkMcz#uh;p0H9p~k3jDaxRjJrb7(2Dx>
z{cB3f-OcVQ+lF*3TYj4Tm}bT?M8*xOGR@Z3?2DBq=x^hH7-je{u3#FUO~s+bZ*sU?
z*x{i@mjB5+>mXL@&an#7OY`n&ITzF8NNuvc_m*VY-l9j85SU!=_uALD5LvXVSi(T!
zTz;>u6NQ|SPpNySENVaeRCQ5%3L{H?c7c`=IixV2Nb#wiba-rLP@P}rUZ*q`NY2cO
z%i64(;U#v=e$vOx@jH|`EG{}-=q5>0*;<uo6NdZ7g?<h*X7en4=9+{DJv~SJ%vA&S
zO~=kwt0JpXLaI3vi|;PFl`@Dacb>jPW@|N)Ow;8NG9eL!jH@Z>H(HEj_BSS8IO0`m
zs8#DEpO|q5m=C3WKP?XQl?iJ4sumHF^#gE?PRmM)UvM+3l_ZHxI^n!S=DuLRzp%E^
zQl(^6rFrPPODA@+FFsmhVc9+8Yi9F$R`lg=4U5$~5ms&5X)9F?iQMSD$}?mOIq!}6
z!?07PxQzbBGbm$t@FTI~VtSzdyrUquk&BgljqA{JO&2FF2^(|+$;X<{-V%}@2azf|
z8>Cfo8LK4rR2f!fsmkX{hz&frUY(J;<n?}`Q)n{xgE?tph0nL=6aNDpn9A$_Zns>r
zEK|4_+xAK4uCYu$5_eQ;k*-tYvhewU0q^duS9W>{gGP66Cn5s;!Z<<IsucMunZnCl
z^mKm;_ToA|b+Yu+@``Vg@xF0fk0b6$k2pCMXBzz5Bt3)ldiBN&0q8trYvrhtdJGzX
zUQGOWV#x08XquDhGXb7CCG<BwH<m3goBAFK@w$8kX%BjhXQ{<xX8$!VnnkNy@-A~9
zdw6UFvuJ{*JEy@rbW<abB@uXH=-C*;<fv29dq7k=b!4kp$(-;2;|hiv*@X>bZynnH
z+j@3?d%JBA;W^ISCc&qJ)|DMuw%T;6hDUGi5~(iMR*x*kkTR-NX<-eS{p`MSbD^a*
z6e&EuLd!`nS2J4LQ!pvOkzhkJqEh^Cb7d6%W4JH!W>_XM@E__lGR43mnQs^H5l2^q
zML48C^N=EoK2Dj=wt{`t596r0`5LbkZ=a2eM`zm8GWE>G#nEaSvA#Nu3@OH_<<#>E
zd1bprQ=Pxv$DlPVB}`Vq@iDR-Wv;n9k~MJjdS45~tn%q8m7#&rKq8*7w^!y+s#!L-
z{%S;AWM)sd^uQLQy8esGRLK}~qcL%}s2}vbGoRX5m8=aue_rLw7#VXPoU5r@ifWGQ
z@6)yD3)K~6i~8Fq6or}(7?76f00kxy5~o?Cb(rVd8GpQjv#fwD*qi*a^K><D+{qc|
z$9DZ~#W+={7$G8ReEPoU$a1)Zm|#dl_U4BhVrQJuOqqmvVzFZ;y9csIL!X{pzjpO!
zk<HvvFi}yC;QwdUa=d-Ln%3a)yi{eS3E?VVc;fNryiZPVo6nCUCXS}l$$YP34^SqX
zXMXlC8A%zM{E3um?`KMfim37I5!m_rFqPHZGulbbEwNtd^u2n$pJZETDCW=9;LCq0
zM^TOf@bNR}Y5XbPw{IU=9^Vw$*uBm(v9`@?zShlWH>BLIch_6cai<&ZlRKAE(hgI$
zySOiuA?El+utKNc`|94dVCgf@H*V}Fykfl#=SWOy$N2b?@4s#SJ$T<=WEOuzN|wqT
z=cd*(GZR%|SM>`NAJ{v38}{}5^hdVG?@}vmdbsGJsr>4hi=F1aIY$&U44P}GfKGD9
zyGmgVR^OVnM{+VnO3YGqiHMvHza?b_{@WMagZHEGpiEX??yq9=(PiS@eL^I>+$AVa
z;y|Nw$;)J#nsE7-v(ELMZFys*QY9LS!L)wLkQB__rNvXxvy>+jOhw#oof16mPfgs<
zge4mqY$c-0ukUZClWT?-=#~kdo?Q`mv^7Wn@?RnwkIIi1ncHkzg$a$|Jl0}p_|)=%
z^t_2guU6-gm5Yx(csLx0)dq+2LeuT*+4U_0mD6LnLgFEqW#!K~8!{|~$X+K6u=Br+
z-ltmpvG@{*%M=x#qtzn{)&B&~k)Z|nQx<L2$dbRmj99K7o5Y<`F8i^sxb<~*w7{gS
zx3@2R-Ikt{UL3891CjFQO%}qELl`^i>Ufd@I+YVyZ;mb*A%>t2iOCc*A9|;h$_De=
z^2%QcB9%S;G%ddS0ZX^?obR^J7&|qyD}VCRm$}GhPcFgy#HI4ib3x`W=Aq4=L2QV6
zMExJ@>|Ye~NlYx=&gRoj7ee%+%*nCrVvb3HN9OEwsMPuG*ecUrHAcc(V!is4ii(Q1
zmX-ou5ig86WL|rN75jUG?dcy)iE%9%85#WGhsu%X9zpssAyfl6*K=$GUo=Mw!qW^?
zS|y9`fAF;$3~uJ2XRf>09gz{O!9-{=c~6WU?rJ!5e(i-^Vv&p&eK9X)mh2fu^?7$|
zYi1J_VO~PQ<F#!Kelib-Bp3iNPot?P2UzJl5NH8v_kOg2Jv=@>-@^kna&ki7FZ>cs
z<_#H`UaP?Qm_<rXUbVL;amEUXeT#IHmxosiM^rxaM&RQgURtsD3#}J7FB{udLG#5R
zk3AZFUQzE`P!?XibSYK*W$Weji1>KsQ%>T*A(NKw?ufd&_>&Vy7tIKsuFW$cW@bd%
z$;oiC9plQCb}c`xXF%?AsJKzXgj%mZe~hmOq>Jn8H4)vyGi?ngW+9jj2BF#*`Rx?M
zoMBXELiR}^swwe7@n}n8FZ<GKJK?nk=QH7L+u<9Y%j>gn#CdXqd&A1mEw-m%vwC{j
zw|3HayQuY52xh#&UGMqL`t6HUNGl8dnbyV0Z|mr_A}9SdKd+p<gmFXT!jj-w+go}|
z#fYv)H=g<JuZ3pg+K^DEk;}r?V6vDHznz$<_~l<J2bA4y46TIg8hFsGo!4Fl-XJ(y
zTlD?3x|&HgguQSIUlAYsU%9or3vcx0XLHY?TFhJ@qUAssj^71oe<_yMUKz<2kqxd~
zK3dM={kK^nYU!Wr)sOKP`_AMh1m7eC>di0A9bz%R`BsAcp^sf>Moz4kWdAJx-K9&}
z41c>sqfkq?&{sscpVusnlm_7T1!nDExuzB!qGo;!HsO4HD>U!Sv9|DfjI(jan~C8v
zX#_j<U(M-JmMzfEd(rOkr-{7$v(1<X)jL_WDn(M41dkLv&wjq79BayS2o>?g#+I#Z
zPxT~<JTjPiQghzNJR0jR$Jy|c;hawlXOd-nN&ep_^&m{F$5Qyp^5NNbocmJ>*CLH>
zR~Aw2b-slBeKOjjQ+%>flcf|sq`yXTnfT@ZSAUfr7nx<SO}!JZd15XNo^7g?I%N~&
zsOM|nB<&9lrNO^@E5SQBcxj+Cz<YNsGtY`q|E~8yFxss{p{BiPH!cp)wwMoniksj_
zF?7gI{&yq^H<Rj@lTR=;E$rT_HLTh*2nlHJ6|fv)V5*T_(Ic?d*QXO=mJY#F$d5a@
zc=4gsWd{5>K}NCbH*w0iujTx<=M9e!z9;tR-Mim}#Ku<lRWezGv=1XAulx9ibvCIr
z)^<2;rD!lkP|-c)9sR{#p!Ut@&KAzi&+!d9c6ca{pkGj!v1+@J7V)X=!UyzU+(_Hh
z5^8b@=qY93{Q3Z^493eo1#|N}*fM|S=MC*hRWe||0+&)qN<zXvBc`sHbM0H(+tgxj
z-@YZFc^eM>S^?%uC3bdp1{~G|H1ERU_V9*3mfT={2CQBsR#)>wlUFf0OUr^rqgN&4
z;~K!J-ZB+mHdNcOXp6Kh+KS8Nx>!^I({45;D#PFeN8!Tl+qXx*>VB|a8HRBy-0ov{
zO2&i_iA<vtD_O!I%8}hsnXdR&7MZq4_v^0rw<|)^9G9pzqb@YUx7_ADMs%mrGC2-V
zd6)2B^f1=@BHaBLH9Z^WG)IJ{jt*gW*)z3U8SeS@9Q8a0x`r4`*bM=@$ZblIg;tvQ
zV(!}>qpuD&X2iwBg5%@k2lKTp>_|~1^*O3VJ6(M1FGAwn_EfRS`G?_p*J}+js0i%N
z3*-?|{TJ>V>3>NxQX{*FGWPuaTQ8R1378C$gt0)SsK;FO7yu5h-R1l`@sRtJlmq2f
zx_);qWMBh{RqDHU7m2(RfoxWt?8;+C%bf*TG9K&aA^j<0cZ8omSAlDpL4^O6A(WOc
z9r6F(^9FLJ1Qs<rN3BFWzh+OZ!c5C`Q@F%@qS71pLWWr&xj<H<AEC)?Qvk2?P6jdi
zn*>3(qlKmHe5IE!{RcKt{U{)u4W<;0QywF+sFsEIWzRiF8DsUamsrh;xNUcGjn=r7
z!N~FN+)2*UtTF~BZr^}_65jhLiYIRgJ-U%clJZOu%Q0rC1M)uh-4Vn1MNb?-T6-6u
zq{GF<4g2&d;c$DteJVae&V!kIXxZyNkQHamQNcan&cNBF06b=m=F(D9N^nK(!7r={
zxBUrXe#ZjgYZnf1C&N)2PW$6FL3vecyxF>_3}H04H;>I+3s6Pl2<&`9nCecfF)%PP
z3kqaD;7|;H_rl5zyqe&B`SRt3cILW=Hs?Tb%wg0^1vAW~S=s;U5lZRkU^XZ`EG%Vu
z`jx=h!L0FEY3|ACsol|z5u--A%(ZLRocAXj>_+q>GJ(chsTo72ne_@YD&DYYQ{n1#
zT_A-}9t_|`$fWzRStHWi+zxz{EH^-|(cDi*vlQno3!12qlyC-MqtAmi8CkVAX+UML
z`VlgJrjm?&BuvI<ukiS>$BhP=8s}X!Av|!iONA7}gfr+~+_<g40ZZNW$YPRT%y6z(
z?+3QW)|a$Qm_-TPDa?zmAC02gT3dsuO5XG&^>ZQu&O4u|^!4=zRG3h;eixY`*tU4t
z*T40romvcN;z}BgQ2qA6?I$NC6_}Ki)SsiumOYEgxaX~tp<ZSlk^)5Q;iC^&S!F}|
zYu%2C+cD}LHYc8<1T7YOQ;$GU!eU}_j&_%`HLL82xh)e$f!Vu(B`5Go0;ay6UPc0v
z@HnKuK<C-&WQ{sd5B32)cgCkrx9pbs#eip;(Wdw&YHF*IQ2RxQdF`Qr78^>?;MWfy
z@s}_AVJqaN$2p4wmpiBoc^F>p(GtTctc~$VVX9#kCMrY9yA^j6<=VeLh2$%1Ii+@-
zlJVO9)IL8_p6k2h+H!V!yo=Od!l4BM^|DFl>@>+}CF3fz9N&5HGGoLNWgH#@UA>)`
zq3b*lk-yN<(IKF@DAOku$DcV;WVj5px(xO%FJ62!uUh4VEv@42?oN<;X<Fj$H>`R?
z*tW~H$7}Ls<`Y2q?FRA#A7(aQF58zT2@Lh8DQFTt1UYQi&I!Dp;Qf1kULvHouaCoQ
zls(`s>(jDvYrm(@o}s;&_1a^pzI5rKlf!K^fjuxOHvam7$|*FT;1!dLVA7;ryC_J{
z!=qYvc4Xo+R#a4k7EWL)1){$0cX}vSYz!@wVD_8IoFcFH?=z6|GxP+l)$D8KYE~gi
zMn*;ti!UF1Op@h<KHeY>KY`eLafA>!NKNu8M1VrndA(*It@UbvB27sc<JDJSeH1ek
z*}1eKAQ8bd0Nmi{-HqnyW>hQwR99DL4Sn)~-SKog{=9#*#Dqh)>Bd05_S8~2YKfab
z^^%7iyHW2CdDLWuO*(>>uds~`%kI4bod$6;wpaBivEH|ruKc#;HK8K(7^SAB7S7Df
z1pDwX;qpNO7zD_yXoj%cEh&^3^@&`#Z~?rUd-v{Do*gaQHIq5eGBZmT8FrU)GorL1
z>O{546n~Th>(nR)_bI9qAtEADHI4=m>$dNCXNp2x+j~<bsL{fv?=D$?g^b%01QN8X
zu&{9WXqv@TZAe$n@wZ-4R2B(LBW(i+Ih9Yo!UaJp78aQR60XlwvAlL!{hT~JJdLB5
zHW;DpF%}_H7!@`#d&@6&RF2FJHhBX}o2h(p_p1HSiwCIM0N7+<_Zb-_NjObEAyn`b
zP4GrzxgOUjVT_n5^%CQ7s!U+8+}bzfxl})N0uoLC@Sy_eR(ui?6EG$Ov}*B`JUkU0
zIPaD7wHU>W2Xj;_yk4O)uH!jgI7spLztI^dpnkl+uI%n!JHH201`Dl(BY2QNwOH8B
zB%dr2kmGx;{vyzp%Oh{zWGLlp<$^|{<>FHAjqKrNP|Eq)5y?iKIegI+<=Si*@Ag29
zn8!K=W>AG1wn}F-R|J*7(L(ZQl|w!l>=d?jn%cuTlFld&3EXac6eJNI5#2yh*e><c
z1yBgpj&qug7Wde63p6ViT%fQ$*w6tbmO5;E!;pl}9(ianjG_n&4o(D`QLzwE`M)i>
z!F8#pPnU}AJ>8$m_xJaIG`6!mBrTuzQs*+Whp{NyZU8%HABa|oP6OuZSQ+@<PR7uW
z1GG#2g-e(Vy%NlS8t+awSlifun!67c!GYIy;nV&5_xnpsl(8uUHU64+auk;6L$K*=
zhFuC%wQjpDltC(FS3DYHq|$-vG8LaeiI$EIh<kZ_eSBm;+$6JIEg8xJl2j$p%hQzL
zr2T7BJtAJ;)QY&!d4V2n#RVg9Ixk=$83~?OaN5kv+Re*MfKyVbdDBU~)O7PiP)<%R
zClH-}>c4uu8nSpAeD*6pVRF)e(^ObQq;1BMmXlKnwCcbm9R38@X=sT=y9Fgp;K8K#
z!I$cc;3GCR0pV6)KULP@T`w~kV(Km#0V}{W7JquY(b!ChrWJGzts4N3^6_d|WF*=~
z8@Gl~Q}3mwrb5X}61Y8q@34kiEQ-$|Pb7$3vyXN8ebU9R?SBfn#a87M3K&N*;WK1#
zLFiF3PoF$_Tj{W#7t80MM)udumbv+N^zVVVbx3|3`|8!7U@ddN3CfU77TMZI_9On9
znI5=rvDp|W+AsnoB|$p+Ls3(ZcmuAXjZu?)r_wuj?hGV8`|$c0yFmto5p>=FI6V`V
z4cHuHe_3b$0hNJ5)yb=<WL&y(^F>Id-Lg_rxjGT73kYG5ZJ#6;9ZcEH56{qYR<l2u
z9YBI`eRf&-%&5+fWuBUn@*>234US{Vr!9m?Fv)_#cHUdb<eGAo2jw>35dy-8c5>Rn
z=$0m`%94_k!T;>fQe=>f;g)4k&fBSU(#}3QN=r#u@S%y5$yUk*5rvX4jllWIHZNEz
zr$4Nc^6*ArQccDqBg=K(T|$h83v|R^yx8dAa_!_8Y5^&L-$H$#l;?icL>_o;PUa@S
zYD=Tt(|t#p->U0IxBxn{kx;688B8#3lfm)&d>y#Sii7Lml5iOJ)BgGW8?VZ3C{Gi_
zK4Q{oo(!9eR{@+2;CiJF6mt2-2n<@0Y=%qV>Cg`%2C%u+>fC|MWr;3VwJ1&H?S?71
zIS?(oXCCjZj3f)V?7e}XOFhXSYQb)JefXgKJIq8uQA-<qH`p`SM2}KbqRqxijX->z
zXS@iDjrtyxz`r~A4-XG37t^AP!GU;P#ti!hi{POQ*r8&BPGX)|Dfq%i2nYyFl*3X#
zj;(%%jDh`F1wm@|3J-emCp~VX<^C?v_;$teNra%a59}4C%5Jc!CJR3kseOa$%3giD
zsA)H#NPXR@ZhvpgB7$};ZWRp8Qg5oCoV+|kSXm1=6$5q1xtdT&Syuk~@p_$h5{q*w
z*x23msXFAjY$-T8#DcDlKEKvRpqiTmGDgqLoIm~$Uvb0+J&<aQzq0+ks9SJ(#2ic_
zxyuTx*<|%$491V%l$4Y=;An)Tz-^X64GqmSeE4I_3&c9%(b3sIemuv%%kns+*nSnQ
z4)ezawZUzMc`*W23?dPRphKJOZP4uxy2Eg#(&Zh-twbZ;ZorbAA9f?Q3qSAunVWMK
z`DkEZU<^xD`OBAqDwgg!lX~gOyY8wqh2~wityUU*a`HS-0o}WHys@K|cG<sv{h|^R
z4I&>Xn{roQ+C<A7+Iq#RrmU|_$-^dfKU$>OUF;J@+CMX*^`itS(j!dR$}V|&vfUjw
z=6-f)7@4NJG7lbAPU*k1;0+<GjPD0>=aw*qUo@O>z+cF{`f%n0MHC$!-TaE-_xIP*
zsMzk`|IY8Ur5Cvh3ZLiA&wu1YMqnS}9>ebS`!if>rk;5UtmR+8HQg!ZbKF3t594>%
zw|15W+QalF9cKelOGoOEx_Ni^lg;Mq6Wxn_=>l%!$oVOutE-)ccGZM=T54+eXg8<}
z{#9^ICgsNPG|RT~{M!FKnw^~$SQW2o0!6v8GWAtgG;&^vY#zOWj#wg~ggxYH!1kkE
z!h|QDz7ug#QCT>M;K5@snDS_&NkNeht0{ME$s2n#?TJGr#tbfh<o2YvVVbw&(PBDv
zY$R7{wO8+2`E1}LLTD$)s`eY4OQxdutzlEn2^eHZ3bn*%g^ZvjCC#CIn7V#pLQRFG
z+Am-30;9D=NU`1WAnq~vT>{D0usFGaWPHfyNW-~~NPUBhoVFOWwx5<n+Y66Jn)W05
zYL&K`puo`t*fl3x?Zlj>U!FdF%5R3YIpyW$a()Tm@WdTztyj#GgU8k~W9jG)AvONp
zyR=mkOrSo%RCIrB4x(UY8#>sW?T)N!fvBe4)L~_qwdQ!O3Zxz>kK(dm`p_`b`nqh4
z!+u2tLID#-T3XtD!CV=5h<)MHsb$zp^JmEV#!N%pDp=p}F=-8rn0KC@?Pl__;XG({
zR3CkyP6j{w-4hcNmxl7=VK<};fdoU;26313>F&_(-=Nm}22qkOu!Vt`*o==fvLSyB
zmDA+Ds@9X;Uw+I?`zo;CK_SzFL3P<*V=U<en>SC?<17fqcE$eKsVx86fSp@ILqod&
zCoS!Zkl)r7;wyYrmg5jOTOYIDcm`&E2$*Wy%#?5c&)`_zff><YX<)$v5@Ga~K+cb+
zsAK8E+z{A2jiZXnn;k6v;w4V!r~Awq<?8CHPHye5qYMHjBJfYq*1H7l8lz<Bz~A36
zJ>O`+Y20i4@x!sS?s$zaGR<jH2!ad!$@y=dSK2#&+gFtsb^|pkgeej$3{mP_Fih}_
z<M!64@;f?SR%kFmmO<JZ0<HKVgrh**PIId5V_@K(Y5ej&mDm`#CZ9>+o*wxL=wO^a
zwV>c<QEX_ays38R14;fd3gpBpd}1CQ(ypC>o^>kqn^R-7J~^<`HvnE5kf!7@i#kp4
zvgE-`nsTat4UQjRwR^0XmA9-;6hO?eW!fFfS2kXXo?gV^YgrL~CkMKalfeDgR6~jO
z%&FTR42a(<*7fVXEy0x3Vh~#7f!HZh?}G#Qa&^3-sNW6pHBszFA@fsp+NR4ZPR4$7
zI*gs5J$KMy|07_%Mc)rC4}-bURErHbLjACz<c#^i{y;=jHD24L<$<g&^du`1qDj^>
zp8b`4T<4vIL<qu!!G6ja8)re{hu<bUDlHSH4jmOe^S;poUZ+4a1V2g7wV7*nLJau?
z1+_wHq=A~gDl9&Jy<*~z^&%MF(Q>QgG^to;Ew2HoSl-3iAfY{flwE}e?B)3WY-MKZ
z2M-Vkohz5JaIxGt-u$dDaywovGa1NIZ0qdYjT!T~L&I(|_z;xy+&cnB1_>~5`ekGA
z)vh_5la#f5q|}7PTs=Lq1{-9qD|Y!1&AfyTP(3lRdkgj3+g54l>D!6~AOuNGPgj<H
z1y5P;Cp!&}5w+=X!4n*z6QtnR?CbzAtMicF?B2oPjFz04nOT~wsUl!h`^_^4d!|1%
zLQ5?vAJ~=`LEUYi90zbs$^bVvXi=&L7D#_VN&)}&?ZmZ-s^zcE<b7a&+B>;G1>@u4
zEtHO#qlF+n2vm(R#FHUZLPA2P%i8DkRyh;q`~IYe0B{Y!5b#M!S9r?M%*avfV(5kC
zcd`~{(q@I1z$`Pb_|Sbhw~W7W+U685GjLg56&$R7j|-Qm#K2b|W;5`g--oy~eK1$u
z<Z$bcZ%~kO5V-)A7^iw`L@LBwOFt7msf@s_^!4=((YYIEvhm}Nw8F9zn$`sDJUW?>
zQQS=K@{-M<gVyGJwQMR6;u=sAN#bi3TfbYKX9Kv}#wmbV9t<i3zQz5gkc5g#w^WIO
ze4*Jyr6LH7erl0cM$+qz3imFhfuXX(8x22yHTp@s46-T(Hwu0CxL}yvnbcA~m-mjr
zNhZ_Q6<Npw7Fd+&#*^He@Y6URqG7YCT1^rzb96vjqT75E9W|s&$A6wMSN0=d><6=g
zj^IHGWx*VGkIXA7Yo-<#vopf6h4Zuhx)Shh(3gNqX8beZEjrL}KRwEXV0vl1!lnWp
z-d~|`m4R#f{E66hki+@-W!M7n%Oa7-zhz`lsEgcU&z~rOd8k^elz>!%KKZ&Ocd>pu
z0XPuSFhM^-w<>~<8z<mm{Nv;8C~m8y?+pzl4(pRxwf0JxZOiKT@7}#D)@#K@j|oj;
z|Cw1k4|0FI5F6<>`S-x>AnQ7>dW)Hf3CZhsM{l?qmwn^WVk5N00cTI1THX+*17+u>
zLx3F(FEScAEhbk~f2{sYd}iF6;?w9Oqsr&7rT{yFS`5P2{@o#M3#<F^&iZ2(VYD4H
z*3i*OeDLzi^r1s{oIoxX`u`kR(Fz8YVU2b=gw8f`PPdSDW^|ItC<+XPs;a75g>@<f
zU<oi(pt-`K$jiu}^}B<<cG8E(5rPz%KZS*bI}qtEZ#Gjfm&HQnG)+2Q>scne#iQ~A
z+G9i&JcpcPso7XsQviwj-za25%^$t#9*09D{e{zXIH6nUTmv1x2C|Xy*+M=Jlz9e(
z@Rg9b;LvNig@Fl~a(wtGD0CY|r48JJmw*2d2L9iq=-&{I^lwU*VE0_l1l_A_=kLUc
N2}?cCd;03#{{#O*JR|@B

delta 107485
zcmZU4byQW)7cYp2iiEVJhysFuNH>y7mmo+;cV4;<ih#7VNP~2DcQ;6PcU-#T4e0mx
z)_U(RTrRlx%$%9MKefABP~uxq@<zBIiAf&v=kI<XJlM`@PT)^}YLjpWm)b*4MGRR?
z+;{MYK>Dm1omAeBr}YAPKQQf%b+*sR>#j{7{E!H>z8+pz@o+lgJ)e%u^X!g<ZF_T+
zFO&FnZ|#!44$}*iq(74sKAqv=P~76dr8*dQV9h!CDlcjm91>>PYeO|sVNJ#!4ryx7
zR8AaR<(BWOzw<_SZEN$Y+;`1bKQ6@-xKDodp0Eo1_=Y*p*IQ}Ip!LiQ)76y<jgv3e
zE_P?13vtJAqOwYCpfkqwaYOG`{M72HD#rzrYbI49Z5K*iLjtc9TE%ChB>_!M(kBlR
zgfyn5r9|!^Zlqew<>nHc)XRo8KzIINo5`hW!d_)>W$uMvk@GQlCGS%fn0LOx5q_J*
zlA>66^^uc>-`M1>lnU{Kgw*e77&=o0W1A|+Jh`qrNNQ(1%0tOYoPqave|{RrvfH9|
zP<~19YrPb!GVf_bnz&OxA>lyVr=uBMu#OCD!<~m<H3HwSs{*P;3hegfAo{IK^j@jn
zM7gjF%uKE~A!e5QgPf22B@BkrA!G$*E2$it^pP@L@_PNzN(;=#_*Zvuj96U!6a#x(
zrjyVvj5qG6Y4Rqd4#V_1MSr(?<>~6lt5*3C4v{zoq<yWVR(CXd7AHyAG~`Rg_@*@J
zH2w+m`q7(vB6`OG{<UvskDxJCzI-3^6h2|0Roz&}rH~GWiid{G_ln?*(U%jh<g0bh
zR32Tu%jH&bi3pC1dV5aV$jDlLtc(hlk+>du6_R=Lp1xtSNcf^(8uS&Bv_D>Xz<VON
zhE4sDYNX0N=V>5$Z<QKd2@=7joGb;ye1n;S5erGIvQTZQowi0Qw8hh{hEOabUVYU2
zdNTxOUqtvNC(o_T7t+?!Yp<i4$=KZU{qXTeZ&kXq#o_#W;`I{k6RH)%@#8J6s7je4
zx$+HE+O_>*)}j`?EVVbjwQQHF?H`n?gL+-uM6~OgYnOxY?5YQMmD_{vnqA|Z%$$qe
zJnoPev%Sa7lS55VtXT4V^Kp$}5@Q{-#U+HiMdy&0XrZ?u(a>YkFg+I2+FGk&=@Bka
z-IBdc`g+)KQriB|v2e6yyVlUZYmGybr#K1&kw2;bvzamv`p|gUyf}R{=Xd`owm`}0
z!1Q8c?|Gy1M=8CZCS;5sedIird?wggL$;OF)Le}y*a}sNyGn7M<D<Qx;;CuMcNk0G
zyuPxu`m?Zhe$qK;JehHa_U7b#<NaMv3l7eG^1FR)sT2&-YJOmss!{J@6+z@aIXhn%
z!Cw+9?{cr$I(r7?zZ_UGZ|)MP<&o-~C}nDir+#=i&Li2F!w^QJm@GGNnLNXoZ%s)-
z7bysFL7DM1;`A&p1G;Ow1+)AA7&nL%i7Y4k*qIHw;|cPn<mI`|Kqq-$GqP%0s&+Z(
zI1eaZoV}UgGZw#)y!L8&E~@&ub!F@QeYE%lI~4WUf;=pmn<o}BeDAJwv(*CMCrF8Q
zX|PkS{T{NHJ7Rs}HSGB?ldO|0bZP!05&A*w5L8D}IQBZ5cO>wdW(|240pTgad!e@q
zKK|9HaaY!E6=<Inl(()m)YZ}JR|1mcj15@kVbxX!lCKPTb?zY;NA%G2?j2ZJwB|1A
z==SAJbb|ZZM=Ltu3fIzF`!@4M@_CDJILu=x3Y>>Qhp5$2<L6JjH8)SHcu*ihNb=+~
z24`7!hD7P(5Anrho)dbq-f-pa_x|MqodJQn{a@7#(EKxZ&~G;1lFzE`*WdGgEnWLK
zR|eD#*T`#hrJlIi(F`So*RS6+_6`1uKCrh&cHr+HM(4Q~pgs8smS;mfW{Vi6ITOA_
z?Smd6Qk|EVn`sw&wiP!AEv~<B5}E!T{}clWsnIH{!1}_a6mfQJO}QFIK=4a0wHsTN
zH}5AXn>(Z5l{z!4M8D>HKapR~T7XOP73#H9eT!9*UQIs~9_iY8oBWK50ucupfk~Vw
z)_zXVlY5`ls5pg293Q-oCZu=&y${H6G>b=rp0CBj4GB6SnQS_BLwepK@*|Ac+arub
zs(8IWV;`a7z$JM6@Rx#3`7t}Z76So4S6#P5+rY@ZTGV*n{0z30$fr(M5i07CoI?GO
zjXqAq*yM{#46AO(()v(TCkh^HVQ<YUW4k|5E_02~=nmM4h0z+DyC-HVj(soXu(V}x
zkhxHl)*)t-w-aW9tACE$jS82(-)!!-^uMAnfFqbJH#d;O#=GT+BqJlkZ)=OUyo_gu
z!H}5~JfDR4x-eaBSv7^#%wiz!dsIYbR((`Vq=>p1t74D8Y<fpYXLPzh5kmWe{w3LQ
ziq6r&=xITX_~Sl&1Q~12_%y^<H)j(P_5D`~zEP9=FUy{us2<>+?->Zqv?s`0>-How
z2*12OFAtH&32Mm!Pq(}3sKA?WTf_(1N8InQ=R*`LNJrRYFyF2N9MD93Yp!O}`C`tv
zzhpW!X3!TlVQDp<?zaZsKRAyn3Ke>Y`cNjFUpu~eI-Qlo$$u6~SK_Vj8!eL9UtC(s
zvu%GRMLhaW;DGMo!#%J={$KD=X+NnH!CU2Xdm7Tiubg^$*G)Fn5ZWXHs9`x=9Fyyg
zVH^=P65ZK-goGiHSjt6wCQk^f?Y3Kwj+n5(W`46DZ08%eeO`rvogYv2q>0-?ceh3l
zHdJ#^VxG}KFCG^U`~CiAK#xMY6iBAGvrNr^*X>v{4ux?uQ28=UKk1-#->?9e>p4}E
z%*P*r!%cUJzc3=7Iy4DQe%{Y;@LSXD&P7dG<ry*eO=V1}#EnwyAwL^t+1rQ75)mY@
z<p0il8W-C15q;~-|3M>9B#-4&4&^6H_NjpZdZvyD5GILpo`l0*f&BaXJpN+IDQ^dC
zmt-Pc6=-m6h>D-`arrlW((ee}=wV8qNYix)rM$U^vP(M1xi?2e=mn?cX9`rk04g>r
zR2UaVm1C{2X<YqBPRB$!<38Nw5Ms+OpY%&DPG_~yO*Gmg7a+#1C^}}7;hK!jT2YBb
zv=qqCSX@Nq4NcguGm;j$+)E~QmQA1GHewu)XurFOUyrN~+pH4ZG4hs5b2*%}Ff_8c
zW#@;g92__#padqiufD&Wdi*N}@;1Z=jm9ywVJI@`)NS3*kIrNTq;i&9yl2>9&TI^5
zS>2OiRB0blEroeKjaKr5a`UyWoSIZQg!CXecD<6bt!5{t&vZUH2&pPIjLgjRD>UM%
zYNZLP)w8XK3z>QQ2H8ph4Hv`$(%C5Sk|Xs?%CWbr4-3Nm3y;DUo4)uvA6`*g!3qo*
zbtox#YXu5j;hIa6_rp|>+*Fd9dy;VJyr7OW2Em=pgI6!T8J!P|qy@fqLM2xDRmqgC
zjBD@7-uwy9w>9m^lkVx?N}jRUCzTnTM3zZ^U1AZhey+dD=fLMewbj(t6`r@T5_u4c
z!8z4<Y%{cGU3G4aYIZlSzOTt=+T2XHZqa=Z{u)mkwY5@Xgzfd4B*lT&)w2dMPss5x
z5gXYXCOLb3iB6;1hZR*%2qD6Octb90kJIQWPWfD7<nqY7X6HC=LA{c+ps4>bZi5jJ
zMq)ajPyv{>9!6ns{j5MP(O|&lHG0qXl4`>3HVpl^_wO{6;aYhkGqoc1U54=)a+u%|
zA|;`;5d2aZI6TEIA~<7oC?9j)-G2Q?9-B6463YQ{c7pb-3M>vbyJihr^V5em6Bx~=
zk)FJvK)|NO2Wa!P`FfgH%d4Lp;ca=e1fSU7=y>vN(|Ph*V<;NqGu)Hp@`NRkF0L%U
z)>WOrno*M6?a_za#F~q>Y0Kx1oz}kMdQjAIiz<pe*IQORaJg7|n4ZrJt}g-SvKr_R
z#232{poe<}Vz^{>3wQNPy7;q)Umy81(4>jo_RS6|H1*qmh~@27whPnRdAUkd+e$}p
z%Ooc*{Rm8-ux0kPd~9-(?8uyAo1-SAYE7LoT)cYOLRmx^{vx2*B25&|RvIryWU<$W
z_e{_o&c#Z(cTe<6-5uPsW1^@$YnoFi0j5F$bCeu)i_GTo-VHKQKm6dWZTz(@Z#EIB
z-7Se$eRjNGLkvghXM{g|Z}(g5G(sAGG9@iV@f}*a@_q&vA?-M+=bkZoa66>7_kj$u
zstSF)M|1?=0Pw?~jtTTiwL?~-xxKUw{~(NPeU3VJ``7}cX{Lb!#eH#f9#TtAoC=xI
z`fSjUW8?c?oL*Tz^I7)ksWyU2;7OH1)HgY+w21(bDRmK998jFDY8LRIvu8euees?9
z?2fkdhuk8|wna}^P}~kX`50_2l1#ROl@XY?84-p1Ev=V)iBB}|zLb629H4|lQhKMf
zz|_{*?Bp^aU6RfSVk48#b9_@d?4l8wJQFe8_dQ+1JkhKXuSxY#cY~_Ce7^Fo+FSMf
zg~DJ8j@|RSN6sU0lUX{RfpHBt;B+zL7Ddv86BqogO!ZQmGYI@4(;=i4bd8rkRVqPE
zA;=+BZb?t#T4Px(x5~=1e%6>192D;7os50R1^EqeNf%zp-{(D`V56Ub%;z%kFML{f
zS%u*a>m@G1@<b`YkvLMTL;UN^CJA;qqQcHybUA8!tqnEwB(2WUF|0{DYMkMDOi>TG
z3ZD?MnMZSTX=`oy(DOz3HOV|<*aEl`otm>L=bSdPT{4jBV$v3Ln6@<h6s27?-24*A
zX5M#6^mNx(X?N-prB$g~tgfQHjCIX7If+UITfAUE>-{4hVb$+s)DJ0B@T!<~HN&nZ
zZ(*GR&Ob7vkhgzzj+i#(Y!SIPXvX_3Q#9PN2hs>*%{Veqzk{=t>a=3P?(%VTq<Osf
z<xAO!GOPMPk!g}EOP<X$2WW5KD&P8`siQn-dOuCU2teg$Cw<bJV@2PTcu(HKRRevv
z;2fpP?r)PDk}VlTTpeOM&+#n}SFM66=zkDf&XF<Fgb}9RQ`_C<^BCt9woD1{`8nZ$
ziRJ8hFevmYYcpfsc3aa#tA(o&((hYk8_SzVy>oelX{$HDvjDClv^sqKM1-3QSDeU;
zCQadHWr_3ukZa=C=GIAp?0&5^nOX^lEVU|xcYbILfdvP6unj_l;T*}1=+5zG-_SJi
zZ|Z1Osr5uy&Lxw}`4yL&TF(SKqBmrJV8T!VKE3AQLK2j(LXl{8s#B2D&iR3bD`acL
zY=lH1kqxO?rcIzz>_aRTA84{vtJ>Z@qF`iA9A~zN8>~)!S=%oS3JI$R?Crf@;Bzh}
zG~-WH(Z)`4rEg&hMq^=8H_yuw^$)?&{79tM`Awy*qlZHCA=2u}uQ7xlvA+~Zlh^2Y
zvJG4X<u~4}RKTRYErX6##eD*4N*(+pAm8QocjEN09MQKTj{Po#Q!8W29BeeZ8LO4O
z^wnp)T-x$5=kddw^*1+nm^|a#|GrBsx#7=vEPymPYMPNH%x&)(I90ndd}@tc^{=A+
zzTJLR06pokDy;0%iDT`!CyAO9tFZ@@Y#tOuoQkj95?N|_mXN6rD4_;LEucg7jvQ!H
z2$or_Keg=F{GCdvy|VfhmVH*V+J_GlpO9P0d=x%>#LrPJ@?`hAO<B%Is<l}BN)`3t
z)0eyt)ie=VkLc~0cM=Rc%?J$8I#r`?p7+~-t-%9kl`U>0NBpus0@Qg4I5*&_h!)n_
zki1mbT5GmCHhWp*;?BOV_-|@Ec>3Ey+hbL;z9Kq{^BP=a1)7KZurlWzC;F{OX+<%G
zd_y$9uANQx*kwW9hQSVGx8D={gL?)a8~)%(kxq71=oXecu;h(08Wnpg#-QwUeH$Ch
z)<Y@fC?7_?vYR=R!vQ22UnYHMl&dlVqNb$c3W+hsE6_f?BNwdYOQNl1^fqpe;m!K#
zS^?V)<Rlk*^1lOFym~^Q#1OkUJVvK$wBs2wb%(z*;IcE5x%z;N;|p3RAJzQ48F<=l
zWx7vt?LqkO12O9R+U{aV$2sLx4w|_Ic|VD4fBbx!(_JikuI`xP9VA`$fCe|<kZXRt
zh-}%GG>yd*Ftf*Ym()RTogRRq1W1`!>MoyrX}7_%9{AnQv#X_~DoQG1cV;A@_3gL1
z;ZkVnQ_imyrO+6e^Pf@i!~(o_pRd$iUexD`i1mo4rJ4oBKgH3x%51uZfhNpn2R6~a
z_E0ACFq4+IDR2`i`dFo6T|d~hm#~nlwx_=x8U*OtBu`Oy@q4l^lPOu_M{QyO&zCb8
z`Xi1DFMHvmrMev0e9G^=)F}=!q;D(HKVp;EV6%4qd?VadtnbQ$d5!5ocD`^r?02*J
zi-SQ&iLO_MW9ddjRki`edHHta9OS=d1^N1mAozsi5=ZF<D#EX#O~)C=?unz889O2%
zT}Wz;J^{@+Z77S+H14NeoZcyp)f+7K+I`s9Bf2KLF?O@F+!ugvi4&nj8x6U9mfU45
zJy{JSZ4#Mk0#%_TskSA1%uVZ9n*LSWQpa2zN9Cb*$<-Cw<Ai#Nb4MmZB-8X*L18Gb
zz@{nH)V${uzHpWw#=LnJtnSbQf92?3ap1I<pnd65x_da-zqqn8KdnVzD$-478>azB
zly8^vC$~6Hy-Ri$R1ObZJhfY4I*7D8$dJ9+k@;GwfY4jnAnkMU21~-HYkkn0sXk9F
z%C&Ah+~+7+|A3=LP7&);D+JNLq#YNR$Gd}BwuSv?J|}{c_u@hdEbVQwB9q~8#(98~
z_OjKW5nBF8fb5S5;ZDuMRW`#BBqaf&sEL8?cJnnROB5At)P`O-!mNkgGGAUw^rl3O
zvmHUcESse{M|3^Yh2=h}{Y>L7^F0yfHJ%`HvNP?q!5;&nb_LecO%T?k%aZ+Bu{0uk
zwIOHZ`?=%*44&c}aWCJ_HWj{u!uDM-H&*-VAho@>H+gF{ckQxe8HIFSPisk{HJ`6<
z<4_XN<UFmb$`i6_m36|~cKp90)m&D({Mm4!4tM57^bAj`f%F&_U<EMep-ce2zj>vy
zxkGWo=EWus8-q?bht(<}5iuP#9)dUQ%F@L7JDFDmH*#HkzqzLAiuFuBo}z7OF0K7>
zoPaMm6MVwrl8p_HcwldXmZp$4X!Of&^}KG^IKsK1JM#-W+onjb(nuo98-{*va7FAl
zV`>>$OIP6je3nNi8ZUe@4g8%s|Cl#Hw2Qx63(fmw3I=>;FiP1OmC!90BlKm{6{<K7
zor*k<XFLT{X%{Oa7uvep6PI?R4Ts^?ShXUg*`%o(sI_<UEQa@We!qw)6S+Z(_hp6t
zap$l)KrrQN$G<BYH?8>(?5Od^Mjw=+w1DfN-CG)Hxi$Lml_~L0RtTbPq0|dg*ll#l
zzn}*7^>@qNLKj;i@IIYCFQ_Q}>|MI_9c3xyI^}p`&jRoA4XJgSCBE`14aOn!02)`e
zsI&4q;ZzmMpWh>&s5wbC!n<qWWld%G0U-S*Vk}V_am{b~`D2Dd*Vaa#<zyhjMTtG3
z*aM;y+^waTz6Rq^2GRSuFDBPnbh1EOWUm3DDJs;x06%qk+19DINu<%2vN4O&a&mP@
zr8&YdE0fDZcdwh;#@}X~`>lIVZ%AZ#*YXFAFi!P<=l;Lt|9;<@<6Ls^df2vSa_2k-
zjcjnG37(5>smZqXfUUh#jGXR>Fd>!A0<vEtVC^*#u-<t)xxu!OsUbbTcs6^fO&gVv
zz{qD-{l(V_34oxnQj6DIE7o&+Uz(qbMpIEye90O#m=~17D${0-;xqa>v+lk6aNl4?
zn>~*1&}Eg_sq__6{8FEC+8|-|x(ToSJwhZ3^sSG=Z(Y^%wb9{M?iW?6gvUSKA|d}M
z_5DFZ7OaVJx-tirA4Cz-8(~RRPYlpb4d%>14e;suZE+Pl6EU^#i0Qmab%T1Mf<9Wa
zPx&uIPaQpo(qmp8^gNbD1o~Rn_Z43IRpRrKS19pAJ7X`@%g@c}yj;GDB`?>N;8JtO
zsB_#l@vF?mA1pV=<-6D*l&9=e^T{QkxRY16w<YeT;@5k*zOv2oPDgHRmDXJSX_q<F
zF?HQL*cIGN^!3zb7sW)Q+O3SJ)w|O2qGqA{m5KdswfP#y$(Yx2PiOaX)wL^-56ey;
z4OWV1qoSx*IX(uP9?^65r%El(sU@0Iu1MN^%2s-f2Ln+22`GT_CHD_Eiius5KO)zv
zOKokVp>}-NA+}Rqd#xBu-^;Cj{-pJSbJggkUimrgi?6eq?DkDQF>F^w=T<rJBRc+5
zDZ=#0j+lq*vC|gy$k#gtqgmC{<$u>{{2$|J&oUu#Sh{TH-fi%T<}7fBU7f%Cjz&d6
z_XMH^<y&>*4^X^RYzW=b%RhXGIF~Uh{<Vu;JD)znyL5A7?N`j3guQ}7V!Tz78%76-
zkI%YS+4eqg9=hHsuF%nV=Ii|k_y#BuE!BMl1YzkSRK4^<s-u!Z?_~1Prgwgbr=dN{
z&y_Dp$&z=cM-cjUH|(XfgM7Mzwaca<#Cb^C+|1<lU}s87=p7~ys|W}PH8*C(DBuA_
z8@xEjXUBYX+cSv`Iq?Iy^B-$NrfIX_7JHilo5e}ryi|k#G2fT}^+2D|(aOH#-H03c
z_=B2Q`D<ooy|w{!rr*)%U&Y^lN30thPwpxQ&<pXM*!P+h706Yc9_;4Nv5F+QFjEw!
z0hBIz5`#xY%;qDhqm-q*!PuQ}mWut&VDsAT8$5-XsPk`AcgYf{AC5&D{-$i!)H3r9
zD@Mw|%1Oz7cyBz>z2_mS&kS2Bvo&%#f&m&_?LijfYyM=#B-&T!+#aXLajBFQu(+jN
zLR?;GmWAela4H4Mi`Y#+=B&-=*Nnyd$t$sSw_g^B^^ao=e)dVkTrEYJnSv~qPt5+v
zVQs53^%4Fl`e4U-MEJJ1zE{gYN6sT`x^p<eKv-Wyz#tMf)vjp_ckE>;R&g)R#%}j!
z#97qU{A`eM&Fx*fVp(Pv`)jplA9~e0($-S5nE3?;D`3wvMRxUK#kM|b<3sN+EV7KZ
zS@{B7D!jOo>g2H-{rx+k_`eFnqG10Hldm_jE}5wY0bKK}Xd+FMxp%HogloIx(Ol&b
zFskDM*V7g5<0I~h1KLJ%?WL~vV*25d8@5adaDz6?sTw8>-P=Np50Y-x%4OSI`wqnF
z8<T2lS@0lk<<FKwm2X^Xie=`8hj-+Q{E<Llz<e2QQ>_*95m-#FMKstPNEagavb*fb
z5SD-M6L^Q;Sj#)R5FfPTf`R#WfwlGZB(*oXxP9qbdnVZhe>SK8VdgDnN+b;VdvYKl
zG@RMWm{T%dK`GmNKsYtvT@^H9Bm0_=emKQg(cr9=OY*{C=8hS%*;3S+G;B)_p;zkc
zq0JMFE*cD);$J0Go_ThiZ*oX3k>OYHjW8<4O=fRReSoZCGe}M2He{Y)O{yGKLhyU$
zu4ebhuQCx0CM2&Zcaxg<jo)qbKu8=F`e<wcxrvs&BZnQ>u%U>evJHEC$Awjul`q<;
zsHh(quhl>^x$7pH`554b$F28bWHYaiN1`UXo`a`axX59+p$zh|roD>{BYhbI2-QY7
zH>W2pCj*{he0z=r81SXZw6;SZFY4&`Iqy6IQdH0xN$X`&IfNQNY+z|P83L56gqev-
zM&@F@39@}q=MHExF4W!Q9EieHx;>^~ZBo8;HdTqMI~r_H+~%~YW{!=|)XoB0diwnV
zSox058?Tg&Z`(~yMHfJS8QU<OUZT!HD%{vogUUY}Ft(`sB7#$0>8Etv*-PQ_7D?CD
zldu^{y_IoDm>_a&Z?w{B?eI(_n#kr_6oy|1B({4oaSOZB40Al6X}}%35v{)VlF-`t
z-!5k;@*fK``Z4&My~Yv3@$f*lCf?`C_Ux@m-Te@)(tXCX>aI=j>tlh#kP?viRUZuT
zMmtrf96ULCRvy_71d-<+ryR*Xd=Nlg0CUK8COvi`gHCG}iacBVT=r2B(CHk0`8DOq
zbz{4lajn?8d>QWhsZZHO|6wz!r||u!3f?2;v2;e_i?#PaynBJ6h;uPgB`9Ek094!G
z=WdJ`#IDx8Dz;hbZb9j2?kOa##qZ<1q!NY=44JIZw_-2C8@@U(OOU>W9F4TvjsL!{
z<^RgHdzQx4leKhj=G6=5^jdoJt?_R;3|IH+fX;leP?wQ$Fz+-YTPiC)pF_5go9h%3
zL%82Z0BlfKsHz>jDn`cQS@&Ev?RXaF&l1Oo+Bdau@pp4ROiCY1Ksqd~##j`ZtHYpp
zEBo^VYJaF|<P=VwL+N08CS5jk_>9JY!Kr3T+N`CKD9r=8oLKGjSp<e_AaF+HS4^E}
zho7E0a9mX68AJta=U#Y+!II>LMvHo%uiFs_8d(t&DLE@=e7tS0>k|K!3Ly~{EXefY
z0TKn1O5kAbm!)Yf>Kw^4Ol(VcXyJS~0_JhX_=Z)t#f~ctKJ*Lp*0F;i{1teiuHajB
zJhsb`JO{nPO8>?usJ8QpBO3`oC!tYyfEcGx0$$K^1e0i3&z@u`0msU!_oJ}VyRNXt
zi$h7WfE?$?I3&3emo+LRa!QdB-LE3cE!egXqaKquyZJK_T3Yt9E$TqeCz)9J*xWZ-
zmGT`t@kau(>&AN`1M#*S+Kac5R(39pSspN!0xkGmw%(<<#&0p<s2~{QYi7-nie??%
zlrUbeTZYQ=+$-zB&8i16^#it2faHH~tM#+i+HX73+*or=+@in`uhKtT<QkE=uS(E!
zoNERnaQA0)d&8mR=xKKUfv6PijbUP-;4`aG0D<@2eI6S-BCnL}iv3|KTGDa$i|a>4
zo)W)`A4=!4UYrg{<ebpY`M|MW5wB}6O=1dS8y5^LyhG=+D|J}gRE!(pQ~hc;p^j?>
z4}6G%kFlJ^-c#U!%~7_XbZ1y{dGPEw-^ScK($sX&$Q?=?q22^)MB>mKa5fjZBRlh9
z)=mbl*MgrAcGeG5rdfWC4>nXcTL@ezVXu~YNmYxk#faZF!ixX21vJaGECz()Uue|w
zq$89rvgs^$V|M8R3p)IP6Pqh}?OiV|t=0O=1DIbF0POtUsjJlXlk7Gev3c$SdDRN}
z1_VwW%M_yk(ErHQ5o_w6pXSQA@CnIiQ#KJ`@U@n9ek$1s${VEC8`qDjJ-NMF_&?Df
zpQJH?bDkZj^m;MUZnRvFJT+kR>238*y=H6a3RdLt+6R`Q_z+I5^)P<%7jk-kr|@(U
zKv}+><SzeNYy2|{C?Z5I<p1=a%uM)<ksb>{y!8wieV=tay~CJZz7N8|TU4rCPX46E
zpdNKZaT9rXkU#gFN+={Q!Z+KU7Rc}zQ^uvgf%D~XHl)mbx&G=;@PfBLIt2>{qG-f8
zJ9{M%G-T=b2Zd}1Pj*^FJAwHIEHTio67%wAw;^MN@nZ$x*+s6s{K!5in8ZRaZ;Dq6
z@djfhc7%%lKN}qs74rax#h_EnSVujSi3(L7micBS^FB@3Llk{G;^-+_XL>-#CFo4r
zIy)PinuK!0>aT@qiV49XAXeq>2&GEBGd%(9C*7VGHwaOd<Q??a6ixKZE1?~?W8-4E
zN4wIZ+YuH+=sw&>fX&poN1Z(>?0K&c7(nG#S9jM80nkTfEX+$$jA?m9aCtAAbKUZO
zL|VzKSEjO=EIO$zp&UxXTg0rkJ^%Mh{`(G#Lxa2g7m`pk(P1B#v{4uCUI8SZK2VoB
zo~=*~R8e4EJSK6yj5RiV`9R1^V`VUi<mWHtH?Dt_xLN0iXwu|D*}_!(0V4ro{(I>%
zZx5GtjGpt-lPD@g@X;UDmu|11!b_VE_)l*UaEE9Vy*t8vo?=oQVK_J7J2*=loMo!(
zGz^(|HN8dnoR1+AR}w*aB0^iI$XM)QCg5*0@}mES|5iJB^eX^pz{-(^7YEJ8&O6Sc
zz>>_Uu$D!?+PxZ%NW^qpKRT}zKXim{q96jR|HDP!Qk95+Eg6LA=(B@bdC(CAICV15
zJ~-Ug8+<9zw1YEtWd7Q~Ii&Uh9u)-NZHmU^4Ac;a8BWT}9l)Cy?*@7S|3)j{%5E?H
z!0DK-fj`}iWIW=<#;+tJGgSJtBdxX05<&Ii+;|W6=dk0ktt}H!^L7np-9?e&_Fnx3
zI34dy>XA<W%M$$e27v13=Wq!_kh(Y`95c*M7Kh@gi8T%%mPfunAYhKYF}(T1l1sI_
z1prAS2^)8d#b)iZJs1HV(u`e|iYY(HP`o*54EZ@WOv$^97gsfwi$#lvn;_M=Y`u1j
z!?Cs-eR6pymn+vR=G_1fWJ4r4ZowZiJAZt$NB3R+)A8;Z0;UL}L&AEH)m<n2N6%?l
zDugUK)eu2+$%^SS$x+WXH`i|F!<-@=;7^?#R<<Kp6BqvE%oYu29|qzQueoV!UEep<
zRCKH5CJo#VDQV9cs<_BLp(x$>8rNs-)D;a_&6n44Ua{^i0~z~u4|D!X+k5}<gnKs%
z0|$%)@^|43BCz$`Dx9!j`aFfC#6<}L9laN>Aay{%gle49JL40Ut^GpL1>$qvB}7I(
zEo0smPBP04<(w|Dh5f?#B^$a?_@~u(y}4!D!*Ir=xv48y+-bRRrSj^xkXi-YUbz49
zeC@BTxZ)QtFNh^S*=RR-@pZ1N-8KvAd(a0px0?>~jaoybC8g`kl8n#!>fsHhO^AzI
z+#C+vH8^CqCCvtTmFEmx{zaq4fLXmCT)K>>2F+UQjO~-9r8zhh77?{@Rs@wtmjnkU
z9V}~1-usm+!I(r+Vp^|;ZKal!6Vr5JCC6XGM?!9}6fk!4m$b=LqYj|@Zl|Ys0kHro
z1;^lt3#`5bsUJ$ZBN~KiFWz1Nn!gYCrTM`5Zwrz0|K8sl<~N#uvOXLTum{ZGNmD#O
zd9-5ZqP+PLR38yuK7Yg{%6Jy$oJBD>137DSmY;$j9kJ!!W|kc(fB=(KC}6AUxJF)m
z?jnLTLbbyZX$CU}ewr(x6c|q(o2YPjwTU}@uPRC%%?Ho?X1kkV*KnjVIP)(`BEGHL
z0Ny45j~w)@Bw;UCAy2irgM2k$$v<Uzi7wA1z8O8aV>S@d(#@6BuTUH4q11bh_<3a-
zD@;W`MHOsk`IEfxPi))QHjwMC0V7z<Nm=_N2S=>8Uqs98&oJLx27%_TX}a-E+*De{
zj`U-;;_usQuJCRB`2DTcA`n`sLh7Hk)Ss5oNgcNOeFK5uQ@9J?XY9^I(<i~$+MUtj
z3JioubknAe^HVAMA0nRJR@@i=Fdliv3&6Y)_8{1w&S+XbDl`Onn6F$bjgNrDiH%m0
zz<crcP4#1F`lV-p(1w=hWCTaX3JhjvKBsDlXL;S!QXe*M1K74fW(D)@)shwaw=I}M
zaxtFWi0^!>6_-^FErm0`kVNx{5r<$B^@M-e-{C5YY%2S%wXp$gpL^W9_aLBt>E`4o
zu338!pBl5Am(WwKZa%iIXn<7NSV|KC|9eDqP70(!CE2iB?44NjzZE1q`hQAd><h-S
z@%C$a-L)k)9es%+Og?`hh2V(huszpk!#PYB)!h=N5gZqR>ITv=AeqzzTyhW)!fKSl
zc043bfRuXkw*t*0UtdxIvXr9X6801%C2OKM0`iPk5MQ&FwkEiWN@S}7*)k{=Q*6?I
zf%w#k1$elUrHI=0rsIJ>e}=m@7v5f>G}Zs@L9j}83NSHMF&;iOxE2`qzY@vievEGy
zf@en}NP!I$cZb>0s!5ZYSm@|EV$DRqf&WcrlhTD!)95gFivY)Cu?_XES)AnR8k+Oz
zv_(tJE;2=2T#AxE?#ntmt{LYO=JK&f_CZt8&bdO%!*!S5dYe;@wVNm*Y(jsb@*KlV
z{lQ|AN2P^@ZEj-6ttcz>J}t8X;r3;bZ|gT4#Jx?P-@AgEL6xAfaeiw?338bTuPqLU
zFo-j$ANuN5^Si-qUy<-U;oN^@D$-M4T8<hor@62!*!y4ur0$UJd8S-kU0NTRsRP-U
z_7AXvx}whRg2yT%X7#NfxJ(-=nz_jqX6GZT^UTcdj=e3jKcwVd!~$_hAzQ_<V}3mF
zLZNS%xT61Xdg6G-NZAWsQ>O&lF727b1fA68aCSu?fd*;BzIZ`%ACyDjW*mQ6I5tsl
z=;twhU1+I2J4o@d(mu=}7S!fH`t!C5MreG!F77)&;;}|P23Zv_MRMsrKRqoka?v8C
z^X$4Cww_=qqE~X@<irN>nAHXZvS|=cI#$N5T#tG7Ptf_6Hq5Jjg3o%?bCc=3EG&8c
z?H)HXB^O;d&&(0k!GQ=fAySR1>!aecD>4^Y!JCi!1FZ-L_NFA1iap60{lF<zK|2ox
z9P8?2u2HU8F%&pV!36kUoO~0QH;&cQwT1}kc+@|61woN?C?&az!~VXw3dw&9G3|dL
zRU=8$&vkaj^q=7hQ9y+(fu>1kbhW+FdB5kpzJ`!+!6EUD9^-aL8{h{2=?1)WWxmZM
z+iFUT=!Ldemfz2t<x&r+i8t&9Gka28X6gsW!$Q1DFo~WdDh2E%u3B$?tVV#s#3Z9+
zVd6zr$1^v{;V0~9!m^|JdYZ@zFaJUq_kY$pc?OTTv6JgGX~zZ7Auj6ffmT<KR?F9C
zAo;r3Ryc{}8SM+7s;tn7$28pBE++<a^qa)SJj{?A=nI6lF0yb!7Ej7S<LgKG*H^y~
zX0{%=R;-t>LS?q?aiPCIMiW{$ca-N(J8*OHh}U@@13fO_jilE)VrKl`(>wG-+TgYT
z%m>9QW=3-B(}~+nCc@(SKYkobplu`Pxw;1;28+7MpLKKu28O3jH&@)9OP5u>e^nxk
z&?CporCXW}CP}bVu&%+l;-JXA(4m-9_#l+T094Y=$xp$-1cCx4_Wu%2<Q<V&H}D)>
zmp~k@RuzfZ!lUh6c`WVr_V&8|<ZYzmvFf;nhC6a0TA(wE+@fZ^l5K?N&7weDRGx~w
zmg3RX)Z35(CRQ+ZqNT+yHH`Ofp{)6@evul2&Iq}MOz-+gc_mV(G3#C%WDOg9NldZ)
z@MrBIqRdFV(`|$dv-zJ`)UntSHgK_bSgBe|%<pPO?~@1udI^fgc{W+ibyU+jCnxQ?
zv8hk_0t>0GC}-Pj1#)94fHzw#;trTXth%YegbOw32Z@Tt9n67=)F9sE1zpc&-$f>U
z@sii)qKY_cE7~%5CX`j}-aty{Q^1irouyeMUDeT31Gkn3gSNh@)$&^jJR(FwGTBnQ
zNimvMVk85@cgT!P2`QT#7&JO5E%B=lWy9upLc@q3xMjN=JA|d;8*}L+6v%bd1fTTq
zjDdc_p(K32()CLS3L6C5?b5{3rfBPsXdLf(fX@cn-II5f+l_g71)1w_Zr?qP-(R^U
zA>#K5bxQ4E*l1gn+g_ltPom7nkNnQo1hw?h4P<<-ZTsd>H%Nuh9CG_4S}f-Boi2Di
z0mAV#jE9%gk9I{m54QCTxD47CP31+~+%maN6;{yIU0Qqz@{`GYr<@m5La_Y~Wpwx%
z$qkOEJ$QU6D#<U8PWzn92AJJnNNANW2U>6JQ5X+BP3Y`?>)6psDF_Ms=u^3RNLNmO
zkXgHQuy=&*9+1xqUVvCrmR;=6OfN)>+?gy8lNXP<$BO`rFF(8M<HMd1kmeQ`g<b+@
z)Y5|f+!kZ$<s0UoA5c^0p>wyYYlat$o_(1s>>yv)k`h7yoLtYqD3Co>##Q(m=gD6*
ze|~d7sac>I)2_%1c5sTsvJ+gr@&^rhu+BM689{FBDT6PN^8jRP4r>%|lc;F^;}EiQ
zJk>)r1An>84dPgk^YvKg$@UczHRtOL@VCG?3rH`f^+}{);`o{t@1n|A6a0xyYH#fK
z+ec(<k(TD&FL#a9L5a9pDa}%SjVzpsWQR>Kd*3vxe(e4ug&)xv5r5_^o*9ZRD6^VB
z1FbFcVo9Zo&h)@BfaR@};yqMe2fSLOD1;kD14Pbep~*a2N=15V4`iL^Q#abrNWhiA
zyPYSzuGJVx8nNy243Mg+;(o~%>o?z249kVZI7c>nTs3kwq~Ymqr7&O`djh8fc+(*<
zc)s4ruAe@Bi<!O&KLPp;ut7L1?2N0#js%s@AprE-E|W=v(Mwz-Cya(a{n!8tT0w8H
zfY>2Cy5@glQx%KrS%R!q3RL*v=rNuEhJJSV2b%{U%-9*g50F3Bpb?f0^}@im1Keq%
z#MI&X^7sjlqxFY~q*A)n%!;q;g0eUTg@qtO*i^@V)xA*rih&{U>(|6fWPd-uK7)=U
z7V}+$&hBnxpJ~+jI1p2>gIclAKs+1-0QwI-!T>%ggP(w;A^p`Qi+)MZUEzd$X{wrw
zrD@!P<0f3Z9q=0No^90+rr=Gft9pnps2^j<;0W}ZERJ?Wv~@})kNjA%hb=RtDP(x{
z6oYg!h~%Rd8U36^L?U^6=~lK8`Fe$OvZE#G3kEEl>|EMx4Ehll(b<}NJZCj4dt(uT
zXgb*_@dq3da;oda6Z2fzA5oy-neXkL`_kmkpNqfuSt?FAZ0qgSZ_=*^f&r~f0FbV3
zEYq8E08@;n?sv+4MtAOr1@QH6pee7hFQ3b`;^QqAoF-9%X1cQ$c)n|P?iN=oW^FKH
z4uNnVsT-c4-aOMR60;}eF-aULSwlZ=FHoZKiJb74vme&=<SRxRwY9guO&vh|Y-FU8
zqB`Fn5fkIDn3>%VMf|wO(6#4m;m$@&>zkGJ3gOkOR~bY4ZRN0)c{sec#h)m%0*71(
zIh|DeHX*kAuC1qMHe}=+xF<F$fl%UgC=!Or_8ky`VJ72tpEfsnn>>u^e8pZ%7kkuK
zu1FUx7C%D@!bb#eC}!&co(b~-P=gT=^%O;V;?bI5HDj`A?gOr}^0ac}tS=7guSeHz
zeodO`<LIMb%9yrW+`f7|z0mGVSA+{sQ0EuX{OqZ(z=Y?sUdj|phEl+#dlYMXhl1&o
zZT6?GE7Mv9xnc}G(+#1(w+sEiEZ<4(2l@C28eSGkf?B?ryR&N}yxDup^mW(_&fme=
zuSbDqK^!F+RS`$(g7wS638sqms*>ri$`N;|-Nwxb+4RAK=)5oCje0zJ<*1+MWCB~K
zr^G>yF(ewZI~r;al>ZT@^d9#wz2N?fJrbMrs;BW|axl4$!>yexaS7dnbEPcJ8AENQ
zdr&~?IG4Js5^`TsY<p3Xdxb8DP3u@&j3%EMOt!59q6O;aCL-DWdlSb4@8A1W!*76Y
zcz4vt1K;$+1aD#Qm3jpvGPXOS6anxVekfj}y)-uMK@JLXp8fz9;AyL1m<0R~BQ{=G
zmh62%B$Gavbz=4g=u@~K4cmohDjxD{jT#=&6AGB=$$8HQHz1ulx$*%e0`Y#iRY`-<
zD>T|qqSMjk#BPn^tsrM6?>%o_apyu&dT~4-_u~ZTRJZI9j<DuhRVg^d$xwG%IjENf
zP5DMhf)(7?sOUBmmORvM)lm_OpObF^K-fQNg*vFJ$|OgX`vE@)K5$zPsajS!R(r=d
ziQ~isoRayM!^>>M`Got|cAgxNhX=$a$F?S@I}zGIVUwE@C0W1hW-I}_Vc^8ca!0c+
zp*!UM_@u5=7($WGzB1HXTOiZSgw6=u!W(&;yGm@+7t{!kn+gZ)-wXtg<vYH0h)^<$
zKnk?tokaj4A%J)vyEyPo@R~LQ049T@md|)v{shd|gvMytQZl_HW{Z0ZvV5%B(Sl}D
ziZ5o~?s6VFeATDwP6GV^iO+9wG?yVL(i#5N-VSO~=W58|P%Cr(%(xt0B5`d?WE)@>
z_Hp+TIrbz9WFhl_VgtbAOPeZ?CIAaZ$TJ%=d5F5#St)dbXTH<O!q^_%z!af!l}qh@
z(uy9m$~l=_cXbX08kXtHnU%iaR1VmDu2$>fk{<SKB?+jyVBZq#s;7}stWKdiEpQC#
z3PYB0&%;i&oknmkd*yQ+dj;9JPL}XS=ubhi9~Fp;moT1ZARqnUV8T79oCK}1Ce4l%
zj!0&)H7XVW|I{0j>#XXEygoNtbBMsa!H3+7zQA5E?{G6gWH<Kl&TzF?)G`077$PkL
zElDXyV&IT{e>g4<)P2aP7%`jl+HV;t6*k){`z>ayiR0R(XoIy#489gHw?Mf7^z7!i
znz|v1&T?Cm7hs@mc6IEDys?t?Y58v|Pb?cljC4jNgtUn~WZ+LUr7Nx-Kg*$RJ;J@L
zVs$ckn1Mwae_{?8zoqG)h_v`>fVG+{C42W?-3Emc#h`zJ&<m^7C+pK%A3=n3kJL@f
zlwV3wB!7efh4cjj8&RNQp!Vmq>p5XPR<luuqG!CDk7P@8<T{i3`EpitbS+LN8T0;p
zReD*Xcfy7%OzXzh3cOmcnnvTRmlMEv6G)|P0xFA<zapF)IE;DLMs3+Y#BZi@PG_Oa
zg0jM9PpAC^Fwb-2fq}3+W|<3XzLE=AZZi)fpm`uo9*!r-S5x_CvbtFY?CG<erkSCX
zdepvzl?7(wsTfJxl>+Gep;itt<#NZIQ%m`XdCQJv$QrN3D<ht$G~A`E8B%Z>ie2u7
zy$l2t2+VlB03UI~mLYI({u97M-+@U0gyG^};I|UIahGkkLBIF`axU48d`i77!d{3V
z5mVxTU}D5V!MOQow_o#^*&cg=Le`oyO<R5NH=Qr1u>b1GMDd8VWq$&H|2`dUlu9>H
zBE*~}K8CT_;|i10)8GhVXF?`ybAQV40RaqxLW1~#VWaax^hZS4*ZgE`JoG~gv%bUF
zY_V<bkTRna_1C3Ex&ScVm~uI42UHh&Hz5IoVv$kKvOjI9+QlmQ=hwJ+iy=7p-jV)9
zAPBA7@gYWo;uh??{P%W)q4Pn9g*U892!36g?*cLAmct6c!!!MXe9%j1jImdGxDkNi
zcC6a@uG+;dT$jU{xFRO0%)uy;t9QF+pU#&n7v?B7#8U}CI!Ns1Sx~!XqHDv>vB*{r
z<BwHs00}Ty90(HOCCRJ^#j1p1Y6xGK-)hji)PE_hth=u6Z>u#whFwNY5LnXI)b&a-
zCiw-GY<lLWJEU6~7JK3>t^|k39;Li9)00I91C__BOr@_lH(!+!ujJ=dl)lt4vpJV3
zF8Vs}t_>rnCc)Y#mqsa3v0(esvK)Uk5<D}(;S&Xt1^b)Z+%wnGvev+8C+FC^18mB)
z!FwQzJuj2%l8>@H<q$)cmAO4;i00o9<`0xkoN_9FDUpMWj7e(cz4(m)SWVZT>NiwM
zGdgL*BRyDTb|Ly3CMLk`^Orjf@`K)%58w^JLNAYg+h->{&aQkkzow)=r(Ub0QOH(W
zM9xJU6WZnrrb49Q<9mezZSSg(-KyocoN~A->=Awt+sMK>w%-L*L3d>~>-BBXi-9st
ztvMDiQzYqrhgibRMiw*pgR>kh-9e!jmGPzdq4`JKe(qA^+%@7D^IDOhYbSs3sbys=
zL^)^lFMxbJ$_^$HejT}>^wRexDdO`~hNf}l$`0{xg^772`Jx1ADuGm6T~~`FT-9zG
z7BI9^{UZour=SlS`OpLd2yx(YwKNVyX)aI>xYYeLDn`j|;lD9>+8@X+>QQRy=w;gF
zET*0Re{XehWvO6u!y5#)?;=ph6LG5?wpJ+0`$?F#wM#tbn7<Q*f87ADqr3^m{?MP+
z1Si>m8RP0>U;Z(5$t7bP^l7D}zi6h?afHCqk_}A2Xk!<GEQHzg6QEOp_<F)?hVxh+
zDRCM=2b%H*KxB5|Ecen55pcs@qv^=_LeW_(&e10g`XjSq{d4Bv_E;JZlflk80iM!H
z7Qwc=Bml|9#H>J1wcQ;vo>eVWDxZ>92K_a;>czZ9zlxzXZ<TEF`38F6Vhf>=)qz|m
z7`_By^L$sTXhPfOt_nO5f_Q_~cvG*%4s-*u9X-U)@j&j-AK`b4Y64I@J~-A9@`LHu
z?~f`4>}OdCkDb-v+;b}>Q{752Ae7kApnMZkdqjh4|4BgJH!*AtS*+jZQ2r$#vvmzn
zV4QC2I&#(Zkf%{`VySLx>u9Yh^4yB;?3iKIv_k_@=~Oq>t#XR}_b#WA7Zm<EJ;+A9
z8^`6D?}0?;)t#srfe&V_rkP4!^Lfp>^I?Ke+QijQY5TSpDK3S`*$KNY`+*J%OeL4Q
zbinJ!CY$#XtVU1Hk(ISpg+_Vs-4tW);c&Hp<Q1zL3e0)Nk)G@#_GW(jkQ)L2>g=Ig
zk4r3nGR+lEv6M?Jwb4$_eXl-~cipXErpQ#(<%R_a`{zOJVgZYrN6M;)Y7Yx@1sI&Z
z_auW^uIotLs^xH2W_VvjVe9q$Dnl>!d@Z4HzbqKFDk9`1f>6EYG1h%5Ex)=Q%73qP
z^&^QZZS60_mO0>5xcR!zzzJ2IWQy}t)1Wz8S=N-({(3X$AP=Jjsq}O=n#01az|*Xg
zqv}Kr?q!4pWA{dvZMZYxt7>6D;!^yK8@q&|xv}4-OStC<U?{Kt%9OV3<d_?{_3mPU
zxGmn0W~z5p5ek58fXb;Lez+oYIwqM>IOVap2<fjrPDu`}RvP>Cq(E3$K?X-wR#^%M
z7lnp;Rr)RtCNgpuofBO0rBg(I_;$7Euf^k54_9G>D0*s(>D&kg!Tm{CZ*{q0NC?%+
z;~OAXc0PE_iFi{-LgV^B&z^htT3$6lys8&b{qO~ub!*q`wr63$xILR{9k82!OkBNg
zwOL20=fa4U)c<jjx2rolZRjb18<MZOXk489I(M&m+<UnuMlepjz^9?GSXR#f`GHa9
zrg4V0cfjMje6GsV<q^ZDsj1uOrU}R4*zQN9u&_G(iv8qAN|Iu0#2V8af;wbVIh9|A
z`gv<nuu>DYdelNwgvXl4^Vp3%*I8wq8?Mh5c)XV&oW`ueSykpH#lS#QP=-3Io`e=y
zUFqw|Gegct<-d{2s%5&tAD7xXnO)?okZXuSh-^neG53aQ1p2VPh2}_taz)_LZ}7Pf
z6^_p#cxlF!=Ta^=SuE2iS`5u*6AI(oArg)?+8rjup@1x!1nEfN?~@HXhB+-}(}55b
zv`-;eHourY)iN9&oqgp-f8|KxmMY>5(xr<1FK;IsC$Z&j#WGTaD5ajco$aB~@}c<F
zvTn*_%wju?#MTqKk6iZouuJ&nM8WUT-4H!<;q2ibIHVohr%<mpbsHN;Yg>ph$H6f>
zzwq_Sw1Bp{bx^vB|Gfi}H{lo2%q2amqEaeZ?dR_UE1QnUP(Jr<*dG-$SXduXTV<7d
z;L>l`xCIer$0bko#Yx7ND`&rlIBWOtK`NMKoQs~|`04&`t{VG7<t1wE&x<gJFcm>x
zsua1>4I;P4C=gX?19H)Jam4f)DXO?}Kz3}?k~{RXDyBbvrH(A=J4Y>2ZQa>{tE2S`
zJ;REkMSDF7!kmkR%crZKwPsYZAFAh~Pffj(^wC{jp`c-Cmy-du-bD|FL3h^!<um%Z
z1X{`I;rNmK;}QJ#Wmew_4^{eHX4}NNZr#B7?ch7m&E%YATxWOtd@e_aNrw~#zP9G4
zim<ut7-xg)d2@tUu{^s0z6Db{y1JcId}epUe2mG($Jn!p)^;_UTSo8~Ew!d}O1f7A
z`{Pq>lZLtZ9F4pk(1@7ATCz!Bu$;Xzt}%GO=@HXx{+StzwBjpEh<m$~Z-O{n^Wcfx
zMOPoRPj&Rz_UJyVR80{HYigTvrSPutpqJy6h;|=-p3d}2^7xT!sns0Y>K-h;Uyqta
zg->aH%RTN2T@>!vB;;wS)B8)TE$<*>$e`|LFJ7Wt`Y4Y%HF@Vu>QjcyQPYB8im^RG
zGWzXEuNZe!GS^Z<RPl|<P|Mx$%d;08V^8{^5v1`jE7>==uTwbfRV=`Tlk(V13ndn(
zO;Z;ak&N~Q$M$^lEwRSsOGh=c{0W`s>be65+o~Kz1XM~4l+Kk{*~)95EN<a}!iT@d
za2rPMH2mWij(Bkjl0_z|Y6<}_2zeITQfWPFWOvYU%GV}B=4<$7m<f75)TB1v>EmrP
zcD`ZpOgiTU8wF`nzZ*S6$YL9IwCXi3EpL|OYKJ$0zX%};U&F}$aa3rCNYaw~+I3(4
z3X@K<s(vGz+USKEM#B-qb6So&?&@AOPXqnFb*hdYG8xwzbsj0ZxAjUTU&G9ITd4MZ
zH>;fnx|>%Ck1y|MJB;~QLH_3wLdIs@0mZ0oy|gY9KIcL8@%o(^<$%BaI=s%3U>r1}
z>%K=VyxZLg<+L4iBukFGt>QDq4Oc9Q9)InzL^im1vM&9;3hygD6^WmS6%)2MXJI4~
zRMoPFVT5JlNWptme0nx=X@7z<l{jxk)3;Q(6?ghM#ZzB0svjbPm{=ZtD4D8CT;~>1
zbHrH4X1H#coD9~^q{_c&REhq`7Ww<T_No=rYeKX|y(ZY@pd%PAtoOk%#&*)Ax?&;P
zJ-oPN)7!dL54hQNf^Krn^SpLCrEkvjMK<(1NrjRVMKblhtMhy*DwQJ2Dx57^UAvTF
z{@x6B&*;wxp@^f)mSmUOsr$(-2F#Oh)Q$Ce_O)vgq+ixnn3-udkM>^2D^IAuVn?Tg
z{tFZQ;t5J)iUP|lf31Htz%<H`$J-kiE3ki5E$(OfdiB9rC3>-n_#vKj7)>rzSY*>&
z>1y{ZAbTDN*pyr6uK2dg_YK>3?T6RDXB$KOkqalR>;93+BvgSW6l_$P9S7NH+oOSN
zz5=J0tq*|06Ad=Q(&X^@?9s+V*JSWjZ*6NlK40B<%N41vb0t%&Ep4HlJ4t;xakbtA
zSEHR0M};}k*&Vg8wNDc@=bta!3HQze;V_+JLIEYI=hjO?vq!CM4IQ3Wy^^12^B}v-
zE|$Nei$TBZZX&LeD>-VM58U*#1D1Rilq<0j<}_0!2;u8<t<ssfnBc0`N<qGPFmiAq
z841DY0}U7s>DlHzu)n~)hZZNZTm@|D7*eNzl0feEd?X@{0w=T)`k1FL81I1~Zb&9R
z%X;zCGHpK4A_TLCe?mBHU=+u%=s9SE+Gc7sq<OK$oR2L{aKX2d`rLtc8}}qdk3@?Y
ze{u3>6^a>?Fwr4EEGASEa>Q-ORFRyHsqkx~)#oUJy#U)s#Yi6v%=Tzi?%kaSzm+T1
zi`Kd0M|pdXJBod&LW$5im90zu@Y4}l^`rnl_)Ds}{~ko8^!DP;#%}$=kZPs<S3xc%
z%d=G}5IzF%wY9$)NxJswuTm$!S*go>JL~F^wZD_}gfDl(Wu>Ih)6>(BIsM;sb{AKT
zyj0-ej0yPqe`vZ2pennriy$EgC`gw@NOz|SNK1EjcgG_lAT82e($d}C-QC^Y^&fn{
z-x+6|J8-@CJo}uz*4k_D^ZmPPf={Jy;6v=QtdXt3E&u&bl*Ddq1-`Sh%dT7s@LhgG
zSu6%h)724F!4YW{5C+|wYPpU*g2`VY9<yq@(f*tvK`Lj-(z&UII2#)~Fe8aI)<Z90
zCfTpTx#)HXHfVWSkLxQ9x~bl<V!)Vo_;WcH4NV;WRj8RgqKu7meC?j-9KbtJ3({??
z5p$}-!t~l-O5Sxp>xl@0e5^s!alx1}0vw{?G}9}R6aRH(D3A5dw@&y$`v8k0Bb6i#
zYB0$14{4~Z2^3m^d(e10A(Madd;Wpo`NPdRn7_cv!6Bb%E~}w2PjmI`4KcUOBF#HM
zFoblcpQq$(Hdu-i8?b36Q<e0s)^hhgM`)q3oLqHrs@ytrNj@%sQ}>_+lK<o+N5mvu
zIR$<YcA5YL>8_hiJ+R`P!FwMO`|N1u%4;CtovuNjESO)tYNy_*8o}A@>=*3-M%z{8
z&WrB8f~^Ja{o_jK`!baiN@#BGmMcXjyZhFyPL(;9fbQh1eHk(K<CXo-@!r4Pd_tV+
z#GoWRIn9xh)2e299|YrE_nP9i?(nrHiu|49{2cUk_~zlEJ|s(Y;*)&0Ad)o4pGH-o
z$jHdasj0ag=I&lKY&2RuW_(b06UZIIuAjm|&%C|8ZI&wU>+Rj&LAU;VwYIkQL`$Qo
zf!_YAYIpvU+}AOGdH1MJFM}LTG(?5uVVc=;A+A3JwT-SYv5O%=y-q|dn4F6RpTMwz
zaRZ8s_}klU=WvpLxRUQbTbJ_$jNy>pS7IAT$n{O?+UPf;x6&KHM>A9H?|L`$MI^9_
zflJOSE{d;j`uvakMP~Ly>8GlbTqBFb?zi$-Hv2N+wj2BEdEJ_h@bK`UYfRU|LSrp}
z$Rd%=1E+N(f=}r?M$25uHjj{t`;(gr+BIk~I3oVpoS=#gck-)B9(U8h+Bsdj-7{*I
zZPtRpyLB*c1KT!dn&YM8Ld&5>bMx%7MyAz3#OXaB7gyZwt{Hzej*^mGf~|oAi5oxp
z4~itj3%XwbQb@P$&R0fz^VJH=)E<u+m$g4voq#D_9|s1x7sL+!hlR4BdgI++2l*Z~
zM_U#;QdaGK>_cc$t8<Fqv{lWXvT%wy3E-`P|Ca1R(?hu8vAntLm+MR7C!Go4L(_Cq
zZf{YlMSM-{nFN=kVaT5LJX>0o%Mp7rHM0KNW>8hcJLed`RzY>`mSJTxCC&f<M%LCG
zJsQcAC1b#9vFsqiy>NFhty+daZu!~Z24kv<SbJxp@Y!V9Cobe=it%B*<rG}{)@$mT
z!z-Bx!nvB@tC0>&B5Ua@Pa$}Vb*(&sq?~t@x3JPnI7)SwTq=(4@c_crcN2L%7STTo
z{(^l3R?=Ok*RbnS1Rd9;e6WiXS?KoFfz7#+ILrnQJAR48j-k2G8!?U&XMC=4+C}kA
zTHfe-{D)+pwsv%o0s5%ct&9%}M$K+u!?g`khKv@6?-#<I5^kbgZf-(F(^K^`J<OCC
z!wTIShBk|eX`V)fT6tFn)D)IEW=O6$kOnUl>XDt)6L~cycd#VGpVfqPR0tHPV!QN6
z2E19G<O9ar2u#Sg{II`Jt6fS13~+Mq4%ZbtLHQfv>uA6HHaL{fnb5F-OyA#j?Jx-!
z`54rTU#l?1z$=HEjH9;?O?izn9a-@py5)6a8;QNe4JzFo%#=|j4)+-{T0N*rI=+aP
ziXdSw_Vr90>=jUcwVzSRJsJ00OqadV9eyMDVCl|MAr?C2{#WL%O#0&aI4&sbvrt-S
zmF&FMh<{R1p?%6pC*M*2#U@e5wCiZ9^#OYJ%$ofp0w_!CiD55hTDbRkik{T?_G0Ld
z1iIU;DJGCZUy>`x=80^flxpAL{I<D!1s=9+jdq-jCDgEeH%efp#u^8@>c8f;h$&51
zK~c2dhQ4$jUodQ<Y~G#5J?yxu{E+B@boxixzMA$LaoLfTaRt}v3d~P81#l*4rYqdw
zaFn}X0BX~wXo;#d+}p=&HdSWFp#U2pD!5?S&Ui4xK&X6;HrHYPF`ZKVNGd9<|7c9q
zGymXMQAFtaCy8{0k(;d{>e#b4y#e2(Ad^qVd^aL=-x^YZBkvLCr>`nwLwzx=zpIiw
z(-G#;exT7h^0K0%;I6{g@hoPV{>cm32%8{T+qt^Q*2rB&Ck^K4r8=%B3?K|dNEh@f
zE3;Qd7vz3z(_1nW8>8)z##i--@GInlLY6=^`zPY)IZH#aaoYg+@<SIb&?iA`$A`Yz
z>CpZH+7!tR#V&EF_1p@CugN`hHnl=&`jeX<V0IJu;!iINW!9nzoxghpXdo_LG05h+
zSq}NHVt)1AhtnVfEO4(RvGMEIQ|J0HTP_0|8odv}{lurP4sJ!dbcwvJv)PuZ<3?M&
z4sc~$bu!Clre)*t>miT`6t`iZOTQU&Js<YyU7+mfKJNS$1)Q{>ljgepSdyYlRMP|o
z@~Z}ZCEy3urBRjPm*@YXCb4k$32d60Lp;xH^p74kSJV1INBgllnASa!?kMi5qgZER
znW*)IDBQ@7K*wAEng@ey9|lkQ(Wu*5!g|7mq}|@qn&1CshCr=zui+o3Y^44My&;8G
z?Kd!X&gOUy7A)fPkE60oI?<g*&7_<rKa~HabRomzx5~Uf#bflT-F?L>zV)Sftc7fk
zeEk-U)ng{&`>mNz?eT}RID1Bp4#YN;l?>@R1<YmIJ?#-kp3XjTYUg)pp&<0E4hM#e
zN)2kc?D57S*7LUKY{xsNN1JExPtotu3aaNsBXkq`3$U?o9}=0e<z^Hju^QZinv#Xn
zWo!Wq45m%`Hk3?f&805MIb(qK{&<xWXxm%aB5LgRhqQXPhef5uOLf^jL$Ruj151VM
z$bqDm!a`5PxzRgDvK7*vyqmvD8MbBduIKN#J+Oe>ejPiYA{8^Z>tc!0EtmI$K~RSS
zBqNlh!jDSDPgeot#}PB1NvDc!XtzE*dy0VGRHD!!V_7gBJ+!@b@I;9fVifX9nLMHk
zyo&_uPtHZ-bEa9bvY5&S8n-J~CQmQv&+pGM;xc$XYA7()AvSb{(MeF-2cyG^Q4#P$
zw;rhX0;m9=heeQ#t<09=fdkosz?2yuP+u>5S<V;5x9+`tG|oFceAzR9+>375-R;OX
zr1m?8YIyBA8x%6Q<51%Rt0Lf36$c71N}LqwV$t}+>PfT>^ar|7HA@U^=qIG|I<zP%
z&D{h%SaB|Ho4Z4HFr9T!&-VpJx1mEut<}r|jk=yGK5i2hqc$$hyP&FctXu47x>8gq
z1$g^kxB7-!V%eRpp8m)Hhkc0OLP1-1^<~7$L8QBC6b7dCGyIuPt{cOMU!N3QdnvT>
z1!RQ9g~=~*{3lJI4FB^UXT3*s{QER{{+<4T`quISVMr{|&ze$SF~|@N=SJQkNT??+
z=~h&0FM}Gei^mOQ(IFmxhrhBZkB&lM2w0VfAE%(nsWOb&E}^sbZUy)7LptNe<!dYJ
zi^r27hH1^-;je0E6Q8iH)7DJogi%0(DGWZ7Sh1`wNN?X}1D#+1%w<>~@gB{ZscpSP
ztSEp<wwPdpzGkzK2G%oy|HPz}sJ+UZ32CathuCXhUHY{9N_TZ#PiI9fh|HbuP4b$S
z!vKM(%%ybst)?ZlS-ds6iEmjN8$w<1-&G_K<$?u0@Qn>qE2XX;v@YtjI1Ufut!pE=
z&k;uD^<EYrN`Rds9s}%He8i%cRkDZo99HyRE>Jlb&BO`Uy51BcPVtrZ%l706o<<CP
z&1B4$<4g)cgDl>r^0h}Zo3jc(0r~Oyd)wnfh3{NEgaM+D%e&k}_TT2k*gQUAhiv#F
z%kwK0&}kK^7lcAyIvS&Aesl<*%#}-*88Ym|<vhi0=7Yp_AI8&L1ruL$9JkV}Kp<(1
z><?vb-t@AY?Im<aEO>#j7Lh|qw8;a%7;aBGm_MH9zdLW1)S<hz3>3}F&=^6p+qrm~
z;EFD2u2Xb=I+Or&$+E4UPk4vV`?!4USj8)e*iIFBF$Jq5n+#vGtxj{q7W{mxu#3Y{
zYBn$&wQ)kp$7hR!V!UZQ`Vb-Mh8w3@(^8t-Dj*lU7bIK(FaW=N(?&65MH}O5F2e{8
ziS9*vo0<9<3doslR<GA085CI0nGdD|?g#25zSd!1P>iG|AWuC}!D6<h)=btSpA75G
z{zdek7SRg#{~H*HL~Ta&S{oX05jk!p7<+T3u)o5!pzDcaqWC#HRu|+?3%9YASyb;^
zg?_~(27<HJBJsY`6)~JD6H!?!!65KP=jAml(NfgF3&0;}`EYO?EpJUnJI3ozHBqJf
z8-f8x3ktKWvMIT!w5FL|!A7-~R`h&Ok4ruc%AdEPA5%qpOcSM(D4-WBc|2NsZP7c;
z5#nK137FBDavgpDVHx*jl$4lf#qaDfd`>>raDbUfnRvW6e+K+{->zj~;aQT)RzJ3r
z%EHtzVLUGlY+-oDB8%E7)UgW0yph2Bre5%^_fac<v(NM4-=NKi*IZOYFrbM8s+*b>
z^$QJ)(bzVZr^HIEXE{7&arb0Pfd-BG4$)11W@^&Fl&a9Iucy;_-~|Z*jlHPQ`iZ=9
zgJOQAGtZh{#Jx(^7to0Z6RkXVb<J8~)7aM<v%?nhfv4?9+umpIqB6%+1uumQPSmh9
z14BwSv+nWbRqS!m!1&#n8?%4%#B9ia1X~VP*bp5CR2jO7sK_Dpt*h<LWq|lEY5)uV
zcwxRcosJ3mYtj@<tsG7=gzB(?2%6T6?bSWdvjh}`c%J%@*2*t-SI&>a6km+~THkYB
z*<Clcp?)m%lptn{SmQp2&l=N*uIx)54*GU-nN#1<VgOBmAyoTMdaJIvZb(249jc05
zisVYU5yYIav9`JmFV}nKeiF^?4%b5@CPxD(6Cqz^jF@4_uir0%G&H{f-VVf;A;1kL
zxvFCcd&ZmaKqm@=pyGY2Oa$)ZEUu9MI%?qm2jiZe2j?qIO@?8+602*F8DN*|2S>aj
zhCTc`&>ZCVT=!@<WU|8knUEK>6s%inux_`*72ouZ?*lb|q=JGFB^9=~_S^zfXHOTU
zjoL^#kMsT+-Lt=Qqkk$jJOn^6fWbe66W@w2SS4x|s!f}Wm~EB^vG1eL-b4-f%sLOo
zi-5MHlS`C?ksUwj=$M6hp``{`6ILCt@!xxsuy9<w#dv=t6sUqa6AsZuW--cyttc@b
z*p}zV`qJ_p1nUAx-MkuYc}iMyQ9Vz=k73LNHIdg7sEq<Typ@q@5v_u2ReNg$a_6{@
zy8a9KpRXQ=iT)H#mdQ09FDC~L_)FJa_HvdSmEzWc2%u&4{>h_=gcY;wuG`aHb<4eG
ze+ti8IPy_d5i-(4w7}M)|M-#kM~18wy|)CFhrvWUAMV9fstjGC-F+)>MlI9N(N*?Q
z726z^tq>_M;>junVO;|e2bJeXQRgo`%t$-VEmtS+gSdHYFr<Pmks_)u6O#cyT;46>
z`9Y$tkq>L>w24franm76nA<a;f4-_XS!D4B1}(Ad2(sjS5%p~yfWfpL338tEx1r<{
zFKx9E8ze~v#^q0<>Gw!#Z~f`q&RlF>1sFo&qcMGtYFr8ROz+l@T;DE9zvkg?Y?6oH
z0dA?=T-A<5BfM@x#_x2pJtb)ed!NkG@TbZ6mt+nz>`Q5b<tLmi`u)<<U21?_IE6CB
zTciYX2r1exAkng9e11r+`dvs%lM$v*!io4_C(|tY|8~r8$+|?{g-FtPr5pGseM$W3
zwC)F;g>VgFp-vGl5?qVxYnsoL70d$+B=n|kyF|5(vgo*%;1wL%LwIK2<iTL8lRL^&
zYu2%RQzY-r=tx$Sm>yV~%Z<6-$AmUf>;R@oX~WdiT~XUBxAtuMWW?cwkK=IHJA_4Y
zI1w+>s?pH9DTVV&HQr>wogTVBW!P}&{o@f*QCP}ym%GJA>g|hR?Z$F0X~B%<@n^CK
z^oP1JZcFzkP5PF3;WSA>!u$c`mYRnJFGmk}6M?Qvcy0PT6vyER@UD93(isuxSvYtL
z<MVjvNifT)`8Eb8cCn{7e?~jttlauCzcId3nSChFOC0DO5Y}SGPbu&{j<;v7abJ9w
ze;E5jp;ha0h0`{pUck+a9mtR>X18Fm<dJK|Qd6S1l)|-(%6`dJ{_=AyQyI+~sXd+f
z&uAsEruJO{1TXYL(0C{X9l&Zb(rsq)aY%W4k14GAnesU;%s`#S_1=La*Th5qD@XIv
zLwogfCA-Y!eF-!bTr3!pG1$CV4%zv|)g^P%eRA&IS0j9G2=1Q23ZL#nK%uBw*ZP7i
zX#6{n@<$AtpP0qFO;z1HMn0(AuB<F!cD{oXO77{?f*zU`>G2O{4N@uKucmfY>0$rX
zcD&Q=Lv@3M-tl>yV{Az6wzM_ix$_~Xxb!qt;5a_CO%*Wk%ukEX<y-!W2yI?Hy89v}
zLcy9j!FLV+EsUTZjFc#q>wtSv4!xO>G6$o1kJ~#M!d+aY%Yf9;_P47*OAZe6`(od=
zBQ3N)ub1s(=%!`DQ#e*^vczU~mwCC9Y;dQgZ{VTB+xv{p{r46;{R$<_Ks|3&Ys|&J
z9e;u1EdLH<8aa#z(Kb_(+d%JhmJS8?-nKbgiFsDglbgqP`iAYDCvix_$2qb+Du$69
zavoAo4$THMN-Q_rr$KuX3zc3>fzKKttwE-hktHCe1`wJs<HuZGA%TuBqu~S0ZF_>e
z-I1q~^{CEzr}#iRp}ybJeOnMu<8aGV;2vD#2m`7+?57h-fe3XXH$!Xs#V46f06?T!
zgq;I1HbSR^Wd01@mG#aqD2y&7SOZ`+W0dCXF%=c`E6;D(QonHtn!b}j^E3`_E!gTt
z!_`~>w5b}NzGJx#6uEiwle@^r9Yq5{t1~xA+g#SaLs5<czS_;e?fNZ(`;*$2FzTWV
zpXxk`+|*D;Dc3!iV#8abb$lC6$1f&1*a2QvB1|t;qZf2};%4(i=fr4E|1~&L8F~I=
z6_k?yV-?6y*3{3Uqz6xxJne<%&|TF-#wxrcK~OBR+1q3scN_X0o;3mpcErnrTdNC<
z{I`)0=PZwPn+_rs_DEiSbVsALFFI=P38L=gRtsBd?w5vRTG+2xQ$-d+#Oi>;M#j8#
z!yx44*~aHCx2xnTtoSIg5O4|io5wY+9{~Riwrk3y`7+&s;hf^$PwpygsRjI6HhT`0
zl2n{1M*Qcu3@30f)6WxvB0hb%3D1!$C16EFF(n6BTS;f=I*o*NaS8K_gUKsXvWn&A
zeGZU%1%oHn{@B`GJuSbVB=O=+Vx3FW8a#ZK-Xkr?q55#;4Iq|i!Q4D?I{IKM@ipZ7
zD=DsT_$n-bXnUWi(RJZ(WA7}6N5CVWy-mQefeVMV`a)UjjNh;Kwa!7d-L1N?)uZ9R
zAjI7IHxpW}cw8NU(Rke;qyFMO&PDIa_EN*4B`?wf0#A`W3MlSKNUBgDDWX}?=;3{s
zkb)h|TL`E<zW=L(;)4Db@)8*6KWp$(fgiYz<Qb&rV-N4{w@2B!>7C?4n<{d60wZzb
zE@Ot`wLH3aPL|0kxz?FKrl<c$M_18KV45%MjVun~a*y@hPtF?e2$G480%?+4{EN1s
zT(tFZ$yI~*zufMgf;a{I4)fb}ALr5jsQx4uHXjq?t{T!rL1GI8lqnBX*QhP0-aMl0
z1$XY0Twa}vc0q2goXziXDtQLu$0*`%Wslfb9~>3zW!jWus=VHyI-$hnqC2?LDenLJ
zde;C<SQOPzz{rw2b#{49S9j+gWAjSBr*GqBOZSnRdpqKIYmw`Uj)u#QD|2k>LdK}&
zxF6ld-coJ_g!rqL92xD;M^|zr2Uts#^gJ<tG+F~nl;VVer^)ggPXh0D9jWuodlEDM
z342)Y{}GO5TBFHgDyGN;{VBSzsmfb&P=0%D_VAr6I8P(@*{6j|e=>6y?(mY{=6)ZF
zJ57QqFIxs4ZZQl!IAUzIDyu-yNT4?BZc$YO@kb3DdBp*;zA}@n7~`16Qqf)^ZAnXV
znXrP$_Q*{9i4N5But%`Hi$R`Lx}TZ&O!(mYbLC;y`Y*t7pET~m`|EtMR1d2^=pAsv
zz{c`v+vNIjJu^Cb3YPM%-a)kvN-c~@UL^Rw+B-SUx94|UdhgZNSFa1$tL}3luHcKq
z?Ebhgk2H^wVEuk6IU>OF)i`Z&)%sHM)u-S%q6OZ$7JrmwhjjSP;;<n1r(>o~=uncu
z*ug@f!O_L)p7{3!5l6xu@F53+ufKzEccOqY|6^z5|643;yK}8pF22K4sw9kR2o4Cg
zZ%E(70$(m6q|S46R3HJgN2IK&%M{v=pS>51RKI4%m+Z>%tE>L(c}I3UYz{yF=Lf~H
zO^J2E5JFIue3io*@nmcdq6azWa;nG*4RxOLt@p(iliu^pV6%<Zr92>F{Ei9z?iEI`
zKR*T_8IC=E1xt3JtE^pRF;vm0pwAf`N62lWYq19L+g%C<3Z;omtbmMk*!k_{58}yA
zA5x|&k26ALyzDF%e6zjytf^!yIG=iN=xdtI8d*-{W~Uga$gW+f%xkLJo4sKU0vv@D
z+v&+8Id>S<BZijTgGpw4w|*r6Bp;d(5e$HNqK5#p<pnOwPFXno*dt3h;tzu@fF8xU
zsHthLAyCe>+r8?olS~&C74w(12sb?6m*^LD4H*41`SwC~o43r-ib+TW<?~$>@Er8b
zTx2-k6qm_f7Q}TZ@9WBB7up=Yr#Xk^P~HM2<LFxB5>_7L@eISObGm(kY__BV$%q>_
zn&5~sB5)4>_8$572?l>d|8(j%Ij>Nt2-bx1Q*N7VPGOW(=6DG%*FC`ayY4yybqj;g
z8wNqMQT_aQR8o!)2#$2nqX67BmOE-qc^ZAp2$V}DC6y=_RgnFEOd^4%Bx%!9_8RAv
z<^7wl1L6fH$aYuWA)({Fgoa>1>71MB9j@|x`i80e?<gqd({dSZaTrrkX+?WlHCrTF
zI+XZdHT~6p94=MV+TldO;`2r%pOW}~9>YOl(&=T-+Ps9&as*hH@E_4u1|AIGPdY9_
zqoQnQ?#9c}!0@5~MZQZsQv%m8oq4lQp~Mg#&ciy{#<3>6vofHcItQyhLZ;Wfy<w1V
zMUnFw;l~Th_vBP}kvH*kwZ>B^bahF@L7G~1j2mR#{YK1>G5dIQ>mss-M%nHlKFKHc
zDBT?msh0|A0!7F$9c+yg2C%$!6UfJQgsR29OgkYj@LcfHnw)y>G*EA|CrUVI688eV
z&97AI|9!de$bXapNg7HYHJ?QT)Q#aiR$zEtfnhnZq@kAi&Nt|5%glbctyL3UA#tc;
z|8HW)m#!fy`u!Z+qX;nDF@tQc79V(t+w)N&iaN0;+|-Q&$Y#vHG+-2qyClVQ2QnHa
z8|?ACggu%c)iwr;Ix?ckS{*d$Og5Q1{iA?7Eb{AF1%Mo&E|e6~Rw4S6lb7-ArGi(U
zG5Fh4o&k89K`)}Djw6g%^Dblha@UW)JVZHJ-#R|T-F=``-#zQOy_^aFU@Gsn@gwTl
zUbq4Uyb#!{N22cuB75~lijy|aRw)umn6Q{^dvsQ=6$%0p(x~tZ!(NxbnDtzqHzN^>
zz$XJKMIQ=~7NOl|@4A-44kZrzd)qklJV({5PJv_nC!vm{FQzSL{M`xbkbAQ%vq$Y-
zorKEc@}Xn>e`pjr+C1W&c%E{^bFxo}8^!R`xajagU}C!ZD)lIszWKs=jT*r9)UlC;
zC}TFiiQr#WXY>|{V_6!xB1O6=$j~v~gSae!Rr1IaA=OQo8ZWfmP2bFD{V0h!G<<da
zE}J^>?q!rD$-38p0uDig^Gyd33@Y)9e_%_w1*Ipr#sqK8p2xsEb==lmo(#xDJ6==0
z1NA*bIuzV(#J%`NQQMtHIU+OB#!E+LRfT@i-QJhHy|P=9|L-Mu)IWiLNWWV;dLL*>
zZg>6s#lx~q6UaVS9|_3QJ8vvSt_m5Rv<w@K-|={f8KN<1U-PkZy1v3w%^QC)TSPab
z@PNFttZjv7xIirG`Qw(cf;^l0#$&#*+?cA#=!b$NeJjEC^xU-@3#?*^wlOou4Cd8F
zou(fzk>r4%R2n2Bri$^X^vd!6Z%w-&&(a9kgIgq{B`epW!uMUAN`zoCyUyYXs1g6Y
zSY%T!xpeJ6h?m%tELhx>uaK)@z(r=iy|`?S7u(k4)OO-ZuiVi>NE-c);NVD{rSRcG
zA|GxBQ+paKcvrb4a6n_VFJ$8%;fWy*ff|}YZOq254H3wO8ri{!ua8^tQwJ=XTF8Gn
zgkbZbu#+ugBn54yze01ZY9V;iy*tNc%`EE=@ysrk1gP=2#}9OzuJ34`Ya6?p4vgud
zr4QU#Pz>ULwxW=TGOQd%DVm0wWjY}SXjXC6aHEM|3U>=V_N|R0gRh}`c*iC*5}?NI
z{?qTEbm>a)Ai0d9T3&chdTgp*!TShI;9W<~Ur!ff9Dq4_3uAQc5<q!!=}SG|Rc@(B
z;eA-)l^oWSfSB`*-_(f?$)&nw1Bc`I#(~<5W{cBfaZX(MU#w=o)$sgJ|KS3hJXK_8
z|3w6fEFG2R37dT46NtA4X^WG1>jC2l3jr3#=ulgjKnuARMW^9Rpa(GC#@3Uhg|<K;
z<om-Kj~=HqIH6ZuZbQB3NL)CnrP`PI9z$(&58K<wnxhD1{(HI5+F-wY8yaNNDm=+i
zX=Ob?7r($UGn)3~HOo3-%?2v`qBp}rG7DAp*USkV`n5SoK-Buenhuvh$<F#>fv^wo
zAR|4g%k;3`yY)VqH45>}0XnU&EM@Rh$|O!S$(zgj4OOu%6=%uiG}7ex#roaF%f1tb
zsaL-wMQgr`;D}Z%#x@}>X!!gjgjE^K=zrHX@V!A*XF#y{sDv-#gr!RI@$;IPp^t2P
zmHE__Kohpc8`<3w{|src{|iB6nQXa7xe+99ZB0fuCygx$)!&!gxDZ0W!Phhm6rtLa
zPkwRF@p(P5iLQSUleABF<$bBruo1`r{Uvh^&mYSOz-7RoJ)NV2^R8!LK!1#6Bpj+G
zcv|0f!BQt7i|8a^tWST_qZj}YZ7=Tj*T_bJ`YOj^2Ssq+HLh36sx6Ma0n<){i7=Sw
z2Io`vX(vAfhs)z8b^!^UEB$oZy-H=vh1693P@2YyUbkZ!9(YGmUlE|ldQ;Df*f?=B
z+kVsbNS~Sfh#F?%StjIXOH)6&G)K<uZ5hU284a&1^~p@5>!y?XTL?tknE!saE*|d`
zD&T)svUm__E*qsgNTBtZ#Yk`Zt*`fOo|fBd7}ZSLmMl8KCe_D~N`3yHEb^1GYknv~
z%Fw*2!bKfP!V4$3WjSjGMOf(dTj4AJivaRP8Ekv}vbD5>32h7MW0P|*khp^EK?Df#
z00Ax~4a62h&=S>qZ(mAxwg*yjnn>FjgTh;{ST){D#C{82H%?O&pp^R0b!hTQrSeA=
z^zNJP$xX|qIL)HW_JULH-DzXHjxVdxe2im*a<F#<YR7~}KVK5_@^KSi=YU9!etgqc
z{#|c%vp9}leG6x=>A1Y>2Lkms2{a?3`}x>V#^2>aF|1Ul4TPoOlu8@w+!h=WYpA10
z9P0GcWTmPQzK$|4`iNueG)#~--<Sfg8bFj&M!&MfVY$3ejIGw?6D8RgZ-q1<otXB3
z9fb%9wiK>dy~m1zLHD7>4G9YHE19NJzV)V{J>Y>$#2#EIF%#gd<mchSpcTX{->_>V
zqX7B`0AeH&<6epd6D77AlH>X6Vn#+33=9kpsvoLxqiEH0{)RtdCF`W4)4Zne1vJoC
zW?6EUq)C5|{j5RoyRWy58zb(UH=4?X$b+$}gN~E9KDF0)2J6^l2OXf2aM7UL^jLK0
zqSLg}z96LCP4r4nASkE3H|nJZLwun(@0M(qSRduaHAlnf($}M&=4|!L^F4KD-mEaT
zEZoZokWkV60m&nlD{ArLI@jH{M7+A%kp?->ckaXoG*cJZqBnQXC)~!{F=7-X_lO`i
zPWAb2bcm<x4j77xlRml>Ci<zEu)ttZiitVYb)gr<jzqx!fgePT4oOSv^b7qaj)S!~
zi?}b;AL}8){%oT?^UTGa`_Cj^`M6{`cZC*Vjltig#ao(J)B&FQ42xz67TDd0PeiUn
zJ;5And1m6#;C&Kq?{s$#o4;B(<wS}Wc8;<JaZ%pLYqJS`i_PVjqq}N$xwwyc=I3!v
zxZ)5FoyH&AjsDhFs8vqL>G5^$&wtH#@y?bAulcwV`=~Z-fa^u)EiD$d9}KXt4HxV3
zxjc7)R}7Kyz`8>KBl&=(c`)xGz)XX|x1~nWcI)(w4HE`(M`h&=_*u>BGXlx4b&+Hp
zl7T^T7R-*+vKHPd?2Sp%Y8)gc!-pm;_=j!r>V$lSke@mMj<wx^%dpPCqgV|SG#=RA
z6f&68!w=rWVmQ$78_4REY~8Nh<tkaaV(l^S!KeCAQYPqKNsON{YqbBhp1<GVg^|Q{
zgx;{OY~h6stwUwIxY)!Iobi<zPVjwIK){UD3s+hh<;z2aOcRUI?dfO(FRn%GRrUH?
zN~q~fmyNeG`{xFg70;>7Gq^_~x!CXL&({<bNZ)?_dV$aK8~qiD0+z<2gqT>Ga)|*j
z;$q$Zbv{{VwwxzKK^ZX|{)Z@U<D~eBK#MkLd>ZK-=y~&>-!3h{Fxy}!vbDd-8>AI_
zqsynDd&e&%=yOz3KtK)pIU3Z4Q{gX@JmHzHVl4KRc)#0vT~9x{)(_ZAjAs&=&1W#v
zhQ2r$;2VE9Qkt361-Mxy?}I%)XhG^m9YAJ3AjY;L^9pap{cxy|9C%5v$4<G;Q2kp%
z=%b1s{CwOlH{lj{IbTA@xo`p1YbIV)ula9zac~9-f^_qWSbDALln@(txAT5{pFr;0
zcf4h{B0%?rS)8r1rB*Euq){!mx;{0wv9YnTv-=|ygwJd`9_X+)JydOPnj@d<ljDDC
zUmz}&_A)_I@tFUpG4~Mmpf`pIC7{OB({p#G0$pl=@jU<E2!l#ETNEm8H&`lnKKwj9
zaJIo1mFc@q>Cr>D;TrV@>2!bQ+tu9PWz>tu<;qb48dp?Awb1)HVZVLXxCGg0MK6O;
z*2BZ%O4sir{1w~$Zj5~6ZvwV_F8BIpC`9v~?cxDmA!1i2m14*i#F@?<1(sM^85yq&
z%r+#z%}Wn@dU}VWvQkCXD|<UHTHD((2nc%W-EX29bP*t!0314l&R=gFxg_4~#~{-k
z-rrp;cvnQxsFJn)8<NC5I%>jV5`^$c=*i~u3hY@KJ1ctl;B)~uUOlXDb$Cw7{9?ju
zd^-@9PgsOTOQZs;;}wO)+N#aUnYs;pFK(5!c*ZGXJr3itY*ppnQ$_SdavIK<*;6yK
zIYO&rTj*VuEZlca*2dZD+Y-H#5}Ri;?y-gHnIeATuemrd#ayXCa>|~sMk}jkX&>)9
zbwKgf4WnFSs2*CG9xsdEBi_d>w?KQZ5iH%GV9qg5ySDj8I3(fWPIeai<IfxlKE9;I
zMXmlMKBz!8+vfJ}PD15Cl>3shw$b%yxvhBy6(V5$yLtGZo<*d}MGmfJSu!JIF85_7
zx_t@1vz)xxK-X=%_^OaY;KM%e8md<<ARL(y<IgQL?m=(QeP4wO&ssa)h^pYMwCBry
z{nVk@r8OM6`&P=Y1|RXcAf>*dbXSIL)QZ_CJJQI*PdM!LW^|Eg<E79U!*^c?LC92F
zVf*gY>Nfl34EA~7)E5n*ej*{Y_1@}tS>!aSQ;WfNE+urV`RAnVG`r@mWN?Yp!bjW#
zFj2+yCB=<RHD)xT`)n2h8fsil8XS)I8E5JBX>kJ+$vIb#Zkt&;kL#_>=H>jD7`z3y
z6I3{a=Z!Nn1bGU%rODm>WVE!wP#C#XvBRDUH4V*y_4vnGb+FOT&dx}w#g#_|(z3r5
zy#3EcumNhq-Sv55TEeL^SqXLmF?S}}K8gE--OqBAJ@XKTkr!>SccnblMGjT}96yMZ
zMLfU(6YhQ@u~b64c~Oz$+nvujHiY0hD>K;c-nCv>{8E=!RHillkVG+IGrFeU7C}ju
zWS5-(ZivfAD&fO2Sb`v#<c4tg8=20zu-@?xJ?u62NM42wo(aBNdvwF8VOKX}BoTt0
za+I5^uzT@uI`Ij6UyGrKf}%cr%lPMy*axe#8EM=sJNCD+V-&0oXXEztC5*z3%vg-b
zjroCeH|QDsiJr;((AiM5V)u2gykuT}5XW%`+GGq;NL3WiV9<ssc!i3B30z^tE|J7r
zcX&ROCTtleo&}_iva+`E@ko3Y6MCz~#+roG7Z+;q%xl$Sm{Q!*PF(NO;!8LmbM_lz
z7@&{IeONRX|8`^;o@%71s5JOWNKw(>R|8x0W&ecC2k2A+>AQ1K1YUz;$<mpu6(?NY
zcp$NtMqOq{+K|!K_y+tS!BWVMK8nug4;pYGr%+L|4OIcvP5<#tp|PlSMHCmc8FB$4
zKZ>w^6C+oi%;w%QGNx&BywKN)$W7U$+vC<KGP<k?;;>#iDYxubjs%lUwL*v55}UiJ
ztra7bH<0NTlBI&Lx5T0}Tk&s3rOR2i?&l*qMl^j!gB1~sa2Ewx#R<>rUV11JFOW%<
zQ>g~teC6MnX1}g%JhEdC)hK@*%2CMj5=B*P`_xFG=o#{x)=z(ulfNAvTEVE48cF9{
z`jc=uJh!#A?d|LH^Ow12P2__Tk5)P`U%dDh6od-psaCA}l6X+4sdm4bnwp|kDg7zn
zg|M`?<{c4%WxqYDJ(?v=$YImE%Fe;TL9JdL{^pa<3-G&O;_UqV{OR^s010$kM85$A
z{QOD5odLpG!6XqFp(Cgi8TC3}z{0^56&5x&HF;)8#4*~eei{F(ESqyeMycS9<0Kq}
z56L^Zj$}w+5&at~wp_mg3}w@(Dqc-lrKGs~hzIbVvd@?_-D#F`s|V*$86&)7tNHGc
z*VixZO?|yIZegehT+{nnJ;R3)m2jU$6(ju}33LeImOpwHDIoe8wnymUhk)fb=_mBG
zFLuSE%eF1y7skDUY5#O&6t0#jZ`!v(bUE_BGFJ{J41UM_%6N61IS$u=q4I=1x@m@G
zQ6D<x<PlrA^c&6kqr<j!YM_~J-g>O0Uv6jI7xe8XVNMxRszIi2=2<LA_)$k}29Ucg
zo{ZR32m1hwrnhI)mzvPJLYvA8o)YgD%N`jD4+UoW+S2uj7yojWuViM67DP|)xUaT7
zp4a2J;4bXs#15VTA*bC>2Hp0>y}iB|CPUG{DsVo`&dxgRS8XU2>A)8i6|JtX+blK1
z8w?~$SXt5EU7x)p-;gwqzBfGE9;>XbJ_gC=XhIWuNlYy1;lT^Qtl4a3DBFS?rl#h?
z406^3Z$@`76B(J0$rzM#bh#380)7~b@sdA2IT4YP+PFMg=}q8@^0>POTP_4+ASR3O
z@mM-J(H+Z?hY2L)`2FL@5C3^^z=7=M-Z3_!SalT5#k~hdiODvaqh<D*S>`&Qvsx|)
z-fWLZ-PwIvO`mLEQQNAs^=-k+OD=qDL2YC6mS`!QWEn{DQf!!U>UM2mHDjZ3MT0*1
z)y9-K>cRN8>GIDBwcSH}jyPPmod+|19GFkecEeNi##;jfyoaSKk`OtJmCX{0|4uHQ
zCrWakZ<TZ-q9my^*L9jc#n$k}Vo>5#8ac>a>VAIe`=zD6X>?}{cKKw{>*QCo<+M)>
z{9zY7Qco(R67YIDa8sko59}N`$vlOTKO&eg=X*24pFe;0ZcSh@#eOFoOnrQRdTJNT
zYAz}+E*`48K9D4}Q19N=-%snHX*&KlMYX~L_qh+(JfH2l`1QqsrcQh@rDY;;vI-MB
z|CpSd6z!6d3uJ7F4fgZ<aNGbgBAHa8xQPknSPmq=(m?{P0>txKSQw{9-CBz9A4<NF
zUyUf)U8ok1@#Yv_I0Z#l6|BHKG!rZ;97$5H+(1o&0<Aq_cmEViH~1=-%S+haYK2>~
zj-Q0-muF&60sE-ETU(IJ9a_v^Q;pvYo2V?@ds-Po#q?LKM-5ds2d>a{T?P4g-wh(R
z<G9+rGxoAU4a~D}0UChNrB@i9A={4^j=aT_DG~@}X)RgvqI?U_Hnc4&hr=o;6pNR<
zY!5r^d;j)2+q9dqBZjA;@7)2w_z;ehHU1W<xmauT!z6A&X<jAfyFzkh<8+KzA5&Di
z#9QlE_$9t+mPJ5q${66i`FrT_75e((py7cxB_(Bz@~6k$*@O`@zQ@%Xj*P4<1MYKh
zC*n&C44FxUkC`J3ZkLB#79?0$ZDMqFPrs$5y+gsKe$B~QZRKDES}}w}h||TQXkiH1
zt-gVO%wunCkLA`LwG((h5VKpoCYOwVO-_#J)x<9#@I*vJ#B8BHAv4LN8XE0Te(Bua
z-fnZgs|<tBYWC&J7bHA9Jn%T1aVPTCS1uNyOzAYC^nU5bC%AlgaY1K|%+7CNL2Frm
zmC<D#)c#h3`9HTp=DE3gYHd5-Vm9@(Ja)apU{g%Ix(FZ{exMgdxkw~(Xoape1bFwE
z8k}Nk<%<e=iDMy73Z`K7-tg~AZoGsN-b{E}0HX2D-x$CPoT<$2oT4NsLunVYuj%yP
z)D|NZdnyF$k%+EMH=#VWudZ`)*I!V!87>ZQ8-@MbQ%)kr^8EDs)ysHglJCL^%P7O@
z?Wm1i0}*Fo*s_Jp2nrnnw!-Nb1wh;ld=Z8CCfkY@69xi|<ay#5=YE-ccjO9OV{iBO
z6qI?LyFY@2s?|6dn$p)o6%OC|39JSw+m4a+-*D@k!01btc`nrd@K%3j4#MPP^dHNG
z`amMCNbn$u6C^vrKgdQi>LbHwXlQU>Z>Bx|wcDG<vb3}eTOsCh*o{oA2W<s_y~KmV
z1sYBjM3GG7?gHuIXJBCKz`=M!878^p)5fkYKPtui#Z9N3x4EDn;}QFVTs54lNXKe9
z-%Fn4*1Xu{Wr=rje{+B{pUCUZxzZj~v*r;b<W*9_q+V_Jfx0<TNO1M)p!`LyLf%_?
z`onVE`}_OLn;X!Wd*SeA(N}rwk{62qcZk950{={RP`Tt9%4k2{x52ohoS$esK8GVY
z^e#MSoY?qO>FO=BR+pUOTiVsHF(eHaH&>em*&5b6);4nGtZ`nOT^O$kscsy@R~NZ!
z*Mho4ziCQcvHN5`*TF_ZEcZ|%Te)P$)FPUZJpN?TbKB;KbNmGj1#DS}E6OjS<{b|&
zVXx^9pL<q74V!Uq+3DS}v(LjF@7T<!XE^NIr$CZ|=2fx77I`Hslp<#*Hsr+^XI-c3
z%!%|$(6WWKTXRsqvN@Dyci>^V6Biwgi%qN6IhgY4=;$a#w<Gw6w|4}t<>ld$Anl}q
zfk8W6!>!EaL0=r(f=vS?nW?+f><!~~(#KYTCGjAUr*LxAK>}w5UR3o@cXoEBRxW<}
zjZ7kz-eGqV<AgTIzk6h006j=3CfN$SC32s%Ts-5z;#`HLrr`SH(S>x>=pRszY$88p
zXg2>U%Z!A?y$rTEM=6*#G0^Qo*eB#Ja^wxyvj#Wd7{D!sRG8w`>axa?v4BlqF~*~&
zf*xO_E)huq|B!kCmX?hqZFI-av9{)LgQdIbFfs%yIy+!?AYsy#b7&(#`!t^YM=s((
z=nF8w$qgUhmd0T}_>vX1HMgFR@~mada^jUop9#I)lBynQhAr4D82N3#6XN~Sy;$<c
zX0zwZiP50>wOcDN-Su-(EQ@J72*S?Z-c_)a0P_6caXl|IpDAztzM9D6nx$MqM!;&8
zY&ei8*bdGKCBK{M56sM^g?&#b1lQu+O*CEiScF1|-NhEB<0>pYKB*T>;LJxwNB;~8
zI>vuIrw%tjgnp85y4$CslmZtUcs$%WCPL2Zcs|eZnoEsG9`n1=Ts4b;fIwwsrJT`x
zs#&+%qWM&@{{Js5KO=q+xEJkvLiuT(<3mr=^KYf$IW<4?VyZ|bX3qZ@&!jHM$+N&a
z!*ALg_h`<-Tgefp?XCuL%_k(Jmhv~&$JZ&#fAfl*Z8dxD0%sDm<!ze>rQ@-1v85Uq
zZiOkT0Gh+RyAT^c3t9=*P>ag!Iru6;>$)^$d3m|r_Nbp^0_V~BjHNC!{xcMm6%cf~
z=etudr&}YwKYl!okKexyiZL0>!Ky(oQ<3=k^@X;!wpzX0RpiZmokzH2CR4Wimc-T7
zgK8-gEp7gxLjtG0crc`$<KK;`1qd*+@$ipq+00_ArH|kb@`0_jwf1bKbs#PSEIN9~
zF*$Ab=jwZuH#yW^0#cJxFP75^ekaEM%uB$`2h|*9K3t{~WG778eFBhmK}X<E<OY6l
z-5Yk~I8cPXmIzP*b?Ly~MgMm~Oncji?-$9RWMHt3Fb;(72DiPeT4&+y;i8$`vjOZQ
z|1-dXeqrMIv25uAc>JK+=M<7^qw)^sYM<>cNr8l7b#k0(oom5tJVs>y!5}3EXa;6P
zUA>~x*gBL0Epge@!TP3}*98{@s0iS)OG!{L`$O8Ht8e;fy)45oZEilkr~ay+sfPi%
zALHX4W#b~BPF7B@X!IZ&J`B?!XOs=5GEAfnj7D)+zGuUxFR`mU;V3I@I69%`sf@JM
zjDe4yQHsoY-25N8jX=XCi|M!?K-QO+m-N=lpHE$Q#3HGeA`~?~*bvAr&v!K#R!4R;
z09Lv=RrA%_d0wu<LLDSq#q~ZYPFPRxJqT)j5S+b9d`W(o<c#1gJZ|>b5~2Gm<hcTR
z3GarIIA&v+CBf6HjU>fsR?tPF2uwFU>rmHyo><qH8|ZwWn>(Y#eeu$fG0a_4AM5>3
z8l46BqH+qAQqcg5QkAm1oShkU#XZ$E`7s%jTpGEt`Z3z_*LrWD_j(H#y?O>7P&^W4
zjJN^W(}9g@8+CHq#FO|NY7w#v?hOyd0~JesUEz+zPk2|&u9vK44kT}1?MI}3S7XhK
z?wO})2u*a`%RjU6=4c|irq{VOQz_rEKY(Lytip7iJe-M&=PE)hyWWr4<*6kYG(2?j
z;pXbgTD6z?_}ml+g1U?D;wWf)YfS6%yqMH)xVt)W=3C}^78dWg&6x2~cRcO_`>5*E
z==pB)&*Gp51n=hf$UWDSb@Co-yK%bE&3)yh@6ps>XrEY8wZMr}Ap8;OK8=V<=)X?a
zw9=XpwPBRqSFyCv7M#h4b2eG7vCpTqNH0~P-Q#Q-(scX{IZ@{8yjt6v#djVykVEGT
z+zV<RuD#~fXC$ms>#UAMP^PpR(v8ar*L;Uq&`mPoSbOF2k>Qd+I+I2H87V#?oS-s{
zz>B8>mBzQ1M_8BvMBGk=-A^q)euPB@%1SXwkkBVHGwMe5Fn?#KS(k?j=)uF~eAoEv
z---~2px|Iu=ql_u!~bMq)rJEVpE(&&KkoBMT{kdT@R0!kxsk!+)(QMZ_Q@=AS<C9P
z`2oI%TAj&?`Yx4E<u9KuEv1c_OqJdeDA2l*L=?Y@i6JaVo@c*7(AD4YDW>p^U#0j>
zx3|_<9&uZu%knAFIc#J8QYP@T8xql6O(=X)J!D7KwJ?eOeYdd<R&?~2n*X&NZDxqs
z#%buD-8M_$#6!tb)tel0^vWb)MgX&HkK)-6Gf%%Kz)n|vk0Mp-ek1k_2jxpZP4~7S
z$?*gUALE*r=H{fWQP6qXP?j>dI;J@a@)Alz@Wv~3{EZ=Uj@(J*b@>+go$VxEeRLf}
z1L=+i9Gr~)aNIU9A9bZT5`A@0KYB9GvyuF*Jb2L<9-HyQk>i`B;>)-FgmX=;Pc*o1
zrZ)ne?(<A;EatMoFeAf7Bi6t}M()nEskOVak=(0%iH5_T3QHLoxz|C278d9dX6cNu
z-X@jOf!lJ#9p~P*W}58UpU4l7RZofv|3Yz?pVu<i3-9{`-+%t{V6~)2tDfc(nfyy2
z1A}%o%$n3YLG$n0q-6~r!-e2>g{7ML0%{Fw7)EVJvG&(KA((}fTYQ%7@!H|BplX(t
z-s?{?HN-;i4SOy|3Ue=$WkF=Tu%M)>VfmJtm6g@)K&9~aHAq_BXQ82?SUu*`rBIb_
zLP3FL+c6MSpG(GHU(Uw%o;Ez((LLgG5#i%l%Kq)>m>354YdeF!Sa^QGPyt=fjc|V^
zkBRqE;KlaSAfY~Vl3>G!@r6(-YGam78Wq%N@ASDj_!;?hc(rM!q|Q8BU_GE!)EziJ
zuZSxwX0>U?Pw9W$nAAyZkm@yAFKXgR7)Hv2LWZ|{Od)fq?oG2uEk>K~iQ{P1%j|5C
zBezo(=r`udpU<OD`H<}4ok=Yhe~W5O*&ko%9kvB(bMu{<CgO5^D!8K%3%Jr5N0KC&
zM_{$VT|;c)N%u>;&r5zjPAXf@8x!&1nv`1<D*j*xZ{z*)9S(_BCWL0bR<b3PEFI+;
z5@8oqUGoJq?N0S0!)@fx^(5-Zfw;c1$rAA!yY&A1njbuZm&cZ`x#4|WM7ug;yXBe_
zpTAniE<L!f?9WiiizHo;*cgad5bL1-LmYVYL@}U-fQqG%hbU6h@M4`*Er_cFS~qW*
zFS3w`4t4nuiiP<g2JXMfKlN;Dy&jyZI{uBL=&vR6*14ltiS?Yoryo!1@Q{}C#*(#=
zh+J{c7qyGNvF_MZGhtPRJ=(jfjP%JImDFK7#cz%OHxnS7n|U{P#}~WD9M;k~;eNHr
z5~pf=#!N6-9^3x?HPZLP&CA`A0m{gkGE+tUnk48xW(yk0R8+Ys|NA!(N7D@q*u1~H
zSty$C?)C?q2jD=c5TxVwK)g>O#5^X1x#{VofDP;IO_u?7d^jS-$6!4C2meAAOSa8<
zd-TKc@i7uAY98YgcUMLp(zSi^_8-%V<QTPV!UwT&e6bI5$lqalHu?-etQ6UwlbVUL
z=R^1M58z_N(zLPU3Zp@f_ciONq{o*Q>+bA{F`Bp1RAm7_Tz485sQ#?xrxV`O@n0aZ
zc|X6d7BD1_#uOg*!{v?R4X5+q_z@8_SBIkFyPWFA>uUQme0Fd3xh0}#*F=t!YSS+N
z=5pY9To3h~2iX;>j?s}5ONg0qnOXc5$_9i+0;g$uS^42|n*v@d^V^Zf;0V?|X5W&!
z>4%eky~UdWT!&=z!rfN8^#Lc6AQhT~keDZ0>AK9e?xs!+&vwOq5Awk=Pl-gGJRwJv
zmRBRdAA+rRzXAKV{)w)8u7xTdP2nMaxQP+poA4e~)-5xS)hMD}wJHVtvO5n<q#;Op
zzol7R6volkGbFW&5UUCCz`=0$HyeZOW~JgmN^>B$LLIY|=>EB_Z#HUrW$th%0<SSL
zl#!}z6(M0s*ET;A<D;IF*!V{}>h<k3{BI{lG}es9Fcw1NJa}w1Z$=t*va^oi{A1}s
zg<w`Y7uvv20p(2(`AvutCK^%B>Wt7VMT)@2>)SKj1ZpG@Bk08d<S_u9j?;eovzFHD
z{{DU?6_vMPU4tfNuU@^f+8QS1z21Cxf4x)SzvqKWkWsQQo~K%3w<Y6)M&zHDM+2_T
z%gUl4lZ@}|3L}4lgo4tt@UY+XPSgDi3%Wf*gaM#>wU2E<Z}^W0v(X?7Bqp}iq6+XL
z(5Zm1wL4XU=>4%N5YG{9Hk!#0&2%CkUh8<Z>-J=Tul8cz^(g{^j)g_ALM|z&EPw*y
z(N00baSJ-1mTB%oBbjV#OjMNc!F-+ZL_QIT$K{&@F30w<F{~JIA!FkYLqkKgx5qt>
z@rq0HH3tJa28Ihcme53Hk&_oCQb=k^hXY3aEdljkB0P56ivHF8z1=wi2BUF}U9~%o
zT2~@2wtKanP~<WiVB{v@bAGAEc$r*uA3`_tVKFV^#cpHS<QI>dr3VwXL3|}fM@~OQ
z+AkrY(Bx-Fn+u1O{6h=fUrp;3QNNR7(@LE6E~Wd;c|jHR5Rw4*)XmMX6nhrLdU<_c
z={bK<vhmkrhUar+&>ezA41P7c*oK-Sc#pf;r3ct2?DP;y%u#lA+yJp(b2nk6y6>Ii
z1Wv(k>`V3O_WDFj9u0SK!;|m;zujvSqUgA$4T0*gnhIM6U%FU<PXzhIb*)WF_v3u+
zD(4$q4WUuLA=J9LKTm7(!PoT}^<u+Ml{Zy-Sbr;u;-70l6~?eQH`Lj;c4+%uE=NF9
zMeB%#Ju;{yiA!B$Q2k@t43(GFskQef<w9oZX$?)^Vl5F^8x;7v1w7Kb`d1+H{P3Lw
zN^=a*(7;n~1%(e&Q`6qw-d|mMAf+rVF7kH-LqrE(_9f%lL)ZJLpjZU0=~82g2vKn2
zcqJqdF~&Ui{wM$x&X+G=ev>nG==x0>VQ~Pa>JF!iQPn>H=Kl;Cxw*R=IYwMzC<)Mh
zA}$9#KuO+{lFp4u@$<thc-)w`T>u&352%+afk6d?NkBK6t8v@_f@j`VHj^?G4?`}M
zs0a8*WlfD@JVZrBrBY$hSESnkQ&>?^z*<B*ik5~^Yi(@}>UQ%-_BYh-AdVdK8m_R2
zi29nm5D%i&4;EVH;RT#7jal$LZUk~Pr}SqZtg<Hl2+gQY)nHy9Q^F|FUH=;XowGyz
z%zj}z>E26J8TMUW@+VZ&Er?R@P(!y+x=^>}?QcvSQ}RyyN@6HfJr0V%7_?-|uYK=D
zn~<RW11Cp^^k&06zc@xkxqbeiCEao!`R~<JX^lvF^;mA53pRqm92S<D`zH<-Bhofu
z#f?*HoQ^Dg5dal;aP#Ntw=mC<Fgi0IX-_I*rd#p`c&|+e$oEyzLUJv$J%zWN$W*q8
z8tm|<<a22wj!>OOtkW($pOH#S=<c|UachBE`;y=cMR#|>)Op$e(R9^eK{Zboq@-1l
zMx;vwK^j3sLAtxUyYteB(%q<ZcXu~Phje#$ev9w#dmjCVzI%6fX3m^*W^pDEPv=}v
zWKBGiJ@<;0U=_Au4t#2(n~~d|?aX4_vn7=miGacF85~3*y!Onlhu&vMSKFK!r&59E
zZZwJx98kSAt+r8A&9>PY=!wFdXVG$cDQ#_(aYiLcmnRMge4&OcchrBe9=DKC2e=cI
z%Z<VZ*+Am4a&ypBWYf^l@BzHF_MxKfrV5ad^3*x(wF2OLQ3*CvGIbE7GA4UD4!SG5
z!)ZvUxNpIm6Ex(;#Ke?H<yAN5{^fut|N9r%5HHtW(m}Gy3|Q(Kpmx`71?em-w);KJ
z(Quaa5$Z4PfMLY5rRRVAuFd_!tyB^yvNxLFQH97<<9<i6nQl6BES7H`o%gq2*29o=
z{;lkgOiYM-bURigPN;`&zHo>I48#PlJ=44(o36{|1>>YSE8ly|(AL)T{qUN1uH+dM
z`bgA9P{D@L0JO9Ggx?`uQ)GGj2roIblurz!e}RmFXC>`na~mSuK;<444iWjNEa&Yf
zKAx79h-oq;d!pw%bnREVHmNPdg?17ZaW$_Xb1YNfm-oerhLA1eqX)jjK1NW&1Dytg
z{bcN*#Mju~k4{P=6&~nIrU*VfQWnCL%(pbj5cM|uFI98SqZ<ul?R-Wh0uDB9Z<5_1
z3Y|6HPg!V!R7JI^6XdlZ09>BTEl4p+corCsux_uSrNoS|(v6DPdte&0tYg#2tSqjz
zKEBe15|5^)ZG@|%+k-MC7H8h8I+jHx47393aSsb4a`)Hi2oa*`$m8HIw4p^2wT1jW
z-lfF&a1%JZ1#cU?09h%i=BtwpPU~fOp%9{_<&GciU0so}u>qjr1PcobkA!qGr5pOr
z=};>vDVKx>{9$oi_H+P0e(YR<uGh%O$WvtGrFg@XrL|bi&iVPq8IRW1)~~|C+86tC
z`5FxgZNNaCgd`N?GW+}cP`TN3#GgM>E@#_c0Q4Ju$8&c&%5#6YlCN470)89@39lHi
zy#Zkmv^<!M{=60p!jn!rM$7yz`!@EZm%iu_gr+j%Q7TY<w}GC1_rO2~D5C84XX8Qd
zgav$qw96T9E*7hkkGHo{z0*5DO}OC3=6|+df`xPs{~9Xp<IrPnBcXe`%N*i6x|GCG
ztW4oVB9(<<DNyj%XV7Bgo1%bUO<U;>(L&QzBej#|{DN-{{(NA=6E2Ql^Qk>+KGN?A
ztuA~;Uh2OLnhr(V+`))N`8Mso9{%;vr1!m^x@crbL)G(*!SN?E9IZ!0A&h?oE`)1E
zY2K4eVO+pKj1C?pG1YJzOIfBPv1*8wHr<SMXWflrf;-REw^(#J&R>l`l*w2#`qmP#
zh5r@TwkW2I-kdYsH}zI5pj$ciIvFWB9V^ol)#qXO<@e#s#&NX=+pDHZSgu4Q%mZ&B
zS(3+}e@PyjK2wWYsNfQ;XTvhXGY7_PVzf<ETtg7N<p%MrXKd?&?$E5&paqAo!y3Ah
zvp_BN-pi4v7H0*|+d*NIwXhpLhMKF5ZunU1I(bgd%#7s7QiB|&9iYu(GmZeX2i*Xz
z7H<T^3jjN{fI=4Uy-}ES$<*J!EZVK`RAO;IFW3P5YgBPP?!wF_0q%JK)(vbz!~W>~
zFHB;kk>^<q2WRcc5s;7o<0X_3Kz=wi^~b|ZAP5l>5~``IpIys+1|e-US~Lsh5N+FB
zj!%Q%z4wOO+)k|Z?m3CN$KEHR;GOQi7u(7aQO^X7g`dYLTx>l5^}Cfv#!@X@P~Sy~
zU{aY-Xew1$MD9}t@_bY_yfM@=Wvi%>$GWET+xm{@wG1PrNCw&yb954IciC@Ii{c#5
zQC-A5cIT%O-!+M?3F`60R;DiZ{-QaSDWx4`Xq}itrq)z^BA`0#CdI8rxXt&1y=Ga1
z$%^7ad|<bEWOt&$K6O=X=x6y)c$FE^HS(b6%P<?qUjs$5*;-H+r|Ws46^}%T<t=)(
z8SC$tZMZ$6J((0uc&eDxNq4ixvw1HEkSe-r8tYoLr)#*G?FLWmApK|3kkB*7MWShB
zd~Aa?-8G{CpDdQ(S~hZ{kE(701o-tNk1q+F<t2T)Kz}H*Z|x=WJm`7p*5bl1Jr7F`
zNAoNwvrY>K76MXA(2T+<HIGy2NZ*gEI#s-d_(X*Cn26-?pe)O?z@(Wl^X(Pb<53%W
z_2kq9^58diY{%i*$4aOp^b136R38d&32NdO3Gkiv-Mk5xv1Y@Hl_d#C;H0c<nR#^&
zxLNS4<75-mymOUk$G<mV(ox~su^$V(VA>_<(Yvfb&;I%}7Ru?9ilHm1ED=;#(P^%&
zd4KrxX0>aRSNY<p)qjt{f5{#hYgFy9mylD34ZlQ6#WKDNoEJ|id5Y=w5oFam6GJ5k
ztsxNnR}8X>qE~VZG^u8W(tfk;b4IG83XFHY!yok%r1#U?bk5;;B@3_XhI<jAkoLtu
z;pQT87@GY0Z?X8R55!Uy&yJ6o?PI-P^Ry?aeIR9vXwCnezl#pF%PS`d#FswB9HOEM
zlBkYNn(KY=(2F8iQ1(9_<UNy;i|YdQ3TFw#63duOM$Lkt$DE|L{rpcpF|k0vNX5a|
zeDhO!db=H`C{~Pzw`&{igAK<o(9U0lVCzF`0+lFgL3kaBg=00T0l}&tuUJLcTRtdy
z?C&0&)RdjjEI|G;3E6#81PppP#pP9n!<%!zcb~kywfBzuU?|*Rm)x4~*1u{nRl~hZ
zIPxu1y8wB60mm=H<09v7uS+-onD;bC-;FQ5i_&_j==cTt<OtZe?X43iLX5Zs4(JeB
zHl#{mXQpFER>?#jdIBZCb{J9~(TSw#?-#Hr3@emQgz>)i|5L8{)aHcioqPZ&W5K_l
zNJoR>&wOv_ik)2nFvEYG@uB*9L0wf&QyA9~?%ALG<eX1f`z7B^9l0<Oo$%(RWsuXO
zmq%Ye1@@07iKeUsmG?*FrZ~_i$U~yD;3WpiM^S?rd*;zXTek-)Ff3qTGnQBDk8BYc
z%jgH%pTPxi8XjI?^ccSh?HJ4Q^Ai?l7^jNEBD)GjR7i$VJ%FX=PuI7eR*8-6*`c;f
zjW7~s7W|n@vTk{=wC6oXEF(2}3pqJ~zgdDDMIkcwSK-TBTA#^Sa$%wfp!%6@N5*V<
zeWH73?f@Kf0c9n!1{ye`0qI!do`!JY{a`V$)K7#gaj(TlB=UYG=ZY(E-PGtBTZ(UG
zJfTq_R1B!Py+1dTUO416;rB4XsB2W|s^JS=AjruN6VanVjRq?p{wlWpNZ&Pkq46Eu
zJaMpMU&2FBOmz0;1_VLYc0Ceu9c#!Z$EI=$nUotYh@6uw91k1szy*MO(?o5|D|IXW
z^6@<LJ3adbU-bIvg?zv(0i}zbjdo&X!5{ad;XQWL1Gz68H`#NVQUt{CQ>-xk?C?F+
zv_r1HR=yW#Xz+r5J7mIL&!K+tb9Q^USCSyQlohc%b?|Alcn0ml#H?ND<RA{Nw+z}H
zpJ$z%C>CKkyRpmiX(;R&&4VUfvGbZ^F+E($*}GjQn?8_CfEpvXc0v8-)lcEni6mv{
zsY5zp%HsHwx^&w~iu?*~cOI>NY=dgHY@yC*?5Si}ax^y-S^-*0Onqi99zv6u*QVLv
z8_jO?Sa$~?-;KNG*5ZX6b|@s_G{=M2ueIN8LPGiDr_V`pb&g>Hb|jU>u`i-wzL;=C
zqy8P~{*cnq8Q)}eru074{ql{?k6dVC0NV!Kz&6@_$WyVjXXfA1UsOwZ7UX^4Z9zxh
zNyDK?sqH5zV))5KE_v;#9Y<|viwF^JIY!`Y-E$>C=!CLgzkOSql(xmY-LfrzH7tK{
zw+aA3gX2(&<T^wY7CafrYj<2d#~p%~QNoi(rUcVjBgz-opSXA@1MUxkI>js&-6<Xq
zh5c7B5@z^&sYk0oF92$-S30E?6cp0T^_cNPLPFr-;s4~y(}R%d8!0>I<|gqBe$-^P
z6vFZv_=|=6vL!l={^=%jVW)d~=e>Z=DvsI*x2-W#C5F$mv#4y97^nERJlt)#ZM}qt
z-(l0fmdO>9BGj}dc}Yr)OqYvgtezMPfGONC0Bvy%CjE21?;m6EXIS`}U<*C9(+VJ-
zwwZh_kjHF@n)F%r5UixQPTw}kpL}s|sOv*>kLUB>^o0U}32eG9+S;;LG>e@Ls!OXY
zB7-?%wEUwJp3JZhdY{Bd&%E7^4h}b?xVFyLvGQG4sJBvf4lo=L>shZEp7)kt!n3!x
zcJ`j~Pk2Ta5eB;=VAV0cV9RLz3Kubf+yQVnQEqw_%!A)$Ii<zEYeVj$D*fnN(8fk`
z=b~y4XK}cIBktCY*ekbJeOq7_FHZFFsw3kTQO{p%wYaFoySv(`sB-4l)0d`XMiq8=
zlN%{x72`t3SEg^Ed(~bMxe-~^$h(XaJ8t}uRV#t*7Ts51h0d1Y*%~U~KNe9~3R^sH
z`f$UlJUQ|+p16JvkB9zBT_?*EN+}$fJR-Ve4gsbA&o?s6rzabqgEjtz9$&70@E$>H
z2MiAj+TX~@N?8U>t!J$62uq~Ykv&ep6yr+E22V(#-XG{Vlj`Y7GL+(%XR*&2To64C
zsoe7?$*(@mv0krXPB)oT_M-QU(9FCYytd3AY3_O_m2a>(S|kk(WrYu}i#DawQaTb&
zH%|_n5x$#k$MG;jU`gY^Es53VtMBmLvduHwS8qbT^KWWDKSNy_Rl58=T`+8O%@K#N
zM5{EVG-BJkNSD|%@bWbNT@v~zZkwXDOpwL(r1+X3vKtBJ$Ekkm9kD078b-X?A0#|>
zdW8EuIjs+cWsR|++(w7YwsSkMNy!_bobx9IUd8d`-)=T|W+(&yv|=ReD6$iT3C?&B
z!4WXbQGf8Sh)f<_Pc&qVj;JdjUW|Nw7g0ot!PNA!-Z=#oN^Z`};8WKmlG3*YDJcAz
zX<J5e*3u4#Q&C{l<=A9uj@4@{Zbu|8$xTHcbW=%d3sT6GYA80MZs@ADtZ|$qQjLmo
zcSp4a(wRD*XXZ(`+<ZIvbG%ghdPv5mDiX`>4}AYwwIFe_mWF7YgBQyV#`0Wm>bb7V
z`06u+|6ZAE_v(<U{mw+N#a!($s0X+;s)~z$iYKyx1}`Fg9=DJ~xhW8#g+1#Zunaa+
zse&wsK~}0(Yu7VIisCKRRlfZvAd9NeCzm4Ce8E55V=Y_b?^;IZ;l*SFO>ZOrkeBWz
zi=NvaBHb95A|^?uN41>Go2xRVsYK~Hs)$==Cq{zb$H+z#A=a|%n`S9iS3If@m(m0!
z=hmYM(LPo*3!Mrrsd*25!Fjb0KzwjQ2obzDMh|yB<S#$PE9Z5Hr6%&y)@+j6Yd2&-
za->-8u^6FQkC(P8P-?ns$8OE1p`l6ZL%l;rGtXIYzzpAr9g3og;tH2NBK6X!%xsyZ
zhdE8*MKk1UqyGM$GBlGH&B^e6l=0%6(YNkl%QmzOJ4Om@tUnD1>UI?Bu*;!pWRS^U
z;lpCzi^MTAy@EQwQ`_{gl81l+?DUOoGv3QlYmTexRz77WY6ZGv8DfTFQhbm4@A01o
zW%<MkZLn_}91bB>kh8`J_H9SbHmFk@I+gI6M1yIw1vT<jRWF9?hF&->@?0tjlRKUX
z@6WHABRqst+5bJ+6BM!eVe%~fHS{l?uwsnoX{mEO&~P}I4^B*!gDVD%uFqe-NXy8y
zf}S*o{WkWOFJAydYZTB=V3LxOIBZs*W@KcjRa^Zc<II=I*_^9m7s)_2kidceB|*rt
zg>}t)^rWMP*6gm7x=W!+*xEuQ!#w&6XVTS)E*mr(G8<(y<3QYQdH^vBQoSN^r?K1P
zjtCrW63848K^~OU%@FFOPa6TK!J!RB(qKoCH0u048!)9NOukXN=`)nMD|p8%W0RZ7
z!=PQxt`h8D(d3a?mJ?mUaL_*&B<b^-cKF<e!b~z!r>MU|=aK@kF-RbDf_c-&@agtK
z=5?GN^Q#A^H{xw7qR^ROR#K!0(yC&()Sz_B(4^K3Lbue&cHrP?@|oV*jq_8^>uP3$
zBf>b^RVxlcjlifEyZZ7JgKu$LZ-x)=UP_rFJD1XfgXv)DQz+PF-LQ4u`<muWWJJAf
z(bvd3`OMF7JCPYHAq?$R;xLjP4>lF{ikP>TS>;T_MQ$>!1Q1HGU(1=)OM1@h9f@Rd
zbG2RP^$m8bL%SH0BZ=?6aPdD2R|$;0HlH}%kiDt*SJUo1EYkexanqd8rG)z37L~Kn
zEmV%gIRYApsAEUJfaf!PL8epV!jsu+@N#l<O{AA^Gp**ssbdyJ(YKYRgLki`{>$o@
zG{TI<)wES0!1$}NH6$Y=+Z@S2eYicrv~gauq+umM2W2GC5QKul1<;cMKmGa$OvFF{
z{hgKuJ5{U=2T%ld{W!!?lM1`dKVA<vO%J+&DSN?9<l~oGzwmsF7&L^Imvz82Fe#e$
zdY_9(g*^S81m`yoN9I7@{gJ9IO&VA2)*C7*waqH0NlaS(kuG>3ltwG?R*`A9MmZ?{
z7TLf+us>fzB1cYP>{(TLdH%6utd-^G+cG}@KZnoe%We4~vW@tDe7i35iLT~z?wT_1
zOsVU`u$YJAa}cq)Zsci2@o?7#b29NJD%tZ>sK4I0A5*X?oZnWmp!?PAM?C$4Py_Wp
ztbV~b(4cP6VhCWvSTv^W2+Q#ParKhia&MLxkIV2=un^E6sGJYoMEams{ic-_4YjW-
zM}ka^(`_{H+i_d`p8PS9R!SJ%8KB<0{`4T!*l93PAr}*L|Gl-JXA41<v)tO(B}Rt%
zE$Gx9FS#JY7}&5}555xKIC4MG9fNAW_>*Y=rOu=2p??TVFDW=((|a4OTP=NWn?B+x
zCRmr{%6!<6WiR1|IMQyi;1CZxh@vz;SWH3rRd~ZX%ewGjaHozp@BG!9{%+lspbj-o
zH))p^3DP^F@v<nvECaeFx5iUaq{ZEh!9z>U<|=VdFWC7UF?Ih^xxk8GR%Qqe^BNG<
zAA3P)Rsr~e?G=<SJ-%hNRr8`<FUHp?Y|M7A256LJgjb7xLTeJ7{JxY?=EQcgvNk}1
z_v%0(vpOBGK=7Kqw^Q&U;CpLZQFZlGzZ%YUuI*z~%*N0snqxJM+1cNIlWiFzxw^W(
zVc6-EC5O@UxT5R|B{2lVUFaVeWD?j@ro1y*L`_Ea#52|7TG-ILy}v)m4;A+XHuh&g
z(8yCPGT5CeiAzi@(`tDF@GKam3KRKc;X@EJTtVy%<-0?iyAIa58)4)qKfxS%OG34G
zyu>{-&B-yEGs%>abcmthVXyl+M$2W-70p1~iAYu9p&3vopl~}@#KQTjZ^gh=z}$s*
z@Y_+DF+sj7b9SN*W75y^YN=cyV<Q{fba-=9*Y6VRHdo6%CKr~v8{PSglQvtsq-?k6
zl!hVupY1tr3LPwhd5k~&rYBBCZJVt{w72W)>DN0=;ED>PuCuXPQg~~@_vL7sha=HZ
zh+CB%DFM?&61q@C)Yg;~GecQi)(piLYFgVY;^w>@UBK2~6SWU}08C*C&{5+#T7%Cg
z+IG;Pc2dGkIJFp9>iElfvGLAEmxZ+QFoX?<E$02JXJQwMuhKNaiZt#txN#Vw5jR)O
zoR*)Zx0?wNmadHU2=_+xGHq(@tT#_(jDi&7i4ZgmwT$Zy8{OC35ID1)qzz=hDkq*3
zplgy2kfX@r9{<_tq|e3OhI!OZ@;%qi{&wIF01yP@Rfv+JGH<D5qSQTu%l2#gA1l(=
zPg@%dj~(klhdo;f0|h0w(p_t@V734IOL{zRhc_^wHIeUh2PlhFK~m3rYX^JABXJGf
zhM;z1+ZqB~VU-d2V;kmTTdU536p~=?8M4^P0%*!_Mq6LEOBQQkDJSqCEa{mj;H*r?
z(<&+dUEW<6*E%^LkI}uX3L5)t+%gG|b$q4V)}scB*qsJpsU3@Y7DpD!4nMU<r-<Dk
z3UKLkpK|+{3P`Hmeh3OhwK}%fJ<I}g0jGp?g=@k-T8Q9Ka{4K8PcN<Rh%~%z5;9?~
zNZ8~lAxWk_PkYS}YpWU(Zq4dfs`FC9!vfyUSmF*N_`vZStInE{Kq`LQ7K=OHmUWLF
zb*5}wd}v1U%k#6OBa1D~y>D|+pZHtJ+=S(g?n#v8AKjvZg#dd3M_X$s2L2o}6E^mE
zC%gd$J*NCnk@p8&$bGMQ4wpJS5=Dj{QQ8NsuAy$*4FgyOgR1MB3lF-&aQTd#Gb(}a
ztQk@DQD)(b|JA11K4K}P(^ZxW1%qc!vv=pu8>=iqw@I}mDIvimVhR+m#emsoG)3X-
zFaBxYk>(TeqI}akjWZ}%GXLdztR#oNp*Nq`nmhKYe2Ug^NAFAwj=8IX8#H5Ga&#ou
zDrVvDXa!gALpUb#fXTv*Da6G~B50f=SEry|*!(*|hR8la%->bla+c7Qa86!W<I2c|
z8!M`mHg&a_qNq5*-?6N`DpNal8wq=i^5w?h1pLOJgcoxfb}*!3t2nS4al8DTEprNU
z7N_n%d_yWLj;F=c!Xs~AN<-sSNJy>W-P<?yX9oA@=9U|aKkHI{KYrb2-|)1$O1C`z
zf^x`MK-v#u!8w#q=5+o28vvKpqf7^{A|NA~6(eQv{pKuoXy-g@Zj0HD2_r+?;YRmF
zg~7o^Ib?z!m{nwT*ROvpYx@V*>WcmU$ple38PDs38;r*aIC0qQv(A_H%+xKo9W4ai
zw0<Jk$SHBo$Z?TsbbS3wgr$*|Sc3ZBA+Z16gH`W-eX0#O0x-S-0Ve<+-!W^_?~CEC
zY{;p;oCw+)NML2Lj(R*WX}KwoB)KtZ;d&avfsOKZT8g8<%FPR3l?ZI-=!OM9+bgLk
zIrgd-Z_#-+-(6XIr)~!2N9Zjzab873Qa01w&y#0V-8X@XUJ57L#%$|t2*cbLX$Src
z^Hm%a6^#9_MeP2|*@jIQ-@&Zga;uuBPt5GuCn!tsddoR(YUOKZY8nHf!f-)U#qVJ~
z1KW06sQ=~gLh;}kBB{i4zaBQ8j)<09E~MZvsL0K*0mTI9XDYfqZ08jzywPlMsnFqk
z|9)&t8wET~Xtuw&fR)|R>x)md@9745n(>QRqOEdZ9&Iw=RO8vc@-x8(tF#SrLRPy~
zb0es4*GGI+#i8A1Vc17@(338k?%Jfz%&M4YB{n2!>rE<q?XKuttY0h&Kf044iLh!J
zFZ%i^qL5Z#7vY?h*ynk`9|U(h-8<RCUgdCVQTaSWGTNXJpP-86k9je;k;WU|HjN&@
z*|f$HdS?g*sGl*tX;gVORYv)o{=DVao6Q3YC!v6J0;taGSM<aE<6C~jtaPWPG7<c|
zSgX=dUMqgN3lT3aW-UztpwHvmm7)g_<sQg%;gB%&0M-CzKd4li^;BE0Jl=3;JCj0G
zk>PQ1-M!KD%B6b39S$IcI^FwQwR{7(#)@w+NqA@fw^TGNESz&U!Rtk=(217)3$bJN
z{Iqf5pMLvyg5)eLOf{5GD?S&{=?<h{wG(qNj!PylBKh#pNgfAZ_;m9qIRz6Mu3R}^
zX7M!X;1r_o$WJ%<1yh1-7OP^9!c4*|9C6}I-5Jgk$A?>C^qI~@@=E9=&QFCHenms4
z;ihBc_LM_kZ{|yK<TH{U==N%#xwOdekD<$)6W?^UsjXSk9t~Fly5jer(gRfA;eRTj
z6Ux9C9VrEfBu;Vseg3Dx&kpl88Brj&F@-ByGcCqNjr?)jFj1a`H0?V+Wu-cJxV*#Q
zclr!U3i`;NscLI-0e>`Ype2BLquH3W))%W-BM#^^mw?jnt`6EGS!FYyIRsQ2m_Lo~
zuCf8NPd^b$t73`bf69@O$)tm89vL|f5q|9#AUc|JzdL2-WnznRYUj1inosY{OH!Gx
zqpd5v!hX9w=Kkg>%wJw*2Fi1{f~C>SMfHL1J}sonMg%uUKkqsGe2+CD4W$b<1q^v=
z5`ip?wdD|5BSmO7Oc3!#dGbxSfbyXv4&YDQIaO8-W^R=T=cisrM$NkC@bavo*CM9R
z&}!+p8|}c8bXkF-%lhczHK-0*f0W?R<o9nKZLgR;n4!BlA~c#~;ILkXeMP;sL${GH
z$D4%(WuQFU9_tzM7OzsmQ{trNGJpF(N5hYwF(eu-AYUH)hoR*EZLk^@d>f!|?g$~~
zo~p5>26sqasXFNGKo589Hx~z8GuML<3L+X_b9Xp7KBvtpaiHeEwT}T|Lz^Wv>Xqs}
z4dSw?RNAr4ukv6O(WYr~OK~>Jr}ki|5-vxL2_Hv7Thh<%$U1iAxR!jF&Bf1<fg6fz
z<6L%Gjq6YLIgMBFUre0%2xLUgFqxS#8|%@UbxC-C)Yy+l1{<@gJYqTe(zHA~<iR)-
zRWl#I!PPI|_3N44Ss`2*b6$5T{g?^e@{_a@`1aP5aphyqu%!S)@iwBFcw1*jipRr6
z)ZqGj3;O1z8c8>!&jlmDzsv*ln%qWf|D~S6zlzMpHZm0T`9HArnOOP+a~5-a5C9Qg
zJb%s)!T>NMSZ?-$dsMgp(Qrg!V&B$i7CsQSloIsM{A2HA{;^iGoC(>7ou-R(fvwCC
zjzpIKQs7r$e7KI#EkAretx@hOIQ9*98?PW{PSY|hCJTBWW->bA8;^F+2xGRIq+;v2
z>uZ-h;M#NrFMi|yQ3hFz6X`)uShzbk`~9KNi+cO#B4ej1Y(YQo&F%JAz!m}^C5l(&
zVUp%R2-44Ba<q%|{LdX;&Z>Y~gW1i19dx~r$Kk{!x?v_5C0EEFD@NHL=$8>l^JV~_
z^+BGf9?DHU&42itW;9lTuSOqJI6Q1NH<>L5*Q6QXn8u&+3w-M7ru3B?!GGc9)0~wg
z@As;zsx0Ht%=hLqry9Ee$mD;POC5Q1*|k!9VWsm>&!@frQYjHl+hQaem`u<jNBr_-
zaW3X2E{uiw<2u9i*iXCfWPBR^wZ<+P0!=t0Be)Oh=sWGhME;<U@9fhpE`;pv{xw=$
z--bnSFtNQ-Qc)Bu!2Ja|rH*Tzh(3uLLwfkJafTJMqRT2@0X9nwJ{%u`z|FgcT=b3p
zTAz_t+s5nA*}E!3x5Xhj!1L*CzDO0cuznCqn{nkhYQ1)R?28MBD{9+N*g@9*7H^+v
z(Tl?<D<K9_%E&laB1V~j=g>QS$Xq0z{aADdDsP^X6GAg*h|`dr$RG3fubgDNV3jG+
z$*&oHV=~5L`}=6!a%AkWEURT3$gk!z6o!2MJ5O5mHxK&?kL$CK6(1k;huwdq*woem
zY6xkUjm=Prmd{>*cprOaOOtGiuLNnIUjLXQq-6lD-W0Tkn!8@Y_Fc|AmKIc*Cr2fi
zsTslt6T9~YA97*~1=O^o_XsaczNF%lXXCD_n&Sa_%25ELS1{2G_xL3f9S05A3%$D2
zY9C&Sf4r!DM>am+7KofTPX=fOfZOotXvvHtN!?hr2C7NIz`7oGKh_18>$njg!1g8G
z-lb%^3rqo!hWU47#_=<G09yF-js~pYEXt?d{R>5)3lZ=Y6h9#B)?e?z#TQ>Q$h=jj
zHW*Ttc}o!%OGl|pw9UCPK9Wl1?wvC^hWlBW^cRL`sCWd9377R^6KgT7DJOnxB!3<>
zzSD?_W-bhB46u_$PWM^TzbyQGO^#dhDFj?9yrjWK1QP_4{wXd2;mL6!D)NX7#QfT;
zLmH^A95LdZYf)U0iwLw=GsbS3;M@t{hW}{a=N6LCLG)9;QsSIiT-*{%D;#*bJD%Ga
zILF<s2QW#2QQm?Yk9CxHuE5m`O8m?(_zFDrIZ&=tqW?5jdOCPEC+_I^zx`f-h5Qc;
zh6(FAG+PfaWXb6?G|qXi9hISvZW@Qu>Mj=+PR&pqM>rWSwk=?v_IN+xR9=gZQL#iW
zjn<O#77!9tJ=jojr~JL9`8vefSfrw5Va^}P3uqi!_67Tp5EXk{k@{<oy(gyjeE<E@
z)&^NqWQ_`y0dVz9mFlBrNhFhTa^e>#Oa@^@05#(bityu}+KwuH@$zbmGy8k=R1K%r
zo?bwI|0%g|pGZ`C2UWN&IhLU&`)uOn&ulo>ujrZMG7*|${4Q^qSy*;I5S>YU<lV)N
zSpx$CbKW(U9rNxRHmLadT@jbyWz=K(Wgtg%{P(wbO9hPea-RcD(<4IzSV5TEF+T19
z%>3n~Cu9oy)Gm|w5rgUty2-(1#9j?5&zXPIqjMqQSCHsSBSTPi#FYC608Z;BG7IwH
z`0qa99p1BGij}f6Tq9CX61_HnhLmkA_2x?kD}GUf3}VzmjAqaybbn>-{MDMhiL6LQ
z&mdaH*vA+YLPPG=^^1f1;-U>2hV;kYYwnm(2j`3bE^FTa-=2k%jrH}BN^?~dR7>%~
z;X-xpnF>?tSGJD@$}44+;Fz?g{l(^6hA5n<uZ*>EKCv&A68NE*@4tONfu10#&B}8E
zd(**{)F4(G<(h-0L}1@UpTJyk>K#mUo8KaV;jC}f;0rJni9+7y__KgVa)!Fbre$jy
zYeB>;S*3ez0=T#`T#8{D6&4L(30^lu^<QFRFiI?Kj;Mp6jtc_hXYI4i;IUIv+-}g~
zASFYpt%LC9hvRvqOb>2!ZzXswKo?i$Su83iofY$PLP|Gmb5U{}h4kaw1{G8_s7bh?
z``hoU7HNXP9Cn+}Qc|uZGfDrw4?_8ph(pjAv+km0iUi#crZFsnnXeTRmUJ{e0@KCI
zT#6W9N}MhYgOd2-5t9?PtH^`>*RUEP>ifelTQ?3NOud9EoU@$#(IRkJCt7FdV;d*a
zR~;&asWq+L&XHmc{h1N?sja!O8{zh>=tS5|GlbXGUU?nK=#iq@t^^6_v_v=N?15Q~
zC#S8n;3?))n-3y88@$8gd;YxY_Qu`uN%pclyB9$KCv0!Rry!(XOmMXLnFLfgj|pY9
zu<-iZ+m*l+VlSyJ5MRJBjv~_j|2-2QpTn(IW+2{iTx-Ac*+%hkCDK7M&}tY>?YN`d
z>aB2SKyGwewEFgIPc`iRq3Kh!lXQ-%FnPxNkJ}la8_p}%D;|V<)J0jGs^0Y*BbpN(
zi0wa#OXL@#zN|!m`i<aG@Nuh>SBD%uSO)kLziJD<>Pvl)5ZQOI@y;M+8oAkDed`%Z
z8Fb;JsVWdg4C03JRVyU@GI4SE9_brsGZqNhMe@E;@|Y`@08P|zGa{a3W6tVGW4QMp
zHv-+mdj_<`U+B@9&U>k#{wRd~vzOuZn#DcY^k1#B4Hf3!y2Y3+^khq}KVvpVnNRB)
z(YyU|ns9_e@<x3yXR7*!`1fnP0nmtwi;Mfdrcj^~U+emBJNzWNvq!?Miluhkr~osc
z$ECX$#}SzZapeOeSR>cd=q8e<xN2-jhU*4BYtS0a5Pubp9|$3$9eRc)ddCNi?qJv$
z-_Szt7<uP_iye2*;@qP=QsOoy*EF0Bh_T|fM#hr89iJ;bs!Pvx3v^w?iRr|-s@@@-
z8;qH)jLBTeB^*``1ufbC?)L+r#ta`8$`_MMCPWcarlka&OHSCgt1H9p0b${DrVJAQ
zJ!{Oe3?<1!V}#%e)#1qOL+R2Y=^)glNlDHgOW#9F!~=N^DHi-%iDJjUJiqy@5!kdq
zit63+NXs}oD^QB*Tg%+ncsvTP@V-nV5XCipq3b)89!;3UB;G+yPU@}+?>W-1%d^ND
zE<VY#A34rL|5@Wca6UROOmLU$*K~j4I#5|a37CI@d^W<l#5^n?xdN`CQYN2=_pm~?
zfchi`XhC}jDoC78TrA6VvFYKtnri;&Ju<(`xNuRmwjVRN{Vgd_1*`3pZ07+=)MdHF
z(_vsv{X}o`U4&weSB;{U05jlu2dNy=8Q4TNo&Cc`eqo4+4gK6WRd%a)qom9b>8vp8
z?4%B`a_{|t-4D41=Nm@`#xeOz3e=4>4|}O+NsL~d8Nd0Pv-06WARF{S0B<%dA;G0d
z;CYbVv15(ohcmej8xN1hoH|^v7`{lIJ@Pjh$m58$JM1<W3wq7;_gYjdheLlY#S_jM
zyA*c;r2z%qOe-uvyM=cP`Be<I!TMyNbUh$DMB->bhmj%@V^L|GJoMwHA5Kr$umsPw
zj7GTQv^6C@7BczxO=@jBbHu>uKlGlOc+r13+|zeDd7?)P@*F=e5+3y|5qywKIs+-}
z5xOz_J3pqQH*e}NjH9C#Y2-j#(P`<8nPl6<fY-YpS<B0oi$nhZG9%2Wv0Wk;iNu~Q
zsJhSY=xBKtRrl(uq!A6p$TKmR)qokFZ5>6WwvM?>mYYnIs4wf6I<0J8{N;e-?djyg
z#m%$*_|j5lIlA#X3;`zCNXgQ3fCwCS;A+AYyWcnnh7cKVt3Tc!t#G6uJ{fCgKR?@a
zFCcCA0wz2ZNkMsbtrcZEn4+kmuF#lS`p#beK>Fv?(w!uwn$VxH=IEi*U;W~*0B_*U
z_{`q2j*}4FtoAVL@_DjA0iA$_#c|aKdleuJseYf|6>||2?y!Ye4<Vbajv&EGtV_uu
zvWvV{CqOq{(-Jr^ZreOGE8>ctZRnX8PucKD?b)|vvs%7rF!E@^Xwr-ogq$bKqIx6s
z=rVlmoyoY14Mh<lE(|0_^8J5RE!gY&CajX~xA=iu_-K|dnny_Ciu2dEY*1^G;*)2d
z?=E$IJ5!D0f=C%|cSalZf4E{-yv7?^C3m{Jy?u&_iL?kvm{cD=%#GDOo*ZAUqWsTa
z038MleL+XZYjCZDLL^`)KDR5>a*^|XQQ2J7oAv0ek_;ZPl&P9|6f=~o<k<PXa~3q^
zgy)u0k@}=^iXPc<@fy7BcKSqKt@c@5<5_w<zZtaO$4gc?Nj*Mxlo||oTOZ8vEO+At
z0!kWo^lUDu{4zKJ@nyi2oA=j=uLs9XjeyR(&#TShX5S#%_TGsE1xdU0ngv7V_1GM;
z8v4gK7aFF|?PyD}O-6b)du4C+U;<e5I}3kO4{rQrG8IlbZ~ctQug9ui_|~?p3MfZJ
zMs7`&kf*wz;~zBMyaiCm1`sAty?KKsBqXFaemlCjwx+)w_83N&Qu-u^!f|*Cf<$GW
zIH^D)?wRXywYB-5DfiM+GOTZoN}qgGR_oC{xRCNtxw3Y_cjmF$nau#hQkstLK6nE9
zqzcB?=RVS8Df+%{$-ZAqEF^%}u=^niaI+d0NGIX|Js&XezdTe4$;8Qj<afa^$shoL
z<?bADc%Vw%V^($8Ep7VV?YQoJ%RU6%oVBV*pF~Zfk}_QFsbNzWwqtp0GLhFhve?Is
z%jAte@?a@NW2|m3twlsdk)J-*T5R?bi(~QvaxnRtJwVk}F4lqtXb?FkXS}B;4A}fV
z;Mqfd^c7$>NG5UcS6R%_^|AnEs+NJn&fREdFroA9Q5SvVq)El?*(9X-My)ywG}F=)
z^0sDcSQr~lUyut1RqLwihJ4A8DY0DO1^SY%`3Bb->lH)`3kx7-e`mif2T1zne>mN-
zK}xc7Z~zJ2g+xTW(17mj8>%M1OMZRaYXP3Q)ow3n=&>jg-eiO$<wnp^#f+HnYa2K@
z`+vXtT4FLruyPH(lj*QdSe+Gims1NBN}B!T1-K&smjG><(XgIk`--lK3V$rr$4bNd
z@DJy6VgVf&M0NLutyOcS!1wq}4YJ%R$P2x-a7YKvfDx;JML^K|OCVt1=?++WByl^2
z6gAx^0L=q;M8)g15re^`9`Nx4r@K}^RM0&38iM#9k^}-U9;pJ~?W;ESAHBXRu0!rE
zWay&N@R&T^Z_ek2FCOub5?v&Kr5Qfl-%U-;UXA@G?PAvLL|Y#K&6FD;zm9pM+!{eC
z`Y4(yGaPE%sYkAm`u#8N!57;)M&U4vhbe*C;>EVM`F0iYpD^gqv18_HenWC_0cA^F
zPM5C<GSxjFKDT-2S&<<{J&3Fth*}<F$E>rWNajt1J$v2-E{<Y{-Up@Ye#LyZqo&)d
zLCd7Sc2Uc-tpsO=8{4lL^b)jM1K#9dg3b;AaUb_Os1fpssWSq-k&xKl=cam~Z|joA
zDxNwr?#<RjK0bB5Gi3Yw_irh>XkhyxA|mqb`*$6n{fOqp=d+yWSszRmJy{<BIM36L
zjt=k@Hnz4v3GiGb<B_qAf`S6G`R5m$N<k&rQesq;uR>nIB>|6<)t7~$hI4|O^S!pw
z(MVu10cJwe(9^E~XSnxfQ(@n}Jq5o(H$vBVs+iPt0?H?)q(q1ok+()$T3VW#nK=dm
zcPuO{-Co)IKTKCo=OM%OM{38CwBCx|4^U%#3G|baLUa(k?P;KCcy4PdS9I*o`8Ogp
zhQ4BFk%H2JouZ9ISSq{Fj<8j>SsHfevN_CUgu+z%3TA369&OyDTy4t58(`a!$x_Y(
zccYLgCaBy;_v1BAozAQq1XXGgqGC;J<BAlC<dM3~HwM22?*F%xzFdr?tmvLN7KMUm
zx`x^k?nti7rMBISZ#(C&=cCKG0AU^mh#l(F+`Is_9f0gBiI$BuritF(Pk^m~e_kFv
zP|+wY;1v1zz&)NaAk+d1EvU?9O+*7Yu6#dZ7p0ez*CAzPjqdn?tJuE2v!hy42&gPT
z>bJPG)beQ1@#6<M6Vsn;-FN_5suyu`a2&Q{Nge`=l+iLn$w!QQt1^@;lG=&eZc}Uf
z?Z0(vA&-1MmG?57FtquX8Q!9W5+7agds3Wdcr&>pRKc|n>&9*;IiTg#7o<zqB6rC2
z`0u~{(}90$-YkJ}rT<KmqQj1d5J$sD9sf{T*4XuRHk4@6CW8}GY~88z+c7m-!$=$R
zgWAs=Vva<C5`qYkB_+1)yLcka%ES}Z5EhgByfI6X>@v%unER<(PAMl+66UBy18E;D
zhs3xiU+Fubyo`r}=i`sZ8-!~CT6)cEt%Z)6-pxj}{DRkn%qPeD#@-iTs4)=Co^B2c
z6q%pz&;1ZgTPmoUuO+W>tpi3R5nvt5fS!c=Mg916gDX3XB;VaBanD?$nwpx^N-+E7
ze%--Ttv$VE{gJ<P*Kg6T5MtP=O7q08hYFvdD5HOYQRM~m@4RwQCPm>Z>E0K;@0NLv
zH&fb6p+7Vk;3<a4&7#%INn7)G;yDgd8<P0XRPOgB(GgAJds;t{+(xkPanT0$NDH4)
z3AHYn0W7h$@x7Gc*WTYUF(FqfBKt0?EwKZObvR6|q9Ug*s{Eb>QHRO+FijO+_s~9Q
zw3eC8vIC&Ks0;!q(L`JA*9t$V$zsYG?oxcFa+&LYl@KNat}8c8>37kf{T&06LX3;b
z{>g98$(h*M4SS<#A1^DgSpfe3O;GlX2%lZ3cQEAxAu8$*ktPc&)wJKgO|T_FCmY5G
z5k0q59M~tpayTESL&IG*ES<pSY6O@%?(Xi=g|tje;CKQjpYct+S1-w0{zc=8bf8gm
z`$WJmLi?sx;RK`bWm^$43?Q5R?ll<n*uSj8gb!Ar>jAo3HyrrJ@gO!MJvdEvA6|<v
z>hS~fnId$yK?ixuwDD5oiK6t+lwW&hof&?Kv!>~)L;V?_5Ax9eMCB;+ol!Hd$l6(q
z>_vMi&8W657+bwjb$p1PjzD?N33x~&4t7HcmjBQPz|fYu{g7c}urbOKyiKi{v#Mh8
zYS6*q!tV6#S8|<|l@)w;3nJikqqW){K31gJ^!wgSMP)|)Z+iOIgFHYkp^MHPDN4l3
zo`ZP96L^_fGHkcj70%!r*r|p4y1SXKl^#Wwz`w1fYQaMkteT&npCCzFTI(t$0iCRx
znwX%VU<aCeFac{VZrw{ej=kzjj{qq{qhT2{X)uch0p;q$Rruqj^=n*4AwMc}1zi%>
z6(Q&>y0K_*oD7h}MA9XzMUTJRu|bZOyKNO%C4YMGGv9-w@_Kwe8UJ<kr*HjjX@Q}G
z>$r|7>j34I8hXKNMn@dk=<ocgHe#eg`1Wp?uJdg}UFU@OgoFdyy%W(+7K1k+^?6_!
zWf{i%eZXgW05u3p%B2aJ4eC*y`r|H#U(%jzGAHC!sq|*BR*W0YDQJQ2cE3q}%ozUD
z)Uf1^*Vfe)09u75M#E%p)oa_A+5)bC<HXhS55`L%I1LO8tRXL2-`(whD05s`l9pG6
z3xHT6nYV%|d~rT~5~cYiA{|EXJj#Z@T0^?UU+dxXu&E{P6SD7Q-Y?s|pL%<WFnXW9
zdhLf35jKBQeh?b@nJfo<KYXi8Dm5lG%T0dSAkZe}@0PJOgBDgileV|KEiJ=*42}zZ
zM0`B5XY6D}L#V0kTWM+OM`J0Y3KgJ1bOnB2p#cF%1i^^F#UEI@4G#8Cn6rCvCAOF{
z#{-teOW<?TIy@W>D(B~Tcwe=&v}&v$wW(zW13DA=kn-l)waD*B11a{Dvc%AEY<wgS
zm14!VV-BaiH)?OQ-KdkFPf)auJ`YoaCyt&j|GQC?V;7s0E*5fMuFBb2W^SS}%b=g;
zy)ok{*wM(yE*mE={BBxL=jK*3YLEI!)R!3(QxVa_vW?r6Q8g~9Gn8u5G#Bd|mNe1@
z7{*k2$be--;i&FYEe1|f!1OyUU+Dv}&L+pYnGKppFCS71js~vstmsg7s6$iRQq5cW
zT@V%4!GWR*dJu_K9_wfxYk|U}Ve)8}<OZ<K0m<AmpO$9sgL+vDHPGaOQEKvN7`NS=
zRaS=6>x*HC4w<erCxn@+m2V6Y5+Bs>f<lO4+zwlig78@ofkTv;nVD2h6ej!D)>cf4
zM4AFM!hCD%CqxXQZeL`qZ{NOU+n52Nu<d%k_FxhxIAHm2M}R>MDM5RV<^??=W`#E<
zH*J~5g<)%3m-Ym{l0CwCdfP9tsiY?)8lF8uOQSz}$@DMS0x-GwSr}qNlgL@1#^t7@
z)T<RH#AH~(OPalB{lj9YKJ_$%^b{^#fhYkKDK%o-HT2I`;Tz@#xI=<bsOZXkC1ZGb
z<?)zwv2|OsewU0lV<KBENN0G~y~D@QYi&fkWJR72@<s}2?bO><C#rVblJyO-oYL(5
z=a={SPC8Gnp6!YchNr=q+8Lr$44>?bNTcSAh!Zo?A{4yXG2O#vU<SgmkMUbB72pWK
z6`M^n4<>Uf=Ed#r_a2<q)YQD3=L3?}(kWnMHaVLb9o^xLfC_`hX`>e$sybW~wXS8V
zdbAM_;&M~xk)@!$%wqC;;}?M6h!{l;Qi|m4@s}jIryWlq#T^LzQQruYdyL2o9DPTg
z{EDp`IXFKlWU7cBnOOW9GZK3-K%?)lJ!@M`ip*@+dnw7?=ED%}xZT*RulSBzQo$)d
zE}8O2%-puI@(AUP&_Q;KwF$TArO6$Sb&S^8|8ogGBD{+x+DNSp+Uvtgod2~`?UDUM
zT7~Hi2PRaM<TpvCy@W6o84@hg#(;^o)WdM_;(c)>0`Rf~Tksp<@aiw0nhM#c)Uq=1
zr5;#**9jM<b58qYLzaJog9Y1Ksb2{dl&<1xoYdd!*X1jfcmwyO(!5zPJq!rSAKk}b
z(0~=nEH6;|^@E7yS*TuD3X&dxCHL+#LT}!zg3J(nBC)~sBIYmq`}f$MkIruS1>56b
z(0etg=U^V$oxnh@+339vb3AY<OP5UL-L&HZG%TaZ!rN%uiVv}%K!IZd?tdUtSE$5O
zRZ$^$Jm34XHLM;)mVyeFB951b+Q3{Kwb>IE?h#ZssUQdzlwE>EoE4j3fVEZ&Xa`&F
zJz9-_bBRq&5wO<yB{MZ1a~nRYfgYLA0?+rY+e?-+Od<bt`mES52BWuMDHOULuPrFm
zF%vWs@#_)^6@SY<XvRNIv>{_tV{Kqtg<$Dgip9c&y<<8?K3PzMo3<gg6rxCNZQ%$n
zY`D@bJoqU?0a{w#6D)gMU_1(D02IWt#QkULh%T9?P<A{XyLmBYoBxA%sQ5tfq@(*+
z`N?UvT`HN6*A-I8;l#dsoM*=nkMP1`$XV>EJQ+E;^IEhbT>~EtO?GSTG%92LM-dT}
zM|(eD=$3Ey09XLAH}_shuU_-;B!R$|2>wHaz_1OBc>?*WK8p?LaEjR3y$7w{ETH_K
zY)3;z)&`c7IH{tb?7?T!f&oI`+~T>mNB1{yW6%XVmX%$j%>zTTQZv{4+pFloS%$UU
z+Y6C7(6cCY-~v~}XP<6S`m4MeR|Ev1V3gWwItIKHG|zaiLo52HCykX*wX1y4?AUH^
zsD2hAQn0>oED<$h!Sc##mZG&p1#gVZ*nsFzj)9i*`5*>-=w{DP!!E0fhgXY*jt3j(
zhJvFOR*JOy|2FDVx7oSrvlbqgO4Gg;`aruI9Ryi6GadC)=fc(XU+e@ubad-RGALya
zTwk*?`kZYqcz4arm7(+H6CNJ2{1zL{Qaxlq0esA*fL7z@u#uIcqai489-Xj3(UO!@
zm~8e-CMS;F@)Za`{YjjW;7$M@wN}>Fzkq*J{pIqH$IdX2q{sec<lu+{+u7z~&T(aF
z$*V={W7eoPb9`?nh*=<i?E}_`_`Gf$5)u*+_=rW*or6NYQVZa=^o^hQf4~Jp?Lg6e
zwHhf2q_q5iAzSZk`^xQ*AL#U3bVENFbO_ZRm4J00S?~QRcO;ra7e8a6d*X;fOZ##N
z@6m62Tl75I?srw{>$TgLQaOhhMk0Ut)veC*)NkNiagp>v5Rdhf&!zC67zcj3bro+h
zu-C9@P|K%P4y*n~tW4F(xe;-1jM(Pt{VI#n`NrrgM$D1j(ueVcDB9bBvTG!US7MUX
zJv@ZoP^agc<QL8SuJK9YCI<#<lVy|fmAD`(8J!p1jS>9X61*X=2CJ)#ybb9sjXXsk
z&DB9jvm2=AnLfHF!pYa@gJz@oj@&;}EC_v8;JP22Z@63#z<e<>({vw^?+GZoAb+da
zHgv71H2?boxT!ihIR#&A{jIW`j=qeR&KrNUJp^6>6C8z{8#=nWZNLjHp4Z*Q??O}E
z31>~Y;h)WQsky7`jmn;?oSfd8+2a8xU?Q02dcY02E&c&{=hp)kxH|&IzQk$ug##j=
zJG-2%J&I>p6+K-rf$S$zJn^5EKcxV)=xI9oosq%Te4Ah4Q>~1exw^G=r9l^M{y7Is
z`^!)4?D}=xPj_Bui_)oeH#txn5@rM(ei&$8SUPZ*`Cj5Up|g?B%D2Gv!f}@ea=%A<
zS4Z~erP>WtCw%f2)>mRvvg)YncjY^1!cM`Bz%}N6a9UibTfD|$qFjjHk}G&yd?BZ8
zxK{bF{y#rz-*N|s{)G2l#%cR{clUQ@vryRPC42-kvz4_~`%qsbtUA}bOij(&a%7`T
zJHO&@<EZuE{-vOzdZNXI*K4q;Fp>KU+|y>z--Raa2N^Oj%XuBkl_}xi)z#OnSk&Gp
zM#<_ee<iMXGw>wzWW(aSQh%^nSxqb4pLC3tZ@!Y#R|_scFQ`Sb-qONO8t~J?J&Bx6
zMgF+MyGLJNXP=6p#%m*iZz$~L2qD72x^v4pIeuKg*j2ygChGRAb6fgI)z62tMTyD`
zy&wIAWSHS6@Qa@oPiVDy9eXO`Nt&B>0v}PIzYaMp9ARdavi^4)l*_Yd{iJ!PuQe?i
z=pW{O<ur^uAuqrqv-!qn?RANTd%1HW9u$A^jN#MYv26~jv_SRL{EZ<=cCA17QOdO~
zZ$&T-4AJ#=megM1Z^%wH7{MVm&2=8eQbRAUi|*~MCcwV*eHNTpSB7<gj^FPtfT+l)
z^o=kIEE_<8^ug`uIbzc?G*r}kq&RfBx3F+cqQ=b3?6mnSfGJ*#gPXg|?aC&PL;(2l
z25%i=)=gbrSAlbo!Nir3QQAGtr;36nCUSI=*N3f6x0nY7*iP1AKI3ZKzJLlawCBLI
zpY6{lsP*AB&%%97Wib=%P1KL9{T&T2exn!jB6qT~7KsZq!g@%;Mt?sS5TGX-xsEwE
zQYWV62}96qZ|#L=f~|6Cb!UijKzm5aPBQci7Cv-;a+?6Zaoi$^Kc+oZ@q_w(RQ~0A
z1)fvVoQ*=70!peXPg3zHvIA&bta*#*a-DgrE8=Q>-bye)GGjD*o|P-4@oU%6&>K>l
zzLJP97YX!3%3s`ru~3!*;87mFYR}~t35Nt3UW#I+oBKQ-YA_XrD37s}&0@B?bPxE1
zDsbPI4e`>05eCK)^0x2d#Q+yEZE^>A%Hp9ZZ2ExY_P2C(HDd%6TojG%cyR9bB{^E*
z$;-1J<Jmt`soiD}1g~|N6E4r17M{6sWcKRXHEk$lXzmog*Pk#k-x4q0B!j6MzzfWc
zV8iqvEqF>Wg&gsHvCDUc1W8MoNN;P?P0^A^#aWT#v#y~)lTd53o=k{to`3A$R@1$o
z;s_n;Fx2Q`Mu?)d&_`A33=S7E!qAv`yV^X1FT=G8Ns8SUp>O5B*L|4PGRCbQ-I!T^
zOS6Hu{KVE9>`TOkaLtql?;khx2U`lAdOCOXebLJnEDUb5_?eaYZo{<(Do%yy{rfP3
zY`ZrmEZ`_>57+z`p-2fx@^GC1{P05|?{w64J146f>g|=mkmLX155fgAbngbk!^4Be
ztoKdc_7TYdh!F_0ow?TF;+&ldav@*fyAB^ohs0!^$bOnB7)P7Q9m>JXW?(xvCDy_O
zb&3T1<(gvI?KtXXKlseA#y=4G9+pJilcnZ8y2Tf8!6<OLP5r<s9-e7E=D1{OSnq;`
zF3Ir~G+X97y!eOY8>~JBXwOCY$;vn9?JYVKB@m|t68EFN-D&$<Z|eEw&a_}F)DfcK
zq<d9c+xl1JU4;srRuTyhvFEUs-neWkE$W>7(q%#u$5EeVFWS<@gU6fP@d5O8J9utZ
z)r|tG{i&;jJ?ejDjc+xrQbY63>MGQZ$#)Awv-jNF+xv1o(elrhoxbj4h>VovRW76i
z)<U;ZJ71=AAK6wQ$;RWdqiOR;FH=!_EZ7>wA5;Sif8u#3EtaU9LyEAWRsL)t*c-Sd
zw6_~{0gm{ivGW(03fH;3FQq3oDn*oCoIbqpkn9{D^$ScL%&x@So`Wj!(+nl0QQ9np
z=vmCr$Mc!<*jpxJm+nro$LQ`W7Nuts71fLHY>zBOv~N=ktbVwjZP>6Pq1O+;GL!pk
z@j*}{7r!i_;0-00moDFd^!M7L@;v2V4<?S+&5Km4SsTR{C%uM~+du5g9u{%e5@UU$
z$Slpv?PacU6UEHXAO4TCw~Wf_eZqx7MHB>7q)VhhK&9IPq`OnPq`NkVh>D7mQi1}~
zA>EBgcQ;6PH=KF&_doACU*8X|g~WdLT{CmfTytF-NTew1zZtpS`^c9_A97sAnJD&U
zH@K9w5V%4p)&bZ=?Y&dr+<};(Sj9md^8G*vii)h8rD;$nuO1X!mANS`RP3Z)FAb*X
zTd<&gJS$Ego_fwMm$?3<j#>>bTdX#NBZq+kE6aky=tBEE3&sFkkSebsf~Q`t8)cEj
zT-zB6n$$EN_xDZ%x&|#mDNBOB`g~|Y*k7tL7ZCT;)$|%}=m<K0+#B6)a(sE&G3|l3
zw-_74z^c-_Zy`|wjLDXq1GTZj3s~>?N|)re4b;1QXGcQ0DMF%rqa&s8rOmIB5`fM9
zsGw&jkdhbu@l)&<^deBI3_OT5e1(_aX%CL%t?5i_APfj)Ps|)MDt~LHic@)o-k=fN
z*lJ<lUM-qX<uw6LQWP~J_X~@nBS)u?<b1Wh7cpH|3kI5pMNxJLjY(ni);s|d>&4Qt
zNUD?w0oOsrP=#m1{<P&6JXt8xx>HEFEibT^V&odPGe1B+CRTsGYs{S!klDS=rGqK@
zD|d5@0mo=#+tAL|Ax%XycD)i~{kDiVa*Sz!CtCM1xaCmCDm2@wLeY%vhO5?<=jXA{
zLaNh<L^0kB*(ox;9$!k`>E}}Wf|nt6$70#eI@US)(C)7BgBt~_#an{c(u>z^KO77{
zQB%VTGP>cxh(z^!@oRI?+ArG~)3m;vF;fO5rpglO-hM;Q)xqu)^}n|D@f)#(cC2zV
zP-HPW+?w^=Y#^egrM(U;DmD7m0|OC|+nAb^uN!wLkXKj#9mD6a(4R>LoKKuAVg?3x
z0`RD^HiVU$=svqweEt6YPTsbanY&-*(IcL>-N<~e8RvI#%82O5%{1LKAjZeHzv^gs
zYg4?bW8d1EfE`O!iTQRfPGPeBQP7heFWQV@onhS7$I-quF{d+(3R*z`62InVJ_O!w
zg?x2$sW&u!L;&AjjU6@?+P1eYkNL#=<8YCdhEoOjgDXUPG1M&76o#_Yqyl84n#agZ
zSFL*#;ydx8`_-gW3~R0uKL>+rc^NbB$GmvP>xPn+S)8_WV&$O=gq(}9FK>_}A%A-6
znzb!$U@<<CSr^MfR2jakt}=J2#=)SIgTKw~vwBfCb{n*$b$s^<-K^_nLppu*qG#+_
zjE8rrv;D;~OFPcpD=KSOmmw}^M2<&UUYW%#<yUJpx3_<Uu)7CfCRpn(_V&z{Q}Wu{
zR$V^{*>$C#Kd0^x+K(B!4@uP)sITeHeu2aY++kaV#h$RR<*wLSzuzfcUY1ek%t#+;
ze)EfN)gU-`N~eV~g@EbpXXpDRsv>G5xXY3w&vnr6k-jH(bhj#)-TsOE>R2=xyI|JO
z`tb0V(PK6F_nCea+A3et+U#;eB#$t)js{ahiGN`q8qopg^^}!{Act{%G837qCo6C&
z>m7fIl9r?%QZO89HASN`u6P;qM;ZqIu++dU=sFT|9(bWTo<~0=Q6|WY++zE%Dn6bN
zd%6E2%&PT+!0X>PQ;~=ZHcpQGu`2JFP$aZaM{NCa{~`|$w&`!wAq<s%9js@adEYXh
zR!l?M;jvk73YRM5Jyr3f+WxNG2I&@IV)f0Rb%VBf64Fk#!apPl3ps!cOz30R%Ynbl
z-e8^VO4zW<`{R2SUEw;kxxZiJT2lyxN??l8o$^BHL*7EV{+57%067H(E5GcM+=7A)
z#a^-cS63@!gTxXZWoCsqVK(5@>T`w?$Z&F#_NI|-?CL+b;j(g}ODhQ9gE0y=hU1|q
zFPB|IUICu(MS(7FGahmwGLv6_cB}Kkx{rKrv9uOTDjB#Q9QU5gr#02j*3G{ec7NXz
z(D4}wiFsQYPt0J$1J>ax>EK)Ey2qAd9!0{%N3M)ysf5DL^mMN!6MprLP=x(>ewJNw
z?Dlu^i;qgx4m54|oM~8uWft?SY-$d}1^3v538=(-SrfY6z`6X@e>qH4NVoojZ1zBy
zlliCfKw#1T)}PXq!dfo*M0+6Z_ypNS>bamVjWcR_eplbnkS5N_-|P>mjPE{>JJYbR
z1Ofw6pvU*a<6{nx0>N4jxnMJ+q&iv9rJrEVc~19=M4Deo!*@W;&_U-^Q5F9x1ADI9
zo}4!Sb(J$*EiQke;Adr@p6=|0SjBF0c>TCG({E0b<CJ41Hvb`IEI#Yxa8V55jm*47
z+UC`A54|xD-Ats|@`HDluCV>bwmjs{+CoW{jWRm71y%<`u(R;zl+4B<(ld@z$`Zdy
zJe<Z~XGe!_t3=(BL%g4Hirk{87z@#I9S@wX-o_SA>H1N6>+-tuh9JUt_B+jY9Oszb
zSUIuHb*3!8nKy<@`tCn1%zYS#_&081Thp!2;ykqJG`vR)(lQVY%gEYvM(j?7!IWhW
zfHGiC!RLX;hN@%p=a3T0y^syubbEUP9qU8f+}zy)x^Kj?x|H)iQVL<oPqmJ8OG$#<
z3JI#|iwZ8yNkYD${vqb@CqT!_)6icr-uj!rLS^R&MU=&(gqMit#iw(%S)7j1il|$G
z!pKJs^G*C>u6>Jzon@6k>lZmci$vq(ub4V~ONh*iq_MwUazzVM;N+#2icr6i!Nh>p
zpx`lsv>5(@K=m`dq|SNEY(a517&dT*F%Abe??HG9ZeZ=uk)KVN@YbbF)Wlm&Y7NF-
z#w&V&&?8Iz%5>wCS*opBd&;C-FEE{^4wfo%XmHlHxY*@?;NqG-U)GmHU8~5uA=h!I
zh+L_CZ1h6*)j{2(L8trduTjpj9_KNKr_qz-+x0jJUAJ7NNUo{y6lSk^Q@<+iB3CaA
zo0vFz#B@`3JD^J@Y&D(FJiaV?Yj=I_y7$LKB*e~>&yG7N!wti9Jv!aSEnsSlcXt{6
z-~nN+pQT9Pz^h#L>w@1+3+m3bPsm2*veC)71cyk)4zx#t`i8`qPIm1BB-)<KUfX0~
z+t^LCrPMurfV+k>p<3545WDxJ^k&H}Z-&e(kElN3XQjCP+K*l&P&?LU_RHmlyPtS~
z{>bDh8|7;^zy9~%ci*`(a^G~<tCBR;#td4h3Exn;WhJLVS;TFkjgilSW3Xj5{FAP&
ztT6TT?5|hk`7~fCY_o97Vzn(kfQ%o*F2sUUMxF4kdNr7E=944HpP<=-9WRqHS;wL#
z^$P^mpvE#0X8*c<{%YPT%fPbpKT!X5;IDniZXf(WgU_o=rbxeMY+cCbmuiCc+SwD&
z)r+2Fn925?gG2YWR-06CKaATwW?CCgf37QTJZJ3`_@bb*_PD0A(eHf=Q-=Ezhw<FH
z)8t;VI19#RHaiQ)N6RNRFI~RJ|9*D-?OU-PW?rM#tHN(}8`hF|WBVOU4adXCU<j@y
z{!fQ|)9UvMVYqoN&Xa|UG+_tpPoHE}IyXu%zQSWiyJs|4C9h#v!~HKddYaI=V6vf_
zm_P7-C+5auPI`L3&lBTEx=bfRklMGnk3ArU(SiBcliIGa-NCAlvQ(}-b1k@{eb8)`
zm_j(LuY=V;o6S!W13y|t1PK|q1Pl=^GY>HgClF$?vgQd%9Dbpyio=Snwr$_*D~j8+
ztGyeF*9D#40l&lS&aWr0-Y=<apDzENrJ)m*uehqqhn-#Emj&R$FQz{G-=~yu(ah)Z
z?~}+3h<>yGqZLwOzjRYWpeMBPbP;WKVrfOvuS8O0wrfhs&O&i_2C;29eFAhh<_6BO
zFC*G(msjK8>g+c1#o=X(KhF9ptn?a!tY<Q^Qdg)mrVF;cabjzo*sJ2vw6wIWURb{i
z8sN!}e=Q3F0?Y$uCL5q3(>AxW9w6~nR{q`VIxh>VYF!+;{LA_Sja3GvO(vVpq<zs^
zR8*<AoT}dYN3=hbMs#Ndp3c?;CWl3)GZP=w>s*@XN@zBWQQsztv`dw=eD7=0s_N8P
zJ(Gz(xY;$Fi{VB34sJUG49j6Qmw6Nj={+{D-7URBQYz{Epr%myi}M9s6Xv{-5#C^O
zf{r0p@nzQiG+wuojdeOXrc6oh<ZArn$%iJhd2S)^LTlFqx)6CIqXVh^&iff21&s>k
zz3-1FZ)m<MRB$&C@xJLTqxf!?poynCm)G^^7A?UIPgA+qmAvC8_yqsLLQcso1~EA0
zk7CbOKjqp5>8*PF4##r#m1Q%EYlc7wL(>sEU(=iCWlhJ&nlc%s{`Mo%na_8>cq?k4
zvG&SDYM3Jz6ay#Kx$6nh;(zwtGg*Vx&G6=4U7WyUA)DJtW9=Kmc^UI8P~r|BGkV-!
zTU(Ivw-#}O-!e^#CM)e&if1~GH@qcYIZ1R5VGIlS1!OA5lq+7r%)6bQcrY?>(d2UN
z=ew>9Q5;niKV~+TuoHG`lD!-#HMX(Rzn+2rH4H_Dnq*eulLa^nFtxZ6iC6-cMdw~#
zB_E9`_So~wry|vynSn${$@$3q;cDhOU#UIM-EuU&dgey+R16}r-@SE*g1k{U{4^FH
zqR>0u&lNU`>k@H%{hMceb$x&+5UZTMeB-4GA#2YfbL6fe9jj)({#dD#llwL5S1D@r
z_{jWzjOsgCp;6Yrz5#sMSzGUp#ptYR>`pgFs$smV#;a0#_yxPXuca_-QzxVBNVUj@
z@vvvt&V_tp6AFim#r!@89^a`^xTs^%SsTynC3KrUXxl@U$IN&CBrf^99W%6H+Imy|
zh}cF-#hsL@gRLq=GwIgL^%rL@WS@1{^b?Tj(YPDC5uP~32c;ad$3(n&i;Nx-XR;@o
z*c%>O1v8$bS1Bgqyzv5_uTRS_&kPQZ*bkKL_ZW7gIeV3Vbtt2`w<1|)5)&H|{?WzB
zv>HEQ!))<IKi^+L9H`5Qzyts>U&{7$2IM_031$fuL$$83QiQ5vhi|B1DQh5C7gc|+
z{4$!)Fs`jNM~fMuc+}b+0?LpVCWpGC3;jy3cV-)A8q6h`7<MjFS4%9qS-n+F>gXyE
ztv*uTBQ;+=*#Gdg@LmO%^magA$#uOek^D&&Pw@q=*eRGK>9#$(D7<%z_RlvK#$Wy4
zE<^iTprmMH&YI)3Wm#x={uyA1{C@VVK|Z6g0TRf2tmeZ%le&wB9=GuqnRJ`V$;YTY
zTbUxx%{G$%5*v_^jJ?Jh)tT#Yn^dc0P!e_518oeh!`|DF3W$H15v=?3{BBHS0<G91
z6M|Dyn>)Tm{@3A#2FxL7KtmW!ki4?#SUBzwI#J!B5H_Ev%ET&Sfs}sIUR-8lSwCXl
z&}|qR`LI{_xJ4Hjm$(OqG)2?|7+4=&<kt|&`j_LHp40D2ZVwH;WjPC<{+*C~G&b~u
z6^pj0!p*vxx}hy~9^R0yr?shpv|m5caZuwZGmoAMZes5>W``h|M{)eFp~B+sE%+>N
zrv|G#x+dn0{E-122|ZnL7q_whmh2VXUFMC_k`fITJotzL{q;oA<+*3i4bX29hf0mN
zIZ;!g?a%3|Hn06siqWgD%;t75YjCz#%4H#+Sy>fZZ}Kadt)U*|R~AVhXlfxK^Uwl@
z;CK$^)8ICzSV3`0)+k0YBCJr+n%1!NfKY=Nq+A=)L7EJF|6I$fVhN4GVg*#$Mn3gQ
zIqRhzdfqSQ8W|NNKHX81CXR4Pr^54!q`>YEj*d#$d0$b)CrA<RIW|K6Qs=i!o59El
zm#5$k0UOoFRPHRzgdZ|hALY$_3<-!nSk_-!QE{KR%)McJVx0ZC>h8*b=WYCJ$dAh+
zRR4Xp**{%no|I1l9!z8+w2Yz;9mqRg=asnT@BCt~pRgzyW0cJlsorW@XUTK6*?yH>
zT8l}&p)W<enC>Ywc$4r|HPJWQbOyerNKc0eL0KC?<rgQD0v^m{6OGa5<f|P%QWyvM
zHnJgTdK`Z0;*fVQhW^)AN_@=`eDCQG?xXum-Vn<C68mqhACJZueda<L*8-}W@ARIr
zWIe5~9ezsOlkbd<b!}r41I-#aTGQN(cySGeI~5VBXTTXb*@RY{9yM)_&00Q;+fQep
z?f909qpBK;e{E%#l}n|P%dDU8fB%peEBQ*M>H3^f8A9u@n{#jz&F?G8#kAL6A0Y<3
z7*`tYlvk;$D}c@KTNsY8=*C_l+A<%f^KmY6$qEttXWST**n91)JHT(o(5@QXbW0r?
z<Qtyp?xi@_T{RQoytUu0YV>$C-so~Rqz>biEeL4F5AZBNRr1oPR$$9q;QI|-PHpDf
zNW2XuR2Oty-s%QiYj9URpO`1!z5HcP&&#eT6pMK57UheidRJ$r%Q~}xDXCl*Yo6O8
zDV!Du8+XXt)7*kFF1xZ|-NZ8INx@zXaLhccbV^zNQhBg%fjPhZ+&O<sGc9#&F^7}z
zLo5!BZdFkY_P<2?_ss{S@5wcP`r#p%ZPShBR1OCLvD%|9;h|qHP0MLduKY+i*<1_q
z<XoUr71`$lPmB|rd)>rx<3hG>Hv2vuoV_{llFxeuLc&_cC*HrDwd*4qW&2a-a3e<q
zKS*`_{rSTEBs4dCzMV1L-07b9h%&{T%uGA)&an>=sgWPrxr9Brc;tM01(|&t&r^$Z
zFtc(w_6dsLciZS?0*-B$5tH|^=NnE+*Hp}Vr4L-A9<)5Sxh#l~vJog(D1E+BS$WMy
z+_$M-XMJ1_s<+E@dVlw@-8R$j*4YtE5&ao5mfuosueAjAO>TRR6L_y(opAOmskv8!
zH)_SVyX;LE<ms34qLB>Y-xZ?+oHE>hv$RX>mFDHCNXte2fF+whHfzNrq-m7iAH=U6
zuujzws{-eY*j<97S-TCb2cuILE5>T%uyHMZch(GPA8=4OC;490+`BUUHHfOQW~}6!
z+_&*C0^{U-Ws}+d@HC?d8VwjXPsJLU^6UnrMSrz7eZ|Q}*cvyC?^-=m{F?C9y4k@<
zZ;L#(-B)MPIPd4hBPXL1@u27HEBazp$`0TD2PO!1YJ>!8DvRk)`=dM4A7cZ=<*L!-
zl7j3!OgC~=Ol~vWs`0&cz>3iw__J^LzRCC#u=eAf$zY`+Xdz9lUGc1p;tlBCe%m>G
z<12#q%2{?6vt-aGNXXI2|5&rS$nMs;qd(OTuiC0Pk;lJNp#1JhlhCu++;?K#PrmfT
zkH{5HfI6Fz5xb{AzW#D~-&1S8%M-tt^ayNi9j5ym*Aw?o{kjInJ#8+ML?}IbE8!E4
zd;O2{DQ=kg{I#}x+Sm$dDB$HDs1rOs??b4nt|Nu*$B|!RlP9BFn$<93%C%9@5}~yU
zid`rEvSIKEn5wNt=ee0DC9X970zLYsnn^PPxa;kc^NrI4`oG?r|Fd5f)kd5@kA{*!
zm9&hkd>ogYQ)T^6ys!8Pb?4Q&`W)weHP_1r(?s2A-ob>mSfnQ*UI55JlDMFLqW&o4
zp$NS<zaQ*>xZ=%p`!d=GbD)SihemKZ`YZFvdi~>lV{;t3xgAS!==qo*4K1t~Xxt{~
zY{IyS1<`IwVKw&T$Ty6A?=w!o7Q7jCe?REg^@rB<J9u7rMmCz>FwUA}r_XgOiWSzD
zY)Z2gtA%}AVj+Yc=Qb2EzdOg=u5g^fGG)!Gq&(PHCCoCrxiHYV_zwai#%3_{2o<0c
z;x=Ds>LYN;<}SWUOO$Z_Abet6y{0!TFDshcoF@&$y0{#%cs^Wa-}oxvlJ!jqVh*te
zffSPQM{UJU!?8HA%}z`_bV@@Txp{)>67O@ra3O!-tt%-s)xA%aUsdE?1aMSSlGWen
z_WRTq=@s=H)gkZiT)chXi?x<0MFztqM}KZVT^E!X87WieYDfcxLck6Cf*^sA5CvKE
zJ1J+aLaV8EfpsHLEkNZ-2SyQ4#sF+=JpduvK^67NwQEMTudWJ0dEM1r6%k}g0*N*h
zQXJemG)G24eIInjZi6;3u-AV8wN)sXqF`hw1qDo8@_3*Vs;+*xRy>3ILpt&?6mzoQ
zk`WUA0?En}$Hkl8-UG4+0F;`9>X#u9rW+dqVKu5~FBZWzAiM|JF3<(aaosdP!B}hb
z;}L+e2yXqRc~8az63=^TB2oBx^g(J~C{u?DZ{nJV6&KPM4y`XdaB9A*!!y!+yc1qu
zOroXpFaEI;D!F+>>L|&`Smw=|^&Eymy1pXE7m6qKub=E{PCD7!8$d-NL$lx=s=lOO
z?Ss|Gs-A-eDuD?>!kTN&AxKj_0AD&z4!2vzD51^$>C;Q=75kJPdn+hV!$5uS@-854
zQ0N3GL^Y3@-Xp$8eC--J?<dGEhnB7XX$pxEbPFzVm=6vQca}jf$vrCq0Wj0AZm)qn
zDZC7Yvquqwiv|*+kg1?_GvozWa%dues{Kd%V=H7h|1Gfdbpd|ad2RG<Lc$DJWBU7&
z`{^F#q_w+4gK{v;6dE?+wPrCzg+l?wz_7{&4%yn+<LA@Qi|2E4PwOSrZ{V3X%<fC5
z8t{5VA7Ur|bKkYhuwfOskKEJ=Q;G!4l##}CaXG=wOK5SPTN3{JrlzKO&+9I0Bih;v
zed#BC4glRY0t5?+-yF0F6y)W7y3A-H0|&hqkOgjqzW=7Sw#=Oa4wKf&8ceD`fKVj<
zASO+Z54{tnfRYE6`~VbJS)8zpo!xzy14s|Uf7;GZC(gg6i3*@nc_0h;22xjW>>;;@
z4K&wwcF27cm?Vm8IVv&_9<2}i3EJvIq3HarXR&j67ch43Az6w`!CzmoKdv=Ho^c~D
zw=tdicPOgz!$VLddj^FnY4N<!NNC}?8|26JR^CeaGT2iW32X@P-ZU#!dcO(y>3y$+
z&rI2t6T3j{gWem&FE;4kG$9ur%qWI^;3mAAcx}?^V&&c+hClHvjA;V}_ffz~Kw!)^
zhj|{(M)2~5i9|qS&BcwK9=Qk>ACQpT0x_HE+1Uo@x3IOfWi0FH>k9+H)k^?CoCy+X
z2SHtzwv`D*Xf&8@VXrFxq3*FU`3McN>DpNYejp)$Qj0*w%iXwy8#6NQ>bft|q2ZGA
zJ_M$YmbJWsf+R2}C+o;LOOc$S;!RM?anoR9)jSzn%K~f?X!Ks8qM`ysmz1>hHIVG<
zn|uD@OPG)@&LdPB;8k%d<Q{u+J+2!q=v!j73gU=_-!12^!6S6>RS^n9vd1asF~v~z
z?^cc+I)dEY{G~qJOszKdi+|<n!6NH<i#4?>^RToU-+lWzLif;WpcE3cD1Zt20Vu?0
zmmmNxLlfMnzBp?QU}^zaag>6Ow%xwP`O!cXnmS0lB{d=dKWDzH8ot*d1fp&@?cXiH
zJfWjd$+PduREA2O5Ee9Gx?xs(!#caT*rX&Ugi#G0I?Jpk#zAUN?j5TjM?2=yKbVh*
zqL^rD>zD3>>wA;F8_EBE@JUzB<jV(^MHL+7!-Ss92g~JHW;0iMPM%vSxpH91&Oel<
zMDwB|Rjq2m{^y^vzB#c*I(|+s5>)@#E#6mCaK>OSi0pJeLCMa+2@WSPM!$LUX0s;d
zi37L!;<XV_AKRQbKP##&7E7=%m<F;779nS-5U`YcK@m<mPB@m~tLsYuop1i839@Zr
zE$S?H+ty^aa{lxChL8QcYHWAy768<lDL5P#<b@9=F9Mmrw#|JOINC3VFvdGOJF&<R
zI;}LecfY{k*Z1*X!)a*h@h{&YUzo;PcSbuVJu#TJ#G`suEoC~XjN3kLolh}XvW8Fi
zFZrR|)Rt}Jn>~18VbR(<{tgW-!SngVhH(wc*gqC1+06s!C<M7R=nVrMPV`BrqF$PW
z7Qu3pxcd6151FRJ3GG>wkiI+m&d$yR<7+VeV6t2nlOohBJ^7-!UtD;Y-nZ4t(Xj#B
zJbYFo7W{*cJ$8x;11l>IN{~~T|A(V*QEzXrd!WY9OyBiHOt9E*qGd-nu_f&WdI)u2
zsO3#tGkI{8_v2CuJc9e~)nvA=42Na@7L_;AU6%8wr2PH})*{a3cY*J0Z5<yr?Aq~t
zH6zCG_x~a({BXdez#10Li6I^lclOGq+%sle?Q3<S#dZ|CfQj+sqPBN7pwajo=%Fq}
z4hpLVM>~sOq<`e@T_b4o+})5T4A&~Lr^Bb0&RG}?GaoPK1G8fY3NA519^v2&azQc(
z^sA9J=!bgy?%j7V!hMu=EiKpwWu031N&fxa->pQ!AGS1P2Ynfk^0=H6k7)x)c!00w
z9YvsA)4<tDequoPDQnwu*Q_=6K9g6}nYMNwrQE;ec~goye(mE~ws*<CZ?|F_T<})x
zpI>klcUh)WlY&iu0bq#ah!4>I+s-x<SeLC9VwuM5Ep%Nx@VHS}n%HM?6$i%{21PF9
z9ttaCaF5t^KGww|AK71Po+))+9VVs9b1<DvYwA9$Aovf~>|~kq1*d6+q#UFw`da~x
z^D2d?^yQuQQxf!#q{M44yyNfLsFO;yW+^RTI}Mt>$V4Lk^B3P$dr7&9goJ}3lS(OE
za%KBv&5v-7yG2R0spL`l=8_w+6-%@oQ5)Oa(9%RnF6{YbVZp5O?8p-BuS{bjuq%=K
z0Cs+J*fJY%=IVN>C=4+gbXQq$dJo<+#t8u;VcWS@5`O0u^;Kb@Btj3}#>PYNs(w^F
zQB@^`EDc&{!-&OqHUn<jJX4enyUz3dV1@|rqu-*nR*|W)s94g+I$cCd&v>1>EDrz6
z7qs?xb)!XHE>a{E)wQL3U!Xp}sEq4rOR3Df=DIqO!iA&tAR0Ca8n~z#z*L~0v3)!X
zDJv6zCbi4)mIKg#u3ZugjH<cEAc@N7Fn0}pU;+|O1?{tAE)&KL?+EG}8omv%1P+pX
z5Ce#v+fJ8wdTZu)?Fsi~Rx}`mXaE5iX#j=C?_gF!0AvGT#u7qYZ*T8WZ~$mKYh%%X
zKGo+W1q{$rF|qDsIVwkf3Sg!J9^Nc5H>6jz^);;!Dx@D(2DEJG2NPBGd*5iqWDS1I
z{``Aocy*nyYjF79W)F#{@EXMyD3`p(J!6;G@JM(dhHoYPS~TLJjnP8jjCqep@mX-z
z)5VM~aN211+_sF!$;p{1&TFyRKLrQZf^<zs{abt%iX!M0(wx6ZLej*Njg%`co`U-b
zYiG;)jhr9r!4MCDGmS*V+tc4r61y=e{Qt$ozQ_BAiREhlmSw<wE3yVOK&O8w#^@zl
z@v2X;=Fo~M=zjbE-rT`*t|^qtq$^QIiTBG@JaYerh)|MJeq&#0>4)|=tezn+8|7-n
ziR4LSMmLvekAyhOA8oVGuB>Qh9q<8-8*^j2GBflTiYqJMf`gI>l3UP+2YpQoLwOAB
zl?QhqBGuoRtSM+WhNGRKSM`-i>Bsf!*L}niYR1N5!87LqcpcbPp$?K}CkUJ(03+c8
z<_M)9(%&E`GebB@H47{n0B6$*$klA~$av6`EOTK;5nh74FfSk9w?2!K)4dUmp9UDf
znb|lrRH8bsLUf-@vZ8IW570V2#p#NDdNin|lHzkKVJ0KtZ}AJOx3!qoOdkssu1+rT
z<bA!8cx?xxf!D1JH4($(;~3d!6H9(@JP}kF!u)4wI3Pf^qP5+>%6?<y1+l80Z=YJI
zK<tb7PPC&c0EMsLtlo%N|7IK%f4RBX0NT5hGWX3#3RxlcH^$EId+O-u7`AU&pb6ct
z8q2!*Nr9m%Kj4_60vE22$*Sq7Y9P3T0!`ex`i!+riz`m5lU$L~Ruf2NEU1;|KCgRi
zUm=n<@;BUM`FbYYEEXEcvaHLh_;#vqI9aR?J8$c7J8yzF$CKr#_8oR>^thHpptbKc
zH*fN@itKaU|BRk?5Of?6pKS^4fQ#;_qPXMUDACPM|2aDNLu?)mTjkopzO`Jyw{Mqr
zR?(QT`{%DC-+lWee0dlc0${D|p|!X43Oh{Ym$o7=2kmJ0PZW{p5~7oqHoEaFqPA>!
z`-C!~EmdiDhrjt$+tr`yT5u1hJ@+BnvG@C-*_lQygK6rZ@4{~<-m%$+%ynCC>DQhK
z{H!5z{TJ*_i_H%A`aw;l%T%(B5FLM7%m~?)dk>eAoSdB8Y3q)!LK=luBo@090Y@M{
z7zuU*+G>1I(L5Nh9FsOu%Vg9-hbAj_8B6OKdx(}F+EsqI7~irW+;>(W0)q`7;{iZ6
zuJwh9SG6oJs9P~hOCNifF3h7VY1I(vJU6*rJZNis$n71ocU8=<>qHRM|M2%F!)c+)
zzOgo35Lk6~gzW<|h<VU@OfUCUTeTIVc>b#gG_%ixs{*(*3XTC`G6{#t`+^=+_eoaQ
z`qaJrtdASw56FLKc==R4(3iqO;dwJnZoKYwP&+t`q$75Oq!u3g<wy8s%`Wok`d@ee
zucvAD78$pD*_LNH;Q<Avzt*O{t!4>h91MBbi;ifXZ0+a!aVMvzhYJZojCb$Sa&m^Q
ztXTb<1??xxbH8J4&cro@pr4;tH*h~LLF1mlW$Alcna4RUsRnlPA#ZoH?<VtRU#I5^
zu=AGh>DTq%+mOYyEqs1Hp!G(?``aAC_a8LyRk1lB;>tYs*K$`sefoses9o_%bCKBz
zI?p_J`_%en#Zk^oK3SoHWTRf^lZ}stX{a-fze@gN#(UM!th9Ld7IE=uWEX(sHD>P{
zIE!ERU?kP#ufE7?5og4~DmaSe^I|<a>%Y<|%o?c>;5D5QfD`4uFg2aeebJ|KdR2yA
zX=opW<Bf~X0da-WV(OQbKz(rMs@%kf^lSmLi4)uzo`~l!?u;&7{g#CkDLi9(Z8)@H
zN@}#IFk(i$yXN)kV^zWM0fYBYveij4pINJR7*8viYCXu0<QXxTOb++fmGk~vb9f1a
z4iJ&CF)=W?<#Wy=A|lXtcIhE_4~@<JrP_69%g%`X2wOxUX)WaPBqAkH*m3UPQ$_W9
zSE6}9Bvtv+#83znw*dJeS70Y2h7Ub4FF#-%smq4~khF^7B=}ityOLv^YxLp&oXKNO
ze%NW3F)%)dhhKh(C@9>}%p~-Ij}0Yo?dj=h!79hVz%V{e>r-7PMr{e@7M6IX86sI4
zfFLbUFqGW1d%IG94Q=!U68q+&kcOD-tuOd>ANzp=Q)_zwnz(I7IhEIke6?|o9(GlE
z*Z+dmD|SkG?!HBORo*KTRa8Ft`J7QaHn`Okh|6dVw5VfMLo-37pxXm95Zz)LDRNmK
z|8K?bo7(V{$`prL<69@Cq+CG*v`C)8?2qufqjS$K-3;jNHyiwK2VJMj5H@xODLvzn
z09*>@5lZl3Lwi7c4|frYs`E1e!(Z>2%1%FvQ~pG?W+>1!9Eu{e;=dd=v-9dH_{7Dm
zu;>#~k0|_8Ud4vkV_;)kZt<HXlQf9Zu##=Mb9R6MH|jf6^`cwd_IEo_y8?A|>z!(r
zI8ZP}y~5KzHNE@pJ7#E}o}R9-XWfohi<&w**vl%EfkP0Yfc(E7%M{X8zxW}5cW;}L
zZ3PhsRM4xuKEJXmSw{VBCzk(Ow$Bfo(4=C864x?cNsOza`WITY4RL%}ByBv@CZhJY
zSX<jo&v~@?wvV3&CroFTRGHhdJYHLo$Zqoc>A;z7d-v~aL56q=;Qz~OYr0c+EYYB&
zh(1%ZU>an}sM*=GSN8r>I_E@4u{~~KQ9s3@i~V4@8mRZ#m0z`_X6W*8iEjkOEfVUV
zpGFs~P>li!BBv4^2%gbk#?`?=ttopCJ0(B`bl%Cw0c-onkuu-mfykh*G}?(rQF_-#
z0*gx?pC<v3kTv79<#!H0a*H)xo_Z|&HrY@@kfH@uwrONJr#DhY*tfUw|2qv``Wd>D
z{>{e(AZJeLeenE;Y%Fw1J%D`4!O>9^sJ~sIZoL6`iu5?0ms#=ZOr>KJH%rTW+eCh-
zn=I-ZKKK!KmHfn608l^=7V{YFw(QNn)HmQH*X773RhMz!Fu5E>^v6>;=J&7Ew=95@
zSLW+Ax`%9yXiYdQlu<OVOx?5j(s@mj@5J!7Ej5+*8wMM3QW24|SM_-jBwL7K*y+EC
zHPg4bEHv85bj7YN1@K5tz?dKt-8|d=_{|ZK%MN+^*OOa*?P&W&KIi!<C*5v5S_du;
z#A`S<Pj4V67g&;!+S8Ueal?VzBEk(gKN^ul-<87^7p%g>YDT?<1?@64G5ougSLlm(
zf8I#pR4Hno*sEcO>r@+|t|fM?4hv-pN=B{a#q%#b<z#JNJG|>^#@<4Cw2w{v-+o-G
zW(O>Y-!NHE!Z(iVJS<@3_I{DsY@i5Y;9viuFzkG-1~>nV7zM@ASY!@wUqxZGA}Rua
z&aZDXuu<_f_V23=xMh8}%W<*O@60m{CCRf9dt|TB=E99GWdQx{y67^Co9n(RxIsg1
z?}Xw`FMlMR-PP2v+mCZEsf+ojpKKy^&;bn)S#$r_xSQFk#=YDd_gIIl4w@^Ri)AaQ
zdZlDmA(ewq;bu5Gch&IJqA0I|bleJ0DacNLh}&I`^1VkOy;I17%tFC-fv51&BZbe3
z?+fxjd`j)0{efFiyTv*(nR>;wfc2Gf&s;+16{^?(*3@EtzTD+79y6+m@7+_E6Fyrf
zt;cl~q9Xy4jjAaX9D-|jRJymfGM(<aA-K%ueF&NlpddY1fmKf5)N-Jj<iBghQ#aK(
zRA@UulQ3R=W~>EJsQ}GBtGP81w~?BvlY2`zu+_@?9BzFGkxZJ-GmV?BnR-H9LHzZD
zk4wVTQ*L7ES<Ga@x+Rv4YK6hzcX1I6(WX^W$An=-F6f2ie9eneKP9o>pz9)YbnXFo
z`S$uY_2#&uqR~XCI~&TbJgq2Wi+UE`J`RAezD57*PBeSA*HGulz@jaG4zsilvoesu
zzUBVw3SM2p#G70OriOLh^Iw11*ox~Ox|1-W^7N#<X4Q?dk<?$al(5R_e&H2|9rjQr
zMMjtwY$B(YRk?NnL5Letjy4-aRaY^!t=@lH5Tv_05=rvxe=p%>v2no$n!Sfmc@MIC
z_|#7g36YiZ+^)K@+=nu&-%7%Fq?&FS%}!0~wut7{%Sa_DZg#%8M?Wt7$%fnJ`PjJE
z%3EHUsic>=_tCv&NSvf>=<z6|_X{6!eXKaQO3dv-#@cdeJxc55f367Nn`@B&8~m&5
zyT86)uz}1Lz0zTe6wc)*L(Xvbws30Dx&2;@hKY#@+9@K&xIGVSrJw=#jpQt?DKlB6
z>O$P2A>O7x<ozXl)q11y{K83Gf((9*B|wvk(Vw|sJ>^$HWdhFJa}}(55>jd=Ywn2F
z!O@<=o}9E?(Os0EMScDocS8kV9$wz!jEcvi{E)jzcZJks4lM=j*w9C6Ta654Ecuj0
z3SmJ^uOgM13BIzixt1#DEogGR+jUQBc$vidMDh%|wY(GFfC|r0%c)?K6)4&d&#Hzz
zo`k%78@o8c<2S>$dk9;Jqfln?k}nsO&>I?STc-STf`5FajOTe&=<h!iJm9Rvj>lpI
z6~Dhv`Yx=_>9-GDECcywD8UH8EBnke20<Gh^Qc;q;?%;zn}`S^Ho|u8n~&ctL9sW>
zDyG@D%(D$IG(8{?19E~he`YOW|GdhPPwqd9sl0_?C+B`cb^~)&mXx{CsL#$+o2CD^
z_&RyTqIDwXoKd+KY@jpp?p27)Rq+Q!h@tiA^(o0GQ?%rF=LgD}I*a<8UC#1E$`;t{
z0YtTuV&mP>VFE67dz=tVjPyW7xP$fod8~WjYam$o{OQxf$H?QC1y3MBH?$Iy55Ovn
z!^;Ovt#NXqV(K?A9~NCL(zq``eb>RIO4zle1YnA9nmE{hZ(8vN`&VV-Jt+;)<azYJ
z9pm)tJ4rdg=Js};^E2vKIqyc|C}rx8E)si7`&e6d77(%$N>>NWQzw0K_ABj7OWwIg
zG=N8hiD2rMxZLdUF1>U{bdky_ZS3x@G)1crW_2{)Xh7_TK1ArhiGTFvz1Vlz)%lID
zA86~vK1YVX>hB_9kDGhtNcM35jK<cvV3L`aI_u)@*43=|c%R;0Ww_>}fx>-iP#ySJ
zzO}Xr;?3cT48)1AoCG7gPz9@^@BSBkWRnieLtYUX%|xX3eiN30RL}%h@B}WULC-gu
zjfuPscY%$ZOz*}&cU)O-1_u(0Osbx7xXO<)fNerq2?~H4pA<L|un<-}er%=eZ^)gg
z`|?93#q;y1UqW~lFVh2DgiM)UIjL8OK4A)Jst0Y_h2B(r5ai0Y=zCXPjXb<?@nXw|
z<Bd8zluj}<4L{B<vfZ_TtuD4NadF{%@QU%zd;q4IM%*rY7bkRsZ$tZ1hD?1;+gpji
zxax++inE)9-(OpClB$;PTo@a<T(e02#rr0AvQL-tCaybvhTKb%<VjEABB!-_qVxT0
z%wap>|3X5Z!rGUDb4VqxZ%N)7H8nTZI(Jd0Wv#YYc}|Gkdd%#a8lK&;p~%hNb#(cm
zoT-2y*R8MD+sTjfzA5o&sjCx1))ETFNebj;)E)C;$_yN?kibE8l|r9+mb2BrHhpyc
zlY`A2s7<gb!hv}3<SSKxnk5zYnTW?e#kf()xgBy6@<U?(QnLk9!iq-tLkx|VZ(iPh
zr={(BsRo02NkMxOP4q^42Zn6u>_cszyXZr?`{VhY=wmN}C#aGLtgv5%1PjlemS#D`
zHqN=N9Y=G@G$p%K?zi-W3LO{k^hhxA%Uj&K?XQ`5Ex_@Cj%Elg#j`M473=F+S`2s!
z&n_Wdi0BZ(^m=x~%tS9jFWsB(ALua5=IQ>dl=<nqozE8plM=1Wk+gh%>Us6<$?*!#
z=fCUnP0-NLe5R6;lNPG*T)Mu#u>7O<^9TRvDUvpx60no)e~sXdR;p^WR^QPaPrmbB
z-f-Tu!|A?=4wuCVo&|wJMRXc_Nk&FSxm<fh08%pMhP+Qv9&Z|R9(mz5Weo?N6Jl7W
z#comOGNFkQ0E@K<=z?%8I=fHH_XeMIm1HG+)>Hi~?(<^B$n={@DsK<l)8*514KC{?
zo1F6huG6-la!1qtHP9lZC&=*_PX~$BJ+qGrw;I0NnEmsxox{U5#ba63{cyajw^SF=
z;Zeqd@f3&GO?6^sk7f&v^rtAUruTXc%5C8mAKamci)UVPJvU7w9+G6KBFWn97|SXW
z(Mww5BGdknhCW{+IH^-)9qWCvDK4Q1UDx@bPvYk4`rC?|h<KwP<dW81H>Vp&?;E^I
zm%89{0pk@$`aX-xVkwbzPYqX@BVxaQBbM(oV_=c!BNdhulM(kc`Sf>XEtez{4SVP{
zH~NVi=;<xGUd!;`B1?XiXR1QYD9^>BrW!%mc8m1-<k|ytc=-s8?Rhy(VdVA6W6PA|
z9KMKaE*%3oBlnW%DSLkCO>~Oe2&F|ZuroGD>ex)xE<E<(_#c+QADOkVb{VIlu|KrW
zht$;{imL5)fvt_=gIgfW<<<z*5<4e8G+POi^kj=hPQ`Qqf%KUTvR-RjxdP(rPl^NW
zsqgZtUlefZ{;g%x{{{2%Vk-KnlS3nMtq#0bC<|`DlOFo2+Y%NojmKNKS6KFlyLc+T
zSLZOk&^!&`&er*ZEZt;hrlS*8O;-VBV4|@E3(&C5)y#CC^u=U5IBIb3=e8P@ho1t~
z1HbD={A(PtA3!VHVo|UZBy#p!fSrRw2@1=VQJm&?;PzTzJ*Eka1zFG=y3@oE_uHbt
z0y5Z#zu2wTw&#VMSH8T4UYy^*pB)|^rh*GNT1AQ`uqxt27i-F#mQ`_}K7vEa`6(zU
z$YLNX6EZ>;P(nUzZ)<Muh6?R27^!sK3iqw;Z6zb4#I>q(?-!%RKRfx}*UqxIcE}Bq
z@mR;WXb=8i{W`yU92gkr08@9ZH9Rbn23c2{{;VYVR)i()wg`p_w#w5M{@erC5^Lm6
zz!Qc-s{XgJ0Z#PX3SV^z{gu;GNLds9sNL^$+J;F)Lz8UyR-xr++hOAWibv_goR=i$
z>wux}CIdm;xgC_9-Fj83J^eb`+Qw4=&hIU<(SvSTDUe~2huw%JXalwOW_W*4UI7vr
zm`@W^QxN}^(nC%QUgJ_A?lmvCqh*3aV`4I(`Nk5;3vNZwKPtQ2pV^Zf=bZ*Rz|1Tx
z>D6-cMBFZG<&d53hevJC1=|ZZR6!m60||#1u;DUqeF-)y8reOV39_87e@o=)1&jqz
zf^NBxLdt-vXNvu7vs;nSZm-f|yNKh$C%9Xs!5z>70aYLOy&+TVn2syjlpB=5=Dt4I
zlJ1HQ6hE&6)6w;xUJaPeckI@Faqd;|K&g9sY;Dn4p_fahcd?61CF%LoxAPJrA!${@
zvB@}j5z8x0C#NpP518(h7V--^#xA*7e$8eQK}}-3m3q+$k%XdAi??irOQ2a7h&70h
zyCM-iP>D~2Qk!{qq6Eg}a?|#hF36~7K+(4VIMxSe@F@6|fwQR_f*zF;`&n#4!*QVF
zA1-l7Cgn7XAIdXL_dY$c=Oxz!ISklq2`=Mq$@TloK=aT^6D0T@RIt+(lO&se|L%ke
zz9SU(mrCZp_Mm<=IXQU;3EJ=1vN4A#gH4Y@%%c_%<I*KY0Rb(@Eq8*v^On=C6qWRF
z=-0`z%qkHhU{?M#WYrHJTmUpdV{Y@Yq+AxEP%K6sv1riPxmA;qlcQ$*S@R*w!!@k2
zT1_q^6O%a#aWS#dQgXD*(>Ve=^M6V-k7~4sUN|F7)6@kx;TFe|zJyGCTKu6uR-7b;
z@2}gee%UY5oaoXJ(j#WfOLAp56(F+ej4>5a(o}qB6x`dXXF+^hCFdE5^WxLj`D_va
z1M7V|;`FI=D!YB>V*!8km{J_q1piKay)5|<iv-xu=2W}#Q<mnc$B8#(SJh0!qB3-a
zku1jSX9flae^ysBA){yp><@6f$|m=qASxM5>9O)ts_-f6GmtLlGHwp_jLQ{)Lku*6
zOo0vG0ZhWoxDdBrLBa3Xxi#C;7yqb<C$<cd<y>xCf6CAI_V)UP_D2&YkaHb4$33WD
z0q1<@Ls)2NXwkIG?oyuxs{TFxf-q1N%BX#?#M?UiK<W19SW{m={%CJC9gc|#5Eyks
zA7mmZiW2Ovg<WM^uNi^FSozHNd~IH&0j>$}(_7H>ukTq!n~3avBdp>Pcb1cI36SE@
zlE~)zw3bbclfsPi?vrjt#P%*VSJFed%-BM8z42{XUlzm*dwYPP%JPUN#J+aSl0`!A
zsnj~x=QDh!F-4hD7Hv`m@u+@9_mWEGI&n<Z)n0~EJ+;xVhDp2EQFDtX$;c$E8Z6oC
zu|H9Mv5t*++kSIrr}q=}6WY}3vwWxp71A!kt(Udy0IFteIsQxEMfpx&Lw3sjQE@bC
z<03?Vyc%{#9XbdOc417C!Fwx1N`HP5-u*pv0vsRXM6b%)%|;3{6JJs}B-pxDk5nP&
z<ulu4TtsuInuS2+YT<~s4lbJC7ar%=g{OtGX)AOU6SNGIyIe$`xqVMcQ{Qoxq~4lM
zeq!mee4GrbENjJo1ho*|ghQM5;BrAE@w`7otiLHT?>4=}>XSqXnItv>r@5yuD=AKV
zyKja_d5fbw92(pFazq{?sd_|`+}JGHLQ>bbzO~*IT;`@}{Uv%C7H&A_xt~Q}8sz9w
zZhiodJa(a~(rw#BqMK_#TWO(SK*vSr?ObRg{0fW|^d}`0Ei5cPHzJxq&_ufkozHB}
z6mZAvfr5E2(0xG57x<WFi-Nh#(D9#!;y)YjvVeWsTON=bA0H1+1YRN*xCWYjzYzX0
zTm9#q>;Um+&)z`a%iNMZ8gcbnLv+DWpueI|@{TG^YPd}o+qLcBaWh7h-TrnFZi_J<
zDK>X^d+p#k<o(fIhGMVEtls<iX4|fA&pN~;C?1E}^2%0~3uc-P{2P4WI~DsqQwjc(
zmQ*{dAreVpFL~YZ$}Mb0cRh7AAaeYcrkV+nwJH<$UK~L^a8X6j3T?M?&{2+96c8Tl
z9vrwzoL|bqxQfkk*ld<pGTESokx5?}I%<uM%-h7P-a-+`kn&h(00W98oD@k3iKda?
ztS;L{+dRG;q!Ig*yW876z*dtUZCxsH_3G7k@7{HT><!!Cp*0jJk6`w8P?#LA)e)O2
z*1aEAj3q?-Do!p}jbA^(O-lOFv@}cC@mSYU-hBl(K=Y%U(7Xs`HmeHj-10*tK6ytw
zl+!A(U;0bBIqOl=L}OXCQb6T-Y;O01ByW{OXaDnQ=1&scJqq3NIK%udAAab%By0Aw
z4(ZAKotvz_L?@!Qs0U)=+@~3@bzX<N#^4J?FbZ&}u=0<~e@E-qpbwFt5&zf=#n)7n
znTPx=SZg;Hcd1<a-4o;|b}-orJw0L0;Q^nhd7V3-a*F&T#m%Lm0?P=oAWBvs<?`BY
z<9YY?tpY4?t4W?TeMoL@E*4JJ2b9-OilHPe$t04E<xhI#cUkB9#;^K%?dxk04mK`<
z={FC(5dZ$za@qOC5b`i>o<T>ao}l1dbwY26IIXyz_B01E5cN$?G%;G(#z6jL{HmZS
zH`+dkN)kwu3Kl*(90>mKcaXLvzq(@oP!XpkXnCtm|9nA5X}WykLMO3Wy5pc?-Vi=g
zy90cT!L%<XDs`Yps0QU3m7fNp5gh~h=1LOXVEC|wOaWcR7FR?Il0{4ssPa;vXWMiG
z5=bY33Ye@*7wJ{flqJ4?{p#lHvW8_s!xHAB7JM0xG6V2Ki&<rB%sVkq6|#iVv^%uM
zOQl~!Gai*=QJW=qOb^YIEKDp_38rn+DJiXvCOWH7SR{#59MNa`{T7eiG3$0@Vo7@c
z@2O(P4n5Riez>@uuim#lo;+)d3>GYnH2%pUI-l7K;BM{eVy$Azu#j;oU29iN$u&dm
zYiO=sgpgyn0=hzvS2!fP;fP8aS3<&GDi~~=5FNw(go6oSfQ|r3CuO`gUAi}HNhbkT
zzoO7)LO0RUem$hYfN5h7c6ZqiULNKG7#NgfcYvCi8KI$h0?G4%T;rB}qo$jygFm07
zrlv*-c@#nfs1rqp#yG>hG8W;`IodS0JD)Z2ML_rYHm8IASN*_Si|P2Ao9|?#?XB;k
zM&;A8L&+=MXVKg9^96Rk$=tJ4;z0YGAFaB(<?IQW#`o)8H8R#-(D2*;c_x86J)Gv<
zv=YeqtLQ93O3ExSzRVEIqI?Z}30k8NO%-^by6brD_9Z>AkpN)aa+sR#XIh9##2^Z5
zA4zO8$wN(yJEZ#A&89j*c|n3NXb~c0_)|KGWVyJ006|9EJQ1L=cf%Whsh+9Ym!Y8q
zmBe!L9`t*k@ULFw!$b&TR$OLG!#!L1b$^Sq$j-GZbzyipF>54UOy$2pmEBAoQMD+H
z8s2RYb{jpR{WWqSuXUAS{&StASiaOf7n$%rhLnv!a$E}a^Vo11lEjSnM!Qry8N*Ji
z$l*J-r?6k>B&fW0KGQA&4QO~!kX&lpKFoX?RKJgo*O4j}2vf1G)Agz<LD(q%Ej~kY
zU4rt!JHjSd&1}@(gM-@zq+wWsrch{iYrKE|{!(vh2*3i;78Vvf=VhZfC7uL?m)kgK
zOm?p~!=WpHu(U0bLmWhc`oT`?@Co|ZGiI~+<*nk92e06|eZlh9XtAWXaj_<qK!RD5
z`*Sw}aZp3t2wM8QySsGGs_dU*yQ;+{lPA{K^(#(Kt@~{(QA|Xbw<>0TTYvUUsYTcM
zWkwbnjUvH~musY3rP0h)pQ@_PF{Wn0^-4NbLhuOSIhmPXACy4j(wIBC56&SfQU+wJ
zJP>-(o#r(~3cxjS0Rg05KU`>>CcO5U1O>HC@-(<CdOQ8_gi~Ol{zlF7(bDv%a9-hx
z6AVWBE*>79O17>JDkcR^+8wx*lx>a|zm`(nupZ2LR5a!s4M@(&^+I^+GEm?)Y2o6a
z&8KTpD@X5BeVTsYFiaKNJ7mRFmz6e!VT1Il(s7!q$u}5>dVbWlizNDZ6*Uu#92Pug
zsdjpqee?KY7ERfD=BNAYX7KO#G(KZoHFLA9NP0d8!>sL}474=Wgs+!{gk@jwk1fnK
zvvUGF;hnfL+o{^eHX{`tE}wsGY;AP|vMwq4t2^_=i<0P!z7&Oo-wt{f<7;C`8Bu9b
zTf>@M794YPi2fo3<$?u;gt&m;@3EhK!urpdovl1z*=FJAFRiv!Yc{X<>{C%&1R-!+
zXJ)in`DpiDd)zbU+_=?}Wvzj1kGF2^ygVJL(u=lHcj{`D?{_wNSE5Gbqj-Z{QzNKi
z%b#9)A|ra8EGfz<y2%LX=#C623w54*_409@qf?NtyG1LHJ2Kb*fM_k84x+{W@1lW5
z1)AFhHV=m;d$Go$atJ6)_Vp?+FRxvwN~eL@hnbE5Kw2UkOUYm$YfJ$T+%hpSQSuHw
zl!d3TN-|sSST%}@0m4AHFKk{4kLG4M<*UBw)Wi`Y3%wa{PrN)s%h~QTB(j>bj~|}5
zh79ceHT&}PeJNw=E;tkkWv@SeuA7qAT+(oJJ8rhBygQ;uBBre!LoXYX0xys%_JMS3
zXGhuHy%ehR83Jp?Pl7oC=M>py!lU-?C2zq?7vPHoxBfXaZwA7d#vO3kOO?lKJqUzO
zPCE+<eVNgDQEVI>dFoOv=-MQjeIFFdUgISCnY&TH`1yjDCR0NBL(aWj^NFl#rSiLk
zr}M9uldJ-EG_D@|F_=hjd~B&R>Zb9sW*j0$O*UE)B)qqFv1vzHea5T@(z7em_??!1
z8~prj^q{G+QM|%^SMl-VmswS3h`Z3<py3)DzP~N=WUP>f^Y~_OsuIC1$EmDWfFUE=
zsr|xb8P3sz&KLlWcieTOGG;~sGs!EpfoxLw$Bvk+UYr`{7%RKfl$U25CGSIO^fSIk
z_I7oD;N^y`ndr3ec9v2v-<WDBqeQ)z&x16$hz=LqC7==s5UwH3b#+hRN+n<AeYykD
zi#ec5Gsao#Vz68XGPNRHM!^_g=zxo$%=jiOb8IB;Lp~>`kYDbLUj=4==QQoKI-5J(
zIQ~QVEbQwjbA##PIOUj=_&__6kPzqXhdp9KG<YEr?xt~&i%Vtn>g_O*Vqf{h9jlNx
zT)??Dg&BybQlZ`p?Q&O!21mvD*)eJ_S#d|B0t)cUvH<f=6n#!IuUzc`pw)6iJ-vww
z9h+NQ!2#m!oWSqcE{z#a?s4{Ugfp=N$DY<$f~H7neKkVX<d&zsVmRmQC+WJvXABQ|
zi=GI`PzvWitFp~{L69jYE=Vnhhi5+RjVzX|cDMg|={J6@Y)2|%r<2~?E10zZO(!0E
z6{Coa&HUP5--hNB$lK>wKXt%xqeMN&fIHf_Xe2fg>mlF#^9PHk-_8|!G`V(n8kHRP
z=0mzW7?K!8Qp2-vtuVzFNCb2zt~LeNQNx98ey!w5$JqUQrYhDq#>OhrE!0wx?;aDB
z*%y~(C6|K}n{^cv(3tA)7!Nzt&dQkDKh~8%rIyYqJ(PX@Pwr#?ovCVy1!kOhX}5)=
z<zDt5e&dn5u|L=~()>IOl7yIuV(<xXW}Mxp*x9MpQl`mJd^`8ny!+XR@Tol2`Bt=W
zT+zd2m-VJ1gTa9jEg?fq3*=8GPx=eLBf3htzfVU?^7kQn7^!rhQQnY}IB{UTO0Dc8
zWnT^c&v#Oiie?Xe?l7w;Cq#X2(EO=IoSvNgdavZsljG5N!{nKQ-D-4-Kmj@`c6L!^
zTqQ?k3OB9H_6h30&-GE9&h^=U%kcI0a>22UhK7bHzg+OIASz@%Ef|eBrpZY0bi=j}
z508DfBu(XtO2b3M*wk36d6B9e_uJnG#Ai5YR0}+>y%Z56Uj1(-(9lGc4iq*2y*Jv!
z$T=`i#l^g8eSeJye4CLz`NDX7{O$kyooB(;7PD(uXfyGm7ytbkf<}A`o&CQ@{Qvzj
zG`S2L&4=YJSSW49VA*<x-_}LxlK%4k7?@xe>|M<NKEK-MiYxl$gF)woPiO1NdYtpw
zVf;4{m!+2@%mM$0sjm);s_Whc1$-2cN02U+k`|CI6#<9t?(WVZ4hEn|NH>Ud4BbO_
z_t2eELpKB8LErcNeb=S_<D7GLt+np^UTg3D<#&La?TN#lo|PtVk1XN6KM&BnSj6;G
za^Yc>l}9afUhbS-lnPSt>jxi)x70a>%A0Pje#G#(+}^tc?<8^E=BDI67=E7jFUexV
zuRjCeXZY{_y*G^fB;@%bw?3NMcob3Knbq3)5Md@rS`4<lOiu(+G7ytRi3p|xtojS*
z5{`PLtu$y_g#G^xPOCvAH`+@~2qR<dO%1Vg%iQ|MkujnXTf|`?JmRTIxLw{0pYeBF
z|NcUC%I8ct-y*udwDs!q{b~EFr}y;0XYu6Ya*_kXKEOU_mf&XoeBo`-GQm=>qQ9l2
zD7h@RIZy)FcZbi6|DCX(Xc%>9i$;L$OrfT9EQAm?;;a~vo%@?Okycn!J7|A+IVAD;
zD33c+jp6S%M8I!2lsIykZ&1uqri^-3Ro4k+JaD?)YY=vh!Djy3i#1$ueybI<B(m0F
z+NlI1ZuuIAAEyJQ(uhzMu5e&EB0@*~*`N2lR0k)h$)5)~lU*L%mlT31kgID;psjcm
zGq*I9w=1aoA7AxSUuu8K@Pm?>Sr&Lu)|{@yH%}AO8u%B^*0dzq0-LERr^#KrLM)_7
z73v5zSUF|F-h8yQ-#wZ1Id6V|SfyP$!iAPiPV>4?&+Qa&CZ9c+)SuefuywJhnu?}N
z!Ba1tUBCt{gfo`}g-}E$H9ll&WOxQ~OO(~L3Dqt%)mcW}UKC-*$H$jW*<|V}g`0wV
zNj$j}2A=H4D7d**j*O1pwMnzIY)<!Gym8r#Da~R#KPYn2T`*`$A-6vvFzv4gVQ6x4
z(#k}DXqjEBahQ%<pqN_<-jAqcUAOu2R@AvLIzL5SZ1p`8-Mg=$TY-HAYrBsE-z4@+
zG+{?MW7N^DZg1y<_b|A}DjM-nQ@h71nDMPDXjvr`sO1awo4Lo{=k4k1QE^inNPNwn
zuey#eME!y(aE&6#t=TH&Tn3^P(i6Sb3k>d<#)gMKha4UrIyuuhFeLWJoayWzzwWgv
zsjYcYtz6lk)fbF&aqdy}>FiN#OiBuI%)TI5uBBzE21}{O;c2Z9m;N#e0L6^?Y;P0_
z)jQVsyuPk(b?I%O+)tqu1`0Ht9N+gI1I?9X5oG}FLJSMU>W3ixp9yaQ_ulQ_^TYO6
zFUTQMbj=_j9VuLmz;h82n~pE|Dn*DWSCdtbe(>kFJ*0mY+gst<4GM3^jMdNHjv5gz
z_F+iVJb?J`^3lFqKJ8SsHv-~nk0Zxvr9_iiZqTIVq=6(Q3j@ED>YCJTY>Yp@UBkLt
zlyR}`z^}nnUncX>2FQbq3rF!r5AiygmD3&*PRkx;`IlXTMPplkpXj*rdYOw`^dTq1
z1g&dJKaNlJENAsLRm_%%!LS#J_aWWAJKS($<}>vW?(Vc+q0?U{@Yb^y4tl~OWKCsa
zUO>QBXo)?Op<Yt?e$*J}%MAdCu5*$TNM^JZqJhaqaqb+RoQ{k=k>6Sm<v_CjfedIQ
z_rSJB{j$~;x84{FJ2?&3yj+lPeCg@>t6YF%!Ba#}ZEhkvFr{&YNA>YTH?hpWL;DpR
zjGs~dSQr*N_s=Ykh21STvsksNyuw0LMOfjjX?cY25Or>>tgg1h@4#2M$KGY3(9m!K
z{D{%m2-v@iPgDktSHOPF>HhtXmkg68xK$-u=<WAo_qEX90X?I0nAJ)~M!&!_A9FzU
z(h|J!L+Zl^1qFp42|TuUx_us4Rf}YM^5?y1XgXQ57=P4lP!f&Y+jewx)T(w!1Ghxw
zfM>f}jpZnECidIq{FN2KU+06r29=XLwl$LZ9b{bT;O;7rv4B?>D}$Be2moZT<<Ju=
zAF!CF7R(vV>9YOjLwvB=<3-$m{u$yL6iNr*o%hcF5%_LS5D^|8DuH6Dp2rd<WCT>h
zP0FTbby)xVSP}Te*pXl2*pLV?7YFLwb2@)m8TkKvdqEF9cHO+@OJZYV$(iqF7PxzC
z;w7)J2fHaKcSK}>`@1gA&vlMx`}$<)7#N5>5?hzLu5ORSjME*vmXP>8IB-<z9%EQ<
zbid<#)xvXMxk4AuqB9}9LW>RHjPfz)cxq=SFyV90!<fzDG1&YkFh4|iuqz^Uw@{5M
zb-V<otn7S^JZNj9girR(%m6bha1+iSAyEJDnvhw|qV4MwVqpd_mj!NjVapw)tWz4D
z1szaibbM}sjJx{B(idDbt#foggFmdC;BQQpmPN-T?;<%BuuL+t;c$$o=hrEIw5Fz}
zjLgivbEZF^nm%1Y0fyKevvf2xpX<N+ITeE0$>Nz^GO+L#A0NPX0rxiDu@5BMAIb&}
zht4mZqD5Ce3XmlxjTh+`fwx8qpD%tEuVs7r@&l`OCCl+C$X$AEczTUWs`JKr2R;uH
z>NVL_^{y^GkA!1$_}L`NGFR=5jj(FF?2K6^L!^~89E!J8V82peT6t~=CL~))+s<Q2
zz0`maz(XV{L5c9ke=~`)vZJ3Gc`Z;W^#b8Pp26n|nbMwHspZE&%Y)RuHa31Y|H8Gi
zk%CP{uSW`wJ9WwHcn=YsC5i|NH^JM`7R+0-#S64~X&q9wlbZ#zvNcp<vkgyd?gSRd
zcH_@^aTE%pXZ$9X9DXOe&kuuQ%@_d69U&C_rCsb6!^6YiS&U(iD67CtG=~QV;O4;}
z%+d`7HuLjHwzNJ6@Z9dP*P1Luo(cU(w(LHs!M?Fu-%D|O+0@+XjEsAOq!$NID{K|?
zix%9&YRfHr<C$Ib%+#_<%bY}fZ7VcP+=h>qe&>!tcem}2r|K?xdf3j>6#+9mJlvgG
zNcvt)@6dBk&j`mt*$3p#dQ&z$+VjrMqoXYO?OW?Xi*i#uiJV-Vt82S-1d0QMtgOPq
zMcJ9Ww}lPMnOT|Lqoc}1gHBhcWD$L_9=E9p33*`U1|$O9s2P9B3+};TRsP6@Ho+X*
z_|XYmJJK<CmJa57p0WdGegtlm{CPL&`iRh0NiZct$N8>)YfMGuL-9QrQL>Ee=mLY$
zNo!%fR=sg{q0hSK+{BTV&nug+bvSh+bA%1rl{#njs6wTYQ9^@?t#t|4yB9q9h|UQS
zh)McI7a00Gd+zGB?UGTyR;^5A03O0exX9Ig^Umzhn$UUx-6k*JYH~_S*hVTaZ<@dF
z$2`2?b-Fd3ovLgk6(k@TV68JDXQh<;uHne8kUCRcpkktr5`J>rBFcSob}{XBUG3|M
zBJ|WQ>Lav|+~`B>Lv9%$WaG8|5*7C83aQum-bkC-H|DjbTXU-0C*wU8(BH{tE~{gV
zsI!3?c6XrrZ%Yh%5T$B?t!a1CHMj-CYQC;U&-*~UMYri80|7o_FH=1_F5;IVO#AnS
zwms9U+5R;{PD_JUsFvg48*c3UoXo_^-{EH>U046CM4QY{pFq720}nY9x$75Bpm6KX
zd#!Ui;O=I<tw}j&``vuBd{uVfpdyQ14}g*~WBqA@go@q0`B?||&5=wiu-pa|jb*E`
z`7GN4Q-62mDw9nXr_66nE3UsdEh(g0ETYDm&0)Iny>x1RhpIKRWmMOyypw;M-RZlK
z4!?m3SJj1D*C21(*sgzke2wMfX>mPo@3*H4YeW72qT*I&URJVDDnK|dQ`%V#ZaY;f
zejf|x<?$?d-GM%MU~O#v>!u%-JEA}BiCEPzsi|dmrZOkByH$?PCcYpiepgca4j(Ct
zs2B0TisF*BZ_h>t1>6Mme745NDMQGgUR`c)kB$8_3*mMhRjU^^j_D9_&}UTn>tT96
zJAo*~8%*l+Bm%IZTitKe#1DeD^F@gu5h)hruqOP|p<V5w&;)<gv31mga?1r@^%(Um
zAC6l*$Or(60By!X)T3(j5Pen+TEQaCKg0L)-!(H#dk>H5^}hHOB&VO#O023iEd;eK
z8SQ}+nc;hqUe0(LwO*(NTX(gB=hT*N8`u<7QUQE9wQ0>(TDeBwIv?WgtIt@emTyfy
zw4H^^4=ak>|IM7E6;iA@VX~ubOj17H6@aRwiOlCwQml??sH^6`fK2<W#xYAVAGuNU
zGeNABD}qLi*F@al10jEVyKfJ|@k0n-VgcwDRj<91sqznNIOJ$)`GjMhBmpuXqG*YQ
z-Fzc5Rt^s2!Z81IriTGUNf+l(Pe_DPX=ySQGaZ+7#OlWG+OdYj;&$@Yq$$6!nv_}=
zEtu!%+${O?%$tk%P1w(<soC@2U87uuz%0bnfPdW;{@{m*Jk^}Vppl-stqtTxLmO>&
z=XC1Xq`!bL(ck#@G8F_FPjkRRk)HCY^3|ay*&P=?3+P}SikYmAg6h;not*8(mB;pA
zYiE*HyR|dlYx~>h@xnr54T@sKYXAIgF#*_O%&nQUWZKQ!aeKA{5{3xnW6*6JmTHWp
zt4ga!t7Um=vTPx;LPpBUPTJRZ!Q{hf=o}!B9Fr9)oA|PAVZr||Ve>ou0&?e%SZ>G+
zGQ;TOnTGobRksG;Iy?+F^`&|T#%Xv@)rM+tTMIGd@hF}_$jXc=<TA4euP(Xb<Dr{_
z(GeNu_VD)1iod|^@Q#Rotb?i5A1tCGGQ5CGUHA&YcUKhH$`N&KpfX=c;YUjS?)b>d
zlD?qOymYU^-K^@d*SOb9#F^$Zpb5py6<%!&jNY`->Mm4<)K&$vwM+ed_wbYbdoRae
z6E0)>zCKYuQqrs&QgX!~;OQSG*uZt!Sy9sREE+GqXKR+}dFv5=zQ-<BVcOd!Nv(J!
zP*HK+#i4uE57<x&Cwg~VgFV>nh!?d&rl5lI)KpW2QLCmoPyY-p2`N~!^c}jL_TY8-
zAa0Pnw0hB|=pO?tZIz6+t<rbUGm;2sn6DKZR&E$iWS(GZ>Xa<ghrZ>IC9BDGIy*c%
z9Um-eEU3iYvhF{Kv^qS9z1$lioF+{Du{R5#zTJ&r2$|2dKM}RiuMTdSlkLfgyP}|g
zp1-{e_qc<;ris~81!n;_lD1@HqdD?fTY<V>cVXetVQHS3!jF4(1wqoXi1{xWbwgF)
zN=gv&&H?ow2j>%{Djn^6>;<ZPV$mxl>g)5!dA%Ar`y!oetDBogc))QiWTa&EDW!<B
zT7=2z{t#FlcDycj#@y7|S>OJbdWt<;#8jYlciqnO1r?d8I?PMQcKu7|BO=@2kY0HG
zI|IK$`xyy$m)&1TR&7uktSOl;ZJn_~XoDDjDA8_=q47G4AApUF-+$oBa*-N?dy=O<
z0fPIjkK0~Ap07tWdH!FslMrjd{@Hb1SI-R&G}-h9*3^z$J00Doj8q;|AO`c$M6ANu
z3s4R|-umfv_ctm$Iuo?o(izG;GBPI6^5DIHTZ|<RzUX})6;&QS{x7EI&yj5H4xrqL
zDA1XjGyOBEeolUO2>m_)#+=5p9fUmCzUeu*tB}ZBH@L2<m>A-hobB9DA*|49@(oqN
zr-2O_tGxLuq-6h%b9`uu2yE;a2LvLIj*nD9>H2knKDDdmeBYdD3?w^Z;P-}VeuV)F
zi!wMwb$QCf9}6qTY^`T&xHc17VGH{g0Ni8$_$G&Y=m5w`O4qb2H{<2AEY?mbFYIQ(
ze7~<gT+UuDH5YW1);p-rW9hEayegWfCT8I72ls^uu)NPyu<E0JVQ>LUW=6L4fIc%f
zD8mHhs6bLfJ#=)nU%M+;m`cBMDxE(!3lKR!U@$c?U4#2f#lSAN!u;<RBmwV2Mz&o|
zJP?Aq8;_=+e|Qcb#p8OOUG4ho_KAoLEW&rLk}H>6_@rrMfAB;?(yvSqPbAt3Rv7My
z=-ys`ht}Q7LRXrvRxNqpXOF9gn<GnVttM9lqFp2+#?sX??Ck!{<WbGf2}&&j{b;%9
zL9p)v@J8QmKCRW!aqD6S8wwn*`$CV>uI6;=z3!YoN}@4Oti<u~^!P4%xrnLQa_5oc
z8Q8P8XJ=YV&sz3R)ys2L?TuLL1NrgNF`HMhxB1W|#GJ0kT{nKl{pp4f&xM$53YU5V
zk$nB=F4^2%d0OACRr9~=8~ML`qF>j^mYeNP!VHgyEL#9~(cd24bq!FrRl*v_xr1{L
zZr^=T5L>F$5B=-o?c=Vi3hTk#_;TsN{NmM{WGj`4oKxFBFvfaKD<t(K89yTOj>*oc
z)O*1e>Lqn0gz}?~UFKAZLgzQ<A&48qr{`lc+KSJcYjgD|>pf0X$$0F;*Y`v-;yP1C
zQVD>L+_&$m6Z8JC^QbR@cM9C$?-hOD(N%tD)#gsg?|a}mMhV<K#K@FE+RYd&*NPpa
zHX(>WHowPih1i?xHGJy*2gK=>V$|wTvn->Qe9O(bs#R?F+x`WsVdms%Ypn%6&P17n
z%NZfri<7fn700P3iE%j~;8d1^)&ePZ6Iu(iE<OzoXA4$Xm|KD9KCk&|bHQ%ifv%{i
z2&Z_aKLDxvjz3pV=4Nm~FVp2sdE#HAvh=Ko)JnQ0B>XX~Py2bhI!Bd@e7R*G$fdBf
zOZBG2Hk@m9uB@nXy;R`3I#DZ*RSz%P)fbhrciVj#n9+H27G8P_AgZeB$WiSR0!~{|
zcD5Sw?Cw@YzY0-2i89ctEAA~#K_z$?)#wbDS+rI1-4vrufN~-lv6uO{yFg-0)>R_!
zIO0kBp5?SqK?IdG0jzJwrvB!nDpSF5>aLItxwA(HS*dkZ$_aBX=_DB`jn`6psH`Os
zAmPJ_K+weinqX_kvKwy-->~P;@6P9TE!nY^b!Xr~UBVGI<6aA>xQVMj$q<E8FxvQG
z^HV&Q)vr^DICg=L(@||4R1TLW^k3RdJu)>lhv(IC?rtEgqf+8?I|!L`+*M%sAxnmw
zYr;sb7#3GkPt$4lUbQZ>s_o`=S(WC_?DxjNdkn1Kp&cjDG4Z)I`D*a$cXxe!WBL59
zlhcmpUmtgLFx5KIiDO_zc_bOIv%3c66lO#{h0T}r`V+?JD*GlgK(>^t+)C7crtD2?
zaj&jfj+cxcsLLb<EKC$cMjcJ^r`yj7L}V<C7r55`5saGF-hv_wo6F*Rxj|so%rH8M
ziW%H0cN58wMAiF`(-kf+tHj{)r*ln~phu7I3ThjSJ9QPc;$yGsoKH)7y#|u2OQE87
zp4Kn?&zK|&M2^DCf7u^gb>3);FC>Lcchc#Rq7)*PlxQJ@(vwDu-Zf!!$n?&;_sb>o
z)7Z<GU9%8ncXayzvvQ-K{|HYwt)LQu=8!zu_8v4OhK=-1o;#%{4+2ZbQ_U=u-uNah
zGctiqY&j85BL$+leW~cNEMMW1JrWvuycN`3msy!*&)(6=_7=AQg}YRPTK=z={LO<K
z)L6M%zflNS9?cE`ZxCnMKPBj9yu30=lFP{GLIMt;b4mq~E=zGTcqgvJKpgr_dtP(S
z^&lqv8!4$=SoGB-!l7BpisqlwqtWK5rywR--L+&p;HwGK-X}av`ZElNU@&>b2ZoX<
zfNkQ-p6^f|rQn(a-(m9T%hCX>Gh}9Sy;t{^<{|af;0Ei^k+SGS_c_K|syL??;PFx*
zFSNnK?8}zFI{Sf8Y;VH~!`ODst<^%w$^~-p(01nJqD~Zau-k=px+hC3VPM+O#eqnX
zaf1daDA&2+(~oqpxUAGosPj!9r(i!%-|KH-vU2#IRa>P=qdDNeDk#sO^UBOqz1~RN
zR?y=E_>3SznrGB(biVC9L<vBBbx@C1&l3NlI?_C;u^Sww?#%FRcXnep`v31+@#m6~
zTGR<|1+-HRqC^OYx`qTRgfyXfLgQRm_{9M#kv!W6c$){^)LW=MU~D2iMJ@62B#3)(
zlgoU}_dIgiT_~HlAhrFM9XHKWLi_?W_Ix8{rHY&PK!F;yU|I#-E&t&2?J#QMWGFBU
z&-uc~E&6KSS6!d=oQ8eUmPkzzMEbu#z9ls{F#fTP>vCn$5DW<0;B}?kwo5Nq$!@a>
ztrTv5t?=ZkDo20%bUZP<atGABp|!|_qePyh;7D3kkX?Xm4Dk<$c<3ql6ALl}!>p8n
z9Y$dR`|x>v_*?yx8o85qV#l4$VuzJlhz7;h5uaB*UuOi|N3LDBPkQI?Lu+-n@x#h?
zx;W^}&2o(tekub)uObSE`zvgE)lj8pe#+>3gI=f75w+q~4yu-Smh7kGW?FZI*>l&$
z?G9oIW;R9$I6+$#)&e<R{bYy0tN6oY_|5wQJ)O%~A_~5Y0eRtcP)n7%FQBWN;ps*p
zQgG+ltt0|kuDwqd;__YOM*|L-47j(4%etW}MckJzD{6Pylj%C3F{WNnj>@#aR%|;h
z%(6A5J|Y7hcWbLXzeuTm=!f^l6vbnoD(5ggt;^U`I{MA6PRZYq00Z?zy>Km|K;=}c
z7^)EYGtBo!Ri6D?+4*co9fY-7>aYTJ!3y<_u76Nhf~!BuFDD<?7n}Is2zjd{H>Tvy
zkaH_InZ;bTyZk7Mjm6&cnQ}EhD};KqyoU)vb(q7b)ZtsI(ys^L`4iF+&<aM+4>qrP
z^(oWU8<Mi{Zs!YtC75<p4zWErQz8p>x>WR$m*w{Ak5vDVWGS&JDP6Ru-AmQVTUX1y
zVCrt<Gi!py)Ld1bB&_RRt1iS!lTJo-Zd%>mtwu|8YN~dl-%+c^r;~02A}6mNxFpZW
zO(V9Oua2m_LJo<sCOjoUO7Z6_Go1eZm1?({|BRg-03E6HeSM%c_iD@PzMiKp`RPtH
z==jIxg%8zwiP~&9|HNGA(C!v^gE+<GKq_zZPds~~4#WH@V4ga~tgK<aQW^9VWb?KD
zL<{_HbRYfWPS;JlnP;W?dhhfGnrkeWX=ZBOlLd>&AV${p*9h7?OQE8wO}93Oj1XW0
zF)xK@)OI);7An+_&Zr~WZN}RS#dNd{NXM(qGA_Z90=*T~_F~IOfA7cf?D85|f;{9i
zaZ+srCBAb#Z#>^LVUq-dPTI%5F+K9(H?pyUaQf5VFRY(HTN7x7$LY1~U3YW9s5o~z
zO)k~@(u1a0$48F@#+93EMWqG-?&J`u1`+YR67Bwf`fD=7Sj_Nz8g2Z2bq>_S3z=<u
zD{T;Az0Ot_*BvRf^#|=UtgDArTCaDr^zLFas3B0Ml6Vdu(X_TaC8D4OUErNY11qHb
zUVf1T83Do4u~MfSSgH5u@JKnmiwyiuT^z_wu6^oG&bs+<fg9J4r8}oD^x^C05aInJ
zA+Rn?`$5;j47|LDC9*^4)rXG_4N&(Js853@c#pFW(7$kLGs>g-=+TI@`+}wb=HG>a
ziOz2Ds<B8nyPcdj*UqAN9$xAEk@+3FnMMBk7Nx4Rq+Iey2EQmFbRHBmjyGaprC`xc
zVj`3fxT0CO^DVK0m9d~k11NFivryixgU1j&NnX`PsB4fbY|3QBocCBIC#))BbmX<~
zy1Rdh;ML$7D7cuH<(DKI2kCB<?aus8AX}1>f(&Oj_A{TJNUYmdSPfV6nu44vco*){
z8cgkaY;!1HR3b=m>D24yn#!Y{t2qVKW_c>t3;ag@!DXtjyaqCECD00}Zcgy->};g8
zj2$WqS{PMgqsAbM7Cp#jx%9o|5KLe6vV#5)#KOB3(6`DjDjlI$TdSJ}!)NQD-2@#>
z61jCD#be2j8^TwcV@WX>tO+@84hr9_C`U?b@-DA7P2L%W16uE*vcP3?2ifOSQlJfD
zGx#A-mUpA8E+)uF&7VT&S5w!eBJ_}Bg&&yRZ^0|O(wSw!oH(n8bwM4Bh?U>veTc~A
zJ9##Zyfw*C#@3k8%I4>r8cwBIcGG$JjY(;NzM4OA5@m&F%15qw_1yO;9n!aFDu?lk
z>T=MisY3wm3OQMk8>JO>*u=o-!_wp{%_@G1X?%Ar<6EtCENY*^J1_C*Bl@N7j&*G<
z#F%&C55;@qSkg^q@GO^<RIA!B{TfCzr%~nht<zHDxe9Ck7faMP4cmF(O~2JdkNZFn
z46LJbe{H9*>{%z=Pd4<kkiKy4=fefmPy18=sBAOlw_cS@m5P9l<)yt6UBC}>_Da8m
zx(##ErO3wgn?+veKli$xD${B(O=P;f{5`D|PM4A`i|ENCCo2G>p5?*;L7r}f)VEH~
zU!k(()_cUt2>WAZtuo_S`l5xL?rmPFo`=)QWu-?gdXU4D!x84s$nh1Fy_0&kX$Np(
zJ7a-(KybXnpnS*SpuXKs!i^?1ee{u^>guE7Ql4LV0+XbZHGga&+|UroOm-Jn6pS$H
z{66jodB5L`kKa;cE`6U5ZWgMMs2)%KiU0-Nb!;%v0w2iA>RGT-%x4<#7fH7MV!b?#
zI}OrH<rO~bwbP`gT4xiWT6sm~^hh;5C~ew}Ne{4H2)A!vi9tOuEt_$beeW8NJ#}1|
zBQWvL1oHg7?Iipx2-|}$*XcvCTHRUb&3u`9k$hrdW+KnFj8zc+2$u<G{Hkfm%RDXB
zx$(?0iN_QQzYrs9s`dL@mv+b4mtKQ<K2<g~>#!^BKx1P<fJYNBY;F!JP|_6Et{Cg5
zG#;xq>%RT>&$4z9y{FgoCNx=u6_Q7_xOc2kV@wey-{n3t?0f{Ops|6G&rOj+54(x?
z(zT}XBb8}ukR>x<7xF9j`iS?b!ou7)h4_uvZ>e-4ztXoX#=j5Nz;VRx6sF%f0#Mf>
zVI&)<I)M2n1t|?AhK__=VERx>mAIh*aixIDA&1jisDgs;w&9_}h0s7x<xDd9;w;K#
zd1=tOgyZIds>UE?mLr*@)vikLc9PJQO<ky3{D*B-uWvk3+0(hELGo^r{CWck>)tR5
z#Q(@4m5#AO_8_L+_J7FH({@>v<XBmJo#wUyazrQ#l@=cFnx3#shZ!M;lIC{9Qe1tx
zZIC22#(I46Uu7aflVvY}QzWNAUcU9>IbgPoe24#n*1`|%A$S)dW?s~3Zq=Cd$6BSW
zd#v-zkK1)WNqiWphHsmgA+Mf4WdG4k8~;U#zp>(y&Yil3bF@9h`RpwXBt)xSTR0R3
zh_|Yh+X=fSWNKFtkP(XIDt)7cRO$EHkIsH(kiapX{~dmU9PF4)O8WFJB4K~p)p|6q
zgM(OG8!h6|qeL4D^{a%l93G~SfUt3xdFuf|TUM50n$mN0r+6G}W#!r`+kS1X>|wa>
z3h!vGH*=6IgPhZ%(mxM3Nmq#o4UG-u0A~h<02gH2YD4`)D@5t=3Wh#>AaHx!d3}rf
zk+k%X1+9tS$B*cG0|PS1nMBMov#~4ZV|^#&v}<LOP=9mpb6OLV2z{Mkn1A~&KCbfR
zDScB_`{$IbXaBkF{nOAt%lfUK9H^rhqrFRgX{34V(=+NQa&-JR4a}tcwE!_2ErAlt
z&1=VF5|KhS@8d}t@I&wcUv1B7V<VHc_zaaE6)PE><DynC7Lg}@16mP+PBj$LTwIT&
zq^hH0<P>PO8mVAp3_rzDJpO{tCuCfrmr)sg1v3DbQdbEVH%((fl2T^Gj~t7Xi_-^0
z1v;&R>>{0C=7qO+sw8K4fVPY*M;4KAbGXaco{3-iFQY=!jc50v<+(NJx?a@mPo4NX
z2_hB25@}Ztx|`c+*F5uk$b8lr5BD`Pspy^=pzJkgOW%-RL4*?6AU=l!&IfFI9Sbwx
zZNC#O?k>+R)`(|j98R-0mLA#c?>3S$KCRY>i`3SV@Q`6y;wbS3YCM@~YOxM4BISps
z)0Zm32*9&ji}!t6ei_>pnZ@*)@FgZHDy98W$X!));-tPU`B+qGW7;gn<>_-7g%x&W
z@5253qdIUCjUr~ByGi`^c1^g4k(KT>Q^HV=g~(j-w*#=wd}_AyHqtc(6TCY8&z~FI
z_n*}NP0)~XdkjPp0Zsnoyc|)zo~6oOS(T^hiHXbss)=vbVww4$#x~g-KA{MeHn|K;
z-M48YX^1<+z!Kn*=i=fqbJ#$0dys103A4zm7`AscF;U}rz8UYqFNmmZR<ggX+z|%C
zxt~m&$nD4Eso2y|VK#jyU>GC4?XRTlyY=KJgS2$d1_K+l7_Y_$17l93nS<z_J#X|A
z9xm~A=Ln9g^Um_!e=;af(iwZSzO{2zRufi++*f?uLH9#VCHT94jowB%+aVTOWem~F
z@>ynWZ0yVOvkeUmzN=p=Bx~EcG+818*%?og6XgZj<?E7M84S6}aE<46`);tKyX)Sl
zo2P7zr5FMtdVC4-_kPCEz8mG5a-#{NW)CEILM}PexlwX4S4E0o&6z)%1`hD=meut#
zeuAdPzK(%p$N9xj%lP`Ka8Q4*xw{I>f*#J=;1MyuC*3FRqQ;&$s<|!UfwbPVU@v!$
zHA2ETGLn6jjmI)JYtrV#*)x!%oe`ze&bA}vE`Shh<iTl>Dh1BNrKV2mXSaCv)s?7Y
zO{$w6>7d@8{qoA?DK<*d-R+Rrkds}nr_qs}1KJ$k?u=2p*@I3juai#p>Y5h^lBtf5
z|2My1urBv*ur>ZgexCjH0;~QKi`Z_KwhQ6U>}-D2|KsNP)B&D6i&@OWW26Ky>z~R{
z4u*CcmvC_`!71S$R-09><G+2(c1Y0X1qo&1rREm-rJ80s2UJeE?RIt(O3p4<zjglT
zHn}Du>V_rNVTIk_pKE_L67=1%qAcCodanC-s{7trbZSgmIBm+->_=<)fh7b|*FrUV
z*1tz}C}mvjy2x4b9X0UP-=`M%jG=Eo*UT)+K04J{2E&!(8IQ^zs{=0U^fn@uwK_I7
z*zOe@jQ44+q>%EG`S~*pDy2!nG(u<(P7iv#j(4g#Ts!c*oYAk?by_si3@*Bwlahp^
zHE(R~7~IjHa+2!mB%S}%moFUjiVJ70%TdUUE8Z8D6PFmb)(aGSM+yfFI4;`h>c%R#
z!sQ3nGYgXsU{c6;Yr=;mCF>VWXoW%?i4(OClvK1j+S;j$&#wKUA<j$dvJ&+uHgH~X
z4(tX_Jn@bkZ}l*01Qse@OwW%+#qVl4dyhX=3#AIk{fOTAu60xN_5>PoWM6({^GbRC
z_Um#f=lmAw`*VUa4>sTx+Fsr&o60<ms35uC=6HNpjk5>aE5EvBBZ<bcuf~Q;cutp}
z{^w1p{uu68T1el`DtO;i_N%_m$j@B#%ci!f=5Rxil%z#B1F!<~85-pMTa^SZLI!6p
zH7dR~dHC5@F!e0I-`nJSUkCa(;!%=hb@3A;N3p;ZF+pH<B!PTbZhe5Q_Tc*Q_9Rub
zL~Agm7o(EeYl|f1Z`~{+suz!L@z{mZxlW~QC<c9Q3Gny9F0)Y1;yH_nCEEr+eKf!G
zq%EPO75&_7`>|WHaCTXLOgr`u^u_@3;-%?+XezeKcx7^hGqtDbS{xba!R&cSMMr2+
zUOqq(9^aoYDaT_$;fE)mx}jpk`KTsIG7xu|^lCm~fuD@*ZV&^_<V<3KL;+q?=hs}O
zqiw@N(`l4s5TkQ)?3XVYho|VD(?o7kdw<eOUz$m~e_jo+%WUU(S{wabtZxuLH*GVZ
zg&2)@rV=D4<dw<BD|IibFT8^dzbezU1}bIAK9|+R5AbjrKxbkoJS{khBm$DY)pd~f
zs{ghD{VNw^Jb$Y1M_Nb$33rscmwTs1CpkKrSpb%|NHLx?)|q&w8rKOb3<7Qro|5>F
zgT;6Y9{%}I@2$lP_~*g0y%uq;Nv$$y-#!+i^Q5H2Z1js_mnkPvfrhLdptwu^vD<`G
zt?rVu(9Nyjd)+{tD4t?rp%+~dhBuLScvWa>$K=otT<rO_wyqE6iOJuIjf@T7zBl8c
z;jy8C<y7l!l(Ex?KGb$gAF+rVZ)7*F-S*|aO$!qhOCNqU(}>INwyRUbktb`n+bLLI
z`yAc5@lwm&M#pM;_sjW;7x1`rs|NuMQT(aG={iFds)A_a=HmI4Q(iEBz7R$wnXB|X
zvip3~iI1I1K(8pe4jlA@ZUYe^XQ~>R0C2d=P7<v`*cq8OR!+##HF%sHK1*4?G~sL@
zolZ?b_P<nS^1SZN6j&iHDDY-WMb3V#LBBbndSD=T#dBwKm~;SL9cqRtx#TFuSK3jN
z^ljn5HJ|r0EZmnBL)}y1(dmKc#V*^Y0+HjlJoYU5Zxt;`i`@2Fg59`z(Z<$*WI;~e
zrqraCcSJBPuT-;_{{<b-X7OG$GyfyIigE?ez>0`@4f1WdB&$(aJht<6)$^k2ID4=@
zhjgZI9thRlV{dC8k3pyx4*RS@-R6Sc!ZPbVV<bsKj3+N)bzaLX^Z^prH{;tFmB;zc
za%S`$7{C$XIz7G42iff2poI&h5(yn46YRk80#oQ><2k*}ym*=t#QxT;!0U)R+4#$Q
zac6E3bZi<i5z?Xj0Yvt6-E=)R=mkyIh>I_QMMD@z1noidbdT5A^gI!0df@zXna)tX
z;4G`CRLnzMQ+XygE%=(-yHIdyuKDz^u^3Iw`&0~UT<rN`EH0he7ISegW%n)}Yt?^5
zaU57!)krcBM3NTUaZ#t%`}v12?b2>Zsy@`Q)w}T|C1fJJFC-i{TZ_E%WEZPRq60o(
z-YZ_|)9gV`Chf{8GUi_Uu9BQ)Ww9V<(wkS%>PGXK2iDDOL>z<iA3oGFRw2XP4g=?}
zd$GZ_Py}ICUc}q#Q>T!N5K?!gzQHbR>nH)8Tt5qoN4~jw{iR2dy}XS~u%^%cE62@E
zSs}6iA`pxv-zSL$58n|0UkAGu4`DC8Z3;+tH`3ij*|A)WOwUy-lLcDg0#<eRFwQHV
zox_XbP7XGbg1(ycR0>ZfCU7>5QsOa7<E%MFV1MU}kCS=#_Bk5s+^ojgxV6int>Pa4
zMQsg!qa9(rzPJDOH%OnT_ou3`RI@kEc%UU3o0`0rfH206jy!|_gBWkvA`col($6pB
zv$uAD?igFNzEk_p^|_Xm!_%b9Ple`c?%Pucx3hcQm<;@Q%WYIhNAE<#82?!AaS~vH
zHHZ{#DKZS-sU>VJ)aEEJt;+P`QrGG0f2o@Lb&!4agZB%6RfxQ?|EI{ZY^6fAQS*L@
z%4CsN>Bx==l5CY+fVQ!k#9yXE66JLu5%$b%Z?@Ba%%#J2{##@X>HP%wik<049MW!a
z@uUwQ=xd6d5N?7GyvBnm<*>D_$0c;KHF9Z1gzw9nXOkK47BWE&-YNL$P2fIjnhHlx
zp`r+pU!4#=I|`0TXcQbs5|5yz+MXoJ7f|z==$IY)QG>pF0W2Q#SEh<c&#d=bw-xYv
zkW@&B;qj{1^PYb0860ki<2zNg6*$u`-c>Q}UuO`!n9vkR=c%Oc5E^M<m1h4^eALIE
zlB(cEivNEzyC+8LftmJ@0v9$LXYn4Jy<Y97nzgBYU{wgWy3tBk-(V+rQ;&Ha{PTUu
z#&h7yT=Rxzt*c=6ix*SgD7BDkxdv7br?G#iNCim5gd;?Gh}Pe;+<pt`ls}v<)rSo9
znCZ*tgz(+>|6~W#F1b9p&?1PKPWq?uUQ-v*JZ{QFvP8xHdPsWqKVdL_cE8qGu|2X4
zxZ!<IxVpN8&E!DH&g{(XT<?dp#c!cEETlR>&4?-u1cZQ2r(4`0-g?E$QNw(b2ops6
zy)=&_NI<Y7Bg4nx#iVWj{s5Kv>Qkh0#rQ3*lk0l7+;03-;mK}DTYv|z`BlG#h)z|C
zzMk(*p+|^>7}Sm}3k&Pm9y#pCtUB67qz6ug$x*km7D1ilyOz9VGWRfv%Khcfqre8A
zH-A)>nM}T;cWViOET<fMn3)C{X}GR}pHgPahrur!qTNeF@Kk%Bki!!*@C32PfnS0!
zivEiTT149#Z}sax?Gu0K1CKzS?ir|5M0NSaxbd{DuSF)stD1izX0kpE-5d_#5UveZ
z*7#tdg2Tu^l`}?`Ff}zi0Q3^@KDd(T&z~Oi^@*D<!t)ewx8ZY3O6=}lefo4{w)<O?
zns4wH*_S$WZe{et@66kmm8JHx6-Ol%5^Rx*Su7J!A;vPNl^nsn@O*#Fw6MW@2CWG4
zo$nzGY(;3!G-BSiHXb4`(Wn#QdmnY^TL%YX`ZAs>l{)V`Tf0ph`~ZIcCR^$9gTTDw
z$9iHxSlq%{`hk^6f(BBp6}Pfy5cb{QGlajYU~9$9$tlG6fHeMtUv`D+=PhAtWnnoj
zxqb}Z*K<d@!g5Zmi5{6I>M=bQJ7oqhV^@rgWi=$|dGRAH_I43DH#ZL>IGKtpd+BQ>
zjncXQyI8*fz~k^m!Y}kc^eZmnD!w_UUzNhP#3m%HD@w}BO$@R_LsG<aZCLu^d7&ne
z$C)as<3|->p>TzAmx%^BdAN3#aI<#uhr(SdS?ie~OWW-#4%HLkJmitw6ecFxWSj-N
zse!#!&zr$ZSan6NFVQ1xYl?h1jcw(<3q8b!@=S*#fY`7^bd~gl;Bfrtwf3p}E+vg#
zY`I#7b_?v{b8_AiJe&uipbXB_5V=!z{i<uIUC#eSIUx@)%3Z@gM6^}3y$|&dh>VIC
z_jTk5<rzP(uH}4P%*9q2qMhv4$fp7;?`L9!IkMDOf3ar^aNuW+?2Et$-KP$~+DAIv
zk{Bph-u^p<u!yuh+J|pya`yJy_S^K~;uV=!0EM8Ok{P{trYWD6_ULr$=MDPcv<k`A
z9?Uorva|w)a8)HO8qIhejkdIEjOD{eT`a8a18o?fcgFIlU9t8$I{{<ZYD{!`jwxoz
z=L*7~D5^VVNfFz0vZFKd9r(_{{i+Gkl9Ylhba8Z)V1vW=CH@X!{|>Xdw7C^&c3R$*
z!}#IsRc+0^AF9*mj+}WNCF7+*BuuG$s|7Zt(!CBmyrW)8b)`-!WU{cq$`F6dhotc)
zJk9*3F;QqkJ#+PMcYBzb_^90cfBd$H;N%SNHmT+p7@C@hZ2%~*Ie1>%9M2*jNLn-h
zQVIf5uYqC4Ek%ARzwKb<f6eUE{uwQPUCY8JiHJ#u9X{$p?ZK0_46PL`x2>)#yumE(
z#BpG0XKP)#;<<{3#QJVyGDK?WTYccy?%tv&)DbD&^S}A@_&kC?>QNPt3FCMv6TFUS
zi|%7t1vp8m<U)a|>upf_Iq2hEC_5c`!Zfx;))^M4zqBFj<yc&~oS(hHVf>~Mda`X`
zEB+Sp(2z8~yci}EmHOd>*>JAG0Bz>jbYXq^{k>ykk47U=mmhO{bkUjiSgB%SRd4cH
z4ldeo%>|1HVP7S=qKd_9t(h8)xFaP6hIQ91!0+3~dzo7OCp#3uxVRvL%YU+v{aJMT
zF*Th~`aw~3psKCbM0eGChYyHOLD0H<gx$@|Za<bcv9(d@$vT)6_8r@BnB$A73Fp|1
z@}6T$e#e26<2ZD7rrm!(PNQX%3$$zXJ;$IZ!uM<IG?Cr=jtG(Q`qD$}a}$(A%q>t;
zG#=d-la<v+$)7ZlE#E%hqqKK1hc{oy_C7NMKP4O@e`e;&n<IY0rH=A^@>jnA|Gl{Z
znUKfgi1Sh0WwT^bX!Rq%#&lsCZI98B0W4-?<5DT9hmb)7keY-3mJc7lW7JrpWHq-2
zYD?lKinI09Plw(ZL{w5hMFPXEP+vek?+-K(qDh>443E~6vJtJnL|B1+<pULv+_tBW
z+vyiCP(>`%{221q(wx##o4X5-nYmEvQalydy!M>uh^m)t`7+*^e1e=Uo!I(<<SZe<
zT`$>|oDjl2UF-&@45>#a<GV@cTSo%C)nrrLU-#U0j5c5v5<untI>^Q}h&?I&*41Ky
zPfyb7BZWZ#wi+Aj%*q^CKkTT}-DfmiS%1ehw>Vm-Jj+T?ZL4^uds%R>x9&B8pu=0B
z7*AB^(F)8sTha3MBndg!FMr~>gqlSoL%CcjaUEau2gU?^iBT=vlJW=AzX^F{No17_
zpZiPE(bNK1R=WKczkf6fp%7JHvdr;Q_5HF_n4dfM_VCBY1><;Fuj9ZtK>4pNBlEyq
zg@S7uQB>DfPs>#Fu*tYJz$%24Xw2U>d$4%LOeP(_PB6cNl<H~D$vFW{U<n8--R6S~
z6&I+W_8K0~c&IY9LITr`>#g4H@w=yhr0T7&Bk-`}5hgfo!$mcMb}>Y6)}#)NA42XF
zClH8TSMH%%FL+msA9?&6fa$ihnE9{UY>=hOC-+Vmh)(wN2EYG*26}eZ75ea26BHA1
zl5PdQ#cQ4mP9I$~iQo)rrIu6b<03n^m?u0o+DLt8D+0+OkR2#6yZ|gmTY51|2e@qZ
z-F%O=X!K@1xWFXoTI<^z_m8c08`okusX%_5?HX9?H_o0Ur|e9JpDnNnzJdV@jh9cY
zDmG>@>Di#vAZj8eW)PvC%pS)W8ELli#s0oJ7inJOaYQs<Mzu8Fqt7v_n0Ow|NO0l1
zVaKBjZ$}hRv~ZO=041+BR($K@rYq?lY4*Qf2t;|6unUAvgr=+-*xVCVw||zu6@7i7
zOm9iqIoJGA@5Yv&(lNS(+}(jz<Ozr>tDx>1piTiRs%V^|`TDibMo$zP(dzi^SQq|8
zKELOSt?k)|G=-tQ<`mFXir_YM;k9CoLy@&C{J^(IAz0*qvvYwp<M&tQzn>Ao^g2Pg
zWgH6IDix&<9Xgf?#=(u7glCV(l?c}!EaQom@GO(rpW?LCGaXF*YboAd@K0D<SCij?
zu$M&$>3AkrmV|^(?|>n=t~)L&EcW@xAj_j=#OQH)VXCX5dl!-BjVN?Tghzp~;QTH^
zc00hG9RqNEk(u+gonf$Kt4FniO`|W2U8r}3CV<-`^sKdQzTU(5;*l@zcLU!lHogQ)
z{q90(Z(nra{d`m3WZdu^d@fM)(f`zuGqKi6IPn*H6%DPPbN{{YxcK%AwT7fP?uwXI
zJG%QdNu2_fD?3S+<*vmmA3l)WUafo40+35=b6xluRSKpGJp|+`7tYfjL?@et&IAQ&
zr<FGGuLIhByq=*e)_C&GOzuxQ>@qXnT7hd!#&=I_j10Hzdd6Wjty@6-0U6^{Aw<pn
zY;*b6iSdBrSs`fsdC&ld5C!EX^$>am?UR&nad3@%+K@4pfU8x|vzRu0CDH?uQn*5v
zMD#$krkqpIR29#B|Gw3N2@(~!fp#~=l3dD<`;)q~g+IHDvz)k>n@>bJ)(z8qePN2J
zURG*}+-W<$sbIjXV%77*(IEAWHa@6`rO55iEXQeR_%{gW-1SDfqNASMn2abZOQ*kQ
z>1gpgE)_#|L2xWyGTLH|BVcT;ySrs#+<LUqsaU9X)OSOZLw+7V`T>8CQb+>gAC%o0
z+|u8yXPu<Owt>ivM2`c`GsdCat{0c@{sH4qTzuUAO}ev72URUai}S*bUz*OHmpV6`
zd;{yAvjUmmxo6umwP3QKcl0a3g7@I?P%+lvfv*K}wQDxKl@U-<+rR#p{`y#3F>`po
z5ha;3WZ_{s@}*Jt;iLbeG>kMi5a$TJ`mSC?!^xI%lY%Uu;KJc*y<~lPR$gfYjtD4{
zmuxjX2j%F-K`vP8qY5(8pFeu1_<W!G2+eLG<gn`Ig@0Vk)jmVVq@yEYdV1!Zcfxa-
z?5~mos07<AwX@S;8R`srUQA6DTlZ+#$OhNolXt3vRiVM|Rqkj^d%FXS#;){JNE1_&
zHtH40e>GIi5tX6QnWltNKRKV&yN+0V(%DuwS7D1R$&q&lt46~xRtbp=f39eiySP7y
z#|fe3@afcJ%QhXertsUI;+L7KCC}8t<s|(Gu%fI|V|C;n*xQ>l{fcZZ_*Sj9=XluD
z8((R=w@Wrm0JYRt)!seXRJPbecHG2y=m~k?sIqc@-^+}l)f0A#(5*7scXM3d4b0jG
z#q_kzYC(Y?8VX$*FnBlPs2-krBJPCU;pqWam)~BB57~XnHByk}cpM+mS<rr?4!{Kr
ztP*2RsLa77>CIIxY84V7DS8_;;$&JsGcS*AF8bb?L9%i#c)dS8duWX7NzsUf2eOId
zaVOnKR?UN>Y%6_C_Wy}4*TUa0iuT`t8KJ}0fw?$RvSOz(1)Oa~h9f!9wR`irprP3p
zX{heK??4?-66-%lXUD<>hN?VK5pZ{{*{TP-6g!NnOrGwD>8w<P`;^r#@#Nni8l>bU
z-mCwv;(DVNW;zn+J~OG;uxZslmSug|{Hv7nh7XG4XG()*tlvv-Spi~61H^Ab(0}XH
zk+%0|tZEr=_?)n!fle#W-RVp&6>sM=NJGIz;9X;B4RdsE1ayLEAta1XSEHl34Cc&y
zAJ-^tOV}yY^qib^Z69lveT({KgtO`d>W2f_`eW1dECK;s>P1GDHeWyhft(ks^Zt$D
zxwa-SxMJhP<TGKjY+}UOpBJ%-z2&i9(BD5*>49EJM8tv}Vlca&{C;D6_RYQ=C=|)e
zmVV!3_cyc!>^e@{_l(SQy-agJ3u|%LH#W2BtBHCIEt>D@Zkyg?eDhJMn;Z<>`-LcQ
z$~=uFILy1ACEGU4lzbmc(&?zOX=b_Q5iqXsEEf}$xPnGTMYrR@-FT}}s6_+Y_DXT=
z;@8}!AaBgFnGt*UHf?~X%IT1#-tqx+3r$H?DPUr)>h(WETS^7FyZ?jbu=GxaJe9uf
zv9$So?mh15!9X@PIh_#sMR4o1tWaw}HzUikni3~Sar^eN7CNLK#Rp^7gZGRS&H}}Z
z+1asp6R-!17fo`L@d>{Xo#mCRmZtW1@}Ad#{t(&GZe?=Lm=%@>^mjA~@W8^*x}W8C
z93(MU+c!gG1gR-+jm+qIK#5_3?oE;7owkf5#b^qHpN)}KZRo7kP{r|oi*c)r%16{W
z?31837Os6B-QbIidIyR?-!T0_;clgD@jC|W{B!3iMTSCdTmQ%>>@YRE0yhE9((%>v
zp8J7l9&mH@N_nc!6+p=oQJu%>{e$@)7efypw{;LiCrWB5a-?m>v*3jojCJhdXjH`~
z^4H)T9jvybb6MJ%b`KA|TIO;odTqCc9Ei7%8JccQ1|zP^wJR&$-t=JBcGcIu9(s^z
z<!kcW)I4cV_j)|!J4FWOdsa5agGcq*(~{LuJQmL7CF>nMfNAw4m_3gj{2^M&X@lbN
zeOUAE1C%~jyd8wgj<0w=UIfYyU2dqd)c^ZQhJIx?e)Ix4mF;@U4#8(Wr9?9@Ohuj~
zMDieP)ptcR+^J<@#@N(y0E!OGrS5zSY6qsssAimhz<7UE$0UnE9Uq2kPT5quMNSEx
zJ=x#Z3T+1%hR4hJ%^NQsiSnX&EeQSxu&HoLx)b)_s4z6qR~4d8j%x2nItt9VHX|30
zTx{p^zL`xpEbLACW8j>ewyAXn7`c)nSN!ACo^RIFtP&C$r<)(b8Fgw2F`J?!W@&0<
zI#$(uy*<{2M3K|2B&_7TwtRY;_uT+1a#Uxo%|b-<k~+UPzsmLjm<YfBD$l=aAOY@5
zx-xN3?BdDtC>OXKNl0f=+|W(6r*vy;9-7+2{fPK|2D6g$F@6!eh7T@wBGgGDgAcMR
zVhO{0P5N;6H@2Un(0?y4w}HX1GL^p<L=#qRd;IECgRA4d)tXukn<arBz@Ex0YnsBN
z*0oo@;6dRc1&NuMd#A__(565c`_8&}iN&S}yi;yNC}Bs}sM&-k*SktFc5?&7i9BUX
z)3Q7|qc``z*L-_eBu_Tj8_8f~n&p#TJzYIL_2le|DdiUVz<0*|#|nsP<#Xy&N<g~Q
zSbvINKU0xkwx`|A2!JeSdCx+g+zXD?zb&^*Noiqf-;4NyVWc#=KWwaejv4~zhq&Q3
zft}aZbaQ<7P7Yq6DY6c!;N&vC8H3clGNf3izeN3@`JW#a(zpwk|4u9uyi$7kdv#|r
z#}`^%qjnnmWAf?f$V_`;^ut1=U5NC49LY=I_s()#Zkx%cB7JiYgV?M%z0a1f$$Fxs
zz#wR8X+M+GP^&(dp0oNC))~oJh71CEwpsM9^P`H!2r44V``5Qz>Fn0VG}HXk1PLkK
z(UW|&=NI#?b<j~5m0h6<afQYtoE)U39cRR17mTz?Q7tc`K%iF>;KUEg%0jbg0IU7+
zPET^Q4;)_O#K`(NQvHQBtw(Q^)Ls9JOWR<1r^WD0%*J4nzLxp9DG-2nx0xsffi#`T
zk%#vVX@P0I&6Q!CbU~|s+})km62GVE$QLXGI1pvg*BuSHq@=!z0dwPy`%3>DY?7J~
zJ~gm6g*Zz}-XZ?tCN?1CGQ8yt+eY9<>e_dPjl-2f;Q~>&q>KWVAgSJ0uN4ez&E&!=
zVV;i57`2s!0Pft@-uDg>XN8N|Igk@Tjg}Yao3W$mxaYl0c-ZArx*D*F`91i*d<m9u
zdW7v*T-wsx9-pQw!1cPg>}B`=krg}!P)*P;tosFg0#md?1uWz`6DN9X>(Tl5MxC4!
zjjW*3z1{bR;InPXI{5>eYD$OqrT06svhOb%*~zUS|E<<gNs;$;EE!+PR6Q|D?kBHo
zIno1jQW~VBia*C1mQ>AreG%SdKp3b*f0VJKr5*pU-;rr<K`DtYL`~9k^znopPm%vI
ztr|erycd-c>#p9HJj=owN4_{Xi*^eQS<&L2C{F?}@-FD>iyNuJ0}DGW72pA!Tpa5l
zhw(J+X`|mPy1aTfQ(^tmr)W;Uy`v>*^Zq4=&zr^v!i4`9Pei~h*b=)rcx=HtwrrWq
zJnpOyns32E0noplPfrhx%mYL;tmGxpxd^m464Fg>VvtBl_)<YEN%<zH7uOdRR-&&i
zU*$sA8g(+qt*PAeL>RDz)e(fzG2sc3pQT_Oh=Iq}&RizAG8r=}bPahMBP6D(6K%yY
z+j%*)MIrTW_|ZqOU7ZkI-28RrD`$0m)2$v6mI~U;{*aJFo20`$z)>c|T4VuC@wm6Y
zmN?OoB~z?jE_w#RpThLd*jP6*Gy0UD4AV>XiOe&9J5j-`zyx7sk)Y3)8P`T|4&3Ss
zq8dK`&u3z!m4K?md2hEVr_PQIq`L>~ts|Vu>i;$M6<}2^(b^jkgFH$}t8^nNEh615
zE#2Mufr=<8Ee+BwxuqL~O{X-{-JSnz&%NjVJ_tOzWA@CNwchovHS@Y!5cQM4Y#BVG
zI_U1+wWQ2&%$h!`DgxCJ90)T%@;yj7VVqX4$mBw)O1DRd#wY~){AJ?kSr}`5WS_+i
zPc@R^ULGwB8i=xS6zdD>^4#}=E#$8ZKN(#e=_~goQATMqGu}HauHQlz(#X&W2+Tmu
zsBu~T@~HK}o{dt;%~nw%DA%iH-5qhO@0(K`T{SYZ$UFWSkVj9w3@x{<5>D%3<%J)8
zH;lGC-hQ)Fud}`NAT-S4yO*lwy_PvdLB|W>)9%oP1jD4C`3~MXmJJuX=0>6ZuSH(_
zJ|mX09I*tnhJ~qJi5tb~aKu+Obqax=*+W9h==lehqmdwW<gh#Xtd&~orIu=@Gdpv-
zChp3DXnp>h3}41t54zFq{9F6k?49;IIh&Z6zNr<O6ZtQvdp{6u$?^s7W8<s-Dw_zu
z*8PXDI`_Soy^Y#^RV0G|wHa9%zW!Rk@u5y&ZEfjug#UX0*kY3!Lx#tN%Zn{r(@<Z;
zY<pE%X_q*#PXCk|H`9AJd7q+u`-o^HADxen7h?Z~`0!N2tS|(Vm)ti1INC2`;e956
zia*|f*9dJi|315?_oX(A03SCcBH%R(?T;+G<b6>3uPgONvmy71?B}*!(R%&7w}@$d
zc{aqicgDiPOvccZKzMsrW}btU?J|n>7t4TN#)%&Gm^#2LNwOn){O|^BRoy6|n@I$O
zcK7T8X3xf*9<iueRZ@h#v9_a|Xq=QgEns+f0I12fsrKrQ{fG6QPf^2Yspj<R3voiJ
zW_(jjF$>Q<>vd={yb{NjV&a@$aOv(k%-`-R^}E^-k1D7HOG}xok=St@g~Bzx&K{t(
zw^um*FiXq1k0geSZ972EkLC|YF31;0xNoRvDuODn&lgqu+#pTS{-w~PN-CvX)q@cV
zE1&;uCIG3~+R+qWL;VwJYds+$?y;kN`;72F67XBw((++%yW$;RU4+y9Fy+j`wAUGb
zK;cXL-aTiFiJx+J{%cIRE23|R22zBO?TJv(igqAl4&9|lt$t<hNdv>xs+vDflXtdR
zdA|^b{{@s(0UA5(n&|#bdqTS8tc=cY+wnB2>j~({>3cCwCe$6J?<49VhL)_G<X$iF
zRJ2au1;^EGxW88919|!7M2z&UpZL9^6i`Q2BK*vF$B0iuQhsX&7Zi?UEQoaE5ozVY
zQF`OhGJH=GHtEhSQJh89p3Hu5@U7xZv`?=d*>+Ikf@2IS)xc9t<HxD1_H=B*&zR1h
zL!^S@dzDsO?$9%vM4s`9+X0qjO}8^TglGxdFYxWKi-p-_Gn5Xdm%u4v)dQGEhxr8)
zb=~Ujf*I5oFU-#@4IaKFAVz5dB^Ri&Lx9hluHWaB5>#!gQ<6ok1T}PSh}Av9`9{l#
zCuh6S&G@y4933^G<9p%GxpUz-r;{XP5&+7@nwnOe4ESDmo^18yeU^+35Fe<F_a*Vp
z!ox4+;loy;q`c>n5BbbrZkwAwQJK5_1|kjEY90k<|72Wy|EM-ntH1kEJ<IIC=@&k7
za962uPCJuWzKOSyV`T)R*33+6Q_ZQA&kU{}I|eMLKYC+oVq|08hM1{gE4N`3DBRJ*
zN>Wt(HTijyNf3%RNq=;gHfLWl&~+(}u{};C1BDQBbX~Exz5}52?{81ay*6InJQ0or
zu)vMlE!`2dXrEcz-n5TXXiKA7>jeHovP?!M{zV$a4mLJl=akC{X}EX8NA#E=g?mRU
zBU^bPhgJ?6j)bqCMGYeAiAZQdj8~&eK@Fzd7{Lvy{{`hxwNBreo*>NqdX0YM_^p_G
z3AEJmJT|EovneuXWg~={S%clR*$^zPxEUdBOX|0FFCVWF$(`3G*8{7ls6Q{3NI_TE
zKZ&t@dpxtxsQD#UpXdvm-v!+$Har=;$(eHuv#0XO86bWou8^PAR0HaBxSj;mTBz^|
z_Ium^sqdI@-lic`9%g&Nz6A31iVo-r(G#|xGPHo|PwbQ*-s8UuXn1dk9+PPAm_uQd
z7(Sm)1Ou&3r02_%JU+u&s#y!ki)Ju))2%hFMWEh}2YTAZ#Pn!0HzXoIY1v;fnK0lh
zgcAi+fBB;4!&wPh8UbM@Ype9AAMM^h7iXe92X;<iiWp?zcVF@!Ebc-6fLy|IlfmFl
ze%S?%G^(vi&z?uoqU=o$MuEi~29#@r)9Rk*X%lgmE|Pt&8|C=?i*|LOvo(}4UbD_&
ziwagEA3wrL%km$#eBk*Hi0J-o+?<(4qXp-&;^L$}=IkHpolEEfy2pp~K?l>EI2IZM
zt^n>ObTl+UBSiUOBGoqx**+r}P56Otv_NlN(M|!oKj~vu#3PF9Qh&kKzReK*o9^Hz
zB*c#FFXGWU`#I_S-|8qeN85PO73V%p3&24|<=Y5>mRC)5>O6vrdv#uN8bywM9H}o1
z3Hi<DOT{r+`J9?RaGk>UJIBKnbE96-5BFefkW*(WSRAhTka>{IUy(YZfh=TWe04MA
z`cJjlSCn9ddrmJ|l+3p3L9c~}=iDRMY{CoORT>Y3o+nDYY)9&#gSYp}>kl1WtQzbd
zo3_!GT)oLjYYQ4)4?{gNBo<cQDL5KDDtFrv%^Do>A_YBZM-8ny{e8`o(@B6(RD-Mv
zLj!m*Oncu@O3BVt1%@m{-b)^4eWrnA`|mR{>ZJ-*irDPNeIH^69C9k?QUwihNXGg&
zA7ut~zo?YwEZAB!cP!GLyI!&+?~B$rw{&$RpjUnZgyB{SANKw78jU}E3GroS8!tf>
zI_Gs!b9-Mbii@}2Uylp@?MC|J`xv=mbo7j7yVc!yr|dF54P^V5Ao|V6gFx5on>({Q
z4X@3%T4Jo|8JHK_)8u8XhRZ}2k>okzS2L{c1W}SZQaP?#$@aqFumuikXMj77U&SUB
z%cE@2G8B3C!FR4@xdM}^SC;pnotAHra&bYk%8jCm2I{aT6)x*wFxFjSClbacdE+H<
zzweEn$z~o0-(p<j)t!dkga|#I*K=GX)zPs^t4|*PIC}j+VN39CQ*F7~m{wofB<RzP
z=Wu4Tlrn6I)xFYFGpc3@TzLHC=#n6akzsAPfy}NqDVxK0_)C;z<VV0T)~htvxh@&k
zN0z+sIEO&${>5s77aWOaKRlG`?;W04X^ZTC!Ok}S%38pPdz&h!Le@;+Rl0Il$Tn1V
zvmpnVo-m(ti{vc66cXW2T|Qn>Zxtm|D-(Z4#;dI&>Ff*+Lid~b7npeH-1V0PC&y<I
z^Cf$uV>VQ#rY}1B19?10bUl~lLHWo!;UzHy%+1ScEJ9oLGW>C8YsuG$xbU1x{DMyG
z)*oS;-n5hf@41Wu7i}#>wB%WGD2utxQ~rP4gDh}8B`DOA3y>Cr6x_rw&jFXhUteor
zWY|-*gx6wdSkVD4Ix5}yrL-k0=UxsPZaRqO%FyZsr+-F7_w<!R&3r}@chL@J$`vT%
z5O2weMLPAzksLl+cS6ZJEdgHHnK2s~q$>l$F9;?{*QO8EO-=ba-<Fw>fR-f7(^oo7
zJtW>jH22VjHuS4E&ZCY=Tspd!9s8h0JOwi|kcO?-no=t4^1n7#cMZkwUbf`&%2-@0
zH>Cs(GXa2G7Z7yC+f|KW%te%x)cyKRJ_BmO0#&O|DS50{-Crw*-Ypc5{P``RPn7@0
z_^5*n)Vzk<`{O>O`V^K!U^}&UF#qzi3&d*b`27u(t%uIxa})O4G-BOB#5MpOd2If+
z;Jd8~13U}n-qHMi*ft$K>uO+ftxMSI`c|7i)GyQqTIVwMzf@>fB~Y)jTIrf3f}T=^
zxYm9Bs!wF_?=Mv8ejn{jRh*vWZ<Zk&-E3}TWY<Qc&aUow7EU$=2ZlLO^20j>8PxKA
zJ1{1Y7jW;*=Dq6)3DvwkAvv-r@j<O?Oi<1;T#vN@oTK+pBacf-c}ZSWIc53v)K>`f
z0hQHXT83U)FPg##j(5Uhl(vsOt&EJUds>8MP|loNX#`Cf{?E6{iCUM@2>dgpkcu|n
z<#*yq?U2Ua%w8lX18B(>Ku4D!jK*1ko1=unK?A7y6uK2zBFp-7)D+A;WZxuzf0B`*
z-3;}!l~8giFABR3h}Ap%cb#|2Z?6ZfJ`!v5OMV3bykLzUq6L70R{czNa4JP$<Jps7
zKFhRxE@Hr*w}MP~5PSQGkHW8cws*g*>NyodLR$}$0I5}pfwhTgM^Oyo+sPkVYhj;Q
z{67)M`cEsk{;_Hd<ed>p%H1&`OkiI|Wr9MG_w5zL(L=<ZD{?qlvqmPQ+xSJyYRrL3
zD1vhT%kPA510BuS5-KW=2*l5bWEoGjNY!U?9Bv~I8oW-{`7RsmukTq@S06drbr_LN
z4vx2m=A4rK0o*EJ*uqO+i`=KZpVa$5JyultmNCVpubW8(-@E5%Xnc1%)FdZv4l8el
zn8uWZP}1CLUEdVl#P}c1&8s`2+mZ42PJw9LH%RU#BV)`iNzb$*ZAy@k??|*0hE005
zB&t<YJ9=3YPef8~(46R9YJov<-H?*X5_mYgY+qMcv{zc%(emqS#lm9{HR$pbQ3kTI
zC_uzW!4{f5CSUdCf8-kg8pS_dx<i94V_LhuDO)2cY8u65hq7!Xa>+W2?x2D%A=>(b
z>xHQ!iEw>q{8%<MD+OQkpRbC5n!n-Z3i%H&f4n6C?J4A+GxNuP-S-0G3mS3hvne2w
z1#nc>J=eY7I8wCh?+9H~nYdAXV_d)BRb?aEku7}YIIQmbyAlfg7!pFfjD?AhKI!GP
zzpSD#l;d+9SqYkAezDMH=6G=cl4sLa5rZgZ^+@IyzP!X6cAr<m$@`AgNpcvo&-u7w
zJY*V|rz=UQ2=F4`D!OZ=%5!teZjT4DrX&Q5fKz&n!<)l7dp&0_1TMF6wm?yLy5)J_
zba_`|Rv!H|A5^uvN|@TIiRf^gOe(B#8$BHYd<`lGjUyX(3h1{i6{bnS$Xj=-1Su23
zT;DEbedrO7OmKJO#MIDqcjV{aFAsu)OS#AEgC<A&!mJvvTf#WZK+=K*KEUT+Jv7fb
zE?q;AJw-)Cc75N*<}=Tded@Sf*uyc&dT{vN%AspnwnwE7ZOeqyVK2ae1)cs^<iwnQ
zAeVYxAi=!RIJDhgps?6L$j8fX2)cBtH+cZd$hE<)|I;A~Jto1QxsF6N;g6m;IvT3L
z%!CKyr(hnt!4gr>Ev5ZmpYK0M&yY-xeAZyvn7KV0$LQwh8o$SsYViSr85+~G;^5^6
zr@!)*7C~7aq^h#guH2DSqx>&Ux}(3;v70<Mk*5Ch!a+wyvT3^QE@Y>mRZ~5xsei23
zr5Ph7Vz=3m+@I`|xwJG=D~F+>f$ekg2nJZJkDVQn@ri+{wftn-?&F$z-y4`@Wo=7-
z_0cmu=j2QHHo`egMz#hnMGq~|ZMEb^%{0|W$qDbhEy0XYioc@R*p><q7X>7@MSD9q
znXq7K@TuOSR}OBhD&z)yL#C?DM|Fxt&y#mL{p3KV{rMv9=g+|sMe4(Sh^P~xSx?%w
z-DOZQ{LWkcmeY)&T1+g{bXD=K^84~+=H9Js7u(mTY1C#K!gJ6?$!tKkF&rL>KarOO
z*p$(?>Asgv#=}1ycD5*~`D?=@|6__oehS|-+zt`mR7!U<@FcQZy?>)}e9wp>=A^er
zy1Tnz0h2^9HqZZAwaGH59bl5Ybi<{jtf{h<8Xv;~7Xs)}FORMc<Bb;|B&h0Vd^&J;
zZ5f_`@&*rcgOjt+0D5F+OEb#2sP@^mtabzBBe{C4`-o9_>BgqM*6f}YZ5+a4cU!IG
zq9)YD%D^8SX7ZuP$>Ur;w9#@&mFc#!wbc*|vrzF4%kLaKbfWq{j~z;lT>ZZ89?W=I
zagdWeUeo<R_)|A{w54!G8j^t#@@vDCgM%F4Bf^7O5mfBb()K^fEbSfP^wp{{l#;_`
zN*bbBgoGL=nl7O?+evo%wF}~>YS+`eqvZl06@^)<MpMVL-5X?fHdLlwT8M>01o^_j
zw=w-nHlX(0tWk2f8h|k^EB%a;6$c#mnkL;^pPEg5q>EZwUU=CF;r=(h>3{yBi5@vw
zItFHHCc!9}nMD2NLwR|$reUc|>7g-lau8Pf&!ait>m&m&=BRZKoGFVX-|;*h#Pgn+
z2K-Tyf|Mww)GrbxQLdM~SbpkQzk4=Q3o8Tf!&a4t-0LzLXRq-XV%`{=8kv}8b9vCM
zl|X8vtDD}^#KTh?jgx}4714myNV2i1h{Za;br`ZDwaRs(kC_u;o}t)sYyI#&f{L4*
zzqv)t>WjSGDDmND>LoyG@4x}LI$k3NdG3Y4NB{eAK4y<_icz4Y`{&ExTh~9za-(?s
zBP;)Us~NSjBV>#Cu3OuBfoDnvRD;HqpW0CdFUtfZNhV>OJ-4P8dzKLB-$C}a6a}CO
zPF7j85Z(A-VQq!HW-(JxCCy@pnY{*;6VW_%4uM>Zd+1`geTq*8x&HA3a3OEN{AdT+
z>j^TtTwC4BF@){*pD=hZ2PlMta?QG6A84a0HeDVV>ctX<n?jzahYD(-8DXiNf_*cp
zAAGK?z(mGWtgY6F_^8j{_!H*?bINU}ADCXkn_u0Vvz?Rbq)n}^k5BoJC?4SayKb{7
zykvf$UKpXNoUcJ~-*AC9HS$x0KT0?@1)x;TYPl=yU6y5)(E}GB9~0Ap(o^c|mKU6m
zA{%h$^%v&SI5=dOO$#c`lAb5~`|%i{2TXv|ya}R_vM{sg0*94;u`n-pBK~+Drw^H*
zt@U`oRxpW1qK9!|Zujrsb3uPv{o^yfL4@ZqvO}P2BKXzzsU%I*i<}LoqnkXZ3poP!
zlKkzh+xc=ua8AZgL(gn=;$59hAvD2nBOpPM&~xaIa^!6HsiX)Cr7uj@av4z;@uB);
zsTCxwqNct=AM~LpK+y}Xsr=_w&&L0znQP<ef;Y2zcp9o~ZoAJ$lv|Z!h}x+I%5NXQ
z`a?)QC^5S@V1fD|<ctjiC{$J*HL44kE$s<;X6Ea56(~buR11Bf<Y@1!$XPYl;+xm(
zpiRok{!-N8;T7KFFi^Asal@6u^z8I_ma-gVHg(VjDOgJ^+s64eP1}{F#{(^a2L8G`
zt(XBg6kbpK+~kw~h`ZW;nCb@&<!$d@i#uaw!l7ENMZKh#<Pug^+&Z;gp#DcJ&_|(R
z_X9c>$789g5)d}9{xEeLB^BA0k;?jvlkXGYK$21fWBgOd`KFMq3)V~T8c0XtK+GYc
z@a+!G%xpC8_%9Q<wf)=tz)#Q^3|9QoaN*nOszLYYhG(S=GlMCpDuWdF&!4qlISK;#
z9KO<Xv$7k*!9a$g!v>nw+zpNC9<^M#@%a3q^kKz`z1d(=;Kt~VF9#>vyY>QBBja~X
z;P6cu&4O0_REnhJN;=u)faK#jq06J~(9RpigW#YDsxQhhK@St%0RO30;`(?`b5E>+
z4<aH`RMsm#-LMdWH^urQqb}OAbcgM1*YQ@8XC-JK36{SHl{0bP0eNnN*AB$D$SfzP
zd2)aOlChL&ug^0|rN=3YVcLTN$Eya$>_iBC4-_ZdC{Oq`&u`xqrYZT<=HbbpUR((S
zGp)wMIFx?p^<M~GeicRK7L2)DZ;19k*t|!9jKHXi0P&BtV<`SwE)4zIF+T^l7lNpZ
z#@qp=4iY=hLM;VQR*f~y*17s)wd6PxUJ9zI@N?F|A{R!MERzAF<9pnCwO_<L^Zz)}
zTVU$I<AtBw%^O-Fsz7z##<uvhn!9qxMyq^lnGp7^_Vk~_8v)t;ur?P^ocGFkPgL!|
zf4;m79MuL?*49*8Fm`8PrWcBd{{H-;WU6hP@HL&v=*{A5P^28L+Ws?xYxd%W;=Gv|
zqBql+n@spzL+(;|^;%+;t<;SbbRDkMy5b7!VoT&L-cJMFA|Z2i!3uqPO)VoIdokvU
z$1N{G=hFc=)Jd<F_pQu@r-Y1GO5Q@VXn!IrtIqN4E_jP9og&jNNW}@+j5p4EqXKNN
zmT?^R!`WVVOHPhty7Rlx%zjnr%E0nJ@*gEQkMaJ4oHtJ34%8;M#oI+HspoWbpU3sQ
zLV?1d(AxT%oX@-d4=F7+t0i;BdscSJ%6izZ;HE4gN5>ugWA{jKjo5=^9?JE<(@@0(
zL>KG~`dMRtQ(ID6nhOn8R_Ju8=tGE_Rtc27%1VQfR2;HoQU`V(>XO1FP+s2(4XaHy
z$F^SYj1AbGUwJpzW;Vl#t&45Y5ttgOt@`ncKZ<C5BFDfmD)QV1zN0dFFmJ@Sp47oo
z&|7R|o;E*acH4<d`AKjs?{*oqmzxVYVMCznL?`u>O+Tr&&5M?C^9g@^;){;)Kyq6a
zu<KOvl^><*H~If|59^=>e$T$V;vG1cm5Cq6=73-kciZg-jTu5~!x!VqR-ARUj##O-
zzwaV5D5GodwaFedGbav^>wNBId<V%2bmB&WeFKW_*1Bidz++(%TMd7Kq6BW*m8H88
zrEKiRAHxZ+1!o+}D_%t+-6?2tydol#a1cWSCy5H(kEv-+c1DkAZ+Hla@#R3>;Gg-e
zUlV~jDoz-+7w61Ok+<8zEudfOd_TY}(K$>`u1H-)@A#G$RTMP@Gn9NU<l}v*EcFZ~
z*3OY0y!hEv*q^zy3aAh>&r(zosFB<kAo(N0NYC6pFz`>SR(Hq4TkN3p3}<pydiwOt
z%)2ld8JUpK(Ayx3Q3MdH>;^h96X0^-4=F)UADn)X+~fW#xi?DzpsE6NO$ONSFzHTd
zL4Y?1`ck@NqG;9x$3X5ft67?W!DBl{ytuy9HiaW+)(3qRfiC5Km0_TtTes4HuFF-|
z$SXZSs9IcKmYMf4dF$xjED+rY77EyZrftyau9M$B_mqTSAg|JNKVb68FtlQ>eg~3e
z3|;<xoj4d%2qwKofyGHLJ10BQ#1!|K)~%*Fah#pleqZ+}=}-6Su_vIFb44M(r+l&F
zon;}Uf56Pkn20lW-JFja2(Z1D5$JLDFkX8%8_$KR!Ovgcah`K|79Ps<wmJbxy_tqT
zt-tJIqR%>aLpV6uKee6drb(Q;ehU0cGKj;uH2?l`+9(1}xQLwb<KEtXd_y1`67qI`
z1Ht6AEH*~L+d<J878MyIr({77xyHtVM*O{zZ%Rj-0}X>lSQ~gvY4Rv<a<a>ATc{0(
zh9rBtJNmkVR^;U|H%21B$(OHPH$=Dg<?ynl;p*^GFh^mZ0v!Ha$SFHIf<soE<?z-y
zL@<a*#1)h0b98*vXnuVxx3?P&p!ZpuCXA*$Gur2c14!<vD|4T4K%%ICz*%1&3(+I*
zWvk1;{0uT@?fbR2R=W+Y8TkjoVbBg$yz$V}07<%qJL!^ynIx|s;ahP3L;rp6FsB&H
zmDy@aO+#ZV{C&;dhsvk6Z!}|`(>(}szVn%mSyD#Q-2ZDzSI_mee>`<I%z`J%qcMfH
z-znZ}C%(Tv9+P8^7jg>Y&L&r~^z<xx>t4cA%T%0203<<`hlh_xKgmr-MSdT_rN!TP
zplq$0xxincQ6e*?e(~X3$DoquYV+mMkS=beHMrz*qFke^vyZ85oRdaeSF38E2~k6L
zB(R!TR1kC%R--eLrhLu$z%^IO{Tnu!aItreia!d9+9RC5cT0FbM$;Q%xZzIhz^+p^
zy053XeLDJ_Tm2N9<he?=qm!cv4rbwTh6`fxL4}q5DT3>#SG&aMs7a@Ma{O$D2z`b%
z@m+q-VxOtQodS^`t;4ivvk8w#i1PSL$W*saaaZiuJ_*&kh^Ee_C?!@}90iPqnIy%k
zUd#;!1U+<Jg9g@R$q>I3qt@1>Pyb#p*bsDwwNKmD*S~6auB4TANKJ7~)!H~yi^t~s
zz2gl{GOU+5dvoZ@IfobwGvn)b%u0n&^p%dj^%q7qW^nk7WTP=5ujXhXQ36T~*FUb!
z>QC08Be@0BWdb}%#qP-`&kBnosShQizYFfUA3uQJj@X2fBoRIPNqcRab@76lI$3~P
z$`VhFpKuN}Unc6F8Tn!DutGjmPJ_QgTQZO(np2KPPFlzon32DLd?P=FvgF!_-fG}p
zLtD?4-HqEEiR0x-7j{{AvBj+H?QnJd&F{)6Tk9ERLU04&OjvTl(FVze<EhVNLdE#T
zZ|GHCF}T`0@@TZ=l5w<ujIuKti1>midU@qtxF?9pr`9_+7=#)at?@P&C7EAcgaLVJ
z$DU2dyjWf&c8=vQTm{B;`}~BDf$Qr+f}CYGjA-YC6zQAXaz*k&g^k5pIb7h*v!E^<
zZ~-Uz;NGktvov;e@ilet2w37o6&0wG@BGIFr-f>|@9cDiSS06ng_8V``|bh0$$YuK
zlAmL5JYUtyC#i1cenvk&TtVIt^~lFl!{o2t>YfxB#zWfjGS02MMtU%eV!QMEHVuc|
zp}PZBZyYottv{tRCUB1(OxX@`?AJ%HG7Z?U%8q_Yu4R(Mj_zf{6!e9r=%j{P=Zcl}
z(l_5<zK!yJdwb&%_r0an`SfeE8rh@V<2$H#j#eXh5H<-H!w0*|?7bK}i^8?1mhJwc
zmXAmZv@bg~M5X^u3NxO&Qj3pWebG!8eNAyY^~KVsO?+E;>fsa7p$x%Viw!@<=w0WH
zSTlV^S3SMsKs1)7-;lq?#=%%o1@Ct5j%Ll(`yiFZl<tAt{>Di9^<B5=zB^WXnmE&J
zvJRmn785rgJPnsEbtS*JAtH<}QrqVwBlXVbq|o<PgWn)Wpr{E(AiJ<5DE;=lo(28k
zn3rgAM5=+#!2YS*W?{NRZ;a!@$Jq;S5w5Je3?2QZK)VYPPXGN&(&y@sWoB>v4q1Pf
z;Hm@SghF)=M|iL_iBzs;hZy`#AeO`!k#F|*Pq>BWAkB}5+jfu)BaVTxF>TsNR_f4o
zXRxU{mA;@UkG69<KfT^_MD`tWM7fJ}d*8o(|GuT~sax#nze-+x>GdhIvGX+cnZ*Sm
z#!@lVH16iVG4hg68lis^sKsOPz2uZkF_CC@vt+Jhv?`-BIP^`A1A(Q%kPz1Gkqp-P
z5=PeQ%6fR~r?sNXod^ei3U8Q;+*EL)<%A4tu9S1d%{9WdGI_s2Pp4K1Y#ahBx0<w9
zsoT;+YEHt#KUw4Grq)OFJHrR{J3~w9c>iiEj0k%3JSGodi6gN4O$Q~)WlUA(*$WqC
zJRiw$C>+bmWN<ho*PX3V!_U6UP@)#qp>?=Y;JHzc!(}w6kx*kBm0diQIS+$$@g9Gl
zZh1|H=-Ze5Gr&Y6&MKRfrqn}-?yf^HuDqFDU)*>#*T`vhj(%-3%7^?LFtH5<J#?Te
zZ^|>8;7!4ipVXUHWu4<v8rPfc9hyc?pyIgVd_d<7r9=$4FLNi9SRG`y6UCNes+-En
zm0#|#TIP8^d**o<?zWjuB4NG*XRNfJ@!aN%F7(M^H-7Soc%&ewVSCEr?Jju@^$j}O
zy}NJ8h;m(s#ow&yle>RC0H6QgdkAn823slZ#yT0*Ee9_`rKNcb>uEWRIWW0`!j!!O
zv59AEbHtC<siDTmV3)twi7^D;VHHqTNcu{pOG>W3-T*v+bmWllsf%t}l?C~fqC!};
z8uBk%-<{;Xo>;*q{b$Y12N_S$lHz0%bqm1>wQ(oh4g|z0XR^ACcw`=z8XBL5{KRQB
zO7j7Mi0#-+486BSoW7s7;X=AGI~ciA%_9El60o`igD2=NcQwiW-f`u94^7X#O<YVm
zGLoa5VMarGEMACMTwJ||)y|cVRsv(Jc)Z1h<mk{JmCx}YbL$2rC^Sg5q{@b~&`kkr
zZCVQACfNS_d0?;5kTXp|%&nSBDfz`hB4&|E)&5mG1^FC>Yy_2y<xOCI(5f1`E{oiy
zq5sXk!+ZIpzN2aKcjb8l+v=8{HQc*WLr1KH`6|IHp6k%gh?*{+_P2i};$X9%!DMaS
z+|*=iYxq=IRyO>u*8o;ltSAh1#cOK0(20Rtt2SILrWoWkFSE={EsqXWD!mcvM3J@@
zxH;OBRm!O}r9sXwmmF(jlH#*WF0n07s!}UGc1vnjN@<Ueeoq}O{T>Wi`S*(GCJ&=y
zTs$Y>z`!-*8YLPH1%`*KLYJ+4_%$;L?lVtQwO<G_@+hEEc<3M?xb<k(V1$;iL4^>h
zYtg56E{*8cXQGjfz1cDw0anVZCt8BO(KlSiUtiz3bF`II?P=V>RG@&2Jm8Ps$2+V+
zL<7}P#XFN?m+~fS4OxuFdh`3YHzshQGmF6IuSJi4<!a|s8`*HLSMpSBkNa#4XGW-p
z)r*lCeMi{XNkYUvpvjxmtNi)<JI5VY{aj|}a4uYM>HFW855L1Iebz{ilV|YeqtJAN
zD2{jiYkBx(U0CAfq5RD8C_B-obR6>Qa_~L{ycXMKV9I<hu5m0MlV>Z~J}I=Xy{-%2
z9(bxnfz>D<d9){NM$f?muQB|0!PBtQn}VUo#1azg!DgxI&h9@=ZD)$lfa3%4uA8*?
zax)Kerr}-Qj?Pgi`i!ky?X8~fc)fP0X)#vHxqUI!`nhjC{<6K!-o6I6Rlbsq_WE1#
zPqUgoa`PAeot`vo!w9bEe5-<&MTsXh;@8(=)H|~on~#Mw%02%OyLYwM{M3WK-MPMs
zPKYTQ?Fy`<KmJUWCsOa}c^e^Yp$JDX$?G2~houfAq(Vx=du%g9G{r5~S9uZ|)LXxc
z6dG?u!b9q+(Q*Ih<FI^h3TD@J92C>KT(P<^UOx3a9zLIptk5X;o!0xV8?|OGN9r9)
zUgPkwg@MhnlI6^GGdvQ?Y-VPO=w)M=RtWrN&V_|Dydo&Y=fJ0;^v6IA|9=E^T>e3u
zMR%;$apX%@@62TzUgq1ag14QHGVlAucyImW-kL0`IC5U*PX?Py8TjJ-MGBUPeQl*-
z7nJKU0x=bZZ?<rTgqv5B@k%{$;u2~M!O!w+R5P)Y`$nK<N(ohAs-Mc&e_Fon+{(XM
z;Fjycpy<!#%l25=pAYS~%H^{?Z0yI#o?6<UU+kG1qqeV}&Ohs{&c)~{Pa$#F^P`H-
zPV9V5wkp(ld0P(MRu<{ayVeE^+M(on(e0@#+S(Sgv$Ge5DH-^mpA&k^e%NEvg*?U|
z!wK>5RK*}FgxZSmyb)0jmDKTvJB-PcLk`*D;L}tBsiC3kbi*W<C5O;%L`l;j^#)BG
zqCL?~G7{ZguSxR^x)Z6o#-y5#UKHkU84aQA-a0ut=D+9)V5Fn_dMQAtaDP4-IRJa0
zD>GhV=MsYMbRSKVEd7K{Zh3W;@#)i7+`^3zy2~TnN6)0B`+EzVlNNc+E<T5bsXuMZ
zAy!*+_6*`dajmuH9!ozm5fi(C)y^zX{fL~}J`o_ey1H7T_~)I&^S52sQ=bc)nqK8S
zR$*_x1g|X982V}Pm>6iwFYoTs3JcHbjhloy)!p_Z5Gs1J@u!Z>&CO-w-^ag#OWb`S
zeD4yBCjLEPMKmDia(?NxyX>U&^MDOX%xwNKv82(wc@X#si&|<bJUQ2xV10Fw7DSZ<
zT|AXP0=ZA(dKCu+$hceQFD`$!%mwiXV0%w%>YH&|3$UKmd@q4dURkq$%WTde`t-*`
zV6P4)#U2vW80Y7<$B$g9`+;ZY=g?@rCkm>DUrl(UV;Z{7XjVC9>%c<&<x+mzwD^}i
z3FOv5Xv3A?n49C%{40<3O81RFVeTD5u@Q~NFf!x=!mD-*?S)xOWTL4;x7@_k{zvie
z6`44A;Pd_v3Pvf<jY}@+-tLC^MM9eGVM4Eyd?nS)othh{CO_;i?1%aB0Vk~}c2Uu&
z3dbSNrJUfc@wjQh(6YB_{lh8u_9wW`x}*NKV7{qoBj{h%3_m{L{uBeU=b#eLDw~Y9
zij0kGk@sWSnXyw11KVT5@T<%M;gpiXJ2@%D99ldl+bW_K|9Xb`ZOi_$uvcXN7(?*Q
z0!Sd|A!~g;>saBiU%1d{nlR_^uPTR)nal`7QPJpK$MtvO*$oqJio3T*!LB)}0Jt7?
z-^vet;K77n{42izn-`<oyQ9X`^JkzgCw1Wd9Md=lMbF^)CHC5;&;4#|NG$Lqfo~n#
z#csIfp349Fi?k*TRR}fFz`P~RzortZi?x}}N823Ri~|}3OG{D!EaVy(XM*eaGYsmz
z@~w1ah&~y(82{^<-JhY?z2ZVIo8rnebHiGnnAxEF_djw!(CuB&HHwUDlckT@*Uoe+
zD8vae$@=>Ht9SYb=xAzogARNuw9x|EMn70Eu5z<<K#qb;<oniEX8Aje)yeY9X2z0{
zA!$#~13qabJy2)p>DTZz&3^ry(xkjpy_D4{MSwmIK`n(I99H!bt5Ds3;p5eZ56i3e
zZ^_VL^BL|Y=y89#Q4Hh4yHP5T-xQz&IG-Q<;!hw&bV`b?3#KZ{FlS6+VgcAP=jZ2|
zWcLXe=;#y;4AQCSfpio%Nvi*O+0~V|l=8cGY3)+)sNqvZb#($#k&TUwTduY3U*1v2
zA+o>&z$NF%`5=8u$I7Z;3)hasBPF$VAb!it!=nM@Tn|oHw<}ql_>v*uH6v9wXSHqT
z8fG>g!19T1QPp0(dGpbA>vuXCzbhEGS{tX_Y_14+k!%ZT<swg!dscXY1n#zbHrW=Q
zK`%cU>^0-K_rj#=OkX+Fnfj?-c{H1rkJfaG0<G4fFjyKfk?A1EKd`f(W+@)x7z9jT
zsd`fcwZUswynF-`EwGub({!70iKDUrx?P?JU9`;dchk6^2zmchR#skGUCjYXS=wax
zUoo>;jWN{&finObQhn4v+{dW`&)08Z!;6a;K@l-0%C<;Q4}|$mKIP@rq*X|Xrm8d_
zd=0z_0qvc4m0MfZ*B6^5k(s}KNpjgv|6X_lv;3f2>zbFA_Ue;CuFvI(dMAn^2C1N@
zszexR1QnOX&>NuP1D-oC1B~}y>+y>-47!JWHwGxW5r}&>F%+m54Ox}ctJ@|L>U%2{
zX~T=t!3!s5NQ7JNFSH}&ZNta~^Y391=PINM)m&^8;N#*Nxd_0xn%12yfsX9z=t&2O
z-3X9y%jbRMH`&4|XEXg%Y0mpF)OB~Z2n@Qttt}U*yb%ky|1dN%DhhkhB%wHuHV>7{
z4o4%F`OHXQ4V|2FJ|`1q7xN!*f@V^Zk_tb5yh}|*We#NK!7^vz&13W(d!|c*q2}{G
z1n;r#yfG+2p$2BRbSf$;Dh51*wc+fw+nf&^sLWd{y@oY>bfboO&jNkAjLqWOzKtg^
z%LtFd6$>C;d~kX?E{Y=vWg=J`bwor&7?_wwfV6j|{h!wbly`EU0o%GbK(HdFA)sX}
z2Nc1<<jTacaDBt{ixIjxGoq%ZrhobJZMg8w&&etqMhJqEg?zLy2w7Bt4DqbbsjjZ;
zgo)+vAL*kGC?gI%%@|Fq6V?Lwcg{){JHxHO03tKtbH3UTkah!C60>89f;Y4rD~!K6
z>l9}9B=n=QSndY;HOMt_t(}=<gIV@0v<Fj5sbtG40d+e3fBt#e%BEVPCx{dQ-=3_D
zq5=~>*!>OWeOC>L-1(LFB=RT>#aD9@IiRKqWLsE*S&VFh`@<*<6S5s0!7)B&tqQtI
zNG_aQkV*=OV{*k}`0aj23c>D?40Bd`pF2tojcVxII1s}sXNq-eqoblSa&zf`hIr4$
zOPJ+ngC`qfMTt}va}7RORZ_CDz)bDt8af<N_H@1B|M>N5*Er`jTx>7`<-7vl>kG^A
z;$)<d^~v@Wf(^4^5hYiIh=_;*0zsUL%lfw>QW(3=<M0TmmordNy-5bD^#kdm2+|_d
zCb;*0J7HudP#%vT$Mg%2f9df3F)r>g*ae$;IVs9Ov>JqI<nT%wE~JHT&N8&Btk>i7
zQDFH|24!yqU=+vKr*lZDdwM4*5S#31yzwUH{0K%-2HpBXf_qJ<L1AHIMex~tAj%vw
zxr4Ohj}LJzxk>%Bz~Cj|vz`T1>lxuf%5UF(YyScpsfeOgusqqCxB&9=K{Mn)*JaAK
zRu1gr*woZiv31zlWAVK^dzxU!)&Rlfks@uL(Gq?APX=`0#YlK;M=C6akA@Y5*TL+L
zLi#rcgasME;DDEm%se`B28wfXl;R-+%<$>&KnxThjm_HKhT)_2VJ_&#+ht=!O1H+j
z^yLd!L7_HWK5`^WhA^nsFI-w)eypVNMr+{&Y&N8EGBU~>Z;X~dC4%XT`V;!~GbHeA
zT0eXCY-6sm5lm2?*KWOr)i_kFrlm!U!HgpG-8?#UYFizcgY#;#o977{8rph-Rk0G#
zKb`g36-2r^yZYl;H6@g{OwXQ)48=dXhI#i7Wyx^)TsVW>y_OZHryNYcytKPpl9WVR
zq*JYYd$lTD>AJ1e`|X7&aDH$;7C*i)u`yU09@tR(cz$j>(~t$QFD_~Uf$4HV>)$_=
zr!b$ey^3_6Z$jZs?j!C(OLLWyl2Wp?%-?e5$Dl(ILPi0=r^HdoCG)A-OjanpdgX`N
zswgS>8AyM7&Ny_Ebt4eNTNCBX@)%8oz{B{l<1v3nh9rl_Fs|jT5+!kQ@z_ZkDyp_}
z-HN9q5Nxi@d0klwX>9TOxO*b^u}OhJ+D^Y!Qc}vQTAzZKWO8i!{y2K%w*g=dtow)E
zoRHK4qnaPiLrhH7#g0&FsT}!ahDtekdHK>4n0GK@d8j1-gIemtJ*={qSTIIBpAVgE
zA?#Cl+fKDJT#*m`>@twLqmnW*!$2_{qLKoiqVy$}UxF%Us;Lsn>M4+|2h`+5S=QCH
zW*=;i3u)2l7#O2J9gNyLI<&~{`=JGmtxC$vvmB!S87<I|iA><KQn!8f1eSjXO%EnS
zPe~~XbmPZvt}kO!Q+KNc!IUBOH9z18(Mi>8l_7vKCuM=*ogc4&LOFy-Oq>s11nH5w
zdwXM?O)L6>@EJLsR%E3Zs$e9rx%~;VRH;buFj5OJgX&+x!qg8}da0$tg)VYWc4im>
zNLcT={9@N|E?#ch8`<WB=9ggbP4*sY(|xdqrGc|!MT#ie98Ej*@8erH-6s?BehZu!
zJvUkFL`N7zh9Kc?x^H4^Z2njL61aR2@q=+uesqIUoAd4UDI+CiMi2M(#qn@7y~>kj
zjI>E55I4$)hK4GDY1#b#Ap&GdRrK^yh`Fti-o>iP52fa4NXFyEy|W+FKn+x@LUPXM
zG#Q(WFZ1)~2QJ%_YOR4d8h>N@{U(V!AdnCA#B+?o?eql?eo=zx`R+x|aGqL$x<z{s
zzO=Novb=nlp`qbOy6AmfukA|9(fpXvM<{0x(ChvUrm*>}Zy(}H7f_rbhPC<T4`ox^
z;0pwW#8C(D*;r<6fyVIYXuN_&<iok%-(dyfvEyN6{Ky%h7ZTEoh=>3YASg6c4Y<<d
zCR7|cu)UCG2w;Fjkwdmc_qsFqQvVqabVwDH91oVd)V}py0-uj?vT@y>bm==WyA|@@
z-+zEqFr8bna{)gT`Atksk%3vO%vkD+kA;PWG%&c0(SqYKZ9A#RmA>SH!3>FU=N`WF
z+*ezU)lBlrwRrf}!Bt4)91V~^_zr|$({^TRtycSz)g}HS>S!Qb21fs9#%=Bx&kYq&
z6(B@f0k{LyUL&rqp=*=BvBi1m*CU&$Dmi2{1j?E1dB`U>or2v6gHhL=Oj@OEx3#@-
zSsO_I?c2(Rv_RnKkRXR=+Fn#n;&1-(8TbiU(=HwX!A6rmYGkJEOic{kj=DHl455oP
z${FWjIWT20Kp-3#m-BA@spVn^6uL}aYl0x?f3HnXPam0?%<pE|9Zg@e6e)uTyfhH7
z-RR|+Q+0Jc;Vm}?76*_`0=r3&=jl$-+L}4AJFvLYxU}*k-7$>ywzK2;4GrLRJQu#u
zj^wL>i<+>J-nU=)woV?Hww{OVF!F(1mLqgx5)uV^^_~zg!EcWrKNdK|(s%0R0#*Qo
zHKSL5Qc8p&^~(WDOPQLiv|8=bI67>@RY>9xHR(rAP3|_ZYZxr3bmHdn-0t1GcQMDc
z0EdJUb4l~^^0wQ~Ra(V?h0b!D^^gz^#8v|c16-#6WwBq+9<l9e1!N9u0K5sRgfts{
z>ya|8IVqmfR%3<q>Q4o<MMV5TUaY^-An0{s-k&0vAm~-)d9;>}3>JU>{K4m4CRU3X
zALo=qZWPf^nDk?&Q&l!d^F^xa0_P+JmXeu6!PmFJwq~6MJSzA1HgeOvN5en{pCZk2
zarqn-@D%_B1hh}$o6s+~)-D5XV8*prY@NXi8+<}Po_BYVmeXNcp69gM7nuoM&L-F^
zF*{BmkP`Dbn}1nY++}|I<CTW4i`tjhv%B@zCsWixwPyT27Zo+m>)%I4M$ng;%Y3gr
zk9X%9$HJG2Ha0JBE_dVTLQ<GPV7<OPWrdWYnE|@_ghQeC{ijzI(5E#AHU<I$Bae)j
z+6gHb2+U68uWj#bmG^sAX@fY}N$w?^$YUpDhEmE2RyzTp3#vQ-SUiF6aXH&F;7yjV
zp0QnCSz&nbLJ8<%ex)kXtws7kMtDCGEBpHT0{<HeOoAE$NeoC?z?^k)WPKc7A0cy_
zRx+gkOEwDK98!?+IZIYoSEoGs4wS~Vz)}_eTtJsHLythjGS|)@7SoS9_@sItNCgB2
zGCq6O?BtUypHpQ!lmGMQOQ|8STaUNlbFrRRz~(BEA<K+lo&WL_Y_POc-wP?9i=#mZ
zRH}hfoP6A=?>s2RE~g@VdsV(#l?vTxW?K+kdH@(zu>E%yN(?|I5WSEdNX||jbM_@P
zl&dqG!p`^XPoSx^bT{v1Je$FL`mT`ApV<^r1T;<g;}9m;#GJpPjlp<o+HTs~+6;Mt
z5VN|aFn|fB0FMf8W0MO&?l-j8wF32?$L!ScJhSdbc{V4zbGy~46wp=a>r<zTBMSuz
zAS!x`MxjqOK5h$71b!*&#0j_yg<60|;3M2(<3)9w3y#%2XJ_sTwL``Gz)=cZx^4Yd
zTB%m+<VCLF?E6Q+^QNf+1Fiu$kjs1^%@CfGoxSk^!Vww(83@QijM_Up8R_ZsMud{&
zbA-<3QKg2EIDe%MJ`|t#EXDW2Ldt9<l-Vx5f`8{{Mo;@(ka!5eIY5+%tsDU$><$Lm
zNM!-kU-36GYShM$eZbVY42luP)UeXfh=UIpNo6xrQ`vR{z*c|mG+3Mhqs}nMWF?;0
z$uypbp@4lAIaGICWROUcjnj3Fk7d!#2M!#Gi&z`4%0TST{`&PFg+vf&=vi2bN1mGl
zlw;N~?zK}b0bm*k2`Q-{c}BZ;uN&wo+rk-zgl3(#fG91Zhs*cs%xQ(C45TvbQy~A8
z1u(N1SUps{(wm6*WA9$}d)qaXj9+~y{#*+K9lg6=fl+n*_}WM5_7v=;h3eZme($p%
zW?mrVE}!5M66Va$&pS@3s+#m6-O(P+jXIL7snj<&&%HR_%mKS$y&t*(shX;?a-<$~
z?ecH>`b_lG+V4`MpwFN8NZnlbd!WC8D&^ECRztD=Y#|sSI0p8+2_hCm(V^LTFBzLd
z5c535D^hTBhUsf(d}p_<Uc2bd?d|P3hGU_lM^aVm6hFS~V<F`M-fP+y06P%dD-Bvd
zIZmm)`QO2XAj1e@4;yBCd;2=r5@U0>H*R1XnWREs^OU-dstIslW+(&EjoWJshMh5W
zb-lT=Z`qngrzT>J;A6eA0eBb&Q|$&IUtmi3YEQJ(2Z)lVY=c>p&$dW8RnXDVi9hw&
zn{RgH09${1qFf?!1mq@<FN0Rbz%Yk^spbHG!w7`SS>j}D;ZsotAiG+1fv^gdb;sfC
z-!n1<r&ccnrG@|=DWFEq<vb%Y`Sz2hXJR_CUR_$kn;OrQ!tFolq=0lxPR~Ie8&eAb
zznilCy;?n(RM=Ms_=w0%UZ<6{lPNn50IfinWJv5)w_Y-HktI`%$=|$C7tat66_^74
zlo`GarWXO}Zf#qOIbk77HgsXgP@Nl%)KV^=sk9n5ynO%uJ*VT6q|^||CklYssdf1P
z)ThGlw!`G2511JmxrC+j(_O+TKChI??;dYuoYn@O)q`9=Mpmo-@B0CJsBzh(k-FGx
z#W@BWDtfYS+jSG9K`ii+a>92Izvq$pXaF*PZ~XJu@;{Obf~!q5^ak_Q3Zj7&HK2I<
zr}MKd7{JsaQ=zZVM3GRl-vMMibc5qEto2J1Vd3H9OI=Z5$SffF@v?QR*5Py6h(?Oi
z14x*nVNj`(BY$zW7%oWpN|z<h_hOwEBswx*zI=fqGYg7q^4nhPV-a)iS<m@iZfjLq
z#yC1T39h;>v_0%i;F1B(xjIb@(LDszz9+U_C$>)<3Aqb$H)<MU_FFJm-K%Oru&H)h
z_4GxuO~|Y?P2@v`^+cKD<(HtK18W74cdW0kkK!V0PJ4!U?kKQV4@tPK2U>B2p+exb
z03MLU4s7M{@$WTN0V2dTZo9=|!2}**)qwJGS<i}#T`SNmH}wRNMc~#KD6WrW%aH&t
z+Qo)R^ywMk5yalS$@Mzj@!TrybUxer<9H|sTrhY*&34rkFoKQs4B%=eD=ea@y88PS
zZZF0grAe*eQ-YKdkV4YSbT9<8`}e*2nECj$qG-VFc3@!@Qgk8^8nWQmB>7bj{NOpy
zO=g7`XF9@S`HL6oV*p@ry=#I=DFdv~5kj<hE(ns*JfI7%)!<#du(&vyA%P<m*{gnF
zVgV8@KIgTkQUF$d`TqSoV5wJ$1@_x;M?J}rT_%1!fbE3A7`970$%jLTIJ&2y%Wb$7
zXrTg5A~*6fQNW{Y)_pOQ{WLHjfF1-?kQy+%+ya(kg`PB)1~7x#aNAK}Fq-Ryg!#Eo
zyo*!I)sBygm|EE?-1iqA4;v+c=<AAncQa#m@F#D(kNlZl<{I%&o#K4xAtqQ9_ZLtB
z*a3c%v>*kZg4>OEgpwe$(Cy!)qQIRDuMg*#z^6vPi($BcumaNdF<^zNAQy5v@fcOp
zpl4>5L&UL;H{M>`A+ZfIy8s*|F{gPNayRQY`XV!EKDQkmTaeSQHQ%Q|K9|#|1Jeh(
zIbY@TJjw<CSqMS`z+^jsGk6={$xn)nRMiHc0XZoZ0L0)E5agTnrx3p|{7568^gIlu
z00z5*iM*;lcW=*rd#3jEeMz!f{=tQMKw8>UX3erU$T3aXH5M%+;tBxQhfIjRy)fbg
zI}R+C2SHnKa4^JJnF=<X`R;6e0=G>9GU5Qxr`k0_1g7PcuE964R;22pWod6Bj~9Uh
zzw_GH1WpS>hv%weeaZZzU`I$Hb61z^<8kdIq?q%bknAA%;YY0T_HFJ%TpIVAnQ}9Q
z%@*))Fuf3@gMsBw1JH344qb^hUO@-JfLSt5$OeHzT|HL4SSJr;UFKk@k(q#7V`O0H
zUT8zUo|_@I9{8zRU0BE@cyqSsd9~k(uvrBhDOj9wMP&0!+KvGdF~XQ0;o@ch1e)it
z*zs=`Qb#}uHXTfH85u&cNVwMy`-R6r&RSu_-ieJQ;ea#W0C~G@{joY05&Ii|z==yB
zOSYh-oseze@~=uYNG<}5{v|=3;@B2Z?#2BWL3vJaEledTtBCy!djLEECM72Srug-{
G_x}&xIejbu

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 aaa1c7f40e4322cca9849a0310c5dc7b01ee325b..6c9247b8df5ed905584a3328de0d98c3baa57ee7 100644
GIT binary patch
delta 89108
zcmXtfbu^#<`#%eY>FJs=rn_s{nC|ZGHqEW4O?M5`-TBg{ySux+bo_4b&+q$(a~$VB
zcU;da9&rimgiq{%FSl?5#NAHP{K+qg-E0uY1!cdzg@zzwp+V<mD>8jmE)`(keiIbT
zl{jg^6yN<EOP+JB{nry94EjW##QLs$dFSlx_nb+_Pj%)h<BXaqXKol$e>5S*^gh$g
zlwcWqc70T9AHDYvtM8yqEsFZl;s$zB+1i-6jAv-W{K%Y*Hm_Ou0Z-zXKE-u*S~{!z
zg?eO|d+g@0tS^v7yYwaQw8`D<{7G{PkIKw0r_PLfm^kt8UHdY9(ygk;N8~#E{~C@2
zbd3ypSLJNtXOG+4v+eAzeS$3iP0G~`I9viIB>c#3%ghZ(x(Dl`y>eB%xBZCK%1-5M
zZ@tkOJQR;PPmE`-f$MdjxB0B<+|MS&S{WL$`t~?&EIE9h&rN<5s4Zf0UKA=a#l6$Z
z!B3IlO<H;h`Ff$ltX2~-hnt~<I`^&(GaUGihxb<RDPp3sWFT7W@1P4-E7Q5+_A}zh
zCo=&v!s61$uULQjml3poWvFDCNPL+8Dj!q2-0Cec_UBP}60jPNxSWkioqv!5@3o6G
zSGQ^eLrnC@k6-d#>DL!9obT*JfB34rBlWL`x76eh_2W@lUqu%>nPE@X5}q7miCxBI
z*E`tkysCjznoRtWUyw;|z^hRgHlIXy8umoYZvA(NYY<Vk-Ih@XOAW_&%Z)z}9V1yI
zgwLYe<q~oZu*^LyeLZKv99))GUVmHbC4ieP*o(>F8B$fG5^P0RVrhGE2`}ki?d}^v
z2C%ws8kcPQ?Ot_|WXfkvo7!;cEw&6Ylxv>_ok+_(8(xuJP}c-1cm5sE?4|KWpu~w>
zN2oroHqm*8qp{G4C{69^euN~ofC-|zVq*{_&=d}MuZmo{TUxS-n6(a8leeZ^hMh;&
z31?L${j{i|c4=V0JpV|>=ChX(R@(Y;KrmPSY%yi!Pfa<w#~BdJUAcJpGaQbn33x*7
zOF%&zFUM*L<kbm}LVU@?MeXQSETMO#$M$b;8y=V>@TBITw*jhi2{y(q8aM*nk2^Rp
z*uZeA%YfKiX;nqVPUp#tpLloNX?O8+A*GsFRPL3{C=N!2F|$T1fq$QAruKvrFPoEj
zl3h4HN1U-02|3v@YZ1SdR7b;6ZkaAK`cFf=B8oav2A1^WN0Bk$+a8(lh{z%7`6cP0
z=EEMNxpHu}lLW?SwE(hvM7QZ^mFY0$7NCA@UFY$Xup5H-&P*Wsx06h~`y+d#XLyy3
z@Vdb#u4)S<wr{FlFXwf<ahtD5?KF%hb~9c#hO-^tC~d9&KBe_&<W|eU*4BQbrA$TN
z$jUwC6a7+jh5(rW0|A_iV6VNj_}^DOQ*bsk5yuN{!KYSJ7zVBZ?3P1`!BtF2K$vA3
zSF8bzKu0Fcm!D~eu>+Qg)1eaZ*z4=ZojlJP5z~JjK6AdLA*ngF*y&4Pof}TIivInO
z%2W<*cbT+S$l`y%=MQpjDedb0X8$lcEx*8SbpdnYQS&NJU1~jGi8f`{dr3e*c<^lM
zgCV6JGu{~1_HS&=_OqIdszac~jZlprJ*Mwl6+7x__hO52cg~optMj~$qwB7n{X|*i
z@xZa=LnB9&`%X*>cKBFkXC?zV&+Vrjr-y(0BRu6ldt^h-nQ<>S51P-{7BMVNw4FVf
z8;xHti(05v<Y-nsVUDTKS6V@NLl6gaYS*m|S@CD9_1UT8lFJ22RPq7T8g+?T-Op65
zc6*UZc-pzuCrq(*5OtrLa6ZauDZRa>pPaSlWb&o{*#jf5VuF2BKj_O?8LrqcR9r4`
zccp7a>ge*d-ncSodwovGuTQ&d{8anEhC@4f>hDT+XQQ7L7}eiMkT8|%Fxj&5Mf>>>
zm6ooAtI-}qVx>r;E4dHI5h3}m@;kiGFy|vra1rHnrOC2se2Yq?zrzAT@kA`c_Hhfl
zxei$VXzR<QF~tb~9e^-T+*l;(QF^`$B{deQRmR7)MHG}HO||XAbF0SphSP;Kk{+cr
z!FDS3<vTd^hqqhmD&~?D8^}jB*zvq<#}ao57h^vqXb5@$;G95@=&oFM`y2RK<X4kD
z>xs82L+nexRAFE~!iaqnR9pos6ij<WIvyxZl}+~8*PCOu{3&(wI=5;~yT_QC7M|8{
zH*`<Ox0EPWY!(a7P5%^r{gagFXtKsRT=Jt{j07gUR8@IwA4ahe&mAlzm*&~E;Fzu2
z(Y+p@h^_!Dt^TpN7)~%Ewii$<J+q<Ayk-4eV7YsG_zdN{N0ox8Gj+D}0;iwFXY(N^
z`+zbA2bU4<_ikF@1sV$lGq$gUQ%(=AZ5&K@s3T}I{zYwJ`lO6R#f203rF;A$&3@7_
zC}n!-20?}72-+`Eq<7xxu!Ctna4Jv~ScU?Puh6huJBqK4wP7F1&oQF%P|q%^XM%Y#
zBTDmOV)}x_bA?K#SNHcsPk!m>?y{UHa?*oiXr@(We9cD`oMC*woJ!aG1Ky1*tjiU#
z)z{y@bRqB%a?DcF^QZ8u?xZ9l$aq+=>e}K}=aoz0ik+8AzmaGODsBefe`<FJ9+NK$
zH`(&W)YtIF(`OY3#Dap!^R252-<65qOcr(mdGw^)<7Oz9I>Tqe8e(E$q#Qi2aX4(6
zAAAuGW`ET#!6m5g+)6jt^9!k&%J=0|g~RR7M*-ZSoBpQFkHXX>npg<$tp7DlDBEv&
zyn(MRTiskc%Ax3u&WH)9O2JtMV6&<ZXA{`9;MaB!z4AI=8m50fd-Q5a=EH*rGc)tC
zBQBP*tgdICaK4=*H<4S*lk_jr<Bt;pBCqxxxg8s$4}OQZuRss7GBYpdBiIDh#R<*B
z-%ei<$L;s_%&F`#aFuG_ACWBjUqe<@%~J}wBcv9t{+caBi{jv5?ZE&S$4g!HrZ8ZK
zq$Wx@sxe52MRQQmWQmJQMSJaC;xI+j<X=zYGE=Cs_f(kt%ZBlh1W0>|ghccib{H~e
zDX*w*6;q^N#C8?+`?_OICjV3rIil$KzILd6>!3Z=Y%ygpnqP;@x?h5^pa|u<+i?#}
zc6sVPGGb2Jtm^XoNCsqIx-XeN-1qr3H}jrfSV+(P?da$rSjQAvbSH1u5TuSn4DV@1
z7h)TO2qZ3!ltKhmL{j?f8Zz+0dkFZ;`=GV}3-0@`P#yTjKd%sx&dVKasj9Rs#aQv_
zD6-e)|ILLwTQTEZc0Vqm&Onrlnvt3Dsh@oo4JUD~)y51=MuCHW14A9d?Q@56|D5}w
zy_aPt&PJL7PwFK`;gI9f=QP;@MXXjG8=Ua%`2`QZg3G{dSuY}R0Op_V{K%up`N#n8
z4v(Wag?j*npnp8UdD9KS*;KEui-X=iuVj#!k3(f+^wEVBQN>P`f7jHop`f5RbB7^T
z#RmI@CP`A90~Z%#<~7|USKW^V`T6-bDON~Gp-p!ZgmSI{0Rb?o%4ZCcT5_;FGta~A
zM>L$CmK<DaX5RAI(+uUrU`NQ-Sc!V`hJ&?9M#Wi&c_RAXKv3r}TV_Ya)?cb@W-UIC
zA`ghgCz%X&n9aZ6RT|3<dWV;2+71m&nDN2UhIL;H&~Dvi(dFcI=@1YVws%tgE=NHf
zE0{50wH#|0KT~2uwK0k=q~>zk>#w-DEK%Ah<7v^Bk&{D@kB?{LZ$DV`JaQJ4*W2Qe
zP`*38EOp)ar#o8Ze4&43qcL1S#sl`DSFvyE>VS82JXg6A8I<qLMk~eZG-PFEQwIyo
zdcQ3KUgv7-Pphl4cP;4&38LEC#F?8&#l^+eqN>If^+>Imm_mvX4{tHRvapUxJHTOn
zUa0#aB~P5DY(ZP@j4EVdCZR3DD;Gr>6vEqn?y9gmZSz4U@*YDpp>lwJ(rhtsU?&pY
z*_h=gH^0K#&APPBtB=)s2Z+~p9Cl6@JxerzckFD;rcsZ<!KT7mt9}_#6L<I^k5gXc
zvHoy^L4(H=Zl)xn(Z*qDZa7N-7ZrJrw*JyOUa<zR9|%A;8#o}{4EbIG8%5?n@oQi|
z9$#*YL$hiAQS7brTOOsIBIhe$Wd&Q5W{^b~cEmhkoqSP%k8<t4mXNY{xyIj?PWcLG
zUb=3<<72s~`j)^62S&{1J9LSp8EkcwnaynIrF%Z6;MbwP^aL~|L5ybbmv3w=xR-;E
z`)t$ncf{V_e;*50Jm2#}o3blznkd|q*Ins<F>Fuvl+D(<5$8o5t}dC;AiTGdH-CN^
zOdFMInzN5!FnE!%aMeti@fv{gYn)t&wO@bB)6$Pz`LV`>uhVqN(wpfzrt@|BnC(WC
zI-kxSP!5T-3Ra2@l8$ERl7~~D!#hn`SBWP>L|Q0TSL#QQF<Z8bdN4gN8Kg@y#JhZ?
znz;&*F)0{wly;u5LMjU1zTxN?sACTv6BP^P3bgoofe3ZjTRp*re)k7{;`&Yti3lp^
za)-~a9`1kdDIl*;Mtx$(Ixtm0Oq8rXUWBMi5Xb)NGh82;BRShCj*KM%@jU4=+IEg4
zqO+Dh>7le=_u1h}V;D=X)7e$Rz)+)1zzbeFreNF9N`3+Gk<3YwgX!mp7C4@faV-Qb
zl$M)56-B2u<2!-CCcSsIidHhzJyI`2=TrWcZ1j35<Jqk>c+%8A35FJ~d%#Ky#&JR?
zs;CsQ6FE+kxtqeAG@A@`lbTn_k8F~Ii0TODDm&;eeL$>?>0j~b`v$^8Cj84w_rod3
zL3+Gy=Qv3s8x%{nV2|+}9<JH>_bNM?^^fCMv)s3`5MUvc_hC;h`O5RxL$f7F#~8se
zP6vcdea)3;T0J|1Ayj{$pisD_v>w|sv9rvL&Lm*e_lWDE{F<vgaZ*vC%iNx7n*W1p
znU?PX{g-&I{9PgGQzDy~*s;2Wu0GLCC8?uh3sVEqs?l3v&((zO(V3XRYD0dH+xGRS
z3ouB4KiW0o|8V7$=|N$}T=}7BF%d8B=`ySb0l(@tcYN)zqq`kC$RK}6E8}NC_;FiG
z^SiF_!{g-)@?gr{k68>qe~+i>sZM-%#o&VHo!28(a#r)bC|XC>(+$}x;?IL*?J#bj
zoX)vjYaGkMUx&{gOv9zx?>+Q|3amyl0N%F4g0er;rh9B&M9{`o*=<vcn(63xR6yjO
z(nOyXKk;swEQg{Se8T>D+v<*N5nET<vSe4GOn*%&NMm{rd;7qiH#V2nC^fI75PMzs
zMAoymzX$I=p{M3?t2~{uI&twkH{1O=uiqnj)d^vxCi#1nfO2%h1ilas#~0v0CB6o~
z=WXW?AO!LJ*0ST{xjnOu#7m|5qt*G5yMU<_WFDQ7F!XS;43~K?Az^#^j>%msPF!jM
zePvxK4-@P?AfS~lI9hNd`u2>z@nulLWGR*&<j1DdPU|Dq29M8fnGF2~T}UKjUy6(7
zn3ieMy^d(iXRM^x1GE6kn?{beTvSG7hRk+g4qZL-V~0=sK1*x)mZTX_4M{!lTD^kv
zOK0zd^yCSS*mtW!7DBrbo&`*#sAn3djOG}0bXoCb3nmG*xPm(+R@x0&V;j=x1w!d2
zexW&dAcHu&I81RkZ5AN-)cUNtFncyVnQKZ{vfBGa71v_<2e53(C&`P{TbYkFo;YUX
zrEye|+?hGUbbVl<lhcg;D$8lRtXQbv`(k@stiS2g-BFQwu-r>*)}?>HDFcg3PKuo_
zLrU3=x!p=+RjW>-VMfJ**7Sas1BNz<2QQ}LvUfFu_*tOw8VG3~beZ-Vi+V)TCgq^J
zi?^zZ{xk<X4pdp*?872DOzd(zC=t6l={FcO<Gj4+GDp$7_f|q6yQ2jlq$af<ze+#e
z|E&d?{!P__DoZZQn#4-a_;~V+z<nCr5KmlVc+9UlhpQXyR8Imq$btk`2#e)ELt3lU
z*=Bp|aNO=D`*v2CWK<UtBm&WhvU^{pm<3xX(4`(G@+fr#{qCB&0>vB<F#RUkMNP;j
zT(EG<)z*cY0dYymh5(aRi}4%+&zKSx1UH}IdNLF#v{{xvO{Ca(D+Qlwz9H-*|5#6o
z2~i6VFX=o&eZao5)Ry$Pd~3S>E=_Y_#Rt)a39Xv?Nu4p=_hFGb^8W%THti3*q*o=d
zU=M@OFMBp0XYE6GRydK=h167UyRTaiS!uzluKjpw;(Jh|+ddeaI>(*{7SRU&=zEil
zK-IYh`QI@@ZifiW>(Z}vcUA`{HlBN5`=yuIhF!nnkO5i~n-bI%w2c>%Zx5e6Y2k4o
zJ|;il2^alFRaIjW&}GKE0nhFlx7WS|%*WZc{;Z~I#?SY4^sh4oBUNMz*o>-M*=TrF
zgt)`j2jfrPt<;1p3!iApUyjdZf7sm7-BAzSe$7vw54C+E#paWG?2kGE#fO?_?~j5?
zsu)E4A{MJnF&GYeGHGBwG9K!2^Rp57m3lky0OWV;Vqqd=`}oX2@J!2JsTo1Zq^3w4
zeq6E1B+Sz*Jljg|z^heb1a30&4;=-Bqcl3?KS4UG?P6kFKsI4h_j<-FkGZSqu=4%q
zvs&LP`VYtA8#nDk)q8G$BsJ9w+i*|$|D~HrL%n%;)}p%5w4vQ3m3wz?GoJuPmlM6U
zWKcbq8i)%48^&0}^*b=PX|AlbCG{oWm@%PqR9_$!i}u2k7`1Jb>7DG#7u+0UP1jz9
zJ(?0fiiT~Jt+amfxW>mAwxV2HT(MF#XG7L-{JYckJX*5yW73$yLWX!5<?Gi8uecDV
zOfj8d%i9RdwS{9-U+00s260}gdtdQ@Dw)@CfF$ofQ1M7z;7aLo;0X#={n^Qlt$g*S
znot~+>Q3M9$v!yLsnh)k=kEvcl<9ckz133Wy_oEhv80AeEC`4yUAtvcxoQbES}zIa
zXJ@&_D@92_nF%fnSz{^dhlJB74+2vqv~5kNjqz$bA1$-gcPkYC5iy}fO!9Ubu>T7w
z`h`utkrW|ANhi%$j3bEK-BpA4o;1858F03T8cxnqg?T`cFbo)?QqiQq84xji?BW8y
zzDJ-~ka%p=4()U~ns|$Q$^C~u{Sy;CB8XP4(_7yJvUV(2JVXZOQD}Y7uIU)=hR9cN
zxoI_FAKL~8cFhj7t;|q-`I!w^)DcS`O?OiDkC3yh5#A=9<+EYWa^9{UE}(@^@BeKj
zY}x;f?-{!rO8L26V`$PGdFd`92SmM_hPjwHp6QpdF0`~Iwm+}CAx^KWvE!Y&2Q{dv
z2Pr|u*IXPcHbLDvg6}=qRiuQL^nwYx(kTD?_&@lYTgI@udRoA3Zwj4NEdRMfWq(Vl
zcy7mnz#ALjg3<*O5p2V4qZBUilW)HkP3fDdJ#8lR8Gd5mK%(K{kX5>Du<wQ*PzlsM
z%m^v<5A?lh>)4*2sml$=BsPxFgtW@2?HNm2Qx=A`WF!3~sB4+fODy@wAgWegq#7JU
zFiY*ZxwE8cq5=HZORfLMjsM_tHo3w6&WKapYdx=uO)eX@){Z+quue!z>W|_(sJAEe
zt={AWV^B)m>d$KVJ@l%j$dWn#Pltyz;e7v|f%^}!FhLc?pwK1`!7M4$xdIaKLgVXP
z`2t%f=qEBZ^lZ6Eb*-~HotcAYZAfcB8T}o=D|%H-#!nGFcu8(~L}+($nP*pYJZUox
z8{YF!ues>yUD(iCb`nzu*yi42ut^ANqx4?cY<BDk=TP?QG46trFKM~7g1XaTXm(Z8
zoaY=*!yow~rrZN+64k)*Uok}{TlX*E-9H}|%(j<gdo60>Y<Wo7C&?LF$_Ni2(cB|m
z$Khk4*y==fghvJ&t_@z#Rk@;5WS_-bfd_g?c>2CG<GGv@R!zHM4v#j4BY+L?64mpv
z4n9(MbP1x=c?w%~qZ{0xq5K#NX2bZO$Teccv-!VND*euMiINnbP%PK$@J{T`9B)bp
zhLZhbiYQS99@#wu!`nN~w*h9vXKckhDt8VB3by_7xrzj|k5n^v`3pIoTXz_2A|mRa
zLbp9Swy_*Wv9Oja!^KPfdPZjPGHok(SWMAX$<(O()#)&vF-85Be{&r7<+Gi3L!IY?
zFwOF>mvsAv{OSLv8>pjqaPxMhONG|Ec_7zlpq9xBP-Oe;0@(m8+H{tTQlQJsY_$iw
zfV3-~pEf*-`tB-<6j56zVi?5_dVeyt-Rrz6y}8Ik)n|e-&&tfj7|OKp+0IywU(r$V
zLFyJ_pelT7u}A=aJ#R`|+&2(Lxz|DbCxU)wcNt2!)Rf+vp3#+^y&{@R-$G%k`ino6
zzcDnwQ-UTASPkc{l-w&*SYck+4?#@ix_w()l7#9V<IML;<9$0_{^x0w^j*Ztf8v}n
zgJ*1A84Eol6BOmH-Gs}x`wSkh#AQ67u7>ZCOuW7x7KDQrR+N{{{@dFefbWxt?4Iu-
zhO|}2c>LMsX?91}X}PYK7wJ-et+n)V5P`SkYqoqeP(@kEcqC<VvfX{SQkyzf26PHX
zh6yT(AjFhMc&c6W)|K-h!?~-_e#I=OMD~t1UKITgnmNDz|DZL;B7<_S$_yR7NZ@n+
z7Ng?+`AkIE;x)}2Rf5^<9~NSMpD8q>p<3OZZ}Ed^#pHn^cb$wR<qh3-herrWZmNK7
zoG%uF$J3pzwucEt%db`tw4oep_Dw#cfG`Ms(fCbCpr=ooMM=v1+d4eKgEbvH>)uF0
z?JIg<F!@`@R46zQv<vzQq+0?n?oml>jTap8p<YB5?iVCd5PG)#cVoTE>$~fMpt>46
zUz^XEf!fRaZ!U-r1plMk6=ev3IrQHU<`Y{0NTpsxSs#HObO&pAn+WsHJOOJmP>I60
zuXA?n9$*Lpn`Y`fXZZGO0ov;)7(xvwdS+M<wljo|W{3GPW(-Vq+|}I0s$pyn4v#h>
zZRU<jmoSKy(e$Ls5AmS&f(iwRBM2C**f%GCICnP?l(7;m>3D(H!&|EwMf9Xljd;Kl
z#C1YJ`1#Qaqc1}9AMz6J>pbfir<~q&a{VKdEo}ixwvf}$&u>@z5(XeoOV;=EI&qt&
zY1pl%ndG6?af>a(Vn_zQtB;MDN-P%rG=u{6xr0aFzNhQXyGI&~9+>j1Kkwcx`f5-(
zmra<Kn9e%T0*@<mztTbl7G|FM$1@}Z1hl={c^O?;8+|RWh5^`^y=fx~@joDZ9%&wf
z;pM*bEH1W^u(V)Kx_R0#;UcJjuWMt3YbJRaAvV&-Pkt6(S==D4NpO51#^g!s!-R{D
zOV>1)WPm(l!t~B~3Yt$w*&m^P-CVxa<f}jl7Xb!nNFEfAq2Gt>_y3_7OM1b^;JZ+M
z&@S;j*gXC@ZSJy&`j`EUI~~n7(6<c{hoJ1UnrDX%%24^L$!(!BOQ5c3!L0@-P$qDl
zYyYp6lb)ErMA7_@e1Xf}5=lj6?bdDxDksFDcJ5&NZ+H3a!V@UNPu#EJn2t6p0Bj=a
z7lyf%EC-<pg1NreLRm@Kw=~X90d|78;c%_@yLgV@-+pFiOOlj)He?OBZD1w2AqD*l
zrCKvMoM#GNMkg$F2@UM0Q&^@1>4f>GvjqP3367wAvH{N1UKdFbKeYu=?k|7*wFU~^
zxq=tcmtHpk@(U!3XE;sl4-siVr~qn&mS0JgD`rQ*l=^)5O?;+u<nL&-p=O(B4fA`7
zlE#T<Er$1vuh;>O-(mF%##ZnjDfF@2xB1}Pf^Y<Srb`KTR3<syw9r>PFG<4S_?j`^
zEDbOTwN$+Jkj0s#HU_xjFJo`yQv~ksI5;^QItst3;ughfGK$6%^Yg+32m#dx{Qg&b
zfypG<|7)55H=*B7nf2eucnOzGmzz9Y941{8zMT{M3>rOlTc;Sxp7Y6|Z^vr3<Oi}A
z_xo3+r14MZZ*N`aMprP^xe_3L%ju)jNwHnH(+sy+UB>Xv58VL@@OEc5;uo+avF<FT
zutR42tFHkN_aFf-kjYvQ)->~2eRe__PMtd^B7p1(8+SCw&xkeFC$MEWlW*UcgL<xu
z#lCzs_GcM-BWC<ttGgzcY7#C?f;{_aDy4#5PIGMKT6#+j7RX=@^nRtZbATbSIZRbq
z%6gz7t%?yJ`nh%nB=IhLJeE&hXd=SsNA4(q0jO!mGKk1+SUw$4f*^~k#_8TQ4&t{R
z!T2xqk1sFwNSda(g3QXt<vV*k?D*+ZiT56E#L?D#Q3RRf-kmI~Q!%#GDf3WuMi9Xp
zP2c|~AlIk-Kz<S1|0LId!4;d}V1|~6azZRn5>TH6bqBI;IK1AuN9tC`*8-a+6u%wK
zNK`r!etSVVVff&me&H;q%b5$EXunuBtye0tMXLzUA<nGSFerxQo<kmRy!=n`Qne-N
z@LxlHNaEEyGFg$`TKj{c3S89&pTQYD%N3jY8|A%K$D}L#&0&pCSNd|-IVwIn=)A!W
z!L1%DkVt6YfMSVQ6TToY8MQ^8x`6UEyS((lc^`u?O@agu&y^yzC}HG9Ve+PUq^=eK
zMF`e?HXkyYWs{w|x4cY$H~mx8ON@KQ9V9&}cuXRWj87hLI4i8FYcvExoR{clx$pH^
zcmIinQygE4(37Z3>RD6gD_#PYgd?qIQUUrO*ao2YL2_~RIsnPFpHlFlCgk^vNgDol
zGm<L2qw#z$Rm<b4M|a$*yQB5}?&d2i&9)hFs)OiKzYDH@e@hbw7}EtS#tnfdEDMeM
zklq){@lNiw*#CTD<mBq(c4rCcs?EH<Vx7C!pmm1c;Ji$?FkQNk8dc^YeR0)tG9YIh
zmDd>u+?7cYwb#VrryC*Eyk?-GPva3cm#4K(_u^<Sp24!Sd=>#u>4xK2VQt8i&g#r&
zJ~H&)Yisr?+|lE7On%{hx^O@RDWZHk(njrW0OuRYedw+MuvA%+k|#i>{SajRC=!=^
z+DH#_6&PQ26qcsRu97#P;1Pv&g4yaWRRADp!_+7bm^e69fHU3ec3Y}4hQZ$BXa{uH
zZ|N4?Z>P4W#2};j-xCmfDa|8_FAeEax!)bfkG5s46LG11RS>n0%yD4XX9({Qhwuv#
z)UGcG$W*S+FsRj>5+K&)bzDtq)Iv*?5-58-#P8EPj|Wx?UIuvm_5Xqe)JvsmwctK(
z*YRF;h!V~<MNEh~N2L>r>Io7LA1~j$oz4bD#5YrBPCAsN`q`y%bon}ca!%BlK78cZ
z7$gpBf8W@c+tUWTb#)1>7Bw(aD3Ef6tJ_#_)d>#ew^4QnZEs?)JFB}$+}>bejbP6D
zq*#Ak`^pJB`?I!*I(9vP=uQ@*2498IG}HZ^YT^2wwMN!s{cv9qaXh<p*Y~tuL*}HP
zeEi@<LVZs>6y<rKR%gns!jSP|1J#K5wfG_Z#IG|t{Wk-Otm*}`--a-79AIPmau&<t
z9&40Vh$9R!a6DTF-!OQm`LW1^B@K4UK!z$F3ApfD)2cUB&jI@?5Sw@Ke2k4Jx|h$>
zvq+MadYL2FtpCF|bx9z9&?0tArAOuS$`dY#VphP63+NA9ZW&ghvWbdHmR)!1*UG!X
zF1?nUek1p6Y-S`pApHnpfBmfwPe*rN%5E8o)Rw8uAl?@Bvc|5xk~S{GpME0km={oN
zI0^-tM0%k2Q@L&@Gvl+f+@r0%vPqo(djq=vy}=)mipmNFY{am*6WSFR@0y{ZY@ae4
zQJqAMy`f!1d-25xE)0DJ`@Ku_aruq2ON1QhY<AFwi*3rFUdrnrRHVjIp03@}p2?hf
z;g2Hbugt@-RZw*nc^^*>qV@o`aBUSuj|dRy*Sp=U@2{yGI~aPV9M2?^GjN2~@BRQy
zjS2_Hc7bw~s|b2$ZrT@!ve^4SE)j&^J=LkA7DSWpVRH7C<H|Mk$ndjP@uhL40)1S4
zS~PjDDAVJ_SSnxbua1s~6z#ei=RZch{LMGVk+bt$$x}N}X5fRG7>*s1rI{1$^IwL+
z&jKv4qBy9m>KCHM!~*pJ;R7aih9_?44pr7JLsOjz(`oRdpy~=*F2%N%f7T@_Fh#FQ
z4K8#9>>O9%5HD;VKD8~;cu#A=Gp|<If&QU^az^)zhX-`I%sII<-&=}iyWD*vBbBce
zC+4ij-|2@R5bwCl750l5&*18j`oAusl!F`hG6>CM;fA=Wv)^^Hv?&HhoWHkaCeU!F
z1nt(v%Syi(0@!wBa1rN+@*me42MG*!eIgj-p?m_wQ-CCFJ;0|Xn6q3>?AT3Bkj7+p
zAC4}yt<SePRnp1o#!rrKQT1UW&ltK;?rQQGNRK(7sU@4~N}CJoGi7xDR1F%J6fKLP
zAWm%q*~&kLexE>;(K#&c`maGmW>j{Ts#jzi#1<W4eIp9Lb#CQ)Iy?|m7|x%NmTnw1
z1+HwgM6#Gj@U?hEkF-9{#=p#8&@rIt#An<YsMrnZm!7|yvG(?>;z=F$A@uBi4Y0!g
zVumrD{o~MmUjVn(q|9RcL(5pkGY%B<4eTesT$`zInC`tHj@RUMNYsG<zi{B;s(wx}
zICC5RmkB0@|EZ9q<{!Xg!)HC|+RJXzbxw}<E)0T-Wpa!aDnKwqvucmn6B8x!xT>t+
ztVr(v3Z;lO*av|e<?IwF{eILqBw<xFgTu6ThTWyfT_X7dnhF2@!sp`x0cAdOa;*Q{
zNd+#7QMv8*y(XvYZ)b?Mn)x&6y@9gEs{PF;b>#b37Vsm8M-xdU6R@TeKV_3ZU3+%K
zqzv^F3mt+-v92MEqa!TAP&hJ+V>v0_M24NEI)4i2(z~3bC;|1ZcPd{*^jbmf@}fIM
z_}@@J3_yUpRYv}^<cV#XtmfS`rGX}4Z?XVnqocsDndOK9?3dz)8YR=Ce+OV{iy!c&
z$=gp6BU<@>qx5~pqL-1^ihn$R3y}g;l75tyAcZYNV($IlF2QL4916#B8O5qSF>M5y
zy9MG3;CcBTXN1uBlHi##rs-w7VZHYEn-kCd{E1W^BB5}-j&H3-{PD{1rUOh04fWR_
z_CiJvV#h?4m6c6dlbNT{t}8n_guXUlcOA`GN+L-M)FEw)2-D=H!~__LO|*BY$`+WO
z+NWVWA5Y;Qg>(x^3|nW2wUjcf9!_LxrZC|HP{Id?S1=_y`-4n78RAv~J8OMiT8*!y
zu}}51Ob=}j0?p&j=Qr=bt+Cd|t}Lyho860u_Ydny(HyI9^vJj!c4tan*ced;_4CSo
zVvrK(4){xW!#j1_bDGQ9lbESQC$e}kzfE!-?Zwnacom9)p+1Z)eqL4w@8OX?KnX;i
zE&E^bUhFKW&My;PCoN~@<(I!B&>&D;B!c?K4EHi$*Njm0(}0H4smDiC@t~z4jPz55
z;W5!2F5nynbH#E*0crW0g_3<igqwkR)z2D4@PJ1rh!*Jva1681x4*p}UEfk>+x9fS
z={PEW^zw;*2aqY^Rk+hof1Jf$1`q>&@vg;O6l-1bsaYrNGU@$0dat%fdK>R*N=IA%
zMO2Ky?iz1!mxss~O(^z9t339LtPrk{XJH`3&E`A%D$x!7{Z{^ztf_zGh;V#5cBErR
zkyjttw<EsP-Aj3$u)9yS9vr{HwV8`L17^F^5d?DW*>DprfevE2$BOe3K!=X<xC^U%
zqHQJ|T$Ov$Ky|*UCeBwP)M!koN{NV{Zyp$Utgx%yc__``Z1Vh@#FPCmNT73=iwxFf
z^+?Z@R{Uvw!j!RcHx5tS{#uU&mqgRv{yHO-PS;%6bpOZtTv748sa}Krhi#S*dupu4
zfzgYe9jMxf)LaesGRN_!03=X$Bz-n$z%scq=W?5rznXANiE4~-PxF>SlO=f-z#-+%
z$gG9tKY!iJUbRW^XJIOSzH#1ikt)>zAGDIp)u9sJrrdd)WYJLm_zep8UlaBDPyBr+
zhgqQbK8YRQ1aO{>fjdbHpk>*2r@-p)SFq(KSfTK_bOnHKNkxusS*SmoI6l)*MNh>n
zM0?d7`iHeLCZq|3zsTO@PSplCg9&mL62wNu4C7TNdC2_&PXvfh(?%x)0yVb2d2!#M
z$=3i)O+i}9>qw91Pv+r)o{;c0ko7U*5AZux<gr&?g+_eXG8xx%w86hbq^a@j8s{TV
z3so{d0{B%z?U7BV3PIAmYAD!s?n#~=PSxArp}B@s7lR%5Mo*fNI@5zjqd-qyt9<Iq
zqox>jjcb`RaW}-3F?A&1jFL2~{LpH1ckR)koD55pL?!TP1u3%-k1V@RQY&Ok*MF=g
z(~OH!3aQ!$p&iCDz41*<79oPoOY;mu{QU-a@o-4Z$u)4?MTS$n-GfjG`~34Fsf=)*
z(m0u6$8P({mY;*Yks99k2QcbiaaK<o3D6gf=eypd>=ayobpX-g*-S6pEYI^>`RoYq
zy`HGf+a!b+1|_jl(g|)GuB~o${Btc^lOM>Qv^u<jKp1!5_wTkk(Qom62>Im+v~@-=
zPSVmJbE}z8famODcqgGct7>NnrF?+uDbOE3hyQsPB$&qwcf%My(jB-YeOX2eXxOXv
zyn}UXJ+n<(JG?i`t~~XiyATO@N{LY}FfnJYusQyFUdAA9!u+^F_%@ya7D!9^%)1`i
zr@PtKJe12mrLJ~5Tdaa$sRiKBZ?#vUGtwabIk!otWvW1X3i6ZdU6qBi7Ss+)<$&ek
zU=giRsmgJv0L4(m<`UiC&fVO5hj7S5KU&d54e_pv{$s61N*e8Qe(h~nM+<5#Zfv3v
zgrxlZ+DWOI?{Vlg^GkofjCQ*Oy8`fG63coMIH6pPE*yV#^nuj`+|x?=Xc-YqavZ7{
zVOV?#I;)I>@7HIbN0a=_-a9@WBAOSex3U<1nA{rZz++-(y+aE+wB^o7-@tudaQR`b
z4<~fSY?LRApM*c6<(ERzW5%5AH?`djwTf>9-nsb1r`{_DT$}(6X&OPoe6#k5F>li*
z(X>~+_pWU&aOta_-Q7<1fk0Fy<h<&ibb~*=01^y50}io=#g}mSYC4wK$pM5MO>tQf
zYWCu9@}&vzgAv;uwHvs|B9r)d?y^ejGQGd!expE-;_k|^rg|~AyKSu{-gxrk%I&|?
z0P`w6^%|e2Ripc<A!U>y_}AyNRdZNk<$fI4g74zK0FG>d5-Y)cl=caKNR7c~{bK!{
zT0#YEH`g!OLo`AG_J`6Y4&d#6HnNpe-n>MS^r=K0qj_f}BTu8wx~sQWMf4i{qkny*
z>fFNU^MqXWy$nOdLqp7k23c%xH|8^O-+|glq#Iw##CZih=Uu(BDGzl`2uim5gubdM
z3h-P3UiqmW-!94baC4nKVwryOXe?}7m}h-~_1|GD%Z7YhK>RRPNQo*9<qcoF|HPOv
z+ja3l?*!5cx%~{4bTRMO%_~epiqy|6YqT)tC4s9b$oVTYFLy|0TgO}7ACu@8GP>^7
zMuHyo3B*aw$9*%ZuFc0#iTp;5L^l9Dz51hAij8I|F`dg=i7`IV3}MAqcay{wI-qQH
zxVa`uUBMUZNWwK*@3hb~ca^OZW$6fX?jf73vp-YvD-td9j8ee5rm=<o>t@D(5$&)r
ziHM|3O~1gbt*x<IQR{qJE-U11Pf1D=Q&q)nXlUpbu$Zg-Os7#BoAeL#wt;tYddKa{
zmu@?9ZY>vI?p_hi_2K1DnIH0{_Oq$i(XtO;Il5Iaq-h_9mX-%q*ME2);6~`FS?*jo
zYpUe83^Q$>ODkyPLVc2p=zut=BbxiETEal6i4_i9)jagDSE#C}tm_$`f>nAeGGt6?
z^aW{I>2!H`08!c~CM}i<VHu!~O>-<E6je&gmT)zsG)aDZ8{(J1vebk6XW>Q=*U`;B
znwbedY}TDqPtQQ@HnC97X}#uPp(wg{cnYJs5ORV8esi4D_8?w7ty;%lsT5AvbMo8U
zTTUJxObiSm@SBNvoN~09oF)sEkd%-M4z$xi(9WG!dKkE#uCj#yEbmR^;ZQpJ`T5ON
znV>E&V-;(C{;yb()Gd%SUx}BwYhX>PBco{df8j6r{&s~OFI*g1K9u^4fNsI1Uh!^@
z6gAd#L~WcG;j*lw)i`@7_06i|QT;B`R;QTKIF1%ix{}KhR>rmcxT>k$xkHf$VO{Xk
zotL`u`JFN>wmEPafW>3|lGX8PX+znO{QY;(0~<FwQc@p(|2J2MbDhpBp73oQ0Z4dE
ze-8Vc0_s&1U45@d1fFTNwJQyS$(>(zUF-mbA~}mm+$v}$%W|Tv#7z5>BWE&#*!SN>
z@S;ek6~5EM{XlIpWFe2LCXnvv_)tNv`{%ET!jDvdyHy4uFO8abX0jork*$}+n}q9(
z&n)!ezQw#?4%i}`eylEHW}Y9fJ>SXxQs@0IjWT%AF8VUj@O<1J9F6T`y>atL`wDd?
zELHn_Ix36oHHgDwHTW}E$7~mbN^I_^l=AdFG`+*E7$gbj$iHpcNKcQ&tdU5+NhvBS
zeoFz^tvKDU4%1L_dl25gXSHahu{$X`e{*tsu@jn*!fGZzwS@FCs}_G26Iu01jQC7&
zjSiwS|2);8Ef#z82w92U=|s>qs03lfBu?O+cO1U7@$1!1cN0vJ<1A3JN$oGy6m`Q)
z?Gr4STE*X0yKO6!mA0o?fM^v+9$%NC0}}-a7(v{&5e5=Y21kYxc{umu#)hcX%G*<n
zPn4cM5swf{{gGnm2jiWUKi%J@OSt^q#$SOYm~~L}y4=B;wrXX4zBaFNElNDL^$Tl9
zfF`YFvF&+@s^>vdxM)7A%2kF^gjS|hH&3Gz#LZuP4;{3^!brSsvN#(o^z_<+1Rbxw
z)pox#DD*q%GBPLq8yFyCVv5@6jjVEmnirRpoY+o+tMp3gtI+FfN@n?_Nx|%~*fI70
z#O7t3$yWZnnZVB451-|D_EMSerqq#9U4=Ao@{L|!;s2ZL?ADx3LE({KOTgWFj@hSr
zhEINP?ksvu`H}X61*Hb!C!jL+(UuO-$CMQT%_4;TSu}~J<O@OC{Z&_N2(yA%|AP6d
zppEIl`zLnYJUNU@9D6+(?zc_jv%E$Kc{9ZNH-jo65^jlPV8#gOJZLBROXm8PT91xl
z30bxGy|PC}i$T4gV6v7b^;mNnw&WblX3?hjNSia+YEf4d84&&%XzaxMT>}3LvGr&r
z10_I2$epclTz!U3ZHah@^KyZGHvN?+%~^M6&3%n^q_NTm(8XQM0Lxst+2iJPEd>s<
zP?7rZ=%}Kuj*^_*pHw(-@fa<jexoq+%S?D6GJ$e)YMNG`9MQ{!K)d#~Hh`ywa?Vh8
z`YVXbSO4M}v;bfYENJ?KNoYPJ?AVl$-z_5qa1F(xLlQ1lw=hA<9Uc&<9O>MYeV;v@
zZ}wy>%9yxXSA=!Y<saP)r~dvx@n^fzcrIVw?axUSar?SR%0U}g26&Kk8+9<lmqhO#
z9iw0#v=#pyZA-7mlzU)3WT~Q@><^LI$kWT9>RakT0T}bBnpNN%8>k+6OyiDfoB~xW
zIv=)AGvONw<_3Xt2_t_qs_JQMR^m}*JfC|}%Wjmw_BRmfow{2xBAV%5tuOgne5sF%
ztE$-F43&yw5<O;9edAi#pzgnvth~X_Kr(g!{r3Gk>}a7<kvJcf*X6DpEgfA&bu~G=
z^6i2PaCN9gA({A9Ru;|4$*J06Cc*~tSnFa<<*>Q6RiWGdDiDPzBr+0}ot=HMOecfM
zpvQK1l(g!;Z)C*eY`r@Sm&rFX^CP&hc?yN*%4G0+JU?D{Gd(LfIy#>56A5@Vt8Yy1
zInwI1rlOOHpb-*=#l>MsOG{Uqj`{9n0O?9Vk#gKYeQ=@D7Q!-k^A8L*v*lTZO*sFJ
z^n?0eH>)eWZ^9^YY1>&~PnEya$1<Hyt>E+9z_jpw6iv!6*gB;*wP^d%FU<h$?q61?
zVyR$gc_12$^g(L4VJ-aL*;-P_--h@jQrft0mLyi}u;)#?Mf<91PX$V~&oi4pJz#D4
zIj3qj7Z2AlSEewRVe)K8VxmuMSP|n<fk5ahUM@^3?P@rh3q6|OaND|y*GDq;nn|U`
zR_?D<1Qo1;cywCYRoNg#Q7;iHn1DV}wfwCZ9rx_08Pn0x_7Ah4W};{9*oo3*9Xq?_
zpXHoW^85{+Nrr6Ev$Teqno{XD(9&lplsAzzvB!uXKA6sz0*8!$3Z2ox#KQUxF2lm3
z5b7@0+uhvV+3!ylFyCu}*pir-nD=%whE}6C3>*=Nh=`od$Al~<!#PqZ?C%LVe?L5E
z#Od8F^vBWIIXmx+iIQwZ3Os$JrVjb@=hOPeh8o0+udMZs`swM3<FhRA+#oZBXX4?(
zTVp=8zV?e`8@zb+d~4tX7FJYTTz7>*Zz7*N`)hcV9Osi|BVragJ`he2prN&eV9~(1
zJl*byNlTaR|ECy&O`5PpD>k4t{WLLy`OgfYI#w8J;9v2jsdLF{r8G;eLp;V?DkJja
zohaM3LGm($8x|A*yG!59qC2zTSumIVT34|wJmXPJP??|i@U`CnlU#WK{u99~=wUFQ
zS5hbBgtq23OQ#M4pvK4N{vxJPl5}4tRFKCr8u~D($u6TNZ|)#!gLCI~>3`nc(YrGw
zbDr#dtD>jIJ1t%N{^0>H*T2g%r_Y<6^>6|)8gXr@>bE>#xE#Xec5_PjZCk!b>C2Jm
zb>Z^Y0JL@DJmn&@<z_Aza5#~2b2p0Lr7;=6fn(PrxXi@JXzb5-Brui|mMqT4nsD!N
zziMl17c8jPm}Bv|Uv-(XE}ug|T)|>Gn#p`$kl){9#+>A`>I47d$B+E@n{QM3w!R$v
z{H<Nj^FXy(=Vo6_-@kvE$|dSA)sxEQ5LjPdFF!+b)Pw$C8EAYK?Xh!v)UzG$xbWGg
z1ijNjT@NeSqFEMqz5uS=$X5+o_*YinM>vQBU%mFQMKRW-@~k#+Y>j@nqrxwVdn^2W
z;>vq)w^ZvQf$98!xKsPh`&TY=M`JVX6eCc|4#SgICZgkVu#q{ypnr{e5NiS6{UWC6
znUxz&la{lEKDDcgDVN=fznUzq_fld1w$%aKSd@y*Rl=SMb(!xqM60`_Ct9@C9bqOj
znGJT?DodWpw0~mK91Ah3fpPmyv5b7b@?r?WY~0^id9LVSkYh%A2*8DRr;%7Ofcr<z
zO<>B^yKu3x#@4NPrI3k6C{mN&b>rw{j+x%<<z@W)_pi;ARbEx~hl`7g9p1Y%>0Xu*
zUWWmwfq{YR)tm{et&@}Sa<gk!cQ;lf%J9g@&gojG<H<6Sx99y;h0ED`%3XuUEpu7(
zx%e66*LysnZ{Oa42oX12<kF1~C@3hfTxyK#C3Y`fyjR1rqaY`@y}LAnNoF+@a&h6D
zo14Q8?^)+B)okb}gFJ~({FNFY%Ff5f^6Ev$#Z}!#v-S4%eWmRQRjRsYOA>dPApalJ
z<8EOm1P%GaY{_{g^e6VwxwsCZ0?oyr;&@uBuN^(IP^G=Bi`|@wfkrHZa&`oDPAam9
zCE8ejRe5y6VQ+;w-B%!$Cr*39XqECkcD(79E@@GAx6Y4bM%VuCKQE|Za*kqXYxK+|
z)WqIe&<9SSzL7yC3I(Lfl3AzXDaJ1NZ4n1^D^6{wgv<|Tz>1L`_P98$8trl2MK$Ri
zU0U2c#2P!3MNl6g^*3otB7_0C1*1pLfXj3d9>NVzdq+=5)P=UmQO;XJt!9G1TJrlM
zbq+qdJX=U9mM!Pq-A|4`zPyPQn+kk5Q7b0w|I!4G*ACj(aPM^=Zq7U|hB$T|x&KNg
zb;MFBz?`0)t+m6GY=S>IqLIX4FH2rJ8Os*6$=|au1I)w*;_0|NZWv3{Yu0bhHY{h$
zeG(IKU%j$^(BQFK{|2L}safy1hcP_NAzf*1WhDhd`V<bEzu>H~T&#~m6?pi)*7<8?
z^!TW5g<d<I-3q^~<!T<e%z_nPQ%h@axy9WMqn5Yu(=ZSppjx5-w!OWb0}3A94_)1l
zGW@rIVS~X3rR;tQ))#~z6h!d-n74q|(Y8QSlt`-#s^TP7zffTJ%--C%#iA}SM}!m}
zAYZEI^uYtYV3IKn=QU+jg@Z%RPKvOrYbPZebpz5fuV`u9L4GIH!Lc&zNby{KH>a4S
zVPksXwr;zV`>cPv&mZ|DOgY!5#f*_goC+vAgM{`Qnw!|$o;v=$cJFjG1AEX<`)ls-
z+<4dwzHSzvRqE&pN~%@n5q$P&l;{E5+VY+T<6{=u%3ncVeiOpT$J5Fe7Vsy;^AS2j
z$nTM;-C|AaiAh2-@^E{B$7N6M<>lqLHy!}i5%azKc9MyBYWYw{hi``0UBL~oLcn3&
zy!I0;D)4I9oAUt%8}a0)7<F)=2$dgZb+N&LgIod$2}!Nm6dh(Lg~J3qV51%_VkIn-
zuU}#C!<oOXf$&ucKRcfri@3eXWF);XO!z-=njUAz_I9!`ByzTye)j6*$s89X19u)b
zoT#mj?^IAy$!<Mp=YYN>(`x)fzO*w7Wdg|G8VkvmC4xlBex{l+Z|5xvTdD&>tY_Yw
zXiS^iIq!f*%^Bb9I;2MdqHbYKQCrt%ylAqGKC%991_j;i6L`<q?KiB5#YsE6C`j*+
z^fn)6`lR~S$I~asb;Le0(7}U<chSOCzDZa()r01C%NsmzKpy-zn+m;7ID310PB069
zp%CfMlxkVdRpxS%HdxHigJZX|@#lfdZq_gHLyO6LIV0<XMzBYHNZ^9`T;S0neI&uq
z(luLQV3V)m?s<EMxew=;byE`?`{D2~_8P{{&aPVkII5YERkCZjFASOJ?+?)d9^-mR
zfEPq`3QVc!oNdO)jF)Qh2nh?@IXD!J*MIL;<ZNF6pZb5HTsCBmN=Sc=GJig^{K!Fx
z1am%KNkgK-Hm0tIHlNFsrDaVoBdf571l>IkrOsG<YfJw~xRp!u8O^rX&`KF=H7EDu
zXVmQWAa#Q$KUa{~?mJqDr&-HWmNf&1i$rG3`9!yGZxh4%;q^5Z#J%juz|8?o5LNHZ
z*U*5}c`KY_xy_In9S^T>Mh6%gOk#oEU4aPn_4U=)FWv<wfnzVO5xGYFMRtgaiqdR$
zISZV5;sRo&^H_>J_I`f<4#><WB_^siYeKv{tL_<9l)4cZ^}F6BbJzgGlLZP%0sucK
zS9F(w`o!71MH!RZrD=0>Gc`$lK3c&YgMWU7BpjI?C8#`$kuvZRSWxn$dT5<ls1t<e
zKHN2yB}$TO1(8ARKQ&OPf!g8vYHrxO>PiR?i67nI@Q!)cXHY{J+rrV?ISz6U;vgpF
zSK`)a5zM1xUq@5}&I&e7HuX33BdZ`hV6W|uLu({4>dLY8nDUZmp!`GgrDJX-?KcYY
zyS-Vo$t_MNknN+?ZtiSTVKOOgUv7ViF;L@)7sxVXPX57(DeUHm0pfPmIgO3oBVU~=
zU0O?88uZDun1poU#WNa^Ve3v3fUL2rs+6SEXDj7{Bw&&I9ZcTe249-&)JmmvjgK4;
z927l{?U7ft`>eKhE2^vF-PORwRZpp0dAU4kBIChCpNtF=OiJl4UB+_Lu}|7AYXKND
zOnECBw>v*(`fqj!6_xnha0=;%5Aaz{+d#-N2ttaW;NZ2(y@_HSNP2_A4l~$TDWp^X
zF+hNa4h0!LrQ8Lq?Sk|7eK}$=Aq;=OzL+yVjV`p<RJlGe%jQ6rH$y}rK(_ODhR&)a
zf1!#?OZ)hjd(a3`knAFTz})R&N>gyqmU~~ARLkDa8f9UyU+vupk@(I}H@!xWU)l0B
znd7nGOSn#JH)UMDY^=&47z%XCSz<v|qy}8m+_<8&Sf*{QFXTW8$=YpczAqlGMPuAZ
zQXutVv!YO6gp9|e*ywpA<;%ZDSL5wR2lhV|qo>b@OU{}~(C4<ZnXSs}c&bnN`S0rM
z^6RVQCFmx`YmGXgL;LqfGwfBxtwwV<#%c{F>c=fuIEb(QeEH~ZbNujV_kj)1nH0#+
z*x6|Be0}Y}zvIvoaB$x|azF9M{av;I0H>h>pVs*E+%9+B1J6*=H?n4JL!tmCN37pb
zDP5<-=Y+$-dP7d!A;iZ0(XGcSqkt?~Q)?05tfuRPb0{jKF2;?yfIIZu_?WI`m$Q^X
zixQ=6vH#B#zY%k4vsND@8u?b-b_&o-o1b)uNG*Nl&3*MMtX?oK+glu2y8REW_s4g>
zA2m5#&o}pjyHN-^Wi7lpK7S5qYim=#A2Vgutha?-trhF$<+EA*vE1a$@OTGhF)ijW
zX#x>U;)9s0Lik0-rac#(dUfXaQ0y0?#sGPUev5?7=ze8MA(Ji%@p|0n1!{#l8C|`v
zHzEZ}rSGTNR4x8-`+i*Xt5!#Ov>|AMs79)12^mxL@)*L<d+C8&^x0n+au~@X!fuvQ
z7hGazArWJq=O5u759Fd+98p77oI=8Fp2fef)x3HxIsDi~tVx$8kBp~eDvfPt2hieN
z_hK;Dlc9Pt&t(qw@MQhT2f~n_)g{+US;kQhhq9`WTr_;*2+y%>TVp$Nc(0zWWg?zy
zL*aL%|4J#=YcGryvK^|IM+iSJ1S*%nI^a3}Kce0OsH^V#9tP<~P&!luq`Ol<T1rY9
zX^@caD<Ghf(%ndRcS(15cX!u&c)q{?yfZN4F#7Rw&$;{Twf0);ev5i0g}I@S@4aSA
zYw(+=!d8f77jxKp#^=oC+dmIoCpKytA6>F!r`0$+c7Hj!B_oU|R+=MxO>f?sgaD*3
z*Y(h*EaCvk$HPtk$yq18Lr23Td0z?y=8m7lpCnLWkatGqvNp0*N%eO9dle(pq+;9l
zNm42jmBmwJG1TZ366vWtZ)sMllHaWSSqOB#MCmICLbl;d3CkCmsDFT?86JVEZ7M?t
z6xryuH3mNke?_T@W}SZMrVW0m4QD}1%5;IERnIR-r}7G3ED319w4L|DG`3gRjg>m}
zZ)O`ERwn4!DVR@A{#@Ft*oVI`w{Rx+2HV0WL@`Hsymx)QYB9g*;0fKavX)Lv8hrVZ
z8_nmq=iF-nj{Nan&{(l?+wAPR8L*B5#P^#z$+<Nnw*D`XiAP^wUkt!AWhT%dq8`{e
zvdMxC?<*<C<oH+VkwwB&R4mjVJH8B9aeL3r{l;tqldG+-`Fqxb^vbG{I87F>S)jr7
zuEE)Ez+{wsU^37*@Sgc45^i(k&#;%!Bz{BHLz;9J{Veq{W27~S|B5p5YCo&1CUoGR
z?q9B&z_HeHEzjgT-xvJm#t_1%yVoNaYaH(1Wr}`KnnfJ;7`2c2qI<3?x(c$bT{v<m
zsL+1CKB;=U&xcfLOR$NR-jMMt{UJvgA&*U=s9Wz6e-~VZL7%q?3tLg@zQ<DQLRr4^
zGBHiw%%lddgojq<ui*!p%x9h-Khy~&C!S8~@Hq~SPeXP3i9C0ss*rjz3YLphi=@Xo
zjh-5#EgDb3%_RS>9Y;Y_0eR8d7875FhjTHZ>=8m0ku=LY_F@hsPnp%TI3GjmzsLUu
ziwsu37CjOWE2z5DLj3nHGkHFYU<72vi(*B`$>?a&`nTu2An!Nn{x|4sdSKWpewknP
zSlPyF_hni~Jswt6z#J01Jnzo-hPws_*ErD+tb<8dI}0>rx059rp(A!M@RX(t!Zqc~
z|0EEE@E#j}`QaD<dKD(8gSnasN1=x+3O3`u?2?ND>C6?71XJ+f*%hiO6ric8$!0aj
zty-wno%SIhhFSj^fI{PL`Cw`vNXi$blBfEdl(heH5gLR7m>RR4geq9Zx#g4irM`V*
z0Vj9`Vv059UlU_vB7Wl*n~!hS_gyaveLsx$;SHKNzWsk<g6Ht5%llVf8BDlHS_axf
zkQVqQqa_4$SgO(V^o`{Gm#vS6_^7{Kn8}}{o_@WW{4%ug5L0M#$#c7@NK<WLg!pKN
zVGR#WlFT8{NOE?O;yHXT99YvH_;MYvC^`N6<BzobFrz_;0l}%RuK7}*d8lDyiBU4-
za1h0>O4BPda6dW!%|+4bxwT}T(dulIbc?nS@zr0S7eje8)o4nH$A-`q?Tfh2&#QRi
zEbY{Y8q=LhyvImX>#G$|4=(I{4qG!I3$!DMqv}s#|MYu3#?`qFgT3pWLEfrmVlqjE
z#f#9-2)#u+LXr%{Td2xrTfE=&8ZuvFKGtia9`MGrQCluYHpltY!9*D=!8j`tlvhyS
z@4ZxQI1wO<G~2BtOsB7kF&gv$myhJu+=@9=7PZ7EETegqf>Dq26On;<;m&uh?brdb
zN;ewBzT-n`i$qm7X@+FWo2!5jICGOt)SRK@3t*%#sndmXybcQ`cegDZ5VZ<2K7?RD
zv+t6_gMW|xqwe$N!Tj}J+2B+YBt%F^s0>zkvS{2K^q&zncURxUgGstUeNb;$$W~0>
z+2i8pPcZ6@QdFf9XvkE}<8ixy=mBA|_U{)i3?K=gV`JOQ)&g8$c5Vw!hZsQ*UKA7*
zWtI6W=%+@(-{Rs>Kufb(FYpUK-kr|g#PT^=ou7dI&1K~mH8=F&?vNmxH?E|m)z{*K
z<N|<3Zf7ptr@*r6>KIr!#GwicHBQSJF)J&p^W&o2+&5HIe(!xCM3pKmtjR+Zu@-hB
zf#xS?c*9(7<kfdt<jZs!q*ZS4i~<WE0jCAE6>P<#SwpOkgpA+dLQsPy%r{Lr8{75=
zM6#Trq2D8?(^nHkBX123%O~!Sjoj><i!grhguP7e2zJRMc1&@dx;d7%Wz3KX`O`G3
z(UU*^AtV09S|rqxZ2cf2rK+FUZ!kf=yq7G|GLIu@f#k_M;bLiqmRt!ZVHeCW?T`+r
zBSY{QXXqKr<8Q}k?p__679-u!5q$SR<N6bwpyG7OV~{z5Br~FSxX+Pkk%{0LJ@L(S
zLv}G*d~BNk)SZ@3P#Xi!_~;w5823&ksTvv$0f}kvIz(|ya^+ceVDMFsc=0B3rSBQR
z5@wlqMoKv&*;m9O#57QePj9f4BUQ&1RH+QMW>^1cb}e_l$WitwEN_-qTQY@Zd>}h>
zH?~RacWYwSavfRH6TAz{)eXJW6Qq}WZ)G+RTGbJ2uOtqt$9vP?!K8d0d(&mxlSSl$
zf`U_1(3GCBv7aneTVEf#goFgZ9323js5A_W$_st0bF7Fij{~~^)z=#ZmrBptIt2X9
zxw(YmV&=5;bl9AUy_NO=3oW5nufFSgUT5CiKD8L3q1a$-@u%Bx{Jy5TdZ<8?Yojk#
z)YbLI@*W-mVID_8yVjm&f2N{sWTey;dQ3=2*c#2k5_-HVYxEy5T!rWUwgQhOc=ZbK
zoa3E~Gc$2VM>fG^0;u18d_)_(98&cOUM_u<mR<ofu`2g@ei>a6kygth-A8Fj$)#p5
zxT?Ow%fWJ|_t*y>8bMK{$<iS!mkDR2!TlP$6^#`>NPh;I;PeKeN3qYT%+~~vFQ3Vv
z8nJ>3jeb6ty~BjHksaY(-OH&So{}%**U@Ff#-XlV4KXd-<474WXYmE*gEIvmvG562
zgGxnO7K$teu~q1l_a1B;&iaX4Zsf7oEfLr<x+xqzD^+vn-GF3f8TrHZ_0pHuq<kI*
zM)&i2Kd4PB@)I8n5<d!P-m>mMqgk2Og0(PBnNpt<9B4kZc?TS65^YT2@mHE^G-TpB
z4(L*R<i7|MDjzQE^A6JLk&Ei(K5N74Ih8rnY=YI|Ns?SSAjqlj4ve}<PG3%9ZcaS8
z!@+#Cc1!%|!CS7k<}dFpj&}IAwZ}e(L`3_y%F!``dZbroB1edaP{9HUgi}CjSrmy^
z`ol<VdGu}0S7~i!jhy}4Un#!+8*&|a1<?F1=tGfPJxm7@c<zqcupBC5!HxjSZB$Ui
z$IWfC@(<<i-r`a7cI=ZpV2d#UUxQ9Y*BnU5re68=vyZ^I)gJfl#eOe1h>9x;+zN5!
z<>loPc~z|Wz)Gq>K2-o=j+?D=<m5z2Y3qQGHSnWAZq<Ibe@4l=s$_E>@~U{(+2`g6
zeMjmvI9-ybBIyTd7P3ci-7g86ITRJjmkIMW@Ox`VF(LYi+?vhl3#QQ--@dd9A~qdo
zn<3?7;r3ogdsUV!EEz5(&!IFH^Te}RV6m<Be^Z~CPl|=&tNFxHl^h-XY)H#E9&@tj
zeoTeqCJ(F|{DQw2JG8!<H(zZ1IMTY>SZkv8IQmv-DoIpa3-$MjG`*$8V3HMAJz9t-
zgTejYW9kh9M&)|*J0_Z#Ljwi`rg4pGEleadJyEKN?BYus!@1VzBg}^T=PmIsGuCXr
zLHWv)=dYQdwu{a|B0smZsfZ7KHxy^iwFh6v{2cEt=SNz_gmUZyUjMynTi{QNK6q_A
zBWdU$+eaw)7oRmY5VYC9Z?axo@oDl(%xgFIO7lw47AS9C3Jc!FqO6+(BHrceaIbUu
z$o{2T88(!3<H1SymodZ8gpq{dhku+@MKYXaCb0;l@hV*@PQg-?%1~L~%BV#&S(eJ)
zkgF}&_X&Q?8A;2?@PP3`X9)QdtC}DLC9SO){tgIu8X&!U_Y)KX@j!x>u0Ist91DdM
z#i+uc@cR=}0EoF}Fv{*87@&Id#`}rx0P`L)B>!5OFJ8*6ui4nb6ZxD<-Jxq+Oj2Gk
zTU%Ceashr9n}|r-(D3d3?Irb_H)#`|WNv#bWX?;lszo~e;Ny7gx3Ccq5CA`p{rq_g
zxFSL}V{|aMawX;a6!?iT`m_~|>F8-A0!BwQU%e$s!0jC^E!cW^XIsNQ$E%%e?pHR7
zt`*NYeo0M;h=^Q4fM7<(#I)K=1#PvzHblZ-z4_LVwT+w>`G7|X_2__~;#s{@PKL*3
z22r-aTxOt}=m(ePFm@cLB&#zR>y<)_vP1D2;)P#6JoXOa4O}4sNGqm%XBQ}uxxKUm
zWu%16U2rQhQgbdhjmSsl%0xMwrRdwkkZ8?k5eaH;)h?)nvTp12%RJ?a8I0XPiBPUt
zq0f}#vJVfx&nTm%xVWX2xyYuYyOGI91H9NIR+m0MLaMn%zLH#RRJL`DJ*1pQe~M4-
z(-ED=rC;J-CWc>RCI?e$ny(lk?32C|(v+QvHN4|P7@y|<YA??}dsTWLuA;QRV9g|>
z%aK44I~NEAhnyQU@1_^q<vj{{bDg4CwAk%F_7@`T@|0b?n#okt-&Wa<vuBQGtsduM
z{i%6j)w#rDV1~D%=9cw8C{j{ZmhN_W00{R)v(dk?<%d6iA~)V&Xg+=W+!mLY8-hVC
zI=SFnb_q8(H?vXC3qTB~o$|uM!j6OVo?U}I?t+2>LLpE7)lTB57)o~f^k?;1^2v?+
zBB*F+qUnFgxcKch&~07%r<qw}(LrlFH#cWrOZNu3wSk6>69J|u6juCvm6TOshl#O$
ziv=Yp{bue1H{s-;s+a6#Z<u({jC@60;cK6RGgKioj>5>ub%IqYp^%riWr3d+T2(Sf
zKZ@I%(#{)?a}gY`_B_D(e4L}?=TMxuOgJ(5K@xT$VijN5zdJ-g<C=LHD*DDj6?;`c
zU1QH23{Gp>-yj&rrek_AXsHXcoOpq-N_d{_8(VG>0vuaSk%a}8s&UW*Uj;Kmbnb{W
zvO~5<tFZP5?$+rqptkK`96f`aD=Y>wbWK8CcW{np<-**z_}gO<l%tgOHn#%<nve=^
zoKohG7^~SYnVr=y$mIChFKKS-nG>q}hXEpzc`waqbWN$wH@&*GF4C)Gi9$Go()2S5
z(OLqW_Ta19iBLOj8Snru4&S-x>HA_15Z+f~MXX8UN=(-fL<bo{X1jqkk))K<AN~?V
z4I$?aCi<g1D?Nw`Rrz2h_eaqS-yp^vvN||Tp6#KHU`5jobbq@uqb_525OS#~H&TJ}
zcJ8B1Fgm|uqrTUb9|8`??eo=6m<XNZz`?2A0a?UCJY^$g&YhsJoo)nyz<Q}?eu)&Z
z=#O&m$6@s&HGI>gA+MkU4AzDK*vbA1UuuuP!6}1dwQS8#Smz1uB+KfgErUts-vcG=
zA8s*y%4Lv6#YnWxWvwSo%nG}z(1Mj1K2d>U_-HSY9%N1$(!C(@cYDK1@&2|mV=}?a
z$J5o~rw8lbfZan1(GZ#QhnO$S9Zuh<%{Vj+jWHkrLdYZdzZ-@9L)_#9rl_i_DyQxR
z{4$yG;ei;QS;TtQmwi3wznN)i*pX=@T#Lq3B#AcEJkDBlOre@a?j|+AEj%W4%g$<Z
zM^INdlV5$?G-LhpDT7`p7TwKBqFxlv6!)cE-LkfM!dgDn#rSRdklR&NjiYSt?Hv-5
zJax}0B$FbHQYdbCyzy>lZlIw<U~9%CiUe27j%Tu{rbEE-B8f}e$xFOXFMn915r+<m
zxS`ak7ft%t{Vw;fx`EJ_$>N_qQH*WbfL7-yz*H(e+eXCCzI9HzYIm$<G~zJbVpGyz
zqa)}wHBI7FQ(>fOb(rV}MZze@32v7<oGH2=Q0VJK)8T@OVXvvuxi16v4hd-BV?ACc
zmezk+!`m3)*&z_?N<p4M4;Ho>e*1Ulf7=S%{*_=2iuX`<Js-Tyu`oU1ah(lXbwg!#
z;mF`&#EpgSCHkqz=sKzE{#+uhUuu${Q-lu6v`Tf=S?gTw!>cgW^!2O5x~a>4DyRod
z+PZ$}-Tg260s0i~vhUz<pJss={)~!`tp9>NH=g)BJ-bA-MDuYaQU#$u(D~n;J&VHa
z`To4%_xvk%Zf$tDpcwi+WyQqVnn`xFV+ms_(ReM)57IWQZ@4`*cpdxes#u1%&{uJZ
zg!^jSMx1Mjo_7V(A^0L5%GGM4nM0YsMTM-^m#{PyB2=V>B4Fk4vQ;#G7j7upbA$kP
zYjth4r<pK)`$*x^)h80P+Uh)03NLk0Uz)G(9O=|GHq3Bw)rTM|gsF4|%d6Qru9gtT
z_G~EqLjMr!oseN~8#cmIX^7unSN;XG?$&pWV?Mb`-4BNkD_tx5@8}8<puVC+ib!c0
zeGNumvl>9Dy}*aC(jLx{_x&6==Yvt7YV0dy3~_j8g!iV(1xS8&OLa|By!AgWM2%mC
z#}baokF@A!KBTpN<l$e&MBSMsW3%Ilqd*uzKgUFT_%)LRCpoF=kt<E7&UFgCJ>aE4
z;6<!%d*y^|2cKs#`6UuU9C*=oLDF-%QF486^p-r>Wb7_$w|~P@fH{F8psFoc%ei&3
z_?||oWmse_YNLG5&Tap8cb2DN<JH$*7XIcu1V=4C$II^9uD&_+D!l%#h6ofsJN-Z}
z`Pb{T8B3|M;*->GKJrNlRFDx2L}5j>9q>E%kkez_Cc3VAC_lPGs75Z7`EeHvG{aKh
zU|LQZahO_okE;AEB5&F7`Pc5hTk3+l#`ht;61W7%2e@4f7En=qY%fYMI;G?2x2bUI
zbIYG+(mxs62l`<bi-p^6Okr}UnMzOAt!a80qO6z$PrMfxvGwTg<wo9U>q?;NX;aY*
z<6pVVX^DGYy<IkdT(|2J0bQpmd+&Evrq(=%PWF}Rz7egY>07Y0$_UR5U%YoXPrj$o
z^Z6u_E&!zrsJM=Jah|X@RmP_*InxiHg_g;TPGAlJbq=!PTni||yKQW3-+<-_G&VyG
zZcYh_@f)jl-mhY!be&%bl4^u{UeHGe-@R8Dlybd<{y69h<*>xtoY}rMROmz`$h4sb
zLRM_h<YlQI_-CpuBfynPo<<gP;(~=uq#h$R{Y0li4?PeUlE}l8TU<HOAy?+<>lqLA
zf4(8wg(IiJX}W*WtE<)Zuf52?SjSCVf3uT<r`C{qMrHfxJ2f|P=<6E^BIVp|?o?o_
z<bX_ItB8hRO?0_X<u{~Y<Kz${P^OZRei?yhi6X_q;(oG8<{Y}$S8cj!u<yq>hH--v
zB)X47rDXsWUwvjrwKYAJ`?h8@7b3vF_}A=5PhcTD#GCo7azoF^`s3ABk^`qogv!Zm
zcyf56vx@6u{W$8KIX7aZ?ba%-M<Syi6vAX<J`Ve0U4{Ro-Oj=I#4-K_C4w^~AMPDC
zfe23w$srU(6^0`&{Hd+1?c#RwT}-+mQOQ(tHP<))Zl7R9B^pO^&|=KoiaTMP_CVu{
zVNH)XMAN1DS$UvHzArk7%G3hOy7E~Ok!SQ&yr>Y@_OANtKADxNFzt6eFZs`oq3|^E
zqW_*T5nN-iia^z(KpQE3dAgMza<cgNpf+~)@`qcZtoYcT#MOTHAkSTAZ#9B_$8^>P
zeS4hp3!~Ca|4vvMh4fyEYmw3M$8mq|s^fP(6@*H7mh$-XYO3Z-_hQ)!qwC%8wC;8b
zx9gu8Fq^7OhCb^o%?=w2i0|qzT;K22lRwUm?zMiet%LF^YBcrOc!PF(e;K(gx1k+L
z%@R-B)^dkp?Xs1WSEMbL|Eb^%*NkDa*)W|LxIz*97}l{>=r*rhsM;Fc4I|d?ngPN{
z4t_Yc)h+Ec`L&K|#l*vtm;WhMaHtSKs@!-L-c4}2Zim68ARyLH1c{D3*qP#IiMY9$
zZl>J!tebCpWDJjNwm+ZZo}HY&8{}z{Y?-%_xOyd4wPf*wf_@LCD47EAZqELQL%Y+d
z6`YrcFT}c*g8x|yA)JOUEf$=|3x6}f+v{yxX!}F<330dNOIyx6S_44pD%o602|2eG
zJ$QumW_<Usv$$#!gE;TqP?CV#o^di|ywtHH1l*iDw)#;bdV6k(`Gn<Ng&4A40KSj5
zv4x{i?$4(`{9k8)$$EXN+TeEi7s!&pbbCE2SyMw}q{Kji*KzM>#d$$2DZm#1dOSl5
zAohwK&r`$2#uj#Wug~#+hb)4NJH_@x@RLMJK18qIyQ#RQ#hW(j+Wz_1*QCYd6Ysnq
zf>Vje0%U40<EZ?N2l)A|f1jzJNZ{)^6<X0nWiKtFV5@mMnZhWRbpURM@0hmJbC$s<
zzhP0l$rd|!a-t;M?rf5DWcA*G;ABHR&iCrQEK7a(D*?oUfW*mX!Ohv%gJXQhpSQZK
z#323=&7LnwFOk$}?Puk_5cUQ8D7_R&CKRzT%(mK{7frgcjABn6J8s>Hhv&WKBgWHu
zrq>*(c)p-q{R#B*tY-zsDI?9ZDA?NBv%hOlw>2blUS-R>xlTN{N01aGk)MmgT{1&>
zG%L_GF7VGh6J(7pxom=@!-*#J*Zjk6E+7Y=`)lS)Z)M(0K@B}iM0YzvP-j9OoXiV8
zA34o<9X|Z!E)(vF8mFvwD?4C3`f4{O&7moz@h#oT8Ux2M>9yG+&rwAbtCIuDfzGA@
z#v}E|oyy%)2W7m-T^7mzTPrzMm+wx1sN9>Y!F6)VLDV~3X!sQnkX~9E28C10+n%gR
z_4M{iNlK<bj}I_Fx@~uLq@SCc`xC@zVPeunN&|(}cM*N>*l=(Njv&?x{th;{RZxuR
zUOc_|W`Dh_{9>Gh`%?tIt?TOT-L<F5Z&qC7VD$CWkYt>YkOhhvjKkKUs!kszIs|T^
z`d0$1I=GpLoF;9jEH=3ia)_#6;NyW2{nwoVo%zRkT_puW@1hq8Bhk?>11~@14VOO7
zHtl>h*EeW4SZwlklVPG6x^PGT+8@=S6HfQ?=>3~M8Iivetapi>D1>RE+s|VIkrt9A
zK_iy4uvK2BwEaCy<5`cd^wlw{Y>2NG0x8)+OUiV}SW}`~Xm}AZgzM?5LR<AM8%F2n
zpJ#iy8qZAddzXYzj@Bs^EPgp>;Ctwhd^17P4zUVH9J=*gR`?s{U5~~n6@EXSp#*2a
znsMK#;HTU9itPI2+I6?^;MO?3{HPo{&-merQtI=kUqSU9_sQu6Z~ROx@6lCA(aP5P
zoADJ77Q~BnuUbF53<v>7db$4;P~5h5Y~q8E--fpLIk^E`_B-0uDeJn7n=U5AE&RC#
zt8)CS*-#u0uij1^3rkgx;H09|J0Hdwo@Q(;+JIi3i(5uGMB1lb1+WGvtOdX502OTR
zdbai6-ku#$%!ZE-_g(88P|BDoJ7AhiOa{q7&MWsfXa#;_W8+<<G_U=Zu!zVr7>mh5
zBY@TalKgEc*I;;3ALu^b;IDR!Kd4)Mx`d>JI4-c`&>1vnd+3VGOaEyxZjgRe)?ffn
zVEe$eg`z60FjZTbk-XM@TXE%ON^br4*1%Ik%7YK-s`^yXqq{Y!=+5{0hEy>y;;G6h
zO43jZGaX%?2IPd|Bo2r)&c^dR_G@fmG)*&+SAe5^6{$!weo`N0IsM>O7<gxcqk6&V
zaFLg*#M-oLA3;Q!zw<?1Kt8ncR+3$nbE1#8R=5YIKvYEv0g$;WJ<jmOe?%oDj!L6C
zLW_Q!QjW(~U62LyvpfA>01@#)Q4PA*&{}g^({T3Kh{1d+Cbk}h5iNe+s~VCoGVk-+
z=X$&!iiBi~O#J39xui~t3`?BvvLe>C>o)e*hK5n%K1;P!yCeLI64s!E0T2MOpHL>G
z+lT1em_xv4D29QYKE3p6=3|(u<4AOwuX6D6PQ~RPB{){XMDpFvVn|4J^PkBp=wtqc
z%V6x%7tCH>V!RA^nl1kA7mWP^&PJUQ=Q=({HSPKF*SOJ`iP4^V){G4y;6I(taY5x)
zdo#ZVqje`haDUBq<$%s=^Sw$bN9A4ai>axZS8O&}gI3(FwVxk-QpuP{e}1#<@qVNl
zbr=!$YdXEMWWv|!Uv0YIl#HzofGE&~B-Ch^YS^4B&Nr`x7FpcuFJpd#Hw6%FHz`(D
zRtE7CrIx5DM;KV?d->M9d+EsJ#&k8!tINd1lvS_gIs^1vuylEUkH?Uk>s?N{9rp+<
zCi2am=$s*8S+ahOGGmP4;bE93Wu*h6B$#`u10mDv*RTKm`&aCKZ3_Vz#7|#eghkI=
z21P|h7@(@n@BXjZ7G@=!PYO%};G5*mldS}j2=_R~p2o+TAL#=Mh0yo&+iVW-g|LVN
z*^KNi1xP=Rdj>*yE_USx^7+JcZ1J=j&!6kt(V=;#iEQ7u+9bkpZ<{cs=VyoQC=wBY
zk(V&>f#mqFHxTQ9c8=4nDkdlY&3QluU~>T!skUdv98dM0ovEohd*d$4R4YKJVohcH
z(a@Onrs)FvdC1uc*fl&&=Wkz{{9cwlX4rpH@c=5)Lx&((l#Qi?<~zv2F($iCpb)a&
zY`%=A8D#pi^Xk#3CKBPj$L#4*J&iga=YF%?R5p}UPKG^@np|n?xT^S@ajnI-RC7js
zMq}9AlmL9%>qy+;mdJDzB|LX0{5ClMR|v*%mNb$Bsmubpv~3DVd`=Qn53Gq9_a!9P
zi=2$)ABD>vFYMOFCz69#Z$`}A7hSCz=gh`iwhvs#2#Tbz@UN^{yOCDS-FdEW99;=q
zf}yzC@6j(~H$TxZy=$MyJgDvaAL40EG&zy4hwhFg&TWBewqf7L$C{t<67Y!3Poh9D
zr%=5=1*+s2n;+h<*c|{ff_{H#fbL?!{~5*X#ou)$MTOrL;iSOM<y?i8eL1YZyr5@h
z@d1c5w!pFhV!yjCLCQ>y1&ZQSo#L`<r7rhfct5lfXwmYjsg}*dVZ|Db`gev#Ty5b9
zPfTZHvR`T1?=w5Kcm1>?pFjI1)=^k_=}^^~E$?j(bYUeM{&T|x>X+1aY{@;XtX~uM
zzg^g+fByOC!4VrL=*lk;Lne>)m%Z*MaMXsm@!+cX;s8skw`*m(3pA-3Bp}uR7#vp6
z&mj&Fe2`N@M9ATZSdRn^bdV?;zPkwAPXE+>)|*d_A$R$&YZ6uhy<OqzVY&?AFVQHg
zb>9IBeB}0B7IKA>Z-YNBS&Ja|bv2PtA)?a#ZR=u}!KM|;gmJ~gL(OsBKyAmQ1C=O|
zh4D*3GQDZ0L-vM1!5)H-<!^pc?hz-(PcGi~6E<=cS(uopH6%nv8ikdhc-uJriS0fz
z-wjdIeC#0mJp^A}dGxQs_qb(nM`C>c;}P|d2DzWCprO9O2jR(yj*wI`P>$SNHI{HE
z(pst<ib^i+(`G*@Dyv9USoAaWm6Q(&<<)kFb|0CpLp~C8<mq*}^o@@Fv+*er{S{J}
zbOF6PX3l(o?0Dujk*qKjp-c|kTqJyMb}wpFOvmD8t=v_k*qaY)zpxC){jYcbWC+L`
z-}azd(f0OsZk}<0U8m2}L0G3;MQ&_htTl=*_e6#rPS@)63eqVNEq+cN^Dl_^Ire~1
zMBwOp>6TX-?bfo^6r$7uBO5S+br{RHh7ku<1zpGWD(gJ^<BvQHH~%hf3KVFF?&Qd1
zBsNM)slFzP)aY5VTo>(K*e@=(<gLK6Hr{j{EwSTT@$<rarwoR-nya4}#$zjesemYS
z#VPUszt%%AsQL`l;xN$eWYK0$R#I-+J#e*HeZH*y{hOHWZ<Xcj#J}_x0A?Pj0n%Ya
zB<x|@yQ)Uru|VEw^%my`Vi=iS?7VLN=1L29j+GULmZtNF(+=DZfjQAEDZ0d>rVvJ=
z*x{UpVSl-GLYNOOFD0PqmSJ5-1sw$q1LB%Yo>YpuVE|TapplYAzV_>z!+?NwY(q7t
z)!Gcz@6rJQy8TU%R$TlRiCMW#4oU1>zEpJM|G1y(WUyOqR|1$+k(`s=4tQ$qw?)8w
z$Kq>zOiUM$vgb8H+nLe1(M-CHS`8$Ggl0SXPb|hIO;P2;R1J>;o+hnPCJ!<9YGrQ!
zFbwp6Z#i&CdbT3-?a02PtilTS6H04hhN8ukpcAD8G*5HiJT4MvlUE_Y9WNS71Ypjv
z$O)5i*>X_37}I=PwJ*~+-?2`>fn}&>OuC>BvI<c#Ycb<ZCTmx2Yn_MP-rVlsQUg(V
zUr%&n=G#2%K11wf<iy0x?wK{#E;XpR24}1j4UpZ~o1CNu{~kZ4tX4Pr1r@<QJzX~;
z0yfO#e$E67!|!uCtZWSZiFB=|>quGI2xrm}QbxRMyf~^>rBfxPPOS<`%GpYj{)Z*K
zhhE8&Es-+o!L<2d7yum(XmEhKGnU;92MG5vTuwK-3|W;vefnGr+0z8zGZpGE0e{AO
zA4C=C5sXPxnL@|}X2#n7TmMh2HH7=*TRmV<py$F{PoSqR(L4Gc4g83fiVio9q(Z6v
zH$Hw?`C)>xki%~w^visnF0N_vFTeGNUlt}wZFCNkpNx^0fs7MXi^r%+)$(7c)K<N~
zdx6_Qq&^8bZL7e{gzTxxi-LDm(9`KY;8lnqG-xeNe~+WacAG+zRw-sBJ?}y1Y;5@9
zU9N>VU`must$AVYUl1U$E`hqF#8X~A-V~0?k>3xWa{sfCgR{h;gc?AQZKA0J46F%H
z!Cm`(65k!0|9IAxh_TrR^S#$w9|hOEI>@|jpk&r-_A@Zga@d`i$H~73DsBQ^`*)VJ
zm6X8G1#|?9ICJ~6RZKvo{I74X!v>(cF2DzWuCb*AV`P?dX22BxxK&Tam(bHUph5V%
zB7Jtaq=oXerxh5d7nsnw?qLvGBsB%q84)h-zAnWp&{V6bA)3TcxKER%&C&EjG+(aK
z7RUHfRo5)Aq`0^qs%az9h9+Cwr;<ZJHPucgP%s%yK-<oD*>|I<zSXk3YU1QTaw;n0
z^3RF~B9NAa1UOJKw>glWx@P98%9KMHEXj46#%g%fAWu?LufrnYKVcUXbt}^yRTF>j
zb=}l8t|(u!=ADu64QTyjg#JwrCjOLg+ihz6uR3_!NW}0rqv=^%_E%cYPBnV)0T^3Y
zp$$9>K=p16fYa2})ZEF;P^u^)pCi3NXK(}M5vb#@kUL5*ZQhtqP>dM#9#V>wl5x;7
zMM>>uSD+`cU33_J@*Wb?;>UjC6o4;p!|!`YaLBl3Tnn`KYHLSU$Cj#2P*=^wE<}&q
zRY-`WXZ$D`{{GodmUiaDH!u_n9wN2NN@xY*y?$JH>F<iXsogWb>T8E@P>?JO6*S9s
zEFM0*mvT%6Dh6R~s<TcmdTbnNeVwp__c{#4d$f0nK?QkmK9A?7iTxg}8bkZtUJ7a6
zkN-pqMJ}QD;`puJGF^R!{6`OBYj{-9k^KMa4D%nvfP@}HR$zz{0~i9-bzt@`vtASg
z7W+<M52c}{T>;t?bYf2QSh6xsw5Ps-#y+0?yWHxnh!>7dfG@-Jh@nV@qQ&yzS7!md
z-x2mMX+nwQ_#7*~RN{F?n|Bo>C*=^HlBI@6O!t!Rdp)|%wcj2uGS<eX-J5r@MvWEV
zGt^qk%Zi0`QAH-ibG#zAk2+kkmD%)e-T{(5-;Q{*D)H_8o)L1T5B2@|Ecy3fIp!HO
zmCP^?UF+PYd(u}`>?U(gOv?ark(R}NQ>yW+U8B)EKmh9x8p93-cNB?3=!RN=_+*1y
z<H8(GjW)y1J|pG8PC3(Sadc@PUcnsj2i)E_6a2}3|Jxly>~nuKkHJeM%FRaofSP9R
zS3e5)#YOH7S;?Fs9L%JFy?a**`fB(*)>l2+JvVk$5#FzAP`S0y`Rra5;CzQ2l^Ax~
z`Po6*^lI!Cn_y5MRQ&?E+}G2}&tb(7Hobm`zCpUrk3y{RM6ckZku6(?=lvT^K`q9Q
zBp2_~Ui>6oqKfpeDT`oTFUwmJ`*YR>Ilcr(y<Y>D7bEV?Bwxa49+o7U{LOX}NcRZ2
zZS9~RcyRfl|LB5VW3+0L&g;y3+80-)&f@TTn<ZM`9R185@>(dwHl+Rt^ZBp$m^971
zIzIedc5Y{0mWMjCs~;6ZW_-pQwq%dTDxCM5Tm$(6@P}`OiQTqIDx5+G{wSe+R#BQb
z7|`na^r|ZKbV)+~IcKej4;j+`c!w6}pvP_1r`<&Kt4~&ZEf4GYR30hcW)ao-+J_uf
zBZfF96o|_FBK^t<wblomw)5Ikq!bKdwS;8o4UG>Yd4MnfWbz$917_u?Q*z4rcN-w<
z@B>o^N!q*$kGuy`0&Y9>ev!K!QomA>_a~{&#gU@;CJj#Map^qv_9EZPzXPxgB&6)D
zN~HXWv88($V2rO1<OIqjwBJP$I4UDGyP$1eBW5u6YMyA<DcbL1=y$ma?=B~N(NZ`#
z`6oW~b)<jxhpy0I^aK{7%U9M7l@$Mj*zB0kmtH<~2X6g!@N7oc#D*PNf2w!UV@qgW
zm`J_H>ojJ3BIsQ82>NHPyx!5hW!LAyAIbD}6rb?3(nsU8wPiB?u9`RM0FjdrN&~T5
zMskB}>%gcXwN@s36>?1t3GiyTQ)rA((UMdNGbQ>fhQ|45<ajG&)l&Z&{osbgts#8S
z-baPIgg|pO0~El#L5C72%cojXc2R(F>^YGE-thL}Uf-yO8Jsx~#^^P#*=6_x{XjC_
zYZ<gpVnFe`rX`uh4pk9=ZUmD|?#cl*?V~b0fS$5dK2Ogw*wsF`usTmm2?JWi|GLkd
z4il9~cKwLR$RrL~gw*Ge8+bBkg?JqfZ^LgCbPB{2b$l}ut#B*4`wXxAkXDGvOy@3&
zSf(H5#w5Tr^Vov}28~!o0}DOty!i-mK3LYu4}v!JfclPW%Msj7@b!Z)tL0g}c|{!H
zEQ-xWspNntQd}HS!%~}#SCUbydPqU^f8_rMm7I*XS=dIt&%mG$bo_r}V+BRkRdQb4
zitzfXo!T7Rb*5J^ZG22Y%mG9OtqkqEei{)VweradAAs$0ORO2MIL#d*fEK4K1c}OY
z3TiUj84Or@Dg^rqWU=X=#C=a!LzVGfu!Ne5YNQE*z{`M$(%hbYUTvZBA_}L$o(Y9d
zF5L6tgrR7<b_#c{3GTUp>u{RHPsLfFRW7}r&wN^r6Yp6I%b6M;99C}c)JK=QLX-<*
zYOVEuzdgYA7KMh#>os-k(9Xck=T4*+TUmh~1$7RVTngkYjU(BoV^d!cb*X*X{Sq>l
zc3g<%Z!9+=VjOv#Xe8yKOiuCNMS|;*h`Gac18;>%<aJ2P-|JW-0*BULyUl?W7B%o!
zmK!t2O?ggEIWHLI#8k{O&qUSGjw%u9*t*n|pd=-6Fqiw6&nFAoXy+xLUx)~Fm+t4(
zNO<=_;+avNdUQw-adEV03F-(fO)uRBNXP-A=$}9gF?t&NnrcpSG?$*%q}RXq7YIGc
zMM7IL!8nLYfwj%;k{MVVl8+_1HbT-)-jW~7mtd}Hd}QUg=R>js6NjISKLiW?oFMyJ
zg0UHKC~$Flk`(ND>G0nv3Md2O38SmNFjr1^;cx$M1b2xSj^qroT+~VDEJ1(<5*FFN
zUYW~o_NRqf)=AG-DE6Fq<Zjx38`BLmUi@naUX@A(qr1{7HpMUxlB**v+@3`D&kgY9
zGCy2$)>=xYFIUm%hPgvq^=BR?)C8Mwq$JjfDbhJ!pW2J>Xm{)=zdjniG|OUQx!A@3
zSULDNEIOgPbhC--(Uy}tuFaX8U<=bKs*1BO+q+=ebi&y7Y%(z;skX}?;IIDS@)$Rq
z_1l6Q|JQ8csq@iTgPcfia6&uxkqE-^*?t8`n&h#uft-rZ6Vw|ad_aV*Hu^c(H@;wu
zTpgK%3wdQdimCLp`@k1bY-|pnR>_wS=O;Vkc^_=lo_1dmn4V7zHsZ}$8Y3(^exKA7
z>4o5(-lKM4oPSTI(2Kx-`N4?Z?V`!BJ!_}D;3X48jo?(!x3sUyT*Y8-U(4|%5Jm*t
z{^<pC8AKV=%iXjN5BRcic0S8DfN;9I7{mW>3UG%5Yv5spb@0qtvjwPmc}PU|q(W-?
zMobz-KI?l++#bGhw8H!mvdu70f5p&WEC(i;%g;y}w2(=s=gZ;Pg+sECUWdoG{I>7<
zw6*A^fhME{n{sK_CTx7eJNaLQVb{KfC@JX|Y7@Va_ff92n_UGaOYU|LjzAm4^{2ho
z3H704Q=%{6(*fcE=tjL^E>#rilM~ZZadL_T+{)_Rghk7Ik!WVKl}De^krG<Wnt!QM
z3+Ttw^0UFa@xE-6;W#EBq7n94p#8Brx0XOeKmTz#`#m=BQyvy}(|HCXR5yN9t^wva
ze^vtB&2$$m(lt4A>nPnPFhxka_>+4U4;d}5N;p|S<0u(tZ>3GOH!yC!o=<zF5O?Ib
z)HnPwt<P681z(#i<cBRseLS*?t8DdO1W#G7DI9*yTO5{3d7q$9Pb>KYh!~_>Zkw1g
z*E4ANtCr{{$hf;|)}Y=syqDi-D8k`3zg#;{qt<?OOVqc1@Q;z{4iYeY&m7YYSO1@x
zz_w;?Z2qaqYSm7!Pl*7kS1BVxaJkU^&@a!S`PkMx7ZT0VvI|Edj9DRV28O1wXEQuo
zRf&uPN*8%}4>wJV=bBYsZx4cBU)$6x^Es=A+M_!dnL&t;lSB#h+yH;Yext$Q8lvg~
zrv{5IKebV=r|t6zV@a=pn*=7`$u9t8ZFasvajTZ_2~kn=*$iyw;(uy0jp^5RLc8;Z
zn!RO;IpKiCufyXO$JjO;I(s@G;Zsqh`@1t21E=dBXO(F4UdW<a_SQ;}y|tVvix_Z?
zh3>XL{Yql>!lG~+Ztea<N-aOU8o5nBt<!08;HCiD?I)odV2Dfn-w<@KCEJpFoDJ`+
zhlSz%zikloq%XwH#W)>eu)9n)Y(|6Cqgovb(X^J0yTIyvnDR->N7a~WwUKW_y#0yq
ztf3gCHuARA8SA0=gN8HE+S2l*GEt*Bi{wta9(6KMkxb90o_}W+?~e!03v@hfugL*0
zc~`in7|BCmjBMM)Up`mwQp4t~rEXhr$(M>`Wv(h+Dmm0`VM?QX8`Uyuq9;XUIH};B
zE{maQw-n1(%t8Fb(cPU90jjc*>%XVEI?wRsixKM^dO1pMC?l(g%OokbD}(}(g+CkN
zpprAD+i*%F26tz<BmW@rmy6>Ya`6k1Ox4pf6WybG-FRX*`~!kWX<wzBuY3B)RC*^;
z6FRQjl!Di5xG<Gg^F6={`LO}zsd`k?z1`zxVBj|t6L1wG{%Mo3V)&efI?O!5X1^~^
zbfgt7>Bb1EUFd6-MHOR`TD=+(_<_D}coh?*9Hl0*(;c8xFrNfUH2WBwXSK*kQdxE+
zEdU6QQ==KaK?lDq^w^$ehttPu%R<L`bkSDu+?v8QsD-K`!4Vi1`=-6$x6`+7#2%?$
zf>`o_elKb$xqV=8!WKv?O+Dp(7DqGa4~<nJC6_RM-xtM=cwLwW_&r@%)XQ6z{~X>v
zJ>sp}qTtuwx+VV;7v8%y@&4{)Ng;p7rFKq<(M0uf;D<{sn{hvtj%M}G3)uc`+?fZ=
zAPww5lM{3u2!bYxbg`OG2ZiM2<brd!UO#=_$D~{(shjCukbmWcXn44gAI}B@0W|_(
zL%Dc9X>9F$opUTKNPd|$n_-1Pc`k-yTA{R!Y_Jl@X^U)^ky{?N_al#rcyaz8SZk)Q
z&i+pZB9%wxzHwu0wP*e(%l0g5?~X7r^qS8U>uW6-f#J+FtYv<;uYjK5=E7NL;R06R
zFM3D&BAgQ4t6FSKeBm#I6ObdPy>m5^{DWWs%A?yyc&GndCW57gw{n=BiPUMaNJ1eb
zO4c@)#-h*Qu51<F(pJQ&;WLV)k7ss7zGBCqk~ezFE&x$n?q|ai_4gOE?A8kj0G@#W
z3*Yc?+)~#EfMoV9cV0i?lvYx;uutz8oIexeXTTqohTZf*wE4$IV4goC&PKOdxoU~=
zJuS_zbKx*0Y|r9$+j8xpImjSd`zIG{$E6c~Sa&}V{_>?_IZ6~LD28cEe%3)zy_f)S
z!C+l)bV_tA9w_ES7l=ATxjL1ibfwMYUYq+8xye(L@F#@+(x?Mh;BB+0Xu<A&RL9<O
zpyqRH02vfQUAwVA6+8uqJC9|$^M#y$vUs@0yT+U=<&Ed-G{m(B;M?s@v4C+TGAinq
zsbYQESXN(v%_nLPfRpm-V%B=P)CdiDFMzn6#c@wv$7TJcmbUfWnV{z#HyBVCG7c{7
z?gqq82Mx;aP86(qA(3T2=;V!`055HcDKz{B$V7escLlJ<^Iy){0p0XQlta)oex_14
zP_L?kz>hko!wqoyR@c>mo7!$s)FAHcT>ZZD%v0|K##&gonc<v|mX6M4JyOkHN0;s(
z^l7hM!WgwzNTuX|#5OGrW5tSytdOP89QUq?U0x^5t)d80NuMB?4Ed7_U5>u(l-o(a
zai7}+`W_(VSl!!Pm=a~dh(+h)LfV@<OHfz%@BDBZIQ}7~WM{j*zA5TN-X;x#GzjPU
zPOom7J~tdTlVTm^j+?)spx|*nT4E3qa=V}hgtF1Y-8IOO0mKdhO%$ia1U(+Tx=rsO
z2yXZl6toUJhZ_(ukAv}j=1oPsfs=U!QH6$w3^Yme0u|)W^nl;rYTIqL((>zUWe}rw
zZL!m#&R0uo5A)$rk60c%D&SN&XW0Lwq@-T&%mQdMW8jJc?&QtNRVsGjUx@#8FwF~X
zQW{-s@AX)8ciI-_b2_5?`_@y&gkyIm0MfL9vSAvf{G4#~T<Pz0pSimc2IG6LkbBKN
z+ZIi#iS*TJ8_ng&5Th<REvo=3CDJDT2O62=dth?B@0rGs08`S@JRvtO3A5LO@VJ6&
zG9FVkd_;@U{|D9$E%{!5e8JDkJ(TT<SY;ZVsiiZ&m{{&{>xfClFInff|1$sz5KUD8
zK^NKi`EMB+gJaoDCaw(co^00M2xz90g{1yi<bTuBo&|}Eg8o|Q?g$QeF}k`Wfi4?%
z3)r(Y3MzS*;**ksRr3lSh<I!b9TtHHgB7HfwDk4{0H1n_GA%F*CJ1`OH8%?b(?%?_
zzPHo`B{OsAOFGriSSa?FQcnJY|M~8ufr&|Hymh10moI5FSl0q=YlQWWfqnlS`S)ED
z)0iO2ujf;5iZa55_+7UElLxk)^>qG*Wz7_L65m$r?$9GRP8v7&*uVQUS&g~N*3wNv
zcIk^0FBP8O9topYWWcPPH;i+8`OOh3N%D?}vp7iBaDahod=myjF0%k@wiRKvO=e%9
zHF}Mc*r9j}Q?orq$TIvt(%gG!Ez{5lZfIe?KihMUfx+}(OIYdsYs!}|xu@o8Y*m3i
z9Z3&_vQ$Md6_=K#0s7D#B!g5HeA3Y90k&sgtEb@QC5n&l*xlOdvKtie2kG&^*-x(x
z)kXlqwGl|mGhdq1m4t%G8+oC7WdTkhp{AY5*0#1^!NDD)qv06DoS_jBVm3C#inpvD
zH)pe+8{m`Lu!Q<H`s1J6G+=2&gLW4V0q+aEEqPBWQqXkOg~*EAJYoWPgAS|!+5Y<X
z?>I(&s0%+a$?@fT^T80?E@oRJ|4~U)sr@I~j4%ujt<)?u_$luO5^3d7Eykd2by)h^
zXMDEH4V)u4!hi}m$COc2?XHa#0>baWGoK60p~;ssQ$7I*%Gs|QP=oPO7UDZgDEWA*
zwh3188pTc-TO2O-Nr}FLnUzrqWoRCZlZ*i7)JGhd*LA3eL(q3(!ttlN2#T`+O_Nw{
z$)UM{rM+$;=Y#@`OMAN$d$XH|PA*8~xVhNR8|4MHNIZS`sDs#g4ri`mnj(b16_9Y(
z;GnZcIhrup97u%WznC(pf_Eyib8uJ(=?3e-VYqbI^tb{3%wev(GXe}jFqL1Y-#n}5
zz@~(V{%g{_!~(Y|8Sefd(!yEvJw^g+v^RyQ1%}dv+xq*^Zap_fSw1=hCqAAQdHo|f
z3xYk^g8hsoYhpKo4HWWnufx`qM7Xz+En@`S%HILgQ?%(f&hr*bJ(^@=m=K&i^7VBz
zuhIkpkE5P72u=XH3Blb2p{0ma|0fc(eoaXqdqyaAJ1{=o1pe(+c9N3i^5Q5upA}4#
z^sZ|1|B*?{N$Wd$Si+tkGKW3158L{W2coWtfcXFq4^K9pGbmNmzZB%P0ZxC|)I6DD
z)%EWlaNPs-d;amRYcU5Sqg>H37&*f*gV-{SS#%<H<nLe-K+pg&O?GmF!U%R6$S^Bf
zK;q-$g8|7x=QIAdzQ{*U;8O(7Y?uOm@n;k;BnoK_H~9e7i5)5ooJnv_MJzb*V70y9
zr$<V&@=u`)qQW-0mSSkkwBA)R50U=PylJpRbow{Z-gx0quEE*&5ZhP%f+ptH;hlXx
zHQ&Uz2r6{eIQA?^x<78Rk(!C8Y(T}ei5VSC-T-$mfwptdL=|5OdIca41O!jk<IVr=
z6m8088CnxZM(V+APw5w^g)d&NezArm%ADT!{QM{<XFQT2^;EKf9cTm$<#%q%K#WdR
zL3k{UN8yAi`@6^<qNYOm6vtCBbEvkfH#|7A&TSv}Oegs*0~goimWRt>!=>qBqd+qB
zz{A4A0_YyX2CT~fKAfe3?Ce(@9Fa27jNiba!B-;nRO3zEH5hVHec4G5Z)C3+KIWq3
z1@x{TIr1w@J5_^O;_laj;T|=Ot%GNy&)$39uAy<1GKYH?=_@&Un^-`Lyz$A4p1)6L
z<wV%%oU1Unw-ooyBJcyxQ$fSkf5?zmmJ?JaK(Uzomb9ow$|g<na(}PK*EgCaJpTA7
zOP->N-Q@%Ae<uxm>Mz0(DO$El8jy=ITDkTjf(S#j)}FP!_-tc$w>OMZ2E5>R2|O+I
zTGgSzB=^)YyF>2oRsngyCkIn(yWXQoC7rju+w<~1Os){gx(LbD0>mC6e~+KtuLwDx
zj{f&gY{BDF8@TH}S6PvQdpeF=U{;~=0u@(Q#{<iv7^X-y&**$-{D+_47YhrS(~W)<
zR8*iu-HLM5^*DcNx6!8$OkCN@xf-sx|5YdXJLqB*1fY3Uaj3^i!7ezhysk*Qh>Rdo
zk47nK7jb-99$9MU#H6maOH29LH$kI9cc$YkN=f)L`37suFwPol@LhdkPnw2zw+2HU
zetvXR7J_tm`L@MxA!p5R+j=Bw!Q%pOR=0n&w;ylX6bk*SRx-bCjPMH5t<f#_${b1I
zm_opz7LL3%q4CJ_e?*H@V&Q8`(}V4n<vc<XX&J6;tRU6iO^Hk;xR8rogP!H>eqad%
z9+M}~0eocb;H(S+X*Hk}*Jo;j`U5*vZWi9%Ih5+dj?@M0j^>tSAjSoT*D>FN3`_gM
zXfR0t#{IY*Kahw66$=Z(>g(_S)1C1+x3;-ST*r4=j)-yutVlS(!UT{EXj2NxEL~i9
zz`Zm88B^dKt$hvrY2Y~sTAAxeA8BRlU(QpY**!OX&F_Y|k>Ed02quPUX!>rL?vd!>
zaxG5LnNmGeSFQM|O9meSr~$7Qx*oW`E}&MR8)u&T1kw8IA(LAWEcpSuGn(5M^RHnp
zf{}9pgxFt^<`XbxgKN+`WInf=6p<(Mp1HZaO&9bas_`X>%4+-fNeSli@GG|}GEg-e
z;MGSX%O`p_m6g~Q>$ip!;T)T;dcMQ|5AR_=BscJd1mx<Fb3eoq68L%GVk<sxM_?BF
zU`GQvQI^@7ojR6{^vG%+$`?%(SnW3W881keeI5VGGRj`T>NGm%q|AvgX83%1a@rFj
zdHc)57^Z#m-u3u|`BY%gWuOnak8SbPoj61zb0rc5LRy+$tW->yoaF21^bh{V%#rvL
zbDwa&sHosmc@OCJV2%Fnjsw#eY%_QwBq3SZd^K4ttaa;~ySqEuLvmhwQ<uH+@^b0R
z&9yaVU@C?Y4<b_8Tvc^oRI@FvD!RGv<2$VXw0XsqhDdW1OfHlJ0$yQy!a+yQIkuCF
z<*?rV3{u?%si;496Q>QS^d$mEo~1_nj~rfvTQTPpj7%))1&>5uR&X0R?j(po>d}aw
zU5)PixmqYJBNoRzB6|yQDoOKB)KO-8@EGdaduua#FG3bI{8z`EqK?LIT)a_na(M=H
zeM9zpKM^lv%vO75%P49v_X%MzC3>rUS-xj0K4o>J#o!X+kI3R`{BYD<sQUKZQ!3c&
zLJ3?yjOD08eUjelfB!0o-*JvX4Z6EG|H@<+>*4r<{4SGSaVmP^gq#^pCp4`~zD~1-
zg@^YR*>c6ozNMsOaXB%B@xvn5(I>~k!ZIFA!Vs>qnpbah=W%gyS--@`$F~DYO7Qdk
zLhU-*J>uujVVgm`=&#1#-qlS_LQkGr7yw8%wuaN!8F0TC7=Va62+Or0DLeZeuS2f$
zx|YXaCXmy@fNTh#Hh-*WrjDrC*zWw#Wif0fa*5=f#DDtZM@YGCz5_We9J5|C45!r`
zFouJMK)FU~VVvoEfo2tGj_ftFo>I`*Yg>C=BJf_&`D$xV`zK(U6D*=0ocqV{b0;Ci
zW=bb(%e$GbLfL(R=XbqX6Qq$Z4a&k;bNMIMGtN*~>G8CBzYS!n5ty)GxgC9b#l?YU
zkdrDWU^@?2Pk4VJX65F>|BLU7BwwVgjw>c@g(2S5*8dA8Kvg4m((tWH_h?esv&S3I
zdy35efjub@q|tv8t(&sQ_Ne#q&dd)g3zL_CFw(=LO)7392!i@%-9jgw_Bzn%T@TD|
z{QTH;M%12QN@lc9?ge;7`4a7}@#Ed55Wu@jiB1d454VR!j*mb%R?vD0+>XpZW>$aP
zLHzV>%1$(9&XK2DK%4xh&iVKYIMAQch39J>@?Nmf(+7gqTit^Zm>f-@=I^jy7aJRT
z9I?>w==EmG&Bi<GhNs;T4^5IiR2v1X7wXMT4s@1RPBx|IL6=+QzzcpGvG8|Y^tWSk
z>A$_x#WU&6FN9(EN#WUen++^q_hi|AyUZoHj<(Gcic-Z0CfW^S<db`EJl;uDuhqsg
zeyc96ShZxU!vIHT3v8%w?G4}hB$KykKYGOQExP`EKN$3PG}KCI?!OmFVPCmA7vIm<
zh&k?bH3sC^KiU)GPFg<QKE@4vJ?7Xh8k%(l4aBm?DT&E;85he=tMZEr`dFUCPH}OL
z&j;kt9m|uTbLHPJfR0m#O^=R_^Z<mW|6KY5w3G>nbdY={a8306e|)`VRF&V?HB1Od
ztCS$2A|Tx%DJUS_jf6<IbY37JAf>c)cXxLPNOyO4H$403?|*-IKfDed4u`-w*FJl%
zz1Es@%^A-VRaM0Ww2hjjIA^^@QkzFdBS4j7&==PQ6x;||K7a7J1q6!8a#I3_Sv!J+
zgoJ<ceFn#gwC}cUx$SqS+5s(qkonVtzP>&nJ_k_awi5jwt}iPee5Ml8*MS=wxEFB1
z8b7C_!vI=UK)fu3_vj%6%I;Nvs)V9N^4XgSu<Q|Q-ARgvg|bSl<+FWz{kSYFISI^v
zw{zr7K$QRdm+-)00>Shasw@f;6yN1F?Zz_{tb*Wf5y4qT^ju{%jQCiE2#0M=66!8m
zd3>YN%B9ljl4Y(cWjEsdo(P*9gXGdc`6^kANoOlXuS$IAYh3tOd-l$uk4|v~g5RF-
z5y(*fv|D`gO&-z0o%F)N#LEVqGnTZ|j3w;n523wwy6?wm)<?F@sdngfQ&(?sG>QI(
z7VCP`)!p_nhdG10?$JMLcYO10ZAVg)X64J4`?CS}{8tZkcM+a3cF};PfyL88_X?Hm
ziDL3WXmIeosOY#;VuRatq_)2{Nx6D2xc30Jz$|`Q{R;z_t(A=Dj>uNXRCc-_^)ZzD
z*^6EWDtr2Qdn>D~@dgSC3XVPzaK5q-!~CU=-B%VWr>VQCvs~{RqZT(7CHqm3nO7!j
zlOl5!;51!lWQ|9n6}~rrMlQ$msX-TptAfh&u7+4DBNJBAG?l;S?gH7Wv0bkAa#9qJ
zl}{Uc1!hsvI#M)dbq@^DX!ln>Wu820D8zffeOE0bLXE}1UDn6lb8P;+%U9;Old3Va
zh!5FKZnaSgp~cIus#0KfzZ$7Lezi2ihT(IG%^_pll+}-$!0yrvzUixh)8M(2cs2+j
zz80_luZDjc=7-Et7So%0$?9eMvAxOGE52*L#velJ%J>Dr(ibi+%+Kp+^{!4S>fB@%
zz?`^}nbyMT!2!re@!(PiRr%jl)+l49r>Cz@dwNP;i}}OU1Z9RA92?wk_|ml*UZW??
zt|0TPuU-erbHzx=*~2$7YIZ^m6tNfAnf*gGqE<oH>}U36yN~}=yIEV&kO_UHm-!;Y
z0k?Z)TX3{z?^a`Vr<20c7lW{$KnM5r%0ByOFR9_<MEM25L@biSRr8!#bkxJB9{kTn
z>0S&B;^@AgQstR~yS(OEbo7vC7Eu%s7Ofb<V;&Msu+{hGD1@rC$uQDGi)~l8-I!P=
zekUH0SmNk`f8_nTmRkl{HMyL>3Gh!gMOTLmai_S|2kUYCnrzUaZ-lC_Qs=V%3}M<S
zIWyH7Js*ejNBSJ!r_<SDw{va6O0HFx=vRBq*`)SYx0}t9l%bE8%l08Q7K!||&Wi#C
z4#oq(;D`bmH1Og*NT5@CPt>W+w_FK9(~Z_$-Q7z-;HKS-V|i(*Whh<3+#LfJyE75z
zEevlP8(Yj+Xnh9#&BKNJ?xy6MjFi78CMIUCXg*H7r;h>QE85S+<gR{{BF9&L;o~n8
zCr?fevQI}=rki591-m)?eSV}ns#REH8O>FZk>$ofuNz_{Op+gTQhgusZfbUhFg_Ni
z-Nj~M_?2n`g24hlqhnF|Q1D_QP6T>j5kF>wm}lb9^WsN4;esE(sc7*gXRm1f{ydp_
zzbwBboL=7MJ;qe#BfnA~L&)_LMLS^~v}7AxpT`UihHW^D?sJoL%BHOQuJcpcFE5q|
zpCg2?=jA{R9n|dy=4kDW21N#A1ZkaGI&J9Z;?u}5HA)cZkvOzJn*#;l)#+Fc^BlVy
zQ8G8h20r&2$Jn(`9I>$GgbouhfM+H$A3y=N;_rc*m;>FmHZ|?u+$svXrMx<-;wYnX
z>Q{HI16{7bqgFGP<Kf1JSDSBzf;u;>=jkEw#<AUBKn~@KC0ch;4bCE-15JtDob;x3
z>IrWVZCXT=1NEr@R#WKb--9;8@%4vh+X}0|nK{@!MAUOKkGN9{lhNK=g<J5N=6fw%
zA5m2$o+~jHNqof8#-Jt`P#HMwtHBd~rfw|AJTK|4O}%$!>)XC)RcpDr#PE9k=^BJ)
z>Dv<iHV_#{K>9ws)Fo$sQ@PK~U{^vSV#$J*#MKH%(nR4l#V+6WAmogx7QJes`1*K$
zqWVH5w@(RntqVOIyqTzL0<u@Cw%qB7b3hvQ9ca0KNup6_+nFj409-rxh0V<_K$tR*
z=>hcsP{Pv%2@XdPa3fy2f(%%qFAR=c#KZ2G=@+w;j?~iGOB9+}1+6mWS8kn$T$z65
zkXY<*G#&gpe?4DEh|(o&Co_am=YTA|Hq25+4?;|O$6~ADTtIvuX(VOqe$=)*v4{4s
z?y4}cCJEt9NDP7W21{LSiB%4oV>J%OQNc8`x{<DhkrUaRhlQXrbXr)Ru04Mhd{Tuz
z(2F@I`#rl{9)9^|aq_vO2o7z+<p;Hm0%TF;+vY*Cds%~rjmPVHa=CkZ2lfw*M$KYs
zLswcew<{81dj&ksH=SdI!qQS!7@Iw*OcV$kto1N9zAqe)3V~gnptEJ`6Nv%-*SwxU
zKUd7%ogX@0?>7WDGHgKKz#3%Dn)e(x4sWhf<R999quQFh|A@ppi8aL|r>@%aZG+5&
zn3O+*v7U@5<wVJiGNZ2FJm?F?!M^GR1ql_S%HF+|dVafXVsQRbK9>~vr|#rhzjRIW
z^dr=~2phPt_v1v;m95Wjx*Q*jXcpbN4fc!cLN^V{R;8l5>tEPKG3%xXcNE(!xFwq@
z&<)t9RYhFUe4&YKTs!S8#Y&<3QHyh*7xb^|;yMdtauPq9z8)-Hoh6H~)WT5yUUxb;
zsJ5wX#!*Urg`$sN{%=8q2H=piJb>10RHYV7L>AomAAArj!7R$<f){poqjnwS$V@I0
zSY(U^)}eE5E}<)oik3A9Y%4z&U+CiFq+^-+C_SNxUglRZb9k<k*)w%t(RxP#H!>3V
zOJv74_y+Ob@lnYeIb(tb3%1GfR&E!McKTXq2ckq95<FdK3%S$rIC2Q|=AAHt>P0qx
z@?5^olexEfTt>?f(GtTJZ3{hZXz&>8^1Q~Eq;PyYPP9y5uxex0Ymxqxn?FhVjnb1O
zuW7eIf`~6h?fa4KMalINoV4cL9hC?U=uwo>6Ub~dXKv+4VaZ$evY8yjuRxk?ai)lW
zSlaeM+bDl|rh!XdhC&1XuOKTpDLWIcKg(%>_}?F(*x>|<;7e#Lw=jnoCKd)0>A5~T
z)o`of0OSS$Pu15Qb<?iyzJ#4IkH9MQaQyJs^oY(?8l+iKD2`JKp-Pf#zu4-$ZgNO+
zhO!9xx3$zA_M18?ResnpA!4n)F7@QwjnFELI+guILpHB>^Y)u0SwL#7P5n^KW<l&B
z>e+e}X9GlG4DtS9tc{x=mA>p|f9<ozLSlhQzNLLWobs+WAO+!-8D9qOJIh0P=WQ_a
zLok;1xN`GhrDXN&5!V<?tpBv@dneLwDzgq%(l^9OvCrC3Y(7kp$<*?Hq`~x`wS7P7
zSF*%KgoknJ$=HumxyeNTZc3F@;=jOQ#-<8+5+niLbh?i<;ox)t)ht{<^6?lnYU}S0
z0VU<qhHB@W&8aNVw3@bTlmats{(*sIuSHBuXn-_())BwWB*jl;FHZmn2OOK3Ag3mu
z33B?;_Gfo}Zt>t{;A9zAs#lbv#4%WTva$AhZM-CLe|9!k3pcMo@vS$HS#=xa51#b=
zFiYKGlZ!6y4DsEY)9n{a93+c3cMs55Y?x#J7{wDG>`jF(2y#BQH)O0saD(r0*Q(r@
z`Hm9B-#^J`9d~v5+|WMZ71b(BM%^bib)K2&%fZDlSs{_LG`kQiJAUb-pCzA)*?p`5
z^}A}~=a(Hi=lX`J_GKb<cM&Ye$#}9G!8&1-&Q@84Z)Q;kt4p+hphm?iy;dPOUWubd
z<8s2z!HaRRWw7+9whmSEl24cE=?Yr$vsoo``IY+=kCO8zN9smnCNB4uOiy!W8t{Q$
z_R)F0kl$LiAbOpOQMvk5xv3}sJ96sgCnu}xeBK!wXDfB8gE1Vy>4a{U$$+_g2S_k4
z&KvJPE=MmdeL6ZiqT?4)prtR~v$hDvxH8?2Dl$@6sf>8sE3r1orr(;Gz$cq4k57JV
z=3MVyU{5?36x>=1ufr0<w1qjbO|zyKkT{%QgEE-o^5ZCIGO!^F*F}_GQz<&9-}=jN
z_Z5o9^+a~~(+pOh6)4Scab@$<M_zROkG~Z;%r4Y_i`8phGPB1<2$h6&)cop)Yv^`H
za@w4PJ^iM{s)P4M*o`(WoQT(BJ$7^F_s6n{=l8<Gr>d%zk2g-Q(8AQbAN9(l;yRk9
zj&GaGOTMw&!i6z+L1kA}tMTR?$d%uhe&&e0wKfe<OVx*%yY<wc2&IbT7Ylk8bI{<O
zNsrMRsNZ!oeG4JU$cUBtWZ`kltWsJOKWSF6X)UQ-xAr$N058(RF1JvQRIk;qVOcM%
zqmwj>UIC<WnC-SkX!coSdtEOUeZk@bd96HdX@!!?$|Ly^gSJ3IuQ?BqRKa|UA9g71
zb}f(*OC*65n?2D;&i28n4o8^l1;$})@Q*Tqgt{`b6CVn$N~_k~ct#Rr`}E_YJuWBq
z-oP$XPZf>yQ+Ps-XzPgVwgZfl{Of+D(qG3YPyDBCqrqRmsHH<z@q1|q&zzw91MD5~
z%*Ds2d-c;#XI%Lrf+Upr?F}ftiKlk-c0z~|8OC>oP8Tk;&p2G^t1g`sFV(FMt*O?9
zGKLH%o4+5^yX<^)%jFL?Ql$(-O&#=$uP33&b*=;5o#rIQ{h{ag1U_1YqWN;EavDw%
z->Nc3&+hKG{7nHz)`)GqZ@am#PkL?}ZSW6!EW$qM6@vCO7O2#f&)=hCV0<2e(!`YJ
zsF3E?GV7X5Gn=J{&F&-~qDP;u9U&CPckIPnrRFin2gk2gPR`p{SZcQO?2O_jijf~n
z5*=y;7jFU<KC3lXr6i$$eX1QRccExJvcvK@_IPWtX8M59W+4Wv)hhyzzIQdFo9kPb
zUj8>6*Dj&h6n9RkXhjMNuG|sOpW*B_bMBUm4YG~%3-c2<+Xwbo<9+?6_eB-m9fO)x
zqdiA&fsapWCu16>`p4mVWWL*aPyi+$*ut+6h6``dDV?bs;f)n^j|=tu;1KsbH!3K3
zn;6k~h;2^W*c0o~Yy`yfKUrEbBF#OEYc_hl-%1NstT{<Ebx_BWsu1#7%d0rQ2O`+3
zNLu;I0bZ@;#|{<o@$o0Y1S!8vIY#r|`C`E<zBxYYrbq{>`9MAr-vd|@RZb-|0Vj+4
zHy6A1V~sVE)UuUh%_Qyv(}?O@Z&SYtk+-e2o=KiOW0@$nBC(DQVR9c-s6uh%w7Uw*
zu{%GTcr&5*f%==mCG_aiqS;zB?+3=_Ik^DHXga1drg!J=6M19r9cgJ7hG#HSw+!{k
z<(t?2a-Q(gYpoTM;v8C^dTIrpfP3KFjSEjjn-=XW$B3~N=hz|M&TsF|DRq7vyI~57
zxgQoBe+uptiP%A*Zu5DSVIl_A;YtnvwFWlImm&w-q>qVxN)Xm2TiUbYXV|ps7RH|A
z7-rQ+6F&c4m`mKrgv>eF3Xhg*2F=jwRGtR0b5<1Yx&yPZxt)(C{@QaF%G6J!QN0K0
zBX)D&m?rtw9-UUg`ANUpRb@$>6^QLrsrL@(4;rnk>+#zozBgucjeIbee45y9Dl$p%
zkc3ce`v96Jxc#%{l-neYDO2h7vmxTTxMqUdr%{WPi63rGdI3~ouFU)~oyrY<siMP`
zqgJQKNyh^gAW1^VRak6LKE?i`hMmihQZf9<dY`^RrX<1n&OAwe;r0pv;~YL7NqAeI
zAZg@%o7Zhlyfq1T3DvVy71^@Y(4(sI$_1LFUFg>j*|%WhJg_y@zP|<eALvq&-%DSl
zWh-P=j(;UH-NJ3d%$a1<g=uty$;#{_)8?;8uo>;BdCF2CJKx(pIh%QLn!g|3vIx~Q
zaou;f&o&;}cvQaAN#X>T3teK(!8eVfO(1X_*rAoGxp%kgbA@t2Lm0fCJUGP+Z3UsS
zV+GASD^l-qXX#q_vDj2rv{$HaFT~4kEQ^g@Vkq+-|71BnCt})|926J42{`T!T)tw8
zUhL(1%?t{+2c+XoJBMSBcQ;v5YKPvP$G=Z^)F$I$Y~4~Ulg7e`wihjLCDtUaw%dA5
zaTj~|!!m2$TK@YYWC-_^d)Yz}s!%QZXI}Dfkd7@D!BVQ=B<5!VE4aSe%-9Z(x94B`
zG#X9|o;&|(DOYGd!FseKd9*l8tW{jvd{l=&xmgw%oREe~BOri1pD~hOWGM1$vzm=i
zk?AbHvk1t`G6%|8F079?$Tj?$gBSX^osg!C$@t)161m|WSG)h|HMHGeiO3BV<vlcE
zYL&?Bc4-;QX?l1c?6d3y_c=kj-r)OKs#c<h3oYYBB&rQga#=8$qglWABq|(uU2hxo
zN+RF+n*rIcD7HQ;FC7;ZX9;q7K0aiX5o++{$MGFr94o5XrTTYcRK1vagalmSje@0b
zzIX&);HDdKwLq13@Wdkriac-?&;iccGc$b4aUN%zp0L#Lm-~id2Q6YBUT3-cRv|<?
z3H_?h`RXY%E4E?IXUf|#s096pcNH@ZVjlQ9nG;OpSV97)F|YBlAxTDonNZ41yGZ%c
z-~+n8ZpkkyP+9&SQYj51<k0>~Evl<s!NOBrkF#k_zy*~b%BHSr;TT$~+?^e=isy#T
zvn*sU7jf2Jr9kSaf5a3GmLj8{9=5o7zdy}7<)R6SvEJvG-u62am)W5TX_he-FRr8U
z5m61r=qL3pdm>~R@_FKq3!Q-6e+6{u$d2;?+KaZJrM7;3R_^`dLI``3R_J4LE^T#Z
zX%%F3f{|bAP^9KQ?)9QSq&d?~Lb?A6VYrz%+F4Q!j=qIN-6#TKKK%(-9J-h=oY4+@
z;IHG?x~a~iME7lsDB+kA&ZAm`tt4OLyCgr5^D!XJ_rXVYmJD5bp#oj`x15~OczHLq
z2Xr^lXr@C8lqwrL2nlE6)<>V1cX~YIN|qo?ZJFpnc&xRi*u%J+OSl*0CX9g`$l#~&
zHf&KS^Ikt*$A0y?G)%+j#So%8w?{n@LljNGcKezaA@OW{wdonTEh)a-`trt1oa09B
z=FQC&C$V}1;+z0tA|-XFaVnw;&u@;VO$4rRbY~ospcwf6ptUh7Q8Cm18Vt~T3S~UU
z`^$c4>hmB=eM5|z0LM^`9XRP~!%3Z>3=+}juyI~TP2U(9Zmm&ut6y*=&2&hpT12aZ
z;|$t+x1Rhs&wUXcWUS4O1rPi1l!!8mD^rgT9`0T3w;ILUmUpldO2y5Rpi2`yEW8}L
zRX;uT_C}E2=N(K#nj3CRgf3XgWbp%PClxmJ>Mlcv3Kk3LpN-I@Ke~C_lxY6m*~ugR
zd)vGo5#uE{2z@$BocAj$-`?%0-=Gr&$9ZZG(S{tKK3__>J8_{yLs98;ChP2LPJs2X
z6P8hmZoSH1?fErV@s$p7<(CHk;BjGE$mTgPTh<$V1G72?!48TM$VICI#h>NDQXY8R
zZt<f4>Mq7_9KqtbNTv^HkaFU}dpLgYT&39!TJK*gz<ro`{H~=!?=yuqLQ`%KHn>`F
zMi|qyn9&(lKaD(l{#$L+$t<WsB_Xw2xT!<&j>vfZWIaghjH#SN64hJb(kl6B$*H)D
zKfd}y8zqBuEvQ!xu>lJ=9XE-QrhQIs+Gt&tHsbq_EK;!Dp9_%trXiw!PS<QvHE9~{
zN4)PdHzFKDHX9gk9F^LAdOxTD;FU1j=bkeGYPjq83oEPNMYr+GHV5H@>NU;&+8Bby
z?5{dg@L=1ra?)yr6OH(NbaG&ZtM=PnDd!KYO6VJ?6_~`u#o?>~x8=3;<?`mI+V!j=
z|HM?HezAOcb@vgw7uB?dwQ*lfbDLc72*@0`u|r-%^b?cQ$>o)91?q<JB^NROmD{RK
zIS%oF3Rlq$D$e(OT+Upy&#F19zNgTAOU(>8EiSMz2>TVjY$(R}^S!+|WtO5Nbm6i{
zM&t*LH#wgQ*qziJ5fkRg<k5Xoz!4Ay-sG5C4*G6R@w_W`{{^u}V?{Yl+k1N!xxC=>
z8}afg#fq=RY0nmx{jA8~R5|iX2#yz_&OChfG_$P8dp7aD<4#}WcmMDFFM$CeJoQu<
zi)*6aKJBYs=V=*Qa_$gAJIt>LyyntQBq~POkBFTrERxOQsO9mIN);R6S;Vc+?AxUc
zEn`>yOS+ek_4SuUz}ArxN>6Fc*VSQaZ7|#ZwA+mkKK6h!#pZhHLT4htEWR=3M#4$F
ze*y(zVe<LvwkO;}JBoXGtl0JZB`;k@>$|r?`PQRE?7P>H&)ee1SSlac3UFA9BsSfq
zll>W&8KIyXUH!(j`bix&&dbYx)#{7)HqT+<yGq(;c1~jQf<u@ZL4D~1LtJ-fX%P0E
zt|v#&uCul4u2!V9qv)@4ms|qZIm|gsFVU%{*8;2CKRK8LF!+N!H0pj!^sn`xtbh4*
zd#An(`OcAMo4MLFAFUh7;K$LnRNBDNf4!l@`1A~ZQLw7==<BjmcNB%B!36`7$H32@
z9Dk($dpj1ACBMPvu9?Y=9n>U{M^B&^s$)kJXsp%<CoFQYfC1-~P@k)Hs~_tH3W5vk
z4#Vg_29Deh_wv#mr_@mx9lo+>N8=|zdZU3d*4PJ>NgA@{wYK%g2KIECGHZv56F;!z
z8crox=HUYR;F&I}nUhSW)<2}nu^#-YRb|~qUoB^Cq}|xMR6Fce<Wl(X*6~q1^6bcS
zCzuB+NKAc_Rb`g?a6tD!1XPS@467-2Uz*C(Q9SFp|0WG8>OWDf<3kCdI`GBuK)NE(
z^>jaLmreY!l?oq2EF*I>w6v)CSkSE&1{7dMZLt4VHPKnUcl!#)JE%R&2^<P#;ZrqQ
zRw~2;*9r7SR2uza)Kk~a#_xi~Q@`EE`yjrtlPUTAu&`55oH-uZci73OvSxqUqb6um
zLZ?vLs^*bFCv9PR4K!s<h))H}gpTI1*_@jH`Z@}+X;Y^9-}Y1-H8z)CSahQhU)Vo0
zbVz(j{=x@cwBO34^@Y3JHqYF%8wA?>L)6Kxs2Fbj7j$DQNLpUavWuX-@qV^E*r=bx
zQd*Y51)|4_q#W0>;!${b@8`)k=~!{Pm~fsTi}r)_$1S97NwF?Od2%F+R&mPSN9J4=
zX2<X4HNTr>Z6rI-@E-KKo=P4RDu_#9y3zN`3BmK)5M!+`4Hx6-t@GLU;=WTNVn26h
z68^PXM2nu0=wTzwzOVS{D`Q{lGix0Vn{^HBL7KXcsVkt_3ZktAukhrh>*?>>Jjjj^
zDeB9j0R+4K_&1RS-nXdC>Po+!*~Z>G1F150fJnXg%o@2=ymRRjKqXerT?JCpoVHj&
z$z2+3TyNeak8k%ha>mk|%L4$6GXp>ldWFL!&SUFOK^L}v(<+wn<O~iX;mUa-ft`7@
zc-~XQhx2D3!;tCR?ch}~DT9<7T2MjVYlx>fh=sPYK~evj^v~jg1JB0k!UB=+xx3Lr
zbwiY~`Z`<?t%?|=oPuA@My&r%-g(3Q6J?q)s=otki=c&XpU=}PAQefBgBY|wDd`j2
zkFwFl_+5!8$$a3o#L1TX7#55R|Gbr^_N-SsutC;jXxGZr|1%?s>c2HOlKI^G5d>Bx
zCVOc{h`;gPoFQJ?$6lJL9?|wo0oU(QELRfH(c+yTY7oyjxejpU(T8{FcZn<Fab}k~
z>J8K?ER;R2m9sIzkkVQJXsvF2kzD=3_E3zJjmneSLnhaVJXfXW!P#FL33c&v*qDNP
z(DE(OZ|+1v5h1iPYj2(PDTgdbnu@?EoqTu1>5DuAZ9vHA^QFDJ3tyxX<2~$GW;|Hh
z2`UXa3p=~r9>xv>$DV>&g@co$4E%e%b!Nyek!DNh!rRnyk*L!wxg=yPDJbtULfUzL
ze<d6Sj|Gd*o|B&qR*wTfFCX~IJ5X7TjMRmtB9mf!WAz=_Bjg&oLBT3N$nJE_pHUgV
zFiYy1pR6<<7Exyf-;G(evCw}@CQS}D{X~%On@Y~$x)U?;;{fYJ>(1PGGH|GH079W|
z_)#Hq`t#7Te0Wf-W<GivLQRd3I5XiPbveEx{7p^Snx{+x#jYYtaV%C?8Dj$Y87IQ&
zRTovuBrBqsFQD0nq>sMUr+V34vLpP3HG~IkBpYfD{zoRbfqPOxN@cMv$DplN)gtU0
zn@HtZAK^o*ZhK(4W&5{ade2Hw>Q06_9rDgpRNr%D>U*uqvmt~b31-(w;1I!}H*};n
zE*V?4JqoCzvD`k-M8XIhad2>AxshIYLq^K3Uo{qSX1;4=%*>5sYla=MmPBA$=ASlq
ze+uu+BX4BLKinRgx?0vI2W1RuZEBQeH|5{M6car8XY4XJVmEWj(}oYc=*&8HEaOkt
z3%0}Q;6ZEIDk+6LF+o9i0G0#PmqEKBF#h8xMQzYC&M()ym8`lg5BP6SpkeWIG6DBE
zA9Mm>{nco7xFigen40R(=X!twd{C6iXqY_EN&X-y`HGy}YiGx7tuMX@sIPpGmR<wI
zS%Zy%WPrrOq?W<I`T$l>O-&8BkXSSJ)8>+KTt5NPP24&)C1nC|<N-OWb-wY)I|M-g
z{8t;~9A*=;5Wm67@1MMfZ8+Y(D(*YI2CRT1uLqdC0J}j?9sH2U@8)=O*h$D~L2P1o
zTj*Z)%HShuVFxVKfen_z98sChG%|2Ji}86oLak=+Z1jASKERs1qUK*lvZVfnnbN;!
zeJFzc1?XJkzt8`An6G1xz`T9NY2rCPISKs&=s@YGb~G4+zylsJhnN#^j({;B&?PId
zs+}}5xH{SV6U$lHvJ04qfWdi?cz@#y#v1`aOpQ9DS?6uxcq!oNz$|GO`9XVX(C9Nz
zp$AH5r7doq!6eH7Ob3*7Um!drj<m9MH6J5MnE+HX=Aipqw93Rckj*!+^d(k{LV!AY
zczjITSXfpT4hT75ltx`Q5zvl{fRr`h)lXKKKLf_85y2`X5_)52j_f|BrNuDZXYj42
z8((-(S?G5MO&X;ud>J0(%6-pk>?Svq@ZHSd<`VQDYU<XBwiMq2LVTdN|L(^ZW-oQG
zt@TqAp-ekZXBQ8VQ^Puu6Cj)oU!>Fgm1e)cw^tvqRjO)ghH_QfbJxM!bpyLs6gYeV
zO&%0sVc}>=N?@6Gaj_giO;4{69vSfGR9tR=1oYu=@(|cQ0zTUYrN3Y%XP|XnY_l#6
z2pn&Kwm*Q#!8d;o2!Q<qx<7~ReRD{6=>Rv~35YZenv}!XA#riQJ`o3eDR6aU2Gxi-
zt6`>8Y{yu>hB27Fgt3qd$Mu+=askUWK|#Tt`Nl*PQpcpeHxn-n>_Su&eo%|TuAA3)
z9}(rUm7=j%U$fesQmV#a<jN<Hq+xt1M@%fi$0AIB1gRvsy1Br2T(x?`4P-6IVwnUH
z-4``c@{s+_BVIiCOr<T)F{M%bgT*6&<l>ENbf|AbR+`*+2h0MEJ3e@|C+ZkDuPp9p
z9%_yid_-?5?dR7>hKor|><6UikjhHV>+}82!NKAieojtKQs~AyP=p*9Y`p@*3wG&~
zP@@V^FbM4Dzs{2{wx4fsgQHU~%MzkRMnORV=PnvpY{d67Q&IU6XgR`r0QMH3PgfpY
z!_3(LedGRiAEMm}i;V1`N_3S^kk}rBDr&(2dIo#V@$o3Y(+d0X<1I+-^PT>;GK3)G
z0V^Y&z)-gN7z0vso(P9$m2Ps-M=zY+#tvxF6;3<aADv#lY4(<1iMM{jq4iZ0(cHyO
zXkF{P$K!A0@qSdH60O9+Iy#GNL)h~PSF^yq!t`BHX{jFI)Wh9?KoYT+$ZCiM2WU}N
zKGT(ElV7_&e*yp#9*9!4*L$__s-*@}<(<&!D5R20r>wP^=mIp~xZ{L-0i*0;6Ywgb
z!F4b#0rS^z9U)}e$&XxoqM``;d5<yRfL@MeS@bBMmQ?eS%B?>R6;rI~hPF{MP@<3R
z#08I5FlIPp6?)d$dc)eCH`~W>@I~|>e`(kWZ(t5J<3Y<r<bgVI+;1;$1h^}H?mEai
z?624GBO%2U8`@agpO!`ZstH4)ql5W}R9|?nHd4Xi4|!v=7EYiKFE20Ot$}i5$FUhf
zSW#-wy9A`^R3faUC|6C|0h?RVWe4EP%1+l+*2~e{b~G}4r?_D2E*^H0t*^yeQa*p4
zG>0Bb!t3}bsik^!O<dP$<u~+mG?<XX?CE~PH3LADjQ;m%2$t{PzmEtC%^!UvtOvH@
zlF7$!!V5dA;?1kP4;{QJYdqeO+%X;6csR5!_=H*UV#5kEth($)oPTULp^@LSjb@&O
zX)pawig;!HQHR6_xVNpW1}7=2ogbp2HqXx|f)V#05EvaPEM$b6cRMG6117?GWikNU
zO2V}+#Ql=1F#J*n448CJ*Vol?0S$J>x&s72z19b!HgModfyJ=|=<f1}*EH<sPk32f
z`0}aG1_lTNKyTlqQ&VkW7Z5mt@JnktbE!WBz#0ymaVzA8;Y$Tke04L~7VWdN#L&or
zIMqQJ6)1JXm)D6H{@1pNe{jmUjakdnQ}Y(Z=B;JQyL(xquRZ%pG<d-92u`Q{JDvv%
z;gPt@w;Y28XJ;Dp9(wHt;?haY<?(8p(MI6vI{Owdyd)y&lC!e>z<mU>vIa^;Ex`N4
zcr*tO=*t%jUpl$CK!^~&^Y)T>@~OC71r4B7<1KND{4q`ELsXSp2nq_KU}hd%(+1n>
z{k!Yi))G>2?DFC~(I{U&-mDYjD>u1WXWsuh2}IJdq|G3&XkOIxV=SWcBfQX3OC(>Z
zAgz!nNC3MS`z=s0fuoX1SOJObjPrKBQXkjuBveXuu>R5vQ!I1%F>o{h>~NUH0T`Fy
z7fJoeIqzC3eSN->%v+GFQdn@zY*`~#ZE_@jKbK=RS4ZfvH(Sf8_9x@Bb-*`Z!mY6e
z3t4PvZgQypjh!B0$RZc2q_4YC@Tq<l-Ew{V#Ana#GEG7K?*wNNl#BG{%LGG9>*I}{
zNQSx=uO~)rzOAlFSn-D7?`k}8b}2;eDrLC2z1^HFr2<>?IX`~_W5Wf42Vjk=ym4>y
zdkhB))GAs|K45~4O*sdHl=rCzaAx2RBIDOoYyCW$qZG&M1ngjbsGk!8O`oJVZQu%?
zeunkt&6_`o0-)_<nHvCo^P3D$6ZQffif7$E<c3U<L`k(SxAzORG0K>aEuMZqO(PaC
zExq|K?nG`fyTj4Ho2ChP*)M>Z;hcAD%L_b+F=_BB>yJ1`VlwD|eKa0s0;bGh(hTlK
zjrf-K|DP%g!ML(p@jU=TYX^Kmkk)kQ$l@2X{*Q3GLS&OmGY)Vg5)Z9<=R?3Umj#Uc
z*SDVi@jRW+EbHWweZ2lZ+*)wC{V`xIE-nUxM=14LGKz}0Ko=oLffVy`DYY7}mkLO3
zt4LQfB2QOL^dT=KobFF6=>{5P|Dn>lctQ0f=MYzrOh>&~^EZs&L#_9AvSV{u!Ce^J
z!@7%an4+I@9FtC+W1H8d<yz~HZ2(a`j3#eX94lMfJo_7g`%4)h72&uZ?@)Drcb!Xe
z4zAx(o~VflRX~|fH`&L=!YZ)0w^q74ErfnT;o)uI?r7mVpHWziruA8-#b?x#1a{zH
zf-D(Kbw+gf`uIeg(-q``4@&Vfav|)45;xZMFIhD5()QG67DxM@AM#(TsVD~YsVUug
z0>B1pX)trVctppY`Db7>HeAr)dHsX2^yaY6g?iR3Ndp*t_m-4oFb%rYUR+v|Prd*+
z=?GD0t|ARcFeNlKiNP^n$cVKNN*!F-+JaevfZzlpC14P;q|H^kf}Ak#V{Ige|4(pn
zVT94vRxjPwZ^fUxo_zZB2{=L**zf`>BRFc4z!?h!=QJQN1!%cb73Mv9A$%Qx?nqNE
zIj{-Y!bn09v??sXk2EemesI$k4CnX(lPv3%-;aTp&6b@O_}sGDrZBYXwY2$zYAM;>
zh+qdzwP$i#YV!Ax1{r8qfkwRZZpH~bfAb|G>Pm*b?rWO%$<BDN4nx+mTmS?uM|li_
z2>3R|E`X+n6gd7g-@k8bD1<olG@TwRYboy87JGdne|l#&|L75En;*e`s#O01Wp|ay
zJV?Fj&tbPcfGt<i%SMG=U0s9D0qGS+<lNUPQ^BXsl?GD{FdMKfVwrqpR=|bo2qrNt
zC8qONn_eEpCCY0u<%orisltXNL0*UF42}FH4GluFfKd<;#D%UhR<EA-q_sBQGX5LE
zI_}Z6s6eU1CM8C@5<xWQjo};6B-hzEv3_vYaOT4IFXUO%f3dRr+!k;?2yR^Io4Uk{
z{LU&@j-Z=B%`yh*sw<~Q6Sb`1+IloniJq0Gn)uF-a_l+MTwNXSH_T@ppjLSHvJL_n
ztS)n7AAD8$$tz9{^da%4{^pXm1QO4S;X=mf!+RP{tnYMQ_GHqhEyy!W<m!5Go_z6*
z<Q~X=p#bW7olboKryi}&uMe2EJW9E<gTfdOKv@mey6v<(Kb^{r5-YR6t%e?fvhDQu
z+Wu2eU;vrV)%IDyE{>#A`wFSwt|vEK@A1HVRYg+a7pm72^g!gwQ!nox+k|~$c-e|T
zV<Oj#5y-E_6^6n;B%93~leu25B-QKkvI$?>`+<L;|DbN<af)kl>@!<?Pxl0-Y<kxz
zOaogu6`nKt**^{n_<))~*RSw*1HkZqgJXn!K!Utn4uJ&VV2LD1XboDtQCHhFJwSRC
zvaw<G1P2G~MUV{ofl;B3KtjE3Tku$Z;#Yk5K7GBY?oz(>lc(4RqWETing&~cHI&n}
zSWzFo6yQB)7O8&>5}iLfM#=f^DCKD1uA5Le1x-l~HA0T2cdpLop8VU153TizoFGq!
zu272AX&}~;FFSFY3x-7JT=w}ucJA-z_Z&=P{ab{HjhXnW+ww-*-1R#u(%_Nh^ZsPw
z=y*7Tr>9q5TvcO5Tv?NUrlKO{)nzf35q07OM#{^h8WHOzku-NEj@v~Th~Frj6aTH2
z!BfmI0qX25(CWUlhg4NnBlHTiKJ0eW*a1F>>)qvgg|sMGsVsQ_M6rT?ee%G=QfZo9
zqVJ|gKC+*^)4L8@s9&?NY*Y_=2M6XIK1Bs%08?bx=ERGOW2Z7D5m=<A=uK-5%3}~P
zH~&`#_9pZasPTZp?y@2vK;p7uE9cJJptSnN#S*gIKbnTgj3&69sr{Q<P&6JWby!?m
zZ@wCD=r~z*QjJHebQJWSPBh155eF37n*`K1$RqYddyd<JFZZg`$1X&nnS8SwJ~%e~
zg6?Olh&`cijytKpD3k);sj(XI|GkM&K%EBcbX)=gm6}<Qe}OfJhro5Bs&!Leb5QGY
zZlRG0te1sGvN|u|AI#18(*r&=0GV{mV~$Q$Xz-FNMOjuZiyh04^Vthvqb<|SMGAss
zwBJjJ@J`OZk|~Hy*vFBc&zm^W{oiW_gVzKSxv;cMR$l(AbSC0+Uf#O8CYQ^j)guH;
zIY>_KvcaUhmj}k-epG&hWzxmBFY^MVW3a>g)Fyalgbr@zw*COyRH9RG5wJmWZT_U}
zxPG@mfAJeNE@xj<n<_qH{#R7LvYg|8spmuMcco6AmD-K&AlpLm`13~+0RaK7YPVwg
z0=O-~O@K-khf)%50Sp7JLf2<|v(x`Jncdpbvh=x&W%6}>C3v$4!0^pWxcG=Mf_`Sr
z*wXEjKH$pySH5~b(l_JRw+G#a)oD=cFwQruC%AN1W&+Fk7<SW()hO+!tOk;oyTJRf
zDu4K~kRLZSJ>3hwGX%qXF5#o*_h2TTDfUnK{%hzxds$yM9vv=WKDM$EU4XG}@f&hq
zbkV*aAZMZ$?U(=Y%g{~wzXS840F@)LdAbq0KiUPp%5k7|fCy&$dqCsl56A@sb#-+w
zcU8Q7e5^qsJIzf3HgV+S=QY5vCce|9GPF-@`JAQdXLD$sIy~_3N}4FxK^E$v?3ULo
z)O4E>O#d4Ggb5M9pcuJHIK8qyR(sav5`P`PWW7f!?q(GZprtgk^MG=yTPZpOf?b((
zBwq0UP81toEP*2*FK>?KXUv82ajjG^0S!|Vz{SN4pXWCodIc-s*@$ReaM518CrOVN
zH#q&%4^WsiXlCIPO8iW-208`h2YN~Y{wS|`i}iu?Mx4t^PmDx6Mtdxdsu``BM}SzS
zE?P!)Df_aG;4PJ5kJoN5^7A5xOm0kpGFOyF;ld=YOxfz>2&}0$G=;~Sh;h`ffPzIf
zRp`X+rRB`o(ELNtx>~*bFIo0BVgWpGq0?Ik4W#rTd}mL?D5YNV@x>1f$%8pCx;;zK
z)s6I<we^nZhX8*|k-CO&1lhUDYgRi$b!|h}jeBjC8qm$rsXO~v?ou-WPGcU^)^3;C
zbMP>|sg@hhz_y6)_ofh6OaPdl0B^j|I&0Dhmh0-?k84Csw#xBaS<)J9UqDfwO}~UL
z&Jd6^ST7eZL&W&(5|ZyZl_zEX25UdcGxY;Ksravt#)H7P*_cSa@^DGI<{q85k!zPP
zaEdzdeR?Y5n8kimZ+9BiS3BK0kD%j?T}RXja#>Y=J9upR_+XLgR|5R)b{Xpzo;x=i
z1m8Ifnn>IqEy*{+QL;jG+~@{XtNuT;*w`!k=5O{w#I46?Nwqc~a#kzx$s(0Gd?ny%
zu_AM%O^7GE4NUcAy(JCE(GLE%BiTI2u-WX8PdPIT6%3zSC<3zhE^Oo)?Ef|J$6~+@
z^vD{-(Z8$l=jT=NP&^{dORjQZx1|4r>7!%VM<)3h8s?#W#;Y>O%T>kAa&gJr9XL6@
z^t?W8k0;hMBy7QoVNQ^K6Kn-`EEy17z+1qOCR#mJ|7d+)=<L)q{US-Jl=%t;{G>dW
z|K1bmXUBn{1A~+uk9#?tz@KNfpqr9aD8N7a7ElgnwvhC0w<}A%h_tpbAS{zJhDk~L
zqx5fL@@zamu`yn`nrT;QqV*H|M%PU_&~P-zJy3OejG`ji0kP$G-1Qrz`p?hZX<e>e
zo;})7-JPJZv-hMzY&J8{J3(WL-L7o>7xIPtDD?mt8UR!Qz&swbsu6<FNX<1Wf7L>(
zsXn#0%!Gkp{{%q2Hn_dv;_dO4eDB+4Zx>5Q6GA*Imlll}p5DuUs8>XzWo&8n)mf&{
zbesKkuqtqXl~bKB2au)DZJDp>pCdbJM6R1R42=B*AXk;uPM$vVe^(9|e=|+-Ptz(y
zX`0hg&Lj|G3j|m$?m8|_Kz8|%8tDdy_Qe%KG4u?Z)pd=p<atCE4#@i*ZqIe6PBxkV
zBj`0Cd`z=619(_n|J#Lrz5D__4mV3^R@7eS=y>Ybg>`K3>t~gh>Sv2cugTm44?JZ;
z{bn7bJr16uBmUnhlLXK}Z+AD%{2u5-;uyU2HtJ&_=Fc}f$*WgYPL0HR1SKsWRoj05
zf@XD+wVtB3)R8Vfc-x=>`plf7?h|4=9j&iOJ~UYsWV!koJ|pAC!pfYUXt;2|%(K}*
z7wy)X=bOpzVID%{7rQ#X5BT58n<gkB&3Su2_ycVY8?q7gR06>dx!l}61YMS2zkmOJ
z&dptOQvn)-s$sz(|5K!8<kOfu7*;3yGY-#y8y1y7j-(}LsZNFBEcH~V5cOS?S;p(i
zM{aV|eRBmyJM`D(aZw4We`ZdLk4|@UZ&?%m_p;kx!OIQ|3>=-IDS_#5iTWkL?Cp&K
zc~gqIh})#Ej6H`DpG9WQQ-y~T*@|2>Jj)w<keSj`y=+0_Vjc6IA0_u>BK6HRT>lIA
zM7`|Azty7{K3?3O&!YmDz%50<y9^JInM~jQmClZx7f|xFpg~y+i_tr%>Y~Q08KYVm
z-(a<z{=YpiPLgYIK+5l0B_Hq&n3#fwCY)$6u^?<*vQRv|rK<}S<R6xrA3x@c0leH~
zb^6WQNiZA<NNS|7=iv+ktiYhx<a94b7pgAByfi%6a!rSt8~gfBNz2u67Y|S-UbFu1
zqQ~sr!pDTt2aUSX>lQ&G<PEx7v6}+|xL{Lt@Ttk_JXp^WaLpM#>(OBuwzlG;DUXP1
zMb{;VU@%Il1({h_SwT}Rg@Qk4S!@ss&VMT~IAAuHvN%(1la`kDBQ*5W_5dtf2UH4x
zK?x>5Zm2w1SMW!Tq(B_FozpOnj(;?^$xDYoB||WgfQpI~iS#BE9mVG0;X#Jm{iA8h
ze?zd~(`Z&@_o}41ySuBvf<QllgBeh}+VW-t4r_5Y_ONER20t}p-0ZV`4j(eM;xZ>s
zH>|3%H;T8zaNNRjcUsoCutcaEFLy5;DLrMPdh7uv7De<dj|H&*t-8?@gE0T9Wf!g@
z8$N0=>(7Q6$u?h_r#U@}_7FD)p;rtR%9==2ZR|;MbIo{o5FQW`paA`TIG{3Y8Q2Q0
zM<i1Cxj@%d__2mMT&)1FC>|Q0xv8hUMH|Fsb>4tPkUAGPaL|P&C-V~`le`(z0~8Za
zDE=$tfFN@{<sixEsq*PXy*I$hy+ihf!59Ahhz6TLPXIDgZ!Ym+{27X?vKoy^E6~xW
z1xbv|Dpc(cMp+%77_ZRz6k*0z8dbG`IR=19)Y$KS2O3J?E>8z7EL)K6v{b=bzJFra
z1Jp_lbeUAEuPQ;lm;&;}6X;V~iRu)3=0d8?4~^FTglOIiQH(t4b^rh#1|+nAm~r?7
zKP7mCh~1%)hw#3@)9NL51}Ox_!<s%9IHPzGf~lyoWviMWO<M6!Z_mh+Lg$A5ZiA&O
zWlhDVxL@ZihGTANui6)f?+l!0d3=A~&vaCA%2<#0ToazOIXpObd8%licYVf}(7+fN
z<_lSFjnL!&A9UNhG$G~0Kc0&Gt62lE>HQv)gNP)W59y(!qdNgS3cd<taDIV-W`5yd
zhs+FYE+1H|_rOx{gI3`SYRsBi!7N(G$#+0w;k&LjRvi6<ra}qRJq<3revTvmf-3vX
znRv&QZK&idoq|}SRMfck!1&3`4ytGHf(gnSWG;n=E~Vk-N#G8iIqP;b?)_IJ7C<iT
z!bo*DZ8}e+cUsehd9I{X-@o1P6UQN}Lv(pEvd#Z;KcV-I%;0CywuDv!i#NgC<<Ep$
zOUslU`b>a87sgp)by{>ESCg{;=BWn$qEQ46V0z+DZeas#)}8s6WJCw(j+ME(fUdke
zKxx^&`?**ow|W$q-RsbID(^!XwlW+$T?K4?JikNk!Ys*M%>+hbkZU&tAI26tjIaP2
z;Ra}VP0wBXm%4mfTt5+Ch7nqkZS@CAQ;qTF3J!7p4eGXZfDJ}g$M5frbA4b8v9>1M
zp~jo`hS?I&YJAyFV>Eg^5jH!|+seau7xYGZ??uwbD3NaUhtT#UC<YdqU<Rw0U=|s1
zqRN^a=r9F|DCm1u0)w;FY0w4I2OwC5wAjNJIV)S+%~|7QD-}^4HWu2%#)n-i83uZV
zrJb|$vG(;mizFi@*~7RkF9FZuaA4>@^*L)uDHQi|{S&3|BodcX&B9y_u%M`2*w~vH
zF$PEDZ~f1DZGY3&Hy^pYGa9KYQ<z;|<x>75Sw<l`zGX1c@Ywo9p{~M857EHH1u*F?
zl>aUGq$~H-|7N6lN48?HVSS_qH(+}Hw~8tsGTJqo7YI*)$lCJzX9N?w<lfH=Yk#jz
zQL{R;r`QSBI&rz-_y+Y=#WsWpKX%rTq2Bv@RQ(LV^x^SoOBCG3ag)o2rrS12>UH$E
z^)cq!#+(%KbYuG%^;lL?5*%Pl$Hr;|nJR+$yvO;<J*gr5>WgPF&ST%ay^}^wy1Kex
zk(2lKWVp<7o)mlmzzDlLBO)V@K!rUB+|WV0$sY;~T$y#4Pfttp2B6Rb@Ku0ZtwQZ?
z*V%AY1+Y}!<3Fqqkdc>mc5d!>W*r>E^zZKuS*5}(xK;H-N*sQs{C*@WsQGY7tLh+^
z>%cmN|2qzfwD<aYi+{J`V}jSu-ir{HoM)7!yJQeH3JB3o;7?If)K8ypbMazyxoMA6
zLB=s0hJ~&Sm`|DBHr`#CW}lbR@i}r@&gG$yxyk}1BL!L6AL<om*`U@jUG4~K52k<p
zdf?{r*sh=NJPX{fXsD=i01C0~BylhTxIlWJF0Ph#v&)y3IiUWM(vu<2VSadipXr^J
z_R2B}v8jm|$(}^LQq<TtG$KAUI|({?<ii;#AGXI_@&L)v@@13+ss=WV?zN!)tBfvz
zrj*~`>_*C5XnjVqSL>bNV80@kn)alo8n-QlIrdUAJ{IN;Jasn+sL|wB?%}UhkZ}K%
z)5nMGSlvD3<Lzw(LT1bxn+$+YC%fNXML5m??TFN`UlD?c*tf#BWyVXPF^kIFq@=fC
zr;h?%1exmIMXL{?B_$;zsiN5CU@(89qTF=c1T4hWN|+3r<ADwyD5)+sZ-&+w;v4`=
zl8D_zaZoe#THz)DpMHq8ur38Wcchb~*ca}2pbr#i*R8kqKQ00AnsMwt1!lRUz$bwn
zs5wx)f?j2THzr)kcl4XQJy^hM!2^Ku3T{`MnM4BabqTIVT~F3o5eiZ2fW-oB4e=w{
z1~^Iqf%%?MIHr^lL@{+1<ARDid`1m>t>WohL^=~YL8{Z*do<xnOZi2oBbF~G>r}^O
z?qgr{NonUG70R%fwu+HU2kn#!0(<(ga5CV(00$?_p7-&t^i{Unxh`qFgXz$&_%BI!
z;pKR#^LeNdjNoUBg;7R4JuvwEc_TB?9VpU(8J(qWzB{m9>|nk=h)h5W$cG8&Gp~_2
z%p`+*G#A)5figT2G+Sw-deXtD?`Ah`;RxulD=l8zfyVR}I02&oB~z?~EIzKTubYDS
zQUm$}Fa&xN{GSp4zkYSP+@6lJ1Gg8od}TM#?<L~4%>tfLpDU-!feO6TFP3ywQBQn&
zbt5oZQ?u|DU9r~Q{8z7mzJY<sVvA4BW=1R<XoO;t@!y0imxwH#dT_pcDTIoO8ae|Z
zz1B?DP|rn<?S+(WtHD(|$AIGDSrx}XAJ3~IF_%c2*-EW;la}yUzDFlJQn<TB0_BUo
zc5H$9@tE;+o!yqT9)#(>#{G+a)#c6f-L-SC(-G@xhM@$d*zUXZGhKP}fbjkBCr9b!
zo7_(uHHP=%)7I82Mh`0Ym5pW(x~q?2``80bf~!me3RTRUd_x+9^Hg0lxjdGd-Ko&*
zvbqHMfm3Q7%jRrtB?2NMJCM`>k1PD*MP`LkEU*K3wVCWo%B?3c5~^uCC>+D0FJ3G&
zN%rQ;d(brh(f(-=C?VjH-t}M40R&pRCyG&})*f%(cJJFWivh6l0)TM&(=6DB(Rh0X
z#WsLb55P0csE_t)zdFQaWEj;CYfVOJ?7*Ji;0S0>0hqh#&BbARp9`2M3izm!t4xnc
z!mZeNa};oOx}6CyURlu2dA&cI`=oBpt5T&<OE{`$=*!{?UL(M(&+Om8VYC=K7>&In
zGa9V5SviTamC!Tpcrzrio5m9ZiIWq3AdmgfUtz5cY1?E+HjcU0f6$)0Que2HlQ>@Q
zY&4##51U!sYqv{h{FLs%BX&P7S(A}$YooV0_;ho7Z|K-OJwtHneB9oR_eP&$ms?&9
z8y>d4NneK)o={{9)jQnN7b({(bnt2bFv|`AonrF-x5o31%a~$A=1}FVa`Itk&@+`!
z=%YU9v$x-XAcUKEl|~lGx-wVXR!_l}jBvaK<HHnUV2F+8ShewfZ|-KVq=$irV&?Qc
zXuMNI+(bX~8AqI3Z1pAh$X_jIW6XcC;FZz?V59oV9(bKc#~m``k=o<pr-*6i+|`J4
zIyPM>8n<b=Dd|vGe7f=r?jnQC2A37R>bL^dzS7Ig(U|)wN$Mt<E6Zn_#ZS-UD44~z
z@TjG>=rFTV1HBd9c(&YDGxU{if)QX(TF1cERu72_YJKa}!&|B8e2Dhu+NA{7B0>4-
z8b0@z6+fUwS_lqG7C4A_M7012q0qVqXImnap$b~~D&Jqf5-LH%mida^Y@(=bo1C1S
zLJVMZAY}0KYlYgLj;ch}2GeU)Y=FsprSXD~ctk`rxuf8>dYQa4P-k2vEzLP_ZsIYU
zU_wGh77rzLJrGRK%&euN(cQcNhMM1<9p8(HYyoTb6WC_ZSx;#!-wHsIk8V71hETGF
zttO@O2%gtZX_-9A+J(#prY!Lj&kpj^Y<mf>fALhg)s&BL=9nEnxDb=62<;0|+-iPx
zdvi0c&w~wndMo7-fBXIZP<Jz5(Pgfk%Wm!lZFtXSG*ns_ZB<KS5W<PrjFQ0JCgJaY
zpZ7e0!#w8ICQ$syI+#$-Er;5+fqvwxTb!!MG_CsYg@GKg5?XaiLnceC)I0cAN604n
zhd0Olxxi?{8FVtrt(Qgg4Gjs*SCp$Rr@LaC2!p`q4m$L1AcYN08yQh7HyITNx@DKt
zJoEXg#X{J;2Tj!Uguo1a`io^Za?^#TbktomQ_pP(GkrSpsw+g2waUo1dbhp0<Ztu2
zhuW>Qqa6oQ6fEeizdElQf~+EbzuG)hmiHEpT`w3Ll)CP5;D<eR(va;I^V?8cqh%S=
zps*sh;++a&yWnT4$#6polwo3^qZa+j7%(nEUq$sOm8Dd^q*ak2t$pRwY@x1>8XJ-*
z8_AJp_-=O!DyN7jgh?nF0g(YT42%(weq;ijF>uTM-n8vr{Ssdy5H07#w-7PI$A+fC
z_=<|k6cDyDAJl=fJ$Aj{sC}{@!hiL_3mG#fP##Xv8fDpm*7{mETgq=FGmzHgKHofR
zgj0O7leOGj%hlf(cV((<u@-M{w2G6XV1l<9GT7`82~8`XPjD}1f<4&6&gRo;ndI^m
zTgPEbah{U&mSwYq`-)eab?~iqO6t{k6-;j6UZ4;IY3V3PN=7R!wMVlRI)Ufp-i%sk
znd{||d<Ym;Y8YS7_X7aAM5sFmTP%sCMuRWKh5!_rpIcBP4$eSCV<RND0DY|bI{BN1
zVj&_;E>&a`*fP~zuEk35835Dzsd|^ny_)TUZW_e5c$Lc`r+erGV|In1`B~`OHJuto
z#ci?YchyQ~bqLvp-ne_(hfY`Hk&;wv*74QrjDJ6*qf3oEBKzs(FV%5Z+ta=)Un1|9
z&rrtt2N$IpbF{qHwj7`?sR&v9Nj9T_x9f2>!DsvPg@9bP1xQ3;ulNA_B!Wd7Osi9f
z0e(n9Ny+#9H>JQik^W;%nfw|<Ay8IS4Lt$rqcNy1^tm$+_|@nlpIAa*qt+|LVJoRO
z(TYjMunhtcsO*~?*WKA#72riQ3_#Ixpq90|vSKs_<*AtfU{F62F6q0(g!6z~a;|Uh
zwoF`N(P`7>`sUIi?_>OIwW5;Hp$IqL;nHH*W5H?JOxj^*=?-&nm#jMECmgC)<FBTX
zPGwP$)e=z8-5tc1sJcyylod&}0#up}bf3lUy<_Lbs;u>37Y9eI)a}*jFc5cL`BDw7
z*mC2Mkrnd0T`6|g;Q<arGGIChh60cz6Wq&q=0j^%TxHO`d;P^yKxUcM#$;*ct8={?
z_gm*T@zP>LpyWCQi5dR%K%XD4YKOADa;{QuET;<4YwPc@g~ND7J6ZoY)g)?Wr~Z%7
zANe0@YW|xP{b9CCSKOq1<YIPE{!jz4K4m&>THoZn5+3a7WHsOF6eT#j1d}%IYMO(+
z_k}e$d%QPH9}|u7_EnU`^^&3KWcW{--~B@eKV`P+JX`y<2Ggr{N_zeuac><KWz@Y3
z6C#Qr79t>^l9JLLigY&&rP2%`UBZJPAR?fobaxDmbO}i3&>_+c4MWGb!T0@r-}m1+
z|D59mmvasC%&xW9y6?5tUYlB)((^ExdL>fFT3Ja+DZTpuIAL>upu#Gw<3U(jp)wy>
z6;if$p8yi17pa%gV}xL|;M{rX?Z>=HAXcFTY7x+H{{WImqz7cT;?lm-$VSq0bC-`6
z0T1~AfLx6E(<P0$1`(h@7aZhC#&DPh1KpnHgG=aU?}C9;QVw%goFM;6&8NqH6fwD!
zY(|l3Iw%$TxIne-+U4-nZU)7aAp<XLK3a^x^B^I}BR|^Euyi&vDk7oqEPM`chrP^Z
zo#~AIZPLSYn@@N<B@W&3y~~Ry`uoZYnRo;QLNqlsKpYty>p;#GxC=oXSbrs828_95
zk(KtQwV-MMtlsld>podYl<)b_-rK7P%BFt@U^x}+KRu9u8aZaEF_<CmIHCtRn}dRd
zIUrWlzYX*jh{XBR%EzREJgN@x7=Y;mpi;r5)|~=XTZL<Qre`CFYCsOy2ztZ-0iY&v
zbv}sIAA|I&FOg!{4ehT`eyGuMF_qBnsq>^1_H~sy)no!HnRv<2@Nk1|K0KOA=TZDz
z=a_md?90jc0HqjLZ^UdNNj=Vc3}L68nqMw9NLrwSuS_q?9;h+G_G{!8Odf}IF6d(p
z6`GR;J?406dseVzyO0^fW+)LIqF4ClWp>3nNcWU8ymsx{@^?v!eyR{bm9_FF`QaFD
zE7)RXFXyH>u^zSN^T1jKX9zFezK0J$whoezFltc!1_ImH16YD|<l6@nh(GZ_bAifi
zviQz4&QyzF%$gwnUz@3lndL|7WW7;WSTH|~XYBO%)K4!aN>FhYjbay!r5mFq?^1V3
z9mfmCM_V#o4oo!t^XD9dzy9H?O}P^t%^6{hZXM_E?b=CS4F$quzb9>Fxvg?6^9w;c
z9XxI}t&0rA8yYkM&KpBbzPF6aV1Z>eqR2n}c+-=Ux*!KNZ+Qj;5vxEt39>*>V12Ym
z3<#r&NrwCV6jVXA0;o%-X-5C?h8I*?TNoRua4sE_NF8MMBx2y*?AOIgvaAa7(XwB$
zu?LonaH%9q_~7JID{=Sj9_QXc$mES?z(az|E-R&TS&b?4gm{Wy{<1{LR`NyT-N26=
z`=C*t&i)k;NclKkWl+*s4T#Rd<|YCp5-`I`rFf=YQmM@WLS>Kb7A|-|vu%d632Lu~
z$Hz0b*Un9Ur;^+oha*-%q3)5x?KwwqwQw28tqnCQ8S5x3g&vD`_Z8Z6a0Lf_-(Py@
zTe?s#EC!3u*peti?x$GOP`oQY&TS(gmU4>P)5baR$t02*4Ea58sxsM)&A@&AvfmAK
za<pOIVitEdA{$m5R$GV-)a2y8GCd5ueB0Zb)9Ck|^o)$v0a;e07@^QSid`8m1Z(<I
zx#^vQ{A)!b3Qr4>j>5u1RkycYaeRYeD;R2$v|%i6rxphb(kE#`d<z01W3;ua8z_#g
z*<7RY$a9~ng9`#>C6~_vQY<z<ES;rU*Taj+ztol^R!hIK%p6^I3qwsAu(A=5YO&XV
zO_&8N-p=l>DggBW<-IsNMj>JZoLO=q=iC>dH-O_l)Y|~*q5*9n;g(2Gzr-YnZV)po
zBWB+$|C~_o`GP<?UAYVtRiwC4_jpAlHa#56k=(P6-@tXloRAQYuz$;WBeK1hy0%R6
zc>9N1r0{CA(@M0}Kr24dW<<*;v8vn>eFe+k+yVEt^}BD@W&+Uhdlq_6&&Q_?jM^}8
zml>yszl~T_240uG<djYzaWoX**6t@FA_trS0RahexJjfa_nC!_a@-Z10Yja1=L4Cx
z9a&-CTG1FgQMWf?;olHW#BHlP#5T`W6(@fPlXLAMC#Ddx_YY(gts=@zBOdQ93Zy6A
zPp{^Y*<Q(+$oW`+eenk~sz3?zA><%kH)a_*xUy%&fPsu_?@fE!JC${8MFu~IbA(?E
zvaMtqs^s3u7B74-%WBM@hFv)@%+$URI<|_aGb~V=ac8whbTFr~G`l@dk=FCL7^RcZ
ziDk-{k6Oi*B5z8Q3N*5>YF-E;+4!cZ($+cWa1>ZU|AaHEnBoe;V|YP!skxk94CyE6
zvmxvE&Syv4Hshy*7U}E!EbAS@3qQW`mkoI;Iiu@LKR1y{sGB8Z2*J$vYKVz0*6-yf
z*1nSDS!XSOH+t_ul7<ie>Zx*6mh#9$qB_MZQNW1s^&%FHxc`_W=N4r(j98dox9xyj
zbv${lL%Ln<jSKin=ykR=%5MQyq(fTLVqjZ=uQthy9TEK_;}f}m`1fxNvPtT93E;VJ
z+e~%{3teU@t`-wR?KT3Z<}%Ay88HPh($muq+J^1{W$czqjUi#mp6wK9hP1w`x1#Ec
z3{(;|DqO>HlHNF<UAu#K)vp|W@c3rUZm?!BMD+*fUthh7pL;xK{`VRqNoLm!$3{gZ
zv|{x??_h+7Nn+Z73i$u?4z;vq{2U4oV=hVjN~wbnK6-$8g~0yTUtT&w{{PQ~*CTFW
zc#*a&s5r?)0su5`SM>nKqx-ZB-ac=-<7j;KK!eu}7tB9QQsa)V;TbFg`y$-*b7mdS
zD}0CZ(}%IEa)9c{dYNe=oO9>j_umEmQhTQ3l;G*^8Z+DrV~HAEs4Ld~rUtP#c<_y7
z<av<5cs#O)<qdLzqgqx5Y<U>ypA9+xXJE}w{@{4Y07Uz<j)Y4&K;X7)^7%j+LHbL_
zbo$Qv+f3gcQbg4W=XHzZ4}T9jacGt#%u<lUo_TQls&RZD`C*Ktd-JA?7OUi&Kbxlt
z%F-08rAK&}C^lpum7n5O46x(8SO?EE6++sT+=&?tC%e`$*2n*Pj(6E}Z_32;SMBci
z(kSi@9PX>QkO;rHYZ;jUXS^h)so62;Ey>VM^`n-5Lf!lftSPGiaDnJ}YEvPLWE9In
zPvwn8#;8|BnA2%rj#t>lIx$UZnLA=#IifJ;E`@;gWOr_6lGxgOZKGC4AJcOnUY=T-
zQTP}c7;z<ZUaNjy#27Wd6Eu_bWO*vZ#&4n2J|yWYtAvYRk`FqXL3f<1_Ps2C)MM1p
zzeQn=?cz<6<!TGL1aL|7)4gKJl=UCp+=h8v4K2wnyTKIle<}@-Dd4?zEU8U+cz7T5
zUPa7?y5DE?pL!Y<Mv!m8o;j%PFt$;cxU*R-BVae_k^C}g2}c}PDs&(uV@n8ytq^sB
z>NPYdArdB}K7e2A^^eKlT6*>J<wrRNNyt%6gwD#l1BFH1zVbe&p~XI#UtsX$BFx~H
zC3o!O+meuRl$wcCwH^z%&Gj5EtZBUBiS>@o&Rs_mB9n^U&bE5K_}hH(>P!AFK6I|U
z|DC-fLb+qFSi_=KyVZK}3-8ZqoNk?|$LBT*UtNXFHKN02!W>HfsSVUUCEqQ3EEeKT
z8VhfZ+6xgr(Xug1cHktX#3H|qi^;0geRoSk>HT}CFeN4BgPw#hPWrWzs!t5fG*Nc;
zcE&}p1)D{k6xtG^Q6tv(_2v~YY$NX2P9X`S=BkZbZM9n_tSuUxCQ5f^N=$V;J?p5S
zKY7y77SRv|hSYE0xVFif+p;?M<&UhBzh3!=SS%lJtwx>Zg~f%3`F|HoIqmH2=WU99
z|FzkM9G_>sHg&s53=0Ewf_XSOOJ&Y(lU!+keg`Zh$;>RqA8|*RL^Z2rd$`~k)lnqV
z!4I~6irZJ979)odOCMt^1=j!l%SNA?%<m{CBlQnSsYe6*Oh}76=|GRWBSf!OTPgp}
z7M53(V|7)<NfWU@gYa)++$lo>NW>uL)min`M?{gY8g60eJN2N?9`t9Gm-5RWZ)+Xl
zYJJc>rqX_S_vxzR(vwZ%8&{vrA-HUqy##5XDgRT|(?rC4K-k<<WlT~3`!o4b=4d(6
z<MDO88@F;w`Tt4p_T4QkQ5v)H)UpDX1?~e1l+Buxe|bApKA<SY`i*uotabg{PCO~+
zK4}@rD4I$A)>SiKJ7z|3Fhm`Zj)zBh`NRcdhQcZvHPq%Wi-&|J&kUw84qmod4fx@k
z%f|rMBhBxeWW-GJ?Wnl9iNF7veF{b`O7mr9Q2F3uyEqcs%Dr^7m8+W%<dnW^JHj=G
zJ#gQ}S>-vl1HN|0{agQ(vEOSj5fM!Eg0G2%Jj<?sCCTx~Ek`Y9oOQ^_MRl&IY1Q69
z+RVIWb}x&9l3GU6|LuF9r$9}W{IW!A$|VMaCi_RyuLj{|;SDrUMY1+#Cnj|2p@;Ed
zZcK>a-$nyy@6zr0d&Y9#k#Fm%&PVxutm;c?5;aL6<bqv_)`2}HS5)&Xy&KD+&Z%*k
zmWS@$5A7T8e@7-FlKwo2HrfFXCQsG~R>mnvtANw&A-NwB3-gN|?|^{yPM|FEBQ6#C
zK#?~w=*^YB$NvHp#J_+4r(Z(<*ING9@a6dbPmlTkJ7nnp*2Mp(6JY%RTf@KHe=guU
zkr;4lG3hcS1Y9mM^785ci10ga<bj)tb3IP>>sapoN1kdj#;?_YQV3P(VVyv|#PIMm
zoLdjgK<%7W;K2ad*$`vN*qRUCKNyA4P@N;+0B4ULs&=-y48?)7hF?B^{sxMh0_$)|
zoNwNzH3MnM$AHB+4$AW!Rq6hxoh_}hl}56M4>RZzTwZ|_J!$~q114_F9{u3Xog|QC
zjfyr+_)aaKD6Dtezqt+hZ-)ZBxdd8trv=I2VbqnnEFfxf{FC&*->DmP(<NTPw12_N
z3xiVQKaXHcD_gm3CA5RGc+}E>#A_%i^_hnNYtjJ{jJCOfpK<RZOw2IVwtWG0F9$z!
zHxcnXM7Dd8f2DKt-V7a=UO!5x*N1d{9#PlXvI`|=Fn-W`&#$+W%hpN#p0<WXH1wBy
zfW@ZV=7k<zZFH7}Yh!yEW>$j5>jroUC_vP7O?tqGP}rq)P7_4)KpoV*(l#SgQ$`w^
z7Dn{ZD^Z04glv`_o+UTY=m$NV96TM*I)tn!(f3Xijl%oU^^UR$554O-UZ*vJ22oxN
zpP<D1-_k%N5I8yytMJuQPVH(gSgjg=Aiz1s65_i`-RdW0xC)u7h?<=*;Ao~}3z23P
z2y4}`-2CqXr@#<@94mEWz=b-XzBdxA+!SzV81u6V3hGW(JF}wq4-Z{HKuVkGp-qI_
z@X>Z<lPYh#LiFh=&{R*M>u;QSOYKlj%wwg}lA$4tR6LsKXU`rCK7sl4t`tNPq<HAP
z^X1bkND(H8*#7Xls~x$C8fNqyv8GAO=vZ-J;Ac?LgO&Cr*}o$qAzfV{nZ3Cbu;iT>
zpy?M=uuK(6d%xG8w1=jB!C}Y!fx=Pz$0ckRR%MR3m9u)FLXYF2MoVtZ?2+ptA|la;
z<kz!?bF3Orc9nae?mReQX?!WA6z*$9?Da`4t*xuz+zXTABq-4hD$G3>6r7BuNmLJR
z5H*NWGX6QQVy#dB=<aE^i9wXA@xqwp@l1#a3T@629-g58mo3DDQ(S1G4B1&RlY==`
zm9T&(WXf0wwd?;mKP*GO`#vVy>)+hoh8`HQ*&C}lVknF{?WZqEOe{dnD9%~ej`clt
zbqs}4<Z;RDhb?dWr!H6i*|dp7zC+uJZ_!jYKLrbog@tYOtRuVMZg=O48|?ayyJJvP
zSS%y(u4td;u1~<F*b`^Zpv$tbc6QX#fLN-pLe5ekfRT=6iZQLp3}F%ZRfDq;8sOgb
z%b=lt7Y5-PuY*=|TWyvVE@3;hxo8P3>AgYS{AW~ZofiE?ak}}kQbPv<PH<%Tdevb-
zrsK(<K=e8G!SjX&#;Mx3`^5`YXF(Ma@llyCa@94RKlgi?$PcAy{E!O`HZq29e-wj`
zZI>uKYTtY}VbRWzCJY{4_hH?4T)7gi28<lN26wxkWMpKyAmTa>&)xEhF`Fuwsyf`x
z43CV@Kkp`mH#K;Is;Evya|PBDGOT;3bdm|D)omPE^ok8i{YA)sW6rx}C(Fubr#rXs
zN{73MY{HhhkUnAOiPX@_M~W%Va?Q9D?=Og*%4|acV_mbs38)2yj(|HH=xnYk*%hC=
z4-%MG8(nqk%?qT3c5dgw0JtM68(fN(3ktitflHQdI%5((mX#kYT+&yAIh9i^_sfgu
zxvK}vT0a#DlO8FvNhPY#&tyTbnzgseR~<|p1PV84g>^wX_YrJiB1&FkdFIV5NHXk`
z$VVFwbBxPYE9;ibH_F6_Ov+Y|STiifQ)o<!-u>&sYb5O-W_;%;Che4-cmX&FS$wMY
z@~1faV@NKeFzmQKv5fxcVa#GkiIRAmeC+v)=*w+_{-}D^KGxh3HM?|mSq0C!4L1}y
zpB@Wzwi)`7beun<Yw5Px@aseaHZ5A|gW7{S6?P6rgwntET#y;!%^<2*W)$FFHUaWC
zIN$yHtP<h@?xz__lMX9UCw(IoyT38WbBj3JN$DHyZ3?B0d?Qjnzt#9amOk{%MP6p+
zhk~cJLKa;;`4=bKbmV*qDU3Q5LqX)wM8wIdtYHk?C@T2aVa}wM%_F0A#?_rh?Vp0l
z@v$iJJ}?4qFd*_n7Z!ipT0ZU^uZH-Brun<K=4jx!FW<Jhp9<!6d9(JU)V5dOCQ>uq
zgYAM)dH_0!)0t&U*qBnKl{Pr5Q&P|`xf+k?^PFzH>DC}JcuyPzcn)?Jx92&$jN05~
zLW78+nc7@yLSy{LkGx7KUuEP+q<&7p>UiRhtnQQjTt~G}{3H3Qv$mxEzB#7NW@nDn
zFUg!Xl`V6NEB@tWU~oy6={DYK;82K?6}U~Qp{A4sllbJ$Ib2DQN+T^mNWLv$u0Gu(
z=1(JSOOPT2QRHqFP1wGZyXu+OCH+^DvoMKVyX~JM<}E~#dmg}cPeoOTqVq;@dWLjp
z6h-wN>4E<62O(|$>=E)lCac#wZ)_}T=`kFqkhYF;BB{_<SlEK#32$uR=D-5)u@p*&
zzVh)Fr1*!T)fBJQV0^iWSIEE)ajHp{6i2`hnC7R#BA-fIXjPZ<$mH&j#{W!o9~cm#
zfZOyN5-)6PUG*=?4NzI7l)89<_Qm&RVEa-8?M@_kgr<4w@w+DH<RIX3vcA`GELIw+
zDDwY0{?v`BIPmK7^jt+I*Ag@SqbSImpu_hFgf0Z72QC7ki|QsrY9nLR33)Zqn<BG5
zzNuPuSj0o{Qc5>&XU)lD`(PL^gabaIK;GM}KGPHb0I2zve}F;M5fGH5*CXi7%nbDE
z>WI(@r_JU$UeI<|s7M670yWLZqIwuW&}`w|h9G+??PlD<Km)G#J2}OVmG5QNdoJy^
z10BO_U}RT=%hCrvjBxrs3uLbI)KlrI^|qN(4w`)3r3+Ti)VlFfb;ssr^gMb*Ln1(3
z!!q~Ll-stiOh-G?5y-b-b!ykrL*Do*!wtc8dD)S4p*aWTgHf5h2UlPHh#ZVrMbG`E
zfe`PSO#0#i$~&8*^!-I62b5sw%#WT76+v%p1{!bkz!ysta}`|jOO&jT?Azz=NUn_Z
zNJ{>c!20O)=4fqGK{`G<P0g;um-f!<>bZUddt-AKe77S0$pRISp7$^W2wh4TxK^sz
z?Q$B8A5@*6B$loHc{>*0qZ%yR**h}SJ@G)mI7eTnQsV7P=w%Fgl4Er$+b4wKApcu~
z=-J-R?&?@Uz6PaGh61HW_63V}GWqlDmf)$_$O27`JPrSly1tY%|DhsRjx+Xi8gXCW
zqa?elQNE!*uVcXwMK0QKqo(>VGJiT*aqiD4qR)b-jy>C~eQ~vr0292$)Sy|IH*>ad
zQ(m5^fb8kyJ>Qp(^H}|zTlHY%XIHs3{Rm+ya-y<%v6VWlRjFZ<r<?1SoNS@Xs75O-
zSzPS+f=V^iXn%JwtJ80}ta?*POX-_UAnsKU{6v8~$XVCDPx3<x8zgq0=!xVVs>)tW
zrx655i&u;k5M{`^x308)#0INl%^oc$4CR%t-YA8kkFx~0j&m6IwtMR>dJ-Y?-MTaW
zP_$yFA9oPx%FeFaYL6^#uGL(taNfgzD@p)Xl>OFC{Gnw=))DX<)x4sn;#S38Y_BH5
z{;JH|=cc1#s&=iLo9%r5<!=WoeIMAieDN$?pTDRmc_Bv@E<rEA-#(Xn>Zyr{h`6K#
z13&w~NctlMua=ZWH0q=Z|Ee)10>D}uwEx&umRgTkpB*k4z)I{hgEP_q0Lpo(7GBj-
zI<&Ba9;#fQXb(~EYcZOiU!EEhF&iE*=KpZ|({5*H!G=3mL#puacxgX&bHR{Tf!+r_
zlZf!R1SmQ29F{&aHv)&=r3|o_JWywp>jwVvB#J4nG*d!xKfmLuE8XOU3wB3RDDzCI
zxZ(#vamEU3MyH8ZA$RmdOqrY2IKE?$dG@yR*CFtrhmtYUuw=rqcVcq|g%dc>TlCQ&
z`P3WBcxl1BCy|kBsW-wxrJ_LC)#2-P$O*NYYLOc`l*d$i-0e_%(HsJ1ir(p@SUwhw
zq`dXu-lHskM4e2Cu00po$>ofBy}a@q*jDPgt)Ohsr^Ko>((4wIE3Ls+TuIvZx$=o<
z!_u0y$6$7UzSWHt>!F-C`_mOga@m>@17KsbKqX@p&JF(il(IvWgkK?LJ2yXSbEQC&
zswzFk7Ou}44pN21044A+i`B1BWwKsOXZT1<lb+6V9QPud_r|B34}ZGNj2F6x_yxHu
z<K7E<u)K-*uE)Q;qrr?!RtSHFg&iImn<+2d#$~!h(LGuv#wvfN!SRrF(7E?Q(~6X5
z$MGsI;bjMw-Ae}A>6-@yuX=QGbWYL3zQoIsJ?qBjnz#Jio;H5TH_#+(ZLwG^k9T`B
z{i>GZ#GPDTx?fXKSc8hEKp-YGCD{2aamSLoEa1zKWR!lIJ()zJogA<@F++H&^%+3`
z*C^K3*#+qvR7%+N#N~rY_Bh<~5zd*e`0Kne;3b);207^`6j<C(tzcQr^t-TVyhHRT
zOXK_9svP;R3G4O+6hA7Ib`L0&HvK932b^iM^uk(&b0{u_Qt7TI*z6cNsUe={>Htou
zuO~fWu38}tS{q2#`svOdmIfa$8!px<P@gMQlGiDHjqnthTO-mH2iPC|SV+Q^JJyZI
zIt!wy&{bucJgZR36~9Dp2>;vP$BYIt>w5;n4Lk*8i8RRr%*{!KH)w$LtZeVPEyw1v
zFhATl*V)~I0caCJN;ILJt*uSEKY2^0w?|l(=*st9^AYgd8YVs=u@LAmpo=yvLyzmV
z-YW9>ZY993oT$}KqWMh(as4p_I|(Mw{Uwu|uvGUWC-wPr3krQ%4nMt&&t6{ai2^lN
zjJ4c_2;ESM;~D?ri(DHUN1lzqZ2s`LJorRK|L&Pvb{@RGn>K8^ZW)WfdQ47$kg}g_
z&{V@B{%YrLf<?w)b%!rn8ZR!c1{aVcGo_E*W1R9eH3?f<I!r1c)I#=*#w%a-u)&p~
ziITc7YWbgqi7DP#okFc1Q`K4S4nhlyOYpU}M7Qc+4DgvE!{hm=88ya+oC?-e&BgZ>
z-@t60_+0;O0d($|$OE+SRWMG?16cS!m(jz)hs@upg*^2`fDz3w&C4?>F=5q7#;Tsu
zSx`3@$%kP1g0b@CsG(uwe3+UUzDQX1vTb6WO8`%2?g04YQ~$y2>MAFd^`s=cpnibd
zt=C31D!F49&&xVGiLT5J!!6(vcj+(p@h=r&Q!YZIa-{PM@j~w3y_4iAPx@R)Kqt3+
zee0dC`bfCJqI771<y7$`qmrVNZD0o87Q}=4y&X<Em@!*{gws?~3Mk@RkyA+@JEBQL
z;8$NCM}9SwlY5gw_SHkVsBJA@kaI&JyXC?}MZd-~S@~EfQ1;%tqAc~-pLAKG_8;@D
z1`XuYG%dwvhUoR%vm~3KJ+J2E(mG<W%JjTHV9CF)dwyPYp+PqaohhHBmB#cFa)Ew=
zBt&oK>k1l&qw9hne$5qqeZog2_VL`rSBr@vNcHe1c_!{vIOUPwWlZFc5Rl9-GnJiZ
z{SN+O1Su&)+2ZzPp9?mTuAu`1T#k*e)_EPIvZJ*vbSnz}j+ggu)zvO+n%2~l@jbpR
ze`HnXsa*zxaKl20EX#7#=ln=HKK$UXFw-w6dL|}DBW><le$PhncgJv*p|s5Jp2)!k
z6$h!$RYSR|?*#lyYd#$Oc$>VC)0p*2@mjU{(Gf+6`r#-Sx=F@@>hq18{pV|z3uEMX
zdpw#haZPQ^DSm=Wmjao@4?-vSJqvc`!vhO$ruzo~|4cYl>K43fb1)lLOJyNTQH~<`
zj;#MXv!B>|qNfNQR|co2<s34Fj`b=Q<-gl&xgU}vKPN$yuw*%A&-qd;;TQp?9JY4O
zG*he&9uYe?DY-Pj!jJ6hyB0ZcP32dsME=&TyU6rA0!TbqzScK?NC5LFoe!yV&ent^
zQ>z?@t@QYR+dZS6@f1Xdv*lJxIIfy?o@0+Y3AUu_7jyfIRS$yX8LL>%eGON2SeTn*
z84$DyRh1#6k5cuPtBz`d{pH}4$bp%>UJm}oQ*#=lVKu=beMx*80Oc&z_;GqB#~&Jz
zpN>li2j5M~afDraFhvN(@bBBUlv$}3xCWaVWsUe;^e3cl=odbH7|3Ed{Te8X2M^z<
zzjF%yd;eb9s+9Ne+GzHr*4_AN^iBBDuGh$BNyX%O>%{Tufyfu{sY@O*>DLdMRZf^H
z@GvJiHoSP5KEJq9;<}4Qd;7u;7vcOBdJ}`e6F^vy1ccB?qJIbj4Ugh;SJx`n9s~?x
zk`5iBXHvtC;Zh!iC9}7uiA8I}?6wNIX=zdCuD8(BG3evf4%s1E1{G*cBuO+^y_UZe
zF^QY{<!2ov(rrM7NOayDoLfR?ZSi*ACRdKk7j<lcyYAr<P?0>EQ19fn8t$z=aDA|K
zUJ6G5{@XNaJ<aeh^Pt_LRlp#L6joxGS>R$47dkMAOxGu`b!a@Q%5qgbik52$4pNjF
zDv>=O+B~_PbwQ=LyxT_E|Jl={gR{5X2N!9-_h?~$DOX>r@O<J#cWS({7E4B}+l`k)
z7I9t)OSBx#Q}jHNhd_?3v5n$410whY)MvOS2CIixP@BK59(F75t+cpLq`k)L-<du(
z;k;5w16e4-s*b<9s!?8;ezhiZc$;k)dv?m1*L!rlm1(ed=$#%@4Mz;@>>^WNden|0
zqK<jsDG=nV8>&Gr-NzBAX|~eXyei`8XT98~pk0ndfbysVY^la&Gq)g?g%rwipN`Io
z@DH(Am`&wJDEr_A?e6EeNDqAb`NjX2guP?pwA*Ya0^L|ukA4$kO|;q)8yc0_doj%b
zSiR;xKjL1_6|GV8%ZaGMpbztrZYMon+m(?~9MPNNHy>jju>jbD#;-bbNT@2ov&Jb)
zpR7+3$En)d!^&I3dJ1xCbuHf6aX9x0p{Elus3qm<8=tTlnwsK|c(ft2fy};-OVAfG
zg0Jo(>IEb^8Kv)uOg&k(qSANKMmeuXA}xnAqX-ya%sTf<+OJG*&UYR&Gt?vC%$!0G
z3mm&|)(!S-A39Ra+i=5IBiCgB1Mm^5GVhnC;!OJV9?}ez3+=Ibl<&%eKVt9EtE%u(
zPnFv`l3K`q5S=jf7JK_(cQ6S1L^hP-+@ok`E!Zw(dn$M;rvAW-$)to<_<f@Dt-L0F
zZkvo*4>!iAj5@5$2I?ETdZ7uhQr(rKqpGvems}anY?u2}4fk4&SL096vGZ0P{>YdN
z0GlAQze$+Ik;y!H(gR|)MQ$E%493=+^ezT+JGxn2sS00d3Uers@bRs}5Cx0+0&0#v
zJ{@XvzTL5AJzSa{@Ae0sJ2jU5iJ-`pnvQbIL6ft>NxjY`P5fa3a>W_zWTh-*3Iaiz
z(#Ra!zF?QV*&GtTe`tbU&DMVn`VmF34E4(+2-kEnx|ifaTlc>Vf++x5PAV=^pZPiA
zunf}Nz;?o54Kt;rVRG4lECWx?kAvEa_L_hk2|zJ)(r$a>cku}D^YYkp(=<08XTusB
z-Q3k{gfHA>D|<X_(EJ|dwt=gD{-{bF>zp{OD>rw9w-@n?A2Y*w$P@xH2+7azYEyr*
z{ufblCdYAHK~<jmtxGRX8lOm{+vJS<f%!^Cck$uTN^GuQ_zBg_68G2PT@QEZ&<%GR
z6~5fO$Cr}8B7N}7X}FM^ZGnHpl6|G56PJ;JUP-&M(#^hgO(i70%%L*)v9pa6R~@uz
zH*#cz;MMCZEgC2(K|}3`-42hs?>Ml081F1sG}u}msM~tNXSn1&UHt2OKxG?PkLLu|
zC8Wn;RkJLQYqJK|EHJ6-I+j|DiW&rY$a&I+Z5%BS3YEC{+O6*oCWIL?_KEEhuLSsZ
z8Ua4rTB4Z!c0H>^?Meh1$N-D*i56U<lOC}5=wU2_)SFg*P+zJTy4DzQN<#gwt)k~n
zb~Vk*3Iz+2yZXtnP1LkH!7DKbCa!!r=BEsUMMIo4nWpr8GX{!@GbOJKUthoe!0H^m
zTDp=>!E@AB8Jf0q60eyNQGKz$iBOimKfxeTUoWixQ6umnC2!hoO3&nr)4JE4gyhO~
z%~R!2Bh8AGMx^l6Pi>F7xPofvT<_?%|Mj-`)FVXQ+JwL#v8es8g9ogFxrIqj<PYKb
z{2qJh)2EF36Cs%$wtwBHX6?*X2W*2b?90YI4l9MV4k|20@CFfc1}cdP<Q^XgF9jcC
zp@kOi^>KcF3Ep^2fL8`{5JyE4QDl0o(?gCS1(}DRUZ2%ECsvhODzE(@imi|d1t_ba
zY4_T3ELVyiTwb?Cf3xdce@6Js$hU20$dJ<WnKbE8O7^PnHDKP!_)C#tx1ydia{wQi
z*dpY+_B=WoR%af#dJT8w@562eK}B7kV+(SkWabk>v)=w|WtMmJW{bXN*(?Q*LuW0S
zq}&@g`P1cMVRJlYcBj{+H<E-h8}uLS%c#5;_0oUj8rX_o-=RsPNpd+A2AlP%bRy?!
z_UF3UnZ&FkgMm?<mh8_3Ux=E!9@!=A`NyM<4+a^XgVjph&>u88FNdGh1Xdk;%ya7-
zAS(nG{duo)ofU`c%?(@VAZiQX&~3re&m7F|ihJw}W9FwDFX_vK23uo(n$&V{jG_8@
z6JV7{a;*cK;n7svRsU<bV)67=(hB{FIu%4Z+ds`WJ^A7k^!n1iccojpX=<Fsau4@}
zbLP*~R=48RgkI8d#%T{(l+*ffr-9;NXb{9mMW>VPdIb1C9#p6`l=Bu_Pb8_HqT2ek
zuZFDVB}p4tv+cd~G{@9hRxal2W%>jJ$h3E<)PBCf6<C01TyhvedbsBr?tZy=dIeJd
zxC7Pp@iCvL?xpb_nk&L<<BbW0t(7;^Et)?fWe_10>u7RSDN&e_EGteoHkRF)HmsoG
z6NezHDSf6tLNTF#k*<@54=+fO)Bt61?v7=C(DQrRQvL<F79Z8}?r0c9T{V<N&E$;L
z^DMJd-J*A*_(l8;9le~xN4|KH^v~<j97nDm6pCZ<3cc5PZAFA!4g=$?ze05q`ad$?
z%MrG|=Z{YiaWanYBqOvY8I4!q>mBfba;$rXN3LS{@AnJ6iL+j}sSI*@Y~SnFtCZ2%
zr|c^lQqyLB2z(YpuIm%1j@@2l->!T6szJF;%XQ_AsdYkAw>r}D_g|aX6yxbsfQN>?
zWWk|Q5Z0o@t7v3)q*2<VU#@#8y}(*27j)BJ#>a|I_vK@k3!eHo_CgC$k*+nLzuieU
zGFtscOj&I|QNiq_<6x9Vy`m(MF|uf`RDC#-#pwh>P;zG)gAZBzr&Dh-6izIJ&YQA7
z=cJcfax}}#w60O`;CHKw);R6aizzb7zFafxwtEpV5$T`%?r`O6DiS+~AOTQ$Fj5}W
zwwpIuCC{pOromVa4!)aM2gfNy$%TSrF~}NEkD%od$&1<}KX!-`Nez}KLt8cq?s>{j
zE+wbp=2Q-_c_55acMuU$RrY1uG5S)zF$lM;A7xzwlve-DscJgrDBf;om!?wl?eNXp
zDj%D(Yj3ud^349I)BY*NO!gWTAs}J#x?b4yafMYF)QVRMEBo80e1+B#E|k;aCxkMg
zRZ({~4v&tWGx2Hx;|Gg<FK0`!2kE)IOR#1$o_Jv05}9AM9WmlS|GaXUKX*(~`DLR_
zZ>MHKcZOrN6MXMi`|+l!!Kx1%p_x|VZZEGj;`Vykl9~6zb&;%--W6Dp0M&C~t93#$
zq2JWmnkRqB`W&stbb$|wS=dGEc2sVYhl}3vQc)nYA;!Ue-saw#sR6@?31byYmxX#^
zi$ObIWykOK{u-|Bm<n3`co^r1^W1sv!+vz9-b4YFB>K=l(F}Nc0t-rFx-YS?g>?eh
z$+<!9`>Wm2C6|t+TDDb|GPue)G~EUm5a(JYaoO90gcM{KFehM(GiEjv$i>P7Ha&c3
zld-a{U4ov8wH~@ZVxYR=nv=?mdwaQvP`KX)KKxLRczN@j6ku$Nvx2EhB&i?EEvAXy
zNFUj(E0NfDz2tsgt6pBjTg6must)K~R&(?|Ik2eaYL=#Y&_qF5kJXT9<o#ID^reZt
zlRE5};<sVvqs~er^`|)-GY-DAllD=kiY2Zvpbc48r$Jfif&oGSXH?Zbkq&t!-_cJ1
z4ODiI(K1tSvK2FLAG?Aq>PmARU+n40i{5EHKS@U`g2xxVzovwq24BDN?a#)Bep1D&
zDcOh=$JB7hEm<7T>f_&y+b_dJv9A5yC<yz0&`ez;N*iW$FU|NOVaanvTPJuyIB(3<
zsxnS~Bo))e>hvf7T$I&VYn*{bl3MnAPL72Ig^yqN50We6-NcLVq?qW9H02)87~G)a
zuJ>xE^k`z7Yu!SZl$7nCwX9`oIWOP9{d-@Um8EfP5|YcbhY^$Dcm?_Lxsh%}1Xc3Y
zj$ePzdCKl?_?KjMNslT+xF<d1ewd~x8LMJVTq)h#2pO8xMb>vSu8!=izgq&9_)CpH
zhY;V+pv9n9V`yS89=%7@tr*2}|1Pm2Dl=gr@yMQ}pXTTE?zr>pvq6!gIPXvu-R^Ez
zzAT~CQx9mXtf<a0<nPX{Q4;btXX2(&`wc9*g4bJZ0yb<pI#F&R_FF2C3>R5bQi|{0
z_wYzSjRtQspDr)2xiFe1sA{y<jb8J-UE5SS<gi$v+x@XmX&*YiJ0R)OT$Z2Nvvo*p
zpsJ!DoEbsG>S)m`C1O*Pmv23lx;*)zeCF%IN#I+^QpnLS^=;JAiH|eMXi>9ILfw<g
zYwa(8`b0xE>b-DgDSk7um^A2Y+6k{-Z)EAwjTQ<s_BkFHY+l!}BW~#U{p}|6=_r>g
z(*$jB>+tB0&5HFumFY7`?bL&o{!|RD5A2`pDpMNm)iCWt5;P9DO!o3)ikwf@R_MM<
zQA4K`6z$!Ur`JQUx6XXu56H^AFDgs7mtAjPHgMWl>`shIRm*lMFcc^d;s~s$a!L_b
zw<8|PO%lpya!odLZ%T0cic?$VjaN5S6T|~)7nDJCWbnuN`B~o>54VQ+;(O>S^5R%~
z^)cQ|e_bhSt>rbL#<-;rm#pFsE0T%$QVF#U9W1A_rLKOZ+?f{KkFWpyJ%NJJk8u81
zS9{kDGe^eCTS;`^``X98ha5XB9hZb~OpXM~RL0rfdT7tiUaY^%|A4f0*)@=!{`msB
zhdkhe2`$2T<FmnKFAsjVjepVf#;V-elbhZkeQm4FH%IL8Yr}pC#np2}#`B2HSr{UU
zh``^mZbh3?E)CWR-Egw4G88Y354brvQ93{RRXUH9?;2S4VBnG4jB{S)>Vf+AIiAWL
zZ{3&a=ya`je-V|K(AfF%ql+eu^GlDAQ>%Q}5?RS-roDd`l0<i8WgZ<hz6$m;@ccGN
zeC_3(e?JaS?$vnry;cK7R7IeZFR{<F@wd0<8%nobw6%xFtUrg)(V5D(_IHJ=pAhcs
z_DsR%aGmgtMEynI?cllC&dgr7SuER|q}gw6om*M{y_;%*csns(LgF!AAzq-yZZ%>3
z#W;e1bje5sFHj*zWcFekNnF;d9}u{-K<O!g^I^OKR~C1_v8zE8P9_Ek2~DFOc2TDK
zV-;HNay!NG<(744i0^*GsP~mDJRc8oOT3i@pXeE_i#pxGlfx%)bzNJA8MbL4)q{U`
z%;q=Kk#gQ{Xv^%eWq*z%lpRjzhd)4CeHbFbckf|vK%FpdW{9bTXf8jGvHc6Kw{wTz
z1^WvKz``eQ_M{e;s==YyUtLTYo_i#C>7kp~59YucyvR;n%rB~=2y}Oz8+i)~N2_%m
z?#aFz7$_}&9@YNnaQK8iTqdK!ygl+}>U(w$<hTTuoYhoZbBO$}$v&^a?;|eGEi(C5
z=YiJLnJ-Yg2yn4^-ea~OM?%%_(iQ?SFjTQU>eglgX|Ua?H0hxqr?Z*VrlxQ!s{d}g
zCxW`Ky*2N|{O<W?za|7>8jvv#e<dAHI{qm<tYP_>2P^w^Q%wZ?H)DYZ)r)}FO&t}}
z4K2fwj^9UIoqAs~Ja9zJt795`(*1L0nE8VJ#hbI5gBRSKL*phVDL7iOvG>16G#8?E
zq*z(9M~>E@pdi!9jQHOX;BJBS?yQ1wk#1;hkX4Bxq6^=w1s*Rj4F76zU*~vNw(=FQ
zBu_A;U%ym)8N!m~H8%4hmy27N`Ap;0X|^7&mdAocZ7Y3v`;Qu9V*gx=zcHjFD!Ll_
z@Y6)e;HP1oV6UHkSYbzy3u+)DYQPGJ)4e)8+=KYi?%f1)&w4(s1^-~~ij#O=i|Mt{
zjLV%B?w9boskqT&Wn)98Qm$3ifE7t9ViR#(`Jv8Gzy$aDko#jVZ7ZE^x250*k&)~N
z2S}J<qTVq%&BDU9YToU^zVl_Cq@h>1bhsKpkFgpCx^+-;u#1tJ>UT6P#tSvbvipPM
zL(n9KQY<llxvA3|j7o8FCf(Y7if*O<Y?Ny+e-i&&3V+l&)}N9(p(6Hy1YTrXU@k|=
zc=@yXv`?Gm)+3}S3)9YELgGJ`hRrdIBg<8*HcpdY#gmh3eDHW<V|u>T^Ke`ol6)n=
zKiWNWzG1OpuoNd36m&y-pTGoaZH-gxOd9MEIXZvGn;u~ow7revD_FAhWbS<I<##!W
zUN7!E_f4yO!%q@z@424dwHLz{MvbKQATwhwhmfEVX+->>c(LJW5F10pmyqF{O=VdD
zvUwSr-VMHAY0TB1e|PkcN@OT)na(?UdtQW1v29CHubzafs&(y5xx=COUdEY=YYlT&
zI&b0cx_YBn1`WI2_FHI^@)juKrDT$4s-VNYF4rys0&5y&k!+bi7CWu1o!Q}?;8<Ye
zuSmz4-f9x5bS9EMGB2ylUM0_oMazmn1^Kz&X2~KBwztZg)&vx|e|#w7ZP_^%66feD
zV}4|@NffB<9DFy#U?PD=t1pN0b#K=sD}>n2*E=eAqdE1D47fD9_tPH>>z<P-0n||2
z-|49$Zti#8AH{x|8gXvp<NQ~beR3DKET+97|Kur0m@H4~c2W7ewEB>qvzmvKAJwA}
z#S8Q1i}37iygSQFvcC^zUd#=$@{!za2NP4s!Iqqu2?<a{K71sVp7wyq{@HWiUro<2
zVyY{PHqI(m{w(Qi@*P!Hek-)*5;AZPLrJ%H@j1j*wkh-@qH*iEHMB@cL$zV8>3jIE
zIEpd8$xzq4u(&o;cc>eG#yVDt;}wmW4;sm)f^kp@2_<Zf20Mqu&5!80|KKtc4f+9>
zmVQl>3H9~#C)i)aw#uH|-qmrT7O-wDo-<)$Ub9?YUG*-Ber{~XcN0hav4T`e7Q>2p
zT4q~7bXc|${W7}kdU+gwH|l=t@S{vQ+SH`9YeE5^6p#1+Oj$Q}P1Z&2nv;AzG8G6m
z$=HK023M!^-J?!>Kx{9_<{SPDowLrp?{v!MMh}5;<OTgy2#}L_B|)2@5X{v&uB{H(
z*dNjO0y5F0FzhoyA>Pn$ziiz1w^RIL>L(NuU#6>U8=VYm4HvCebDLbh-gVouQ!=xs
z7(E0IEx<LX7DKT8mFOcJ?Ctwq4MmjdCEX3@#&pV7CR5*ZN;`Hhn(|*8(&kU)=5CcZ
zg%DQo$OJ;;la-Ph_JVzO>i=MmCpVg!#i(6yl5ak~scIv~*)<}YNtA_zjgI}bvTBL=
z=~9e`_r9=rfOON^ruRES*wFMhU!?VPNBd`~^GeSW30+(fg9H+`uNMRqpFiGir-a*c
zJkvPs@-w7!aX+$2Gg>O-;jWX2tAi3K^kC(0x&`bxX{ovRJdf}&nmuaC1Q8QKTj<3p
zD!R(Yn)dn_{w%&zY&{2J#~c-srhfS|GHMFb`4>~&N{S(w6S##&Np9)6D(Ov)L!US_
z5y}8@MHg+8Z+CLrU?;^^GnTiuj*^nue^@l)mK5J@Zh8MoWq@<|g22i4FBE;-DtbC&
zAwTHXbyprkRL~<~{R5u*y6q0~ZJG4YuQh^ld+3#ku2f9ToQ&Bz{m0T$e-rQ1anZUE
z`ul0NaBWZ!9H9TEs+#gSJpX!-DW7fo;jDj7o?U^EP?HItp#3@1YuYHoEW@VVnR%tr
z_}+>Fi9^?ty$|&MubMAa8}uJkFl+F){sN)@@+V<v@ere?&geL24&}|h$>Ervr&ub*
z#VENg?<(5OENi_-$oyMCQ+^U}3yq2O(M|q^9FFAHPxv`D!DXS3q&-+^nHR48&C2rj
ztlw}_O8RL<;b)WJ;A>@DrmvUYq!DCA<0&bn6rd)W3T#Ju+5rE{e*LO4@73J)JQKQA
zxnhQj*xtS=AqKPII9GWsNcM?v{T&W5?9;q_R~p_slqiACk7~~vXNLXq=G9c}?Y5om
zYv)|d<m7^EsT)2$^prE3l~m9snT;MxBZ|5t2W&i44b!(x?Mq1Qt@V*@LE+(~ZTHwq
zlrVwjB*-~{@}$wcE0tXLC;J?Fozc9{%Pjv!#H6xB^pBc0iZZkS1h>`2loA?hj#@`%
zou3ngav(=aUGGm(74^sb7b!Q@`%7)%y}D1IHU6QiV>GMZV@wU5Y?WFzJ_^(7s~0<w
z=uKuBUNjLDuyYC9qY2AJ9l=PrHpW+^sUa;Al2FY8zDAvRse*M1dGjgn=lQ4#Hg&X6
z&c?12pRX^578#ty_>JkiS5WV9{?D=k&xc>WSUk>E&_bzf9$6!oR3}<(kxS-dzl*x?
z3HqXB;2!0HogEm3*7GG87ts1WjuMK1Yc`(JH(QV0f)w^O4Jy3dth4C<!3_xfC+PC5
zkx@_$;K7m5P+n4}==nTLJQ<nyh2>3cExwH}$!*KcCs#||Tk0&*t(?rYgA7S%(wVC7
zbfoVw70bx7fva8Ecnp6actR@RL9=>{`W&)D9=tG}nUesNXtrAMy2A~KTu8!B;~}uS
zyjt4*1I<2BZ9ysNlFc0u&Lgr%&vL`V^k5oP+R-X%U5$sY|L)#uXJBj|9(_b8d+XS>
zYR=St3H|-~gZT)}B+X8|ckYa`qPs#2S>J5wg}<ZtT#drR-=rWy0xL<Tru3HQ*FLG)
zna&2Sx~okst?RB2<))W#ejQXodhnXNY@AK%V*F--pYlel&uN)!LE`wUO9C_k{&B^2
zab0qb6m<19Sd}W;s1hqHYlU|v54GjrD`=;uW0bykwo$77(jgP<Yw(nl)^EOxiTd76
z>)ZPdd%LO5)fiv#@kxa$+#4G^LYIf`0R;SmY*n~z)l^}yg!%tPc5z_OxzN+AGYnG-
zwYH(o7a=zlyL{U-fzqYsHq(K0_0ALC>Z&6(s~bcK(T=G4cm;)pl_}hSZq}^3lri61
zHh#G13%w6N*Qdy%;JYSp-!PnkNVB;3y9~7pFaKo4(TKks%TkN6Na6mUw+uvINZUg6
z>Y7-qoeXV^+j)9LI!ir0R*{I>2pZq?^iQTHvv4R?A)HcC?(J;*9R<3~;qiieA0f)U
zwJtNrfVv@q3B7c-SY|o2xc=cp$O1=HP9HN+!LN{pAx0)$#ZbW1o<pB-qoIU+{B0h>
zUoQENn*xi(_3J-J$A>>}B3Zl~{e$20m24;e^w+e4jWi+_TRm9H%cW@U-NXCx8<Ax;
zlrsXYE|qID_r`^kaLlq5fiAI)PN*)~vS<d)Ev%(&|EwtCh%%h}8TY*fY;OJ#Ri<2O
zegsCqYwwW2#iET!J*D5Kkj<NxgxazKqw*JQZ}+#7Yd9R)ogyFd_ipNMIoZ<ka|Lxq
zYdbhQ=U2aE&Pl9XJrMsNc;|8m1zbaHH5kt?a2%y7@)Fq+d12rR_?@^7P7agGQc)Gl
zcHZdzZDi{3*k7!f?=fl5ZmLA#eB~Iw)gvcX{N+2J)t<d|mJ?+6^)}sq{O$}%ujW;m
zjM@C5W_e@dOGzaO_`@gfSGQZ~?AV`2_}B)xoNti<cG)!8;hi_Y?smNtY8+5IhzONm
zKMf+o|A6w4=|Bb`MCow-j$CF<;X%I1Portlb#tmDqj7w@PS<1s&yARy`JlaYU9(vL
zclWHE+_Ije`MrUS{-pB{Kmjn#Ow1dlL@3Ak$KL{#@lqUTq}YCp{C;mlo@IvRSof1n
zlROfx@O7S{GMnns0o^F<AWRv|fGC9^OfsH5b5g^axVX>qil&MVf&d*OYJN?rWg%`h
zgo#~;e<isGYEfq@l&niwrrxWW$ZaJdeVn#^I2e@_g!vRxJE|!N(;Z3)4IA5@ZSwyc
zW-cIm>$>CW@48#W=GE_*Z{qka%cgg&3<Qh_I1g#jq1OvuzFwH$p6(A@3l3Nch(EE8
zApbE?ad0N=3tQ)RPuRXC@;X^<08nq%lY<<tkEISt-ZRZq(a>LfV(ZCY?yVO;D_42s
z%7$n??FU>uN9)3ZMyw$_E`g}d=rP{d*lqJf3OVbk$3oKGoHK16?u9(G2^Ar6e?)bk
zO^xKyRj-Df@`cfwJ#L$tPN2Zd=OeO>=qD?;<$Zb+z_sbku?Rrjxjcu@^()>yE4Fi4
z`hD|u3thka`gK8O<jslkjqeEw6b1WE084DS{>JfqJWxMuKLVMwR|cvvL}<lEzg3sj
z{-o`LTll`4zfRCSc^Ix9<x#Z#$=K+vSyJy|!W38Q*NtJ{5zo`;vb!TL2=^Bo>H)oe
zfSA2qSdhX;qe&)eV_Sbbjpu*&&g^$toGx(Ly!~2l<I|>`qxQjb5eq{VRrKXFO3&S8
zrJ54ZRKc-20%+IWq5tRXRBj`=6f=uSe?UoIon1u#?!)cDT-NQO!lwiYbG0w(ObMs;
zib7|asD2El-@ly^qgf0OQ&ReE@Ll-_DG6IKJj(rr?0+HlwOM_$QKtj%zE58|Iv!QO
zVUyM#17@~un<NY_y^*sbn||4_sfm1YSuHaJ_KEA;F~TghoD*KuIJ4)6>q!csT4ZEJ
zrX5-j_g8Zq@8>;q;<k)=@kPA0IvFcvcgME4e6c+_MpI?l#&(57PWO}z@IeyHeBRU1
z*ZH~F1n9j>|Bpg<oOfan7m4f9p3Ay?zM%^*Zy!$~IbOsUu-XoGPSi%j4>HQkA!T@z
z3Ksq@`tMNQbA0snGnRU<XWBHL1uT1nc^+Yvgp`Rm4YQx9AIr(}@_GnvIuj1VN;n%D
zgUQ2Mx2!AP0L87_-IE+0Mx}Z(oFF(^Y({bK4RRs(v3yqTyp$D7hUw1w(F^JZ8+&~O
z*T$K->I)vWSqmo?13z7}E#ArnXrab-ix-16ghUTzp(zQ6>yrrh;rV(KACOD*H#jR~
zt8DLlObyrQaP=IzF}b$R*(#l`G94p$*Ne|z!7MLIH_E2h|6W40Z*~zIr7)W;e63M7
zx#n~#1aK6vaX(S|?~VMJ%wGh|h_(58q1~@Hd(`*zwf3<ko0`Z3+uWh#=v3?yf5LPv
zjdsS9kiNe9FLvpn#QWqKD%tbsn{yOZ_a6KY#D2AYX^!4J%-5DQ`hevjG^NdDaU<h8
zT1RQa+_*hh+7P%mjM3*lYxsQ5A*L{q2t1s7qKXX0&7*hscHZY|3f$aM7<sbju<{9y
zo{70dRB=-uQYf32@GSHKE;4{tPx&n|B+zkD_4!-^v_Sv`e`d7Z{!n5bIPBsq;4bQ6
zQ=v$j&(-cEAS)(zPAc#<DThCo+cA%c84-@T`g(8iCe=-+Y=d}Z;PGhE*K;iw(@_)k
zRAxJd6^PqAJAac>%2AV>Dvy<2iAIy+URs|ph<o#OhLYFoS4#F5<4ZIV13eRDHw(93
z-+*50<{t1K5fRCa6`yDXE!zrlB69NgdxoSrot+I0EU^&eykBY}V_KY5v(#!F53d0x
z_IG{l*JN9UMbO2cqU@gnqhV@Vzdm&fv@?(B5^9E_N335QjG9w79Iy(7)J5*#1IL3P
z>-NGuV3S2|ue8l;bfbiNO1YV}xPP(&k*2FV^()5HuAP{cy=5XS!6AuTKKVK7T^Ob^
z=228>7$z0;jNv(nER~oBKINy<kBJxVIyyy0UTzVnCN7lHaOfQ-7S<&Ml_aK?u;mk{
zGRsg6GT<512zl&Xr`F+1);}qwkj=F}W>bg#r*@+et*|k@(Rp6GRbHM_SQ?Jtu`b}U
z-XD<#dg9UQ%69|(m-UVpTe8a$FdjQ>G{cAp6#^PWg#@KNj7nmSNKwiDfj-hKE58{M
zu&{^B)p|toEJEr>?a33H$ASu;;=aSXykVgdn<CtH&z|Fq<QN^-@F(ywa&VM0#rOjs
zdS~;qN4vdF?{+M#pSV;XdcSJ58E+&cZY}}>g-SUZg_#+UFht1kU7S?_g=5Rj6`ug-
zhTcER-gn1Sq7U2N_#P%$XGkd%@x@d!=*@5vCT!wEy%nx`rR!I#jn1!L8ZbjH*AzW6
ziDTEWxcKf60VnRCXSHtt%8B&dEd)4V<w%LvYE1my+wa5NJodXJm9}Jztdj3c%^MeU
z_Wx%C`~&WUzYHXUpM`ZoK)|Zq5W_VSQV=CSpMV|L)9<&Gd174r2*PyKy1M5d04^3i
zC|(g11vcFxT5Y6Ow`@~XRY8%C5^?hXH1!s6QFhP!@M2+rD1s>YC`gA&36dfrjdX`d
zmozM~7=V(JOQ&>qhaw;?-QC??@2t=7`Tk!&qI_WQ-TRy~XXct~<~|pRo6<*aZIic|
zaujl^oGI?nxL^+}-1eHft@digK=m!mFUHHgC9Umm@Vbx11nb~jy2j%B<l6mkDm^_N
zGt*zFmgoz%#Ep<smW_VGIxw2gR-LwsR)}3bTfAdvZX)FV;ztIyAO69xDj&CPXm?W@
z)!n-RV3LD9Hq}?l1U8y$9IX1Wyxu(>VQl3(|BTcvk_!kZqHa`L<-b}xvKBqaZAqZ`
z%Hza=E=sW`bI$iu;~)iEz($jAvHlT6r99L_#*ep2in}aE`A2@Y=>iWt9~&mcDt^Q7
zvwP6DG?ygH8}{~6ENU;r@$507@T{Z=vq&U4<5W+dd-?>*X6w)A)6qxJ*sZ3lZt)Kg
zMY`2QGNe>%ydcRrGW#>9Z6@54KKpEV;8vtoZCnip>&!aS2{F?Bj}K#)m-v@<!j^f6
z&DJ-`Qf65)RFHfGKyHpKqUS=fp5C3wzG&*_HK!`Y47tZk{du5<arMp7%xJ7%B+?z)
zolOPK_pLsJRbGIp@$;85VrIbDD;XL$42zPasS#6zfFOIgDf^6HvrC<#ukB|P5VDUc
z(c)oP)TmMIQ}#Cg+_3v3+wGt8DxX;LgeFIDdS&=}XEAM}vgbn9Q8W8h*v~@4TJwGl
zE#E(W_I&$zIH$WnGLrkaIy_L!#+VrErd?rvf!Uu*^ZvdcF%aX_B_*ghn8+x|6y3*f
zWbsB?o_SgvBB%8+G(~Jc{~_yEd4saJoZOrQDyBALx<Ok%a~O~7|JmlJNBEULFIqA^
zxGluX%l(XtEj`xUoO$UCKPcj@DtE<up=gjq>MOnE4_q}_G#OWT@QdHO%Q(18Jp|n*
zkCS9Y%OpH}^KQ4TniBUa^`|=#L<gpCI#?<0ky&#4Pp6y5s(n8U+Sv$Go<N<)-q%pM
zf4{5YhhMG_WA?B5?xyi|zv<pcb6acc`0~{Hk1-ktaRx-LrMDV7;?bf^^uQ?QG_ynI
z2dv(vU6x~slYx28%5ZO+yZ2&vR4Kj+<qtLHpxcq5ex6B}%@v9~N10j=Z3hJI?yNLJ
z5+g;G^i5U$MV^F5P&l0{ilYx_e>WNJ$>34aduwfJ?q~iD9d+sSagX<N)D6xX1M-^p
zIU`(%n5`z>(!zduM|2l<vLGuqHU?jQi<9y_3V5MKKijCJIhA3czuoPpapJObuW4|Q
z03D(mm*W|dc7Ig0WTt3ydSN90jNK;VT$61t{Nk=PQAkkZ?swF8PwF2xmI8^r&pCrQ
zsr?Mr4{wl`5{N<i2TaF+n(EZ-%C8A2viqbxZ-U?bFMBvZ-?^qaS)GxZNtIIysjByt
ze%q?Zf^ejQHd0elPD0XPvFs4RjgBmh7VqU6m5_4OzBX3sgM)}_Yc(7pyI^3wdq@x*
zY{)~csi?rx*OaNlz_YGz9Os`D|G@oe8;PbKA%T2b%I5y>!B(8#59!hPkytdBS85UZ
z*d=_CSRq(gIOv`zEanT+c<4*fSe`TnA~!R3?^mepIyTIYJNhSBZ7JU-|Ixni49bgZ
z&*pT^>Y)OtkB!cc#$rJ|qW6G`im8w9H9O0<ZfNXPGP1`#{gg$epC8RLPadqYWM&LK
zHbiBXyX<phEwis3rjppWZYm@@@6f49MZbEG^SvtZ;>AiHC|UjN`s1dH&e>SP<L}=3
zI+)3itkgVJ>=)s^h5l2^V3|@lh0koDpGXse-!qcaegH}Im<B2@P*Iwvj)>%JQuQr<
z8za-g<eFHl{u*0J(E~b$$XNR|>V<lx@WVl+83N<%H%^?8^wSeum!dMtj#kq;&d3(c
zLA1?YxNz!>cy<dt@Y}Ly)g<E+cfQFy^?e#{mRO*zASK_hvrD&hR`)l5?WX&cu0)Y#
z`9+%SC^9B_7nO79SS9~3-XtS*MoIqt116z)%Zc;#H!xV@UTHa?uyr>otnZ(!srQj{
zOX!GRGh10*SYH$*QZgWIli;GT{p6U@H*tO-$2eL^7$T-%P$S56(%IiVE%O2&K<UG9
z9}MWKTVFm{dvVY5@bZDOT4zh0uO_)Ue0Mui3snm?e(9_@6unOoudE0tunAx(p#H_4
zO3TPg*Z=1SK4$;At~vSP-e}bg693qV{@km4ye=`YO%V<3`?DbvH*M@}U#%Ateud(}
zUgM*U!TYu=-4!?A({qKGalVDJi_ZeS`h3{(O|r>wi_kZ3R76D#M8+1;qp?``Py;Di
z?>8GH`65rh_Ynq8JK2e7N&CYNRu}8|o>-75)jzx65`ZY&GWC!b67?F^o116V>1>QD
zL*fGTfSTs7%;Kz#g1l-<=Bm=qxcwGe!oM${aa#p{GwC1Lo4f#YaB|W4k7(}vwLiBN
zKWpf?p*^MMgP?3ZmuiU6j78hTVNqedMTf_M6E-F|H&Lpgpjaz@P8|c$gi%7$Xfgiq
z$bW4tB<OYkBc69$L54YrbJAU+g6GuacZ7tl*rcLdCY)AZD?bb3KMAg!dh48-<p4!d
z%2SuzYBPnZ>sLE|Poe)XU~(X?n#_rnB3jqj?ed2FyS=8ZZHuGKZgugKm=Hv$a~lOw
zMO*LPZxRjeCsnpLco^58lTVocSMpNFjZ#+rHS}efk=Op~tj=vX$a=fR&Wtwo+gIP_
z|5a8$V~b@@5R!k{<aL^uo%t1C`V#s9a-ncBKl9{--R{}wi2B;=wU5Y)S$S<PT)ZaN
zDfZvwgr4&91V-hfo+j@uf2yde>f3}WXv=xE`ims>c7s!-mDSw4hJ)j=fRW38f?mTm
zX+E;78{{)$Q+)M`aj?+cJp&G=)}jx2S}eke7~`_S7v@#GJ;s!j8urM51q;JgsOq-R
zzw?PH`2ys0XXju*3q(xIb#wN>inwfwBQcqOlw!l{QM4$EsV>%gM{C&M=Vpw_rZWs;
zUIEx|@x3SYXt~(SezU27poQM3xAh`FCz0=y#B#eIKZd+!65fVtT<tZAg5>)>9}}6b
z(8aNEZc$yI_-ek3&us-D;Th|cV>`Ji2;>pJI*!N%=i@2sqJ}mnQ(^=~47a$T`<&!>
zAzcuv<#1xa2%$;Jo8MPT<B8MIFcJlX#<i@Yfba2hWY2}l)G_Rfw!SGKEJ-FdZV$+6
z1C-QTq5A!qclRFa1=UfQBHP7Kpif+hhhpUlk|)e{+8d^69F~?XmeiY`8%$k7jPY50
z{JlbY#AZ!nwZ}NNNGwlTcH>WL*GJTT1@@i8Rz{>_spTVvNPGVn#6*_port+#|0_0K
z62!(ZrMH0ND$F@z8~^#|8>PSGyEyb7VN%=~<7pc+*`doXZGq^C`{Nm&eH_=q!&^^I
z#DQ345V<*5G;wn`U5R^OI7q0QMq9`5Hw`MwpnuR>by}(NkoXnq%9W_sVmf1a-MC`^
zgvu*EHELlNqi49bvm1}ncFNkcILnW^!D-$#*c<+-o7zy~^=MnypKrUmo~&7tH-HHb
zKiV`<4br~Zr^v$*M}g=rS$$BJT8d?G3H3^I9`*pxZhe)kA|@;TEHz&)Dcd~!EVL=A
zs=Iy{>cQJico$L75TB;X*`#)Ml<}GhfA80??Zw4yT;i0t7Y)DKo&){LW^MC52(|i%
zEryJXl+)?Cte01KnG1_mQvTR9P?|C0PG6R}%9O*y^A0xe7d5CxMn17<uz_qSsAT0~
z!S_PJ`5lu5!aiZm#<8wI`(!C|nPI1-(#$9zdcI3gqjUvxvYiKOl&Xvc?B82W%(OO@
zs)ZFy4qU4qE*CO6iIXCGUs#+V8~+Xph0kgHe~cBbDU?Fn%}NB1hu~OSsywY_aZ$V$
z*?Pu*y!^+1M#TRvSkf%v3SWj)?kV5fTUo>hWhao8Ii$%s$=5nDC>r}}<Qg;8qo5`X
zMMJu*2u{N&AOT&;_-?tCpiXhi680j}=NcML3bdK=zha`}=GGbI@4{b7;`EGfYhxoz
zgJZtBr(N@`Ig&7OhD5r>e!U)2U0hwyer5MeU-^fVDGx3hE^Buz{?0Jvb5u1s@#561
z@8}HyCx<-G`vCsd)-N7)0qS5FvKK3i-x`@Ui<9oVJQxl9QG!g`y%HmAlwz-Pb_=lM
zS5C^l1~!W!i_GedyHgB(#~n^C(T7}z>R&yLki+rDSL7ot-R&C-CFYd<R#p%I-VqXf
z3qa=-wQWAqdYhWlaaMZimDz}txk_1|gBt`YBengF{azFa&~v1xN88-uwv{|~1b3YF
zsnT-4Xqs?ZJz}V6w}f?!jpFNI!1hbXNoj02T-R0dr?|G}X~AKE3D|GW_$r69*18x<
ztT{M!q0s{D?e2M=U**&gD)^b*;q~yt%MPk4`T9GzaF$BjOIUllr~NyHEK45K@N)Xg
z-Bv=8uu1=qauPHY`@3-4iaGJJ%qcl3&%veJEKPrk(ExWkr7uQ5nU_bb8(2`FVt;V8
zn7W~|zp6lL8j?eCX*|w^K)C@|>X#-fxfB0XqZS>%%U)yn`Oxrun7i+ldZLWtHtXs4
zpAYXwUyVPtFuB(9`{&P|U{+L1sJ4-Dx~<Qn$l^HJG=~h6TkUiDxO^14F1&n;8JfE%
zgX1KH4M2RipXL%s)6mFzi!YouUIDWY`@IrH%n-~#aNVx)7bMZRl}<BM^dVmJBK{Mf
z(ycDr9<H|a-y6+h06Lh)o=p4TD2r5FB0IoMAzfbmb<0GanbW;8ln%*dLPDFZ{pnhw
zSa9gzID0)LU%Rb)?9>hY{-Q_pbK&4PHnYa=(_0liUk-p2m8cy8GL%+h?~3}7BXUFj
zm?KzGeq(wsTD5W8#_HWU5rytM?b?9PR6~qfb%EAsw(R^iCWiTWar8CiGwfmW2JJY^
zLRv#<OQBy6CG;?0<2b7Set+U~;A(Bus$%+Yl2g*t<TUWo2ek~^v3i(i^cNhwyPEv!
z)d}yu2C}HP5^L;~I;cgZ+~~%UEM4sGs_5wQipkQ%Ol_;1;tV=C*Li<Tb-H?B0}pY-
z$*#UDcMn-rl*jS(Eo^m!bG2*YO;*DHf+y<M=i~<*DtE}gK;s%>Lqc0EI`CPXRIY?k
zZ<N-HoQ<~RtW4yxFY%M-LhbQDPxO^vR#fHC=z4X5=gz&_rtbvBv3xBDF+<`npv?Bn
z%%$=^P?a!CJV07pal-3v?<6O(W(5d|f5ZDrb5lica!?JwXY~=mHn(vceXtnH&Lc_~
zl8CQ%W;WauZXd?lHuCeSAk2KvH02?M9w$>H<K{Mnmhi1j15DFg2%1vGkqaLxRL*9Z
zxom7y1%*Bg;e)~Q=|SJAJv>e}n!Q_z+2RDZhblKN!UGYMH_nwC0pi`Hq0-Scm0)mE
zBbAW&ok2@OPvh^CPL+m=k#AnPU3YMnG{3xZe6ma1wY`N2_Cc#de}?OP=l99yUdmqg
zUmx&<4z*R%Rmi-OB$#R9<Gw7UE0n)NA}fcP!tjtb)YLg{0B1~}-}(8FY4#2RS!^V<
z6EVKPKd)h){C(@JQ?8X4>BVu=u2fC=p8x^lH`kzm^hPlR)$-mx*WlcMqI+aCYplPm
z=l7T0?Xl?$-^hcRj+(>6E<M+3oBJ#b;()OpbFzUrT=17nQ!$2FgpKi9$5y{qe6+uP
zRbyGdpH}q?K~Xv(mrCeu>?(^jfCkCkkPK6PvY*32G=0xBl~tPqa^FMd5jN1QHE)?*
zXxvVM?T@miW%CK#+VV{H$oroI-!QxlDb|PA*+)LkarUFO_C@zsR-0Jjm7Zy|&~)|2
zO5yG8nT|Z;TiT<ce7H<FUS-Shl=G~<pM-OJ+E0T@U%!vm0<2s_&4Q)cb?rW~C`&7!
zgPtxt>gU7qii)?ort<cmRA0-)cIU*Sl%5gZZt8J$DQpl=lI3W3iVZ7di-CS3gB)RC
zSzgNObdtGntE#b<NwZWV%{h*64`xp7)+M5UdFzU7cG>giFRyTId3kZU^ix~hrKa97
z8V}Xdbh=D~OVVOwm9R9XFfPtdVzMVq(($46QlEMmrggJDMQ@}txhxBp@|3aCF;C73
zsGHvNwssHw(7P{x)yv7qehdlUQ|N)RnfmdmQgJjNoPLhF6@u=|z<P!?fX2(jBBd08
z_9~MeJwC`u#L7ONRgdA)x4q{C(f~@Q1*iIqP%SBu%EY>kb)^sIZhM4aiDnj-GZxC4
zk}IV3#JN>c;(b9$qWv(!!}jLSzJxUj>HoImt<N5q4EMuhR*vM-OrH%O#tYG;HfuCg
zk8lyvB5e$!YDkA%))~@y*=|T8yBwp$7sH{GEW)XDc|=5R!`d2L%6n#ZvLrG0!rScf
zLb6?0hMjJ0m1k|LNCy<+J<xtBL)Z%=($MGhQt@VJax!s{=f^cw`6?}V%kUx9{oXY$
zJ-wInv(xW%v+_)TpZf>BnN}8xmQ*;$HC{W@KIh<}1&$C`#J@m$R~1_H*cwiX4JbU0
zjxau8T{6-qvFJeO<$oD_JRQ4ar--0o+>-3&NK|5i?V8YK({y8gYi*S86DX(EnOT;n
z$IO3W{^7fnlD7gm=}`J3;G3dEJEJ7h{Gnn8yoI)|xl~DCR>XS`yd0>FcLs?K5|T-W
zt7lvDS0qdWOv1rk`&=EPd8b|GULV(0H#1J%RC#*aH3R7FfEY?yZqujCQjz3|dOp@J
z1qILB(deA*XZ_YzWf9Th)|{Ltqw2P6^W*8MnLb~=`ScqMwD5yMUvY90RWFQt*LBm}
zdg4U=z+YN#G)xCt^xUg20X&M{>!@@vVdec_{ERpH0(DJ+)@VLLHwtEtAO+QAIf`T5
zJI~M&egMp+amX^!eYsea1mqMvx7>tjgk%;68dFj_<kIA~2GF>9SmbcV9%CLBC0=Iw
z{z6@lfo&vXZfW{zD_S!Ti-=TmCL~S(HqexTt|YVw-BCTrb3PEr=G}hAdI24fheeMy
zhOW%RBBl7nh`_+Yl`GUivto~=*ZBmDlA`8c`K1;S{5st^K=0>aQK7;CZu0RshHkWG
zUMY#SKf&R&+P=E8tMghW#p$uo<TMQhG>}cFd~I$rd4J61>4%}=bK3=Xa17;4Mogv~
zr?mCzk4!iR98UE=H6AV-6GSE^SQh6^hFB(&jGT>2K)cc}wD?u;uSt=SH&XPal50``
zoJ;8^Gg9z<%fUwNB+F)N4>iW#35jx!)o1*#TE8()jUnI)%sQPREgvKy=_qiz`6gbb
zCqqNk@|y11pp?~{C*(J8I-l*_QJtpQU2FQR$@`j-@og*r@N!k2EYP(pyx3f~(650n
zz1xn=)LdO-zX*(d_3jn0P)v;hQ<Shr4Nrfi<0&>{0ZU%jM4t6E==ytooWh5+0h+&{
z4X|mpbp=@@_qV4r(i;5@kwi$yofC7n@CG<uaH+D&4|I3<hs$Du)N<_LP26c-D3jm)
zP?hv(yyoO%pTmticS376=>fYM*ik?&TfJ(&bX6axW0mXk%ZEDk8j(3~^-a~+i_b5s
zNyqB6q;&&M+v+&DJrL2ee)6c+J9=pQ)!!b=ttGsgk>RiZ<0g0>{;ePJM3W9y8j>Sr
z=}aiIpKNqUMu|)2uxkQ<+3zJm?-oFAzj{TCDZ`^9Ik_7(%6~$5-@U*WSM6?FV{@y-
zytSFJhFa0PLz0{ekDc}fV@9pNurYmIei7~ic=lek4g<iVN=~yaN|m?TXG=v08(ZrK
zMWtcY%N$UOlWCqUIxoEyq*NQsR1t<c!GR{1gW;j~lw9t<jZz~yM*LQM&Tmn7r=9lx
ziI4j>X2Q9(oz(KX@C0GQOM}ikUzsRioqxjlWG?oy^>WdW>%0DT=O@Q>%`;ZiW@Amm
zp3>jcpoi5NT8b>lx!d70p!ZIBYsA#!3ihIn-?<{7yu*w$g!W=B-#~a6TdBW$RXL_Z
z4XSpeHKd(Aas(J>io$G6hnd&%=~rAy&cDDeyM&Fs4Tt!#I&RHi^O8l=?X4sK$d%5H
z>_Qj7F|QZ{Jz`;Msr5a-#fe!abUs4*<df?oVfV5TKS&AT=I_lqUwwYpWc!Fl15$gQ
z{Z;$S=QfuR5MN#$9{(BZ#1dh>v(>^-8W7pI7IvV9`Z^akcV?=(fn~mDC+XZLIX#gi
zlk9TB<yc>TZ+V$@pyK{1FYiWfVXc6)7$AbiAlZRCcSJNDFJC!AI`4Pg;Ndm>{VC?i
z1sxG2Ua(m3zTh#Y7k;~U(3_DjK3jLiP`{a$pQ5~sYx<r3t8jIx7lIFtr}3e0&__M(
zSKCiil<)MbZ<?~r2a?b44HmJg5w)TIr0m8f$<T+_EaCBK)!aq^ID2(`{AaNDJ`JlJ
zB)(r;Sl&Td4h9V6?t{{21&%+kFF81|8)o`8N^kTiiPL4J)<VOz4xi@Eu1#axPtuU<
zkdbmlgtf+mIlvn{Vdc=DRm)+mHt<I!^KW{pE8E}*vQ&$K!!jU`6>DQ!S(@M9c&#~L
zA2wL$xdruc_C?ajh!^p32}|d(JGeVf*d9CAvf`6dmA&|onQ-Bb+#M?J8mylvLBM_K
z>9~TDl9Ch~rk}C13NvvoUc887Oh-a<!nuG;9>42gWJH$)T|=Pit^Mwu4T?Wmfd%AX
zxaQlVb>Y~0JS08jrj!qV3=2H`;balq@{Z?xb^pY6=AIQ3!!AT*<;T|CInwu1ww0kz
z??8K&xLss4mA#?6GoZ|0QV;53c&F6wxZH)E^IR$UR|jkvL|r+d`<>3kso;Gdj<b5y
z`Zf;5&5S*YnZ+FRmMO}!*!g)X3*xp47vnQakI_!cqEPi*j;B+-?PLyRa{tVET#=_G
zC}mdjlYRKWpv6OVng535bXS`ppK_7MDG_rou`X23qD1dF8nx0&_}GY7CqdUkXL*fs
z)%ib9%hYv!0zP1!lkj}Q)0e2L?^8^%JX_^EHMzL0n56?yzY^8Yq&)AMFV<qXcX}PF
z-e0v!(wpY&Ss+Nge;zPFm=0Yym~ULz;3xlF#bez(7x*dkW8lxX5R0lGTbx?c>vbZB
z%L`9r3#@6Gp)>5~1qvdA7!vl6e5*qo$V@D<3-|v#_y&Ltp5S<Gii8Kxxdv2KyIMN~
zu*inGf1o;uzRrwq;&DPt3SaCg)eVETiPsL#=5xmXVdo@*<;{p@!qU{lxwh03*V}$B
z{3I<6VPkW4F1>$|bY+0el}aT^Z^hRa1J!}cmcZSy<ls#7jAY>DQGZgJg@bu!Bx}`-
zD>hV6)qG7TZ*vT#B7)+#Lsf<qj;QZNRiJ9yf(g2;wxC&{p}swI%<SSqU%5ge>WgJT
z$#ja6)|9tF%oJ?N7ZwfQqLRkGbkubQ+;pwRSsB3EM)pOX=A^s&|E&N~V9VVZVPbmS
z+@1{215sZ5lpGfGA6WiLDN5zw?&$^9V4n}C;ZJhUc_~`@>)nZ7`J-OCd*ydM#qsqa
zKk6+X3cGjzX{Gh!y64N)>1#EN3|bmvb%J&DB=F|?R|?bZwSZMAIU=|=b_|OdO5=I6
zdVX3tEbMW%h<(HEf~!gITwUSOlwb=z?IpGv5!_ir;}^0zw^c4*E*&1`Ea(3Zg?ttX
zfwst9Ma2gw%7=V0JyGiKw#3v;&5R}|{XT{Ceu|0sr>7~O{(*w>0}~r9_x$GPM+Bcm
zgm+oBdZ^X|%Kbf!F8;3x<ui&B>}1L2>BGQB=cbdj_2R)zYbPRAFCWH>x7_97b*5##
zvU98isn6N)2x$+)C5yQ^_H+uuThVAdw+z&1UD;vGKO+olH<t{~UYV0auY>JcihA}x
zX}R~H1ji^L5n+t{ALZ^(Wi<NWXCi$`>X&QYd<pB3>$#^ASMfsvAMnMr-o%-xms-6)
z{u2-U)QfD5dDvlHlV#|u-hPgMy6;}{J;E-=)|NJPPCNo?;UVzi?Mx^&o~>PFcM<8D
zJ;`lyGSSxw>dYxx+7;AcA^%&wL`k5I*}S0cFf(%aG9`J^+~~&%xklJXI#<JyefA&V
z3ZlfM^nQGWMm_4z<=qnH<FR3%#j#ziHuYO<H#W9@!CH$O8NrK^JS#G!PN75ddhG!D
zXoM9K^>HKTPl}6LW?L&%Wd-D=$f3d9X31dC{V}3pZ(%IB`_6xu-+}5YVdV>42n;|N
ze_QBbgMMM_#iqqp<KgsIRzw2y0e_&30Ssntsy$~jgPz3sjCH$)KN7)8^m#~>3V$q?
z@3LFT&Dgk;^P!6Eh18`_zSAA3Xi;j058v*o)L+mE(D)|&MBt75yKZ{#s|g#bMYbPG
z6qfrGGcNDTVKG(82!4JQY^-w09bY)~Rvq>A5C5oL(4nCUXABBIKKl@kUw{}ir`|q|
z3bJkpA3tql<(z)JJ=4B^x17bv57$U!r(<1?fbfkp9`1eh-rFBkS!f^2)R3SY(LBj%
z=Nj8ZBgu|~)+Z(%d|A8>%2~vmeG6xnz2ERL53!!yxXdDUu=<>}h=xg{q^Mk#V$`zY
zA*r9Yx$bLZx;>sG$@|1D4d&6)DTd~zmW;F9TD{W!WUZz0Vza_W)1~B^MRWPhX%rp$
z#yu&qEx|kJ<MruFCL3!0qfby0d6JQayB~y*96D9ZDO01!mCt7e_2|Iw{U~114_M9t
zf1g06_x?y!PqjXnaqv0ER!=!~jKa!$#luo%n13iS!=s^C{wPfjIp3Ys8;qZNuc9s4
zU&}$!mxQAyVO`Fqu~FQ%E8MAbRS}-@ucmPRen%>WdeqJvzQl1+-!B`d{DA|i4!gI#
zuHyC%i>le4=PNX~ve4R-$aPJPI@xyrln&OI=(5%oF)@tQWWRF{GbKyUu4@v=$v2<H
zci5&=*lbeTquRgipi7taUPo*W$~w|->}te8gQZ_rME_ZOaa+Wq_bGO-c^}W|=CbdQ
z<;lA|%G=S|FPKo`qdAil=W0THNASGm_D-aZ&`!GrMW|&s>Hx76v8Tk|zoew~reMMF
zbla2HnBOEO*<7M4rrW^F_qmcxoV2Vg`};{6;{ElIX3Mgj#!cPbMoV2LvA=o-ws_;d
zpOxZ=vj`)D2($}~Bs}d1V^YlIza!brG9?#9LMjb%P}wHK$=BsBg!Ct?$eA4slOWGd
z@z_ib=Oc<2)g*^^vR+)gXrrebW_9C+?Zd`Jji2$t5<Z&Y&jb^dSqL`DIf{!oBvIn1
zgBq0KkBFk9?s7@YSNi;WT586ZDwmm~e_JBY-*=s*&mmWTd@bKCCl8(b``m50$44Sf
zY(ErP|6(TNu4z~53z2nALv8Q5p<q|}aoLT8&~%Es$u*~Q4|px-uQHuTe+)GflW(46
zTjh_Wt9xuCe0sDrocGB?V_e+6aWplqRv?4&Vv$th@7Jf(`*J(S0rp0nLzkGOC1j0{
zZm&KMqKni?6}1VT;<oMDSave`h3i6~&ElAklESQ}z^u&1&*g<Qxxnuk50?~>+4{F9
z1Hb#kx@ZuknyGS_f1hvg>jMMGs8xuwR^hU`bcL3YP2R4Vb|O%{dy2k_O{Y>b;<0A3
z5j_oy)&s$tgJzkR{+<ut>kFea_ZJCdKK^B#lt)q=J7=SWWR9bDzk;noh6Osr9y+u<
zR0B(YjcS6E)?u>uS$aGb9TP)EmvxdsY;5A2I**W;7Id^Vcae>5La5nL<~kmEPadnm
z%+;&%)2SXy-PO@B3u+h3?E-v*?EW4kf{42S@3iav%9nr~KXb-s+N%6IA+hpiN@Krg
z*CSC4)w6{3sl(tGC2FW{ezo)(X0QG<Yn8F(i)I{zNFHwFQDO0lWUmj3MGmdVw7Yc9
z6Zc`XTs7?PO(GE0UtmyvZivyY#0ox@63y7v!IURtLyCqmg_)*V&!e(E;24z-_SKAM
zdaL%_4C1<)G0E&|B~9OK>?@Q<N2`6!4f~XFg<j0;IRv9ZL$Z30@r>5|){FGU{mW+;
z=XZWq?Jtx$HNI#2KZO$J`v&}iPrz?6iXtst61GMO)%f@8EjZHtOn3eozH&Hvzq`PC
z+on6H!?3N-tem<$9-qdi6T2?<TT#z{RZ8>T+<{Q9>(>Gv(ADa9_wG`%rJ0CiQShV0
z&hI^at@Ggn((=a(UM>^A-hah05eUA|;Bs&lhJ-oo*ON~TSa*jNO!PG8_Rddq{O3?5
z;sKMBNUNUF-oQML$ep?2q0MR6(Lb4A@j;U#S~sDE%;Gh%wWQjk5hwJ+PoH-70;P+z
z+fc^Tf5oRjSBo6uQfuf?(y4FWedik}6Zdz?Q)Kp^4=h$YI5p=zI5fWR5q~IsDg~d@
z8&znhQeYd&cXmX=?wy+AzjiP4s^@fI$*PTTAs?x0B-P?RDmF25WMs+f-@|d;xZ&;}
zv4Q(BV5yE<_;R&vtXR$xd|weZ9@>#@-QR%KcDzdgsf{{B)9*O2opy@YsCAIDN0f0K
zY-8VeB+9a1q50;)19a6XiyB*h>Yka+Lsq$UW)02WP0WUYBek$C1KESqog$UZBF;iz
zo$93r?yvNkZN{bxpI#lc*;Ob{cjdi2xXK+7owy6pXSGu|@H@j&e}5&-xzZ=3x{2jc
z1ln~A70ls${wYGFr+akuEdx)-`QB3Yfriu0X8-s?Q-HmK&WnFPu1A=llD4~7Ni)qC
z<s?V($T^FXPctDa?U>ErxE+5wLRiFPYC>)2S2U%3&IOU|zRTx>#TtQZrM<oR_g}fs
zFB06zn!CMQsArKb72P5$|FSCG>u+E|A>6pXd@T7mVqOBL{`5YfZ?&mlaa<X-bg{Kp
z*Gh_hmexk+6l{4^2=R^G&r9|=BO+s)B|CIeE^V>e$I-+)A&Ex9UxZK+^?3T^O>6&4
z5e0hh2Wc+yLS%gCB%wyXG(ly<q`A7;C%oqxUw-LE1mf|<zOj|qKH2_3q}4BjvgMgW
zX%R{IH*Npu2<MMR*RJlRPNCX`HMBK1hH5ctC~UkuL`%;)g!#_34vuroTs|~T)9saC
zj$4bL$ch8`F*uoCl`Qe+$E|G^r$~l}ON)EU<>=DtxVSj_hbq}m&k?b|c^E{EZ;{F`
zlHv-rcLcy_hB(Y*p%L&&Hi~<z^409D(!?&lZP4)QT-|dj(aV+06BkaNV~Nt+5T?tx
zD>e$SH`n-doJ*w?5C^GO!<q1o%$qZd2<JPOEiSuzc%%z{x}w;a`dIpXZtrblyl#0{
zIB@*pMZJPy1~y>{%y}I)=9)><Q8qELiIz5n9}jqYQ(vH4)W$AxlcMqmpW!9(GteJi
zB)jPyR4YdA-*~K^JG=_QW`)mgXU-fhe#wlB`zH-w{OON<P(l{xC&O(#W`{%Q8UCqd
zu88*Uw^jKKb2C3XOUJT)Vo>(yj|`V_r_*cCx(7~|`Ij$j;GW!&7X;-REUfB^;}HTS
z%{ggrwtF0rD}*B~5Vm?=2NrJsN)yu$sOEPR^Uz7H4|zh^{>jU77>6*cTZa{AuO^n^
zB&~ANablkNfICrV+V;mdp-WY-eo2#dz?B;Pdmj6Te;4X8{h7*z>L0EF9p^;kq;E@x
zO%$|hKQIW4HlEK45Z)We%D}1n<K=QZoBeC(Rg;MSqM@B1CjP#HEv^uD>t92+y?vQQ
z=5n@=X`E~HbIdFFM#?IFZdVr5sRXj+kNYYR{ymWmfv~*?FBCrGInxyF+Sao8)43AW
zGUlqX<mysv9~<$Kz2{TOxZ?lYt#39iD9m{H^)FNlaxN<S$=Nfh*s0K{qlTTSgzki9
z1yWU={>~vJCXm;7@^1){@sxo}3>y;tQ|*uK#M+-_FV8oc=pJ=iH4-6tj*k1*ZMG90
zH8a0oJj~OzF&HN?re*qf2+kU-Tv?Pd*q5G@9-ZW{82!w!SSY}aWSgk-XI~&<V#{~Q
z%(m!z>04J9ybh<l%IJqykDy*}S04*HNyP@VJ>EIWp;_E~?xA-2e{(V+MC<T@UXg|B
zr8nTZ?Hct&{tm4;mOY9dqGP;JuCoRyGIq1BQ#hR13^)1Tv!n$y1b*C*zZj`w_%kk&
ztsE7lXfc?a6p6oE_d|%=ZqDc9g^RtZIzqcBqj9Oa?Ded*yez$Y9=!qoj(y@2e4sJG
z%6r4t>oLS#6)C-M{vDbO_M43(y5qRvEI%0&2Ojd-SUx6^Dr7$PWfd~)yH=pKz|SHb
zNT6O*z4GD1U7C*}*E^zvsN7$@RF1l&xK*m&+|e3Lb^phYd{0lcsH=+ft&cmDToprq
z5_1H%HI#XshA+3?FB>%%m91JZ^zN=TzMLXO_Ya%=-nUyBiGafi)+Nv~lUNjJvL>u|
z=#Jl5`Ivv;bpFhn6B*Tqb0pDxE<;QAQZE1!;p9{j6chxJB*n%L?jYABudGZfdj00j
z(6)y0+7lJnqf~UXw7;jQaswNu-)L&a{MC{H35=<ixq)yMyx_%)7w=$FEVbi7Nzjpp
zTvXQ3Fty%Dy9m4v2pQA=<L2Lg);ep2f)mt4WwVY=%dF?6GAI$L`mp7&0s;bticHXU
zt3!IW1oF4L*$(5bz?+08(}@->;oi@Dv!gyy^rI#`d6CHyr)~m|e}eZS)n*J<J$%V-
z_Fs-{Q(TV01Fg~$?h9_l@4|$01P~MRXD9RgCW~E3f#2c7(yiy)@6ucl(k{ZNFaN-%
z2&VGA_vA$f?|xMMS47VhkN{OeBFk6oD)8^z$@04xcQ*{wK*wD6HAK5W8#Wb0Nnjb=
z_$MC|paq+-1dN&=OLiA>UxG`GjEuOg=X|KrKoc4?J&T^*Mr42>>_Q_2Dj1^!Eh8Dp
z0Y7vb`sdLjVGw<IiX#MieTCdN-Bc$)MUmTXB@6`P3-rBwe87((P%nP`d%`*Z3$I7S
z2=tptEON{;-aygUf^wU72<Q(kwsDR#+PC(k$@PMU2(j3uxMD8XlTt;H^8j&2GJaQe
zC3Q=8^U}#Hqt4rHUSGfFag)0d!{N_tRx`iFiJd?=-WZPj?~S@DV5Sq192gk51Ny4Q
zpupYS)bs{47t8I}voS&}meUQ2B9|s;XdSQ)P)yxyvvw6M+*jPrpGeBZI&XcQ>xk7_
zIVvn<t8Z+?NL-7DGQJT&SR^4(@ALL=2!ouwT$?DO=Pn!~E6uwb4C)^s;1U{|3|F{-
z-~n<W-jmH?L)#nw!Ul~qJ=VlNx$A+>sNJxcVk1Jvn^Ch;`NmyF836&e7>B8AC=IAy
zf?X===*0VxawULTsS4R`0o-e#wOn(4y8hIHALalwPpRr2_hhM+5KCOC74yE!Bx`J(
z{z6D-x0SUc6@+E3U%mPVETYhK@G0E&a@GD=cTvA)#w%Z96*3|LvPR-c#7RtBgZKlt
zb8g42?{^uM{}GMgvQC~tO-&Kt<3Fr>OaP+qAQZd6S#yXhvVP$k)gvmZ*Ma6$yNYls
zt4SRq^$CMw<~J&g&^#y*D{5=U;nc~1P%k^^Vc0JG5d;C%;CFfw2-lB_aMJL^1(3@K
zwQ*&V^ee@ZA@61@de?+6IyEfdMgg-7x+#G@=?WRLjXS8NUhzQC;?&SBLmaSN*Mu{P
zV#AWt*iHK2!>bSVy}J>qtFp4P7#(0j!X&suWio65+$T6fyvUtzRIB4PH9b$7nwkPJ
zny`J!%!o7Z^}81~oH7)%y5?h1*{~?=Af8OdZSzNxbM&up>|>maX{=Pi@bGYYT3XaM
zDt6<Z*G_vYofzF?mt#A3clR`LG&3=`jbXURGGc=z)@3i8YB9TFZHo8egaJtNrq0Z~
zs{-8~<k{goAqfdX#V9kJ($>L$u0C91p#;J|RpRWR#r0G|>NZ}jG)9gZ312OhRdOj(
zN7EGNzmZ5c3a<APZlIuodLk?$VzRT)8Nug*Oq*V-+?J9IdmOJ{hn1=jC6)5ZpPWhb
zXrm5KAxD$f`wmvBkwJS@=SZ0~sIF@fR`4-0O6-qV`(ME$XRH`Cf5VM9`%DQ|odq!9
zbg!gF0R&5W;sh_=_jmwz2)hWQ^8*t1(PJQQe>#bW%0+S0yAcP*#<GZVyBrlJOUGv4
zrhK)qR<W)K%0-|EiuoL)2A}(?`R~4#+bkxSkCv-M-E&jmFdL#0UFypsnPr|x>FVm*
znfB)oE#;%35lKr+Yi@5>1k+%8eWi9B!sv#Y@Up%JE!^wr@jhmLc7Id)8%{U)Hvj&O
zns_Ph7E8<B+%UbFs+6ONocMrUS#lX0dl{pW43aEksB@RL=;ulp8AydUSjvpBBfAbM
zkFXA6!8c)U4Lm_-Mox9QVwNgK4-Aa!+O=y($K&VOph9E}>PW_a;y(0RV)a-?y~9O_
zb!MutT3}T=?VDq43q=Q-(qk=k9=Hs)t!`KssYJ2y$b(yE(*x8I#^{|L8!}#`{7uhK
zcWAL(X<1mLS@m0C6GU=If{v2hznef=In2D~)NT}YehP|w-LN=JQa5TTVBiAxy`&Os
zj&_W^gAl27M6SO1I&~gNe%Ef?*w$p`en5ga&qo>e(u;Qei5F7Qy0amK`D`k7vmyC6
zZ{8&NEe@CDeBYiwo#+ukq?24`K?Li7*mvMBrE~;8J^{gZ5F8N;Y>(kpTe=6@gQh%)
z^C`QEB5(nsU2%e+7^N=OUKbvpc!HDI`b4<Iktc?@=~8VCARz&|r%#_2I&SNW%2uH4
zR^f$l)Pz|nD)JL2BjvWsRHEQ#6a^3|tca7gz13j_W#v@(sC10mY{b(3kg#q2Jt-;V
z5F6Bm`^5uQ3k~0x-^M{GeE3M@rq$fq+5<CSJ5h@TaMdZN17-viMRpIS{lx>btM*jc
zjk<oNhanjM<Y@7`dwQZwz))0kwLgH=-=k=LRRfsU2wtaRD#GVjFh843G_kJ(U-^+5
zS>huUzxw$pf<pNtQH<Y(JEJKXW?DR%|H@KxB5OU@8t}{HvRi~s7~_?TwRq9IPKwiw
z{w($q6?UsTpsu?~Xj+R$#%1+VE?I)En6%D)t!iJB50wk+%E~GS))+*EarjhJzouRy
zGOhMjR4@YTFdjy0i5k~aHoay7KM*8X#YBxPA0HgEk<vGjoECj&r^g{^bPBlIG~Gr&
z;>X%<*=2LYV?J1}i~vGf&@BB*S}sWtiy59il?CCa{*B4H-bzPna6-AYT`!Zw*`4-G
zcUOnu`1-@|y_N`NfD}fcyQNxTmj$t9798K5JBd=!+>4;Sq!7trmg#r3RvN79Urgra
z{}vZ5(1az3!9q__^q`F^Wc7>`{to5odiGkLjP+%yRn_=|DQ8S$E@A(O5x5Hmc^(bq
zViFpoqz)p$^i)(YF``}L)h@dOy8d*q%P_`h6rm`RWtnW$JZl9gN5R1_Gn3JRjti=j
z^b+&O;Fl|psUJO>O_*jmaXFaYzKxNMJEOydbVQsT2!H*R=eC+{#xK!(_l}m5QV65X
z{d~QWQdG6{eLQSlx$Y;+pA$X~tlFtXeYc~dH}z)598^#H_3urBN_rd!!5i<djVg-#
z1*gHE{{CWP%xT@RWwf?dplWZ(1S6Rq!Df`;=7vH5zyenXD`C{=N2XeA8vgOqCnl-C
zS;Lv<zt>^}Xjy1zXfnVb*PC<V2u)5-UWaorAZFM!lG*pzyo5VIxIJqE!Lx(#pFdxa
z@i@o?_9Tk<<IkGw!8!dd2a}#hy*aK)VA)_WFTrNJLDf6qJ3KLBRRU|5vHw|Zad9#5
zJFK@!b1+@`(NYE>h_GSw)+|uRk)Sih9PtRu4r1z`U^+P)hz|_h+^P>AJTUG~_Qk}Q
z&z~>Fs_hr0?^^(bi*eagDKhGQ1;1bFO7hlpJ=6pH-V5A<Pzf&t9dxnI&Q6d?M^(dG
zCWElpI<l^0#LAoWrY9KY?C9D+o-UZg(x3QG`{g&uxE;0$PkD}43W(Zd1{*P>?Xe@p
zKs2ws!L#H2aZL6cX%9MI#O%hwQBnqZ%W|@^!GVDZAV#hv<atxU*|`$nv6v}p@>&TG
zxxao#OcpkUbM62t?ANcCkEp4a4(DSr+R|QxpNSyK%6z(lD^_K;K2|kUVV^VC9-X0C
zg<J#-D)aK?OJQ;=-}L5FlZfO{7Om_DLLLMp1-gwlF*4#9b+zi#J+&f>@#<Ed!^6YB
zvp(B`CaQ1+6s%fqlLihe86?Pm|25Dw6zF6RlqN6=!%gv!OXn>!+sT$U|2_M^mgxN@
zpwws^^b69!_yfO#_gaR{5)xy*Hd2NO_g}X*=R=t_(m!0p4yE04HJ@t@f7CpB1K^Gn
ztZ8>=q6kB9$?Lao6OQ)RwU+&7vX)MtTZ~oW*Dtt3sK)#-0Ts;jHX_*rk3utHr0Ngs
zJa(G}nc3OdkjSJ2`s2S0qLv_LqQx9G>xrt+Y9;1bDY34qYQ?5yh98t8V3%T4%mXn&
z8?e8$h#zS{%Pb~EF*nZ+3<_e`u0@D;f!P%~Y))CtM{(Px1H6YDO$3eX=IKooEkD23
zxa)D2z34S`OVaUrO-(D8$4#<y)akMoxAQ?RxY;U^FZz0VbWh+oc#uZ_zAZen)SJO_
zY}p^YG*E*{9v*_u90kStXbm9T;bJp6Q`1bUML_jO9jG(8r9FtQszt^j-?wQND>nA!
zVT-O;A4_qLy#J(_gVM|bFA_R%*36It&g1Cpc)hCM<@|L2sbs-G2SEFb)xYN*^VLD3
zN?U(Nnny>BezUvOQr8+3+ULNAw3Aa(v}|HP&@e2p_4n`B+pXcOdVbawqjM0`Bt-S$
zhQvo=qN3CpVShgti%yA)t6do>E3hAl7xDzzF2Qp~=~zA&3DLP|`!SWJu#4C?9o7ww
z_f`qZV<BI1oOjsi6k%~XhOt7n2cJM|8w1|Lv5|M{H)WC^-H^rXDpAR>SwnrjV`E6M
zzy0daY?TW*OFh2gB?pY|j9bpcRlun-=_O-Gae$u0Y=$y<%Yd}f54TbC7}YW3p~AGO
zsj1+I2-PK0n8fY*4m1_UQR!tnF+^(-Pq`CmezS2M9oUizwu8w*J8qREP`>hZ2Aj}p
zmm`mCO9lZ$`M^T(oY&jljHa8Mga21h(sa~IE2;nWD-}i<_}y%{IHZ)9hNjU;el6^Y
zZV%x9r>O1L7~YCjoDhZZzuWo%9|Ngnl1*}|6v4g8Dk_HJ<Ktrl?~Mj>b*=*t@lz==
zkAADGE6yYcLs-&S-<|ZNPy|c^`@51Ka%f1N1M4Qg)lJH%t4G;GTak2VxCty4QJcm)
z2*!GT&n>4ugFEKMrQlOXK_)8y`t=u1@4Cr9kcBWTo*=UzKb##)A%sz!RrZ{h0fcm1
zKc9zeXR3<Fe5BNB{;RLAf}x>dnJ#_$Q-6$`-*|+jh*2&o(gGBEw7Z<*dU~{IUvnHS
z(Jc>&&lqx<sp@#GBJewSi-lib??9TlV{WhGIES(-o3ScmKICg*kk2V!O1CNj0ZQZ1
z@*;?&7$+jzSP58Qj8c`w9kx~7Wa-Q%dwltx1-PNI;VanK;(;UP_FspY8U54%)~xXW
zAqK`V1mU`Dnm*yboBQi3nIm=XdHMSGf_?6yYfgl_4ii=qQT}}N9NFc~taXp++4aXN
z9nlJFeTs@}<N0C9D!Ti6^_4Ix0W%?t3!vBg`T5#d%eeElWZv-Gw{I)aHAhPf;mM9o
z{CRnKdSmlo(9LZ#n+}Ec>zbmnFxTQEiiO4IAYF?A#bnRk3#h25SXGQ}Iq;0(K$cqe
z-36r`vNTi^a+{27waPhfJcw!)wub6o8LG3tabI2x63zYm?nOpBwF3RNN;DCZs=VI*
zAuM{R6G*;1uS9$IsWMq@&4}vh>4^qHrWgk^EzP7_C<AL@%AuA=$)BOS=$0<8aa{5g
zqV7~x<@?E>R*h?bR!2}lRT`;=SvoLEL_I)FgqxV-;VileT>)cQDV~_n_9f+#4cq~H
z!C)K(t61<`eQTo?;#FC>xiUf2l3Qo{9n!_NliWqZ%ekR*0G5!`zMS;og-C_|vn-#D
zjrY3?okD>-KzA&7Qn=FEMm0`%3gPj}uc;HoxS>|TPbm;#>b9vsn_(dYu2ms-yOwim
zvM~(9LT4gIG&?)XI3hVeKOa(7j0G9)tqhK>zVf;E=501$wO%JKW@ag{b44*RFGzu^
z`XRN#z)4Y2(Fn+U9ua0}l-qPB2zytGxRsBZyX=qJFLkH*gL{625(R(5Xqk}1kX`!I
z>9G2Ets4SVaAf4Ni)o5fbXOFYbx*cNMGLo)ipo%l1q;x$o^@~5zP8b)gB9vkZZ$KE
zwP1qHPxl*DH8lQ>i4f)u|HKKZRv~T6wqFVhJF-Ya_K^aKT1+jB(hV<h0%~oKOvfr!
zfG<F;69S$Cx_}mNY=!fOHhsC5FFzyqR?;A7I+Zv=c4gDX-X9#6A?$q@39hH={1n>F
zGC;Z%Lyqa(10n|t3k&v>Jn$yFJKfUDzXK>|=OTF=;b67vHV__YLErPa2sAA1u9l38
z1w!~a+w?~T5A;pGckoA%^4cy1gYP3CCSGze1yBMO5(s{V9U!CEO7&a#b`~~h8lx-F
zHI;fB^{!s>jb~Yzk>=&17q8%aqaq|AFkBhP8ygQAHCF;>U!>ngTwh<G2DxD%Rb<x8
za93{bW80-3QH*yK6BZV3dW^Hq=K#b@#;)aG+2(3D49t@s>Xzou0tza`@fcPpr8h$f
z0C>LGn#I;k6Y_Yq1oCkJA!1TeiraH-;#H@>KJDD~OuBaSW*^W`J1tsg`CTe3N^kBv
z=f&Po*geZ)tb1Jew%!ZBMOQr)Fpz4QRnq?NjW1um$Tn&?e0B&(aG9_}Rb)^<hGm(k
z15y(HH;!FdF@-%S2eva*YMBUhU=k$3%a9@g95e1s5BWcJV3Xh99zdS@LwKI_+50$5
zeg@+{-Htu)0Tfk;I(G#py6DA_Edx=n7gDF)^>Hm&G$woIy2m?^TxDq0xJIy>phEdh
zcjPc!8T_&rqQcI^C%kkRr=B?lK7IlucE+1i_3*+_pae0A9UxMZUtexVJhM!IuwMu+
zvSrL(F|GMjjSzry5lk`6;VV{y_8zzh6YIef8^QVzv+Dc6((iwE0AH2V-kqVuvN~KM
zjp|O3?gwTET-F%93O5GIr!wFaEaXCA&P~-6eArk}DH)%$a$5v@!L!@+CV=dX$E&NT
zFru>MV2Vsq7zX=*Jk${W6YynVWbFBT?M~%60BMWmpYSsTpATkv5enwJOT9<STIcB$
zt_OmU^h1)uCW+Dm68gad!Dwz<SpaHOMh6>{US3}L&oHQCB2xL&-%FXFKUeGE+4hNX
zoC~*V99*?)3wx62A&ZGTws<0#%&Hb2B$hD=6l{27TZ~v?t$`&5J|faba&_u3$+OGR
zVhScl!z?KP3p#4*S1!jid%fUe*x&uW1?#0kMhF*d7<m{)B=<gmZ~y*5`~Uw5(7!zP
mUCPvR?Cu0l7kI%X%t5MHP}BJ!F=>gpKvY=jW%dj0kN+P;VoVeO

delta 88855
zcmXtfWmr^g*EUKC2q-Bn(jiE9Bi$fKmvnctRRpA^JCv609ER@h?huA<7~<R9&-eaA
zokRAleXX_5x{}&agE~;lZM*^VaUTb2q?l20hu?B#<6S}Mp?^G8t+-WGm}PK}abF`#
zik>h+<c`S1{|UuW=Uf@kCf_{p{$di3>7)LJ0P`cKz)tqN)Xt~ju({OMy)Q?6CH{g^
za&Fg+za{k;JyD6GbFL1}PXyz)d+;M(a?W^?j!NCa=UWBd$N&p6CGusUKgtjqXi&!L
zOk!@HFj-89)VgMNr1C3Ke~*{-AVuEfT2{lF%M_}k`K5_)`R--BT-R=dVJtaNGPw`}
zbqXKw@o!%oPkE3KX0FtqdFXI-J^3-Q`|EG6c@xjB(8GreH~tl#1`MOE&w<<$+Y>iS
zHAzXj9i>PJD436bCY^+V{=@yFn(P4;D+SNJ3z}-cMLW7rAA#plQ}BMwVDIYsSM$Dx
zefrY(_y{40A?Za{JBympM!w~%$>j5Eo3tArxx$$ld`rGN_=|IIGDJfXt<4t~s0YWT
zpX8ip^e}O4+Tjl$={%kh&$jJ1j?Sc?Pg2uH?aG@hmY@<kvJuw-dtyVk?`0zedv3^9
z;=S*`-|1no`k;Q@309OeHPKn7M5p}mN?Z^xW)&YSEIH!B^x)2H%f)Qt?S$VTcYw{$
z`3B@szkT=OeWAItD+kiWmgd>Z*zE(+lIdS>2-;gdmi-wR;o$B5$w9(*JM%{LBUEcy
z+EmjurtmKybo-JRP&9C76xcqe5#Z`xJ;>=YFq^@?AH%owx)CwlSlu-EqRZhTpSINe
zOz7dgVcyVcCMJ1<c8#hC|I0^Q4aQ00Sh&)gE8e1Y-6F?MkA)e3w(DrGcpZr<i+uI`
ziKegQKGHMoXw~GkJj;(*E*MNorJ0>IR8W%U3<Is!wH4`CfS(nQu)o{x#d<G#__$-C
ztkYmJr^ZcBJ%VR{Yhw*9qh~(9s|yVcTTp``W>NiHd9hrzlkMOYQxj{c<It!V7}}=C
zjg8OWqx$=$U&EwNux+L4o;iFn$^onO;}v!SwP)Zf@t@LqO7674u$Gj1EdawFw3FlE
zXpT3<pKoaa-h36=R;)kDvGw?@Gk<)8tt*<lZhnkjJGif)qi0!hPq?|`>^<?teRwz6
zKufd)<-M2N<~hy=F}pQCrmiau&E9j87JH?^EupTmtQlN_$`s7lfL^(XJkv7Db8?Nl
z-ck8{Y#i@3-s#BYYI1gHuWHlB6RXkG)?unK{LLy~U`;Dm=SsECeD`Iw7vFUqaX2ub
zB?s{z4XbiGMyKX~7q<cD(GdNSt2G5B&u-BRmCH3*X>UU+BbN5JH?M$wq5W8X4`0@s
zgTB)Bu@Fv&kh@?Mq2f86Xg=Ix&S~$C_K?lTIjgbbfB4JqndEYMtT~KwRQl_UJk#59
zJ}O#fK(uV8!e&`ispRqdb33I9SPtEy5v<r=PvjOgk|iVj(H)aEFPF9L=@h<D)t3FP
z1=iKG6ZvRGRL<n(Cz7k48t-4Kei~@|q0%LnlF~n%x35c07Pw#RwO=ROs1{Cj8VI%S
zp}FxM!&Kf4+fvgBVL=Rnxp@L|eolbhgANpllzi+>LHl>%O^o7Zy!S@}r7LGF>A0`h
ztr~WMliXJ~E0|5SVD3S8KB~qZ+YTO0hTahxp$Fo`tB1jEhLX994bjs|wSQOr@MieU
zpLaG#%T5;TjX{32CMM>3YNf_UXh})h6+Rm{r_S<*wd<-m6ZqwR2hTMaxfE~xtN;xP
z>l!r0TPE2PyCQ#<k>mcHPx;@%QyT~{<Vl9@GrIpOTRU3c6H+g}luAuU@fPAELX*r`
z>0VoT>;V7im6tunCtn-3iy4V}Erv$Pu7J3*A=914k4Ark5-~}}0Bm)xmzvW{wT-7~
z*^!W-;|CtV#N52C?;`f|g_WX`1CXKsj<{;R-G3MS&f^>PZkG;NEJWRV$~LcEys%ri
zh(EZE{a(nv1Wz&8{uJzeDc}z$KLuKpiSGZcoSJP2s1_bOoUpScUOD@&+~QayVq4E7
z-jLX{uMJ&xN-b^fwRPzU)nB`7S)Y3^=^rR640*auzVuE!9diRaTM7U!1Qd0yzRWd0
zeJ!cY8@oWZvoc9*8d?N|w+OP5Vj3xJV1I&#o-?KP;FHM;zy7l}RPNuJsVb1Uv)oWv
zCbO;H@!y4r<5(L$AKyf=1^>yw3=t)3HBxe}1>+61p@06(B<Jmu?Fk$`9x<`$+E#nh
zC^fKwVl&FS%MMQf>qw^ytOjp`KaS(kg`Ksz7*tEdd=-{yWM+5u>s`ytO)NZatVs-O
zRU=5J$}ZjL5>ttEQ9-Pwc6_qLZs7pg7jf@w0egeD#Q~AhhW1sZ`E$o3KK8omycMPm
zq7R$daxM{QPp3DIPWJd|m`2s7YK<VQs#fVI-<}izQ{-REN2^e3Je;o9@-Jr4AdYE<
z77*osT}EI(f-gmb5#PE7g|3d#6)@*T436aCXNi~l2+ueC?UuO1>*a~jb7i->JvBj1
zCz-8j1FoM=k#ZujgWK`91sv8j-C6#z>)A5Oir^Y#FxNCIdInRi5?Z61!6ERz43Rlr
z1{$^(2Rw(Y6b7Bn96o~O@09$ZR$|_9`l3wFQ?sN^dO{1w>HfL&Rq@*3i9U6w)UV`>
z@7YzTho9~&$Mfjk^PjWU4xe>6pY>ar{q`q{9Zn!Fuu{AX<XksxKRLc+^>N(l>|y<-
zY+kan9X~<3RF&XDclqlLg{-?2G1l;*Cvc54T0Jy@RC5%Z9x%{>_0-<4{<O<{pOKlH
zo6E_`5#0X}o)Bw#QZ7;8TCMZ`=H>=l;N{7y-A6VYm7=LXQv~|K8iEgpdf0q?T{G^*
zFJo~2sMsh}lsmmw7fHY}kONLsbLSl#ptlNl#<~@D{V-l`%dvEn_o3waX^HvBz{@5N
z)9m(V<?h95+x5lv6ixL>Kc&}G7>cfwULX{c_8-l5NiKC}TbHn3X)WtP?*846!_WP9
z{m#9GydmT@w}*@t@h53?oy%$jdi}qri>uhJdAJ#0q9#^mo}F&1))3qW`6@suDG2BC
znn%V&RyZ`e&sfQsAqK3OG#LmC07KUbjDP7dgS5B5|D8b7X`EgE&=7=4n6h(28mC!G
ztx&^~GbrNK-|&a~woO_09n41dd4<S<ScUwFrHgNhrNMN%Vq(u!PG-Jwvj#7wF&EpB
z$rlsq;ze{LK}uIId$l(nxpy&6G2lKpr2sDOmff$?tbK~!i^jWl;Rqpj0Qp9zl6|7Q
z*6o8Q-f4&P4%@u|T%W)r5*iYsoeMoYyh~Rp)R0;!!w!{lf!q0iYq<^zQt>A<F|pR!
z9f{UVl~>Y4;jWD=VEiyG08oi8v%WwFSbJ-ZJ(+gT9-$)A8hNna>|O7E6ugO9<%$Um
zM6Ru^1>azHa$jd#;JG&-1it2^q$E85SwiK!=H}*hhgwBNMT%WkC&{_3?d|7An#0f!
zHC}xC3#W#W2igOH>ZrXk^3UUlYrwfTR&B@1FDUee0xB8+cXd8_$Hfp?97kWA?i(KY
zo<o)>es^>4(;-}+sKHAU`7H-S^Pt}npF7K;FbahLM5*WGo&?}+cEPuZJfh~KyuJLV
zf?1l;!fCY>P`hM6U4Ei&7FpKZ;7uQ^5kF8NuNj%1sx>7CEN$)V#3m*NCnsy?Djzp`
zFFxqMO4Pj^UXK@E6l7fCW2Eoh8I8mzor46OQToU#h9brx)dkmyC@u&pHf6AG80K3w
z*z?fQ<$p+4p93O8JdZc4I%C_WYp;ETn3%$fi<w3H2$ND%MShmQG`W4@D@ON&p5z9D
z`U!`<h1cG-jUS27vMG7F>Xfi`b>|Y(nrnW^mBc0RP5=i<u{L`I7C7hU3y5a<67<}E
zOF`b+>+yY@>>``A<e2<-{fEUC_3hj8hA}@uInp#Cz|rZ7lJb;PyKwCDoDcDm*5-k!
zR(7dz&tZGtJ#!jc42hP~v;G_vu$7Xu`G2KGi`s6wahF|34>9@Y7>AWM->h9}6lyN~
zyYAPGXcR+`)Sp$&^9T1yr@lZqh*=GUx%#O%ofgtvBxM+z*X?|z3w2%$F*X}A#^R!K
zS6BkmpZ~O%(Tj{0v`)P&=ETdf40+k4@MYF)`jIU)Aw6R);tE>i3V-|G`X{GuV*-EB
zDF6Q5mtcsrV`Tt9Xw4UMu=XJRW^oW{sqm<UEA;*no#(rjqyK_}yMtc)%`b>AyNyuG
zvx~3;kpg|g+LDZsDMYAR_Jxg6XRkN}FzA_*2#bH<KWDnQjd}-;*0W2$xn^IM->8HW
zt5zGd-{qsAO8fDU*;zi-CgW6`1qTD+#K4tE@qDJ%Tpf-*caks<NQQ^vc|ts~;&q=}
z+5&s}gmm56+ON0WN)G|%$`{;>-KX0dT7({Dd!0d<)C}u*O2QH0dpoxkkADV$k!)Gm
zv!tYc`rcEGZ2MPON{2aVQ31Mht9Dj;(hInvW%)Pi5T@!)F)I+FJm?%F14EDEH)4na
zZ&sfv?#Iin9sV2&M3gOmmqOK~>2fmZ-#zr%bkvXWsbDAcg``L0;7>n$;ta|x+NwQn
zd~UFl7I-K8_PHCw7r%Pk0~sDP9?$j{ZN|%I@&yI9_@`2^s9E-CahLbNH+^AJH&*b;
z<L~wKyURWQcI7f2#^F&3dD#}X1Y);<f=|$RRX@Q{Z=pvP#+sC>)cojJa;;mebMNn3
z5fWOzl*<TL;D$>R?`-uiaL@9g7>rycsE?AF%28{Coc~HUS-!?Y0J<81<{=J4!#`~k
z8-2ra7X9#haF1*4nieB4#)k^CMxd--hhmSx<VBpqxCJ8AVev@%nPS-kT*A-5-)X=a
zi3A}h2nt}!^_)0e8====j33vcIoY|B1^JISBCd67xTxiZtA{gT>k=u7cl)~3aWzSy
z?1}G#(oB2wnK<yYZ=x{%x0>-SXMzmp8gDI_ZS{B0c2%37Hzm|ZCdVU^EgL8uU^&D_
zCv?VL!DGvYm?|MsrtQC?ClGkcE$OYCe&>d=F|2b+CqNKei?Z{yZs-4O70T7jyTTr6
zz2*w{Z%?#cZ$47&z#Hm)Sa<@oR#!KbRuX#%7J$qChHU1f2XgBOv?VErJWco-O=PWR
zYSP$wvV-MCzo2X5^E}{=G{h-7H?-o9>+U11bxlwKh5#>Qtw^n;Xa>LBqfH*{f8(81
zagfO|(f8Bo9*Jc;Iu|_;jxly~-PIhP>6p=at0>>IT$XT6-4=k;3{VXa`W8~VSRji7
z0v7bUS_y%~Qa5cz4yVqQH|Dk?mF#L1GR^;EFyH^jV9;aI<A=xjHW2<@XU2o<kEFR&
zE;p|$>&U5W>}l@U8a9>BUWy&S$9zWVovUKz1WU+Hcp{0a)Bnjg&7IoOot{;`;k?7x
z<F;~e437$%=^o7S9+K(=43<Ww=a%~VWazzUAQ|8pa@vGajU^?Wj-AJU)?UD$&>JbM
z>Q+<vr96sJby7UhpuskMc<gT6{#bSsD_3LxQ5)jP2!cY!_>@=}os-Fw=Xv9IoS6yp
z{0f&t1Q7Cz=c>gf#Pn`eNt26qDwAW#+X!d0pyP75T=Ttr_z7^@QagAHE7T+9j~DD2
zMj2E18?IFl|9NXT@Mo?LDs$6H@w^z*P&-g9tiA>pjcu(M?q?p@N&jDaeUSN&jugCI
zm|=eVo{t=wCFVJ4tF?DAMXavVd`*VHM5riL#`M5;JtGA?)juYlcxXj^!STyn(-+#n
zl{z~eK#v}k0NOU4tlvAN@Mi|&=wy)uSB`9BqqTiJMH#yA)j`zHgyqP1s8f{XFfD+a
z?mnZ|8xr|cRL%niu+KUM2ftrIZ&=-kYu>w)>?rN8jVDX+dQRnng2x+ti`ZweY*p13
zJ+~j@P4JFx&8iSACoSIPCAg=}3C4p|5qLuUu8ns-^E*3cPG*m+#W%$)uk(3~b_hg=
z@|Tg$?>M&(P)YA&v&|s{zPwwkgUrmvPDtUx`3lW>P0D#ZO$gkb?Wy^c)cD{MruD6R
zEzh)oN^$K+x>)i3-*M6CDAqiGd<!D}aV=|7!RI<#0=~KAE%NLqI?>tjHWny;0>}89
zy@y-~NbYPc?zvSC>WemlANQy3^@M=pS?M`0V%%l;oz|T-mpRB1^cIB@!<u&1Hz!`9
zj))GHr1$p8yyZJ(W>6(W08z|p&DhF#?h1)i&o!^3`<XdN=ms+{v6^_^Z}kGw&bD2s
z^*43h)49aN>KK)3e?ovJ_xf_pvM9s1*)5+#f%H6@OW@0lHNu{&H(|Vvx;PBZQRm(c
zy?6yyU*h-y)0N`-IOfj`YdSTZ0HRI)U(UTVzNr*wYfE5&>`Q^R%STvXLk+I|GrMx<
zV{xr48V-iE*&<Z>(QSm)g-xRQ=Ekqj?XVKB?iH?UT>iQ8PXh5@aLRfVRp9;#x~@q_
z!Xyf_T~!|r-_53qXJPWDe$zx{Yj6}sJdJ0KnZNK1S=-pb7ht~^6Yz$6Fd_=du?(o-
zVY^sv$s9gV50fs4sle$>)nvhniVH6P|7D3_XTXv;OxoVwjhwhAX0|ub)HcxxcHP|S
z63sLt@TpMaoSZW>zY*JH>4<}`+7!9sb-rvUub);EV@r+Y#D&*wGbaj(Ko5J@RegDR
zrST@A{68T}JeYr`?D03nPQ7J`!L7#{kZB`v`-cwu8RkorwH070OlJsGpdw_VK>G2r
zmGAQ#>Rqrb0z5<}k&cEj!{R)N1u+@%u`=9%e;jLXz=7!KwesAA^@wKuJg8}!jjOIL
ze^>Nrx26KH_DlgSyC;@99K4_FUlob%Dw07xWzGF%nu2bwZVTJJ_&zW!A1g~B%~y*=
z-bNtBeNPsuB@BX`If>9}OYpY_grKMJ#x2(Kiba--gjBUP(fo%ae)fE)2L97$(0|3e
z2{6d-m>!~t|7)9#bFgTQ@av7kwdu{JlKo2}zKFIXMH=o?{u8#fFF7>esQq3P%xZ5i
z)sHu^F<9n~+wBVp|L*(VcxD)v1V4zI71e@=)%PA|Vybm3-GRGd|C^1iXMU_np<=FK
zrG<dpIV5Pdm!$QvR8a-klhSf{cs`X16$OQ`Lbk*{r3fhy*NrdZqLztr(r#a!o9!3`
zbRNh4FchFeTSm{*wE1^@ZUmy<r10u5Aa?qd_9~(<nNNwuz*DO%Af#}v1M95i)VWGl
zSrN&@#caH~vcTk-sL6hb^G}7RWw*W-i&O@-pXxIZ_aguwYUtq`L*@x`6mvvYxdvLf
zYHMo;QNeX<xXoyR-JN(VUvE9HCy`n^JCQi5#_6CS@6k@SO41bw=%3x`$M-1B-ya+|
z#MokBqQXZ|AAlRKm?vD;6Z1OX4-4ngkbG!Jr|02zkDM%?SvL+gE>A0dY1TGhk3;8C
zp?2Fl+z8YS*ZwyPvSo0dHjX5+qoXtGTY>krgMMcD3|xj}X-fl(GubRcT+zP*QX9N3
zIAu>%3#N$E^dQHe*tVg$G+@VG(7b3nOuo_uk^|YF5u&MyVcB7*Q_uG%Kz;yz@a3UA
z__e^5D`&#YuP<y}bK(F!oP311_{9TK8rS&ZSoXA_(lxosTU>kqPK`W+s>n(ES>lN9
z<9bqk!EdJxEl2yY!F7hoaGdxj!jm6qyRjFzLER&*s`4u;JB$t#N}g!!EMNVb#l#2*
zE@KU`UuEt7w_JL^S%HRl2<uZz{*)8_|KG9P#j$OCK5{Zi9<X%b_`p@i)>zhAZuFJT
z8_(WVk{%dL?QJ6)7=yW9**yu3JHirPYb7HXxW97alQY`)XeMUQXGoo^IKpga>fNvn
zymm;y>BwDRsGNxqgZoDoT_jMRYr`g$c9D6hCTj-9=jSQ_eY|FvVL@7fj@`+Ek90C<
zv+?g{{{^5hbof5|*kQ_=a}^{l)IsSVcrV{Qr(ScX;DXgL>KgtTmyr!MyJ;_*t#GHD
zZ(fAgf~{V?$N4K!+BZULDuw2<P}AnPU|kw~ZtBf%@<ej7da8$Z%0~zr@TV91qd@Zi
z22s3@V(hdKZ@nL1-LUN+?=!;iJ!znf>FwPnKSE0Np>_tPFnev8cI(K*^6$5X7`3#c
z^T@=r-;4PEoxjS&ptRbgkTr+xWzIEid35{M6i>#|>GS*jvZBDErePqc2_gDKa*-_u
zIh<|OkT$gz)3oMv(!$1ZcgCc{wp-i${l58nnmU7m`*nQdbxIbfIc<Ur=LeAf`BNT}
z!{=V>TxhLw;;d$BVn=gd#~j3p0pXIo=GaaUVaHmrqTEd)HC9Sz5FZ2-?`}`+{nNP;
z$ZQDN{kQoA{X&&}FMREhfI)b8@7coWc8;mF@nY}pHAzpeTx4haco1U?4c(38mdAQ4
zt-1egjG9!ZIO0M2?+~H-WFSB)m)c?^@3%<Y;BPV}-bx?Lj^)Ql@{tJ9PvU5?w00Th
z6PYU=->Cl`8S8!ZUNyo&P{gc{dV(L6<jDV1<r!t*8D)-+aAzflyA*51%#XP&eh0my
z9nk0u$AH#u<+Z(C-Dj$r_5=QDt^FTk<2y%<96YCSnU7pZn}8-UYhmW5kHhnC@|ssN
z>v~4A;jML;^TQQ>u;hoaD!fgb6AWTWP)|2G(74EO_ZJJWd}vo-Il52AEfQH9xrzPR
z;1&^W$jrdaH&wfUad(wJQV(KWw<c&*V8mKf-b(Zv0`cp@i&?(L@ec*^6?SLNRX4tp
zS}f_pKjR^EfEep=o=(-yVft~|v&jSQ1W=bHy?4QTop^N$QmB>(FCK8zGgks60(6T-
z8dTh;C_8mSe^ggS<RKym`P{zG=&Am6WjpNLejL2FegjG}xAja{#R0nz%%=KQz_py8
zc<HDvN#tIC-K5<Ul+I1y4l8G7G;K|S=-!Rr8i3caCCK@$TQ8gcPIp4xC&tTIpK7a$
z7Vm_v^VGC>xpV4}LTxBbU-S;|{S{BHfd~~__-Z`Gv*Uh&T*F7zx}BqyxXscUG=-9E
z>afGaMM%dol9Wb32v?ir(4Ezh^Y-j8PTdnd*UoH_6l+N!2zZFPbYN+6aK6Hd07Ce_
zrAv?d+!+^>fr93hL-B3R_1T6&R}{;A@b4knozd~Ne-~G*9IA15&CH-Qi_&`5L2F`f
z{OOA>#9##5u%#_`0@dl#PCPTDO~c`#9n!Z1IG$R`TwQsnfDXu)JUR|9Tf^~~jj)K_
z^bjb5qxqQzPKUT$jR?Z;-~?v?wr|b}0<8P}M71(j#zauDxN$mlJhwLTMQGN#f9P8h
zV4gfy!%rfA!A)DN&mTYr3x#4?PhzS`ZPRVG-Ib4{CMRIK*drXD3dG8jdi__ulY(C!
zT>lMo)k2NCo!Csay7KkS`H}ichi`2#bI>x_fG=<6^i13*18b93leN#ON_@dbOcuEy
zQz=}XK(Ms5xb9cxYb&t+=ZZXdV?saQoc1nt@4|cApDTQEUxE1IJ`Hp*;_ZlA$C`xo
zrR-Zf2SY{tZ^M7W2t(~2$AOmip(0{q$|Ano9n15^=K;0ZLqkR*%B;WM_{Z=^%#YtE
zuXWXm)OwHs@VW;=8)=8e$i*M=0miuPg;TWEq~}6YDMHfX3cT_yj|H=r7pZPJssr|T
zmM*hw`K%18JDBrWqMf^#q<W{c2m#9vI7~3LG(*+hO`Ef%vo6>_^wYH^>(L54s!7;f
zy9Jk6iAh^{G$_y#F5H1~M#uCr8i<)GqUHLKC4ii*j-5nG+QbAh)X~2aq8+Aq{|(K}
z|35UdtU(F>wrEAQh>KM3mH;)eSystgZ;a{R#iC$E45lFu9p5SGMXHC%Z19Rm{a2=t
zPht5WRl|ZTWhG1O3j8}nc~2_yGs0+~KQq+U-S&vTG(#jP+XIpK;-+{q)#((llFSy)
z{5`XKy^$Z%Ra@iYxE8Aos{n!Y`Y@j@yp#@vBw_k=hx3%&idGVx$q}BO@UxTiPf1Ax
zLJe^S`5RXanFyoS$AZiU^4j?txqpU!-6ZMqRr*Yf<HSzmLRZ3l@I@-s-pzezKu)|G
zlFlDst%QI)9=E8wY9rYHAAo*pId!9rf<<Q7_~v77mFW^|r<8$~^}(exs~>8_epaS{
z7oT?VXU>M%JP>eQdv;qZ3^5xyA^G|tuOEs;lVnkU4NzZ&xHt-@@6`}<?1kmhOv9hN
z${(I_GMENn3;+3Ev(Nig)}@Z^eq*P;5}@RXybp1h_~tnS`xXHdQ9S8q`Q5KP?mvJy
zG{zV95+NPlGJNbLLXbM@L%QpVynh4dQ7KwAbJ!3dUPqKxpRKZD@G~IxTcK^@VR@G2
zv@#O*-dH7#KYe9mKVg@2r72_Z9Ix0eOi2cgGHG!elkLlq#~MvoTeV%)c+L4QLb*ej
z*K@jius=HO9Po^>Pp4)6f(aTZ2r6~5KEf$2I0C`z(o#VNFSn;7!s<0_f|TzKqz<A$
z9m1_Cg67MC<fWZ<_W)I>Qhni+wZEQQGMa<O=6WF_sSbY{NTAgrH)kLX4UP{W1dU0e
zkTp<~>yFr90tqHkeCRT3<az{l3kRc=YQthf6!1nZ2!4^vSWcPXr~9x99#4sEhFep)
zCE9}<^ELlG-_gvR`)EgA<avJaUsRsSOlJBqHI6hA$-fTNP!K@*-{L$GYHVy~>YMgd
z^H=e|{-+n9X8E5f4QhLRi-#(h5B$x1AyoYXRE`c8YC+cj(%%C;LdOkwC!Gug8obcl
zgL0Q6=9^f1CQ!E$FAg>M-%x{k2XE7-2H2Lb9XW`$Iywi@kGzE<mcv7m@s}~Zoeqv2
z()cSQxb-txYat*E8Pp?a&l7=KOsyCG$kdU(E_A!HgY9n*Wz|8mB+4keXqyjT2$!RM
z1=k~2{bc98hRr22r41bp;A^B_fo6bE6N0wFyf_kBgQ0!56r9szxG}h*&U<zh2e0)g
zQ%k?1eV=E);V#uf#~4+;=?OLw3X8c)&-8WVzT)q1WXY6$di;Jr8x=#L!q1<lp4=Pm
ze1<Q)<i0y0Yw*D_R$lj8@cvZNH$c^r_i4u>vv;NNzXN(F`yKet&lDw1TO_(%yK3gW
z!w|=^+Xj&dh%KX2V*>-iP@i<}{laC9qOs!;C2`51B^F(krts{24N)?ZpF=eK>w7|I
z%bnLU&MuXXXAXhggOv)Y&Sy4#H675``2Jm4m)iuIf@kx*-q^%la~n;(ezvG+Y_GrM
z(3u|n3k6cO+(7Ildk|EDDYw7;8b8l=WvVPt748%a@p(VUd=P@(^xdG0=b!ifIPgaQ
z7Mp7Gn1Gu#bD)HTuunl*tYpl{6ebhl{r;cGwVy8{aU27;YJ=oQy^m!xOjkDAfZbH>
zw_h3{O^HidMSkMvTd!liwriZ8UY<*vj|>nRMa{&4iAv`o==<Dqe%s0#i8~Q(>S+g`
zeEuI%Ir%y_%Qs%yBM%y~;~iMJYI2I<p9uQdy|1w7a+^QOnJ7drFu0}9!|NxcWpO3%
zQ<EEL6c;Z4Sm|!`Gu$nt-!q>H7ixT-_xVNalDt32S!}zHX;33uM9yh7l?~V~5JrW^
z30xJKb}g56Ra95zg8nPJ)053fneq9508290#NMiedGlZYvbOs(1cHRCNR*bU^t;7D
zb_c1{Sk+DUm#=ZXB4C8#GIHVFLlcRhI0`y(%x)rEjmix~1saTl?~COi_TcirPI6Sk
zF$EneN9T+7xgj_ZLY{0e2QCiuk?4>@BBfFT-UdyE3%*KE(JLFXnD}HV4NrPdul&5g
zw|=0D2=ee?vV{Mh&FKH0ZGCeumsxTv4pQ|mtqI;d;y%7!8t=uii#8LssHD1{-H`;p
z_`s6RFJo_iDO9FAH#<?SsJ#5JX#tX*m%!ec5>aS%y$vs%n~OIA7#JC%7T`h=kk;a)
z4T>5Rbg`uaBOi2|qXd;46NlbIeN$oE$``+9!l+J8FgW?SjRheF%Dumb_CH@?&|y2!
z@W2mQpX)99SuVG*`WcW0T1*k6Txfd^No(!j-<DgQInTDZQ=-=1cja;fkHQX>0899c
zl^kq%WPGFHHOflae?w&JW2owrEc8dMaH2edJ@f-F^10bR`XYqkTvIdCC<oDdAA{N&
zk8aRd1Vb=6UBsFKb@b5|JczF>MVC~*KEiEZbig%t&d10=5qOC-g-_JS!qOJ|GVhC7
zP9*@#e7U<h7;4zUU9i#$NGUU%q%D{%MlKS#IiY&%e`~}2-PQAhV>jL2SNIPGY8@+`
z04w5iyEO(n%YzpW)6_JQ6>b+J;ocR(B1-;YY5SE?>;yq!AP5fW5IDXlFd&_)Nzv2i
zo`m>C>hOKf#7C=mZ3|ik9+@Q_?L<^#+wp1D7o`TMx~PQpzn&2RH{GZOsFC>yuV$|q
zHsFN_ZEzlT6H!G11t984)To=Fzp?_K-Wt2qT(8OV`-=aYkp;i|*&5@M4lM1biL3mE
z=znU4T-WeeJP2-_*Eaomf8|hTg;&sEonCO-EEM&s7J#FgZv7US<?tbzfYsk((j3R0
zobI6;VAB)&s|GR8Hd)x-*<riP?UvMoW+})_P2p!;{IJ&aP*%yEU_z^Ulu~mP3<RSp
zn|lqWzU)4J756-La1c4F+as~&&sC{|6p6z+ljSf4LlhLVXlfzOmoHYDSE43!c#k=Q
z+k{8EUo&%g@F<M6PkWvNA1nNZo{NuA>;Mx-+^hTBb{gUBj-YsHycdkvt04c(LH^Bp
z3ANkMBiA$)8+L7Hw(je$Jm|LL#T{Gsh)!@otWUIG(G>Cb`3vKnTSH@Y(q^O0e+$`t
zuQDxMy<$Quy-zg!GvOiZ(RvsL;f6Q=mnD#Xvx)3xd(>ih(Hg=0`Sr)!QUGb&mo3hQ
zQ2E1tz+-IR(tm?Tv{wKNtgN^O|41{plJ;jjgU^Xo8Zp(!(s-aJ@vO{W?@N^dYjbcz
zAb#TIAQjE6Ng+@6#Foo`Sw$dQq6XCUZj>85dCPd_(}CFczi**2O$gTblU0FzHO6`f
zNq_lT07RD-U&!o(9k8GF?_tpK)P;ku;x<<82>S_Ff(ttmgiZ6Vel4@LV{2MB&Kk-z
z4GR;P1l&JZSgvS2o{=iVZ?60aQW;4A{kfB=0)x#jW^eHGKRn5b3gi>IvWHm5Uy(7h
z0?r^{n#_X3vmw-(+-A;kYf&{X{6K46X89pNL<1)o06vO|8|(%QPPMf=B0Qn6+Y6@N
zg!r@TgW!{YLmJ0cnAd+3*xBz_u8Po)tG-7e8hO8+O+LLkrFa62*i^PpbrSscKeYN&
zE?*Xw!D&5*`Od}YmW!VLV@Z%6nXeb}c(v>96$+>^&{q})RH1?~2I<7io?$+}KA=zL
zgRg+MP|pLg{QeZ8#zmPln4$}{Yq33~`(f}B#RhY*w20afbF2@!r4-p3ziIlhe@w_G
z%<kc?JY;rWzy*}7_2xN7Mj_AM(%im-6JVjN)~IrPVB=z{v=b8(o)9K3pvKBP`7!2i
zWZkMOkE%|CHr*OoUS#q4!v;jH3!s0QKQv(0kBZpa*RO$7_gIKRj7Y2M!WQ6wh!NWy
z28EZ;ZZA*4Mom{2<U)dE6Xki<Qk1Y)3<8dA_qYy>Mhi|**Sh1?nUmPm_zYrgn6}=$
z92><lNBJHf6d%^I*4YMS;#PSo(u@DAkxZ@{+cwOs-s@9vfGlvlOC!UP)ncK?5-x;I
zI1UELy34VltWNRYBzh*mwV%@!Sn1@95#x~l-U-||3YYinKf(ldjO<{&V8AIXJ)D`9
zHMj;kqVaH|ir}dm*Ub7jIZJM!E=M#7-##!&od>lZ&`Po7q$PduL!z7U4^q-Ptq2o@
zMnLncDOmx5crd$|W0r3bY6rFHFliSbVWc8nDfT`p=w44MFfvgr=P|qLrde!uZp354
zkg(I{i%R;bVrhac;VV~H*+5J?g~WuZt2s=?e`j4W6dai>3%XEia<t*XpmI&)Kt)AC
zYY+qpsjr70x5Yu_NRc12nC~A<U6F4-6f<(t16w&(EeRi{vA@zhUcKf|f2~aUG}Of7
z{^vWrP-U3ot3U3FS~vk`rnuT#jry_lCM1MP=v79E;QIZoReg%=0ZK{o^%Cvi;CuPS
zNS%_k`@Tn-0zZ-Qjk2o4f$k?%+*$r=bVbOSSAZ)BtX=tDEU})zsen@Hr)yxn=K!QM
zgiR45K!#)sw=&Vv%y>n0u=q(hU42S)3Z8uL+B%(T(M&ZW#n9)#hFq?EZvU>v5faCP
z-lI*k{N*X7`~kY{Qv>&5vesJA$QqK%8<Ec@x;u=+>CpC|GrrgS)FBo4kP1P2c`9Bu
zQ{_Kb3~CIaxihdwr$}FUAn_}-WA0_f8R#b)b!0wj*{Z8^V1PNgZ;gucZ5Pa*8=QmK
z0p<HXbn@Xtfg;&Py$rk=-$B6Z%jST3snk6c&f;MF+k(&S3yEivgb4NqzgIuebGwjw
z|BTq+e(v4hPgp@xmeDq*=KJ^Om-}-a-4P^cXlRuD{5r{LSxA6Z8G#7D=zog@3)=z6
zdoSF+ek5$v(`Y!#8u_*B^{Ra-b*8KL&k=Rn;O7d1gS#AG@SZ5II)_U@ovE=|M?F!I
z_7z+q;qjCUBzQZgEjOQM8%;L$!@4KG7spwGZvnq_0%W$!cb1W2xZ-5X4Qkr|cFpzv
zV74k?>H~Uf(CEAS8SE!8HoFZ5m)QMIUi)Y8f5)J^s&ZH}nzY8XU3y+#?Lx&W5OMwk
zW?40X;OJb^4!W!pASR><!Xa``?oCaMiw^yUl8{oLX<u6YHf<F#m7cy=Iy?FX7#N5c
zppy-6LawokyDjhD+eMLo$J^o|<lxVnKl>V}b;n{+NNGxk68N??u2}rmC&FaPZtKI-
z8;WOrj&l}=#G0j6v^(iRFVQpDK0n=d{bpdU0s@N*v@h+qbTKVZiChYFjXcm*o~VWN
zzZZTwyFY=VUmjyxrFy?5g0Uo}rL`ol{)7A9K|hmied6Za$BTJkntAp50ka?2Sg&{_
z>4TA|WNw3@Ex%Pqy+!#O`dd5D+<t0I7wF7ft+#N`Wy#K7LpG?H5Y{?@0<~9@u<M~f
z+}GA^l&kbtxWAb>-<#qTB@5m0#K028Us@Z0X-&hCm4U(z-MJ<+B08))Me9KtHB*qT
zva#hFs}SnAl8WxMNMHiQCodhh_FtR*E{2<$KCiKp5y_Nsz5ZlYT}w|0+D2*wDf5XX
zRCMacRqxikkwTp6<eH%hV2IDF^NY8jz<0YMVEeiSbFto8F8TVH8$3ZNbufvU2Z!|y
z!q2tWjICQ7!SR8o^@iWT9BMq<JFq_M5A7WtXfIwMyyfLh0#GnRGkid|s$ZxrH6=wZ
zmR7NMaFF><dm>NKYQCCbdpjs9=k-(huJakEtT|i<6@tJSV_?tShc*V<0APegj!UPG
zy7qGTB`JRfI5cAuQzi5ikmj37)0X}kSsZ=Q$h<+;DOX`KL+fu>oEzC6bPdn~btz8!
zrGV}dzz5XF`}$>QxcIP`Yy|d6^+Ba27mJEPSWf$xK6vBM9&2YpR_SK~k^>*(m5ZxT
z*yaz*0H~;wdr>>cEx(K|7{-(hQSnixfSDA~D`2am@A$cnnsq$eka#l7m;5axbQ6i_
z{N0aNTU+a0Va0k4F$W8EE{BVJl~q+vj*dX<%1S1!V#;WqB8g6|t?bm_PfUvXSD`sG
zurccz8XmkTh_mwXaJoR<hX&ZrPET6^>dEnkO85AuzV7r(O4eOB=qhWk%VuNi>Z?IV
zpr!4G*p2VH%Dp$77^hP^GR|vR%6E$8F4SNIlVSaLyv>?EKIhd|e7=tCpDr0#0oMF|
zzYy{HW`Ugo(!9lbza{)jZ8_RJ^Lp>w1ffXOoa0+=rj3h|=L%EKb{LeBd@{<)J2yL-
z76S=Bu#}$8&R4j&9pFPHb68*@0Cy*-0VtT>qsHhbqj5gpugahgjCuwQ&bibcPXW>s
zN%f+s0xn|L-J7hbrF2L6>2@Ic3vV*Ilh?0gZ1=oXE36^1xX0p$=r>Ag)|k4VbwpMn
z^{Lrs2Ui}lTOUj|FNKNJ2Cj`|LZjBNTQ1Ttmlb^#c$0S-c=>%+LCKWX&U-xP6E<_P
zZGUt6ezB_(`MfyDO*$7Fm2~6;r@67}J+D=u+H~L%RsU<M;TefD0IZQNl{)Fq?wAgb
z%?0T4urT%v=e?vzti*WRgg}?s?*)G8&g^%rtWjB6Z&~yk&&wqWU48xhFjJ1?1Ruad
z>$BA#-@bKw0Za7wughOWk&@_2kE`LVilcB=2WY&%BS%X$p?q9%=&xF1E`M(zdnkN_
zj40Hmhq)6s{u03U;mq;4Sa4GxAu(WFZ#lhiWJ8PyVaO1EW`<o%N<$}k24-3IKf?!v
zsPd>B4AvJzg!c8IFKh&)GyLp8^5u)I&Jm{OK*!3HiV67kfZkAVzYL_Vjfl7WfOT8h
zC%6=Gc{8w1xTmcT(&_Ja2V=z8nOJLJr+(w&=E#K$xX9b+3H~Ks8KLT2VEqMpqw%IY
zHlOwP;EjIJTa7Ybx1hjR>#C;{$|Cr*9$+#rnavdC9)8s`Fo1!Ji<^t|Nt2+Wq9TpY
zB`AV~$HZ9m7TPB+*bf#4c5t9*X-SVjtjH%A`oFxy&!UxLTvQL{;3YJ<KknmZX6Hpf
zAhCD77>0C#H5xau%uujRvLijwC2jlqi>=34YlcVEd6D~miKWbWOkGpE9hvQ2JTs9r
zg$UTcFN|L{)yqC=F_+j5@bt3tQdA~dQD9OF(WuH+dR%-Zz*@CKn<l0a1GsRpXk!~h
zLTFZi&TnJ`M2`nL;9z*uoDNd)yl~(kUqL)iO$?-18%WPIjU3REkzYk)O2;tFMz?}E
zn77#mf1^cT0>l_$2Th>+oTU3X+UFhe_*vU@Z{59g45gOzhSJhN&9s}@^j)DaNc9LW
zsaq%PyFFat{&v%)%6uPbL6VlIU1g5XlF93|p3(M$F&>?Sgv8d?_V^08vS|JK44Yq)
zlDyHGzTd6RBGa$dkxvK`p3VOR8o?_U;I<W)U9}4duGopsv7{sXJpM*XQ!6GoV_Obq
zctKqDJOcY`AZGk)C{N{TdX*y!87{7RBepO6_hFQjsrb%vnb<dSf2f%rop;<O*J$nv
zY>PlwJ6*UjedJ-{BOrTt=^Bt5v6gN<Iw+Q0J~%BYd+zS~4UNmlAF;r+qXwhR_fBWv
zQE~Ls_aCz_g#Al$V-`y+Ffd43^SQDDs>Y!uA(fDAtYZVW&+{)D8)H)NdOFw0;$P==
zec-6e(T_@udtDDm5qo?<KR-@&_HQ~b5!@;GGvzS-nrFn?krBXD$|stUmE_)+X$2io
z&GZb0nQL%I$i`%5Du4d`24N_Li)kWuolZHUb8zq_2M5PHs(5HbB8$Gs{%nOtktU{%
zjSV<B`>s|1L*BYL%F4>h&$YHhh^Sbd6BD@d^77U*r9(eoQ}O!W)wkV$v9PEz=|V@r
zB$lkPp8CMU1AqrPEJMiWv;$iNulr7UZEY<O?z%mgj2;2bs!H=AV(_y+GBYW*x3^~(
z7AESQOqac3@40PfRsL?!_G>lJc}j=BQLS%wgH4|v8a&~s^=S269XsnO8#I21T7ppl
zwl66{Bu%*QMeSDxx2#mAFPn&?2gz$mM5^dn%>btFJuY*}K^!}7aIE}F+xUz6ln}!M
z+Ung?-y2lRA4$41Ny3MGPjLAUarg;h6OL5NtxkL0x6$F`&U#&idn1IK`DPQXvF)u7
zTlWv3$-oryu={OQU7P{INBz2$V)>lLFtz7ADuae)&%C0j`FMIqZtmwNwtmaqah+FK
zz}g&Wz$Ab^Bg_456g{pwhXMIX&1n?T%-h^Jw{RD?4>WkL-WgsOrndWO{US1;tS(+K
z^(e2KoSJwbivux;(k%K2=CCJrKbIRlbK9FV1~KN3z`*GE_^z>RX(BFbJOcxRG!dYs
zQ<%qo@n=yHQ(5yVdXZKcEiG+OS{gvgV%Rd+>~;169sQ5PN(=ac2p2}caq8scv<B{(
zoe_~oYG&r}=4Qb^+1YcIW*A41EAr*#<tu8H#|CA5A`4en&QC>}Tges~Yv5-$L3q2l
zwH2L^&|RpW|MX0L@XK7Et3w)c4pp|Nv9z(#F<)cz%oA2p_cJOAU_DZNdTHW<&U%8)
z@qzhEt6hv)?2aFC@xw0zP}~uCHDW?a_eu3UE!v-(Q6I*s@LC(iXkp*sBM-c*D6cf1
zB$QVv9Jqg9sQTmy7oTf*cH`bAuJfk;U84dD%sQsddce5cj!CM+Eey5P5dE8|E{&Cq
z$!b@Ycwzp+43Hkqnae5Zl(y7NBz!2n6&A=^Mx3!Zg(&aPwr5#<K<ar^^|3%B$e=tb
zIN!Tj62>wjrpMyzjD{1cJWtr)8$7})VMz;yzj7Z}u^FG3rsp8&3J`NyBejBS84Vko
zHPIj3%hKSfEqQ0CEH9t!w9yTMQo=d&u|FTrf!&GV;^JbHx*FpS6n>AxaHagFBbdw1
zaIPWLpXSoExU#!&2DUq%i^0amhL4j;Pay?Ojf{+BIfa4?kybfFK~4_+ss6zf{;MFs
zk^J}(zsbOfC-OghlFbN8PFM$@3&Y!e2WE~p56Ng5-{_M~GEOt6D{1CMM@0aMRiK7%
zaonW-HXlK_>7~drvwyXzXKl>gEe$EiF@&$TS^wV7_!ylK(4DtCfj}D+!unsp#@W?7
z3ic3~4X^P*ZxGHsFw|!<Ri++iW8~`A5!>|0IY>yV_>^YiDNAFiwF058*X4yWB?m>f
z;df+XImP8adL!SJS=xc4x({{L>l;=@$Hh9?=WN33;Dw7LzXv9;q=BOfZ=48fI~mNF
z4yCd8mDS`5f8?y|L$D~64FkcT4Pt~YIG?2r{cp_HoJB<uHo7CIsHt-l(8j={_h-!V
zdMnXTMpjlVe<C_HRo>Cj(U|B(iRQrQZ;7KO|1V#t9hy&200b;7EN1P>9|Z*k?PG0V
z7(R{H>3bxZz;nmzV`I(Y&rZrBVBf4Rd91v@yZZVJk=O5*6Z{mI9XmZclT}doGE=JG
ze7zo)yF7CGlLPE23_=zJ6{x@8_kaLdWo7C73H(c^vu#Cx_yG?wr=`l?n~ygiykocw
z4Oei2zaj#go14eWK1W20Wd??wskHy`7wJhbeG-4gdoL6x+Z4WLho;OI4dIDa53Xa(
z#$ftpVL%dGl+Wq{^^aYwVLSF9<*e0`A{9twM60M5<c)s5Ga-fzX6$_hO}U<Z8zOhV
z4#cvB%WJ>+SE%sAA{@cSa!qu8%oDr^mM04XGOjqE;weN{vi)DIr;5WDA_Iw0Jlp>`
zVjX@~s!72Fm7dd|W#Cnkjh(`yEdAG*Fw@_K0c!A>IlO9bF#5cTSD*CkD;Tc_w{(ER
z*OMA73O9m;Y2T9#tS9tC2J`=>JrmzJ_MWV5gm_;u?u%DF`)hKx1_Vxrncy));2|hJ
z;$C7I5o{d6bFdk}p>+lp`}dS}GPfNu$U+cXCnjP`^_#|PY_#~@_I@NLCbr#Y2zs5C
zrnH!wn{(RD$Klc`jTUQj45#sROqb|}{QTM4vvbG_&?zJ#e9g|Lnkh5vX=__6oq5m9
zY&o8TCg8rGWigUryHFc30*E}O5gK@|mAZqche<Dk%Z9+TJG=qr79kZ78;eUO;Lfbj
zo*?Dbc{5V^%4VjtYh>goLl^|0Cl?o6@ml1ElC8>>{U?HH|Mi>cKA3aq7q{uDbeoha
z+#+($hsv!N9tK=n+0hGS0;nz#x!{d7u{R2spRAKuLcY4}0=&G99M_P-yYfA?T)D$`
zkr7oQSZ*!L18HE~OwFjKw#eqHc|eNCA)Nng+8`J9m4~f^6pN<q?n7>Kcil`}-z1Cq
z>gHY-%+%uS{?tX;6$~|iZM&RY|BE3EOAYEpr_p!QpZG>(@oS&f8l0q@8^y#K?S<`d
z=h372-Jlx~vxqPL)0x}n(pD*rSFFx{>G0MI9O?3jEM~{6Z5oAYf$j(MuZf9uOaPEg
zzjtouR##W|zu8J`2OJh^4@zfhbl!s>MmFjWC%U3me3UKGJe#w~eCBv_y44hu+tAR!
zY}i8jRVsp*&HMZx&IU`wY3M{x4{C%IQI_NWEKuPbp&Cb#EqzQb=vgkf@l?hip0-_5
zO9(an*RNAOt{GrGAD8HZ3;Gj5<<i0L98p3V7-(1(`k^l%?JUkoOPj~r=&4p^fB}Xg
z{&HTpKgxpE*G{h*BDq;=Yi^E<#<dFotF+4xrAhi5Uyr#7nvy@2cQ`<~d1L=XA9#}6
z)2KLHz?d4Cr^kP&N!93gIWM2GA5CH1IRKMQ2@}$_xcIN-tepkHdp4f8*JE!BFiOck
zQMe~Pd1N-=fzqwBr=g+>0FR;L-L=!@(aH!&7%^~T;8#ca;1{=DZcZi_^2VE+`9&!q
zw>V8kNl6)Kh9xS?xKaqH!!sZ6f}^RVmp|2A4xgseO=6&<qZ`W0$h6MucwxW7$NzMF
z#%9_b=L3O*m0Mg)Tk(fSBpP^!hy|dqxoj&Kcx~Z7F;{Q^4i67+5!Zc)979{)9!gEL
zAbl$4t0)@KBGQq2KARHPa(M`yOMU(;MbC?_f{Uvv%L#yybK2{l2mkGtN==loSJIq|
zZ9>56%FM2gLo#a{{#Un#j&I~3ZZ3|FjC=#$gu1dYx0eN_7R&istvM6V=X|^k6A!yu
zy?uQQcSg2mpSoG?pVY(9hzLxOXXk^<-}kr;6}&njE+LUD<imZoJ%pI$akSjo)3fo<
z^N|iP|Eu}_L7uNhM-lu2zkmPMYjOt<)dL?8r&+{R;JZua<@K0a&B4uG>(~IEM3gJz
zg40l3qtPrePV32^yA%0UuDfdBti>>Nad$T#Ok#UU!u`_2<Ej#sC0woBBLU13Ie{5|
zaaJT;Lbm|$TiM!@Ih_>9ftO{HQsO&+V8@f*QP_o~o`^Dd>jm@%!K_UCTnReYUQKKN
z{GoBMMO<;XMroRWoSH9Q1D`f8Z_|c;2^J_x$HD5ZZsr9+FIoy@!QcP|<?xNePh^j+
z=DBh&ghEWH9LhZg*#ESb#w4ajfCC%6FPx_uW?vxB)~=eXW^A`1DrF8-vcFc>R&oep
z=Szst-=#R3fw_WWPAG^wWfVg@#=EgBcpR*!ZN4>hWhZCJ$;-Y4w{XmQHhu5R`s%*e
zr$=wkTco(Z6rgCSFf=%3jt9nUsq5%IR8>G8`6_E6_nu$OpVh}X=@KD-)|u;vFoXCw
zBu}ldKs6gh<l&kM`1Du9xcvtj9v)t~?~QX1Hif|^f6%#)$Rp{l4#8tuz~5%+Bk`Lq
zFX?oCx9f`;L+yG;%Ga-7>(x7cd#_ttZaMajRw1eN2O93@W>2=VmRsuvd#yKliz}E)
zDV&AJ?{4>s${20Qz}s|8l-%E+6nQY^n3GSteAXBkk12Kl+@6Ag`{PoB9ZE@h@~2Wy
zzfbKtjFA5sh1IkJntBYFDIV*|Gu^6XR>7zgS+AdBrXb1Y_3<S{M!p*Ua<u&^LJ4A+
zJu+b0AJ-g-4s8z&G<0!2;(zi>y_V*Oqo>nRo_>ep#W|O^Kom*3QV$ry(}vn|C-3%M
z0nIHBR}LVy$^|p0;SYN}`(#Tm_hzQD8Hu!F-twb1Glye;5F}BdW#G`zaTY8hcogyR
zASd{AA>w!^G}u~x^!CLIbVkN^hfM|b7P$#JdGRGWwe@~gepOw}x_4HZ3Au$<rB+)k
zDR36_=YI<5;Vm4nqo|_#iop*7Ss62Dz_B3V_M*#(zB#GLm5a!Zi@Sr4l)dJjU74<q
z_C)EzTXMhc18*(Dx_91?42h=cvzc+eG`<GHd(0?egALqQ3XjGkY+oh^KI?n}n$6gW
z4l(a!WaYlva@Ffh`+l7nj@IhhDY#>Mkf&L7P(M5PQjnz)=7ZZ&hAj39NvwWG2)I9=
z*uo7vn5)VkM>xurPqbXDcgnp7;mvS{5QgJIj!8zw4D|6}z9!57;yq<q%wyiv-PM(R
zFYDzJai;0qkoNT}!t>|P->`pqvEqA%>9Rd&e0#nt2>!R{ds9XIrc9GqRj8gK;Ga*c
z9V7y3xU{8T{qTtbuS4C(uxi%=tNN3F*}o7tHwVC1*GWQaAOe5a^_<mBWI~qu4iyKk
z9R?@_udiW(H+#&{5Layea+mjK7Q%o~@5&_B_k*@6bh(;eYC)Hu$$PFGwyz(CTn~l9
zwU!NCg#=177zRhSo3Jysj}I}&DfaP4mob~;v*ILb(TDM3G915&_}_g6pjr<UvCHse
zB?vjYwE%y5LhjkJ`BDt)*lu-J`B$uAwYPm@&u)linDgiv6H8zOm1jp?;ogV(<YbT2
zeuf?#yS^_SL{?z8Tss9{-IZ3&Fg4j3jtTX@(pDtlqShV{iaKO?-l$6nUP#sW)$J}u
zbzb8-zD1StobmmI>RfsW%&cT>FHAoUju=9ez<W$_e2e4HQs!jMkV2~+bCE78e6gfT
zH;gv<-;ooKU^onR$XSW6RuGuw8X6BTMB%#JiR@wzCyWl`6Vr=P1~B7G5;|^)GeiA3
zA|eK$W~YO>)wT;&^DPnr7qQQqQC*D6jlaRsKe<GT3tX7O4a0H(cOk{3(StpA@GpwI
zD89da@!c_M%(9eD1&@$WYu+AY9zTMEwQjvNG&F9U7i#SsyoY9H>U_Px_0ENrc6)st
z#|pp?>I6?Zam|M|xn_8aOpT4(;uzHbA5C8YRMq#jjRMjo3WB64-Q5T(-QA&dcf%nB
zRHRE9q@^3AL%O@WyF0#pfB*NJ(HZADT<$%4pS9O|>iiP);)UCtDExA=)?Hb~kZ%9x
z%^PHqhO|+$>jVFs>GQ$Eu9uRU5SF+91bj|LMN@-*-QR-nDdN*_H=Aq^n!G`1CUPem
zYCsFeV%JKTzMy_OOCQ^pOn(1-8B^{?!gol^><l{|v5C@-*`8Qp^UIa~LacjSL_%y*
zR}d3cxStqCH1h&h>->A&7W|quMRm`3KjL^#L8n)84V;NVow6LJe1FQGLxQP-eyg{f
zzJ)}<fo02Nwvwy$(p+B!HDR2F{`NtKF)8G8PhVEoOG-n=g@2*@_d#;m{6A&$pO9K?
z6P??}P>R-#XFb1SK$(Z8>ErqQgH7aE-8RZ-Zx0ENsdr;NQFaxV$(d+gnHA8#-0S*i
zVD>4<+&l*;De=4n0x)_Tp@@s+mS@ZzNOJ|&HU%Rs8|dox@%s?NibyNXLTVifQqB#B
zK}Wz<l;_!wzCEi_=4-q`c0WeOmvQyndg#HCUHre=>u_&aw;NKoPv4J6<vE79e(lf4
zZ%%meYD?j`wpdn@V6R)$?2DLs+>@7_W|LOTs*<mjBQ%bpnOrB*)}U}vrV!k%nM<72
zNqhuhaBGfHrFtqh&NX`7O3s42R!%yeHLG=|nlL1AGsUCm?(zw|^#j-Z%uIqphG?n%
zx+EUGdh;)qFI&^)H;LWVinK95iFhJ~9`B+k#6ph%s>5v1ijZ-e_p`RcZEuPN3{7Ar
zE@p@G`ob*0aee_&b$p$n8eLRBYeNTfa8J1%#{SI=;%756H!bqb8`0v)(b6ws50_1k
zEaqdmf7P8kPXE7{fJ8Q=A+@BWC`jIBYgkE9+b?DWHK6NhA)A&-;pfNJ{;Arl=PYK#
z)(<g{Up!@mJKU=Jy}r$@nQCI14`wVjAXCmAE9XHfreh<-dRdC!wRMH%rE{Hf9(cQE
z^3a6Mo;;$l(zg}L!>7J$Ue+c=xMlVJhhAVQlRv7hs`^ZP(vL>@Q`Q~-2^;GG<p&dd
zS2l6fMpY9vJWn^w9!LI@ezXerv;?uLGxrJE+ju327(ws#RW8pJp*-r6Crdpkl9`dC
zdl?htzu+}jLR=TL_ST5mrnzp5U2|*w+33@j86V4yLV}}nx1p_Vj%{K0CRL}Z^AVHA
z<vD2hD3j^ud12YU)md|niGdoPP6t3SwNxOa>Bm#ND({D!=*oL*He|N6Ik752!JqzY
z0SJyL#(TdfOXbJH23vmjd3qBHM%|4OZ5={SiaDOq!4C`vvj2jf5U_SICP=RqsYvVp
zo-E%wW)kO3fMdF8qYyRL-1WU&S<;uSoNT+8&Y!+rrQA9|J>+0(WIQG9KI(n<^e@8c
z+6oQj{Vdg(w1gRod-54X%+j48uc-L5b>w;P?HT3v{YqQf<{!_7nwlE^uJxrQ{Sy~R
z+#vzXTtS$`pU-#3odlu@x94ivz}CqG7zhF);_}LhpJLue%Bb|oT^f4&x15}@SVDK<
zO^*+Co()@Drl9lHc(^V7(rJx)qp6{Byj$FKGV}q<z{Dg-HUAwL$4ye)vOS-de*1=8
zWi{(O`-@SxUi|(0_t%Gj8{x&Jgrv#H$kOiSEGjiY+s*B9K@X<G_v!M?PCNbIUvCa1
zQ&4Z+%sl}!;c~Le0y#A`GJrIJ2L}-mF^Sjd2f!Wz4jXSV`RaDR%W#FRu60L!4Gc{1
zN{WbxfK9U(`?IGlNTfqqQuuRobK%{F(J?VN`1qY@O!Y8$=m`zP=MH+-mq`-ooa5K!
zzj$~icV{`QW-2hnK(W-}Y2pgJN8|cpwA!9#Hj9CJ@RZw?-Qegfd8+Ojw|twrQ}s@*
zm*S<lmaaYL>xCN`RMq;)zN5~#{g+1u4#-Y#87^37jSK(G`w|+Q=Fnv=S@Tg{rNtYx
z=!CO+E?T3qUhP8tdNks_g=KDIf8iBvC+ATcEzQ(z@zcIDcn&`&dP$^N?|(&1g`ssw
zQsxRI>bBMS!H#_xC<lHoOz%?90z5D4Lk!<#tIT_Lh`*SKEx9Ut(aaT1a;2l$neNLb
zphu`^bUMZ+VR`%crMhNwPD>!7i|`-5TQ@XQ%iaL^(!FbFS+A{Ay_d88Jn_Zc{wgEl
z670>RYwygzoM{s0>pe!4wMO>L_86jjda&C|@!UB|(2na<3zQtaA?j=SH|Sq9sH!89
zaU1M!cVg$?FnweZS<#q&Vchmxs;lsYVwxU;3utw|ddvYX7<M8xxS#(vXbYU0n$kBg
z@Rg>7+FDzmiiwE<WWiC%_7b3&wKO#z`Q1@e8H#zY*o@IHE-tzZSzn+cg4txfMW<GK
z;P<CsTJ1$Z8|p##m3Nmmjn{!-D;E?NMsypl-`CdE4CN_Nfk{Nz-u}eu9^u)ulW{e@
zn8!xTX;!dXzsJO=TtD#e@Bp9#%ljHy$Nz$rJl%m}d$y%O;&~*@&(Gfz)>K&u+XOUN
z2fW<e%$8Hk8J#r01~TDnL`0qf<f37FvqR%%Wo5<M<9<d?Tbsz^Vz1WBxj}1m0a3W-
za@)zh%8Ptn;(+(SgXlFoo4J<-5>~68I_CQ|7bu-EZ&B_EvP$rsvp@nX2lVqSFjRD6
zM5SvfIa{A@bL+>vp`LhgvM?d}sWvK=>VCx2Asf!ZnztWR{?;@j+l;-Vb?(2<A2HU9
z2~1uri1K!$=(<ArT_3~zDS`j<ZJfoBg-vS3j;@pW{gac(@|J&!yv`jRJ??ca*m6U=
z8f@1&w2AJVrv_8_gW^q|_IOaN_A@7r&cEVl4g)Qbz0?H^Zk8=?EtB^1wT*7~#ubg1
z%e{x{cdDvOBcuqL><MnY=RVpw``YbJ`UMBAeqSVb7sC}l(thQ@K81Tzqm-o6{E%9$
zI$dF~3l})x`P|Xnm5D@^QfeLhPh@dKS^lGROV2<;r}wX~BaJ(R9TTBY6@S%|5xrG`
z{`JeNH^T1E(`gAC#x62khuS`*3WTVkv9u(Bo``JD9N-&wSF2$vRq_08tf1AE$tvdK
z=7ufX-5u>?^+jGz%4g6Yf!a3aS^%e%Eo}f=Bz4V+0VTNX&~mV9^y(#PGgPYl0s=gM
z3vn&W3jKA)54rcoGW7PcC};lGq*QSOxQYxxC>gR#dN0*L6rlr(S`#y&3z1HSAU}NF
zNbWu}Be9;?T-EAV{;BD{j28!A%lw|p77D>R+q)JR^z+(PeHCjMW@_vkep$Eu*pab8
zoEZ_yxcClnM%3M33FEPwcoqIJ+rhry`&m5?DqY<y8|1L+A}S%=6FM-|(QMhT<>VBg
z@4@;xs@9geaYt<Wki2fwDjJXEl7+wSMs?NGz~UJVOPu`6xX(sfZWoTdEyS*lML(TJ
zA<^<RR?BW516@Vtru51e^<z8BA}o9TzUO)JX)OzF5elcet$dj~HzPc8teLFXDUx{;
zHZoV5ZIHg4^6SxWnl3(!gReN#wXAsVZe`633W}BgW>I$BwiMBa(g(lw9FaFX>A5vy
z`~8|V{wIdbW1#c$yDMNn6HC@<mZC|f2+^grDWiJ*AQ|wPA#|Y#wx;|EgcsN2WK&@|
z-l+pWzMoQ0rJx`=ET=zH+~4e|8o?kJ2v?+)fpV(^Un=&<Q!}`2o#vnr%;Xt&ih$7w
z`+zkdARtN5E1@k2s~mdt2H@g{&!69b{2O%*4V+i6!~wVZArReQYbc$YPc4=DO;oq*
ze&r-G7S?7bDP%bR*C{h62Nw^oV>L`@0F-+lfBz4a_a8rg3{Obt0ms5Mczvqc1|t3b
z{X2DxB)IL7fhvffKWU-C1N@k(nNXO(#VaiSBe;Ck!q4FC_GYWYl9GtDwY8Da(7pg{
z?&m9(mFyGr4rZeyFoxmlc}C=<xc@lcorE!LZzw3bdwVI!$b0~4&g_0>`V5Oi+d|kp
z9n`-iQ(IeR08`iJ9ejhxkS~KQ!u*k(h=Zh6PlHhX{)U1{VOH)3+@TJw!KIm_Cb6Sk
z9ILJcZ7V#U9;Nr~cV4Wcj>8(>TkEeL{d{)v@5M|i>dLsQsRHrfXH`m)qq1$po}Hcg
zOo^;ir`9k%t>+C1-3#6xX}n}vo~qRl)}rE$?oI#4x5bja^BfgoK&tGd7)mTmeUxZD
z7(HtHv@OOMWo`Syv`DCoDQejGBiD^AF$fDhf7GC9!fTy3=Zu);z&R*k38~qq+F@_b
zSFzx(zOATWYCxx-R>Wa?S7D!ar7RdHN4YY|SkkkR-d49VJr@Lf|CI;zUy$%k-V#yw
z8eWB00lt?mDHNEjb^eG`uh2FGm!fxx$+uCkx|Yze6{q8Cq?TzhB1~>%Yf)^1&9%J{
z`M@SRLN?V!IPl+;gkN4>o&lN^$Ah_ei73kW%EOWp=8eJB!Mp2Ii@Bhfn3w)YSjJ$T
zC`10WwD=yccCvZjx%y8YUtV&#KF}$3nps)(Z$ZNu1Ok7JziWu>v(hwsth9fhWy~BJ
zl6U0Ywrh)`l2hfCh>)gio0!1k;^Oi@_yNW<8_AGH1Rph9(x{tb%mcb}e24Be78Iw9
zySV_K_=$+>=Y-VVlF}b^en*n{7Qq_@yaTPJxJG;mrEJ;o1d6j!1ffG-D_rq(==`{5
zA(UqlDZRVN7=f4g)<_*`dYr%QCV?4zlI<f6X8n^dVc8f2r>7R$-zDEVpRx=bhL8(!
z1Vz?=-s4ZQyKthQuIFGts_rZ-U5>AzS)UZaV5==QGr*Dz=6#gzZ%C5RFrfL$Ext}6
zjkAHAb0QH^t8C?i-<)qm(Q(fikh!|~9h$fQ+3sc1CD)%`k%qu0XQF+_Z#q*4-qoc!
zD?%(J$mnpn9aQ2^VfaY6r-wMtom1azDpG!$MpWmUUfo(3iLz!uCYi!qETQCB0EP3v
zr8#7+{n%ExX7A?ab2%}63+`rXm2h$yBBRaEnw*g_g>0x<#E9E&8?R#JA*7zc<RF5l
zWe2L^s-gy>DH6W;O251y>%G$ntE_d&_y9&q!RJm+1px$?oww8!%R30q4~)6RIQX3f
z(?gC?tfbq|#H2~#<z1~<1(k~V&<JWY6=M}ObGRgXsPNC2+oabxc*Ix<y!W5>!nYC@
zK2wQDeXaG{P!u1&D8a&q3KF2iOz&*}!`A#CB7=>vtZK|OY<s-&$-#vNxm7^gxM;fR
z+<jZD#zUv>MV}pEFR2YzI$H8VX7{#HWr<KsXs6{N#i^r1HTB~i%Pxn>LR#vu`Il7b
z?Uk9^-0aRQPpFUO)o5K!)ETb;Yw%1{inO2@>>`mNizKso_&}4hv$JzLsD8Fs<t{F7
zLV8~qQ5o#ZpGr%qNH7wruviOX^l|ep%i15+XzG%c_ni%}s;J!-G#YPnfSuKT=MjCi
zPbL1n@vYHyx8YPB-wN@H-03&Y3Zl96R*3=H&w7nQv>8EAVt!|4{h$`NE3ZP54jhlU
zEeClp*9Vn#XXt?Az*nuQ-0;_OLT`<8|M-4s`DOj`Ukl4t^bTGD!+ZtasaG)#T1iUc
zXNy7K7wsY|rod;!*C(%TUT8xB%g!EV`r*jOnL0X9CkHcS!;@4;K<wEVYY`d&O83V{
zugZ^Y|Ij5*ew37zcXZFjbj<sqkPL<T$o5NPw&NLXKn^#DZAtTXk!R)IFO1eHs+Gkz
zbFIi)?psY4gwSh^8h-ux06Pv4gqeM=HYmYGJF(s$tQ8o}Y(Hr3f4rFa9VjVSuCr6D
zGS1eAb{^-_HSXZTPx{P#qCR6yQJkr*;X6L+?vB5>dvZ!!)1m(5qg5t%&e088=!j6n
zbsi72ma~Cfy^k<!Tucs5IbQzZoS8n}tOksthm?`Gx@Ja0&0od(<ksM0^cEKi?<9g7
z(5X(+icz8hhzIlS`7RTlwRm^Y8*bqSlgnYqQX+K9iFXz$Y;G76>h!atA5)p{dz<kS
z%OD$eyzr9@a^Y($LM+s5YfR-VzwDIB{UHfu3H88HNmURExT%$O)kS^+SJ}>dOLCE>
z&Fo%&)7(FX(VK*t*pfP!;p_84Gbh(#-9_C(!twerP5JQH-YO<lJM*^m-S{5VypMr@
z#6O9Ou!8Fj8yiH46IhS-hbM)YLegkdP8TmB2mN8yv3F|l8)}0Qq8U_|_LQ}85F1a!
zG8y<CAMXxvz+`UINc{4@%Z+Uwf_#MPC!dpWZq=4@_(*Q>_*oO<m!Gt;LWcMAiwDYc
z{NawIHr#`>=9Iz``L1E&t_b5?m)vC`1uZtX?U_Ha8U`p9IS<lzp;3xy8DV^94ngu_
z^12Opy^gl5q*cc9^>o3I5Tq#!j)lN}@9Y`N1!J#STksjDVo~I-#w?)|c3Wor+894d
z0qbHf$cm3j^)P^efCKAj$Ja_NY~JfcuuNWk32(@!?=-JO09SPEh$d4y<^_Lr_tw?N
z8aE`fOz?&!+p9wQCV2eoNU9vS(hU!Sf4l_)eiuA_NMcWw5_=A1&NTQ7WbKy3_`k?S
zpq^+g<)4a{T}Zctp$2G3tFKO4aJvd=p<*rSHH3g5?1p3JkH!*SlOOGAKH`7g-0+bz
zT}?PwK#L4b{<7YCO4HL9XU&X$>UY4nWJBjgxgAkd49`=-86zb&aMGJQ_=@*bV&><T
z@sTs?h`k2O@gpQI=!EkE??@kh`;>9&;f9BHOnh@I)zLoGsAMMw5e8AfFA#<V)CW#R
zz^cX=1%QG-F!kJ^Tv7vEw=?8*kZ1;rIdQynAfQWqeSIse=Mw<?wYsM*M4F0=&{FR?
z%bpFo@kqJkl{?=o1ZQFsM^ZDBM+zEqicnof2;tH~3k`I#2w=}!i8`)c^*riNzjQ5q
zrja>qt{(LxKt%SIMFnZ5!W;vtKZ3ORiLfvVmy@%x4b@tj>C>Xh&)CYy`+ib6@7g{5
zVh$?}*6kEga1%AS?QG+zapk)aidudmXIms3xHZx%|0iqz*Mj=AY7*jHg`7o_*IXAQ
z5c>|%bY?tFx*5{I*8AwMuXo4C$>x1ur+hd`YCN7VBtAI@hgDMCqL#53`97nQs)9we
zDlq0KmtmmsfPJ61XS$18J@pxd1$ITF^3y~%`5CQIW7kRA#6LeIHBQ#lfR!;~LCA&!
zaS=5XfQ;@nl0sA`%cJE)j}BOWR#+jD>ziv@HaEVI;P9vQO?t6@krN!qoDY!*?gGS}
z=M5B2Ru~cXH*cg;Zct_NG~LY&em6NLS-<^|Ky(0b!*<+>%-=e$-_LF6pE6T-X(g!j
z*~s<9rcjz1VA*gY7NZJE0o=P+=7uL4t^7H6g)7pZ&o8mOZZdO+>Ha$_A^EU)f#y2U
zF^-v6cHi^gY-M;}&b$21E+(7XHDrC0Ez}xG)%5vsBo5?LhB>tTD1Skx7#SzI78VAt
znAQ<4b`o%tf0>VE=iVkfW}%o#KU2gQSX?mCB6ckQJF6`Ur{+pUXVh(X6OwG1r>tVG
z%Q7Hn>)%^?)NHGG#YprGGMdHDCpenOl}xPTuzp{sAi(q26TgoiEw#EPg~8;Tz;p@Y
zOFFH04U`y#74dV5!i#xr8f-@7o)Kgr!}CXf^R#IKY@R>)_Ul5t0mWw?1=crA+i=uM
z#2;xq^_<i05#bn{K%YtiqyCq6uEEKL_aWFURtPAM$)ZivD4NsI2PvdBHO2|aLyF`S
z65K%jRZBgi(<b9PT11yw?$%C#Sm?D88R%NL-m?!wq#b<isn&xS1r@b=b?^OdtKWfb
z_bVChke$u?=|2p!5<WB!EU~{o4{h3iROY-b{`L}gQ=(;_%4->AapzKkG3x0>NWt>G
zD`#W*t3hKyl{gCsHWaT@(bUXN6D(Lfi+z@qAv?g#g+M9h@Ohksu<9bG?bkxb-tYhG
zY>?vb-SQ9S>qZ+qTzDO~Y)2Oh3kpnUs~D_itNzX0Z1jbK(E^MQfm!S4&3%7V@{cYr
zsW>?&=03bYNKY}F<M>5_8<fV#pg*dHW0)kQKLoj7n;qZr)p7mfQA2hsVKWfrFQOfj
z!^B8(bDePAE?tuwXdRGlQ>7nj=Ci|A@^Selgfl9tJk2lgfEpCL7tJ~$Xb$oXkvsEu
zr@DErW;wTyA865C9o}O2d6CnSE(AxB_;a>j<UbM9pV~9YPrjlzG^J`K*=96ikqh<>
zeg_3wQ8glGcfL;#EE>clqvmffIX@tmV~D4ep?2b~zN7RNevk;_*?h9aW~5%p_444w
z;D#W&rv<yI<B>BFQ8$O>Kj|1p25nmM5;3~JsyL2_;&Y@@ec2gHd0$-;3F0(%4c;|`
zy<m0rx{PDbGk6&|KonFS_WCXkBQ-fjGDjH_!zz&QeXR2d5E;l=lqq0Gdyt(yy%z$<
zviNjaAcBj{*fe5_#aGorqf;unqRIWj_j4CfZ}$W-1Sb?XipX2$EeJK+Zbve$`5HB)
z%ydg}{9pX<{5p6G>Z8lSTsK$|JC`_fIOoHK<X^vDNlHoaIPbmc3?<H}srd!f)Ns3B
z9$@&Cm$Sj-0tE#>x3;!UmP#}Wv*l6+bCr#M@n~RwbuS#;Z8PO7Lwe_ycgL2GkM!u8
z99jPZ-UN_kvX)ohngq?|bj^u`m&t*MMrq}E*yAIZ&0k$zn9A>F!EN@Hk5H_>RM=fw
zbaW+04c%NGt{~57f(WD%p+DtnkGBstG}NpeVyLnNy;3L{UPHTQcU+H$w27KC)NXr0
zYPL2j07B6H=GVVHGDT-!iXa#oq=C{^;r^`YlDh>Som${c)N!-e7c-J!(LY?g+tJFr
zX=#x@_C<u=f03gy$dx|r2GtGO+)#|$Y?7?4t#<qhD3)C3yjX7zh7=Tibd)5P_Ox;8
zUn4ChAr+>Si2WS~*?xR{Tjq|0lR*ben1OFXUaTcM=qBtHzscfNPkyt1IsL|S_Ss1n
z7n8D<ixX(7!tqj$=~APIsC}K8E1v8;e~IKht!bpBrG`ZExUE9epq;5%kNaES1<#a@
zTg;uT2z!Aa9*yyh8l=wbc~Q>3gKc>C%0gH|7YA(W@Er=dBQC}okA7RGqw?jEh;nW2
z7kXypCk|&Ki0h>2!y`|gxymQ#^`$SmjOnNSubXa7CL2wxZLLtv2P7&Z5s&>>pjkRu
zgdWM`)b0Wlb4CGaE(}Cx07fJNT95z{U$DYXyU;}Yzkg3kfX)?A@74nb6TftebG~$j
z-W@`ZU&eBKNP7zp7l06V@<T4y*%l2AV~eC&fC%FTae<P%0Qju>w|gA9QYvIo%hQ>3
zMc(P^vmWxQNgu3x2u4<1(nrlp!h5$b?%v<jKQObhNyB9*<`3m&>Hkv9D>y5cs$pA!
zTtrKk3l}rHgW#T8W}*0MGi3QFOXex}o|;z~<s<R3I&N@BUvR2i<kKjz)vX$i@p9Zv
zryHO13(p)-h{?~Wv+XYt(ZE-r8X>;3blv5mS;c7<#k3-B>D46sP(?WKBHCnO#)JE0
z$YeESAa{KCY%ihCR{-rnIA>Tg%qfT)@@Wc*&PyuxZU6j@5kyG_j6^OJ3rSa-AkAHJ
zNt7{Vmztj8zcA8i*w|Yenm{3>&eGSeLj8@lWKRQ!XODK%|4Je*=XQKKHy^LkS}0La
zSJv!dJn^l}ekQ4jR31@VO#UH7C8i<kjL<}|@FBnIdb<gkOu>c9WZu$mpY?qU2rm<X
zW{dxTn=ypn)o7M#T-X0o{B=R{-|*Ah0dc5AP5dz9nIsJ#hZJHCT4o}VPjD(yTa!E9
zVwEok(NaIVCdMDumz-6cUFJ2TITNJ%`xU7MX-Lu>2HLMjm+C(9hVJ$*U6M5Iug}qG
zcusl*Mt`rPF4)-I`i#O4Juk-d20hKB2M6h)Kz3#YLWAs$O>|+Q6>*qouEo1||ETF1
zN|nDU8eEoP`KRQcoUg2W{t=v&Lrg?OBu~5BsYC%+Zrr~D^o#|%4S$P*0As+w;=yhK
zh!n8@5PV3zuhM(qzJttPG(38k*(W!b`r&F7t0R;+I4kSj>E<BS8|ckbpY`6j$watf
zGUQ5e%8bz{95B53HVEBGoe7DF3sr}mIow*C8}A!R*)|QF{o+PYdn6plWgPS9zxjI7
zb0NobHtQ@~G%u$5nNYjt8pi;vVDmTZzfIgzOK`F$d1$#(J(=#&92kHyNV8fMpC0B{
zl+`P6zLi9JO_Iz9`DSLI?A_bpZeDhzO~*NC@m*5tRuBw_$jXT-Cw|w)i4XX>P%Q9q
ziei4Me2H|xQc$$J+mkyUQsU!t5!1vxcn^Sq)uH#&mZ(cRi#~ZsjVsMeTGOX(O)pIF
z#ht=!(&aUuPnEyK8YB#>{Bn=E_Z4$jvRu$=s@j$<hm+_nbZCi}#PBySB&IMj8~(%g
z@d|cc(zhp!BtN6dmiX$%_-&CD^{Yx5c)k;?QJP2@z51iSdC)bbM-_FX$ByK^&>G-x
zxAjnWPvlm4l(C<iQt^7ed?t~}MbpjpZ#fP}2i?d&#73#Nb~e%hOjNqFviP`z3KXx8
z6eHa!zpo^TSVB5~ihAk)YsJXs+>9SP>KwO6P5?KyWIl_82<&+<2L2~8AD)kFGH=YI
z*I<f(F;O}h@_BMY5H9?p+s`L9A(RXUthI4$G~G1qGu@bCHu&{g*J`(0CAKytWhx_I
zR<PZri9{{}9LpFp6@Nv9#e_V+eJ0C}j>61~oB*ZTbG%y9a>_0izO!61O;MH_9a5P=
zI-bIck_VnX&zm;^W^TDsy(JDV?S-Xr)7F~Yw}06jH=@w5ZRzud<GQ5!v)QYWUz>O&
zdCi+*@44v7WE=lv96ZbEWxJ=59$X#k@Hf;I|D^Q0V|p;Txro7Rtkg^gYnXaMw(2aI
zHwkJ~3XF&d7m-zb_DKVgdcV-lbV;NQ`_Y`u*(Szh;4-3^<1JoPYNQgtjo{*crs<L+
zoTH`}q<wTJHCjtou_(E;ZKcR1m&MSLAys3bd8|KTJD#f5@~!-ol`1qTv|B*uIj_%}
zS(+Uik(xm9wElm^`aFCZz8+9?lYEex^5T66K_gd@g^7t;V^VCaacn7ypRG%TM%R(~
za*!HY(_fMw@>n&`TvrJ7-H*!e6r4QR?r>cz!xE$g4b3vzo4&A(cFJR*)goeifH!?b
zD8D=VcXG;ucX&Gwxu+*{rV>>OQ8l*1L}l_dD*=hj>y_b(^lqJ)4ZVhq{eF3<XJhAW
zP}cRQarXM;k(R&?oqRd!T)CdG<Ya=-<c6mz#ss{D2UV|Avk)tfG~$>jHF5uIFLJ~B
zNR}!q>UCYHg@j0{a{idyGt&NjFnZ{4vRMPXXi5tT_^OuzT9f(HCT`}Ve7!L@B~A{^
z4NSthM&L>(&3Yy5NfEq&jMg(Ms#}J81AG|!k1c10atZsKCyU(J8pHPHe%?0bOzw#D
zdZz$3LLodhST-Yu)isebQ{pmKeR?8wCWx!dp#h+}iMDO;vDW^SI+qn5<;9*z&uX)!
zP^cYiQ2n>D<w?PaF<Wh;_i%SA2L}|ool0^7Dn_@!u(SYG9B{mQH)$emGyFRsUW^9(
zz+5rF05by+R?87>9E!HgQxCE)-yv7N<=J`4*QDE7+rXkYKZgnyWW-?w{;S~JzcdND
z<l=^~f%sp;c})vJ!s`TZDs}`&(0oUWUbRA!lDffbFvW+bCvV;FeBdUSI!s>i+&^7W
zHy1~TgZ}R+2ihzMGb-~nv=Z;qhF4d#jyJ=kDA<xIl1T>eDWbH-c1Q64w{tGpKvxHZ
z#quOeUJuus+e7K_g_@OrN~}CR`Q?P}9j_0W8qO52PXf^Jd*a&K+WG~g0eG}5xIW&B
z#m9y8<2PDX0sAK5fNOG=-;@XuKa_lIX*|`g>}N~0UBO%BT@~ryWEJ<YWKJVfGu}s%
z^in=EfG*3HId0Eky8!xvBNqFK)qFu}T(%y6sFufG3UQ{bRz4fv2xIqmDwAIYYrPgn
z>p;T|`U0K8_X1BdK}54!eR6+hG1EhOnF-1`;VM*F{4{~2-foUt?uKLR56}YXhgk&y
z3(iJO>@m8p6jBrhHk{~3xo%<}S%h`M%Y}5FwP3Yj5EIshk0{m;!il+1ny>sEA-o)I
zJp=9TjsLqYB8AKzKGmROMJHtcCX)yr5RngD&IUkC28vfK5W$j1b&E?%eg+N_dd-Sl
z&jvuq1DA;!^l%GkRFy`4&nsI8EG3vIO%8_6yQfhE;T`%2MH=`){k%661ZP7#zM-GX
zSko7V|8Nw5N&oA7PS#D{>b-u6$v@%7oO?1_QatUop6-t@Z9`71Y-gdJzvOK-sDS*x
zF<25qTlMmo%L1Le;&h&*jStC=vwtm*$We!`PdqDVF&rIyH}B|D-tQSOQI9qE7*^0J
zO&wi)wr4aMagn^g5}eM|lCsp<qY(^;-waOwUU-}wZE7Gx3s)HESWxo4Fj&A>KvQC}
zxoP`Ie|@uDQf*M-+mlUk7;xApQMZ2hm&9*)4W1--mmK+$rd;j=Q&hJ)(7DRA%s}s?
zVwr81-fLEv3IkRKIQyEwELd(L2OLg50sqdIE39T?W9Zb%fYJsGeEJ6lsIdFG6gH}w
zeR+qN{TcP=;gaJMOX;>ARf$V{q(sQaPye?J=_5Jfg{Rfhqe$kGt_nd?dgzf1-8b7D
zW{NN&-Mc1psxee(pL}}*>KLI@tcCls-yir`R(fgsdllE~k7a@#XZ(2aJsF})stv*+
zS21^uM@Yld6f=a>YB&!>kH*I!-26*Ip$^TolB&#2DJ=1Innpn^+GIQ8EniYbk(310
z4}5@i$r;IR(DP8&yy8Gn?EOi)cOlIk8C+h>XAtJh<?RInRk0z2<xsG5@SB042&N}>
z+EtwamBOqJkSP!MmwG_*8xRuGFH?*`#2pS)(yaDtU5QVdV4F^#)&q;`O}4-rfvP(L
z{<nk*87JBi6``tfj|3FOa-8pw&)We`OQh3?mPEf?u5W6(^yOR?k68elmp3ZrS5Ij1
zgzEGh1P+gRy;tb}G+ns{YT!OA3-<9MHHTxRgj?4rixiKRKZJbB{n5LGLQ11GHI#YH
zEaqBGquzD3>7Vr?bCXc8LvzeP+Qw7l^#^V8K1pW&u6(3eJzr>~)X6FB@sR`wD*&;?
zD=>|Z%6?=`Z}<nDbS*;rmjelCYmfv9gXf@;J*)><sRn<H6)hr^;-y8KyzjZ~*N_3#
z*xu1m=5c8aCdYnY@a^jDUA;P91$P*7JWe?$8tkWdU?Y-qhKEQfp>A-X%=U<(NX+og
za;l@N0O84)qw8<3eNudml>m3qViPbbjGSb`<#86W&>J2#y)TJzu!R0u+cO#|PuCCb
zVvW}8y?PthnyGZ8SM}m+!B0&jx~Sz4Hh`K9t{EU;nQE<b4p{Ptf}RfB5J~QapjrqH
zJddRYKS7bBZqvG;@t&BhT9J9@HqQ3UJ|c8_>{Xuj2bD7AqiGGW0YF;_<A8_jZX^Z)
z`TT=&UbFe%^KFQTxGV35C}w-<vKI{WxlRB(c;PERmm>FnuzN1)L06$G5)x7ZmyIk?
zj~}*gzz)E4cUA%{Pj*ZsITPppc{qoGwp1GIWk#d>S}h6TaOsyo1h`LNTl>OCFe+%+
zVI}-8aTA#`8OH@H$kR(JWjM3X*PxP(W$uH}=F8|l77D$;qQg?1QKqccUz<a*Ot=nW
zx0~ubRwSzmOH{sS>6h$@mSghIa2ug`+%{70Vt947{EGMIw$8q-Y{3i#1OK*#>&8JL
z$)zubS<=le$RzYauT_~h$>4S6J>A=>R%fvz=~lqk|NO$y$@icL+muQMP9xJdQ;`Ou
zu4BXPaRkH&%gwC>J30olDIXpxPuSxb?dwIIR#2vw8n{iZnUXV9F-Le5cgYK-duHZS
zP&8=QV(m%p)@Ve6Tnfim{|66|{VF<Wnuiq#5n|B4AVcTmgf14{K(8gv>n!_6MdQ0D
zvER2}57dB-I$WYgY_T-~^PdFTvNZ=*psp4){c^=6serW^O%cwn3rTLMK<;%FRGR=8
zmvDJyv)gPSFmc&I22waxK6Iz=Kia)Y_345z2Sg7=dBhMaoJ5?MgV2IjkoOdXSQUx+
zlL?ycFaUdR2sA>`d%eD1JTI$aw0gF=8-)Pr)iUvKA$PwEDty6zcRk8}h~EKj$Vh^?
zH!(rEACG)MbI3|``Q9d@isFCG@fZfq5{wx*@Ce~WV0j4R5)WyBVf&e)dT)eeP|LVE
zZCe01XYBRCY-kSuqXv08NiTOIJwpY1DW{_k!&Fx{f80~|2H{CBc~@j#0j&tnou_uQ
za!@#lOF(NM{~9XK1&Qp{SB`YIkFVzlJqY-JCS{@n(D^I3ihx8Cxh79@3lZyJ)<hy4
zlX%PIfZS!q{Fd(91EIF*E~Pa}BgJSPpUMmdfM~OCPxz%R6oz2#EqtV;sDZQ+F0I<u
zaW`Sy)lWb&8E;TjD=sZ9f3>N5pCAT(T^{Lb^W>xczi-L=zi){E7mp66L)Xf1bI;fq
z30zT{7|{vlmtT36f5kH~)MzG5L`d&^@XC4t8XZ;VK$$Vh^6=%+Y!*Q)!o;fk{Jtt7
zt}}&S5hm_>mCYj^KL8IgRA7^+IUg!*?W=c7b~l4GzT*tLk02n9;>($MTCZ6c+tnq`
zX+75k#B?y%Tye3vC{p)yxrv2^IIPmh7n@3szc5_;&8T2-c<?QxT!0nr)CAc=QYAB0
zhVoJpkFoK<mHkv$0n0obpcryy!YBa*Ze8^BwnvVv+;DEiO4P+EYB2+614S1Kap&eQ
zRb#nA{y}fmZsn=_vFY((p9Xvw_e|Bs($3D#(K;6!MMXvP3=i1rH5M^zc=m?Vj4M^i
zWY=Q>iMr)ZrBt~t7OkR=6lqc=60-^~<4z8ddbwY_{a8twwXmA0<#|3Q^walH(0dZW
zV|2C1*72V;*xqjsB`0KxYdJ1{gN2?6gHSsE9X@!~&$XPIC(n$rnw6C&C~lK!g@7~N
zzV~mPAQ2(A1X`X^wo5GKxq|o4|9wh?WfFr6n>{x-w+f5N(BHpzqbw?H7E=Jy7u=~0
z)0^GK>gH^ML3B#MhQcTyYlKvXp=FjZ8ZW#s`$`c<!Qv9P(pgC6Z}l4F8`e2Ntxj;g
zueYahgvA@8dgEV6bZ6PDi+N30qb;eRDk%0i)&G6`c8kx6TPKxRzoUVpc%=R_XaAjl
z?m1GXdkI|Q#F8~4FImRVf_CK#2I=+YV(>Vak2c_F^lQC&=p8$=AZ<Sux@?Zn-dff?
z4V7K<SJm4LJLbEs`Em#?e>4ywS^Lv6{nf#p8y~Le;&bH3|5gHFdExT~+##?A_xRVz
zB(NJCE;P8h5eUWy!ivqD1yF)b$lS)WX@&X*pa;6Pe$XA7QvKuPE|dQ3VxxJaH)-*8
zTkCQvv;D%yr@UNTmLGVDr8=MwD~)U;hx^I;${$SkT~84p6?0;u!62ccGy$FFiNVLU
zlkcZLllWRJZ!H-$zUR?H28`4ABP|L~GG2@5x!9Kftu>G&u#^?6ri=lNGS&7K4~Z5t
ziU(HCUjkQRvl+e@@d4=f2bQiyI8&2H`3s9?aO`2Z^X?_%pdOl%;B*WNIk$i=twye&
zmsbgSGZ?ZnTk{o5t9C<zjUKQ+PJOL&#6{|*zdhYR7|5F%7*8@c3E1=9Y(0zvPWjfm
zi6eJhQ+p{}{7n71_gGi&&;#8RBGuYYHXA7zIJ?uV3SF-~Xln8+e5w;$O&HP&NBdiJ
z;H1#D*3k7T{qkhqldl4N8U~h~?o~kXFN`+I*Yb*iIMc-BiUVDgvtFPEZK?2U&izsx
zC$^=W{TO1bz%9ehb2k1N2*1?9A+7)0!AQT~yEuyp0EJAY`e($Uk$>Pzbn)RBT95bc
zTEi=OLr|vN;gUAq7KZ<1>*0<pJick5^1p9)uI4QKB$(62Xsk08htuFWgdrA|H9a~w
z?6LE@{&M$uXtE6GQc?Z7!FU-G7RID!J^#0|w(dV^cwLXLUKPkM9$z|`!=@?GA`U>;
zMfI_31o$Ip$g!Lx#(fJjdf$AXb$&{3#GH@#wN9mpn1Fc*SI`_2k2SP-=EL5`m%u&M
z{n7t8j}xV`EHCJ~q}E_M_dxZ?NDA4G>|FZ!qi0FMMW^QAnfE@q*+^|=UJ2W1P0VPI
zem48E6<~jMKgFX}cj!90@)P~21bWG60+!_jlB@0KAMux_`+`a$1kA0H{#(j1rguiP
zf;V0SG94kC*F1}5#)G=DRJvqrOkt-qd3!v>{W-mm+K}1EF1q`v&hG)LiSuy8^bYXN
z&(F^p^kRG0Hx6`RMd^U>A5XwuO0@!QFVNMl?goXP2Mi3X>2|LSt?b8Ck8V%h{G&Cy
z&Vn=D)}M3c<(8T*cE7s4UkuTzj%`2u`Nj?^Q08;hOmjrXFfw{_$OTU&C0jHtX5N=I
z1HCpCSle1}{Z^bB$EqSEgakOW_a;sQFIp{z!*jClzbu0ZH*2EPJQt{sRzQVsf7v=_
z+Qm*5rIzK-vuPlM3V@WK36Ib5?8~9^|ETkH63L#f%*p6&&k93=EF;>EfC~M63M_lK
z5r-GD9`SqKxIC=h=xl9uI7@3?P}P^t?UNouJe$H5zDt2z;M52DHelaZo&10!ZFpf*
z8zy*TdTlQ(?EC+BLC5fVT_(!-Gq+c<wH|Z?N56&CyQVHjH+G}4=U0YC#Fal8Ve6y$
zgPC$ltiD-ELP1ryp=)IxD8|i}HH7lkzLtW193@Y8cOjKC9<noZ2{QH)A>g#&w+}$1
zuR$(O4)m{0t9>_;k2HYarnb3sZrY`a!&^`1zu-}z8^a2d>H$vT2*cv(@u#Cc33X3`
z5jyG?h0K6{f}=f&sie^4I*KoNje+?P#^US$Ho~0Bh`s$bhZ|uh?!fSPXJXV~=z0$M
z{VSqZK2TjD1|BOxRV=Cq|J<*MS^<T@5F=w(E_O|ZLC<Rrp}Tt<ox@h*jA2vov^ysO
z^cM6!!R{F;riS*-R0y*f*enTn>@#b*$YG_!C7{NNDl;HRI#=w_0^T4rny7bmWcVv8
zl{R0kduqjzPmW!D5m#A7_|xHFCI?rIw=M9{?6_wiC2bEa+4dAq-h{Q<WP%I(5WIK2
zxX2m}Xg}<w%Ax(dNzmo}tF(g$yy362DLODoO!Z{Felid%U*JoZ6;5u{rpY%mVsgUF
zD>z)Ii|A4e%EVIlVovFOF9(R7qx!qZ)OHw`MSu8AGLK&(RuWZR55o}Y9Mxb8>a*i^
zPqZceZv?TO95SuszZs@_Yh=h;uR+!%$c=5pAs46By)L~wmm)9cB3iJ}WpRo()YF6`
zAfh>+u7KvZ-Ivk=OGf(Zv<v6D9Dv`|)x~^C2|Fj#7e=H1NC-MT(Pl`u!zcq(S~r0Y
z5F)5v*&BcUQ!z9=U9k^o>j*J)lt|2*ON|U84DDu(;AFQGSgm32hy(4|ww+oZXJg%W
z8+BY=?MD)d4Sq53Yy#>EJZvSn+p^`4f`++An8!I_JgNV?%ElYIuf*hEH%T2G3sZ|T
z*<Al(T)kag*?c1L&h{<-#pZiYoC#;n3hx>;xd*`z7NI6eKd272Lu^Sk>Yey*zYJhc
zH2doXV&-zzcw{^YxxdcV<0dSORoi5EpTm{e5koeHWh`eTAt8~jlq2hT(oI!=e=!U5
zp@KFE7YD}|NFd70A4Zq<_pJ`dVVAU(SS>t)j=@zU5%3I0ijrR0>#A2}*~$!78NMSG
z`>t9Ld>U=uQt2gMF4_H#w|ioZknR0c*5*n`)As#*fT2B_2+5vwmw)o`AMk9pO~q&x
zpOFPZF!V%Yaykio%&5I`V_qs$fDx{rrUPCWul-grE4_4VZD~V^&#nBA;=d#=l<n+n
zlEU5do^9gdL|y%$I~W*nG@JJdGgZzInqTnxxivgL*H#d#nr{S7JOIyp15R%RRaH@a
zeKHUy_4)4VxWaV!_lOgKclv?4ak|0`mxhK0DBHm_E)Me_YwHv*o(XzhG0)ezFfk1-
zt*-h5?BY-k80Ob~kw~Z?b&4lXK|)Nqe#=wPBmrv%j3ncKygP;dQ?WY9OydIwFrdd^
zVprh0(5|&dbOCy_H*a8~T2l}M5E8n5-@Nxk7lIxWcU+=lV^4bMieJ??>duklz)JL!
zoyh#C_Zo@hLz&U6SYuf+()3xA{;j_^)(Lv5DMD0o#^Wc#zV}>?(s#=4WZZeqZLX=J
zCV4^gK*m_SoBk_EIxsmA-q~7Imj1t9yj5V+R-0dfAl~uRN_;brgD!#+KHJfHaCP^=
zVEjP9g}*s}+#Fb$Oh8-2b$>sWrv&qZfPD}Z6XT>*f|sd2Q(-pxi_K&}1_}!erR3#J
z0Qg{ZgU96vup*pHTK}9ktJJK3H1}ohbzoitZd)>dNOPSx81>>+Sx$=pOUzWI1%6c(
zmr$j8lvg~j6D?4iHssxdOQ&98iW7{-uu?uK2%ISd)x=!sc#`b@H89|*8L}fHu>?<p
zrxi&r4rE_8TBe8lYa)G0NYtYqZD}OBYWnEPgh?sZc5BD{_!J@<KpbcAog><oN)a(L
z8#@B7i@4CYC7^KJ@oam5z$&y@flCHpE&-2z!;+6wGgO8ytV3bHfzq=p_6G0*>;O}S
zos!^wpRrd?pmsHt{}OexO*({3S6g4-#AgOffP3?Gaj^_q&v9}6C9_0wMin$QC8ed&
zm2x22!lUl1T8B+mlK~=Fj?lyXE!bPI%#OIY&M={eZ<1MWfGWDRRRoULdG8x|QaIKX
zB^6c1<Zdq%h)JxuaM{b$Bn@^Bh`Ax+wuP(I3=z#(ZvT$UZaw!4_=uQ|`(BmrNr7|+
zjndDEd3kxc<2xr_>ebf7&d$#D*U&~Hu%aiPhA*;#I_6<v_`d^RtZKk{>hF|Nk}^U7
zVF}lA>Kxk$KFcK~2JJ#@FibnDSQk2=sP7_!S|A<x#wuVJs8o_(9Mxd)PrXktWXOht
z^MLnH2b#Yv2($cj*J(}rPtC>yXu`>=qemZ(7tTDlpZK#f%)>*)u@<~a`a|5bq-0z~
ze7KZ<|A+5m?=(o6ZNpXAyRhYvdeBJ`^6-X-M@b1wpkQ*k6eW!>w*@@`={hYk>uAxE
z?-(>ImcV7Gw@L=r353G@OiX+&7E08z5qQkY%X@NnI+y|SG`Q{F0JFhr0KP6YbgyX^
z&?8Bo31wxK+Ag=AlwRN5#5PTU6ueL(9>1I%s@=(ADspmP5Hb^WVFWCk)%X3w!x125
z<`?vM9|yd^$3VcYT61teu=ivE<fSCkNx^nO&25(<!nw7~*vx$i7t%BLu>#Uc*#vAf
zWK<g+gDA>*EqnWY@2O-&jKVDXd8v1bP<J3n(a<>JomGo2CL*xe#&$S@C$edOwTcI=
zmYEls>-$;rpe$Niqy4dfb}J#8`}7$2Gsl39i>1#f?=;03{r!1NU5xBZkDX<0_}g>c
zn1*vd$j5kj1YLdEJP=oKd}8>gdA{cxytDt62#{m)wLiLx{;w7Gvz6;%gu<=&>Sojv
z=i>51IM&{)dOr|j9O)*;(bCfX0I^hr6$3!bHoWavf86n!3(>{qtcxOy*>2_I!_97S
z;?w%^=|(R;IH1&@k^xE(Z5<uA^KrFn@P8mW333JhmYIqB<G3@1;sY@8?{GGVAHP3?
zU8dZ31{8lK(4>I>FFTN9<54!qk^_i0VxG}>G@Oui8J2M1QYJ_G)$J$g8&KCzqThJD
zXXB%}nw=wg=GNNTzkDlix($c&iJ1lUVx6o!e0cv_SOrr;MZ2T;i?LM8eBcCVzdh$s
zMO~$Q7Xl{$-h}9xgV<_gW#EJKZyJ%F<nkR#ot)ZqQy(NEnrNVtHPhOpS2S%0e<h<?
zeaHV_Aps~ey&94ST$z-l%)zcEsdcDJRo^5?@+zsQFdWI0KnH*<2p)=vj{YjEc1=Vu
z$!cWM_sbX9n*xL;{R)AlUNnS?mKN0q_}tWHvO$~?91uQ?t$EdQc9@N34}-0?x4#ed
zUbYMjVERl=sb^d)dz1f<p_7*acEVNM2?WL7rOx0E7u!+<?^=ZSi+L2ZRvRmuzEBX^
z&~{cNIu@sooL@_If~7x9T84P$6y%xbZFBlyRq9k)>`B~^ay|-?b8JMu=u3Y6)GEI_
z)TuK1X#*2t+Ni0<{R|dxu0g7`k^nVNcP#ORMwj(v$M4k)zYr32|C{a%0kA4TQ{#VQ
zlb@~H@nvLBeDXjSYhLQpm><Dz*ELhKi^F@sO7|HA@ey!Y6KPdh8mx6ifPNsi{=w4S
z{l;oHGh1>o--(ylyz*~rss(QDJa<F_h)B!)vVEX?jnFVj2}$(!^|8C@3Emxj(5y87
zr>5uiNm=>Dr%#_K58LinLJ?f0BqafHvI5RskmB_qYpDpUwL2-gVc8$w$#m8+zKqq5
z-DxFU5n-t(M#~x@J{Y-Mp-uE)KaarUt!vKgAZzkE2|(4*1gMof4PQ%nyj)EGpryb-
zJVR%iFYrH`Vq%Iadk}6QQoL4FRqtscWh^es1L)2NAwnwR;=$*D6keX4jzBqO1e)um
zWu_EiYom)F%CNQx;EV!nY01jQmS6nL>~a{s$I0(HqHhfi5Vf%;goR-R<1=MvW`14m
z3}drd;0GZc*Ps%aO_eNL*>T|G<HM;IX+v(KAI)CgYuB4<k91cq?N`3H`DE|l@Wj@k
zuHmAT<@Px#=^zrRrw=g1fo9^JfIx%iZw`=pCTSGdoquRhiMKSqdtRVf`6E0W+tSL)
zY%CWCh)_XmvM_cqYtv)|PDkLcgbxu_lCXh&WZTQ{^X`5XHN8t#7?Nl-&Csnb{o5Gj
z#z;*h3#H{sQ*kDLJVnm^0UirFoh&kjxK-$Z%Ib?fE=`&xdH{Mh`;*$A@J*Ik?8TA#
zWx9g{JiD3e5+PYYQ_Hb{RUZSxJpO%98T6SoHx+YldSHHGv0SK5lUQ5w;}CV){Az>W
zGX;+K6|V!)LvjLvkSL7*0Oer0Xq2UfCxl;M3VIjd6&GbO79?A|neVTd=PTxM&b)c_
z)7Z*N2Bh%;Zt;0%7^%s@T&-!Prtv_Mn^GO<r{GKMRz=Ih(w!{KFvEcoP-Vs#$jjg=
zthBEANdotz!08AFST>HfKny@h!UOG*H0Av!B_H4H7&Kp-GcM^<T~QHzbYv?UN+jXv
z=on)a92^`EY`HLb2{=kCS&05CTDUn}MA3LDRtSrM;u^)kOjR4jyoz2R>Dw?3`s*s=
z=IZav2i%^Ja6+c#Pb^R3rx=A2H_*iaAO9a#kEJUYV@1k_h_BdWU%$rcBdh!vA%8;)
zeNvuy`BM2~$pIFEj>q&rpRPY-KVE*s!9jTP@(GRK83|d)a}n`(4YIfQc2VNesqt^I
z2R7YKY}rm190rpKBKzVmQ!Z`!HRAp}tlgaz2QhD<Gxvy@1?h*XYQ9?eB;k!yIvb|O
z^Y;4e)5t2*hZ5W~JFY`+PTCr|_;;Sa_l}{wmSzGMW=?xlLM@B}X+NY;YQ#C2yb>X_
z3*%Nfnj?2~en$k(tj~ul&G%BNl%67^4+@O`1w#^sX=bg&NIW7e#V<OC7mtqfbx?`P
z+H&*69Hm|S++G}<q0uz#3Mcnr9X&S(jR=B1;66dGF-ezQS9ikY1^ShkODyO$X92&n
zq@-llE$|b;!d2ih#KMYB?ecl!%7`qj^1X#l59$slcpgG>g;DE4#1JfS%;zL?f2tI|
z^6$=vm<IKwV)F)&n`GC(Mj+F^|3gue?95s1X5M=SBfCOnhLC7L_2#6JB%@C@86WGl
z5}8uY&&MrF*P4)%#J{PK{j|zlD6R?#oweBZZ=-5g%~;3nhtk<y+oYR;=ZzqrL0(v!
zqw~71#O~hJ{#48H^&+VgiX>WKfXQ+e+mR%@r!A(lUW;hf@<veL@mOp%3dz{G9|A0G
zmc0GkbmS;~ZiFKm*KAvQp(Or|*Zggz^$(%JKC8MCkk7W%O$Ev4(_i~f^4m1sbJD36
zkg>3o-qE^b42uJ&E10nm_1nMUfB-cY7ucSMbKIQ>ft@*tiCW%}f&xyCT&gj6eGjCU
zv^Ik1NH|Y$<Zk_;{&stPecTL>1viu<M;b~hq`SZh0=E-DrU3{_{Pyiz+Lz8BY6Ih!
z+BeYIHhgn|MmaaI$X$Q7Bp@K@i)RaRJKY$+wyVd{XrP$~-8!sWgae6|hM;}Q1u6Q3
zGBI@PI`>bfz~K)X{0e7Dk#f>n<NqH_00O@==xQRRB)lUOr^aKwyUdCMzZWi%2z~1b
zh^h@&_<c5gUJ}5G+0&;U1~qw^&kJj?aI%pX_I%Ni5HDmNuxzO$+7vMlurqe3Mzry;
zk7Q$s-r60bnCxE+W;vAJsEVU{-bD8!n(g~qwlPrj*&aJ*sHpSJTw=9|T5O;j!!qQM
z?dC^w#|zxA5pRU<nI<|@jM}R$t}M@NdEMQW>*PvXn*HvB+`ddgy3);&CaJ&m#N#@o
zZfJ*xhwINqq{bIWiHNj~=J!AwL&d-l;<D%!cY9D*JaZ4r`aoS2Ive`%QB?F~Bhd!-
z7i8@LbfiG1uF6yNU?fZGdq{}UVxt#WtwO~<U}jS-*0nmWMSsQY;|q^e2BsL=SpnFX
zNx)_dDZf4xH904!PlX=bt}px$(f%aYYIRIs7svdSOW`-$cLZPN8)ZbQ!cAVS$4f-9
z5J`=(J#WaB7TXxHCBurmwa(t|^3N-`ZJu$`URexG!rn+m0a<U+@lKi=#WtStTXjhV
zu186026(zM7iTg<Wu`!WuGqSVTWpucdoq70mH0zv`g~>$?7`;9<VSWcCYv6o+vBD1
zNQ_jY_vJ687i!|qOHFE%;=&e8-`3T9rJynEyFZ1~iDTS6*YPMVuQZz8i|6O1{oQT<
zxE3WhRbl`Se0!A4%-A48uC9&`_(x5^oezQ;m)Dft$^s;_K0n-Dx3{-5TTIYfLw~m^
z*$FWi85u!bPB0iBVO~?%Tw~A#PBd6vypGF!Ku{3UGyf;avO(QvlRZO#suA$GfN4fS
zz5Wg`2k-)DvS7AP(1@$1SD75tf@j`TS<m+XuR1aY#-~En)x*O<5XR95Y=~}m$DM`$
zkEyQ!t7_@~1{6?Qx-sY$=~Pe@knRQ%>Fz^p8YPuZL0XAJcOxR*-5}lF@XgWt{-5tY
zSMLQB_ROA{wPvkfoW0=_Q%h~uhDLSBqM5zORC$!AtSoP&AF&l|kV$}<cxlv8?GZ3O
z7ZK%r!j-Gyie8b~`k7PbdSmRSX}0af#BcbevEc}F*f+q9JFqSYxzko1(1~JILG0nI
z4Wo3A`w?NZs^cs-!s-~Sh7uS2PzD{xe6Pgs`RlAS-Jaqy<A@=iH~RTY(82m$E{?1W
zj)A3(&2as8UjB01@(sJUk_2XUQh#)MGZ`z9UgMffYgaav7VvR5XZe6<^T!*jRnz-w
z37t|A7izZGsm>i*sc_FpzET&uIT5Wt-+h8hFv*^_ADQn9Y8ppt1IHUC?&UaCf_P+P
zQt#itPdQx&nt<mqXCsPG-bg3~UpUa@91$oKSbb+?)CzPP5=1>=F{k>vuEa!|^-b>J
zeH~c2yI;b#wRLnXmn{)XQjn8;qGtXz4t!e5(AjJGQeW4{ua-b$m!`YlWXD&lP!8Gl
z#Q9a0qLDN+O2++RV!{n}B9!^bk`0+I9c|d>6vtOgg%x=<5WjZi65$Ap5(z&ftv0Dd
z*V6yEqlk{Gmq=x{V?uKMk%tY-J=-&j)H3|7l=5)_v!1?6A6jaWuGG{U>FHiZ$AFky
z-hEt5+N-a~j;Xeo<`jn}vL^bS2B(Ai{H5odZe}qivhHQa*_%|JYkbDET6~ya)e#58
zM~cs*z2O27!LH#XRb{E+oG9<M(DSV<S|zS$c6PQu`?{ih40Xw-8fgr-?<5K8K4N`$
zi~<Mixuc^4i4-gFfUIVu+?)~I)#q^TnNG?}gYP-*ijVX2pvP#J<=Xy*Cek*&`d)={
zgFK7Rk^td{<-_*HFiMSz7|81ctHAy47^%9Eg#<K3-K@ipcgN)5;m@`!+EZiX{z97^
z=3VQt%4a0F?JP^IOnW<BcOMq%uQKoAY>g#|$Ur{V3AQ;LTYbK$`?gt*F9OxH*0%D?
zY#F&2QI~Rc&e*UkGMlLyIN6ZGSDt_R()2=hXewV2EYGW>K`uEf!Q5#Bc%a@+|Dc|k
z&lgA$MX_2(&?{V~y>EXtCy7kx_JLMVC43CA1cgYf<|86_-}RG4Yt}qsm&W~Ig8gUh
zk!Wgnso9oR-IqK1Hb#F!P9J!kdZY1M{GHX6o|%bH=z-YFK86w&G^{_JIyX>fy{<2c
zowv1~n^n)RuBtDc<(XXJkQ!X`u_q)z3Bm`zpTxhVPfkhEZ1AWai2&bb@xsw`S&p~<
zf`>jsiYZjiI)l~z<TMrqElTO?YVmO!3C8&%&6nx(6A6F$-1Ael)sNKDJ}$B)=cu!5
z!XG=^%Cpc5YAW(B_@B5JS?o)We@0o-P8`er)`ZlU=_CCrXJJ7V$IQinp{y_*BHZo^
z?87%u#gF&a7GCffl@<zYlvnBHYsw}(<Gt@C?Q@1wq=@E3t?0JzpB?GR<VO^76>)Ud
zC2jFZs=!bKGwZiI4}mQ+Nkh%=n2hfpP7tlp<DCBEa8W10@CJ=`+&(Yyqu7kBkezdu
zK2256u&7`IWBgZ**~MPHaiDh(z59cC`eq6ZF?rz7K~>X#Zw2a&%Tqv8vB2pHQ0a9V
zM8JJJZ{;0V!pG3im+&zoPEt<N{C;#r@aIcyy==IX^qAOoW|uKTY?QR_tBi1{W%%BF
z-BTnDJ!yN-y+730_;^7<C#-$VZkJ8EaWsXjGk@e_Fv(!w4aqtMdP=U!V?})>K7}uw
z%aTdh>LPV4hSOB?^7p9SP>ScC>89>7URVnDHBAUol$1bt0lD5565P@MM3vA5Ihos5
zp5=@dOOT(bZD2{jwQN6oGsTG|n71tQ>TyHLg^$woHn6*Bop(;{Y|~ZfeL9Rfd*8<l
zne<4~EBTyk^;C5Z8Y)Qc_!$@%%hEQZrggSQA3vCV@oH1cY(an9(Avbd!f=<QvDkeb
zPMng5m?2=wbzThfaRI<O({gFC`kYj+#H}sM<>#tprsJh1xB%&m)YRJQ2Q*GJdE#_y
zogWYpwT%}W6_5j?yPwBBxqJuP!>2Ly*GAecQ9jG1khI174BwnP3vyY%FMFZ`47k;V
zrbnB<J}6~ioU0?c^C`ede(-J`gd?{)z?I21!!WJ1wD+KN7i1ckT?7jeEKXpy;NOc@
zxK7w{u+G(cFc}=F2Xr<0RZ;7vAE3F9Y?HEW8F;;MqtUyy_8~romuKtG-5{Ko`+JCH
zRf3`K<x0p)*YgR_L(!h(rQT?0t`(xSkKT0b)7+|jxYu|&W~7+6W4$LwQlwn!bVH6G
z@wH&e>+eK>Bg}UMvrbM<4!AntD@;DQi;jK>q*lHSt#`JAlrRX$A^Z@R2?Xf5QFq0P
zL?BBIN?qcBXVD9)82QA1Cj_!Ngw(&Wu=*G<xYm8Wo>Ofi*iNf|LWkOB8GiNc*MgzZ
z<-lt5H);KmoQ%%P=}|^)T)1Q_$9H~MjqTj)XXb2yFc;Id`0A*bmc69nPAqK+w!H2Q
z*}dgguBWWb<-xTia&0q>E56$_GG1yJOQ#jSNi!B=ad`r!^W|~|8IK{@CJ^`Ui65~v
zHY=8*yf&;<8ut8dG=p)iUOR&xSdlg_&F(H*;N@HZHq-9zb*{(ZyCS7O1ra!M(zoxT
zJ{5csV=T@6n{NB#`JyrIp@fn>%dR^XhrtAaP?Bu`$}SKsJ_ABrU5)xRj!fVpQf7_=
z#2<c+j>ZC+WQh4$7$O=dy@t*tCR{VEM%c*AjWT?htdRK+W_q=PUe~$OTvL<XcZ0bK
zkgp^M&ZwvD{TO{gKX2)yWb3(hZ>#6$v!<Y+#)JTEmOjBOBEcHIrCCqHumZ{TFG9!s
zmeSW256ZaK>54>Avnc~?*bP@r`|3lo351I*!uBz_phNy*FSrs`%9?&oGV}ge%eLqm
zh75nn1@7THje(oxsc2p*0&Sn91Q~+2)_2H>&>L>3G8#6V6-;<(=othEtWOk(_n);h
z(wG)3xA>4o_R%JiNFcU}_`QKG4bLy-T;LXAJe;eD&}jC!J|0sIClm7$20*e5&&!Jf
zQAjuGKY<U>$A~#_Mp&KPeXi9_Cw&kS6N@Y|-HkguMI~&>)XVyKNe4sEr(py~)oeI4
z-T}9u7QJ{mW#m1zoL}oir=<a|HZT<;q9UZ|YCnzV|M=}$`@;N+3cSp^X7fXx-}TvY
zPYf+(Xh1+#++poFqrLzvt6zs1S{qe(eB(*O=4-A;Nc7P9@_rU6EF@=8draqnXpv|@
zd_jAa@#OYOdvSEf%kB!J&O^@hjfPgIMGM>+9VWqK*_LLF$>VJuvo}YhMxB!WsO<x}
z2Jw6zPG8~+UYkv1g`F4wJz4$ndL;dbT#-RjqV-_8>wK%U4^dSbgtBBvvCx2?u^Au{
z;@~)KiKyd(;#1vWuaG0)tARRi0?`uzfdtSIQguuHeiZozh{@X76;HVF^3=c3zf7Ze
zIpF|qL<3JEy283$RXJ2M20q@ezFQ*@d{KV1>h4ncg9J5?g$Qp4tue=8W6bZ!oi%+_
zR}c#~<2ntm_X>jfBZlb88GaQ?2|})O_=&5NmDJvR(!KgpzOTrl-fm)WZAbIAJ6)+u
zR}G}SSFm`8TKT9HM^q#02joH&)05};ju!WHC66<mBGLqD81^;HipL{c_!$eHCR@D`
zw4+XuQ{P^+!)D`POPY>mHTL0;Feetw%cHio<xlzvEv=f)2vrTvNHO%XP|UD-Y}M4Z
zsN1fext;6I-tF{~2hz3G;#-m~s!4?Bjf`nW_5O9liRF{Cx0}Zo%$0CdG{0|Go^ZT$
zH}ifJ8!{mwAs}x58en^W)XjBte6I_TkdpfJs$Hw({e3X4SDX5`TtdG9mJ?AMwxCp2
zfmB{vTNh5Wo0`+L#ftRye9?2Uy2dA#-Q1|)N_XqvPffRU9V<mQSCuw;pDbT6);EIl
zq-MbIB_9$vFVRw{wXNDb^7kETR8OC4IpL}_JWc2)iZ$3YKvyTq-f&1xjvl;ejumxg
zrVHHHL7w=Yzz^fsChmKi5XTjb&C)<z0!6`9yxz*vXMNtvEw)^F<{eR@_rlsd_KMv`
zyaxZA>e0>PTM|_77dA}_vPskf!o#cAkfEUn@u<~Z!IZ+5_r#B09u6_0&Q1cw&^$tA
zppw%Hbk=|#Aeyx`aqnlaod#GFu(F_OEu&FWv(SV1761NRMT>j|gZwb*fPxIjI3yvq
z@oCYbQBd*vhp2e+d5^avtd=HR3HdO}3{<$le~H1Ny)2<?0cZ>DpG?^h;NcnU`&xlQ
z-`iBt`p(O@^aM1XM-M<lhacE#IVNYnx3E~q%I*1hp8%FZ4X$BYGY>a$f;at}O&K{7
zO)a@|RmT=Ko8Ld)UxNhRRZ>4$Fkr1_(p<v`wliM{HW8xnMRUp;+S}oW@%DsvSooM|
zXB{J(sm0sHeEYd~Z}0nMpK8>nkf6m+6R>FTpAOj`aeWQns7eZOJlw_IkZymlaJ*AL
zMXKW|93AF}Nh1`6FOj9Vw6Ps7J;Z%GHho?9ck*>SXYwEqHh%ZoFP=;i>!p@9&F8VA
zC+woCG_Sc?!9Hp>X!<pZ7V}C(@8_``d@RPr?DK61=$?HW%^k>PtCN<{88Uajg^qvn
zxVyV^Ct9o?8yFcq*>wsqT}hPnIZd)@cqAm01Ymt3?&+v^?{XS<Ml<h5q1YTM>&=%B
z7-X37-a>T#WN!tP-F(I*jT~gJ71c<<MmU_B-mNYY%U-0aFc0p{`fSGzE`st!$0;;x
zLT+~}-%<n=_!&PFgukW=a$Eula`Pw3tbWrSI2M5|E4sIwMD57>>PB1Z{uS@3JATB^
zTc-?+B1L6K-UNVu{AnrjYxFnG+U>a2k^6gEi1%cyx6h)nzoPTkK4sO&(AD$%0iO$6
zD`hCQ%4AQ;kM<5*23iw*OV2;&sS|&DroL|RP1UrPSs#<FwoDPDuH7LT^@3G6T>p(W
zV#6MxiGYi;wzf9Y-t7DlnkPs*z^*-lQIUtCX;ZkdahsIa-*PZAGIH!qPD~iuK+yJ8
ziWQi&`lY{<#<ThR%M&Ek-a5A0Om1L01nz~%2zh?hDFTgqj|%yM)OX$wS+GQ-I9aew
zFJO93f6F4~W@hAk?nu4W{k&2jOgDrNzjT|FAWryGQlq+Z&RnjRi%mpWJ=;a*WBw>~
z-R|Tb${5V=ZR3LD3nYUTD^M{#fqUb0i;Y8L`+M#e-giPzR53goPC6Z;;Xy-gMNj+|
z+19UaW+M>uR!<|_-q!Bxn?~eBIT#k^fnCcrKTxsK_i`q&@F&G&Pw>P@s&qy+xTl8~
zOx&4zHZP}th+@1xufSV@b5gcB^}X{ERgm)&SeN4RbPwBJqW4@EL|w~y`0j7O)M4O$
z@|w>HLtKXste(oyW(5bM%gG%Q@sFE5H^$oMAbv@IvDHEfHT!O<-y?NLWI7rSl7{!f
zObqzvR}Iy|6(@Ab^)~qp%PFs&SO3UgE*#$di1q!JRga*v;rZ`ey>gseqmDjH?IL2!
z;XM^0(q-jp8Yr(cOzCWM8;9U}A%&lvN~}GO;E9`KPY8b%pQko9iN1><_NF9=ZAI*4
zJ65W<57ixuPPoh|uoq6J-(AygeqR+sV?v#^VQ;=?)1RnVre}9wI@DjRzj2am>`?}+
zgwCEPpfQIfLfidWA<AR|vfX*v)*#xg_4a0CBO;%Memeabsgh3;41l?doDF2ly_il*
zlqHA-{GAeTZDVdEX4AUs9TlZ~ZpG!Ues6#>dCol$WUz8fls7rgLQnH_rmuzc-Cc7q
zu?`;d(S@iEk<8g*^3EKm2KA4dM|aJr>-~Vd;RFls*+H<9Y!vwZ>l>&@v$W~D^K|c~
zcSF+!R>cLQfPHZH0`}SMY1oK7l;3?-kA~&-Ayf(P_8^^WsY>*S-jxL{ILwipI{cD{
z)4k5g)m<=RtMI7blTVLncC3Ae1x%07Yj?EXs*Fj)lU|%@LC-S$+X9W>Lu0%AQm^(J
z9gwh+ocNe6SjT}I5+Z#zB}TktddRr`L*{BIUb?_jve2umUzO4ey9v(y_0*-dRw|s&
z?)j#?5r)NP<hk%^d>wmCQ9uQ5cZ>m@9PBNfnKeWwZyBG7#4-3{{HniX<aHj=$1>_)
zKR-XViLv>B_Tr<=OjMb2D`qp^d{ebppRM^&bItCQH#;>un=eR_2NMpy7D45md`)=+
zwp=9M;BXjwI{-VPh_ba<zAwaO;3XVAyniPmg^ZIA$NT9QWn1$Ft);yr8cNB%rozt}
z_c)TjU%qoP!8ABA`#m(|a&XVr8vFmd9JF97&I50Qm9=4xpf@41O2L6oJU3{X^swbC
z?ij2Fi5b>3f7pkd>oagGg`a9M9e903;z$mxy4rC0*by|56P=L9OnUSu#eaIV^p_E(
z{w0Ivr*&zI&=()N)tvglH34b-=BDNV)iW`8z~KFAI(8?Ygf=5)RF5cjNS{P@VPeYL
zA~vdQs-QcXom+oS!ZUyAEOy5}35#`?O|H{9YiE#WDUkMpWc(Q9TW<0<LT;GaI63qZ
zK51LN2Up@jy7bot?TR4Ngm#~Tf`mHkotPbpm$YK=vMLiNxSzVN&}I@6*t>HC_I)+D
z`GoN=P9|o381n-c8*9QGFZPrHH@N<_`|sHMc0Psdf0DcZ4DLIW?NGljzbZtRNTmpI
zxr~MKlMd;AwN7we`<EbTRZ)}`DPwr7x7l)|<6|I$6-+9$`#}8lX(7+ecYPctij|`u
zY(uTszA3*y$~@wszn?01v(jGD-_~mPC(BrZPd%O>|El&(_$nT2Q+b}A1VCP?E2QlX
z>>p764%`}Cv=PINC%PL{3Yubaw;mtapIAT~p}c6|yC4n^HL=5|B)Dx%rBT}BFF0IG
z&2THstCZ4N3rSMY>1}Nt77MF!hs<1EQz9~)oc=V!O%4tz=Iq~(u%FFSdc>wt=7bUJ
zSU!}&H91^(gR>Tz%VUlvDio-lO3GG1_FoblqoMMbw^vYNnjoj%lTQP&n9bDi0I-8i
zzY(Ht&y~Pmdl{|dxUp`nneX-0mMgBbR({47RnI%=*HP9`k?m;x9^F&z2*WCcdn6bv
zVMF7S{bn!uU9L|bZBdq&|KcGJ1x-NgaPoV3=Y!W;$A{ik9=$uFBblhPrBLz(3ylJC
zNbQ8gx=|gQuGtvyAB)8{k5RhUFn0^*eOy1&Ecxn+MlmeTgR1JR=GpnwszCkuJ@?p~
zmMo*y`MBZEC6DR*shr+75t0VGiwe6-<Kt8It5dju3HjXFtIFq0SCQ2Dg9^FHYGzhf
ziU{f1A1Y9k;N49A-yI<J73wTF+IU^<)WB$s+uWd6gsVg^DA2P;eCbf4QB<7=iCEq{
zdzLti+V`OkA$^=9dIj1;@9DGe%uoDE!jWjXw>6ig_w;bcfphH$Z!PmkxZ>qh$!f0s
zgQBM~jgpfBH%=KjY;SGQhfwC<QtL4LvZf5`r3X;?*D#hU!B^6dY1}cQ4hH__sAHz9
z3J^<7F#7|MHT(g;x-v_AoVG6s(c#}-MufnHyiz9nOJpdD#Px60_{xnI+vSxa^<TDp
z;|HufrRb=Lj2z6J0`30n*LdXG)#K9tE`5s6+{PY)`tY}wtwFtFh$A^6gIm&24g1=}
z=HZLVKyX!HY-?*n0#;&4L^)Ucj!?y{gG&F;KE<%rOf@Dpv5lI~Iuo1hLbsF5dwOjX
zQ(<yXTDtTx0mFZMLQ70th@qh5?5{UHn9=y=be9w}Gqcf7aN~R_!e1rCL?NJ76`al;
z=oeJFTwLftpYbU(YE?=!=DC^@TCPgfrQ0G2zPJ9EH|$u<X^+u8{0NF=P$O?G{*;>0
z*xP{tuN`ct3*R}YAwV3_;GC8L>($A@p;V0;d}5<{Rw`t)=S%t@aYk;&xVofG9b}}P
z8$GqooeO0ph3;7nuci6mxel4G@Ajdkl@y-z*vl(HL57<Om-%`O9O{-^Rs#&gKE1g|
zGQ49v2b6AQ*54q@WEK@-)Mx5nsRyNP59~o*$0Fs=f6HYN&Dd~B5<G~kc-CWZQSDEf
ze;#XGeDWSe+QOG&AmGkZ5B{rH6&yOR^<agA&rU#FhF;;kW{yT&=eJDXhbiB3@O3(N
z>fk({LamH!&s%nd_ud(*Bw?FB?`|o==FoWdA>*rM_!lv@t+6^WOss5rF`9n|nZGGc
zW)tYijD7E)a%Z7niHd%lCJ?^dlpriT>AFNhZcsj5NW#FkWpbYB5~EpL32PylnXgxH
zt-=NgR0Fm39ZZRZQroJwsOe$W2AF|p4lh+ckJs<*`sgqlj)Ysv?Jj@JfHTJfDCbog
zJmE!ts~9U@r5f1p>gQZR19#`>kny;F#1WPL_aZ)0M6s5D$T{|1uC-+!)2eGZ_pU}9
zmXY=-$MbK<qQdA+^Cmk#)S-!$`NePiUdhJ-l2P}+V)zSLFnz5UZRw&i^yK>0Mxr6K
z`E2^Ces7mpW1C%fw=O{!3gP==$3@1QBknJM9oaq98i|R7<>t(qn|LCiWHyWiyI>#f
zJ6CesX5GS=gn6me(UbKO&C?LSX>aD^dGqz(GKn4_S^qfa84MILfAgn`d%<wPHM*+I
zks;0HwIDWbdwo7p0^S%?0>Y<$j44K|EqCnZ)1+huA@V6Syi{`j?#UGcF*_mtIQ*IJ
z?%I&;Cp|wbI@svdTn_0mT0F6WyIo}{2I&xfc7^IY;Yvck!i!hK!{j`bjW_Zt<UCiq
z!jzqb?vVj#u20Y+gV*?JL*-g)Mm*=xhOI2cP+%_prKp^IYHn;JBZVwQ+{mDsnxR+U
z*nv3;o*qIJG`Yw_DRqidZm|<uU4}&e^mkGrvaGTyT0}mW(DVB2czf3s`i+@?K9&P8
zJA6)^;_(F?n02uFhL$)`ysELFn*Ay?o0zF{as$I+mCvuAcP>P_Aws<@0R`M1|5F-u
zjL&?Z{pKtgE|t(ga~z}M7~nNvRbSkor(vyg+du3~4H|o2f>PI7*xv3lX&G!?wmwB2
zQDIN~@Rj+QZ$wMP!mu7f2~~2QQxAUGgykXzf)8JPF#~vZRV@8EI9igP8gSVm4r1nf
z-Hbo~dWqJ;0h(?e*L&`kFe}vabphK_^ll7a`PydO`uWWrPL8`f`*-*tZMmDW2)YdY
zCm*JYv*qeF5$;vJz2RMP9CQ5w+=28q1M_QZALVjbZfD<fRZ0Or>FM~EU25-w(Qqe+
zhUbmK54_|8raaOjgwD*GUtlK)f?Ecp<C8*W;L9uGSAE*bU~B7Qy`gOCxcq9O!zx;{
z`Z$Yu=G2xqsB!AmE;<xA%wN~%CsFKBVzLBuhygx@LOFUhH!kUbaw1Ojb&kE^YRtjJ
zxX*Ph=fZV+v5Pb3la9cpwe^;0RR;|o&!?YW_q7Z!arP^(8%=kR30uaH{g!Gg5+m@E
z^JAeKH{T`&4V9XgN2VLff74F99u-A7Wc}$=oOP3^Zz)fpZY<w$IU@AxaqkEbM~47(
zt=6$m(H>Aj%n-5Nsc-we`V5yL+PUsE#~KWtMhc#wU;Tej^}rbsi%B&@G1pr#AqZg)
zXot6qed<&vqMS~yTK+V(emJkdUBNdC_vsEX%WR_q@$H`T56_?tM@P^sQtH*A6<QB2
zIn%&y*@dXR^-YnqS8tzBhez#=yjsq~kkezwO147>t-R0+@)X_et?geDEFX$(Ce&^;
zN<O7ikrh43nB3QuJTX}7wsS>Id}^fi_pk;nK{DBVH|y`9>c4;ncfN}*+am<(yv0cO
z?0sDpHTYIBh3|NR!($3C$_!CuZ+aVKlOW!~#25D-<)^*+OmphV?FSZpdgZnr^T7YW
z1eD7yH>CE;rHmacEfZ7Nea;Pzl=zZ+X@#H{m?StuA4MLOX*vDpo#sO#0Hcfy2AS@U
z@+QprscN26=mLw$1pm`4EZYB0F~ACdI(IYp9`NV-f;ihCiB_<rg0|SY;15i<EZ1&L
z?}{+s&{|(gy{n)B9+fRby;?e+BKOuZSak*-|Av1bBKmvU^ymy{-}$*Me7}Gy8a2{_
zWk=7-6!Q(+i(5AovBnBtro0uMXI6uS>Oc$IVuR3{R8|i`B(m8c1Up@GAICbt#g@f7
z!P#9o$6mHm2I^Act=nS7RU^+3{y=r^Y#nB^IvPuJOK!Nk^?Z%0ep!k4Z$KDud`k}k
zO^&751WxjX;xwBBoW@dyVSi2YolA0B>pW*(wno&|l?TvK?l&(-z{e?)e#kfjNZtY{
z7?}XPqQ}XENz9Iz+W{?rrLDy2do6%E=v@dE)vizEbZ=N$eqmknI5Qf^CF9Z1$ZKhl
zYieqCr;4Qlm5Rvt_}{?Wvb)Hzm6H?;i0veDG@SVYuD-I4&RYcqyviA83efY<2AKAk
zMah>h_bzuE9Tr+pfiSrI*~j4EMF5rk07}lwMnDLm4SXP{p|RW!_H(fvv?LkJs@Vdd
zrr<jtZBNCuQ@e+PCVht5*zE?;fE!)jH@5^+h?skFWjAdMPi&Dn+2^rx30a#Tv^`U2
z2TA$f<zfB{nuT<68TeK9k5n}Y(u0!RUJNtGk7KO-dQiS^kU{L4lAMeXv<KFoAApJs
zfNvIc+F?Q@>@RbMo&w`gjLM|9^Z-)U0WcWM)6_&n2pr<=c^@=h@6J~bJN`ka3?It?
zKBE9e{p>?`u5BtPPZ@A>n)W{g{B+%g288|D;?Mhl67yP8QZft0-s##VO!7&e=p`!+
z5VjAP>Pr!72Nso@z$gkdk<S3&cupNsQWAgp@-9#Zffx0GzdBLp_6+>6cL2WzYI$mc
zdQRP8no~<7goQo75N)5AckK4<)OnSHS!TcJr5~|<wL`I<LqDbRb;o6zlK<*W6!$PA
zb5ZoWG?t0Yvs-V%ze`qJ<JEL;RHr0_+l6v$haR>8cMV(XU`V}dFaV{$Qk_kWifRSE
zrNHv^o6ByndY>n-ECXuo-T;AtkSxK##C%E~2Aok3!QTOX7A<Kn`~Vj}@8k^3!~8-J
z%?emaq+ybv3<=sRe=6s^j81oE{QdoHH-?!2w2PBGjF5=P4t#L4dwYYcDl(~w0M7tl
z2m=$yq`kzr>vwPOFH>$ncNGVqcwnLQo!h|~AQvs?8d3mizQko$&ry&H7(fAN7(OZK
zZvghFzP-8JO!nh>r*`VfgPLGZ3%bUEi(@=UF<JrZZz{Ny)lK}DKNazkNr!emB%#=2
zw_5nDhVt~hbF5Q_vlo~C+)onqb%~+{G<+6-#yxTVFWr3fkmafFgsDxlcqdXX@4SWl
z3nliNmPJWj3M1f>3?PSU1zK$Y*&H6F+~zO=kS<7l!iR4IP)zmosKG9xL}2iRUGdP-
zgAOhcVCb#lcCf4w-asCQJ$x}AZez(rJ@^q&pR9M=01Jeq0hPR$3Q$A1ECAi3fb+xE
zv2#iQgmVK#BBb#~NDK&{vo!XBUKK!Z?9JWYRJMo$$brYjdd_lnnKfvV)dGT|-QN5Q
z4-XFnObfh2&;}e!4Eo>9AaWisGiwWPlc<|NfA=de<J=_lP_NmW6!9ve#e&pMm2T=`
z?sYVziotcIa;>qBzXygYTwa%1J2hgx`8O?wo~Q{~&y;8KIpd1CC2wLjOlo=W0tacJ
z(1+8+(Rih_eghC=$a}zeM78P~@LbwWxw(KR14i3S-i+tz3{t*+W5eW9<H@9jDltpf
z#VWO!R~?RtVkw(8RDlH4+1J3IOthS-A@cr2z@WzXGj{*z=+AmiDw+X;`9fS;)1Jo^
zfBi$uX@ihn8T%Tuhmf4y{22HTKKeOwAxHKCG=Vf+<r2!LI>_{PkVI8I1mP%?LM>%T
zq+|Gc5kAV*cyYMT8(&LuTOa(}5Fe$>3bUfv55WBcy-W2XSFnor=4JS6YBztN6~<{Z
zTX&Go`W&oBw@vTHK6<t!Lm@!VxVhei)XBz7RSMOFoi<d;+joHUSWIOEjo4S<!*>Y4
zPoQK5qjukVYTa=D;p+N&v%Y8oK~!K)*0$=z=*o}``{CTZx6tYaJcRij*Az0^@}Zr5
zbBqrpt^k*sXBlnzJ&=(cahR!Z3JX&*$deqv|A@!4)Xk7>ZSDnqf{WCVV1OD;R5hrX
z4;c2#Vy71U%PP_(t-Z1^J?IF<Ib06!5K%nJA{Am_JL%Qhq*ce&t#NGb>x%$hgkKU9
zg8?oXY0mSM3JG{J6qZ5q_V!$9gmzsR)g#}{9&ic?)iT?hg4v(0cBtEH!35@&$vsJY
zchW}8-9P|w+o@jdJj|B`;o*4z>acm#moLcA0Tbs3m2yRsHVg&>Lqd#bFdYQ-pV$oh
z;{sV#n`W3CF8SIZiLNKtTfawL0cs^tjtAncbj<$H-{;SbYC_JcH$45IMZ9LqyB2yE
z`<@Sd)s<eB(o+BuLjw3YY{sZZU6apLcn(KeB{Vgx0Z}XdZ8mu$yH4)SwT$dG8klZP
z;JXb>3lQ3auR@3isvS&LdQ)ToqYBvCSf=#$0n7NaynbyG<ABkUbW2s>)Xn@8vI&hJ
zlXgcdN^7E^qLTISs2gbnx4(~pC+)AZM53uJjr1beU9?@FuMZuyN7fv>;{RJf+{wHm
zzIG?h0T}{#Nj7?T1bg)*qOaqoU>Y3NutdN#K>}nZB!qtg$YdRW*Rz_yQW-G8XwnZU
z;-M{2lW}$BAIgzmg+RR>;SdR$kO+rl!R@(>po6MH?aIR5_PTZY<sTg-H}$F$6;>2T
zfLmr$5-;1aMT`wPKcn4<Hm@mRP0PE*+m9%(u|tn|#T5McSY;ntuSy-g60|I<7?n}_
zw}BNo%)=5vA*iR%#sf2?xAh9I==&;Gm({v0#1%)pt&MN*zKhmT%6@fq1$?X+0QKrI
zFd70o8{ZrB>eK|h0IVO0Rl6coUC%8bEiLVrbj%}(KqNfi^q#Z-M3T`W9)KTX!FYxM
zIZNx5jMtJhEm<ER&9Y8#fLSoBP8FU4u}xk`;KzyR%vXM3eknRZMTawb4f<BAeSJG$
zo0vl0q6@BZZv@?D&Cl=ui|5@#MvKq20?Sm1U+K9LhaPLJuD@rmb0Jn?wsv2p$>ovN
z=3=oJ;)HHI5LN0cRmy82K$-(D9rHOba@8m|Cj$1X%cmqD_*@*BfzNk=JYyrJ3jpJL
zf{9q(ynXu;f~yl3KO+yLrKJUyN5kX}dCX&cv#xi%fepnW01PT9C?GLj`*7$~HGy`L
z=C=Sp;Khz`r9kq$I9dQ|H2>lrz>R|Bg(OApH3;UwG&f(F8v8*hQ-TPh80lX{y-XW-
z^2gjb0F^-62<Yf(Qp@~zvO*8S3a}B0uY7ystD6<>-^0zfs7w|%D($Ae0ET^rvKEGw
zD&li;N1U8v-8{O7{2ClHNb%fe_`tV$Bpd=7Tv9SJOTaOBx#4mr&v_aEiPd&>fLFQO
zdRE+zdNF?fs>z8q)pOt=CJQjdsa_XM1oUrn?cZuHRCNz#LH|Ekb?~gezyIKCo6`v1
z)1=^TMABs^!NkO<#!H#M2d=J?c3xkW|JP3^cCc9P_!Hw`I?zBz`P(-ryeSg16H7Nm
zk?cK+Qi!1uhOiyVtXNQ|=_0`E4&HwGBAW=2nV76%wDp10a*Q9bxuF}#<K4S=@c?Si
zH{(bEnpA)m0XyDk1Mrmz5dA>cy}*S%)#Knk1+S$6U||7b0|49F0yy;9-U4Dl0!hc@
z<m3%X`2sbB5h@O~@NeKjg{Wx)!#2=~Jy^?#u%B=8k!B+((_#ZL$*?Dp2RO9hGcYiK
zPxwK=qoOUCh*HQI3p^VvMP1^jC36_BpL)(vjQmoelx(Ub;!VZcAK0t%ayiAKvxF@S
zx0LJEGI@H-+wCh*3NR(CVanJZIA^@NjCCZX3{+3E*?#d~0ZH}1a<%nIp8orGbTPa6
zO)Jt-oj{(r7@-uawBJY<jgx$5P!KBDL7ThT(6g#fT!oV$cIxkX0->wtVDp@#SUEkQ
zpwM2r1^`!Jmo!#OVm0yDcw`-HQA7@rvh;n#9OP~7Rd!4Hv3>$iE!^u8#R1AemYx6@
zgb9Ik+!*lwh+xogMJhOF2NZ<wctnShK2sm9*DAwr%}bF_fdWa4N%q>?#Y&-bwWC+6
zB=LN=CPV<*@zip~73D(bvi|G8u~ZRH)bewJM_m^(zMJDE`ey5dKt~`|Tu%$fxi=hC
zYYF;u2vUNA2tW}fmsKNG;v@a&L)6*nX&nUM2(TI{9@l|qFyx(+yQ!);ddMJ1pcLFZ
z2F=zzj|+JuN^?5O7TwioZntkMocDp1UQ7hta)h%V6IS~|hRt;3^Eba|)kT{apI$GX
zZ=@csiI_1yxNU%UkhuPfe~Yl?$w9w{sMq~2_A|uD1dFiu$1b7tQJ|Xe@nF#!u2-)C
zf`j?7e$WMA6Cl<bSZm%Ush&`5xy4us04G}HJOKk$$y9p4Taj8)zZiLL*C2QIq2=2-
zEwSy0882UB+-~LIS_$nY^@Tkp^pAT!)`k5YWEbvD(G`1|e4wn)&P%D$5Vo7m)?&(;
zn40tN!Nfm)=m7}G6d{)bkT;@m8Z;rD?96m!N=0>sR3n|vLhc+OPykv4ASh}AkOYX;
zhFjysqjImovdhi9&3Is{OSLejGM=h>iv)dWiRGr8PFYj$dW^NcA9ubmA&($@upxaw
zm%yyua8bxJ@aOYzB`6!kuf67P>?weD+5R1+{&;+7c_glJY}#rKBGAnhM!-xX=DmCK
z5J=Z@a&tcd@{7H_eSS(fcx=T_pvXF<f16VR+kxIKXoz^-Qg#GZ+Zp%{*a&;q%V3B2
zCBC(EH4~saf5kaK^Xu}3uD1L6M5}g6^F*MhzV*Ujv4c||fG7T4IX|&m!2|##UW`pp
zV4#}7%0|F&r`mTGX6D5Zjs~d6SrA0jmoYW713^CWv-eTa;vKZi#uQW}Sp3EBu*&<B
z^gjRQ)7%&S=)DS@JmNtmOR&FRasMx5YIEINT5Yq1feIu`cNf{~Q<8s?8Q`8Y0qzN0
za${`(93{>suW&=&lq;dL@$}$;x$zo6D*phRvtmG606!OsK@)QzDQwuWgk4jF>g)PD
zb=Zb5_}AWyNIojK8@|#+L-x7gwN=JKcqPq4j~2G_Ut%{)dd5qsH)&5DRA=wmXngq%
zdV8Ovl0R9<;2tjiH)wy;*nGWtCwu$!+}zy3Z7(k`ohn;oZ(m>EgT>vEfH^>e))Jxu
z69lD80x>As`IYdihWXcW*Y@{#Ga2=v9Hj?{)|MzAPM@<V=&B%yE&Ry5%xe}e6al;w
z`Rkq$mz|K@;8KHjPXyh6r%n0<Kt}Qbo!ih*ETrI&kOA{@B>#Yb7!p^wPPLNWJg|Rf
zol`HUfVCr@GF?s2hb;lxYvw)}EcpZ$c$Xr{Z>iq9O+9@C0FHn-%9Jt31lw7=J2Mz_
zkWi?ai!Z#6oCc-NQ{0w4hyqOhZ((tRg$1;``vAQJ38)`5&CStxPg&Q7fmn0j8w5j6
zvq4%U92!xb2Afn-&;Qn#N$;$+W8*4BTr=l@`F|}57LIzy1zue3E&`S?MQY{$%_gJq
zF;gjEZ?VjYy6zdAp8&h%&dMCwq+dWn5m<)llL|Roo^JyhYJB<71At8jF^it(0)O>Z
zSYmO(m5X2yRHn*z8DH{jB}cPpFmh{9=Qk;k4|TtDfV6ny!c(4rlqoVil|k&u*unpL
zMdCUH`J#R<;NG`9rSGwi3W(;&P3`SLpmLJHH|G(Bc$ZUM!0pf;WYe06mzf*eVY)b+
z?FUede5)s4n{(kBA01AdvzCf%FJ#Gl)D61mJC$_cI(0M8vwSn(aUy>xEQZjUjI%21
z^K53^&xw}_i+1x0l5dvrfJz(l%p3rmeyco9Kh69$E|93sPyD|HeuP+CE-rBUFFf93
z`cMiY&H+GMWNNCu*Cnu?J)8H#?KfrJQe^bUbBea7|Ke2I`0!lC*WRxb+tKk3hY%yL
zxcvjSVgX7$s5<J&-0cfEg!S%cvU^c+Wy|u(VqjWv={L}02iO$}OHoHhF0t_&RwR@Q
z$NZtaCwb2M+JDT0OE9A;w9LMx_#FuL4YXFHM9R~B_F(p@*T4E@#c}P|F9b_UMFbth
zD)~rnbd18GG6z@4QrZUrnajA7P)S)CTx6S}v9YnOn?Mj4!q*DgH4dj40Q@mJeag=&
zEr??L$|~5n0te8~v|@hxz#)e9pKkPCSA?KKDH~N)wc|+rppejYjIK5R;h}FboA;6c
zU{7Q@xn7b7C1Oc>c`a$<c$MlVh)^zIpO*Pn;6N1jl3BcC;(pJ4ZN}~M6hBA^fv_$c
z?K=Mh)N&V(`O#Q`N?nyT_Zy!|Ui0$AShi(*lSeO?WQ6hiHgZH;(w;T>6)dC{<TX0Z
z_7LAJ;6ABeC?{p75^Ddk9^z_y{+8QW_Fzd_#{?!v`Ew@%`erLb4uCSc$(tliroMEm
z&37w>MKNNsS4PjLW-k?)pq-kD78a^``mrDAKfF)E_uiqCE@0cVms~ET+S$6Kg8*>c
z_3tSCh_<)%ExR}E=GTek2G-B8&W(`#dT^z#yE^DX{KBq@z!M)<2b90O57ERfIjBu{
zcnDqKE5_)D5)a1#UTbOMVkZ%;Pn8=NfmL?1&gU{ZfsRSvuoYhR($7cT<xlieH;{qd
z7%7CcE;%Q^KW!KD%aC-dS5u$9uASZahe4)0Kxd~8hr5UWw-_HKCIb*MqE6-tZaIiw
z<c^?+l?xXWn|iCK#XG%mf9YzYvJ_nXcJH2;pKwe`s=b4?(lRBc=$+y|hAef&s+3UV
zQ~h(NCtaAiTI%>U9LU^B(CNK|Gk)Fl(ffUwn?2V%U#UL9n^)K#XLhm=WI$<p|4!zQ
zIcLH~<zEoX|CSgWPh2M?6n5t_)(SNYgW62HGNBe)wL*8fG#sYj-#gw_>r>=bW+p-~
zpB7wjy;%fZA$E+}@PH)&gEQ?%JoR@$eg+cRSVtX?>G-3O#yyPu>H@{99p%h6dM37q
zXt0F1RV&f{Q6GSMt8!edb9?&V(JRhfW0C$TQ!eLD)2NzS4v9E-sK10A@M&TU`CdMl
zNVCREbGCvH>hxUiPVQ=cmLC4}Ycvo+!h)<|cZ;o>wGLd~H-%%__IpR9a7kPR%Q5Z6
zVvvo5e2m*^ZELR|Jp(OwCW^}Q_|pYcMj9_*S}9NPan_#jw&?k@JOAC5{7v72LTOKT
zH|rdtFhTI)vA=0AJC#VD<xzgUhFV7K$vk|yGe8BS^w7>0zM8JN_$x~Z8n~X-z7H@U
z?Ox-ufT%!A87u*PEAsgGn!-j9>wd;mmDiaQcCN#kOuMxHoX9k5S8<$>N8;YtEy@4R
z|1seZj5<(vioY4n=`-Mrwt4;fb%vVy3~GxhcVuj=BqEcwLg21x5oe?t09TeQSeUqR
z(zsWtyo9XHD+pa*)lUda4!I91z~sp>=m`P@P{Anxy0I^eEN0=$)5Dg=D}|K-a2uN1
z#>M}Qwu&9}hOzJ8XR?7S!~l_kF_u{X{brQ{iPI0`{ur;b_eKGuOT0gVp9XNoNeB2P
z5#U(M5eEP&VGJ>5Yy}q6^{rw5I>Z)&$ou{V!~q?GkbEa#XD4=Z<_8qWv*ljh8+a?=
z9;a9HRZ;1%sfICZVbN$|c~4y9Q)vXnn(Sv_^{u8i_1{jgn<+L$+XZ<#OSvhqC~P(5
zMzIBs&j_~U<qs8RN-C=TgM$_j>LR%j(N{F(Dkdxns<zF6pZgpf**4$-o)<vI6htNC
z5g)9t$!l5SCwKcGe_cTrOt|CiQq&m3^}n|GKtHSXz@7_vSaw|#1u^jlDc48-MTu$1
zk4!DCIJdg|J?`+UUN0yYIO`#2d;*{ah3{;2lzR7A)<JJFHw=CL6srrPuF-^XKM-;%
zRr>J1txTRcqRkBw6`#yZ+Q`Vrd@9h`>He^qNp}=Jt6<>03&d|gyN~D-5(wOCEFT#B
zjS=0|fmSl-56^WP++KkG^n&H|b-k%rT8I<Df3@d(0|u;#&Dum_Vqz9S;B$GTSKXxP
zyj$&iy7-PRke~IjX|VyyjSKOhG@^A{QJ-m2Rpku2J7fU8adogFeP%;hH&^aeI$Y|=
z$wBQ6QbDjRj>T~Qy*5*P`iw8}QJ4H>?gUczpT0NFOmlMl)zzNbzjJ5(sR3bcA+ZiB
z!%v<x#o1;eV&cfJU&FyJO#{WSFguM${!nhy%v^{fe7Y|%F0OU+>7G4$R=+ld-~Ct}
zZ?<(%W_x;3AcdWpk9R!03GOr?O#&Db&?4P4f4x^!;MqR>&XtZ_Ddaxbt3^*eDMU5@
z?F#bvj`xdw1$mmqXLOPH?|x90y7e%Ok54|ZJLTT-{6nWBQb!>m^C-U++F&E^1oS6R
z;*tf#W;;j6Oh8}<#?KE32~`ZIZ*NY;R=N{flLZ|5H`#QqweLML9{@!WG+|L`P1fyz
zXM6jozrQb-$g8klv0#kSEzPg|{7EOjxVSXydLL8qH7NVk&Lsuwk0J5LU=GPQ*_at{
ztJeF?-lb47_$9@!{TzC}8y@PmYN#M(WA^va?L1kSyG>@d+~fIH^}wU=Swqk2cN5xk
zS6w)73Z6|LAG|;lyP;~nra_6prciI_Q$7u}zgVM@O{tjxaz#a@rAwe2(6cx$>n1vu
zftJv$t)lW16uBnKEz}E)FflM(S5sV&;yDdbK%2w-BNPqx!gNu>(!Tj>YxY><i`!P?
zGbYrUTJevVu9M=$_mzU@pYtYjAT?DwSnTKuQ$_F|d@8JRHk+YzJ4;%f*aMWd{JESm
z-7{8X>l1GzblYMdOA!#|uM;|X`yCb^TDu#kmwouR!^i5=TRil<Jnq@$;LJ~_Q(ZH#
zLCs>IO7#g#4_V~?mM2I3&eyJH8sJE`%A)s`H=1Pj`pefl);8%|Jvo3?15dWWX}9h?
ztfuDqEmRFk)v+WlL0Duh7V!FCpC3tpK1a+i96)IIaKAj+Rs!V{?%=)qm3Z7mEJ9+_
zO`n|HnI_^uPo6pK5aY<WX?W@$#TZ0Wp54q2!)r0Oq}n1q-N^tB6Q8MeZ-1V94(vIY
zXW21_C8qi4l7T&=6;N(*_3|IprHCu4<7be~+ebUY$aPb5-FrD+;+sl`v{dMovD1Rf
zGFo?!-tSh*AMS+Q8qBI+?%9l0e<>kuu~*~RqZV}h=(IVCj8NUKu!JUjwF)FN22?cX
z0i)jN`~=7;BnUWUmn-9oe`Kyct9o`3hF-Ou^$Cy^C8RAM3v_6Xn_Jl(9;WN2hP$Fx
zSZ;6!aLRy1BwTfM`@8S=@>QzI?}p?ThH9eW)V)>Z)-M3CfWJzB$0_mODWlQjqbpW?
zS*gnJR$+Xm3~gGl?dTPB73DDgmR>yTb2z6xkkZqy%R+uu-T-z2Po5Fq^0`h#z9`l9
zVV@YsoL-Ruvbj@Fkgc;b`<$rTLFV6#U(0OE(p2oB|JPf}?gIHa;F9&gU!L?~o9+NB
zzgsmP?6u8>m@`G|hZ)(%*H7+M7vvDS{GG^`P$}TmoqgDsK{~3x^(<o-Jq{o6+@h$u
z5pzGKntcz7P*-n<GTlo$soC70r2WPNY5?~Z_4ui;n0?Rq?G}Ifq(2rGBWe!lRCz$c
zNbyo~!hKg<=Jf3|FUVi52z8R4uD<KiiIO_-PXl~2id#0RR9#g4nBPj{JcqYXZ+;se
z>SJbMkx|Q21?+v3nT!pf`1E<id<4-%nLIYx)SRkwbI>eM%gV|!yS=$4=e3L{=P{?v
z8~Gu2dsVu}pVX5LSWXH+f?w&)n~y*sDi^5NGIMaqgJxtj_&z|ED<l-ku9i<Z1$&TO
z)T3%+3=Ri<J>Zcd(+I@dAhKN(DcGL%Y@H_H0jUGQ3qnG|Owgt^1IE6SMTQbzj1CVE
zd4PP02)I?44`zN67Z+D)^zsB1M^7!F)J?X^i2;Yti|`oQ0G9WH@emoW)&A*ij{nb}
zFKy!PG&ND4J5y=ZzWyE*86A`SE$t)TtKnd^8JyyhyQr=<kD)l3``Wn7hL<n0Xyjuc
z?^lZDKYr*1@dv12pUw&+fq%U1?F>%ox97?g6`dw?^`<tvxC2us(s|V*h6nq0#}@T-
za-G`s^a>BU!IKR)LQqfqQwdOXSC60;2^QITYm(ajKP&;5Rdp|~hCnPb`@fTGQ8@+0
zq^81|l(S92aC2r>Z24)W5)u+N#)~s5Z07V_rxJNA$^g;a6ZAedrYhkFhle58jyDM_
z?C~z(Tqfsw|7Jla@>2N<Fv^B0Z0f!NiS}66w^H_kDdbPz`Gew|%j7Z8lpFy!R&K+V
zyL7a)xnObwDn)=1&8u)RTWH=J=vCzdzjX`1(}t}im;|Tv-K(SC+W|jq&;I%Ib(HwJ
zI$;j=J53*zaA?i<&jcnf-0Lhgt6y24;ubOPl&Ed&|8@#C)2r-0{F8063GZ7mpXoO4
z<X0=ZEqqHqVm0?B?q-#lr@qeWOZL^@2@N*vDbc5uIu^p(TZ8JgXPn2n-XE|~?s~xf
ze0j}n^b0GDMyz4nn@>bU7qQA(T3Wy|%2A3C11&3TYGDn~g;fh9=aJJ6bpuylGjP_K
zkLD+U*7^5r*`&^8d$Z-vm>M7ubqt6DG3Q=0Krjoo_ALnL0x<&d3R$2UoXBgH2s%mB
zBA3usDdKB{WMrC)`?Fq`Rse1w2et!PzbatCgNRLEJ5vavwHbggyb{FVwj5UgjJ6Tu
zINfY!)gQktd0C!3kpQgEekzw~#O#}cVsB1)Bo96u9pP{y{pQW*KQ}DTn=xjdm75n$
zSO&xau<P!&Jre21uJyr4=)ObQS%!eUU~*)e{01cx4@PVYn^Lx};O8foiQHC6wQ5r<
zg{tjS?mu=9D}P%u&2&wih>e<^JWFL^V3XmJGN)$d!HxQ%ktv)L{h6UaYN+{aCh?-*
zm8YueJ6ia4gm~Vo=)M$m#JopqwtSn3%=~f8abTL<)c*scKuMnol5#ih>iWX-@HfSB
z%o~sa5RsDR0(mIJa53%+)E7*G&`&D`L`c^Jw9|=KgINmuLGduyqw@xA=tNP^8dl;E
zwS0|lA3s8~r+TpUEFdD6kjTSwne6~z8A<fOs^e_8sp&bQ4|@SNz;O@XoGI9;H7ki0
z(%fp0mX_8+@EqWX{s5<KK&2)tF^}~$;uW*o&l*LDsQsw{rNQIfIsL|)i#*V&{U$GZ
z{u0zen6_u~@aks$DnXSwAWmk`eqorX5mfn(oduAvTFhKbS2AcU7SKzVkK!EFsmv~3
z4O;YMUT#SFnM?a^N^|hq>EGJkVVZ{Y^BOgC;X7XN@t6jagR!4;JjdVmcWh5P%?IPQ
zvv9SfZq@Z`zQ;P6P_2K`SWm(k?2^rby_h^bb(W<(tHF1nVuA-DW*138(yXJM?)%$%
zms>W@x(hR-uN*fwH?XsXAiD44H)XqZhvBsAU)^_@hR#4EmboHe!xweO3u*;D1`)Pt
zwtk>iICRi32m?Dkq`m>Vy@||36fIrz#cphEyj*&C#*&G5VCg`;BQhB34r|#&-fi-^
zgUOLPnS1_L6l~{~uF6?hUS$`u0iMHa`Jvdk(KkFOnYh~3cV-P7_R+PkGb!F>-Y>|i
z!lTltSDq6SRb<cEgpv;}Z`@*5>UktE_M-GmAB&J}el6vvm(oqo+%L$|Q;qrhIej<K
zBQtrpyJSugF;MR@)S#+U`Hnu>Q`h_}%2U!8k5(c1Sl^x@jFxE>AbU3eq)3Os(G9K|
zF=kuf3?gyA2EWru#pyavkoDDX{K)I%&Y=Y@)Vy(MGb25nRthxf=%`slMfD;bm%{jB
zYr=sS#o8x^uL=OgL_|)`oHqjg)v8j0fseDOOVjsYO+uiQjf#dQ9SQ3Dg=J;jr!U)B
zsq`B>+zfFeKvzcB>wI~Z9o-vwOQ%Yar7@W!;9SP0@-sPF+}6(x<iVt#tVEy5n^g!_
zaPB;yiQD(~?a<bj)LOr(RdSh`3>jW|@6@BW$&O~xN;vczadC}I6K;P8f0%9_>JG)0
z83_rKw;T2;uY9`3xQC_oMx*1h1ttSPMyjPBWUs+O7xcPHCHe`@iWpjv>!TrK>M8Sb
zVs}q_`;truWo%Ieu6GhM@Bwnz52*~+mi1sqok5YCd0s%P=XGveZaJZ}+i+P3Oi0r5
z^YdH6rZIRSAtwU<jD1)W1wMW{_@Z27bYx^YurKu7ZzIXs_NeqYx5uX7-71U40;i(x
zcRu4g1|IA=GEjKJ`c3;sNu?g81|0uOnZBA^SheZ2<wQweo#qmhZYRbh*1n>5?cR|9
z*%Ecpu7Bre8cgqTP+MQizRnV&L5y0#TFYsRPD-$B!6f3B)Of=i0PNCo+0e~K8dRo_
zk^);YlhzCXW+G|hx>T#uD_L38<*91<QJbyJ&7USkq^d;n=C4Y`FsstW6n+I2SyMm=
z#=ycF23ZCK2l8YrVq*GoIm+yDn`PI{p`Dw^@IO1>6#HS1n3$x1i@PY>JaE0u17}c1
z3j_h|$p&ckx<#X;`3jLXYPVO?#3uT`nUStaKbcDml%C0mae6E<wepL^{f$?+>}I3>
zj$W9%drC<Ufn}^UoJ~sWaAras^A2KQy-;fGwmT!YuvQS?b>Z+6?M~s|6*nWWRFBB#
z3kIbN(gM~&x0X$6vc@SdoLb}u@czs*ANiq|3I{oCWTtUfY|4H-0H3&d&X#}>?nk>0
z>ieww0w5ko(}n>x7`|y+Kd{j0fo^I`Q>_7N>6aLpCKf4)K&$}(z;T4YZRG2IJosO8
zfLVgg`=EE`VIdJshl`C>bLi;x)Js?7v%^=YZEgbJ+00&!sZBT>X7zqi@`ol?*EL^p
zr6jjI?R;%SjK%w4)u!0INkqP5*?wbpmpZ0P^CDu3MuSfqj4gYEha$IK)g@XYhE5r*
zP+iA?SCfrUBjmg@{T*QIbRv&_639n)c>rabXj%jStFxQ6X0;nMtjpeaMA|3RmKD|=
zNJ|ZZR|YQty4q{WzL)QdHG=D}K_+%AXD6y+jSQz9DbNNAddD#b2M1*UHN~b|qXN(w
zgP=ehosv=jdig&TGbEQaQM@tO4L4t$22?=Qyl^o$ua?m}CaF1LL5DB3GB27t-@a8I
z{e;!`3k5MQ*B+iqEad{2tmFGG8ZDmd4kfluDS@&VAq+JnOtY8cwIv#!Q<7zAvaNvj
z%I$UKDs}6hG^>^)M}-I*yh*Ymu%mu3jz9L*e;fsgXlUR7vJ}%*XZ(PS>^lSy1+OAO
ze!<X_EKq72#T(T64FC%_fsTPlL4X91!lkRG83OORW8G0YQZAYQo5@$V?B9XTl7E`H
zDk4j@3e5Y{e8{;?ZAG*17(7gn+6g07%{u$AOGo@J*0;M;_^aBL@Fxa1d}Bd4Ql*D^
zyGu>-U84ApueZ9Y$Y5{~(ks-vXqyworN_e}Bo>Et#hp@O+3YTIpLrM;yBVr=CP~C(
zWEim3y!5isaJ(erj;LL4f<^r-3acd}j{qL-P2^6^COaxBYRr@eV&(z;IR~^9nR$6n
z%BDb`^W%jdni){5I5=b(D^I68n0Dw}2XO;HMAFf)$yK1Or(h{E)lGxz@$uS#1ZZ2S
zfrKiAQV2{WHkDx6^o9sc-G!=v$5|0E0J{K%B#VWX072KiG`=8oh5JOjZl`54J~EL*
zbW!^Ip41qtb}-&B+KvT5&$oz|_dGzM)=Rnm;&vrZ`dD!C@RTp{DVN&+<L#}ZqU`>+
zQBcuGP<#YQ36<^!X#@f3uA!9f?z&M#6s2RNySrlu73r=S8tLxtcZ1*W?|siX|D5yB
zS&L!Knl<;_J3et;*Z$mluR%V#=J&@H_?aPL9<eiI*HDq6hvm<!yL7@VDm;@{b(>zt
z&g5!7wV)ng@K-0;^|V@N2@4Br5LDxFf(>Q4tjAPM%CvwTBCkO&n4TPy?%43l(fKYo
zs3F-0rCSz&VbmT?d8z?Bi0K9gNoy#8vI$>MK$ig+%o8$O_LzOJA_`f6-A*h<OEbV_
zst6bBYkUdT`uqGI2`x8y&XMYA{R9<me3PIyW~{;6Gi)4OoDOOnD(!eV(c<aLVIPC7
z`6;Q!CN>^*&^IwJ@Z!N3cE9bt3CG#%5R!cV`UpGj-Ei-&#@p+`@KxZnK)y0;&?{h>
zf|S4dT)=jqY2l*dtt2=Vt<I|%w$ddNTAGXRI?P9y$*_Z1v+d7pt3z{q<QhJ;pk@Jn
zJOT>zl6ES=YHqaIH(y`0YZUJRwv+lgp<dh$1A>OMz){U&{cIq^JYSPRQQ2hDEdpJq
zFlxD2r=B^=_UUPTz}B{j{B}8p-#sBK`jfx|*(P3}_hePgQ1g@xEHgEqf_h4U1WXYs
z)ja3hNB$MfRqC`Q5lbkG!Q^9i%tHOR%~1g<?v=N}FDx{a>{PQ&4X@1@twY?%>7n3h
z2iyZ%m6&wfdc$QlAwg~0Fc7k#DDUy%MQUE2c9Qn~@zd$mrj<Abooel{^@$o4EC`<w
z@J*(H26Q|>Nd`AZ2M5l#KYV{d_wL<W3zH;M0)(*KuoW9H0zD$4O}WoDH4FF&FYQHB
zQtrG!yX@cJb%C+Zn40UQa%N{beZt3D;zUh~)SGWrHcw3?JE~}j-7?mWSj9H)9Q*2;
z=$S}eKQ3Kvb~6{41$t@GwUFcT(^~LysL|<Y?@<(aTims{SIQLij*^<%0=NyafLrnE
za2^c=K?{qE7Qpmt0c)j_0su97)auYagEkm68x&T7bD$ZeHc5}b_Pwb9hsglbE0PEy
zD*&uE4?M*jP8rv&o}00N#{8fHhSEqlEE%|r{}oyZ4IwjR@|5wE5P`9J{e>f9UFwi?
z-3EL6X1EZGA+$Z@Wj^*e=3I8w$+W06=w_(5??<$!{|UAGHEHlNLB9TUH7_tID5}{Y
ztldaiQ85%ONoGYxKGF3)81~rjm-N=9?)j1Cd(9U8;WU8ief~Unru50*KLLpaU^GOO
zD0?rNojt~>S%s&TC1jdrqn&y31kA9UtA5f{$mFxKgD&B0#HGG4hWN*7zxjbr?Jn^B
zF?qVqltUg@jbRD4|6l}P%v``M0VS{h^zdmWz<Zs&y$Wi%s{LU9lYNjw&B5LT`BMcy
z1TfuD`C2M&H2}CblR<-k+WqPTMbU>wd}vXs^}#{Id;Mz}0$Ka}>gI;`&*nEfqc0I!
z(mfjE@8P)z2Lwq6+knJB^tuveKZ!{{I(#Ec8X-L(DcHm;*eEv^T1eczn!Q!V1J4(_
zo@TTMsbJ^dQ?UO^1u!FC5^O3eDsyDM)-{=yEvL<g`apXgk@Bg!@q$XF5yM-zZp|+(
zB`$)>s_TKlS-ae>*sQ!~elifDo$=^Ux^*{D_$8Bmtdj_FrId7=N*TlW+L*=`y^_Mj
zhz-;5Rk!D4k<8qobgxZRr71r7{vvONez45d=(Fam9GH&W<mnEp)`wdHh_ai;q3OY8
z718fQR#FY4R?&ki`$qI=QSt4)={_uzg~W1eW>eaOQAc}hD@%sTI-iGFH3RqI>?8y4
zNz5%X=N;Ou4m6G)qjJu~P}Ld>(TZ{Cx=*yKVtw3pIWh{R!jb7x0RQ8^Q;rlb<(-__
z{p94@#*nj6k>IqE=SugCGrNxLCcmQjLdt1*oW7VDO$CBO`L_2ixOQBv0Vz;+O$c={
zPH>TZGi=ATHTX)Gd_C(ktlu~`ORx3i!-TWk*+%Vr=d$QE02Ia5A83EL^#qlb!A$B9
zWl?3Vxp;QxrjLqx|G1&6zWk|_QhVP1)cFMWwzI)BTLu|l%KH2R42h4lsU}c)E?1#i
zE*`?vpz=6xP`7?~rEsd_Wn%6}(tyas&=)evYGD<*+3KCAyN`<9X2@>b04?%S5k?(Y
zbJu)VGzkG?W9hdK-v`w-9(j~#5~l1q(oKYF=7cJnD*kFAsv@ez>nAQ!o0l`||6<;>
zi=Gc$zSVfupb-pF{K5D;AHLzdd&;oC??LWb?8ltC){%z)-VqIrmMY}d6gWEZe?EB=
zZd2Wc!b59$_4V$5E@P#=rT72sEp+qs|I2ldJ^>mU!4tTQR(lM9Ls6Y8%4QY+_^aNB
zxA)Gkb=+Lwi~e}~)(`j4>b1}63R>A4P@%?Y|NNj0RsnutOZ`~xoHabhB=1xx_ureZ
z6Fx>~L5rl&Rxk8{ymOJ=I@IvNg@#g}P*q6qhc|C~W|#L0jNkeTQeHbX8s#)FoMxk;
zXq3Gb__t%9Uj4vATcF(7zVFR%6d%0xlF|QEh6fdXsv42Xbj8Q253>_JT`cCls*{%F
zULUkh!d4j`iEW@G`Fr!barx*XyltIr{gG!=t^o+^ri}KDbZtw#hpto*-LDsu|5A8Q
z$=7m9)7`^8IH)&_ii-5P#JUkK&_0P?-}+D1QiJM1;+-7Bbk5qECrTe4$=->}fX7PL
z=h9!81)2yQ=FB@pK_xf$`%w(6?nftf%N8<PUqeQ%V4%!CyfJ5<CXVu=DomB<&6&E@
zU@*qhjsp2>5Qc{K*6U6k+8<*`?tIdFrjT!Q3Xy^Lhh|b=BfR{4aJo*D=;mj$t*(Fe
z>w3+}ScOppe|mkL#pu0N942vXCsv=T(g=<^jc0$qsCx57oJz7#L7wUQlGtHCxMoE9
z*B1*8wN7S}WJSY&yG9RZ_i}*<!;Q*D0YctbWA9)T7M?~?sS+~R@RT@=kUH>=_N|I8
zI`Lt#jFTP?Ug|36NXl2U3_aMdOa6`~r99Se)L`oUW5ofwa9?M!!sxx-enI?mZ|0Z!
zw&Qa^tEi(=8~NyyRJdO)IHQNpJH3qBnypccNm#qM$|=GC=omX74XTr&XG2Uf6M#bN
zC}MP5qR|Z(knK)uzp<B%pDfHx$IA!A_XO^V5jHG1y_{EFNqXeYOQTPauje)gIbhi7
zxnNJ13H>$<xy78Na>#x@b1y2S$>zxUezwRlFFO3|5e~rz$NDQ_E|H2+EA|-r9`akH
z*!R#`6x$zfM=ROcLB*h6)LT~~27~I}S?x#6CMvi?siRQ|Y{7O>dsft>v#LFD%V*H0
z;x5qJ;2z(yVZ##y@@_NTdZh(+UCGZHOna+<NGYW(BH|mu@ScU2B>30ThtX+$^{>ON
zct7Fm*a$0kZj8GaB_ED%!6V{SssrIk{cY{-(T&Y7zBV>PsElM1Fd`7x!Dfbj8h}pB
z1AZg$W1t+l@4l*uSHNvw0yXO=a}C8*V!Qhx^V;l_hAk~_FaJ7LcK*#=|1wN{GMJhO
zWVnqs#xoJgz4VP3%p#%!5Xaq$L~g7?G3VdyTfXoLL`mM+FL*$QP#gT-q~SxTPBmBL
z0&(*dk>+b-*K_GF5PYJA2rQ>pJ6+LSxu@wQddea7D(v~H14O<>@XyV6=^uhfv&~*M
ztk3hUM&HLo*!*0`T+g6gK`)UpXp|{NcJ&HbkNFQ4eHGtL0Q21e0Yij08&y(L=s8Dr
zM7FlI?h;Tb1yyn+g+{Z;jnD**JzwA*Hl3HV7(i^e=i=YohA^tJ2n)@(BX{G63?58S
zv`oS)H@@b}z&T}(*vGR*3-nDt`H}R4m4Q`Nx?Vb&P_>Kc1_h?V+P13Z8<LGC_UsRa
zufvn^f(){gJ!ht~W1fOblN2ikm5ity@!-8jP<i%60z9X+&k|3W>-?PHZ-9RA1splu
zp{4lxbD`U+!~t=?{+YTBs$cFb%3eSY7}~pG3bizMU9Qyuoy+(5`tbi;^$A?ngA2x_
zWD-VSsV$=q|C{CL?{jPYZ4Mfm-;4hS{F}Fb<M03D<D1U^|M4mGe{azL>pSlrIiqg}
z5gZNeJKx9qB*VO5wA4oY?=z@u6t(ap!0Lp1bZ;X(9WaF$cOjiVz{)9WlCY}n{yIow
z^M4$K3wTI?mAgA8CU&?KwLbhdc}IIUNI8*|829d-RTCe>T;zj+&dYdNS6?_Z>!pi^
z7Ei=b^;i?#0=?>AO>|3iIQXv#oqOy0@?>OYT3nnSz{*U5zz(>%0E&vgew_^{+BM_?
zhlp;j;ru6?{;;H%`w^FzScuqaK(rlv`SK-J7;0GTO)3KA6qcampFOTyDEqIEp`q2~
zfqf8x*aSpTHBe9c<muBqu(J|4q_y3o%p8zCVh+gU8`#l4=UN+cov;rm>ikbqM+;Gl
zC98Y)?wtpaj)uWw8EQGo;($JaA45k+Uj@oiBB*AT8UPCHyxmJL9j5;bzgr==<I+H(
zF27WG=Yq0?#HWzF{{n$tT~_p>!4<|jl7IjG(<FU!1u&x@{Pc2<rX0cZ5^<C{;KX=L
zOiaGD5)fs<qE7SFqC*Yl6=VT6z+$fP(iQceak|{&TClGvu8}|4Mf^T$_b=Q&NL>1f
zu`yKC(Am5<BdW1o`^Twv-t{MA#E+WoS~Mw~Eh_qY5KCPXc1==K7cjDrcat4zpWV^}
zD{&B;VauCqIn<|T4$h)qNR#L`o8M|=Y)r??JIZiyA}Zn%TqKiiw`<M4Fz%&`L651e
zmslhtf)UoVMkVhsve8j8?yY;5-nVW-M<4Es`7sGMG#xzCxv;c!R{J0_txmb_N_g-C
zn|^y|ry<LW<nX+G%SZ9@SyOIv4>Qu!m!aCVLZB^md+Fl*cLc>V^nR{6)Le7+=GN9c
zaD>3M#3k*d_xklmP-Sk{GCnsqXUrXET%uX0w(7prx`e4P=fr2cyBE=p@+b72oaqZJ
z7_V>}q$ZW(OWBVnQ49H|YSm46P@nBbJp66P`HS(QQw`=BXJqT%bBKap63W-jI@U!*
zg@(V#8qE2=Oj}_-qUg<RQIy@M$kO|d!b=4qf^*qAcAaHRVFPpa9k1#iHIy~+@;)KI
z%o{njz0kI(m>(>8Ty5FC#Dq*)<GS%yE5O4`3e-N}F{%pbr`xqU%=LiV!D^jy_*^&h
zX_CB3f3uivj8~y(pd>Gc7JGi@)pT%b3ov@OviZG^$ArskIP?$T*Yq+BkhI3p^)(0j
ze4RMCtFsk?dO_z}oWX%W!Nqn3njFWnOpl8x#BiSO_?}M6(vAC0k~dbu7}-5nPV8Q;
zeAJ8EEf7lLS2#JG6(%-5+#6&P^fViqbS37aERuzG<tta-8unntHO<g^F(04B8CPg}
z>pb>q-?VFshG+TyK#Te8z--1_Q{yklbo-?K^aDI0f&x%mt#gKramzRCNpsgZu(GtQ
zj7sX~&!2d!u5W<so^s0+ZAya<x`p=<0n|KY@N6j8^}$nJXph!kI%(dPtk1Q+_o$r`
z%F(!}T_qS7u~!Pitb#c)@=IFc2CoH>A}SolU35?O9c!z$qE5|5nlM?nrCn{mS2)FN
zdYk5#IymBKl^rntcbU|rVzmXoWL2%PaP?C_JLSq7lW2U>m+46^oBcd{1Qso`8N|X~
z*_#`l8AURGQPf?<m0MFPP}3x+9$2ecO57Wx7x*Ju?t-$>JsB#K?(R_TJ$2^HbbVBj
ztDezj&%+%w$RJp2c3PR>VR4Bgcj=wzLN1uAuLp_0M`AfPkAEC><n%hfilBVBHX=`f
za8h<s>{k<VEw{?t@3ba9bFG`wD}i0PZ`~%9mL9-zop5Q80U-H0joekqiLSV|5wKl5
zI8!`qtkOD<K;J9Oj$r#oNu=sSw-vPv6JvADR+rP0#ixM}U9pa1WNS0rahV#=@GElk
z)V}gq4TzC@sUAOW!D^ov-XUh2F~s=Z9n1H}qJ9wJSu$hu-Y!p##kujdqs?X!D|id>
zimMPX3x(i8f72U9E+@L}t7?SXjzNXZBn&Jzy_mod*#F&%>r6{0RmEwWTIJ~quleYm
zZyDgf8t@85N6q<K|AI}wu@I))6Q#~L>hA9Gr~YWcsb;@1&m#|-8^K8$&-2msw%65<
zp%gryk(BOfqiq!Ziz#N*V(}jaJR(QVLd~1QHgVk61K|1G9=CO%+GT^4pv0m9G_kRA
z*Q5sBVnwwNA2<tbi<E7tR2Vo_xAg5}Y%XR91bJ4l4%UWtK8)5JK*AUK6g1(^-g6SQ
z^~Pn&NszPE?T0pl$`L$PNd7?+v;B*sXtAi90F%s~&B@@VumFTzyO%x3(6s-ty?t_d
z!=>YU%9WCy4LrRTvf7Kd-??%JxntF?N!;tBYu&r_<t$4VbW$EE<IToob2Wtt#^L)1
z<XH3bTv#{18viOo>!&u54Z9Ns$5F;;Q7tS`Q8?v8D`JgBHrepIoI!)1=A>fxj-)Gs
ztmiXU=?gDfLoD|iE+7jM(j4S)ZnS$m-@BAm9R>-AT&nMM(sM|taNe5i=Er~drzscy
z!T3Xz)tNyW@Q8trRx#)Wg#hu|EQ~EYJs}|2Ev+2Wb}6MPJYAm4P=GA&PszBw!;O>*
zcsk&zim#<5GC{gF5d9!Xme<nGhNG-3K-T$Ba7H>L9S$c${3|ukKW`M$`WN*?;i9wn
zzW8wa7NwNdiAzXpN9lr0xah6h*Oz>_jYWp=0-tqMNrj90eGnk~I{>QkfC>fvgIP>Z
z&#5ZOQ5I=<#5g}05z8!<tCkVSBb~QxkyMxDsXQn^W@$TPNU*V^dDAy2*#Y1x#eA?U
zQ2U~tS^s-6vY@@QMDOr)GHaN7N^W40wH1Z`T?~u%pA=*Tv(D8zv4|tzE+3!Ofz%oT
zhQD$_)P2xkgM-F>9I1h;uhW`~Y&h&?`%S@<n8lhyOR3mX^aY|=;BAo#&BEBmslyzS
zJ%qz?Tt1<#9a1_%7CgZtUTwy_F-;L;0Hu>PtM=DZJcjdDssztCUAHdTgLckm8zT*?
z)>76Am43HBY+}BnFv<QVe7|KkO1h(m?%-sFJ5CQ)DjL4uRlwt3!F|&?#To$JY1Na5
zb9u<C6~di*6&fu~9V~KsuJq6Ede@uh<%&bEmt(YTnL}_G@(=AxYR@`4qrNAkY5N?D
z*SJW2>FsvqbDoZ`8H<a}%;R?*z#|=l8)p19ie*3`;`_Vd&Zup46ay8H0$eVp5aaAo
z;vNZNFb9r*mC<^sN%^%=i$l*#UX9{cuUG|MXSF1m^7x1Z%w_p=;kipDwkc$kxEG(k
zPNK_mA=y1zp<!QPvAUOa=}AWa7YIp6fPE?Qr!IRhh&?VUq<h-c8gqn~k>o!81AYO<
zu6#g_=VDP%;kH#dlTz771`cClVGeY`6E#rkOkS!rw*GhSt}dbU)t5}<Dpr@-Do*zq
z>WkfKTqv`i@VeMq#6nY=4bn_P7s^;i<jS3Q4-VFrQBO$m7yUZ^MvAcGvsyGXytu~E
z{70WIllZtSQ^Rm+uJgkInBTyEigRDR{`l)wzI0a)ax=30S6`LqntCj&S9UBaZ*6YX
zn}M2MdT*MjfI#?4n*8E8VoIw-Vk=)W4VIFUi%Uiy0i{#KXz_hdd@jX)c4FSfb+M0x
zHdQngLv6qQM@J9=_V9YMuCD5?Q@v_Z?}eB!^OcMYx6Q!m+!y~Ujr)%VhZFc$4yWp!
z!~<N|V`PQ$D`)>HhaDV-o^kDeTI1Y-Pjj9XHCDJ!`;fMuXBmq{j8!;qP$f8D4SNrl
zL-M5L>VIA68(<+KU%-G_EFYk*UmY_}^lVIWD_WzukMVN0|GX8QEX&rfuhk><s=9vd
z?(R8r@@mOv$#A8m;XQeK;C!>(jP@fLm`3E6(Ov~MC8fYGfxFIhpK#xnWWNd`Zeac!
z?^SOBceNv{)`6G{CM&dsFAl2Iw<dDZ+UY<QP%aT^flEn{({&^5_e^#T@thv7jE<hZ
zMTW%I80BCmy-Zo35IBkJ#(Z>NJ&Rh7)9MGn(v?9)o%x~R0|N(p(+mfa=X@K&J~}x>
zF@e~RRQ!m!hO!Lr9jr|#DW#+|FymgE@E>%*L?K$!PHyY|L0lU1W~8;a&3LhCk&k}8
zgsogcVP#1W=V`;w)xi?KN4fcKRxSuKlCzYCm`26mh^NPRZJoUu>NbTYR~4dVSFn`Y
z;oOxn8i{OIM5Rz}rGZp?*HGNF{Gg=pWqRHD`HL!+$#O0ID_;HQ@85f@w>;fHmg6Zc
zA-|z<gio$jE?jtL2CCsi{yI5^59Ldj+t`k()~}+}>bF^AC`o&|24-&u>~BPGz^Xh$
z3Vk-kg9zRHro8^>fstsazb#aLHDa$TVD%9U0hN?DF%lKjJv)?R@(o1YsHmi(`ka+h
z?=2zk3i~*0YU<oAj^4FK9&D&j`>Lk5b@&VkVm{~Fp*%i20j2ksm&6m9{)Iu7>G*1C
zctnKB^YJ>?(gG=?{6}V|NKy~$4%gTQvz!L#1cNI_WgN9|Gpo`!<hMNyR|pBf$P=p_
zlqXz269MCapXf6Y<_6r!^A55gp312=u^P{L=SZ-8p(PA0n>my=_$;<#In33jwYOWG
z{J5e+nW#i8(|UUx;MG>EcbT`ozK4-(<M!QSD6b$)mv3zYFYn+maM;7Og)zg(WHqAo
z(^<C1bl`N2Onsu8gjP*WZHiR>6e2ai;vx@2%;|0NZ0ZYVu?Wrjkqv`~aR`yULo!8_
zBsnVls5KkXn&cD8y~%Tm_&yEHXlV3l3ebrhu1T*_RPr;Xx;bi)$Wz}`d(uaw$Hj6X
z2M{*u3B`(Xy49MtK0Fuacsj4ZiiR^8pJ)wtTrDyuquF&*m2%>uN-<BoXn#4?_Vg%X
z)M#LSdeLH_s4*!kkRf<3Cx`guDwpxX;%W{g!?!yAw2&ER^HSUIv)u*jvd2U7X&7Bh
z4Dt2=gLG#Hp90>^;MTnd8bKW+9h0tt_ijuVbwuV=pMrI+2GdRrfUdFba}}7oN2ok~
zD9yVyILa^suUAH2z8_>&uFwCV_R&$a?9Y;aFJW5|XH|K)g4(0BtFzbONfv0MU|qtS
zS3Dp*+{tl|+M`C2Aj)Z@3bc5Hfmvl*;-F%0`*A&owL}f_{<YI=GhfR$rAH=R$`bOy
zo0j9sldI7US?1%b!~14ghoz$%9^M_U3ySZW!MZh<%i>XZIve~*(Mj=bPODqB*i+}v
zIfe0c0zvUFN5`quHP4`6(m~V4%ffB6ycWIwwBIu}=w<EcA*cJ{!&b<9Wj>7S-Vd(t
zn|*wvSLrarGnsW5r94){gN?~dUoRb&#&*Y6IMwWKxY$N%q+HZFL^)V<SLDOQQNOtu
z2MK+sRc7XdtosdYqcu{xVYNEgSt^gptuyFfy)q_FvvGsIUwauZciiH6{Xi$)4FKuD
z0`BP6U+EW6esyeLQKTg2+|<+|Aa5k%>oHVNP@d0NFj3j`6;E9!!Y6@2s<y#CO}R)!
za5(DW8r_fSYb40>o%>;95sae!V#GZan$wgTu+3BV)cG5B&VjOo#xe%~vl%m<33+gZ
zw9+q6B91gav>Fn<deu0C%vjdaiU0SjxP4d{SmGOVoX@_-FPMq0i9#;^y)_@P@{TyJ
z$ApmRx}8eOwFp=9;XeaMAs1D-KQ}|hXfpK(FBAt$ODUz)I!6rtsA7an_B2Ue!>90p
zLA+&K_0P?-{Vy>>#_xMT+t*$8eRq@QrW^&h)OPq?YlLGr?Xpdr%3)kq$)DJ`>i854
zEMhae^pi;gu<58MRGf3wRI)PVIEy_sPR@w7wj{6v#3Fo>vh+0y2Y&vh#LzAX%u|t}
zxQng0_UWNCm%6H1HYFVwN_~CfevB5(lTqoumX<@UPT>9fkW$>L+E?qP(Ac;oG7iRa
zRW3rc<@UEcTr9nFC75Ig4?RsO>pNl3pN}O)W#}Np-d`@Vhp22<i|ak><-Hz+><7Rw
z{2Xg^Kx&AYN;>enoT}7ihUc}hF|DEoLoH9Qld91H%URV0f#X%5V?<+L$@~+%1aXjW
zbG&t+{06Vn`yhxRy7uB(JcFsTr^{H(dc)GhH+xH&v=5pmgcoeN*LpfV*XACFv{?zc
z+NoE?o>6R$%Tp5QEjiV2I(K5rebz5hll*$M><K!<m9A?!Ihg6sNF%npL8C1Hw5&8w
zw<5Q4T{zjACR}ip%F^*=IX{<L_)xhGsVxMIcDO)K?kt2HbvSQSaqw)2*=<UX_W$VI
z?Q!V5+Q##O30R$O_!C^>)Tx8DTs^n%=NGbk;V~Ih_a0exUf#YYUEaqRA+hYcbj}uI
z>8gA2Zxl4(GgrfjX`6b*&+)%9+bU{Y(_F0LRX6p@S=?aDg)rz&!BmL~6Zx_5q^laa
zw}8QS)K*#5#95D~_eE=x|4$pxBWL>p!sd^kXi+w(h9Remnych+M~8!x3HMWz%Ueut
zM@+gC`72tj*2Bf`bH!gT7d1&n#Yg1LPv}hYp79j$N7G8>lH%v;-84E|mVHH0A`_on
zApc=&#C{tzfpKbgz~}9Kp=N<Ynjw5dt_a{NGgB;l%TMXm#8_FtV!;;_3*u-cM}`lO
zl#GC2+QI0e+{ZIy7q6P$cvRWuWyEo>!6(nankY0?p%L!ne&#Yo|M1#26nmJzxi*MU
zm3y?x>mHrcQe|f309xg-Q7Yps?&q<~5fccJJWaCGVPU?Mf!6tx6GMaxSQ0&S{5+f(
z`US?Z$yAq8C?XT!S*y=~dZes1*CL!9F`ksg9@t4_PMupu328TKC0(2%xHl*BG7MYW
ziau0WFTdPZ0`iY`IA^w&;dRR4`Vvg<5Z|;gkiUW0ADC?{r{R*Dln_Yyaz&cr37zWr
zw3KxF_z?GPD0}bj6<w42YapEC?!Ky7_DOG}UtDkvHQd-J$}GhYLb-e*-dQpK=iNa(
zyKYs5S*EOZ!!<7aiH@#UcZ7!`)!0Y+ZG{X5Hc3{@ZY$|3BTQmih!E_0;edRMiW#KU
zlxS<z?+){vx@|~2+!`r_Wb54`932muTMEE%IC;ZmI#O<wK@s<C%^WOW9NzEW(=S;X
z@_(0e3icI7MNc`GopR=17hgz`8sx5J#htAc=uJl2R;_JTwU_}6wG>ao!J`4khI9ZY
zUsXnAA}2f%mxG0BB{PNddI@}m3M>~+Dv<b{J7x=&q)_d`0gqs9(G>S`6sp1oWuw(?
zBG-_qFk>+~`(2KDawxw<9lJs9;%O@=l3sa#G=&1&M)&Zws#HA)2bi2ryOIIm|0e#*
z!C(A$b2Qjy3MpeKbpefaw0)Wluhmao_d-xxkARGE8LTacXu3PQwUjmPc@#-0c-Z=4
zVGJV8Ae%cQMR3_}p=))S1U)@0cE!s70prdtdOul9|C~dnh)G*(bx$`uQMX#w+}*wQ
zC|8I;cFW*@#~%9{o!X;>BPT{GHNWxlzG|U*Bc@ljVBt)XzXCO0kLWb1T0^3yROR({
zp1#MczC2Nq?(I<r36pba@e<0#Cs9o$W&y5QW?QR=UzXIMU#kNke%|Y)sH$tzZZ+P7
zHB=WC*;u+Cx?o|57T|k3iU}WBtLihuWhTizck^3OMS8l!6cnIXj5FVmzkU+o`Gq@)
zLyCAbBSLT0=L<-hFF=`ps!u)6u%?zQp^$3ItM}JYSj^`F5<8JM{nD%Jy#Iak%Ds?s
z<xAW>I!@N&x{mzNscNYBm!BV(i*a4UNOgwll25fofzWiZ`0$Gt`vbbJkqYmxN5R(t
ztg;ezZRMfNW=<bo>j{qv*R|FI`^tzsl(cnmQHlUuaQYl7m<rf{V_smT=4FgNKBtGD
zd@?TEprhsrgMEi1=J50YVG{ig`p(v54LyKQ{5G^LMIuI&6D+x7_D?Fj!67NxUGXIK
z{N6~SkOUCxgX8UNUASsZQOc+jS-5;{z8-0YvGGdT^kVd>7X?;1tm18#_BE**Lz33I
zjD_&Uhq)si<C)<NW94cZUK}c;CoZ~3sImf38I*f&_3Y<7H5L63FJ&L<2(t471;IwM
z8i)16vcw4UPE^_+{^|;bjPVYS<>}YnT=l5sJyo{r&i{CNoql3i&rGETWG0z;4f8&h
zu}NJ=7=>UiNsxy|<`Pe=N)c4A5b73Bh?Fv6<ePo!Tw-FtTq<5OJ%>65fy73Yk-^Y6
zt>3FVRwHZp8sqL0o6Cd6A#*TbzZ5_vJIhgZ`fCQ)tBq;>G%UHZYR!wCLo*B*6U^yP
zD9md&<MOQ4fQjGfwL7-!_+2<)lrgb&@5*Yux#X@+79K6|lB|xeyy*7abDo|;C$Ft_
zq`BDc_<Bv0zc)gt?BOi^mC44rii_8QBs7gEQf4?eRbt<Aqk7ILjsEgn#ZxsZ0oleN
zAGt|b#r}@qb+6GP@xjvg-Ea^YPu7qOLl*tcK~JOH;}_LZho5KGUZ)l?(-wo-@4^hD
zZKvp^u4OUyUfg9{WU@&Rm(2lRz3|#S@1qG0$e65CMZsq`hB8N?$s?BxM`zLA#<E|v
zmr*l!=$kAkeu=rlE2S|tG8@fg$|WXAXLyQa*%L^xKq}5Hzp);LM6PqTQulZE5aZXA
z+EqiFEHsusvNg&r#&6Q{{hc+c8o$GJFVA-)3{oIWdKho(60>dJ)sND7r!!dtMO<Dk
zdkb>dmtOO@UM5)P5<U|E?8>j-`YJo_%?m6%;=dMScs9xNPlxl)3U3OrL}I|A<lWMX
zKhGD&tX2n$ePYzN5_3)K`2rPQ7FHl!2=2Qj$Y-&amf=}cET<CoiY6r<teQ=%I*$Q*
zDLLn6(X8(JcCyTeGz4nNd5n_}SXKoUofc$0sczIfmsx+dV8`dIH95wf^llIlzN%Fe
z>_VMoF}XQOpabe~xl&(LockSJOU3n<-KtDe#Y!uqTGH*^5ulpNw>h;M#^r2Tl<2V=
z{mK}|q^kwcjdq!YU=e7TC33(x7GY9=O(tA8|4mjc1p20W@tt_pQ|aF1NPQlRoA!+S
z#mx*iNCKyB16l{AR7I00knP@n1MFPfhTqE}5r`MNd#ve<-)X4!*1c6bwJl7dDYpb&
zw8O&>j*uJTx26D-(@SU-H0dwu_fPlM$v3Gx&M)d@O$Qp21vx>U#=Yr{HRbF3Kxhje
z7vz7q&pXxhjO<lX%F;Bc!#ji|-hX%;*lBwq>!X*{s^XG0ScWb?rSORCQq=V6TvAaO
z!6F^updTAp$XN=}m0H?u@zcKLlBLqK=afVrw|BsGgX80c^Z#IU^aKBr(=OL^RL>uZ
zzPJbcPF($~I8ZE{r1lnTq051<!&d{$rKj+9?Glr6PuH_1qQs(4OoVA~y9SrOgi8&C
zePiU+We!o_!SBMwgVo{{ltnWSo|XCZ)#`di8kT{|Wjr6oM)&$!k7Rv)LAQ-zDl!&!
z%Gd2HM11qjt(&upa$hI#aq&PYdSv2Eg}a41-6<CzF3`w$GQF~(oUiznTm5$y#y1?K
zKn8w7{3Kll#yMD4HJxh1BseY0;IqUmr(gHl5tTpu-mo<)JSL~D@897(JzWhKv!Yjm
zwK*z-)MUM725a2)LJRy%)1_)eyrT|$y6jTBM@Je!`QXf!<f*q?vZlHestrA_caAe*
z*2;9>v2_09mO22HmICz=Hj+A%Nj5pe%#1`C-;ABW55&7<lKC+!k3EFBhi%JIM^9(O
zyq>=n*UL92*jb0vB!g2Q4NqNuigM)=rSWV&%?C|MaGc0=UZd@^7~U0=jIwm4?Q2~3
zb|csnrj=S%l&zjyTFNT^^ESiO)KcyjoAYk+)!GLrVz2xw(`cq$XO-3PhHYJj*cT~(
zlvcM-AU38t%qG#niu+n15^I&~!%BiOLL?sy`r5z~9VY-pwCwG_!!!XYhUg4cTDeLn
zi<(4NCccT5Qid;z^CVkNX1@nIEzD`Vn1op)=4`umzt<4e@DY7tBxZW8xwla{xLe%~
z()~W1`}S*Nc&TG#VnRVp-W->!bKN!X*^;ck=04K8<AoHG2SYST=7;(Wm_=o=-GTQX
zDoM*10%bn$q37Z1BvIK=k)vci$2>Y$p7!Zf>pfgtu0I(X6+TCazlOU&I<fe8hk<{(
z@Ijm$KUi~Ew#~9w@6z&^2_8S#hYEP2JQ6$1I^W;+*q!U$KU?MUM`4^I8YS9!d0|zS
zRmhF8f~rh;)I(jwIXD)ZT_D%e^3O3uy}5bco>Rp4j;kvQf4M+$RXQrwIW5v7MI6)m
z{p=?8s|b+~AFB~-M)Mb}_gO;QDP!)Kk2WsV=zAce4Qn1*9%{w0m<V3HfspR_*T;!p
z{~fdZdtdLBnL!S3jjZPW<0aeI@18~NS_M?f+N>xL@DM!6<$h0}9cs@i6gJ&N@$yO1
z@1YRkxs1<BUR-BH`xRD+CKg7&e+0i@f56*X8OJ7To-w}K*0xR}%2B!v_DOQ~%Sc2o
z81>$~1-;fm@&44NhL^2C9-L@JLF5{y=cP>J-}gE(BkA#iO)~2y=tBE1g*dBJifGl*
zzhL;U2Y#75!<zZMNWglMrr>5gLD-NeC}TOT-{leO91v{xOek!B7AL;D+lepBIgAIb
zxP(<wp{c%<jVDWE94+{hzKN}@0L6Ko<?o@^njt5HT}-pNg)i_3=#VmBImc+|Tzs9s
zNPZEc=H-q~K;?s-onJqP?x6Wgi)%{YYukYHsxmrX|85>UT^T*+fS4(2@1Q*O&X{%P
zOB95%lGC4PI6CTKTlCUG<X+bKwv%bq=eJxWT6ZS>Sr6k8KfW+LKHurl{<ezDn3^N0
zd(!g#<ZE_z{_ofNNzf8srJ)y=FdrZPlyvM;WK;<Y_xmKVB0FZTgc=47$7h_87fC)J
z_a9H$S+C`A2<$YYA|v5*=Em(Z31b`sy-!&z^R$m>`FN5xcnnW24G6Ur5@yUsj4v<$
z=-zc%-%ieA`RDo!`|Cf>zO*xzx38-2?FN1&+D0Pc8y|~9#a}rzOuFBWCO7Qw+aaPR
zFpAmPjYP48UF66*E5D?baB(l5S@XNuthhQOm%Z)xY0>bL9;E0Lnr)M){g_D08ukt%
z^3F3^pA_<3q@%>S-xwh2z`@h_a}Lc5dLF+Yj-z8`y)vQf$$kn?K)CNMEzqjbC$5bI
zTlV(M7ZfHyW=i0m2lQqN3^IDlBZ<XX<fgBR3X?zB^4QLG>qlE|Ot-fXOIk9}&tnP+
zILl}4wKFj+b(RsJedoC-fz9cWaNzi2(DW$gh!tpNNH<Ryu6phdd{=UAdE{<OcGjc+
z`;9^fUei4XgBGLAd*!v@W_=NFmgm>oB)w}Ly$>PW2S*2~%FLz{ADZL8ealDHVf%k#
z^O$;pDJ|_@?<n~qDdFc@0{^VWbt?sTODbohl0^yyIQ*&*ca?P1dyd7)OU@=vJc1mB
z=mQ@+E~zMKwLjZ3(Mh_vlXST8s`%?5DMRXptgKtUoJ_D>((l|bG1h+65}2XW9``K0
zIq00gS{V0N`Xk|#^t$*X#68E8n3Xz`C!bx_90-Pv+NP1QJP0RB2c)bXHYR0b`o}PG
z#(A^jg?qf-|M6Pjxc6TCWcWw!HP{?*uB@m!CcNWcas_PJry=kt<AjY*$8HV|i$)~f
zvRV~a&xE%o3b=gUwu;<!`iGLb$QL3wBDL8z(Ua-bHL$f3NDK}Z5Y9a$t}GCY>(BDm
zo|mQ?tlY@dR3A;Ym#5AsUQR#7y(HK7R<c}=s&>=6oH14Xk>%}K?DUqr>%oKaO17cF
zxq6$Q=WHIdXr0`)hOIGWuyHFG=cD`E+HQB&io}P{bVPK=T}}_{=iwS}4a}gnPD4o)
zF@I%lp~Gu+Y+F*H{$zVJAtC4B2r0T{sSE*Lqfq2XGj!e_R%|D`f53Y@;s)k1*HMN0
zo(e8^FnN2###pjy4&mvh0H%ygMWs|%m%*xJQi>w|iH)7G(8l=f2tJD`;<dGsYF8c~
zQtp4Fq?`8@8Y!un<SgSHVxVvDZR<lsbLLC9nMT>)J&`xbgpI%P%_)d;Jl^^8{VA~>
z-!Ac1p=M4+H>9NMfXN|zg$aJ4<FQ}NbQyyLtl=LtR<6FK&dYjD_F^6f{}1itKGF^@
zl}PUYdfh+2n#3vHFjBr;hsCKqg6)bDgYN8<hq)nZhStqci#<iCG-a2zuE<l3{}vO}
ze5#ur<GB^_hBDu4?ICa=Eduw1QBQ2;lx$C`BFy6#^S8Fh`TfJ*qpvO12b83wAi2j0
zn0|~+#DC5w`BX41>R~Tc-(w2qbB=rCb)hPIhOMoH^YEql<y?}_dl2p<L>TneCQ9@S
zRiU_%fuu(sPOmV!P8YQcN_Naws&Hl%w)lRmGH9%ENu<cO%rhZg>ZswOhT1}BOLt*&
zR{{<XVswIEznEE)fxB<~De)+n2Q~AIe2;&%X7xUgdZS1i<+XFv0-KZRWH{d6dWgnI
zE>!#~gL9t}C;H*0jH|0ghcSENrAH3(*d59n>Rs3)9cN=;O`e)V>#`aTMDe?(7neuY
zG!`L=`8`)$LA=9VpZ%>Mc9u5ntBge&Pol}?@DG{{k8J^m`Gx-8W?<A4c+{RcN?z7d
z#qqX?^>rN4dXGMjk5^)MFy_fA_CKC7qj!_xUXT}_Vfpr+9{%ot-gwTcMlculT;KUT
z9H(KtR@X8=FM)|)NXNG<=%HnyPTE>-<ET9czW=9PZ2CV|7JOotQ(fnCrmWno^SwvY
z&$DZw1V?r*?iWU*Jl~0)ItnrQ$unHdpM9N)RPW$wX(f2FyPoLx?acI^I6INx?Xlv`
z`l^cH$Hb&f!yjKmT@N1~=j(^?X(U<ff-bnhw8C{2+hT7AaJGQ+*o9zu4U$Y7R-p`0
zT~m&qrL<ihhef581j(lS>1|uPb8$(7Fq{p0Gjv3kmHmAIafIYc(rCE4?@KdQ`3w2`
zdlHLoW^3%h`d2v-txiF{ut?EH=d10Nm_1<F6Vj<Z%2Cjd;rSr_Xcap6VDF&c7pAp_
zBv+15X)X3-9<*`!q?Gf@?ok_pgxHp(fDi*ygI}prlf(kpKkf<g%^8_=2)8OYaIq;F
ze$&aB-=b(m13uuIkf*>!w6fH#NX{~kiNT^V86f!+#EQzK`#*;#K!kHe=7coYBMhyr
zeuG{@pL^TPW{EKW$6Jm9M~vkDjTI3!p@!Q?%mEj91Cb}3xieMY&STa$?rb0ILMhMT
zJ?ra*hi)~#JI&H}V1E$~2IifqfvB*8s{kVtYpApo9sS~KWy~-B;%J<%j)s$Gwu3)=
ztEL=x9UY&)o3!v~IHX2JgvU58LoXde5>oX!9)Y}F%*jbvR{l+9h3KP+5&E;JH*W8S
z{%P>uso^`*FEz(P<G2L_H&K=>H<9r05ST6@-6G>{FG3u$I`;Vzv!TMntHa_&!HH@;
z7q-MK8~k0iGhyr1M4iELCFR9sDS3HVJtmov?F*Dop4h*9{v|axJu55`BBL>p%@j>Z
zMv*@qaaiRnQXteejAxJe_8}Yx_>0iWxSuNmgTwGluP4ixbunq<Wo*nqQ4IfP^x~V)
zP-UI2o!?KC1Bdb|$4AC@T9;Ff$23d0XZfhuiS84{xbg8K2`x?4qyjbkZadf7puLyJ
z-rw8*^F|b>abvXgE3w5TWW(=$7X&nw4Q)fM*4(YJ??i;G48=_334EPM6;rdxe667G
z!q_CDDXIhA-Qr6@7a|vA8$m&)w!@i@As?7%hPmRBHvUyKUHLXPLilv2v9dtRLnv}1
zt9Q4fdaN__f568-j~<2n!TU&UM=4v+*3YWXlAxZfZut%}{aRIlr!hO*PEY&eXNR<9
z9im=Pvf=bw!t*7~T6YE-iWetKr`^Q%q~~7lF27kXdEB1e<)ei1ySqPY?r=N}PWZF2
z&=?k;&@?r2Ol*JZ;o)RH`hjb!tN(lG(ERZN0~*NhHrsVJJikzJt|9UB4m#cL=p?mI
zTv}gFKU(R9Cdd+qDt#Du9CmFkwroCui5(R3rMf-Lpc_Sxhn-0sU*i!JzU$ESEoker
zex*CuQhGucfT9-U$oDx8>ar6FPGOw~G?Y|wh8NRs31|{G`l4IM%Ihk^MgR=33JRQ=
zbB6ZTo)%3!E>fj$Z|iIVps%4@fY>g>f_QVS84CMIdC6<9Z1~~G&dg0sXKu6KPM<}H
z&pBEQ{>u#-izIGEkqb^y&h2m8HUH`hjsGQT<iPuhk6PeL4%>h7#B0ssV-k-{Z-4_W
zrz}V_{|$AQRF`L0t9T#Jp3%wFJ5=}=$fcczp><mqeCcF)M#X^Wxx<oexf{F+A^nvm
zYeThFUu!xs=G5@E;YsUQTbl!H99OTKwy0+9kfbNOIr#N_76(Z#D4{BiS>rFx>F7k6
z?JrEhcaOqaq+8YZ1CF!qG`>p^!bON%{S6gPhcO;;2#skDPjKx8S<Ux2+{1k*G^Bq0
z&#CG=vwK)<9GpLT$6_@;+}<gH8urM*73bF0ZPM~oTr>m_4~joHpyl@vq2w)DvzgY9
z_&)QIQI8X2<Y3^K#wbhzM~++T8+P10d2CCIYTmCi(uZC~hsOA0ZkLx@Db}0H<+e)o
z7R3Ddl<C=!*D94`j9WBB;+*rXMF{JzGMx}o{iCUgJq2A%<&<8~>ee*$j90~5E{>;P
z5a(&Z{r^UxbI;Frik@-4|B&0=n*S|`coHtBMn1+3%L^*-$GG4xx7{aMHL$j6Sr2Q9
zOJ|AE(d|WbWZR?U6<+-8XbY`#=)OO-`r5(z$V_?-U)Ef=F(@Z@OUXH@s*{fK&W|xo
zN4nemxVWdDNZV$p!1<s|QB`$L)A<(+?$PU-CzLeC9h@hm&?ymYY+~lypbA?n)74SG
z>+4LBgmHXj3I85*@P^WDPez&*iL!JmZ9J{T8*&sA|L@G-nfZx(UvDNQRXL|RI`Vga
zaVMpzv5KPd2fqqDA!g<W*sGt*<mS|~CK(RZ7_6Z(nNACmPFgzSQGZ3|4AU?*Sm1lr
zz^^q#O<wixsk)XAwqQH4TB(~&{bQ$73Oe_sGs3O2#E#dN*}}_8X_*;Xe|7{$labX4
z*gRnpBvxnsG4kVIt&I5ffl}i3j>uF4>=qiFB>gw)T11uL=>ttzU+*BPa2$V8h-}KU
zqPkm9Jxu-Iac!+zv0F;7{J)9qe9`G`r$|e`gX`_B)78DaxhYCxN|PSWk7cVmowh1u
zyJV52$CQxqF8)*QW|LHt<JQvREfE*3xz_f4;#TP<dILq_X}VEfU^5^=<?g*aS-@+H
z>VPF#TRJ-8aFn%Kd~ocbXTnJKM|k>;si{Kh4vgoRDQs+%rY6tB-q<-7XQd^P*N<bo
zHG12(Tr-J<tseb4D_W5;@Arh_5wE(iV{D@J{~qcYN}Rf1Vy|+POD>n$g!GPhKvuB~
zTANN0Bo|`yV=jPAwT=#VXIY44Z+GQuUu`%!(D$wjy|MCn6?EM9jnSnQ`n|XMK5hSK
z_w7O3^N|?BG}WBy@Tefw7eAZ&XBxU2g@CuSFrU4zxc5jUM{($-&-yN&&uGO{PX?I=
zcOUM-(}Nt1#zXb!C(+S3i81~~L3~O+0&M>NG-59k;$fBuaEPl{Ul~W*g?AX5M(I$}
zQ5>pNDXwa7<0u;sjm;ecV&o(G75%hdl|G7ioHs6R$HVTPiK|&aNdp<X|A#f+8bTt%
zKL(cC6b%{(G$bkCuZ6A_XqHT`!d&-WW9-wTnvt;=QE%SR(!OX+O;yU%%S#$JsSt1!
z>goB|`Wv*U`K_%&LTil<S1`ORqGfKXr$;9^XwE?b9~XAKA|PJ^<%ciV-2%K(h&i^k
zR@>#(^Q4RV_jsO%$lCDHU%x+s9hP=`hT2E8V9NrHo<GN%W0r?FnN3alCnqa5?^a~{
zEapxV{2%i{1Los4JPh7DA0jlP`dba{7Rfr{=9-+Nwk2cGw>D-b3QmfMgnoMx?1xKs
zW-oi85aOTVOC~Gj??&oblf|AXfB87cw$t_mGjY=K^MQtK@+|<+2p8wX0cr*6LMYu6
zkiH#&0umvWL-I}L_g=uXCY#s4iR$=9QPKSz$q%D5Yl|YSESAzz36GIyk&$`0`XK8*
z*xUKlizLPp6JwW=u2FRkVaLLZ*%^g<y+OD$v^JiU_B<^7<cfVwwMTq7KFu^t)m3k?
zuDvL_-PbknA<Rdp+_}7|=l91CftZEHOQrnOi40n!vZJMOuco>+IV@<$)8Ikh(3Q~J
z;nUyHpYj+Ea<;hf@;;7B>`7d9%dE)JcmIOOQPjuv&HD5JNytH2jjJu6V4|&Rm4;I{
zp4UPIg!P?c4mev7`g8OcxJnSZCuBoOQ#RM+|EQT}&?OrG!Rk<$Hv#j2OEU}y6SE&K
z9W$7LTjaW#T_V3auIpDF7SZY{sc_ve8rMyPn#)`1>8Yc$Cl(cJ>+UTq=l}XEOVOB0
zHmpzQA@2+fyvOu;6k>Gwg+u3a?U$qdD|^Rb+wT@Ra_5WbJLkpTG*7|Pk@FTV%y5n}
zdv2b&K_S8K6Sce^;j!ioA!mngfs|uQ&inIy_3PRqEd`fhniEHAxxSad06GZXf;rO9
z(x>rQ&Dn|K0Gw6_Xj;)742$#p3qVX?8!c@2gyE6|Dx@BbJ#>TR;#O8nQV0qP>TGPh
zY44fY3&lErnGO{!6++$Njs!WX^d;OJO}Wvf|8U%U^mZoJ*~|S<v;dXMd}c6rzq9RE
zpGeY<Aj0w+(PRNu(Zw-`B}L(m-qG8MJe;J5VR|~hIy%D$SJyZyJ$ZuI#Q2<l|LzO5
z<gKx4Bx)jrrtgLTZ$Cco!a&PJ8seR>AP%<&T!lg0cFXmC3N0Qjr`k6mH2nBOVLDx@
zxTad(3-oDAOtV=U`m-zV#E1o2J>PltlGPy$s2cqxS+itlNMyN+j8N1m$)SxeY~KPg
zuB}ja^C#o~fP#;nDeuN7u|#p{=@|^yKQvPs8*`-ilecCm6jIgXWwJ!=s-AJj$ME_i
z4=bT-5Ain^PFBv@oXfYYI{vca%O_<3P<tYsq1H%V9#XCtHZodl{)-m>VDl5^=bp~i
z_OaNkPZoMo?v2eErmXfmt+bKSfUM7M&*4@SVsip*peA_ZC6^(l4>!_ieqnH~f~O*Q
zX6BE`^bZ$WReIpogYBbXp)o4ok(rQq!F`$gtX-(DDjZp?L#<zdq!wrtbzD>1CtQ3e
zw<zRgK#dlF8*azSHS@=C&b`sI7#W*^j_tDko%0ib%qom|z=E*5<4I!1st)I8qQ&<@
z(Ex*3FcaONlC*GS6O6F!4S0_36SwD4V85jh_o3#?eT_256ZUygzt(mph-E3ruNB?R
zy=xh!f#8E_?{$o@snR~V_RGM#4UW=Lf$&Ep8Rg17dubAXOl-5*<msb!k@H{Bx_F&L
z4eRG?y*z~?x4Rf+gMv=B+hbsckMuAX7PfqTi_jZ)XFmsK*4Q#{!v&N>A0lqwD2lsj
zG-2DnO37d1;7|SGor18sNl3qQdC@kPO6!e6A3t5HMKVOy6>qUdNmBuqaMTx-{8zTU
z{%>&;_sYz=`CiQ7F0ZZh<SL1=a%E3{Z~(fRi!8BUL=0Wt5xJ3RTaBl{;AxI3Z77<D
zGTqab;AD7!cKw)?uWRjHHef=kBJa%U?s|9}K@PbG+`E1lsvQdeLC>Pz-Db^QsynwX
z7-ci(<ZhAB`SNus4y>!RAjU^;uX-Z92h0kjD#x{miFZ*W@9wA#YgObJ8k@UkJW!-b
z<xwD%Dwere{U%L?gMYZ%v5sx6gnlC4ylh#6-#t0`>IKN4Y|i600DvGZNqj-tk%h+U
zaAQQ#aqhqBzOYP!t4j0R+EV`bh5P*cv3<eR+`NV=t8wS1HdkuvL0?^N<ru#TP5;DX
zkax-bI+s7RyD$s<X}J{T9pnry;9y^?%5WXYlXAH({aUkTH{U8Kzk7_!y+v82<0z;j
zht&USuvs$r#&=8KgUp+0fk?We)176vlL|<cJ|qJn?FF$Y&v<idSh1Cz{u5~`1)r0L
z#(1S3RrPTEF1L7-U~H=|f$+U>5PuKg;t}W+(5!?qd$h!-d_tjr%x?={9aUBuw4)u+
zqx>zADfv{-HFflQ^d*CD6?=_{tXqsy7I0Iap5~MB-to?f#8T>iO_eo7%zqZQh2~b|
zX;&YCs5&c!Sn3K2Xc?In)`H?JgW?VjlvU~R09yenTA%Ht@Hvi01~O<uq`u<-=mOCG
zsUby9Gu5(82G9^pd@j`rtbZ9A@OmH6<4b%3aNA@<sO{;!kO+OHlmmgD^z{C-kf8WC
z(myNcV}y!=O?x`JzA=4Wd}nCWvs?U6%XwMC_qy1p|4cDAB!}+AxokHz+vRu3j@Ft9
zRpO6ne3H&k?t!kDG9@J9<>a!KVjn{BZ*ngZZ$0TSs`H$uR9GJgP|h$o%zoTO!YN`I
zw^g<AjEN_tq}w}&tWVl9pIV{PvgwVmZFpJPiG6F&-Ks`{^#aDxQadtyYHCJ7@{V*S
z#2ZG&fw4ss8zBlF8^eW#H|vZ{BRa!+|KkA1M?lz_d4IgtrUU>JRK{PPcVy9x2--hN
zi72Dcz{RPtqsaYYI>K5b-aH{I{}YffUR|egHz|b|@AMR$Cp*>7usIj@opyI>8`0gU
zvTs*Xx9}aXWjH&67^kSL*Hh@Y#MR<eWy}vGKA%J~KB^XUpcs57>%#%$IbiK_fG{F7
zOiEcDAK4;N2Tjl^X4yYD*#AB*Nazcm0m|jFL1`V-Y(iGIx~;9V6=~fLekk37v^yoU
zdK~z8qkjK-5svxwWn63z!E=It^iBvAHuyxsWIx6E^DZT++UC2^vzH?0k8>~Xp?@E*
z=g_7J{snjjIg9_`Dtx%ke*5Lw91aM;H1*XVi@!0dLm$Oi4Vu(X3(f+vCj_nq;LG+_
zGZXWlKtsi0uPEZQnc3`FF_tk~ndwP6rX0JfjHYo}H=H|&du_55s-)a8yspDN`FXKj
z_skq|{=bU43aF_1=Dn!kD+&e(Dy7mbNJ}f=(%m7TbO=bt#Q+2aBv-n-rKCj!L{d^3
z>5%UJ&-(tq?>k42d)&J>elhdRGjr#;ywgZGu{!%B%NEnsl;cHK!mp`gXWVBWYi;$m
zo&8CdLCnpKT@o*X8gVm867xpdM|0O_9B#9P@cRYzY+(9e1_K`5HC!>qI7i}elJw@E
zwW__^UK*>*DJ|{a)?s*D%3xxJsf)z<M%F27`m?3j<EH(~kJV=&1`Z0AH)mr8eEIP@
zCLV)BO_tTH<-z%;^y<L=x-y@Lezz`FfBV<ieW9v*LYNT@4jm#<bP_#Fr3ulxtLazG
zv5$O|GAiw4jeSjx{gHbj8k;|MMK%t{&XUW>&T1MhZ@>49joUw79#|Mmmy7cpmqpVG
za0jB!KhPc=<T2APw_JLnaPa|IJ{{e)n8=`D!kn+GKK=3Se5VcN9yNCNfq|TX-F9@-
zRnIs#EsN2C4pErsvopSTzkk_}{C}Xd{u*{|evbU`HYuJG=*wGu3cS3ej)*^6j)OH$
z@2xWZ=gLZMwtC~-0x2m|Dd5?!*dGf#d@aD;xOGc5OI4XI%+T2Ck;8)+$?d+<MM{Z`
zR13lQ*a%Ol*p8{s*~4=r<|i~lf^;N%$f>EMhPtG0iCrNz3-vo@yY&|vyH!&z+vz%^
zBf6Pv7Db7sq{8!qK2&UP)*a@bDZdK{VeHMZCrDL}ZIi`@dt=LEQNzrKPL2-0joj$y
z^B?s;UW;sf`Yk<e%Fz6!j@q)J^F;@s(Ldk{RbkN{5t=cT1f$3j_WP+xKmroT-QZcX
zF!MX$w>9^hs>+JOKjJqA!WPT#z`NHRRD!}08cl&F<38`mKi`lRm!uo?@d0Gt)mHyf
zEKY!vCQ~KFE42$Fy}urT*;^dcIw~<HMSyj%cj9S$HUlpu8a}g-EbnJ^wLan#D9=R*
zMi=Wy13hO&k4Hwow@kSyBi#?fCU70mfsLmz^<+fN#k)tigu3{&9QWUiB;WnR%DQgv
zWb`~E2!db+Y0m$0Q%9}-=L(*WtEsAoYUsZAOOtrLR5{i@C|kpHPl&yMF+II_Ys3Q;
z*D5#Ck+v9XLzf{~c+OT=5)aj3V#2jT__Da-zytC{T$FU|IoYI3Px}k6$2;HW?rQGt
z`XC-POHl9$A3uAxXU~1w$^`GJ9W^E;8iyV?_Y2Adx4bhNCjtkD5c4b+hwV(~B}_yt
zjvxIXwO`-C`_a4u!KuyFYAGP6G_O+Dcg*umZd_d9o5hl%BDN;Dn1^vV?8uDdB?xkX
ze8aNjDE7-k4+NfM7D!pxU6AR}`?@}Cps7)6*ma`mc(`J%JzzPq$(uZXG?^p2Kd(|2
zD-bBIk9L<jkLTig<<I1sv?QEnHiO@NR9@*bVa1xrmP^e+UVgCKZMLMB8k6#{yLCI-
zBo&xwGb;?@)~`oIR97KJw0ZF!k?cH24W-{`qG384Mk9pVN6c+3A1bT!K*C9`>s<jS
zWTw6|_UfkW*g&)jI4bsWx9)@T+8qtOFnyfpR@`$3`>5x)*@{A@GKUA$&pSRH=#b--
zf1xjIw~vdBL6`riPVIfB9Q}Z5zNTW6r{C|RoL0FlD7S<dY;L+`Jrai(ep(7LC&aaV
zJEp#zj;FxGa%U%`lK^);;XKE=H(HS^^y1^!&}&1#hJTLtyavu6xK#TSN#afQy!uoB
z3@xA=nIf_~yk2YXr4(c|C%vH|X1WfXJ^P;rvVP+O*laG0|5{O5GRBXrMsNf^-Q}r6
zyq}SN{pZM<m3>u2yeKym2$F^iXh8ivS*}w&7l{9pUB7Zp&;L3EryP~;q&$=&^#z7k
zJ=_ii4R30p-dnD;GS^eg|5<!LX3cy}hI!^su*^j9{l#CVP81WW_DsXcMj?-B-p^mt
zo%^^%Nb!5nHd0AR3k}9%Yyo4FfO7-c%P@G(u}GKD?k2s2-vTiS_6ZudWC2li{~T1x
zSlT$8d5qF_<$0K$MANv;xw#otIB(pcTjT3rD{SjbU}Q3^zBi~jeiZ{zi|-QeKIE7X
z3V!MDrIuEQ&vEi%xO;|9C*&!#oag^x{+fU9OY1{O#WFG*Zw8v7^b;_+i8wSk_86u_
zFv5!jCj<FjBEloq&a#3xY11@6Mt65V+}x%bt-<oST9gWQeVzZwb&-gfYK-#iNwcz<
zz|nNRzL(||PY=HA4uP+tL-rV58yojDb=JLhpGY1YKJgsRcDsHhKBSR7NkbFjAqOw5
z`aw(HJ<pD%Tdmrv#gOiIMygPRe!iQQ^VikiNL-1EdmRyS$M(g-H{(2=2}9R8iwW?y
z{gS$-3l%afm*3r{p^ZB&wR|Y%BWtBf_g|ABsAz;+uqOI`b?_`ZJw0I~?X}1dpVFeM
z59ot`7Zf!t%+A_NL4uadEYNzTbs<V+?{lzQ>7Ui#A=BYV1V^3$)yVR4X;Ar8o7Tzh
z%r&MNd*P*lu(11d_nXFdgQnTM;~$A%cGl<*ChOQirxbB=8nvXevgp>RTb?Rsg%=4`
z;y{=h#;{lvux7(3(a6H@e!-izv%UM4j`|(rt!fD@D;?Akr0ZoO@Xp;iCu9WI@wiLl
z)tQQeKeCAA{^1o&e0`U*$Zc+Vjinm~LWh?C?CT^O@kQhgGmSOUcPg{wq^jM&d^DoV
z)%1RG2=}}tbLhJOEvGxG8FoDS?<PI81)6;Es_#37Hr;sW>51pG=&hVKS(B1swny;_
zoU_;G8$svxb{gj_0%LafxKNOqk9>vS9#d<sLWf{B>x+6&)qHMG^XwD|ZoUt!;0Gpy
zNPC+a{jmLkldpY6?3`5Km$qM;8N$)pW-l&5RC~AJ^GD3<GZ)+s?67Hi>WYht-K;>I
z9C4M1WSGd;kA95}Y*tv?8W``b2zBoGb^%rGIeji<kRLP(<)?F_s(ZLICGANQqtbAZ
zRp-le8-r0>8_%bt)p=Gag>b&;gwb)e`(OMIzx&TC!bh((TyDR=bji$M+q~8%>5dd-
z!go>jWoBuyW37W2J%NqGxAoov_4jKvF6UWa<GdfWzK;PgtsX?)W?4_Ms8;Zzu$<nH
zGR7Nzn{ZSbuV>n69%!&WYk32?dE@zAMjzK^d0vK>n#x)&fL|H1a%s8-{7zmjmp0Yy
zx7xEk8ci+Exqm)GuIuB6%9ymIJ4?-krrI$isk76Z+f$t-boJ!9c`hEJA}Gq?GQa8M
zNxayr05swZ#0wxXrMcNbo>wGNdvy2Olb`-q2pj3m18a03f8Mz&bP4%#*pqHCN@Kgf
zRB$)s@h>8x&WwysUbEO&_&14oS0AM-cD4Nskv}QLbbS=E=d;;yN+rII%`i<%KdWh0
z%2pMu+r2X2U|+;}-*}y<$1<>#hu3&)t64EyRd(#tySC77yttkM*ngz16z(!IV^IPX
zV%Dn>lb`H?==s12DRtbM%8h5I`rFw!qxCH>VbxFB7<Tv?)9=|V5$3JG7I6IXb0|RC
zCp1Y||4*zSo_kNF#&3a-(S?>9H}G6}Q<Rgw4fgwbIaU|^Z@!+hgNt4h4Ug0)Jj@lS
z-#L6MR$?(Rx6w0HGLa^~tc0_gr<uj=RG(c+j6gD2`}z6w5{t1!JD~se9QQ;~V<+3Z
z(*Z-4+0kTVo1L#J9z!lVw(48!VesvqYX+=Yg>P{Xn@MFiI+Aoc@{`5;dFvtT7g)w?
zNXd+;dEQg#8(;s#1KePO@1?}N*FqW81)uh(&GXio$6T(9XlMYGT{v#WV@o&f=l4bM
zH6a6L>O^(!$;ta>3szG_Yu(P3Hgdr$KGKb!<sR%6XE)jF0WZ!u(J2?Ms@|PrO*V#D
z1@!O13J1Cc>QgiU<(6Mg$kf>2=Be}cTH>*1r6=`&B(5%obJKNSdzjo{gVy)p0KCML
z-lpa0$Oy~8^9-RaAIlssBL+yIxB=75{6NBhR`%3PQOd_(RrF{di7oA_5Fia5gRrF2
z^p>kW|8rtbAt9a)gsRHz=<#zAk7C!0XsyBb8@s<pZ5%QB6>-HCH_4iVA2Ed#?gcwe
zUmrM0`Bv1b@x6`xOV5u`?i9Z#ip`F-aR6$O_mlxuKKu(6uCW21k)cG_|K)SVRsN#i
zkL7Lz5%&C^LZdeqR#Q}{sJ8yR0&~i%qI@AZu<u4#Y+QR~gH$0m$860?oZ~H1D+dEN
zx6gLe{65R3;w4Rl7YQd1G4WDN5rubIUhUN78C!qj#(%~w(xbq{Vgy?N8^b{8!y+pO
z(Vx|Y$n<pWW=%-iGDQTGSK|G)vgx1L<mexm3aedI>ja8f%raM%X=dcI%=!!ayQQDs
zzywU}*Z4s|l2s&LA)fREoKelmuP;?n@>oL7YwI83OePIQ%=nl`;Lp*@yDAJW76UlG
z1>--8zD?h{fGs`pDY-;BUgj5qkd-MZDv8k}B7`36S^X!FM>Meh_2%53&0~?=!1tO?
zcisopowgiUDiOCbS(ZRTZsIs4V}9>YAm3!J8?x{j!{tdO$*-u&S9=BGGs373>-KI?
zpFJByL-W&Q{}wFURJ#ak8!dhjgG$*i8zR4)?&F=sPGo|Nj%`6D8$x!?|IPSZbqAm&
zGy_BG@Zxs0)7lLHkrpE<oE5RwyvFloZbI)Ae6bYaxWjmwhE6-##HcNNmlji&2sEJ9
zczkb7M{PQxJYm~EW)fcL3$4zpdE6`S2u@Ljg{B%T1eOH@I%JmA)FfMr_+RFhco*-f
z{3qZ+&WV}bG797V{qmRF;w4lkURfWb|JlWJt>NnJ`?Op5<5ojIvdhOZuNL2*CWrYk
z&(nF5_N<52n&wOKOPp9)X~V;X;`ggrH8kHKzh=HP)&2|j9d?wWM$dEna{uTQwDb96
zj8c-~2Tjdyfw9#()&l#74)Zw*90e(O+2UCUuv}VOY0B|d1Ckk=I5*};FR?Srh^H7o
z{_&%!A!+AU6ReIA-O^Y_Dx?W=+?uZZp%EDB^}>=b15NlcSc<;&b7+(tI!yp=VPy9G
zazXX?HnFp}xC+`YicD<Q@qSthb|ja7JP8CdPg7ILr}D=7Z*Qst>}{r}LzE(DkiIbK
zC~l)0<8@meah?fvUaz;R1>HVZZKzi9Z0$Usf@Q`Z68@Fw!H!{Sh7{BbK<LF7)(TT~
zx6&daKO~-V)#>8wYsxtP7Um}sn&*Z)-V{ET*+1tUZ#{ISW|sG$(-jlLm{pBz5jzCO
zCkMW!SRo;KO!Glo^=m|~Rk#{5!#P6nXYh2enakNZXV>vQH>c5Y_3oa<6rJ$?rma`)
zizuiIX=?p(jmnZ4vKx=t8ENk_$8<~cqvE0ui;uQ7%;u)_R_#_wr!WPIipoErk^{Jo
z7b|mvgw|LRdRW14ZD-n_JBTzg(w>He984r#*~WltEMs2uW>Y;fMUGF90xvx$Wf`*k
z4^V0U^|+xq!qimrdd{PI)_Csj8#43P<p(zWUu<Xeysi-<;|&U~JZ^dW-L28tb~K<Z
z5Lw@DYCrmHOzIyM#r=cq#{9z1{^5583GR{l)=QY&nvif=dBdxtqk2UgJlR!ZtTdUO
z3j;_Ob8Nudrxta>+EZ!3an98!$ZFpz#&|UYQhdzBayP4RK1i4%FE45&gVaEX_u?_0
z$A(vxbGrOwV^NCBLKW4Z;|{s2L}P5NWf<>8NYs|twz3!0ybALToZ;eEa$L<`O-l&m
zKFkQ56V*Kf6+*@aR%hmVI9JSiM*4k2SiQ!oe`N{TkjyTgnwUPkFLUwk<Rl{lBeaM<
zWU?UD(aAqPVTRfo!?A3U&mkC-92`EJJxoFl4mtI>C>>@Xv(*EFk_da;W6v5g)#%^Y
zqJ*u@7p8~pngAQh$8vD+aqwnekaWb#m#^gFc;FyJ5wX%HUc$3<be{)@e^yze8U)T8
zS}N@Z)Bo3jiZ}fApHQEqqN^_HB1r<+N??~LG!)M}BV{%sl~qXa*G>B%hpdHjReTGA
z9cf##Q2W;|G~&3D!njva0l85=OmvWPn)uM$TiJIQ7f>Z;Urh8*P2h(J3d>yeET4U`
zlJS_1`Y!W%85NblmWN`_tl47)a7tP0yhDrK*iW3vn;UkYiuyHqrcaw&GVOAcI?}UG
z@wlClLWGQx7;LC-O0!m>8!Lyo`7L9w;o{4HJxzkiJKTHdrp#EqZe0COFl8n!r?RDV
z<yK~;YW*xi_ETUt=?srrX#z)KUDL<kaj$=u@#(2`Iy}NmNN)FNLjLN7bzkWFQ00zP
zGx_vhDRuLk^r6ay5H08|smt*-_zg@X20^vqpManYS+;?S1ihcw9H1|rv$%mm!39%>
zOH1VEzCcYgq)7#Ki@BNb;9lRd#c}ty+ARD=f`DHB9i{bXZe~faV>?zCYsW|02@qb5
zxWoN&Y_erCIXLQM5c4fH?Pq(t5BtIH<eD+j+mGC@KhURtvT3_#a;`+&;LN!7sQjU+
zL3L?V{QsBi{`0Df>+imq73R6_O~vLLrly^)LUm|Ngq@9~T44o3id>9IAr3Wh{0+e?
z-V*c=6&HRr1Kv}~Qr+0Y#1-OD;_zdt;eg9+*}L-%VWy5TEQO!5m)~Pn3UNru6{ew{
zwX&*e($rWkG3+d4Ws5xaMDv9X&-I+e+$q9g!WqKI7U7UnP0OO<$KBul!J`YCk*0+t
zwy1ep?3`p!W-0Nv<E?KPzaktO<R-u>VL{hONOt3uCpaHnQmSx0cW`izCbLKAW768X
ziUMk_e!XYpdyD1sL*NI)R5ia9c8G0VGE=b}v7Bn2eDbUj{GwI0S8@H+k>fUAjP0W-
zE8_;{DM{YX$8}NCX_GDNm@6biGC-iow||sH6abxr{nQfnxNUG5nU_n`tuF}{nUl2N
z@Qg0{&mew9Vji#IUv4Xt*yQLPz}7_;?~*5rCuM3>l}uj4^4rtNMuOzzdq+|9rA-Qp
zqhS|n+_2u|BT_ctZk&ei^GeH|&>lF=NXiwK*nO9pp;1dbpzNVK^V0UV>(bhVleeGM
zSCO)2A{7^pVSOlKAFtUu$_s3!ypG#Q3vt7bY5S(*;;`1J!EK9-GqwLzeBqsAyMrUm
z%=AKEh}Dd=cO;>lULmzmylEaOa}c60*I$N{TvT{dD9CaKP&T&JLf$%-5J#reAYI<M
zBK*B${<B-B-EjMj_+f{C{2zS{(Bm~|Z2R?j%=+|VS-fyVBnZuye40*Jr!M5q8sC@k
zCVfyk7uZ3;2x$ahcq+<?uK#C0xjce`m$ffYF!ssyr{GqqG6cxR6A3Y}%JeO(A@p0#
zQNs8XLvYWEpBAasV5-uyP`5&L5>WjTiX?SN+M9p0$<8@gTj=SOKHVoPICIZ(SIlIU
z|50@0H%F);oeaBkX}mB3($`}#Jo_Om9QTlIZB%7t1>`nm#~sP`cRUzoHii)0x8U=B
zMC!GPc;Oj)3PD-WTWNX2sUo6X%c0=d%L%ba%U-+?lpCb(4h0^0EO2tNilFp5Tb}(9
zd9e4V&9zPL{l01ah+AG6)nMs;?8^ZFYS1yx8#}*=045LIN>EX6it!7&|E$%9i5Ndp
zkoR3uy_i7UR9%F0n$bzN2uf3+@Ujyf9l?sRVL`r&j*i`$GD9=F6xTNK-!v+MO*v5V
zUyI0oIo>2P2}u~}IaUqrwPoW4_r(;XOaqRqD7T1h_??9UUfR}4nM>fJ^mOBHuFM72
z1#&}j;Av#|)yR;Rpm|Mr2LZKaXE!^tJ8opHf0}Bk<@0{7_BaS$XFyGu8n?sC1VH6Y
zn3;%sx#Ge``V-_|7lD`&TC8-Y<5EG&ap0TjC>NnV{<ERxgfJj3i)w{m*Tev~b@sB%
z#Y<bMsU#841sQ#WIP9h)G@c|DG&{4`e7WD?D>$Du?hG$^aJgI2ftS*ST6KfCn(vcD
zivmsX?#26!&7_!^`|>6x{}wBL5@CBueWB%hg!BkY8#2geb8dk)?frPk>kz#M#g0FZ
z&U<Q~YT>7a?&q9LY;Kd`Jm=swE~&d5rx$0(c|R@nQ|RX-aY!LNXW_V+ChZga?tToZ
zhp2bC#_Ogbm!oqENNIR?0RI11Sr{ld!$p*`Zl`5_O!05|<L{r5`svzzV`KH&ZGkEG
z9W&&B2>I?K(R^iVGUuEC|MOY@tlp8Y`Cd>@Fzw^%7X=AZk(vYLu#TcEMq+%X?2}XU
zxG2@bvkm73seg-wG82DiV!Fq{Etb4VcKe7lJOY!FJ#?kFw|5JtQ_`gi`fc31ckfGL
z<R`nY$m_9*tJ=Ys$Vg>vZG3MZA6!XE$z8-)dwXMlF2P}CgrCQHXe%+}?mu6J?p~`V
zfJ&fU$qLz~&2g_zNbcPk_@vx9^6d@oeaflr3sWyBmlVxzQ03}2?2{>#Hn|80aPF>W
zK|*XIY5WaKM&Vobe%t8`iYgo0meq9n2yDeRIQXF8cblH^zZ(quPF@RcZY!HOBWA7T
zuMf^#35jTUMP&Mmf<nWx1s8C*{_@C6%mZ8_8#7TK@s~@VW*BI%*_T;mmYx)Maj}iQ
zpM9-n`3w22oW}vquq}!!1k?2X>$k_{(>l^lgj9Kl@A?ie-wkW!ZUU}#t*-v;{*wbP
zC_Hz(H8;M}Q^#}9w(k}|{U7eg=qtw+`ka#YCgTNP+Q;Je9CM<rymnuKeJzKwv!v*M
zKa5Geh7YgS?h&0{rV%85Pe7ox-%rMZY;vCndxX~oQBre1@SFU^X~oKd-2xR#)bl<0
z)+x=my6$|yp*Z7@mN6dRfA=jtBP=rRbzT}R<ic-uB?(a+As!)*Pbf~G7Z;aIY-%YK
zJ)eCM^v2KO#Em$sqT>5?eBEWM3u_xjj}<fU@|;fC1JH7~i7A_>NF~+8ljihw>J}ze
zW3c}a3_WIMtM3)1GIhA`_PL_Fq2c=L;UC^N78ZUEjYV#n-%OpE6(%NIJ;pe0?Hcbk
zzeBgSKkKZxh>Xah@37Keb|5m={bRK-%HwU5r^9{VcZ1jcftTAhZLQ0WIc!PVEt5+v
zT8n{ug5(qn6LYO!f++44nAQxQW-sn$z7Bhsa$t9lyMDC$Rq7urzy}<h_sjwo{#$-1
z7>f1znUKDfrTUeyaQ1L~v&chF*Trz_F`}Buz|7jv^?g3cATKA-KjcosgY);Ri&t_M
zPg*~1^vwv7bLD=5lN(Bk=D6+7Jh8CPQ3#o+5rN5cwV0o97Y_3AzcA(Y!c;eSrCi-w
zST={9vLyGFxag*PuVPR-9nE!;5H&^bsv6IfL!$q>2US3iinh<Lb0T*qr_k*!5_ARM
zAf-k^xx%NZB{ZLPK|GEn=bEz5^-q$!Y79BmDGFjPs;VNJ5^@`qS%)qkc?$o_gYWSI
zA&?}GC4x6PNeFeTA4UD?Kze0pu2q&;)!%RM<Mb4*L$>iXf7i9Qi`9s)nb_#394O)Z
zblpdN9zuj-V^%MGt(ja&^qNhv--KF+d<T&kdt9_nqh@rpk)ctyYR~`fn+Cl^c$HGE
zLP7%jMw{ojhab=hA2P$?1+S;-Pf+Ym7M*(}UVKqe*X)-CIz4K_2|H_`61MGieA#WY
zUXel}l)KRsS)T&D1-8&g`@&`rnG7jtW)gHeD7B5=orgl|8GHWYGsE2Vx&IlfE0oqi
z9VaTiu((A%*Fi79P2enZ5sG=fPOG<b^w&A~<byi}PXD6(o96-?{`v)VjYGR*CbGNv
zm(6BhC2DzN=xf>7bfsmc;2@z9J;y~;D9uDQ?acRw_Yff`1vL*Y+h)uD67fn%mD9ja
zEBzF3HeA9<jHL0*>*TvJWbgh-UfWbN)t#Wg*u^MGogZPxb)4`~dK4SLDZujTm1;<7
zxEwLQL|J5HTsMzi_w>=p7n4jqZ<`GB+JvP=$Hr-Xg_YQ*4<kaZgdXB`Y;nD}S!7!g
z{={jbj9wll<dgVjHFS=nG5^M`q4Dgm+TfZDdHuU<^3O~ye-$HY&^2sy<iL5zV>Ry9
zc$TNXy+){&Z<1vUZ{LpxwmqA6p5b#$U2*IRJxzKoBsAxLnB4G<uXG59=7+W#%mrPZ
zKOJ)AgmRJN@Z8L|Zk5RQiR_?%dfmiczC7spK0JeP$Q`f#Se@Z%e-}SBUbSn{_wu1d
z-{&I9-QXPGaO@k!69x<}N?)nQT{WsydS0mHv~Dl3@-lz@&9}<qn@aDt+f{-yz%)M7
zO~GEB8(Uqyc=@94o2vZ7ntY34$(Z{=i!W!G*|TF0W?rOAdag&vqLHdO8~H-Ky6rXA
z2I66)#9#f7NM;66|GW{?7^_s!!YS)9sM-vj$|^rW8PH%~p7s8bJR!k>uG(b~{_U=J
zizNH9@5t>cL*3q;%iHtWNkyy?PLprO_KlRd)6Ow9*%U7n&&8`x`4VimD6~}06iW!K
zv<a;`G4D{Rjt2|(ab&23*;I<b-z2>5Gf1h5nc@eMlWi@DH@=H`-&iY6*X^Ipl7?3W
zGMd-t3su^j%>7_{z7SQE!d;?J5Pjiq_asMhT=oD)re~)kb4-1g1x`b%>YhD5&>89D
zq-o!?4-ve+cfh6?7cD7eK9hrJ44brXZP%W);N5(9Jagzc<hA8eM13_bUz}Bb5XtzR
za4hu@8s^bEY;O4(KHTb{qWWMLg6yFdqjxae0|6TP>c#VB$4hgg79yJu<@7L_qx1^b
z$(+*@MHA>V%qEUHxmQG5!>8;Q*r%Eo*{7Pc!~eG1U!<%hyHFrrt|P-s$MnUd=Of`V
zlc2(I=HqRPA<d<p?uZiTdS-zPr{k-fi5N)NQpxi?bQC(Qmx`{KW|q^bL-+LCDw{p=
zW+S+$Vq=&TG{aXT!&maw|7R-^y&3V6&l!D<rn3Hn1>42W(_MvjoeF70uSOsHcblBB
z@r@5d&X+V}e)&}@`5}XT-GVV*C0_l$hVBvUno=B4T>ENy8j>P_L`9^<%iU6&pC5G~
zu==SBH&*KZinLVqxbR-hvESkLd>%E|>F_yL{W`H0KjqTZ7wMN8x@fnTa%`IkZ#g-#
zk6v}odv)e)UeKjp0qCN+-PxQzUldZl6sC4T`_mmIMsvT?UDb5hF?2tb0+xEw4FxNC
zR|mF~ylbXgpd&iAUtGk)`iliy*f;Yr61_QFy6tzr(xPwH?N_oX4wqPsJnno?k!RWz
zr2aBT)bx}6^govrHO(^rge!f)tw?Ts=4SWRrH0S2TDAu|lGKX7BL6h3dEsv#@@z6(
z_UaMJU>dzG|2}DBUaPu}h5RjDQ8|19!cPKcCVDP4C?YP83;X`<d|?Y?l4Ga(5Up&X
z?vyqs-75S*D#e$8SqQ`aVCsd%&q@OAYOfS8f#;%9@^PWmw-p0ZZe=SRI>9K|63h=t
z@W>DS+vCw2KXJlmVD%QaGMqe=Gw8<xld9R1#stc7D%mg^NAvxv?0`8H{81aZQ+p)j
z<OE$UAKS8_Pi0n|)itGb&=NLOO)qi(q*zl=yR?`2bXA|=Pk-OnZ`g^0hGi1O)ob?W
z&dvK$j&AQSav#|5>IW`#m5Fbrnmn`IPU7Ou)ll%vH#2Btw5$u$(z><T&6PLuNzFHP
z_1m4y|70m23^gtW?O4B%LyDU$Zd|)t-kj2#YnAuQS<#{FG_~Cw`AMK4oHDmksnR*Z
z?VIBIw|OM$c|G6Zg@6j=`xA8cMN1xralN8#2ZKGs8hjy2xoYzd<8d594|o*@gigZM
zZ3eTMlgTU^-f%#(;lNuY!^%o`?r_rJ{<5m9_Ig9-xSYv7CAE-%Jj7#Uxu?k97E{n{
z+EI3J=ijIN;-|Ql{1Hd5YW`qlv;-F2$!Cv<ws#{SfwS5_LgmcFDX1G9T$!TH*>4gy
z<{~qU3}#qHb1Qw-bIx>-eNiz*fHuxI%Xw2}qw)0lhcWFvpV_>lGLx@=djX)n%qguU
zzd4QB$h?I}NG_&OJlI|v$eG@7VKb}R7g_H1G|2m4cq97oTLSS-O8s`UB_=|rY_WJ1
zQ7nGzcCZy$AMl)y&1F~~^Jr=6H!nJr^03(oy5Fg5{M#cI)q97ynjy*jMR={pZ+;1>
zM`r!~6ak0xDtQ9WGuu*J5&pQWDpJ;g8kt%pB&j)N;j&O>jN4A;&$iO~O{#VbO-=8r
zg@n%=-L0KE+n>5AL*+k<+GK0W&CqJi|F)YHrTB*KSIl0qHY)Wxh`F_0o+Uk0BgNZ@
zB#~@WzZjjtEPOiHCVpDZEClVJl`mVW99%_r!=_zoXg<|=?WSvPcMy$nI(pN8f@~E>
zOZE0pEE_QkcAoAm=I_N-Z4d5>%&&)(YZ!|EI~%>Q<`{ZrF#6e!wslDgM70A;39P-5
z;Wg1U)>8g<?a3+Jc54J~Tm4tFHC|F{EWNTVw}HORHjWimzDmDc1sWgaxIPlH9t(r2
zUue&uRFHl5--$>hM3N`+ZkG?UVp^X$zQ?~mx^j`yT}szr^y+{{R&$W0@rKQ_Sivk0
zj7(rka_`Bf*~O!()@MiB7M1KqV?)Vi3m1B(lq7nnzVo>gZ(jeY6!F98u?1D1qF=|;
zg^3Vqiz_7BS^LlJWIGYkyXzm*r(buPs#5d_9(t@b4sfed1n(o`1GqiL8y;&8vjuas
z&#Q#8_YxeGNI%Y1^I%)ObIM)r^bzQAi0?d|J2|n`F_>6)38=Bb$#iFMVP@dv)qtMz
z>dl#jiAP88#%4po!S83U(B`u<Fk}Wij?8Hj`O?g(AQO7)+<7$tsfG{54o~4}%H(2H
z-Go)&=L88BAWE+L*IjBJjZQmN?XMA{m2nC#@|{1G7Y)VN^I=j=>f=xh++d&Ftm|3G
zZK5G9NK6!szbNP`fEs^PuOpU}l!TSO>*HW*BO<qQd^6O*h1D_!sn@|jJNR-}iEc8Y
z1)^d!7WkUuOy|tVDe`W6DR5_Z&1^*Sz67}eNR~ZZ=nRlP5%p`Bq$vz-o>~_s{o_=1
z=FC}E`E$&!UP5=785kaCe+!je^Mp44-33w&>s|!#3ICtMR-3sgM?DgI7#mkt*B`A9
z4z<|0xKx9*w<x=RKmV>@-&u+BID2|V?L)hXim3PMiZE|u1ZIEfE`e`5ojdMsD_F=-
zgl3t~DLGUtH6bVg_MC;CxvbA-qR|M?Ly4sqzb?H#UY)~7SZ!S>_bMD}JIsE$=&d<R
z7}%B$T61cq))(CL{p~6IccR`0Ad%j-g$Z+;rxoY@el724_mMuIz*pQ|NB9+S@tl^J
z8t#?6SQV~dE-F(}KDe4@IyawPW{qHXxX*=u?VBR0(%KodiSF-sivty5-Ng=D>p^<g
z<LCOe(V?j!$N72Lf4k*ClfDK9mzPW%M|)K?=uWRX^@Txx(wgJRaL)t#PJM2Fi_Alt
z+N|uUf6tmom^^`-{hgblplkGXeoqK_hK<4Jy=3{%W$Vv<5jqxqHniJP{i%|fqM~Et
zzYUm0Da7k9swUk8!!Wiy%*Z&EnIBJ9;Ap-u6U4sdvK7oKvUs3C!c_Go&9ph_XP#$4
zR#M5oTXoH`bgm>wl;X;t=kqyrb(t3ms`J?g@&~+wMS2p)OS``B6jfGg5phEk+1lZ}
z*HMVpfyua-=iXH3kmp7?xaolt(@@HaXP^5n|9kf+;Ww{HGmW;z#Y-kCGPrkRBHX-I
z1U+k3y=%vu6Wi7YoE9_9!&a0!ovbbI(f)fAi)Zll6L0;x^pFToFXQL&REg21I;(b_
zm<SbLCUu59Bf~w#z*{`3&`^L2`Lj5-NI9(SYp6-_ue=XR_$^#evL0$)wpOz4)VOO?
z{OeXxj#qFC5<Xq799;aT)VFy`>7*shzBh;7UGYPo-IF`cPGuXL8#L(p?WI=A8)kNQ
zHM7>*ZlaaV9=&NqKcNk7O*?9nMTcG`XK{D=-01%#?JKA%n6f@&%r$&CXce3DE1?41
z`tWijP2anqGCqP__2-js&WJyJcKJe)06yxA!g>dB`9w1gHx8!yI}QuN^korMy9f=|
zyFj=*FX$!HljwAN2Z!s##57fqZzk8wguUwH;}e=XG4W*3>WH8%0Yn^RLSMal#Ug)(
z*;SBoW@ZK*iWNJ1&c_0~tzy&$4q&FCd4wH|^CUl7Tf>^CLTG$^eYIop)LYxz)t)>_
zJacBxWw}$A4Z%PgG@8E3&d3K&e+mW8cw02piC(_~a+;t_XaSm;>QZ9YIpo4FTYP|%
zm`8`I=8S!)za`O$t50?mV;((rt@+-ileM#B_eHO*-c%%Cy?OgOq6gyGr7QbgAUK~x
zjmnht)!M>8hw@qhf@mOF;vEgzVYW>P&JD)l(HDytUfX`p-BEfpR^~k3b@t{vBAiTu
z|Ca&=`6BCIh4qVkD2taMSh)m(4Pi7jYhN#7)%+U$QF}y$4{<Oa-*H6EF7<H9<vc`*
zT7QU+&IYkwbf|WzMOs&vBAl;KTV6g0WQ0l4$_>5*eI@29e=iam7N(}Cc)Q_VZ;`Q7
z&J-#W7b^@)Efw@Q6x7<&L6igzTcang{&zv5CQ>w$aWF7@OvssBLKI3}tH!MqG&2J2
zn`Pq!)Ijz+FX<7=A_gP{Ef%`dq-AA$f2K$|w$Nhcqb>7)ZVwm~eEfJLX9_112fEb3
zb*dAKuu9J$v8U1QkLo2!H)rDDP#X8x40Ku@%ZrUAZ~XQx6GY#!PkW2k#>(*#%InoD
zQtxquQ%=RE{IJ>>wMW&%wGAU{xiV7Gn6Drwr{>{NJvlX1`uw*z?zuH}{I&CQ?sbZ(
zpq<wjdrs8)w|xaxAC`fZHtpATR#uj@oLnEqz$@!;wax$pddnE@ppdT_*HB)oFV9nU
z$?=Xpb-aWs{r(E~BA4x$ATB`0pHftPh?<688mvfARbO2D-Jo6VQq*@9r}qQ+jN>_x
zQ^w65>==OZIOnB@qae6j{}h#r)i{OAN`Sj@Tg=Mn1rbi{>Dr?1V}A~}ynS-Gy7Ys&
zKox`za)T}=1{o9Rqr5szrqZr_K`5YaHE1%!?1LUd0P^mg%Hzjx0eZt?26cuck)z+2
z$!ji(hK7Ima2s8@-b6O@oG+K)Y+@4&DA$H^J^MlIxH6*k^5sjoqwS@EfnP<&eMt|`
zqm_+}k|Q{9_AY2cb4+gs7R^G@?!i(n{z*JhSJ|H4uRa%P{cX?UczxrN<ByX^E9a=^
zNmKM7IU4L}We}MPoH;X5$KV21&g)rpI*8UVqAO1hmRRM_CYH&_%Yzc2s;X)@+@YZ(
zIg0z+y-QeW*3c5$@ml3(8d_R(sDRTSP2&pG9(%3MKVMc^4Qh3<+ONtgDt(D!XC3Fe
z#0K=7<KQ5MhI_d|=M(XsnV4iJCL*GtAiPYyJXWKFRjdb9USblGbT++8GmtKRMwxf^
zOP$C53??)c<Qn64FL@_Kfl6#>Xz12#jP2r}X@Z)SRW5jF=oD6au*eJxR}K_dci8$+
zsJ@na{1SpFk!=vZUMd~XXWrbAj}us6NA@o+E#P=PJv})FC3M8^It4`@EN;t#t%bCp
zs_lLqZo{u{q6y9?s@n;~-4b#up;?#;7EuQ!Q?N>j>2U&%(y;A+I*)s5{5dEryf^AP
z7l&Qea2FB_u#ba4N}*i*;lu4_krO4*`Bi+4NAUE9FWliZXdLBv@ZxxHD%kCKdte*Z
z*C4W(iL!X?<yA|wFx?Us4dV5Z?{6eMSo!?$_wV07Iy(p8IvF?bo0x&j=VEu-J5aAM
zu`_}V3!;`ZSvuu5;Z@!V*rm^2cceUd06Pt2Zx|jt(1E2<e)}d0fnf5B?e)AE$9%X~
ztokrkp%$cMvGPZZ)YQmpB32x)<aq|N;N}$l8n=Zp_tot&5_0lfGMmq0{QUd}6R$6G
znRbxZY<4_Qm%x2ew72JiHVvLXKY>Jopby*{109`YLG?xe)=GJMqhN8-($LIPo$mSd
z=jmi%1(reYSG^fmvDq$^8^s_2ay-?zYx#OrEU>na^hcmd!+x2JC%2d5EZ(ILG!%T#
zMWL~xq&-fiUUzqQczisYUqk(caFj?m27?`n$D!$RG3idrLyY(CDf${RKvfX$xJU6y
z#T?d!U^P8SGH@p@!&*+fBtfqv;JT68h(`5>JJd-I78*7vH^1@mab+is`#|80Xd$&8
z6*T6s)^ZCj>?NH&PY4zh6pAx5GnJdSmWMUE2RFT*FH)oW0$+jhSE22=FnW?%&70~1
zKE85Kx*RP!Lm`39@jb{c@lbjrc2%QT6F_6$l_*A%f%C*C*>>%QuhQ(t<Y7Tdg}|q0
zCdRS1MSSz-&7seBDKuGnNR<;Ds2CQ|-^0T6+{+R=-G`>axG~@M?zxFNF#&OTCKeV|
zxK!U~J7X~Ig`z+A`C(VJ9=%=z+fIXK)h=Nu{|C=ITNGxR9d?xki~v?gdt-C6W?&9R
z1nidEbchrU8ib|OqPfhC)MQa4^8*D5UmF^bQj_*u?J>Lx6%`es9h7;+7JW?Z(cChj
z0dd|0-luy<o1L_1Fz_XO7f?Bmq@;2{UOY4vWSqponnwUUc(Jqj4Te7joSMvJ=^4@X
zd$J_2aVGra6NK~Dr<ySj(m@@|4fNpm@p&fv2&sf~8CBAWb%K8DI*&S%QX#)!UZbM2
zK#vw33a0b=wQELpcTwv6o(CnAZu58Gc_R7K4eGq2IiKN6t)0nq0w;E)4#HR1#T7`#
zYcXEy3D7_VOfA;=5!|>Zs-bI#YukhpiHwQKp_Pf82Q9Cjq=$Ym>RF(bZn?Mi+hVM`
zqHi6g`Q+!LWC<9>$b+5b98gIg0sqVbl2|2f+ZK7+rP7RySiyn-e}AP6`M4KTfxOw^
zeblASP0;_flvr>2MXCr76Z;*<0lEHSZ9)t>TEZXe@!HWvJ}=wWNY=4&DQ)fOK07$n
z`<7TeXfgaKb(oW{*_(V90Ja~zI6QJPIL;g}rC^Z!uxjA24!;8kNa^S6;lfJE*jF$K
zN~x<ymWRg*y4qh)-}G#0X&Lhji;eXdTc0}-K3d=Z2TL8D5<hW-e-s})d-lv6)XKEW
zthD@xIXOAQLqd|cE&J!=oX0|F+7m<!Kxa2=;iIFtk{H-6R(a(r;_GD+7%avOe}uhB
zX$A%c#m;yk(Cuc+NrC@{`~G|P(k`dRyEQX5;I8yNx77k~aEQxAb18|6BHpq+;XMnz
zA=FYOo`+7X`qi3{J0;2ZE7f@lW(ZCS2!pp86Bde9H_tFCC-Y5OTYK)~=a-|Fqh2`^
zsbe8<7keeq*r(vbjwB5YlL*QAZlf2$KKt93bAk;OKG`!KbsDw?A><-)=z0rEXhM4v
zuK}STEY8EMI%O^jpCck8>ChlLj!Bn|N^WJZ(*>Qi#nGzrF|R}Gnd#|RUKtaU^hh?n
zY*>l&i#Y}AFJ6>j&y_AmCKt=6bm77U;qnzrwbk(wXo`Hi)nn0<9*F%uO@_2)t6N6E
z^I!{ihC@yT4g~4#a1e@AtMl?i2*`M3*Vos_wta0s|Hgw=5O|CGN7X2d2O!#BaM7!d
z6ap{SW|<FMLG-ONCh1c!cDyr}`M|WGh>eSncUSAg#KhEiRwo<Zaa;5<@W*u+4baom
zGxPInt=8<u`w~)QgZcG&^CtY)$q11RfM)Do-5qr{DB?h{8sVUGy!EHJ(_(q3v=RR+
z7JX-e@l6{mGwq~Y9xRr^0#;C}+X4i~eD`kh-OCk{YBML?aw;;BEI&Woc^D8lSnF9;
zW;>n_P)ZAglaxVuy|pb2qUd3pDmgfc(dihVZ;9E@yWl|NKs_y@bp|U%XNVXY7THN?
z1-J-Ox(db{4aBXczKHnS-QSyi;vbNoReTSD??ApDE!wg__h<d<^UOkXvHT8PFXp?h
z%}m*X9eWJxo<(W;xF2?`PV$J*0634{JRRP&j0_D1+YlJwOyu;W^yKka_E;U2*Pfxj
z(Uz`!hRX5bF1r%)g9PRTw5{2dR^lyT=iL%`kqPpGKiZX*{{KD-60sVKwABE+$6B#|
zFib%>3yfbvxhRVq>{2@USOzqfK@dJ#dq^Sdp$XuYGFS>sqcn(WfoL@3oBH#507CwV
zb6D)n`ZT;8E#zJXrj9kZ8A$<MI_sg5ma5b@tux(1yCW*I<PZG_`|3`QZDHac!)eB-
z9~FQ4RAShGr{}(08rne;!mOVAdA?asCENH^c(d((z}vTq0DGNJL9fetv{J38s3<fQ
z5a=SnA6Cc1H*d~ifmQcl*jC)3I}-y#SFp%Qre%JGB(pk_+Xdi2+6q}n!~EcqgLSJg
zw&ZoNKp&N^^~V(Mwr}SKc5@{d0@)Nk3E-ETHe$~l_7T6=k;`UEurOy&cX!8w&Nml`
zWYH#1pQgIPk2sql4s!r^mQGU)=!(`xL0XzA(im1nT8E0F;uEKpk--Nm6ZNl2k&{Fu
z47{OmI?ZOkjxDQW+sW}^%`Wy=_@kXLlhqD_TU`%HYdjtQ82B|9!WF^=I$?KIPf>cT
z)hjm>FPCrZyc=HTnVOueo{}DJe&D>aHCF2|*TG4p<0y!=BVRP0pJNBf<GvW%+N{)6
z8iHF~nZLFd78ar(I3xLBL_xJ6hKY?$9t?L{2Sf;5&MUg;?wo>J?b*um?!mM>@$M`9
zXw!i}*40PmyCV(?QS=D2fKkc!0Ru#DR&b_h(8_H-$&UsN>nEp+yf3eEXBBLfBaCUJ
zn~tpPdq9cQ=%GqSOaEaIhOd^I3A%Y-5<*P2`$%W5R`HKEMXd-|Y*NOXJqf$mo>gxA
zU@rC8w+D;Nc;}8-LG3{{R*W3L*y_i=91Uh3p0eI+TtY&lRbxwMFAFBZ7qGuw5QvM3
zA$Im_xZ~H5^!YR2YFTqI75n`GPyz3Zsro+K9@9$h{O1=(&FRVEEn$z(+DMsI)`t&Q
zY)Rnws+CHa?!j&Ci}?6IY804N2TqQyvrcsY2yBj>LK8lZU+%7;BtO$Z1d^4YEo<Mb
zs``u)OQuf9{0#_E7Ik2m0YfT)Ze~xy$rh}lEYQuY*`JMZ15-qUe&~DYXOVF;IMs#G
zm646M8N?o1<?5|@Mrvg0I}JXYv9T;3?1g~>JZiQ&R+ARw3Lv0Hszm=EOe?KlgI)cU
zrMTnr_}jN{?aK)W2_tDLMpv$t$LWxloBd24afOHoMEyzu7uYweJ^Ihh{WX=kTn+cI
zTANr1jvNk;)5Nj<5I#n>dW@#Ba|G{SN>Mh34czdK$=Q;bdDm7;wkNmf@RJc%0Xwnm
z@2o^{8h$;uIs+yo>6bMJ2gg|JxaV%1Y)|z>!mnSyI9E4eLC=_MxCjmvn+DTl0p_8#
zVb<2+<>S+&@>qMhvpQaoo67<_11zwuug^tN{VrjoP#PZR$J?Nw9ev^PRdlG+pI=I>
ze?=^izn+sRt+24meQ@CgNmZU!vB_9#g2x6rVss90hwcifHQMg#vuRtFeh1VP5fMR)
z24T|pmD)p_QnR1p5Da`8v}z1G$_+|8o2`t!kXd)C|L(@o{$?vhmGkPDJ)*GVl%HJc
zi}fkQQdoD(G?RwW-_Tzft8p(Heg&Q0K_L;{4e@aSEIf@C&kpb%RZh!H+ht$BVx4%)
z;dq_UV1LC@FG*g5b~M<o1#G1Mw0lua@35Q`&|+VT)Ma$H6qRRc0&-H(YX`=0cN{?~
z9OB&1AZ1{u)b$=?N15>NDAR5&_Q@2~>_n><=;v*1y#RFXI-m~DMoUWzODTZA;j-*!
zS#_e8e9yo!z+6y0Swl`rngIk7*5d)KI<_=wQ;tQFpm+|*Yxz|`AB2YUV7AvOaTcAb
zwjnah29pEk8K_eNa=yZE{<Uls)7MsKur4G=qd?upCV#gmj?YfK*tAm}o|bJf6|m8G
zZ4Ei0A332^1wBtr=HgIY#QH`nyF0rx__WQr&fSxELCRn(bphVfyfNT9?3C-*XKDnk
zhq&SdT|c><!R{5|_gKdrUT8NZDYWwoF~;J#FgiKnYu6_AYQLBiGv2)`F66$`8%irv
zJv`#RT2nb0z%1R?mn?D7bFKdT*s0f2UO5)QfTV5fy-PQJlOTMOkd(|e?S$^$E%c6n
zRP8tA;sv#Ead9gl{<=;{xe}KG263pwJdCErrhF;k^vKlTnu5@8_dD{Fh+dunY#-1P
zW6dNG)uO}bvI+|o!{`<AzppTZ$x&p(EWP?M={)Aztb>=Zn(^52&-jnZmC7ZqCtpCf
zTTJX_QxL_{u@I0|rPjmh%o_Qsy1KD$+l#d5PlM96tOOS?0v{Fk*E*^I8*r&%ZRK@F
zf>)(B)p=~{a^fNu7K^3{?N5hA^E;SCmBY3g)h;sfxl2u*Gkrp*5TA>siQ0yj`f@VB
zeskN7Rrd~$gIKGron|!9Fj$`l{&JN$ZrkX3ME%?$zXss2A3$};VAxoT1e+>AqBRlr
zqigV`;J0rX97igo-jXs7owq4Rsx1DU6&7G3VGZuIbaeCwQ9KC$_z@fHngdTaRBoF9
z!4WlBNcBAJEEq^^;O=|rW-=bZtyt%dmuJs2QByyJ=kBdL*=Envt5OF?2X5{Ic1=0L
zeg<o(0>nWdz+(G3xT|2njkgEeOV}tES$xM{+XYA!zSIZ=uK=(tVG(xP?9Ls~xB2w!
z9SMtOA{=M|m>(O{fl=f(`t}bNa$dc9bqF@G#Oi|_)tQq#d;rb>-J`J5!B``Qod}p5
z?1jMOTC{!S0OT=7G}+kH6e*wpc&@~GRUeT{5Y}tOun)uwxo1CJ!1|!X{CjQ54Rr9y
zq6lD)yTYW?eAgX_1KK7ZW#&R$mgRZ46Pnuij&up=lrp}naO8wLP=oL<H}D0DM0B!i
zCtzG^@-AW*{N6o$tPxeMxX!FTzlxkzd7bQ=LcAZ9&e+3+M8w7A0YGtF>_r0<{Xj!N
zC9JEYq(nlG_1qprmRKpG#2?R0O_#DV;=5H8gw$5}sz$L4La!nd9Fn;58w(Rrns<Rv
z0CT7TL2q6VHZ!)z0;CFr1rc!Gk32Rbne4TD;=o^Zcf>t#oNc>_B^Pb!$7``{4R$s`
z>oz9^`$OUOdvZM`0{><a;MpdN9-YO+eaB(DRJWfFKmuitek7)f`O=JINxFTC<4Rz8
VrYrnoD13oJOUOUU7k~2V{{Y*+Xq^B6

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 60419b9743198108ee5e624b18d88eee7e3a7eda..ec3205c69fa744b752017b421a331fdf4075aac6 100644
GIT binary patch
delta 92101
zcmXtfbyQUU^Eaq~5>nEo(v5Vtba#UwA>9oZNhPI0x^wC7M!J{o?(TXnpYQMagTtOZ
zm*w7h&&(@kHnbDrs1u>g+8xM_{NukQax5#v9gKoCNt@~WDs#y~sN46I-k3aGlMfY@
z28__3B@ZO>3qdMLiQ{IO`zKDcuM*7BuVA6~y!^z>_Xm}7(k9S@%9=Z8gpdAc!oLbx
z@WMQ~ZmEaGj@n77l0O|2Whjos_sb~#D7O4D=Yru!NzyS@`$hcP%HRs<HDXGXF}CiU
zn+t~@ntNSlbIOTK9)R8J=DbfgZWNKJRFrrzw#GcqclF?ruR`<r`5|Fp5((Re@^V}#
zvjzwWOL^TX)~smKcpntjGc<wo+2ND6xZlwT%MQfrhW*H4VuQh_IY~fQY$#`BfOc$-
zvkl4$Bi3~Cv)gW|ujCLo4tO1+_@m14ep=X_x3z8%yxhf%68^yhVTi=lFRH_`Bv~V(
zISa{;IL)@GgF!hOx@b1mf?8jI!`X7B95T&;?|5+kgN`CPa&@FhbL;J;w$(CpQ_6l?
z4CR>JXIgkb?&wvOZ6M>{4m1+`N|uDXJ+yJ<l7$AJfw6{(*Q|iqQsm%Gbn@IoPWL{+
zMfZU=?|UC{@zxMA3)RI>h@ZfCE?`HkU<tqdbo7cqeQ<MSSe4T)VElM0%w!tP>FRyf
za<=SD<4uLvVZ_1gg?9a6-r#Jg7UUD()>|GN-Y`XdL+4Pt-sg?B($ZHnf&7>F`1=*n
zGEO2%UtGIHr4WJ2*@sIs#LDQwB{{h>w;FF2<Se1yXhyHlqDJ{dD~Q#>$;o>d8GmQb
zC<fAJ)2n|&4UoX`^8kIp9AU`Vl4D0*V<Mo?IgY-pn|3>VAikw)j$%cUDzv`D_?kjz
zp-GBzv|%yT<51et@{p^+91-X0`TLbSMb=L*fiUvZ19sq*$Y17@(^EFwoZAW(r5pGD
z%6@?*WIt;YC}gV8K!<N<|EroiMkPPKu7i-4V2;x11A%@<bp?gtZtHKZijD)PFnFTI
z)@Pgo98|QqY<e#eLS?_m|HVz~|Keu*HQ@GvSshPy)@TsJl8`<LZ)s^Lt>pO*)`uK$
zpjdWPs<AjYAar|9x=dHH<1(b)7dRG?lX9$o1P?^CB(XoPG{}${-)tt6R6Zh2@taN-
z#bCN|(0!j2=g?lneB_7f<_9bp3d8V+;k|-G>VI<C(`|~#_=->eMk^yde(`&~MW}`0
zj`*v+mA_c=rUq9unt%NZ{>20PGie9_ZJvN=T;MfCeHP{kBdc0h#x7lcI%TxIv=BDb
zVGVMIl>B^4ug+10hsm$wFO_4s{SYZi2Vi>HdDri6n0A%#76~QKvyo8Zpf24SVzJj;
z)l#jU(ceE|A|ly;D?@cGw^d@sV#*+%EMMJeK}r5Kqel8gi>6qTr$;U|sP8-QeLs1X
zQ2P=VURM$Pz+7F!&cPNP-m*6w^7>16$B0#Gaq%l?VD|avL&(}`U(#{)vonN(mhX21
zBxqH&)GO!i#KvC03IFQj7lyy<K{avnHDB-VUH8vp^C!)tMvk|x`@&`)A1k|vo})7t
zKFCmzHmvb7uYKkM%f}l4@;-^H$@F|@9;<ltlF=@kE>2*@r?(k-@H;}~74qJ~f_gzi
z!t^gl-H$(?u^4{Acf`Hu-pJMaXx{sLb3QR_nZlPG<xBUH*o>{M>4S5P(>h6^C7UZQ
zV_NDA)YJB(->aZkmiG>y$h+g>A$6E|Vessb<RGaW3eTOO*gZf50>$8$kqoxcd1~E5
zE|ucGBpyymEA4^~M`G^hRz4iwy+r3nJn1ObmQSCkq8|Qf9(%=cUcAPWv)7t@Nv}Zu
zr-_#3kO*LwoW`2Gxx}9ypo5`h<NdMp>xn?4;9GXZQO~l)2-hB)?T1pYGbAXqhESI+
zVDp&uqy3Ny@Y3UG`@6q0E5#&r#9~^^R@$<*wod%X&OdxJWz${-=TRCyhx2j(bsqo7
zEWj6CL`S1?#ur9ry_B@ezoovK1yM={&$VZy6wxH>ws>8C8r1bl>>IiM$hFG6ef`jO
z%O{|7C>7TCy{V)b+W$9?(zE9y5{YhYO}WvUI3R$&R4)*+Ekp4%DYhnWMHiMxFwR8_
z1M?9^QdCF<m_&Y${G)$jJ#Qj+zYX!89Un#XK+jJLlA|{-P9MB~Y)Qm-YB3i-lSRvA
zXI7*+-!kzvja$@N4ABYSv3HE6Zv63074ENC2xc#;RYpEjD|8ZKwe0nl1&3s{*JjCZ
z8GvrSL?f#x7luP`Sq)l)>~$xA{5(BB!^63p`^(00ZwGu<YY%nXQent!$<awS`TJwd
zAfeJCHYbb5E3eS>vaN@kp(Mh&`oE;K6b|3JzpwN-d^g%|8_io<Ygc?OVVh2P$zM^Y
zs5hnH3%j63&nHe8@b*fTz^nzXU_MgZ`D-5XdA+)Ip8A?8PqKEOSc>jN@k&<`ogx}N
zRw_UX(!MyK=8f9ic4E3*=%_;z8%+E<HS^pn!q#!w{r#cN(B^sM_gwN!H4EGG3-T&S
z?Ku|`;gBmYg(5bS%cRJ^g4WL5#6nhOHKkf8Gu$ZdXdaH_U&bhpe!n>esx5}|f~S+k
zEc`&&0S|zTn&IyXJbe*qlliK4E&p?Bd1zfo9#@$c92h}zdam$?1-b>^y_IzmVb}O4
z#~rTX1@4jBflY_Ve&KT(ij+gv(YBu*6I|I8-Aa@ntM=Mequc&aH1RoL#gBZoH@KIQ
zBVLkoH|Gs-<EnGNQ2;2JLfLiUM_YOol6CoPT<Tsj4UJ6<qbXKJLqkGM4SM)KYdkpE
z(#GX&b+}xfNa^(Q5mT4d_q%xs!4$b5Idx|uCA;fJXTA?uvHJ7n?EfUB1tH}f(6J(^
z$C>&&>c<=kjjCnXPeX3kNnx8c1tse<nIZyOj#M6wc9Or}15C$DvU0!R+S-R&`|W1+
z`JAaG?OeLplvN8P?O<ja)t1YprRro-8c>QB^qtPnkiLy#mVVj!Jq@yk1R09Sz5Sx(
zsUC;aIM1(|C&htUmBjt|iyQv9I($Xm!$sQQ-{EoZ2_O?)rZLZ}Tyx#sdj!t8K6OKk
znG-iEyZAl=v_%gdi)Ih|eLsDDUm+q{$j|1kt*z~^VG1vJk+o|}(!?T$^)#UiSC0t@
zBrFV+Gz+PVCH2|W$K!?d?DLoPUE12P;35kS=_7zGJU9f=*4DnLOwUpi>|zvIt9$=u
z`=70t@h);67m=o0)ry)>nDGmrelC2UjK9*HH8KT8_o@Dkw72(79Vq^*9jNZTC_Qvh
z)!}*4f;d51k4hd>6be*gTl8(P!nXG2J^czE0=E>ri88Az3NN=M@@RAOKcb3La;UK3
z?pL>>e>}!{lhcZ`p;gzYdU5-lmPTqmajB4Gh%P*dEP0|91%c4Z$;s97gi2M${q~bi
zl%W6+k&wpBt8+-Mx|<8Kv$HRftY~SAc<!VL6`i`fyJ0lc&lzRFO5mA(9&A6P<?^!R
zaO1-BQObgXeb1`lefG9b<eNa&FOBjp@&X`$Q~i@_Yv(H)FW)q!%3@w@ekk;6wIH=w
zZh!&7O0B-sqIY<La`eE+lo_9^%7X14sG-}w$)E?TM~{f8uzi5z*U2#IX!Y?r>yJf9
z|EVfRDR)1a>XiKfbF9(!p?WP;-^IN-H7W|D7y_~I@|`Xko?3>Z?p8gfQ{C-slsn`R
zhxbcuPKvJp<`LzHW`)2`lulN+|HgkjcL1j08i*S0u_|<G{QC84PG9nzvkcG$p-dw`
zLXnf(T;7L+gZuE|1B_HlTTjoHn!ho25zh(|_b1V*fBGN6s#wFGt@nNx(sGBoOWmI}
zVIGDRp4x5(0TEw{jN<MMSkW5gw%mLO;cJQS(pC@D>}CV*hUOf07rtN&TjqMlo6-{L
zckIf-__O^`-{n6dD<O6|0WRK=(La&F&He^SoLWQaO;K-&xAJ-T;%-)B!t<;vLkbT(
zRkZMI{oOaKZ7DlF{Jv8h&00tH{kY*=kb<~9g_WM5TC42d6ly=_9>Pj+HrGo%?gZOQ
z)|a7rh~qrG+E*7hTPw8K{zW=_q9VA-(+?GL6xw}MN+Mip4uoR?C?nB4<Zge8!RbK_
zO56YBkn&Rm-y~3}(ck)pH<PT@Pq!31-M`TKw~9pnBLI_3enKO@Bl^aNHp?(*Glpnk
z?qS`vij1skz?A5SB!6$tubkw$8?N~$oqq+SJgy#FG`<~w<L80<BaxY#d!6Cqr_{TY
zw4YL;bgYR$X!7#C-?;ur26ibxsZ^j+wg1K6sdebH`v__ku0s9q6?<jIC{JsI@Zlp?
z^Vyt^t9|dJ?uTc&HZ#^0yoI=2FG}>$;%6t!W~}d&7#<g>awTa(8$V^6z7+CB=Q3ni
z)e6N1mp8x#vUc~D68Dvh_x<7t^e`xrx^y^Q8Ns~-$o*Zs(IBy*zQuBvNj1Gwi$TT8
z)V4WjuMTLY$8uSj63wQq6vE<o%l!MkuT8s=oa`^4#*l!F9${awE`*tSfvvB*jOTS6
zMFp^boXk}d!Xcpj4eCYn5-~f<$T!D)d-($o^F+dsZyll;PM@IcP%}1r>4gC)NAevf
zDS%#rfPjE{%r>G?Ei<D(VmeW#*v;lk5HSBUEDM3J_aRh2`rc-4ayAc^9-&2KU2lB`
zBna`SjHYJ^sSnq^A~!cdG4;RIQtnnsCjZ==H~tFK`MRhUM`@OI&G7lW|J@#km3-$`
z3Z34w9NWoQyk@KAdrGKmo>1*}nbTQKG4MM2_pePGL)IO<n;TkSijTfCaC4d2rDq_Y
zP0f**D<p8ScY*bj&GDLaRx0Co`hF`X$l`maz5DY@QH=VF!&q&m=PPh6d_Zb?f{B^<
zy!?((>ki&af6+(ESo2I?J)yH)Rm#`;<kxRD^esNw95oh@?bD?`z8{PTxF+xga+V)A
zc2?uxsL)`&CLz%$wNvY1bjUlIZ$fQXdtDkN$PtlroD_kAd>!zU|K!lPVmkNasqQ|p
z#(?n>U*$;p-HSr{HJ9(~iNg3oPIJQPmpfPWoelOkx>sD@$0RO1@>qDNySLmBgt@m!
z=*hvSwXh>Mmt+rK;fKX3t^;|m%RJgEH?l>wKNyMWklo`%yk6OT$71efHMX}wvhb^=
zGc_-pDG5?Z(v#14&$&F9XjNH9SpGrXg(Q%dt6-sKTaMUZDO6p-qUs{0E<BJ=FW<JR
zTFXK<DTqH~LfL@{A7$`ngfu&^3;YMN^)@r-`BV59ayAt9m*UtE3xFAa=F9Or#0m^s
z=gVxN^KiBgo6T0Ezuwh(+;=X?-0+Q6#9mU|+z@W;?6B?ZBBqhmseX2TEFg0^WM!&+
z=g3G}fA3Obo`Cj>^feVV%|^7O2=%{#7V^{oNg%}kNg#6xNpJKdfqNn!Jx2P@KHA9x
zCsi6YA~DYZK>o3$OEk@3qW>(d$!yt(%1=xtz3yYQ(?nym`)<1Cyd-a9*yZzKSWC7K
zI-{rH#0jxUuWPOA*3?MBbUIewwoCbHi1&xh9Mz_%zOLU+6SL3fuB=+LPkcsrSMP^6
zSvgK02PJWl-RA_Z;9A1&H9i=Dc@#(S0HpAAXF6E`{l@|_oo1}*N(C?SFett?pwBGd
zmQVNFdu<awiG@?Pd=+#1it*pv<<Honp2y|Y4l4pIaaq+hKcBnXqFk9&n~3Y(6FjlQ
zHx)pUl(l?Q*ki;7@<XF`9Q9TSH@*)YgoFB;0ZJitl{`Nr%O6fBw%IVUUKnB5;m3D+
z0GL0SfrPhCIB<CN{#7JGDe#N*dis#^%+7WFS7ny%Dl%e5^@nMRH~Q`<?tgs`M4@u~
z9Z%hi(P`)&6E$4aV+T*GT<o6SB;B`o$essS#BSCV-CveaEDs4?w+E))1h7$3oAb~v
zS#>1R>u1sSH)lZlzLSB<7b<fs)^!GOl-Ja3R?B<m+7*>-TCZ=c9wzg(fVZB@x_7IT
zvxW@V(m<UVEcL3>)k4>Ft6eEzAn>gru_-LQ)3QdFlUF)i<=9z@fQW~eB<!}cS8wq3
zUksdjeA8nld`Ayrrqjbb#-+8LV<MElPAg&ZK=&N<z4tG_-;}Yxr$agVE&yzN11Dx)
zl6T@2?~s3VY%o4|4KM9N5xg-y*cr*cT=7;p29_Mfxun0+Lw*Gb7Qi0K23h=^(T5kF
zSVxq-lK+AXBE#0)uLPs-Y4jV%N%K+oFB>C0y>YhXiRi8sox{WiXpaGTiB2e8DoY~E
zyosI3Jl(=;?X|u6Hv|BJOYqtFzfb`owT>S!gvmN?CTl>w&5lEC_^I{EDp4CKfD(DU
z2<M@6hR-AGTy+;tO(Hz&8+f5T{c*m<qDl-0SyUWwJ|M@HxiLV`2%A@#d47WQOC0qy
z>x<=-=y=%>Ah-zgM63<QmHwWo{_EXoqpx(aHk<VUYio5y4~SRuTc)gAG&)EYh`LDV
zw7hHmIm21EP6UPXP!|Y=C%VNeCreun6H<R8sBk0V=;lrguZyn8^DKT45RDrqHd*~0
zqqI@>o>(NY#ehV=g(E6~g>q$xWyKIqmax4O-%(70CQ#WeUA;wO@2L&5=>(c=FP^(>
zj!*+pz$Pv>0^ImkE%s&vDLH~&jBOuQOy;iZI4ELY!rQedi_rxB7DZu-+4;?5534OM
z&%0k;s-zD~5B~>@FXiVNRgEH*9HlR2U2hsL3ED{nyoQ&Cq$R^c$VeH^8l0bmP<HK~
zvyi>lR0&fS3qLF`c0iXRPf6&$gAL}T5=0LGJrSiz=cvpl^DL#JAL9)KUV#rGsjtN|
z*;zC_&Q6?RUTGs2D~0|XLa;R%m<H_wXKl<TZ8b)PwI5$#RA_X(2v}Utgc!u>$j!HT
zqq3aQaO&?GoQf?8-&wLGgW$YMG*^GIP3)u{P3#e)EbctoWL2u&Kyn=gYMjLS4Jv?V
zn(^Ct|1qb7B}b1};&gSngKe(t@e{Jwo>Rx)s%F{9xI@t{_R+$R3)SXC)h>Rzc;HP?
zywiK>`wLkA$4GpdCM{*eFy)pq@>Ns`Vs{R7NqjC;&PfJb?5<5FZ+1gHjbDh9q--!b
zBjsyTb@`^fH>-y@Iynizk;$j94!6toZrlrQLx$wfT%X`Ux=Xy|{U(6Ddu;YhymLai
zBz<gdWd&OyXC~RJL=UO9UR1>XfqbRZ0yCyz>~fAztl2q2Q7J*zS?NQ->~9b_XX<a(
z_NL#2LAL^1lD+m`o_z+1lDu8JbJ84z4<Yo0HE(Qwk3#bR!UwBFi;JynuP;c`^^B_2
zptU@MNi@_9WE?z1o|Z9^6NMei*gh8<v{Pt_|La<m|2=#r{D*Doh)nHSwZ~Hv3!a>a
zjwZs8Y1IOi$#-va<a};1o8AS8S$tK{aM*jp4kK7JWoTvPb~f;xS@BPrUzi_4KFz&O
z{n{LGsghuGTdk0Scvb;hR?&ESX=SzT8=puzU3#c2mx~uY|J!_h=RnNQgvi<j*Myel
zpE5Qfjr!UK_mvb$02~K&7W=<WfGO##{$dirzE4%hwBzbiJt|@nFSy;?TUFj{IdZI@
z-6fo@3?JAg?URrdgg{r0HpR`NYpRZb$<jFrKk#`sx-uZi{53&U8=Q{h-F0rzPn^YE
zpup%_92hxh0og&x5&!obN+Z@i#Uk40t9`9UeRY?C(g6P%hnG+8sxzx#Q7&__?djz4
zF|qO(%QrS<#Xn!SKfi!0+`en6k#~9}T(}UQuN#bb##sZ9Zp)K=W?_Q`{DdNs@)(6K
zwMRjL07~GYqk*w*qFAjkYGw7ywu%^jorUJ2rTCZ7i_(LD@q!odGmTt^>t^E4<C>1$
zZW^<kAaTD)c<bZg{=phF?WDJw22zJwzcJ(6%!yeT%cixDoXsR$OJn|t>mM0fyAGJp
zJl~)GfU^}pTLL(3Be<F`Q2Mi4TB)_R{QL$gp(u0@<dK#BX?EF0%h_rJ=EPsTWxvx*
zS~~+K&^c^aT@V?LmUu4)lP&;PIyERqoZ?O1${qboqZ{?^m!P5W=1>^P#yN4nt&~D&
zd_G$RbJVTXlencG7m2Rqsq8%H%Ds48(MAG2rPh7`;=M`{+vF+gr!Nfn<5F#vh&g1#
z)YmmOf*PCmFAwdX|MmZh3?^3VtJ!sKvK~Qu?7Bt%^ijFgr4cj!FTOhjnAMbfb<<xi
zp|Z4Fr>(o|Ogm01*#X0;qaE*>EZfX{Ck;8H{Ja4Q0`F%M9_Vvd5ea>b>sdDFjsgi!
zJ+|O2Dq27a=FO>|hVE83G*QE1=|M%5oxChEIEF>I*qEB-F@1iMK+Vg2r>nAeX5ky?
zax?-eq+eW3O;Ycc5=dKVADxI4zG;J{>yX>hR>KwoV_PusCmo8qhQ0u=O=;u!m1~16
zh4knKA0w;gg_3~rNCWK^;PrCb8K(Xt3En^y*D*R^PTG*EQ*24gVb2W?7!aA)LBi1`
zkiSNjmjC$vSg&*Ym^UK^y?ja#74@wPtPR0ve$2+fZ$cRlmRT_V_q7m%3k&_THc}%+
zRfP_+rU4k}!x`4Wu+KLq8l3cZGtsbu1T06S|B+UnpJc890LWhP#|_QI<x0aoo{dDj
zvp#2{sotz6&ANV+u%09TTXhr|Z=--Bs@X`N%9m)2luPqFBNFzcjB#dhgY)V<(2|d?
zMR=^AjNi)&m$8K{$h%XY)q)4XrNpqOm82(q8hzaR4QR**`gtr8%msi1r?Hox5;RXa
zAMWQLm|XxwNp%LMC%=GAio6EUPD^X`iU1*ctZLa}hWqcc*&W;N7x7<wHHfAhj9N9V
zr3*FW-2qJkSKcqfou%&hs|Y~Ic3#{%`J@z%^^oUxXr09S2j-u3`9v=YjO5vdpdCdk
zZnib6l867Povc@XZL%k^3jeI7`~J19<s|~(!~9um;kKq&tF1Uf%!oPBpe(|Eq(N(O
z=7e9~weM{!rQyRaqJwFZjFb}ci~DZ<%y%bHK;{Za`#dcfDF5z*ey;dPgWB4F^44my
z=j<}zC27qwXyk-m%<5kitUFV_K5C!7_oR{8A5vnMro|JBbX5SaPbN7BrduZfEB*Vd
zs0F9dj;bE1iXD@_MGKuOCU)E<D;O;{hrHIoQ$KB*1v#0EcP1_8Z{VB5-;+K?5RWjn
z{II{q4}%6$c=fzYnmFs=%D@Q>-E+&s`T9CZhgYW<$n}hmGjW&_t@TVY>F%OlUC|Mu
zv}I*2clSVB25Tsyf7|eTh67*?oV-OeJ!wRrw78xL>sxHKL$#D_oo1h?yB^rsHeq%e
zN3&;E9tiuf(6|h1>$%7}ZWOLjB>*$#fn&GX#SQvQjIGxx-TNVPo0E}5gQ<6~DA-Qa
zaGR8lx4?(7oI3Fr7&LnouBFCs7diyQ07R2Eb9#7*X=<+=jOb7SW^q(~{Z}h}cFMGb
zZZ8t*tzF(97q*vnC#zV1dow0cYzg6NIO7{WOg6nLyv4J#d@@J-T3>MCu>*%+yQk&%
z6Uxt(STLBv{$7Q~l@eKtJE_&4!zBSNbj)mJA2s!>x{`{Ca&%u8c)rRy3GzjsjwiB5
zT#b`VY>~`{0k3s+kj$I*)gGg}oH~L`+jW<;>CyK$*BK9jcjB);GbRfja9~ga+~e++
zi~m?Cq<tiI&sexZaCvAiK@nJA!ESq8pEfu)U-Gv#uZ{~RZTVbxNSaOFU2#$b#Lu_5
z&FfX|&EWWz{2J0n{Y1MOqd5=zD@11~HDKg5(^@xxTJBP|@(`>gUP`;VkNOlamL3%E
zmk|#~I+?`P?tvl*F<dfT?~{?HtG#t7_4*wh!c#04NPa5S2j>9l)<-G{np~q&gxyJx
zy4x@RX|758|I=K5<g;LiA3RxUs{itZ^pu&do`2ZPjcNAn>)Pxb(0H`tVP2OX2by$k
z#9B|MAV_s_8c&_9FH@U8992%$O@7`>qwTRE96}lIYX(hTp7&N6!4AnxsHpu|vkC9I
zO|1(5ftVsIl6<*E+gclBHA|`WhzOMG!<M&y!}PBxd8i8)oIrwXJd~&Vi|2_FM8MmT
z&`mQ)f2vOZch<9s;#yc>ht>YQGXpl1TD4$@*_?s>%Gy%9GQT7HvQZz>*(>1hTkNVP
ze<(OPT#p-e$2`A}=)l?ZK+g`ae8|f1m-8Yvxv8|##LmihVt+m*0y$Q!R^w|gy5sa%
z7Uau}?0labq};gq$$7?vXU{7h(BD51A@uZouVl?1NsvzFLkd)^#75iF#Ji(tv#k1k
z#e&|yj5_`6q9suNgW5p0z4^a+-mbC6rJe)zy65$dCpf-y$du``uQ}ZV(yMFon0S82
zgxrIYJamw@7j}nbL%-={>%bn<-Gwx3tnI2=z*h@S9$uv-OYx`e$B;t{>@PRJe{)z>
z9F7FQR1pJb-u2|j{S$Q~w8<I*ej+?-k!eW*e#*1&cEQQQpoXiDydIJ8!>(+xYbo_l
zdU*61d!~|>u(BE#P<F(I(nfX>Y6^i28s=}{^~y3!-0FeW)4zW^Ea*D*j_@NZHE-BS
zGzuQeYg}@EEK3q<-ZFrEcdM`5m~vs#oUJlm9QbnEtq0mhNzEa8D_sqEU=yHv8J#Rm
zp=Jq*>$%gt%j2-$9aRcGpAgD199_qBJ-TG1#mdUpEm;%~E$n!_^A{+VGugf)=e?eb
zH)AyhWPac;`N?k`uT4w`MV8YNUl|LjbC?~&d$UVv#>)^T4=7W7=Se~k3`*@xxLE}v
zX6AU*_~rgXBrS>m1V<OWLa8!oPs#HiT?Z&y`F8}c)nGKctPahyn|1$-7sye~^S<7o
zs@$1CmwKGW9$wLtLm3J7=kkd_!O04bO2vo>s(p51+;<`;oNq;kRP(_O-Ko2?Xjyo?
z(n8+oXd&SO`IqhyP_=O74LA{;+^&%p-)_>zJu`*P7y)P=2N6j%mnZKHx<W`s9ai$~
zG4-awq0?(Y+oYf&1~xmEb}N#tfBi1MXPR>t-Z1)T6>Ub}HqJEJ^9y+1!vEi3KD_kt
zBKOeiVgIy0hQw^H!6RoKvszrvga#C)Z{#h^4peT#b&Asi0G3;EW-fyI_}tY^eUfjo
zb7xbfl$y8erp0<^<lXZV-#qSQ=kKPh+CaXPtDF{e+_zltUxMEMkMP!gCFD(UTUkWE
zAQXt%Num73<(S$OL+2Pk{-&X({lz11MEP9jJ>bFp4xPlI+EOTsCcHu!HjL;~eL`2=
zIR`-II-$Hy%|O$}Q-z6d(MV&}&y!7xbM+`8e85#XH$O9XKFVUl3<^9s!a1pp_93vd
zMaC!n4oSjU^1gg~ypSeF&yy58(@gP6Y_>&RB1r16myLlRrM$-B)2A&$is-)L{nM6j
zS)lP`7Ji^T(y(k|w>ge?=1{3%!OF>94=A>3A_RtchNYKbs|L$XGQ6|f{*oxu|EM*k
zuN_9XD#8A)e49q*<P|4p{n^G$!3*yMPc<?ceZEg3EbpTL0%(Noy$Kux+4CO94@%6!
z7e{Wb<bMin4ZG5*Ja2a8JuAV3I!1}p)y_Pw;$c)uYB8OP@x>9_krdz-QhbrkTx(>P
zGM7{b75|DU`WRguM)WdTke%+2RldoyxYykDQYt8OC=huOShrj=JPirW=B{}NZ;vko
zr)n6Fa5blo)NB7X>kk;29*gEJbsDd+z9jTQ<^QPbFRkvLVIfT_X0_%@;ny%V&K%+J
zEj2{o3_TX$0CAeufoDr1xV5c&Lj0eEeAhh3FYwUUw{|V;1g09~yWPOO6L5WjUT->m
zI`Z!aX>cFGJPg}~QGmDU_0ewRZgrTRx{`Z1sNL#3ZdZ>sZBDr=n!jFKSSkHU6A`v#
zoy9BQ<ZMJoQy~#It(_-|dtoR?-~NXZfGA|`XfuzF?1O4^Wmo3@eN&&ay%K|p8%zBU
zoPh;7O(jgDgv(R2Vd8laJG)hs(s`bS(OhXj&+{GtN2Ezp+tth;xTtYX-kKXZ(>-T1
z4ucKx9Z1$#wa&#&hz05b!f3bbzaP6Jo;#Fy3@r{sI4{D#Sg_#GrohoM6F8Hhz!bkK
zF*+w%wsTyD4@N3|{@gY}`Nx?rkm=_1-3=6Cqn4I4?P)l)R9c%<lKo4TcDvGJz4XPU
ziax#bcQ#vRcRV_0Lgi(K9^9*fy&EmFLE;O|OZ+Ek!nB<g#n8pR|13Zx!opqo&TtjZ
z09s<LtH$8M0=W9|2S{@Nnf(3Sd^?W(sf!!tZ-OGnhn43gf^H;{J#;sN%BY3Q@}D&&
zqn)p{8ht{r;M`^p=1!2Z+&#)J8Y71GY9))TrFbc))vd)#q&O?_qQ?xbM9m5$%VqGU
z<F9*{dQ)Xv6*Zy|WRugAcpRPm6cCulF=kHgpA$}89ZckjL;%HaQ{s3?=j6KeZH<f<
z8|&|0&$eE=1cwETUhHx0ZNPuJ$|mnEKdRMidayOx0@#UlCA%f(ecZ$B>@WQ%-2UX1
zdiqeTC}6|LYPl473b1`NoK6e%DuX>YY3fKcYW%`w5p!P^givJ%%V@!n%rzLX+|(T+
zB0SS@7c@^)fd;(f!WZWM#4}oTaAgusn6}gA#h7jxkBFq8YR}z=(Im5ZCxgT+1)d8+
zv4!r5<XQHl_Yr_*!BDm?zvy0$pX*eCbTd4p->PUjo@w_tv`d#4A|v6{io<(=)P6NB
zuVBnd{jM@?{R?5gVPCQf7jX8`1}HCFc$#-4@&Uvwx^sRlo86zVTBbD7R?C~#48KbE
zipapK;d3z*NNZw`ISBo5QvEaX=V&*|0qZo!;COqWicBj-*^h4COnIWU6h`&+RU(A~
z+<;qfv^-qJK};q<iS+8gK*c9&QozHg@c-`;RNwY$L|kBKqy^jkcxB=J`1d^T{TToq
z6tH!=rmnQ9))Kit@|&&L^U#QX6`^W>arje4g1*4fE$Z*tJ?g~G59oZqs6Ekwfj<eW
zbHkH28*DtH!H-w<+7mZ-ex~)dTTEH{&jN`HvM(uE>kn!xkp63}{NM0p+}_ecPM{SD
zyGvL*o*+5^t8cVD%2lBNFyk%pPVB8cGIJ@8({OOUC&a&Ug;ihf!M5r#n{+stmV~MD
zC?M$7X{D`wd<e>ovOQ(4(`dg(Nl%;@A}Cbtw@@Z%hqB&0d<Iu_w*2H9nZD0e8r!tN
zb%;Bzk49*7fila{ZSHKJhQXc<#PU__oq*0xH^OR~#)!}X>OmllGdA<!)y^>rBZ8kQ
z-aC4ATJqcJ5hQ-z-kj)OS5qa*)W=KN;zKLC@N?%;f{U0B@dl<gLBTv(skZ&~uGmd?
z)Z}p8Ck8~gm~Z|v8(x<MQa;8yT(olyr|`CIaB!bqtZ|Ia#po7ht<JbxJ}L+7PQTe%
zf0qi4gVVJ1zbywq4m4c@ec%t(s7R?^rc=XMEe)b3b+!?M^PlMiI#yt9ZgvPQfcrqg
zh>(laixmj6<^n|k6wVA%Tvm~u9aeXaEJD$RB|NEj7F+S3_DZ?q`-?tTP2Gjl9*Sg8
zmaZmJj?T}Ez~!zbFpiZbW*mX$bYY*dFa4TnPkIJ8yF>#G;KAkazOHtCO3*guSVDPt
z2UI=u_Sbs1`1)hUw+p@2XQk(<zn5?3TU`%KSF^Lc)dLM@E$j26b@*@Fd;jWnm+iol
zV{s9SjCBo6ITwFryPSlYRZinU=QC!oQ43C9`xb*dzZpu*g8Is#+80kYJ%FXpw6yF9
z@FU0WrW!P*8atn#nU%7?mIPd>>kc-Kmff^Efh0i@@ikJ$@*j5<XGuYK<0hi0Iij60
zR09+`EPq#8aX!}kS$U80!%_b>YsDruo5im9`NVY7Rn|(y+5S3Bi@S4Mwq39Dh|7Yj
z^-**P{=cE%i|PmCq)1Sp!prFZu$I84mQlW5;zHswnY@;v4wTGWab)lw^4obn`h}>z
zdnXLwUziKYNwETQhk{$N!I;c<Yo5;D9KkHwEj`fLpS@e5A`p6ynDQ3Uk(Oh2s^=RQ
z+&437O^m?w99*g*#+1d<W5Gi1dty1>v!RfDHvFqc*0?O%<xaEWo0)DPZSLRP4_zwi
z!QAs5Z`LCMJ?r~_n;sIsgVoIs`So!^;H%74Ppn_tmVdVI=8qcb(JZTAQtyg7{N9x<
zScepfSY>HUz$Wtp<_bS2ZGd`{{rDc5^`;8(r^7ANDH$1-dQ%uLUlRcw&$!){lj+TK
zGfd&!7iRZ?&`n$vF<f07Kud%3ht?k=8AMov^HAdiWNfmg|0Fz((0^RiKR@<hkr1!m
z79%yI#g>d9l`A(!_r`SX-<vLMnh4itT8!asOkCprUv8MAb7~KD#Zp;Y8H~N7CoO8=
zme@uFs&=Kz7*cjGA?(B@X%JFnG-u_eMO>?H8qz(`(bpRWL?Y>TLrT=^=Vwj<z+e<F
zX3UFdk&f>}aFA#Fz~6@pLWTowP3sBpnBBfZr8Q<}Z4X8wZq^^Mb!MJ1$pdDmt6A_b
z$-W*M3=|{G-h34ZK4;lI8T7hvd(1%WoV{KYGyW!{doC966dj^gU~0}>ZqrUNFpaax
zlKA8t?n?(qx9P8cLy5Tg#*E8VX6sNdTg|S<%YTjKJICe|0(YWAHW>CZp@V=@T81G|
zAH!hs3ujD#co`}wj$B83=b+*t8yz`KDpI4o5D@{cR-S!|<GBxoTz0Db#|Gd3L*qj8
zV))W_1B3gk!-w1LG(s+W|E?}kV|!o3WB~wZ$rZ)`F{q2E2U3DFLhlhGI}vB^xN9vV
z(AAD=w_=n<mKI7OL9a*=^okX##m&1MuZJ0wS<e1!6<*zp5;2TF&;my=%VvS1$g_=6
z?Z8kYr^o63CZoI}=Ob+-xT%}7&w7L-_k5k!Ha3vc@ibz|nMrqXU80kT(x!!S?FO`u
zoD%9(#g1AJ^;<2K$xEkdUtwD;>&saBE4d3n#@05l0#hnE$OE{*-9%CKgx5aN5d_!I
zrkq3YF8NF`7t)g1x-v>)H0%Y!N+qQ@gCUy_HS1t%P2%HuDClCy_eMs?ybB|Yy(^2I
zS|-5mL9Zr{dO8lHa`A^WCy~n21D>%&`lTk%zOp;<<Q)~FwzaPq2{uD~E-=X3k@6>L
zwB8>JFV~WEL+cU0s8#0E_r?UY==GbO_bX%&Hcr%wTK&%U9>Jc%PJ3y6T>?`QzSj%&
z4i^Wr(Oq3#e|Pj=z~~x{K%58Oi#Is@d2b+6-^JeABKxe)^-Q1Ctr;M=n<$f~o#S01
zGS$D9vJ%fx8*_6M&=Ob(=xjaMScsD-5!K72;;60)b0Fftzp$1pTGc(VK(};W1|@30
zFu}A`w^|_t>48vwd{?9LuLsA;b|SeATJ!z~N;RU7?)UF1qKBhW9I;`l)kGw=dsz1f
z<iJlD#6Ly0-Q4+V3}h7EujdDdg?~uxCttU(VAj%B3859wj16GN0$>#xTxj2xQ*Il=
zb4ey*E$QRa(l=A`Ni<S=*xhUsCC}qajwf;&uXPqJ?75GnyGVDC><6kR=ufQ`!K#Pr
zJ$}yPK2l%lk3_;^eHm!cP_nbfm6b8UeE$5|Y)rUCp-Ej8@bKl};Ym6<u{}FKmj<A@
zvIKKgKm5XmQoe}%4^Bu%{%~J-?JM53TvFYACaN#c%%3tRY}TQ}hiIeosEaqWaQc)m
zz86&25f&514>aT(Zn5TUP_eEm6taQQu2zamr)i44R_b>9x>cH`gbr=S`?i?$6Wrvq
z{&c%r56p1Rt*spS7h+_YI4D=d1aMECEDiU<%WXy_MbW=oGLK2JuO<|CO34a=+sHLZ
zdV~v(j-chEe4n%3m4WN%ex_9M1wWM2z23msRQEE$T*3NmYJWa2xOaFFzpAkH2nRfW
zRp+fiyf}L8o?O`^PPa3%fPjFyh6W6Ld?{B~fK#`@F~g`gd=jFKqe@?}uNMsxa_*Fp
zwZOF!^}}~dnw2K86fPpd!ZVeos7`w-70$8$rJAgn6~q>auJy>uNx_1R^Oh>>rb;c=
zbN)Lk4Vwg{Fyq0edOWDSBAUHGWrllR{npM^q;E=czPNDJZ(?dalQ9mJa!OL)9=LQ>
zTX#}-U!u=0A75KLR=ni2Mt`h+tSVW%guE)A+zFv|Spcd0133pryf}$h%;o;{@`;v`
z5{9It<iTuZP-*KEz7M%{9Fw7Fcl05mU2Xo=n`7`zt%fGcS!T(j7f)<$DOW8En*U{M
zPsp^1e>JYa?=jJql|V`AqdtH-thQqD`cNnK`G5_m9#Jqy(dJoL%FC46e{ZTHqI9*#
z&*i3G9wUg-C)>9MmIK3~fV8tP{8uhGFz74TJ6|<WjlWk_drggqgF6mmJn|UL>tu;`
z1s}d|eA~ASsx0NX<YvzlN{cg@@@UDevT|oj&MIOnA>y;x<|?!f2lN=SbtP27Q;e|q
z8L;rvM}3q>Q=%8JUv^5ev;D6Q=N+yNX6K2dAebLMe203fuN}Rdio^DBECYwlQd2|Y
z|8{fN!5}ODNkqsyPtRfbzQ2;^su7KR2eeQ#`eU~BsF2s+=aV2`!x6A055l!|A74>~
z@^+TQDRBa^+STC~6r9mPT1D-A{~Gsxs%C-1%o?CtDskbMgfw@8kpkV}W3bVHv(dhx
z{4@^2sP}i9D)y}@rY9b+$nZx2OT&?c%MaVwOWECUau6<m_kKgL4CNx!;n!Dob-uOc
zR9ne!c95!4c_~`@pWlKAwz8&Pe+PBJFJ;na0FDKU2pc~(_PaP<dS4<&kCTGb1Z&y*
z-s*?_bbo*F2>*TPx6Z)T?d>47B5+GtQ9(h_-k!NPf@FI{;JFJ5y<!)q?}Dg;G4b(%
z$;^tCBx1#vg@gQu;YZ%>Dfo8oR&Ok&l`*Q7Gh@a|Ys_8S2)OHnba`wi1#5^IfDLfV
zWhU~yKo|9M+PK24z^JEL?|g&6&(&`;4KvCg60;718aIE-3gXo#=TX*4V-^#ck06vM
z_-$+4oe`2#+m!yi-xowse_Qq%AuK9gAPv=nnR+g^eQS4sv-t`wV#uw&Dm#d)0+adX
zvfCy`@}COL&MK_#!lyxEo#KznFo0b^nE>Yc7VGCY%k9CyPF`*g!k?=dn^(G}Mx@0D
z#bZRnzB+uIs%KWuPcl^O7|_}u)zF5LA+QtoWu(=#U$NpU1TYepER~|T*B0mgdr5Xy
zhpoZ!&CVdyWu^p$9HrHronKj5q!}3*yf-U;wcMhgtQ*a#`kk7CNWAL+M=p-q;4JPJ
z;~o5l2VXC$;+|cJe1K$LOtzVgyv7WE&$Mz>Pz$>h21A>g<V7<U{JAy_sW&55TOFn?
zGfr4+V2%7m6XYJOuOF8RI7|sEd9<<nUH(x`@F0j(-}x40hIq)Gujh}%Zde008k6`v
za3b33#vRO*kf@zQVDnw{e&+)A=uD(P<`>WF6G{{L{MhJ(b{AA2lgtJ5z#p(r<h*uv
zGcJy^Rcoe)vGfWMLbTc+^)~XhD|QM-V$gFpN9VN0ENu~t-yAWX#`568V?N|3FRMj3
z7U+<;q_2dd*j--x7;&K`D%Z=cA%1hfXAiGo-NVsqhDH1Ybn%InPGe$Xz80LXu`0UA
zz8T9B2UlWx=H~w1xNh34&sLfyvi`tmZ*S-1<Qy$G5^uTND;lrQ83Z%2ZLK>5kKdEC
zrKM$Nc6KaZk+dhAXdpu*Xz(bntn5#YbUd?R*IQpdzw+|(_K}f@r^oyAA&%vko(C83
zot%5eha2D&KTJ7qqTH|>Q7jZ!NJ8Q*J3D(_UEOGdlSNM`UgzFa5k0qcrn{<#ugk$q
zdCmX<yJh~-(f`t$6Xc{I*M-ZHq?SIh{P;kLyg-9{i*JCG#va(B_a{$ga&oKu$jqW$
zqQ8U@rLAX7jhdyLapLSfc6a-&r)Ls0rw{O{nIbVD|K5{ZVfj$oPv&gow>pJpTe6~o
zFxw954;NJXDCXP4oLOAQP5HtcM(F9jbmc$kAvKIgmcN1qa5*qK#VeNzuKaKPhTF#D
zTghqpn-^5wo0v7I3CejT@#w7z+{)3kB^GPGO&;4F$LLLH{McpN-P<W0VGxW9Z~{U@
z!cKm$;>tSWO_1>51nr?Jfq95-eyRV1mLREcs)-{}X2+4+k()>?E7I+JYpl-BsO4r2
z$<WY{ii&Eb?dNNm1SVJk0Rg!bZq;vTz|$?Io}M1><)re%)BQ0SIk}>Wil~ap2e2eE
zyzZ}%@tFNPI+9|)-3=6~7M)yN02^cCB%AU2-k&Ha0>QOj3`|VTR&T!2rZYrXSXd4+
zh36J|Hat^9!_Q9p(9Mh>;x+Kv)$ttZn3$M9(b1jdM!g9fHpC)<$Qk;r-dMyeiY$eE
z?F6{E|N3KTVR2bZkP8Y5rkek^Cdd;Gawy|(J{-J9rJ*})L27WAfJl0PlFL-o?;7(c
zkP!Oi@k2tp9$E3<CnRpL!M0Mgt+|5Y8<%@sKYwzIIP`DzX|z6dKea!oCKKQ=;j{}@
zmor93BX8E@IUml4?UP9E>}Y@I8D{*Bj;fmaTmMWG<K7_RNi{fZdVXQN$?j6J-s6Ls
zs?wrybCP^3sfP)a?I1p+{(^+?iST==QGc>S{lQ<yCwo^L5M$_iPnR8=P-~b<ycFTP
z<GELZk~Oi@!bSkDTrF&Z;{%9R^(%vsRKBjB9yRIKO4AXO!GtfyEveDbBKuH?q>Kz*
z<&ZBN5-zVR6S&#DX>D=2H+8W)fsi%kIx%fJ^tXF7Lj<5b$g3#C&jXiRTkelYsi_OZ
z)-b2C#@{W0{FV9HV!4G+IZuw9mNulpY5xo?9$S#hU|?YM>y|bVh`+GgHMv;)l(yZ=
zp!TmwM!tjcM|xSR?OpEaYUQR&?!o8%Byjdctazhw{G|eKzQVgQBAS4xQ)OhC5%qA#
zfDJ&xusWY@kU`OJquI16$?2V#PanBr3<)i(iK)h?n0#%pT`Vw$v;4rhgD5@oc*wnp
zpP(`F01EW5_pdl;<w`?7-dgI^%W9(>3)hyvj&*ObNeOWKA|<<2p+o0WJJKevURF`L
z6<Mm^!B^)j>*FGwuY#QNgzPjQDCEc=3h=@1E5}q|aTvXFHm1r^3ID#1w>Ln#9QxJB
z>0W#u=~GwAQ75VZo$h-iq;>G|H~vMEs!IKmOJYX}6YF$Y_7Yc9!v;%ZZOss5wNe9u
zV6minUCkzykdTm!IE?Ti_!Je<BH=KQefpH)=xF=x+ovE@l3evtJ(z>}+K3!LI(}(%
z1ng9<x0~_O&yOp9s2F&7lHg@xS?JAr^6u{3+O^i3r@bV#t^!4*!`azna%uc1g6CU9
zzkmJu$iab=HKsAwyEzbl3tsbMEHj|{OYolJt_z`%kPy~;SOh8>nzdMUy)TXW(FLI#
zvLMR%qq*cySdg}T_x;w=4`l1cq8lJAE>JH;4hQJw#f$}13&!Kzt*;%o@73aDYA3$2
zP)@H7VP2c*l#=|F6VtXz#~F)6VSDAYJ*LUHI?a3I9HhTDIagG@9e7WJxvDCb@Y>Nf
zXGIV@{#qOznl+0e%W%Xc`^~w9IZ><3YqQ3vC=kKxn-E<dRBUA;A9(P0EL9HzZLz*N
zYUxo%S%1~O(H_-;$Ng_`+?ZrIBBI#IeXrX<Z5lB|9OTkD7wOYVRD58GTYtg^mD3y?
z%W3?!^)<)ZQJmwgokB}l-xRIB&cTR(wxM96%3=F6e^8h*oG}LouL;H{EA2ALoRN&J
z4;U$Oy<jCI@_PZ?Fd%v?0MCyvKNm-f1Yj?#yytVj*l}X>f<PcYCUQ|;PRwe)hETve
z8HIoy4F^ZuVC539pp;KV_DM=2z+u$un!3D`RI{?OdWVW?3ZBiy@p4)i0mm8G1J7rS
zB6O#Zz}%tJP33Z+C<Ot`snmiMe_?TPcM2#{|A7I;m_pxP9uO|psgL^e=MTZ%TSUaq
zR8+t6Qr-L4aKB#WUtys9cW`U&Ai<?U0(beKia6IrkE%MDM{4Hm`Fy38xBkxZmmN|@
zIWZOqIIkx%WASs5)<MJlJ@N2`n-S+ayQfM~RO89D!-igUIJZ8JI3?|j99Ah1P||-l
z#$(-*0+?Ugs_QNB#-8*xWT^fyuz?>8Q|z~Rxb|gEoJ-_QZqnVo9yV&Ky7KejuJ;J*
z(HdBhPu-f;U(VLc6Z=!3-qLlmg(1U-s;}!CEBQwyR6oQZH9_EKZz~aHpDwsYWTNkR
z3ilF8o8^7!Wikqi3bV1#Vn8_I@77j<#YPuOZtjMTr<dAcrnIcAs90+}ZlQg=?g=9(
z_0R<o8oU~YAN;a0K~ye~r&i>4+<Xo5D>%4;K~qadXV9kg5nixJ4U)iVXBc3oo!PL4
zjEpRrr3g<&$m?3BW%OYxZ=BTi`e0T$#qOm9gcqtSO0s`y0jjs?Nz?u_&r|yJ>pPpm
zpi!~%#6hxp_LE9!q&qt4Z<tt7``A3zbKK>LjbbTua=tmdFPl)A2W7Hg?hON0O4fK<
zkeb~ZJ-}0@r0q!0Ic;Y&+MDHSaP}5dDkA2NpWUcu&5Elz-5N1Qx!)}33RRyIE}DQ8
zGgRjS|MmyKrL*AhcXB}TKfPiYVXAAh=Va|0qL0-|_~H_lfXbNNaz;>3j~I;I7TB>5
z=Bh*JwQFK1z!mVhua9?^nzhzMsJthFiC$s&Y=s?Gx(;BI8fQjGCyOneD)8FPiN_W2
zesFTKD5cl;5MDH3ifM_9iwpGgJ0t%13HX$8u??mqO`&;-i@vVi+|10$dW67P7p6QD
z$es9~9~@HM_o%s@4<s}+aEXbDr|@g@lcgwK=c_H$|9>b`WGE1s&)Xy|+zuB3`|?-m
zA4zNX?55Db-VSPQbm(Kx*_-Ov7LACKvh2M+U6|>!5k;X~rB|@qGyv2@6_!COcpQ-l
zKQy0hOCTYyx;hiIR3?w9ozCbsS9WF2Gl;0^Xu7_u@T1MotDlAtfX9bz=jfO%9zlGu
zGv+s~4-iH7i8?y6zGN6kG++%z(}l8gbJw<<&nZpzpb+gNb`UPVefuUVF7B|}5wNd!
zfB&G!oibc}2INe~#5}P5FtxY0-?h&Y3mZJ@)S1GfZ2N^yYBp7f#bPpOFqZjMR8%yk
zWovb{y{NRMe`hS~^z^jAeI-*Vd#@EO6UvMiN|qe@iIySEcTmStm(cdqp||xbA1m~1
ze?QfbIntj0=Y@3NM&2f95`;xZFn=EHIUV*yuHmfx1JFj|^hx)LscZQ9<*}q@IGrF~
zLZGvIDBOIfk@0<!GI>n2Bs?7AI<I+8Er=}@g?IE~u;P18r|u<NV?s;v-%P{E3-fSL
zR7qnhXxkougUXZx>On>Y3d6$w-63U)-U_E0-i+_ee<ipv#Z4>`LCfRC@zCi-d5-&}
zF{=rvf6AZF`^XaEza9@N?2denm6hF4m)F38A6?Q18H=p1Mv(45rulVtUrXc+O!n>1
z9aP+p&QMk~yR9~N%bm+&Jv2e%3;be-l47g3m&r(K7+B8Hl=3|WOega;t*0GGj4}1$
zT#rkG%i;}c-OJnQg-2tJ3SM@46E9Q}cz*KO8cCDa(IL!}Pcs42nNFk3r!Sh4#qI37
z_suFos`uTl`C<u^e6^`Wu1unmJG$4!m^fQaINI9Li%uy4&GqIZrT|ShjwWj>8xcQ#
z$PT~tN06D<eiF7T0)=q}=(v7XKOpDfe+-<S(n7i9X3HEvdVF^@dcrOWnAJ<)4_CMx
zxicK{_>CDftvaa#s4uQxujz4-(i!)o5_6+T=_m7Z#o+T`Yu80+cWTe$+^E-qpw7c)
zd7+XRDUWHX@$+>ORe!ym$=2gJM`rmr+Qaj~qKl5|<#XHV^k&6%9QEh?{I_+r`E`{_
z()1JKHO8HnLwom!)9gTHaf|WnjfrNxsn$_577pU8OsbC_Hb)PSb|2XATt4NeZLha?
zzP@(g-*)H;0LQ~4_Y-{nV`&0@kRm_1Mef@=CA$`6W}5cixs%fz6_{}{;|7`$#?`j#
zB(B<1svNHF+GcOQ^_r1MblDOH>#!=VHJ9YAe<Smfe2FS4U)=zf>o`3}!s+-DOaudK
zV?`j18BcDhWJA9%(38AwuT0Krk4s!`8c0=wg%w*!?^hJefzLBg#1=$*BZO9x#Or2%
z*=30Eo-m)SEh#EW*ucO5@@U42&!AHWV>_R1n3~FAO?JH481MafhsO#{s?^2Az>s;$
zZYt=xmA9`?#s=uMMl-)6f!eDbB=&tET}|NgU<dOj7o_t}#M&9(5_sG0L29F(=e3Vt
z%=Vo~2<=Eni$3V7Hf&M9SRA78i6wfSh#3Z_m8tk}Q85NbDFIzv?7@22idX7Dzjue<
z4bAWAGXI}m146eAPj^o(kdt|$_48TH8O;#2Bo0t02n?lRr3|U76Nn4BH+qM+5E;au
zXsKYbo}!?r(h~^KQ`k-8?IK)ljn2X~LE5C=-y(AAOMO!!_ISRh>;>k#{TwJKC}?y_
z56!0Vo#{1tGoGzPil9FA$PJv*KgkdrsMLh)xnjR`Cu`NGB{&7`a2yKVHF2)Blr|<t
z#{lh&5hIOuLtKAFC@D`W{vT0q9TnC0{SVVfD=7_vq;$7{igZXf(%l^wNm05R5s>bd
z2I-RS?(UF$4)4$Rx1P0zKb(ck%)RIAv-hh8NjeGl`a6?on)Xm9e7V=6BMrQ1_^+$8
zj=ckcGFJii;oSBpxR-{1!t$vQVGu6Nb3dJZD#R?i=-8m^E6g5Pal!mME5|^k@GGbg
zFHa61r{x~^WKRCY$w%7{jkgG!lOGUxi4`PODJ^v$zr>pDh-gE+S|7LalH#0-x9BNq
z0_LTYtR9W<B%p=fM8x6IM(IM2Tg?3lRn%iLCf}7CW0<AYSw4%IhE2!u3Rf+;Qa-xs
zJ<E4?m$NQ#p4D89SfXHA)OF`67wm4Fp3d1$>Rok{L(d5bHTL${*w`leNd;UQH@HMZ
zT6Q;#`{M~oc%h4nCbpy<H9hZ69C<;5Q5oG&VU@xrCQqY`KbZaD%Y5!Ve#vJnQp{mn
z(5BntBV<bcE#NAG4B1S1NJMRH-UqL6?65j<t6tWc;xiOVg)CJ^Cbn!jtZy9{b>^N-
z|4b+^V-r1imnq46-GT2iq(|kAvw<mpBmLzsq-1UY$&=khOWj?5TG#$)y2>(^qW!_b
zK$E#e{y_|tjH}X*{^9u!(kYQ}YmQobG-sddM#bk{F{G9gI-v|g2hRqgJchU;uYX$6
zmuc`?e4%H-Bf6u16d#of{3ZzVri{vNzh8iN!DiA{nV2;7_fpu-;7CKY@i;72<-n3D
zElaor8vSGT7^?)^^6Te9#_-TT%Tm&Yt=;{Wg|Z=%2Xn@v;eg=AZ2j!!Q`|B}g8INz
z)qBhIi~3~M{>cb6hw}u|QDMPpbed~FEsUI(xJ1**@GUv^9)qG4IAd<BTT|usHB(S1
z1|6etoJjh^x6#yBJ3*aegPFn7gRfbb<2qKAp>GkJl}4+kf}F1G23ekyGtitt8eh2;
zOVOZjCK*@8?TwLg+ZV4lRz(&V?Z{VqzV+3J1w!t}vE*(<$F<Tqaa|}?q?7sSU49hL
z{OlgDx3eM&5(HciwJjUSp}QzT4wG^9TU8L-v0pi?3W*r?6lZjTu9}d?hJ;bA#ApQi
z`5piWUz)vdcZF0L85y4d9Dodb4G-VEJ!;wNB+wgg$e2yENCnOI)N`w?|0gEMuB%Jz
z>FLpg|I33s#=*gXK~||>M-z4GBIG&M{?pv#`!m14$1Ywq@s-5k`dh2r@U~>x{Ztv_
zFXs`<$|`a@j(Oa+(_(mUp(@&cf8Dw7EZNL#AsMy0+tmBBJly$ecK^q!ARa$U%tJq1
zYT|Xw@BDAoOi!6j%kMgEv#mO5#1vI~E4EtV{Tdc@mTy^KZtFVQ5oC8?Z6S2b6}KvN
zUCOJzxScw=;mMw;I62VESuV-I@EQ|E-JZamN2K?6s}S+RLme#o3eAxW*3>g%ER8g%
zi&|hZ>T1&=y582exf+bNR3ly|&wZTQ+^1kou(w%byY#v+O`AK$z@pBx_3SC~+fch8
z&&F4GLM-5PO5-hC9DN=PEixmzk`AoI)ErCc54P#RNORD`G}58Z&c^s$q894@BMa-R
za-IuSjMo~d*lLdVgBtvW^bL@=HKgqbe&jHD{YBD6dFLOGVf!F<x^m55J5pi#>><s6
zMGCQqV_Aeob9t`+$LQ!;5@mI~dbvwQ&TWk1kMUMlwd{-YYJT_<)zP04&tUnDXwkrW
zXkRGuygKx{oH5;BYRL!<MP~rk^1UgGYE3CADPEsDUMh)jXdM)YB3;|@JVobAVluL!
z3gdn{#+dX)rNW6+A<y`9Ay0?ZKPWIifBt;->{*@lg1pFml}9}|9Aq+t(*gnz3JPrQ
zMBWvveo|NOIa%vY0$|V+3T1sikS0*;;l4YbFL?3?2QZz@{Bpl0Ea;h;H<#P|1<wXJ
zCJZOy=%FqRFg^If!f(%3Gp2A`zvAJky}4FXQF(!b)A2*pzhEvnc<&0~r#?{_*)kIy
zU7oJZgp}7?ynx%09Dn<#EgH`)hn`Z6Q|mi>RJAzJmi4;B^r&l*$+wP)mbd;h1RtSn
zX7+yPpOx31nt61&tLEOrbwv|j;}lXmN$)p#$gsO{0z*JS$m-thH6OufPIu3eHIwn=
z?yFH*)YHD@VIC-xy0C0$lWi};?s7P5*#;$}8eW8Sud${oP=4-97GHP#{%=LGRpAP-
zi<MX_bH6(c6g%I!aZbZO;GwrR2d4B$4;3{0OD$L0@qcYzc6FINT0IKaBE84U>G<mi
zlBU|VN_Q8IBh4s5LgTpRMR-NxJQrRVh>MO{wQ(BdhUwL!amB`67G`@dWp4tDZWj#n
zt_qq2Sls?<@FSoSyWb-hNHJR2GRCl-ADr|stsfhHQ!Wry(*!`YT@?%ggykX3Q8%>N
zr1z3p$lQ*V+(&YPrN@(%96y($mJ>4V(OQ$?r_TcAgM#yYcQ~mb;9Ji3rmz5hVr6B;
z%E2K{6Juav5(FUhh2PF=fLbch-P*~`_@&b@E3GM;Axs)5))s(*Q&w9m0e;iQMh*-r
zBO@ce%9buCtDRgW<j<eCf|+0G{vLD~k^teIoSf`2s@Q$d*4EyJCX0!|$UrG1bYJ7+
z=jUgGlE<v`JJ#i3*1+6cqX`0W2>>Vv21`m~N!+x;9xj#F-Vd4F>p2!y#O|7{wY999
z8!s3{K!8PC=3?vbzpSCC$26~B+k-JftJx+!hlP@IrDm!WdQDAj2jEgIlc`(N*#?Sd
zSHi^~`ora+9ElT8xf{%Gli~}_rP((Bo~;DfvaJ^AXOTjpS7>`$<yIj^UO_DDB=K+l
zz1oU;^oIphtBq9k@`$_s84Qk(LhHH%%jdiBN-#Hrs@0|klfQz|LQg(uWa=MHRhxb-
zz+-TM>UD26!}RZIzn<3k5st3E5D?K~oH%TfPhGHs0)C!**_B6zr!|*XJlL?d<5X!*
z$~~43{Yo$PCcM}g8|5L_@m(6telzqtHs7G_{WTMYqa_-d7H9F|D2l*VX%r2aXBYYG
zwuVXD2WDg1eq)jw1GJ5QuWyQ%U`j~jsqI}ul^2aZNZ${S*vY?KmUH((34W|?8u}$n
zs>yoV0nv(P1-fDz82?tr%w+ClxW0xf>T4YxO73pXP(4BjnGCO)l0wsHuP8h=k5=Z+
zVP+^Ezr26<w_V(kc5Py^<<J@Qp$IF&oTfu0U|b01pL?nc;C5F`PLYz3@Y=5;JUmoC
z7JKjJ=T4U!A~Q2HBSqlp18Dz^R!vb`I%L7)k_v!nV=&}R&rS+Wh2-Ya=E|iN)TV)3
zS<p6DqSLTjb&@=;+@H<-^k>;RULdS6$GY5LC)v~2-AFv2c5W2)c0;<zcJEfqPSuoe
z5C0S(JiYKI3m34;(WpZ<yl^fKZUfzAa;Z$qindkKqrByx$^Dny#;g3t$Dj~n&z!e&
z>b7?5$MS~=def<cN!7uGr<=TQ4s3=8Wamk#5a}EE#p}Xnk{`#%@Wl3eL*%zk%iqCy
z?q2cS*;L!E(3r2->n&6kKAK09TMbiEF(bZre;r>_Z7N?&ycj6a5?}6f`KWLok2L3?
zADe(Xa8(}fmteu9%^d8HSO!tzNGVTy^0M2kNV^3uK73ssL{)MC&9G|0-c4kJ!R?wU
zMu(81Ro^R<`xhR9I<DLn;sOu1Hs%IDGn9XKKL2!QU`sY4ef6Brn!19YW)q!!3@u5f
z3$)qq??0j5rGX~rka^kobsANiYQdQ&1^L53fCE?fVu@hQwy~h*4s>~2sPR=q{$4l*
z?-PMi1qVSme!KW!_U!PNno$f&rbtD4*}NJ`9QFd%+KS~&dBBZVq!@IIJ&hlS>^V6(
zsRFLyYuyo!*T)7!86xt4n*^~X>YAF^xVY~?A)xvYst@20UV*6$x>qs1ZU@y<UqeIJ
z0rjR2W=$hg$Q0+ri#C9G44n;&MEHv@ye@ysuo9V+)H{?WpwaBb7mP{zijR*Fv?hXr
zg3mB8UbC}D10D(dFGWj{2%*cD0D6HbR4aWNT}k+zzJe4PP*=Ww{i=U`yb?;tfg~Ux
z&=j6V-$kF$@4ga9;P$i-%M9QdRg0dmdk7Z~4=g=2fH$^XpR55ktB$T|_G|JVYD-E=
z%DeM1c?4u+o=XMLR(F($V3N%{VzVc$VBsNLrBWlhxu(B>U2ISuZFW?`7$-Ya>MG3q
zDX1?=6e1wc<KfF=KUF>Bnl+Dc^rwyFH-FY*N_Qu`eHE$4ZGhB-F<$6QIQmOy9rh%L
z`cTkY55bL#AxxQABj|j)C3F5wW)=<!<N9>8&bfQ}AvWqkkaHJRPs%|?QM3gO1iinB
z?$0JBd0jvpvEOeL5V6zQnv`XKT=XcyI58!bonnAfuS2kbnNsm_+Wog8e4|Cqx9_(@
zlDDJs-O)^3Cl!8|n6Vj5GzT&e`$L-Ft`!;gZ`yQgubGElhJ2T!+Ju}^qM;}mn-AUH
zH*CuniqrKIkK}DB{{pI)>n|Qnru;5aH9Vgmj24!i7dXlCCrT5}?-CRKVe+gsJWeg=
zL^~;L`t-k1<cSw-@^t;}dNfpJJ|;8k`tBV9V7HmTPjQ>Kx+bK_?saV!ghg&Mu1-Qi
zlE`k1`stU<h<lo^ukWr$-36OVDmX4e$wksn)-a?8u!-UoVHd}8<(FJ2fs7+_<YNSo
zq&XthH*Q+JOp*x#0XsG}Hv9{Cphzsn5~jfUs`bER-JD<&(p?ZFd_GSM=u-4Pe3cry
z*E5HjtPzy}r3O}rQ3}B=JOhU)-N(VG3U(pOBx*S|(lmWrGEic;klbWl_Dd!#1_d)L
zv_)J)imsLEJNT!Ej?wq_US-4B_MM>p=veP2EA~n}ZkehhijldSjM3(QDFIGvy!RW^
z?!0UJTOaI<kY_QDLbZMzCe;y6O!<9DnxCq!MrMR6KRl*sCdNyvn7<C-9~U<GQSypp
z#flBQ@&%F`Y~w7K>1Zi9Wt>B<t!X8Nozhna;25MC%Fdj+I+y%a?*;zaxDjY4d%hHW
z<B3H}mVArrt@~1Le$xz0L^%|%6Dsboi_-S(f#B1)v?c)kAVBY(w0|uXIPAe=z9AH1
z9yta;KqG%EF+RSD%oTVNS;`?hBx_$gwEsgzm2=m8UwHKXt54|CW(X5oLmddP8m`hk
z1wT1+d-7Vpv&l4z@Lc^Hwl>C-N(yBDAdKy7U6jLj)qP7xyEBV`eMrMAfkY;-RQA@Z
z11==_{_BY$j8v(7m+8K<DGj_DS^>n)Kz#8*@?zbO?8728YzJ6%K~V0GUEeaPVll;>
z)~-pUq{U9(z0KM77~xM6e@!hG2$%nR^UwT_<!M+(>^A!FKDp9AYZsC+DVS-g*4&Y<
z^VDg<m9Zx(B)8@Hi551Q*SlkkQ6^Fy-feV5HRj}4oA+>UIBGdmB`VScas3e{9P7Ne
zipuQH@qlF5UamFN#Z(b7u~!DQq=yPi{O`5IvN+s*r{?D7#&uBr?64|bU0;3Y>^uEb
z?!qcKrlhF&RZ1~D#g+pnT&^yLM6e`|pr@F<xl+5cT6nKi!$U{K9ML{SKC<qNldzm#
zDLz9csHuT!F%z~b1VYVSIA-ROb5O;(-Mv@b7P_0$7ND8%e(?9t#R12-!Vd&BstPwp
zTxXjt-8C|&0}knMl#=??y(#TAG6B!wl#+(GR3@yGW|Vp%asFYtGYuPvIe#9UQ~&6g
z_I#7ideefyQBQbDCU^b5DqB{Wmr|VHz$RzTd3V{o?`ovl5y}>4TI-68iIv;eo08(x
z?#=v}hb}E(3FroteLS(ANV%%*$=U6<RvI>JTMztzqB8Aij+E2TWphAt+#zvBe`TbA
zV2va}QEP_%$@zcF>KXs@jCll6Gv!56xH>nA_|R*eI>dQhEoHTxEmx`g>Qc`j_spGB
z%qEvZIS8*qLB2yksx)AqVP4C_z(xmpE?GF@aCfD%e`|sRJ`9hpUF^d<d{cadRE@Xo
z;bZ%WE%80(R6D1Cd_6e!iuSGTFpgtsPWlS`Og;#H_WAuX@BZe}hWxFn7f~v+-<J<E
z_$~!`LkIJg$${5Y95I=)%)zf-^Af%w5n&>2*n}4563v`~PEar?Oco5x1$Dxn8Fj_^
ziTS(IE|efvD|Kd)*E$>rGtu=i<dFBl(4vhqFuhkhqx+pK0njb`Q+wm(&=DuMJkO>!
zDbQ*ckyzz@V+lm9ws&?b)mhR#NFiR7fT!zaFzU`N<ZC#Y4wEr<%zqd3Q$|!yiJwn$
z2O`m`_`a#5`P94f_~`A5s^l9zx<d9wCoZvgQ*fJ8Ls3#WIjl?x?N!<UQv3G31zGXY
zBfzDDKgC7@@u|i>4?v=VYXr@E!6bYK-|@l{HL0%HEhR@}?*RHP?ap$k5~R5DizM9+
z8S%n8BB`EaFbd8*$O9Sv@tyTBER=0oD6Eo$SCsNk1lu(&3XT+h-XKySfX{ED+Ok{<
z-t?P+C+3Fd+4S85do)%zhsDF#g~?eB*~_7*&N%}KX2BK9=?C2WWiv_#X-Cb=`lieK
zi-S~M$LBR0FUgiC-~>uY8*bOVjZAEUw7#3H>-Ght)w6cB=0-Z;DIF`&WI=m4QNKvD
z(>1ZsP5Y7i4P40tH?ZP^?&X?JgbN_yz=}p}RoUD4*llWXF4@WN@LL^GATq%AP1sdD
zt1<fI^d`f(@}YYp3nkR;3S~XMx@Br`U;{xS2)kj2cfU90hWnkz>p$_Iw>Euk)qfax
zbNY_Q5NP~6t&QElm*#o-!5tE{C3sdA=R<2ARZ>cYSjHR^yNz}yXv0h;;uU?&*5tG(
z2jDu!Uq0eDGmQE+NLbu4K1+eqOS}>}3_B$EX|ip-B&sbHUv1<ujld+u7DMbtiUMke
z7|v+^C)%o?MHzMx?C!p<quG+c#ndzz)oaAd$M*s>N1(CUUTlH{hK6L%PWVKh%k;Ad
zKF`2p`Y7TVFD>yvk+((Nblda0EMq)T7D(-$v*#JYhbq<KfEE9YeC+!t#;cCe!ec*J
zjqHC?>M0>f^{GQRyO>q^aSZVyj~Hs<w0;p>cMW(5a}%tr9NU|3yu&WOVBv%;`Fz`8
zVK!P_onqrC70QCj&eozxbSiKT3Y^?paZ)jQp5IepFibe5{w0^5ORn_Q!r1n$EF})4
zJ<EoXrG@W9*dO-)!w&Y+i@(*?4|_{FH!TmWg?jh8zc+*2e4Aj&!+c~B<M2Vne`D>;
zxH6)UdOvSIArgw_`B`5aKa`8YEKLbk%dxDYmd49zm>)`E=wtYI&^z2K*T~QWEm|P@
zaFGBg@7zAFS&pEin}D}^7v|e31KGdTdVUV~2H9`XaE&dwlA@B9@P1(;xqhnBQ<CsU
zA9b{Jj8YrgZRNeg3zX=Es#&?-^2_nWP|9Ln>2thXL9ygee2i@m;=zaC#$7Ld!>~0)
zyD2S?Qw3EA^Jslh!)TG-AI!{1Id?94zWLhTs+Spe1C{LNT2e7f?P-r8WTJorz_s%m
z9$lPN)~!3BK>i&;kBk;KsO?R1G*lLsRCkx_TO<!_#`AFeYx|!Z<3GzXd}E(E>sa%}
z%^TT>^2*%VecTmKopZz8m_?`mUX%O%kdcuKI`(;`oVRmogv<QkJms_q8Zi30xl#Ph
z+A$`>x@uCdF<6sM`+pw=1%!1vB5S)QX6Kp82-CVQHvUvJ9j_o>XCH)F@#&EC1Yabn
z0~wg@Y>#G2FVg@auOp|W=^a7_jijN|RFe&Z8skd85zJ2$p;03^eOIxD6&<@vJ#hob
zf_XsJ4{P?LM&9q)ba*Yt$X=t+f2Z|9#Cc>tO~NS6ANi`jf68nUs&f46z2Q)OE<9?s
zbGG`-D)nH`>R~HYFMHXhj>_x#PpxCF_80grzEi^hce8I4hu^-~|ErmD>x%ly$WCIl
zHpB0q<nhtujqHy)HnhtPgQJy?5%m5Skb3!wGHzTsKy~TsZnMb+SJWPbdySP^o9nsm
z^j@VrG$9O6HajO+{<Y8$Sd#$PW`S_g90Q*}HOpq`*qKfyVgn4{*G|uV^aniL?Eidf
zYXA_6oOI`2xY*=5*5GRU+iAz4f0&7hX=l2e3IO1LW^X13!ohd}W{59PU}sqC08p8s
zJbOkVEW9`_j*E~Pg0fF2mXS}Tpc-GcaYpCBOgduTv|99zd-zD+y-h`od|xhrETRxq
zHb$Jolq{X+y~mq>S!iVfYPERM%Ew0tUm4p3IDPXp3<0#0<`!qzauClxxoDfE$M>w;
za&Nkq?`98k=j4GO-Q5v-GbnaP@$2QHWF(pYWap42tc3pfo#}4c^&2N6o>yc$OvY@(
z-~1Wzep}O)qnmV#=6_2cMyKRuSFXJ{q?EtsmQGgKbu+k-g-g3j!$4z(W!3M>^RQ)@
zrw8pk;mxF*g8A|R!dK>T>rU-JC#G!W=i|9TLVM4C!%>rm)7tbyD92TPk5!W_m6IJx
z9^^^=)3-|^Ig%KA02`3xz5U(x;Fahp%8;f+&Q!(?-&`LG>WsvrK=tICYk`ASSLjo!
z28D6|F7|wrhAV+v4WC6$axr;vonmb*5{kfZWYrPLwWf4Z!*2xF(zEiQ#~2#txr=}N
z4Bz`MUS>btD+t^y+9bi&2G1)Sm&5sxh6ZK$GS{QUKb@V58XCQTh~xu`?Tw8MYHI4g
zEf55gnwG}rb<L`xqVgoWLJp3TdJT_+{x*}fZJ-w%g68nL<nO!7z4Pe$){orpd4JzI
zGkmkaOr`Hb<7+zce(-AHc%Os<4v5pg-SQ#b6423$gCJLf?YTA4{5ZH&a@|NW!qN~W
z_>{_(e=m7e0wfs8yPzv}d~Rpc(!<9?7J54NHZ@eao&J8*uG^mzW;2fmdd}R{W#y~o
zM|y2dISIICp2A2qCentby>Y156fZVkQ+~1}d>C2_Kp)9G|0>f3e<TkYu{vd2r&*1B
zC+Mvt6YPfA>pM5ZU)^FQewbKS`hWK&c9({PdU~hAP(+#-;^ugjVOairgmqD2`cnDb
z_&5XM`V>2_`kmnDcVU+Bx=Q~NiCRhc?fZ5+x=)g#i--x?y$=>j`CqsQM}73@I!X<n
zM|l17YI0V)@~%|WbenDQP!elGm}%=Vp<{$_N&dNerpa3Ckh<M<tARlw%HpK?CH(E_
zTb&NRQ8<*G<4tz}ao5)qW^IXxpE0-HM3lpV9c#Z|tE1$-P5^(7cBsoUi*{%IF{HV5
z!!V?pyHHqScuL_xl;D6w|0{A_=LX5GcLFRt-cua~r~yLK<R3f$1#<zB4+X@}pA0bO
z3h2>i{WMub!Ge<rK+R8DdmtA8SRb6DqhqYpYf^qE3Q9^?-)hS#c8_zjPncu+n#J(|
z8#*n8?!ViwSwL@U^KV<SGl)E&$p+yw;ceilYifLECOE_7W+;vaI}ikSSJN@#QN*mw
zRM3_@7^u&ADyZnbg$%mfE3mJADi*TF^zU5ED<CyKu(IRGz+@>fDA4EVC&tC86t^@4
z2vFI(q7yIOs3dzm`w>{oNf`+s+Q`1VSPOUE{u-;tw`w1Jcu^QVQ}}OHROpkIocX1~
zuvMMeiqdI6&UsT4F%nyvkm^tv`%Y#SWcvXym%&&NZn`HE%A#`9kT&d5YyTI$JD}x>
zB|(-f@csxS$baZxEah>~A6YU}w%92@R6u@K7R4U-T_sW*vpna=#Ah30(Z?zBUsD|7
zhqo!z|I{kf1s^{3otEf0x;SiaBMcU0>uWc|eU_#u#)bhSfd6J_PWQTK81L@y&LWY4
zc&vCu4X*<lNRm-tr+-F_q1*0a_V!~wY*Oe@#D(E>3wfdADn}%Z>TTnej2&CBF(@ud
zez;EQPN)f+WQ`STSstZr74x4b2k%0pgNLL=%6JkMCBTdBj|(EEGDg`rz%W0Mg+AO7
z-ti+5^0sQ_k2Bz^pjug3M-tYJ0qW3Tq3E$q0Sihr|5L|p+y=$$@S9U~EV<v+B%qaU
zo-92$X3NZXz6i?5V#GV1Cv2R(IlId^dMUi<Df8YKQ8Vm(iB3yv?gLra-)RPhKDOGj
z)^_b!)P{s6f;+3U(u9PBPrA%P2g}RLzHusJZWajoi;eCi0xqFkmXn-MaE=)TO@v9Y
z22>fey=?~5{P=Lo4^Itv&o`P?Un3)*1AVPJK%a1E<^Hs{`vG-Rd|n<M3^0*Q8vWO7
zQzV7j$bT3C!#Ac+OVV^pR&TT0*ZuFoHs&S=M$$b$4$qVL;ih2VO_d3f8z<r0-Hf-o
zNp78KO(#iTvGAcWFWr1CtB*qzg%mT-eGKY`U{lX46BJdIB`@eOFaYQE1H(0Y_uUJ^
zt&;HuZ$@NdviqC&N&w9TqX@tjf_D|#tPjAv^g{R-!9sdy0S`_+4Lb+ny`JaSuM!^i
zVAZ59c|?EX=(PKF9q;0D7+wXSNMRQA?#!gbY1~$>AA}qcUPDnCwUFcT?O^Oy!bj)G
z$F})CzaPy9?rw{?MWiV%`m|Qc@|sxBw*Ja%fzgHTJ;CX+*bI~P8|z{-F3wP4me9z)
zknQEZ5d?a)=7#SpU&JHfD!<g=Qg10FrnAHu<oTyHP~kvHa-$FG;di#g2@y0uW}sWX
zQ;R+MZhP+?&Su*t9!T%7YIoVmRmdWADmxhMjwfl|-=3>E)yrEzC{$2l(7Lt%+rU|V
zEJ5t&JtbEilmAKx!pacGKiWl+hwCg4K5FXh#mlxf9<qd603!nZ*((^pQxrb@`vQ+F
z!z(8GJeeOLBS8<hY(b4&=|usYsACTKbab7wy_gcV1W@A~^7!p`nVoG}RrODx(g63(
z4<P&NIz!o<xi6)O8TV88R&Zv;jwsCRyQ=jg^%jzv`A=%U;JjC2lGRb$AsH}Sk{8!d
ze1B8V7;A!;Sya?MwD6+#ElbQomEEPI>3Xm{`)X4N5SS~(CdnHc5ONP)^4`{uy!rFk
zoGc>?kqf1h$wcHLe@fQ*Gb1a}+msqoAQn>DA|NU>H_~WX*hThJw<yi(sh@*44ShvR
z@9#HG(swEa8tAUfBsAen*b9^Whfh)TOA<blk=k<hA4HpAbzzLy{cZ!X=UHL_1is@P
z`zb&(zkRU-B|@J()k`~!Tn#?|!F$Day&Q6^MX;>YgdXRwbxd{E_V{Bc!?-HH0hH;*
z^8X69nV$^V@&ZsdhYbzx9X)KKFQrUXS5_A44)yi1_f7*~w4rCf$gTRKyRaBC#CmTA
zr>*6-I@^A|Y<iKkdr5PR<5r9%PD5T%R{rXTrR7?;0v0Go!UjyoD3&A3(A@8-X;1!<
zgM7Mzg6!GqP5e!D&Fu`81*0A#x2n^&thk5~9Tj}NhRp+6g}%*M>fpFe;|tElv^al^
zFG73rbY@Zo5ou{C;b~2L-{u)OEe5T5gc+rYmR)tq6U}7**Sph$32dr;^}|9YkO`-z
z?xtTf6zz8f|4W%_f|9bbvdocapU|)?E1go~3_SX<qauafj>*Bc>t=+^FL4`jcJ@nr
zr=J|?`Wd_W7Lg=a2sF>l{7o6jn=Dx8bXqoLuZ1fd$Bxd@Rf{Pv#$S)h>19(FJ`a;(
zGCr<3%``r8QJ>{bn+@ivbZpw0t|zwdc#DS10La_oll*vSS7uHOC!I>|q(Scgf3vQD
zsVdT}M)d`5ue0$_@)I*YW4~l;i#&DScZLy;0jGxLsv;`rbgviz=wMDM`DoQPr~Rg3
zGPmKx8$}FWGNOLqkY&>GL_bxTrxvM%NpMHt0=Av(Fmo(BZfI&rf*@>Qdwg+QJQq4e
z_x;)UEvsdJd7DLmj)ktcJZye6^>5U|Hkej;cczchR=w`8Iy)VnRTMVZ>nJh4&FSnc
z-CPo39+Y(o=hUoI4wtc+%$C;sANDgp3|5QxNhd)eeB-dt70Hvk4$wPXHL@}?MyKn&
z6H6`UqtX-oQDe@MPg(-#>01sGSoBJC!$U(kFIWMj5pWx;rUmWKlhA!<imMn+#YjEj
zmj@_0uN(i`jAm^}KEjIJXdahUMniN~%VDg~vIV7V-NY<e>R_e}1{ib9DE|>u&K!4r
z{+V!SFyEfAo90&f{))SYX+V*jC4d1#olu?NUWw31Wo4n64{|xYFg?}{lQh|oUQ&uy
zd1k`?)h=LYXw=BsfpDF1SsP_{Ef9d+Uma7z*;XHE{@QEnb#zU^JngRgZZh6PEO9~f
zqwOEchCI9#;$bYc`wN&!N$3Y7a0=#a^R3Bq_BjgK+3vNI>a&Cz!~dHlMa`bmU>#$=
zWc4-ZfP%`6`|&}C1MEpM$($h}IJBdU?v6(6guplP1~8$W-wD)83GnbbpUCi*N5GO4
zCIG(Ede`8;`%j-VA9{)9v;l~MGC@*Zbk?G*jh!EX;3>SAI274%&HT@u-_ayzUF8Zg
z_&dpiJSR!kQ}(ggI)U#&ZJ7EDTHD5Q-;J25L*y>3hNQI^XWs?l)NUZ$C7~E%6zw&f
zGr$~3xD@76YEL`H$FV@)H8cHmnUAgB4&r<j+DFV`P)`c?Fh(L?(iIccWQI583>%JN
z@~rLvGQCqOM7w+v&Dq^0$v~n>6S%)V@1}@SfU68=Jh#aS!)ho9*-`|`aQ^KQF6PYp
zAJD2VwQCJ=T|PgjAwPX<h$UJlpX5d40r)Iney~BAe7KymMJMHtt+Ad17KnDBBd|p|
z1Qx*rAoc!i;$p}PraPmJzPKkHH(-$yJrcndY4<J~{!Ud^RvruY$^P|?t6d2Kx3Ph_
z<@+OopcEg-*H1s2iTq6hoTm@>u`H0D6f*jA47_Nz!cshp-NY@w3$)!$Q(!!;4$e*?
zhP31Q5V!aAh4+WM7dt{q<k9mg&4&V=CL1rCaOb^C`0n&llKmT|W>PxF27w4f-*)H=
zD4FNU{!d+VqL$X*s~qI)Q(Qf5xWErBm{qM;klK%rlDAPhCsu1(mSYd4LT{MvqEVAn
zZm?WTncoqt?r+Kc$!faL)BCSDh}Kbo?S)uYQ)2>ZfYnlS8W>}zW|?2T@*7OysdPQk
z1(dywyY=?p%urH+cwo+1#J>Koj2q5)^1k5Od$Ci#afMB(5kVA3(4Xx*t_dLx6M7m}
z(2JVAVkZ0g1S%{W!8(e!N_2PcG7ajxJa=-YVd0C>$tW$XtOvgLrVMh-4U21eKOfjX
ziJuc)U--T_5~f_R%MVn|GVUP$t{AJ<hSRK6Q&u8*_9FZ^Lsg$<?a1}<2X%u2P%->q
z#<^J|#(##A!^$FgMNx{c4!Phx40TOTd}(|1s2bV?$HMmI`dd21HfWVnE%TNl3(cVI
zwUG%snH%)V0WC)4`l(WbN8(5S_j1PUg#JhvufPZa>iX*1+Dxr2IdIvp-TgZQw4VL$
zIDHTXaga=eM#>XX+J^gEmgsHniFf8(l+w=jcq?=E%vmU}`q-7bkqqs<)NjF&jG>6{
zL?>c6SrZkG51QugnrQjl#lizd$E6fyagHE>0dTXsR>I2{!R8!vaHEU{vl`PGEQpkr
zS>vsJPHt<0$_0e1-nal{pR5l%o0E(7xnxI}Ym0ffIly`=LHqI`ut2G|xs+#O=>aoR
zE1Bc40ZOm~BUkv$0B^pwxXvjq>lNHx^f%BMuHiQaWVznQq+_@{l7Jl*$cQ(}$FHPt
z#aA%edSrg=0!POM@B<Q2Z_9Lf@&Bz3Gor(uVmD-DWMJVI2V$DTKb!wyS3c~7SdAhm
z$nvI_K20z1wf$5GpKmZ5-`8tNL86h`h^fp$X?&5YD6PIbS=$26@7BWgo?Ao`W_&Qw
zMn#wv6XEILx3BP3)HRFOeQnb5;+39z+2k+wdP2h%pC~4B<^&qdTM9+gKbUqgeZqAb
zQc0&XDt13VdaR2ypn{{mV=?7z0?Pd!>Cnyua&{o?z0>ggeQvk6fg!%>BP7^f_9z%M
zUvW~YND7j?Oa`@z5`RaYC-$>)A$k0MYpeFGB8JBI%_mWd9NgH2G&B_=5cu`-ae>pT
zU30sNl*5rP;_;dx<>8Fjbdt=^75hh6lT-Mtz{jnR`GG|Y^p>us%<J@#T0QQ^S-H=M
z7b-;~{Qrj$={I&gIxzzLf#O-%I};kJ(~00``xD`5XKFDXr$3boEAR&!5OBC|74HNP
z7uvvSd&s;W9)<sE0UawyR}b1z0dRRfM=P2VjQyVm`M(}-d4a4$9Iy`5aTBx6Dvy}x
zseC~uls*@j;u@J0{}cqL`!c^*E;c9*j*#&4XYy)-VHpJhx-cgq%*RY0lF_(C0z7^h
z&=W9V;<aY_5zX~6tvjW$Cj#0_fA4eYl*@Yjrr=$#(JcBm{i&_J9@(DAKJoEaMjs`R
zhF*qwM%L=iz4+gVO%(j>_`lPkX;a59sp1ZUeQoUHymQdNH14;`Y>MxO90UNMGZSJB
zHG!<}-p1l#Czqw7?Tm^EP+5xCmLU@J)e~@l$l{&fSmEfef!J+ySeM3N%e(@oL@8M-
zU|u`CDHx~u1(P`rUbxnegrN>VoQf7Ji{^b=i21Ygvk9JriQ?+BNes%dAFkH{PynZO
zT~Ylzc4yw_lQytmHvacoU0`+<xIOxHA#0SwCDEhSLnb43F(lXQ!B#uaKfMByUkB#w
zPsv)L^P?olShRQih$A!%aA5TOH7EMN86dMQ1g7-qR>FWz@xSg<d9AWQDw(yfPimOx
zm*3y7QjXtq5Y^Du_@gDg=}J|9(v`N4%GHL<f^y>30e^yl#c6X3O(j}EHoIAX&b)3W
z@(m1oU5p$w=ArNAZzDm&)6wF{3e<PRP6GcGooDT7WBRbV)tZ9<vY4(i#{r^9YAP!D
z;+hgdQfdNbZL>Ve|G@v4p9RWF(VxCnh<yP>AL#gd24pfQiwcw@y(!b$3*GCk>(@qR
z3C?H~ApdcFO{Tclc;Kv|<lyLN8<X6ES#KJe-<@?=F$+D~*CY?qD4m=iU5zh)xuHqE
zsY>$<|4G~j9)M$@$e0G+-JPWt5kf*jjFG0N=jW>#P^Ck3e<uWx=1Piq{)?Zo(%A=B
z`O8<fCbJQ$KY0gqeT*}Ahaj1fPa1M>JE-P9m?uFYBx(EmDCjeV;8}@N;_$gs7_9Hi
zZHq;w(t1P>aPa}X@XG$A1tMp0J#JLZ<*t-U-XBmv#aL%7cP^K(8F8Bm7U!Tfr>yL|
z9;(O|E{KTw-#(+5r;>RHY0As{3M{IMsxF2OVc^gjt9LX#Ni6&i|63Zhr~)Vy@h!g1
z&w?tYo6JH)H!c4-(0<X&(^COj?(-z+FVIE{9=E;o_lxsvzKMbV{_r7737`&f{w)~O
z{h8{B=5etbc#fa{2?J5|A0UR<`8V(!=djpZ1wXEGqhsTb74NGFf5hWTBv<LbFnlWO
zuT34cb7jJzy9)mj`Ey!}NvcGxB`7Hw+-^QVIlLnfBUkGX;MYO7u+9?cI@;fRTJt%Q
z?-lY8OsR%;PQ+mD1injR{~rbC`Mnpu5x5MCFb`|2W}^L(u~s*7ZjFeZZj@cAqAH!b
z$R3os<bpk51SDQ$r+#m>@+Wtka~L3Abr>kyen+1~B<$g9Ia==MPfy|n?(C7{xmXs!
z_v_v?SK@@8eZdT`l~c$WZ_WR#?t^zxk6i;*e5)EuczJvAoo3GTQ?K-p(af1%48%uF
zGHB3<h3<sR)2~HrqTDuV#c9a_0B1QRd7bt0_hV&Ob{45$c%oJBJ0g>M|3ZiT!KNwx
zO<%|_9{S*Z$SQYC`WHtG$b3l2dnWg02@T+&_Ie%s{CcPfefH*uUnC-H&5`M!Rv&nT
z4GfHvmZ+Ei^l%5()>n0mPpfYVkmq}*dT58-m43DYt$R$F8zO!U6o_j<9<9SKSo^+z
z`^g?JbgNgf`pJ2A68e_V664pPYI5wtm^g>e&%Zzf6vC|j0Ni)?Oq{_UFiu@^c*4sp
zlGC`O0N~`~xr_E63*ZytTP|FS><ZghtY0B=M}i5G;<~j>b~yAhi&)#T?p<F9@F2H$
zpU2za-^H7#)0%!;v4aiQWey16x0!_{`q4e+AK5zy>A&Y+r$0Y>TO3GU0`vE_-hhI`
zJI(UH5|dnV7mbZ?yD(ITwTXSQ%IVc5pM9}zx+8LivG{L;M^5*n#Xcpf-w0^Qz;sij
zd-ylgKfi$h)Zl=IK7g&*`|MROA5LW|>DLNxvb*IY%AAU8<m-ceTv2tLbeg~12@3@h
zkn;~j->eS&SbI}T1s~+merw$-)0IDg{B_AgDPh6FO_K@mI3MjlG*_`4RgW+cPb^>y
zTOwy-8l}r`9JwZN->Ii0g1(VjmcN!6$7O4`)<G4|i%!-}z;1J!Msc&9>F;;u2nn%z
zqD9#N1|b^lF*uQ5Z>>;(yfEbDTrSks_21nT11g4`<@=?CQHN37bgg4NODvLg?Neii
z@89ZTqk6wMmU}f?AujxC8Y<&?6d%Z0?;@+9csZ~RtNUaoFs|3O**$#P3o9jxx(a<j
z^_pFQ4&H8y&-Oc!?L(QB3Bhot84XW?u%LsClK6N%3)lma^Eo*J$dvw81r54iI=U&I
zdyDPb9qGlZTR<wv_>n*8cLk^aqa}tM{?q(5P!}6SY=*PGz0sb?k{-jZB2xd%k|X=A
zsrOJqyJa~Tvd#>n0)*zX+PVu0+Z?-AiOHVYOb&bp=bm*HGyS&L>M38UW0=;Oy-_&o
zdqr-Z4oJVh5JJx8`lxM})#9TyeVhs^O<;}j0u@#?_ko6#X_L&udHy^3q}Jf5CbO0G
zp75C5?UO#Ic&pMijw_)&KvMvrTXbC9#I*Q}P@_}%`Uv40vkM1F$&mlI47#4=g(wq2
z?%VMoKaG7k5Qa6ze!IJvheq}`*bv$W_@62wla)bTu7SVzA2h*YfH9b3YcwK5)32BW
z*I43G|0cvB3Q|s33SF<^<Am=Yf=1(I)}}0QULbuhX|>z}%zNI$i=7F$gua!t@5df*
z))_&Y6Ju~BIGHU)n5d<3JbhKVOfZ6t-M_rUQAUkXVM*7f_$#QG?|6W4IvEOk0_l7_
zFaTD$MEu{jo>!?rMcv`?#?Osr9vz!X;TV=#?|$L`l58B#Rf~AQh~0{50Oi8YQ1xo)
zkDx#l7L{kFQla}zCCl|g>0nqBnro<3j#$1j_a!R=4(mT^OO~c{><q6{6c}KnHU=3E
zzZL_Ys>eJYwo7U}gipr)ldLHN;~lS(eZ|*U97!&@Iv0~cO7j(1`FnO9TN48s#GeV-
z_Fg+=VnHtKUIQN(`WY!N*5Bt(+Q@?nO|%OuqD^T&tT8Ty2L=Ra-+Z+F!_^<IGD5%|
z2JBxK9%s0f#YywPr851{MpR>@l;{hpcBwLMJy&${{s}=bsE4|yTbr#hEyJI;OfC6<
zwvw|Yx9!MWDc+TXF<($JT?|!S56uYq3e9*uIL7XrdzByism)p`dHUGr_YofldySp;
zD16-Q=u|HRb!_MpI0`*Y+XWX0D@|}pnY7>C`fi@1SVO@dn4WcUxFeMUgWld$De^K1
z1EYm#X}iZ$p4#^S8hMp`>id0TyK3LJ@D;AMlY@l5K#Tyu(3DhqCMRcZ8-c0G;csKs
zO0_}J9<NZUH1nMT*z7+L^9ArJsc1^Nw5!B~9oeiBj6@S~H+DSvhYFyVn-&~j8NZ1C
zv-|zh0oN@UiB<oGPIrOp$#2HlyL<>aZ>r*>dwUA9*xve*eL@1hbJ%>m=Q_`V(J0E%
zEd_}A1?T-~p^VOe$0i!ThNlH$N3K!fO&h#6gflYWpCN&n&r|1FN%;lVi;oUFPIIsj
zjUaPmS7}Y^c`{N=ZiSJN(cfY%9^sp9szs1$lgwi~1je(FZ{ON`dr>rmphxrL%S-Ny
z$p0>BIAbq;EPi8A7@88Cn}(#f3Et#&1>p0ovy$TM=T2&n-YP0~7#_JQNM2UopL07v
z%U$LD%zsi~)@~cnde=>W@Hg!*azmDC9$hbU_M1`!Mo&vHE|fczneS8X4g$p-rEQpD
zPl_RZKT8g0?(=FfD6A<L2W0d0yermuD*AOtAfZqAS|?fms<7?8oNY>E{X*s)T}yh{
z+I5c6B)Q<9MBye`p#hKl2@BOy-9eCm!soL863~$dh=^=f)2zTFDfWcV={)6-dEFk=
zTFq1h0`CRTw!ai`dGi#kQBuNV<3SG0?0}$#)DldOFD^O<_Pe%nLBz&MD?FAmR61et
zZ?CjPy~^wbkco&DtBz$!V5NH=@&VnnWB>22{qIWoMnJt<q}!AXtlN$kyDGrV4le3y
z^^`7^py2!uLk|&FcV8X~iskql(zv*|#?wAV{?bxB*N;z2&C|DHb(UITlI9u2sUew!
z+}Ec`6F+LvHa~EDias$H)m5Nq8M@c~apKlI+K~1_(OdY?5$Jn>l*9hg(RM#GF-RsT
zEd}1i#zXE6-BTyfF0|z)_9~yR{`~YTvmwIZ^{7#6=wX_>*Zoq_qU&-vvK!$0qN1XJ
z8T>dtBLf&q0HDkcJ$ON<r>6kf@xzr);j+X>!+%q^u?4bW0JnPzOoxug%c6k3SGu1a
zEElNO4AU%b&S}}hMMV@3#xw?>zxKSzdX5{wK_3#Bb*39!S!!x(g{#eang>$(Vt`ZO
zCjOG1p8kz)V<LdjIDji^b0Eoa&K`&8$GiVcf%?$V3N_Pu5=B`e%Z0(wq4HY4$)<yX
z#Sa<hJuqql;&q9-LjZ+U2>*cvwz&zt8%^{i;;qe>bIRfQKC#jj^9drQpE1+_W;B6H
zDO`{2F*aw|2bdgvHui(4086?vk<pZrRTXLKheCduBfLLP8g?Z8wA%pzRZ)``3BLc4
zjF7}kS;Sw9vs;O|^zcY-w%J<%8H+4ir@;k~;F$_^f6X|RRaLPG2)YN7I4r%`AHd2m
zY4by@vYZMBAxEDS6~BDQ$^`v2xz9ChpqQwU5jD_d!#Ewx;uWiEksiy+%S#kas6t^R
zd=<ZsfCnQ<$n$b}W1};QN<sl*hy%JL5NRQ+uTKG78^8?KMtzG7@>{;)G4>1$yvtWs
zJ?gl**ekEBe9GUTW@8(P4Zce^I7MHCK5YJX<og(bpf8Z1vXO7BPypvi^{}3vApEty
z)=KrPMtdLZ#OPUP=XkgVOAdWw{@8p=ZJ<?IHo}a!ZTBC9vLz>1rC`dY*j^S)n1$Tm
z@;0=n%Zo}4a$U(6^ZU+=$!&Gtjseunsjtcj>$hsjj#r4>TAsJ>ywiSY;`aYAb5S!8
z16KiSc64Z>EOHr-&*|z!k^dty`Wq^V)uA?DA<(D8mzj<H)aoNp*U%WLvDSJDGMJmB
zXJUekZ_an>>g!*Sl7<b+t~on7t=Dg*ix;bMfULK8W}Q9&xat9E`6z02oO(}R-)e4#
zPXSp<Muy&n<;uzmFto4j?DPVd>7$g?D_-8|JZ}(cabqpw2>#GYFuBQ77RBCF1h_Ss
zUCu|Y0QUvfmWd}7>61T8f(D|hgOI^|6XcJ`I^I#{6H8<+_J-mKN|x2UwCPTA4mQvu
zyH=Oh_)U+e)EgvR@^=go&Z<op^EJaq=7cV%A{cy~FQoOyy2R$*6h<GrJ%V%S3XK*9
zGJx>=!;PpX^7bC8^?q9?9K(2|E}{gQ!qMQ`baYzkoKvVhn%+-$C}ElZ_7W4Lb!Z`~
zVj8<g#kV1c40)d>BqvfQ(ABIpEY#v@v4Z0+Jy!44!nA93IrncCHDXIT-^SHd0ixw5
zVHOn`8FYa0Pnf5HT9h?n=C&5bV}o+IxF01!W&jLePpOx+2yT0oj$3jtWS;xwT7C^v
z^}m0g?iQ9pr@&%(e6<9ffqxKHvpDYnLlBHsjTI2q6Me<Tcsg+YzJ1ko9*zI-5kBq<
z*XCm%a7MR9WmY56-!`36^ca;15&TMv^J_>;UCw27%jH79^qq83I<5^9LO%<@nr(G8
z_&Wp}-KTGXPQvikgeGgsv16Ac8jv1(Gy@5cUxV-@Yblf02w9h#^cP)nOMo}QHCOsA
zPldDo@jt!8j+eO!&GrVKP(1(=c`3{1*Bl?22a)pl>g6~W4*wsRWH;>KT!u{1@^f^%
zx#VhP)6KNsD-4(q(9qC8=wKI!jnHWF-~y1o8E5ryl)ca=m!VWX&^%5;*Um!IL<9u6
zQ*MBAhIuLM<yLHsgh7PR3YdV<#e_Y>R8SYeN}C4R3rGHGX{0b9`@rL%<6aEG?P>RA
zeyQ4i;b;3!8W<AEOSYHVfa-(~(f12S_xS7pIPiR#+kT8ly(T2HLLv|J)f1XQVzM^y
z(R#U!*{O8D$PM`meBS%4cgwHIL~QaXW1{^Hci^5Mmwz&bbkBkk(L>MT%ZD73LCd9>
zgj9_!24Er!0?Zr0?z0e`uk>PMHJ<By1@b@u@Qgv;|F=>~u$N!QaR?;Dm@13BMnzQn
z_RYS$wk2z(fx`F?4J|E)`6x9g*<HX6WWESBx;x*`?C<Z-P4*7N3af!!mRKp{Fuf^S
zg$DOmTsrF;lHH>VO_wMhrSj47DJhoc&5c)!o>jI>=^+0uH8JsJUteG7V(@JzDax^`
zvT|f{avw<ZsdL&HO`HBtVZ<q1jIR<XmK^TtO>-AdGZ8q^x#8d@K*o?GFdx?mZ8&pZ
znQbsKIk|o@|3cB?eHuulkto?V#Y!*GR#|PEHaKi6()rJtHZuvYaSn;(JnxdT53TD)
zK}EB7^D2H~X1r@AJ?cn4I(%xLnpZLV*2QI`y}duNchK#6EH_Fkx$#xpQ@0CN@gMXZ
z>Nvu=SR?Iub1er{sjwhQ9eyb3v%1IG#f5Qi^lN2hWjlab<1=f21TML!mRVR>xVCcw
z_*r1o>m4?jaB?Qjoj1PuNYYOE*G|DVanZkdgU~v7a5>w3vo8%Yx{jJ}nSouRNUJs+
zT+-c;$tq1I)Ib!Fc@xZ~?yFWfQQ>i6Ato+fUR|97!o)z!0s$$ilXL?}Ao<4UxXB7!
zT*?e8OifQHR{H|2&#id{_Gs%IQPDs)%#pxJq^9DnloUo$f56md$_CU~shO$<f}%p_
zxacA=x#C$CYeNrd`X5DDj)VhdC=P@YK8te3icD=r#qkShlhXZTa5<8*=UkoSjm33N
ztl^B^yXg>Y?R`(lIdpG8r`Dl$qQvcJPMU#t8h5<$`kESr^GSt<;AC0+`11dNmIk@0
zNN&!n^X20Sbj+L_;_=`Qg&SvCK<y-Pd%?DGd~O0Pfxu((G;{zP*(x|IKNPE;fl|CU
zS^`odVXS7Wde;rN|B5EV8v(Z?R9!PuZAkz_DljP=5zKsBVLD6(^Axc5A&e+6I2d_z
zbF;^26tV&7s9_6fo|^s$ZorBZ5*lg*CK=GBOwQCaHl~7W$!5DK2#(Q(=fIx^I;;=C
zFsx|4lC*iOk{5g7QT04s2;On1V;3DTh9iMxLdeZhPf61aP(MhF!0}r7u5esV1Ew0(
z(n+taySqu8yf}-OhxBpctN{^R=yu}3?i?t@#>L$$m?3r8iT<nxAS6N7-P%ZNId4`f
zry}G@Z<++76*Gjx=EX4k#|&2K>3wfqxk(rUfT)=bZBaTPSFUYoMxAe(b){P^#I4F+
zB=!kLgusvM&|{_R(u!i-419MB2gpBdiVAp65fBiKAfBfbi}^y4o_i7xPx)QhAr`*V
zMD^k4w?-<@J5fK2B8a|y&+m6mMiGWB_K`As%DhFy20LVj0|#!!E1edLhvP`-YCegw
zH`V>feS!h6ROVU$X855d46doUjZIkCDbzn7+IfdmYwFa}Tq8Ha0AMqu*rXCQBDT*6
zFw*gFDB0M`McIS|V_SC4&dv&(Ai!(km5+^%7R+aW>^U$;z`o?NP;QFQTBwiR(zn&S
zzfE#obkAKpv3(b&v`i}EBPigq4--oZ!lSJ7CML)W?mFQh)jg0N_;oLB#)!s10ty|4
z%S;F!J-Uo3Va+cdom|!rA5FNb<~4TNO?{`HfTZDJd>72qMr9qlH0~YOUxHmjT6n3B
zGS`RC)X>@gJG=ifVo5V%jngUqIPS*X4;9;jcS!Gj=t2J{;>GN}I-fjgMNQTL5e$|z
zKef**_Z($s>@Ex#JR*XzxjfD9d!SZkRcgBb(kDwKaDp&a;+a=S`s@DusV1i59&gdx
zyZKWov(yL=fRn2X5A5o`WDz%}vq>%6a**yC9TUS?;s6o?LCzl=$ZLTCJ}O;vax5$?
zkOPD9cB(|@$wBAt?v8gwKtSMlx-JQx9|Wp7h67?kLX=j}3qm2y$$oWUBE=HErhsYk
zyxIimI(SStpA8Ln#&WU1cYOoh6TeH5*M_$Ea1KB`V4i{@K-;PIrYjz}^olep6FE%f
zn#j9}qXtt(NqOyjCkoZzS@m0CxNYXY0vTJ@h;oC{VjPQau~sd+_l=WQE;abL+RlLq
zXLL+-kSc1!?+PqF(s9;iy!kyWHg`KgEOzNQ;fD0fzg>z+gnR!I<ckI97`sef8!LMj
z^-J!=b!oiG4qr=^F6A8hOm?}v_Esn5P1XMkCa|%<?@|(P@>u_k?8b3(5Hf#x>W1z5
zC(?x>sqvgd*#fUb_^|4mnZ&J+!Sv3XRH*IE|1Q2KQFZbLOUv69(}%X?>oN34lh+yH
zL?0wrhKa7|*@%*A0bO}C9Y&~rK0kJjf!5?9sYeOoloND4^1E(9Ja9pA%Kcf!BcKEP
z!4eDx;R!8W_Yj!>q?SbPjypi3c-BR1dklVljF>4DP5^mF!RHT$5EvjsA|gz|FasL)
zYN-K`H>FynZg5fm6k!Gnk5mRaD~mayr&Jygg*Nf>=!5AQVPl5(rAFsvqx+dfYQ5G_
z=f-f{?+9`cn>{D+VZPERMrvH<*Gq6AN`y&ijPHMeapQkxg3w`TP1%=gVSn}Q<-Y_6
zOW}A`jBp}oFPd3EPSoV0o4!%IpO5QWT}rWb*+G{Sp3%|Hp4!St9!W$bsNvz3)wTKJ
zxqmvbY&WG&PTkXYyFH^kzBtyuZE)O~DTlvh{xy0qW-Oy!`-STyw?0wNx6SBPZg#A=
zIpM&AA-N7d%SJV6!^rGhrO~Dn88P96Q9X2b(p?1UG{l2W2e?H8K4i6Zbohg?+j65`
zY|sqZxs>qFh=Ffc1#{c#>S`jJArjAm6An=rlA4;D$K`%67>8!S%m=b3^Q-iL*R;Xq
zV6~*-uzhR{Cu`)ZZfUpGgQKw-h?)Z{9-N=Affu!PW`+<v0kjnAHCBWY;UrML@?7P)
zHu!!^z?XgoQ&m84aF*(m8hjaaum2xYUjbER`h86dpi(L+DFV`+(t@CLN~d&p#|u&_
zr8H8~(jsx`4(aah?(Y9{o%wz1o3&=m8pM0w_sMhi*=O%Xuo>X;1>9bliR6j>3rw~}
z7*co*NqIrJyDlzJ_SbKUx1khp5k`y7I=6%irnjBE-cMBMy*_FV8QA?}bBAdOEi7#P
zw89;^|NV<7rfs0=)q77^-{`l<+84%t+@1`OPz@7*xWDCT$6$4Isxl^evT<~C;D<kB
z{uxBI<&Re~amy@uV~cL=t5KgxooEXrycr9WCX~XgymVifFTw0%nz8aADeaVz5Peid
zP)gg!(AqhS5;4YfV6lxXYiIrgK25EsMKLe}t$FvvO9nL#)5~)&9Gq?*wi2%JhC0ZV
z<?eceG605N>mjKsmNU#|G4d1jw6snM01rR~<Rwsro4Y>QorWz*-Fyei-*ABt5V>uW
zjz>CzoX`-&_cF<R<x|(ii~=EDR=NuyxwLAl@b?$<KVzb!E3xMWPw+-CUdHPCIVD=d
ztem3U>(T4`I*hm2rXB*U&?k!iITUB@WTb_q<shky_9P{YmmEI}X-9Nuks1q)Pvc^%
z%1@|hR(~sLTdR57SNL?$6&>>;0A)fRSL2%2e=L;@mZ^D_)tzYw__Kl!ybGTX9QJ#&
zOWE5Era6ykq}WnQ3VD4O7P^cH*k#)Y$-A?D`uNG4mAczamwu+7yAWe4SF$5TRB{Nk
zI1=Q)VycqNwL3W$%<OuSy`TAHW2Vqe_pjKjjR;?1tj6-baTh79)TN(nY=83YO(5>j
z@85RLPR;YvA&5_O2#1SVp6vR6=hLAm_;8F=4B`h6@*@z2wSRe5%~fte%fwVTfrVi;
zB`*c`&x~8-SsQeGy1Y+!>+o6wVfQ7^!65y-X>{|T_Ltx27v76}XH1i=^TOV-W3yrw
zVfiJn4D0j9eY+w4%9oGVEZieKLC!m#qg?I^+a69hM#G7*AH#J!H$r?o0&^^PMLXT?
zDQatIoMWsU#y&G?seMzaUGZFufz-)GQIi*!IOe0h3>kD2i%86|fF&IN^g*y=+Q^%M
z#_UaN)hMt&eLTz`(v0n^+1M$F$zz*s{wQnw!x$JZz)T|85}09nu^fdNns7~9NIA{3
z!D|wgu!}x+bRE3Sx)s*NILoS*g;pAZ>VWxGU1V$%!-Y}1qrRA%W2x=>b8QVonN$v6
zm~W2Ppv9#gf3e_$g#ij@Z2ToY<6|V-`Fd}Satpj4W!eYl=f%!9F&sYE=dfGixPk&k
zGVc?`9up2ABLK)zBWoStW9+^84t8C}?ynhoUY05mCQnFctn3ssp@f=bU0hzeOZDL$
zUe2di9Tx_FuFW}{*GkT)W<Z&X$&Dp&GdzYUUSh*yQ_8l6imsMJH~PW@xAm=wLM_cT
z5`#o3Zz2)t@sfkYRrxb#dy~QDRjlPf6O#S7N~e!%bNCTWi(!}Wj!&_w@xk`+LL%-3
z9p+Q?Cb?%L?rx)PINzoE0%c->TP8ELu8O9#j2Whw7OnUkLxLZU`@yO^Gr8P)HyV`w
zj8yP5SZzh)_88P2K5l!4A9%=^wVYT)-23vjEx~q}&sMmxM71>We(kY?B2jBsTxJa)
zJg|aPJ{*gW0f9!fk~(lV@O40-EeUTLpdB!CW|PtcKy1c9MFu*i8H1?akHISDL^pNq
z_<a}idudO{=hwl@tQZl<`St@B{l+>9xkMwaaL-E);oEKLNGz%mkI=pI+4K|3*_7NK
zO>0{JXZR^73uERP3ZnySO~SQI6v#Cqnh(9=h7Ohy7$t*91bJpBiTf*rMzIFXb~n!5
zaN~Qs>~}^FKA^c}QI-{FSE4Tz_<3qIbvzkp(x3|`>uoCy!v-dF1$v5QB{BH8$Gt1(
zF-tbsuIeO+uxnMOy^jvzi$w=6gfb6X%~%hYPvl?K<gBIG;Pse{)t9Nz?Ntmlq46|j
za&xh0j!U84n_-vo`-hO$2;YZ+y<EPf{dd;}JnoI?MXv?={^B{xSDS)Bc4lQ|g%&L@
zTf7C_3h_LUEg3>eN(x|=jbvAWklr6;lmKKk9IgaITx9#w|FiAOh8)gK^5Lf`kMiQ*
zYclZT3E=$WN-sWFOw)@NmD|2}7LJ9H>K<{_n`r-4Tjidrx6#vd!AOXyWol?QMazOz
z$ZM@`Vspp9oc+e9>aGLvf{RpDx3y{XThskdZ)h=~ucd|#JtxmMj=%Z&{uOCPApTm=
z`;$e?4I-oW_nKN)r-l3PB5DwHW*0EC;x%dTSo_FRFq@4yOO_HMq67CV?2SgrNp26Z
zje`#OAt8L{tu-v$;E$@k*;NeS=&Sa*{wu!xfWqbhH_HJZKmYMAY#vF<GXSvM>u;bV
z=@Sp{ws?Rx9s;ogHH*#z@W<S4iWO5U>+2i<p#|7h`@v$1H9qi{ZauYaf#Nb)=mQ^{
zx~2Qw&cAPr+Wk?7lNCs9_318cZ1IWV)zEbsg_Pc?=?D6&nKAkfg!}M7y&?M}ht)gm
zoKX#o)6Ib|gQYa|fr~s#G0!TXc0SbvuK2}}<AXkP5S_X9r-r^2*+rS2C%7%ZZQr*<
z;qe^tW>#|;Y0OV<loQ{{Zo$?5t*=@dX55-%cM+T{rFfnde#&s7x9OrsIP|f(+kNLM
zb8WIb=&7LA+^yo}THE8SKhuv%lVa}LOgIwLAq5jfbol0ZK=DtQ!UBXsDCprNz8xqX
z$9(vQGD3y^;0y#_AR7L$bPWfMfe2^f8Y>uwo!#980r%hR!|%Y}qjkIpo{tH8JFPk&
zna1owA3uWcD>iRk?C3TUaz1=M_~fgtZ=9jQnB;`}voA$vq(-Q61oSB5!PrfEjI~JK
z?kn5RDq%-?YMRhuh6meOa>u}gWi(k;R7CUDi7>%y2G?BNvCO8-mV;B2GC8anm_9uP
za`gyN11m$Sp`{B$+rS~>X;<zS1k#zkkoup|Vhlbo#vsgHooWs9VYYg_O07!HrdIAB
ziY5E`Oz{|_Fn_-C#pl4f!>%;v_oq~9i7Qb{()F{pZypf+<Zr@%Pm@n39cv*4F+fw0
z9SB3M09<YbsBPxC^i<j}gBTxccbdoQbh8>j_+LSu0O|?FDrF(?aYSHWVph;pNH4?K
z*t2o}p-Z-4E-$IdmS8$`fOV8_|5GZ%<f6dSe3mIb%(nRXwmvFJkD>{(=M+PoE(^4s
z)~do*e|_PHksyTs)^!~<%XwIbXBlUypT*Wodg#tAUeD!KLw@p6jHSY(UgM9pm^52@
z=ff{5&%_lGpH`45e@TP6(bYMFmLGJ>p-<CJ0b1EJHtr;r3B9e>Lg@n7E&2QW+AKa;
z1EXO}{;#NZBXV?7``!n&+VG`Yz0rW|Nx~ilT{Xt7f1Y-Vk4>{?jG+FzI0DSLi3%Ih
z2Y?|890mrkFA&Xo1UNp8W}Hjw>u*6uY422Ex?hD#3nZ<;p2-H~P2f8-oK6Vtd=g%}
zA5+&o#&K_mFi^o?Wb7-=eHI#2jW+wdD;1dW1|Yg?+ZbgtctE+|WnT9Dp&sZ}>?t-D
zp0BA=YoJy4&h>rgI^pmcr(rK=e_7FaZ^%IXpIx*d`>k=YR%^cI%iD*zMr$9q69)NS
zEzD+7Q~2Awxare%L%BkVy{)ww%*DfI_D(7FT;e_5e%-etASu^nEaTp0R_0O(JEpuC
zV-<aS_QUyn!Gj#4xqPQlGV#5cm$r~id*|IN7yLS1z2znQFeYKj;9B4=<;UAkgwK~>
zytik^8oZKWYBtM|*)>bo(ML}_Ti~@*UDXt4x45BksWbOa7bH)Jfm$WHBjGBZh<n*U
z`(vjrpcemodG*{qF-DLDtQbWiqh!gu_;I?s#R!t^t3p*s8bs}8AX6<r=jZ2FE3*SJ
zQu&s82tHFfP?&>mmdWuuc%8R!@9)kVEJ>BMwPkd5Q`$PrEV8J_1p9iwy@srBpW@4I
zcU7>byLSpCW?Nlw{z9{G?)qSqB5OO)KZ?~-Gi2bVh9Dz>BT<Lwsk7>3b?8|A9`;<*
z8C|KK4N4p&i21PpGmzntGUh8e8%>RSeM@rwn5QjH*<X`R7?T8a3PMcN3eXd}f-v>a
zU**Q-3uT*>D!p?VQs+5-)~KdY{G3Hl%O{S9!aMgS{4o@HAA;0q#`>WkO*X=ejOggD
z;pb08^le=fkGlFHBYY6^IKF>W>pZx9%1xJ@#0NQl0nA+v>nYBe*}^0WO5on8!A{(L
zS6SXxF!-AgkB#YJnqtP-#E(XGf|y61G~4(bg>HuWEge|O!@V!rc$zQcE!8LbthlF7
zcwaK;I{vqn*H*vN9!!1IEnY8#&Anh{9oRyd1{KJ7K4**a^Mqa#p_>CSfDOvYL2|03
zIdqfLTJCff93R?4U)zdg4E|8%>8kc$;xpiU#_;7O5kIEIsPlyfYa+N#f-Yp=zMCSK
z*VW{{YJQreRk=9Y7q{T2v%y(FET=2WVR)puFsf?qEfR!jqHRpiCV=d{eeXDGwRYq8
zW{H~={-2|vQa0R}rJdz)*RSLdh$9B<-HiM__jNIC>QlPfse#^Z=~J;R#pShSKdE8P
z`*DGrx)@C2di<Fifjn=f%8SY<M+)W5(jEP_%UHp)qWV$tx17viPYs&2_RV3%ycg7M
zJ?+yUP;)Z>dLJTMJPxcm27eaTN{ML?Z+YB9tF2ssBFp*Ra{W{v=nv8CI|rJreV6q+
z2UsD1r-MP*l<%32f{My+K9p&(ABALltgN_DKCGW(&V5fEfYRgPDDF2dXXq+>tEfs2
zHp0>L&i*SHPtg(SnZ=j>%x~^5!!DRQCQi<<PZD^Ow6a56=LV~vC68x>``5rD<VE!&
zC9|><wV-Fbkr28_4u#J4spIKQ(;J_}TfxrYr}wprC=<DKLdjBVn<`+?43tl>Pdsn9
zuY~xJ_=7Fi*UKR{E#qY~=9w*H_aB)r{tPY?OraJQZR+}`+|BX2S1Y(xBOGCLy>N=9
z_hc%2WUIDonPzN8oYm;bsiyIxzias-xIAWlRkl%|6Vm4-DuKb`koGbA^f^^d7Gv&1
zn)L@j73UW$YFyM&b^Q9ixw-iZQZcZfX<D4R<e%>Alk@06AMD{uv={?k9|(EL?pi*Q
z<~}|`Ak0Epn01{mETQ=YARb#RpPJ;;VrJz>BZba(!Q>}rzOj4*KYt87D7pM4h;thR
z(gH~P?aUfwsNM<g9^#yA^4<H<YAw$lk{z~8+?HIg2jzC$jxuDKar*l|+k^|OcMW?@
zT%eL~^CnQlq*0K_Tf)^G+^ZXs==GnC2sFnAS}&=ISxQ!C$-!YBa4nSdCBGOC<u&mZ
zZzaR_#UmG^W<zVH)X&wMkxy;ksS%ru+tEW|-?98Dd4D?)SoH?*-!h^{7X~8(hg-gZ
zBKaJPU?-BHn2E697)MxHqa|uZw6M#Q*tWmXccq+}M5x6UZa}3CXCgStGD^<c&zxt0
zgZp@5oiz7hH}O2Zcr%?JYjxIdZ)#IaCs(!lJA8z}&zYpsYo*tw#}|2Vc866^$yKZ?
z*<jeR@NQs#<a54)#ZtFk+G`{OO3$%ZH|OHQrS5a|zm!{scl9%EM|}#IU8EL^<IqB}
zePKA2b|)Bmp%H5(F)T|PgD2Kp+P4B5oMj9piQ$b~DwW0WTl}@hTJl1E?sC4F^9<RI
zvid;=5&y(>Yp6$P@fI<CJ?JqBq0n+YxzukUGSv<a+i7awz{9n!$`r8KWvYB+=J}Ee
zcYwVN6H#ZH#|wf7O`o>^G~fuCZ{ZT{lz;8`&WB>}Z^13A4iu#OzjImU-D>{QM?~oB
z>%)U}?&wgwakO(er}OdN=FhUbNbO_oYXIm|@eRqst^(nPEI2~Cb0Fs0MSQF3ZDC*+
zIZhi_19_$`A_8oD*&SR~p&o{!g<{JvviDzVuq@(<^Ll0qElUc=_<+s*m0P5<kUw6(
zS7y&m9V&*0OPSK+`=wg;<w^<+6E_YlDH$gs{LdX+te&;bMxJ~T8`&oV-ZYAIb3Cnm
z-M{$o3Zsybvu72dWmYtTw>HUqQlIZl1XhTNT#}+cT+Ba12<odOj$y*L!uj_^<wWnz
zZq!OV(&ZR(lz1?h6d(4|$LE$f>)_mqe`Cy7qY}h`WEd2EROCXvi+KFRO#PPMlI6NL
zHYBCeK<(hbTI%+Zr%7}|igeU#6vKIU41Vigf)>J42-U}Dmv}XCv+41g<~>!pg&V~`
zNTjchJLu#J=JoEqi3czJjl~uGpXu_wAH=DULy@>}WhCiGetzt<#id+Ec6#;CjFx*-
zg-m?ngRIk=Mnlh+J#q;8PA630Y#!yiSqlz&%KihLil7NSXRASdc2hLo{pk@V{9szp
zYT1TCe~t41fE8pXh6{`_(_ir&b|bPcDMzO-K9~<evr}kM-bI+WYm$$*{iRW>v$xU9
zBp<Hyr8X7#^U+YG{a7PidZfSh(0DEH5vq_gjREZR=PBV^wcFP&N7Pwd?IP$V`Fs`&
zoh}Gg-nPzdABAMwK<47zq`8_%!4w_X37Mn>1e#Jv>$S+4On5!aCJFSO?Alp5lgsST
zSr};-Q)4=x;`OXKf<>e%YQ_MtLY_3ZZjy`Oc34VZE}b0**4p}sV2_eB&ojA0(R?J-
z^W0wZZH1c02Vr*%A1UQ8tD|(B&5=fKXS|s)k6k?FgL-Q_j&S~Z+=F~rx)&-!D+W*u
zt=x?^cgC^MebD(=CW5s`I@+CrDQy7dFq>tXFuByh4iXSXO+91M5Zp$WL37mKVlzf8
zd*^DF_WqGqoYaCHGZh}<kCJg#lfI0Ah^_agH#iL>3fzpF@3o3*_yc&O623}!Xg{h?
zMq-RG&H_lypZ-67MK-JqZ=g%pMU4wKJP_^677cWw2<*lzt+aiZh(ms5+%{jI!b(Yq
zwbqz;m~?ZAkVd)19IA^6KE=&zRJy?JjIxjXoq`DsY(Nx_&mf#xRBGoj77-+3o7g$I
zt`$v244A7M)kS1DIxeLZXCEJL&)b<@V!7R8czw?#$qB>K-hB>DBjlhjC@^ox{^YVi
z72x@<l}bWg`0U>h`0;!7_Y1e{!uRG;;@u}#L?*Wr?|`$7MyZKx2ALShkLgPjhWX`b
z#dFBW#@e>FQ96!!>yBG7)fP}H&7H{kFvE;a(e)m~HXpP*irW?a?BWfpBTWu^>o5+H
zOKfN6a};$#W=lJJ>UKjSZ+7GM#&e5lsBrB()<3^wI!L*@ZB4!O{jEOJn<7QN+#46|
zf(cnq{!pb0TtGv*v!Wd8M&41Ox+8NujVwk}6UKh&b1W}s!`lilFT2G*&n~Fzws`fh
zU+iznH*}=i**RMs*n^7&jstD5`5;USt4lA16GEZW9QmQceA^G|O8Jpfs*(G-h47Q3
z{smalFH7?p`JuZT36?(KqxUn9sv$n|=7ZVt&oa$l5<irFbV*iO`S%pbYZ8#2*9)Wg
zwvw_@S@BD#9#JFIPF)ND2>A7_C^2s7f)X<bK0hYKq;!m#Y2A^$XF}4I#Vh{iA_qZp
z8^+km(@oJLyfN4WUnsPS%i_$^0`1g{RkvbDxE8iW6r~Wa(v@~d%hh&CwlbBT5RE1e
z{M=hHUHR(t+^-;7not_E@(Q5AxtL61)4rHjSFq|_(q&<fFGC+c03_$DCtJwU3{q09
z_MzGANA6kw!HMoR)&JteDX6msl_H=L3vV_0^vkW~V$V}!wlBXeH-$ZsAJ%ih&j>)E
zLImQfP3gRN5ji2dDH+W#eZnmSY09O{FBRtlSGx3EU!jbv`kg;w7TR+nBGWk3{8b7Q
zv$Va|F7M+rKIHm0*$(*S`+lE~gPq+fCK%i!sCFlIs@x9NkR6`52F6=?f^9>cbpSun
zb@T&n<L8}d8HL<~gY#x##>e-}ewbU()`rF%$+H6a0>&U^9O1W6N))E9tFgCxvcQTn
z&40^v>{njn_THXNz5w_Er>}o0R&otaTduf*kqWarY$QPxK>(@NI(qgrr|eh2Z0ddc
zoiU^Zwql`6Xh;Nq9Sz3f+S{NH`|9Tfx~8@~@F;|ez<)0DM7m;x>xjg?!Y0isiB1V0
zrBtOJ5g^QGE^TtAwuvkMeftusvGKA5cw1S~tQQw!cC7IRdKJzkF2;ybmLG`+tGw;K
zSa-xL`t)S)vIMeBcaRYetiA&OWf6pcsz`_~xc6A~u<2JEFZ4XEuhJsy``N8%+Qft4
zd={edkY^KFC-s;%OWhX?x|ERmWU;#~H({@=kI=vK_+XIaogctwij)^&=Stl5vWj1P
zV|gRdbB;h?p+DsIqt~HEmqf27E(>Q)(PVdS3nkSx(6wdp8#lz-C4cFWiP?i4EKj+J
z&jNz;%cG{MB=v@dz6%4`xe9lEmHwq;84Y?L-r_W!FaB#EeAJRK-%ros*Uq89YWLHZ
z9A0rqA`^d7xO?RO!jb@LlKK7TJBR>tz!A_2z(NAUNPTDeQ1i)O2;veK3m6FgDRud;
zZ;ccEp<}eC7xwL@@jWJ>NVQjx;Wnj-#^M^ll^c(rVmKNqZ;yTO@`t5DSdCNFu?c{Y
z<<|~XCca`R)?-Su&m)BNA+lanvHh@^T7Q?N$Z?QRqt3C7zFN-FfWAR=p>f!y!mIpF
z<jrGJwAm3_ceocO@O0@@D4?-le+a+>%45ii2qn$|o<>g%`8Vp4ZgMg{jW|*@0?%%Q
z3T_7N8~{pD>fQ2N@P{e=Gc)Sjy(TRKD8;;LZxrVHoT@Thm#IB4*<*>+FQHU@c0vOw
z{n@rz2YP)c>G&Dd&xbc3La)9PC@Ncy%-l`QgB$9gq15hq&YdLwv&B>1cXUwp^1=f?
z%w0`z)hmv(c|Y74eIiHDyyLc7uA;}!+_FT)smU*KckTC*Mg1>q^=YjeT%gdT>K%_8
z{=Dp^FY>(2^3I;J5mhEg!=K4n^R^b-dUtsG>*=ow3sLZ2Dk<kM2DE*zZM0-1r1u6`
z;LnD!Z@X2?y$`#L_lu#nrlFN;bQ{x0i{xy-Xy6F$%<&wK6yg!;@B8h-741I(x@1yK
z6>064Vy>>g`%LW)YDmkmFt?@?eiH1X`9HjKmNL&xD_(vjMeA!C-m<9e#aH%Wr?#Ug
z;R67E*5<IYX-i35jGInyzy)xYEG2W$=I%y`)nl^fwN0TZ4wco=U;dq!PsH+yE<$>O
zN6g!LIucD;+Le<yplC^>vlyIm@SZ0g2lf2QzecbFnrcZ`EPe{Re5af<SMWw$JZ;pX
z%I$CoGr47SV1<b&DJf}S`xI|Ju{V~R4p;OOvGU67ZOOPfxVjxNq@xzsoho|?JUo3H
zut-L~W<kNqmj~Cz6Dbc(%!VmJ&fxa+Ep}YCP+)#hIxT-m7&`+jO~v?{tY>k-m4D-O
zVL>za+{f&JrYYK3T`lgispKzaS@&>{YKH!AVv~+eFR&~0isHjaM_8gg=RA=>Xo((a
zB_O9;B%>)5vZ_oF>2<*=BlHNlHd?4XS-Wvq7fSF;QbO`9U0BDc)c#T@*BxIK@swz^
zE?55b-@U}C+*TFn<#Ru+&6oUWH%Z@v)yb(y;Lj+Q7wM0DVu1;wda3M_>kuz~V?@_}
z&!hu>4=&lG-ca4*VwK|>MaN1aF!7*R+w1{w#>Md}uEez>Dwk!03wn*qyr7I44hSaT
z&R<7P^s)V}{hI452&qE@aLtw<Q1m(PD@27e4@kY8$~|7m)K<kO&ld+Hd5Dc|G(1E8
zZo-A6UZB|Peaqx@4C3(Kj}Byi1-*>Yd>yYDcehE5Gsel*{g5%`S|*$d<1t<#r#G%a
zGUOqyUr{)+|3&71qLp^`bV2I`3~jS}dk6dB3qNzcY2IQAkUePX1MRK(rrzOdU&-$l
zt&8P`L?=X|8cJ=4v%|_vz!oMpd$QAiOD;!ANf8LrGmw0T*Pd8O9f#N-+IQq9lar#o
zea!v~UDH%YIcM5vXjv&bEK$1<Jrk+AQdElV*&&@Xz8m5VA8gGJl24)DRbekl#EL3o
zNdX^nf;DrV@;@?xcosk#mPuFMf};1sXOF|{GW?w{bx?>3NpHzQ(mO5?sjLL$X2vqK
zne{50#e)-5Y5eM90vR;_x#n@M{(Cg1CxmD;L<H=1q?Kp<7IDyNn`OW>N8HRJ6p~(A
zSBqnV%r6Tpvt_4FEdWzgHn|xyqZ7UM^XJd9$sfTL#vnvdc<|7?jY-+Yc1xv5^77ea
zR}gIMO5d1`{=-<5-ul>uvyJubVeu0z=J0^|L1|{=fAbaD4l9%Go6)W_<=w_AL~tOT
zU$(4CgXBc^rozyP5Z7QB*HT$Q!8ZrWq2Wso?~Y?JG*$ENv{`NsoAJ5cg<C?vh5oy4
zfzhlfkW#-n>Socdb^Erkpubmlt_w_o%xi1&QlsG}%7AwUnh%XkP2mJ<VD|JDXg~|!
z-*5nqcAFXJZ{X|+cd<I#o0XQ7d<X)G8JB5mxDVXGzyQ5HlmXP9ejB5&_a(Q?H*5t&
zfcgu(R{(f!DX^IU<!E^M{mmN{@10%~4iI}90<mV`z=fVM_%2n*+YL|%;8qqE&Mn10
zWmG2KWX1V~U;x3;w~nd|#x^TR_q%d@EmSkvtezg9$$>GphfQq#7yDRtw#8FH%&F39
zV*voT)_7e0s2%ok@?mSiH1r<57G$xZWP;!E7*u~*jx(cT5Gw3eULLP!jBkT7fJ4?a
zc%I=Lc~c-O6k`+d`7_)<^ZsUCcxR?63Ur66el}+%1m_S?IS&Q3x%bumW}F{5iJtM=
zJqPX_1y1vzH=h?Y4nT16dk&6LDa5MmJJ`#p&+93k;eZa?Ftlshmt+8jo>9B<>&S@8
z&CLz{9s?5-H9h?oz@$-~GHnUMw3_(k2UpJmBzSP_<&hM>jBPL2KD2l=Zel15F1z$^
zl~*3L#OBm7`;{trCrl|W7&i0B=j6WEShUmfr2fVP(DTx=BcV+x3<7fWdtbOqIJhDw
z-ZY;rcS#7#ZzG+aA%t!9yQvK-%L}430(+pmI$_{M4LTJTx3;8;bz_QCJnwJM;kFMz
zunSBFIq1AMU{Cnd*|`Yhc(TjNqJa)iB;ZRgALaliEM<g1GbpRDCCJ40ngG`!s0nB*
zwS$Jxhrr*l^)pU%Z0ra21URG){^w@?{+2uswX76wM{;{U4`}Sa0XM|K!GSeF2K*+R
z6AuVTc|A$V+}1<@f6oG@Hal~5#C5t6P5KL<Y6TdMB|ut?%x%HXV_iQao=Z?y_H{N1
zxCJZIE?ghJ$%0f3grYNgMn)6$&Djg20+piJ;fHINCa+PnsJ7qH=`IT4HZ~5DNXAF8
zS0F{YX+cih@%k^~`<m<}Ica9wCfbF-?6R;fVjx9!l^&U2%`I_%cRRT0#Hd%F1Pq)p
zOxL^Cf4l5zBEa4Cq)-NKZoE>UJNMeUx=E-35)Ka!$2+g8sye(_jxgjTazE;%xxCsg
z&hNQn;sG5caIQLi+Z5mdA;L-QQiX&-C>toD8}H3joeFg%C6Va5EutX6HyFUt8TLwH
zupF@WZq6I-3(Mp+HHiR|R8pL|Q0snjugG$|2$+G8>*?u%?iE>4Q3wghoI9@>{5Lcv
z2jDKQOamC=K(^-tqfG;4)E-{M`52~JmttYYFTQw!F;MXg**f;i=okB;QNIqKt`M0`
z=Tq{XU+JO3d0k-r{Hkj%_??iu#~t6Tj-RpM&|9iQ3@0uFA5)YCKsn2GxKDcJh~Ng?
z0vKkYdLVxS9M?Ha`wCQM>V^P%h}j6BBnyE%P<+y~F8N&x?Ey>TZF~Ik=}LP_1Yi-D
zr!rI63ux5{z*o;JOVsKW3kyq%``R;vv0^=F^uySR8y8B5-YJb-QlnK-7L*p)+|puA
zrVJjfQYlrkMRJN9`?#OB)P=vQu^S?7gngG^e=p)m0~ip7TFsXBb}z7>K0?^I-Z;ol
zto<($e3Yy;nHl2cf7)~uEtD5-qXxa`$cPZ?0t_rT{Y_jE|2Z{vR=RVeo@rEU|F2}g
z*}lITZ^#R?WT#VONa3|lZ#Q%W?8T|+jxgemkz8eWAVaTSYJv$x2;cGY%zJSFw_Qx#
zi5;M6f)LYJ$h;c33feQqcujy-eeLyL4Jarr<h2!Sg1wDT;6_PFnQt9!RoG5T1P=<;
zwgecTnzE=pVqSk8KwLZejAZLD_;V_9Eio2LQou53H?JHvSda}%1*j~1gQu(R-5H~M
zQRBUqhp@YmW8$9|{!>_rg+`EU9ck%1FraXAdkeguus1h1FRreT4#1=!vVwz({u;RY
zAxS-f)qd5An!-mMwkRK++XI~>2^^+wKwDqE+KCSMGb4!P3?I%o_8;CH_p7aF#DVT7
zUn%sTqtqN6IKKEynj^f~UVbo8aMdY~hyLzJO-}gaEK-rb%D{E4+gV%%72Um9M}-y!
z(x%~*n@|#^@Knp%H~*FNPoIu76!7fqF>WJ~_8``MUbr(D_$O{YUxgmHRv|Wm#uWr(
zb|ZJspe>Vn4-b0uZYI5YlF~`-?eQu{t_CQ#a8^V_1kNu7O*7CZuvJR*n~?Zi_j-0F
zOU;4jATYu!8V9O~p2!fs$96{TfTnJAp`440OV))N<y!er3St56L3?|9WPH5YF$;V=
zOK$HBG_bRRr_{$Xx?9M|RIWI0>wo$Pru=tMgyqPdKi2k|0YFhJfjWLGhF5Vx{}T@_
z*J+4)Tm%p{(4O=05%~h!HHX6`ahtl+F$>eAvyDOI3&$rF?U}=yz%#)JFs<MQ2jFrb
z&lvoZXj`vQd|TtHMm7!zm*dy*JhPeFTpDu?Fup_{Tht%mG?c|JrPG*6r=X?%6YCD2
zceTL!p8>7&hK-WQlwOS3xU0w^P}I8#9}=+=GXCVc-N^sWf6h;u8Ea!SZw~=t)cCQK
zl?P*oh_}Vzn3^}`ZO9qZ!<ss^j1>*?ySqEM|06OQ)??tcOQ7e4*a*B@@=1k3%^V??
zQKOlmxK`B8?qF0ZrpX@-8ThuU6(6<)Gg!t0^5YlqvLJeZG}nc|44_0GhlGb04b4%2
z0t~oik-L_xPX`Iv5+lpZz?!uetFl|cXjejZa<si5_fi|mqZq;zD@lo7HnooZ55a6^
zpY6c@VXUrYthjQS(=_7ioD#m?RFm^MYpFu*E=Y5@jJouQ8xwHAjRDQCF|}>5;ppH1
z(g$`o@Ls7HVBs4T7pG<q+1s-LT?ZrZa}a3cH&xZ1j%mTK@s$-L3{>MM{J`r{N$Dwg
z{30&cmV@{0={VxXXg&d8c_IT#0^cF>3()yjd)Q7Kqse=5eSN&(kBu|{bZEoL1iy3|
zHCu7x5EDy--~~hnZ^R+-K@bUetp0u1HV%Nm0*qbJ#zI5BNR?ZiHH}e1*P`OI<iB;K
zIyf<He;+)2VKkG7YxT#MQSe6Z*Y7h4Ss4@cR4nv<zoyn-3Fia#(8Q>8HR@`q1dzxx
zM0hC^@0*cf7`A+W*$+j2`_{z2U)ut%Q@!QNc?j;qPz7(d8<J)jb%Q|XL0D->{xMuu
z6L=j9U9RBlfrcoHj%b4LZYI!3cX8ncMn(+WH2I-`Xk5IZp{*D8q%`*rhAvoPJkdkV
zPmNm!Q5%rDKSbl=YLQFB7oO+?1FmybU6~sfe5NT_fcNCV0foAgTtZqt#6#!bBjWL}
zK-!-sqRG=#A)yD(*u4_uOP4%mSzlUG%Ds5yyfvcP0k-#AU$QhXssx^(3)zV_q8Wn=
zAO-=<*#uzg3!JWu|9qvu>{|t4hY6t4!b8j-Kb~_EVFK4w3}Wt3YU$X=I5_^LK!$A@
zbZ_Fb==R(JhHe5-3<Loc;3*G)#xT&H#0C_l)Uu2}%H<e>w}fj+J)x9X*x3mMKLEz2
z@TCCw5erL8LqCo!LgQsu^LiIXLB=^;X{kJ&^8|jvmM7KCMu#!hoi=rS;Dy;)4%kSM
zF8~v)uOc>@_(ADPN!~N&gZ0#;sN&E9>57``<G*7>@us|X>hfsg3&G^so&3(V1xI@f
zfx)WysvH47E%(R8{xeL(C$B@@4uz&p9+;<x=AVN}<Z(;IBISEbN~%*LqUGJ%(J}7P
z03H<hd13B}Yt_HgkcC^Dn5KHG?z>r(e+P?=-DO8TaaCWkr<$xd8yYY}(Jh_q0m_!@
zwQg)6!p<ML&M3MCbM~oy32!6cs#T-%_g>707{a+-uWhbc-kh)V!dD4((mV#VMfHuf
zi^AP-W=Gm^$CHWcsr}P`gMilZKY-x$gk|n6xNu(`9{Vi8$d->LIgdCT2CI1&x7J_l
z@5?!n;Y&gQMfvgjgEtn%cl8Whx8tKDLT)QUFqzh&2k?<>1IjN8V?$9h6O8W|5QWjY
zk#rdST9E`aVG7Lmc`cXzpBYmtzJ8Rh(k~7|oEzvS)!Y)v36c>^>hdRzlN5^n69=V=
z6+aaFbTxh}Ti@va?*%x{062f}8i){tvkiL?|9G4-vB8itu@}Gz?{Qr{0l7GmT3R9h
zUSI@RjCTMV8PJ4@AaFk82tT>r=c`Wgk%B2pXOH>viOoA^*O-7Y;`lu~!jp{a5#T9&
zGiJ!>ZWpdJLpp?19~uzHClbx#<m@lUJkXExvOG|}ra4sKuH0~bsMibPF-Clu(Rn2B
z-`yt4X$e?bHbB`6Y+Ynu;B-?8l##&>R#5CgsRaE!a8528+Y>-yP8Sb>M}dHkUm*u~
zg8!L5Z>YF2le94IG?6%afB+dawPTP7CT(hUyugTi%5v1=NNzYnY#{Z#2g474>bt8#
zzJ4Wy?PZ5kHuj|9l(wPQ#%-VJ-<1e<GHN1$TO6Z?Qi-7J-i%b>Y{MfAjFr0cxs1Qg
zfA1U|9j%kY!Eh=D_RW5t*1N9x?E%>Ai|>$q?n%1*CLihtDHaQV^0#CP@&Ddi;B8Kq
zh0?|<gaB|PZ!OPD{>H@QrnYcFNY84e`5VV+zv92=1Vt-m%NoiY5OPzo3Jwld<6r9v
z9O+WaWdIogXvKs%^MO_Iro4J)FcDFH_F<qzyt9(|xPrDcL%<74M%f?&x!2#mz3vZt
zX=T)}8S(8GWl-kr;XwNnGMm{0Dy_|hk_rSXA5-I&bArE-kSO$EaA(r%Tlw*NznWD|
zWu=N$e$j@!8FwN>{rTBhQROYL$6J{+O)Ynp`uoMGf}R96K4b;ZuBD7#O7Li|T?oOO
z>f1>PiVR$=_2&JAuZAMlIg8;iGOD8#t*ub#q+_<LeI8lFa51k|QRo+A0tb}zf`<zo
zEWp^dC*b@44)`GBWNqX2&W?<|y?xB|?d|O=*F6#haGl__%`>1k)VS_f{QKsGfafo=
z&`UbWzW-M4#sOh0n@$hPQhEX}Z3zhD7rWFuw{s=-fGPy43;l+nxT0#t$2;8vL;8o)
z;lsCgT1@|=H$6_DEOGbafvBhgdU|9K07q$H(+L6iEUVzUvFO!nD=sk6)8|d76;`-+
zz@IT)ZtgWx)%q<RC_FrFI3*-7?E9AZ9mTh%fdD`>o2;}*>-aXO;TqA&^7b|q|Ge^3
z?!DdE&WbwXKR%kYc+~&za06)A46N@n&=L!H;s{?8pae+}iy^>fU5JHmPjC<bt;(=D
z;2{EFnf>}O4;tjB|J`YGAbQ$#*=KU9ImZXjieP7F_ls8bF6(#Tw7_m5s0v7k%lvQY
zrib?C^ngRL?1LxB!|<lTwv}%?i>6bC2iy%eN9cy_p}Cc}AXkMsLts5AvH<>KW{2*=
z4tdTyq4;;IN~k7LRePj?KmEousy+t3$T)3!6lToE#K>*B0xvB+tstja-OvD@+XelS
z|68|QO=}!(*XR3sCAXm8l>?|$BY1gxyI-!xxr2UKT1Li#$~tXeRtHb+3x}VGej-Yc
zHIah~=KB}PT^vl#>12Ys0(l9@dHArxGWRnV-O$Rd(iz)MyGA=mRp{T*A#&(15?LZ>
zt9VS;T#h!x*@9vAte<J^)WiBp=9<=Z3kauAeLUqoWC=fxb^dNH`S++3{giF`mx}(j
zJfr}z_H4?s%z8Q&jxqo~(3fe5M)rFvXjX}kFf@)31&2sZnQ~`A%H^wEn#gEk?VTSP
z{BD+Mt^ee<t#qtA98B+$T-~{P+*v<wF`W%b_Qzy~5%ByIHWm&bWpLU)I7As3oIKL^
zEJb{<Gh3qDg3UDFWIM<}+efEjW0vLb65h2fFpgmQNK)|Io2}1OE)aA_468fgY2M#3
zcK7+;8u=ayIJ*jt4WcnpZH5to?U|BM@8sok%F1wsg@r?|i9wD^{eg$nVD*?Ba8ciX
zvB56)+uPGWkE4ry<)NM$?(Y8Rx_(NutE4pae#1PePABoDz5VQ_fPfO+L6KbGv!+;-
zS2_>gI@;STLK545&dRa4h-GhXu6TCtJ@VN-dArc|tWm5<@7W*oDnFRl{h@F9Qdr=g
zEjOo+<K(`b{y#$PV5=^=dqJ-9<E83BT9mnJ&>W#z5#5~}_*hNvDD91^z%7MD-4ikk
zOvqN}IO}-t>;Q&9xQJZbZV#4Fth`R6m~>7CG|1f<A#b0&$>#H%U-&98J?t4^$^=w%
zGWvF6g!FxUX5;>Wie7fN@R+CJIzws385XR@*C=Z<imigUo4nA5#X5sF_r-Mmt?=^3
zq6{c>62)fxJMm!)9mulDuUo=42S=WLB97@L(C|DRRr|Xj!J;H~CZ`B6`w4*}VXQm@
z!=Bv305nUwJ>5<%bjOzGt?+h&$&<b=iT?MT*jPp37(PCs?1pFMyzf|Y@B2&^mZ^A?
zLR=N3J#MV^3tMn1Na2s*A0^>lH>NUko3BfD;6iIQY886t2R>*!^}lBVy-8p(?n8IS
z<a1ufvfy8+S7H`iUPN0F@;qNtzahgsn$+)uc6WXRjq~~We+NvOsAR8P`$*hgxqdX%
zr)(80+RTBb;bJnnv;268`f^AGwV(X+DVK%6t6JWhmJYVs+Cs%%$6lTfoC@r%%K54q
zqx|%_OW8b7;(IZt_5RzeZ=%522d|fjLofZm2}*PONuWej6wdd|HXIE$&)&sjNSt|>
zx2?VUVj`8b)CVJbakO`5r-9wy(WrH3$4L+h(Piq=2gVRAE{XvRq^N#R=HiIl4Q+I8
zj22Xu;PeUn$H=uz-N)Q|X4`)*ksoUFx{UpgXbZl#+A{rfrfxA>BQ&S0nnNJU8J>y!
z5O*DyJ|wq%NP}#HTkqluse~At!)s0O_w$&a1fX#4dV6j#<%Df=T>Bo^syo0sVc{pD
zULm{K&2L@0gV=|}lTkn8_Buw#Gx{#<6T@kVVV9a`izu|@KA{JG@{ylr-Qq#j4juVF
zHh4hgcTgHggn+(%$t<98z3h1dXe~OtUYe5|imoLEs!XKMMLJd;H*r^S7d8eP#Y<bm
zq_fhey<ty%-r9!W{E7m=yCmON-$iZ3w#T4q?nFrPdc4H@aQ~_L{s2LEbX|Q-jaHnY
zJoW|EX8%3Q|Mqr~yDFX=Fd=8X&yr6aG=mBz!o$ME3ybQ}13HaffbO}Cjg2qU02wrp
zg&)ls0FpOuk+<|_)kCXPkkR!B$8{6;H%}(n;4*kn_YwoKG$DOP^h7)hkVMi;U_0Xu
z;8)lk$2+>lVZ~?tZvXFV(ZScYG&j>vF{tE#91!yh58zx{4ngpP_}Qs3xzOa~kJHNJ
zJO>#H#w?}9jlIx}n$q_Mp;@R|Z6C{-#JcVv!c)}krT<n?K&3^|XYnyOYOvpMF9*^l
z<(wA*X4%?aUCcWDjD_C%-qCG#r{zZuZWLCnZR4aECV$F*voaK0|94AO?x}P6pC3Sr
zE$=ls!L_T;o`AsXu=0Q?EPKlPJv|;UK6Uf-T;AHUkR<}+Kr=6g*wub0P&wRxByx$$
zs}&HHgWfl1Qz0q$fZ@^w<2j`eO8B}c#L|sS+K<vmL)9pe@V}Whvr>hbT`+`602*W1
zr%%6L0#s{4#_pb#^UMVzJU4pn_arI*{vcOwn#I3UCb;Rz&k<E!ES7*UClZp^J=Na7
zsE4bO_G4wJ=LU+nyJY^ihYU;ILxR9iegnSqQc_<E#Q+r5R@TP{&T^DMiHo`BRC(%`
z>aiFiFi+cZ0~b;CoZ(;Z0H0PLPAgN#UwOw&($fS{HYT3kidiTA?Nozj9TJA47;{@7
zah~SO1kc&}!}KjqsS8W{zS56GekW^@5EPDshT*S@QXNfC%691lTiVLSwhIg!m0b@2
z=dKyTS*T;iOw39t41V<)0mH+FlgKmi-u*c|38i=w#hNJ7N;==-m%J@0`Gu071Hw#>
zLZ7Qg3~(hDIlrT{w=Z3y=cOn@R9hw0R#bcfFe*SGMIHeD*!lMA;CgHY&-<-dpm8b=
zsE*kmIG|+>Xc1&Q!%y}P19U!n(~(3RB4G^*qMz&YXgDBD`po}fK`&=v9^hE@d!)nC
z3o+{NSSUbS<qL^(^Bbc;=<i`|W(3-Ggt-g#C!6EmNq)(9=?RJ<H#9A#-rA#^Ip2`3
zbidMGWeog<nFuqZHRVp!DEWW_l(m~WJ3qs<kV!~nKs=OXe}6rH9}l3%Qj^|h{a+5J
zw%~!q&JU3{?pxm(my{96HN9Pb4vS3{WOF}=AXHEOl2HS3EE5O@<`;L2N~?eJPy%J@
z+dZS3ntz%<3&eY1_;e#gRb7F#tDcBD9Nw}iisb2RbTS>`E9D{mZ#8;b%U%)D(PXYW
zN#p!mByO<$b>gP;K8eJgBk}T_K${NBzUctY{F9-51BJ<wz3Fqet9m;*vb(xmWth5(
z3a%zp7#l*s!X0ABvwEo=L1c_-3*Z;HX(;H;cMfM_WLyE%dejA=eUt`zWW&Big03e*
zK5r6`jB4CZjT8D*J`m1&&BOD+5rxH*mQ75PZKvC>G!OyQ#d^KbbCxD0ga+}(nE?+Z
z!s}KZbPfd=TAtBb6M!bJL_vyyClKI3b%21xz&7?b<_V$j*J<5nEdTMJTAhn4fBMFy
z63R4x_O2{5d)x3Z<iP`CMdKI#uKL&?J!aqb*O->x2WwF6Tt&v^U@Qi@`eeSnpH8Ip
z1XdzfUl`UCTS^j-yo%n29m(0;TpZ%H%>p0{u5`lgcAziwQaPps0^nES#2hdzfV1z^
z#8(7ij|G0919nn($!HDTY!s&F@BGkXtfP$kmFZ7})7a(NgUlX}n@s>8DAaqM4Wt8&
z73V%nJ1OcHuj;yZV<cn`e3=0qlC$u|l+Nvckm)Zrkm92i%fDD`74n*z@A;JjiLPs2
zQ<1)f&SPizHqkZzInzJq=|AWTfeicNJRg{iU(y>h1)ReBIqeqU{;H_@jlPIhGnG$(
z43k7gZbjoXfD*p!{{KzgqYkI1`Xk%e4%5=huV5_UWfw*c%B3aalgoSivNdz0hgcSq
zBj%3`z5$+vyV>Tw;@5-@Xi`}2n<HCP`VJB}8tiOmfdfUdy_2iW7U#bi@vj7LexDEP
zw%zt;7Tv#c#z!<}>XbGHa_D;G7B~7!FL|%%bhRo=o^X`c0U|w5{=W;!N{Do~v|lKH
zA_WwtESz_P?d<JyCxm=%D*uMweCA~?W&7k35e*Rb(wOPqkNy=}KonzX(-~w07YVrG
zSgr(ph{a52Wam55AJh~E<GTKRoHwN(xDziu7NrqR+!q9$is~`L2d#UyD=&lWoc3Ra
z@6LbN#<hG83afxE-P>C@#+3)}f>+I79ehh(=;UcFXelNpHfT`}n$Ya_W?rTEKv(8d
zFxv>#eB<G4S-fQ_X=(TUy7P^pEObzR1nneI0Dp*!jC=|Q;s9ZP-uql91~iaecP(dU
zW*+o#^gkvf{8U=Xao<1Z9n7O~wofiRT(+6*Kl<BzlK3CoXGrv)&|lA>6Zqhv3CWG1
zJwd}tdeGk=h!d0hf?0w4Qz*o4v_Cr(yO&^cbN8T7U&|~p`ow0wc=>>p>fBo=HPd`5
zsqUx>R6l>{LAT{2*BY^8wE$WkZNU~U;I^KkWo1=hV`C!{^eE%Dozvc#0QS2~+La3D
zv#zO{<rZ0>Ud&}ZrQ$m490z_|Ws5{&@er>z=j9G)C}yd)qBn+y@%82*DWINo<n#&w
zqeoYN_#5VRv<sESBwfNI40Kg>|94z__zVpqP`~Vo6TQK;0rieRL1-)I<<&J^`fF0I
z%`wp?XA5^5G6e9CNEUCIYY5ifrPh-$zFf)md<M>-h)MZmX7YTz)Y(Gg1{t-EJ!AoG
z<IRu${#;HQ3J{R;)yh{bRBG_Kl@i!XcAZmv{n`(Re>hK_nr>-WmQJcDDEt6Y2$lfq
z&Y3lVO;by1XlQKB)ztvIDh;P;AdOY-bL-&^igrXGXMObOk<^WokGUg3AHQjz-LsN{
z|8WUO!ic!?%$Pe*>Fxj$xe=qLI|%xyp>Ba}s#hdfKrBcCK*~b9SGU*amH_Ji&gYy5
zMDA6<bEQwiMUy^skn=gcOZB<R6}ml9mQCVS1QmLx&&SXFI*{i829K+AcFv-Q5S5y`
zzi5F#sJN9z#gxddHw}eEaq$;V?AG7;SlBC~oq8V7i<et#FS;MGJDjXTH0WjaeNx6`
zEOU_B%P1n(u))A{-Rv;}SLU&3a`=!ca24oXmQv-K!whWdTrFo-EfQpe#Fvw$;XQId
zPoj`cIXjy0z{JRC12jn~nVJ0rfTQBxAuuT@1SKqsrPj}RdJT8xA#D5A+swN&Rm#>=
zWrLssI8yCw`3g#w1AI~mH9hb0IgjnaNAtxs`}JDz9JxJ$yco#-(+dmhf#To@sq0Ld
zX^Oi9{Et9AuUlXav}+Urf3WTkB^v<p9RnImctk|>iM`oiFwB4gdK=#(POv5cT09_5
zpaZ-`EW53^88|qU!G$x%1e-|*2bcu}T%vQzBe|YNpw$AEC8>AszOGy`h%7#u9n$1d
zFC0tnmQCb7g)hz4O*@Or5fD?~dA-`f>g<2J^^CdgAIb8YOvXefTD6MP&8ZT7e;3Ay
zAA3CK;|uFHUnh&bSWNk$`Je9cQ1;A;RVP)h^`{YTmAqY=RsPHXFLUo|EU065H1s;>
z=j3G}*Uhdco)S-i!=3<q+<!Db4d02O8a4Mw(OUi|U_|lVZnl){9E+t-5#ya-3T?5B
z$N65NuP;Jx(ko?fcqrQ)Y+8>3CO(U#$I7?2eN#`%sp7O9K7&H3>W=amxKD>ctsFFU
z=74$;1LIEWc^!eS8<eJ6q@BxcO<%t`BMVAMxym{CxBSnai@@f6Y5)_22VURMta8XI
zs$BZQ9u||5qFc7)+WVbr3lup9t$loaEGB<vy_8E*->H2|?p-$FTy8Ns1m@zjo2B8W
zUW3ooK?^n)XrwustD|pP7IgCjT7^zkj7$V_I)~FWu8^&K9CG7x67HHcfg``zT7G#f
zJxLf>q)9SO-H1eW{%6h?nnaW8MB`n~x5&Mjn)RCJr%%1zueGR)PMCUA$?vnDoy*$?
z*2Im))Njz&O~1Zs$R2XW%+w;5jOdWkbB)q9W8!)^;@9A^wY<o}VBTNS<?#SMya;&F
zDUr025Z+eZwJl=#Rx<Zm)zbEMJ{p;uNTk$`{54SOco(@>xg7o|OxWk<;;Gh8X5F1F
z9s#hrhnr_b!Nn)Uw9ils)?{|;HV{jomz*j!l~3rnx^*32b5+x}Z!=#k;s-hvbh#@X
zGqeI^2Uz4nwHrsdVc(^Xo6yK{b?QNl2olu6Lhx5-wa>1&t=7%kXQ>-+!aG1skQSO)
zqSV2mmffW-e=wHjTyr{WNZ!3^m*SjKC|o$&JZ>U=Co}|UNy@iYuQ7g>R4RU8F7?)2
zgw&WSb8A|0pMi%df5{aW{-O)^ubYa18T=9Z3|ap3aV`n{gR1$cEW6{fW>y$90Cqu{
z{jvlX*eXn#Wj_Zdn6xSw53d^T@3bPF2E_?tJob#^bX3z`QRlw)N2S&!fAWMvva7R`
z?pPgwm64g2#dQ@k&Q{YEvY?b*2@;|L&{(X&h_AsIn;tU+^mk~_<KG8@GYTks(1C>&
zc>-1hgP`Cnm*PX|4oIy~^ZiH*<%<`hpnzDC(~YQ)P!o4k$fd^^De9dyJ}uen-{tZ_
z5ZWBAP;&Bhxhs*}ZqV>(Zz+u&;<DW_Em}C6j8&Td^SgI>pHgsEw3<~kH#q9#7+o)J
z_wP1pe5FX!RW2-jb*N(#W(Zql*lvA&iS`d2Z`qEwfnXK{q=CMKhmV2VEvvN^QhnSX
z&n6am0z7x#yIF;_DJ`7>{&VErLv3lLV%;`1iNs!NmN|NJuER-O7j#}{<nrI}!uLqM
zw`bkuTH;i#d+G1e(hzP3hMAk)5!r>%_GLs{09DrCo@q%cSy<$0mYJmi;j){u<bBmb
z&8%RE%wwZ076aI!TUYTIBkOy~1uH``-<PjC;tA=hlZuhcy|PTTp*n3o_mqNHS+--9
zPDc^+f6A*>?{yWfu#281Pzg?oJ~|&ByR^h>Sk=GmLxaz`%6F_;w0Du=R(G9+_D@}y
zV|u@5gbSjo()v2Zbs7!f$O}+Nei!ttpn^I#o$w1c3jcJCDt9nT8vII(Rc&5drisbI
z+^$lnaj}+M%{K*QFd|`}S|lW-6F>!^V`Ec9K|yigC2`<O&e_E_Q6a;2qgKNXYaT=I
z?Cd0Z<*;IoXuCgGcY1fdPjGt;coIWEaD_et7vcTPVr8Awj62}eGM>^cf0=?HOVSso
zVJc^zWp>pct1VHWRH<5D*~D0K?e&JybIX6#sIE$*GQWkH>M~)#UyRpv%5^vAtL8Vp
z>Q$aB(rML2`wC_+uu_|R4N~X}XC7(A*WX=jM`nTmgF#407qDjDC-XTA2<d?XkXn*K
zt%&h7J`kP2!)xx###mt@M70@A?n8V03eW+*QPtg^O;@jdw^V`I=v-c1;iJq2GQ@`!
zC?V-e5^-5X157zn%eJn3#1IHA6@ez`Es&Sw7O*3z*w@@mY*gFX>x|P+eyOf3vaA(P
zl&*EZPn|aIX87t6ByCWSmzo1vcRD$ctxtO1A;I5;KD9U%5^vdaQ<BROgS@XlC50bB
zGj{{+?%?IrMni0}ed1=79=?y2b#f}}<KGt`6>w7krGPO|&3h^70mxPi#`T~=Bpo>i
z!a`$Xym?0xpq-&v@8zD=qoWI#CZgH|*Xb}QXm&fvEqa4SB!<H?z&07?XmeN0LaB`$
zfhz3{y#XB9)!#o3KB^qB^_X<3Rlv4Ykdq4sa;X{d@$p@Zh$47#zixkRi<<Hzj;9I;
z8(gPx^bdFTx;NiuX20%|l2QXRc;5c^wcL8}cmy(0Wi>Y^D(1^w#&A|OqoX{GN@Klu
zG&9}SGi<y1mu9)s1FeZ(%6^A3LHO|Qie~1d3<O=~fFko-vv6y?I2AlIP0lb~a7$Wc
zCGRN+z?9YsKLK$h*?891kt0A3w>N<!$1IAwX0`~(hHZgvb%DY_F;EPqt|A)-en4G|
zcJCtPl6ccVP*Ob+<RY=s&R+4mRGractRMjpj9i$rk@cTzK%GTlgzsxsO!kfOtzwaS
zyQE}T+*2~6#>ed)$t3eEzhSM=a?t5CIrrTb<-sS-cj*TC<4Mo-@yQD5y5)8=EGE%j
zyUr$=URqy$ljc=2-(wy%hU_qv*FLd95nh@E7$iP>UzpqQ)U+ui;dddmB&c;HA|WF)
zSl6>O+(1eoj~@gY3%JC@o>SJ~RvG*Bgpw18d0k$pY-tX%yPr*(I&Tb6c!TNwKjPj3
zD6ZydA0<LaAcQvp37!OkOCY#w2u^S(3C<z`vN#+fKm-Zy?y@*6u7M2@+}&+)U7S0R
z_xs(de|`UZ>#MqwVyjMV&Ym;VGt>R_(>*<)E}FzQlrO-MOLky=_2(G~17$4-YBw-D
ztb+Ik2e3^t*%;l;C5?Gdxk$Uv=-WbjbovF{1i5f=WE3Kts(o|g;|@3q(s$D3gq~rX
zA-X8k^aMu`b>+@3+m(wBs);qGA@*j@by7z5NX#b<3}E|1npQ{4#;LTc0&8WTs^rKu
zg{5lcNHUtvNfSSvFzGpB>?`&@hF7Z|E{lnYDd4eEf^<8XSXK@WRS?eU1?>F=)C0SY
z;j2ThK>_6Nz$eN8yKKR!Wp>blJScK_w3HlBY(365?z|Wa>JJ6}F)51TFi&z_FP8!d
zYigkW>JNWV=IjXIJPza52PHO>`rua<L-~m>{mC+c-?vh8yuLG2y>&YvXKK96Hrcvr
z>m`7iMMGA=J2_g2w)q|=eyuK34DG#q;7_F-)Zc%Wki>4KTmy+e!Eca_l?n<8G1%h8
zeOB%;bQ|I5U3>M~g)FL`rpa^5A6di-L5sJTPv;q%+%YjTHLeOCA4?u+GRl-z$<3RP
zh4(INU+GJbX7nR_AuV{MH$UI1$PF9FVW+GF`FhGWqt&jCk<Fl%%^(P7q@gHWbB3(E
zqH``#(I3b@-fBXYrCg#znaDME`RoDTM%4}nK*&WDw9+*-H7-*hab)mYWwRUU#`Nne
z-%pFdL2kpZ_ZoJb??77Z&K^hssN^T~w8#)endIV;%2*@7j<8Ofu1u|#od^20Pa3P8
zgAhH_x?Y?)h=i~I{JTa~zi<9NaxrRn2)URxPgd-A6j%%t2Fv+hH71)(9rL~JB{`rH
zFUq}Cpi$H9T2!iD{vOFNT<wYkNfS)KD(1Lf%esM7-vreW!63YqJimjB+w++~C<8co
z7n}K`*p0ulztD^RXZ(F5KA-K(7csG$5?Ck6drf-9ZfM1x25NJOVpW+RUPX}hltsdX
z%bUe}Hr4}0B~0~K3Ud|hy^0Cfnku_E6|+CaKGWv77z&Zl0qV4BYx2|Y7oQk~bUS<Y
zdCGsNiw<KfWsM(6OG`(A>Kqr-0ce%qgM%8N?A^rzu&y~Yk?sM>q{dv))w=XDGXuC2
zpj;z8Kfg{83{?$*J<H3WXm&5~{4zlyGBj8>7l@X+#iAq(X3pywl_-izv*_bg=ZnyJ
z&=WS=eB0bAs)1xceMPn+#OdqsO!hjr9`$fck163QL7e*Ib^)5;0BJ#Ma-{;ci*{q_
zgKXFb4|IRLM_#|<BhFfLkOXS$%`Yy}2DhZt)j=Zo#OxBFVnhW<vEBm(NSrF=LK+aB
zj-WoQWH<<ks(5+DP6V)XaO@WAKLSOw9ZdPK`U6{1WaFv^;LfdA%ncL{I}}T0@+yvw
zZoBQ0L^7wIC#;`B;tw7Xt>Ol?k<yID^vG7GT2@x*Piny5CAB&VUA9Y26)dX~$h?<E
zaJNhRqR(z_ZmPiI<CAj9fzsL87n^#~;|za5W5At~;TOn#h7BCrOIJ|^)~aLLH6?k@
z<q73e%5v>i$5)D0;b$y{@nhF`d!SModTg}7i5R@<=;Q?j#UrAe>5VZ)AQu-{{O+u6
zCX;*L3#%S`{*-ozIBPO3@P@Fx`!M=)$<;mPgQHVGiC0BX7$RugTEot+5aj)21LCYr
z;zt8RJHiw5JCI5;{PU+zG6Kz$lLkUUx#suo-ye^>Bf>d=`N*heMbLwnLR%`ku^#Tf
z81DAo<s--1&)qH}`X>ffENfYe22}_-L0iM);=wKrXyZdQ3-GE0i}Xe!)$F>WH-363
zgshJ<HKk@-|1cXrs2AVH+jZRNxHN58PG5E8rD8gudx0;W9hwwr-SKIR&BRi*r=bk?
z7J!P-Nb3}@ymsX?MO13Mzbx~o)CE!Hvb+%U@H_J|qeax|DkeYjj{M7I+&5|)N*Iv7
zd=UOh>Si;1@AQ5=ot1lGv5+3{0ZQ;fpPo=>HxjbgbiDVf3K}sO7T2h8iars)%DD36
za`dq1xGQ9?WhhsS4=GyRaO}PCgn(n9HWGKgCvLUZ|5mPwtQ8JiW#ibJE=SbJYr(vy
z+Uq5&(`(yFmfq&_><0iV77#XiKp)R6*fOeF$l#GA4%Yh9$JZ|(ucUR;>53|^=-nq3
z2j5`ut=UHSOr32iId5W2h#dA8Y(^$NWw;I(-Qf4}o`_So{@K!xXxmW{W12Mjw}W?q
z57Nq7m$Z4Kq(8x31rRmPQxuoB1xxi#0Cv^IDH*Tx#vw*JIy$J=HfR^<XO}E$te2+j
znGQR2{niv}%PL+2c;S%5PfKhr>D;sF)qij=+m*!!eh)OHzmz}K6!u?9cox#T(=-3Q
z#!8T}hhf_&D+i^c|NRClBv|K)jizSAf4-rX9*kLB!^N6Skh}){d)K1}S6Kg>%Mcww
zr?<MZ`;xBkvs-Li`mvXk2o_ex9nE^YfA7B`S@+*w66j;`gvTftryJ2T(#r!}{{#+d
z7gdJI3DMYS+_?Pl?XQsML#&REDT1hC>!W*XCx*cwcBoLvoMO#fLe`}<Zd389n>@-T
zQ|$8gHxk#Yu#~>#3_BmK%@dFCn5M?uNndGA`=~4k4bB^77v3>4Yy2}m^4z|B#GqR|
zJfK$V<MfjWMv}{4Ke_?Bs2a4JOuO5j(IPVjSH;~5g>r2k<7Y8S9-Xe#T{YjjwY-u%
za@4C$KI$gI?Y{dH{p+P``uo2xe`HyB11t2g@C%LJ4Hp#Qx<l?W_Yuj80@1Da8bJ*w
zDX^rJ&WcwLU()Tp52~ryu`;GLjX7?(9Cy#ee>Sg*JUiGLW(}!g6-869$Xhacg-&0)
z4~d#A9}e_*^0j_d?2nNW4vOh$WhMICULw$5ry<}mjO@4ZW%^6Hq<PK37>-+Lq|d(v
zlHj20(BvZ}PWGw&w=VDZ4FIwfSnUje?DfhJf!P>q=fmSyj@)6?h@&M90`<Ql&mz65
zkG1(&NB`qK)8q^;WX_tavh~qLQYO%Yj|{27OCK4>_|N6@R2hM^r$ld!610SAOpR!^
z&os;CNcT8T+wsNl723w5{wgf2epzfG{@c<3w*s$t%Ym41K}Bh*{}8WA;5;+N&gz{p
za(MnjexjQE!h4?YETgNXy=Qvr<|u_S773g($UrHimYQBrC{@P2v+n^I&XrC>gCas9
z0_s87^-iBC#O9(HU?k-jq#(GeaJuEE)|IQ&EpHt&>RRj*`S?eV&s7ZYa@4a|9bApW
zRZX~}<rTPWZe(#@oxm*|wrp){+qr0wX^}JBBN~A|-JX3vH2+>Tma<`!#b=zw2WmZc
z``sycSn+17j`Td3(R0V%E_)6mCtq(EAg16WUW$7%jnf}6-c~D>XfeHbJw{<dw(q*B
z>p5iMX;GqW%&rzIe0^li;i`T@6x9?W0c2cCzI^FY>5E&0bHfFGR#*2puLEL8k7;dh
z35{=hYVcz69%p1p9fQut75U1h8Hg1mX0Nk7CFz6f@l1c@cBLOkV$1FU4G4?}Ox89A
zgRy(INmDtlzpMNF_^5Tpbjwb1lO}qc23DhX<>SYkDa5qhj@=LSt?d^i<>kEFdpp6v
ztwPF*KH~$j0lQ+7)M~&M2Q{?gD+q!P54sZXXLKCg!NDTPx_^JZ0?KDrvBLz+jTY(}
zHoq+A{X6s-xj_BZeYTL4q|4H~5A4QBsM^-YnVr;Ftt>+F%p(M;bFRqcBvC?uneFKV
zFE2%|G(41I5c>VlN6!$yRw+yDwDR%}k$y3tA1E9q{7xTC<d+t6$scVw9pY%Zh#gUB
zeT1g#(QwS;&WC?VsVw}9+4;PvNn=p1LPyAi(i~X$x5~qc`k_zdM;IgJNRCEgw+Z&v
zy}w>QJi!8}K~I^k6jxcjy;kyE!GXykQ1Cw(ksAL~|FOK_W53B~^ZXoUt8fd>{2kt`
ze^*v$9uiNaox|Fo7(&X@&4j3w^OhGL9JdWx?mO^uc6^C&vI$o*=Od_9da~ER;rK_Y
zRqrP?vrJfpSNU1?$_C%k7qyX)CkWZ#ZSrlrVzZ?c$INl#MQ5bbWuxBQyi`-~fPwuC
zRpo?Io@BudSOekmQN=WfA={cO9%z>pNi_G74E7+D317$itH`}2fSf{^o-=8N3pgWg
z_!4JbgE=sGw21<%0+=tlV*N5S(kdrH-)f$8YH>~ZuV$$S@~`84S9-uus_5FZZ@#gn
z{8#Lp^)#@{D=e%xN}xf1|NXC*|LXt$_3Qus=KmMIh5l!6u9;xL5^3iG0|&<K9dmxG
z`kP?t!{cMXY`c~x*&B+<RUu{nHk0rduKwREERdY{zq|am|Nj?v{C&p%TQ@-e4?ozz
z<{=i=X0GjVQnj8~XvqJmLl*ze>dW5H(W-C$#dBDYr|8}Py8i#;@~`~=zkc(f%Y|-I
z;`AMb9<Ht3%OV3T3fPn;9h4WJ$K(+v6Kf+T=a%YE0B%Y+dW-$vA*25uFa3|d{oDWl
z4f^^YKkagK|3?pl{`+@?00jvi=Gj)}-aWx`#q>?lsBwFz#v7#sId(O)?)`T&jBqz_
zgDPdtf5A|!&rkpFF0<65bHLF&U_Yq5cCANssm)|A1h#g9G~Z0%hF*jZM@p?$*6#hq
z%2;)VS@Iy;BrtK*rnY)$54gC1Jq`pLs8Yb5PYY12ls&p#KjW{(!om{B9=Tpc6D$pA
z$pOzR*8rcC4d}ANgM;4|B^Kiqa6qU84diGggZ-B3$T7Pg$hHf$dntdUS~Nu!IDp&*
zY)b0`L5%)fJssfOfQNxCKIrjE<i*ib(mkNwwlQ+FTKB)Oc3B>HqiJud>_h3G-epx2
zleF;cf18el>5uRnoPK{1l*C}(J-Zwj5Q-q~Ps7<522B?a0y`9_ux~&&;<hxi`?GR#
za_-sK^TkS@3v+UCk)+2=uq3v==Q%r!+Aiy}>`<vWQP^@<?j@|A*b&I+RyalT#|ltd
zR5T1?`!v$W%jnb8+qd%^Xp08@>ilZ4VYzW0n_d-}Vc}e#S#l*K&O)LNR-z3E7i#Gq
z5V>|6m<3^nRFK+z0{NoA&Sl1D&zh+AoIgVfxk%XzJKP*~Ld|MD>?~ZEM-5U|gfOse
z<(Wd^sAX%(aFyPTEYFjwDw+7tgA}t<Fl=cCIyxnQ(2#8+&loz7&wGe`on=8WG}pUV
z`gKcex~Jar?1EaEbex=O8hY?h5SW}@3;I_cbyQH+9<F|RZ>DrG4Lu8s!o}c$GZR$O
zc0kd!KEMVl+XVdh5j<=O4$X>Bml0faV>q_=x-t8n$7p*iuw~Tik!QR0AbO3z#IlWu
zTvjO2cTBe*k3hrnL&%NzL3y_B*ApxS4P*+Yuj9R2orH`Q=Fk{?QbMF@-aTcN`S&~z
zAVT6qh4yGzm$M$>-Fq|B#M5_BEsLK5zG-+z<t|Z$(cJI!A?cuRS$-cAJelnF=OkA>
zl$YE;X+XgZF_q76w9VFXA@MaJ@ihn6?v2|fMddQpi~Cz!(}%ZI?fbc=;C!(lu!}4P
z%<|squ!^F<mPDjMSB1k`7J=|dKGbLHy1SHES0@5ifpg)UeJjO3Q$Qo#{ueRE$Gf$g
z*o3PX&ubx$CIf_5_K1r@LgMsi*c9D3<ON5I2ihDba<Ay|#n0l7n_d{isiT9ErcOl@
z0zpcDn)oi+-r8<cJff^R^=+x-KSDwkiIA`#SOVfcbN$k1Ef5#E<sYr0;kEsw#ZP&8
zT13J>!Wlnt53f3C?d;4`PO+*yJJ;f=i=XVzDjcgCOA^N2DG{@__6Mom^NX-CY!RNy
zqU!U`mX_CmE~loaqsx#vy<=XWnWZ{)yKX1TxK8KMIyj166`YJT2*Ul&Y}dx}SAM_7
zr#w$`AQO5T45eI=CxyMTd{OfCrUvtmc>l#Y3vscen#mFkn8Rvcs-j>vKaN2bhSEg@
ziEpd@U&#b#H}sd+Z`o_)^u^j#Ay%sdytJ7rO7(JXa9HIumxvh$ePCkLktHDS4yeQ|
zuSm*rxDMmWJ3AGxleuLgWiLZSST~f`u1-KIjpLvYG8$}g*c)OHNT3<XH_c;l-zH8q
zHqC~5Uq*4{$Mf|e^j6DYI>@ds9@QK$l(k^FS-H@uRpNd#YGpB;mab0ug+`TpEVluH
z;vp22;#1++J)|%AfLPNlw3B&lly3H2_QV;ZLSCMJy6dEihE!1S-lUE--{7-rfA!Dj
zO!YSHScUY9oZyS-LoC1J04O?p)A0&7nWB7e{-S}ZwnHfedO}iI50d{jZT*BYSem=o
zDpd_RnV_(THXB}Gt~ejZ%rjXkYkX$Xe3;IrBK)&<IET2AWsvmRaoD4EqAcTrrNf$s
zW@mfigcb!$UaV=b<{;xmDF5B%T6`nXPcy}Q+cBJ3C(WOn-vVT7eW~<}rZ}H6B$rkg
zz_-#_-uvio%wkBn;+^>9$m5sVcidY8;kB$itgt~fyHqt<1&^9_=M{3^7=_b~!jB{)
z?4^GeK^}r;yaA^sjWj_Orbe0l1!JDl-`6?yg9_AOUo%u<83`*H1>3JU-u?Qj9N-SJ
zfBHb*?bS%$ywcgH?Bf0rp5gE&frfxWX&qri^5x4?tSn29-{g6YX}Wvv6TynXh*+I-
z%&a1x!&6BOq;Gfo+Gxn}Nt<D`xl#K5(T5%NYBu{)nhAH)XR3ew<gCvYMV~OL3h62i
zmDNz+xy1~s>ErehI-n^iMatDd4S~DgMLxoy9*gepo;)Tn+o?H?W6F~`em6}Slybh$
zTiakLFj|=|7gTfPUO-l5@)w6It40%6ffSb>uq7JkNF}S2lR<6ip3MPf1_5vBp8Nsk
zVqP|_opbHvjsiA{tXwo-L_`Jx2fCAGY*fZKfKZvXCGqmkF#Sw-3TG0H-d|T4Br>+y
z`Ku@`{4MTpUfm?~>CM!W3SeM=F2Xkb*4hv5@tID#XPpWq0BXY_kAJ_GRt{LAL!$W3
zjp{1FDg&rD<8AZWk9~(*oD-K$4S1XqO@E=}H7rbuiAjqI5@KZF`Mp+vjwhpf!r8x8
z;5+4H5k>91RDc&o-IShECivH5%7_CF;e-2g8?4f;EcxYv>ZK(jK@Z5LHug1N{_{ZT
zV2H3xP~G99TF`@gkCyf|FI%190sw)MTtcnn-6v1JcebkMwpagjG|={@&J<l6iK9cb
z!ko4tk9ApDN7tNF$pF2SQNzEoM1S*qDDv&dVfAtLcy6kXe;H?o&GDUAK}T@()W@u>
z8X!U3TerKm8&PB0O)y8kk~im1BHi8kvh{Qb)c<uP?}N@iBXOC=24nUR<LHHY^}BTd
zQ@P(d*{S%sV`U+;CyN6T7KZSbyTzTkqH7LZUA<6|MTSU)xrHnn)Sd4)Z*9Ht`r8q{
zC!ixqq(>N)U>Eq_O3gLx{PQ<n%}&<C@xUT}I9Jo4Tb=WM%UK`H?fa9uy3oQe$lk8a
z-NU;-YcV`8j9Sor7EXb{zkHY~EAYi+9=s`7>UUiaqK3w|W^G1dgVb|(&L1MR46KH8
zE8f0!Ici>EO#a1?4f~-v?y=k?)7sLH52n08boHM_A|A2|*xA)JMcKi5LD!`qYdyw-
zzdqla4;RmdpGvy#ZRTf(Muq47NbBsG`Rvo_;@>{qwbORimt4yl8lR_Cgc8i=dt1^W
z3YDm}&EaflEDk+F2-wMmd%Td6dQsHjqF|?xn;W(Fd0}HXj|e|$Z#ejz7bo<u(?A7$
z?)NYACS8#)4ct?e?RGefnlTky$MGdIo3Dq-{CoXn+d2n#5&V7dkq_0SF0o?Qu5m6N
zF7{Q5^Rue0>(yVma<n}Eu&+oShJgHpbMH{PXP&d@r0UUUHu;aohvjOhBh)GbYPy0x
zOZC-gshqNJJ(KkIc1g6ml;9icOE0j00mij5-hG}<VWn-pii$@c>$&co=2qkGATzWy
zBg9Ubfz-Mu`gW7LkYGHcjyBCota^sp`du|tCB6_<v%Q`bI#z}@&Nq`MG^_P(a`f|V
zJb7ZF$|yl2jVdgJ^H8Y-8SQU%XE1s%mAGywo4rYJ{PEXvc#G8GCCFJ<J0*DU)hNX?
zALt3^9I3ROPbA5Ila@pd<`QPLx$bwhU&FZ=9V7P0Dov~YyI{!oUOU2eIW?AN<Dvd4
z;q#t>YY+~cX(c`kA3?&^7y59wZ#<kpedVQJmF8cN`J4nyV`Cz+e-gSi3^KuozOpef
zW%q-z(PD;&M41IP_FZ(GXs^rO?#%DeJ@5%NP)axmdB97iASv~dp=)wxZTnJammXHe
zKSconNl%|XAiO)hrhFR(mIBm{9>h2h|DER~j{%utU$^|hqA{-eXsfff*cxLEv;c@D
z)A6w@>%BAlFpn(=_Za2ceoOd)jhL6$#c`b=s>sdDKaM9crpoOwsy&O4EQJE0y`1<~
z117SlFu_tJBcssSr3fLqH*87a&)Hs^M{0e~A<Qz_TY)KIAZwDIqROXYVpV!H7mcXH
zzu-<<)UAip(uLyE`bQlfR9XQId}KB5Qq5%;#)oP)o6)1<%y8vqRn^+lwBj1a?7JSh
zbl)3Rw4TzSt4$CYc6N@!ELyZ5+n10GBC-`JD3d?x@wsM>WIf|#l+E*ap>i?i+jzL&
z$7|j}ZR+lOsHfOC*O$6sdEJJCdTq@f7Y(`OC6uH=piTeiNT+(U>4UU#^hjVuxjPIW
zST%VIuU6`1GxgFia=s*+(zU+k>f;ZGMY9|x2b?Z@as2ZzE?3MQr|$aMZ+z#8Y(NA*
zsqrl)R0m7V6Dhi1f5*L}>>&DaKOHh&LZ0I6&QJ1x;GviP9PgoUUhWW~Ux{I1)!5Jj
zW}r8JixsW(F6gS8J7=K^zuxfEmuqMV27x*Q`f}%ypbug@4ja8WYn8h_yEdG#Wn6bm
z1Js}OTHGf5!h6M{qrET!)B$p*A}WjR8b9OnCMph|jt>+d{+wIXE3ra6uk=<1`x^x-
zEcX4!UCip=RlmTvf-zspPLIzBBpn2o8K;8h=>pkmh^JJ$-Sgo~CpTWFliw$PlmL)c
zf1b3s-&()Q1-2Q~$m@+RT4VrHrW88L56{BnmW{h|Ol_ry#)hZEw42Zf*`v(*TJ7f&
zH?C($N|fcA6GGXx?j3{Tcto)2Gja6kZ$s^~4(bx~vEQ=&m*Fsm2Gl4vmI!>eyR+5;
zU3Hj6{wU)aQa3#?c7zbmQ4s(=z34D)+tEFX2P8?vvSG{pHm0$apR#X)#EH0M6kVr<
z+Qv)#;?>fX?_QPXMDMnG0fyV%>w3Z-EqAm#hz&*Ux7jRa%w$oqG0QiiYEuA~6m%AK
zbhmGxZ)zBaU_|P&lL2Mg$rTQ&7V2T%g2JN9!^4Xim#E2;_LBU3mTtP#LLeMCY}Dfm
z^UhGI<}(&Tg6=Kox~5)rci)@QFj%SEnhn)7@83Cj*h?56lw|U@gabB2h>L+RtUKk6
z09wMX9k<di2cEmrjn*FRt2@rtM&wepv{cksF;;iwg|}Ahk6N-XuL0~c6#;h0>_$vP
z@K!3XT!yQh{zeLPfb3szYdFHK^Mw4`CVPwRt1Dbmq%^dwxT*zCM`bedEkM$ei))8t
zP>S?JjEd14k#Y#-l%PWk|B;nGcH-uDdZ3~hu4zop<DnaWvK?)4sI$BO?6j3^od}4_
z1?s}m`eRbfkMT*GKwV5O&#cRk!*8H40G13&p=DjtpSEq@U1j(&oKw!Io>^zIqw1_S
zq6iyc00VK%R|1#(u|P=Dh%#bu88c<kTWEo{vx>o;E$i=>;yTDx`Jd;D8FB#q89j<r
z&+X4Tgb-WrFRid=>m`j`JiB?9Tux4P@Ce>trf%kix`n}L)HdUX#>FBy`5m0|8?s;{
zta#MY01DD$aQjsBMykG@xVHmznA%)yeqgocBAtusAQ#tZmI?CEtAsk%qAt3Ewdj38
zd3d7dL_TGkTBp0f$0@SGNFJTmJy{cuOPC-ZDHqgN(-^WTGvsw43a1|GRAg>wEa5!-
zvQPd)Bm3pNP4HU|uTFkP%E|_h?&JFw>Kmq5bS;2CNu3Cb4yRpg5q^3D8#^{TXzCfD
zPpb8REj9gx3jw-2VeYoZMh~Cp{08N*4g8ZXWr?CpHu0%aZ&A@RX*k2syD3$#fbsQY
z)EGtE&7c?T52j%1`SI~Ovk!!<X_x-S;k@f{^%F>><z!XH)&+9|(ni<5`}?Z;EVTCt
zWYtJ5O3TO&?n?eBS^0f8)x;$8^>_ukyCT#Zvj#C6P|lu?|F%ZLMr`o*+M-JMVnT8r
z|Bz%R;p~c-ZtN>?F0%6juJKc8nYN-ZCs#4`QZ6z@Wif|RGKKV{PoMiU^<(di@}|ts
zS2o|qr`N!LNodDDB;VEvKwSeW_ZyD7GO4K3`}{j5ri`5?8lHLO4Hxa6=A7fbFs2Yz
z|CKAOZ{7nmZXcG{$8TRdXu$8#3X<~p)AV|2NZ?KX_*K07%2Y#mUV83Uzu~gq+X)wV
zrq*N{2%K5;$WwI#fxWUPLPWK!LZaevhk0N$SREjB>(SiKhbK=AinNgW0!oy=&Ppab
z!=H<q;d+#|leSM&m-qmdCLVQw%q_-f5PqzO$r~4VKZAuFE|u=E%}*S(ArR;?649^M
z$OkW8n|iXw6F8&Jss#XPwYhmv%77oKb5<1t!nV5N0?5bG)1y~kX6PMMA?Q1KFXmIm
z9e|jx@q_=1*7uMHzHA7!q}O%Wo8D#%Wk%H+AsX_w)`9ItNU8)p+%NLz<HYegjy&aY
zic7JzTCTH#kx|BA<9T<@!FB0eiu8xvupFYh)3|#`IyQbO+gsnX?U_VZ9M;s2UC$I9
z^|)Sv9{!%i46CfW&04wd9aBoyU(DeTz(D|S1%mJWSbY1`b~7KKwcX)QUcg|!WA@nL
zjD;rC_J~nyT~(22Z?EOf<Sm(&b`=@@IO|wUHW!6r5H;s*{BF-Qzy1{UuX60muS~+i
z$c3`nzqPo$2c`mPom^9K#9s2<w~G>ofWGeC>pTZ5MofLiQ|r}9MZ7`O@?tg4?i}_w
zqd>=ZDpZIuvr=rnw)LrK<EtAEYdM5Er4Sn2F^1~|jT9LKeXt#5iDS6L$x1C2&HM%l
zB-%(AGS|hrDyXLy%rx@q58YlBDO&drBjlR={bXPDl^H{5%r&3B^Mdr<!{7jOiCQ7!
zC{xsQBbwL6*KT`9jg{xs`I>u;^~R9;*|QuI!KlbojdR^>MJ}s;+S-#es1~y^_#H{x
zlK#EKYDwtmuGLtPMi57mysY1feT}x{%Ln`&>mMxEI|N-khT})V!*YcTDF#vJ#L4_O
z^mNffCX|%;EB1*(C(92nSb+<;TzujN56I1Xc)X{4p60i=k@r)jC(7`ai}f-Cv#^`$
zX?kam{Mrazn|8w6YF0?~XKgLchI>D}?b$W>B!6(pky({GrQaYnzU<`Kg4~4ueOhUf
zp5&P)1>L%%=|~#4!xEvPO(ezoM5<V|lgjq??&0L<ydnECTK}P{Sm?#n^K4ZrQGSSJ
zY{x^dC62020C;h{H}$jnm_zjqP8;oaWsK6}N$6$NiVwO3y5g?-rf7M8kj8v(pWHN-
zmXTHQq0mVL5M6)v+v0&q=c4tsp2UlUD}+L<fSIR9m32nL$!UoRie-FrJKK6(Ly?qi
z+W10jL`7Qh`fM3yC6>~EXo7QLy~?nc6Df(zhVviz;GH@fOB<&+9XzNDCn;|@XL-~C
z$OAGUaNJ0GhV<?7yw{*7%`VMXD;h#H#-ZDZsTSeI=8_++QVZH0W9F`K^uhl%h?3$a
zSOp3GyO`A0qTx2dX2;5X3xFK9Uqo{rCgKR_ydfL_9s~YHug)#4VHflHD7*NLkljB)
zp)w%BGd3JQi>>;{g_D`U2e9t#=4I#ux`TV7XRD9HZ?#V$7zh7n>h0>c@<9~K-D-IR
ztBl$0X`}`&YUPK|Y_6*>p%<GOsc5BQSNi);R(!IomEJ^JRF!ENwpXrO97ekDj(=A>
zdfOjw%$HY)7%Ef~&iT;4eW>;su5$bWsYCYwppU`9+ugu`gUn&{@zQ-$0jS2j+9m5D
z1r6XpFL<LBz}A@W@CK(8RJ&a8HNdtqF!5-vMRC!q`1LpS`A9KRKNlhOr#0_mt<A2R
zrrRlp%?7Nj#}mB@qNw%wB6jsEL-rqr<<Ls0EOn)yVEUwYo=>Rf+E40Kih@+Tv<vU@
ztW<{~UF2e_h(M_qa#%pZS4xa}O-g)k4_1cnv)W27MfwWm73G?B>jJil(|95vl>&n~
zyQW=ZoPFZvw%g7r-g+9iu|q1c>{U}u4d@;2)EG5&|GW3~!&G|Q_4q9`vxn?rVO){B
z!=axE0anZz)2-%(S!bd;v}%+=@Sm$PYN*z+99nU!&>+z3O9|{LD%xww!)tY~3B=Z9
z(iT|e1TH;?{V|Ik?+y-m`yakp-aF|k2G+y5vNdtZF;}Bm&||Mz{j1i<)O6vXN){ro
zS?Gcn!<=sW#OYMXlyi*|ceWG~Ed_eX?)J?PZtrly2FgtP5*+4uvmRY^e(KisD#){5
zE*29^y?iM2Fl2;bT23xVzhbE&*svXBH8O3Lr3)YS^#>)$;xv8nQOg*>ser40c$M#;
zo)((<Ov0EiE-Gzq_cr#y5qcQURo%?@E7o~Zd#!#fTT$O^+8K{Q(5_f5UTd)(X7cQm
zGt0c9^lSe{o}#K}lOiA|Usn>;LS2xDL0YJkKmFvo#pqNB;4Nk2ttYaC3_AL%h0k-{
zD)m0Q#-w}Y)vS*5Pf3PjRuea9eN})nN^@GGPcC$%Y;#7K9!sliEPPmP4XBW3%U@{D
zKoYsoi7ut(fldj`3qAOaHB-@NU%%$kd-dFY7tC2F0v33Q*(?p@rFrr}tt|pKptsZ^
zr9!e_MOQ58YyuLAeHZnR-=u$FEhq(GvCK_@D>Wv5X1&V6gI1ST8rdnNH(fBc_sb{#
zRS-}sNiOxGgcOTu4hZ0j63Vpw0l918hXzkvWlYj8yg5G6j}^tvy||sdR24PaFTG)-
z_2MHB<DrCzr3~BvV+nWu(!mK<q5k}>`^63@7J0wXluX-b>he$fF;{h74>3xMo}EQK
z-{$fo_V7c><g&AD42wh6=l*eL2Pzk?UYuTqawm%(eNjtFJ{FWZKi_8CO6g#|G^v<3
zCa5gG(!ZnMeKF{sxPZO5ZL~N|)mo&@nE`<&AtnUX&T0G}g#XOIcikt#b0WXv!)|YP
zv0R;BuX_odMsxJc!AVP;Cq0nOwjkP&TUP<7VQW7XVV+mO8MDx_`+hc7lRWet#e7WA
z3ZQIKGrlp4ac^ZIdNL)@Aw9)>cA}b-Whyk~=<c-k-nz~O$*h0o;vHrMo#_zs4$HK+
zfo%yvwej)V9A5|gt{ts>BYmlAR`LOU;mWm>pT}^m4rc?LY*BxN&leMs5%P!WBFxd7
zaYW*#&#gC#M2>K0A5cEM=pojGR2|-<-v#$+sipD7#lfbR_r3=qeeaOug1m~KfBth=
zMyyQK5gY|y5XuULn3)-Mf)*(%ty<`hV8<QruGqNAe3x3Z_0u|msRn<rx|=;b8wav_
zy!%Qdb1vrZ4=Njs1O3H&JY^c}Xs5{Rcw$rctC;b6Tk`uezX!a|c>g76TZb|@0JUfv
z-ah@_#xoex>*rJZJF@v;5jOYH85i}z@U~Cm^qHZ?08}opH5K-;efsAt*?2m7LD${y
z;?Y=(cNm>M%6t&ulz$P<(tFYmyTD-Nit?c#w9y@Sw<5txN6&rwki_LJtCm~aKH?Gi
z^2qp^S=w>-TRapU+0L%>QzH-rc%gGAy%(WY<+UU6CB#N@WsHz<uVZYZ#I{ptUK;_l
zFeFLYW;2>nuujabiP3X*jzr-R+)<Z#nda}bS-@Xwwek)rDJcPVk{B7eSjH`lG}XRL
zE#M_{F+M^@6?}-?N6}`~nFOl@1+=96tP|1e%JqvYKfJMLzHw%vZCi;k8z`LD&iD9y
zhg|W(eQXf_)Ssd+s5gShe!07nOo7mW$^|<?_-c21*QUf`)WH}Ww?6&8mAVBa@B(IL
z@T-UW?xc<l!w%-P{Py5JIpA7SiJGmt5`R&rg0S6F>XZ!0M)>JbhU^bboA671`~9^I
z?~gJ+O-s<g6-Fb5C+TU04T4~VCb0=3ghS*b@$Yqy@B6PrH_1M@Po4^CtdnHkJw1wn
zj{!&v5L|Z_)fO<Vgi|_OLRGSND8i`Wj8De4wzgiIDg9)(vnv*wE^{gCK8J6$I9L1%
zvr;Kj8!cT~(9rJ<OqoksVKSkVym5nGkkz1EJ^NzPb_auVSK{3>37Q2`-nz30tbla3
zl70YdnR%Tw3F4r^@^_9em4R1L8viN~Y~w0z;!aq3)GG&o6|<n5szZ8FgS%7g(PF`h
zb~D!c8t<g}n5hB7papXUOW(WKfX@I_YiM9_GWN<$@f*DuJ`S6-jplQW_GEN;WvL))
zW^c|X-fSX1`$uk3L_3ICs9RXaaczv|wIA=Vv=7@uoGGS`)of^r5_8?N36xK85Xb7_
z!xHn2nxy_eLNvi>931-ex-_|LKuA*b|FSVzTI039h~5mSaO-QVJ7*<Z`P^ClMj+H}
zvoi2d@8Q(S5lQOYg2ilH^+*|5<k4)v@mfzE+^DX6x$k_Dg6E0$^nAFPZ%C{BVGWNJ
zGGFV^+ui{*BiO;L_fiTD$0s}Q{53B4#Q(<4pHu7WBUjxSLQ4B&LzCdiAufrMxK_4~
zm;~v^qO<F{!GV}Y>M9YMV555}^Ctu7@Q#|r7{>eR4*TQjdN3L;OnPuM-i^>`k@@g8
zo=4$zl+$;~zRMbI)5h=E{kmBbcSv>wX>%#Q;ABsNjzt{2V+=<TW;7*T2tIap#^|nH
zxvCg4*w~v)R}oo<_{zHYyMu6r&9K&a%L)qYx)nndIFRu}8<)v>1zD<$S8^nM-iMn;
zDYXRhYkh=G43D4Gdg^r-oV<S$6H(<x<8eZt=TF?k{le6Y{J4#{vv9xb?ww!Hpr7ih
zzD4Dc3X8>D+c&Vdh!&R2%QUza7Iw|Z>$*4FR{i`9BjO!PA}V;ogI<2H7D7}BY}xh1
zeNkfw^^HLObYJXmse1g7R^MzeN>oUnCn5?fK`Ha5Y^PD1ADu_dhsdu(JlaD1!G1dn
z&vR7X#}1ZkG)TrB?BZ?zkbTd|y$FRSl=^OB>Ia*T_65JUx0#x%GEExXH0&(Liz&NT
zAHPx4{*@&bEznsn`_?xK`E&a_uH5dPbD}$GGuKXXb<ylvinD_`{qeh*!)Lp=i^@F_
zMehi4#Kecshf%L6XYmJ1<M<j?3_QebwTR=?!>Yy-nHx{5whBut+yj#Awuc}OMsG?M
z?}EKPVt;SFg?NI=Wa!v|!1eIc<C}l~!~_LvYOmlmxc}5qm8eBd&Yt*1U$5k`H*V$#
zyR#(kNPrw|(Zgb-5^v_*F&zBWZAmJ8Psh;EW2p4=!!S1kSipo9l@wO~lF?0LJIU(7
zjH?S|f(Qy&-|taD!;L-Y(7y9=;Hv{Ffgl}T=e;<bDIdJ)6vBtI?_jd1g^}Ry+V!z^
zomwXGOyw=HV=i4+4?Y{mb&;ahIhKs<1VIFobEdmaL!5iip&;L9+?vrUKct6V8r+SR
z6h?+X(1kM|EcNlmcF?95`{s4%W4&4LqqXpj=Ct6!+EI*26!U@f3uHpS^P>7NOY^U(
zEc_^U&vf@=Yrh7@QMCE&Fa9e2CWmik%U_O{KpWUqKO7j~3ZFmvll}<3BI%{u5pno~
zhnQF^9ZrsmZ#XyRHTN)l^CfCuLP*$;o39j=$6k(ycAYb{YRVS<sOxM^vhm?gQbw_F
z0iLi4rhEq;wRi%xN8gTq5denuh7<QRc=ADxBljo0nDwl$r7zf~zp1zLa>1`M`Vcfo
z7{n*P4PO>K9_MS{F(^KjyLn&eC}gIPU1*8q&-`NAwv5u_iO?23rAaDC2R4E3i~CL7
z>->!FXVuI*Z>Y>Oc{-eOVH;cYpX-1B$iH6Cd&~jd12w*iJ8Kuz?QlQ43GhBQX3|#=
zE1}^2jS3F#rJXS|X>C2lIW%6hMr@sq;`HR{J?<_){4?TMHWBU2%<K5An$6aUk8gYC
zYn1Cy>8_<%UMV7wF;s52H2SHQWc31XWW=9_V6?^o<YE;zFvM73TeUsQn&H{$;%F#+
zt{(vrKQa?+R#FVWR9@d%{9TX5G+yOLZrWiXXwutI{>pQ*kxJ8X%_O!HyRX_R-Fi}1
zUyGN<+q;V0#mKK9nemuG%E)AAfBODj6wXwz$|6r>gm0W`Ywzb9W`26Riw0!3@K8W-
zpA~(LEku53a~s<~=X~2ZG%YZmMhM2eegkSxO-PYgVY@{fDR*{Y|3K#1B*AK<X?5jW
zW!X+0Lw7-<z10RA@uOtr%!s354T<zL3&Auwn&k6Q$^<0_=Oi`zQeV=K!I7B*hSYU)
zVe-5ub54Ko@p;Id{V-v9T`qs1?)DU;G-!*HA@n<8;|=yaS5x2ODt?FTAD6iVkc6n}
zU92}237s15uK8t<;v6paec>bDlpH1m>faRz)?olq;1uDPmxV9p$Xq)O>IR#?_fy>%
z%offKC?O<mzAerW_}sC{!tRH~#7()Uf%!9Q+_nWJ##VE1lXc!#<w^T8`h=Utzp;gQ
z9B`?wT^ne3r!*87N1!YXARbDXz5P@6BH}`8Jafq=IA1&%QQzU>`{l|lDHDRk-CD=l
zhl(Ei&ty`A%&#yzRU01jwZArYvxkIM2W}BQpB>#OZf;IZ8&~)$T)BdkovVxJCBtSX
zzK>j;d2g(!t3oCIU8c>#_k86yn~@#wy>DL(0Pni@8HHAD$yUvTG@YJ)QB;TrJpl4N
z+>XcRYz_|z-5C~MxuVHc3{`8JFG*7~kq<@ZuIGld7zrG=-$QKKYh=Z@Jsd^8efdTt
zcJWHYL<P3Z2TMg!>09<^nyH))3qc$7ER0k`P2b{iFp{+t2%-^Q-h8`*$5>&O6WXwz
zquPb)c}6-0=2>X1gFK=q>*$8(NL61A^NXw3e$z7<)n{>H%)S1j0vk>r->dr6^-4(l
zm>Vk|v+f@nxc@7YCRR5#61>^DsmZMtUOAwt!LQ@s#zARHN~&&9{&F$&JbzaGHZJ}e
zxoX{T2*-`nqDW^{=Nrp6zYQp>WR+CTwFiwXAWMngJvMAFxE9<j_{h~yng|{?(r5mJ
z`3AJHP8|y35Et3vc(D_YU(+twKXokbxJyFV-hXf%OtFJ1MWlANBByhuJ&pi^sil?m
z&E7|G5<^u6Ldv|?lgSGXT)#tQ!1(|A!d&_F_n!eLU%AAoyx%L=Pq!XcCeQwS)rX<4
zgz~={S7HXr;xwQGV$XDKWZl582rxovqiFc*k8e_1je>lE5pwY*q=&12oWX0YX<rK?
z<fbgP7xLONezq;oF4rF{luUNI62D4iS|7acX_elo<bYbRbo?$OJ)hSoRq%rC?b}Y4
z3)S9eO>WEn!gzKoSXcg4*CQ$ohC^uaHyaiaW<v|Mmq`OG6Y=V-mu})zXzTGNPuV?E
z)c%VnWl_;PqJ#c?egzq(8aaM`n(kBo*e);oRf+vamrZ-zjn26xM)kGx_f!vp&MgP=
zXniAjhbIpQX*fThZOQ8#<*e;eb@lqu;l1rfq}=H$qryuG_P1>;(5jr=((JOoPb&5e
zW%?E`$(UDiG>HlSc&1rf=bTQFj~pHjILmujv;4}_lf(>8&bYk0S4035RDsH_`OL>W
zKf*9?Z=;Y9J9_3=09de6?%iVbd(`Y$q@MvlX^Jkzdd{4H?Lb3QA^|f=Nk{!y8{M1;
z3_dA^5%l8hW)s;%%xr<A(k;cd#fxL-X`9<MjVBZoE&e)et7U}ZFZfCMDl4WeO=L24
z72G_Y_{`p?&yjN>Szfk%HzLtWi0g7v9><)nu+v=D-M>H|+VY!~<s~RG)H^ZmlqdVV
zw{&A3%&KHAYgG525Fz<7TO^0nr!R9!PyK1BgLN}<N34w@8&o_M&MeHZDeCLWgEpgI
z>84Z5bDCDUv7yZL`gnKxmMgAY=i}`{^?s(>`gT%dN?F9Qdou2DbQ|b1(~;6&g<)A6
zQ4z$Co>S#>QpWSO4XBu!4W~$#L<kd*i6yG`G_^}E`kjiL3XYK8M@_WKFH+#>C}g>E
z2~V;l3=9xMoCR!LGY~?jaLx@Ji4i~F&&G*-a`#^wCHfBq`O|vFn|@QTB3ulUp45O-
z7!%aZO-z3vO9(aI&Hnz=iK=gvfmKyj3rf|T#l${Y)GTynq@{W~>~yMLQs&@MBOq*f
z^PF^u@)~)N@W6tJ1k<_K9!g=eqi|BXqt(O)lG82WaTgTi<Fsi;JDS4eURl_kY!hO6
zAGr*=>N%1uthBpIDcJFr9#vr7(Zj8T#ZlhaGgd(Q-_}Bgk`9+6&nv?d<0v{y@ZZ=%
z@YHPO)Q0-LFE2FmUhArRM;2`9ZOL-mTub;-$ynOlJVZij?=~XAHTiW0^<(-Ab-KkW
zbWXZD^J`~Kdq*OB0m*%jbZYzyf;;)pXifNJe<9nki7`~B$KIZ;dobPcet?T^CK|NN
zmRcy#k5<7#Zj<5;ex5IENRR}*(u7Lyuv=1oL@5`ky0!zC3gB)P7#Gkjc+NpxnSd3k
z>!0X<s}l+qf9ohbSwXU}*dDZ#NQMjH-@m2*U+d@#^zHo2rux$+7PUzhqek@b2!|Qj
zt**X-sG%yipch3J@|&M3Xg2^Lebk+LpV!Oli)7Q!pkP1U#9#dbVe0yEWy7PKW_oHg
z`{?>#`hve18BE5z4aejZXwnmY()jyd=iD*XM7>GbTZ+W}X^^wNGWrL~eLL8JYR1i!
z5R1&won4fUMEfPLS4pufEPN3J`)c7DHu>CKab;Ck|J=skEGk{-pAvD)*{H!${&pSG
zA!QF%e$~=ZKiR&Q+bN!$oZU`!<pPH4z7Q^ymA%#Z`)NtWBj6kSC?~$M@4yost)^q&
z+sbB5L#uJjBt&(ea0)Uv*2?qfj8bh64pR}V$M?gyX|xCHHqz+R52MwI@!g4Trzit`
za<lXI+{<_O5Y^Qr922#o%HBs&TvjQr)Lh#4!Cd`v&du5^xw7~yHTBV>M@y=&p7gap
zrCcN3!~OJFVEtFE$fvHG+l<_lMeJM_o0?pqhGwH?2%1tNqFu;pu(TeQ;a5^l{KAL8
zDx|>34lZoKnCz{`FZk^`K-G3_9JEf^_gXl|lckxX*)omCxze&Kq|^*2=)tgJ!3^6G
zw=Z<qiWc&_&PtYPVt)RJsoE`=-jU#BHrmZ%geMm*z)Z>$9+`OMNZH)4)8Vx_J=Ore
z8`0Exo}L)=a=|cgs-2OgS<@h|zU3@yCCmVY4-W197bbr3QKWYAlKKwHaB9DMD%Hpm
z91<z%Vas-BK>eVs?D_Ky!459`Y=ugLRYpcDkZO_b9;UC`(1Vw#E7L|@lyd!A=7Wd!
zFT4+2c2uLo9!;ohMZM0UXzU(2I%Czv0gwW+N*6ykUSbPke?iZ{+*sfGeSw0airwUT
zyC8ayrEaM1A1$w#oS+_yme#q3r+1p4{S&SsVq#7YR+E#aHc(6C!qJs^0dk3tkvU}5
zBn)5I5KLFvF?6-M&532#b4?e6^mTo|)d;aHn0$acCVnouuiP^3$DOtQx<A6vu0#dW
zkTtc4NUzCie&4e>c3hsxC|xK+Dx~;wUO=5B>#OLg8?C8vu%5OzfOs7{bD^QYXu;e{
z5$)~LzkgE~-R+<Uo%A$WU*F@)txEc2IIVHh&!2{#>i##0Ptl(R*G^BChH0><u6@3d
zx90IazmS?CHR7oj_JVgsUI99?cNen6`wz?rpaCF{;ugTAVD)abj6h{^5vqo|lMYsg
zOPS^$EV@sw<mNgEg_4SSnHo);LFvKDPSzOFciY?9O+IO+h5UT}gLlYM!STe|-UJOD
zzrf6I3jM<bOcI6v9RkZ!_><ta$dX;D0WtwOFR|BaJBHzT(B?|+_>hj)_S%mZkv-`e
zvCSpO!jy!RFPpw{ZdQHYwr6j(F!2yKET@E!1i;qbboz|*v-eCsy!HN>%Bf4OGd`dt
z`6rzNw6Xq)jqYp+eqvdESJF#4bBo8+{>8*iYt$bHtLFoQ;%Q3n1^ICR_SChs%5K7#
z-LD%Fmnw_thBh_E8X36}utg2`S)qqII-s>%hhB4`;oKSM5_Gwha^s7|*;P5G54Wd@
zBX95S_c_P4h3+`<@|&NGjSfe)u@&elbQ*yLAG=5&r1no}bb!RudX(gjE~d4Wb!rD1
zD01tZAYqJTmXnKM`|x~2A?|Bzs#nKKC1sUf%r_ZGdfx6Ojo!E(OgQ0&TBmM5M)JQL
zBvGf`Z`boSyHXby*tTXBc?oR^J+W~3f&T678z7OTqAHY;xh3EE26*f~;*GpyBppAf
zybBJzhy&!}vpv?X2Fg&c`ul^PZ=S@7=Ty2}BYb0RR`(AjYehxnjHR_Vn6pB)kmqOH
zTQPES({C)V#tENwY$smZEG~RploGOj1(DCuU#(nN5L@dUd}afQ*hcbg5iZSLJmSO>
zaiQ3g%f85lV(6%9J<a2xCdX)`-N|U4#rTwk8S2f21HOh#B1W%(>hC^Uq|W4?G+{)y
zF+1QSG;)AQ=EB46nHeFU@orn)mOt-$!p3iJS~sBYejfHv89y#n!E3i7x`FM8^#+A_
zRI@btIiAIPr@mVSV9Fdn-$yemlC&D8G41Vip>5v)R28wjY8s^Mg!wJ)fmhk!&DJQ*
zz?7myWB3a7*sniEkzlFw$E1g%rVGv1A+{4*^H!veddIh4M$}Lpj-p)mquKd62{~;N
z*Q6N$#;w<$t*45)%VWh6b0``9Ntn80$qSFN;x;4+nHtyU?VOr=>=!T0{o$Wch4{C}
zh(|=bt4TZFUdM*Bf%FfYczF6UX?Jsr79-A@27yEV2EZ;Q16CJKTU2`knUY2jrTnWI
zCEce#xUHrGs-a`8owHHSI8@_vx{PZ$EZ&7{>T;j&RgGF_)=GV5tY!Vr>TF2f=>F}0
zD*!Q6{Nbp>BmCDNrmJMD2U#`*M`o*vBj0U~G4J+5VQx+*9KstzO@1lV4sOL$flkIF
z9X+3}kdG7r;zyg+K-|p(=wbRIe)yE1{~9N>FYTUg{||Sez4Q5!w1X6CYq!H4IeJG9
zJ-#CvFXrbEE7&=NO_`{~uspT;8~BKQ5hB_>2KAKFfq(dIz45qhrUmz)G3`VCG;Npr
zwVZF6gsmrAykW?)sL;lFUnVd@2RRztG^Li`<5FCrHi9R{he*j${89?<A9H=o&4va8
z)edFOR&UL6bj(`YC?Y+>V!oHP{@xqqA&b9YK`1|L`~mC*i2)3%m3@9Ua=6NM?z+#+
zz2$Nf!MHH=7b|%Ts0~h_;mRLWRCxGw?y4|8T4%N#R&+2X$q*uw)BYzvE0MWj_v5K%
zz6-NKY_-yCNmhGJ1W!ZH3pS1}$kY;(OJexaVfojJ@M%NTS6I~~=Oaw(FH`pvLc*dM
zJ8JKvBj5EANiYxYtg*3i%xOE^8MTpo(K?Bo*r8KnsNu1{;@JYvczO;Wq|iS*qXqs?
zXR%dNm31tKwG&4BJ-_nz@0w0COu#cW1{BT7$w-?3zmZb)pp&rg+5~jrrcYruZ<;uw
zJwInI$T(Qd?BE^nZ{J9J`SSM0*s@TL@_!UU86_GyUj3LOsi8;tWv79}I}@VWs6$&f
z^e>X2N((Tm7cY)n4mLBlX`AkPudjVX#Sn_Vl3QW&BQ6RzBba)i`Q>UPM-;XoA8h=3
zc4RDmG}P?(26#sqo*82TF*yilll%plRvcz7=Gd(tLZXSP?3@YMKJjFEC460gBzZW3
zobD@@J{Z;vj*FH;4o>#)`S^UkR()+0kd9?Bfs#0;QetU}?_{z~IBA%k)%|0@svsZ#
z>kTFAyPa+N5VfgAwAf@_6nEtDQsy!inZM#M2%uHKj?90~t=D!9yTJqYNHow*tlX0e
z%=wh9HOygaXJoe#73`x!)FBCKeY=GwZ*d~YGFzI8Z`bfsdS5?q-zVj4c7JYsZ-ItJ
zx4m^b`zPLB9Bq1zFE;@N&Q$1YDp-yKycpoN?qoU*x`>jE=igmC7DE^a^T>E)q`jb+
zhf=p2+b0wZV^kUD*S^hOiEv-~((KVva=7kB2ULCp3F%IEZ*NI6Mb(2=#>w+9A5~j{
zk~%O~GVp5|XUJVxIN?j7Wc0?D0FF;;s<0h3Q`H2liZY6a@44a9o9{-1RTbwq_f9kE
zG#H%CC*OiOIq`^lArL0|<bqy-F4uRT8~`~!Ce!E=iRPx>(cailsvB=Dl#rk76d7e5
z<Q{niN785SQyQ86E<dj7smN5K`-$bEOur&UknH256YC$0wc&8*okRMa49`zN_ipd*
zt}iODSRVN7Wd3Xe3b+XnNN-h9@WWB<BEyDPmcD)$t}QaU)XWcZL>A%nVH}RX1ZoFc
zIu>$p%x!dw-^^MqvQUvVHrMjHtwfu|9YCs(FFStZEA#9g(eBcpKMsi3SDd{hmtf^Y
z^5f&8Pen78*_=nMSyGMr%qcuqhI!?Ns_sXMUAcElRQ?pEAFe7xYGXrP_v!wV6JlG$
z1&#@SVcxbrDwqVJKJ@CU>CBI}+&4q7H`hcs=?lMq;EXT$V<S~5)9HBgmc!Wh`qANU
zyow3~B<It?!M2Y1v;_%2ZKEB7RMN^Rb7=s8FzP7iedS51?{U|id5!pcTdUJib`w~f
z9wdD}-<YEYoa5*bAYApv2JmZ&&>nN@d?6pDt8jb#K13}~Mm}WsA_@}_BcX+osM5`x
z3q6(MeGvZY2MxuaxQudK{r{{#Cyon)tVC^OwB5TO8$q&EWH%&o8{I{_qhL+6aPmB#
z0LxxIC(HintcT0Oip}dQc{n&dMbhIzm5$m&MR<tSHM-E22&9GlENgz;zCqP5zh4kF
zlizJEp$}L+E;VNxR#r_j9%zO^?Cim$oF`8-Rn?m(rTnkvTfTx;ErzHOG&Dy?IV3_i
z-7`b*xRc8wLv#&S|Jc;xOr7S|!xh#ima}UFH%m*C0Dk&En0gDas<!TX_#mQ!2v-q7
zQl+Ftx}~H$r9--;n~kKjfOL0vH;90AcY|=~?)=vA{_pSo?xS8GIkETJGsYZqt~D%K
z%!(K+I#|VJausklH^P*aq!1(u^RZYke(l$~?i0Yo9G}}j$dp3u>*AR?3RHk3v<vR^
z7Vi9Vxv5AKgbwgS@1)yz;71h24eDD?38GS&@ML_pu$UlI{r7Gbv;zr|u`hvRws*`X
z=<7TUnyyJ)pWVcNG*3y_>`@wTzrWJ{rJz@0s_qL>)r)CrflN6{ICPF7PeRh4l`-UY
z;w(iv4!KGrp?!Wo&uJZ=ZNCjzb_SbY@3Ju0vrl4(t9D*7Q$!C*y2auR@|zZrOII7W
zf8n(ZFDpAY?(BCahIj#REYojQr;(6QvF7=Ks%9T1Ba?Fy1DJgDg_2<GXp++M3T>O!
zga0MbW<QaAG|YCes&4=oohpJP6Z4t(m}eLKF#1E(gQXXL2FC|C{JiX<q*+#ydk@yT
zcQ<3EL)`fps!e_Vm1fbbLkdH5c7*i!Epkgf-%bfL!?<w}sEmet`+Wf!ZlB0m%&>Mv
zI}=9z{P^H_kdgso5s(|B9&2EQ0e@HB5p#Zw{)~~al!awE65}hx(WjS_-I-9-3cGHw
zQ4Y4M@6Q{syTe*BGBFjX<!?2881n4=3Xg}!E^0)SlJxoWa21)~fX!<{ul=we(ClO|
zfrO_2ewL<}{jS1UuTnlNjG4Ssd6)N5OF|Lh6o>nLrH{VXwv-7knv0hEygH|d@ud|k
zhyc;*ooQIcMdJzSY*h~%?tHXZ`|}=9?nlzRd|2X-L%e;kl2aRpYhvp;v!m&<YW+At
z$~C^a!b_Si4Sh0U`7}0C(%A}C890C~NkBF@lJBI9cM1n9v^kzSyGc!gy134Ka8291
zJvg%GOI7J>a{6Q>$kJetSM8kk7%g=pHsb|0R%CqVS7`FS0WpWvk(P2PAYcGH%PuXk
z<`=Cs%F!g2soOq(1MA%zs+BSw&e)t(yu#Sz`lx5Rwbhop*VNkC9;+c+dY|Sh!qA>j
z`zZKrkv{ypraJ#;B^E?Tm_BX1f7|4nejr+LZ1yuYb76Te>*|5||4gCnHOnU~h$tZM
z{^a=@bF;f9KI*%oT3(!y_meNJLqhuEnTIzw_d3lvWsr~?xl4!2QdP_!Tsi8PXn%0F
zpmP#xSJ{fWwm`?R6heBJigYFdIPj)e-sA+H&H+w)2L84V$*quFmli3r)GMydmH4v5
zLq)*gjGSMD_o*S_u5A@o!Y$09kLOpquec>#ogcifB@no&hk`U4u!TgO*g+e_*4k?M
z&_@1~;~|ZfkNPVP*5pv>#FZ8KxS0SEvS$YOAMDS0%)@NI*#FU0eV`_ttYda&F}CcG
zs2IxFRLcuOlDg}f=!-;j9m!vI?sd=mrK6e7{Oro#8ET$Mfwt1uB*CB1dR7xn<3Alm
z-)$P0tZT-P&KzvW{#=m(J)+2BR<qA@mGK4jSr?cz1|1y@oEA)$A;f{1xog48Cd;5h
zHmjBIc;c6QuD6F2LPlyule*^K!PDP=ebA3@YdMxTUTR5}{ZC}Ln1>f3?Y7xgWf5nk
zeMvUJ9m-hGzKKe<3Hw2L`&X%%hFJY~=Y$Jplf7VVE~pPVAR)~OTvKC{?rr}SNAupd
zy+L+Y0wMpk1U>h^{0+qFfzmAimDo#RJi*Hjj1P2f@cz}I64g&Pt3Xl1QO``%s#j1f
z>|!V^MNW=oeZ%49#lh5ZxG9;dAV(gdqh}^4&k4mp2p+3FBoZRaHoE4bWZWuZON3h~
z5+}>&Ttpu4uj4M>CBM}8^z!9!cWX$=u|(0Yj-l3ty`q79mbuO4HGW0;z>7J~1D!Le
zYsc3?TKx18aw^^usYTt)-^Bpeyqe2G>r`wiyt}<UGu7u4X#5g%RX@?Ts*y)LjfnI0
z%Y;xYA{+hYXnQZrMW8i@z24n*HYrl_2ss6bFbPofQ@XQ^qo15)`T0@%IvoGAyS)a;
z4IOpqf*Id0YvdiS=YKwH5;BIn;m}zx>b^bLc=;LoxwfG(AJ)T%4{VJFQc^IUT$1=S
zDRfI0H}y^o3@pM4>(AkU?<gw+?T5AxVE&L=p0TdNtRM8V2TPsiT-KFt8jRK%A8x{T
zKibnxU||u*-lxw|GoiU$V45f3jcSC`e_51!3mR5`p1;twH1>}!$3o_;Fw~ZD84CZ8
zwz<EgB?TbOLj)%=oA!o_hIVsq;zYCGU^B0!Vs^GeC9Z$i`VeMo0kK+9M0eZNMgF)R
zeAZC9u>oU?j~y!Eyx+gJdDKNuH-7G9tlz9cQGTiDINvk+fmmEX0PS-eYeRSMK(V@X
z+g@0C?91<6!QQ`2Q0Hro(SwiHk%}Ln9Gb|?%?RYczy9hbJx?4yI@}GYjfnSQ2mO_p
zr?@iNPU=(>aZ&Qkdr(+rE+v-psW$%r5ui|-B!Yt-O@FG)$GxkP?a04$O5K<qi!Pej
z<tt@j)w534l-M2F(xLxfjb0-y=&9_uh<<!B6d^_#(PJ8fU;#jMG?8ThkB+Ibyj7;{
zqnSP$3arD|m!nW9RUteqSy)c>!70l)perF|m(?b^AZo1(`e<rf)y~Ac{x0>-yt=gE
z_i{GY->IM1Oit_S8w6w|`P*)F|56@P#Y-zVjKdxMqH#jCXDU#fh<4&^%ApJxp6-*;
z>)A2ISblC_AK#3M*f81jK6y>{P=DDI_{?O>eo>5Xj%vOYvZ^xIBS4;Sd-4Nf7h+&$
z>1ZGSn7!9zwOXFCGmZWGbH<AoBw;xbCsEA3M|Tra)yaJS7c@NwWdI+}^#!n>hHCh7
zWo6)QXLtSLG*ml!t}ty#dHfs74(mh0Qi^`W&z8LmbhhVfeou*;tOfHh<-RncQlzR%
z;OgTQ4k^7`CIdLl0)nzAB8RzCu=QMQS2p;N-!bQ@@oKQTaKe_UCPw~IgpYQYTj0wf
z{WAQ#zZEjK{fJ$yP{YU1iyMrk|EYcL_2~GhaKzC*DoTVjyj$sBK)oisxH#{qw6bLI
zytdXKm27qO7aV7{$ahr8S(f3w&$9iSGfldxDsQ1qreSWmYUt~u{lmP|*yeiY^qP(s
ziXv+wV)$n9EvJfFYH_wOzx!*k$UMHWeMR_+iDC8;?%4<^tdxk^e(l=8^D7r0S&aV(
z)8JsVIscBAT<W5jXsA9vceV~3r>Bi%@+G#ondYrV@`Z{T!|Xu(%ddq8)s;;*cXk&q
zN&jDO6@mYQbDc#qUU9)R)T#tg2m+*GizmpPMu%<IRkPp7iwg+w_4GhTejUEbrzZRm
z1OpLDDuaVsaW?leYEx=f^l9RBo+i`$D7&~bFdAW-aDo@Xa~lNOGU?xy2girLx@!`F
zc?lh3qn6ShWyhUPEu-6G^l?IN&1|duODLV_TW~QvvPKOUY&2upV1yW|$)>WYtCJAQ
z9<3H%V0JRR=LEkt>bnTQ1$DLc!}<Q4lIG?jy}irPiS?*)CwL<~cX8YI`nXN@^ey#8
z0cP<R>}#~yGJ#f^H=d@{YJ5mZ#)Li8#lm9?y*!V*Ou^WjEtOsK@9raEp(lm_CD)Lc
znCHfmrG`V=QtJI@W~`@9CQqR{Ih)cgzgd$$Qe=HLjooQ=Rb51K!inho&SkVxG4Mg~
z$Yh$F+{^QMf(K2ke$0Lt@-pfY)GnNt&d!5fV=sjn<@*<Ex@IHu51`FYpz8`4<UG6C
z6ocFKtqrTUJWqYZgi<9s#GxyUk=@WrSLk&R=rI7IeSK-^BPJ#*_}jOx9>WcvclRKz
zHtMTT8Pfa|{@g!bT;Bjfz}9HGMDpJMk<|D3l+!>Ic>B?Q=a=sa#W^jF=YUi)1r0Zr
z2ZN-W6LSqv=S&=`Pk&RI!QX#7>Bz&I9^z&(fh0wi36<;^paa;Wljv!hXtR?_!gaSJ
z*G)}jw731cUc(c2an(P!^$vRSoC^DOS1t`VaYvQf5ZA?2w0eV6;D8}ldL_R#U@#Cj
z#X##NZv3*s!5;L?uC3UgdAvxH{BB|-YLy+-z|}!}jpz#Lq!kZvCVC+n)*w1Lv}Zzq
z<hZ@01#q$jVB%-iJ9slZane2Hk`*DmB?2zvL>xa4OQ?^J4hJ_}K}%!Gl?2U$wIg^4
z_h|W>&UW(`e?{e@gB9q<e+LsB+QVN3oos2*7Tk;S5rVxdRhABBkj5nXP0f4V3kWyB
zh!3iF2J|enjh*FaJQ;%qf$y?SO@EJy=;2ki+;3*FyIN^$1VmJrrn;w3E@dA#lh$3B
z2B*#YH{qbju(cgC_bUF8_6EE{N7qQC+1_ps)3uop^|9fyaB`TMf;X5*AhvyCP}7V#
z?w~N1@3DU@;r^uYC+H`|-fUU-Yu2|CmoE%Xj;93#*Rpd%Gxxu+F(oYU7ncAu{JX%c
zAPC*5T>fE_(G)snPksJZEHt!FSXPN-aJo_&8qe>4m~JPsurF5Mm&(Te501$AqiHUs
z;?6YUYf4EG1W3LDkKR;|#9oS^eIeZs^dMJ@z`PLaAl}?)^vDRsfhfIQ<u;(uK1V}Q
z?=z+8XLoig`>R)aKi!^<=~Df%=Ghxbj2Iz*-@R}K-sq7e$j|DaJRbbA`xKR=^!<C?
zgDhGamCBD>=W@9}hUV&E`@3NgZ#QLx9E(Ad;ze@f{g!1M1EXMC+GqyrUH*Rc#*Y95
z(ebiI3USvfi0^vVLF<T#AUAgoh~(Pso20nmz6)`9cw)L;2sn}&gFE*JEo(Y^WQH2{
zURs19i1CAI{trPD8W3f+yWU=v*rGl{05#@~aN!rt{c$zH>+pTh!0F`Kfh*wVcIjZ`
zCzrW=lb)S(N9ege>+0Ts8*=~LV*D|4N1vND8jLuRL)^V~O>&`vG7{>!*q=0p83Yq(
z>di8GJ<!#&=%4<*<*jrZmeI6E#3>M~j-a)549p5faD!=?sN%wGaKhCTeo^y%_2#}J
zl}A>cjhscX+lu-?@z9-UHd=m!V#z|MAGo38X}qOr;h#iqcLLZR0XrY43mW5Uu-NrD
z09le3HH%`TYupMw)bR3mb?4ybnym?sZFdothV7qvTLs%J|1amIAS3!Bh{2U<rKUJb
z9i!i26rCdr;x0mzIa=aSzP!sVF3UeM6pkII>sco;=RxRMZU@*P#bUQTJaqn2&v8Oc
zxW3bsO3|SM-kS{->rG7AsIEjepKyGH9y|aYe}mo9krT03FO(JDfti2+YN|)aC#ja&
zYJ~@uH+4aPWE~nBi}jN4FqROR>lq&WZkO)&K2woVuFNLpx(luvp$*z$TvNrEFln<H
z0XZF6?*2!&0S*1dsmkBQNDJH@s>UaiHDi75BZgfBto&8=_OQcZ`vP4sk)i$8A6=@f
zU23NtrXr-zP6ay;rZ$dF;wV4^>%5N>5ZX*uwi#YfA1#vs25~@qBO#bkkeF%0sg?T#
zcLFpT{Vw;flWl^6-d>gC(NR1-(3y(iJV6CN0TVcH(rYp_;m2z0*yA5HJLiYP9Vny=
zCQ8r5WsP6z;kXLsN<bagM2Q!9mZ2Oc9v1LFEnUV~R@LteZ%VPiOpD+6IUuf=jLe06
z_`3j!2>i*X(RfF(1|y4nJB}>?X<jl%=B-s#WvVpog7W`iPrjOD1Soqe*4;hjrj)3C
zr8fyv9X+zU03^_XwO>_^$60joARZ^QiWdNh^*D0Z5Jspn6;5y=Vc1`@RPe&~u1`6w
zh5EP^`+|d~*dPoLpWESd2i><Mzbk$zi_*N!SkQ{E`4RQiCl}SP`BvT3^|LxxkNsYI
z$K<z@bU&qh{_&4gVf#lZ?iKbWaN9qMEXiDj%<dQQ#Y^x_F;F*i->DC@R|KiTWgsN+
ze!s)nT`yQ~{Hb8@4tieD#(d-k@>RM_E^fK;LrC#drKqp9=KkePrT=mO8rU7z4muXY
zc+E@as0virfR_lA^4;%W!m<-<>iDxdVz~1F)WKk+5zHn~aI;EpF?8ZUKB-MSpgr-j
zRELv2Q(tuj!5L(k&9(6=LzZduk=DoY^d{zG2i@yWgZ8Epe|hpd7fiK8_7<qZW%CqV
zoiDkHiP!>~l8Erv+&nZ0Z$<WBxz;;9_K*DpMn{(#=_;!=Ya_~QU*dOQ2vdcXtPfQF
zc`<e^><{{T-@~Fqvm?p!g)x4nL~XfJ2yk=X60qYsJhmCOwgxRNs8Vz5`TtE@cz2LK
zYxpU^;4o$aqM)-&9pmhRa5e+a_B<*Grxu`?B*IE(Ag81btn>@jHr4ID(%{@=8UQt{
z?b(-=k8<y42E2`e37--C!lZBhra-0Rq;lUJ(&_=U{`22?%~{ESJo$<By5h#|iOxo>
zN=%$JWU!%2BPm(gU+OsW0<5Q{1IEv(*@AgEs2n@98^;s}lJCkjO$z`yJR2&$Z)^P~
zGbbhqDuTB+`uX>E7b&Hx7s-@>&L<K0w?+~Q*<6JNHfvmBr7G&2zCFvv{@-+oz=^oZ
zw_pMj=Bq0Fa+UWBKy>6c;gGhNY1@ZRW`5JtXL~ghKWw^|*g6Nz*Ww9F{(G*xIhf-g
zKY}3p@rw6W7nI@8yQuOAvEri#9tukj-G){IvO6_2`{6*9If5<VHDAcJ<B_HiK7}q&
zKR$Rs;4hsqRyIb%;~=URX)a~_vRS=Oq6Kg?_SAUD+}td^@dBon#nW^UJ}_=xgXY#4
z3l0AP`jnnLy9W1tT+k27pG%jkO)#%c+bA<31U9B^^7$99&3wABX&?R~x6%9B1hV+5
z6#B?Rr@h@#O+zgWPKt+FC>5`YYhn-+UASkfm%%Wa_4$|LZE;s8<G-eVe*-hJ$ouVe
zPdEw`{^L|3zJKuiCO=C$z6FEA$Sp@-3+=>~Jvdt5=L*Rq(l#Ke%MN!~^V_%4+5G?`
zB-Ue~VmxIlijTVQstu#SikNc0q|vK4wM1pGIhRqra%6PAeFK^ePM|T3eoB-ToX!ne
z??0%Y1M1A0g!Al_f$;i<{rJQ%W?~pwF;xKqBaJkma{{=<%#jP&@tK8ZbR=1->=3De
z!P1h94W^#Byu3xBdPjydkKtA>V?hl;S3pk~==}p)&nGkW^-Ho$^{(G(HT;RB%%D5e
zscgvL_MzFi(DcX?^I|#q3C+0;Ets2P2i>gTQU=;>bE|g@ud5q^!0b+jGPx*pk|Ae+
zj{g(btIF0Sym;XiEo-#JZV}f=7B)BctfSSQ4xj^$Mnl|iH#UYkhKMK_<;7O5L668k
zuV9#Ycx0uKuGMw2kOWM{(qIN^&&2-o!o)ZkHrS|0ou+grwJ^13Ma}Mm8Iq*iKO|pa
zUD7u9^Pc~GIkdpq@K~R+GDyR>-i!N+_s%<O*TSyJ0(6}U;;(qLEGK_uLpN>o(R4>W
znLUo5J}v8+`_!f)8Ml)Oem0Pmd;Ks)Kpu)}JfJdnZP?-(xUmOQOJJrP7tGHU2S@g=
za4i%uBU)Zp8~@D5vNcV^QOL*;)36{!9?)LfB^k;EW;HnNLu`KhZExwQT|PVG{{JQs
za1a?cAftQY_m#`Uyn}N5*ZHJtJZGN~=7u*KC!5C~EC+o1R?>NHKMQ_NT&}M&<LmHk
zzA$JraGx#>^bQI#*?9uGAVxbo?X2`%(xKya;^Xro5^w5LzJoE&NnHFf_YY4<a|Skm
z{tb0Pz_-t!vR7RJ{51ZW`v+SSD#mEnfhki~5t_H8r78PG41|S+v6J}^wVK}{-#9lM
zx@Uttraiv?wBh<jdYi)~WcKG9sfF^ccprb%#_6iHi3zQC(Y6ljxP+Vr2gS4DfsVKv
z`3~z>EJ<U^MWyuQNk_RG4uqhK@4E!K&-ZQ)zF>W9b8=T+p6Z{8^0d_I4-Sr=epX|^
zbo+ReDRkXYHDVZ3e<*E1O8zt~R#MR5bO$$A&=XLtd(yEAdl~16dJT+LTwuy9Ik{@@
z^VY^cZ**6__0ZQ4>H(k4DdIyw1HC#sAvr(IQOx0;y?p65ce)B2>z4jskazin<*SIE
z{1bg1XoVcx8pX}6aceAYNz>Qa`S`%r5cId{W*dZxSqkwhEdG~b2YnDcdV+ecV5)+y
zrAgXI{*c3Am6Gzs>I>7|%GiWN#g~RtDNdkOU&6rhhKv(GQcMy4Yf=rX<8a)&6b#a;
zhxo6k^5261uXx|lWq1^b<=y6;iS^4-pfncZ+4DGrgoT5FYCTsCb++^NY#0~}{KC!>
z)jP{|JvV2(wWVR|;$Qckm8D4|EKy!XP{o#sA=kKub8=+76CVetvY9;)=~GJ14n4V3
z_7#(%xO?dAA_uq^B?SppD+cn*2Q@MAx`uLY&<HK+CY&xc<@=BB;K5B@g$cT#&1sZO
zInmFq_nulEynT+Q>f|&$-KhrmK5qmg{h!{sQ9aI*Cacc*Ho8rKCNB1+tg>~iUuuio
z7E{gIe(b2w<%BLLVs3ss>T`Bfcv8G_A^*+k+3z}?qgnQkV)pj5VgEyxUm%F9u6g6}
z-6t~r&Au!N{MSPYm8~)AMNW+6;bn*2Y+err^wTUqSbSo)_POWf3`S)JzXN7^xYX%}
z#i?Fd)M@O3siRU}D<aQSK`Tir#F2xY#gVYGwDMgxfRC3Cs>}!3s*5$Aob`X8YHu69
z7s_GAqGWd?-m?YmWXMq#HbRY?qgvIeX1#r-46Mz2M@&&2u1%6udp4p@Cw)74YYfIg
zd$MC(9PXk2CTZ{KzCbIdNpV=SUn2|*p2Z=E*p^Rmac{yzMCj@GP&m#OQ7q1MnOO@p
zjZqiPOp!kTVqy119Yr00gfGd+Ce>8OIcsNP5~ydMW*ARGIr&Y)fq`RsN}B0+9N9Ay
zMK!H@irfw%c4nBbM}x~H$IdDllJ>&uJ4TtLRb|_fee(8}R}Q)J{P}Z9|7_ZwCh!jQ
zWjud>c=(Q$p|x*5B&iS?cbMUkTHy4Z<D35*U>{9Jn5$20DT7dErgygCv-LkIM{J;?
zhG32yCPk#?yGyP8-)$1aMZwUvDBKBjGG8fE8E-BX9pMJI4)i_o7yMcKg9+Ti@PlKS
zC!$iwDWAAmjxJeEcbAY5)744l_}~M<q%@dNP>G8(ksv~=GgZ2{;fN0J_!Cv9TyGd1
z^(->pFCw_Iqw4h?ue5XJz~+`&0CZ_6BY)<1RlcMHpjH8|bHYXb_&<qpc?@f3+*j{L
zCk)s2QyX~o--L`WW0L%z<pk$i#qpr=b1PO-O}dhZ!5I`<vbtBYPxkT|48ZJ74T+Sp
z;ugU;Wewj7(~rbVlteWqipCmm#;&*sO1rR6a^0X52hV(C7%CGYDcT=m27`2<p<JJ~
z4CZO<hzyk#2;Td6wzd)uPK=oT!bTI|cz{&0qr`+Qmf62z*7!15E>l+}I|T1apY`lZ
zCuq-taji7#Bikgnvpojr3K?y#`8LPr_pe;5VMj&1cVD8AxVffIlb7FEUi%sC=JMs-
z-%7AI=HqWjKH?Ln$)k~bswwne_o^+jcaF*}EVC&7@5wP?{7nU*E%aa!46S~S`mxE>
zaU0{<pP8YPQ#>%@_0S|}V>%C|iH<HoN^D6+Jx%H;18jp4U)KkuY|LzSw3HrGR`!F_
zhuTYB|0r;TAnF^1w~-(w4(LT>KoylG+JezipQED=k1~5Wf?tE7@Z)2480bMx#Pndn
z99&#3D<6~AmVCor3gZZtEt1*4hy{mdpNYH)d)IjAay-;kUPt>vMZsyiqI3TfAl1{s
zR{t`vJ!XDfX*-E2jzl^HK9=$UF?409Cz%p~3`RCe&5WQT?z`vm$dKyy*ktKDhXun0
zpFqs3|3Tqf#=kX;bEUDQM>S46Ha6(R{^OJ9p4%O?{LEm76TPKCmgHSQDt!^a<;2|F
zgdR9&L~rtwx;}%A6E^jjGIFosK1*#TW+(@*xS3OkZ!L-lq(#=U1D8&CZnf3#HG6Az
z^iIY=XW0^UnAEPew1D0xeTGV$$R+Oh(V#Ij+2tCY)y5k2_(VrW#_y_xe<{K+>E2@7
zN_K5nqT1eyA?Vua^1TLzF~pss7;IG>N*L@FiH+#IwVkM)B*FLcS%GPyf!zq8K~Xl}
zc^99nK$Qg4CUk}vY^`j-9Etz&$uTtO90a-?J-(;#cz5pH@6<7eT{yEMV$L6^U;I}V
zoLhn!;ic56sxL%<d)p57*IP>x0Fcg(gRU|sMi#$1LP5nFjG<_8omGsVqjJ|X774RH
zLqmI6Y2}j0sze3GK1$)Q00U{XyVUQwAw9lPbJh1jD;e9e*&hOxktH!tE7;q&>AHR3
z{W)p23J1(Bp3yfux~!CT3c1TNfJz=9VFr8$Yj%=+sm@#vdI7rqomz1bkMKk5#QbcO
z9nZMSqyAq9&wupb)>?5+l-q3DCIAm<_5TNuVA__(U6{oU8O-xPe|hU$Hu(y^S-O~e
zIQ0@+4bBW%)MNKOcENZZCZcjXZW{9vf8t{D&a*Ikpk#H_80@Q<VzbAsBDdp&P~YVX
zsLgLrCV>eL&8vfnt9qIL(y@1^HzK?RXGf64<MazjQ|U|oV!phq^tYijX6~R?q`Iwj
zj-boGea*y;^;jN@?)sQeF88!k<xSW=)Va;NKHBgSbUoM0oC&vz=LB;zP0r24@Z2UW
z{_Oq6XPcQU(O=>M^}Q|6M8|lokWEzM1s;sip7=9(3K=G%>NvkY({Q}}Ie4PhMNbdE
zp~NEYd4YVSREU&P9{-q9Na$Fz{_9@9pB#DdMd)JyQzfR$0udWg3X1)j3!x{h5#du`
zwtso4p6bEVv$GkiTv!3RU9UAj1|>_7L=UkC;0)h6uVC=-+7Q`b>9O=D%V(QhEhTau
zt_B%WM)t&(I8p2o@bF0)e+oXB%FJxATW5=?+Rjo>)joL4h)Y0_F~WYU28?g5@EXg5
z?8ZgMP*-k`0SvMPRaZjT59xiZudN03=w<(_NtQ}UWOFw!EUxvWGUc8U5cG2HHM9W*
zA>vbO+pNiE(EnMgvYD_PzDqo#q@;hC5|Pg+0QNLimv&4~ehXTb1CwQwPq78SW=$9m
zU4O#W5@$i*R+24X;|_j0ER=(dpRjAQf|pBs0&2q0LxMhu-_+D_11UPNDaL{#&MWS+
z-1ycHG0J}Ui8aJ4nm-n}#Ge<~It9f8ANI(B)3kt0j?7I->C;XiEny&SOH-@Aio3pb
zaMMG7F#DYk(`5rwUCbUsGASkm^t>hvdD<ro>C4u=>q1MxE>)6i*Y)K!;B_xEt=S<k
z54jy03>R^IW}W#Az)+T;mmgKFQ1w`Mv>o9gW*TgJI?3Zu6o2mL`EK3|-MI%a_1-?*
zjXeQjFn-x9L0OdZilx}Yjp2YmT!cSYZfRw?ckP1s+2W33aV-l0Z)>?0eQHJnfsFvN
z@44g9%m^X)4h1R^I+|r8{%0;}(8}AQoE8v>fHSsE?#|)e{UWx~ksZ3<=Nt?ey?!p{
z*iE&`u|7bSHdO4rWAw#XgHLE$j54#Z`0(>Ntu{+vKDfli!{zydjfxqep87#aGsHg^
z^;u#w(ocxb_|wb0?k9Ybl9I%SCf<I985Pfwk&)3&jPNP0S=XNsWSu}qKwMc&Obk|1
zQquDK_c|X<zEsos^5gzQ&Q745N^~lBpAXUFt4<R1JiC1x(dBqhPkes$o8gZt)k&|U
zL=5i^DwyZ`2(QKP@jX}y7)xljD(}FR78YLPVvpTGzT`fP|E`bo{BX7I$#sL?#1832
z<8g)@`N?jV=Z5@&6=db!LhZ^_FbjQ$H+}jECaW$iK3KxQhPYU08NEJT*VL?5+x+t{
zQ^J`8F~1LB0~DBI`=U{Yo+-$G8g;^a3cbMV%v6(gt+8glqF2vj=m686Q&=5GhdEDU
za5th{#rwP0Rw|)#lRLi8*0mkDfW^FJE#W1WO7sjYER=$du4mj%&z#qa)cXJv=hQ4-
z(!T<d?C>AIRbYZx)%N=7g?sM*fB-B7I)4gJ|NYCABls7vJ4Sz5>(x@2wwclIfa-<M
zQ)<OW154@GZpXNkpQfHiM(LXBd<_j`qz}UVh$tYql$-6p9-HEbfD-GfUyGESHW(T$
zFf*g2HC|o^RKB?}93CimxOd*(2kwI2fU6Cx5&JWu{);VE(XKAOBAG=KLvfF294}9f
z(fsL)=C>SVC?2Q%cXylaV8kH41A_t#E`$Apv%mYvz|ISKWH&H>y1uq1#)_$ozH5Z6
zn~^On?xNP8ofQ!maU^2aa0Y3O&o6||CDXiNrtss2(|w_4j~(4s@0AR)DlN}6+=7AD
z_|PKnFR&H|N2c^gt49OY`*LqntJ6fg6)b<jm)1o%d9^O9E&ZGbuljD472QI?RzI+9
z<)J**EG?}_Z+BWl+C9*n?NCYjuV3&}PxSAe2GErsg>Kzqr<?cOl6xlFeWE{N2>S#K
zhs&V{2N7MEXqZa`p}(AFh)E^1E6%&3@9CAVKKn8#qrakU^0UI@g&k2(>*$};m^ZCg
z+MrE**t_`k%#G+V7FGvn6$3uUQrVnGYzzle&A=Tg<-$clD<q4Y^u_b=*yPaqMmMic
z#Lf}uZ<2LX2QXg7Fmv_&o8@J&p@omUxCPXn21L7TDxbKwGg0{qX{fQ}P*%$xgetzu
zuQ8S2b5vL0KJp1Z<dzFPcx$dr`5$20w*(X(e<9f3R_3_!0z<t)Fn4=bYm@95YGNGg
zN^PoF>~!%!bKO0qb4wc=8D;3ve0LWpQm9^0v0}oM>xFHOxcO?dW(h|JuKN<y0R_0l
zB^9gAr_1*=Dl1DRCWdMJ@IPGe50Uurw19V~qLj|-l^5<3d)4#$SF!()Qjvk3#Ia9t
zTM97DU(WPm0?LHHNCRGL88x`;6O9@e;6B;&;(jowj!?8AX$f_&4pi|0(!!h=P+4)G
z-l+I=N+(wZA8~U$OBRV$mr8D72`-`fU#^YMS;S@nn+x^N6PyGyGnOrkTfwkz%w!s6
z$EAbvI_PnOzTUe#!AIFx&o&M^W(u6{2tV5GTz^-hdlZM_=(N1G{4;Y@0la9iZw^w-
zwu;hxgraR*O&46t#dof}K32s%ZeRd5Kf*y9xo_L-wz!cK(=tzDlGi;sLsA(nv$8Az
z`G=4_-;4HpUTJUQ@dF>F?H?%`sxJ~oTgA%?Kb)mUv0L<!6-)?t&q3qqsSsBAL-IMA
zP*qe^;xLQG@Sp437K0oO4~uN$retVq)2{swyYy~C$M;!oCn=vbCzlZ=(cy2VMMd3z
z<y;n{**tN`61w}xOAPPweza}W)t>BCT+DMaH_2Z;TP-4;LKm1|vCNBfn9qW*T)rJp
zOySf@$_lOzrxn5Y(f1CW+YSy+Y1UgPQqoMlN`tG!JFDDXQht(sdSZL`9coo1e42CO
zc@Oy{JFL1#FYa<7a-qjO4u`w5*<o)i4;*clXQdasORW5a5^-)-OS8;=nm2FLQIUOi
z)<%|6^wx-)g_3uh)x0^nsPyq}sxfkrVpC5hr>5g4b|<GX;ukfK95l{cG{&leSqurr
zkb9Ih71&RDHY9>Cm#JgJp<DZbxdiPe?psT0zLZtQScQ|2j5I~TP`g5dS_r|&o!VtE
z=|<&UY(@*LN#yTolaWHLfx6-Ld$O)S&+3?H6o}Dl`Rv<ja(0L}hZF_$)TnlTp*9}1
zr<sVMDg@0Lj2I4Z7j`_s-;78Ox9c$sw4IH0ncq>`=FMp$cFTkGq_WB9Y`ws5fhQR#
zCfS22uV}Bg$^#xf8B_CDvmIDo?e3+2>wqsm?5W{mJ9j-0sj)cBs3&Y{l&r~&_4gd7
zO*S;m=kf~Ox3vR`dBnE2Bo0u|_v7lCe0`C)gVi@qRQo3yDICnfA;aCucSTiwaT{B1
z4@WK5{4RUWpsk9V^#fhuHyj#1stIM%ZM;{ZzOI*<Qfq4)m~Fo6iM%|oR^Ht0BaJfe
z(3=)v7dI3b3=LgQ4AhUguldE*(aF|JCsDENO7MN<k(L^2xkxshg`y}sSBJGaLdUc^
zf-B$tJq~b`+m-d1rjJE+l!b(XM_(2L;k2S-Qk5(ewmDW&I-Z?LZG1)q9+D(>)_tmC
z$?_~!a6XCisaGhIO_y?9m3u^1;X=k>3rqv=Mb_+3e!?Cr=jbm+PqEUYDWnG9jz30e
zC<-z!JQ-OSO}^Vpj;pqgysdL?MEu@lCkQ}&EUWZYywf^io;RVQxu90F(K5}VA<r+f
z7W#V_T2bnv&%`Z>u-$uVaqf7gyuqP*INoGhEx0aIEEIZ^+)Hq3*cIm4){vU8(QP%V
zYJ<C1IA*5F!iL04-&FC<3FU<lLZ<Ca%{r1@RTPv5s)R3d?C^v>ZYGCH4(X9tPEh1O
ztN1LCCO3>>x|b^fJVr7t{rN<0gfFBur$$n=kq&d|im(GmVgglQKTjOZ4GJldaJs-8
z<&D}nA#d^U_wGk_8YxxAQ#D~pFF92Qf(Q%mDSz$%y!!hoZc53Z%oJt1$75@uz@Fn}
z?*nJ6pN_Fvx!^ovpZn~et4p8#l>t{^MbmVw<t7rXTgS?e?68=0PaM(MfNDMBp5qaG
zw!7bw4Q^%syLP@F;wn5AQWkxbS8{yAGe!%x#Rejy-0VX>KEYRVrqi<T*ceT-!>@3>
z`FLZy=n;~@K_>hm1y7h+0jTzZRhY*ctPC_us6Z&Ffd8;Rv_39gU^dABA5NxE5ILfl
zGuRTq%iY8q1NiUF;X?;S?1Aj(b#rmIt8Z?`?i8B-8z#7ka7<*2SZ(APG>}j2Oh1mG
zsC1GJFt|1>t*bD}RlZ$l#G5S4_woOGYC3E&@U7$0>Z3=4Pq^#0j>b%nS9crnMpDgv
z)<pZd$2SFxO;n1LF4c#c0_4ZtDay;3)-sJ%p<4ZXbWw+jz%jy|ne{k>;IVh4f)=5|
zxu2#~OZQUN%<e?v&u-K3lUzRz%-Llp)taIUAN_q!utiU+Q>^DUci&sXGjPVuWSZ)A
zx8P0tZO+)vTyc&gubMgaob4@=id?QK3uns|l?7QxeNpv}hf467X{s{Q`1wDJpp`&V
zwoW&XBsZsJO*=w|dHvAkRJ2nxV_3C92NKLr1pZw?56sj>AqM0+`=OJ$dShU8H&z!^
zu{gCi&F=GswY<9upB-z6i0y<cKNs|oHy%V)-IYmCFqx)#*YC=}W7e%Z@xwdRhm=qq
z=8AqVzG|%8e+;H9&HKDj5a*AAXFZU<D1>^)x+=>Y5Hu>xX}`HttlYJFpgw4S1WjjN
zHL1RCd&hL<&rA89t0698oQ}HMY|L=sF2&&K$?*oQ7xvN@_5G?uI{)#`Hp_5Q?$gkz
z?(V+AqL_cLK1blKqb&c(`dhb)pEdy@-v$c`IiH^mWvkjP%EUs#N4M4-@(%Q>MJ~~<
z>HML>BgD=!tS8&D>9M9>aNam@*kSI%7&y47c6|q0NUQx=X_>AU8|!R26rLCI?{VZF
z9N!`eQWfTAewo}Hvodd5%12hXgjxjL@zZV!uh;d^kLaIlQU112cD4^0&mbDlIwlFv
z08TsM7kUR3KRBrwUOr}~-#JGuyxA~h4$aBtoNiZgOuE|{(>UOC*ec(H23JTdi+_6m
zzxx4pCY#km->9%P?=Fx`?pi#-tCmAc>B8(X6Y^{&PDt2vc@h73y^utq_?~dlILd;>
z9Jwgo?56b;diu{J+W>#1qi@dB4HQBY{huL9sx(bSp}&a*_JL5aRgKT0T8%EcXA>9O
zpl%L_>olaMsg9(mST2-sqT|n8H=C2(1zj|ruu$%ee&5)ILv+u1<*2RJuZBvYp<%47
z&h&sR-tShYZ`Y{P(y1cbzmM|drD7y@4Y!gAkG`U*kULshj!C+s6I^DH?0T3y0OfFG
zIh|fC1{Ry<a1b3e7eCJG{ByAg@6{R*U$l2oIN*h+(Y+gRGGDYq3jZ8Ol=Rpx0nU|t
zH*j+^QPtG4a`4<`?w-t<%)XM0g+<YOM#jPrSslKM3!KJ2GLYbp6z9a%wPa*uu)}DS
zV<C@F{iG;T=0_NLwjsW+4dK7pASXHVc$VB_*Bo4BLMP4i!$azQXaBmc9s8pt!tjiH
znmAuDZWBF7vFp$jbNKW1Qt(!_e!-#akx#**beceRhYQE?&=%I@?^*51FCZ|M!iSu%
zrw~KIHo5tz($AQRmsjiS*RN$$JJ@k=$luF_+rDQl9<~bM9Oxfn<51<hhLTJaQhQcu
zE$<OxOqnJ44JTtqUVp{#YZuif3+-HSqnq5ZN4ft%HQ|xc;kC1%l8nrE1%I;G#Veqt
zo~ae;yK;jE{)YXJcJy3-=T;rfAFwt@hlf9+9Zt*Djf`>#D%ZI)+LpdIdM4W;oj<rm
z$$J&wm*oLlm~*oTpr${9CWP?N<0VhgcB4UvBo&>u+7=`$7vUx31wJ>rJRE}t=WDMq
z_YD}|D0%tdYGDBTR&{=#Yu^U$nmAN)P`9`j&@^vSKJ&Rm#7m12z3s~N?}Ng{zN0}}
z(B>V0-`QYdl+Nr_=Fu;<V5jV^+*zlfp>K~X`sx^M%)41b%DU$18OJ1IU_lhkXWjx2
zN5lFzzNOHmEu!SLWQIK%Ew-V^G=#6O5Y|(14dib{kcyIAX{agwJx~xRYWT_S-&6c^
zao`{to^<10I9zT+=X}*pd3C&+)M)Z-Oy+FPRbH|4?|J5bZTzu9m??*5YW+ivY>0$i
zYOd+afl%TX^OYrZNtFim@Z)i>yY#`7l+A#@2fD#LIJ`Y^a`+jEoMjqks^#T6WHq>1
zHMfR`dkD4G57vp*+(%E6X%y_giVuVkrmJ0P$_p6(zm%|LGGI*jmE+aq5|H|ZWU(}z
zG_;B|ijjz+EWT#b)UX{$&(5x#+=PHA%e7kVA!GS*Uz6A6erIy!)gMZbri*U|NV%<h
zzT^}CyCyK#M=10%?VCQ{My)o|rn{`vC)cx1m(B}FZjE~%VpocHMrB9H=hBmQrpL9#
z{*4!-D9;-{FqgdaB}cE5O26tfZeJ@@t#J2p=lLQd3`r%*$!MqGF`MP^?93QXjx~;P
zS5>07m2kZO_ua1lfG-U8B*b{KF()tJd{)8nuM#e7^A(yVuRgJD`PfsFC!w9`*AWz}
zIm*(fDX;QZTKIRj%~s*7qT!xf6Vcr72A2g(<d$L?!!riIJqnCy5hIEgzle<uCne!z
zQ#sxL5{T6*=5uk;J<u(u(&k$>#im@!Ym>-9Ry{*2nukMZZiS~(xcCbd#oM===$WqF
zbC^f+aw27P-*2S4XssnAoM^%&q)&)HSRn<EF(lV9AnlYAK_RV5h^*McB)Ph_21K$%
z?Ca+?*-8*vS!j<RlZm#rww6tu<Hzv!3=Bk*qM)FBO1Q6d=*R^$OCv~8?%xLo0jdDE
zv#y`(9vDy-5&4E#6%3jm(a{v4og|`YPo7k=Jt>HakIx6{Q7HHK52llA)giiCbvn^J
zyJT92?MWTG`>?Sf=quH$_wVI69d<_nt5X=_Nu?V{PCWVwe8>nBS*}v!YcJ-2q{j^`
z=d2@4jbtG@xIpr>!O(+ro-u`Op78_KZJl=cl*!ytD$osA5{AhHqnEE^KZ0G)9u4po
z3kALq4PWbxO$VQ?mhec>V~oXezTj%NUOF5o`aI%3>ScUEaZd&xotQ`6*x0yiW@oxY
z$7{?AHhK@Jd5^p7HByK=FZtjwIqqu*Vj=BQ5cuAU=37EyMl9!JqnBC6aY}wLkLg!H
zKP+1+F`Ts0WK0ZbwSf=Su1rWExH;^h56|#@f}aOu<^FyXDFsD-O3DXcog5%4r__lg
z^O%6!SwT32AdD2804os|76x9nDE-#)Umt<B2$u>XBc3G{grY(SgB6rHpO`crbl|Us
z#B$i>0ZBSF0e{#ii!e9<a5z&u5*+EBE*!cUE>1VZJfWVxRotw}z`&r?i7bPxP8SUn
z*q!+FEE`Xhg7I04;;@i%XeCAzbam63F4pjO<{Q=6+1Ug9{bkQ~=a|7QGh56~ZhnJZ
z-v=Ut&5(Xv)~v;uQzvYnH_LbvEbqi@txcdta$H>A`MG0zXQ%1XPam{Li-pe@G5gQl
zQ$^xg%=6e$J*I!IT5tGMB*_3KuXIaZUOp1MgTDD%?rZ^Set=12Cd&`I^_C`sUC$$9
za;3yu)OBs*!c4+Y9=|Y}Z*W2bpfNp^ClAUmsuKfdNkv05y42=Z>3E=<PjWBkHOL^B
z1JJGf+74n3GDIUj<NDUwZ7F$m!BjRPq<Lh;!~#I_IT@9g&UL#wMSNT=qhSSjJGkVn
zo3n+9NM+63kHL7K2ys3TcH%X_u9NyY#9w>dakyXCto6kM6F8f5z=m*evcG@-2jFmE
z(UnF+{LpzDF`F}v2Px(2*L>l&E5WQlibMtf!H3{ab+);r#@#<CH)m_h?;#`W<N9t+
z6c?B3b><29VT5P=`XxxMcW)o70-^m3S1Hl{no(|QZ2jwr$6-uXuFqo}v1bbp?p=Q~
zjsu5)hG!@hsz!2TLCVFN;ThZ8+ju=FEf5ImFxYiBlVqWKEl8X<ClhK34?dVjCnV5-
z!<ut)-hwOqxyc4I`KE)rIh5{4YP`_omKh={CKgU=2PFPFtdaJ$T&w@_SWs(nt!=v8
zq|VIB%H#8TusN9ORaC#r0bV5QtD{1Q`Qb@-Z?6I$AM}+pS2q3QbfZi43mTOa^O-72
z8X5_WI@_VaRDm8Y)rTI?1u%Cq(dq?v&b^l~UPh~BK{7J3jXu_e(eeSF{K-j`sU3Uv
zL@BqEF=b%5@v7ApI~sXNu)GKz>i|cX`{vbN^Mq+}qte<=B8T0u*6ktA_z3d#@7tTJ
zll?XblbVA=6&UyU@y3Wap#zYuGx_O_%8ZZkp>WxjgW37mNOTye1{OG<Y<^03^g)u-
zZu1Ai_xGBviY8BVbad*D`Z*D14Xz?^7~#yVk6SE#e0?`Vq3d*KAawWO3EuB_=->*;
z$jMDEk2fw4I*Cs%w~DubeU(FFvPh%f5kj29T@e(_ZkM*fxD5ObvsQjT+u9<NlZ(Ok
zNHH4p4*v9cJO*SJ?Y74ImVW_<b-!^Im8-Uxd+UJ2i!gmH_lwM5C&1IhL`7xo?W=%k
zSn|Dx-THdscq~w<vTmt^Yu#LEZU*a9_>kCDW(x>#eFVd(6sq^G1&3r=h(d)Molk}v
z&*rxpY1!CRSZBX@SzB8lw>~5`>Q7*Ih9^1{1faF(f^$*B`-z4yTkdGIn$*Fi$w){9
zX*#W9Bcc=%5)v>jF%gkL-rECAAZ94VXxOjd_25ZC$m5id6;e}}On^aOoYUh)je7e}
zz_x(4+qZA~nnWySD!F<O&~aaha(r3zP^VK4{&P>JJ|i=elABut!_N!rEt30NdU|Q_
zecwqX<5|FQ@|E)1+9~L5GC=3uqUpk5&Sroc<dKmCb{iV0$1TJ%`}aP)43J8*J^UjY
z7y-+ngu4fq>9qM~hM+!ve56Uo@fsg?R|dIWSc?t=Tl=JiwJ-1xgWx+Uv-z}wtgLKi
z$i_@{aYi}(W^q&)mWL0b)ejwD?%}_b^I*Kso<Gk3%>uD-sbn6FO<Vv@&A4E9%l7q%
z2qHS1!skUJh1`NBw6Ms4p5DtKRPDr-gx7Vyr5U)`2g>*F-&a_zd=!<dhwOL23scCk
z3$s5dEP?|~$MVTU!MZ7ZfMqknuKTLYXB1Raa==&RS*>({(60S8?Z_r0D=X^a;Sz}L
zBH=(FRMz4tI!y5^c+y(Ht46^JR6uv`t8lnW@c>v65VwwhlZ1+d7d0G;0Y%JTc`-1W
z!D205A3s0Hv@ec1j@2^B+Qx<~kz3VpEKf0x*))n2n-dAF&%p&eSv(vBcw?()FeU$x
z5T%|dS~5|f=;(dAJ4X$`!p1h--?HiABgkw~zf4g);&ftY9w4`1Ci1wcu5^adJ4n>q
z?;I}&(jQ`*wO|tRxT?r!Nl;c1wAJp`ZPD>SCE&R-GRm)X1V@m9&`C{7`fFJbR$zU3
zY|MQ;C;;w<#zvyq{f0@u`*{e^B-;T=?7T222pwTu_^cMAO>S2JA7q12`&Uu*owi3&
z%`)125Ry9$E32}26wQ#sf?GC-kYLs25SonpOKoEfPB!@~NF^<4f<l2zrsFgxJ2T3m
zytg$VNvrg(S{t8ij;U6fM9Rs@{juf1#==UMO61t+z_(NY*{2}nQHwONu7AOrN2u<T
zXq_b_?7x5r`kP{@))!m7fbI<VHQnS^KW24357hn|nnNopnDSQ;?_>7#1fs=1OyT{B
z&Ij!{*dPmld`?A5DhN`j60kB5nlPz(cr+0f3E+Tmp;eY~_o{W*s<onHNI?OOB_+Gl
z;U6jq_*HmBMB^1XJNre$%D@fecD{1-6nw!gWy>b`Uy6NqclYq`Fp^5Msj_F)`kz0)
zB_R<o9WNLH4Nel#I9{l!Bj~042*Z|lnkT4x)Dkl6DXGi?=@7{Mt^ePP1f^UGLVb3k
zNJB~ZFM#;r_4fz+w^w^EJ#!#&HC}G!<1y%Z$3aZSX&xT1*eptrFGz$#jIK_1R2v*i
zJ-xhWrT(g1+ybr(-1$COOh<YOGO}FxY^in&R&;@-r6n{>Oyzu~LfanN>QP%ka0M_A
zwl0v(&Vzpf{*Ud;hwxDlsh_P{=(B-%e|^wdw^KC(q!aagyuEWkKyluOS|0+iqOx11
zR%LdO^o5$1Rx-T8bb?MY_PtQFmU}HiVzu7Evhn70rn{?)-7>1TR{|kF32bs`Oj)aN
zeccH0Ss=i~v00@(d-lxLVs0XDz8Szc=gW=ku`Ee^#LvZ}>6|w*qisPnf84DXmjzzP
zWHS0zNEi~%cfYyBV=;?aURi0}db04P?Fp~~JZ6(04)tI?0Km-wrZ~UFVY|NAr+j>L
zJR+UUDxgn{SZ|pbtNTEA`E7^6)!qW+<>{GcHs2nIV+v$*K~QZkl9rNEvbHXpciH29
zLdc;20x=pYDrM<SEyXOr&<L;Hh3fa3boE%wCR3i^v#bM!&#~!B(?D!*I6|=&gkiBp
z^MNGp>2kwRL=Mwzblw11PDxNuP>9GnIIsK0G%hQO=2O~h0pI8N<!)E@<_k?4092K+
zm``V%?#x8TQh@5Zdd@x`yh?8S5+GsYp){e*@xu7=A`R;39*8WuClf5P?n*$}^8-gN
za9!<1oC~5GR9Am_Yps^2K$blP7Zk)mf>i9UI<eP7OeN=jcAGV$^}$rK?#=~-<naRx
z0+qDiSB1cLs|9Oaia;94QLDBPla?-)nSX=NVkUEUdt-VN)}Yu3@m}w@J!@aD1;S;}
z<=LKA+7DXYxGY~xV$HN!?g@m>rtP}S>gsA_d}i8z*TIL-5E63)D7+yWd~P8{va4J?
zt0j-+SCu&6rIHD3vf-nES#UdF^5bwlF99%@izpNX$;cl+GLVoTh=h+9s$~N(`GXTa
znkx@Pl9^pj%@QFH^*|!MxVQkKzYz|is9-r<c5AbH*?lnx^Psz{1@Pd%gO6zx{$`-B
zZ`|hlr0#Uulq`|<jdFpC70_<YQs}&OTZ=V({`6@o0Kj{<cD3r$b$0o>dU|A{2rLSK
zklN-cDk`cffLIyj6dRmpX9~bYfwHme1f*^gtW=<q3SdIy<n6FHeWY<5u-oK@i5p;P
z?<LD82f)?l?=T>oi>D26P;c7-3^oy?Iv4?dWCJu%*My1PK4)cSCORacVv)y$_x}DZ
za)h|W5S#kBHIUdOL=9%D%*AJue*93N=>%3Go|v7I(y|RusRK0I>h+NRn(vNWI+@2!
zLUgULdR}!i=H}|mWzP`23%f2EJV{5m<9ezeAPhDqn=$;_z&E_LO9I~E*vWmq=#?2l
zIc>SRgHt_Y-qGIf)EQY7>aaVz(dY((r9My9rXxBcFcdF7(gZG($GYGu4`j0On2cmf
zw??3m5oU*7br3|QfDbFGt0#aEE|%B^Li+A(ZFq*VmRn`7on>aZxB%DLtW9`-5x&~I
z|L|@tJn6mkt&;68qp;%U+qzz-ZoTp;kbW>l!-4%MrLyO#+3hTVShWe(l=h@5CI<Ip
zV?;Q-0xa~iYWSZ&j-lOzj0Qmv2l#e_q)+7rk+G>|R7d|BJwL+Ox@}h@l}#=VS3yLU
z<LCF(bURT5PAUOl**rtOa<P(`nOQEGk+XB1L*v9lBF7Kl2f)7%ZttU@0Fhx43_k(z
z%gPG0y+Vn3H4VTXyv}4+hp@hQ)@b%!aEOlc-OU-q?0WtM#Co~cuV2s8fY&?B*+plR
zZ&ki{{``kavQkdb&DD(DT{VCNYmIjc2+KY@1lqIRi4-g>N}$GE4av>v=0Y6W-V%Ku
zQ5wQV-9cmqXfpZoWGkEZW&>D}JS8Qi^HCq`2?#}^;lAiE(?Hn?((7<NG*`PC+p`+p
zvs$-!MN=%(xLwxMJvVMpo(mQVAk(4wMi+6yBLD}SLya2$p^FyB@LTchobcY9Zhrdo
z>8b@j0RfFDsQ<|<IIGz#Ky0Z!!u3r>$nye(uo{y8?Z1orf(Vs0M!q_4!G$Bh`0Z(R
zvna%&_QMSre0ztYvXcSs$4szYe#s9<ad%uvX?LM4ChL>lnSMJ6L9h?z0*I$Vr4P{l
zd=CgA0;My$y1EU~%~!jeu~|;7yrre3`txvL&pmyA4jd~C2Z%)F0QwQ1RjR16`0}NM
z=X?=)XQ5f^)2C11f<dI4M}G|A&^4sEX-9(FiF`^(xYOv`v=BkMGg+Fh`d7*G_HW)-
z5Yy2~Ju3rhEB0#T27L|Dycd5WsOkq02xNAq+HzsHQN3c^Bn;puYHDina3;%nt~s}}
zIg_C@A5a$Pmrn%{UVI`Hf2fd&_@bc<QLMeyiz9e6CuF}j-?*Bzy`wH<`s8m84?(0`
zg>G^B#y^d-Kx<sL8s*_b(?7q%!ZSdntOX3ODEuk`<PQ6tX~j)*AjyqTjX!IKT*?Y<
zCUR+2{VJLKFbIbcHGfyeSbp89rM|bHUp@#$;TfSkmm<7(*EYE;AUN(e-`zsiaE`P4
zX5fc~b>9)IO!(LK#n#92qcvT2m<u%<RW~*^08T$DDm019Dl04d0mucAZ;XIY{rt|y
zAY579B4tgdb{BS_jSl!EM8*SjYvV-yoCN|?`uqEXV=$*tu=5c`A_&{d-1kOz?6*LT
zrjX17x&8R@gH#lc?TfZQmYark7ukFdB84awYZlJVs_*G>S}l7U3?vnTyVV#K?dcwF
za&uuPz+fi`v6ePdk{zprs8?2_wN5p6K~7s6NZz~QcH$yJkQg9)4hLXU15cr%>UMDg
z@twoz@bk%igs-^bg6aVFR%ZpAt>wI3+W8#dIdaLkw*bu6yMFC6kQWs6c0S$CB@>Tw
ztgbzqcd={7R{jx_5Ax^bq8Dbv-Oad~WiCK4Xc!n#I=JuA1wbJc5EP^UlJ&`-2<c6L
zSy$KMxtz+-(9mo_B}T=?HOnas0%E1{2)QWGotfahyBK&j191>w?1iOMNF~UC*arxT
z#!y;N-$sy2$^nxjmrha-n^OiMCsmzp0wuD2ui+@98l7iBI_&@J>HI>P3d1n233;V8
zD2D?@25BtKwv4Rh+H5XX>C|NEmM;Hvv1y5|IdrRuktHU9GE>QcBbQS*200xKNyuF+
z#iVqL)Xl;oHcTtjAeTK~LU#wwAI>@7_dU<^{x;4FCDyB!R9^W+V|e@1rlvh@acs|T
z4cL`pVbmhFUNJiy=}5&F+bf^zwNhG4j0utWpE{br1!9{&E=7vYhIF^=oM^m|2`UN5
zjTbaO+({>tU#3|#SXNFu8p3A@1Y&n6+i<3s0Hf|P?@*x#`mF$yF^|0jg7ST`r2)tg
z&$|P^EIb`_F&KB>Y$>hCVR}H1&Z8xg^7@BL)z?`b*5l9k*mw#hFW^u<+MlY+AKABf
zLp{GcDn$U3n1ibwy}gz=cHLnnGY?hoJ77C|ji1wdN-2OsS|s*-yhu{QN5!)EQM9!5
zwI>mg(FFoQGkI=ZC@7OS>nt!j*jxgcQvs}JqJ}^SLseeu=Na5peEkY#BdGI=;BRwO
znYIr}br3sgaD{KqHO@R%+VtwV%*>H%Q!yz;HZD01nV(d55XGv=2NV)H>2&&tdd=7B
zQ;W^7L(+o?&X98!;W0$y$PHm|HiBLeBzL}xSy4H*FrGBo+SZn$GL(+%8srdw9?tC-
zovAWwhGw(C{g!mP<(8CnW!=xl)?7Ja&Zc2<6*aHIGu-WJ$t^~OmXDx*NZ3kE^#H6?
z*V4%tZrS)mLKGkeeTkR4mP5F-MoJ|Ue!c7f5z&F0N5sRk{+&PS&^G=hh2AnS2a}w^
zW=j!J0(0gLM9~BxjiYrKjmG30;@rOyJ<IqNlB=JZneki}W%NHYGz>1*);UX1+Yj^=
z;1KT8D-sjKY<!@F(Hg%M<*y_TCxxvS1FyjS87HgXAG^Nj$dhf4mU^+{gAiY}*M3jG
z2dvePX7YjQ?&E|xqLd;Ghm0(b#}l6?7t!{V^_b+U-LBbqPFF?C-p5RLxcEQVPNoKS
YB3v1Ym4#fhA7_LKvvT;>%&Sd*0TPMl*Z=?k

delta 91989
zcmXt91yCGavn3E9cnsGF5Zv8@2X}|Xg1ZxJ2pT*I?iSqL-GaNr0*kx5y&>QKsG`^{
zmSLvv>F#q*-+R!39NK|gYV8Ij`+YnkAv-H3BSsWemM{4cjQYc^1Q8pJI(I7v9`%Qh
zk7O*=oib_!A^qdKrc37d_`?VU1;3AxNYAEjdHFF1_;(AHD5R(MG_*W(-n<lg@lIqC
z9|Ng5{_O2$`y~uU+Mae9F;;pjm4%bXY8=D5&G+kjB<AYY<PXDJ4(8KHfC(?GTk%rg
z+FJBJI~{IM_lo%@;|_K6oAo@pVrE3TVnO`Xal2v<2JqmiOpU!_!T*lu+Q9!Z(&PH&
zD;}@b_P&63nNV7%1z%yY)wLbOmD!(pnO@VI_(vCBjRfcQMs~QZbG?DGna1P0(b1-K
zmU>#Ka33I7rS0FKgT<tWK$kC?f?1{G^{_`hUrXr(jNaQ_9{J|pg>sO@Ei&Uyx;SOQ
zyQ&BQfrJ!j7^7ENRn;*inG4xl3rSMH)bOw2*mBx?w4xQu{GLACsIr58K5z9{Z6s+o
zErfA68941L$#4MAS`%23*h+>~&N3c%GDoHuSqyFV>mSMQ5T39Brc%ykl#-w)GX6Ew
zxVYuON<SPg`wbtfX2p)I!i#r)FJa+7;Y+~4+F{L98n6dC7qt)3{opULVy+9!ZE|Py
zq_FOuYYQRX^)fo#=rRgV<+VUFuL_vR&D=}4WDS75k8}v-Dhu!hnvrtx?k09k2c#I7
zCsEA%T3@~S)He)#?RWA0J1oukO+X_Ecjn^$CAu52T=N-9_z+1}I9p9=d)Fc^<LikB
z3GT4F(Y5$QCr!-kTQtC1(qA*PY2C-}l9}`oiJYaGN8mKFS2tcQkF64)Jxq*G_gN?>
zXI(UHno|zn<qe0rx~7BW<X3u<tHHJXP$a%vfKckLZvP5EWKVxwT{UNv8aK)&KVDoj
zi56(YM%R(v(<)GD>vu}8A;ix^8MtzF93<CxMjaizxpaq96wjqE^&mAdicMi-F|s>z
zdJ24LcKn&vZMAiO6h|<i{4Z`7zyFI{EY08H^@!O%%RXyE9DRMQvgnnfWRI=8R9!fA
zR={G64%et*sIM;ut9c2I8e6SBrgWtM@|9dP247cK`^(iFmF2xPqnK|B(%}5*S*_Xn
z2AlaEt7fJ9&(K<y!iPU}D6e?>psl7Kced<R<xt2OSW-yt#hLU(W?6)X{SV1c2J|hH
zJhnN_o%1T3rB<gs;KJO)I?@=7j~60nDuMk)>n<OL5p3+26Z#(o_fzwaF47}_uuK<2
znBjY_Mt3KSuj+PAJ_*bZnTWZ{)Xsy%5;NsBpXdkZ91<5Q@z?<#@6(1&gU#jk2HleO
z_OTmwBkk%e0YK6V!j69CEsL!kC5;$bf|i0qG~e<oQh(m!V^1)u^+HTp<c1zA0WiaE
z@u^46^9MvteuI6MDt4B$i6a)1zP)!cq^3pt`&1!}j??ccugz(wj!;27y3~~}7W96R
zNs}#jS38MOq>P^0jRl)cBRppMmBiUtG*C7uOuD>ZqMD&tAkkPFxE@JRwz+8%1cK0K
z?_;PKLVY{Ym_a*N*#S#T%RJznNkgH}N?`Wt1}=4Y-b7KBAS}iu`%!c{Folfco_e=s
zt1Y|a_GO*TvpTa;yff~zF8{BTx*rWTBMPqzKg=6rW_@DtZIf#nq9Glisk%IHn#Rqv
zU^?Qf+=!;>tHNVMDXL?epB~7`ErnUp9Ec(LrVP?-rYYv3;NK}^8RP<!nnD4eM2}_q
zy$XlU^8W}Ar#GHYO41fta@BcJ_e1k~j#Ws1+moZ~c;VoMy<PkB(%tRNE#MwxbU5+h
zVgkwIj_~Mu`1a_Cs!6nKU%8ORU|09FRT@Jsfj#q)!q^-y&$7Yn+kTxG-%SMUT<)=T
zVd1x70Y$~$;*DQPvXD7&#^i`Nix-B3S{_;JW>4(C8DV_gfbr$W%y*z^!-L;usY>G(
zjK1H?OH7TgGT@8aeqw0WfET*?<JWcHT;=Cg?!WzhBojfa@)su8g&QypE_P_7-I@u5
z?9^&YFLCMj?$<Rxv594_1RQI!IHZs~A|g;<O*%QC(NKUl54<D3Ew`QeXdq|v{^BhA
zcha@mp1!aJ92^;(govQ>uwF|m1uvg9brH`c14ngR8l~m1=7c$&(>$+gdgDXy*cA35
zMN<%e_3y~Qn)u4Y2A#F-j<4C|D;g~{)y2=E!&m1U1&m{yL10At+%0I1*RWUal4UdS
z_LVHP3_e)(1Dw_7%sr|yNY>3Qg9CK<cV~}yH>obA)nzTh_)B|dWjcL(i}K>O#X=>s
zLqkXBM&Wxp5zXpq7B^GQ2Yf`2%f@#%_gdi7CxAGzbL{%KykYTG->73b2fDGM@L0rl
z8TYR#IsJbzXs};-4wrJ?ph^{P;%f~Dh=``Jnl%;#k_r^oi>$?W-0|K!y&n!pr=e~&
zWd(5}U%<xMa67X33m5NZZ$5fRDob&+Xi3S@V_6}5f2nmvr=ZUMUD8>U>x4OmY%%0H
zo3*g%v>+?izQ1M>6P2~5J<)7@UhP#)_d>lnpB~??%ZHm8$&B}MP+Ual?qw1nj(x}#
zHU)o~hS`|{6%msD`0QLuS=?>1y+D@lfyau9DR#)#wg**~2r4FNfL=F?&spaRTY2;4
zOHOm;F8`ka{$ZEC=+l)xUt<+oZY{I4W{hi-@-MPKy2m+kcY0Ocw6u0NvUp;v4Fx*b
z_DISS&kCB`(_V9FaMrsnZ{tb>yR!E$_Sx~=#ix0ai~ofx>=(QZ`TFK&<gZ_LIpevN
zKJfiK$wEb$<J+I;>6^#Re(?Ogiir2-PR2|wKI8ubgSy;%efspNp~leaXUywBO~Zh!
zBCTkcI+q*G@$Xe?$A#wagRY0SGDhKf#iuikI;06TG@j0O3coNJ_p5+JJX{R#ut>S+
z)ADBLgWs<W4uajY^MAiFfGc$_Y<7r!>4G@Sjge1OQeXECnRz*A8vm$s7cWdKRP2NK
zd6>?S`6`>7L}Na$ZBYQEg1)bCv9tCQkxqfvU_qiuG7>t;xd2kcjc%S{{w=}%DW9-a
z;{)u1nIP<>q0^J>L*^bZUEg+e$$@-wa`Jb$gwHAw6%EbiuiOi}!>1_ZDO0l&j2pk0
z6nlK=TJv(g$EatXmmccBAJAi5I6_NYz`EpUf?!SDd8W}&r<Ian5%T2N-*&}_WsVxK
zZj}rs72UPSh<VycG~TW}Oq)m72qJk-0`E8of~5>pFBY~#oPf1;V|~L-(-Yb?p7opQ
zvl3Dl6>Wh>&7yBbEr-RAJn4#g)c0;pI;;s6HmB#^<;V#_*LFXq1`4NiKO`6b^l2or
z+)t99)GMBT41@dDLiXvpj_k6%si<-3@$ueW#HyauT!~)jN5VTqsAPF%W#`b)fozU~
z*+#joElTF-7canzU$dxsE!NWxpa#FLzs?T~>=i$}6j#>K*MA4MF?oleq^8_jjM3s?
z+9Nqcv?MI6N|1Jzdwy*5d3Wf38*1~>>wB0?`1an8jIrkHZ}@gIoVP06H4Cgps@WBZ
zE>)y1R%-_;$leQ;Thu-EUI*HIchdYWV!q!y8q_L)U4R)xwxpZnUDa$l!9wF5%J$XE
zWOo5skZ0v5={h-_D&yqk$2i0mC-Jx(a#uYx)6mmr=jOg}a5&Y}U3NNcyr>|{E8S`P
z9gTO8QgukMecrV<U+#x6uV#C*CCfbi8!oGLf16MQg!$IorGyVbRq@n4&fB+9Zyn6%
zxn9)(sdw~tKxqZdZB<HatT+Thk~V=xOG|4lu1UevfX1KpMo2mA4(%h@0v9M6*l7*g
zY92qZNnGz#f;ca-fYWraIAMkb1@lTMuSuv5q)J{sWV#yP;m+_xFFB_}#^vc#1!Nay
z=_Yp1EVE8Scq;I{m%FWKh*zy)Td<&lK!B90Eaz{3XwO_*!C09=Npgdmj5LpnN%+3;
z0q$z8#{*{?EvnJ_R#;v12Y)WAkKGW1``__}y3CO{`y^}E94%dxY5G~G1=i;Ye(*0{
zjHRMsTd3s{^|hoHTF8y}J1XMlu!P;|ADQt-Q8w!cMmuE8C2Qx3iA1v%G2g~1Ujw+F
zkZZIs)P;dIO<#UtaZBarWeTqfHP6`g-TB)cN&uzZaKWeclfnJvyBhc+SKrN(>9pA`
z%qgS=xwYgPj^F|>M6H9wbl~A2C?(o8rC(o)vuT^Ax+H#hJCo;|$qn_LZ(=mQ09g0g
z=~4&%`2>_oR$1hO;{8F=j0W6#DZrdawO95J#Xxpfy@*uib&UT0@<Hg-D_ZLXJ<s*)
zpiLJ8i$m!Soa8>@7VwuFUR)%;f)N&<2@}pze%OpRbI_i}#Nx}Hm9=nI=@YCVbDVDE
z%2UW{h$PY@P)8d&ouXO#1iI+F&W!kl<3ME65)&Qlusw2!7aa0p8@OWBWQj-ANZw&X
z=^CB&4IP)TWh3<{qQ*6KK*bBI+-FV`(;`v!8#P@U<{~}W+{21)F}23Rau@pHTmEUm
zDagN+5+kP5L2urEh-7I{OSh15+ll&#FxheZ`o}*Ne8E!Xklozgy)Iy&zbkaf)UvvG
z)ut6*b(t#{0z4d!<@WUNB?~0bh{rIQNxL-Poimc~f6qF|Gz#Ruy5b<aXt&PjP*cb;
z>O!mhb&K=U>-;0Q*kfrLfv7e>&CcTtjiowwNaSvgVoSe+`dpnFt@r2`kRvi~g={}w
z6#$ofbY0y>0*vl+G`VcXjznC(NIE?W<zzPd%c32?0fS=8>Ax;aIpx`HQi7#_5-P)L
zF<)W8H3b(f+0{x(!Fbt!OawWhXwsekA*oUZOQjbGSuKcyUj!wD`TN1!1j(Xoki)BU
zp6bLAVvw?W2YX#as-Od%VDZ{|ys9PWtzdqL<BRWd;D?vOXngfTN88%dk_3lw0__HH
zZ^csks~y{k%pmZ0^YlWY#%rBF>SfImQ(&D7x4->*;$}J<OUIrjU~gwb(7wf!?DJA6
zD&N@yo#81ZGpmap;=}rZ&at7e?}sF`RkgQr8&lzqfE^P?L(3PiR;4!a`x)pc<gKz1
z&}DYZAnFY_n)FHr7s`j5F|81(16Ry&auKPjyY3;Xl&RhUXCx=QqSx`3iCpNycH5iR
zszcVPZPBIVoZRk=TfIR>;e5ajnL~C{ux4h~rQbR0)`d`AMbV_DgQma<6>&;}e_NW#
z+#?A&8&$Os*Hjd80B?k5(Yu2IoOU%pJKZV_iH~QKR}dmbbP--WVKdbX1e7<nVBj&4
zn7EM6#A5cZ(p1}(z2fJ>(%^POcdcKrdA&fd-!T2DONRzS+6G-mit6AYGK-)ax&4c7
zFzU&*BTVnap4mpK`5#BD@qO#+|5E4&!SB@AL7zh?d|Ho`RG9-B^Ea(%NtuRMOE85J
z7KLAwvyJ6@_|oZ%NGyO}rz`5|n1(<VO>Zn#d}ZpD!~E=#kz8=d3EpvgAS=qYv3>Yu
zunFWgxc*%0x2rw}VatKApseIIA%0krwm>vOON4MH+iJ^rjDeM*Zny0R0lCy45MAl5
zqmpCui2@-m_PqxKiMGe#*+`m7D;bT4{!Kh|H)U!zko0Z*EeC3t2+r26bNYu3c0&U}
zk$?l<sHzzCD@v?|fY3vM7*+N=n>psn4V)vxx3s~ChS&4IbkHM*+Ky>=p0gR9B`!_5
z7H5y(?#5l7S79kaU-;+|j&rD{pS?fWl9m)=1@13Y!qz_hOGwK9Rj_x7Y6I_R_~Khf
zUbzBOhng$<lLRbtd=D6KQkyburi@Ws5Ai6vV}`BcYewA_2dA<`>Q|_s`lumFjDFjj
zqMKs4s)i*WmninQQgX885J{c1+SEGM7vR#O5eC?@yngRo3Tx_4TD&2y_t{JACI~}<
z4O^M)Ptg0(xuZEaNI@RYTnMW6v^V%=iGkU+x;MABXVyX;2Mj9I!#@La^sLugV^mJf
zyJ(>*XgLTY_RH>Z8nX@t_nU<&ulpt=cNy`5Z3uL=qk66v1<GLe6rFx0-B7-`8=Vy7
z(7U7AkQ%PrTk-qseb>_p0j8QtA6)~zb}FM4whlV#rfWC#<QKz%9i!5B@PsVei2$3r
z#{A+#r;Yw$40)%xt~bub$ZQ?#nR08$Lz>F*bo=((wM|v`@PF`f@i|+@ev*QJzwgt)
zncjU~cUP+iscj|%_cg=bM*Fs~kyo@%=IF@|vcQ{}G;A~z8Q9(he$pa8f=LSwryE;$
zdw0A~(%UgZx!<o`kl+Y;#DSFfBXTTsui`N-D4O<y=d}lv-&oEwe%}JucKOndNb>~g
zK#;i10pqiwu>?enR|5A^>1XnDZ(TjZZlKmXO{}Ot3;>HS9KpE<Vzj2x(3Vz3sU1b`
z>Zc*=`*2XHZnK;XVT&>(^N0z0VX_Z7nQ|Q=z|UctOfUTUBSx$boIalSw0GlQF3YO(
zxEYS@{*?)#IBq97tA~fX6DCO?A9onpMg1{z{$UhXPM;A^e)_j);{n@K>Nr8Yn!f=b
z6-?yXwM^RD&<W;!Lb$fjf6BkeaY$3YCd=CumOE;nujosP_i7++ykLnA<fDPgU2)5y
zZ7qwe9I(m~%?SafHMfQe5@3TBTA2USN(26!fYVWUw<){V0g9{X`gE1x_4!89AY5k$
zW=r?FCP@`)Yy`2;9I4b-cjxnJPh_Vv&K4BZHh7z>OWv|r{N1gTkd7&6p5*l+4Q-JX
z{^g1XQYhZAZi}brOiI;9kv+0^utO&5s)X*M`Vqil1lWQCAI;x{RON*4!L^ord!%09
znl2<YsqyORE!}siY;ABl=Os_wxc{8hK3eb!fsz;|IRk=x@Ewx@CDTfQ%}n<|J5I}#
zu-TaVdvLL&9-vvWG@yBB2QsjPs%<;1h*huG@-2wtxWtCj0iu=~*KbTB4%<ov?gN+r
zRrZU@gV|V6!3)~1vb%g5A4^7`$JE63<5Z=S&waBeyZUv;`9J3YmO_RId7>gnU;j|(
zvEr3FU<qaA&lQTeU%J8lp-6nF@<29m5erM~`>Q5E$#n6$7iDzqYnJ}uR&ZjDzyb7`
zO5ZjFXNJ*lnqT8bv%2_1*Z4=ETl7ilbuAJ;rcunysofQ<<VmQFn$nfpofIirUyw{6
z`134eykojxKDul|kzCrUw+d9M*|!ZoVd<6rK_<XLQq_%4gUmqGM3_0Mx$w3hnn{v3
z+Vi8?RT8z4O}c1mq-p6({UxD<tSYzN=%$2RP|v^uyos8P$1b~&g)!ijGv1h$(e6>s
zqSV$Qq`_K5?LqtOi%g>?Mm)+tRD4tXFXui(ta0FsV^kj9D(a$Ne-NN7XrIqlrsnnZ
zJvw9a{LYa+Vmf$)WgU>C>j_t>Nw=|q&ESO}CU#<9QCe#EOvR{0t+>`pV0}~d4w5VT
zjVm55PGj-r0bsAOZyoH6-<^)2qFMCo+83!Rt8&Tk8h3W##)!mTTmOV*Nj(i)Lz<Ed
zy=zRowM6@OQ(^xPRdEw6xRZ+41w=FgQrp)%TQG%nb(e;{0p;)mo#AM-l|SS?>8cjC
z&ZiW)c7&QvTS`_Tg4oy_4kt(KpqP-71PsvYGlY_~`A}Y1i-Cf|&iRy`D@GDT8py>a
zM9>2aCU*K-?(3nCmpAt+ufPa$)w}S4R@2vPN_{KCh>!*bnfbFc{g7}|*6hu7&;eE+
zzW2P3<Ml)+Ad;G(NBqV`Bv2ZJW`Y@+LKFzm4UH_oeK=1TuU8V-Bi4}#2aI-?LnVv4
z<BAL14Dcn<kxef1dtN_PyIjQoe@1t4wL;6iUTMz2Vn$>Zw0X%M$uLxe+7gqM?jDhH
z(<%48#d5sNHrswC&|q=H@Q7J4<o`tVSykEE8L59$&S*b%Ng*y(Yo}I~m=u;5D?nZ<
zND~QjyGD#mc@4;Ep4n+5P%q81u1j_uAcC6gs%`L&{ZxlO1*PFUZ8p-*)_L=n2{hxd
zv@9-uX+dmd34VvoWGRLmEMH7qmx`}Hu4{(g400>e#pJRvkG<C5y)>*YObRK^J&G|@
zHfJMt!(8tzOI_B<@v38*G{;Ec|K8anA|&~$Ry94k6Zp1XwZBfNeU108!dLzHAAv|y
zEO?S)?i>soJ3&<wRy>W0cA#xL+VcQU_Q!B(NBX$9>{}O+SJDmhjREnDS$ZZ7YYhSi
z8YAAjx|iX`3La$1AT9g48L;HKzv_<-;oa80f@C*yMTJ(1OlUgnR1$^Cjq}gc0aKz+
zh20n|>=&<{ED?Jytd?h*F?jinnIpoyu%hetQqn++%6?<UIwPF6K=C^VXtR<-Hoexv
zC$X^^E&9+bbsON|kcW}3YTWQ*y{A9|oJsTdJ<SR)6Xn!v(z4{#9J#OkeRYlZle(_g
z(K=T<w_EBsgKnua<B22KUjaj7jk@@8ko)|KB?-xjrsIED$GsNWclkQ3+3a?Sk`?wN
z3#q>D^AChtEpM@`w<dDN`U|l`&2&<nrJjhHx{m>YB*Y>4NjMbml5ff`B5d^K2+)rU
zwSC0&YzW|rWL$~=%VLcGf$68UXQ}~c)cciO`*}}w$OCyaP#$Lc$C9r(+Szsm^tIov
zr$5J3&9}!#s&m@}E8-Dsqjfa4?%-^eMeuyZ>1Z!_TPVF;F&%gWo3GenITMY%a4?;w
z&c1iS1J~hu`Ug{%JqOFA-33;9M#tw?fq-b|5Ghyr9q|(Z_bb2!yTKiEh*GjU*5)NC
zbaM~5z-Qe)<&LS&2H5WF!Kr}=sjh!hBq5y;=_dzE1$}<fP|u$ZEAC~^NAJ&$OHfQt
zw+7K3*a9h<D1YO;8EO;ge)Fa*l0)48q&H$7C3h)f;;Q;$$o1;^PdxYd@x<(L`w?Vp
zeiHB10)gjI_@l6c8mCEKV-H?}%+VJJd{NUtVdm^00;#QkEP5#~Nr#PEJl{U{@7Ecw
zZVQ#nt1i2@e-xK>h>$%O+cI8<6!{x{7Quu%?Yx;Nb?U0Y{g<Zx`L)lBo{THW@$7!M
z9S#dQw>_%%X-|;WIVfhn(7MruOL41%%`fdc7v8E+9n6`ooQg@7ZK-xG?h;i3jO=jG
z0xV6oik#;5AYrZkMLAH&XSduw2gA89c1OCHqU+iZJ8&W>LIiz#!`AuRGP@I_o_bx4
z3y0b1o({^N$N`~ZYCi{?5gGJYq4k3zW*~Q-;Q?6To&by{wX4`e5GL0AxWzT(f7sO}
z;P?H|U!i9IC^4kraT)ZVeE3Uj(iMi){tn8T^mPh;&7zErAMZcBf8H8&o*GHuR4LSh
z@MM3h$rX!()T-}RK*iNvp&^OCJ~Cbj(wd6|#SBo}h1S=)Cub3)9DO$nYg6s-L)0x@
zFDXCtb(0Z_j=!c7W#q|+Nf={yxWqnlJz1w(A8Py;y;?x)Zv(+oPYnQv3lO(Waa!6U
zuIdxdp~D2jlj=G%Cd+GQIK<iBvu|sNPn^{^eu`gXyLaH>&PYCQc~Q?)-{z)UCrRG}
z4s_NJ`8{xS6a@2x!^A$Y@RP{sd(qP&dwyPptVVezwD#8EV})vkO)Vsd3N*?MWU&~9
z127;f!tN523`}aw2c~D4f;MAbj{aPGSNfb_6E&SEXYnNYmOgMC=={)i5Y}S2y4LTz
zI%TkLI_G0+UKMlkp_x2wF!Rqx5thgK#~7|=jYN`4%O%35VsE-7jQ1Z7DkW~8{Xo$j
zgkBFLFv7YbN~`l)&#|jy%Zh&-^50=0c>+w0ui^Np8HJnjh#PfDC#W|(n8MGZu9yP_
zz^T;?7iY(gi2B=tN^M3_RVI&@&rtj;504qan##r}{?rM7Xv(5Cg5VRuK+T;_LLq$&
z&@9qF^kLdde&j`*mB_NuW3^xZtEdW$#ej1@P*+Ee+IYF&zAh$@xj*5${szo4wi3^A
zs`7p&bur!i0$RL;F%R-PrNTUqFpy7kUbE+{w_egT5`wlxCD9c>Xq<u3Q`=Ann-Vm>
z85N|^PefN{>5QD^UB-abZ~6nneUs!LTxCddOJolmR`{9~W4N)*1n76{tHf~Jp-KMH
z!icv~`O1z2Y=meQ0A+QFj=c5gO_R;^DrhT1#Apk0DnA%ZTcQWR6nl10Dmcsctg1Z(
z@O+zpU^WFXoz2$J#AN0=Ro|Vz2`Y<?f;8OoQIn))@u6R&<&s5^pJqpKO)T<wKYErC
zu{cOwkt2a(R-y*-7egMkRUiLohl(><W)Z0f?{US){`MYt8Yu}a$oxTmnsF+0%w4+1
z^|G)t?E?|^UYFYRq&EnUQ^LnqH!rKLzncyg^$tsuZI5{s`_Yl#@1VfBVccLbG_q9;
z_&}Tf>Wx$wE|#RJwna8aW3l#OVH%ej?!zPTGp)o3gLV|w=K?{}sM+>3+Az1KzFv%R
z&PE4p0Cd!6W}+cKFR-k;E3c8S&#DAFQ)UN(FB6x+QJohZ5^QVorGo_}IG>SdAdm%T
zjv#}EOOyD*@#M~5gc}WqhVl}2PxhPGVVL^rRy;&=lQ>w2gVxmF(dff6iw>a_A}zNz
znIcsd2aV^EkiE9qRri^iWklKHRI$ZsAmV9MG8GO5M_cxanL#`iLH&WTptwW@X+8<*
zu!D_<QXY}$zWu#ZXD)RkB=zX~WYFGR>5omwBT`uQnXXcTpX2IyUBmTW*t<Xb;r{pd
zBeN(%$*uE3X1l+OeNVaid@MKZ;Y_DJU%cVJ!IrX^@7tsTL;gYmO#(FZ0C1p$kx;eY
z>moH=dX__FseXa*>vrQ!oy`MX&s5RU?_VVzV*Qw&2mKnxzf9Vr9!duH+v-9|Qr1zb
zpJAcVixPccqr$;;NhM67pTf4gYo(~%n7U%$*V<kLSL@gcni_HotY3YXadbN>WA1UE
z&Kp)kizo*hrN%}7rU)tU7_zkrEZP@>MD7FR?&rb1c(le*Jv}&Pf`iw@VyT<#EP7sv
z-wnp{Rkev#JR$_HO0_@7!KK1zUEAw=L#8~E!QNoe1f0gt()GAsO{`D$wv7BynU>t7
zG><4e`%0h0|Gao(<ZtoGccOu`I=(vbaW35E2#R+`OQdaK^T5%5YL5E#E^Z;@48xyC
zQD1zAuE!0$%Sc0kbm0>H%k^R9e+a$(uYoU6fP?W+uTw1&U`06mKx;VOut7pDJQMW(
z#nt4%f!YYzgy46oYD+?1mMKin;jk9MXnC+l{^Gnw3yP?-Rr=1s#ud}<!@J<ioyJs?
z4~!KrAj~`1J(hHergE4y`e(64@Sfs;3U705#mbL#PDoM+oojX*oTaZE=*0dE_cWWy
z$dvG3+=XRC-8Uds7fje`Z!V9*InW@(GKLIU31sR?N%5Eyz^=5h^?*^R%j;VZJN!E_
z`MYLj+tg5yo6J1Z*R|a%WR4O=>kMZgD5TDN)KnQ#qH<q0If4^z{2JG@c>$imJ1qHw
zTzGQ(hvCMqV}wLb>}uc2HFZ!;*g9Ucjx)qfCEqBN*lB6VKg}dxUb6lt(Q4w0n%?M;
zy7?4i@!91l<wY|qy@^IviEnaGt{8*9|G<_$aj#q}w@;@Xp#X(fcr*sA^+iG2Z?;%x
zzyJ1O|He-BI!>9okcju6xZ*$qUF=8>jP0YB?}$nA=N`IrKCIsCPUacwmKc?lS7(Cb
z+H%{3r|y5C^`A}>DKgME%H;jTsLZF{Oadnbj)4$#m*nj9t)pE)Vo2pAkp1OH)88x>
zaS5rgo<*QP6>O1q_UHMwN#=an?L^=v6P%bSY^bVk;RmjF3853uCkW_<tf95Fx6o;p
zaf?LIOzdSFONd|H+X$;jl=Cn7wQl9sG)3{PPNP#){C9jzXjs#m=7;mrDwq0kvnTck
z<EQwiTZJa5!3*{I*81clhcMJKZO~mr0@03TeGR>`b;>(ll95EDU7>Pf4hL5dRJ2nY
zH|X5dcQDQ;FdhNWT(eOe*tPmp{+k{KTbf(bekiJCjJ+FpzL&f(Vk(mw{|z>x%3c&V
zIO<9GcX+UVh0BD)X3iP^%VST^3hqqpSY1#b3qcssVoeGiaNk*WPe?V;a_kGhsbLAs
zITLoa_Vlc(<fKS2NgZ5YC3rSL&FVAY*-Mi8Q)B@9gHHa?07A4m!7$D7vS~&VH^;)>
z61Ati__6bTF!x_MDPT1pm53L`i+R>f{k(6yBhRwXiWn56&g0|GnDkeO>=ZoL?=^z5
z&xhz__Il6ZSlQEUKD7dmO#;EVNR#Au%Q#uJyZX<ph|&|$D(&HrSwLTNw&@s4?A%em
zS1vW~ZH-+&y(E7QyYj*3;vk*|M35|;2$_N!U)rmV1Ei_mO^LXXx={#W`)MI8j*dQL
zgky@Znc6$?ZUJa$yi&LPPu+8k?si8MHHH%?+T!(B4*=5=(2`AmN1CABAV!=-Hj)t4
zi8sZN(nixkz;qd6w-wySFn2R$<(HgElhpe?3D!qx`W({)ZlqQvwy!7e8EDKhUaZ^<
z#9vT985*V~3Xq&d?X)U{+IWZKy;olM4^|MOw|}cATNbvXKwtKSUFt9VC$`|1s{$_D
z6TJbzNlyTundGy4X7-PQWxv>yZ)3U9s&p>jb!u~W%<i+8U0!iS&&l*fcYM_4_e(S(
z-kGw|*qDLA-c#_82FFR6L?PypIQr^CUg~+S!mA)<Dl*>k;oPsFgBm9SEeG>>Xu6;;
zbgAA!fK`))K9DYZuynYr(xGI6DS43{!5bK{_U$Kje^`6Q@9D`HZzi#9^z6fZTI3sb
zGi#!_bl<REyvI<((_k)<6#jvRrNZv6#TrE})bV8Qa^K-4kGO?G^rI>tg=R^GX_&49
zpJIrfFmnL8UE%#H4_J9S$2eA`|9>yR{P-iN>N6*1!0kyR#R#+YJ0|Lh!1E3ouG<Se
z?vJ=C^?VZl{NX32V)U-gS(dS#kYIN4%8d;^_Eb-TXm(!l>Y&4<&h_F&=F*jb#=^hS
zQ{(@iWYt`b>sWXiRF4^0(hTvo5N?PH69y)R_^(Thw`Hiy=b4__Hz^L=B##co(J^l>
zglmB2Jv<N&38Stg^i7Ww24tV7?Mvx!9$YEKuWs@)``1dTwYH)2=VHn1Tl2{mDW)Wj
zy{yh^Rg2|pbY~;=nhFw370|lDBqIANf{XcBANHYt3G@rGTS^St)!w)mv&ec`&!G3O
z#A$3ZjTL`na%q2y%6ytu!UHl+<E*!pM1xT<U4WG1YKu2kj(Br1h}iXFxQ4N&I@dPE
zMX0PfE~EW@VPTQHoy*x9S#Hw)#%cb{PbDu&F8r7t(x}K?a2;I7#OuUJQ#SANe5Ktg
z5abv<-V}*$7jYAW7WG%NQHKKEf(&Z5Y$fWy{gJ@Q%;$pG`H6(oK>ZNs(t}%n$#-{Q
zdVo>)K~b*vXNxOcM&30{(o~a0*kP}Ioyc$y*3!`bK>Iu5-wTD^iQTEFv>_R?YPf8j
z=bngeY;OscGf#i2NEe9ntd5qR;b&A^V>)`}k%Ifqs=YXVfL`zTr6H^R)^N~A$c(SM
zC#et#&8@}ye6yzDLNAS8=69eZF$%J9Uz1XNB!M7bz}aj`y2(yT<KQ|Si>$Y9+7DXX
zcGf+#Z?MFEy#iu%9K_*|TD3Yx;-dL+Thlm7ucG56?Ji$2#JyB^v%5?s8r0J?G_{t4
z)fE&z+A8?lzs(-U#jbGOlg_HW{e#u`R{rTz+Jh_o-6O3rGe`ZFKuUe(-ijckl4MlI
zY?P5J!vL)qKlEr_^v4ionhpX1zx)O5)jau%B|0JhjSiQt5<$$<?c!V<sk|wjDm9tn
zrr+b#s0jK;WB;=!g(k$Pv9aVx!!k=~_GElrh6Q6#?)f57^<ie5kzp<@z)dqp#)1rY
z%Sh}EACM3U&J+@;_P?}fdI@HOC0$1#47N9QSXx=5e^nT(U?G)iq^+$y@Q}(*4P4nU
zUTMYvw2u>g&r%dnCoPt4iw4r!>799R?n83f2reGkV+?7Qdh|X$E_9_pJL`ve6rxfk
zdPY4>CwTr09Iy0W{C@kY;B3yb8528UBwwWsa6fNr3q8WVmEgry`X1(=*N(GKMn{bz
zM5Gs`+r?fj-?4V1SEK;bmY{ZUJ**+^s4*f-8umxYAxkg4^IUt|;r^<W1>2p$B;a8Q
zEiIpzC9_6WD?rf3d(<iI8zQdEX^l_NcR0)V>bIHA@Q615T889*8yer4lUq;%T4F<~
zKP-OY5)Jy9gwt5zKIm{V!;fC~sWm>n!Nj#cMnf^^Ogc8DlX}#=cA^eT5O2m8_C?nG
zo$L|R!f)OMs_vfzPH#gwhKPoSrI*3Lsb9$Qf!ho*c(5qNh5(ElZfm0h!_CKge@Kp+
z@nw}u{#KqmFAULoV=dd*&JMWDV-7^Bv}ny_fwT|w`K(&-Z<JsqI2^ayyj?eeG7jl|
zqIhROsb#`D0B%DAdAV$Z2}#;CVMJ8o5V#~%Slts$R;pVqT&B@p*<v4VH?4}VE;1~|
zRt1WG0bQs^({*tjnNYlQB+Q{D#SQr)fBcAHFYL$a^A$xBUR=xyAd`&xaii<?8?==O
zPCO?fOq1o;%WZ$J-;~)g*{Ql?NCp2`hoy`!n?3X8qrv?Xg2r0?;aZ63*Lox5siC~f
z&+m3S+BtV~-kWJlV6)(KJNtBeeB1|nzFup)A|#Xbv~l4_L#PrlM~c=Bd!m4Ddl|Vq
z<HKu&mQL?Le?<>~88qy^Xb_=}#95eKdf2#0<AruFoYin`OyTgd8`6RgWN;Z;C9;<w
z(^)(Vnrr&Iu!!z~-;JTJAoGr~9M6?|;VIw>uf-8)znZ$}s{Czay2;+39nOJ`cNwXr
z$KJG!-KU<F&cW3eLg#9|JzsdM_ij{Qau=~mEF@Wr5m3E200aFi*&YSlVF%|-BFlT#
zlFmvzQ@o(8$Uynio=Px2Z>APiv<_;`NSS~(FQ+#`C~xugK+q{j;Ek}V5I1Y3%1wu=
zs;)~Msc`3%?M<^l#sGfxlx#IG+asH^S4{%KDZ5f5D@jbZZv}p>#wXI}e{iS}A)m8U
zM-5CV0k@V@yqF7>zhw!cmVCv$Tl?7q>?P>feRVvMGT`NQoNORg*Z(}cJFT#{uEVJU
za+e2-K@kxpK}S!Wyf#98Mw4kdQlauu&L;e!F64B+J-Q@>@_{dMy6HO>hFjI2^HrU)
zOZ(ypFJ*IVV)4uL$x>OOYeU@4ng9nj;Eow|SI7<b_>yy%DiG-co!c9}dH0sFtSUw$
z7ZKA$YXz}m9=X$^m}JNh96*)=O>mkd)(*y#w4HwbC8h*TX>sFrF1QIzjPJyRqopn{
zz$-ADb+fci8pcK;tMHL%(Xm#NKRU9TslAS#oy*YF<|D}CHPv}QeraUS?)w?AD5uZ*
z75x)(uq$PloweB+Xm<M*EBvlG5_}E`PQpxD57#gdJv*0{f`5<bYZ2Ym=voMD&u6Nz
z?DJR8QMLKp%96f=fe=|25f^_$M1=lXT^(;Qu^&13vcR0-$k^EU_3PI{*4Ea&OI6m3
z2~$N{hWd6cg+FPZfH~I#u#H4QZ3M)?(vn&GS=`wk$-IgH1QY7u6kSAgCS9`bp7<v2
zZ8{C@dpdeVPkp8fi0llvvUPQSvpQCcc@~|ZN-fP{7ODfP^paTC`&tVnp9Xfs%wIlx
z9`;HVWu;Y}-Lse~<i>->lt!PhcnemVk^>QwhA4Qblu#QdKk)$Dnmpk<^<uFne`SYQ
zuMytzV2Rhu;YnYj_>S%d_e%72rg7-)Zg|H>f17m37n}{6KCp?PPT?_e7|DO|YI>q_
z<rOBw<&N8m>^9JWJf)n!!^1tJqrcM9$b^K1wkGo4?9bQa$R=~=noH<jCVXWj7~OVU
z$9j9&2u5cp4q(dsQ%i`4_ZQmaetcH=^^p2siPRd=Ojub{b6%R?Nt-e+J(V{;HXEeA
zuI|;7cl(*3B7FPIfjl6+JcK<@S$$4R?Rxj<Cc`46I$ouNL3Vmb$}auRUNJd1q&Rul
zyzx-7*d5Vqppx^dR+%qi{?FBoU4q*2gX%M5^I$Z9#svzl#2p<EPht;=dng7W`@)&F
zkPtiy3d-JML+oMmEvC0<2o95h$aTm*s$EI$1^i(u@St(xx;>VqxbQR%?!1>N5-U<L
z{7a9NywdHXRga1Jr=v{z3$@RJK`93U1Dfe>?wv`|wYx{cn8nP(#*i0H{+~lzO}4j2
zB>-_`vOBgG%EZsTKj=d?7@H^G9lmyFDg2BMEJ{QG8PszZC1bHfEwy{Mx9d1B>vrG%
z;*tAB5-APQ#EiL6opI+<s__)wxir^TY}wqE?9Nv0V5^^6otvt1hGgy-?{0kV>vCOR
zm{nbw&hp?}+buh7aSxsc-48-+Z0uJA1VCS%!{&h_byavvON+#=PLR8974B1(>z>qV
zR2WYZL!RU<tk}@ZMh4=B390~`i=TPEO9uK@@83S&%6#Kbdd^Nx)3w;ssGp0gp1Y#V
z4u=VKsn%N~@F<FI-`>kdq-JlQB?1$4=;1&`u8?5;qe=Pdyh<O;>oI{T8E~Wep6mOH
zNVgtY3fY3YWhP?e7vHF6+*HR^YA~83(UlR4yNOL2BhMqw2YP$Aef^0@+}F&FG>Z`8
zA7KTab#_v2^C?3C;sql&86$AG_DmJ!t!)kV_eF>7)@=}NE?ZeHmj~5oEXV!x9UR&s
zanR|d_PSaL{0%FPkH^>b0FDub1JK*%=Lw5cLq<(MGB8ARc8bU*aXxjE(#p!l4h=Lr
zcNO>V)6>(jf4>+hAcu)54VnG}sk`pyz<?wTT%>uWD53qe7e*HaSp+2ZfzGn>J*bv!
zKQ+*C9rSg%_oV%G4Wb_9tkdpDP1$SjiTOxce(KQ@7gD_#Rs_O3>R|pdpX2M2l%zlB
zbjQt4r08~LU?!XN(o{rcG*%&zocTG2FIyn86J&h`-r-(kE0ywu(J95+8!zey47+9h
z$qJ-mULCtS?K+GW{i{s9wE>@;^lpZLXn;%&i|N{yjJPA_Ey!w?_qspuvY#5=3HkJe
z>VTjbC>$6i270~s3`z^u<{uEY<xR<IT_Vl*opcf>4`+QkTC4(zsYXgrNqM#1o2CwD
zGZL8J2>rMnKieLcA|d6nU(4?GqmD+^($ccDvO+;cg{=gVHZM}4+V!XiD-{!hsZ_XT
z=j@Da;`D?ep#_Bc2*N*Ss|!ET-^kI3j;0hdwqOIH1qE-hnrNfmCzWI=Zn^W|ZB)w}
z-$90aW98cOELn{S=6FZ%!$~4EGqcw}B!9N?TPjwi|M4oz6{C`1s_y1@L57t3@X>B=
zNfgXl*F}8<6@ux_9^U6;<gA%BGT(tsG3h%ynEs2-jxzJgY$k_-AvYR3wOg)(UnRIf
zW?v7mc5tmBadc6D&6`EvS2_8q?Y8^G9HkM~<;~K%?w}b3nBiB(7=sO2HvapqxaGQR
z0Q5x*^>od8W->MID*fA$=4avj^tQZc?G@CxA1O$y|LEykNObVOn@mgtpIJOOBqS&}
z7(PuVkzM5``>M>ahmnJ$XJH{}Mtyy$+X2`dVzPsYz{A6rmX)P@-d=dz?$?l8NX5Sg
zk?4<x%j~!%-`Ln#T3()^-w_Z>%oEcQh?Pi0DIkzaE)~OQ(D~x&PfJT%=jRifJ#SC$
zFPCE1?&L;CN6C8}VVyw)gGxE$Wd>cSVxf3~dU|9yI5?b~oFMb1=?=y3*qH<hXdIy;
z=lNh&n<+DtBC%PhEvxVNZ^-SW?b~={h5sBWb1Y_ALjY7roY5?&lhd#HEuJ?ml8I~N
z=A9oEn_9F<_YE<eceGC_nz57#G{K(Xbhi3yHzrJ^+qJMyrs|RZ5vQzhT)^sk?`(*$
zT=lt>VN?57t1UJxy#Ni+K-s4CGNW+QnGG?=#@)F(&B9;%XhQA(0M2nrl2qX^jgQ1$
z@*YM&a=?rb`1vYPbAtj*{@lti^_aAQnBrWn(U-|`$?R<tyUvf5knuRPZKmV7y%Gj`
zAu+^TJY0h1iZ6_K+pO6Ps+i!-`~la5a$={YUlj*OW%rv6@UI_jIdY>TAvkOdq$^~K
zdEB2*i-?N;ijD1Q^F_uN@ZyC+p-KFn^$T$pQ-y7VgJC%-Zo%vQv5`N2eijsb_UhHE
zfdsaYkr8<$3<43ahoU*gd#lNOmD>9HlNrN^;~rArB{sIGg2F41nOe@2epl1=#Maea
zN<__l<OQ-523Yv{Q@EXWj{bg#I08RxdwpUQ931?&uMc*0ygCri5-8}4G@2uqYCjyG
zi$4ijK0iG*T<?p93&m#%SX*1$f&OE{ojN_xafxpu!^9+)3dUy%LIpl3-$%5!8#I6U
zSDXK?iljrx;Vl~Iyr+GQnV`PL1`l+%wW|XIMRa()`vSQW*(ssmW|eUPs=s}K;*EM=
z5OGl1l{x_5wC{9fl=Kd3mpVq60(ZOYFgbByL!K<t0=DBTl~4DM$lrdZ#`nkMN%a$1
zA8w1ZV5KVat6yz3Zi_-blrB@#M)ZueB?zYS6;%GLjbJ6jlU!}KBFlRT<wB+s&9(yK
zl+Z~<K7XZmntML3?|7IX10G&Furseph@sPhFLgPDWMpO*kRO0yz-KX*&Kou{G8zTZ
z3tL^4B4)9H@*-gpeSttA@wb{yZfs3%=k4Im^pk7wvy84s%l50vsw$_f1Yx{Ktq31K
zew_H?mEopb-%3bIs=EWEr>9E;@y!0Aq3Amkc^#Ka!0>E^Nl-`#jYC5@g!75rzH}*A
z`^!0G$qA3CiMMJ4s%P#ZREi12s^BIEcKJA^3)-LaSs=>K+*V(G8tJ%w#ds<?TS1(8
z8V#G419#C3;N1X*wk3xS^Cxst3qdp@mKew!4n_e^qVWa#9y)N&z@WOY-Ad80FMZ1o
zK7m5DgYPI#G7JX~<EDTu^XkA2<LlwBD6Y!&5m`t4dEM>J+Y7F^H(6(Ebc0#EuA8)X
z&e$w84kY&n<VgtT)9|h}{*qnGdBgHKGK?Y{?K6?3?<CB$>}bUFToT44*G(yk1tFkR
zR%bPK@v3eDQ(+(AzuC(C-P6|A)_6XpMa1V;DTz2bEbvGaCe|S*a4#YzhJ@t%y}ezi
zy}ey>?lYJuDV`UzTGph5g#8Y~e8}FJnIBF!hbigli@!SCdU^4ShZCiXN07knEwwyV
zsD-ziC+q!<Cw(;Jz*FBv!NLM-7pZUCXFEHlyu3WXa=tnY1bt(Bj|TBKijN;{HwN(G
zz}_nc(&*XQSv>TB-*{@p+AGt#K-Ra@4eTKW<p{}^)uYQx1{YM$r~E5vzRB;k15Z&x
zS^wK>kTq0@hLHIxz?CY*StXzZ+tDuD)`S|I(Wn~ZPQfQbwnBU(&<d8~Bdm?!fkG)C
zRm{FnLtK_DvKZy(;>^pA?09X^{`el_Mn9JdZ=PhYpz+<}^Ay6JELuM`{5JNHp<j)@
z_}W}A1#N4xIWX!fFVU~#>-r4LpG`&2M}Q3A!spv)5(GCuyrUd#j1NNS6PZf0OTm!g
zD#2SQ={FTFd8muaGJ^lE>}5MMuxyD5C+oC;$A4<HKsXT&f0wLhYy=A6k_zh+p>Voc
zgmDHc&&+uq!L+cG?=sG1Cv9Ed9lv~f9wj<S@^HdMyVqyd0OGX$BLXt!*LFV?kJCX8
zaFw!Ullvy}l%5@(oWR<UN!P&dKPJN|uI}!MM3-Qt=QulK<KW0~0(Pf}xa|Gdt>$cD
zoha<q3tyzA(Y?Gl9<+GO#y-L&B_-85Zodu>=b$e)H#3v?^5unAwPm`^QnTe;We^ts
zmG??V;1c6OQ4N&(8?Wb$BgB2biZEr`l$Dc*XFO9p;`e(?QDyOzM@zIa`*lfp1O%Jl
zmBYh`mKFgyzH?=N2jJ~w=A0JmKZ5XsQ1pO1i+5{Fx$1LrQb`w03+2N8gyZK=CD!**
z)pF;p(uUcU7IAM7Q^l$Iq&It7^DHg3_qNYB<RQ>2yCUj?c7c*#4wLXsjb350(G{uG
zP_*Re{SPisH-X0~dvz1YEVKX56`hfM)=~xxV{xi|w&q44mp0p(9ht{P{HBp7DR#=!
zRTtN#WIWCS|7v@j_jXfjVt?f2SV%M$Q*#Tt5`Ak;tiBQa$G5OWUT?~c@XcdVNDK@-
zHl$op((;h-K|Fk<;WDhq2CPmRrFzf6_nkdGfyFuvukrCUtQ$>IJ*y`W!4oC}SKB$M
zk`Ik$<3KhTt!+nI^-?%*hUP0+cE+|q-lRDcc=UvO|Ng!8rx>tfm%APtz-d-n_WGf`
z=^Jc#pHCwvARxd>VC2)DE*zN8Ca&N~&FAZK2yL2-4}5Y(I{9c;%~5kzqEpMh>7<=6
z9h>88aUkD#9X-ctKL4bKDfZKYKW1-DnpG3Ps6Gh4s>}_MJFTzEUOK=C6<bfzG;6Ie
zN2`r;36p%_&kaVixmN!kUoUj&)0FcDfgit1I8w~^x>e)%hgDD@iWnAm4K2P}QvL4A
zc3hG?y@h!SdRA*b)4ejiurVmhM_mar2D&G9uf1EVRAXpJ)L3w};ti@_W`93Fpww<v
z%--If(|+wmAQq`H=q*?-)>r)@t8>4!-kB`uxjEncwv`?z8OCNlk!yeHv^bh6E;UEw
zt*G1#&AG^mg%~W<+Hsr}H9cJ2a7>)LdG<O`QBlpzd^&svgTW*p9;&7x!YYIJ2L&{8
zXlQ6&_vdOUt}qmk4z+=OxC>B{;}gBT*hj&{?C$Ey(mch9R@#QQnXk4gV8Qy2<wuTw
z0hP~;84c9^a*1dF?yppTu^WU<#6n-pb~^jHOH&=6pp}ZYdMJHM?oO8uTu(VT2dN)}
ze|AfKVXX2DQxjG3gmDWamn|D$lCv-qYdRRro7TBmQ;7;I-~$GcYA9RCNHYbm{b%Ot
zgGm>_RLFe^a_V}gU9lSLMc36Z4n~47un?b|PwTQ+Owqz2q2qnFw`Z1>t$FB{PUd#%
zTL{!FqTu9=`}_9`J*3gK!{Jd|Kr&}MexmuTJ|m+AcVAjsn)cg5slITeVO)Ow$#Xm=
zgXeTw)nDf-0W%PK-XMYey{WCO?YdtzjmPK4N<u=SBB!k!?0AnWG@%LZZ28Ng|FKJl
zg60FF77wL1->!U##3$p8$s?)SS_|+nngM+>1v|A5xM308K4WKh2_F1$R-d0)+CHNu
zXP}Hc&B6Q~LnN**NV4sfV~cUpN%yxvQ>|cnH#i8uXc1ENpuWbhQ|Fq6ZS4kwB^MRj
z??4=%i7H9m(|e*nXbUyi_RNVq*YJZ6Hye@@;!v`re^xTvb=|i<A!(JmMy@`Xddvkj
z!zX+UvDZW0KS`2Btj*bRs$*u{xmU*Dzht2~rT*x&xt6v0(glV(D}j#}d_Ge<w+dN8
zHhBbs=}@z)MlzVDGloZF1!{SxaTqHcYtPxJ>zwP{Y>Qyc_Q6awWl_`RAYj3h9$6WX
zCbxkkt~aElq!cm<evlHAV^~+?Q9D|ewV~vK*7c_JbpF*Qf$}7uJ8nM1@Gb$Hc_-LU
z2Nvoaf`WopKpyi%0l`kj1lCsy86csPq2;c9Tuwlz=oULW59Hx%y^&JZ84&^xwcXXP
zgcceUUmL1(A1Yl~X$UzR@5l^mY0kf%b9x!U&w;X3A9nCL0Os7=-@8`gKR(-R%-?&g
zz`<8CnlRdEB}ejYgD#=^!0Yx+p&xiYwanu|S(XGE00B-VFF?^!i8mJnTnzJj`&+HM
zb7n(izg`YhiCA)p>C3i(|At!~G&e2k!}lY6BRq=o{n}cK<99iaPK{B^ts(K(tqU@D
z0`8Z@*j<zSws#GC?PEM&uwN-FKY!~k+(i<Z8<c=2V~|ggJ>$i6F2j)>2P<G&4-<P6
zrjseL*A#+*{Y>F)6qE<h0)WsAvAOb)H0Ur_5=*Je0k4v?PHNq$c3oYPv5pVk@nnTM
z&-`S*7BYj0<S@3dIlimL6kqcn#m@<U3acgUQD-kZ1d?WL<1z+(<{>Tou<Fe}>0!n6
z)>&t61~tiRBMe#+{3-HG!6(MWgM6j}^|;C1=c4JXC3SM~=*WPai9Nl^=Wv1~i!xai
zb53#0>BkghR4=g!vq5FSc+9}h3OJu%J{Nph?0&PILqmZxUt^;ijRO7i=i}Y|eZii1
z&bZyf-8b)f3Ty|L7P}AMzyfx5u-G752IZY&qo*HQlj}>Q9s(l!*I6y5LX8I#m_S9>
ze7!-qF%aJoiVv{3g4f+%Y;dXSQOzZ#kUR1M-(z<)TO<}1^>h1>Ex4b#@$JTX^o}7O
zW~fGgV2NDCpTMO|eMxJjqY%CJE0#dmd{6emQnL?A2OKT0mYl8?RQ&4iLO{7p5ayX?
zeq7zO`FO20^Z6$or`G3q9L4@olHtwHw!tLM+IV_TAmFqQz%ejXJ70T-k(WyrcnAd&
zGTqs~0HrFQYtdY(1*rmwu#;%sGrUV&T)OWaCFgOqof@wUMrSkSaVrWZc9^nX2DC;j
zoYH;@ii=bI`kLbnkMS`^9!IA&2-`gf;_eNNB>l75nSSZHI5^;^jKjyDIM!$pDLF%v
z=8ND8sQdgMO<x&S<@0q7Qi6m^NrQB%bR&XvNOyO4$1QEpC5=c+cem0Z-QC??@0{QN
zd9Um6MUcap*|TTwwbovDkc`w_Dzj*^dV7$M+4b*P_8VCu;RyWTv8fs;f-@^sI#{Ic
z?MTIeFljFm`A@TCO%9<lN3vw02Lu(#I(Em}Fp|p3sNSbIR?C+3NvlYIYz2h(Ca?nq
zMglSph5Xt_z95<jFvtuUo$m@o<51s+JQH`?CgxC(n)Iwj8-(zpUEYNir5CLomCO?G
zZ1N<zS}PbWKFn)Lqb_vRz5P@AqP3>hH9YAu?~a(`a5yQZ;8J?ulOQUkshZ^RU_~sK
z{HZ^=cJA=ptM>|uj<*9hyS|V2Zs6+v?JdB|TW?qgdit8ijn!49vPSFaa$Z`wrY1WX
z)0_R8#gjkS@=$^^RZ0Fz;a7l0DdozUfUX>_P^%`KUMXj+uR8Ge*U&HIPs9A>a@QL!
z(@ifITPW1%EV$M%W-yZcMlLR*f_6?`8@lyV-W+P&UEdP3qze5_)+_27>Fq|f3KTOM
zD5=n3b;~)E&SK=>W~Bb`D4Q9gGhIrT*wHW%<^3-B656zB8yGztMAi;Cixm}g_aM*j
zK~;Qh>HkKw_H$$O8+&?_(knD-zH)Uz3sQsy_gLE@%*G}iYIYM#)kgmGKmoH>P0UOm
zgTN4Whp`r13w;%aneqI+J5Q4BGPPG=J^OLuDx0uU7JZV|-#p-_s-i5!kMQhlGs%Wt
z*hyv|K~}6cO+40*h67bbm)=$PH3Y<iZRuPc_3ZB5B2r?63|dHh<&nvK7Q?wG8tIH%
z=hhF-NFn$(F&E2iG;b$}v>E9C{v2uBkWqIG0$UUF0t%w_^&nvs*Xl4T6yGlExqIlZ
z!7<s8=3+{-K+{QOc;n$vhKNU_WX~P)ceQy9%FvL<X8olF-kFqmu*{^vK$ZdBvEXiQ
zlB$aMVqM=hyO0%?Y^s3-P1=dQp?sO1%)UgVhayb(8@9PXtq4XKlJIlH<%i=Q+H_^6
zkdP3y>4V&KUqjGO#!WPjo3ry@&N<<>U=h3Tx;aQoON&L24t&F9R0qYOQCE1jS_!0!
z+y$l>2skWp7}biY7#O~h`Cj>b#ik{klUI?!-8^3HqNk%HfBpJbFXDak-{$`>CO8}D
zIjgL!<n3Y_)5(-L(j{xMvR&X7|33F!j{9!z9|b$1=UJ>~tYJpwdkQf>&V-^0HHV|a
zfhq3m=R6)HEWRp_M%KlZ+4E2&m-&PP&g_Pu=vmah_n9POgzx1qg(_(kOI*S#=PtE;
z<<~zJ)4vu6yWR|S6sF#?(G4s;J}Dqu&WRL;)V6k>R=kwpAslL8^JF!+mSHLkD;U+*
z)y0&U^na)P*96S2F|o!d{rryINI*xeSFmD&%7&1UjP32=2hQA|g%0RzHP+Zscf9UU
za19g}-5{uXWWacopM&$c=t<f|&;rX*I=A7&*Y=5lw6;S*-5Qdn@9Q*X<=!j&^mVhd
zNI58<O*lAYlQr8Km<_CGynUkBh{BilCIbaa1eDhq7BU_X&n)|bn#`6X3f>Op_EyJ@
zAUtCcsSB6GMjlfqTv>wNXl1Vzr+)axQJADTj%TX)q&9@j`?+}Uc3+|QI!Y(AchfM{
z-S-MK^@}@g`Z~T=4BgWoE5%YyR#$nr))L}ek1ux@wF6T_upj1sr?qBz)R233AV><v
z!&!|ky+9Dx-1WQu@jf$m_K%-Ryn2~)X-;2E)ZU{5wl_quU}fPB>1?Rv+gF8}>I57W
z{M|AJ(Ii}(?6N%)1dyoVY-xPZGC4RpegNAL)AMil90#EGp!G%}^KR9H?p4Ybj+Q$L
z_PN;EV*yEV{7(eSz{n^-wSbz2CfF>^BNzl3{Br_=ujz~X0pfy>cTTE>>hSMDqu>ER
zu0(E!*U&%`N5R!2_%GR|rIDbMuQ2Mu#6d$E&5>=$P6rGCtZjn>mCheP(0FX7x;;fl
z|1&p7+@vcH=Hme<wedsGce&xPxuY}Z0wG!Eg&F9`LCT}%=T8BEN;>F$KpE~YLzCIq
zm{MEIt?RxVq!nIZzsYeGie6V&2jTWHYkql?=Wj1mUn>+GAY|9s-s{&kTBDXc*OZv5
z<rLgM^|k0b*}LR$2k+|z&v4~DyK8zLYxi9>dWA>bz8Rrr>w?c%n!;ZMK*E{g^isJ;
zw){MhUQB-K{q5}4^(}LO{J{cKJJNzJJnPlzz(<^%O_f=G&v^>DM~_$9%+OLF=QypK
z-NYQBpdG>mS~$~!@Ljl4mDV9?x%=FYEp|=%{0G<XJ24GhL`yXDfp8c7vl6a`p63H*
z;n7lLOwCUk>`LC~QVSd<SBC3gi+eV)SI%r$Y(`ICTr<4o3Xl)@4c@<hnpn^Z<6%OZ
z7pJe8>>=ML^xV2bt3@#~9(v}Zla1)@(lehdFomdlE?zQTc5_Zf|AwKczDTC?XH6G5
z*N+Z)NyYr&nGpmbz2J^I?(6#D{b~4d&coxC=lK`s55(%ROhtNHw@&&7FCCvDAuJNw
zj7So|5>v)IUeH&unfCt#&7r%8hk=Pn5V&Yq6Vch&*!K4Kx96a`Q|6v|Sg3H?{ARJ>
zFzLks3P2Dxe<vpJ5@muvWRO!)1?J_^z5mV8V87fhHF*ip$KhWeaa$mOG$0}&8CY0+
zm7yuv8tUo!I+)7G=CmQBsHkYWOG{4v83xU)TxLuE7|D?(0<HoK7ej4rerW<;Cw*^P
zUU_cG;6b#l01|gUYC~~waw_0{h>wpC3=Vd@Ix>LU+uO60DSUR`k%wpp2)WIn^p^F_
z4UHQRPqtq7H}cxr#Gr9&FmY{yu)OdQmltv;x527g!qta8lHSrVV`wgdspv2G@$jX)
z4MGY(e#?+%+HdaNxC$gWa)h1)${|mTs6gFIsd)O#n^PwOYudee>EsxZE);Pw1s9U2
zTU=Voy3rZDI;r|uv^$Ep4Vw=dCTvAg9fNigk}Zud%&!m?(xeg?WJ$~J`#2Q$7jFM-
zpNDkI_rTLSZBG#-G48HD5t$b*F&l_Y|829gFp|8SE<|~M(&M4-h&0AQylcqZUkAZS
z7ym3}-z-Clv0_Rsz5hNw^2ccC<^v-_ACr1d_KOxr0u<jlQAxQBgWbu1KJnGBiv-MZ
zT!}MtM^5a2NN(yCqm{bb9&L)M^VJ$m(QmIL|7|ZfxvNLLjuXj~^FtOCz+zM?D=k|q
zBcqw*7XFpp#1X;lPT<7&Ox+b%P(Bjs%~2N5{4wME3d_^od}7qS@}kI><i9M0HITv+
zLk8WI&2TE^$Sfz@HRqqW@bU45lkf$X*o{9<2%e8A0r(DsmNu8S3+=7TjxxY(Rqf=f
zh=|Nebemw8+{DDhc(s$8+vcCI*c{c^J0S`~nKc%pIlme|m@pKWD5gRneM~3VgQWLN
zhsPtTGJz_^r{n05Jx9)`7R5Q!;&z!TRb70p8XHk)>hB)~G`9~P2K2d;WKW2A_%~E)
z<cJ5B739ypm)KihlOi`B=UD`>{{y#pw&*OKynWuPP@PfJayq(P5l?q3m$;FE({1<M
zRqC)}ba@PA;Dlr42Gru>6ilwgYNS-_Ow~LiK7B~-wP6#BM|xR&zqJDq9U^G4Nrp*&
zP*DsHav2AC9DTbx)5)YT`B;cuy0|r7=G6fO^=T7R_FA?o)(<Ith^0_I-~&dr?L}zX
z<;%3T@yQ>9vc0jr?`s*eHCtSkjZj7A6l`Q@K?155RJu2Z#%^5CYEnnqxykaKrFCSE
zg65}1ndd+LSu^NfQ@O<EE$3rR48^c}{C2TPb;T#bE3T8$Dng%18K(RDwBZfPgG|6n
z=8LpJ{^})@InA+y(*nuJH}dzK0qZdv3bg{yFgM#oWV2?59wtUGC<G!kX=L-N1u+!+
z6zB##wkmQ_A<~ID+pZ6=4(EXN<HwIFJWjy?wsSmM5(bA^+|ckf5)zWSy883y&));^
zPb37>WUADFn|Ff$lPFEh@r+dopg-C`v)T>titg@iaweuw0OzLQ>$=19HS9<m`YE?l
zFf)e%5x^;o+gNIae~FE)zz|nkSNDRLSOyXmMfABjqhMkx?DZw}+IvUpz67sYr1=;8
zI&S+VIB-R9^YFaD!|RNs5P8Mpm?gYUAcLE$&9eCyUvDixH_h<==A22dIYp<zH7Y$_
zzQS~1Yo?0Ps58`{JCfqpnI(cH4w86)zyH<sW;!Z5y7ks45NjR9LYSmeHvFs_&^#H#
z(>+dIsUuq<>X(gf`JkfBBFxj#$$nAdg1e3Uz2tezy>S4o`>LCEvJclt)j{eM{m&T6
znfYn$DSVM#;TrioWd+r?g#GT%54O*>MDH$QE<gIK<qk(V@rcY^Hj5GS;XgA!gcCpE
zE_X)}pjFlTkTI}iKlg-x`es@xA2L0``jJgmCsW_1kXLuy-7F)v)qQZ5%=v`xGU>5G
z(ScdCcFg8=&F*U*10!CCS$R%#f3<=MdxIeY3q8-ta${)|{@8yL;nAZ=+B|Vz>_R?U
zCWnM>1$!~tRr^*){|P8lw3)yD!5|>cT1T)?wkx|h8Dg}mcnTrfZQzcCZbFky9pOB?
z+*PJRH`!_i+bY{hl}ZwnwN+eV&$$kVgW=i4hpD<-2?zZ_pJR_-0Z2;eA1uzGKIw^}
zO|CpFEq%K&ls<HKeY!gr5)=d@(_srQ_SewNtP8-E32zNP6iptVo{E%|=oHZ!t%9P}
zxIf`tD1lOqpT>uKh?Swu>twBGmL)4P^0_nbwnGPKgH(7Vqhx40W@ZTS@qcK}<Gujd
zJ=T>I6)xguv)DOkfaQ1I(DZC21O_WO9JCXM?shAo%ElVg=47&{NJhNp`h4<B(J_WC
zTOskFQ+7<u3uP)*R#QeX8&x-kq`|66b99iJXmUvBwE!oiL1;i?u5|fgVlqe)wDp;r
z8F(>d4OK^mdyl%Smd>xsk|UfPAao>)y4Z~$`l}mXG9#Y(z1TNrqawzCoQe1&TU3j;
z!ySEFKT-XY-Rfv)tdUV87C%n7G*wx!{guImef{Ukge*erMmSO%U}86!PnBLPu@}Ee
zManHMtjPkI1Z7Xv@~y1xihUOZzcq_ePIi8WI39>Uum(K8rM<zV*hr^)d67cHMgM&z
z<pUFmb=jr;FOSO!M2wx!^6IC&cH*$Z3xe8>K<C%;;+mt!M}ICVsy_(_NW!&PIHpqA
z;?vv&T!cOCGcSThg4;vAl>(`OM~Ty>sefQ0RDhKALn$`VPI=|;E1Jdh<lDQ;>XFl|
zs318p!Iv(~6TkXzSAAj=?QNc*s$r*O@{;l$K8G4b?!KlRe>r{!_i>VTmF=RQEyk~c
zXFcBp!!t9>&{3F+(k}Y_{9u$!FbK$Rfu&*v!Yak>y)$TuZGS%(Y|izkro{x0I4Pb}
zLanem1-zCQWpsZn1Fk6NK=FB?S3<usa7Q{HiLk${_oYu*VOQl}uocDUZBmFC?b|Vc
zeus7$A5xAwJJsY}-M!o9Fq_FrZ`(0S58Yn5xy>u?$~J^#UW6X&%t?Btk|t{0>&c6y
zQ~vLrGTIXP+jh6NEp}LtgrVAy>3ZJCLELXn(Pfndw-VAS7NmqKs4NSzGOtqYX{YOS
z7~e#y*m|33{I>F$(5^VE2eD=Nz#|3@dRCfjwAgiz@l^fHD#4z3LHDN_^lkNRDzmt<
zqA9Ytw&lf*nRoeHr+RyHBy>IWf!ZA0MDM<FDXHwc9&i-HPc~N=LrF_XeT}{Y9D8LI
zE-l;jr(A}}MyaHZV|4q4=!8vL(OQ3|Tf$$qtm1(3mTr*#kBn_Fhz&u50SjCLsei4Q
zbg@+pa{F_#*Mjw~;g$-0wG_rQS@sttR@8(nW{OFAp+RBPm}!GoZdk9)^-+>sY%QE$
zS92G9W`ju9L7@v7r^sh;H-J3*fmeJV;pq;USOL62_W;pfWfIttCx7~Z`HP*cKRG+y
zkgKoulw17R6iMyXyslPdwc?=7j~gqkd-T%}tLp^b`*W#upM?)x4(YPdS$iF94*c57
zH>jd@b2qo__@+)5!(!L)=H1(LZ)nTOnSc0KYOhqM8bh=AR$mhML&N3IiX4nnM<&qB
zgxoA%mYg}78~&l(pndlFcZuClwQp9Ju3yaczo9QZ?Q&~s4XIwTG30v2rmSaQsUu5}
zQ7EHVX|s(cE2SB+KdN!*w)u*d2+;i<?fD{VD)S=G+1)j7zUrrn75CAo>BVC$s)|G6
znK;3cez+=>N&9HYd97B#b33U(+?cY><$}0Rx}#`RKGHLZk)4XWUiR_RZ;oQj!>Zlc
zOq<dAO9d>n(bFkV-OJf=I~rw`e_)8pU{H-pjQn8w0Zkn5@TWx=9=fMjf7p;Tf_d-Q
z-u=THOy;?7^=F#xcQZUA=4Wd<J+BX?@XR=gJ9{Rfu<l2%K+b>;*_zD%Tkob9sJg<>
z36=YNGxlyD>|SEB<m|WRcN?3!?9duKmMB6|89C_kwE<5&tNzMRh`Z6LC1orWhd+wi
z3?OD%70|4I%q~p55L$6diGE4qqm$9X7&sMhGE=2q9H4|$$`KPiKYe$)%=!|mllAU9
zWM)Y`Fk7xi7lLOSWTSBu=r(^rtA`MB8mUERzHoXmb2(Y#<aq8o6|HfJSVB7Gx<55f
zdq~L^M8GnT6zTE~quQ*kn^!O>Haj{xgq0vQ?aN!6NMy_DH?gv0_walb%k+Asd9!9a
zVV4R6r}Q{x_~ni#-qF8aX3aE-%w_MELPe>ud5Cnw?Z@6zvBD2g#1nFlWIr&w-ksgk
zMrtYF>vAb|qlP}+*z3$JWz^bfU{UN{vVH*+-|it-lSL7pk=}Pt7<JR+9aR0m5J)Y`
znf((Xt8F&P2xlLX-k>TAs*kCgKZYw_T^z>*hVSX_?t}iX%zFfq<nbZEF+I|Uu(wZH
zx-U9;7)B&Ew+_)ABh1Qo;*gwu%gI0=6_D5%J{bi&2JEj%c%9$B>AgX{q>G_-?}{Lc
z5X*dLP6Aq_fB*z|c=$@I$#52(`lYQYZnAW2#I{QB5BJBGlcBMrxpnSaZcpmNxN0NL
zPQ4_<7rr!wU<=>RPKv~~r$Rp~t6n=ByU7(!J(DhJ|L*eYV^U@?heZPYT3|$x;BbzL
z)zxayqvN$Xzq>{|A2a#Coaom6pB0&BY`xC}!)aZ2Q6CbxBOIRTb;&54e`!DJqU3IL
z=evocKl;qaJz)^Ib*T1JM*6nQr?j-#8^`{8^S=F=E5)IEb?ndWPa&)jyjKN9a{`U9
z;FXJ|#np{lzF>(^|8F%YvbfIAR$nWmM687-KrF)TixA#K72~pk9<h(M#V~E)X~QoA
zAKuX4A|t%{IGkp`DyyiRMQbdRe}C&!KUq^V{-YPSAY@ksmo63VS4Hf}D=kF(Y-#i~
zMZFG7MUp4N``){X(t|GoiimyL*rF2^9Ubo?!TAu4?kSMkdD%eeVq-;l-ft4me~z2P
zT6H$oP9?O>Juef?&NCMvtDnbLC@FI51fUo7r=JR_>@<8)TLh1T&e(;mwA?F@+?7u>
z3=2Llu#5~v@h7CVM*Zdc6qo0|HA?$23>+<R-2{sDuc0mg4G)y5S`c=yIV?`L4nPcS
z4j-gtUD}uzJgYC_`LA(NAcz3J#h$o{fh=1+apaEO_5O;RoZ8azRHFHx`?<$C1(1e>
zddmK0|5lx>w6CVRyVT)MH$HOQBaz}=P_o^4UJ<A0SX0O0J)Gl88LTza3eC2iy<#Ex
zW-_~3KyWp&1u-NyTybS|Z0QnfRTBvcVx?C#ztb${yjX9oXJ;ymJ4@p)>@{_CLb}fh
z(;pa1Up_^o7n_`!lwG$Z@Q!)<e(g-0Q1;259ISE(Z!YMO9`x5hJyKw_z_GNk83-Im
z>^5c<^F8&xN3#6fy-g`O<%N9TZftJ#%x2HB#qp~u<Vm)Dc9z18!zey$nf%D(6csy%
zQ56#5Z~4O;qaws(f9aep<$jkgt<BdNvi%e{;KPUboc*UjnU;JjKI)O2bO+6+xkyZ}
z@-ml-+}%KU<U(~w>h!yYpha5joX-eK+EKno`Wh>}!Q$A<2kfCvH@ww_k%k*b3{L^8
zoChju$gP39-HW-c5pOfEg0deYnBI)1&XP=b|89%BI)b8s4t50r7DSi^TKElgC1+d1
z0iXe0IB^4H?o5rHJb(boDjQvl#ogVx4Gattk{2cV3pK0I06}hOWF$-S3&7{0Pz|}K
z9O1Y6!bn-YkI^2-pNb|JB_cM{-`qf-qP^mlP#T__wyysEuxdQ{Co05DK9W@WVU2&U
z|M&d2AVPOqOhdF|^b1PIn{LZwvp;(4-O;HdU#hUQ2vjlKA7*$t9l`XItY>2jA;Kc@
z#~5<5FD#!gSF*Z(S{dC|1ya7f`)+2Ha>eAXZq`n^O%Z08MvCx5+Nvh?$}r&tgz%BD
zUVtyM^~=&sksFqWWskGPhx|+rTrO10^+WUorx%}_l|m-nm6qik?A8Z<)^vUrL$T`;
zR@U8D?N^kX{Gdn=FEBe&l$6Ezex*&_kdp+#jk<Amq%^9^7H{NMg2L$14U&o>R5hCF
zT6H*c^=smJyW6PDn@!`XFUxRu3B5E$<B9O)%I{S;a+_&Be9R$dRwHEAN?tn{sON5T
zk{$kCW&L1$9ai5I_hr`i8w~KDRV1u~{eX*Z{1KUjVhxF2Q8p8pry&1dDHq7s1I|_|
zpIf=lt*f?9HsNOh&ogoWNS2qghLZ^m!XRyWdI}bna6z9N2ySU>s~hO-=-<5A+?%ez
z?J=m*lwEB0QOK8h>r@UlPZM-3qrS@WQm3O=UaoeSkH@^(*KxFi%G4eyb4L`A?<+Z9
zO=7<nZTx53N-M|1UZTkqBm`#{uRiT&BW`wWW{Y<+W=^yT5gdJBPwDO|3c@#Ph6tPr
zcEn2#?jN8x^z^I~T2gWZy&|Yajl;WVz_~!JN_L<qmj9xIv6k~P@}<oi(T)vt`iov|
zmY|3g1J@_4MQId=0&@RaHDgJmdfb;f8~`5{w-Y8O$Q*}BSNJ+3B{w*or6aZ?GW*wj
zht9UpYE)8<6PDwVtHIt-iPSnL=34vrVMTu(rH|0c-iLRML@0~eh(&3o;`O6WEC2Y2
zoH2MK4y7aTL*AE(Nu3PV(aBz@C9;wf%EevWdcPCgM{r+D#<whNsNZZM$_g^`W~8BJ
zm-vi<ZslLE{5M0dT6K18W;(nJ^2GJRqRq!U{9`^9gE1pD7)APp(VuIm&A+c3*EIBM
zMI#AvMQ7*l^!${h*B?dAmvz&PeUC1Epv%>c;9Vr~jdv)U@>qqlfLr{Hvf@Jq=gd^w
zd#+wIDejZma9%`)|2vzgtW94TWk6e1@47Fgtc){X?_vZX=I-@4C}V8f8Aun(ttMH(
zGyu2j-Yc}ACt!mcT=&0x|Bj$P03~sRzIxXlYcP?hTapV}zq`f9d$LS%wR?42qs#Qb
zbRynBF_DP0k!VGxSUkEl!-Y2F>`5o%9?WW}4A~_S&6ssA*{x%m3sG8&t5@2wM_NGK
z^Ul^;6T{SzXFi=wba9eQ%+w!MAIzQ*R>h(#F~q6(x}v&?-3^*^Hl7zi?AOoVvsqbu
zAUAWOr-;*}89!;{oGIT&NChQ>du+Ap_(cJ|+Udt!`aBm)_31L(bAI8O14;$DBBn%N
z_i*+nm4xd-*)TZT;t%ELIO3VvfRJn)k#NF6T-DIVDwm4Q3=M2_QiTG5O$xXn+~VDZ
z$z{fo?qXIW2r*h@IYNHOFQq|6vvp1vyK&}mZ%{A9chuY>8STWA8e}Ax%Y8e;VMZBp
zf%#9Ao_f_$*}nRDaV)$(4Oix4zk;(*J%k{p)3oJ+GJQ78+KW~^Pi0|1dC#3Bz5c3L
z*NZIQg*@iPgSs7Z%c*)5cDf&X*^JtJj$h1^_@GH<NmRuV59kgxC8_?nw_4t^5gA5v
z@a>zikaE-Bx2aMzMI_N>59d!aa4XTxufe5w&%?<0t4gdqqt7)X<{WvwD(V_q6a}kt
z>AKLDs?kT3fI0o77Y;TuA%`!w*_+8k7QF7Nf1DPZp6aA0j!4V}8Mp|E|Bzn3Rst~N
zdXUg!ffv+xnJ`kWc*eWkq5!}`4C1hV|D>d)!JwH@@$>BV0~0btLACu|YHI4(SP6N)
zsp;tqWzO(iF*wtqw1Fr}arGuo7m<+HPd^B}^Wd~P-x&wWUHGF8EFtA&@^@@(Ms5B`
z0Hc0ELNWl5%(hyGRiM5$d$_v>LHVQ<@Uf6s%n~kQ25FD6voC8jrwDKPV+h23>xM&s
zj~VH(+{!!vg!-sEi$S1aUD;FXA{Y-6D=}Got~>oeFYY@(m|um`EhiHqT_gCq@KX}@
za{?J~YR7o}VjOQ??u<C*lbYznG>}u9i|o@2#mny`_Ho$iS(yFJbfS&+X7alif^1h-
z9$vAdINDil&dde;i<;lQGd2#RjA8^#|L|s2$WyN@J`gnnbpu`W0xEF1Xtow+tCsR0
z$8BG)Jl{+D@>Lc-<v!`{{y#P18xzJ`->Z=5ln?osCoLeLVB}Xv1+BitVl6fwMMuMM
zxg8yf??V4JMo5oPI>zRAJIKPwpeXC_&*z@a+{z5JiqC#vbxCF}HE@p$I6PG}oLEen
z`y+Fr=$OTdK3?ijWj6PVo@!5^FXPc>d1RT}*~?ByoV3Wt0;noTI?sL|Z}g?AMx`7W
zz7IR@b$0PTd?`6crnKGnb+Tr3yxVDP42?e)5c)Y(HBYT_cXnBy`9uHpm~Lr3EyoWJ
zw}-fW!otEF)mZ3_$F*~ZEsuj>$KbYrcINNm<^HT$`%@%g^F8Z%;#X&bM(EyF=-Uoz
z-3`$nEo@)JV{*duKGPE=o#G)uUI9T>(f?r#BcB{uOd@^g!+!NYVJeKB;&f<btW9i_
z{O3Pr)kB^R?TFt@d@I2v?)zbBG2t3_^q%zAfEDbV;-Gd8odB7O1`lWPMeUVW!d^~L
zW&~r0pl@^rSt#Yc^~-Aup4fF<qjmG|01}e*_dYD>8ABgFPlR{&F@^g`aGkQ?o=_-l
zVJ<b!eN+GPUlE2ZUzPdRa}2|6ec2WZ$}KJz?nG25e~rdWF-7Og$i&MA8aa2Q+i`Bg
z%f%DW&!3b9{&|U}=wu~x^aLE^DZYf(g|=Ayc&Wq8e(y?4fH7m%B6%(^ty_C*pJ%O{
zwoi5&ZCb*|o2vLe%v}{$6rP=H#!&oCQk4_dES5Mz`>&8<{{SQ=2QyW>Vw>NR8d^M{
zz{7#vnW-sdrrg}zWGYSo1GBKQ5r92CHs7E9m+>_0vlE+JN-{zEU7@@3`~>17qV8@c
zOAJ<`7X9(=*vQo@OOT8@7_$g;DF$|Yv@Qv?+)#Lha3wwJlp9nHid);~jughjf@OlW
zp1qmVbSIcs(P*o(T9`o0wl0=+REJbQX^V8s((e}@9jcyw_V3JEnVL>mnwn<#WD=J-
zx?|gLtf8uomF|LM_>1FzrF&=h8mg|-DUnME8XB6teV*^c;mQh;-iDp-rRl}RMGmSQ
z3=KQf;L{`Wi7(-7uHetdvcl(Ey?q(zshXnG9VB4ZE!mp|K#%MqT<C&Sj(v9q?GjHH
zd0fA`&UYaB$j5lqaOzT^Nmcyh>vwG)55M*^erLiWcK@0n`~k1Zw)XntF~P>fuv6Kt
z!SOFd$e!f#Vc7vL+45D&p_Kz{%RG*F=`R0k=EI_Tup0qF{hwYRrt1Ks1HWf3h`4MU
zP3Cu{r0t;ogoSz+D{OIkSe07>a%@%*Vu$#%g^dr#Rf0NUhtkh?B&E)%U45IvpWPZ|
z)z`(@+iu8}NMb!dv7gkG)!cT!_i=9Pevx`Z<TLi0AK@nh{pQgx8)Dtcu;{s8QnTI6
zIEYxgGU;k$?HTd+3-sZWLT^}N-*bedQNXkW5bP-jpJmU(3SzX;2nFN><Dds_4vt}-
zk?mzeRMt9;Oco!C|L%EvAvO<xH!xCyi@LCx4yA!Ae5BfDR%*7htLyY^R9@}wVMO|k
z3X@*FbG{+Z>(}jL2C$I3q>q?ro~oTv@UE61jN4G}A}Fa)d9D?h+CX{0Y{dn=izA<t
z<Shy>s!+ARn%o~DOBu*Ut<jwE&wTQ&%DxI6b2G|`BH-MPiOO=DMQq3`zw_&5l8Z&q
zoVMSEIce<fdyTI8E&T5kO~gahvSkiM9*BqITcWCs?CD_F-o^+g2n6y;#Z>i}L}vQ4
z1(`=|I9LwA4_foL|9cG0zTbj$UA~u)B3@w9>|E-U3ZWp^k2huBD~8RtesPU!o@*2*
zf`LpSPQsZeA4#liyKsb?lX6nE3=NEo1@!(8Pt;A|Z-dFgkg-Vnb7Yf06^tXIps@Mg
zxp#&Ui31`%s>hg$7EHiwE8?+etC#5-8&loiUY1<l16?3gs80IMFfhLZ2gblC0g&E?
zZ0iqbSeGJj61LQ8!XG5y1uA`h#v~NVm^*`0-$y%03Bdn+-yKA~S1Y<6(T5g#px;3H
zQRm4Jl`iJpuhcTs`nN?g?p{ls5rz@==NKKC5r+OY(pk(_gE!|Zbi%+`zAGLrboUU{
zjDwKt`P!U(<K)RNN>Oku^gp7_;O1Mqd%deuE8pU(a_~;KUTDGq)5NjFy+Plp|IN)D
z$w^g}(qMe97U5Hw`7_fK|2fSj+iEEt-Mw_@RHMjhldcwTQv%jx%Sbk?mjq+B>7V#v
zu{%F)_vu|5{nuyAI#I0SC0*{?6HSdy2(f<wBeB>xIDbCFBiKqFT<na02ZLk!rmmDJ
z7yH14`!5=hEC_dnlgN_ffV;w;Y?sG@ZbKDM(U2kM-x1~@PJ5>_tWS1#__B?FgGgy}
zvxP-pwrPa)h!G#{H6FHvUKV0E^C64zW8(GQ@d{*>@`-V!DkLkJs6{WX7YTZ031RLH
zv`q~7$pHwm+G67xN44c505q@7^mn-`>pk<k*0@J9N>_y(Qw&p=_lqgoO^Zjbc7d=g
zMfZ&Wd0g(*&FehpV3(gLzP0GrmNS^UfnJP<49OQB`vw*U3anXiM5KEcOka9ChP&cQ
z#l2oNH~)9%`P++wxPhMOTUhh}-gK(og>iX#S*05EwX0yj)Bp$x)ipHcPiB4!2fPCE
zpiHr_dg^0F*jel>cX1+jyrCR1>LdFj@(Z4W;SG-TuK7870?$sTv6BBk2^dgo)aM!y
zz|)WRq(oCt)y@E!?FrzmroUS3U0K>x4a`4j+B$B34gu(`!6BIXdFC_y3naMU=Ep$i
zj%+l~)HHD9L{$_q<=v$-gRJ<>1^Y=CY~W2<AABF((AZ8a$&3VcxT~=iC?1}xAy!~p
zrmu}7^cA-+5&*I0yQN3cqUt&|2{vA#X)H93BU>ipd`mK8Wzzn-Fa0u|2xQ?RwzelI
zkBGw1f8alMeocQq(1<{f51@Wi`1I+$wKbF1#h$@XngDq4LHOOBH*6$xsVyLi;H^~!
zZ2b~GY##e;zs<p4`@LLMQA#h8B$LOWDR7qxUy>f=%|VjBC33a;ORVu<H_(hd8irN!
z*Ci?^Pdqney=s9gjd$|CCUuNqEr4tmF3&?O#)Z%od-YE&rjjYCDGGi~w0*;IUnP#U
z*yc0g76uPfXb5xC3WwEnh%>D*&B6Ov@4Zl7E&Juj45ZZ&S=mA;AkBt8JW%fmq+}U!
zGK(=xeoD2*?dQe%9u1gQ0)a#D9jml6>!svOi-jf*ZI-yf;f%SRAqA)o)3@P?U6-F`
z7a*vy{}r5Ev%}~R08w==htGYhsi??QDNwDn9N&?%hFxsWEdlM?8nRmE<Ys(1hw*mv
zb)ig<gqp=zdO=VuVvk~5&y%S%RJC5XE%2u|(O}q2T_I(>v?x=P_%8VbvfA2L1f+RC
zKuYVg)xPRN(3dZ2q=n`AGPa$0b=cEq$7Caa49NaUf1PS(QJhA>e`b;8Eye??Bhu?3
zx-RfjOIV#F4%X>}9RQ|DuRO|<-xoi4m|a~crAQrxNDYq~A}DIiuZHtn55oByZkXfc
zlf=RGm4JgWpYk(=4RJ~W`#qi&d7X?@??J!6BMIbsfEjx2%L=Lo%PJDKpcK%w#)TcS
zERQ829O)8Qmd6b$Xz^_m=buOFvV_gdry)%<tZg`w-mTGp5OXe=T>al!B7aq~+%gZl
zZ$2PICH8!AKc~2Bs4=+3FOU%Q75=TF5tKuM04OWEW|Aid`mz3>yf)p$q@TTDfn4N+
zb@iZa6#$q2=4^YW_{3rBR{kHTlV@ABWj*(BCFqLFeE>DL#brYUiabkn5uM+!kKZ$?
z_hn>ewlyfU>B8@md?FcKABp0A;>j8QA|HHm8TAWf1!=c3Gw{$FthIK(IjPy+fpzDk
zx1T{xYPfG~_0M$;4~_I;&jQjs$1>_tv+DUK9oH+OOku9?bt6mSXZ1N(c=pBr0)a62
zf5qEn6Zi~?M_oi&e|0H*yXZ#-X%r$^s5SPIxt9U^)aOgVZn2K=t1$B47n3ykD7=|h
z8btMp9O^wi^^s_vK1Id1tzYy3X+XVxM!iy<PDOdZ@KFQ0?#6Wl1@#f8l^o!&wC@Bg
zNyKa&X*8>l{YieOQx$X0%(N|Qm$PJXBq6?$=$T}Zzs&7OSfC>1@4D8>1$g5x?%e0n
zUREIT%|T@Q`OFZcOf?0`i2eI~QMrTP8yAEBb<yElKLGztaLH>v>2Gd)mJfl`{207E
zHa6f4@%<mdH^vtQ8Zi+0aEa)kVsNjFxAczxA&o6YABQ~g7B>)CIaY<7nLC+iGD>Dg
z#HR`!pvZWfRmpj>rZgRVV`PV)pw((9tmk4gNAMf#Y1m-_X@Zu<Kfr7--v&+Wx*9vO
z0b$-T23FzZ{(toh`BOK!CThsNN|PC&5r(UsVZ=hdDPv>GXE!^h&XFlXP;>K#m#~8<
z_E9M(=PetSOFo*GnpYvQO;lo{Z(wPT%*W1aoRebH)wR`cTGDnnVLy>AQn$}zapM(Z
z6C?Wj(fQm$G-=5XUC`F6PJfROEKXMm5|-;0wYT5JtA{!kVFisKLOq86t0IKWMR(ST
zyu7`^6xA682S=4)JwEKs8&oZHUw~M`uGy!e&wVq0k_09>!79JtEaw}jM<%EuQIRBg
zY;1`=ZIDxn_VOx{CBpaeM-F_qHQ#lZWNarkrxMP*C<3UyHNAo6bq;(83#|Q1LCz<%
zi?^4@kR#uvm5b$QZ+6e%j#PF4F$cKo0sVJyQ);zB&I~TEMUDC~@xSk2D*^j^yaIb#
z!1K&DM>9lFzJ0foDSkmpD(CKA>)x*dzF%v%?c#}fUmhfI%BiTUOmjaxP0Y_MY=O*U
zDHZ(<sKFz*s!%KzGBL2c*^A|RdUc@1hn{96D09?2TS&dHZ4%9T{nJlBMiamk+3!UN
zq-gtm_8k@bw?FMiGp9Am(D%k*v<rIP^bo0xh1c5kQk(fbigA2QYC@-9`Y3m$IZ<lb
zmI%PaKdCQ3lpH=C2s}V4F1ju(zgh3*Tn<-xgA9c(4$tC(X{c{6PblOupm4F2(M%7x
ztKjdFFaHm}^!g5p_S21l0*M3QX^{eIwbf;X+n)dO^0F?F7?j!GJaX5pb0_owCW3zr
z{$WC~Xms}*d#Mn<Z%g~%`*(NzE7)QK$)Xi3MV2)bso~Vv%@oJ2JlmJ{D6jUB_V?^k
ze@YbvU?UxMK`NDIXfP#6aY|!1$iTAILpjml<CNoXL_5(V?BTxce0Edk|NU<fhEkt_
z6LLL=1@VImUk@$I(UgDBR6P&7%o0Y{$Mom0GS$m*sV+xftE<El%0~FdQ0}|iHX&+C
zlc{3-s_S72<I%Wvb-`tKOZ$a3Q(JyMku7{$4uqiy+g9)*Oi0Y<({A?m_<QnOr9jkC
z4Yr*nU^^!UYUAf@eg$3wE@0!|onDVlWVA!T+|rM?uMUJQeBbq@-G-k{t!dg=UH2_Y
z_vmC3G|oRUtqQ)ttf6s(?;rSCLFqk?m&1V)6zY^5UbbAMCtEQ}qnM)78lBq=#YZ>j
z5F04G3O%V8sC%}QADHX2{NvNpa%$?Em(uC?crN*<r28%3K?m`Iw>-qTLcb@j-o^1Z
z9Oj1M;j-vaDN7DUDQ3OkC|TkhdVKEt>fR&FO3$lfRyct)GIZ9_$vQmP$ICeGrDrl;
z-Y5!vFvl~%rd{PQC1y_d4Htw)7nYhnN^|=PW@L!qAZ7w-0Bkt8Y_3>`Ps!~KEK|6}
ziL_WR&uv)s%-6Lk8rIOD5j2jWci#SFQ7QvE<gx5=+Ujs1*_x}4(Ps&mE~049-T-PK
z;RB##ejDRpI4)?&MOHFkJ1*3e#-RlLBIgjF`GKW5Zp;3=015J*M)u#A;QP_Pc=)F1
zQJp9eD3_^Zd22W^Tulk89X)xxnPlB(?Pk7+DQvX=T?`115P$y_Zy_SY5rIZ~6hVl(
zZTiEUmM1pnZ?mSquqN{9hgC(*;+2lY_z~|r#I*ENW~+<+O_>$y@{`lLgrv@9HnVHz
zX-8gp-_$m{W=Q(%@1I~yjpsXn5Yr_iOaEq4r)|3fv?D;0QLF&u2WcKe+2!bDm8^}v
zTij$@D-sLi)<7kz>%x~BUX)%E9j|AJY_o?xD<>{)aaKV3a5H2VT%;wx;wBd3{#Nxj
z3BSO5-ry<|1LQ{wRLqSOO(&Qh=H^-Di}Fi*!!>AfNSabbr~%GOF<aD3b`ZDC%8Wu!
z_R8o*Apcy|f>BWsP#`Un>|i;3e7nzNJZ?x;B8B3_(w?LYa5(~jN9z7Hn#Z00s^Sln
zu_=ik5A)fs?0Xb@dsRZQYzJn<N!NPUHtCWcSb3CaUT_#Nk%>F!CI$%1hNu7BWmHm!
z8mp2H=PrMe3V7-p7zlfMdIB`@RnOX@Bbm?H@V9j^`TYOGAPf}L(%8Q-TnpyA=b;$*
zH#dHXp$I}Y&F?H`T8OWG@PDs3nww&X%<basQIakd1B9VeA#`Q_ah1kmSwk;>Eii+)
ze;GAjcNZ9ECB7|h5b0~CfY8$g;OpAa@w9^{2V0l<|J^Z;7ND@ix2oYUomWI=zBU%+
zLT<TK&tx$9`6=mi3aexZdxCi7`*NZZn9DF$R`kzQf0oR)Jgrp9gFvL_)%kEoIOyx*
zZTH_DXRjtXG>~~2yPYCY`<}~AUQU27JU#frOyWJ8c2UvWn4VIIQ|&o;a>*IXvZxSO
zcm!pA=CGy5MqkPGatTD}iGKiw`vNqJY`5MbV!yqnvU<~W_A=e)f&ea=%g%HxPa$U^
z5LSTA;Q}oO5KXfFB5jJm4V+f;<WJ8sLXDHFb+tT)w^)XE>kzA7=%1635`EL$KAhd~
zlyEHL;2lvuH`EZhwu+yn*)X57yD&`71x_w%gIuPHRb+DVot1nh9_ufn!3Wu?t3g9n
z!;<!fGJWPBwuszGB43e<<ArnB2`5GtlA4u*N4LKUhB4H6rG|Yck$>P!-PbE#2C|dk
zdsigihSy;yw!dO~V~h*30?(^NUG)$3gMIi(z2qs2q5p}34hRX0;>CsT0@wPP#46GC
zhXVpWO&mGb1!)h0r3tKUV0Anf(@n1VT<t5nO)z(X`H}f)C)@Q?T|GTL(7&RzK6}&n
z-qbWFpX)!cBrGjQMz(hw0Q;B^F0Y!rDh9*hfD)nPY`yRgyG2G!ro(JqwXdhlJ`Lnj
z@TRBh<@=H!QsS>-%qpWrz6h*;G%^08!=d5rhMiuz(OCN+AU5a`j_32?llcI7&U4JQ
z)}ANb%CMSmk-o#jrHhF-Y<HF}reTskb^|C+(BIfye<{XBURv8l?#X^0hl1in(fmr9
zh;c?>E{+yT9-Mk?qKR$*nF4gvkVc~mABxf7i-|vBdchcl@9MlN{2qM0M}?k*srD6|
z;{6<*%;}6HXUC)=>y1uIlG+?hy<UrvcLz2l7BiLfdfvxKQBhIAQ-r)=SqLL7EiDIA
z5_X*&)VBBYTeNkI)KXD@4pH1;6DS$^!C_))K_na7Y11|O5>CZ}I-&_$6b>Soyciwr
zzV{>QVxh#6zY9G)WTA>lesQOd?@zYpubQ3lmNEj+WU*kPvySgqWrZaGQ8(6hJ6mhS
z*p3Dia5ixiViQADv;R)a!}D>W82_3m{x2xqDtSrB?rCI8Z}&e7s?(#(dk7HW^_S!t
zaeKata=FENTW*F0DTQcI7OD1uD^I=Fo&x4x00tF4e};pBIY7T?4#=#y-83Gj;Oo=P
z3JY*=DJh{oeX0*)A7*O6&9_*)?Q^!J;0eAI(qKM9Y!yU{=>X#BL%`zybljj@vI$1R
z4FL~9%xw=>q$)jGtmE-;y-5K4bC`Dtp^7)a>jH>Z%PhwkTo2}Y#tYOOTwJJGS)(U)
zy#N{VkC$Zh)(;dUzP`RSV~2E%jH_{Ky7}ifo(>_RuqzrRPr5|*iHU{3XpADE$4{05
zLqqx|QXEs4H$$JwNa$5lRn4@8#C}i7#}M=7M|?YVNRQE?vp;kAJKR;3Ap*U_NJJ;X
z=d;#EPf`B_+qgil!Qoo>Xts*2gVm4tCG~L<G4M(afu~pXEgDs3izM<#mQ|WyzHO=n
zdSr%#d~VobQy|~pG?Vl3-NqX;OVm|vkLI$O4W?*4_+IR345kV6fsv@*hx^-YU_9{G
zifi)#JfB6g=pz21SP9BkMki$d0v2F?<Prvhl$$&w;d3>Hp&wd{H)Do67bY{67X7Jw
zm+=>pl9Ei?wUIDe954rMI+)@y>wvwJAsX;s69{a|$eE3k$;%Qb&kc5WK5h2_YCF&6
zQ(@hCkRu6kchDnEyO+b;b*4L7%nd#NOpl*8YOf(KwVUYNPf0{LX}SfJ^bCF=aTu(x
z9cr9iZ8G^BXaSCp5y;JdvA0b9e6ZHp%MP-}mbr=q@33lV=dA}jMBv8FtXfZCS2!wt
z4`tpE{41o#10qbwZ3JY7=@v-HW~P#Yh6XuKDo8Fjv#>BUA>kE(301daj-584L6gB0
zo=S_+-~bdXMNQ3ba=Dbi+Xs+UFXE&=0&z6ZWWzb{jG;E^>O~#NN=gcsOsd_X18b#U
zhd?@?2o?^YMs2_b0Kkss?1XY@{9l2&f`Wnq@Cursx?MRu90H<BNoi@3B2~4$4j`)r
zMnlU3T#HnKf(x-B&|R9r$&1Fv!Tx99gS6jw&kQ45(P~LWf6tIWs2016@shiFy)n}$
zxjySy4i2K8<z$b7)LyY<5H9_v!XjsqiEO`sC#09bD9I8rg@19L=xg;E=K;;%-lF*Q
z^*OI&3`0pBWVkK}n)Y_GLy?jnq<P~K(vzR9){_QtCm5jc-w10tMrMCR-Twbt{)kP3
zf3Ky6pLtVO^)@%Zi3GLI11CP-(9C7CSA#^r6X!G1GrT#)=J)x^BLF&YzB}ogouFr6
zFzk(`udS^GUOPeKvd1&Cv!{2bLz&~2iR>0Hfo6F<2wykP{od0e=(iM8R&MU;>5!1x
z9|$&?d8}z}P6w8(wxOZI_udm2_JJuUc1yh+nMB9gMKCAh0h~*KejM7GZ{QMoxH8J#
zT{x6UTiPGXkI8R<A-91I5i(->>8?O_`1IDvAX01oMqxJ#wd^wtM&^VyJh>C3fdbP@
zCiGz=vJ7!joM+C#wsK2mz<<G!ZtT6`SDeu(ZFPHsYLhmUqO5bjpMLvR=+Y|Moii3f
z*fsaDXCc+D|N00tr~PAiVl64Mi7_{S&8H4T{hj5hIDVSB*(5p|?l*q1Pp}sPTm}A`
zJ)@9DD#k-5E=~S}FD%9b<U3qq%CId4ugk2I58Iv16>-_aJ`j(V#s?N-e4M#q#KUus
zeBidT4N_3i#B`k#EpKW#*Mj?MJkWxWdaS*GO@RSq(nH$6h6do>C`)Bsqz%;I&aJ3m
z1Cu@sU?7bHdc5a?P5mg`DyD(~0)$eKGNYD?0_leTd+T#5YhyQy<x9q&`9@*aEv)2_
zxU9aYeVk<PLJ9v<Rs!n*|DN3JHCq&vD;~rJD^!HJpV?utcQz)^0_K-9RgZk0R+a?>
z8Ui>8#S!2<TMu_jmLiIAu@&Zf_==8`eJ@RGf=g!GCFJRccTG^?IThR8JmKzUNRjIm
zz?{s=v0ilPrcBj+U%zo6<{`~E-UGBRH6KpYS~60PyZK)m1Tv}_&+PP*NF_~EXT8Z=
z`?8h_lV-7S;;(geJTbKLrvH8l1AWoAwKZes`rn(~wStPSV3Zq-NGRVp2}~1WV;g<q
z2liWV=me};y_}$VL|Fo9PwcSo{1r%)av;*GtE+Ku^^J@W0Llp!3g6ndJclVh5ry+;
zK}Ww#GC@4jyWG_m>}Fqw<16{(=VE1)3a8VFOGk+j`PP}DBsgoWw8BeD9p(WW6GtDd
zwl>j_6z>;Q5oV0`jkV5s!MwQH-3LvX=-B>LBJ_Q$S0uaEEYzNQmceqFCg)Z{k*Nu)
z`wRlQ#GxaD5ZXh}=Z?DZOTEOf?dNeJnKjP<k<01r^S{MdE?(KBh-2f6NwiTG3Sz+h
zar(@@ys9ZP(@mK0MRGzyH*lz$1}PhW)4#x25AL}c8!M}=7A>>vjc(ECHpoeF58CPP
z_cy+dSyJ+8@x~VytH}TJ@`<i2EM&^OTn~V-1b^DhR(qe1s{#K^C2ei%Gr_qx2NM1@
zm6fqzje(y<_uudMhks~b9ZAOW_PM!~%7}C0?udR5Za#~(jW&Z{jZub7eTR<fCq)~l
z8oY&(5BRhRdsrv4w0e25b&0Xw7*S%hwKZqO)z@`Jw6MONU7`K9sCDqOrF;h@pdX75
z3a-1;wsOm4O`aLny^iCN<D@vat`n=4R1%H1hc8VX*+?Jm2-fguN+hWNtE<V`SrPao
zOgcGfUj1;JiB1UA%r(;Q!pwQvrw_csm~<NxMn)82h;^gK=~G~gS1}~m(bu<I-Ix>;
zgAK>1T3BoHtFqUZMUNA*{?~K9wtp^aX96C<3}^39*K|=var=^tZ0Po|B_cbU+G?_x
znvyaT(4S4Z5mF}pJq3pb=kZJ9yXUYlTw2mEEGmLkVnBn6h=@2haxiPxVge!(W>NvX
z3P0Mx4!!;5*LlzAvs%U_3sfmK`nFqLCaNE*P0^YV3Yb%^s)$s;EXp3P<l8JPa*PAS
znOxx<qm7Z26McsD(;)O5PL#S(&;$dO*Ei#G-JwCA-K_PVhY>29;H0HzX;Lf)5z8l0
zJ7%pme1+IPy3i}B$|Kc6_N2PJq(d|w%d3so-tlm}FhR{7Zw))Y*#D2MQTjXxS9X4$
z+npm0K}~UIEJUtkv!FvU-%s4@lHJ4|xuUX?3Jete(Deip(Q<Fy4zxk#0Pe<P=Kc1L
z4*02Bhs+)$^GUfR1Y_XXqEbCwW`qK_KdUAXA#QVF0xB&yn3)?onIJJ27Z;dnWlcOw
zPfkf8HgelcQ+<QFP+ktPf>H)XuK;tTqG}sSNmSrwYI?CZox<-?F4zu(U;~jIOgjuL
zmZx)$Y)aX8&h1J0-Ox<a0#xz=U-;`pAB-I*cRyCREiNm>O(S%*u*I-$hfxK>QBYMP
zisUKAV!lwI|5aX`_w%*C3q-y()5n4=s*0pgNd66s_CtA3C0=A*!db*V|G=>AGIJP_
zENXbuPvvJXI203-oOqe$b<iJaJlJ^ID8u`NEL9DUzG0DUVHd5aQk$Pxej@v~5RcIP
zE(JO-Lg<OY3`Ig_9A2=yLbRUf-o>g4rlg=x)Y(<v^F2hLxdFcLFth_DQ`B@iFK^4Y
z_MIf5VJVP*rz!$t%#t6a6ug@f!`_g8Q{oY@h9yP7BJq>WpK9Q?e;mn0P5)O3#s23C
z_@y&bHQ(K2L<rt}qlRQ{*ZDmBxF-smSWb?bKTp8u=CZ=|8y*3x@zbCE&P_AjQIzoN
z)z$>S`R@GN_CfKTZ2FHMYU}qM!5EAYsnOAj(DuXLo;Bb_pSErdXYP8>{T|F_LeZN2
zy*Si(d(hw#4pCZ0J~;&AgVR-3_&@nA_+0ni7Zu-pGI@h{RiFub2(I>_j2N~<!>?Gp
zr%m&5zBTEt0rWO;)yN?gam|e-U33*nn}0hxak|IoSB2_PNo_Olat?B}U2$txMdP}l
zZ?g*Z=F`k+@zgr?(B1_vac*-(V9CTHg7R~<0fz?`Yi*&lxjVf1&P#C}{}F$KWexXj
z^v0JMWGMmjYNZMmwbvO2{8j5V{wOY0<iaq=pr;N3N$lQqHO?68qME!hT-G@fdY8UM
z$ty!_PO~i5?1X~uC#Yi#(ho}<L$i=w3(ODzxNb4E^v;{JM@L4s!hE}cLyu^X&)occ
zBA*)z986IM90k`lHsTlcl$HO00vq=4fFa8-wFAOeI6a?p+<FGI70@KWtnT1|MQ6xr
zSHzDRnJ?ktIl!(9Sb~8--<m30iE}y^_yLrcaBo2k9S)HSXg0sn0yb~J{m@{c(Zj`;
z*iKZ{PRecTAH{#2ZL1#wEW-m^VvtSNr99MH;HvEw2Z51IQED{!w8L9yQy)I+{eNSE
z;x)x+y(}Eps%$dHnF0et#@a5dUyjShEXJ{PKCS}R8DRmeY!j<?>6&hzR6_?K8Hs`u
zb+wm@jG=5=_RL*%k(vf6({na$vuP=7GU(Jdyyx96r<aEPd@*(V>tFvIetW$x^BS0Q
z=8Z9EH_n^mPK00Fi<Ay0@dRl6XCkJ3$!^I}pzG`RApiLNyv&T?c%M*^NFXIVjdc7#
z$@*=n{-eRdR24pU>DqMe7CtnReH`5^<?PH3pbf>UOB56oH4QE=(4c{md3D}xalQHW
zpuqKU(f78f{Sr&y>^bOm8?QH#VPBx$9<$^C37R=QssYgZKyW-=V@C#ds{n@$jO2qt
zO3RxC0U2F?snu_9uCAij73(`5vmr3~81HNt*6=|-gg{lP+uWqz=5J|s5bxmnV3mCU
zstip&5)ewiB-hMheJu}kKej1kD&lc~lb?nbs3Cl0uJz@pSbMICK}^?E?!2p9DIlP;
zJWDzvCSBp&0j+q<UhZ5lUU6)oC0eVm_1j)5$y0*y?-WB}>&1_Pe^Tv)x|htl{yCU)
z(I5Qp`BP85=jNxc7rZ@P_FDb=dYm7v{~uFd0aexWeGe!gjY>(1bhk7p2nYxW2uOo;
zNOxTt5doF%lJ4#<>F#dn?)uNu&-b_fEMq-bym#;1nK@_n*=KLWFm4M&_Z?N{jkPq;
zJI2hs^j3<{_FmxCM0$_2`-Ikx`DfLauGFC+UyVy<g)dj1zBHd~0?I|8#qv$s(Bj@+
zKM*ER1uGw<)jYgPiRQoyJ6aeLVq;^2iDkxF)Sv)lDJTd+1DcJWz<vwU3k0Va<w+LM
zZ!YM1W;LPd2u?O$%{s3#M|9b5%h}76a}7W}I8U|I9}xeRS66+(Hx3q?^F9`PgNAjb
ztlbz_X5gWouU=VVJoG|DR8&59?Be?R7|;Q2Z!WA!p&M%kq2a#JF{6V>E)}HouiD~D
zP~WHnQCy?KTj^Yvj_<JP{LzM4oBZs2<yS&IIz8^Z7t<yyCAv!ERk?7HqxJijN828%
zAIqimj5SGe>(sZGUs+qEl)Q?|M@*WFB<-9}-&&dEjY?4DLQ6=GXi9(ad)Ui|ux{u1
zpgf(iiY?>ksTDhnVf^hXH$Jak>CHdV&`H%*&B{bEhAfjtz6){K&>J<4C%jP{blpr%
zL#@xw$C(byw=ra_On{gy&~-a4h=`(@S#S^VGaF~K96NU-!tJY+9~uj3te3OAyZ+tk
zTY+X=N12;EX8KIf#jgHpN-H1q>;kqUOo4909jK|Pcu;xRgJ}y2UV5`tkhOh?S%3B9
z<RmLl@5_)&(ptM!!A}$uQ)W3x1SK%Hnfdt!t!q4de05$T*p$TlCSJjO@uQW&Q6V}(
z`|?gjWpQ!u&A>G>$4N|Wa%8R|TztKzyvb<v&#z6!2<3==HPU`?RTX(x!|W50r1~_P
z@?UdzbC~<bcX`ChNs%GXwikc)GR*Obblus_B?D{`BOGli<Xs*u%%CC=-%ZoxKXf%>
za*$O!KK}eXP-5U{BeU^IB?qde?pmYFO)COQqY>O3&8+uy$BpX9R22JU{Rw`*xe$=Y
zUQ-<*{anB4B0+W~^DJ<@&wE1&wm~c-{VFfj>u+<7cIezSpM)wyWv(%?UFibDd*5jg
z6CWNQ>Faz{={Vb$_;}rvA*=Ez=wp4;_THW>P+S8Ze<^PM7WVmsQc_agl@AbN&%&O7
z_vB9WeNgOnJb5kEsPP`}Rc#k;-ER2)h}5DJrl0ENO|cUNa^&Z=q$EMnZPI40j<Lz<
z!3qghe9X3{7)aoeHJfmkyDD7S_Dptn?ab+JZ+q)uH;*)G-8DP>qn(NB?%D^}_J#w)
z+By*R>?YcEsm^Nw4LZ&5DG_*l{0vOQU(uzSdFb&yq<xr7tO8ixp?Rk9zW+pN^?u~!
zWv;BX1-Wp2**C>FOnUCg+>|K(vN9Dl9sTfEADE*;l2)MML))9nO~a0U$aj2cciSFJ
z!M1)eqKlQZU;;ZNgGh@nLBl>%st<Twz6{VDn-kXJS*|us%n)2IyZbH(Cd2ou!A3RJ
z+1Alv*-KZ{S!Z%^8X3(xp~S_jQoGS3PqP;9y&Chy<>eU9-PW_l2S^|9gQX1(sY3;h
z9#My(Kt@JJku2I4r+VORs)R_?s>5UoJ%YQ86;RW{W@v-rZ|l=Nn}6X5H&(qp^b={g
zkf3yIxW!(2YI;{c#+w=d+%dz<J-oR2K#;Niy!kq}U^hpo|2OH5THWK9znU&0*UL?4
zS({i)k`;zboSY&=r{`u-@^dM~u1l9nAAXF09#Svj$vGvL4~H!6KaDuXD567bknv7Z
zc-{bK9{xH(;)9kzO18dsw@y37J%!+1$*|fd1v7$i5xOI-zK8MMeud07YsU22b7Wo7
zzv=Ai)8rCD(pbh1TT7F$>(`7oQ(f=6ZIzX@SAM<t81rVR&?GPEm@0cHir}(EaJb?%
z%zq9nY<aISd{rW054X!R5Od8)?ty4<l%-re&rnWv^gHSo6cj_-rv&Io=tvAFuW`U8
z#TdUp)R&<m6;$hZ*|HoLu8g2`8Z?%b+(_ewgOujZ_DtsHWWVq&501(To6D?g&~Tx>
zkw3dxdBX|yM>kSWb)#?MN-}ZCje}6dL#!=Sy7mK2h1<{vMU0P2>}Tza$ONWcRCwR$
z8HwI^zG%`{2+_;+$LDWiAW3n@1qz2W7_%d&LMT3%vQpM3PMD;h>o#4C=Dwz4<Wn|&
zvCQ;H#&LA5X&z>+n#;7Y#t*kexA$@s(NTj#s{cV|<k4~IK6Y-$V8mX$l)eT@Syx8j
z3Du!wlx2Gi)hMf7sJKSYJd6K4dEIWgqji-lwm}aYA#z`1k36kK5^f7(wfteZeW0I7
z&CjpS`yQ8x5JZnwZ$vb>gM$MdAWop%2h02ckBH4-=VNXMQ0TXI=hzB%<3Dvzm%N{h
zJkZy<%o`@0%39ofRRt;19eHpVkP`%tVOH7~8MW=8baPv_zxt#*7sd&{pX4**T!wBU
zeU4H&%H!6y<61N{-E#EiVnu2$^ud7{eO~pq`k)7*cS^@CC<zjFDtNL=@GOqL&7e9a
zj>aENrZ68c(Wss*c59ZuIZ@e2!L|Ox&eo4R;*)9qhbAXsCu#^gt@-y>rm!+zO>T2>
zIb?IwDID>vW&iH<<7iGTQQZjQ&Zv1pqkmhP(RM}B-+{nh_SZQ@C<G@-pcwTq{Ol1B
z?}eS8|Jc<Wv4NCu38*6E4F1#w649qx;rZ0g&dxB@6bQ9)0J-qocG!U2&fF|0{Xvy}
z4o#wv78>0<;vT%=H{O3M`I;?`0?M*>nw<1Fakk1DRyp0wIufen>K<>+9XgPv4@qR6
zJs<j}7L+niqQ+*pr=QMxW0=R@I{v<gzSITiIz65j43f%U-t$wmJ}<1+x$gX#6wz`~
zpTB5$K0UBzib?2{|8z3z%e+hYjU9&}7o@JLX8uG1n#{+s*WUVADRMt%aqgAEL|gT<
zCs8gpNi!ELrfXMW`h_}zJRIlkG@r2H#!3B8W~*4;jHZ7V54mqLJZc>rkbI!CtpA>v
z*bmtM8YxZuI!$mwHwUdp`%%oA-QXaSAKTjEhgMu(ZUEG|y5CX2wyjZUi0tnDOHvbc
zdujja$t)cyb+{0>%F^+r$1L*_jS}J_iKJQ~{=AvG=eV?WIriUCc+`_|G|2c6TsaQW
z50;iJ4VT8dOJz;`LiYNq9LdD=a^+oCWd$+g6z<g?=D%}6$&>S<lURzAp5&dj-A~<o
zh_$1pcjMg)PS^{%IHEMr)dXFp)J?fnEmlp-laRc-);a>^GU)PUx=d%A@}lAT@WZef
zPw{rILZ~yVu!dwbrZ1pNvfnpOFkI}Wm#rIF&8*t$x&JCh;ClPN*&)KoudJ<~-wRF=
zFjX-iD*FNebM_H>9ie0c0FN+UXz<9Dn^*_=8OXVyBo!^^^g2!08iW6iXB+tQ=2N=C
zD=hBg2d56mSn6Lnf`0Qvd8;H_mWwJ8a#5_$if-?{Hg`7SiU}B5w5nhynrv0xJarL6
zZtplc6r{%d^rzRFr;2xsAwO=f=GnfEu!plP6$-MRuI}dA4fD%xl@KjD&M4FxuJA9=
zf<zFM9E&f82s+C|==|rGvropkRD7}W<j4)T*(_olYA>xB4cH%-?nWO8e9NlYw@SE}
z7Ufp?ifI&P#Tx10&aXoFr2aIz<gsqEH77m6I)=6;rQx=`nuCd+2jjoV@uqZ02T{ut
z2pku=UD^m;?{NJ5`4gsTjN}6kig(a-#pfU&Q4-gIF;Sq<zF&8YIalkLsZywA*#1^l
zc6~5KRB7IbzMfK{Z~}%-IV8%+SPo7I{XvpHiw7s7liw>g6dYP((g1w_B*N@v5t8b`
zbbC2Jm%k;}yd{O`Tid+98Gm*9{FXE*%RUOTwbfXd)|aD^6n2Nqq~9P4VG881A>gnl
zsHLg?MBsv2WPc~v$P-Iveqb9fLG7^VXoIJb#UK2}`t05a$e)^{)qOkqkt!V;9?lVp
zM9B-9%u!_ZcS;H<t1(vmV#Sm)LpLJ&DH{)zDV1uSG(>x(d#Jgoq*b5yr^Wj9Z{-9K
z5%*IcUht}&V;0ysLc@W_-QtdDWVL$*2e=x19$Fqz++HOMPhIM@3b1R3G{>mL?pRQ1
z-C4*B{S5Of&hK678@hMRKweG@Eim{RClZ!5sdx1EhdAs`4*|6=`h7<m8+yy7_X-N1
zhKeH@RYkzC-d{~L_J&vkX>Tp}(=ozMpknqTF3#+XEf!LfEH((Wwf!}rsov--Hd0fQ
z|IJv7y+4zdX#RA^OD$rU;Ms=a_Ni@o>lX_<f*}mJwB|3bGdq>l*~rXp;w{Dmzbnr@
zh_HM|ZRbZd$v4@o<d=O#Ohc1i1{I};CP)S93aGqtFr^M5?T>8dP+U((H)#D@=(n@V
z5J9FGCP@cT@|!ZoU7|0h<*9s+i=}GpiThsg#zW#u<L`Shc;L0f`e7tPf)nf1`Cj|u
z$520eA}`s@T-F$|M?*K9QeKrW`4al++rvS*tLXF0OT*r1uedlr4t=nkIHjpYH_0l;
zqq)=w(*!bj$vI6&2^nw*%+eHew~ePhWvp4E=UqUPy5==glo?1oWZ=v&69bgANIQ+2
z&17L&6$tco8cz|m%mz2@>Tb_w0KWIJPDdTsIE-Y<6&?wNb+t*5x%(z0kc5PU;E`m~
zO--vg(V235YVp;|1ZLUE%G|$ee3toiSqZy?s0oQMOvmgmTv%h?3b=$!=6Am3A&`bF
zr0{CIKgFw6El>3aB)`_&<gDWlNtEHxJ5pPmQZ{ja6^m=AVL;0!fZ(=uXFI09yqx0n
zse*J3{09t%2Js^oS9kqYl3AL4YKE=fIMuuEzeG|Y!!n?lYw7I&)%M<Gw92J6ZJ4v~
zp36~~1vdDID%$7t+Gzdr79g#H-d{1!s9QCpht>zljdec8=mG19m7u7&{I!96D86<5
z$hQ5OCgJLRDj0(-E>liZw8CNRiL&YWdu>7MzWI65;Pq)|L2!Qg?W##0TWzSgc)K2F
z`MbZLnw@=kInfxvqVHnZ)4PFx?aWO$zSC#UN#oc)2t^kutp9`7B2+#!si6pkkBNbW
zaP#uqm3b#L+bJMFHr19+KVBLf1fQC_Zy;aEk~TAqi0f|aOnvL$mhU8=0|~N>)e8wa
z&e&^))`!GB@W~^C7WP-wE{9eEjeWO?Z4`&lTo!`)C{BV5^@=$=n|f$5*AeoSumjO8
z>9fG#+w$Y%X_hi5zr<Pq+2WK__`L|Z{`SG{FcSagk$NP-n>wo#tBgf+q1CMju|K^^
zS8B$%(*@lVB0bEX!x8omi4DAj&_Y-6hn2yCg|B}d+N{$wKCX;f$IVm!T6;b~1+3b7
ztpIUgdMmDl)~EV;{~iq$wHvL@!NFm2Mw=fpnwgw@yAL`29kf1}PTZ>q4$B^9y1t$s
za4@unt2>`}9Xor7C-P;xZz+O*>SP)<{neZrn&X8C@aoP+3`Rvo$4)@z=NSAft2eP6
z+-A<-(MBb=yuLp+y2+h0J~yMMXBi}v{iuKSH9m(Chs2tQDO&}m!-Zd2zH?qz+dNC<
z!3;FJ50X{vMYuHg2$7`1`Xl1C?Ik<Xo!`k$(tM4L=M-SvtoJ2GtPhKmpBsyj1lZ|O
zXtx%iJC)^Ma>!R^FdX_jlzwtJ%^c1G=k#xSW@k-j=v5;Vb+1(Uc*2A8eFS%}9CB>l
zcBagInYa1C>u8HsN4<Dn(#hOgM*z}F$QOG8rSJEUOoBJ|L@<~=OnR)Nvy%-K*M8W`
zE~U_vGKgR43Y779aC&O=Gb*;jAU3XTliguIlj{zABh+Zg9GdTrVzf_`Yw5i4V~+Q7
z#N$E${)On4aL|Nr2@5uRj!J&({l_DY8N571A})b3O!3PPd!1Y6X*hQ-wXM(_$cZLI
zx^v)gt?ezT=J5;6J(&gN*7k`6+18rJ^```R(*CaMW5U02N%HEyO^b_O2v&X+?sD0>
z41F!-bsi@TJVMJ1aq?9(yq{1kVVx8@blCsSN;D7DkXhQ_>|Pn4|Ca0NVE9Tyh^e-$
zxl#T4Y9EG~KC&ip=Gq&{kcCp_wXi(OvK6tpvkXUk4VGH`^0XTpx(k7vB>dv)>IaXw
z-g#URPUMaA#3hiyDxwD5Z=$c<0y~BK{t43gO$491-!fgF#!lL~XT4<{Z#T#}D+t$Y
z&Lg6>$QHNX@Wcdl^!pE4%H~)b14LTNdO{21xR#c~(M;Vuw;XRC!tiP!)z^BbcRE7G
zzyTG)`y@R}!)a@*%j)GRz^X@?|LXOLBEdgk#RH?4)RJUAv7e%rLC|g?Smo>aY4J5`
zo^(h5P@)&dUY0|2ed&WrS+e{#hym~D<dk1l;qlErF<h(RTyQrbTE$;)h)3kHe&8YJ
zOR5$=+vKJ2HHKv0AyEi79_A}txMq;p6?QUe5aITsd+X3V2~Ax4vo(hixl#UG+^iFv
z?xa$Vf`W?Jvn~7tpLV|+SPM+{w%!-|8mxJW-%4QT;^A?tz7`kr8hcscoN{;cK!o)n
zkJa7)NBp!Mup51u926iOl<e}f&Mu>l#ytNB<G_?Nz86A#zG8mFrC3IK+d(<2=G+p3
z9!-WtWq;G8XN1%n{6Od3Dh}A3$9ZnnkpmtJ6fscZwV)YoHGG_#LWp)JJUTf^5b#b`
z7ybVZ2UemmL)cVazSAD^;ZI@VsY`E@u8@mVTElkXCNG!dqAiM4>ooWOv@*$kQ-;3Q
zCDAmYWs);DUXgq{(mkUySkFO45_F~^uRP5=B<P&DlT`jxGfb%W{aL9J2`4xvz*RIh
z7P!4m%J{@dFjZ~rbJRD8*WIFDD%ZMv{5?C@kmjkmbbv|c(yK^z{_1*)Y_=$#(zUl~
zYg$ifSnz>T?ub<G&xPT|uCOBE+$$&*l49!(QZZb{BYJk(Qg0XdM}FJ-10mPSg6lT`
zNJJz~%EfyoDTIG=RzgJw=hH-96uNfb+4-QQQkrNe6?QTN%It>OQ>Tl#B43Ewvy-Ha
z?B=&s?nz4wL^P6f(RgV%s_kxw^J|e+nHsm72RAyF=NhQC*WCgStzN!^gi{zj@%E6I
zcES;~UD$~GynIFAzzpu_cl8|{sy7RUwxUG4{dwz6drqu;bZFR@72XC*wq%AiUKTUj
z9a<4Pn@kP%;jTYqxMvo^5~gmw!l~*)<m2@T$XAH3(|CwnLS9>TF*ESM5V3j5+=(G_
zM3^kK=l3*PT5xk_m~beR?;bcd;MUtNYTdoFiU(@^Bx=(>`Gf`Gv_acq8(t{Hv+-4C
zd#!eaq*@y*0CaRd*p9im>8K=DEk_gLK~AEg?KDY4#3%Ya+&bmP^YZ~8g=uiC*Feb9
zIE}cp8Pb1AFcluI6RdQf%czzyqQtS?DWz6v(Ff2#i0A}tgF{*GsQdSX(s2@8YSi&^
zlW5|5*LL<4xi!4-7c<!0H#r&O?OEI;d&l(%m~mSCTBKGuD_+_WKJM4NcYnk$@j2f*
z7>Q!zT-+xOsl~Lxu2K0`;5l<rX{{WWiP!lKS>|0ga(Mf)#o7BS0!wL?X|9`k10S_?
z9Pj;e=sq|2t)*iT$PpzztWaqqGdJwLyE}81wG?VeOmDQ>oaDWrDnH^Ob8MkOPrgrk
z3@TEi*E}!JDrPUwe1LGt_XIMF=N>9J{t3?_I;MC{Hf@Zskc6C(0VvVONAut99~;Sg
zIfAVXLn|Jfe&toIwH~)Siw$<XHi61@0^!nMMI_j@qsC|^?ZlVQ9yfEsEjx-mu3UK-
z6SxWIRB@Qm`kfWw+A&+D_vVBcolo!~ADv(*(@@0@Z;r-8^LJ~7KO91!xj!aoBQwoq
za6DniQD2`+XO=}7>msKxcjt}f@)|dFCbs-?`<J6~P4kWdBLV3*ST%2;6JU-J=PPQh
zPo1tuJG8RA9ipB2vTSD<DQ|G1aZF}6`r<Gwd=>wP6>Ug-PYll5>dK!@Z`W<Svdzc7
z9m>oVPyVfd(RcG)03Wu`Vy!Jx*;zm{TBi+`k5&@BpCWoYn4bcU3R1vu2M+Yd8n)9x
z0txx&#JHS?wjZDs-fShQp3oX*N|m#kp8m!P)H)UGGl#U1F9$~x-og184i)1AWBro3
zj1APVzyU_ou;`Dbk(yd;`4M<6kA=A>+bte~TaZc{ytKGv?wq$B-(~z_bJh8$9M$)T
zaCmy$J&(i-zlVzxT*+FuawG!Xi4^r~fe{^6=F1z<pF-P`mhFNvmcq2!y1oBCq2&pf
zuXD4rc@e<Q<cmv@<8Z3I?JwGDY*dVHs{M676*$k@#;Ayr+PiQZoYM241)i#9$D?7J
z9O9LoUGXGI(dx@3-ATStq1=hC9$X?<^^-ucWDfe<pMCbh0-#}Cot9_=HA1YpHv-BG
zD0>1HD!_HrXn6lk7vHMas*_dnz|%AAefFaoE}Ro;p6>>#d4e&X!?BCvc1y3&$uOZG
zyJ!<Xq7rjYzSV=W49bUwz4HKU{G;L0sR|t?@bkc+fMwI*-0>`vg+?@-XN$&TVk>-X
z@UG$Wm9J&oo<Wypt8!@95ZC?yf=_nWfW*IqpFH*f-2p0ikn|^K^W7cYl<AL*L-cs9
zsIfv{&R_T;ePpWKZtudlRu$URG#KB@PdpIgc4EOHwRiv$sKS7mrk3{?jKmg)l-yjP
zuQhrgTL#%In9RnN#b9MddR<uJnte%xTqxnj<nbRfL+tGX2n($(KLV_GrG{Qz&SpnR
zINxtu2JJilJ&5HF%Ah<Ha5*_6MOHT=Gd8^R@Z(gilaZ%r?~OgN2H;sQVrsX>v@3Y4
zoM=PHjNC(p^u03?mipDcX6Bb4*hqX#`euEIINpDS&JD5L)FLX^N>A`x8cr<B=*PeF
z@5_d;mlPW*k57tL5#U4<NzigK+(W?b?~~XNbZXo1L5m-)*1IAx3Aa$v-xl#A0CCv*
zKnCMy=D&$(a3aRH?3nbz;u-~<7{4ba`ha6JGt(T(pSd5ht*zE9W@6NB_ELs_2UIFu
znQf~aO05{86>Abatuk+zd~_cO`JwqeJllRZ!GFc6kvj2QG%%6t{+v#{9zFwcRBM(g
z4WXmVH~Vu0!2nZxa<0c?u5$6d-NO5)5#kv4{8#}c+ll%q@bh21e!bwsL`fSdvanyM
zgtsS}3H6*OSu%(34OY^UwlCjCWt4fPXVjXdvTu3hUH<p#jT^opQ6NJq<{Oac`oUZo
z7H^piEXyKiVh#lPpta%gV-NRh{&}(^1hczSI@Q2UnS>H-L$f1<LaoBW11f-C?WqM+
zZiG})(?!)cl0N8pYQ|EPz~BjqH8(H<0GDIwwle}@6=QXgd#0-)93B+My+6ALpZkzn
zfP`!1z_Hepm=KyQ<N#0?vJ1t{^|xUKQI#HBoG2{0B;Pb%Cr9J<D#TDNj`sHv;(dZ-
zvWkYhJ6*mpnTsO2ef+pYOFcmchUn4nU>k86s@VkX<zEx<FGO;>%B0%DH>7b|>aRUq
zFlTq*OT8CIEDl6jE^7FD(Y#!5n))Kcgm?A2{!Bn$rX2FJ6)lxPb-JjubOmq~x$Q86
zQ@YeSdHDEJM>e|~xnp`iE6{=}OU6MhvQBQf-dsG(Ad8%DlH$@!vOg=B9%@$aMKdt_
zSDVRHbc~RqlVt{U_HGYA<V=XW!XeajmUplrGElZ~&z3qDjPG08cNn_%BIm-~KVE#l
z1I6!5T^!k<gCwW}XN`AjEki)!EPY@oTVv8W-R-4=&OP%xr+aHqLhXH9vc%(3!)h=_
ztSqDbZh5Gb^^^zw<iiz#qYV<4x&e}n0^81jZ_>!36iS5}Kdhj_{8%vd3g_=qdc&a<
zuGoZsUN3qzrEk3Ph{Gp2O|ktoGxTy%4{J_i$!q+;{sq;7w&7Zt*lhhGo5=p3InFL_
z4!(o}k-mNLz7#f2XD=17N34lRoOe{(npN~`i=ITL9oVNHDIp?r(^xn!J6s9y-~{gm
zpPM~D{zPbEEjic$@10<}kitijkP;CY&Vk=0Ta(Htr0fF*auJ`fY5Y<c!meDQbhDSg
z%<&G3LVI8|`x=z;(i+5#RQO$M4+p=h?x<Hp2uGTy8cQJU{LXT4i~1o*z-LRW<pr<~
zyuaZf6%+6;%o5de0P0`WoJ-RDTJcyG;Jl1*-tmhGJmY<r2m`OQ!RGN&c39-&>#)XB
z%aM%WGDENm7>uFEtDWv-eb2$9|HC9vT_o)9G&^VdKQh7WD5KQTSF1CJ*3twPk^tYw
z0$B_WveOx|m0iC&DCvl}t3TqsMYOXA%MWC_rc4J27Y-JEw@E^x-7#-cD@m!AUNKf@
zs{)`K_8ts_Q6ZwWj|Et9)RA3#j?5ks@)JQ<!i28}0v`m<c>Ijs?W|U)u<|R#Q_;pk
zxS~Z<DSLEewq^I%UhxeFe)%SxHNTvQ8M1-b>h#xI6{2Yb1?fR3v1y<%U#B~Vcj8@K
z+4dOFM2O|{BN7k8cUoCliQ_{Ar3>XcAN3`Y*{|vubMvFw8evDnCE*w>{5?JqPzP<O
zszgJA`pQgo`C4bA_#{6+wu^*ib-2vmfWS!-76<|j-HcqtgG?73$DY}3<e6bf0hh>0
z5}H4(JiR$|<CMTq4+{$mP@{lRUUNsr*3UV>q^<&h5sdl@+BH<uujLX2x|Z96F4v;8
ztK2Txfy&A(u<sK%g+>*fT$BarfJQqr@G1+7j1*T^#sfY8MxZ`SNlkqOShB#eh!xml
z=BgCJXx=*acegnB_^_7!&y{l>pym&ZkfqbBfDZHxP;k+xvVs?G?0zbgnwolXu@X8}
z=WKf!4a^J<!HcJO%9q95G4IT0t9}4;z#^joh!ngt@C#T2TvEN8i^Ht3)5fcz_-K3Q
zIzn|YgkbEybbOeP6wwwPaD4hbUD0r}W=eS>vy!tTcyjB%?BA?wE$d-{##^Ov-#<%+
zTG-{N)D_;<)^9!wOlBc!(a_NR^E%B1?v7Z%CNIxs78C-8Htm3U7F6D{OC1&%!(mhc
zI2fpv5@KSHadF$ByJIqk{dt)}t$O<Q(d|<}p8+xi&A@S!e%6SUfRz9p###qRu}T$;
zn-tGu?F}rQ05wED+fzgHUOjE-H{kriY#QD#o!M^m7Xt@3JQl5YporYi(6DwM5gv{W
zbhFQPp&5pqQs-ks&}jG$d>CMgA_FWhB1DD6R&V^wiObKdv=~<Q8G5DRHX9KfzWmkk
zR}`&#kWswn%z3kWz~a1U_gNT?PCM0a+UzR1oF}bQGrhvk|4vz7Y~{z5$G=L37FtVQ
zxzD~%yJFCya5?Itf&W-$EFW>cvAS9W0sA#z`~`^3%VV}=ZU^vac#JT`f8qO^Fwpz-
zO`HOlFuH-?2@elvH6DBpNQyavcVHZx042H_%p~!&#{h-|<&{g|Xfs*w%1usA9?xh0
z_2<tg;Fn>F;^4*ob>qFLQm6wS7$G;H^$%p6ip{6kOG-;2Ho{==*4R&<w#M_?()YzD
zu<0QIa&e|XPbBc)#!rw9038&78ukzg2_{Ou*LZ)&x1kdpJtn9n^E!hVoa(DmF6=RY
zoxLvKPf+z3d#3CBB2UKuNgDg_cBDFKb)H`luiiBrl^;F%cI+d0mn`4>!+eyG%OeF|
z#z9B~`j>Y7mU^f1a$3FUJG*Bo#>KSa_wb;CqAK}`4gyvkIJNvp(1L#m2;eKL9}YDD
z(*>@HV?aPL1uhnN#DZlM5EX@wj78ek+?;YFq@tn%(=i9rT2A2b;ey+-O!`!#fj<^m
zKaA>={m~Ujpu+)NTp<P5r@kH%)U2`71%g!eLY;sJ2kbEr;m-CK<bd;A_kL-4c{*Tk
z+?+Sw=NHLINuefrT$wmC=d0h3?G>n3S`Z63VFMpBP%@SPW(c4RSiY|N-^jpn9xyUI
zE<6CfX4T))F(d5dLtKbpu5$<_V?^;8s}F|?=SlY=WbyD9`*|ZpR3tVhWxI2a^VR2A
z9{hKx_^AEbaI^OVH$$%hWAG0s{c5TJqq(EQAJESp8~_K04>b^AjKK+<42t4f+ym=j
z@GyH51#3mD&rKZ{+_~YlCkhMCE(+_<)FRruOYW+HY-(Yv>Zo~7Boo{>Qo$M#Yp9YM
z+z6r$OjicA*g!x~&>8q&!B3W%D5kaN?H!t;8a8gpCGh!meGt7D;qdeKR};(t>%qFb
zK~em>6)PTIqN3rhgN=U_zo#5-Qh%=JSqLi=Tt~U@gx=x@=C-v5gZn=o2glX@&AG5s
zpX8fu{t6PPUu(9B5X!+UfP#g8to)E5)ddYYy>Z-yb+l;T!^1z!^DXX|O7<8|_5=G}
z=>B>^se_9!A{Q52iPCN{Kd|sx+XehOqpOom0niQwAhSf6dpz(+cm_fDE-dB_z`?>3
zbwfF>q~$0}J}rlY!|lRq-gRFPlZ0>7m@U_}F4lVrasB{0IG`t@QqLj(+5tARwlO{|
ztj?|xzZ-^X>~&O2N2F`M8{K6ytbM;0v`j#v1E9Gbft_M24dcHAJ?lrD_9lWhir-+*
z+|8?Y@wapx{gOfTrQ>{EW1^t$c>v5hfSBU1j0|8&`Jf3r6yzqLvdFL-79WpJD>hpK
ziCEa1sy*!O?=R?{5fl{Ub=)InsXIXEiDEGVEzoer<!Zq2RTZ#MtB(ZCKzKlUuM1y3
zt*@^SH<&6GWZD_dbP$=CIRD=*l^MgHO{=7UgDl;i%XCW*$+q$=lU|{}3ZHiX6TnI+
z6Y2)70{<q8kI1(ut)N#XmXq}{k%)Zp7<?43ABZHQFgF6zAwUvC^92042i^L=OU}@h
z`rJ+r-s458mz&~Jhh=vZqT%O2_X>)&@DK!nU?&dDXh83lE-)ss+nbF6p7z7QBp9?F
zse?X`qa%bn`jc@n)s*r)TJ?y$ygbxHB8v6$;Y4^wi1t8SLISw?S)Q_h(_d=J?V-F0
zRyy)b^h8?E93JWWE6&^EUlxLi{~d@hGG#(_0K;j?7_)6X4!`Z8TVdY7&{_rQxO71x
zwQT&OS3und4%QI`GTB*|?R=$vp53WZnuCcK<`|N>V^2s)`vEx<G=DvU?Y2mek5}cq
zORnto`JWVlg6;AA++#M&8p&$Yl03*#3FHz;uRtqQ9HXHkd##wLj!rVyF>EFi`V|yK
zUkIl*$v>c|@591b!jOLYWybl5`wvW_{QT3OfFlAs%Z>)ArL6-X==hjup(XX0RrsO#
z7H_MKI$wNmy4ZCjfYG4Z*GGA1KrGeeY^MwG@4f^FuYr~QPzNx`l|q2YJ0%=gtgx0d
za$4G~Vg&;r?hn`2);4N#^W||jAvGXBu7Udjh5+f~?*14=NVFUHfN7RpFY4g%5)GgE
zVR-w@Rx0_YmsBQ0&jI_&-v%H$=u|F$_LT31FA5ByOn)mCaE$vYn^>vl@$af19iK$=
ze2*mm93l1tKM_$+aNL<M25mn9G2-it)+Ms3%t&N!_cUj3l5i3XrG{`N?)&HF62L=8
z7|d7a2JR7J78Vwlx3J(N<Q@yYUjo?OE*nQ@GnOzxAz(l;v>hR~1L6}3gn|OHfbx#n
zobJ<L?s+(vQ4SytVBz54z)3-i;0_KzIZ4@gX>ntt6Lj>!y-E^t=kd7V`5X7%0NyA_
zUc@p-K=uO@DcTN~5N$J;G8_P4<RkI|3Er-B!D=;C-Dclmg6b~S$j<-mtZ($pT0!2s
z2#p+C{%GXIV7EVjUUyFujj_e*DAF@Fz`0Pg8p$Q!dVcSat|2ri6^C&wVLnX0K0c!5
zFo7WF;o)JY^Et=r{f6rtl5@cNALEbu^T$}rWi#D$9}5$+z`?;<>F#zP0zTzUuns)X
z&X_~2sLaC*lcM^Ms|vx$s+>ZcoMm-J)|y;zdNW>dz>==n_fn`|BQTH|ZcrJ6<>R&W
zo&T$+`?dp=n%#MgSgp(w1-WD~g`#NHn_X1J=mS!W=jf-&Xas=Qo)4t3$5sKp?g^x?
zAGWUDF*Q{!a%c;7ORf;u;s#F+G{FHg#CAxAd(IaWsoLB9fUzc^ub6}AHvp>c+X+21
zNp3A*N<|AevwL}Y0kSQiLq8L8;{g+rf}S4rukDOt?iizHO*m+tV|PB%2cTxKcv=c*
zC=md*)}Y}sS)u14=<WoLLmx{GF<^eYU0o>P8$d}&wfGZaX9z^n3a(9SV+@-uPAaW$
zact9WZKTRfVI=c0Y%-<;Xec81UBX(klpasHRsMv96xpJph5(L)tQ7TC3qX$MjE5Si
zNRR@SaQ~>gV*dk>-b4t@Tps-1!k;>9mLtA!A@b*n((N-?lQkluW%s-*Jj1xt_4?*^
zs5N~8p<ASW%RK|JbHX(V#0Fs2rP}aHU9+vDW8A3`5L|)a@#&tJdc|8+d)N|%<q#WA
zpk>s0rGw34mM3>jN49sGv@l~3upB@l2GdLg>G)KYwVvNRS@|vv*abLo2H?mHC^~v0
z><jVo+YU@O6f>8^rVwuq4+o$f?FL+T`7y-xzI#<MK<z+}<4yDwn(!$64?8UoJ7dO#
zUF((Xv*zq1tS7a%4^;rz5*6u_P)Y&N+J`5bZ#I^e0EmK4k2b!6of^pX)(f;-L;=E4
zDG+q>e*`NyMfX*P2^TO4{toQ-4gYjd&doBo_UX$&Csc2bwsusg2AUZLa-Ics$qm4(
zbD^Jj&sz*Cps8``|5<#;yi+5x{KKi?YD2(9j!>$tao_9+k@VkwT%}cAagw~h`uHS4
zqf)HNw~AEtmL#o9AD{`TbyYLDDl8y4_^aJcqyQQ>KbUZS;%V@UQ$|S!yVK=>n>Y?l
zZ^Cv;-0Tm&!MmB}g;n|<@r6&}9}$gIj>}!^RVL|7Jva_u-W$X_P~xc@g@56e9LHd1
z<9|oWn$F}lj$w%C6@-BtX}MBs62waQZGD68--ThPs+PKVIf8EK8<01`p%bur*1Mjw
zn+*Rt`o!E+e?Bh_;sof5fIxtSkt9H@Cg69-D<6SPJNnIg<A-W05*J*!XGUYfO-V6D
zR2ayPG*@FKk3?N=3_P69o`QxC!1SDI9&9PNA{nJ8Si&Hif0$q1W+B{fX}i7PD*Z39
zC2tFE-a?!Q*>r^EK$38yi0=750s?{)$kss1<O=~v09bqzn`Kgf%l-gB*449#`|P~L
z>!go7%ApIw5+%A8#L+q@L|8ie3a!h@)K7`D)MO<+9=|&YIP`cwF|OrtcA#uzRiV2O
zrCw#XqcLauU+4i@KoJ-Kz{FddNl8gj3aNG>FE3Hr`ug?j#c~i!UP1_nWAf=xf#Zqp
zVwj_d7yNp>@-!FD*dS`kAY9MWB|nb$#ySV3dVso~+o9LC`}M_=4r7_MXMoy-aF?>0
zQyCN7%)F-J{q28e9T84fz<b&ShupQ9{{H^8nVj6Kr?Kxcj`r*8>bfmqluGCI7^A$u
zO$+kr%>%Q3xrO{toYp<kiVD82Cj0C%%2EMTF8`LnD~<O+0!e>ce#W?JWf%ETl9hD+
zVd#C8%G5=suyL<O1a}LOGNlcu?X6N8El&@(*75uoxyWOwa`BxK;^S%Q>2vIAlak2r
znKj^&!B^ddtNjuPF7@NGFsL_6J~Z$FN>w#P(a3V~L$Q(A8-uQUVgXD;P})iv2nK+M
zT)wpoXl&t^^$bS6QhOghmZjrKa|7_Ukg$^~AeO23Uq#c`*a0vjF)%TyDJULzE-!xq
zKmZ(&LSI9i2LwStc>s!XJ?Bcf`SWbBtIUc}(UGQhV-2n=GyhZxB`w;FBHH}qfVVn6
zZeuB9f^B7^E#>`eXr)KA-E^L>B&_lp$lMd`6g-Tyc(uir{<T?vTot1aOf1mVU23@?
z1jo+KesB(29V=!mRzR#l&;h~lT`VVxXKMgf)c#DS9NE7)HX2h}Ni?@#?YDP6D*`GK
z(7%#Iw=|1pRbT7)QJa)rc<2A75V@49d-qO}ZfkcX7OB9;20`5x-=kg7XIKonbUiND
z@H;|C)IF~CLIn+X?LhJ-3u2aJa6{Xr@W6K`p2S{X&UfO9yZor~82mJAsCcLtk;n|J
z;T@c%lkt&#>3SO=V}gGETG{9t_+2jW-+l-901tBq7D-rjUR|EyYr8c^v+2Pd93CzK
z<YXPRP5%PV^5%>kOzByWP5*tEIR@;pJms6u0gJJ(y55L&ive#Lq==h5jPV|d;)}SG
zC4nV46Abv4-0!oWXA`{hj4{iDw^H5wyuYOmooh9S@gnYAn=)iVyl+f<)1}TU0paw|
zZv7uY6IzFmb=;8uo8W?AZq3cj<!b0-ECMs|Ojw+|-YH$(*oayX0?vIx7mE)#3DI0}
zP|4qrq{mC@&+V|h6^clnT{M3{GS}3D`8D{v#v3hj%m9_W)!*mP;5HfLL%_B{YQB?i
zRb5|24zzWRL5&@7SFq!-ND_LnDumHIJ%vcNId6XFXx`ve$ENNetobXe8E+)l_1m6U
z6?m>HqEIGD>Fnb-TF*$|KR*eXU*CWQ9oY7h+6B1tXk+ILAF@Bd0x}@t_yK_}K%PH1
zJ>{xxOu-g`qA6Uq*~SIVQ2_lh{m+>W(Vr!xmr&fNn0z0Z>}X{w>&5K_Nj~euKr&&!
zuqf#fO6CIo1PNaT#CiKJR)*ho&yCOk%^q}5n2^jjbfi%G{7723i1b-7S24traB3CI
zPtI>$Ktk9+)mfuDsx&%iwN3i&w*I2^&N99P04smb|1vwLixM!Tu7C(bVKF~gEkDCR
zx_Vai(a^<{(6-X6>`>M}=m$Dx7e?D#ED4UKe$YkhTT3S<f9cNtV$-V^q#8oN*E;Ox
z0%sy7@hNig^@=v_>`x>>l)>M)LS&8A>2hXmW(o}zNzbmfzpFU+`0%yXqT)abxvC2)
z-}0hQuBWJ*gco;eohjbAyD$R4^`Cp=H@7zb)<G)#jZ;=OM^<C8U>x5A>$H-q7iQo{
zO$dC$*}-2MpFOa7pDO*-Zp(R=-acmnfA9*UeJ<B|s`&={k;C%xl2b)PZAP;#Zl4c~
ztDH!b+Z*jJh@=p--B5}GT~NY@tb;lFGx46_!^%Mw3C+~);lN^+yjI~DuH$_{CI3GT
za^Q_uP{7J|V%on)EAoZf=oDcehLg>^Y%_ts`ABnMW82yu7<d^L6^T*l(!$etwQFAJ
zKf&)+AA8$V@gpL7o>lUzVdliG&m*D<tb*{S5gLe`zVa;L>o-;w%yEs5`F5{BvVpAP
z`_e9*)k?<(l?LYm!G}B__$#Cm;+F#9SYsxiR1?0-xZ3?oV)@?`V*rcxMcRB*gMs?%
zG;fLN@z3jf3(hmma~=ezNSJQ7wPUt8ec@8=p!4bN#oU9%ub7EJ-j#YwPM8GHo$^~5
z7B8--C6LHck>HbYQ@fwC)FGHznBg>D<CqjnN}+x6(pq10j>9UmUtCV%VNT$&)I$H4
z==ov?@5LIw$R62*D=Lo|+w{Z6pjdiO405Y(!0;|xN3Q8ZZm50(aQbi2g0Jy@J+V(M
zE?)voSa0G;Uz@8JMczOp&x$A>5OXf-W%gn>rh?APbP@5>7t86_R)^MwTppn0-rez8
zXx-fjzB_=7qJC-QVQM9DyD;Nm@2mgC9706^lmbJ~o=N~r4En|Zl1r(y!~i{!pczZ#
zSXR2=g;})?c|^;O8}aORwQ*#Xb@m6l#&cy*V`znn$sC^6uhF6eEDvhaGLkJ1b;Mzc
zw)}ZnK<1(OV2k)_xOjNN5sSOWW@imEiTdXBUhIE~1=m0-f+fJ6G8#`eISA0#W<b|_
z)|ir`7b{`#uk`fCAd~f&0l*-WB$yXD5pl8U<`>hdq`jS_%z*tF%Y>*UZ|MMqoSh_c
zhR=ujdw0Fjm4yep07$t9Cu7gec)7XMW2(15FKy3s?iD)y?`{jwKy6r5M5KF$;X^ur
zR!_fj>vf<@4?h<9V*1Q3olI!zW6Eh&aL%W9<h|(%ON;BiK5=1Uy5P*2-tL{wBr0_e
z0kTfd^1s^v{TIfZ6KyZdfM3NrurYkj!&8;5_*jHfv41{2s=z3GRG^4iO0VkTqZ4l(
zw(|1M3X;py)sG*eBPJ^TTL~JDjIqtGAWxTKB5Y}C0e25a%<q5>>y~Vwk3f}>kf^Jx
zv$D1>ro@0nUpX4cU`3yKcGJuwSwIkqXK4d{UNkH!+mtlEer|kMVR&&5X>4^oGcUbl
z*WngPb}Z9-|JzSRI5~sVId2lg)96?Z{Qdk?pJ+WI&!aYe^@vcN2>qI!d;v~Y3A*@@
zo1oHUoC-(_A576$0`Oa;Ky8x3bOruGu(bqXfB%_Oo$g#-cns|N0sl?pi@)J19tKEM
zp3~6<0%Lo#08rQ|1zZZ0`ZR7Ff<;#jNC=lmSL_8G2;5^|r%-fbVAa(VE5NzHj<VGH
z4(ii9M2-;4G}fO_a}WP7ToD8r9w@gD3=B+@`(d}xBMEc(XJK5kxkWvkn@d&~RAY?x
zyoVC}%AUYVr=M9ipD`DZfKD$Md!w<g`pF#;^1fsK0^=zoNpBD`EEOD_RJ$E`@NenG
z`gR#puCEGV?#{na5J6XcMrTZtaCme`Q+EBz1)adJG#62&8O0jIINZ_7&(9ACex!j_
z3dry7dc?{g#kM*>|1`>Q4wWvgtsS_a*Ui4BI~niL@utGpk>&$1+cyt0g;pAdH-e{E
z8`z&JlVBW^n``vC>Bj=*W(#cP1X_Xqqn}`F)%@LpzZ`f=*hnH7vd3K5ZksAIdIbLR
zji&E3*E&XdS>f<zRxsF=%o#-mYhS!y;n8-HOOQW1s5^{7xhf;&W-7!aBqlQzOrR_S
z+zONk0N{HY%4@sMYSd4B;)14f4@j75Fi-?OG*kgk+I0!t%?;kh^kV911ls*qZ~HF5
zxM#EO3ah%P;?`;r8T|oZpS|;p<q0l0*|nCFrw8hU1Tsqd$=uEUIMLQm`th&_Ybo^R
zoe4fb!|Hz<xI{D-zM>w-a9@+I_f<#z@+GL7S$Mbay3V$BN%9TaMt7`t``#Q1Nm$(n
zyfSF*h}b_Ktxt&;x{^@Lw5a&<<q<IJ1$8BxQXU7z#DUYS*M<jpxVUUKt8W1<TxomK
z+Z$otd4mLw`d#!ZQ05?fK}GIM%Wb~y1l3Jn7Hytd4&GZm^N{k|E5KexNUV5o?j8Pu
z=R3v*XzJOWjdjJB>r4xFOJyj~RO6N)y<2yFSy%S$>GnC@XM7=P&p5&7P4F7#&(*O&
zl!r1(FaRy+%ZyeGZT_>r3#ZM#FpD)Qii}KTHv0ijrLn}lr1SySGgXrpJ}S^FIRw`Q
z>B8C#n}Yz}w%yCGQJENPevZ(u4bPc4TG!)E&{e~*5!YHAcjQ(efpH{j0if!&_BnD<
z{{!HD%BKGt`<u94dNHh{ePO=Ay%t8k3x46&!t@Yhr@J4E+v?MtmeiR+b3>lgN*#jx
zGuG3WEgh#4`6fj{q`pNEp*k{uNBuP|Xr0jSycV|h(B`~}S2Ky1N~1Q`QGu-)W!+su
zVflZvKCf4w{cwp$@e98_-2NFi$4z$9V9lt2WM^|eRi*6ZPds?Rsq5X5y6a`(=<_k5
zn;CnHUwM8-@ndLp0hlD80N4YEPR#S2NxdRjId{mIQwhRz0v+`q8|r~q4-ix`*$oG2
zn-sIq-ZzgGcnepP7obb=`;)l0vNm`XYU;B$b;c8!6k1+PaeOl5y-&a_!eYJ`KH{+9
zz2A6N!eaD3IwCYJfWsnR9UBk}FfC+d5@2XOxOin-oe8;}yU)0?iJza{jb5n*)v9c>
z??7~@B|mxYjx~e=-F{BsTbeYz%eMqt4t-mbGR{I16uEK|*FeUnrxKs>qo<nYF?oDK
zLQf^AW}5uX_?1BkcN*(a%7!qq?e>V{prh%-rKd=5tyMVDWVM)m=ZE6%LgzY)R2@h5
zhaQPoY}2)C{vF7nJkVCxa0#73d;E1>CHL1CG%{g(UG%PHJF35E*&lwu#`mtm4<UlS
zKA#6V9@fcOkIvIxRK1U(Uq&$S`AzG(uOy{jV-$OM^H{AJ&p0lebM2N76HC-abK_Db
zHlr`osA1#R29;#g>|M%@sA-hzw@D%`dE;Gz&uX`sFAFtS_nr*U)6hs5LZzzUC}A|0
zwgrOE_bKynfCNF)r$*bVGeFxYrq9qoA7CKOK7>FdK9<Xjx^yZJl;%n=c!Jtfv4n5l
z0~kZ`{rm4VcAKfdS1C6+`4#9Mj|A@nKw0T;P)4;}0;tHn2FTY3P|(H2!TAMR|Nm4j
zc<`^>GD>63L!evQboKWS)3KZgfD4S~sd8YD32DT={7i3cuB)4Z$M`WBR>fpeD|-(<
zhmhMM56Cje0Rsbp72^4pT?QvPnMv@x)ZM16{AJYL-0$hgd3mSi*5D34Rfyz<<Sk^L
zP|$Tt_9?vj_`PLIEJt_4I0oTnmitF*O51Kg1XOig#2z-hwr!;Frasnmlvn&Um*^Zf
z7999WRv@0P6+=EZSAOGzX*zm%g&Mr1$C=$amR5dqd|iGrVPd+F9vYV`P@0_l9D|tW
zeW_vJaCZa)C>s<igt`KVW6f`b?SEVXrH>!8bX)xs9B!vv#?|tbaq;j>r^_G{LLs*c
zlgcUcmy!w&4i&J01gQYtLWRW~9~c{1P(F{jI^Qn@kzGMJ0Ssu+dV-*91;8bh1)^4O
zaP9%%aN0p+X$%qo15(AGI}Wwt!G!53PpRV_Zo%o8%Vd~RCYDn&FfcF&*c%;w2G7|k
za0QEm?^m{BrYQt^(Fg(XlL)X0g^gn*3oIura3^k41BGtz?_R(5ipzyx3?9a$GTUr(
zp1yOfW8}T0vN%8Gpj$6-*aF%JUrpzVT8>wZ*0%}=*3DXXaJ%`mO708Lk+{Bn&XB&h
zjZo8?FMgxw0UOX)V?`G2EI)sBBp_Fx%;)Q7ZW}<X7dWAgLV`!XDZuoK1Qm>rrcH4%
zG1&lNRt2TpuR%c!=NI4*KLIdVF0fOKWh-u!wMPTsISI(sWdpF?1oU5;PL%*jssQCe
zt^Do6My>t&b1lF92{4xn14v!og4R6we6{iq@82gs<1qXKib+GDI-U#r5d?bt4M_Jv
z?RXr()A}#txB!d`dzRp{OVslDtQI|i3YS7^+ACnBzy(^=H65l5a^%zezlo=nl$4Z#
zM*lhMP7*>vmolK(=cdM>YPo#o$t@tD0S5eMJcLNAmg1VBjDZ)FeqjSvz?>!~9LsVM
zf0_BP0Mo5;BDC|eR{!WnjlXmsLi+o#@XryLYh5+WqWtTH6Csz|V{2#gTlY6i?6KYF
zHWkOA7_ryA8{Iv>f1v+VLzd1L{Y6Nf-Na0V_ja4zj>n@gLzq)uhS?&!dF4KBQB60(
zb^{(Z?zoXo`h}F)hATlDgq^}w8Qx}`%4*#ns;Ec$&>5PQgtNhLy{qHp^|cuwD#WrI
zeD(I0waf*S00GJ2v``_J{>ZHK>gBHtoCe+DPm)UH-o70K@*4*1F+dic-p}v$xZxaF
zZ;SyQk<lEbcJ74Q-HI7b05vW{*VotEd$s%H02VH=Y`7E$(nEt>6B>#c#Q?DRs&Z2R
zY1OL%y*<W;K7ICVKw$uL(LC@N6tIh*On_pmDG;u>0#E~3QY`L(1pwB+hV#Bw#Nrug
z9S9~24H}V=Xqg2UyYuz6wLS2QR6ELXz)89o4`Nku(3W)M`fjL2SwAjSQ}~1PR{>{P
z%E>3RjZ-Y{A0qDMXL0RUThQLQl;)-@H6(MVIYJiu2i>tSkIMZSgZrTDRb$7n%;_(F
zqPKEyq!;WWwJhkkPzSvl9ZioGCnD--Qc6e>{?2MdgJ0B=3Vts8cX5#M?nsA#wwt^A
zDJbGdD`elAEP%X2a#-+uRx#zU6WC6hqK%1xwsQXfEHV$ZDoFhoz(zCI+}BQbN(Ulc
zZh_nhl_cn*r91uz-7L-C?8TZD@G{NXivc8)D&j8M6|EAv4wM_18{ACO);TT#t0jnK
z0UUq#n2|lBqJ<s&D{d=fb5>~FC;OirAQ4InPwr66VN=PNva~<k9GsrJ0u*?gNs^6t
zqwN8+GH5Yh^=DG(i3r+em+?k?I@{YVJaQRxmecI)-ZoBm?j4G?bR(6UkiU02%gp+_
z8j{lW$mr3-UFYq5^oZ}?wFKvLqB71po+OwRU{@LimGkztwhzWT>wR%iF1O%c5Qf02
zc{WhLm6J=)nMn!TA4j(-Uc8``tcI$q4M-V;gtUIztpsyN)`kE~X)}-}@jbxraEXZ-
zOQ*mejXDJv@Oq83G{?%di678Fz+9F72}IGdzkNgBc$3aZqFrrcV2u?DGC@rM%)!P2
zf3n`GR2iy%k9rs}Z$ZI@!8j9Ge*24CUD_V>J`Dt`o$vlEC4~|b;!~>AgdLG82e_hK
z#3MhG3X^Td5Nr-VN+*#WXJLU9D~@4L8E!+Yk9OmZ@mZ&1HD^elTJp|Q<h2d5tmN+{
z=YIB4fCG_hF;{a1M6)8oPX<tz0Y^dlXQIc|CTi&CmvPedLR?%*!@^!>CZ-bw7v08b
z#V76R>G{-<P*kb~e5OURG9JKEN<cSgE`aq^K>SmXm6bgdw-AeKOsGMLue5e;A_CD9
z6x(EfhDSzbfclmDVLN{2u3M$enp7|W+j3bX`hxqpW;M6LgT-57hk!d`?wQL_`MzH@
zYi=ZuP9qZS?@n#He5*@V6P-&tsGfQ#=q9#9)mP^~IqnzfBVG?HvME$as&C{E6*D3C
zI10gTI(fKkdsyP9o&LxGb8uLdZD3xcMtSJAdJ5dI5|YnrIibs6lqTbOsy`_MTIK6f
zD_=0E$dS}?iZxrfxX{IgCVhx}nYSStOvptUQJ4s{B29qx`qBCM1V}Op!ATDn55nuV
zsO(v?imwokuC#G&V&HW(#wpi&D=9eyHheiWr&`DTf>5EWtcwLeKFNmZhdfJ%%UZ_l
z+*cazPg4kbo-OfAW%N$HM<?6pmyN8xN$1Ul0v26wsqxA=Lqe{#rRDo#XN2J4-BT*b
z?FRm9Q(f47HC>Lptcda!rGV`1diJH{CY6`0E0_sbBx`;RJa>#xwh&`_2PAhsOAZ4Q
z82g<`w$q&{MJ}^(yKA@InaU{2{Ex-U-#QsT7B%f2E589%7Ai@YE@RAqYMoOtr3oOC
z)hziFeaKa2y1cr=L!K97h^|*ne^~dE6L8splkZzca2$CS`b+?FDmvSn%S9%2N-pV!
zd*8l$W$Sa?vQ{3StIx1?q}G3kXF-3uehYn$;%a=qMab-TsqB_G&qdkV<xIS>cZmkO
znJl*k>Qbb+<8Wfk+Pk|wdETA1`>NX%3|nw-mN}Z0fi9Z%5>y{lU)pbvkBQ-H{sP;F
zDX8jSfy5S~k_0rNH|f0-wT_mO!ywg4-r4!-+P^bXDF^DDx#874+uaO>;gs$motbWM
zca7<%tC0f~h*6*dbFO)gv++6t??6fc96x{rfOa-W0I$#YCjj}NT)*qd?qsoOb#-+t
zNLWC0SB{|qPwq)2nS<7!Kc9}T!Zv6uAYr>RXTB;PxhhFh;;fxj^tJfkHlG<`x8*sb
zVcYX)%xqKyW#(=$HbySeOkQ^Shm0EERQ21#`Of)>uAQlusbZfAfzMGU5NMOSqM5a;
zM8Y6622rErBjeUsZWMTYGy$#TwQ~g*7fz>u$EC2I|K;Gosvd~v{`^%{1@ZI&!jMzg
z%ecM_Ff3c(9t89KL}8mPFWCri0{U%HDGj=rc~XoLoip0}9p-=g8JHUwy^I_CCHWL2
zSF3LCkeh<^GkLf_eFz?otB+FsE|1De&rg?j0vZssv`5;VDYGM0`M&4ot)e523EbD_
z{M}2Xj*hWLOIM(ct!!83@o~hJh$!d(N84L~RrN)0!YGOfDiTU6-AH$-ba#VvNp~I#
zP!N!o2I)&&>e8i1m*k~8FWn8Z!Qbyc-^@HS^M7XM<B>-=+_TR<d#}CLyWYL_S~}zF
z@f422LoxCW+)mGYf}`Z0!hU^ti%D|)m1&vp@D$e@1F7Q}M6nz$FBnQO?u^vEd;4})
z$x-bwiryL3FA2h()SaAmEy@sYhm~IW-R}XAW^PtF+e);3u3|9}luYnvzl1j+Gi1Nf
zC4%&-fIJV@FUKj0Nf!nVv0^uaS5*XX3tr;UeXGnzHJC&I2J#<Tt&de`0nx@H6<qC_
zzNYf3QFcPt+I8xiv1`kevPY8tjQeW7EV*P6j~waa%gIxq?6)0Lf}hv<!W66C@1LCo
znFa+9HYMq+i_LWP1=DwQ_r7;72IqguYqfUkp3^b349Xq|!TCsi#$QDyF>!MrmTYcM
z`;+GPAA!Q1CE#px5FwWw@Eap~%vC_m-7Clogbt=%mY)W3^6XH6Ks0X<x}{a+lD7`v
zABXuLVX;(2pn7^i*hDf|yIFDxyp7F-_c6}Ej%X~m{STmQ=^=3209;=mgt-lXq`13N
zA0K#ZPL%s=t8Lm06^LBM#sHY9Ia3PuZkIuf1EU@mCMbfKkDD>e7c9|=XLj;2psh&i
zrpry_J)v3tu(Y`k0o{re!xast5^Ke$xKAzzW9m_-mh)ga9El1%Na_#MuTwic=%znE
zSv0u|-BuMgkH|gJY-{Id@^7fPo~4(HN5E?tKjWdhINc`C1=6Cil2+MG|6-dc1#`t3
zG`+Z(*-Q%QT;cf908ULSkPA9yQ4m?Ju4C^@WTvs|TzaBB_al(0H-c8V%RQa~s2~&=
zN<kw9pA6n=-*$R8;LIk2*;Z=Mg<6L{AA$YQLEy|yp;XXdDIRNa+|0SmHHeon3H)_>
z{#5DR%1&p;GROtVLJ9vJP@T%}I{61(!YkFIK>Le)WvNHKw*RD$Yo*CkJV$Z|i*@K;
zq`O!2HtS<tJWXwZ<MAKOlZR@gU1M7Fzs6ygsg<=#%tJd&9lz{jLRkblT0i5WqccT5
zV&Gozy?CJjGV^@aLRh9>nhU^nw6!_dymV@3EL#GBu6MDh9$woFiHdgZEs_+Hz89AM
z+&P`CM1wo?s0#zdc;dpaw?L(<-9|LW++OKqCZXM<SA8#jr(|hV@rNk-<y@#(1jQl-
z5*jA|fbVY&(6oY{P1NZrx(Rz?v~_m1n1e$RAc>N}QUNGmY<q7{)yJn6gwE!I$=3~I
z2Oh}NAh8GP0lnoVA6&tzXlrc^og1@f22QSR&0f@{=min;sew|_svv%iQW5}v9qIOK
zI!{mt+gyyop5C9;Gv7+bIE|&OdU4h2AU`d=p6vctUrqI$E7n+BQQrjNZ=lWHmlk3l
z?h9fxjt^%V8eR4qjsMr17c?(uBJ9d=T&i?0bKHVlxd4M3zj%W4@L}>bbPO6sMp<=z
zZv=_=uA+SO7dAWEDc~x)Y|R5lkB(t=*CzrvHJN+Az5_>#zVBLA@O=zjXsbV{tNrl<
zi4%u*SDye;kG^4K=7tehi@D&=1JwEjm+&<Mk*;SI)doca&udDx+X`v-7ZPE^3^1?7
ztMOD21s<6=@W=Z<KO*nH47NTz%mN-L0lWD(pOZDN&N!}+fhyJS($_XX?&knzR}f>?
zZ!|MA6Tl3=jDLt+MpID^s$^H+5Qj(KUG41~P46^W4flMee!QVZi%=lGr4QRj>U#|~
zp#S>7et34#GG+8_VwC#KNrz6FbdMot_Y})QIRjN<nS9<op;LN%HCUV2z{vjbC-un!
zQq*c%OiM{A3RaRLh#fac7Bz@mP^$L?tiwz@1R{IVNoE%psYyu}*+GKue3@Ix1S?0I
zh_2FGPknpCN7EndOE=$>65>u(h;@)4C=``qP{^S-a$d`Ejm)ODaSACVI9#adLZ;t2
zNjEE#<|~$d;BjxYmszqG;*{Ka3d2=aFZ1DlnSmyRT)(nwXz$|6`<;GIA5uT!(T2io
z*$GU_Ms;KE3UiokvY6LPm*vQ{V`yy-%HW=Q=^P)G<2<I|z>N6ioz4*FL?b_v$jC#)
z8tXma*zb%w`oqzsT4B;}pupwJm77kQcq+5JqLk+RREbz9TlO=KK^TLJS$Q>W-_va=
zOHb?geD6J`qno!>247u`5}JYy<P?daknU8Ah}PQM$gQKg$j=jkD;_64?oEvE3%&z;
z3L8A*%{lpqj-$q0^m2zCg`GORqpy1EDvdDck|G{;DEJ*<p!_b$@E9?W-jJ_!7A#Be
zmGrlDx_j4{-_cROAk&Bza!+M~3=8Clxbt_(sUtw5i;IG1z5_ujP~O5`8^Z8F<|~zc
z<M@EOz3ZN;no=dUE3u-fVXVx#sV_1o#lxDg9oaqS=Xv9*VFjKsyiLLVA*03`7Z8|S
zP$1JR*;}<yyOJM<=EnC?gd<igTdXfjDv2)5rTmoax(Q>6l*9f5Gk=A@w+~b>PF=#8
zaR;Jn5cA)@qD6GZME71<W^e(<9<8=XMNM6mgZ0m!g?TD)F1I`?{=Fg`=F}jXn2)we
zQeytkXV92M>#wvT_+2mH9sWO;t8uRV`w%zU>w5oO-Bh|!4Ss|!f1OId?RMp;17IqB
z<WY%+R-VjF1V}427l%rK<Ud&X(q@Te0ev8^j&XH=^+V8R{H<0`>=Ks|!?jb#RMwF6
zL>FzlJOz%8K1E%dDjDZdj(r8+t54SxKH9u4Oj$HHw{n2Yyd-IW1K34RcuN8SoyA8h
zGqeeQ%id#)vN-!{V(BRxpHJdgb^X^z0d@AP+cyb7R}~~#gA@OhHBix5AQfqc)GI3z
z`{7YI7RSi<zP&#l_i~9N_7*v$hMa`Uym?=8q^@{m*j?Q{#W7@ziAE<PqL!F3eo)Ic
zw)TWhc8y|O3vOIaF9{h?E&I_d@5uG0Ka%2$&xLOg!{Z}5cJ#UbKAP_G(RX}2)}_-|
zr6VdoNgHWP8<-r{zfM;4$WHIi%mD9gv2T<lg(W_Z|99}K@xa)W4^JP~jO?#yWoKx5
zaEfhnM4TCaOuOI=C%{;$hY^n;%$3sz?))A9YSOD4^ur-nz4Dy*#RoR-M@JF|ToDv`
zr@LHSYJZ0uMhNE4j@-&8WuIYm=Dr0pTyyJvrKnUW6>;!n4mc@vfP4F@&oJD}52gZk
z(YTY@$tKs-WtwvE+42uNo>eU3;eu}Ky{ox?hf2SmL=5>+Xfoi<FD!Uee!WZ2z_9Cd
zN7e6Dmu&7^wKyUd&XUS+!)yC9?P_s*B~Y}g$FDn!zLn?s_&D>_l5N~kZ3nrxz)Zr0
zLe6~imfWB5=##q%>I?<=FyY|gwIw_3s*CSAwozqOGYN+QqUkO%KR&G3vF1W(4^eMe
z*kW3&3b{y{m?E_1-|UzsYmraoJ<Az@fog~H(TWCY^mSgRapN}O<9g0Mw;PIAIpbF8
zE_PC$ZOpBF7{;_cmCr|GYU!Ny9Q^3Es`%TWtX=E9&}s5<p=Mqp(>2xGLex`sn|d=Z
z?eY&UGBL970!1-i4II5n?ldKzx(VaCJ^}J~+=o<*jngvQvkKnz3(H3<m~hVldz>g^
zTI?0%C9TuK;EShuFTr-C(^TkbP?JN*BSm-=NOi_lJuHpD33>Xk{$TY;O+n$A<o3?a
zjveJDMO>%asm_bz<F>b~p5<h2VFRX->Gfn6@6gmX^`^GBcLLP6w%G{W-7B^^VvM&r
z+@VSwl^<Y4x;r}=_2vnQh;-6@{C9VD!b1Q<`X2Rl!wyVkZ0C=7{8AIi{PR#R?NY~%
zB^vx&e^tc*?xpA`nyY;JF~S~(eJ@RS7?*TO<9!L6c8W8(YNiSLe=8Nma$GL1dWYK*
zjkk`r3mkKQLeWyHS<qiqk);LHKz%1uB_jIQ3i8`vnzRrx^97o;oCK6X^Ne3&e0^i_
z{Q8yf2qQ#*s<AO;KSBA~AG9|wk<0QwW0YVzX{A?AxA&y{D1ts&sqXXqTBI8F--YCJ
zwU7v$E%H|;%9|^ikNG?kQmg}@`3`Dh9I0?D2_qNDn=X5V-r-uiYi^0Kn^(6@v?2R$
zq6s?>*l(ua%_q*_NflI`I5oB~2<YeOIMQbnS`?vX=xvg*u&z{j6HQy6``&gklT%%H
zZg}RekgGqNy7^8DXFN5J<Y>n^@9>Z-@65E2A*U~WV3{&_XAk0{-S2Aew>m`Jg+28+
zzxVgCdS8I{q|*eHeZCt?Z0{+D!Y#Fpen3k^Mn*jMAJ6|bG7amiOo?^2V$_GiGo}Bz
z%MaYu^*D7{y+>e#u2NiEKKk!0!R~NsUV#<xPXC>bzS6(o^8ZpvUk=58qoaX3y}Sl|
z)M&D)zYqT(TK@Zi|6Rl2`Txf)N5C`xYCIKL=JlMFWo%gWST#RL;S}TL?x~yC=mrc4
zH=El3vajIz|5;OY`9lB4ef^)=K<Ls4UVi<*Zn-v*2`zDJrnWMXDW%P>C8c_dKj!Ro
z&WhcmlF02_`X4f@6xFb~MQD+lyxaet7TD>aSaiIG&vd0gHaU*|s{#p6PZqxR{u*n`
z)Uv~r2rvx}Lu+Aa_4UFJ)PYbjIpzM}NWlMS|Nq~~)_-PDq5p0@{GE*doz(tkTK>;9
z{LlLP2MzuIAqZD6>wnrX^nbsE|1Sn30uE)~v&-Dwb+R5SR|K9>AK<GZ0lQmJ@B(2w
zsRycesDrBlx@7;%OSvrK&Sj{TGuZ!%E*IqC(F7+%`oU?^0T9Fy1ahg1rS_oX{;Qlk
zRpZ_v-@M4W<aqirLx8`COG3f~7!LqB@^q`zz<w0e2iTuIkY{{3m~R~v9Q)@vQ$?12
zAGs(fUds!el~lT{!3a1kTg=(p-n?k(?v@{~az%o3xy+KWEJl8BIph3u{?&m<@WcZh
zUS|}Aqys>(ZEf3j9RD_1d5>xeF}l_meRg`vuK}>v<rjojE{<6K>B$750KQRNk~Sb=
znwh|$VId74UseAo5tjJYg*;|ZP0%Z>OMEc2cHt%KuEX>;Sh@BTYs=dqh2E!T(cMmj
zZ!c%VZquv$4{i<9&>Gf}&Wm;mJElwE^V{=_<%Z721#xC|OqQkl&uqlCtVE{I?~sOt
z=3`K~eOMYu=bgB77)h^=0y_`fD=a)bJW{f<&eGnAQ_*pYK(EHX3>5XnlUIKa`=z8X
zwT!JGI=AKVCH$>g&UUVveDr&|<Yz*YeU!;}rx$NfWET`DfQX+&CiKXxWTK6W?vghh
z%6ISH9SXdiwE#OCK?h_cC~Nb|!os3*9jaUbduH_V(SKbVG&CE<JoQapBCJcg3&iK6
zz{VDd+thp31bjTa3{cS~X=@)anSg@;dfqkCW8!GD;#c3$n#)r!FXK(8=X&*)#TEnm
z30jpVf!IoNG;GZBh=vAST`s9JW>l`Jb1J{Cq3<wp@Qfg4<d#=U8`RlHy23@{)2&CD
zk=|-kMo-83k~o%AwJy*Q_Yu+R!qe%yC#{PgLf)(U-P7F?XTZcre1OmGPvJk_wV}wp
z<_dnDs`>3Z4aV1+Q+sRUzF2CZR1N3h;bBYt#tUJDnl-wfe?tmL<t7h~^r-`0bq`c;
zi#oc=$jPae<h2d8K5Y&P3{(~teg}5A$27OyO?D0ru0Sx>{QN?zF>5Q*eAkIbRjs!1
z$Hm!h;6$RJa_iHl=_EUInaceLRp}a^jd4_8_N;sUALpo-fw{h2A^LH$R8Pluy|`l?
zUczNh)RMaOlV4gblLoOKA1*~r)UYqC^VRTDz~%1BL8Z%YjJoc{b(+%qFdZHfm~h4j
z=se<YwJdH`Qg|i=2|fMnz+?`Texm?Lt66jWAv!La8szPa@yb~nwUS^>oIQhVZ1No(
z9VsONm7t1pl>qxy>DSHB*3}uI>$tvE(oqbG;B*DM$F8|v2O9n=uY=0R=4u$Re@M%-
zY<XzbkKp0>YnDgTcRAvm__^d)9*6HX<kx7`7-i;r9rKKjUPyVf&lnuJZ@urOmHaoD
zK+4tid$E`Ay0=-i!}boYA?HdHXAi2cJcYo|ujLu-b3SotEaD#ds)OY?-#`wlWd~Uz
z9-V{zy57F@zd(ptVeR^aa>F<@G^7FcCLL|GeQ)V}9^4qtp-=KqTjt?GjhDY~+FMI^
zL0Cp`uQx@pc%-Xyb&eHk%QNa^WY;nOnJ{YJZC`=6OnMpQyT!4)CW^_Of@t<fy1W)n
zW!$0|5jyjZ;-hM9)nh-lpyrh|Rja+E1HnI`5jZt=&DM)m#(&rUlnHF0NDYX{<+j)5
zSb>ecJwQleuU}{4PK?SIVXlGg#uEiC^-2m&dcDe?ws}PQQKYQICMD8!@&^{e%5ph_
z(PQajNPy5Ux92f+oO-EL-V8#iihZ|q^IVVj892rmu`M}Q(i5Z2CpytDhiaZV)8cq0
zeX-E+um3j{1e2}T##v|5tyBTKJ4g_5f7|sMzQER&F)`SOV$9QCTB{W>VK5?HIez~0
z*Z>4Q+{NluhZy?dR)cB|>8jE;BJWAuHoS+4vCcMNwFE=_l^u)w7GnzfI!JvC$p_G3
zs8Vqe9sDo<FtrDh4fIPD53yWUT7)di0I6*bPBohG?}Nm_DuC1OR=0SkK2mh?u2#21
zgX8*j_7o_|!In~)nOkE+-#*6XY#LvHYyKYQ=M{&at7FG?Nj_@7zNqL<R>#=bj>5(k
z^@+z{zv3$y=$F;r>-wrm;CWFKsX%!(+*w$#&bt%JJVaQ-Z%f(EU7tffIz%*Z_=OII
zq(`n*?o0ZiSFUdY&7HG!mE#Ub|73WdjBDye?$3V^qLgIbEieCN#E-1WxgS=0ZWOv&
zZTc7Fvt@D8yAiURCxfGo9+mf#M-JyVIKt9?G8)nG`-$`>N3o1VMEr1v;A+iFrZ%vA
z6z}(lw7gqbFn439d;|&8oVF+UbR);AhUP5cTUL<g0WD|7sx`?~kL!I?weqKKP7zc!
zr09x=(6QF*cm8V7Ay}F5Pz*~O7snn`)QT)SX;G9^(ix|TZl?2Cbx9z)+JiuwngkA2
z#J8wO#+=TKtYqQnW|*VH*WV@V(aqZ!&3(O;t+G&v?l+2M4NF+N(LY`}a6N?Yy!Y`5
z169g$+KuKl?ovvMocl0^%wt|Z3nZy}8Qt7QTOtwm0sXoT$v^m89)Jv~+na3kWZW9f
zMUD|>&Wv;8(TTK@F!fYFKB=5-R8n=)NiUlYi50YEgulG4arJJ=9%RteO$Auu+Gwj`
zu|;bNIaAQiNiBL<njY~;MG6cCH*#XTdF#+>c=|@P+SUjWR_7!nAx28-Y}Gds^w-S4
zLwy1r&c*;GiYS&Z8g278pL`kSIW|D*9mU9?Touj0Ze}|8k>DYIrdN)tzRiT*w7u!Z
zP~KyAchqn#ltA|3<G9B2ztSOhIbJBBA^Wpj7|%*K-YskhX{4|u>K{h~Lnq)g&T*Ho
z!eGdEDqGu6B5XpVLW{P-eO^$g&#*_jT}?79v0{&^Z15kx1Ru~K>Dyd!dkg3Xl40vd
z%dTs3vdqY%i@?b=c-dR^yn+`3j?hYHq@rOYF5%<6DtNU{YlVB|yuW^gEh6%0>}8ei
zL1!HNZ24(rrB!_X5|Q?o$iL9WR=ob+)K{C@+)Od7N<E9Co0V@zWEgqf4&xU!)?8{z
zc9{kCo3UfOf@1e(9AQuKJV;=9T2%bo-+0Bi`U1%BRBn$B^TtMccR4~z4m(cc9i$w#
z-9C-|IXT@)#~bchI9J2r#s=QNY!AffAvagxs+M}S7%~?kW#^lLQKW)&IpaPf_RBYK
z#+Ew!8)nM%Y3B^mQ9Kha*7wjGA4O~_%qQZ`ZNQWCAx&LvM1QPN^T2u^eD_yx<Va+!
z+RPAmD-WZ20tj)mE6phA_S!m*eGqP>uWT&w`4-)+tLIPrzI48`M!bT=ykDGN;2>eX
zV)BVFQkS)X2UYuY9o@N91_pPx*ywq<)IIxH171x6L9*y=nYJ9yO7YJ*$j$nEo6*ZG
zSMT(c-g|de-otEwx2D)K0d@0Mj)-}|{V?nzN63krO*+hYcdMts3S%knVl4C9JG5@R
ztN1sx>g#~OWv$rRt|jO{Q}PzPcyLy}&CEDjElL<37HA89MDQc0s%d$@*4s&RQI7S=
zeUjW#)Fznm{elyYQ@<7Vqv%QY19%-`uU>@G_@b{UMfB{raz8A;@&a{kIh_3dRu1$R
zIZgoAT{!0E(TUaJ9?}wgx0#mF^N_!a;i-T<ew6>GUijsBvAIrFsUERM+2!}$yZ4+K
zI8<G#UOAJK<4&3FJGoztS-mvypRyjNhJ^eeBg4UcG`*};9h}sfM&*ohtGs{p5e3YL
zo7d~zr|h^(e7YgMvt8t*R+VR2aC!#7KRGtWX}qFbeTZnMfyO-TR)5t$l*S7SsE<5C
zmitBXydE>39Cq#jH<FH>M!}v@%lvop%0Q;kE%|6o%5-C6)MB0bO2{GRrGnedeJkqc
zmnE6Ni$bOsV>qq-_&oJSKKcsy+Fb*@kxCJ+Hozatt;!{qC&Acd)+3-|1zb0?ETdZm
zC`tm)Nrk<Y-eGiAH37UF)d#^B3X=>v45O`>rvW%Pcy!c4`i>V`=@n3(QiB#^cldE?
zf0?rQ166HAnu*y;r>)siDk-ESVqoxPG}?9;26TMZGYArJnZ@)t%^(~ex|(C?iuh2$
z1_l^ednCsU@yLet!^F~nB~kJw;8IEb^YAGJac5Ug?<?`$wbF!<0@tK`pLNk70<Vt~
zUQ>EU1VM>4=p;^?ZedI$EK;;jW5UvWWC>TA)K~&?2-AVmoyWnmggi2QCLv*c>xq<x
z)>hO-$IkIa(8t|DXdT+I+e<`%G^UsYJ1{It%Jxa5I<Q?m2$s(N4AZn<XJ-+y(Qw9A
zYIfHH!OpQ-P;2%?jxCx2dB9wkjE9Nxtn|8^RAHVghTGk#w>cI-k?XRx9vj>3PU<Iv
ztmip5U#%qzHx6c}huRm$B8<w!I`q>H7EhCnva^q&;}~k@mY@|d`>-|ii%hRTQARTI
z-DObD02r7|;>S;?;C?C3Q3F}EVy-K&GWEQH(VRilxTa%7R6?FS3E^@IN`T(yfMm~o
z=IQ+0;))+$kH<!U9_#>)cR%#}JmAZ1BqrmJJ;RVd4#RUY#vw>zoYY&VeS$Po4U^C<
z3-Y+?zUbj4S(6nkcs9=^8_OpTtbWO1a2$Kdt2pfG_3Zm$HkJcA?HiSvm@hwYcyB4y
z9F2GEsN3Nhl#POv+BEap45*3Z#Ha~p<u7&zlOH;%YI<|xYM*ioi+vPPhT`C~b>}{b
zbosnty?q>R9ae3v-ALtVr=Bt{NJp1!?5vE+AO+x>Jp#N@J&WOCg56&Fww6{NdE6_o
zIQAa!YGVwP<jV5i+c94;K4wh55GJ-nGVD|U+#2gWi@3YFH;|(brJ^^7s2irMc1!Ea
z^%eH3aNM~!&>@1P8S933EMnO_O)Wd#k9gIadU5i45ZYVITW|6L@+PjN<m9-5=TV`~
z73sWI*7N4iAw84wg=rAt#QeiV3`>sE%9V3ej(<hddloOy&@;;)IKRyy!EB%GD=}Kr
z@?d9D14{%hf0B<)mm3RWV4J$n^}^WKxAg+<Z!#Y|dEmO9+@*+!R98su3D2yo`c1b|
zzcqWjii9R{QD%dO^d6BsV~il7?5<*+AS+?e@tdxY#+r~TNT7PPJrLS3{b#pP8idQ}
z_NM=u{fZ)>FYd|WX;+8+v2<2<R~_ljLDEbbDvri-ILXTNU4{z3z!6$!i(+s#1AmR5
zr#C4c-4kwhfjxStkir6R`qF+WV<co&Xg)ESGp=)Gfs>1XYI;Yw#_jd`R(KeWc4id#
zoQhk#osiI;rH1}f9v<KqL=)2IR%V>?>W4Xb@0iZ(z4w68$8m-w!4>aND=~lP#H@pB
zQnwxrRT#e6F_-MMwHmC{xxzzJr+|z_<Kq3cv>G#-;=%~P5(*lxx^Vl^Zsh*)hfl*=
zdrxsWy9iad;je~ypdOj0zh8ovfJE1PL6^qvir@Eg4%jZ`^~y5ZrAEk@ZS=;zpWh%o
zvE@KdAVtW_)x9`QF`b)}2(wfxXspi;AgLM`f2>2&PMBtF+BS*IdKTWAZXZBW_IUR%
zYX-rmSC+4%mCGx6RL;0=*!!@Ln<|gDNFc06UdyJU%AFJEcJIq-mrGOy6Dew`l@j#B
z(#xY5*~`sbkB;3YCxZa}ka1coT$itx+R#dSmpu-*FF8>RlO%0lHZH;{7pPoGC6C(l
z{?W(9jwR=AMc9rG#;)8rxHxIRcZeE+x}FQv)$^E2*i~3=s{N$?IKKPg`W1>Ks4$>J
zaNT}zxA$i}(Y&_x_5p{QTl|k^#?$NWU6+C(>?~Y=&y5FY&f#{or86F(r43>}Fl<sp
zeHr%Tm{q3l{B+rbO(z4OQHbT@fdJRuJz{xT<yK7!V4Vz&e(cTn$!9^LACGYp7rlmE
z85GeIy+uy2gCom@=u54U_&XO7@SZF_=8c>XaRIW_t+Ah)4O(3x<s00gT3!j`Kpd^B
z*lB%I2Z`v<Qa8YQ>@Sd8bC&n#WKFo=>Y6CQXWfCy_vwrsaoPOl5-;JL9i7`~+BdQ5
zB&pQyhdHLI7dcqq0x=THNcDF0t2jG7?`QIu(4*<*xt!rFEC2<WU0%>imerg`scj6=
zN&=@xX|f4(fQJ2btsn-a@uK<gbn0c}VzS2?q&^p^89_c$_KeyQR*8>eVo;phbtA4F
z!}mB_jC<#03FxTEbe{W|QqIrO39?A$JjO=2AZ-xd(A0&~r0Uu3R@Nh%y6$y{4ZhjE
zJyHnoR@LPTHSYbu!6ixqxb;>q+uq{O2C2?=<l+wg67XtEmd4AR>~o^bh|6^Qs9?Bw
z8Fj%V$^fJ(?d&)&j!l+KyA@Gn`$!ZVv)g2erTk=VNha4E*77*a;-pMp3Ox$kogTpi
z`u{kmpN#b-M8Z!2&O(fjl;n@yN)0GSxH2<p;KTLzN&WyunZ3!~9zsDgs+1LLkiv8(
z(h#>Bw3aomU5bjXxutX04VowhtI_=}wQJ-R_{}oLZ{pKWCeW&~D$Dou`5MN?+MnVg
zTjMgEUim|k{OHMj%HSY*nchHY=%SbO7FPDbto%w~CJFMJo<JLWcW{IMl~%{Y$#PLh
z)?4x^T(_Cj@=4D%6^M<@<GRpF?dG%@FsnPy?>srMIy$u9&Nj3rtOT2Bg!`+J6qptR
zI?Vg!8o-cA5t%u1vzslFTw>xi>28G(MS1gWNPIcbCpyXgZE6S<g0WYwl5~2Slc-fL
zv$Mad?jiENvOh0X-P2ilqX~0w1ODw*FY424fjZ;#C}Kg_M4u_Dlrt~FnzlcAJ(})M
zcX!Z1Evj-hpw?sa?2eZ6?sqj}fedLBGVc%fta`I_7{+f^eu{lguKlHa_1a@GQut(U
zs4lN+7QU^ePi-wURIMwB@iBkHMUqnG3L0hISEYBpra<K%uFSsK7#$fOnsSR5(YIY?
zqgIy@B#r4xneC`zede?yTr-gAe>YEU{A`drZ=`lMe#dxmaiupjguX0UcpQZssPwc&
zLc9oij<Lx29A)a=mDD5~f~*%pLU|K~(q+}>8PW#>(70dqb|vTP@TE4-M()z<F=`W}
z{!EV3;y|KO$}E5i1<ta-=ETWJ^+mnrq8lpe`8EduhxMpjIzEoMakQM4SM$R?TFS9V
zijQd^Wwyn+^~XJ>Zs`bZ^MMFiZA!`RN&6ft*P*~ez&-8P)=)}@oE5%4Qq!(zKJ-EA
zL6?YtDpm2Z82{?T^_+rISpqA&kaek!t{y9TwdfNs|1>jvxPq9-`}G>w<juaStAH{X
z`gOhfc5Wi9@slo3Uci#15RaI@1ePm0g)3L2H`Tt-qsmrZGgoc=cqFckp2nNJfD9=l
zBo<RqfzqgoS{Ub7KwL*}Q<<HIWC<+SWs&X1yFCZ(>Zs;O5Rg;dR_fKRc2@pNvmGC~
zhqSitp0G)gC>Tz&wRw_!-bnz$b+Z<Y<P-M}n-EqxE^gn!$qO;3<+QSseTKdTM-0i-
zs>Wr&x3o;#(J;_RSQBv3DF+QSriC`(5>3JPn!`C(kwPlwWzk1t98h!c%6ZZv2$e2(
z|NP<Xj6SD>ZaL`0pN{>70uY~-z|PS&=XHOp!Id4^Q5AXy8Rl_k@p+u9o%Sr88`bZk
zkG$7-87ImmJINdNen!?v>pt?}Y+y!ZSsBKnB%}hUy3UrrHd{|X!tJGRBU_urCnsJc
zkC7hLmg`06tq()8VY{3DkJ~j(=AUf0tvE<{yeO@~KzwFGRsiUfq(7=3dhNWetG%y^
zZ83gp2xxkm=}(TuIfXS<(N$e$m5s|qVwe%3;i=s|ar)~!H8FL=xuUZZVLi>_Y^58$
z2sWSop4`>qL?c)3-GQjJ%9*&~HpUWEQSjOx>Q7adGSsLlE>lx^?6DiZ&GKY-3J#%;
zDCr~35Q%&7Z>3X6bPF`(nYEb0;fiefUTS-muM;x^u29MPrSXZ8bcCWPYWiox*h;0i
z2|uPGnW}c@G<N2I6%tsF>CWKQ#&)vIG@UlUh3!pu5vy`>O90-Zy&wArE3v$^x|A4{
zaSF~VjHT<+vx|MI)8AUqx0Oq>Z<?h#W0w!xb%0B^l{g$so}Y14{`uYXdL^W+xoz@o
zZ`dpG{)KSny-A@c%f`?fxI^*QghardEaq_TQ`<YV^gG{aC#5SG?iX4oV|3eb?w+{z
z;L?Zpahn(Qy>Y{QY;vXl(i>i%gtkW9$z)5zCs#QAUulQQo3H9y^`pX>cejo^VZgtf
zY^u#5SsU><{L2Ql1<DShF^6d@B>NI_(ha@sb?%H)b&DV5ARFL;mCig4FLA%P=4?J#
z@|%Ue%qE&*e4<p<cX8{0!%`w=A-2Hf<3!TV$6C+zn7bkKFlwPA87QnDZb3ePbvJjD
zy9{BGZHC9F?x9u+P{r4ayTi)GMJ0yqoCEjo|A~|(RCh}(uXOI`XVn?SBI&I)`eLcZ
z1^Z>>o(kM<%Z+Gd4)YFrtzx5Ssz%%Xw8NmVco9#V6a@EbWSX#$Pz~!Y)1}jtDV5Kf
zzfO=@C*+~uEt3erfdAI(D6O_sK}m1S{sd=XBXjrk`mp>_Ll<L>=jBp>*{_Wx(i2=-
zWd#Eu@%Y2;ho83@@%F4n%3tQV$`4d2qBl}gS)3)-P+r#qn+v&PIyd8FD{0yge==QD
zF)m-5b8U2^mpfkzG5%EJz-FvI-5z!0CQP<d0<xBt9MB(ijnLRCH@OfQUUktOEU5iz
zHnf|c>%nE2Jf6J_T)vH)_#nnR;j>-d>6Ad;WU)FNV$+r8vQ{&-3GUs#2SO0%%vfnN
z^mNBZGRqRHO(cBPq}A0K3Hp)oEZtV1W3hm&UVv-E8zd#AI@{FNHtjfqVX5c%&d^|x
zKvj-)c)-Q;D{5!T{qhYZx$R$usR&UF%@+nVaYq5Zq%@U!F&m$VffWB$+Yk5NqV5TC
zL{YlgqJ+bwQ5hqpDr7sUNj4=A6^j~lk2i`sCD4@s5Q}4_(M&+sPdy;$i`rY6l1t(s
zQ~GN8s9;C&T?6$}plNQS?$M$b@%OcfVTVsd1H@<;>-Ja<%e_g|HP>8_7~Jkcx^9Po
z3D#esS}~jNndUP1O$DQC`9dxx3GKuWY@PzbX}G)38)nmmL|&;SyYqq%_*t*PI^!^E
zRqE76XkCiSRtbBIEY(2J*FThb+AGJ>%!BpPRjsR#cByS+WoXy<T8L>yEkUXI*B>nM
zNz^YfB=2bQ6!8%;<;^h0Qj?W?(Q7R-=%-w|t(czswB3flAE7zd@L3VV+^c_g?y)(N
z)6TEwrtnRPt=Q>zP{=qSFQUpB>BwG&IZ?2+TkrJeDQ;UiQK?nw+RB{B^+huboEv?i
zD}l7;zs(uFpy7Td99?3OZf092Dlk%68>4o*E)ZMc6mvOgubRUzo!!l;XB;qHZFQI6
z0RU5QkVU@jJGmpZuYlJjM_*>GuBMm2>f(Pt3><zMNRuPRTpoDEJm#()Qy0QS0jB2?
zC7X+!dJs!Vsfo!21KJVf5*UN}bNqQ5Uwc-{T*~*mdv09!RXzYG+}4Nd)kf#BUd7Fu
zVevUHtS~UL{+J@qf6tC2x#>(YX<cq*?KXM8;z&~uaSR?UGC)6y>(p|e`@WsWd32&y
z{a|C}qd%o&g7h6%FE1|~>?c_rV>)@#{Q2DY$8}7+Jo1@KMb>xA9H3E`@$)8a-mrKj
zuY}>H)RuR`CUf{S-G)8K0-hXRzbI-pkJOLd`fp>R?3a0(*9J2X`^3`ypz*yHQDS-*
zGYU%6Me~y{Do-h#t#lpGEB8BNTb5>1Dg8s*7VK2YAHZK-@VgY6+@HGiNSS@kBexl%
zW>#KynBI)_iVTZ`GC{vO^$FoeR<i^77LUD!7H)C`!IalW_9=+^l#{$NfPSE>?Rv{j
zN5?H;;eIqyVp|7RGu+jzv*%sD?rvAXt$vunktZpsoZMh?wB7(X%pIW4cJkyt>@iN(
z*vp(E2D~*v4;7=jg#4}I#U)=p8)RSEgvz=*EY;c{3tDas5$fBxNUEv&8oZ4nc;VIk
zdpwBZcMsY%?6GlO?POcu-l!-7)qugE-YB{fH|?=;0ds7LKjl>`#&ahucSygrQpDV_
z;6ZAgHLQ|{$g1zhn<WKCub;*h)`H!%=!k~|Pesj*arE;iTe9*Kjx3LN<c$-|L>_{|
z_LsHop^NLGQ@EcnqT(YvhkTRlrWY9Rrzz>Z9SykCdAhWO`XJIsrQmMDiq+V(W2V9F
zZ}j;4?lOKgfndC?Y4y9>sLno_ps!0vrgyqO7l(cL0={%QYZE^WjS-d5!bp@09<V9s
z@2tD&*`;qRdx)9vxSG;|C(icu_06X5JseFJ1<nv(F{B-=l>V!yBLX>c>&36`I4Tzu
zZX?|@LFS}!MHYsdCt^-EB##A;zG|k5Z?An_);c){FOkdV74flfD-92C`mIeyA4^_f
z`JHz`zC0cT`H?{tJmcC&jtr}JPktCaMXjh(W^i^<XJSW&i%ZKt;7=%_Hi(p>Y*=J8
zzPJfFvoil_{1QD<YM5L`jg>)#Mx#RavX6DhgDp%S6oTfbio)yee$!XYbARoL%8L71
zo!gg8!IQA+ABF@maXVUX2>njnXN`=!7YQ2Y77FQE{YaVpiVjXi<@7MkIz2aDR690n
zg*Yxb(2y@OrsY->J@)q2vqZ+S$a-5WuDVe}7Ky4#zs{GqeK0H9j^<Pray1%%etU@J
zb$+Lp+N}X$o!PzJp+2dkq7~pDNeM5tZFr&2%%7LHu9Mobs^c;iY`UDTP?{jvUigRp
zbf@QJ*9y6>D6jv|^xhE3N~Jzvek0<B=5Y7k@=4c;PPu#FGkmjrxumuSXP!Ns(D<I5
z1Acv5%eO*V?_mK0oYCf>`oW>^>7$^YMi{A8ZV#X1Z#y)+gu59k`B^(es`c?kX8qnl
zME04ct7o-t(>xT|cOXAeJ3hX$u35Cp>9SOO#8g`9WS$}L^lUU=p!#fuHG&S88DZ7A
zho8A6m$}cLC$av)1GnVzg`g6vS4%tcY;~U03O^6DOLz8d+e`L0jE#a)5x6D}H%XHO
z;eA`i(GlBw6@h3S3EL_Ndp2<_^f%|N#StqC^lI5kEQ{sG1U0-4Zb4h<Lf_lZ??u1p
z?K3+1<`W;!`=0i^bA!V5+Aan52p`|JWH-~;_pSS{y!dIO+n)+Rk1`=LGD{A4!}FA{
z->bfXNu5~z6Z{!JbKBJ4<xq{nRm3}}Ixp`#?vwff{ETgzOyRrIQcsIFAMeN4DdgBQ
zja4@O;n@LnSZ1BNkxhr2U$>_-WlQ67B5sch6&3$L(bMM4uV(WF@4qjtI5O<stud`r
z&#^xbbcx3_J#9`>RMCRk{B=K+Z~wy4yLZ1nhpgfr%4*Kcs&kZh^bw0U0sJP$H*J;Q
z1tEBL46ALa@wLxxfB5?^8U1S6{s2|gk3r=vIVscRvRk*PLw@$P6*ak_*9dTPSYHn|
z&B*1q`>gM})<ZA}*G2Ol#dsJQ-rR8}j;fA9hV4WT<3q%O{;A~558$<a*7ri1cHj2h
ztx#&6Tl1ueh(ub%E8mf1;7|AQy;ww=Sq9Y*JZ`rVIt#daSgo!ewY%~q=KcFYPCpM1
zuRlDWImj`tZ8k)$t@3-yS;ckUTO_-7e*2!;*`78+5KjSFnowj!T9Iz@{RzSLNY<l)
zreIiB4pAZ$+Q=s^H)uihPF_A~QROAJhZ1c!)+(ICr6;zIn}ERYORb=PB0&Sr#R7*-
z+rk^mUrj6xZl%uBK%@Yel=iVN9OT?uekyZ3cMnI3QV46QbFrq<Q0Y~twvo30-riEj
z#>bP(+Vk5-$k*v<=GMw|;XNA(4-{!4FQ-QvYQ}1A?NG5Iyz?Sk^l>o10G8t`^sd)B
zi)ZunC8bBqm66|nmoV3fq$?E~E}<<ruY3xP;<}C|gSmTDDesDRu<~9f<;}0=r10-C
z*@&S3n4V3yM|bo@%%4?#=q}#)CF=DyS3|%{Apte(p{-qLzpGF^9QE@mmH>J!qLPMv
z$5B0(Dzo4?s2;2)`U&@Y_}bcoqGNBGCSFC*TI#jtT(7-b+v49WZ)E(Y{<wnl&0F&<
zL}#$?P9Y~3H9vXa&y}(;Ee(~N*<pO*_NC6MZ}gEes{<dW6H4l0FLoF8Q&i|;P?Pao
zyd^mUPa>N*c4kxai50K))tx|kR&U=*p%OkF?o>S}LZJ{^L7~jyECwq!!7BZ~IXGC;
z3erkO4gs|zP8CfaDsi~CjXS~ZRINtu9!ZCLPab1uIc7(K5*v)Y{#kM1$l**ctKfwh
zI$FEmZjLf7rofj^q!_zD0=3DdOjr{yT)qq*R?rU|%(t&;u?f`I4MMYMGFAieTzn<u
zb>-iAp!{9}KBLwUoX9BbxtZV0ACHsu{gv)nClG1yynTiyUHUeM#o|nV=X@I%Cn{na
zC=(eQO=omrYds63-}1F{V|<)~Hv=QL*kAo?;ciTeJiM9;QuqnsJ7y2(hBgV$8I|zh
zp!k?U7!On-DtcOO?LRmq=2E(P>L%F5$(dX0kswEZVX(MnjziGKm-WuEHT(Xl*}WaP
z_{O37Z*_uRA@x)~99^ybL!YNeYa2R-zTcy@DAC`-muB`=)L}wCA&raf4)lD$;?XB}
z*O;~PsLXi$)`lf9IR$HgmHpOtLV``vjeL%L!GSp_L2hlUpooHvIceb51FNr!*(zQ|
zYTiMy!^6+i>0$~GUd)!QQ}L5a8xh95{Hgzl>-Omm^Zx5bvd}EJvYN>A_Wjiy7s2AF
z^Xv6&d9g-Th8^%|_#fvzkA3mxl-vb^1R04R2bHp3KKs`f{4iX@vXWFO>^R(4%1k2K
z=Xdw*ATyofSoq=@F#320YuD=c+01BbtQ<?LM5!`2k)cLIG2aUc#20x%wQEHr7I#cc
z(eiKF{KjUhuc7pjr;fYaXn`ktS@wI`R1ID=Pv+LX=%h>`0#-ID%7Ly3F8;~R%C*?Q
zMC^nzQVe0zAS5h2o-#1(>m_jW`>V(x@ZYbGWSp9hKVdwKP6TN?L*2BsSvtD47kwsf
zTei2+BZoXt_jgMA-EQ9A+<We}>?(6~Y9;3@{=#R)W>rRJX>HoCD*Yyg60PJ--*_1b
zGQm2ucU1ZLxhI>tBge=69&)w$OuzHAMa#G*7Q|kX6+cLN)3w#NYqJ_05`-HC`T3>c
zN89mm1br*Hj|NsE{61;h0h>Oo@S-uU?D~s=Iv$|OIbK_tre!t*q1`#}Q9iBpWbQjY
z$isFNc>X@_i)4vQ$>Qk+)eh>M{PYpt&#uv))oNl{Wj;b)4u_w^9TC_S@#WRmgMU*h
zD1`2^_x8=s_1}r0#c|RvexC<rc5f~WDXKQnhPJlQ)4h}@n=aPVb+2=LsO-8r`%;R4
zoNKKfV55nkal2x6@eiNUe!Z0YO&zL|tcMuyod^b6ciPCPab%Mar`7m(diw12oQ6rR
zE-odl-h;b;maDJb;O6-=+K`ZHf9JH&RfvtF|8(4Efq^bEU~#0RzQ_+!og83cd-29Y
zm|l>_I~c%?-XHd}bSzd_Bw{Hly~6{di$Q0?XGCj(!6y2R8IJNF@-q85;`HCH=o>A6
zMGZY6+^SP!)NpwIiP)@*#mZ^VMB*>9yaboc{t}yaU!rppz$U(%7ldTW(JRn%Ac`B?
zM*?zK3XSLHr{6#(=^PWC6N#ylUO)4vb{K|>f|REpy{)vblG4%EC)Fe3{eFSD<o_P4
zmDI|S?N{_go8K(8yX!D1g^6h)GjFL>d~G5LaU6%qo@X%!6$J;6{X8AoE3ax4KfDTa
zZ)@f(HM1OB*;x<GpV@zct*iIF_xC;o`bVVgtf^X8!o@mdmX+Cj&-zqzOunX9HCt&K
zSeRieBrP3dW5>{8;vs42+gr)S+T?G8f(kh}7k>BOd0X+cRX+OHZ8@nPjSH@p#xMAz
zr8C)4o$GHY1}Ae^Mr}m5mMfvdvpmg%-}I4Ey$%j9dXSmwMhCp{*)YFDHwSj6_p;RT
z7BZV8ckbjspK~TT2xewmsWZFn4;^So<q|dKclMg)*ju=D#|aG#%q7%(1X(MHUJ*Y|
zyrA-bwNX>&%+Zewq{M&8O@vK{UHuOrcy{{&`{Jz5<L2~pi53O8j<uzgGTetk__4N?
zp~H?uTT3p>&wHrY>IObnW4Fa%peTW7AnO3R;5qn)iItX??C#bNOyukBz}QpFT&Z?f
z<!lTdt%MLSswrU)AKhekaiS>5CVj<s<9XhzPL+4nWX`&i87_N#t`nSUK>tgbm=Flb
zcPQ8rIUjI1uy9fD9?_!Z*13dxv)KP$STHg*e+Nf!6o0tBSz3EQ0B$&aYuzFtLk02B
z6AB^44!lUPaaB!-?z||?W;TjD4pUWnso_CI>nlp5N~hD&xmi}$Fg%vq79XdwYfn#`
z-DYOL{4@*g0ew5~fahSELwx@#?2e6j`AL;Se3~WamQ<HgbX(R&&6{VDGuGCk8Xg#2
zkFhQ0=QZKy=a|@i1B)e+=(^;EJ+aWgCi_?2Utw*@8S?6A8_JsS9#+{+u}(#xhO{*P
zWLn30bf-`aE|_*a;Y<@b4r?9Q^UC{*a$N_=GWYBKR@#C1sXQ49b=M%OQuDUoc!<&O
zux$d>efP55Q!ljKAdYC(0ydVE5b-s&F)j6pdmzi3V2+H&3Z%4g<P+JfJl<p<*=(WN
zUD$CS^G~!?PEFZ8X(_aV{KXoy3ut|&k1g@@4O<(<M@`2bdvg2giDlJVHd!?PqQPhU
z9%Mi1`PEeDaFLI(x$!)Fe}9$P4hXO-DE&3s=zHz`nq6u3B&y_AW36GjwW+;iWOr{s
zz|pkfZ75oqt%ZiENtmv&aqR0)K|TJy%A0-%d2X`IN(a;C)@M`H`RM6Nt@~_0PrX;U
zzdU^k?C#7P6Hzu-S3N?3c++)fdi*zU-+a#1k-Kcy*hER>xxeA;m(EV=)U+GgMn+?d
zjF&nZ-dVFKo}utfS<6q4gZ}ec9>cXxBn=#rX__8#voW6krl#@<U_GV<Qzv`Jl$zql
zO06}%Ty$_nlO$Iw<4irqgcO7LO4w?HWP|s{3hKJBspMI+B>YLUR0Ex%!sMi6jHYp9
z#Eh{iY=;}b%$CSD%MEKQM@KMpE#~ix+Gv<ClC$Hq6GkE<qaFEI>7FYpQe>DtgF+?j
zoC`7&;LHMACjeb3Y*Yx`#!`!VO@*&WL*G26aKNKB;TXGa{a^9K76o?oFC&p`rNZ+e
zdqLfU2VB_LIYBFZrI@5bY>~m8aRR@G2Rf<@Uu6~Se&xGxp<;=_-zOsD<$_Cj0k7Ee
zhoSq~6w9$&WwGUuFtjT3RS~A5)7kQxE_7;rY4gv5J|SyGe2mcP?~zyHH0m@kFQ3yk
zRiDwCTb}gNwVv)6;hqP%s&)HnQFEXQQ&i&nkRV<~A2Pu&H#0x*gn!quu2RF=_v+6>
zCqMQ2hNjNe4?l=gFIJ3;-Wf+`@t`U4xHItMb_~XaM$ktRj6ytdaa*Pe+NLGeL4~y>
ztiB%zHHV)(eKKWerA=AuX$1`u3${h>9@ou~s1;i1YEQxl4=RR6$RZ^^x6?l>%gxox
z(WBoOZ5kMJoapNMBq|De^b?KwK>{eH#S)xZ9NyGj6Q59lP23P!a=(dlZ@@kNr6x}(
zkGljIMP2p~$hoeg+N<`^488qeJ;#Nz*!RnwKL7hOHIk1Dic9{u`#ks%9mp2iN~<H;
zogAp08cV2>x}BKgrT71_GD*P7gy?9=F<r6*ebY#^P1gdW&ms7-Z}SS-S;>maKYc54
zauIE^qoDXo{@kFN--{+tKQ_P5&)~sJXtMMg&_<VtN}0yZe#2l;zv|1~7hLRb3d*Q;
z#|Qws;3t#vjEOnvRpC~=l*EF93_pp`+0X6gX8^W6OD+FobIw@EKH!~h{Hn@s1@CO)
z=-Uq^=NbC?*7rXr5)VhTjDQOFP-Kaw&sX%&6I8A{jNK@EJNw&JA6fo$;2VS`!Z2PN
zPF$AyWM;MzcxD>-_UP}LqVVqT>}5oz5(Drt7a3_Uz!oQ<SO}!VcxF&w_rS6PHT`kO
zURw90)X1^Z1P=oPaqA6$tF^0r2}#VpThoPoEs;{WL0_M&&Pq0mb#ceVDUwA$2B`Gs
zcegGos+|ddVn{q9gHj>?Pz?p)_cBdWSnmnc9R)}0NSUDn@)V1wzvt&?Q)Z=F;(J-;
zs3@La6ZEnyYI-71P5nbzd6@8N`&<-)p4naq7v*?;bFBUvff4A0*~&88+d5ob_>GO=
z0DFv^H!V+3eV?Nb1H{2+BVL>=ZkE8s!2j^Ezf4N~_|FtIjugl>iLV<tSkK{EPsjxX
zfPL&OE+ovyTX^R|c~p4#8mpPAz->qEe9ZM7g0xzB&7S`*I{K*Nrqb<ETX7`)&#JT1
zV;%yPt?K_!g||cMn#uX7_zMk{`j*&E1h83Iw_FQmR|LnG^9ltmq-$4@;REXMot`gj
zry%~0i|g^@;iIh&%`=e}eA4%h%WRv^1Eb9mUX1qcM@!P_9qU7hz&q*Fpxzsodq39E
zHvQ2kbLcT{dvVqNBrV^ILiG@T@7n_--s{(m@Vvgh%ucH*Cf7e$*0a8DAt`hdP$Mrk
z<}{fmDua*_7xtd>q~9&GWl#4HSx!rg@Co72<bswdr+P1&Jtx=fPNHtipC5NT!eAD)
z@}ABn`YH??Bm%8q&MI^D<V_ptkBXZ^2DpRcY6Nu8wC#9L3CKC$E=uTJ7XvOhh$t<-
zj?P{?crLQ|>=#>++s2Y47`;D6YZ&|t19wJI`Jsu4Btffd5^Wtb?NfoL`~~ueW+=aN
zW+$3Povx6RtpOHWbh{5@*W@S4*~@#~5U~iOI!QIUBimm8M>vUPywd#Z)1C@)TM*7@
z)d1fSiGDUlhn?gcs-^Kefd0`F)CwC++AN3#eYC9M_n&Von!~H*#ElX{NxUJcD%pqI
ztD<OB1BSRqwp9n)AlqT8Qnp*ZEF`_)yq0C(P=)+B5~ick@+*8%!+DQ@e1T~$8mT{j
z^b>uBzxkGdOj&+)o!3q1lP_D9ktdN12izD*1r}u;xs9Qv@$UjTV$ynsoB!(QPq$KR
zM#t~HM5m*nHRwMTr7$*T1-ODbLne!=A}516a$6@ZDn5ZT6mK6YCAxEGHj>omWxA5U
zZ%uXE9yVn(=dj-+L!WCNix(z)Qs>3J0hrZbXK0#XwdX1B!<g{Ea0Ev;W^Z-awn3nz
ztkKvHdB7&fY}s~&W;R>s5}N}=DZQrU=z)6eWt}V8-kQEbXlq5>F#hws-nBf%EES;2
zE`{W09~El_WYTB(I12ksK?rqbrmXR|GHR9zcbt~CE$g|LN$*fxF(-#H7HC9Yr0ak7
ze|jY!G~K`?qGc6z%r{i%cq%cmLG(H$H{L2nlikI^SP6~__t(bn61|>V?GR*SKM+ln
z-Id<&R9p8j;KrfRe%wHlH@7?40u_}`d&k1cRG>2nHB?m+QHGU(HiZZ!hv}}N%&j|f
z70GFpX}(un+z#Lsbd-t=Kw^=kmg{u2H-Lh+wwWw)bdlR#pX|{(1ih4uYUX0SKn~@C
z-TN{@K}Uzf7FtFRbg`xpq!54<CVQLjftZCf&Do@jEi9aJmNV1)VF?2t2MqX(xw%Hd
zB6Ro+K+8AB-$ZcQ+G3oiXq?Z~fMKdGy0|?rMQIyMd^QxB{2y%QLl`x%z2wDqd%R?Z
zvrF^SO>NWYEF4+m9~>UgOS=S<ehH7E&A}ovRNp+h#osqVt4{L`4KPZCVwmYwRaSW;
zlJ5?9Z?XarQW5sCXTk9BfF3x&5S**2N5CDvzqK_7YU;7T=BrQN<_+`s+~8~y(+T=A
zUj;UinyQ9}(NR`aMyI#)BulF_2?(|ivVPuB>(@kN8<<#FRNPmf{!#XoSnQkh8tSVQ
zB{rVZ#_zACc?#DmZu%}+@!-pjjR}(h)WBkKLc)}pfwUxg2TU#Wb+YJQ%*$)ocizjM
z!jhs@q^y{kDe6Cj|IEEpr?+D7(BAgF+_^h}peMD;QcqQTxkhr<VrHQYFvb(yZui|e
zI082^6|&tHTO?g3f}7fhzwpS*_Yz$6^o83w7EnL0dsnFK5~>TUAjzj%flnHCZLvwb
zONZBft&Q5gFPfeX@t^D&0xnJH!D{EH1zAghuDK}LBLiV(K<IS)oDox+6ls-te_N3E
zj%^CYdRv3rcuQP9CD(*ZFJEPQ<h>D5uxzrsk*lfvpX$4M|1nQ){9kUsHzj^&nr-d`
z>wYI#Dt1^xyKkvTy<-DZI|K0_!goCMY6HBXThKFsgHG41cU&K?FZGnWw2oIXuJ*)4
z?Yl&@7<XtRN~ftc#BVX_E||U~ixTr=Anhf!AlAP)C@M9y*WaT<-poO!<WfD?(9mj|
z6c4#p@YYUf6*f$f$4hm5oJ+v@&^tRKu#BuC^1F)l>emgQ8&e;dL=bZ86KlX6Y!#J~
z)j{RoZMkf+Ic<*1q!qGNo}iXPwY9`HoZ}7|=PVI~|Ancq0IO<?wmzt!fY_jfpma!g
ziy+<INH<6~8<CJ!y1P3)G)PK!cXxOH>)iX^egF5N@*!u(+H1`@=9puzQT5F^VhgY?
z>-7di9p0ZoKFF4*%yheus4#CzO=Y*)!NC&r0r(+vkG)0BSMsNtV&(~3FKI@<ZjT_y
zngiK{F+6Y|m)_3qzykC&9bJpw(*Cc47L$DPk|P)o_ORlhsJ`eP&WjD%H;M+CW@>Ct
z)zEjS_%mcFUProz<otTR7Am*#+xb4(%iM-c*aZXj9szeNBj*_isL}MU<bZ`3nKalq
zO-R>Uqs#SsDt8@c8Z@#zF-ncSy7gJdfRuM{T4C+<sb^gTiQ4P#)}|IOP&dA5uwy;A
zKvtca?(K#2$J=)DD_?v|$Us)3oJ9Y>Y7!q&QM`Z%xP8QF3<ak~D>Mt78noSK%N^{O
zfnCpQ%UP3o>`@7OzW4B#VW7k{v=^FT(HsmDj-HaxII~{4{f@!To#0x%O8_%K{@ps-
zqMXG8YNJnp7x5Prn{d1JY>Os<L%Safr2#>qk=4TH0SFmz2Y{_Tcbu8S(On!I6#4tN
z5rksU8CquwPD#YKseXX8-&Hs>(ze5s#WQ+eDk#IR07D-Qxrl&Eu#v#@V)>Z8mzeZr
zlGevh?nq$pE-23l`KWGJt7aZj{OtHzLATaawiz@cDI_2tho|K<4?Tm~RNH=;s$5XU
z#y&h+p^Lry?GFOt{6}%TU+h(&zw|sd^3P%aS09UU#I&f4NNE0O`ILt1>QxXmH6W-3
z)vHe<+9YPC*^ZiZVykMb*jT=?K=~^%30D)2ODxZ0RMIgqZKxUJdRU2^?9Y%?snwdn
z;=NYKwYt1&Er*WtXHd+aj4Dp6S3New#Z9$2TE9EPLb)xAnBy;e5gN%Ss{I^X2wtF{
z;pxIlrYmzyF&eLTzl5WYhKVEVFH1*n6(}*OPT#`5xSvT@#D<YZm@Dn@rrQu;6a<B^
zW;NVlS-u93rft5#45M#{=Wk@(FZI4o(t_;hAKf2DwV;O9;kQ?h&ULBV7hf1J3`Cdl
zy80&N?c$cB{x=mue8%-89EmAkep!lDxVv<cAZ{Eta90<VFjU%i;$SkJ{8*M(lk|H}
z0QvF$VWVJ|j}9|7U;=Z#AsMr?FJob87w8_=yjQP&i`sbZtWK08&{Ub}?&`eC)v!Gc
z^?b#^Wi8+>EJWf=soGu2o?_Kri7>^h$ICr9B2oK|L^s?nX4*3o`}pkW&qrsk%5qyh
zYj&T@Gxs3P>bD|RD%gpNk+!y0i=F6l%8r|OS`W2|Dak^lB$L*{hGM(?5o`0|`H|b5
zCJZrgWlcDydgJv0oKOJ9;c{n;P+<%K2ubSByT+i$Wng}e;a-o|))&P*=4)3$XQ&Ae
z28lg5QJ^WCl;`>@efZ3uH7ETRx`Ot-wMjt!MiuLIvD8mg`kjnfmfTja5-4{E202-7
zlh5L?qKL`$;I=BXw~j`RHz?AZZuDJO?Y(PXddZHj<;JaRX=r%sdIm{USD2~j&cVY%
z{lq2z%>4;owf}7wJ2s=rK0Y+u>tlLwAdr>5SkRoi|1iZ{NSEgNn0wA*x+#=;>YQp3
z1^PVQ{|4-7*}09X>*RJU+#P6l^<L=z;&8cur;;OaVSB4?c9xJDnXmK75TNvCv|K;&
zi;Ee)w#<QEswb##Ma|QXj-@4(5oNsyeN=Dto$uQYu8y-^raUD@L`N|_OR8UjMBU}4
zyj2ikSSYi5V)QR3srcR{CRunNP5-s1v$8zX{@qz%Az(s`H@*lNsiG+9>KN~>x%90?
z#97$c+ohCdw_aPky~tJ*a<6{VS@d2>n1);+Q@RN1Wd12?#lW}{GfM5)Aog&#aC@p_
z#9J(9Bfb_#hUAYLW!UqCP=CPi7GhTT&V1;;aQn);>af+4n5_%~>@Rg?ffR8ckoD_X
zMnV2gt#KiGyqEia8XSXw#>bi{#U8gQChpA<$cNF)&JV6=H5cbIo`J~H(cRbHJLn0u
zJ$Z6$IT!rx+s@Vrk$1g(XJ47-O5E$LqpseWYa76L#KZu`Lnpb&HzEf+L5*oCh(2wj
zKmU~3ma2hv)1~j*WqnOdM01drS34;n*g|mvZdoJXZ9&>|-T`>Ec%NLIQ6`X)$JJC9
zH?~_mPIuls8<z~%`|k-s=TR?FH5XE-sA$Quik~@%1#pbn)g_h1I4gWHk(A^YmoUtk
zsDg1s#Z|_MwbIW>iaYC}&PJCr!Qy+mv*$4H8QPp2EMsCCU)QFmYUYQj%qTj}_6>g|
z5aZ`R)Q@GY?fl!77cF&D^s6ie`{$pYb1zEN+3|AQprv&rP%cD6n95g&&o-^zr4XXJ
z!^@|qU37L=O|*NDVGvk10Hj3uS7jQE?XZ_JG3&dA>;~=))Jg7ac7oOiygg)L=Rg-H
zy`ggtU1F`U(3acbX+T94c9`>-p6p&?h%@DXA*AU(ApmathRH=CiSnX3>YC=_0Jnk&
z1j(WW>{paz`Xyla;IQaWDUieHSeS;5Y0xr4k89Om$PJaiOWH<e>pQ8=Dzw!PI-wZ#
zlp(HNMJMr9j>2534`638PTZ<b-vSD#K?W7xj24C$3cD)wX7Gqo_4@HKuPGEQR*#-d
z=ScU#t96+f=(3}wPR=yUBPfU6l#PTT(Ku{u=5aQC%%$iz3Uv)Z7?1b-azZ+*aK_SX
zTbq+*bN+VHzN034D5tBdZf0)ld>TZxi)_9|qeV)Tk^IO{AR!-~ZxDaS>rn0X@c%;Q
zCpQED>K7q-5pQlT+}oWht>~VGZ=xtdQH0OdG*BsEE!MSbOvG0Q$y&$;WPaEjz`pTM
z|DvwR0}I-e7Lo(x(#bIPExU94{U{nac~$2%M(&(&Ks8_w)lq@|Xh@FC&gMq-<P_2~
zHTtB>Y_7~m{EDfcXzysa6Xxu9wGH>HC|4+_-)>`1mfWX(j4$j_Bvo)(R#q-R_)9+g
zh4l0h2c)5ZA5s8$d$;xVNQjfuP!ZA4-u0RYYoudhy_X~3CAQvM(=in@lD7p+sU7ws
zNUD!A>7pu!)!+1ol@3+gfqmiSX2bUi%u8qImNm>|`JBxPoccpYoQa8z>GtoofNTLB
zt+KUiZ%Pt2&sY|1fi+1$o>oq}Yy-j*JI-Z~xjH5SYCbIf3b>hPVQ~(9hO?uJfcMY=
z%?JR%_pWPb=J@`bLZ1CUb7}ludHE)b1z=@+G+Q0OYOVI!f}&`^%U0xIJVBfy1$2CU
z;%>Q>ZKJeS$Z``QR#HRY(pf`Ep-yx1{ejI>4b|V-BbBc$W>cHH$70Sq)*v%(Hx%#M
zc$VNN$0xRAe_K(-`0od6RdsIh-iVeNW|iHJ`>CF&sSuoe5m$fP*HQ=OYh`GnADG2o
zUe4hxb#-<*K^JJTyjO%;=Ho3V=MQkyRQpE?o^#{Y477~;cm;~P&s6v}#uPE01X%v<
z98*s_bqCxX=>Nov_p5V4@ZV45<HL4uW4n+aoIBtDN=I6s2(qA((nCAttS1-@l6sOf
zpd>_4ax2^ND@OwwoJ2^$?PN15b@}KSqIPTUvC~<bTPL!9UXnd=ztj}lTEm$V+P~d3
z2x_RZ6e;myP&2T5BklERH(uE&yPQ9|`24xN9G({9%lQ_UeF-u$==6?*N7q<0yZ3SR
zfwTfpF!L$y`0!IVZtf@TB+1e_<Uczc_3xff-c|(zam$OaIA3@L=FE)7#_p~^_->%e
z36U8+ct=4&kOs;)3MESR7f~K5^*lRq!<Jzm|F<lEz}Fw4i~yBH)wX&|eqq=6IK9xC
zA<eVcNTkMfQl4%!6ER=G-u0OZ!V*w#ev4+SXUhwA(>~*qBqJ@mQDl1I=ausIy9u7n
z%tBi2`OMSsh^V37iC<?N4+rOl?#q3@gC4)4#QY;dasBmu_zpbH@$#ss(sqkQOB|Bi
zLhAqwMsiVBJE(|3{RKWKlXY|mtwlV7#kw#F=B5z0&J5bvVAt%+^=j4x11@^c9!*a*
zZ}ogqdG2KtAUS++;RrYhAhF>7$(GrjBM&tW>*(Yii0aA?=j2`dMT28Q8d*ynloZ-s
zJ;JbAJWf^JZcsK`&oAmLYSguHCfuVREdpC0V-|m4!t+4ob}@nio$UKznrfHmWG_Q2
ze?>?zzEx7ceV$n_Xr8BnVQy&b7gm6gGr@jsvqO4ks!>BgunQ-nj4V&p`5-P~yjpm1
z9kTBE76LD5y@QDWEy184XrFyR$H?L*nqfH?rTL&ICI5lc$LH8w?dr1e>Eu8^9jWUz
zJG-|xXtN$@M?vp=E}V4T#&Vr8pG8Lgv3#$j0_Yc@$?C1zV!g0Ajl=0ul`o}d%@Q(&
zg1x*BT>dIV`ue>572_T8>%FC&S)Uwr?4pI!p2__A^}SJ_xi1|TUEuU&Y!nf96&50z
zG2&+Or2lbqT(UaIpe_vD7{vrc;*<84mr!ivw~CVIHKc*ddouj%+nd%>)~|<gje<-T
zn`4Y$UX*7!S8OL~KL^}CU6q5>^!83gF#Kqn{r2ocr*p-+etjlLPL7Ojux~AKnZ_eg
z;FA(uFrR?i>+O=z7k#~!#U3@4Vv~Z=TToG@RokP?*QI4i59?qq+=QZ`skrVkKrjRq
z%=|lDyzNI7I=jmb$4>m)Dxk*!(h+qMAji2(0co=Gbdd?@yk3UIwk4nO60%dNZM~VR
zJ--L21vSghRva>Lh1vkggfwYI$)mR2^J)ZBHk(5PU_CL`tHb8jX9;Z+jyG7ZWo7AE
znC3wV&)f#+vY{e3H3-z+1bCbhGeyq?R+bNjGU$jK5$$Idi`}6xr1SG*_QOsl%gam&
zKlNUr_C6S$O-r%KQX(!J4A;g6NJQSHFA9ZL$GTj9$gpOoeM;aTQ&J-7##H|EFS%g1
zhOnged!QkDc6mJnV5~c0XfxFA#p9jpOA|;+#r=r$hS+hX;+HQh8}h73_O`8bksQy~
zMF<(n)+fe;eryam7Rh~ySf=@?D~8Uq{X(E6U}$}Jw{s*=e3sY#g@;31h<Wt=*Z)!R
z58nPjl>H!9^=B@QxT5p(2lH?TyC+U6g?nCB9K$((FZHF|huL`=ik}s#PPJrYF+wL@
zML_eA|Aqxn0F$G+nj4^b@PwBc3w)U8%7we8FJTO);d+Oaj)87o@6!vLkfa>BdnMVm
z@ds)Sa`Kvz9wDr5_a=>kpewk(daYgDlc0EZCa3&1ui={y7#J$NCXd3@SV-8Z;zW|%
zQ<za#znHfUxw+imA3m)E&t?=__%UDf1NR!s8WYqdn`Z{5K{po_F~T0+WZ&8G;*yz9
z-50P`YHRl%!9Z&^-7c(F-?BbffKb~~{zFnym2{TXVW(c;6oSLy(-jh$yi(Bmq}x{f
zzH)Nvm%Q{}P+I{lP6fVDj_ko<ee?9+&Ne^=eC`v^VHusA4y=-c5<dY-2*_;)1N^sa
z#T$Ip^)*^BwyCL|EC;<fD%_<15uDBIis~_}I_$NKxmAzlqS&gDJp|4=9&z66xWP%w
z?dF^nZD*EqSVE>$otUC1PA;XUO~Ts}rkdVcpq;&oO0=;z5^e4M!cYy@d2GKH?Z2Gb
z^c_MpvxsEWRofD6HXlqS@m~$q1{ZZnSxkg0O~Lt)U5j14N5?Oy#$=UcJ8}8UW5#Pm
z650@i`q}#%nB&<Q?@y}Sq8t+ov+m3bo1enOA9$dcu_t(+GfC=YK?&}5H`%>1Yzu-F
z+hNaZ)g>aGCFS3S1b|pGd&8XGNeSth1JOmp<83jjhb};OrQ4OmsJax@aKcd^g!4;L
zUOa>uw=z6dbLkD`mk#e|n&UQ)Je=f`a(~D3Wem(z)0uzd1tC9;b#J-IAE%3%{S~g%
zTmuVYVw{1!-}iI;RF_y2+5AK^fU-h+Am__0w{@vv$JqKbmvhn7-;;dE3@EO!ot)BF
zmQBoUbPmSAy0m#m8oYq(@KZxJml;3#BdPA;?>Z2Fu`XpP+Usvh+hpGs1>y8o;?D^G
zrA@w?3<7PLd@H;R(N;r48oI4{`s0)1akf4FZIc6{Kh66ykp&3;dAcv|jB`xgdBvi4
z)xZpo@iCZgKpzkNrO3Y(@>^0W3{8yl1&YW#p~o%BK1Z&s{M_85fp5IPGX~*5{z-#k
zmec<|xKI2qK-KUoy2gZg#hSg2Yx>RVT7ewcm}2V(h^C;JL=6ZYeFH>k9o0laCCv1z
z-+=}P$58$yL3LG)H6y*>Tf^@ap!7k5^fvNni6TUm@o^;Mct=ZEt~oABbBi{}mR}d|
z>(QCtDvN<?4l6S={zjWVGBZ=PUAfkY3dlzsIFJ|3EELz4K;^hprn#xmr*&P%XPPfh
zxpFNfw*S?du&Ve6zA4N8VLg$6o`Hf&ih8DWF-Tpz;AcQY19&t|eEM)Ggt8NJcsT?<
z_XXg?zpu~|r`&$N--26l@sx<(1L{sIdXk_l_@4f_&kksW4b1jMeb;g;tbo99v*NN4
zA-Us?<C+o?xmUgst>L}X0dhltK=R7{I(n9xb+e4m);J3F4TgXMW^JlGy)r*A$zgkt
zBhfdE=wbckE28_*{opo|e(MPnfE4qsBFybYC>k{DJ?~1H%^Kc%e@n#2UF%7Sh%%JO
z!tfJlN-CSvvmK)Wt+64e5oe%-i;q)bS16PL!TEGb##XZ;@pG0ZJu|~%68qjzUBoKJ
zK0Zr~dT(ft+CY@(MXD03iKd~J0$9D7m#k#)>>v6r(5V9M*@%w>?8Evfs#r4sa*FMf
ze}7!ea3bxmTn2G;%l}s@KEwH!hXHC|^u_?B0$b5V4~{<!B~CBh(W^Qg;@~BR3Gm#O
zZ8V+`(F`5yfJ|lM30o0d)O{TfUQpMMTdm^hRvXiz((8^tQ+0BoH@PL$v4)Nuwfo*b
zK7E@~k;V<0?*Ny!tiArqNVO&zck1qrce`rSZJPZpSDJDH)Ab3UQw`PN!yAsUODE>G
zq0Uf{W<Y<<$|1Ynrz9hX0<J`(yO^a_ymPFgsP6ZrHR2KI{_mHRG!6FuJYIZJm}w|0
zUnm(G3pAiZ`7z>Kb8txVXtH#ksQ%iJ2z@kv3G}jr7GD4)pIKE${RH$r!y~L2seH@*
ziE|R=JC9Sot6Xk&F`v3Ev`NU?Sg&u6{#a9mr~B4W(e80Jb$AgcPE-+qHlix$<dU9&
zsGqu|x+c;u%sktm<6}plsUoo3`6w@m-^JfN;m#86t=dTj!E8+@LjR6H{f4JkKt{U?
zCQXdZHWT&R?XP9U9|^xLdlD8J=%wb9M=h*=>H9{@li?R<TF@5<xY?~W_eNG+)}sln
zyZx;ds%z&9G(biHT%~Pc^wC6yegQk?Ra6&}(OgT2zGZJ}1=zfx-Kk$vy-HM5hW>jk
z_d!X?AXJqE^g}u$Bc|JIFS~1Awkce20U=~siZ+l|(HB-F^)88<&3Yji$*;oXfUpmd
zz|j2<P%m^K%4(ut$_F@3yhX8FyC#f{UEk+E6<TM;Dy36tCE(Am$TV2hj_gQz3k_4&
ziv_HvTVBTj#(QBcp~g>2stA&uZTu(|qNthZdNmkeYpGD9x6|V#P<n{IzBu{_R~ON6
z=l8sfc7Iycxv~F<^kg@OS(7rN)N0ok$Z6?-R5QNDJ<RL+-it%@-HoAv$yikGnxh`E
zp@E`isyzo6%RxtjV4U7O-rMe7o3#!K^5t%~xLTFy=>9SZDt8xM2ul)&C30w4ueBnt
zHWH9Hc+p%w-|4Lo8qAvkD|KO!nk<QzYPdcZ&h%w&xa!p1BX<Ud`~Z(M>FRV)Q8smo
zUs0L>xoqLx1?N;>y#8({^R&1yF%ZS>2j+9q%lbV4@*-C{C!K__MTfJ4T`#L^%a6S7
zvn2uMhMuS)kRJ&P?VV&Gc$`Ev4-ReTW&?L8kwvIYWO&6L5|#EY-Du10tQ&h<ylhx4
z2*bMahV|>e9$~FmBf9~SnSAVmJCHvE9i?vjjuB|`O7x8uY%*K^_kj5Ny@5F4X@Yu=
zK#IJMrAg{g-hjhVsbbM#7sa5uDI*0$RYurbs=1_$i><9|CSa|IMH!Jh>R0kV<O+F2
zN9BvRT_@*`fLA;?J52V=GYvXX*oz9!kwkM=8QF8(-6dk@tQj?TRaa*_UNM9L4WJDP
zsZU_M`t|g*(bAHFu9M#+11n2?86js)^3VQk@>DrlE1Tl1Aoy#V-*!;m#N(f#89b7f
zTX$O6lza)-1oJC=Rk9)`+Rt-7e~w<qo!7DOPY@ZAfD>o$)c}wmO2zUiF9^Is%vdUB
z7%ock-0Hxqy)8wW+-PHikIIrZ;pO6)P5EzwKjjo3qFKD-n?A{wmQnzTQOBr{puN2+
zYl_QoXes5M<&4svCkHgKqi-&vD2i&#UZtRzJ?9jhzs1@+<3?<#Kk|&?zrx|s)3&9n
zsQUdR1@+unv(9p)7ZMS;fgLNctc+AN)OkM`v(J=CFHx4mBi}2L(`7GB2Vr0UmJao8
z?1*_7{0cKl0bhzgp7Vq~hfBzAWnuY;Z<ub~n#@Zz2;JuLL3xVQ$2jK`o3&{Ln;oEp
z1wrd;QLBM&#9Bjo{_YxRcn)@Ul_{RsZFRLx<KfyIJ>-$(H~xDymA=UFI!a+?y3iUg
z%*Qd9vVGuy{l$N9*y|FJq*Kb6m_HN#0{W&vG~2>I=+il^CKjr886+NM1sE_0NZcw_
z))UudbLG=tzt*)hR#{uN;I}R$7y6m2omN;aN<}GqreYO*yA{71Uz6&5rr<QkK3dA<
zCC%B(647VMKRwracAR4zKiqUVhaS<l7<P!KAe18i!gUA5g;srDQHNopswz4zdfvoR
zC`9r0@$R!91)qC(6aO7s2=R~IKV4OA12b`L5maqsY@iukpO*S6iw46Fj#Ua}v2j2!
z{fJj+mKW2H*rMcyfcTfm%3-J~#AkkM(_esa!PS~lGy=_-+aw!>EaP5(`I9)On>c10
zhllg()|7qlgR!cldoh9gfge6dNHCK=MV?_UaI6PMgcx$Xck^b-L`8peceHl2sp?Mr
zydxkh=rpvmqfxneJKVA};Lgeg4(^Z9E=dQCsHV!wGPNPJtUi~QfKT#zN-EVjpn6S5
z_`j2-|0ho7dfK~T8T}kjYE~k<U4CS<LL=S^ErK9=evSu!G_0y~<JbCw`A#;o>Y?oq
z)a%2w*v6k*<0S^;#dCrJ7heEkSzFgaat*5o_p47x5s=UIhlx(S#TN5ZFSOwW&Z)mB
zUOh%L38_!iyx%MRX8Qdz55P765ry5Z;Zy-RDCzh)?faE!Yhr3}HG3t!G7N!gAn%Bc
zwY0P<GDhO1k`ieDY)tnxDDXids7qMxo#C|gq?9Dyy@fQcBUzk6i8x~C$2kvp8U5^t
z9x?(A^5u(jtEnfGwQNVd2G0M%_fO40+73`4Xa&LCt6vEUTKP%3;r$3+0#5^s4i~49
zM_V)vb==a@VpB|u9#Dg<GfAq-&^P#~|M{pDB2wqF2@MPJI~#oDg<&P431;XLfSTqb
zcmV{INWAnSqH5^qJ`TIWiXx~D>h-B|1lAd$eZBDyzo_2TgXSL6*D`pP<rEB2N<YU+
zks|<oIIZ})NAge*0M$d3iD@BH2}9<kjj8^><&qj*Qi{{VXDu7rfh^onMXkhekvy0O
z;tp2%hGc7H>QxlPj3oIa(0sG2^!-ic?II(nSS1BD@o#3YJZ@goa&PkghlI}i{<-h<
ziE7MS&BRv`QH*YV?_cFsXHX6W_{RX@0t~z&)LtB0T?x2J*XwO2nC7fQbT<cS7|;~Z
zZnFzpRmDg7U=;j*)dDpii{I6sSFtP#xW3@kLd&zc0z$8A9BX7oU0t4Nt^=g*{ey~i
zNbZ*~v8(GQp7sW|odxe%i;9K*+Ozqkp2o)0r(b-Uvii91HwuyU0^pR@yJJYe!v|`g
z@0<??dm8jBtSkeuy}w6bBKq%3K~r}4YkQ`w6V{9`K)(+%2*Q_HJH0aiklL_8Q`|_N
zDM3Snjoa>?8`pHjZ_pXhpZMwSIt70SQcf7}O=z5--&5nZ|E9p5IigL=F!j2(Sz0T}
zZJ?xk+$Ea9J5_N95S4GLj34?TdslL@?XRHsnFJBcrQzz<IFRr(c>Id9A9mKwV1T?m
zH(Z`t(UB$^>{@(?KD8~F<TgS9&2z^!>c=N1PC!>M>ZMg=H}&#BgsaYt0IWj-Ew_u_
zHeD(I+9pDIN=)!ZUmEMqqNuY$!DNm{ktglp%te>5$6_f%f|3;aBr?mv(hTa-U+_|i
z&d?JJJ^iuV1lbIoKOJ8py?t~oA>)q0G0g!SOKqK|Qn;oN;jjJ0ME-66z?>62_+0Q}
zMx8Gh`47;Q9Y}ztGp7w-reewvoGs9MumoZI7l`ZxYu)QMoAZHl{PPOQ7<9SwNAf?^
zyny;1fslYK{%@qW%|Mu_3a)#CJ7m7<YHd<+AQ}PrLxnS`N4d}=k*>z3VXpvQC`3N@
zh9y#%_?rDnf`i4h6Yzx`MVfs+TX1O;6=qi&_6VLWmIH?tiFk9f@#?SmiUtN81g893
zAc}O&I=}VUaCt(nyQeK%PF#4T&w%JE{`D~jwadYuWmG9Wzi&=?qxbqje`>kulXC$y
z_4n-cC-MlTyQNP6nZAVY8qE5Yz1iAYFf?s0Xd!07=e;P-ylFv5YXEH)QCaKioImro
zD5~h)(Bb|5pnNky^U03ahO}QSbPyG|NyVf);Ei3GZ=6Jzt=lXoT5E(PLp5Yo&0WDN
zwj|*B(^B9<E=R6>J!lK`d7F@J;GH6vdrnkiu;O$rDey+c=5)K~Zo+=)@5TO=XOilu
zmg^M%6`ovH#FPqe=YpW79C=DlPftg%4{{8hc2innaPO;!_Pm=E(^j`_)R%QvNUi$%
zFk7?b?q^!G*Jk-8EG&d|*<SrXh$$%eG9v7u2RwjrEy^7Vsc_l}D<sip(7e1A0%B7%
z4-g2NZ)%N~);-c<P=S<py0ceXN`5;*z~;+KUY5SQM@lX=C_G@_{dMa{h!p`4M&QW_
z_WVdR5B_eYn(CKiGLov~A1_fwwuAiNXO%y`@5b%!dV;C>hHGC&>Lx|3w=?rVZU~_T
zXbXf4&sV2OuoC3Q=<S&p^(%{56EqUQs!J1~Ap`od`}9_LL1F@h@{6D+);7Q|LQQGV
zy_)cNYxtWv`Wp9U^JSkbc|^Bbl@INh<%32bp?>Z@YiqO|2s)CVS5{h<SIc-gf|m~b
zfQ-FXBYuM6j_A6MYl~^L?IoGhZ5+ufC@KO}3yoharJS8A7fiH+MUlaKEO-G@dVpX8
zM~KG8En7@r#U3g!Z6-L^rD0<52fc)s^UIABhNLof=rxlUj2}FeP6@PugzEb6G7p2J
z-6vC;oILORF0B3=j}MAZh@)Ur8Zl5pMMP-pn0N}Tk$pU;C8wbNtWfek8_LPqAW96{
z&&G@j|MKOF2OqQklGhzd9TI<qyQaFjxRn(>3=0eE4K_A+pz%m&l$LwP#nIYCt<zCj
zqW`-`Pk#vSkNy(g=XsOCfK3ZEgg1oX3l-{TBv$#M9%b>g#_0-eag8qgl;%+_iA86%
zRKQ(YSpOTJMMhr$&9%F~EQRu~@71d|{L-@Dtq&t^NCH(}izc@+ojKYpP481DpAg@O
zhffZg;9}7OzcJ^|_K-h13Su2eCF)D*iC2)k1`=1tI({`xQv#2w6eN&;g4~sTWC`D!
zZuyvO@uFia9qtJ8s8+?^e@&}|t7~ZdXZv8u(Qf&xc%ZFVJ&C@-0n}05pkg=wjG;<h
zH}Katq&wXHkdRFAw$@R_EUVico;#QOHcEvCKwAOcaTH#2XYdy@s<mY=JUwkGRGav6
z<nj!3*el}z$dCA+Z+Ni5H{jium~-Br`^uo1sn^PBS+c#udcO9T{+7o+e!2StAoYqX
ziW{d|W1uLXrx}J@iYNGaAd_DH#&`!{z|Cu#=LxZ*q6S#MBO`%4F5Pm9rTpZCkMP3b
zdcgqPMbuD|F2L7#80bq)L2AT?hD1u_MtX(;4OwJB9Xn=RaRkZ8TQk6R;_vQ8+@Ob?
zEFstc9iU#N>Q==>@B>LnCdwrc1rLls70(f`nS52keuk)<$EU2{lLUotZ@23#)!mia
zyVq}D*gmJ2nAqlqeocvpr0cDKKob~{K+UasF;w;W$ocNsl8I$IZBav~5-m?(O4oa@
z>0m;7{<`N$HEBfVZf@&ih19DR`K(u33Kw`4cGlTSyx=YCN)^?gKVw8xoFH~V6ch}4
z?*FZryvRZkYHm!xG*H}J(q(8RJ>-cD2|zCjeB*<LNzzL41_zt_G|#Sro}NKZ*!P`}
zZNHns45T7hSY7M;-A92199;awQgf>wPh4X6nBqn4Q(w>;?!&;@Z)`82qt^!X4XE%|
zdUJc)jZa$BpvYfQPmSDzH#-N0y+9euaCit%O_2lQy|jP7q<}6T)1)9>T+q?%Lx;-5
z-{qCc@yWsA8H|Tic|zRGeA&PBKYyZZ5Leu*Kh$42JEnLf<=^R2pUjs?WJeG(gcAAR
zq3#4hvL0TScYF}QYQ8d`(G~s|+Pz8BB>dH1q9UjHhT#W=o6B9EQX|@(dEt{yOiy^3
z&E`l58i2;g$-icnsxA`Wagkd**>t&8w0{CvKil0b`NDVa8=i^D$$bzP$1(XoY$!!+
z-1ELEjIt)#qXeBdWo&A%WXX)~MMUVgS$y%Q`-TQrEP=+G0T2jOf5t5L!3s0F8qD(&
z6y)ACM)wp1oI0$SmE*VOjaR1rPgP|BUjvDxbLae~uq=l-AE#iznn_be4RI#h-R~Ct
zUtqS-L8`cqh67$5q%-T~A`FxJT2}r=b1YcgX!e2bjf>Jcr2-{r<uTsr(HSnjT)$oA
z%!h8~H3aQ9j(-@?Jc_+=ah?ZI^{&EUZ9yUe^{Ra@tb9W>rCd?b%A@oz&0j+0FY@ww
zHoDLGUb)wA2BOz4F<3Beu6tZ%BtNO8^Zy|yBk22C`X^;+WU1e$l)s~crE-&*VzKda
z1mlk@(EE3odd_S#>I$Q|);bb~-h|eW^-VzWiG|ELz9VpzFtnbd(@bixgSOk1ygpYW
z+40_>2u4#=`E}HNQ64K%^E5&#roO4T-=d*{$92_|ONf0;JhT=cX@-BL<Wy@;yvdV&
zlC`xu+Ay^0vNwD#xg!={6ROAfSYxO*M{(9EE-_>)jZ5j<uB%fl%dY6}ckI$E&2-T5
za6%ffNQhgFs+0F_K%%bsruNNNmfq*j!$UXxC`xt$$)o6KkB5^4XK6x=Hz6f=TDY<1
zG2Y#5oe~hb%I38NF*|=$XHNfTi^jW8e*#|Ip7+*FCd6jOscIBTSyrh$f70#*rzhqf
zsPWCjs1>}RiQ08s&|g_zQGk@^A;M|9nLRs&DVE82EBU7%w47vChH0}HnKuK9^h7VG
zxE*Xfna>M+EwMAlGbPG(?)bP9iUnWG9aykii{Dvpu!g3O-5Q48yMIpj9I=ClJQgyo
zzsO{r&;jb))?l%;8{IO>1jbV8Zzx+x2_winr~D<^aqsJZTnZ9tdiyvDD2e;59Ea5`
zTgqK@lsbue(8_Sv39eRhmwf4VoD<BR5NkVxf%wDx6DvF6(|6DE{h~ffLMC}j5{m)u
zfG8@)hU95;aS?R;odLTeUM2f)Vts*c3sT&iN5sD%E+~76+)+kchT>|CM=8g>{~~1i
zhMIqHZm}@(y>VMq0JNObIOgzebC`)kluGNSW|GA;v*_%tbWIGwChc6&&}mwGz+1m{
zTTBzHP9BWAZC;$ijVO+oxGdhsxTy}e=N^Fp*Ub;5$I2V>s}ySwOJhF^ZE5lDDPF}8
zm~2)H?C?0{j#n}wBVTf{e;(_-Q(w9*6KAK1x@B<Bzh$t<{|t2?rcv_lLd~yzO52(W
zTiw)qW-LJ){e?|O<w5@PMVUbrJk%N@HD~L&zs+;$NTi+2W0^<5PPJoY1g_~ZV)cXN
zCvi;XH>BPuDC>j8bSqd`o65mMHD3Gy(QoRwtULuvG|A`f^vz5Ql5R6^uc5o)+K9a4
zv5~D4@J9jaTaXr4{!xmi&JnqEe7lf-qu+C~>GslYNrC%A4+dfc?J$sOYflN-bHh|r
zHYd_$yn|=C$112}P2E(hS!DK8BXbF`GU0b$8}RGwyjG#O<W^v4a3P(0n$R1r{iZtI
z-R)#+oUgfXH-mGrP)7Po^2o1ms#!96>9-f-uOM!gLwmPUN<J|bVoH*o;?u;=@E*Eh
zVh&NO1P<f#kvis7O9JF^vBdm;ZB5x<)$WV(g58O1RRrop`$j$xs)vRj#1(DbtB~kw
z3p(x|A8<C<Ho~k#vn8bx3T}jXd;4@sVk4{nq*7JSd`%WO7VxsIi%9}Ze=6d??`xo-
zh-Ab~CRfYsH+AhC2oHWvQZ8V!Z$1Tli)bJ6aC-mzyF4NoPY>N=Z{G)MZ~xhm^%?~X
z362nRIB|x7zLIlc`F5ENlgW7Tpv^@oqhuCV!4zT;HnZ+Ov|S#iV83^7ge^DavfMID
zTj^}Hu~0*ADP03`R<QqlHVjSE@upCQ2ppY?VktYm@RK)+;`PSC%^xNa*{H~dyj(|~
zjQ*X`W7^jndr2)@EioVVPQJ0kp>gA9Byr{RxX3as@dk{5UGVD;?{-HjW~x=A|JkGt
z2CM(!`C2pB12)s2R>PxJC#bbMn@9h4Qqd@JINz**Cp^>V1lr8$!`#VZ)8BDmpc2#7
zjO{LbC-T2yg3JiD-<>~gg=&;%2I_+@SxobN-oLkY;R0JojrzMiWH!8#nyfv0J+r^e
z=)7CVdD5++%NF}TwtKGw%aW3#uY4ba{n3gIz8}tDs_~wA6O88dFAcQBp<K*h50%**
z*+Mf*QpPKh1=WPrDJ<e|(n5S@vb9U7F)?2d^2>VkWOi){8Oe?Xt1`NW-?HodY`S;<
zq^b78*>%+{kKli7R$L!}kb^y55m2}{3pOV<+zjbb>Ztv@b9ZuN(Q{*!McCYYa-PkZ
zOQLJevJnI?zln0&oXYomv0KfbpbMpTy|om^vH7e<GHq8w_#JGD+(gS{xdLZ?iS_Rj
z;tBD5O>#XKE;K^UAWN__p5xn~NfOm=!#CA)XB@SAW#av~xqq7MHvRAZzKNu2P5yML
zLXz9v_gE0Yv*3>vy_&85Oxafxs}ekI5D7gV=QW_BVue@h{Co9K@{^SL&!`e<X~&>M
zr{Ey`RQ7vUdRi#SWaJyaK7P+t*M|5^ol(`U#$gb6;u6q0-^_5CP>?|(M&g}OJ@H1y
ztr8U_@p3Bs&MjaSXQVzTSl?i??z31f$G<za!8YGXmioxrS-#9W8JZi;ZjewI+~uyO
zxJ9qXCP>+WXd1+$MvAiD)o5%?u&ahSE5{9n2#3<-XlW7f+&170FNp>Wjn$Sfk(Hlj
zC5>>UChi)&R$Fkog)2F<?oAi;)!*@jgw)odz5Mrnl4(-sKgZBC9LpBCTw%Dy?$>GF
znWg+r-yYdE=M)Vz#d=%d_uaBT4IV6AhlW?B3cYr%v5O-%{+idN&lrf-H$*|I+C#Ky
ze3bYqfh{Z}`-igv=0AxA#Gz-%mdI5z7FFsrdAjn^8}?fJ<D83^8^M+CqTKfCxM2=d
zOUoiFHR>jKm>H*0%t~u>8Z=8|vO{jqk4CXRi%CV}hfv1eer%gai!S;RoGXvTFM9=L
z`*7HjiaiX$llflUFWTc9pzC~F^USZEiG;wFCv7u&bz<#1ZGB<NrepVAJxw%z*xL6b
zMu?mC+tiLs>s@?6btBhW#kVnJ;ZS2|N19=)#3R>R-m^2UsmpsI?Q+|8=pSeJ|ETc+
z7w*pP?s5)6Z{(FVtz(ubf#IPZB|BQk%8ieYPeGtmUH=w|C(=nZbod=x_8>csPkaIq
zxSa~~$@Iaa6diTDWv8IduA;^*l1)<hRFAKNa44FiMkD*V!@W@XbxQdZw1VJoCgITY
zq`L%3LuS${M`UE=L-qNNA*6FsG~o4zfuZmn1tULLMmur-OCHfMer66Z8ibHhP{3aA
zN=E+Ce}zQ`JEm(z+u3)yIE9whjgSvqa%H?}qTsctwrqN3R`{ZjqWs}Q4m>QfZxLMh
zA!k)z;8U2--WXXCDt8=hJy!vDB1;mf=egI_*B9l^jT3xk3hauMK6+@_a}pG7mHq@{
z)7RA{r#U;(XPuIgVm+J+o&9{`%YTaIK8ie#d;U7x6YTtV@1Co}SrtWLZMW8IV>&aI
z393ph_p$6>#K!F<FC)4MZ)~r<9ydz|i0M+sKg3-6jZLYV)ok!)8{$<uH+5=0YV1om
zvJ-4>+wyFnx^&L#u?YJ^h~W+b_}<Y!4@z79y6l1L_ZzOHyFQ4Ut;GU9mRs*}U&*67
z<gV)*<yUifA;V*Sv9sN<nir<HBpC!U&i28DCR2jH78VqupZyya?4BdjN+Z(*nI+#Q
za8ht{3K8s$D=au^Y9tYh;7r`b=KB9@!{EUY2zH|*><eiS3q9UGGmS`0uo;-5m7K_;
zJ=<0Y1M{zXZ#C}!TdioC$^O6r(0do(HX(SK&F-qH^|f77MUXEellFnN$e`NO+Yo?*
za;ZrFy^zgk;1g&)j_Z6U7ITs!*g|$E(~bpVVzdsfDeY;H*@es)R4P4zOvBA-A84^N
z2KV_No7Kj#k;Ea0<DT8e;^q|aUVJJk*on1Xy+-ps@2&WE_pwhzQ|+6Z1`@+;T;dlN
z@FxZa8*y$APZmOWP0u>yO0j)3uaZQZmm?PxdH(^QFxWZbQv65IQ3pKN5=W;k6*n)c
zj*d_2?_WC4ePd|X8p=t_l&BWso~kJL6aMejL)qx$j^9kaFg})eW1ArO^v94tc-hM6
z2Ia<eJ~=p^BO|%-BeTi4wA&<ns5s9h&&A9X+eh0@@83_KzXK=4LFUB68LY|jIo>t4
z``0-FHq7^`*1eNw-2hB<9Q+WW3o@xrnWDsP;DXxz&3k%cqO-3nj2L;WNSW5<OLD8<
zmy9MxLt>GX=g<F{+cZ&!jEp2Af?tcb7Z$L{JR{JKjv|sGGKzgL^vkRYb)Z!aJ5dX`
zTBtlN$K)bYvFaaSjr|BCf}6NylN<))ZRM1U;#ok7iE>{+Qu*&MERgK;?w{upxS#Wp
zNhOwZpyYZO$9DGgC`e2H1be2BSba(H^t^|Og(Vv5`Te_o*#RpZ6q%GnD+1iBO4(5J
zmk%7nqoT+}fb0Cv=7nOm=lRR_H4#ZkML&N&LyQWZGC3`+6u8~cS2#E(RyfMktgK4F
z6$$Op{89z4O%<eWU8No>zr~}rJ6_Vf`3RPaeRogcBX9r(9_u*?oDQmQkri5)BzUs<
zz(F#NMP&<|`B6+?aldvn-?z*(b{4~HraNzb>evSpsWtp)Rcq)We{J{2plqaijUTve
z7X-tSaj*>rsGh;jtJabr%>XRwP~x5#Mj>$Q-)zqXV0q@VwJLx!Od>XJ`SIL&HOBCR
z<b!0EH$1MDgM)*5@Tm&zW|X`oST?djUt$5mwJP+u37ytxfBp{^*}Zuww#Ud(-xiQQ
z8Zn3uF%>*Pf>rRDnB)Ruc@hzmk=%$+{QT6C4;oE?f!O6rloqiF1}=;IP^JX<VKR%b
zq@*IS83d67SR54u-%rM3b3V=v#$zWGF&WK|AtWL~d^Sn?KflHh9Y2()#114%2(6d{
z22<8Q*&K0R3S#Yi%V@VLUtvD=+vf#Rws|l<cW#DQ6eT^q3~(JEA&F8?t#`ZHyxb~M
z5*H5~Ie-nrqPjAG$C+mTxSD010&soHTNQwrpbs%cL`P>^)?WzLIP57(N(Srb=nU7n
zIAGv&DFgel-yDQ6UcNu*jh0YMCIx)TZI&K(9bz1Nd41H!46M&}#<DUqDS3D_flZle
z<=2#Z_Z+E^&%YnlX^{_RO2nB+-b?=dwMyN2(T^-VG*ou`a(GxiBQtZei?(6-=4`I$
z&-aIF;m<sH7vxZ34QGw_pRQ-dS(I|1c&O_A@kEYgwM@Ks@80oxTF54pn^!3N^=mO2
zI=bEYzBaXH?fB*on2GS0FUhw@Nq6B+R|9E6%*KPn*gh#1Gu0zs7+{K*z-A1bHnqQB
zn)_r}&Sb8>4jj&QwHq6JTp3)lGjM>mTZvFY*t{#YBTKiJpwPk^49gB=AI&!e6Yv%~
z9j%Il&bpmxv_h?|6u=A5N1Ghj(^IRK_Y>UiKMbBlekUR-Ec}yr#&IbKxZREVAW7>+
z(r8eM#4_mnb2#kg1CK~>MVBXArrYBsoBb)@i7Dvm2Rfr+*AEInbZ%tF`k?%7V_P*A
z&+2|MSEjMaLX&rB3;Y|5S0^?zUNVa~M<Gu!BX|QE$%`VU0DgFF)=2wWuMn1*4Ga=;
zR7!>6#2V1e304-8^b`phS=WXwB_*XOxWZzNCmuLDZ+x7O*WZgw!Y#FObVEZ!iACOW
z+UJ+3!>(V?Q9qbJi=@>m1V-|rIXO8d{mI`rY}b>w3Tq_6FM+~IvHiaEcLqjA5FDU)
zA|}JxzwGyB3y{&VjHZ5@gC`+}Fg3GX?`H@ZMVi;DzdjTb8Qg)-^}3)m_KC-{$p9~P
z=c69hlkKuTO5pVdJnJR1<h0yQB7U1sWsVl8xkQ=3JT~>B;|cLSk}hwr_D&{Eiz1v{
zvLq9ZqZ+QgPmW={rO=%__w_#1>>nIFx;~J`>~?u9+dkvGRS-+3{i(<2zH}a2f+WKC
zF!cDKogh!M&KWFI=^?NnGuUlb*MTWGmA#v?a`bNPktAYT+t<vb+QP0;h4%05s{sUX
zxkPZ6jnD0z&G#Jz<Xmutetv$jJZ{zQ*9Ss%x0hR&V1AiFcJeT0VqRXb$NS;3h(r>P
z29aQ3|67OYl>Pm$D`3-caGs$+Y{v4h3!m&Z20zbK+m*4=!}9XA8Z<+P3)EOUH5#Q5
zQy<XR6jp9vrOC?5h9@KxfT^MO^RY)m$~^Ks+<&x(LkV0f%Huc5V0^azJR}&dw%ZE-
z^$Ysr7YXB)u?-JuUy>=(th3##J9D-~fAVA$gLOJNJ|53)|Ifo%M#JD)r?t0vsrQPN
zU0rL_%Y=^a@ACrJy)kfUuU@`PQ_^($T%cMZ3%vHg<a+Vq#c>F9&5v-M?2cg^N#?ts
zw{$OiiLC<%7M`k7W-xot5!m%1<`yJJNJv;4<%@x<xEcNJ{zK8wCwOe{qPi4_s$j!3
zz{gNjN-C;(Vg7-r2SI077h))gML^h-tFhmC_2vzEGaay_&Wy9HR{>6QA%};DOoqL9
z0YmrpAA-3o?UNctp1|caxJV2$>8#R9GaE0~&e3gZYO0M2ko=3%*zf_;s#gyMb4_}5
z?7?+ExBy}ii8vMod3pI9T`+`X8{1C2Y$wk>HU<Bn!}Q1DmPy115tNM~50H^_jE6Eq
z(}%L8qBz<%Y_V{kHa1kwSRg_`D3ekDyMQ5B@uvY+$H&d&HsS;3_#s(>3`qNyh|BU1
z41A1m<qRDzgfuw8FJ_B;-ZI9*WNWB!J~0uI1>=VKta0Np3@S69VgX4r7o5suy0T=%
z3Z|UAevJoKmHsWCV^1uwLc^j~4Oe?`|D!@Oc+yc|RWXAR*qLk4y12NQ-8wovEVr29
z;6D6I6q-&|R4XeH%Tz*13PV5Mo6}lYSb$W^41!jB;<Bcv)kO?LMpY_IA`E(BNkm#D
z8&Mh|*V~J=e@rY@;~#h^0GH2^Qhl%(UrKSM)VZ8>?%0cQTsbX*;UCzZC?h#gMM^e!
zLmKc0w$Gm>l{9*Ou+;T-s8EBm?tC%86&ZJ*-}foOXgGzOKXv=p!kXRcfTxfdc&^>u
z!<AM`#3CScqM@Reusw(6#q+qSKIK0C3GSy0Z>Rx+Bw`r6i5!vSIngv>?o>oX|4gp7
zAY9;O%CoYvk^@5R=6FdO$`ib7s=`W7R$UK*uAZKeP!h3Ygav7T3O^Wi`Lo>_X4mtt
z%$Bp$n~ty##YO@HgkTiKZZ1!TmO^+f?C@Y%HdX4Xb<P!v1*#e0;jh>eBwJfrvgNWQ
zs}I_6M=~XF^N8;4zx*dsB9PjAn)3H3j|C(p1Qx=DJkU5C_mjb{_2BEb<s3L*16;og
zYj2N;n!0Bb;XMB_2s|hlY-O^_CQH-x;B#whYmRoa54dp|kUmF>w0J>SYW5Ea83VIX
zFE(>T#x&kI7N8sOg_%f1Mb)JyDw)Vtg*J3izr3usS1%$d$+LHE0}<SwyCU9}@c1{U
z!?F`fOG}xomxX}~V1A1^Ha2$W-@pB_%%;b^oO{Bdt9?mD^tx@sw%r^l+3!H~!XnGj
z@JE|U?Lyp^Xe#0&NnX8rm2KFUI9XvDH6nxfIc(*ht0$?wyS<vVW(7$LMEP97z&9zR
z60zj(pm!fLjXO7NH~;=3nyh!L!J^g7pP&B{Acx?VJnld+0tc-H13B9Tg81fej$FwU
zMzge|V`Wow^9XP&ow4qsW`6rmSqzGMkjdQBfamGMCwM=JPe7Cws8YiQk(2<sF0rWD
zVPUqIR(}8fy~$W%0>0}$9|<W$8d<XwNv&2(HZ3Y97Dl`?Ss^BuDLw!YQz0<EJ=vX6
zvuwEeZEbDc8%M&(C?^(0GXmzL?a7D|6$ghZ;#fnOL8OxL*<M}`lGM&Ba?dPPRbvy~
zFO4gVhXld@Zw_VzXgY6@B=ERhdLBk+=73)e!eJI27#PUaYJlo}FmN-0i|U?pGwbx%
z?+E*XAe@z%T`(RXk0TQGS@n~*9F9krH(WQzLu)<K6YYp$9k#~<0S4w@GI4u#HkX{7
zOd02Ju-FQ`al_*!4NJDd!onir<MR=)k>B@;0tmvNK7A53gLYBjpj-tjHsflyrf3|B
z2C&cw7(F_lP8e3$?@a0-FH`}DJ)F$<UDk4{%rNA1dm;_wuWY3v%}apeFrPn{`Q@vg
zmTP?%o0&r*7K!kSY;7}}tJj!<TaE!hl@H=!j@#9#{%S{G-<UakCKLSDs~Y!#)X%Er
zM!!HB$*eFP123#}tXStj#l^+rm8l;-gNx$Amnzxnb&##@<U!Pnjjz8xpoPK6+YHKb
zOvegC)0s_2*XRFyk4UuXjc3Q<ur=~KJb;5}mTp>9_gmoCnc80cm4%=6nYv+bybRo8
zCOk1wGo|)3i`jTG(tVWSOCqN{=R5cF1z!^X0UNMPG2YN6Ljn|P0Oey4@YD)axfCM&
zFW381K;occ75l7kN{<-odQVS4<c048@a;$oj;kugbCJ$sIqN!muY?I?oLDxiRN@?v
zwJ-OZpCUjZU}c=R@K;t>DJdvK?mv7IPF$b{mqf?<K<?0+BbU`_0XGGs%Heid&IRFz
z93CBksY1FJ`hq4(3?wC~7@Nm$Ipptd&l>F3``?4ZIEOsz>FgW?GsU)tc}$o?O`uS@
zWHFV%K_Vyk=Bk^qkDQS36NrVgH(S<K<#t<&t!-_hp$IIFn7hL}FJHVUcfGI{F_ao~
z*xk!XPX5Syd!BZAx|4?qQ7IY$h(PkY;doF?KQ`0-a)WHoj`!v`Onm6WDZuu_CN2o8
zI*&d&K+cOE00<l#DtgUiF+GmQ>t1^?AVd%YhF~qhu*nNGlbDf#VUP<H1Tjm~7I1cu
z)Z{`Z0LFR&CnF)j>?#4XLUxUUoScr+VR!qj1E^#mbf~|f1LQ;kxlpIA0(c~c!QCy8
zIe|~9dGD?PFfhxG@4TK8Fk8=YTGYHi2hJdxum)DgtkqMV^DjBN{-<hASFJntaaLAV
zvsM{9OQ61Tww$Z7F^mZ*;J`=3k$Eb%f1coG59^$~tZXO>29EspC2+d#Nq{(&KoFG!
z00Hy$>#XqbaM?9bC&Y3(6cPj9H7#nJ8MTe=!DoHkXYuJ}TWygJspT%aRaxC(4VZ7~
zRhPGl8fa`1568_|xR3k!L(}!UA{C5+Hx)Im96+q924gB#M@>p92(aQWV)L2mik2Nv
zkIkl8w+CXU0SDzMD8_I!Uq#;_JHM0bpGhBusJT2CG@Mo<RO?-<KtvAz`7>Eb%S{0+
zQaJ$09J+x^HYP5vnh9!bV&bj_Aiy8uON0La>${d0&`x0WhBhyNqF7p8J>FqzsXZhr
z#`OC9V68_(4-}^Q!NA@+dKS8K1Cw4I0yAD-SojBP0PqykaF`+4*mf=eY}v|e7G|#L
z-R<w;y8=+gIPKlRpAhhD*Kt`)S6Z+61qCS>85vdSQsy%JV@$SBf~1HrimcEAk@@m$
zujuyXYD)^b1hYt9RMdag^+=))grd;&_xwKV<0ZNvwH{kISU9ah)>X^aRW#>3T~zY4
zHQS}u@QMjD6EIK!m6ki~&C0gd0w^#`q)O}6wE}Lda{hohrTWxTKtP~s3gzKLvCwG?
z2fwLa>R`2UBbi+;AMI@2_Tb4%E8VIk%GCe8xPL+=O|14C|Dd1|0Fd(<Zgwm~d$IJB
zAP}fatVR?Pv3+E6=m8tC@HwKUA|m1n{vp$Nblgpa9)uv6^wd93MHI+;JkJiF@?J}C
z|BjA+v$p~wj%22>=Dw7fnOV%46PVD}j9}^SOq2<Q4wve8y?|pD6yW0F<%7BU@#<_(
zNl`HptPBwBSet=UI;{~zpqv42M_)il$b~j(?~GUkJe}-FJj9Y~_$PBl$SNzZ*KON!
z1k8Y$!nn7Bh(4XWM*w}$oSg!XU>Dlg1r@5x06PJXb3ie~u}T5|VmOGkh>`$!>*@%(
zE4B|JDdZqd#AH0!;^K6BeX$Y@tkW%_Dl!*Nr2*Aua6*`W8Y10B91e=nI?;l`uNyc2
z^XCr^yLCG82B?z9z{Ja2<}n1hL!-t%!dXpQ8<E79Z}#RA&9<iPOaex=se4lReUB0K
zK7;W|ZsCm@Ho%OV18Kg*8)+dtgMGYr>YHpK?P*tg4Gq&(z1$Ge6sVW;XowM`?EB9s
zzi>X;%ox*jk;qdjQgn8%1{l58G!G=3{QM6jq@*Sw-vCS!=}_~WNH9ElD?&;0Kr=WX
zU}(AHpNt2H+`SXPM*%#ksH2k#s(*XFRufPl9(Rx>IflUOYGy7$eKrC@4habfBobO!
zcRKlk?6P_}RID}%kwOYI>r??pJ=<=-*n--V$gKuIK}~j8WJCstUmz)iua(5{{8X?$
z6x%?;+pe1ibyZIi@7$Hkp(EF1oeLa9m~>1k#c)n3dHL|O!*b&xP_1=rq9Q`+IUPWx
zFzF_MMNh^d-4F^I8c6`MWxy(QxVRVBllkMvqm%6kIbcA2U||Ewt4)8ZLUMW~Kp-81
zI?VCzssKJKizsLC@IExt>Amp*WWp<K?3_eymrZbw!wb#6T|KN$hkBT=UabS@CJ)x6
zLokRorXb9P5(8Hj2v8LyzwOxYW+5UWsd%=O%1SnGA0KA|HCx+q@JQ2}nm$PY$VgaI
z6b-(uQZGcO444B2oh7zweG=hF2}HG3y;V5W`0YN42#6Lqpp-v51C-SW2-^kfRk9!!
zgUJ#_++jd<@tf(x2k`_ByHMi)H(*)g6Brn1#B+EzKd^=LGOJdm9vC1E`jCYF4LD4a
z_3X^d)O&TnVX^F^U82BZV$G!Avu!Fj-2?Wsjq=xrub&4TJGcT|aut+c4;+5lb2Kp%
zI6Sx&Sm*r()|cx`u62iou+4w}JSKE+z4mDX9^KeoTaJ8^cn)ev0h@WxY>jM`lonkn
zd@iRLw<JS;NB!M*t01wm0XQBBbjvMuYwKuW9t;4c9bnIQIj}YXj@Q&$6+i0%w#}lp
zW-SE;Bru~(8KrbcnPe2)vs}OTTU2UFO2*SuQ-59G|8FU<bpz4@EQc?F2Eh7ctyk6m
z|0iXT(6FoQZPdG%w!{|`-Q@y-^_*#O{WG81k4MF=fTM3f$yLDi^^1MK-v#fj`g-b)
z9dMi{dRxxLCRT2f;%7d&H#Ru7wzj^yyL&sR`P?b2t_7_2tG~b7`=T~|>Z&{YLlrG7
zB2G`!y?kn__Sbj2-)8|Gq@|$|0J5m;O+>Gxu^X_}UhTB!#*G_6&tC%L=i9sbyQ^Pb
zUY`B))6-j<)BUIF%mq!sUlDDKEV%*f893-}tNfhy-t!Ky1-b;-W{jM}F%{@LJzyUy
zZqk;Dk4blHKA#1Sasc(z@Ozs8sRhS?z0WLQTj1XRcR+`hCIZD6<UxI5h6b(eNk_T<
z{(WD69as$LEWfOI*(9}o*0N2TH%|ptalmnsOkfo^X$dg-YHMkodUOP+vmrl;7noTX
zLUz7+lLHz%t9rF^`6=1-?YB*nkM&4-0gEe!E4JX`ox%Iaks}$uzrD@6wPhuw?=k<V
zAZUDQMG?^4%b@IbR6ITd*l7Uuq<~Gx#lRw5^Xu#D{=oasuT4;N2KHT2fTJtGmgjQG
zWtp?W+JQ_kh_RWL!7x4XGNU@r@)?UkzO4Z5FQ4vklCenapZ&9^`X2jj)=veg^>p=f
JS?83{1OO(BO4<Mb

diff --git a/dev/articles/layer_list.html b/dev/articles/layer_list.html
index 79637a97..44ab26ef 100644
--- a/dev/articles/layer_list.html
+++ b/dev/articles/layer_list.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/articles/lazy_tensor.html b/dev/articles/lazy_tensor.html
index ec0d85f7..2836a1de 100644
--- a/dev/articles/lazy_tensor.html
+++ b/dev/articles/lazy_tensor.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -402,7 +402,7 @@ <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.771e6d.x.output</span></span>
+<span><span class="co">#&gt; * pointer: nop.f74754.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>
@@ -449,7 +449,7 @@ <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.771e6d.x.output</span></span>
+<span><span class="co">#&gt; * pointer: nop.f74754.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>
@@ -499,8 +499,8 @@ <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.771e6d.x        &lt;list&gt;   poly.x             </span></span>
-<span><span class="co">#&gt;        poly.x &lt;&lt;UNTRAINED&gt;&gt;          nop.771e6d.x</span></span></code></pre></div>
+<span><span class="co">#&gt;  nop.f74754.x        &lt;list&gt;   poly.x             </span></span>
+<span><span class="co">#&gt;        poly.x &lt;&lt;UNTRAINED&gt;&gt;          nop.f74754.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/articles/loss_list.html b/dev/articles/loss_list.html
index b31b9eed..23f2d13c 100644
--- a/dev/articles/loss_list.html
+++ b/dev/articles/loss_list.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/articles/optimizer_list.html b/dev/articles/optimizer_list.html
index c6c6a226..e2dd0fb7 100644
--- a/dev/articles/optimizer_list.html
+++ b/dev/articles/optimizer_list.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/articles/pipeop_torch.html b/dev/articles/pipeop_torch.html
index 88f56842..2195e3b7 100644
--- a/dev/articles/pipeop_torch.html
+++ b/dev/articles/pipeop_torch.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/articles/preprocessing_list.html b/dev/articles/preprocessing_list.html
index a86c6c09..f23b51da 100644
--- a/dev/articles/preprocessing_list.html
+++ b/dev/articles/preprocessing_list.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/articles/tabular_learner_list.html b/dev/articles/tabular_learner_list.html
index 5ff1b76c..be521854 100644
--- a/dev/articles/tabular_learner_list.html
+++ b/dev/articles/tabular_learner_list.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/articles/task_list.html b/dev/articles/task_list.html
index aacecd35..74c5ed3a 100644
--- a/dev/articles/task_list.html
+++ b/dev/articles/task_list.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/articles/vision_learner_list.html b/dev/articles/vision_learner_list.html
index 8a87b660..e60eed51 100644
--- a/dev/articles/vision_learner_list.html
+++ b/dev/articles/vision_learner_list.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/authors.html b/dev/authors.html
index bcf8d86d..57d1852d 100644
--- a/dev/authors.html
+++ b/dev/authors.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -96,13 +96,13 @@ <h2 id="citation">Citation</h2>
 
       <p>Fischer S, Binder M (2025).
 <em>mlr3torch: Deep Learning with 'mlr3'</em>.
-R package version 0.1.2-9000, https://github.com/mlr-org/mlr3torch/, <a href="https://mlr3torch.mlr-org.com/">https://mlr3torch.mlr-org.com/</a>.
+R package version 0.2.0-9000, https://github.com/mlr-org/mlr3torch/, <a href="https://mlr3torch.mlr-org.com/">https://mlr3torch.mlr-org.com/</a>.
 </p>
       <pre>@Manual{,
   title = {mlr3torch: Deep Learning with 'mlr3'},
   author = {Sebastian Fischer and Martin Binder},
   year = {2025},
-  note = {R package version 0.1.2-9000, https://github.com/mlr-org/mlr3torch/},
+  note = {R package version 0.2.0-9000, https://github.com/mlr-org/mlr3torch/},
   url = {https://mlr3torch.mlr-org.com/},
 }</pre>
     </div>
diff --git a/dev/index.html b/dev/index.html
index db181637..33c57c11 100644
--- a/dev/index.html
+++ b/dev/index.html
@@ -33,7 +33,7 @@
 
     <a class="navbar-brand me-2" href="index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -100,7 +100,7 @@
 <a class="anchor" aria-label="anchor" href="#mlr3torch-"></a>
 </h1>
 </div>
-<p>Package website: <a href="https://mlr3torch.mlr-org.com/">release</a> | <a href="https://mlr3torch.mlr-org.com/dev">dev</a></p>
+<p>Package website: <a href="https://mlr3torch.mlr-org.com/">release</a> | <a href="https://mlr3torch.mlr-org.com/dev/">dev</a></p>
 <p>Deep Learning with torch and mlr3.</p>
 <!-- badges: start -->
 
diff --git a/dev/news/index.html b/dev/news/index.html
index 7b367844..5486760c 100644
--- a/dev/news/index.html
+++ b/dev/news/index.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -62,6 +62,40 @@
       <small>Source: <a href="https://github.com/mlr-org/mlr3torch/blob/main/NEWS.md" class="external-link"><code>NEWS.md</code></a></small>
     </div>
 
+    <div class="section level2">
+<h2 class="pkg-version" data-toc-text="0.2.0" id="mlr3torch-020">mlr3torch 0.2.0<a class="anchor" aria-label="anchor" href="#mlr3torch-020"></a></h2>
+<div class="section level3">
+<h3 id="breaking-changes-0-2-0">Breaking Changes<a class="anchor" aria-label="anchor" href="#breaking-changes-0-2-0"></a></h3>
+<ul><li>Removed some optimizers for which no fast (‘ignite’) variant exists.</li>
+<li>The default optimizer is now AdamW instead of Adam.</li>
+<li>The private <code>LearnerTorch$.dataloader()</code> method now operates no longer on the <code>task</code> but on the <code>dataset</code> generated by the private <code>LearnerTorch$.dataset()</code> method.</li>
+<li>The <code>shuffle</code> parameter during model training is now initialized to <code>TRUE</code> to sidestep issues where data is sorted.</li>
+</ul></div>
+<div class="section level3">
+<h3 id="performance-improvements-0-2-0">Performance Improvements<a class="anchor" aria-label="anchor" href="#performance-improvements-0-2-0"></a></h3>
+<ul><li>Optimizers now use the faster (‘ignite’) version of the optimizers, which leads to considerable speed improvements.</li>
+<li>The <code>jit_trace</code> parameter was added to <code>LearnerTorch</code>, which when set to <code>TRUE</code> can lead to significant speedups. This should only be enabled for ‘static’ models, see the <a href="https://torch.mlverse.org/docs/articles/torchscript" class="external-link">torch tutorial</a> for more information.</li>
+<li>Added parameter <code>num_interop_threads</code> to <code>LearnerTorch</code>.</li>
+<li>The <code>tensor_dataset</code> parameter was added, which allows to stack all batches at the beginning of training to make loading of batches afterwards faster.</li>
+<li>Use a faster default image loader.</li>
+</ul></div>
+<div class="section level3">
+<h3 id="features-0-2-0">Features<a class="anchor" aria-label="anchor" href="#features-0-2-0"></a></h3>
+<ul><li>Added <code>PipeOp</code> for adaptive average pooling.</li>
+<li>The <code>n_layers</code> parameter was added to the MLP learner.</li>
+<li>Added multimodal melanoma and cifar{10, 100} example tasks.</li>
+<li>Added a callback to iteratively unfreeze parameters for finetuning.</li>
+<li>Added different learning rate schedulers as callbacks.</li>
+</ul></div>
+<div class="section level3">
+<h3 id="bug-fixes-0-2-0">Bug Fixes:<a class="anchor" aria-label="anchor" href="#bug-fixes-0-2-0"></a></h3>
+<ul><li>Torch learners can now be used with <code>AutoTuner</code>.</li>
+<li>Early stopping now not uses <code>epochs - patience</code> for the internally tuned values instead of the trained number of <code>epochs</code> as it was before.</li>
+<li>The <code>dataset</code> of a learner must no longer return the tensors on the specified <code>device</code>, which allows for parallel dataloading on GPUs.</li>
+<li>
+<code>PipeOpBlock</code> should no longer create ID clashes with other PipeOps in the graph (<a href="https://github.com/mlr-org/mlr3torch/issues/260" class="external-link">#260</a>).</li>
+</ul></div>
+</div>
     <div class="section level2">
 <h2 class="pkg-version" data-toc-text="0.1.2" id="mlr3torch-012">mlr3torch 0.1.2<a class="anchor" aria-label="anchor" href="#mlr3torch-012"></a></h2><p class="text-muted">CRAN release: 2024-10-15</p>
 <ul><li>Don’t use deprecated <code>data_formats</code> anymore</li>
diff --git a/dev/pkgdown.yml b/dev/pkgdown.yml
index e4251fb8..4d8098ef 100644
--- a/dev/pkgdown.yml
+++ b/dev/pkgdown.yml
@@ -15,7 +15,7 @@ articles:
   articles/tabular_learner_list: tabular_learner_list.html
   articles/task_list: task_list.html
   articles/vision_learner_list: vision_learner_list.html
-last_built: 2025-02-07T06:14Z
+last_built: 2025-02-07T09:18Z
 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 e619d1b3..33efe32a 100644
--- a/dev/reference/DataDescriptor.html
+++ b/dev/reference/DataDescriptor.html
@@ -17,7 +17,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -276,14 +276,14 @@ <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.75a3c3.x.output</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> * pointer: nop.1cf78f.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.75a3c3.x.output</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> * pointer: nop.1cf78f.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/ModelDescriptor.html b/dev/reference/ModelDescriptor.html
index 77c3437b..6df22440 100644
--- a/dev/reference/ModelDescriptor.html
+++ b/dev/reference/ModelDescriptor.html
@@ -37,7 +37,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/PipeOpPreprocTorchTrafoNop.html b/dev/reference/PipeOpPreprocTorchTrafoNop.html
index e8911eb6..743655bb 100644
--- a/dev/reference/PipeOpPreprocTorchTrafoNop.html
+++ b/dev/reference/PipeOpPreprocTorchTrafoNop.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/PipeOpPreprocTorchTrafoReshape.html b/dev/reference/PipeOpPreprocTorchTrafoReshape.html
index b1105f9a..6869d7f4 100644
--- a/dev/reference/PipeOpPreprocTorchTrafoReshape.html
+++ b/dev/reference/PipeOpPreprocTorchTrafoReshape.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/Select.html b/dev/reference/Select.html
index 067c6e7d..78c844d0 100644
--- a/dev/reference/Select.html
+++ b/dev/reference/Select.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/TorchCallback.html b/dev/reference/TorchCallback.html
index ff22d3ba..ef900f30 100644
--- a/dev/reference/TorchCallback.html
+++ b/dev/reference/TorchCallback.html
@@ -17,7 +17,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -230,7 +230,7 @@ <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/RtmptZtedw/file6e8513f33c91, freq=1</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> * Parameters: path=/tmp/RtmpCZgcaa/file20ef2e4d4b3a, 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/TorchDescriptor.html b/dev/reference/TorchDescriptor.html
index a57fdc3c..dfc33552 100644
--- a/dev/reference/TorchDescriptor.html
+++ b/dev/reference/TorchDescriptor.html
@@ -15,7 +15,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/TorchIngressToken.html b/dev/reference/TorchIngressToken.html
index e5a998a8..0e482d83 100644
--- a/dev/reference/TorchIngressToken.html
+++ b/dev/reference/TorchIngressToken.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/TorchLoss.html b/dev/reference/TorchLoss.html
index ff5ade0a..aa5f7282 100644
--- a/dev/reference/TorchLoss.html
+++ b/dev/reference/TorchLoss.html
@@ -17,7 +17,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/TorchOptimizer.html b/dev/reference/TorchOptimizer.html
index 2671dbb5..b8294db6 100644
--- a/dev/reference/TorchOptimizer.html
+++ b/dev/reference/TorchOptimizer.html
@@ -17,7 +17,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/as_data_descriptor.html b/dev/reference/as_data_descriptor.html
index 3870b3ac..e3b9b02e 100644
--- a/dev/reference/as_data_descriptor.html
+++ b/dev/reference/as_data_descriptor.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -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.2e6303.x.output</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> * pointer: nop.7f82c6.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>
@@ -119,7 +119,7 @@ <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.6caef5.x.output</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> * pointer: nop.170f25.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/as_lazy_tensor.html b/dev/reference/as_lazy_tensor.html
index 242fea7d..12d00dbe 100644
--- a/dev/reference/as_lazy_tensor.html
+++ b/dev/reference/as_lazy_tensor.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/as_lr_scheduler.html b/dev/reference/as_lr_scheduler.html
index cd9c9e7f..34a367f8 100644
--- a/dev/reference/as_lr_scheduler.html
+++ b/dev/reference/as_lr_scheduler.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/as_torch_callback.html b/dev/reference/as_torch_callback.html
index b1dc5767..6016565d 100644
--- a/dev/reference/as_torch_callback.html
+++ b/dev/reference/as_torch_callback.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/as_torch_callbacks.html b/dev/reference/as_torch_callbacks.html
index 366de7e6..474945e6 100644
--- a/dev/reference/as_torch_callbacks.html
+++ b/dev/reference/as_torch_callbacks.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/as_torch_loss.html b/dev/reference/as_torch_loss.html
index b6b0cad4..8f11e232 100644
--- a/dev/reference/as_torch_loss.html
+++ b/dev/reference/as_torch_loss.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/as_torch_optimizer.html b/dev/reference/as_torch_optimizer.html
index 8ddc1b07..032f3e63 100644
--- a/dev/reference/as_torch_optimizer.html
+++ b/dev/reference/as_torch_optimizer.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/assert_lazy_tensor.html b/dev/reference/assert_lazy_tensor.html
index 7e2d0447..d379aa3f 100644
--- a/dev/reference/assert_lazy_tensor.html
+++ b/dev/reference/assert_lazy_tensor.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/auto_device.html b/dev/reference/auto_device.html
index b955effd..078d219b 100644
--- a/dev/reference/auto_device.html
+++ b/dev/reference/auto_device.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/batchgetter_categ.html b/dev/reference/batchgetter_categ.html
index 2aee0834..3db6b08d 100644
--- a/dev/reference/batchgetter_categ.html
+++ b/dev/reference/batchgetter_categ.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/batchgetter_num.html b/dev/reference/batchgetter_num.html
index bd356793..f0fae483 100644
--- a/dev/reference/batchgetter_num.html
+++ b/dev/reference/batchgetter_num.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/callback_set.html b/dev/reference/callback_set.html
index 521517be..d38b22d4 100644
--- a/dev/reference/callback_set.html
+++ b/dev/reference/callback_set.html
@@ -15,7 +15,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/equals-.lazy_tensor.html b/dev/reference/equals-.lazy_tensor.html
index 62ee9ea3..8df96b0c 100644
--- a/dev/reference/equals-.lazy_tensor.html
+++ b/dev/reference/equals-.lazy_tensor.html
@@ -17,7 +17,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/index.html b/dev/reference/index.html
index b2a416bd..b54fe9da 100644
--- a/dev/reference/index.html
+++ b/dev/reference/index.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/is_lazy_tensor.html b/dev/reference/is_lazy_tensor.html
index 6ab5fb76..d5452ae6 100644
--- a/dev/reference/is_lazy_tensor.html
+++ b/dev/reference/is_lazy_tensor.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/lazy_tensor.html b/dev/reference/lazy_tensor.html
index 8bafa3a7..0bd697ac 100644
--- a/dev/reference/lazy_tensor.html
+++ b/dev/reference/lazy_tensor.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/materialize.html b/dev/reference/materialize.html
index 2ddad32c..b80341f3 100644
--- a/dev/reference/materialize.html
+++ b/dev/reference/materialize.html
@@ -17,7 +17,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -145,86 +145,87 @@ <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://torch.mlverse.org/docs/reference/torch_randn.html" class="external-link">torch_randn</a></span><span class="op">(</span><span class="fl">10</span>, <span class="fl">3</span><span class="op">)</span><span class="op">)</span></span></span>
 <span class="r-in"><span><span class="fu">materialize</span><span class="op">(</span><span class="va">lt1</span>, rbind <span class="op">=</span> <span class="cn">TRUE</span><span class="op">)</span></span></span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8098 -0.5025  0.9536</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1845 -0.4500 -0.8570</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.8274  0.3922 -1.0590</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1568  0.8758 -0.5632</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6920  0.7932 -2.2409</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4585  0.7990 -1.4406</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3679 -0.8739  0.9497</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.6540  0.5656  0.1746</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1380 -1.4313 -0.0537</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9234  1.6273  0.2737</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.0671e+00  7.5651e-01  1.9241e+00</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2200e+00 -1.2343e-01  8.3400e-01</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.1294e-01 -1.2217e+00  4.8949e-01</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -2.9355e-01  2.5931e+00 -3.8925e-01</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -9.3131e-01  8.9302e-01 -7.6513e-01</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  5.2385e-01 -1.4898e+00 -2.3702e-01</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  2.2191e-01 -1.9253e-01  7.3084e-05</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0812e-02 -9.9583e-01 -7.5198e-01</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.8363e+00 -1.1356e+00  7.3768e-01</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -2.1744e-01 -1.5434e-01  1.0868e+00</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{10,3} ]</span>
 <span class="r-in"><span><span class="fu">materialize</span><span class="op">(</span><span class="va">lt1</span>, rbind <span class="op">=</span> <span class="cn">FALSE</span><span class="op">)</span></span></span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [[1]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8098</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5025</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.9536</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.0671</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7565</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.9241</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>  0.1845</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4500</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8570</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2200</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1234</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.8340</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>  1.8274</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3922</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0590</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1129</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2217</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4895</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>  0.1568</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.8758</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5632</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2936</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  2.5931</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3893</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [[5]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6920</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7932</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -2.2409</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9313</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.8930</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7651</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>  1.4585</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7990</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4406</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5239</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4898</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2370</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [[7]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3679</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8739</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.9497</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2219</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1925</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> [ 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>  0.6540</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5656</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1746</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> 0.01 *</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0812</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -99.5832</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -75.1981</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>  0.1380</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4313</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0537</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.8363</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.1356</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7377</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [[10]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9234</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.6273</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2737</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2174</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1543</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.0868</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://torch.mlverse.org/docs/reference/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>
@@ -232,184 +233,185 @@ <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> -0.8098 -0.5025  0.9536</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1845 -0.4500 -0.8570</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.8274  0.3922 -1.0590</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1568  0.8758 -0.5632</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6920  0.7932 -2.2409</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4585  0.7990 -1.4406</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3679 -0.8739  0.9497</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.6540  0.5656  0.1746</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1380 -1.4313 -0.0537</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9234  1.6273  0.2737</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.0671e+00  7.5651e-01  1.9241e+00</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2200e+00 -1.2343e-01  8.3400e-01</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.1294e-01 -1.2217e+00  4.8949e-01</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -2.9355e-01  2.5931e+00 -3.8925e-01</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -9.3131e-01  8.9302e-01 -7.6513e-01</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  5.2385e-01 -1.4898e+00 -2.3702e-01</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  2.2191e-01 -1.9253e-01  7.3084e-05</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0812e-02 -9.9583e-01 -7.5198e-01</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.8363e+00 -1.1356e+00  7.3768e-01</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -2.1744e-01 -1.5434e-01  1.0868e+00</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>  1.7508  0.5892  1.5608  1.1568</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8770 -1.1304  0.5787  1.2020</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2732 -0.7877  2.2388  0.1827</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1450  0.7184 -0.3360 -0.1029</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7497 -1.0479 -0.8173 -2.3177</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2669  0.1360 -0.7766  0.1551</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.3982  1.1425  1.8144  3.2104</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  2.6315 -2.0763 -0.4969  0.3670</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6687  0.0809 -0.7422 -0.2983</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  2.1002  0.4322  0.7277 -1.5068</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.6287  1.3658  0.5398 -0.1572</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2219 -0.1518 -0.5991  1.4036</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.0065  0.4194 -0.1043 -1.1464</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.1741 -0.1886  1.4039  1.3031</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.0056 -0.7701 -1.5744  0.8223</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7315 -0.4518 -1.0262 -0.7027</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.2091 -0.8113 -0.3490  0.7018</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7377  0.4125  0.1465 -0.3253</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -2.2844 -0.1151 -1.8222  0.1563</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.3287 -0.6418  0.0915 -0.6035</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{10,4} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-in"><span><span class="fu">materialize</span><span class="op">(</span><span class="va">d</span>, rbind <span class="op">=</span> <span class="cn">FALSE</span><span class="op">)</span></span></span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt1</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt1[[1]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8098</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5025</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.9536</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.0671</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7565</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.9241</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>  0.1845</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4500</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8570</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2200</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1234</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.8340</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>  1.8274</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3922</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0590</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1129</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2217</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4895</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>  0.1568</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.8758</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5632</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2936</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  2.5931</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3893</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt1[[5]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6920</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7932</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -2.2409</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9313</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.8930</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7651</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>  1.4585</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7990</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4406</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5239</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4898</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2370</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt1[[7]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3679</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8739</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.9497</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2219</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1925</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> [ 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>  0.6540</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5656</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1746</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> 0.01 *</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0812</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -99.5832</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -75.1981</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>  0.1380</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4313</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0537</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.8363</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.1356</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7377</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt1[[10]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9234</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.6273</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2737</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2174</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1543</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.0868</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>  1.7508</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5892</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.5608</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.1568</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.6287</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.3658</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5398</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1572</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> -0.8770</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.1304</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5787</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.2020</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2219</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1518</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5991</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4036</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> -0.2732</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7877</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  2.2388</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1827</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.0065</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4194</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1043</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.1464</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> -0.1450</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7184</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3360</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1029</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.1741</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1886</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4039</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.3031</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{4} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt2[[5]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7497</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0479</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8173</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -2.3177</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.0056</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7701</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.5744</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.8223</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> -1.2669</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1360</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7766</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1551</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7315</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4518</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0262</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7027</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>  1.3982</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.1425</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.8144</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  3.2104</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.2091</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8113</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3490</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7018</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>  2.6315</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -2.0763</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4969</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3670</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7377</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4125</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1465</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3253</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> -0.6687</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0809</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7422</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2983</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -2.2844</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1151</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.8222</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1563</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>  2.1002</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4322</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7277</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.5068</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.3287</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6418</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0915</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6035</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/materialize_internal.html b/dev/reference/materialize_internal.html
index dca6f3d3..2992c565 100644
--- a/dev/reference/materialize_internal.html
+++ b/dev/reference/materialize_internal.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr3torch-package.html b/dev/reference/mlr3torch-package.html
index c4a26f8c..ce0716a8 100644
--- a/dev/reference/mlr3torch-package.html
+++ b/dev/reference/mlr3torch-package.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr3torch_callbacks.html b/dev/reference/mlr3torch_callbacks.html
index 9cf465a5..38c7b488 100644
--- a/dev/reference/mlr3torch_callbacks.html
+++ b/dev/reference/mlr3torch_callbacks.html
@@ -13,7 +13,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr3torch_losses.html b/dev/reference/mlr3torch_losses.html
index a86749d9..a1258f76 100644
--- a/dev/reference/mlr3torch_losses.html
+++ b/dev/reference/mlr3torch_losses.html
@@ -13,7 +13,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr3torch_optimizers.html b/dev/reference/mlr3torch_optimizers.html
index d67b95ff..f3cee22c 100644
--- a/dev/reference/mlr3torch_optimizers.html
+++ b/dev/reference/mlr3torch_optimizers.html
@@ -13,7 +13,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_backends_lazy.html b/dev/reference/mlr_backends_lazy.html
index 11e58ce5..8efdeb80 100644
--- a/dev/reference/mlr_backends_lazy.html
+++ b/dev/reference/mlr_backends_lazy.html
@@ -49,7 +49,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_callback_set.checkpoint.html b/dev/reference/mlr_callback_set.checkpoint.html
index 175ebd44..fb13f2c0 100644
--- a/dev/reference/mlr_callback_set.checkpoint.html
+++ b/dev/reference/mlr_callback_set.checkpoint.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_callback_set.history.html b/dev/reference/mlr_callback_set.history.html
index 51d05507..36f2601e 100644
--- a/dev/reference/mlr_callback_set.history.html
+++ b/dev/reference/mlr_callback_set.history.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_callback_set.html b/dev/reference/mlr_callback_set.html
index c7965db4..37aded3a 100644
--- a/dev/reference/mlr_callback_set.html
+++ b/dev/reference/mlr_callback_set.html
@@ -17,7 +17,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_callback_set.lr_scheduler.html b/dev/reference/mlr_callback_set.lr_scheduler.html
index 72cce4c1..c77c10ca 100644
--- a/dev/reference/mlr_callback_set.lr_scheduler.html
+++ b/dev/reference/mlr_callback_set.lr_scheduler.html
@@ -27,7 +27,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_callback_set.progress.html b/dev/reference/mlr_callback_set.progress.html
index aad39fbb..711fee62 100644
--- a/dev/reference/mlr_callback_set.progress.html
+++ b/dev/reference/mlr_callback_set.progress.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -184,7 +184,7 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 <span class="r-in"><span><span class="op">)</span></span></span>
 <span class="r-in"><span></span></span>
 <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="op">)</span></span></span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> Epoch 1 started (2025-02-07 06:14:43)</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> Epoch 1 started (2025-02-07 09:19:01)</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [Summary epoch 1]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> ------------------</span>
@@ -193,7 +193,7 @@ <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>  * classif.ce = 0.69</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> Measures (Valid):</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span>  * classif.ce = 0.62</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> Epoch 2 started (2025-02-07 06:14:43)</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> Epoch 2 started (2025-02-07 09:19:02)</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [Summary epoch 2]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> ------------------</span>
@@ -202,7 +202,7 @@ <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>  * classif.ce = 0.69</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> Measures (Valid):</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span>  * classif.ce = 0.56</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> Epoch 3 started (2025-02-07 06:14:44)</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> Epoch 3 started (2025-02-07 09:19:02)</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [Summary epoch 3]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> ------------------</span>
@@ -211,7 +211,7 @@ <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>  * classif.ce = 0.45</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> Measures (Valid):</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span>  * classif.ce = 0.44</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> Epoch 4 started (2025-02-07 06:14:44)</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> Epoch 4 started (2025-02-07 09:19:03)</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [Summary epoch 4]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> ------------------</span>
@@ -220,7 +220,7 @@ <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>  * classif.ce = 0.43</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> Measures (Valid):</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span>  * classif.ce = 0.49</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> Epoch 5 started (2025-02-07 06:14:44)</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> Epoch 5 started (2025-02-07 09:19:03)</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [Summary epoch 5]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> ------------------</span>
@@ -229,7 +229,7 @@ <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>  * classif.ce = 0.49</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> Measures (Valid):</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span>  * classif.ce = 0.60</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> Finished training for 5 epochs (2025-02-07 06:14:45)</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> Finished training for 5 epochs (2025-02-07 09:19:03)</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_callback_set.tb.html b/dev/reference/mlr_callback_set.tb.html
index 14629280..d999f41e 100644
--- a/dev/reference/mlr_callback_set.tb.html
+++ b/dev/reference/mlr_callback_set.tb.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_callback_set.unfreeze.html b/dev/reference/mlr_callback_set.unfreeze.html
index f4417c6c..24841dfe 100644
--- a/dev/reference/mlr_callback_set.unfreeze.html
+++ b/dev/reference/mlr_callback_set.unfreeze.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_context_torch.html b/dev/reference/mlr_context_torch.html
index 162a7143..28ca07a3 100644
--- a/dev/reference/mlr_context_torch.html
+++ b/dev/reference/mlr_context_torch.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_learners.mlp.html b/dev/reference/mlr_learners.mlp.html
index 3b0fdb2c..78c8c293 100644
--- a/dev/reference/mlr_learners.mlp.html
+++ b/dev/reference/mlr_learners.mlp.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_learners.tab_resnet.html b/dev/reference/mlr_learners.tab_resnet.html
index 7e8ab113..8cee8285 100644
--- a/dev/reference/mlr_learners.tab_resnet.html
+++ b/dev/reference/mlr_learners.tab_resnet.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_learners.torch_featureless.html b/dev/reference/mlr_learners.torch_featureless.html
index baf2fa30..bbf4e03a 100644
--- a/dev/reference/mlr_learners.torch_featureless.html
+++ b/dev/reference/mlr_learners.torch_featureless.html
@@ -13,7 +13,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_learners.torchvision.html b/dev/reference/mlr_learners.torchvision.html
index ede883d7..2ba9b4ee 100644
--- a/dev/reference/mlr_learners.torchvision.html
+++ b/dev/reference/mlr_learners.torchvision.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_learners_torch.html b/dev/reference/mlr_learners_torch.html
index ec472b2e..23190f86 100644
--- a/dev/reference/mlr_learners_torch.html
+++ b/dev/reference/mlr_learners_torch.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_learners_torch_image.html b/dev/reference/mlr_learners_torch_image.html
index 9115005f..8ab89b18 100644
--- a/dev/reference/mlr_learners_torch_image.html
+++ b/dev/reference/mlr_learners_torch_image.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_learners_torch_model.html b/dev/reference/mlr_learners_torch_model.html
index fb3e2db8..58964304 100644
--- a/dev/reference/mlr_learners_torch_model.html
+++ b/dev/reference/mlr_learners_torch_model.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_module.html b/dev/reference/mlr_pipeops_module.html
index a76fe2dd..5ad25ef1 100644
--- a/dev/reference/mlr_pipeops_module.html
+++ b/dev/reference/mlr_pipeops_module.html
@@ -19,7 +19,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html b/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html
index 2a94b740..1ad0bf41 100644
--- a/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html
+++ b/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html b/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html
index 27e93c33..4406fcb8 100644
--- a/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html
+++ b/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html b/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html
index 960330ce..f39f5a5a 100644
--- a/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html
+++ b/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_avg_pool1d.html b/dev/reference/mlr_pipeops_nn_avg_pool1d.html
index 901498c2..432f4c77 100644
--- a/dev/reference/mlr_pipeops_nn_avg_pool1d.html
+++ b/dev/reference/mlr_pipeops_nn_avg_pool1d.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_avg_pool2d.html b/dev/reference/mlr_pipeops_nn_avg_pool2d.html
index 3b659c9f..5bf90306 100644
--- a/dev/reference/mlr_pipeops_nn_avg_pool2d.html
+++ b/dev/reference/mlr_pipeops_nn_avg_pool2d.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_avg_pool3d.html b/dev/reference/mlr_pipeops_nn_avg_pool3d.html
index 0543feb4..e4b820ce 100644
--- a/dev/reference/mlr_pipeops_nn_avg_pool3d.html
+++ b/dev/reference/mlr_pipeops_nn_avg_pool3d.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_batch_norm1d.html b/dev/reference/mlr_pipeops_nn_batch_norm1d.html
index 3bf15a6a..17b8c517 100644
--- a/dev/reference/mlr_pipeops_nn_batch_norm1d.html
+++ b/dev/reference/mlr_pipeops_nn_batch_norm1d.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_batch_norm2d.html b/dev/reference/mlr_pipeops_nn_batch_norm2d.html
index adc7f69e..c22b162a 100644
--- a/dev/reference/mlr_pipeops_nn_batch_norm2d.html
+++ b/dev/reference/mlr_pipeops_nn_batch_norm2d.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_batch_norm3d.html b/dev/reference/mlr_pipeops_nn_batch_norm3d.html
index 315eabf8..c11555b4 100644
--- a/dev/reference/mlr_pipeops_nn_batch_norm3d.html
+++ b/dev/reference/mlr_pipeops_nn_batch_norm3d.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_block.html b/dev/reference/mlr_pipeops_nn_block.html
index 8c3f8167..bfff1bcd 100644
--- a/dev/reference/mlr_pipeops_nn_block.html
+++ b/dev/reference/mlr_pipeops_nn_block.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_celu.html b/dev/reference/mlr_pipeops_nn_celu.html
index 72883d34..0f0f24b5 100644
--- a/dev/reference/mlr_pipeops_nn_celu.html
+++ b/dev/reference/mlr_pipeops_nn_celu.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_conv1d.html b/dev/reference/mlr_pipeops_nn_conv1d.html
index 4dad11aa..e92127d4 100644
--- a/dev/reference/mlr_pipeops_nn_conv1d.html
+++ b/dev/reference/mlr_pipeops_nn_conv1d.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_conv2d.html b/dev/reference/mlr_pipeops_nn_conv2d.html
index 4d87eca0..3798c5cd 100644
--- a/dev/reference/mlr_pipeops_nn_conv2d.html
+++ b/dev/reference/mlr_pipeops_nn_conv2d.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_conv3d.html b/dev/reference/mlr_pipeops_nn_conv3d.html
index de904c24..b8da56d7 100644
--- a/dev/reference/mlr_pipeops_nn_conv3d.html
+++ b/dev/reference/mlr_pipeops_nn_conv3d.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_conv_transpose1d.html b/dev/reference/mlr_pipeops_nn_conv_transpose1d.html
index ab251c61..34104bdc 100644
--- a/dev/reference/mlr_pipeops_nn_conv_transpose1d.html
+++ b/dev/reference/mlr_pipeops_nn_conv_transpose1d.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_conv_transpose2d.html b/dev/reference/mlr_pipeops_nn_conv_transpose2d.html
index 28ab2229..a3d5af77 100644
--- a/dev/reference/mlr_pipeops_nn_conv_transpose2d.html
+++ b/dev/reference/mlr_pipeops_nn_conv_transpose2d.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_conv_transpose3d.html b/dev/reference/mlr_pipeops_nn_conv_transpose3d.html
index 5b5cec70..003b502f 100644
--- a/dev/reference/mlr_pipeops_nn_conv_transpose3d.html
+++ b/dev/reference/mlr_pipeops_nn_conv_transpose3d.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_dropout.html b/dev/reference/mlr_pipeops_nn_dropout.html
index 92cdd1f2..67db997b 100644
--- a/dev/reference/mlr_pipeops_nn_dropout.html
+++ b/dev/reference/mlr_pipeops_nn_dropout.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_elu.html b/dev/reference/mlr_pipeops_nn_elu.html
index bfd61b68..0e5f4349 100644
--- a/dev/reference/mlr_pipeops_nn_elu.html
+++ b/dev/reference/mlr_pipeops_nn_elu.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_flatten.html b/dev/reference/mlr_pipeops_nn_flatten.html
index 79790b20..17033173 100644
--- a/dev/reference/mlr_pipeops_nn_flatten.html
+++ b/dev/reference/mlr_pipeops_nn_flatten.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_gelu.html b/dev/reference/mlr_pipeops_nn_gelu.html
index 684e53f9..346daad1 100644
--- a/dev/reference/mlr_pipeops_nn_gelu.html
+++ b/dev/reference/mlr_pipeops_nn_gelu.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_glu.html b/dev/reference/mlr_pipeops_nn_glu.html
index 702980ed..44879fd6 100644
--- a/dev/reference/mlr_pipeops_nn_glu.html
+++ b/dev/reference/mlr_pipeops_nn_glu.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_hardshrink.html b/dev/reference/mlr_pipeops_nn_hardshrink.html
index 53208a97..684ee4fb 100644
--- a/dev/reference/mlr_pipeops_nn_hardshrink.html
+++ b/dev/reference/mlr_pipeops_nn_hardshrink.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_hardsigmoid.html b/dev/reference/mlr_pipeops_nn_hardsigmoid.html
index 666677b6..14f75dfe 100644
--- a/dev/reference/mlr_pipeops_nn_hardsigmoid.html
+++ b/dev/reference/mlr_pipeops_nn_hardsigmoid.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_hardtanh.html b/dev/reference/mlr_pipeops_nn_hardtanh.html
index 8062132d..97d4ca04 100644
--- a/dev/reference/mlr_pipeops_nn_hardtanh.html
+++ b/dev/reference/mlr_pipeops_nn_hardtanh.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_head.html b/dev/reference/mlr_pipeops_nn_head.html
index 7e5cd1e6..5f649920 100644
--- a/dev/reference/mlr_pipeops_nn_head.html
+++ b/dev/reference/mlr_pipeops_nn_head.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_layer_norm.html b/dev/reference/mlr_pipeops_nn_layer_norm.html
index ae754771..f737ab09 100644
--- a/dev/reference/mlr_pipeops_nn_layer_norm.html
+++ b/dev/reference/mlr_pipeops_nn_layer_norm.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_leaky_relu.html b/dev/reference/mlr_pipeops_nn_leaky_relu.html
index c8947141..76e659ea 100644
--- a/dev/reference/mlr_pipeops_nn_leaky_relu.html
+++ b/dev/reference/mlr_pipeops_nn_leaky_relu.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_linear.html b/dev/reference/mlr_pipeops_nn_linear.html
index 2fa973b7..1d940e44 100644
--- a/dev/reference/mlr_pipeops_nn_linear.html
+++ b/dev/reference/mlr_pipeops_nn_linear.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_log_sigmoid.html b/dev/reference/mlr_pipeops_nn_log_sigmoid.html
index 9e4ec073..30381827 100644
--- a/dev/reference/mlr_pipeops_nn_log_sigmoid.html
+++ b/dev/reference/mlr_pipeops_nn_log_sigmoid.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_max_pool1d.html b/dev/reference/mlr_pipeops_nn_max_pool1d.html
index 4ed3b683..7f725610 100644
--- a/dev/reference/mlr_pipeops_nn_max_pool1d.html
+++ b/dev/reference/mlr_pipeops_nn_max_pool1d.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_max_pool2d.html b/dev/reference/mlr_pipeops_nn_max_pool2d.html
index f88202e4..b60a5bf4 100644
--- a/dev/reference/mlr_pipeops_nn_max_pool2d.html
+++ b/dev/reference/mlr_pipeops_nn_max_pool2d.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_max_pool3d.html b/dev/reference/mlr_pipeops_nn_max_pool3d.html
index e8c3fb45..17fe9e48 100644
--- a/dev/reference/mlr_pipeops_nn_max_pool3d.html
+++ b/dev/reference/mlr_pipeops_nn_max_pool3d.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_merge.html b/dev/reference/mlr_pipeops_nn_merge.html
index 642f7a39..bbfb64ef 100644
--- a/dev/reference/mlr_pipeops_nn_merge.html
+++ b/dev/reference/mlr_pipeops_nn_merge.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_merge_cat.html b/dev/reference/mlr_pipeops_nn_merge_cat.html
index 90307489..764b3d58 100644
--- a/dev/reference/mlr_pipeops_nn_merge_cat.html
+++ b/dev/reference/mlr_pipeops_nn_merge_cat.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_merge_prod.html b/dev/reference/mlr_pipeops_nn_merge_prod.html
index c8738314..3c964546 100644
--- a/dev/reference/mlr_pipeops_nn_merge_prod.html
+++ b/dev/reference/mlr_pipeops_nn_merge_prod.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_merge_sum.html b/dev/reference/mlr_pipeops_nn_merge_sum.html
index 6a56fae7..b6269e56 100644
--- a/dev/reference/mlr_pipeops_nn_merge_sum.html
+++ b/dev/reference/mlr_pipeops_nn_merge_sum.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_prelu.html b/dev/reference/mlr_pipeops_nn_prelu.html
index 312592f0..101ae439 100644
--- a/dev/reference/mlr_pipeops_nn_prelu.html
+++ b/dev/reference/mlr_pipeops_nn_prelu.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_relu.html b/dev/reference/mlr_pipeops_nn_relu.html
index d33b1e4b..fd747918 100644
--- a/dev/reference/mlr_pipeops_nn_relu.html
+++ b/dev/reference/mlr_pipeops_nn_relu.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_relu6.html b/dev/reference/mlr_pipeops_nn_relu6.html
index 9558142f..c2aeec1a 100644
--- a/dev/reference/mlr_pipeops_nn_relu6.html
+++ b/dev/reference/mlr_pipeops_nn_relu6.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_reshape.html b/dev/reference/mlr_pipeops_nn_reshape.html
index e7b33e47..b7e614d3 100644
--- a/dev/reference/mlr_pipeops_nn_reshape.html
+++ b/dev/reference/mlr_pipeops_nn_reshape.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_rrelu.html b/dev/reference/mlr_pipeops_nn_rrelu.html
index 50ff315b..38fa38e5 100644
--- a/dev/reference/mlr_pipeops_nn_rrelu.html
+++ b/dev/reference/mlr_pipeops_nn_rrelu.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_selu.html b/dev/reference/mlr_pipeops_nn_selu.html
index fee36d3c..dfe8084f 100644
--- a/dev/reference/mlr_pipeops_nn_selu.html
+++ b/dev/reference/mlr_pipeops_nn_selu.html
@@ -13,7 +13,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_sigmoid.html b/dev/reference/mlr_pipeops_nn_sigmoid.html
index 55d3c77f..7386fb97 100644
--- a/dev/reference/mlr_pipeops_nn_sigmoid.html
+++ b/dev/reference/mlr_pipeops_nn_sigmoid.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_softmax.html b/dev/reference/mlr_pipeops_nn_softmax.html
index 3d6ef9d8..c8a0fc74 100644
--- a/dev/reference/mlr_pipeops_nn_softmax.html
+++ b/dev/reference/mlr_pipeops_nn_softmax.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_softplus.html b/dev/reference/mlr_pipeops_nn_softplus.html
index 439d8de0..7a87d2e7 100644
--- a/dev/reference/mlr_pipeops_nn_softplus.html
+++ b/dev/reference/mlr_pipeops_nn_softplus.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_softshrink.html b/dev/reference/mlr_pipeops_nn_softshrink.html
index 414fe39d..06bd91e3 100644
--- a/dev/reference/mlr_pipeops_nn_softshrink.html
+++ b/dev/reference/mlr_pipeops_nn_softshrink.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_softsign.html b/dev/reference/mlr_pipeops_nn_softsign.html
index 11c7ea2f..6448e8d7 100644
--- a/dev/reference/mlr_pipeops_nn_softsign.html
+++ b/dev/reference/mlr_pipeops_nn_softsign.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_squeeze.html b/dev/reference/mlr_pipeops_nn_squeeze.html
index 91b8a6ea..0b9a7645 100644
--- a/dev/reference/mlr_pipeops_nn_squeeze.html
+++ b/dev/reference/mlr_pipeops_nn_squeeze.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_tanh.html b/dev/reference/mlr_pipeops_nn_tanh.html
index aa0c9d53..e635720e 100644
--- a/dev/reference/mlr_pipeops_nn_tanh.html
+++ b/dev/reference/mlr_pipeops_nn_tanh.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_tanhshrink.html b/dev/reference/mlr_pipeops_nn_tanhshrink.html
index 9e914a5e..db117dd8 100644
--- a/dev/reference/mlr_pipeops_nn_tanhshrink.html
+++ b/dev/reference/mlr_pipeops_nn_tanhshrink.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_threshold.html b/dev/reference/mlr_pipeops_nn_threshold.html
index f5861033..a0a25081 100644
--- a/dev/reference/mlr_pipeops_nn_threshold.html
+++ b/dev/reference/mlr_pipeops_nn_threshold.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_nn_unsqueeze.html b/dev/reference/mlr_pipeops_nn_unsqueeze.html
index 3e85a695..46123e6b 100644
--- a/dev/reference/mlr_pipeops_nn_unsqueeze.html
+++ b/dev/reference/mlr_pipeops_nn_unsqueeze.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_preproc_torch.html b/dev/reference/mlr_pipeops_preproc_torch.html
index c1538830..6aa67666 100644
--- a/dev/reference/mlr_pipeops_preproc_torch.html
+++ b/dev/reference/mlr_pipeops_preproc_torch.html
@@ -17,7 +17,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_torch.html b/dev/reference/mlr_pipeops_torch.html
index 32181294..a93e6864 100644
--- a/dev/reference/mlr_pipeops_torch.html
+++ b/dev/reference/mlr_pipeops_torch.html
@@ -35,7 +35,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_torch_callbacks.html b/dev/reference/mlr_pipeops_torch_callbacks.html
index fa5a66c4..e449ee84 100644
--- a/dev/reference/mlr_pipeops_torch_callbacks.html
+++ b/dev/reference/mlr_pipeops_torch_callbacks.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_torch_ingress.html b/dev/reference/mlr_pipeops_torch_ingress.html
index 1feb75bf..18a75bc1 100644
--- a/dev/reference/mlr_pipeops_torch_ingress.html
+++ b/dev/reference/mlr_pipeops_torch_ingress.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_torch_ingress_categ.html b/dev/reference/mlr_pipeops_torch_ingress_categ.html
index 87df2a71..2b6bfbce 100644
--- a/dev/reference/mlr_pipeops_torch_ingress_categ.html
+++ b/dev/reference/mlr_pipeops_torch_ingress_categ.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html b/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html
index a59112d9..2e4545f8 100644
--- a/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html
+++ b/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -287,35 +287,35 @@ <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>   0.9970 -0.9383 -0.0510  1.4438  0.5061 -0.8022</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.4175  0.2638 -0.7721 -1.3055  1.0725  0.1492</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.1705  1.1516 -1.1165  0.6934 -0.2440 -0.3882</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.4484 -0.4592 -0.2580 -0.4509  0.1134 -1.1358</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.3275 -1.4557  0.5706  1.2162  0.6805 -1.0252</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.4683  0.2398 -0.4668 -0.1763  0.1691 -0.2560</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.0584  1.1540 -0.6838 -0.1345  0.7166 -0.1001</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.1030 -1.1843  0.4625 -1.2743  0.3107  0.6001</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.4018  0.1173  0.2082  0.5757  0.4001 -0.1134</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.2773 -0.1097 -0.0896 -0.3457  0.4073  0.1018</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -1.1756  0.0686  0.6384 -1.0616 -0.8770 -0.7203</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   1.2721 -0.1520 -0.8474 -0.3209 -0.4132 -0.5879</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>   0.1049  0.7440  0.4347  0.1592 -0.2501  0.0542</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.5176  0.1912 -0.5836  0.2313  0.0562 -0.6430</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.1808 -0.3858 -0.4339  0.1317 -0.2483 -0.5749</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.2762  0.4425  0.1593  0.5774  0.0714 -0.5108</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.4984 -1.4006  1.0352  0.2719  0.3794  0.1281</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.1655  0.0319 -0.6908 -0.1547  0.5281 -0.6154</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.2584  0.3291  0.8919 -0.0781 -0.4938 -0.3368</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.4498  0.5942 -0.1619 -1.5562  0.2340  0.4918</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.6124  0.7222 -0.5876 -0.8403 -0.0998 -0.7715</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.0641 -0.1189 -0.2512  0.2063  0.2547 -0.1059</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.3604 -0.1723 -0.0245 -0.1846 -0.2950  0.3261</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   1.2149  0.8136 -0.1479 -0.5586 -0.1958  0.2617</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> (1,2,.,.) = </span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   1.0034 -0.0084 -0.5612  0.6336  0.6045 -0.3766</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.2037 -0.6614  0.0176  0.9215  0.1562  0.0783</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.6090  0.0246  0.6608  1.2319 -0.4807 -0.7408</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.0390  0.4402 -1.1336 -0.3919 -0.2745  0.7000</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   1.0377  1.2182  0.1394  0.0664 -0.6851  0.3546</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.0136 -0.7515  1.5054 -0.3123  0.3507 -0.9329</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.1901  0.2382  0.7609  0.7862  1.0801  0.0396</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.8384 -0.1858  0.2758 -0.3891 -0.2032 -1.0331</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.2101 -1.2954 -0.5540 -0.9116  1.1030 -0.3491</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.2743  0.2570 -0.3152  1.1351 -0.6616  0.6196</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.0025  1.0044  0.2625  1.3346  1.5293 -1.8108</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.2873  0.7334 -0.2273 -0.4304 -0.0558 -0.2041</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>   0.3226 -0.5212  0.5643 -0.6904 -0.8490 -1.1094</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.6671 -0.4693  0.3308  0.2793  0.1941 -0.9796</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.0677 -0.4034 -0.8510 -0.2002  0.4505 -1.5667</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.0590 -0.8131 -0.9660 -0.1422  0.3739  0.8411</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.1460 -0.3893  0.4009 -0.3064 -0.5538 -0.2872</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -1.0807  1.0639 -0.4367 -0.3011  0.1753 -0.9030</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.7590  0.1862  0.3582  0.6190  0.1483 -0.4890</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.1560 -0.7729  0.8068  0.1980  0.1058  0.7656</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.6955  1.2231  0.6754  0.8966 -0.4761 -0.2140</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.6465  0.6770 -0.0757 -1.0318  0.2228  0.0925</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_ingress_num.html b/dev/reference/mlr_pipeops_torch_ingress_num.html
index 17cbdc68..a31b331c 100644
--- a/dev/reference/mlr_pipeops_torch_ingress_num.html
+++ b/dev/reference/mlr_pipeops_torch_ingress_num.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_torch_loss.html b/dev/reference/mlr_pipeops_torch_loss.html
index ff8f14da..649a4d6d 100644
--- a/dev/reference/mlr_pipeops_torch_loss.html
+++ b/dev/reference/mlr_pipeops_torch_loss.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_torch_model.html b/dev/reference/mlr_pipeops_torch_model.html
index d0dafd67..7ed36ad1 100644
--- a/dev/reference/mlr_pipeops_torch_model.html
+++ b/dev/reference/mlr_pipeops_torch_model.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeops_torch_model_classif.html b/dev/reference/mlr_pipeops_torch_model_classif.html
index 822612d2..e1af3de1 100644
--- a/dev/reference/mlr_pipeops_torch_model_classif.html
+++ b/dev/reference/mlr_pipeops_torch_model_classif.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -284,17 +284,17 @@ <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> $optimizer$state$`1`</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> -0.0305  0.0118 -0.3143 -0.2423</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5306  0.1713  0.9039  0.5068</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5001 -0.1830 -0.5896 -0.2646</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1260 -0.0209 -0.4384 -0.3011</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1005 -0.0250 -0.1186 -0.0357</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2265  0.0458  0.5570  0.3368</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>  0.0048  0.0007  0.4051  0.2395</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   1.1423  0.1193  3.3175  1.0430</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   1.0133  0.1359  1.4087  0.2836</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0645  0.0018  0.7808  0.3683</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.0702  0.0056  0.1150  0.0182</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.2329  0.0109  1.2956  0.4669</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$`1`$max_exp_avg_sq</span>
@@ -310,17 +310,17 @@ <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> $optimizer$state$`2`</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> -0.0674</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1564</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0889</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0876</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0196</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1072</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> 0.0001 *</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.8577</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  9.9196</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  3.2051</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  3.1150</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3254</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  4.7708</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$`2`$max_exp_avg_sq</span>
@@ -361,7 +361,7 @@ <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] 0.057</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> [1] 0.051</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/mlr_pipeops_torch_model_regr.html b/dev/reference/mlr_pipeops_torch_model_regr.html
index 30566ed9..07598d91 100644
--- a/dev/reference/mlr_pipeops_torch_model_regr.html
+++ b/dev/reference/mlr_pipeops_torch_model_regr.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -284,14 +284,14 @@ <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> $optimizer$state$`1`</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> -2.4873 -12.0392 -29.9780 -1048.1003 -19.9276 -19.6820 -628.5255 -97.9720 -3.1496 -15.6931</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -2.9507 -4.8365 -12.6544 -136.2684 -14.7779 -14.6720 -174.2433 -67.4318 -3.6596 -6.2324</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 3.6781e-01  8.1886e+00  5.0809e+01  6.1723e+04  2.2420e+01  2.2421e+01</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> Columns 1 to 7    0.5200     2.3228    12.1758  6475.4131    13.2938    13.9329  4650.6953</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 2.2605e+04  5.4466e+02  5.5291e-01  1.3928e+01</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> Columns 8 to 10  285.8094     0.7534     3.3957</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$`1`$max_exp_avg_sq</span>
@@ -307,12 +307,12 @@ <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> $optimizer$state$`2`</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> -5.3484</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -3.5004</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>  1.6237</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7837</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$`2`$max_exp_avg_sq</span>
@@ -358,7 +358,7 @@ <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] 0.04</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> [1] 0.038</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/mlr_pipeops_torch_optimizer.html b/dev/reference/mlr_pipeops_torch_optimizer.html
index 03fa2e8e..6bf6b3cf 100644
--- a/dev/reference/mlr_pipeops_torch_optimizer.html
+++ b/dev/reference/mlr_pipeops_torch_optimizer.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopsaugment_center_crop.html b/dev/reference/mlr_pipeopsaugment_center_crop.html
index 12c22fab..99bfcef8 100644
--- a/dev/reference/mlr_pipeopsaugment_center_crop.html
+++ b/dev/reference/mlr_pipeopsaugment_center_crop.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopsaugment_color_jitter.html b/dev/reference/mlr_pipeopsaugment_color_jitter.html
index 2006bdd0..3c8e2331 100644
--- a/dev/reference/mlr_pipeopsaugment_color_jitter.html
+++ b/dev/reference/mlr_pipeopsaugment_color_jitter.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopsaugment_crop.html b/dev/reference/mlr_pipeopsaugment_crop.html
index 6e2e80df..02aad8f8 100644
--- a/dev/reference/mlr_pipeopsaugment_crop.html
+++ b/dev/reference/mlr_pipeopsaugment_crop.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopsaugment_hflip.html b/dev/reference/mlr_pipeopsaugment_hflip.html
index 6e6c2be5..4d134df5 100644
--- a/dev/reference/mlr_pipeopsaugment_hflip.html
+++ b/dev/reference/mlr_pipeopsaugment_hflip.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopsaugment_random_affine.html b/dev/reference/mlr_pipeopsaugment_random_affine.html
index d1f916d7..8970b37d 100644
--- a/dev/reference/mlr_pipeopsaugment_random_affine.html
+++ b/dev/reference/mlr_pipeopsaugment_random_affine.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopsaugment_random_choice.html b/dev/reference/mlr_pipeopsaugment_random_choice.html
index ba22359a..7c2125d8 100644
--- a/dev/reference/mlr_pipeopsaugment_random_choice.html
+++ b/dev/reference/mlr_pipeopsaugment_random_choice.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopsaugment_random_crop.html b/dev/reference/mlr_pipeopsaugment_random_crop.html
index 75b20fe3..b1cdd7d8 100644
--- a/dev/reference/mlr_pipeopsaugment_random_crop.html
+++ b/dev/reference/mlr_pipeopsaugment_random_crop.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopsaugment_random_horizontal_flip.html b/dev/reference/mlr_pipeopsaugment_random_horizontal_flip.html
index 99242bdc..e2d28584 100644
--- a/dev/reference/mlr_pipeopsaugment_random_horizontal_flip.html
+++ b/dev/reference/mlr_pipeopsaugment_random_horizontal_flip.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopsaugment_random_order.html b/dev/reference/mlr_pipeopsaugment_random_order.html
index acca8c22..d23349d1 100644
--- a/dev/reference/mlr_pipeopsaugment_random_order.html
+++ b/dev/reference/mlr_pipeopsaugment_random_order.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopsaugment_random_resized_crop.html b/dev/reference/mlr_pipeopsaugment_random_resized_crop.html
index a1f9854d..44b2493c 100644
--- a/dev/reference/mlr_pipeopsaugment_random_resized_crop.html
+++ b/dev/reference/mlr_pipeopsaugment_random_resized_crop.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopsaugment_random_vertical_flip.html b/dev/reference/mlr_pipeopsaugment_random_vertical_flip.html
index f47c9a5b..f44823b9 100644
--- a/dev/reference/mlr_pipeopsaugment_random_vertical_flip.html
+++ b/dev/reference/mlr_pipeopsaugment_random_vertical_flip.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopsaugment_resized_crop.html b/dev/reference/mlr_pipeopsaugment_resized_crop.html
index cef55c5c..865e2a74 100644
--- a/dev/reference/mlr_pipeopsaugment_resized_crop.html
+++ b/dev/reference/mlr_pipeopsaugment_resized_crop.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopsaugment_rotate.html b/dev/reference/mlr_pipeopsaugment_rotate.html
index a007848c..4625261e 100644
--- a/dev/reference/mlr_pipeopsaugment_rotate.html
+++ b/dev/reference/mlr_pipeopsaugment_rotate.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopsaugment_vflip.html b/dev/reference/mlr_pipeopsaugment_vflip.html
index 8654e66c..723ca7d9 100644
--- a/dev/reference/mlr_pipeopsaugment_vflip.html
+++ b/dev/reference/mlr_pipeopsaugment_vflip.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopstrafo_adjust_brightness.html b/dev/reference/mlr_pipeopstrafo_adjust_brightness.html
index 278233b6..7d66a6f0 100644
--- a/dev/reference/mlr_pipeopstrafo_adjust_brightness.html
+++ b/dev/reference/mlr_pipeopstrafo_adjust_brightness.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopstrafo_adjust_gamma.html b/dev/reference/mlr_pipeopstrafo_adjust_gamma.html
index 22686ecd..db17d2ad 100644
--- a/dev/reference/mlr_pipeopstrafo_adjust_gamma.html
+++ b/dev/reference/mlr_pipeopstrafo_adjust_gamma.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopstrafo_adjust_hue.html b/dev/reference/mlr_pipeopstrafo_adjust_hue.html
index 2e9811d8..9e16c738 100644
--- a/dev/reference/mlr_pipeopstrafo_adjust_hue.html
+++ b/dev/reference/mlr_pipeopstrafo_adjust_hue.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopstrafo_adjust_saturation.html b/dev/reference/mlr_pipeopstrafo_adjust_saturation.html
index dc50be9f..55c83439 100644
--- a/dev/reference/mlr_pipeopstrafo_adjust_saturation.html
+++ b/dev/reference/mlr_pipeopstrafo_adjust_saturation.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopstrafo_grayscale.html b/dev/reference/mlr_pipeopstrafo_grayscale.html
index 89a2c873..c362ef07 100644
--- a/dev/reference/mlr_pipeopstrafo_grayscale.html
+++ b/dev/reference/mlr_pipeopstrafo_grayscale.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopstrafo_normalize.html b/dev/reference/mlr_pipeopstrafo_normalize.html
index 4762ee20..e14d5efd 100644
--- a/dev/reference/mlr_pipeopstrafo_normalize.html
+++ b/dev/reference/mlr_pipeopstrafo_normalize.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopstrafo_pad.html b/dev/reference/mlr_pipeopstrafo_pad.html
index be2241cc..7b5e15a3 100644
--- a/dev/reference/mlr_pipeopstrafo_pad.html
+++ b/dev/reference/mlr_pipeopstrafo_pad.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopstrafo_resize.html b/dev/reference/mlr_pipeopstrafo_resize.html
index a0af46fb..aa755bff 100644
--- a/dev/reference/mlr_pipeopstrafo_resize.html
+++ b/dev/reference/mlr_pipeopstrafo_resize.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_pipeopstrafo_rgb_to_grayscale.html b/dev/reference/mlr_pipeopstrafo_rgb_to_grayscale.html
index 2f5e2f32..b52f3ce6 100644
--- a/dev/reference/mlr_pipeopstrafo_rgb_to_grayscale.html
+++ b/dev/reference/mlr_pipeopstrafo_rgb_to_grayscale.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_tasks_cifar.html b/dev/reference/mlr_tasks_cifar.html
index 42b71fa5..908acde5 100644
--- a/dev/reference/mlr_tasks_cifar.html
+++ b/dev/reference/mlr_tasks_cifar.html
@@ -21,7 +21,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_tasks_lazy_iris.html b/dev/reference/mlr_tasks_lazy_iris.html
index 1b5ccb0d..da2c6977 100644
--- a/dev/reference/mlr_tasks_lazy_iris.html
+++ b/dev/reference/mlr_tasks_lazy_iris.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_tasks_melanoma.html b/dev/reference/mlr_tasks_melanoma.html
index 2f2e28c3..682dd8ec 100644
--- a/dev/reference/mlr_tasks_melanoma.html
+++ b/dev/reference/mlr_tasks_melanoma.html
@@ -41,7 +41,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_tasks_mnist.html b/dev/reference/mlr_tasks_mnist.html
index a2c6b884..4a9be340 100644
--- a/dev/reference/mlr_tasks_mnist.html
+++ b/dev/reference/mlr_tasks_mnist.html
@@ -13,7 +13,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/mlr_tasks_tiny_imagenet.html b/dev/reference/mlr_tasks_tiny_imagenet.html
index e1a24893..cd858637 100644
--- a/dev/reference/mlr_tasks_tiny_imagenet.html
+++ b/dev/reference/mlr_tasks_tiny_imagenet.html
@@ -17,7 +17,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/model_descriptor_to_learner.html b/dev/reference/model_descriptor_to_learner.html
index 9d63e811..b8f7e7e8 100644
--- a/dev/reference/model_descriptor_to_learner.html
+++ b/dev/reference/model_descriptor_to_learner.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/model_descriptor_to_module.html b/dev/reference/model_descriptor_to_module.html
index 76200f04..2d8c5867 100644
--- a/dev/reference/model_descriptor_to_module.html
+++ b/dev/reference/model_descriptor_to_module.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/model_descriptor_union.html b/dev/reference/model_descriptor_union.html
index 6bc65862..026c1113 100644
--- a/dev/reference/model_descriptor_union.html
+++ b/dev/reference/model_descriptor_union.html
@@ -27,7 +27,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/nn.html b/dev/reference/nn.html
index ad47173a..d0528994 100644
--- a/dev/reference/nn.html
+++ b/dev/reference/nn.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/nn_graph.html b/dev/reference/nn_graph.html
index 251e7763..eba86564 100644
--- a/dev/reference/nn_graph.html
+++ b/dev/reference/nn_graph.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -132,22 +132,22 @@ <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> -0.3465</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3795</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0351</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3785</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2675</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2843</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6249</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4385</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1047</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2959</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1331</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0724</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0725</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0181</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2933</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2035</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2769</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2779</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1570</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2491</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3878</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1224</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4271</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2581</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3228</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0445</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3487</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6967</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0038</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5004</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4031</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4273</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/reference/nn_merge_cat.html b/dev/reference/nn_merge_cat.html
index 0026b982..d4b684a0 100644
--- a/dev/reference/nn_merge_cat.html
+++ b/dev/reference/nn_merge_cat.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/nn_merge_prod.html b/dev/reference/nn_merge_prod.html
index 31c08f58..e9ecd853 100644
--- a/dev/reference/nn_merge_prod.html
+++ b/dev/reference/nn_merge_prod.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/nn_merge_sum.html b/dev/reference/nn_merge_sum.html
index c7e7f441..3b75ecfc 100644
--- a/dev/reference/nn_merge_sum.html
+++ b/dev/reference/nn_merge_sum.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/nn_reshape.html b/dev/reference/nn_reshape.html
index 6f10ca81..e7e88467 100644
--- a/dev/reference/nn_reshape.html
+++ b/dev/reference/nn_reshape.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/nn_squeeze.html b/dev/reference/nn_squeeze.html
index d49a5a7d..0c8991ab 100644
--- a/dev/reference/nn_squeeze.html
+++ b/dev/reference/nn_squeeze.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/nn_unsqueeze.html b/dev/reference/nn_unsqueeze.html
index 59d62f80..480c2b6c 100644
--- a/dev/reference/nn_unsqueeze.html
+++ b/dev/reference/nn_unsqueeze.html
@@ -7,7 +7,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/pipeop_preproc_torch.html b/dev/reference/pipeop_preproc_torch.html
index 68a02d0b..80e0bbf7 100644
--- a/dev/reference/pipeop_preproc_torch.html
+++ b/dev/reference/pipeop_preproc_torch.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/replace_head.html b/dev/reference/replace_head.html
index 37c331e3..e21b1de4 100644
--- a/dev/reference/replace_head.html
+++ b/dev/reference/replace_head.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/t_clbk.html b/dev/reference/t_clbk.html
index 0050eb0b..9a135277 100644
--- a/dev/reference/t_clbk.html
+++ b/dev/reference/t_clbk.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/t_loss.html b/dev/reference/t_loss.html
index a9b69dfe..9e821073 100644
--- a/dev/reference/t_loss.html
+++ b/dev/reference/t_loss.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/t_opt.html b/dev/reference/t_opt.html
index 09757bb8..30a81740 100644
--- a/dev/reference/t_opt.html
+++ b/dev/reference/t_opt.html
@@ -9,7 +9,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/task_dataset.html b/dev/reference/task_dataset.html
index a540502f..57fad6ac 100644
--- a/dev/reference/task_dataset.html
+++ b/dev/reference/task_dataset.html
@@ -21,7 +21,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/reference/torch_callback.html b/dev/reference/torch_callback.html
index c1b46058..85386973 100644
--- a/dev/reference/torch_callback.html
+++ b/dev/reference/torch_callback.html
@@ -11,7 +11,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">mlr3torch</a>
 
-    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.1.2-9000</small>
+    <small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">0.2.0-9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
diff --git a/dev/search.json b/dev/search.json
index d8cfc4e8..101a07d8 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: 0x562772bf68b8> ##   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: 1.18 ## Epoch 2: 1.06 ## Epoch 3: 1.12 ## Epoch 4: 1.15 ## Epoch 5: 1.16 mlp$model$callbacks$custom_logger ## [1] 1.15931"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/get_started.html","id":"quickstart","dir":"Articles","previous_headings":"","what":"Quickstart","title":"Get Started","text":"vignette show get started mlr3torch training simple neural network tabular regression problem. assume familiar mlr3 framework, see e.g. mlr3 book. first example, train simple multi-layer perceptron (MLP) well-known “mtcars” task, goal predict miles per galleon (‘mpg’) cars. architecture comes predfined learner mlr3torch, can also easily create new network architectures, see Neural Networks Graphs vignette detailed introduoduion. first set seed reproducibility, load library construct task. Learners mlr3torch work similary mlr3 learners. , construct simple multi layer perceptron regression. usual calling lrn() configuring parameters: use two hidden layers 50 neurons, training, set batch size 32, number training epochs 30 device \"cpu\". complete description available parameters see ?mlr3torch::LearnerTorchMLP. can use learner training prediction just like regression learner. , split observations training test set, train learner training set create predictions test set. Finally, compute mean squared error predictions.","code":"set.seed(314) library(mlr3torch) task = tsk(\"mtcars\") task$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, jit_trace=FALSE, eval_freq=1, #>   measures_train=<list>, measures_valid=<list>, patience=0, #>   min_delta=0, batch_size=32, shuffle=TRUE, tensor_dataset=FALSE, #>   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  #> 294.1396"},{"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. 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] 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 AdamW (optim_ignite_adamw) SGD (optim_ignite_sgd). TorchLoss objects can constructed using t_opt(<key>). optimizers, associated ParamSet interesting see : Parameters TorchOptimizer (also TorchLoss TorchCallback) can set usual mlr3 way, .e. either construction, afterwards using $set_values() method parameter set. see optimizer’s parameters added learner’s ParamSet (prefixed \"opt.\") values set values specified. exposing optimizer’s parameters, can conveniently tuned using mlr3tuning. available optimizers stored mlr3torch_optimizers dictionary.","code":"sgd = t_opt(\"sgd\") sgd #> <TorchOptimizer:sgd> Stochastic Gradient Descent #> * Generator: optim_ignite_sgd #> * Parameters: list() #> * Packages: torch,mlr3torch  sgd$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 5 stored values #> Keys: adagrad, adam, adamw, rmsprop, 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 11 stored values #> Keys: checkpoint, history, lr_cosine_annealing, lr_lambda, #>   lr_multiplicative, lr_one_cycle, lr_reduce_on_plateau, lr_step, #>   progress, tb, unfreeze"},{"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 $network, underlying nn_module. full description model, see ?LearnerTorch.","code":"mlp_custom = lrn(\"regr.mlp\",   # construction arguments   optimizer = sgd, loss = l1, callbacks = history,   # scores to keep track of   measures_valid = msr(\"regr.mae\"),   # other parameters are left as-is:   # architecture   neurons = c(50, 50),   # training arguments   batch_size = 32, epochs = 30, device = \"cpu\",   # validation proportion   validate = 0.3 )  mlp_custom #> <LearnerTorchMLP[regr]:regr.mlp>: My Little Powny #> * Model: - #> * Parameters: epochs=30, device=cpu, num_threads=1, #>   num_interop_threads=1, seed=random, jit_trace=FALSE, eval_freq=1, #>   measures_train=<list>, measures_valid=<MeasureRegrSimple>, #>   patience=0, min_delta=0, batch_size=32, shuffle=TRUE, #>   tensor_dataset=FALSE, neurons=50,50, p=0.5, activation=<nn_relu>, #>   activation_args=<list>, opt.lr=0.5, opt.nesterov=FALSE #> * Validate: 0.3 #> * Packages: mlr3, mlr3torch, torch #> * Predict Types:  [response] #> * Feature Types: integer, numeric, lazy_tensor #> * Properties: internal_tuning, marshal, validation #> * Optimizer: sgd #> * Loss: l1 #> * Callbacks: history mlp_custom$train(task, row_ids = splits$train) prediction_custom = mlp_custom$predict(task, row_ids = splits$test) prediction_custom$score(msr(\"regr.mae\")) #> regr.mae  #> 8.543908 prediction$score(msr(\"regr.mae\")) #> regr.mae  #> 15.54688 head(mlp_custom$model$callbacks$history) #>    epoch valid.regr.mae #>    <num>          <num> #> 1:     1   2.310486e+05 #> 2:     2   2.009898e+10 #> 3:     3   6.967264e+03 #> 4:     4   3.220922e+02 #> 5:     5   7.638778e+01 #> 6:     6   7.588778e+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 #>  0.5186  0.0007 -0.5029 #>  0.3135  1.2928  1.4883 #> [ CPUFloatType{2,3} ] module_1 = nn_linear(in_features = 3, out_features = 4, bias = TRUE) output = module_1(input) output #> torch_tensor #> -0.0060  0.5752 -0.1935 -0.0095 #> -1.0660  1.3540 -1.2291 -1.1676 #> [ 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 #>  0.5381  0.2576  0.2043 #>  0.5026  0.2841  0.2133 #> [ 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 #> -0.0060  0.5752 -0.1935 -0.0095 #> -1.0660  1.3540 -1.2291 -1.1676 #> [ 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 #>  0.5381  0.2576  0.2043 #>  0.5026  0.2841  0.2133 #> [ 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 #> $module_list #> An `nn_module` containing 31 parameters. #>  #> ── Modules ───────────────────────────────────────────────────────────────────── #> • 0: <nn_linear> #16 parameters #> • 1: <nn_sigmoid> #0 parameters #> • 2: <nn_linear> #15 parameters #> • 3: <nn_softmax> #0 parameters graph_module(input) #> torch_tensor #>  0.5381  0.2576  0.2043 #>  0.5026  0.2841  0.2133 #> [ CPUFloatType{2,3} ][ grad_fn = <SoftmaxBackward0> ]"},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"modeldescriptor","dir":"Articles","previous_headings":"Building Torch Models for Tasks using PipeOpTorch","what":"ModelDescriptor","title":"Internals","text":"PipeOpModule represents nn_module fixed specific tensor shape hyperparameters. constructing neural network using operators, one take care output shape operations match input shapes following operations. complete Graph matching PipeOpModules can constructed using operators mostly inherit PipeOpTorch, making use ModelDescriptor class. ModelDescriptor class contains Graph (mostly) PipeOpModules information. PipeOpTorch transforms ModelDescriptor adds PipeOpModules Graph. ModelDescriptors always build Graph specific Task. easiest way initialize proper ModelDescriptor use appropriate PipeOpTorchIngress given datatype. use PipeOpTorchIngressNumeric, used numeric data. ModelDescriptor S3 object contains Graph, information generate data ($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 #> -0.2818  0.9781 -0.2101 -0.0965 #> -1.0439 -0.2501  0.6842 -0.5891 #> [ 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$ingress )  td #> <task_dataset> #>   Inherits from: <dataset> #>   Public: #>     .getbatch: function (index)  #>     .getitem: function (index)  #>     .length: function ()  #>     all_features: Petal.Length Sepal.Length Sepal.Width Species #>     cache_lazy_tensors: FALSE #>     clone: function (deep = FALSE)  #>     feature_ingress_tokens: list #>     initialize: function (task, feature_ingress_tokens, target_batchgetter = NULL)  #>     load_state_dict: function (x, ..., .refer_to_state_dict = FALSE)  #>     state_dict: function ()  #>     target_batchgetter: NULL #>     task: TaskClassif, TaskSupervised, Task, R6 batch = td$.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 #> -1.7275 -0.8973  2.2747  0.1551 #> -1.5510 -0.6123  2.0240  0.3031 #> -1.6239 -0.7723  2.0707  0.1194 #> [ 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 #>  0.4332  0.2392  0.3276 #>  0.4445  0.2243  0.3311 #> [ 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 )  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 * #> -3.3103 #>  -3.3103 #> [ CPUFloatType{2,1} ][ grad_fn = <AddmmBackward0> ] #>  #> $nn_softmax.output #> torch_tensor #>  0.3040  0.3772  0.3189 #>  0.3040  0.3772  0.3189 #> [ CPUFloatType{2,3} ][ grad_fn = <SoftmaxBackward0> ] iris_module$graph$pipeops$linear1$.result #> $output #> torch_tensor #> -3.8395 -3.5382 -1.5806 -0.5244 #> -3.6195 -3.1991 -1.7572 -0.6355 #> [ CPUFloatType{2,4} ][ grad_fn = <AddmmBackward0> ] iris_module$graph$pipeops$linear3$.result #> $output #> torch_tensor #> -1.0179 -0.4841  0.5960  0.6553 -0.2139 #> -1.0179 -0.4841  0.5960  0.6553 -0.2139 #> [ CPUFloatType{2,5} ][ grad_fn = <AddmmBackward0> ] iris_module$graph$pipeops$nn_merge_cat$.result #> $output #> torch_tensor #> -3.8395 -3.5382 -1.5806 -0.5244 -1.0179 -0.4841  0.5960  0.6553 -0.2139 #> -3.6195 -3.1991 -1.7572 -0.6355 -1.0179 -0.4841  0.5960  0.6553 -0.2139 #> [ 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_ignite_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$ingress,   optimizer = adam,   callbacks = history,   loss = xe )  lr_sequential #> <LearnerTorchModel[classif]:classif.model>: Torch Model #> * Model: - #> * Parameters: device=auto, num_threads=1, num_interop_threads=1, #>   seed=random, jit_trace=FALSE, eval_freq=1, measures_train=<list>, #>   measures_valid=<list>, patience=0, min_delta=0, shuffle=TRUE, #>   tensor_dataset=FALSE, opt.lr=0.02 #> * Validate: NULL #> * Packages: mlr3, mlr3torch, torch #> * Predict Types:  [response], prob #> * Feature Types: logical, integer, numeric, character, factor, ordered, #>   POSIXct, lazy_tensor #> * Properties: featureless, hotstart_backward, hotstart_forward, #>   importance, internal_tuning, marshal, missings, multiclass, #>   oob_error, selected_features, twoclass, validation, weights #> * Optimizer: adam #> * Loss: cross_entropy #> * Callbacks: history lr_sequential$param_set$set_values(   batch_size = 50,   epochs = 100,   measures_train = msrs(c(\"classif.logloss\", \"classif.ce\")) ) # This is required to evaluate the logloss during training lr_sequential$predict_type = \"prob\"  lr_sequential$train(md_sequential$task) library(data.table) library(ggplot2)  newdata = cbind(data.table(Sepal.Width = mean(iris$Sepal.Width)), CJ(   Sepal.Length = seq(min(iris$Sepal.Length), max(iris$Sepal.Length), length.out = 30),   Petal.Length = seq(min(iris$Petal.Length), max(iris$Petal.Length), length.out = 30) ))  predictions = lr_sequential$predict_newdata(newdata)  plot_predictions = function(predictions) {   ggplot(cbind(newdata, Species = predictions$response),       aes(x = Sepal.Length, y = Petal.Length, fill = Species)) +     geom_tile(alpha = .3) +     geom_tile(alpha = .3) +     geom_point(data = iris,       aes(x = Sepal.Length, y = Petal.Length, fill = Species),       color = \"black\", pch = 21, size = 3) +     theme_bw() } plot_predictions(predictions)"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"torch-learner-pipelines","dir":"Articles","previous_headings":"","what":"Torch Learner Pipelines","title":"Internals","text":"model shown constructed using ModelDescriptor generated Graph PipeOpTorch operators. ModelDescriptor furthermore contains Task pertains. makes possible use create NN model gets trained right away, using PipeOpTorchModelClassif. missing prerequisite now add desired TorchOptimizer TorchLoss information ModelDescriptor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"adding-optimizer-loss-and-callback-meta-info-to-modeldescriptor","dir":"Articles","previous_headings":"Torch Learner Pipelines","what":"Adding Optimizer, Loss and Callback Meta-Info to ModelDescriptor","title":"Internals","text":"Remember ModelDescriptor $optimizer, $loss $callbacks slots necessary build complete Learner NN. can set corresponding PipeOpTorch operators. po(\"torch_optimizer\") used set $optimizer slot ModelDescriptor; takes desired TorchOptimizer object construction exports ParamSet. works analogously loss-function. also callbacks:","code":"po_adam = po(\"torch_optimizer\", optimizer = adam)  # hyperparameters are made available and can be changed: po_adam$param_set$values #> $lr #> [1] 0.02  md_sequential = po_adam$train(list(md_sequential))[[1]]  md_sequential$optimizer #> <TorchOptimizer:adam> Adaptive Moment Estimation #> * Generator: optim_ignite_adam #> * Parameters: lr=0.02 #> * Packages: torch,mlr3torch po_xe = po(\"torch_loss\", loss = xe) md_sequential = po_xe$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.771e6d.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.771e6d.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.771e6d.x        <list>   poly.x              #>        poly.x <<UNTRAINED>>          nop.771e6d.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 (2025). 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 = {2025},   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\", 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! course (mlr3)torch. can check presentation UseR 2024.","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.75a3c3.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.75a3c3.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/PipeOpPreprocTorchTrafoNop.html","id":null,"dir":"Reference","previous_headings":"","what":"No Transformation — mlr_pipeops_trafo_nop","title":"No Transformation — 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":"No Transformation — mlr_pipeops_trafo_nop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoReshape.html","id":null,"dir":"Reference","previous_headings":"","what":"Reshaping Transformation — mlr_pipeops_trafo_reshape","title":"Reshaping Transformation — 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":"Reshaping Transformation — 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":"Reshaping Transformation — 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/Select.html","id":null,"dir":"Reference","previous_headings":"","what":"Selector Functions for Character Vectors — Select","title":"Selector Functions for Character Vectors — Select","text":"Select function subsets character vector. used callback CallbackSetUnfreeze select parameters freeze unfreeze training.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/Select.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Selector Functions for Character Vectors — Select","text":"","code":"select_all()  select_none()  select_grep(pattern, ignore.case = FALSE, perl = FALSE, fixed = FALSE)  select_name(param_names, assert_present = TRUE)  select_invert(select)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/Select.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Selector Functions for Character Vectors — Select","text":"pattern See grep() ignore.case See grep() perl See grep() fixed See grep() param_names names parameters want select assert_present Whether check param_names subset full vector names select Select","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/Select.html","id":"functions","dir":"Reference","previous_headings":"","what":"Functions","title":"Selector Functions for Character Vectors — Select","text":"select_all(): select_all selects elements select_none(): select_none selects elements select_grep(): select_grep selects elements names matching regular expression select_name(): select_name selects elements names matching given names select_invert(): select_invert selects elements selected given selector","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/Select.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Selector Functions for Character Vectors — Select","text":"","code":"select_all()(c(\"a\", \"b\")) #> [1] \"a\" \"b\" select_none()(c(\"a\", \"b\")) #> character(0) select_grep(\"b$\")(c(\"ab\", \"ac\")) #> [1] \"ab\" select_name(\"a\")(c(\"a\", \"b\")) #> [1] \"a\" select_invert(select_all())(c(\"a\", \"b\")) #> character(0)"},{"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,   additional_args = 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 $help(). additional_args () Additional arguments necessary. learning rate schedulers, torch::LRScheduler.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Torch Callback — TorchCallback","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Callback — TorchCallback","text":"","code":"TorchCallback$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/RtmptZtedw/file6e8513f33c91, 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(39)> #>                          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:               jit_trace ParamLgl    NA    NA       2 <NoDefault[0]> #>  7:               eval_freq ParamInt 1e+00   Inf     Inf <NoDefault[0]> #>  8:          measures_train ParamUty    NA    NA     Inf <NoDefault[0]> #>  9:          measures_valid ParamUty    NA    NA     Inf <NoDefault[0]> #> 10:                patience ParamInt 0e+00   Inf     Inf <NoDefault[0]> #> 11:               min_delta ParamDbl 0e+00   Inf     Inf <NoDefault[0]> #> 12:              batch_size ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 13:                 shuffle ParamLgl    NA    NA       2          FALSE #> 14:                 sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 15:           batch_sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 16:             num_workers ParamInt 0e+00   Inf     Inf              0 #> 17:              collate_fn ParamUty    NA    NA     Inf         [NULL] #> 18:              pin_memory ParamLgl    NA    NA       2          FALSE #> 19:               drop_last ParamLgl    NA    NA       2          FALSE #> 20:                 timeout ParamDbl  -Inf   Inf     Inf             -1 #> 21:          worker_init_fn ParamUty    NA    NA     Inf <NoDefault[0]> #> 22:          worker_globals ParamUty    NA    NA     Inf <NoDefault[0]> #> 23:         worker_packages ParamUty    NA    NA     Inf <NoDefault[0]> #> 24:          tensor_dataset ParamFct    NA    NA       1 <NoDefault[0]> #> 25:                 neurons ParamUty    NA    NA     Inf <NoDefault[0]> #> 26:                       p ParamDbl 0e+00 1e+00     Inf <NoDefault[0]> #> 27:                n_layers ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 28:              activation ParamUty    NA    NA     Inf <NoDefault[0]> #> 29:         activation_args ParamUty    NA    NA     Inf <NoDefault[0]> #> 30:                   shape ParamUty    NA    NA     Inf <NoDefault[0]> #> 31:                  opt.lr ParamDbl 0e+00   Inf     Inf          0.001 #> 32:               opt.betas ParamUty    NA    NA     Inf    0.900,0.999 #> 33:                 opt.eps ParamDbl 1e-16 1e-04     Inf          1e-08 #> 34:        opt.weight_decay ParamDbl 0e+00 1e+00     Inf              0 #> 35:             opt.amsgrad ParamLgl    NA    NA       2          FALSE #> 36:          loss.reduction ParamFct    NA    NA       2           mean #> 37:      cb.checkpoint.path ParamUty    NA    NA     Inf <NoDefault[0]> #> 38:      cb.checkpoint.freq ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 39: 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:        FALSE #>  7:            1 #>  8:    <list[0]> #>  9:    <list[0]> #> 10:            0 #> 11:            0 #> 12:       [NULL] #> 13:         TRUE #> 14:       [NULL] #> 15:       [NULL] #> 16:       [NULL] #> 17:       [NULL] #> 18:       [NULL] #> 19:       [NULL] #> 20:       [NULL] #> 21:       [NULL] #> 22:       [NULL] #> 23:       [NULL] #> 24:        FALSE #> 25:              #> 26:          0.5 #> 27:       [NULL] #> 28: <nn_relu[1]> #> 29:    <list[0]> #> 30:       [NULL] #> 31:       [NULL] #> 32:       [NULL] #> 33:       [NULL] #> 34:       [NULL] #> 35:       [NULL] #> 36:       [NULL] #> 37:       [NULL] #> 38:       [NULL] #> 39:       [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,   additional_args = 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 $help(). additional_args (list()) Additional arguments necessary. learning rate schedulers, torch::LRScheduler.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"method-print-","dir":"Reference","previous_headings":"","what":"Method print()","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"Prints object","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"","code":"TorchDescriptor$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(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:           jit_trace ParamLgl    NA    NA       2 <NoDefault[0]> #>  7:           eval_freq ParamInt 1e+00   Inf     Inf <NoDefault[0]> #>  8:      measures_train ParamUty    NA    NA     Inf <NoDefault[0]> #>  9:      measures_valid ParamUty    NA    NA     Inf <NoDefault[0]> #> 10:            patience ParamInt 0e+00   Inf     Inf <NoDefault[0]> #> 11:           min_delta ParamDbl 0e+00   Inf     Inf <NoDefault[0]> #> 12:          batch_size ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 13:             shuffle ParamLgl    NA    NA       2          FALSE #> 14:             sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 15:       batch_sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 16:         num_workers ParamInt 0e+00   Inf     Inf              0 #> 17:          collate_fn ParamUty    NA    NA     Inf         [NULL] #> 18:          pin_memory ParamLgl    NA    NA       2          FALSE #> 19:           drop_last ParamLgl    NA    NA       2          FALSE #> 20:             timeout ParamDbl  -Inf   Inf     Inf             -1 #> 21:      worker_init_fn ParamUty    NA    NA     Inf <NoDefault[0]> #> 22:      worker_globals ParamUty    NA    NA     Inf <NoDefault[0]> #> 23:     worker_packages ParamUty    NA    NA     Inf <NoDefault[0]> #> 24:      tensor_dataset ParamFct    NA    NA       1 <NoDefault[0]> #> 25:             neurons ParamUty    NA    NA     Inf <NoDefault[0]> #> 26:                   p ParamDbl 0e+00 1e+00     Inf <NoDefault[0]> #> 27:            n_layers ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 28:          activation ParamUty    NA    NA     Inf <NoDefault[0]> #> 29:     activation_args ParamUty    NA    NA     Inf <NoDefault[0]> #> 30:               shape ParamUty    NA    NA     Inf <NoDefault[0]> #> 31:              opt.lr ParamDbl 0e+00   Inf     Inf          0.001 #> 32:           opt.betas ParamUty    NA    NA     Inf    0.900,0.999 #> 33:             opt.eps ParamDbl 1e-16 1e-04     Inf          1e-08 #> 34:    opt.weight_decay ParamDbl 0e+00 1e+00     Inf              0 #> 35:         opt.amsgrad ParamLgl    NA    NA       2          FALSE #> 36:      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:        FALSE #>  7:            1 #>  8:    <list[0]> #>  9:    <list[0]> #> 10:            0 #> 11:            0 #> 12:       [NULL] #> 13:         TRUE #> 14:       [NULL] #> 15:       [NULL] #> 16:       [NULL] #> 17:       [NULL] #> 18:       [NULL] #> 19:       [NULL] #> 20:       [NULL] #> 21:       [NULL] #> 22:       [NULL] #> 23:       [NULL] #> 24:        FALSE #> 25:              #> 26:          0.5 #> 27:       [NULL] #> 28: <nn_relu[1]> #> 29:    <list[0]> #> 30:       [NULL] #> 31:       [NULL] #> 32:       [NULL] #> 33:       [NULL] #> 34:       [NULL] #> 35:       [NULL] #> 36:       [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_ignite_adam, label = \"adam\") torch_opt #> <TorchOptimizer:optim_ignite_adam> adam #> * Generator: optim_ignite_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_ignite_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 (set_to_none = FALSE)  #>   Private: #>     deep_clone: function (name, value)  #>     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(36)> #>                      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:           jit_trace ParamLgl    NA    NA       2 <NoDefault[0]> #>  7:           eval_freq ParamInt     1   Inf     Inf <NoDefault[0]> #>  8:      measures_train ParamUty    NA    NA     Inf <NoDefault[0]> #>  9:      measures_valid ParamUty    NA    NA     Inf <NoDefault[0]> #> 10:            patience ParamInt     0   Inf     Inf <NoDefault[0]> #> 11:           min_delta ParamDbl     0   Inf     Inf <NoDefault[0]> #> 12:          batch_size ParamInt     1   Inf     Inf <NoDefault[0]> #> 13:             shuffle ParamLgl    NA    NA       2          FALSE #> 14:             sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 15:       batch_sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 16:         num_workers ParamInt     0   Inf     Inf              0 #> 17:          collate_fn ParamUty    NA    NA     Inf         [NULL] #> 18:          pin_memory ParamLgl    NA    NA       2          FALSE #> 19:           drop_last ParamLgl    NA    NA       2          FALSE #> 20:             timeout ParamDbl  -Inf   Inf     Inf             -1 #> 21:      worker_init_fn ParamUty    NA    NA     Inf <NoDefault[0]> #> 22:      worker_globals ParamUty    NA    NA     Inf <NoDefault[0]> #> 23:     worker_packages ParamUty    NA    NA     Inf <NoDefault[0]> #> 24:      tensor_dataset ParamFct    NA    NA       1 <NoDefault[0]> #> 25:             neurons ParamUty    NA    NA     Inf <NoDefault[0]> #> 26:                   p ParamDbl     0     1     Inf <NoDefault[0]> #> 27:            n_layers ParamInt     1   Inf     Inf <NoDefault[0]> #> 28:          activation ParamUty    NA    NA     Inf <NoDefault[0]> #> 29:     activation_args ParamUty    NA    NA     Inf <NoDefault[0]> #> 30:               shape ParamUty    NA    NA     Inf <NoDefault[0]> #> 31:              opt.lr ParamDbl     0   Inf     Inf <NoDefault[0]> #> 32:        opt.momentum ParamDbl     0     1     Inf              0 #> 33:       opt.dampening ParamDbl     0     1     Inf              0 #> 34:    opt.weight_decay ParamDbl     0     1     Inf              0 #> 35:        opt.nesterov ParamLgl    NA    NA       2          FALSE #> 36:      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:        FALSE #>  7:            1 #>  8:    <list[0]> #>  9:    <list[0]> #> 10:            0 #> 11:            0 #> 12:       [NULL] #> 13:         TRUE #> 14:       [NULL] #> 15:       [NULL] #> 16:       [NULL] #> 17:       [NULL] #> 18:       [NULL] #> 19:       [NULL] #> 20:       [NULL] #> 21:       [NULL] #> 22:       [NULL] #> 23:       [NULL] #> 24:        FALSE #> 25:              #> 26:          0.5 #> 27:       [NULL] #> 28: <nn_relu[1]> #> 29:    <list[0]> #> 30:       [NULL] #> 31:       [NULL] #> 32:       [NULL] #> 33:       [NULL] #> 34:       [NULL] #> 35:       [NULL] #> 36:       [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.2e6303.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$iris) )() as_data_descriptor(ds2) #> <DataDescriptor: 1 ops> #> * dataset_shapes: [x: (NA,4)] #> * input_map: (x) -> Graph #> * pointer: nop.6caef5.x.output #> * shape: [(NA,4)]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert to Lazy Tensor — as_lazy_tensor","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"Convert object lazy_tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"","code":"as_lazy_tensor(x, ...)  # S3 method for class 'dataset' as_lazy_tensor(x, dataset_shapes = NULL, ids = NULL, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"x () Object convert lazy_tensor ... () Additional arguments passed method. dataset_shapes (named list() (integer() NULL)) shapes output. Names elements list returned dataset. shape NULL (unknown, e.g. images different sizes) first dimension must NA indicate batch dimension. ids (integer()) ids include lazy tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"","code":"iris_ds = dataset(\"iris\",   initialize = function() {     self$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_lr_scheduler.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert to CallbackSetLRScheduler — as_lr_scheduler","title":"Convert to CallbackSetLRScheduler — as_lr_scheduler","text":"Convert torch scheduler generator CallbackSetLRScheduler.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lr_scheduler.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert to CallbackSetLRScheduler — as_lr_scheduler","text":"","code":"as_lr_scheduler(x, step_on_epoch)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lr_scheduler.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert to CallbackSetLRScheduler — as_lr_scheduler","text":"x (function) torch scheduler generator defined using torch::lr_scheduler(). step_on_epoch (logical(1)) Whether scheduler steps every epoch","code":""},{"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, ...)"},{"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. ... () 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, ...)"},{"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. ... () 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 #> -0.8098 -0.5025  0.9536 #>  0.1845 -0.4500 -0.8570 #>  1.8274  0.3922 -1.0590 #>  0.1568  0.8758 -0.5632 #> -0.6920  0.7932 -2.2409 #>  1.4585  0.7990 -1.4406 #>  0.3679 -0.8739  0.9497 #>  0.6540  0.5656  0.1746 #>  0.1380 -1.4313 -0.0537 #> -0.9234  1.6273  0.2737 #> [ CPUFloatType{10,3} ] materialize(lt1, rbind = FALSE) #> [[1]] #> torch_tensor #> -0.8098 #> -0.5025 #>  0.9536 #> [ CPUFloatType{3} ] #>  #> [[2]] #> torch_tensor #>  0.1845 #> -0.4500 #> -0.8570 #> [ CPUFloatType{3} ] #>  #> [[3]] #> torch_tensor #>  1.8274 #>  0.3922 #> -1.0590 #> [ CPUFloatType{3} ] #>  #> [[4]] #> torch_tensor #>  0.1568 #>  0.8758 #> -0.5632 #> [ CPUFloatType{3} ] #>  #> [[5]] #> torch_tensor #> -0.6920 #>  0.7932 #> -2.2409 #> [ CPUFloatType{3} ] #>  #> [[6]] #> torch_tensor #>  1.4585 #>  0.7990 #> -1.4406 #> [ CPUFloatType{3} ] #>  #> [[7]] #> torch_tensor #>  0.3679 #> -0.8739 #>  0.9497 #> [ CPUFloatType{3} ] #>  #> [[8]] #> torch_tensor #>  0.6540 #>  0.5656 #>  0.1746 #> [ CPUFloatType{3} ] #>  #> [[9]] #> torch_tensor #>  0.1380 #> -1.4313 #> -0.0537 #> [ CPUFloatType{3} ] #>  #> [[10]] #> torch_tensor #> -0.9234 #>  1.6273 #>  0.2737 #> [ 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 #> -0.8098 -0.5025  0.9536 #>  0.1845 -0.4500 -0.8570 #>  1.8274  0.3922 -1.0590 #>  0.1568  0.8758 -0.5632 #> -0.6920  0.7932 -2.2409 #>  1.4585  0.7990 -1.4406 #>  0.3679 -0.8739  0.9497 #>  0.6540  0.5656  0.1746 #>  0.1380 -1.4313 -0.0537 #> -0.9234  1.6273  0.2737 #> [ CPUFloatType{10,3} ] #>  #> $lt2 #> torch_tensor #>  1.7508  0.5892  1.5608  1.1568 #> -0.8770 -1.1304  0.5787  1.2020 #> -0.2732 -0.7877  2.2388  0.1827 #> -0.1450  0.7184 -0.3360 -0.1029 #>  0.7497 -1.0479 -0.8173 -2.3177 #> -1.2669  0.1360 -0.7766  0.1551 #>  1.3982  1.1425  1.8144  3.2104 #>  2.6315 -2.0763 -0.4969  0.3670 #> -0.6687  0.0809 -0.7422 -0.2983 #>  2.1002  0.4322  0.7277 -1.5068 #> [ CPUFloatType{10,4} ] #>  materialize(d, rbind = FALSE) #> $lt1 #> $lt1[[1]] #> torch_tensor #> -0.8098 #> -0.5025 #>  0.9536 #> [ CPUFloatType{3} ] #>  #> $lt1[[2]] #> torch_tensor #>  0.1845 #> -0.4500 #> -0.8570 #> [ CPUFloatType{3} ] #>  #> $lt1[[3]] #> torch_tensor #>  1.8274 #>  0.3922 #> -1.0590 #> [ CPUFloatType{3} ] #>  #> $lt1[[4]] #> torch_tensor #>  0.1568 #>  0.8758 #> -0.5632 #> [ CPUFloatType{3} ] #>  #> $lt1[[5]] #> torch_tensor #> -0.6920 #>  0.7932 #> -2.2409 #> [ CPUFloatType{3} ] #>  #> $lt1[[6]] #> torch_tensor #>  1.4585 #>  0.7990 #> -1.4406 #> [ CPUFloatType{3} ] #>  #> $lt1[[7]] #> torch_tensor #>  0.3679 #> -0.8739 #>  0.9497 #> [ CPUFloatType{3} ] #>  #> $lt1[[8]] #> torch_tensor #>  0.6540 #>  0.5656 #>  0.1746 #> [ CPUFloatType{3} ] #>  #> $lt1[[9]] #> torch_tensor #>  0.1380 #> -1.4313 #> -0.0537 #> [ CPUFloatType{3} ] #>  #> $lt1[[10]] #> torch_tensor #> -0.9234 #>  1.6273 #>  0.2737 #> [ CPUFloatType{3} ] #>  #>  #> $lt2 #> $lt2[[1]] #> torch_tensor #>  1.7508 #>  0.5892 #>  1.5608 #>  1.1568 #> [ CPUFloatType{4} ] #>  #> $lt2[[2]] #> torch_tensor #> -0.8770 #> -1.1304 #>  0.5787 #>  1.2020 #> [ CPUFloatType{4} ] #>  #> $lt2[[3]] #> torch_tensor #> -0.2732 #> -0.7877 #>  2.2388 #>  0.1827 #> [ CPUFloatType{4} ] #>  #> $lt2[[4]] #> torch_tensor #> -0.1450 #>  0.7184 #> -0.3360 #> -0.1029 #> [ CPUFloatType{4} ] #>  #> $lt2[[5]] #> torch_tensor #>  0.7497 #> -1.0479 #> -0.8173 #> -2.3177 #> [ CPUFloatType{4} ] #>  #> $lt2[[6]] #> torch_tensor #> -1.2669 #>  0.1360 #> -0.7766 #>  0.1551 #> [ CPUFloatType{4} ] #>  #> $lt2[[7]] #> torch_tensor #>  1.3982 #>  1.1425 #>  1.8144 #>  3.2104 #> [ CPUFloatType{4} ] #>  #> $lt2[[8]] #> torch_tensor #>  2.6315 #> -2.0763 #> -0.4969 #>  0.3670 #> [ CPUFloatType{4} ] #>  #> $lt2[[9]] #> torch_tensor #> -0.6687 #>  0.0809 #> -0.7422 #> -0.2983 #> [ CPUFloatType{4} ] #>  #> $lt2[[10]] #> torch_tensor #>  2.1002 #>  0.4322 #>  0.7277 #> -1.5068 #> [ 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 #>                   <char>                                  <char> #>  1:           checkpoint                              Checkpoint #>  2:              history                                 History #>  3:  lr_cosine_annealing           Cosine Annealing LR Scheduler #>  4:            lr_lambda Multiplication by Function LR Scheduler #>  5:    lr_multiplicative   Multiplication by Factor LR Scheduler #>  6:         lr_one_cycle                      1cyle LR Scheduler #>  7: lr_reduce_on_plateau          Reduce on Plateau LR Scheduler #>  8:              lr_step                 Step Decay LR Scheduler #>  9:             progress                                Progress #> 10:                   tb                             TensorBoard #> 11:             unfreeze                                Unfreeze #>                     packages #>                       <list> #>  1:          mlr3torch,torch #>  2:          mlr3torch,torch #>  3:          mlr3torch,torch #>  4:          mlr3torch,torch #>  5:          mlr3torch,torch #>  6:          mlr3torch,torch #>  7:          mlr3torch,torch #>  8:          mlr3torch,torch #>  9: progress,mlr3torch,torch #> 10: tfevents,mlr3torch,torch #> 11:          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>          <list> #> 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":"adagrad, adam, adamw, rmsprop, 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_ignite_adam #> * Parameters: list() #> * Packages: torch,mlr3torch # is equivalent to t_opt(\"adam\") #> <TorchOptimizer:adam> Adaptive Moment Estimation #> * Generator: optim_ignite_adam #> * Parameters: list() #> * Packages: torch,mlr3torch # convert to a data.table as.data.table(mlr3torch_optimizers) #> Key: <key> #>        key                                 label        packages #>     <char>                                <char>          <list> #> 1: adagrad           Adaptive Gradient algorithm torch,mlr3torch #> 2:    adam            Adaptive Moment Estimation torch,mlr3torch #> 3:   adamw Decoupled Weight Decay Regularization torch,mlr3torch #> 4: rmsprop          Root Mean Square Propagation torch,mlr3torch #> 5:     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.checkpoint.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"","code":"cb = t_clbk(\"checkpoint\", freq = 1) task = tsk(\"iris\")  pth = tempfile() learner = lrn(\"classif.mlp\", epochs = 3, batch_size = 1, callbacks = cb) learner$param_set$set_values(cb.checkpoint.path = pth)  learner$train(task)  list.files(pth) #> [1] \"network1.pt\"   \"network2.pt\"   \"network3.pt\"   \"optimizer1.pt\" #> [5] \"optimizer2.pt\" \"optimizer3.pt\""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":null,"dir":"Reference","previous_headings":"","what":"History Callback — mlr_callback_set.history","title":"History Callback — mlr_callback_set.history","text":"Saves training validation history training. history saved data.table validation measures prefixed \"valid.\" training measures prefixed \"train.\".","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"History Callback — mlr_callback_set.history","text":"mlr3torch::CallbackSet -> CallbackSetHistory","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"History Callback — mlr_callback_set.history","text":"mlr3torch::CallbackSet$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.history.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"History Callback — mlr_callback_set.history","text":"","code":"cb = t_clbk(\"history\") task = tsk(\"iris\")  learner = lrn(\"classif.mlp\", epochs = 3, batch_size = 1,   callbacks = t_clbk(\"history\"), validate = 0.3) learner$param_set$set_values(   measures_train = msrs(c(\"classif.acc\", \"classif.ce\")),   measures_valid = msr(\"classif.ce\") ) learner$train(task)  print(learner$model$callbacks$history) #> Key: <epoch> #>    epoch train.classif.acc train.classif.ce valid.classif.ce #>    <num>             <num>            <num>            <num> #> 1:     1         0.3047619        0.6952381        0.6222222 #> 2:     2         0.3142857        0.6857143        0.7333333 #> 3:     3         0.3523810        0.6476190        0.5111111"},{"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.lr_scheduler.html","id":null,"dir":"Reference","previous_headings":"","what":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"Changes learning rate based schedule specified torch::lr_scheduler. writing, following available: torch::lr_cosine_annealing() torch::lr_lambda() torch::lr_multiplicative() torch::lr_one_cycle() torch::lr_reduce_on_plateau() torch::lr_step() Custom schedulers defined torch::lr_scheduler().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"mlr3torch::CallbackSet -> CallbackSetLRScheduler","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"public-fields","dir":"Reference","previous_headings":"","what":"Public fields","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"scheduler_fn (lr_scheduler_generator) torch function creates learning rate scheduler scheduler (LRScheduler) learning rate scheduler wrapped callback","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"mlr3torch::CallbackSet$load_state_dict() mlr3torch::CallbackSet$print() mlr3torch::CallbackSet$state_dict()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"CallbackSetLRScheduler$new() CallbackSetLRScheduler$on_begin() CallbackSetLRScheduler$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"","code":"CallbackSetLRScheduler$new(.scheduler, step_on_epoch, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":".scheduler (lr_scheduler_generator) torch scheduler generator (e.g. torch::lr_step). step_on_epoch (logical(1)) Whether scheduler steps every epoch (otherwise every batch). ... () scheduler-specific arguments","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"method-on-begin-","dir":"Reference","previous_headings":"","what":"Method on_begin()","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"Creates scheduler using optimizer context","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"","code":"CallbackSetLRScheduler$on_begin()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"","code":"CallbackSetLRScheduler$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":null,"dir":"Reference","previous_headings":"","what":"Progress Callback — mlr_callback_set.progress","title":"Progress Callback — mlr_callback_set.progress","text":"Prints progress bar metrics training validation.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Progress Callback — mlr_callback_set.progress","text":"mlr3torch::CallbackSet -> CallbackSetProgress","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Progress Callback — mlr_callback_set.progress","text":"mlr3torch::CallbackSet$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$on_epoch_end() CallbackSetProgress$on_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-on-end-","dir":"Reference","previous_headings":"","what":"Method on_end()","title":"Progress Callback — mlr_callback_set.progress","text":"Prints time end training.","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$on_end()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Progress Callback — mlr_callback_set.progress","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"usage-6","dir":"Reference","previous_headings":"","what":"Usage","title":"Progress Callback — mlr_callback_set.progress","text":"","code":"CallbackSetProgress$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.progress.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Progress Callback — mlr_callback_set.progress","text":"","code":"task = tsk(\"iris\")  learner = lrn(\"classif.mlp\", epochs = 5, batch_size = 1,   callbacks = t_clbk(\"progress\"), validate = 0.3) learner$param_set$set_values(   measures_train = msrs(c(\"classif.acc\", \"classif.ce\")),   measures_valid = msr(\"classif.ce\") )  learner$train(task) #> Epoch 1 started (2025-02-07 06:14:43) #>  #> [Summary epoch 1] #> ------------------ #> Measures (Train): #>  * classif.acc = 0.31 #>  * classif.ce = 0.69 #> Measures (Valid): #>  * classif.ce = 0.62 #> Epoch 2 started (2025-02-07 06:14:43) #>  #> [Summary epoch 2] #> ------------------ #> Measures (Train): #>  * classif.acc = 0.31 #>  * classif.ce = 0.69 #> Measures (Valid): #>  * classif.ce = 0.56 #> Epoch 3 started (2025-02-07 06:14:44) #>  #> [Summary epoch 3] #> ------------------ #> Measures (Train): #>  * classif.acc = 0.55 #>  * classif.ce = 0.45 #> Measures (Valid): #>  * classif.ce = 0.44 #> Epoch 4 started (2025-02-07 06:14:44) #>  #> [Summary epoch 4] #> ------------------ #> Measures (Train): #>  * classif.acc = 0.57 #>  * classif.ce = 0.43 #> Measures (Valid): #>  * classif.ce = 0.49 #> Epoch 5 started (2025-02-07 06:14:44) #>  #> [Summary epoch 5] #> ------------------ #> Measures (Train): #>  * classif.acc = 0.51 #>  * classif.ce = 0.49 #> Measures (Valid): #>  * classif.ce = 0.60 #> Finished training for 5 epochs (2025-02-07 06:14:45)"},{"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_callback_set.unfreeze.html","id":null,"dir":"Reference","previous_headings":"","what":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"Unfreeze weights (parameters network) number steps epochs.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"mlr3torch::CallbackSet -> CallbackSetUnfreeze","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"mlr3torch::CallbackSet$load_state_dict() mlr3torch::CallbackSet$print() mlr3torch::CallbackSet$state_dict()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"CallbackSetUnfreeze$new() CallbackSetUnfreeze$on_begin() CallbackSetUnfreeze$on_epoch_begin() CallbackSetUnfreeze$on_batch_begin() CallbackSetUnfreeze$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"","code":"CallbackSetUnfreeze$new(starting_weights, unfreeze)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"starting_weights (Select) Select denoting weights trainable start. unfreeze (data.table) data.table column weights (list column Selects) column epoch batch. selector indicates parameters unfreeze, epoch batch column indicates .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"method-on-begin-","dir":"Reference","previous_headings":"","what":"Method on_begin()","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"Sets starting weights","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"","code":"CallbackSetUnfreeze$on_begin()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"method-on-epoch-begin-","dir":"Reference","previous_headings":"","what":"Method on_epoch_begin()","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"Unfreezes weights training correct epoch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"","code":"CallbackSetUnfreeze$on_epoch_begin()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"method-on-batch-begin-","dir":"Reference","previous_headings":"","what":"Method on_batch_begin()","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"Unfreezes weights training correct batch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"usage-3","dir":"Reference","previous_headings":"","what":"Usage","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"","code":"CallbackSetUnfreeze$on_batch_begin()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"usage-4","dir":"Reference","previous_headings":"","what":"Usage","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"","code":"CallbackSetUnfreeze$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"","code":"task = tsk(\"iris\") cb = t_clbk(\"unfreeze\") mlp = lrn(\"classif.mlp\", callbacks = cb,  cb.unfreeze.starting_weights = select_invert(    select_name(c(\"0.weight\", \"3.weight\", \"6.weight\", \"6.bias\"))  ),  cb.unfreeze.unfreeze = data.table(    epoch = c(2, 5),    weights = list(select_name(\"0.weight\"), select_name(c(\"3.weight\", \"6.weight\")))  ),  epochs = 6, batch_size = 150, neurons = c(1, 1, 1) )  mlp$train(task)"},{"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. last_loss (numeric(1)) loss last trainings batch. 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. device (torch::torch_device) device.","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$new(   learner,   task_train,   task_valid = NULL,   loader_train,   loader_valid = NULL,   measures_train = NULL,   measures_valid = NULL,   network,   optimizer,   loss_fn,   total_epochs,   prediction_encoder,   eval_freq = 1L,   device )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Context for Torch Learner — mlr_context_torch","text":"learner (Learner) torch learner. task_train (Task) training task. task_valid (Task NULL) validation task. loader_train (torch::dataloader) data loader training. loader_valid (torch::dataloader NULL) data loader validation. measures_train (list() Measures NULL) Measures used training. Default NULL. measures_valid (list() Measures NULL) Measures used validation. network (torch::nn_module) torch network. optimizer (torch::optimizer) optimizer. loss_fn (torch::nn_module) loss function. total_epochs (integer(1)) total number epochs learner trained . prediction_encoder (function()) learner's prediction encoder. eval_freq (integer(1)) evaluation frequency. device (character(1)) device.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Context for Torch Learner — mlr_context_torch","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Context for Torch Learner — mlr_context_torch","text":"","code":"ContextTorch$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Context for Torch Learner — mlr_context_torch","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":null,"dir":"Reference","previous_headings":"","what":"My Little Pony — mlr_learners.mlp","title":"My Little Pony — mlr_learners.mlp","text":"Fully connected feed forward network dropout activation function. features can either single lazy_tensor one numeric columns ().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"dictionary","dir":"Reference","previous_headings":"","what":"Dictionary","title":"My Little Pony — mlr_learners.mlp","text":"Learner can instantiated using sugar function lrn():","code":"lrn(\"classif.mlp\", ...) lrn(\"regr.mlp\", ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"My Little Pony — mlr_learners.mlp","text":"Supported task types: 'classif', 'regr' Predict Types: classif: 'response', 'prob' regr: 'response' Feature Types: “integer”, “numeric”, “lazy_tensor” Required Packages: mlr3, mlr3torch, torch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"My Little Pony — mlr_learners.mlp","text":"Parameters LearnerTorch, well : activation :: [nn_module] activation function. initialized nn_relu. activation_args :: named list() named list initialization arguments activation function. intialized empty list. neurons :: integer() number neurons per hidden layer. default hidden layer. Setting c(10, 20) first hidden layer 10 neurons second 20. n_layers :: integer() number layers. parameter must set neurons length 1. p :: numeric(1) dropout probability. initialized 0.5. shape :: integer() NULL input shape length 2, e.g. c(NA, 5). needs present lazy tensor input unknown shape (NULL). Otherwise input shape inferred number numeric features.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"My Little Pony — mlr_learners.mlp","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.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  #>       0.64"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":null,"dir":"Reference","previous_headings":"","what":"Tabular ResNet — mlr_learners.tab_resnet","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Tabular resnet.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"dictionary","dir":"Reference","previous_headings":"","what":"Dictionary","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Learner can instantiated using sugar function lrn():","code":"lrn(\"classif.tab_resnet\", ...) lrn(\"regr.tab_resnet\", ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Supported task types: 'classif', 'regr' Predict Types: classif: 'response', 'prob' regr: 'response' Feature Types: “integer”, “numeric” Required Packages: mlr3, mlr3torch, torch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Parameters LearnerTorch, well : n_blocks :: integer(1) number blocks. d_block :: integer(1) input output dimension block. d_hidden :: integer(1) latent dimension block. d_hidden_multiplier :: numeric(1) Alternative way specify latent dimension d_block * d_hidden_multiplier. dropout1 :: numeric(1) First dropout ratio. dropout2 :: numeric(1) Second dropout ratio.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Gorishniy Y, Rubachev , Khrulkov V, Babenko (2021). “Revisiting Deep Learning  Tabular Data.” arXiv, 2106.11959.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> LearnerTorchTabResNet","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"LearnerTorchTabResNet$new() LearnerTorchTabResNet$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"","code":"LearnerTorchTabResNet$new(   task_type,   optimizer = NULL,   loss = NULL,   callbacks = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"task_type (character(1)) task type, either \"classif\" \"regr\". optimizer (TorchOptimizer) optimizer use training. Per default, adam used. loss (TorchLoss) loss used train network. Per default, mse used regression cross_entropy classification. callbacks (list() TorchCallbacks) callbacks. Must unique ids.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"","code":"LearnerTorchTabResNet$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"","code":"# Define the Learner and set parameter values learner = lrn(\"classif.tab_resnet\") learner$param_set$set_values(   epochs = 1, batch_size = 16, device = \"cpu\",   n_blocks = 2, d_block = 10, d_hidden = 20, dropout1 = 0.3, dropout2 = 0.3 )  # Define a Task task = tsk(\"iris\")  # Create train and test set ids = partition(task)  # Train the learner on the training ids learner$train(task, row_ids = ids$train)  # Make predictions for the test rows predictions = learner$predict(task, row_ids = ids$test)  # Score the predictions predictions$score() #> classif.ce  #>       0.66"},{"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$score() #> classif.ce  #>        0.7"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":null,"dir":"Reference","previous_headings":"","what":"AlexNet Image Classifier — mlr_learners.torchvision","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"Classic image classification networks torchvision.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"Parameters LearnerTorchImage pretrained :: logical(1) Whether use pretrained model. final linear layer replaced new nn_linear number classes inferred Task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"Supported task types: \"classif\" Predict Types: \"response\" \"prob\" Feature Types: \"lazy_tensor\" Required packages: \"mlr3torch\", \"torch\", \"torchvision\"","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> mlr3torch::LearnerTorchImage -> LearnerTorchVision","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"LearnerTorchVision$new() LearnerTorchVision$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"","code":"LearnerTorchVision$new(   name,   module_generator,   label,   optimizer = NULL,   loss = NULL,   callbacks = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"name (character(1)) name network. module_generator (function(pretrained, num_classes)) Function generates network. label (character(1)) label network. #' @references Krizhevsky, Alex, Sutskever, Ilya, Hinton, E. G (2017). “Imagenet classification deep convolutional neural networks.” Communications ACM, 60(6), 84–90. Sandler, Mark, Howard, Andrew, Zhu, Menglong, Zhmoginov, Andrey, Chen, Liang-Chieh (2018). “Mobilenetv2: Inverted residuals linear bottlenecks.” Proceedings IEEE conference computer vision pattern recognition, 4510–4520. , Kaiming, Zhang, Xiangyu, Ren, Shaoqing, Sun, Jian (2016). “Deep residual learning image recognition.” Proceedings IEEE conference computer vision pattern recognition, 770–778. Simonyan, Karen, Zisserman, Andrew (2014). “deep convolutional networks large-scale image recognition.” arXiv preprint arXiv:1409.1556. optimizer (TorchOptimizer) optimizer use training. Per default, adam used. loss (TorchLoss) loss used train network. Per default, mse used regression cross_entropy classification. callbacks (list() TorchCallbacks) callbacks. Must unique ids.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"","code":"LearnerTorchVision$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":null,"dir":"Reference","previous_headings":"","what":"Base Class for Torch Learners — mlr_learners_torch","title":"Base Class for Torch Learners — mlr_learners_torch","text":"base class provides basic functionality training prediction neural network. torch learners inherit class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"validation","dir":"Reference","previous_headings":"","what":"Validation","title":"Base Class for Torch Learners — mlr_learners_torch","text":"specify validation data, can set $validate field Learner, can set : NULL: validation ratio: proportion 1 - ratio task used training ratio used validation. \"test\" means \"test\" task resampling used possible calling $train() manually. \"predefined\": use predefined $internal_valid_task mlr3::Task. validation data can also used early stopping, see description Learner's parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"saving-a-learner","dir":"Reference","previous_headings":"","what":"Saving a Learner","title":"Base Class for Torch Learners — mlr_learners_torch","text":"order save LearnerTorch later usage, necessary call $marshal() method Learner writing disk, object otherwise saved correctly. loading marshaled LearnerTorch R , need call $unmarshal() transform useable state.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"early-stopping-and-internal-tuning","dir":"Reference","previous_headings":"","what":"Early Stopping and Internal Tuning","title":"Base Class for Torch Learners — mlr_learners_torch","text":"order prevent overfitting, LearnerTorch class allows use early stopping via patience min_delta parameters, see Learner's parameters. tuning LearnerTorch also possible combine explicit tuning via mlr3tuning LearnerTorch's internal tuning epochs via early stopping. , just need include epochs = to_tune(upper = <upper>, internal = TRUE) search space, <upper> maximally allowed number epochs, configure early stopping.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"model","dir":"Reference","previous_headings":"","what":"Model","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Model list class \"learner_torch_model\" following elements: network :: trained network. optimizer :: $state_dict() optimizer used train network. loss_fn :: $state_dict() loss used train network. callbacks :: callbacks used train network. seed :: seed / used training prediction. epochs :: many epochs model trained (early stopping). task_col_info :: data.table() containing information train-task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Base Class for Torch Learners — mlr_learners_torch","text":"General: parameters optimizer, loss callbacks, prefixed \"opt.\", \"loss.\" \"cb.<callback id>.\" respectively, well : epochs :: integer(1) number epochs. device :: character(1) device. One \"auto\", \"cpu\", \"cuda\" values defined mlr_reflections$torch$devices. value initialized \"auto\", select \"cuda\" possible, try \"mps\" otherwise fall back \"cpu\". num_threads :: integer(1) number threads intraop pararallelization (device \"cpu\"). value initialized 1. num_interop_threads :: integer(1) number threads intraop interop pararallelization (device \"cpu\"). value initialized 1. Note can set session changing value within R session raise warning. seed :: integer(1) \"random\" NULL torch seed used training prediction. value initialized \"random\", means random seed sampled beginning training phase. seed (either set randomly sampled) available via $model$seed training used prediction. Note setting seed training phase mean default (.e. seed \"random\"), clones learner use different seed. set NULL, seeding done. tensor_dataset :: logical(1) | \"device\" Whether load batches beginning training stack . initialized FALSE. set \"device\", device tensors set value device, can avoid unnecessary moving tensors devices. dataset fits memory make loading batches faster. Note set datasets contain lazy_tensors random data augmentation, augmentation applied beginning training. 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. set, epoch improvement observed, can accessed via $internal_tuned_values field learner. 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. initialized TRUE, differs default (FALSE). 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. dataset must return named list : x list torch tensors input network. networks one input, names must correspond inputs network. y target tensor. .index indices batch (integer() torch_int()). Moreover, one needs pay attention respect row ids provided task. also possible overwrite private .dataloader() method. must respect dataloader parameters ParamSet. .dataloader(dataset, param_vals) (Task, list()) -> torch::dataloader Create dataloader task. Needs respect least batch_size shuffle (otherwise predictions can permuted). change predict types, possible overwrite method : .encode_prediction(predict_tensor, task) (torch_tensor, Task) -> list() Take raw predictions self$network (predict_tensor) encode format can converted valid mlr3 predictions using mlr3::as_prediction_data(). method must take self$predict_type account. possible add parameters specifying param_set construction argument, currently possible remove existing parameters, .e. listed section Parameters. None parameters provided param_set can id starts \"loss.\", \"opt.\", \"cb.\"`, preserved dynamically constructed parameters optimizer, loss function, callbacks. perform additional input checks task, private .verify_train_task(task, param_vals) .verify_predict_task(task, param_vals) can overwritten. learners construction arguments change hash learner, required implement private $.additional_phash_input().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Base Class for Torch Learners — mlr_learners_torch","text":"mlr3::Learner -> LearnerTorch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Base Class for Torch Learners — mlr_learners_torch","text":"validate construct internal validation data. parameter can either NULL, ratio $(0, 1)$, \"test\", \"predefined\". loss (TorchLoss) torch loss. optimizer (TorchOptimizer) torch optimizer. callbacks (list() TorchCallbacks) List torch callbacks. ids set names. internal_valid_scores Retrieves internal validation scores named list(). Specify $validate field measures_valid parameter configure . Returns NULL learner trained yet. internal_tuned_values early stopping activate, returns named list early-stopped epochs, otherwise empty list returned. Returns NULL learner trained yet. marshaled (logical(1)) Whether learner marshaled. network (nn_module()) Shortcut learner$model$network. param_set (ParamSet) parameter set hash (character(1)) Hash (unique identifier) object. phash (character(1)) Hash (unique identifier) partial object, excluding components varied systematically tuning (parameter values).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Base Class for Torch Learners — mlr_learners_torch","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Base Class for Torch Learners — mlr_learners_torch","text":"LearnerTorch$new() LearnerTorch$format() LearnerTorch$print() LearnerTorch$marshal() LearnerTorch$unmarshal() LearnerTorch$dataset() LearnerTorch$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$new(   id,   task_type,   param_set,   properties,   man,   label,   feature_types,   optimizer = NULL,   loss = NULL,   packages = character(),   predict_types = NULL,   callbacks = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"id (character(1)) id new object. task_type (character(1)) task type. param_set (ParamSet alist()) Either parameter set, alist() containing different values self, e.g. alist(private$.param_set1, private$.param_set2), ParamSet collection created. properties (character()) properties object. See mlr_reflections$learner_properties available values. man (character(1)) String format [pkg]::[topic] pointing manual page object. referenced help package can opened via method $help(). label (character(1)) Label new instance. feature_types (character()) feature types. See mlr_reflections$task_feature_types available values, Additionally, \"lazy_tensor\" supported. optimizer (NULL TorchOptimizer) optimizer use training. Defaults adam. loss (NULL TorchLoss) loss use training. Defaults MSE regression cross entropy classification. packages (character()) R packages object depends . predict_types (character()) predict types. See mlr_reflections$learner_predict_types available values. regression, default \"response\". classification, defaults \"response\" \"prob\". deviate defaults, necessary overwrite private $.encode_prediction() method, see section Inheriting. callbacks (list() TorchCallbacks) callbacks use training. Defaults empty list(), .e. callbacks.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-format-","dir":"Reference","previous_headings":"","what":"Method format()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Helper print outputs.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$format(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"... (ignored).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-print-","dir":"Reference","previous_headings":"","what":"Method print()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Prints object.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$print(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-2","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"... () Currently unused.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-marshal-","dir":"Reference","previous_headings":"","what":"Method marshal()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Marshal learner.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-3","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$marshal(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-3","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"... () Additional parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Base Class for Torch Learners — mlr_learners_torch","text":"self","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-unmarshal-","dir":"Reference","previous_headings":"","what":"Method unmarshal()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Unmarshal learner.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-4","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$unmarshal(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-4","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"... () Additional parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"returns-1","dir":"Reference","previous_headings":"","what":"Returns","title":"Base Class for Torch Learners — mlr_learners_torch","text":"self","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-dataset-","dir":"Reference","previous_headings":"","what":"Method dataset()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Create dataset task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-5","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$dataset(task)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-5","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"task Task task","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"returns-2","dir":"Reference","previous_headings":"","what":"Returns","title":"Base Class for Torch Learners — mlr_learners_torch","text":"dataset","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-6","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-6","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":null,"dir":"Reference","previous_headings":"","what":"Image Learner — mlr_learners_torch_image","title":"Image Learner — mlr_learners_torch_image","text":"Base Class Image Learners. features assumed single lazy_tensor column RGB format.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Image Learner — mlr_learners_torch_image","text":"Parameters include inherited LearnerTorch param_set construction argument.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Image Learner — mlr_learners_torch_image","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> LearnerTorchImage","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Image Learner — mlr_learners_torch_image","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Image Learner — mlr_learners_torch_image","text":"LearnerTorchImage$new() LearnerTorchImage$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Image Learner — mlr_learners_torch_image","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Image Learner — mlr_learners_torch_image","text":"","code":"LearnerTorchImage$new(   id,   task_type,   param_set = ps(),   label,   optimizer = NULL,   loss = NULL,   callbacks = list(),   packages = \"torchvision\",   man,   properties = NULL,   predict_types = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Image Learner — mlr_learners_torch_image","text":"id (character(1)) id new object. task_type (character(1)) task type. param_set (ParamSet) parameter set. label (character(1)) Label new instance. optimizer (TorchOptimizer) torch optimizer. loss (TorchLoss) loss use training. callbacks (list() TorchCallbacks) callbacks used training. Must unique ids. executed order provided packages (character()) R packages object depends . man (character(1)) String format [pkg]::[topic] pointing manual page object. referenced help package can opened via method $help(). properties (character()) properties object. See mlr_reflections$learner_properties available values. predict_types (character()) predict types. See mlr_reflections$learner_predict_types available values.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Image Learner — mlr_learners_torch_image","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Image Learner — mlr_learners_torch_image","text":"","code":"LearnerTorchImage$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Image Learner — mlr_learners_torch_image","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Learner Torch Model — mlr_learners_torch_model","title":"Learner Torch Model — mlr_learners_torch_model","text":"Create torch learner instantiated nn_module(). classification, output network must scores (softmax).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Learner Torch Model — mlr_learners_torch_model","text":"See LearnerTorch","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Learner Torch Model — mlr_learners_torch_model","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> LearnerTorchModel","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Learner Torch Model — mlr_learners_torch_model","text":"network_stored (nn_module NULL) network trained. calling $train(), NULL. ingress_tokens (named list() TorchIngressToken NULL) ingress tokens. Must non-NULL calling $train().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Learner Torch Model — mlr_learners_torch_model","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Learner Torch Model — mlr_learners_torch_model","text":"LearnerTorchModel$new() LearnerTorchModel$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Learner Torch Model — mlr_learners_torch_model","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Learner Torch Model — mlr_learners_torch_model","text":"","code":"LearnerTorchModel$new(   network = NULL,   ingress_tokens = NULL,   task_type,   properties = NULL,   optimizer = NULL,   loss = NULL,   callbacks = list(),   packages = character(0),   feature_types = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Learner Torch Model — mlr_learners_torch_model","text":"network (nn_module) instantiated nn_module. cloned construction. classification, outputs must scores (softmax). ingress_tokens (list TorchIngressToken()) list ingress tokens defines dataloader defined. task_type (character(1)) task type. properties (NULL character()) properties learner. Defaults available properties given task type. optimizer (TorchOptimizer) torch optimizer. loss (TorchLoss) loss use training. callbacks (list() TorchCallbacks) callbacks used training. Must unique ids. executed order provided packages (character()) R packages object depends . feature_types (NULL character()) feature types. Defaults available feature types.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Learner Torch Model — mlr_learners_torch_model","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Learner Torch Model — mlr_learners_torch_model","text":"","code":"LearnerTorchModel$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Learner Torch Model — mlr_learners_torch_model","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Learner Torch Model — mlr_learners_torch_model","text":"","code":"# We show the learner using a classification task  # The iris task has 4 features and 3 classes network = nn_linear(4, 3) task = tsk(\"iris\")  # This defines the dataloader. # It loads all 4 features, which are also numeric. # The shape is (NA, 4) because the batch dimension is generally NA ingress_tokens = list(   input = TorchIngressToken(task$feature_names, batchgetter_num, c(NA, 4)) )  # Creating the learner and setting required parameters learner = lrn(\"classif.torch_model\",   network = network,   ingress_tokens = ingress_tokens,   batch_size = 16,   epochs = 1,   device = \"cpu\" )  # A simple train-predict ids = partition(task) learner$train(task, ids$train) learner$predict(task, ids$test) #> <PredictionClassif> for 50 observations: #>  row_ids     truth   response #>        2    setosa versicolor #>        3    setosa versicolor #>        9    setosa versicolor #>      ---       ---        --- #>      134 virginica versicolor #>      138 virginica versicolor #>      140 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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"Calls nn_adaptive_avg_pool1d() training.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAdaptiveAvgPool -> PipeOpTorchAdaptiveAvgPool1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"Calls nn_adaptive_avg_pool2d() training.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAdaptiveAvgPool -> PipeOpTorchAdaptiveAvgPool2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"Calls nn_adaptive_avg_pool3d() training.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAdaptiveAvgPool -> PipeOpTorchAdaptiveAvgPool3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"Calls nn_avg_pool1d() training.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAvgPool -> PipeOpTorchAvgPool1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"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 $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"kernel_size :: (integer()) size window. Can single number vector. stride :: integer() stride window. Can single number vector. Default: kernel_size. padding :: integer() Implicit zero paddings sides input. Can single number vector. Default: 0. ceil_mode :: integer() TRUE, use ceil instead floor compute output shape. Default: FALSE. count_include_pad :: logical(1) TRUE, include zero-padding averaging calculation. Default: TRUE. divisor_override :: logical(1) specified, used divisor, otherwise size pooling region used. Default: NULL. available dimension greater 1.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAvgPool -> PipeOpTorchAvgPool2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"mlr3pipelines::PipeOp$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":"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":"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 $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"kernel_size :: (integer()) size window. Can single number vector. stride :: integer() stride window. Can single number vector. Default: kernel_size. padding :: integer() Implicit zero paddings sides input. Can single number vector. Default: 0. ceil_mode :: integer() TRUE, use ceil instead floor compute output shape. Default: FALSE. count_include_pad :: logical(1) TRUE, include zero-padding averaging calculation. Default: TRUE. divisor_override :: logical(1) specified, used divisor, otherwise size pooling region used. Default: NULL. available dimension greater 1.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAvgPool -> PipeOpTorchAvgPool3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchBatchNorm -> PipeOpTorchBatchNorm1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"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 $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"eps :: numeric(1) value added denominator numerical stability. Default: 1e-5. momentum :: numeric(1) value used running_mean running_var computation. Can set NULL cumulative moving average (.e. simple average). Default: 0.1 affine :: logical(1) boolean value set TRUE, module learnable affine parameters. Default: TRUE track_running_stats :: logical(1) boolean value set TRUE, module tracks running mean variance, set FALSE, module track statistics always uses batch statistics training eval modes. Default: TRUE","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchBatchNorm -> PipeOpTorchBatchNorm2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"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 $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"eps :: numeric(1) value added denominator numerical stability. Default: 1e-5. momentum :: numeric(1) value used running_mean running_var computation. Can set NULL cumulative moving average (.e. simple average). Default: 0.1 affine :: logical(1) boolean value set TRUE, module learnable affine parameters. Default: TRUE track_running_stats :: logical(1) boolean value set TRUE, module tracks running mean variance, set FALSE, module track statistics always uses batch statistics training eval modes. Default: TRUE","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchBatchNorm -> PipeOpTorchBatchNorm3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"mlr3pipelines::PipeOp$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 concatenating (via %>>%).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"naming","dir":"Reference","previous_headings":"","what":"Naming","title":"Block Repetition — mlr_pipeops_nn_block","text":"generated module graph, IDs modules generated prefixing IDs n_blocks layers ID PipeOpTorchBlock postfixing __<layer>.","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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Block Repetition — mlr_pipeops_nn_block","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchBlock","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Block Repetition — mlr_pipeops_nn_block","text":"block (Graph) neural network segment repeated PipeOp.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Block Repetition — mlr_pipeops_nn_block","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"Calls torch::nn_celu() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchCELU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"Calls torch::nn_conv1d() trained. paramter in_channels inferred second dimension input tensor.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConv -> PipeOpTorchConv1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"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 $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"out_channels :: integer(1) Number channels produced convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. default 1. padding :: integer() ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘TRUE’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConv -> PipeOpTorchConv2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"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 $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"out_channels :: integer(1) Number channels produced convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. default 1. padding :: integer() ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘TRUE’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConv -> PipeOpTorchConv3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"mlr3pipelines::PipeOp$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":"Applies 1D transposed convolution operator input signal composed several input planes, sometimes also called \"deconvolution\".","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"Calls nn_conv_transpose1d. parameter in_channels inferred second dimension input tensor.","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":"state","dir":"Reference","previous_headings":"","what":"State","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConvTranspose -> PipeOpTorchConvTranspose1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"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 $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"out_channels :: integer(1) Number output channels produce convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. Default: 1. padding ::  integer()` ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. output_padding ::integer() Additional size added one side output shape. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘True’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConvTranspose -> PipeOpTorchConvTranspose2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"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 $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"out_channels :: integer(1) Number output channels produce convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. Default: 1. padding ::  integer()` ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. output_padding ::integer() Additional size added one side output shape. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘True’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConvTranspose -> PipeOpTorchConvTranspose3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Dropout — mlr_pipeops_nn_dropout","text":"Calls torch::nn_dropout() trained.","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 :: logical(1) set TRUE, operation -place. Default: FALSE.","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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Dropout — mlr_pipeops_nn_dropout","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchDropout","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Dropout — mlr_pipeops_nn_dropout","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"Calls torch::nn_elu() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchELU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"Calls torch::nn_flatten() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchFlatten","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"Calls torch::nn_gelu() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchGELU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"Calls torch::nn_glu() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchGLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"Calls torch::nn_hardshrink() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchHardShrink","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"Calls torch::nn_hardsigmoid() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchHardSigmoid","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"Calls torch::nn_hardtanh() trained.","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":"state","dir":"Reference","previous_headings":"","what":"State","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"state value calculated public method $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchHardTanh","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"mlr3pipelines::PipeOp$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.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Output Head — mlr_pipeops_nn_head","text":"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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Output Head — mlr_pipeops_nn_head","text":"Calls torch::nn_linear() input output features inferred input shape / task.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Output Head — mlr_pipeops_nn_head","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchHead","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Output Head — mlr_pipeops_nn_head","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"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":"state","dir":"Reference","previous_headings":"","what":"State","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchLayerNorm","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"Calls torch::nn_leaky_relu() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchLeakyReLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Linear Layer — mlr_pipeops_nn_linear","text":"Calls torch::nn_linear() trained parameter in_features inferred second last dimension input tensor.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Linear Layer — mlr_pipeops_nn_linear","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchLinear","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Linear Layer — mlr_pipeops_nn_linear","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"Calls torch::nn_log_sigmoid() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchLogSigmoid","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"Calls torch::nn_max_pool1d() training.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMaxPool -> PipeOpTorchMaxPool1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"state","dir":"Reference","previous_headings":"","what":"State","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"kernel_size :: integer() size window. Can single number vector. stride :: (integer(1)) stride window. Can single number vector. Default: kernel_size padding :: integer() Implicit zero paddings sides input. Can single number tuple (padW,). Default: 0 dilation :: integer() Controls spacing kernel points; also known à trous algorithm. Default: 1 ceil_mode :: logical(1) True, use ceil instead floor compute output shape. Default: FALSE","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"return_indices FALSE construction, one input channel 'input' one output channel 'output'. return_indices TRUE, two output channels 'output' 'indices'. explanation see PipeOpTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMaxPool -> PipeOpTorchMaxPool2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"state","dir":"Reference","previous_headings":"","what":"State","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"kernel_size :: integer() size window. Can single number vector. stride :: (integer(1)) stride window. Can single number vector. Default: kernel_size padding :: integer() Implicit zero paddings sides input. Can single number tuple (padW,). Default: 0 dilation :: integer() Controls spacing kernel points; also known à trous algorithm. Default: 1 ceil_mode :: logical(1) True, use ceil instead floor compute output shape. Default: FALSE","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"return_indices FALSE construction, one input channel 'input' one output channel 'output'. return_indices TRUE, two output channels 'output' 'indices'. explanation see PipeOpTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMaxPool -> PipeOpTorchMaxPool3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"mlr3pipelines::PipeOp$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":"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":"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":"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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"Calls nn_merge_cat() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMerge -> PipeOpTorchMergeCat","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"Calls nn_merge_prod() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMerge -> PipeOpTorchMergeProd","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"Calls nn_merge_sum() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMerge -> PipeOpTorchMergeSum","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"Calls torch::nn_prelu() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchPReLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"Calls torch::nn_relu() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchReLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"Calls torch::nn_relu6() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchReLU6","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"Calls nn_reshape() trained. internally calls torch::torch_reshape() given shape.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchReshape","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"Calls torch::nn_rrelu() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchRReLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"Calls torch::nn_selu() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSELU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"Calls torch::nn_sigmoid() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSigmoid","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Softmax — mlr_pipeops_nn_softmax","text":"Calls torch::nn_softmax() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Softmax — mlr_pipeops_nn_softmax","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSoftmax","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Softmax — mlr_pipeops_nn_softmax","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"Calls torch::nn_softplus() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSoftPlus","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"Calls torch::nn_softshrink() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSoftShrink","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"Calls torch::nn_softsign() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSoftSign","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"Calls nn_squeeze() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSqueeze","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"Calls torch::nn_tanh() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchTanh","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"Calls torch::nn_tanhshrink() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchTanhShrink","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"Calls torch::nn_threshold() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchThreshold","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"mlr3pipelines::PipeOp$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":"Unsqueezes tensor calling torch::torch_unsqueeze() given dimension dim.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"Calls nn_unsqueeze() trained. internally calls torch::torch_unsqueeze().","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchUnsqueeze","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"mlr3pipelines::PipeOp$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 #>  100.8678 #>  100.6844 #>   99.8373 #>   98.2156 #>   98.9629 #>  100.8301 #>  100.6073 #>   99.8778 #>  100.9331 #>   99.0387 #> [ CPUFloatType{10,1} ] #>  #> $x2 #> torch_tensor #>  100.2551 #>   99.4546 #>  100.9304 #>   99.4623 #>   99.5476 #>   99.5607 #>   99.3838 #>  100.4416 #>  100.4826 #>  100.5421 #> [ CPUFloatType{10,1} ] #>   taskout_predict_noaug = po_simple$predict(list(taskin))[[1L]] materialize(taskout_predict_noaug$data(cols = c(\"x1\", \"x2\")), rbind = TRUE) #> $x1 #> torch_tensor #>  100.8678 #>  100.6844 #>   99.8373 #>   98.2156 #>   98.9629 #>  100.8301 #>  100.6073 #>   99.8778 #>  100.9331 #>   99.0387 #> [ CPUFloatType{10,1} ] #>  #> $x2 #> torch_tensor #>  100.2551 #>   99.4546 #>  100.9304 #>   99.4623 #>   99.5476 #>   99.5607 #>   99.3838 #>  100.4416 #>  100.4826 #>  100.5421 #> [ CPUFloatType{10,1} ] #>   po_simple$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 #>  100.8678 #>  100.6844 #>   99.8373 #>   98.2156 #>   98.9629 #>  100.8301 #>  100.6073 #>   99.8778 #>  100.9331 #>   99.0387 #> [ CPUFloatType{10,1} ] #>  #> $x2 #> torch_tensor #>  100.2551 #>   99.4546 #>  100.9304 #>   99.4623 #>   99.5476 #>   99.5607 #>   99.3838 #>  100.4416 #>  100.4826 #>  100.5421 #> [ CPUFloatType{10,1} ] #>   # Creating a more complex preprocessing PipeOp PipeOpPreprocTorchPoly = R6::R6Class(\"PipeOpPreprocTorchPoly\",  inherit = PipeOpTaskPreprocTorch,  public = list(    initialize = function(id = \"preproc_poly\", param_vals = list()) {      param_set = paradox::ps(        n_degree = paradox::p_int(lower = 1L, tags = c(\"train\", \"required\"))      )      param_set$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) #> $x3 #> torch_tensor #>     1     1     1 #>     2     4     8 #>     3     9    27 #>     4    16    64 #>     5    25   125 #>     6    36   216 #>     7    49   343 #>     8    64   512 #>     9    81   729 #>    10   100  1000 #> [ CPUFloatType{10,3} ] #>"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":null,"dir":"Reference","previous_headings":"","what":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"PipeOpTorch base class PipeOps represent neural network layers Graph. training, generates PipeOpModule wraps nn_module attaches architecture, also represented Graph consisting mostly PipeOpModules PipeOpNOPs. former Graph operates ModelDescriptors, latter operates tensors. relationship PipeOpTorch PipeOpModule similar relationshop nn_module_generator (like nn_linear) nn_module (like output nn_linear(...)). crucial difference PipeOpTorch infers auxiliary parameters (like in_features nn_linear) automatically intermediate tensor shapes communicated ModelDescriptor. prediction, PipeOpTorch takes Task channel outputs new Task resulting feature union channel. one input output channel, task simply piped .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"ParamSet specified child class inheriting PipeOpTorch. Usually parameters arguments wrapped nn_module minus auxiliary parameter can automatically inferred shapes input tensors.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"inheriting","dir":"Reference","previous_headings":"","what":"Inheriting","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"inheriting class, one overload either private$.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":"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":"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":"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":"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":"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":"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":"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() | NULL | \"infer\" shape tensor, first dimension (batch) must NA. specified, lazy tensor input column needs known shape. set \"infer\", shape inferred example batch.","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$data(1, \"x\"), cache = NULL) x_batch #> torch_tensor #>  5.1000  3.5000  1.4000  0.2000 #> [ CPUFloatType{1,4} ]  # Now we try a lazy tensor with unknown shape, i.e. the shapes between the rows can differ  ds = dataset(   initialize = function() self$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\"),   cache = NULL )  x_batch2 #> torch_tensor #> (1,1,.,.) =  #>   0.9970 -0.9383 -0.0510  1.4438  0.5061 -0.8022 #>  -0.4175  0.2638 -0.7721 -1.3055  1.0725  0.1492 #>   0.1705  1.1516 -1.1165  0.6934 -0.2440 -0.3882 #>   0.4484 -0.4592 -0.2580 -0.4509  0.1134 -1.1358 #>   0.3275 -1.4557  0.5706  1.2162  0.6805 -1.0252 #>  -0.4683  0.2398 -0.4668 -0.1763  0.1691 -0.2560 #>  #> (2,1,.,.) =  #>   0.1049  0.7440  0.4347  0.1592 -0.2501  0.0542 #>  -0.5176  0.1912 -0.5836  0.2313  0.0562 -0.6430 #>  -0.1808 -0.3858 -0.4339  0.1317 -0.2483 -0.5749 #>  -0.2762  0.4425  0.1593  0.5774  0.0714 -0.5108 #>  -0.4984 -1.4006  1.0352  0.2719  0.3794  0.1281 #>  -0.1655  0.0319 -0.6908 -0.1547  0.5281 -0.6154 #>  #> (1,2,.,.) =  #>   1.0034 -0.0084 -0.5612  0.6336  0.6045 -0.3766 #>   0.2037 -0.6614  0.0176  0.9215  0.1562  0.0783 #>  -0.6090  0.0246  0.6608  1.2319 -0.4807 -0.7408 #>  -0.0390  0.4402 -1.1336 -0.3919 -0.2745  0.7000 #>   1.0377  1.2182  0.1394  0.0664 -0.6851  0.3546 #>   0.0136 -0.7515  1.5054 -0.3123  0.3507 -0.9329 #>  #> (2,2,.,.) =  #>   0.3226 -0.5212  0.5643 -0.6904 -0.8490 -1.1094 #>   0.6671 -0.4693  0.3308  0.2793  0.1941 -0.9796 #>   0.0677 -0.4034 -0.8510 -0.2002  0.4505 -1.5667 #>  -0.0590 -0.8131 -0.9660 -0.1422  0.3739  0.8411 #>  -0.1460 -0.3893  0.4009 -0.3064 -0.5538 -0.2872 #> ... [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$loss #> <TorchLoss:cross_entropy> Cross Entropy #> * Generator: nn_cross_entropy_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: classif"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOp Torch Model — mlr_pipeops_torch_model","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"Builds Torch Learner ModelDescriptor trains given parameter specification. task type must specified construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"General: parameters optimizer, loss callbacks, prefixed \"opt.\", \"loss.\" \"cb.<callback id>.\" respectively, well : epochs :: integer(1) number epochs. device :: character(1) device. One \"auto\", \"cpu\", \"cuda\" values defined mlr_reflections$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. tensor_dataset :: logical(1) | \"device\" Whether load batches beginning training stack . initialized FALSE. set \"device\", device tensors set value device, can avoid unnecessary moving tensors devices. dataset fits memory make loading batches faster. Note set datasets contain lazy_tensors random data augmentation, augmentation applied beginning training. 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. set, epoch improvement observed, can accessed via $internal_tuned_values field learner. 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. initialized TRUE, differs default (FALSE). 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":"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":"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]]$weight_decay #> [1] 0 #>  #> $optimizer$param_groups[[1]]$betas #> [1] 0.900 0.999 #>  #> $optimizer$param_groups[[1]]$eps #> [1] 1e-08 #>  #> $optimizer$param_groups[[1]]$amsgrad #> [1] FALSE #>  #>  #>  #> $optimizer$state #> $optimizer$state$`1` #> $optimizer$state$`1`$exp_avg #> torch_tensor #> -0.0305  0.0118 -0.3143 -0.2423 #>  0.5306  0.1713  0.9039  0.5068 #> -0.5001 -0.1830 -0.5896 -0.2646 #> [ CPUFloatType{3,4} ] #>  #> $optimizer$state$`1`$exp_avg_sq #> torch_tensor #> 0.01 * #>  0.0048  0.0007  0.4051  0.2395 #>   1.1423  0.1193  3.3175  1.0430 #>   1.0133  0.1359  1.4087  0.2836 #> [ CPUFloatType{3,4} ] #>  #> $optimizer$state$`1`$max_exp_avg_sq #> torch_tensor #> [ CPUFloatType{0} ] #>  #> $optimizer$state$`1`$step #> torch_tensor #>  3 #> [ CPULongType{1} ] #>  #>  #> $optimizer$state$`2` #> $optimizer$state$`2`$exp_avg #> torch_tensor #> -0.0674 #>  0.1564 #> -0.0889 #> [ CPUFloatType{3} ] #>  #> $optimizer$state$`2`$exp_avg_sq #> torch_tensor #> 0.0001 * #>  1.8577 #>  9.9196 #>  3.2051 #> [ CPUFloatType{3} ] #>  #> $optimizer$state$`2`$max_exp_avg_sq #> torch_tensor #> [ CPUFloatType{0} ] #>  #> $optimizer$state$`2`$step #> torch_tensor #>  3 #> [ CPULongType{1} ] #>  #>  #>  #>  #> $epochs #> [1] 1 #>  #> $callbacks #> named list() #>  #> $seed #> [1] 1338150431 #>  #> $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] 0.057 #>  #> $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\" #>  #> $param_vals$jit_trace #> [1] FALSE #>  #> $param_vals$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 #>  #> $param_vals$shuffle #> [1] TRUE #>  #> $param_vals$tensor_dataset #> [1] FALSE #>  #>  #> $task_hash #> [1] \"60543108f51f96bb\" #>  #> $feature_names #> [1] \"Petal.Length\" \"Petal.Width\"  \"Sepal.Length\" \"Sepal.Width\"  #>  #> $validate #> NULL #>  #> $mlr3_version #> [1] ‘0.22.1’ #>  #> $internal_tuned_values #> named list() #>  #> $data_prototype #> Empty data.table (0 rows and 5 cols): Species,Petal.Length,Petal.Width,Sepal.Length,Sepal.Width #>  #> $task_prototype #> Empty data.table (0 rows and 5 cols): Species,Petal.Length,Petal.Width,Sepal.Length,Sepal.Width #>  #> $train_task #> <TaskClassif:iris> (150 x 5): Iris Flowers #> * Target: Species #> * Properties: multiclass #> * Features (4): #>   - dbl (4): Petal.Length, Petal.Width, Sepal.Length, Sepal.Width #>  #> attr(,\"class\") #> [1] \"learner_state\" \"list\""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch Regression Model — mlr_pipeops_torch_model_regr","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"Builds torch regression model trains .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"See LearnerTorch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"one input channel \"input\" takes ModelDescriptor traing Task specified task_type prediction. output NULL training Prediction given task_type prediction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"trained LearnerTorchModel.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"LearnerTorchModel created calling model_descriptor_to_learner() provided ModelDescriptor received input channel. parameters set according parameters specified PipeOpTorchModel '$train() method called [Task][mlr3::Task] stored [ModelDescriptor`].","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"mlr3pipelines::PipeOp -> mlr3pipelines::PipeOpLearner -> mlr3torch::PipeOpTorchModel -> PipeOpTorchModelRegr","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"PipeOpTorchModelRegr$new() PipeOpTorchModelRegr$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"","code":"PipeOpTorchModelRegr$new(id = \"torch_model_regr\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"","code":"PipeOpTorchModelRegr$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"","code":"# simple linear regression  # build the model descriptor md = as_graph(po(\"torch_ingress_num\") %>>%   po(\"nn_head\") %>>%   po(\"torch_loss\", \"mse\") %>>%   po(\"torch_optimizer\", \"adam\"))$train(tsk(\"mtcars\"))[[1L]]  print(md) #> <ModelDescriptor: 2 ops> #> * Ingress:  torch_ingress_num.input: [(NA,10)] #> * Task:  mtcars [regr] #> * Callbacks:  N/A #> * Optimizer:  Adaptive Moment Estimation #> * Loss:  Mean Squared Error #> * pointer:  nn_head.output [(NA,1)]  # build the learner from the model descriptor and train it po_model = po(\"torch_model_regr\", batch_size = 20, epochs = 1) po_model$train(list(md)) #> $output #> NULL #>  po_model$state #> $model #> $network #> An `nn_module` containing 11 parameters. #>  #> ── Modules ───────────────────────────────────────────────────────────────────── #> • module_list: <nn_module_list> #11 parameters #>  #> $internal_valid_scores #> NULL #>  #> $loss_fn #> list() #>  #> $optimizer #> $optimizer$param_groups #> $optimizer$param_groups[[1]] #> $optimizer$param_groups[[1]]$params #> [1] 1 2 #>  #> $optimizer$param_groups[[1]]$lr #> [1] 0.001 #>  #> $optimizer$param_groups[[1]]$weight_decay #> [1] 0 #>  #> $optimizer$param_groups[[1]]$betas #> [1] 0.900 0.999 #>  #> $optimizer$param_groups[[1]]$eps #> [1] 1e-08 #>  #> $optimizer$param_groups[[1]]$amsgrad #> [1] FALSE #>  #>  #>  #> $optimizer$state #> $optimizer$state$`1` #> $optimizer$state$`1`$exp_avg #> torch_tensor #> -2.4873 -12.0392 -29.9780 -1048.1003 -19.9276 -19.6820 -628.5255 -97.9720 -3.1496 -15.6931 #> [ CPUFloatType{1,10} ] #>  #> $optimizer$state$`1`$exp_avg_sq #> torch_tensor #> Columns 1 to 6 3.6781e-01  8.1886e+00  5.0809e+01  6.1723e+04  2.2420e+01  2.2421e+01 #>  #> Columns 7 to 10 2.2605e+04  5.4466e+02  5.5291e-01  1.3928e+01 #> [ CPUFloatType{1,10} ] #>  #> $optimizer$state$`1`$max_exp_avg_sq #> torch_tensor #> [ CPUFloatType{0} ] #>  #> $optimizer$state$`1`$step #> torch_tensor #>  2 #> [ CPULongType{1} ] #>  #>  #> $optimizer$state$`2` #> $optimizer$state$`2`$exp_avg #> torch_tensor #> -5.3484 #> [ CPUFloatType{1} ] #>  #> $optimizer$state$`2`$exp_avg_sq #> torch_tensor #>  1.6237 #> [ CPUFloatType{1} ] #>  #> $optimizer$state$`2`$max_exp_avg_sq #> torch_tensor #> [ CPUFloatType{0} ] #>  #> $optimizer$state$`2`$step #> torch_tensor #>  2 #> [ CPULongType{1} ] #>  #>  #>  #>  #> $epochs #> [1] 1 #>  #> $callbacks #> named list() #>  #> $seed #> [1] 1879465332 #>  #> $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] 0.04 #>  #> $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\" #>  #> $param_vals$jit_trace #> [1] FALSE #>  #> $param_vals$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 #>  #> $param_vals$shuffle #> [1] TRUE #>  #> $param_vals$tensor_dataset #> [1] FALSE #>  #>  #> $task_hash #> [1] \"907e8c646e24bf41\" #>  #> $feature_names #>  [1] \"am\"   \"carb\" \"cyl\"  \"disp\" \"drat\" \"gear\" \"hp\"   \"qsec\" \"vs\"   \"wt\"   #>  #> $validate #> NULL #>  #> $mlr3_version #> [1] ‘0.22.1’ #>  #> $internal_tuned_values #> named list() #>  #> $data_prototype #> Empty data.table (0 rows and 11 cols): mpg,am,carb,cyl,disp,drat... #>  #> $task_prototype #> Empty data.table (0 rows and 11 cols): mpg,am,carb,cyl,disp,drat... #>  #> $train_task #> <TaskRegr:mtcars> (32 x 11): Motor Trends #> * Target: mpg #> * Properties: - #> * Features (10): #>   - dbl (10): am, carb, cyl, disp, drat, gear, hp, qsec, vs, wt #>  #> attr(,\"class\") #> [1] \"learner_state\" \"list\""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer Configuration — mlr_pipeops_torch_optimizer","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"Configures optimizer deep learning model.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"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":"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":"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_ignite_sgd #> * Parameters: lr=0.01 #> * Packages: torch,mlr3torch"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_center_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"Center Crop Augmentation — mlr_pipeops_augment_center_crop","title":"Center Crop Augmentation — mlr_pipeops_augment_center_crop","text":"Calls torchvision::transform_center_crop, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_center_crop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Center Crop Augmentation — mlr_pipeops_augment_center_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_color_jitter.html","id":null,"dir":"Reference","previous_headings":"","what":"Color Jitter Augmentation — mlr_pipeops_augment_color_jitter","title":"Color Jitter Augmentation — mlr_pipeops_augment_color_jitter","text":"Calls torchvision::transform_color_jitter, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_color_jitter.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Color Jitter Augmentation — mlr_pipeops_augment_color_jitter","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"Crop Augmentation — mlr_pipeops_augment_crop","title":"Crop Augmentation — mlr_pipeops_augment_crop","text":"Calls torchvision::transform_crop, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_crop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Crop Augmentation — mlr_pipeops_augment_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_hflip.html","id":null,"dir":"Reference","previous_headings":"","what":"Horizontal Flip Augmentation — mlr_pipeops_augment_hflip","title":"Horizontal Flip Augmentation — mlr_pipeops_augment_hflip","text":"Calls torchvision::transform_hflip, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_hflip.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Horizontal Flip Augmentation — mlr_pipeops_augment_hflip","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_affine.html","id":null,"dir":"Reference","previous_headings":"","what":"Random Affine Augmentation — mlr_pipeops_augment_random_affine","title":"Random Affine Augmentation — mlr_pipeops_augment_random_affine","text":"Calls torchvision::transform_random_affine, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_affine.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Random Affine Augmentation — mlr_pipeops_augment_random_affine","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_choice.html","id":null,"dir":"Reference","previous_headings":"","what":"Random Choice Augmentation — mlr_pipeops_augment_random_choice","title":"Random Choice Augmentation — mlr_pipeops_augment_random_choice","text":"Calls torchvision::transform_random_choice, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_choice.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Random Choice Augmentation — mlr_pipeops_augment_random_choice","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"Random Crop Augmentation — mlr_pipeops_augment_random_crop","title":"Random Crop Augmentation — mlr_pipeops_augment_random_crop","text":"Calls torchvision::transform_random_crop, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_crop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Random Crop Augmentation — mlr_pipeops_augment_random_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_horizontal_flip.html","id":null,"dir":"Reference","previous_headings":"","what":"Random Horizontal Flip Augmentation — mlr_pipeops_augment_random_horizontal_flip","title":"Random Horizontal Flip Augmentation — mlr_pipeops_augment_random_horizontal_flip","text":"Calls torchvision::transform_random_horizontal_flip, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_horizontal_flip.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Random Horizontal Flip Augmentation — mlr_pipeops_augment_random_horizontal_flip","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_order.html","id":null,"dir":"Reference","previous_headings":"","what":"Random Order Augmentation — mlr_pipeops_augment_random_order","title":"Random Order Augmentation — mlr_pipeops_augment_random_order","text":"Calls torchvision::transform_random_order, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_order.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Random Order Augmentation — mlr_pipeops_augment_random_order","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_resized_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"Random Resized Crop Augmentation — mlr_pipeops_augment_random_resized_crop","title":"Random Resized Crop Augmentation — mlr_pipeops_augment_random_resized_crop","text":"Calls torchvision::transform_random_resized_crop, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_resized_crop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Random Resized Crop Augmentation — mlr_pipeops_augment_random_resized_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_vertical_flip.html","id":null,"dir":"Reference","previous_headings":"","what":"Random Vertical Flip Augmentation — mlr_pipeops_augment_random_vertical_flip","title":"Random Vertical Flip Augmentation — mlr_pipeops_augment_random_vertical_flip","text":"Calls torchvision::transform_random_vertical_flip, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_vertical_flip.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Random Vertical Flip Augmentation — mlr_pipeops_augment_random_vertical_flip","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_resized_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"Resized Crop Augmentation — mlr_pipeops_augment_resized_crop","title":"Resized Crop Augmentation — mlr_pipeops_augment_resized_crop","text":"Calls torchvision::transform_resized_crop, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_resized_crop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Resized Crop Augmentation — mlr_pipeops_augment_resized_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_rotate.html","id":null,"dir":"Reference","previous_headings":"","what":"Rotate Augmentation — mlr_pipeops_augment_rotate","title":"Rotate Augmentation — mlr_pipeops_augment_rotate","text":"Calls torchvision::transform_rotate, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_rotate.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Rotate Augmentation — mlr_pipeops_augment_rotate","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_vflip.html","id":null,"dir":"Reference","previous_headings":"","what":"Vertical Flip Augmentation — mlr_pipeops_augment_vflip","title":"Vertical Flip Augmentation — mlr_pipeops_augment_vflip","text":"Calls torchvision::transform_vflip, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_vflip.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Vertical Flip Augmentation — mlr_pipeops_augment_vflip","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_adjust_brightness.html","id":null,"dir":"Reference","previous_headings":"","what":"Adjust Brightness Transformation — mlr_pipeops_trafo_adjust_brightness","title":"Adjust Brightness Transformation — mlr_pipeops_trafo_adjust_brightness","text":"Calls torchvision::transform_adjust_brightness, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_adjust_brightness.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Adjust Brightness Transformation — mlr_pipeops_trafo_adjust_brightness","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_adjust_gamma.html","id":null,"dir":"Reference","previous_headings":"","what":"Adjust Gamma Transformation — mlr_pipeops_trafo_adjust_gamma","title":"Adjust Gamma Transformation — mlr_pipeops_trafo_adjust_gamma","text":"Calls torchvision::transform_adjust_gamma, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_adjust_gamma.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Adjust Gamma Transformation — mlr_pipeops_trafo_adjust_gamma","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_adjust_hue.html","id":null,"dir":"Reference","previous_headings":"","what":"Adjust Hue Transformation — mlr_pipeops_trafo_adjust_hue","title":"Adjust Hue Transformation — mlr_pipeops_trafo_adjust_hue","text":"Calls torchvision::transform_adjust_hue, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_adjust_hue.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Adjust Hue Transformation — mlr_pipeops_trafo_adjust_hue","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_adjust_saturation.html","id":null,"dir":"Reference","previous_headings":"","what":"Adjust Saturation Transformation — mlr_pipeops_trafo_adjust_saturation","title":"Adjust Saturation Transformation — mlr_pipeops_trafo_adjust_saturation","text":"Calls torchvision::transform_adjust_saturation, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_adjust_saturation.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Adjust Saturation Transformation — mlr_pipeops_trafo_adjust_saturation","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_grayscale.html","id":null,"dir":"Reference","previous_headings":"","what":"Grayscale Transformation — mlr_pipeops_trafo_grayscale","title":"Grayscale Transformation — mlr_pipeops_trafo_grayscale","text":"Calls torchvision::transform_grayscale, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_grayscale.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Grayscale Transformation — mlr_pipeops_trafo_grayscale","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_normalize.html","id":null,"dir":"Reference","previous_headings":"","what":"Normalization Transformation — mlr_pipeops_trafo_normalize","title":"Normalization Transformation — mlr_pipeops_trafo_normalize","text":"Calls torchvision::transform_normalize, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_normalize.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Normalization Transformation — mlr_pipeops_trafo_normalize","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_pad.html","id":null,"dir":"Reference","previous_headings":"","what":"Padding Transformation — mlr_pipeops_trafo_pad","title":"Padding Transformation — mlr_pipeops_trafo_pad","text":"Calls torchvision::transform_pad, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_pad.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Padding Transformation — mlr_pipeops_trafo_pad","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_resize.html","id":null,"dir":"Reference","previous_headings":"","what":"Resizing Transformation — mlr_pipeops_trafo_resize","title":"Resizing Transformation — 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/mlr_pipeopstrafo_resize.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Resizing Transformation — mlr_pipeops_trafo_resize","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_rgb_to_grayscale.html","id":null,"dir":"Reference","previous_headings":"","what":"RGB to Grayscale Transformation — mlr_pipeops_trafo_rgb_to_grayscale","title":"RGB to Grayscale Transformation — mlr_pipeops_trafo_rgb_to_grayscale","text":"Calls torchvision::transform_rgb_to_grayscale, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_rgb_to_grayscale.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"RGB to Grayscale Transformation — mlr_pipeops_trafo_rgb_to_grayscale","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_cifar.html","id":null,"dir":"Reference","previous_headings":"","what":"CIFAR Classification Tasks — mlr_tasks_cifar","title":"CIFAR Classification Tasks — mlr_tasks_cifar","text":"CIFAR-10 CIFAR-100 datasets. subset 80 million tiny images dataset noisy labels supplied student labelers, asked filter incorrectly labeled images. images datatype torch_long(). CIFAR-10 contains 10 classes. CIFAR-100 contains 100 classes, may partitioned 20 superclasses 5 classes . CIFAR-10 CIFAR-100 classes mutually exclusive. See Chapter 3.1 technical report details. data obtained torchvision::cifar10_dataset() (torchvision::cifar100_dataset()).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_cifar.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"CIFAR Classification Tasks — mlr_tasks_cifar","text":"R6::R6Class inheriting mlr3::TaskClassif.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_cifar.html","id":"construction","dir":"Reference","previous_headings":"","what":"Construction","title":"CIFAR Classification Tasks — mlr_tasks_cifar","text":"","code":"tsk(\"cifar10\") tsk(\"cifar100\")"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_cifar.html","id":"download","dir":"Reference","previous_headings":"","what":"Download","title":"CIFAR Classification Tasks — mlr_tasks_cifar","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_cifar.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"CIFAR Classification Tasks — mlr_tasks_cifar","text":"Task type: “classif” Properties: “multiclass” Missings: Target: “class” Features: “image” Data Dimension: 60000x4","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_cifar.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"CIFAR Classification Tasks — mlr_tasks_cifar","text":"Krizhevsky, Alex (2009). “Learning Multiple Layers Features Tiny Images.” Master's thesis, Department Computer Science, University Toronto.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_cifar.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"CIFAR Classification Tasks — mlr_tasks_cifar","text":"","code":"task_cifar10 = tsk(\"cifar10\") task_cifar100 = tsk(\"cifar100\") print(task_cifar10) #> <TaskClassif:cifar10> (60000 x 2): CIFAR-10 Classification #> * Target: class #> * Properties: multiclass #> * Features (1): #>   - lt (1): image print(task_cifar100) #> <TaskClassif:cifar100> (60000 x 2): CIFAR-100 Classification #> * Target: class #> * Properties: multiclass #> * Features (1): #>   - lt (1): image"},{"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_melanoma.html","id":null,"dir":"Reference","previous_headings":"","what":"Melanoma Image classification — mlr_tasks_melanoma","title":"Melanoma Image classification — mlr_tasks_melanoma","text":"Classification melanoma tumor images. data preprocessed version 2020 SIIM-ISIC challenge images reshaped size $(3, 128, 128)$. default training rows active task, test data (targets) also included. Whether observation part train test set indicated column \"test\". labels test rows, default, observations inactive, means task uses 32701 43683 observations defined underlying data backend. data backend also contains detailed diagnosis specific type tumor. Columns: outcome (factor): target variable. Whether tumor benign malignant (positive class) anatom_site_general_challenge (factor): location tumor patient's body sex (factor): sex patient age_approx (int): approximate age patient time imaging image (lazy_tensor): image (shape $(3, 128, 128)$) tumor. ee split (character): Whether observation os part train test set.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_melanoma.html","id":"source","dir":"Reference","previous_headings":"","what":"Source","title":"Melanoma Image classification — mlr_tasks_melanoma","text":"https://huggingface.co/datasets/carsonzhang/ISIC_2020_small","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_melanoma.html","id":"construction","dir":"Reference","previous_headings":"","what":"Construction","title":"Melanoma Image classification — mlr_tasks_melanoma","text":"","code":"tsk(\"melanoma\")"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_melanoma.html","id":"download","dir":"Reference","previous_headings":"","what":"Download","title":"Melanoma Image classification — mlr_tasks_melanoma","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_melanoma.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"Melanoma Image classification — mlr_tasks_melanoma","text":"Task type: “classif” Properties: “twoclass”, “groups” Missings: Target: “outcome” Features: “sex”, “anatom_site_general_challenge”, “age_approx”, “image” Data Dimension: 43683x11","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_melanoma.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Melanoma Image classification — mlr_tasks_melanoma","text":"Rotemberg, V., Kurtansky, N., Betz-Stablein, B., Caffery, L., Chousakos, E., Codella, N., Combalia, M., Dusza, S., Guitera, P., Gutman, D., Halpern, ., Helba, B., Kittler, H., Kose, K., Langer, S., Lioprys, K., Malvehy, J., Musthaq, S., Nanda, J., Reiter, O., Shih, G., Stratigos, ., Tschandl, P., Weber, J., Soyer, P. (2021). “patient-centric dataset images metadata identifying melanomas using clinical context.” Scientific Data, 8, 34. doi:10.1038/s41597-021-00815-z .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_melanoma.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Melanoma Image classification — mlr_tasks_melanoma","text":"","code":"task = tsk(\"melanoma\") task #> <TaskClassif:melanoma> (32701 x 5): Melanoma Classification #> * Target: outcome #> * Properties: twoclass, groups #> * Features (4): #>   - fct (2): anatom_site_general_challenge, sex #>   - int (1): age_approx #>   - lt (1): image #> * Groups: patient_id"},{"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 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 #> -0.3465 #> -0.3795 #>  0.0351 #> -0.3785 #> -0.2675 #>  0.2843 #> -0.6249 #> -0.4385 #> -0.1047 #> -0.2959 #> -0.1331 #> -0.0724 #> -0.0725 #>  0.0181 #> -0.2933 #> -0.2035 #> [ 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_ignite_adam #> * Parameters: lr=0.1 #> * Packages: torch,mlr3torch # get the dictionary t_opt() #> <DictionaryMlr3torchOptimizers> with 5 stored values #> Keys: adagrad, adam, adamw, rmsprop, sgd t_opts(c(\"adam\", \"sgd\")) #> $adam #> <TorchOptimizer:adam> Adaptive Moment Estimation #> * Generator: optim_ignite_adam #> * Parameters: list() #> * Packages: torch,mlr3torch #>  #> $sgd #> <TorchOptimizer:sgd> Stochastic Gradient Descent #> * Generator: optim_ignite_sgd #> * Parameters: list() #> * Packages: torch,mlr3torch #>  # get the dictionary t_opts() #> <DictionaryMlr3torchOptimizers> with 5 stored values #> Keys: adagrad, adam, adamw, rmsprop, 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)"},{"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 $x value batch corresponding names. target_batchgetter (function(data, device)) function taking arguments data, data.table containing target variable, device. must return target torch tensor selected device.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/task_dataset.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a Dataset from a Task — task_dataset","text":"torch::dataset","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/task_dataset.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a Dataset from a Task — task_dataset","text":"","code":"task = tsk(\"iris\") sepal_ingress = TorchIngressToken(   features = c(\"Sepal.Length\", \"Sepal.Width\"),   batchgetter = batchgetter_num,   shape = c(NA, 2) ) petal_ingress = TorchIngressToken(   features = c(\"Petal.Length\", \"Petal.Width\"),   batchgetter = batchgetter_num,   shape = c(NA, 2) ) ingress_tokens = list(sepal = sepal_ingress, petal = petal_ingress)  target_batchgetter = function(data) {   torch_tensor(data = data[[1L]], dtype = torch_float32())$unsqueeze(2) } dataset = task_dataset(task, ingress_tokens, target_batchgetter) 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 Descriptor — torch_callback","title":"Create a Callback Descriptor — 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 Descriptor — 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 Descriptor — 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 Descriptor — 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 Descriptor — 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 Descriptor — 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 Descriptor — 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: 0x560291776688> ##   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: 1.18 ## Epoch 2: 1.06 ## Epoch 3: 1.12 ## Epoch 4: 1.15 ## Epoch 5: 1.16 mlp$model$callbacks$custom_logger ## [1] 1.15931"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/get_started.html","id":"quickstart","dir":"Articles","previous_headings":"","what":"Quickstart","title":"Get Started","text":"vignette show get started mlr3torch training simple neural network tabular regression problem. assume familiar mlr3 framework, see e.g. mlr3 book. first example, train simple multi-layer perceptron (MLP) well-known “mtcars” task, goal predict miles per galleon (‘mpg’) cars. architecture comes predfined learner mlr3torch, can also easily create new network architectures, see Neural Networks Graphs vignette detailed introduoduion. first set seed reproducibility, load library construct task. Learners mlr3torch work similary mlr3 learners. , construct simple multi layer perceptron regression. usual calling lrn() configuring parameters: use two hidden layers 50 neurons, training, set batch size 32, number training epochs 30 device \"cpu\". complete description available parameters see ?mlr3torch::LearnerTorchMLP. can use learner training prediction just like regression learner. , split observations training test set, train learner training set create predictions test set. Finally, compute mean squared error predictions.","code":"set.seed(314) library(mlr3torch) task = tsk(\"mtcars\") task$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, jit_trace=FALSE, eval_freq=1, #>   measures_train=<list>, measures_valid=<list>, patience=0, #>   min_delta=0, batch_size=32, shuffle=TRUE, tensor_dataset=FALSE, #>   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  #> 294.1396"},{"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. 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] 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 AdamW (optim_ignite_adamw) SGD (optim_ignite_sgd). TorchLoss objects can constructed using t_opt(<key>). optimizers, associated ParamSet interesting see : Parameters TorchOptimizer (also TorchLoss TorchCallback) can set usual mlr3 way, .e. either construction, afterwards using $set_values() method parameter set. see optimizer’s parameters added learner’s ParamSet (prefixed \"opt.\") values set values specified. exposing optimizer’s parameters, can conveniently tuned using mlr3tuning. available optimizers stored mlr3torch_optimizers dictionary.","code":"sgd = t_opt(\"sgd\") sgd #> <TorchOptimizer:sgd> Stochastic Gradient Descent #> * Generator: optim_ignite_sgd #> * Parameters: list() #> * Packages: torch,mlr3torch  sgd$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 5 stored values #> Keys: adagrad, adam, adamw, rmsprop, 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 11 stored values #> Keys: checkpoint, history, lr_cosine_annealing, lr_lambda, #>   lr_multiplicative, lr_one_cycle, lr_reduce_on_plateau, lr_step, #>   progress, tb, unfreeze"},{"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 $network, underlying nn_module. full description model, see ?LearnerTorch.","code":"mlp_custom = lrn(\"regr.mlp\",   # construction arguments   optimizer = sgd, loss = l1, callbacks = history,   # scores to keep track of   measures_valid = msr(\"regr.mae\"),   # other parameters are left as-is:   # architecture   neurons = c(50, 50),   # training arguments   batch_size = 32, epochs = 30, device = \"cpu\",   # validation proportion   validate = 0.3 )  mlp_custom #> <LearnerTorchMLP[regr]:regr.mlp>: My Little Powny #> * Model: - #> * Parameters: epochs=30, device=cpu, num_threads=1, #>   num_interop_threads=1, seed=random, jit_trace=FALSE, eval_freq=1, #>   measures_train=<list>, measures_valid=<MeasureRegrSimple>, #>   patience=0, min_delta=0, batch_size=32, shuffle=TRUE, #>   tensor_dataset=FALSE, neurons=50,50, p=0.5, activation=<nn_relu>, #>   activation_args=<list>, opt.lr=0.5, opt.nesterov=FALSE #> * Validate: 0.3 #> * Packages: mlr3, mlr3torch, torch #> * Predict Types:  [response] #> * Feature Types: integer, numeric, lazy_tensor #> * Properties: internal_tuning, marshal, validation #> * Optimizer: sgd #> * Loss: l1 #> * Callbacks: history mlp_custom$train(task, row_ids = splits$train) prediction_custom = mlp_custom$predict(task, row_ids = splits$test) prediction_custom$score(msr(\"regr.mae\")) #> regr.mae  #> 8.543908 prediction$score(msr(\"regr.mae\")) #> regr.mae  #> 15.54688 head(mlp_custom$model$callbacks$history) #>    epoch valid.regr.mae #>    <num>          <num> #> 1:     1   2.310486e+05 #> 2:     2   2.009898e+10 #> 3:     3   6.967264e+03 #> 4:     4   3.220922e+02 #> 5:     5   7.638778e+01 #> 6:     6   7.588778e+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 #>  0.2528 -0.3532 -0.2106 #>  0.4273 -0.6144  0.1054 #> [ CPUFloatType{2,3} ] module_1 = nn_linear(in_features = 3, out_features = 4, bias = TRUE) output = module_1(input) output #> torch_tensor #>  0.4542 -0.2453  0.4121  0.1119 #>  0.2307 -0.3919  0.6421 -0.0936 #> [ 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 #>  0.2538  0.3499  0.3963 #>  0.2639  0.3569  0.3792 #> [ 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 #>  0.4542 -0.2453  0.4121  0.1119 #>  0.2307 -0.3919  0.6421 -0.0936 #> [ 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 #>  0.2538  0.3499  0.3963 #>  0.2639  0.3569  0.3792 #> [ 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 #> $module_list #> An `nn_module` containing 31 parameters. #>  #> ── Modules ───────────────────────────────────────────────────────────────────── #> • 0: <nn_linear> #16 parameters #> • 1: <nn_sigmoid> #0 parameters #> • 2: <nn_linear> #15 parameters #> • 3: <nn_softmax> #0 parameters graph_module(input) #> torch_tensor #>  0.2538  0.3499  0.3963 #>  0.2639  0.3569  0.3792 #> [ CPUFloatType{2,3} ][ grad_fn = <SoftmaxBackward0> ]"},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"modeldescriptor","dir":"Articles","previous_headings":"Building Torch Models for Tasks using PipeOpTorch","what":"ModelDescriptor","title":"Internals","text":"PipeOpModule represents nn_module fixed specific tensor shape hyperparameters. constructing neural network using operators, one take care output shape operations match input shapes following operations. complete Graph matching PipeOpModules can constructed using operators mostly inherit PipeOpTorch, making use ModelDescriptor class. ModelDescriptor class contains Graph (mostly) PipeOpModules information. PipeOpTorch transforms ModelDescriptor adds PipeOpModules Graph. ModelDescriptors always build Graph specific Task. easiest way initialize proper ModelDescriptor use appropriate PipeOpTorchIngress given datatype. use PipeOpTorchIngressNumeric, used numeric data. ModelDescriptor S3 object contains Graph, information generate data ($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 #>  0.2274  0.2261 -0.4887 -0.1051 #>  0.1440  0.2850 -0.3803 -0.2081 #> [ 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$ingress )  td #> <task_dataset> #>   Inherits from: <dataset> #>   Public: #>     .getbatch: function (index)  #>     .getitem: function (index)  #>     .length: function ()  #>     all_features: Petal.Length Sepal.Length Sepal.Width Species #>     cache_lazy_tensors: FALSE #>     clone: function (deep = FALSE)  #>     feature_ingress_tokens: list #>     initialize: function (task, feature_ingress_tokens, target_batchgetter = NULL)  #>     load_state_dict: function (x, ..., .refer_to_state_dict = FALSE)  #>     state_dict: function ()  #>     target_batchgetter: NULL #>     task: TaskClassif, TaskSupervised, Task, R6 batch = td$.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 #> -1.6916  1.1106  3.0460  3.0379 #> -1.4601  0.9800  2.7398  2.9830 #> -1.5414  1.0404  2.7707  2.8034 #> [ 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 #>  0.1791  0.4450  0.3759 #>  0.1817  0.4483  0.3700 #> [ 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 )  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 #> -0.4005 #> -0.3776 #> [ CPUFloatType{2,1} ][ grad_fn = <AddmmBackward0> ] #>  #> $nn_softmax.output #> torch_tensor #>  0.0855  0.3679  0.5466 #>  0.0941  0.3845  0.5214 #> [ CPUFloatType{2,3} ][ grad_fn = <SoftmaxBackward0> ] iris_module$graph$pipeops$linear1$.result #> $output #> torch_tensor #>  2.7784  1.0401 -2.5610  1.4984 #>  2.5208  1.0563 -2.5473  1.5691 #> [ CPUFloatType{2,4} ][ grad_fn = <AddmmBackward0> ] iris_module$graph$pipeops$linear3$.result #> $output #> torch_tensor #> -0.1651  0.7799  0.3149 -0.0282  0.6354 #> -0.1651  0.7799  0.3149 -0.0282  0.6354 #> [ CPUFloatType{2,5} ][ grad_fn = <AddmmBackward0> ] iris_module$graph$pipeops$nn_merge_cat$.result #> $output #> torch_tensor #>  2.7784  1.0401 -2.5610  1.4984 -0.1651  0.7799  0.3149 -0.0282  0.6354 #>  2.5208  1.0563 -2.5473  1.5691 -0.1651  0.7799  0.3149 -0.0282  0.6354 #> [ 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_ignite_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$ingress,   optimizer = adam,   callbacks = history,   loss = xe )  lr_sequential #> <LearnerTorchModel[classif]:classif.model>: Torch Model #> * Model: - #> * Parameters: device=auto, num_threads=1, num_interop_threads=1, #>   seed=random, jit_trace=FALSE, eval_freq=1, measures_train=<list>, #>   measures_valid=<list>, patience=0, min_delta=0, shuffle=TRUE, #>   tensor_dataset=FALSE, opt.lr=0.02 #> * Validate: NULL #> * Packages: mlr3, mlr3torch, torch #> * Predict Types:  [response], prob #> * Feature Types: logical, integer, numeric, character, factor, ordered, #>   POSIXct, lazy_tensor #> * Properties: featureless, hotstart_backward, hotstart_forward, #>   importance, internal_tuning, marshal, missings, multiclass, #>   oob_error, selected_features, twoclass, validation, weights #> * Optimizer: adam #> * Loss: cross_entropy #> * Callbacks: history lr_sequential$param_set$set_values(   batch_size = 50,   epochs = 100,   measures_train = msrs(c(\"classif.logloss\", \"classif.ce\")) ) # This is required to evaluate the logloss during training lr_sequential$predict_type = \"prob\"  lr_sequential$train(md_sequential$task) library(data.table) library(ggplot2)  newdata = cbind(data.table(Sepal.Width = mean(iris$Sepal.Width)), CJ(   Sepal.Length = seq(min(iris$Sepal.Length), max(iris$Sepal.Length), length.out = 30),   Petal.Length = seq(min(iris$Petal.Length), max(iris$Petal.Length), length.out = 30) ))  predictions = lr_sequential$predict_newdata(newdata)  plot_predictions = function(predictions) {   ggplot(cbind(newdata, Species = predictions$response),       aes(x = Sepal.Length, y = Petal.Length, fill = Species)) +     geom_tile(alpha = .3) +     geom_tile(alpha = .3) +     geom_point(data = iris,       aes(x = Sepal.Length, y = Petal.Length, fill = Species),       color = \"black\", pch = 21, size = 3) +     theme_bw() } plot_predictions(predictions)"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"torch-learner-pipelines","dir":"Articles","previous_headings":"","what":"Torch Learner Pipelines","title":"Internals","text":"model shown constructed using ModelDescriptor generated Graph PipeOpTorch operators. ModelDescriptor furthermore contains Task pertains. makes possible use create NN model gets trained right away, using PipeOpTorchModelClassif. missing prerequisite now add desired TorchOptimizer TorchLoss information ModelDescriptor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"adding-optimizer-loss-and-callback-meta-info-to-modeldescriptor","dir":"Articles","previous_headings":"Torch Learner Pipelines","what":"Adding Optimizer, Loss and Callback Meta-Info to ModelDescriptor","title":"Internals","text":"Remember ModelDescriptor $optimizer, $loss $callbacks slots necessary build complete Learner NN. can set corresponding PipeOpTorch operators. po(\"torch_optimizer\") used set $optimizer slot ModelDescriptor; takes desired TorchOptimizer object construction exports ParamSet. works analogously loss-function. also callbacks:","code":"po_adam = po(\"torch_optimizer\", optimizer = adam)  # hyperparameters are made available and can be changed: po_adam$param_set$values #> $lr #> [1] 0.02  md_sequential = po_adam$train(list(md_sequential))[[1]]  md_sequential$optimizer #> <TorchOptimizer:adam> Adaptive Moment Estimation #> * Generator: optim_ignite_adam #> * Parameters: lr=0.02 #> * Packages: torch,mlr3torch po_xe = po(\"torch_loss\", loss = xe) md_sequential = po_xe$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.f74754.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.f74754.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.f74754.x        <list>   poly.x              #>        poly.x <<UNTRAINED>>          nop.f74754.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 (2025). mlr3torch: Deep Learning 'mlr3'. R package version 0.2.0-9000, 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 = {2025},   note = {R package version 0.2.0-9000, 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\", 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! course (mlr3)torch. can check presentation UseR 2024.","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.1cf78f.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.1cf78f.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/PipeOpPreprocTorchTrafoNop.html","id":null,"dir":"Reference","previous_headings":"","what":"No Transformation — mlr_pipeops_trafo_nop","title":"No Transformation — 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":"No Transformation — mlr_pipeops_trafo_nop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoReshape.html","id":null,"dir":"Reference","previous_headings":"","what":"Reshaping Transformation — mlr_pipeops_trafo_reshape","title":"Reshaping Transformation — 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":"Reshaping Transformation — 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":"Reshaping Transformation — 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/Select.html","id":null,"dir":"Reference","previous_headings":"","what":"Selector Functions for Character Vectors — Select","title":"Selector Functions for Character Vectors — Select","text":"Select function subsets character vector. used callback CallbackSetUnfreeze select parameters freeze unfreeze training.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/Select.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Selector Functions for Character Vectors — Select","text":"","code":"select_all()  select_none()  select_grep(pattern, ignore.case = FALSE, perl = FALSE, fixed = FALSE)  select_name(param_names, assert_present = TRUE)  select_invert(select)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/Select.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Selector Functions for Character Vectors — Select","text":"pattern See grep() ignore.case See grep() perl See grep() fixed See grep() param_names names parameters want select assert_present Whether check param_names subset full vector names select Select","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/Select.html","id":"functions","dir":"Reference","previous_headings":"","what":"Functions","title":"Selector Functions for Character Vectors — Select","text":"select_all(): select_all selects elements select_none(): select_none selects elements select_grep(): select_grep selects elements names matching regular expression select_name(): select_name selects elements names matching given names select_invert(): select_invert selects elements selected given selector","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/Select.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Selector Functions for Character Vectors — Select","text":"","code":"select_all()(c(\"a\", \"b\")) #> [1] \"a\" \"b\" select_none()(c(\"a\", \"b\")) #> character(0) select_grep(\"b$\")(c(\"ab\", \"ac\")) #> [1] \"ab\" select_name(\"a\")(c(\"a\", \"b\")) #> [1] \"a\" select_invert(select_all())(c(\"a\", \"b\")) #> character(0)"},{"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,   additional_args = 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 $help(). additional_args () Additional arguments necessary. learning rate schedulers, torch::LRScheduler.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Torch Callback — TorchCallback","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Callback — TorchCallback","text":"","code":"TorchCallback$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/RtmpCZgcaa/file20ef2e4d4b3a, 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(39)> #>                          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:               jit_trace ParamLgl    NA    NA       2 <NoDefault[0]> #>  7:               eval_freq ParamInt 1e+00   Inf     Inf <NoDefault[0]> #>  8:          measures_train ParamUty    NA    NA     Inf <NoDefault[0]> #>  9:          measures_valid ParamUty    NA    NA     Inf <NoDefault[0]> #> 10:                patience ParamInt 0e+00   Inf     Inf <NoDefault[0]> #> 11:               min_delta ParamDbl 0e+00   Inf     Inf <NoDefault[0]> #> 12:              batch_size ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 13:                 shuffle ParamLgl    NA    NA       2          FALSE #> 14:                 sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 15:           batch_sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 16:             num_workers ParamInt 0e+00   Inf     Inf              0 #> 17:              collate_fn ParamUty    NA    NA     Inf         [NULL] #> 18:              pin_memory ParamLgl    NA    NA       2          FALSE #> 19:               drop_last ParamLgl    NA    NA       2          FALSE #> 20:                 timeout ParamDbl  -Inf   Inf     Inf             -1 #> 21:          worker_init_fn ParamUty    NA    NA     Inf <NoDefault[0]> #> 22:          worker_globals ParamUty    NA    NA     Inf <NoDefault[0]> #> 23:         worker_packages ParamUty    NA    NA     Inf <NoDefault[0]> #> 24:          tensor_dataset ParamFct    NA    NA       1 <NoDefault[0]> #> 25:                 neurons ParamUty    NA    NA     Inf <NoDefault[0]> #> 26:                       p ParamDbl 0e+00 1e+00     Inf <NoDefault[0]> #> 27:                n_layers ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 28:              activation ParamUty    NA    NA     Inf <NoDefault[0]> #> 29:         activation_args ParamUty    NA    NA     Inf <NoDefault[0]> #> 30:                   shape ParamUty    NA    NA     Inf <NoDefault[0]> #> 31:                  opt.lr ParamDbl 0e+00   Inf     Inf          0.001 #> 32:               opt.betas ParamUty    NA    NA     Inf    0.900,0.999 #> 33:                 opt.eps ParamDbl 1e-16 1e-04     Inf          1e-08 #> 34:        opt.weight_decay ParamDbl 0e+00 1e+00     Inf              0 #> 35:             opt.amsgrad ParamLgl    NA    NA       2          FALSE #> 36:          loss.reduction ParamFct    NA    NA       2           mean #> 37:      cb.checkpoint.path ParamUty    NA    NA     Inf <NoDefault[0]> #> 38:      cb.checkpoint.freq ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 39: 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:        FALSE #>  7:            1 #>  8:    <list[0]> #>  9:    <list[0]> #> 10:            0 #> 11:            0 #> 12:       [NULL] #> 13:         TRUE #> 14:       [NULL] #> 15:       [NULL] #> 16:       [NULL] #> 17:       [NULL] #> 18:       [NULL] #> 19:       [NULL] #> 20:       [NULL] #> 21:       [NULL] #> 22:       [NULL] #> 23:       [NULL] #> 24:        FALSE #> 25:              #> 26:          0.5 #> 27:       [NULL] #> 28: <nn_relu[1]> #> 29:    <list[0]> #> 30:       [NULL] #> 31:       [NULL] #> 32:       [NULL] #> 33:       [NULL] #> 34:       [NULL] #> 35:       [NULL] #> 36:       [NULL] #> 37:       [NULL] #> 38:       [NULL] #> 39:       [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,   additional_args = 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 $help(). additional_args (list()) Additional arguments necessary. learning rate schedulers, torch::LRScheduler.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"method-print-","dir":"Reference","previous_headings":"","what":"Method print()","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"Prints object","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"","code":"TorchDescriptor$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(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:           jit_trace ParamLgl    NA    NA       2 <NoDefault[0]> #>  7:           eval_freq ParamInt 1e+00   Inf     Inf <NoDefault[0]> #>  8:      measures_train ParamUty    NA    NA     Inf <NoDefault[0]> #>  9:      measures_valid ParamUty    NA    NA     Inf <NoDefault[0]> #> 10:            patience ParamInt 0e+00   Inf     Inf <NoDefault[0]> #> 11:           min_delta ParamDbl 0e+00   Inf     Inf <NoDefault[0]> #> 12:          batch_size ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 13:             shuffle ParamLgl    NA    NA       2          FALSE #> 14:             sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 15:       batch_sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 16:         num_workers ParamInt 0e+00   Inf     Inf              0 #> 17:          collate_fn ParamUty    NA    NA     Inf         [NULL] #> 18:          pin_memory ParamLgl    NA    NA       2          FALSE #> 19:           drop_last ParamLgl    NA    NA       2          FALSE #> 20:             timeout ParamDbl  -Inf   Inf     Inf             -1 #> 21:      worker_init_fn ParamUty    NA    NA     Inf <NoDefault[0]> #> 22:      worker_globals ParamUty    NA    NA     Inf <NoDefault[0]> #> 23:     worker_packages ParamUty    NA    NA     Inf <NoDefault[0]> #> 24:      tensor_dataset ParamFct    NA    NA       1 <NoDefault[0]> #> 25:             neurons ParamUty    NA    NA     Inf <NoDefault[0]> #> 26:                   p ParamDbl 0e+00 1e+00     Inf <NoDefault[0]> #> 27:            n_layers ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 28:          activation ParamUty    NA    NA     Inf <NoDefault[0]> #> 29:     activation_args ParamUty    NA    NA     Inf <NoDefault[0]> #> 30:               shape ParamUty    NA    NA     Inf <NoDefault[0]> #> 31:              opt.lr ParamDbl 0e+00   Inf     Inf          0.001 #> 32:           opt.betas ParamUty    NA    NA     Inf    0.900,0.999 #> 33:             opt.eps ParamDbl 1e-16 1e-04     Inf          1e-08 #> 34:    opt.weight_decay ParamDbl 0e+00 1e+00     Inf              0 #> 35:         opt.amsgrad ParamLgl    NA    NA       2          FALSE #> 36:      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:        FALSE #>  7:            1 #>  8:    <list[0]> #>  9:    <list[0]> #> 10:            0 #> 11:            0 #> 12:       [NULL] #> 13:         TRUE #> 14:       [NULL] #> 15:       [NULL] #> 16:       [NULL] #> 17:       [NULL] #> 18:       [NULL] #> 19:       [NULL] #> 20:       [NULL] #> 21:       [NULL] #> 22:       [NULL] #> 23:       [NULL] #> 24:        FALSE #> 25:              #> 26:          0.5 #> 27:       [NULL] #> 28: <nn_relu[1]> #> 29:    <list[0]> #> 30:       [NULL] #> 31:       [NULL] #> 32:       [NULL] #> 33:       [NULL] #> 34:       [NULL] #> 35:       [NULL] #> 36:       [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_ignite_adam, label = \"adam\") torch_opt #> <TorchOptimizer:optim_ignite_adam> adam #> * Generator: optim_ignite_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_ignite_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 (set_to_none = FALSE)  #>   Private: #>     deep_clone: function (name, value)  #>     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(36)> #>                      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:           jit_trace ParamLgl    NA    NA       2 <NoDefault[0]> #>  7:           eval_freq ParamInt     1   Inf     Inf <NoDefault[0]> #>  8:      measures_train ParamUty    NA    NA     Inf <NoDefault[0]> #>  9:      measures_valid ParamUty    NA    NA     Inf <NoDefault[0]> #> 10:            patience ParamInt     0   Inf     Inf <NoDefault[0]> #> 11:           min_delta ParamDbl     0   Inf     Inf <NoDefault[0]> #> 12:          batch_size ParamInt     1   Inf     Inf <NoDefault[0]> #> 13:             shuffle ParamLgl    NA    NA       2          FALSE #> 14:             sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 15:       batch_sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 16:         num_workers ParamInt     0   Inf     Inf              0 #> 17:          collate_fn ParamUty    NA    NA     Inf         [NULL] #> 18:          pin_memory ParamLgl    NA    NA       2          FALSE #> 19:           drop_last ParamLgl    NA    NA       2          FALSE #> 20:             timeout ParamDbl  -Inf   Inf     Inf             -1 #> 21:      worker_init_fn ParamUty    NA    NA     Inf <NoDefault[0]> #> 22:      worker_globals ParamUty    NA    NA     Inf <NoDefault[0]> #> 23:     worker_packages ParamUty    NA    NA     Inf <NoDefault[0]> #> 24:      tensor_dataset ParamFct    NA    NA       1 <NoDefault[0]> #> 25:             neurons ParamUty    NA    NA     Inf <NoDefault[0]> #> 26:                   p ParamDbl     0     1     Inf <NoDefault[0]> #> 27:            n_layers ParamInt     1   Inf     Inf <NoDefault[0]> #> 28:          activation ParamUty    NA    NA     Inf <NoDefault[0]> #> 29:     activation_args ParamUty    NA    NA     Inf <NoDefault[0]> #> 30:               shape ParamUty    NA    NA     Inf <NoDefault[0]> #> 31:              opt.lr ParamDbl     0   Inf     Inf <NoDefault[0]> #> 32:        opt.momentum ParamDbl     0     1     Inf              0 #> 33:       opt.dampening ParamDbl     0     1     Inf              0 #> 34:    opt.weight_decay ParamDbl     0     1     Inf              0 #> 35:        opt.nesterov ParamLgl    NA    NA       2          FALSE #> 36:      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:        FALSE #>  7:            1 #>  8:    <list[0]> #>  9:    <list[0]> #> 10:            0 #> 11:            0 #> 12:       [NULL] #> 13:         TRUE #> 14:       [NULL] #> 15:       [NULL] #> 16:       [NULL] #> 17:       [NULL] #> 18:       [NULL] #> 19:       [NULL] #> 20:       [NULL] #> 21:       [NULL] #> 22:       [NULL] #> 23:       [NULL] #> 24:        FALSE #> 25:              #> 26:          0.5 #> 27:       [NULL] #> 28: <nn_relu[1]> #> 29:    <list[0]> #> 30:       [NULL] #> 31:       [NULL] #> 32:       [NULL] #> 33:       [NULL] #> 34:       [NULL] #> 35:       [NULL] #> 36:       [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.7f82c6.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$iris) )() as_data_descriptor(ds2) #> <DataDescriptor: 1 ops> #> * dataset_shapes: [x: (NA,4)] #> * input_map: (x) -> Graph #> * pointer: nop.170f25.x.output #> * shape: [(NA,4)]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert to Lazy Tensor — as_lazy_tensor","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"Convert object lazy_tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"","code":"as_lazy_tensor(x, ...)  # S3 method for class 'dataset' as_lazy_tensor(x, dataset_shapes = NULL, ids = NULL, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"x () Object convert lazy_tensor ... () Additional arguments passed method. dataset_shapes (named list() (integer() NULL)) shapes output. Names elements list returned dataset. shape NULL (unknown, e.g. images different sizes) first dimension must NA indicate batch dimension. ids (integer()) ids include lazy tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"","code":"iris_ds = dataset(\"iris\",   initialize = function() {     self$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_lr_scheduler.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert to CallbackSetLRScheduler — as_lr_scheduler","title":"Convert to CallbackSetLRScheduler — as_lr_scheduler","text":"Convert torch scheduler generator CallbackSetLRScheduler.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lr_scheduler.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert to CallbackSetLRScheduler — as_lr_scheduler","text":"","code":"as_lr_scheduler(x, step_on_epoch)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lr_scheduler.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert to CallbackSetLRScheduler — as_lr_scheduler","text":"x (function) torch scheduler generator defined using torch::lr_scheduler(). step_on_epoch (logical(1)) Whether scheduler steps every epoch","code":""},{"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, ...)"},{"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. ... () 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, ...)"},{"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. ... () 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 #>  1.0671e+00  7.5651e-01  1.9241e+00 #> -1.2200e+00 -1.2343e-01  8.3400e-01 #>  1.1294e-01 -1.2217e+00  4.8949e-01 #> -2.9355e-01  2.5931e+00 -3.8925e-01 #> -9.3131e-01  8.9302e-01 -7.6513e-01 #>  5.2385e-01 -1.4898e+00 -2.3702e-01 #>  2.2191e-01 -1.9253e-01  7.3084e-05 #> -1.0812e-02 -9.9583e-01 -7.5198e-01 #>  1.8363e+00 -1.1356e+00  7.3768e-01 #> -2.1744e-01 -1.5434e-01  1.0868e+00 #> [ CPUFloatType{10,3} ] materialize(lt1, rbind = FALSE) #> [[1]] #> torch_tensor #>  1.0671 #>  0.7565 #>  1.9241 #> [ CPUFloatType{3} ] #>  #> [[2]] #> torch_tensor #> -1.2200 #> -0.1234 #>  0.8340 #> [ CPUFloatType{3} ] #>  #> [[3]] #> torch_tensor #>  0.1129 #> -1.2217 #>  0.4895 #> [ CPUFloatType{3} ] #>  #> [[4]] #> torch_tensor #> -0.2936 #>  2.5931 #> -0.3893 #> [ CPUFloatType{3} ] #>  #> [[5]] #> torch_tensor #> -0.9313 #>  0.8930 #> -0.7651 #> [ CPUFloatType{3} ] #>  #> [[6]] #> torch_tensor #>  0.5239 #> -1.4898 #> -0.2370 #> [ CPUFloatType{3} ] #>  #> [[7]] #> torch_tensor #>  0.2219 #> -0.1925 #>  0.0001 #> [ CPUFloatType{3} ] #>  #> [[8]] #> torch_tensor #> 0.01 * #> -1.0812 #> -99.5832 #> -75.1981 #> [ CPUFloatType{3} ] #>  #> [[9]] #> torch_tensor #>  1.8363 #> -1.1356 #>  0.7377 #> [ CPUFloatType{3} ] #>  #> [[10]] #> torch_tensor #> -0.2174 #> -0.1543 #>  1.0868 #> [ 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 #>  1.0671e+00  7.5651e-01  1.9241e+00 #> -1.2200e+00 -1.2343e-01  8.3400e-01 #>  1.1294e-01 -1.2217e+00  4.8949e-01 #> -2.9355e-01  2.5931e+00 -3.8925e-01 #> -9.3131e-01  8.9302e-01 -7.6513e-01 #>  5.2385e-01 -1.4898e+00 -2.3702e-01 #>  2.2191e-01 -1.9253e-01  7.3084e-05 #> -1.0812e-02 -9.9583e-01 -7.5198e-01 #>  1.8363e+00 -1.1356e+00  7.3768e-01 #> -2.1744e-01 -1.5434e-01  1.0868e+00 #> [ CPUFloatType{10,3} ] #>  #> $lt2 #> torch_tensor #>  1.6287  1.3658  0.5398 -0.1572 #>  0.2219 -0.1518 -0.5991  1.4036 #>  1.0065  0.4194 -0.1043 -1.1464 #> -1.1741 -0.1886  1.4039  1.3031 #>  1.0056 -0.7701 -1.5744  0.8223 #>  0.7315 -0.4518 -1.0262 -0.7027 #>  1.2091 -0.8113 -0.3490  0.7018 #> -0.7377  0.4125  0.1465 -0.3253 #> -2.2844 -0.1151 -1.8222  0.1563 #>  1.3287 -0.6418  0.0915 -0.6035 #> [ CPUFloatType{10,4} ] #>  materialize(d, rbind = FALSE) #> $lt1 #> $lt1[[1]] #> torch_tensor #>  1.0671 #>  0.7565 #>  1.9241 #> [ CPUFloatType{3} ] #>  #> $lt1[[2]] #> torch_tensor #> -1.2200 #> -0.1234 #>  0.8340 #> [ CPUFloatType{3} ] #>  #> $lt1[[3]] #> torch_tensor #>  0.1129 #> -1.2217 #>  0.4895 #> [ CPUFloatType{3} ] #>  #> $lt1[[4]] #> torch_tensor #> -0.2936 #>  2.5931 #> -0.3893 #> [ CPUFloatType{3} ] #>  #> $lt1[[5]] #> torch_tensor #> -0.9313 #>  0.8930 #> -0.7651 #> [ CPUFloatType{3} ] #>  #> $lt1[[6]] #> torch_tensor #>  0.5239 #> -1.4898 #> -0.2370 #> [ CPUFloatType{3} ] #>  #> $lt1[[7]] #> torch_tensor #>  0.2219 #> -0.1925 #>  0.0001 #> [ CPUFloatType{3} ] #>  #> $lt1[[8]] #> torch_tensor #> 0.01 * #> -1.0812 #> -99.5832 #> -75.1981 #> [ CPUFloatType{3} ] #>  #> $lt1[[9]] #> torch_tensor #>  1.8363 #> -1.1356 #>  0.7377 #> [ CPUFloatType{3} ] #>  #> $lt1[[10]] #> torch_tensor #> -0.2174 #> -0.1543 #>  1.0868 #> [ CPUFloatType{3} ] #>  #>  #> $lt2 #> $lt2[[1]] #> torch_tensor #>  1.6287 #>  1.3658 #>  0.5398 #> -0.1572 #> [ CPUFloatType{4} ] #>  #> $lt2[[2]] #> torch_tensor #>  0.2219 #> -0.1518 #> -0.5991 #>  1.4036 #> [ CPUFloatType{4} ] #>  #> $lt2[[3]] #> torch_tensor #>  1.0065 #>  0.4194 #> -0.1043 #> -1.1464 #> [ CPUFloatType{4} ] #>  #> $lt2[[4]] #> torch_tensor #> -1.1741 #> -0.1886 #>  1.4039 #>  1.3031 #> [ CPUFloatType{4} ] #>  #> $lt2[[5]] #> torch_tensor #>  1.0056 #> -0.7701 #> -1.5744 #>  0.8223 #> [ CPUFloatType{4} ] #>  #> $lt2[[6]] #> torch_tensor #>  0.7315 #> -0.4518 #> -1.0262 #> -0.7027 #> [ CPUFloatType{4} ] #>  #> $lt2[[7]] #> torch_tensor #>  1.2091 #> -0.8113 #> -0.3490 #>  0.7018 #> [ CPUFloatType{4} ] #>  #> $lt2[[8]] #> torch_tensor #> -0.7377 #>  0.4125 #>  0.1465 #> -0.3253 #> [ CPUFloatType{4} ] #>  #> $lt2[[9]] #> torch_tensor #> -2.2844 #> -0.1151 #> -1.8222 #>  0.1563 #> [ CPUFloatType{4} ] #>  #> $lt2[[10]] #> torch_tensor #>  1.3287 #> -0.6418 #>  0.0915 #> -0.6035 #> [ 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 #>                   <char>                                  <char> #>  1:           checkpoint                              Checkpoint #>  2:              history                                 History #>  3:  lr_cosine_annealing           Cosine Annealing LR Scheduler #>  4:            lr_lambda Multiplication by Function LR Scheduler #>  5:    lr_multiplicative   Multiplication by Factor LR Scheduler #>  6:         lr_one_cycle                      1cyle LR Scheduler #>  7: lr_reduce_on_plateau          Reduce on Plateau LR Scheduler #>  8:              lr_step                 Step Decay LR Scheduler #>  9:             progress                                Progress #> 10:                   tb                             TensorBoard #> 11:             unfreeze                                Unfreeze #>                     packages #>                       <list> #>  1:          mlr3torch,torch #>  2:          mlr3torch,torch #>  3:          mlr3torch,torch #>  4:          mlr3torch,torch #>  5:          mlr3torch,torch #>  6:          mlr3torch,torch #>  7:          mlr3torch,torch #>  8:          mlr3torch,torch #>  9: progress,mlr3torch,torch #> 10: tfevents,mlr3torch,torch #> 11:          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>          <list> #> 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":"adagrad, adam, adamw, rmsprop, 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_ignite_adam #> * Parameters: list() #> * Packages: torch,mlr3torch # is equivalent to t_opt(\"adam\") #> <TorchOptimizer:adam> Adaptive Moment Estimation #> * Generator: optim_ignite_adam #> * Parameters: list() #> * Packages: torch,mlr3torch # convert to a data.table as.data.table(mlr3torch_optimizers) #> Key: <key> #>        key                                 label        packages #>     <char>                                <char>          <list> #> 1: adagrad           Adaptive Gradient algorithm torch,mlr3torch #> 2:    adam            Adaptive Moment Estimation torch,mlr3torch #> 3:   adamw Decoupled Weight Decay Regularization torch,mlr3torch #> 4: rmsprop          Root Mean Square Propagation torch,mlr3torch #> 5:     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.checkpoint.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"","code":"cb = t_clbk(\"checkpoint\", freq = 1) task = tsk(\"iris\")  pth = tempfile() learner = lrn(\"classif.mlp\", epochs = 3, batch_size = 1, callbacks = cb) learner$param_set$set_values(cb.checkpoint.path = pth)  learner$train(task)  list.files(pth) #> [1] \"network1.pt\"   \"network2.pt\"   \"network3.pt\"   \"optimizer1.pt\" #> [5] \"optimizer2.pt\" \"optimizer3.pt\""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":null,"dir":"Reference","previous_headings":"","what":"History Callback — mlr_callback_set.history","title":"History Callback — mlr_callback_set.history","text":"Saves training validation history training. history saved data.table validation measures prefixed \"valid.\" training measures prefixed \"train.\".","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"History Callback — mlr_callback_set.history","text":"mlr3torch::CallbackSet -> CallbackSetHistory","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"History Callback — mlr_callback_set.history","text":"mlr3torch::CallbackSet$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.history.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"History Callback — mlr_callback_set.history","text":"","code":"cb = t_clbk(\"history\") task = tsk(\"iris\")  learner = lrn(\"classif.mlp\", epochs = 3, batch_size = 1,   callbacks = t_clbk(\"history\"), validate = 0.3) learner$param_set$set_values(   measures_train = msrs(c(\"classif.acc\", \"classif.ce\")),   measures_valid = msr(\"classif.ce\") ) learner$train(task)  print(learner$model$callbacks$history) #> Key: <epoch> #>    epoch train.classif.acc train.classif.ce valid.classif.ce #>    <num>             <num>            <num>            <num> #> 1:     1         0.3047619        0.6952381        0.6222222 #> 2:     2         0.3142857        0.6857143        0.7333333 #> 3:     3         0.3523810        0.6476190        0.5111111"},{"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.lr_scheduler.html","id":null,"dir":"Reference","previous_headings":"","what":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"Changes learning rate based schedule specified torch::lr_scheduler. writing, following available: torch::lr_cosine_annealing() torch::lr_lambda() torch::lr_multiplicative() torch::lr_one_cycle() torch::lr_reduce_on_plateau() torch::lr_step() Custom schedulers defined torch::lr_scheduler().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"mlr3torch::CallbackSet -> CallbackSetLRScheduler","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"public-fields","dir":"Reference","previous_headings":"","what":"Public fields","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"scheduler_fn (lr_scheduler_generator) torch function creates learning rate scheduler scheduler (LRScheduler) learning rate scheduler wrapped callback","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"mlr3torch::CallbackSet$load_state_dict() mlr3torch::CallbackSet$print() mlr3torch::CallbackSet$state_dict()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"CallbackSetLRScheduler$new() CallbackSetLRScheduler$on_begin() CallbackSetLRScheduler$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"","code":"CallbackSetLRScheduler$new(.scheduler, step_on_epoch, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":".scheduler (lr_scheduler_generator) torch scheduler generator (e.g. torch::lr_step). step_on_epoch (logical(1)) Whether scheduler steps every epoch (otherwise every batch). ... () scheduler-specific arguments","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"method-on-begin-","dir":"Reference","previous_headings":"","what":"Method on_begin()","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"Creates scheduler using optimizer context","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"","code":"CallbackSetLRScheduler$on_begin()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"","code":"CallbackSetLRScheduler$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.lr_scheduler.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Learning Rate Scheduling Callback — mlr_callback_set.lr_scheduler","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":null,"dir":"Reference","previous_headings":"","what":"Progress Callback — mlr_callback_set.progress","title":"Progress Callback — mlr_callback_set.progress","text":"Prints progress bar metrics training validation.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Progress Callback — mlr_callback_set.progress","text":"mlr3torch::CallbackSet -> CallbackSetProgress","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Progress Callback — mlr_callback_set.progress","text":"mlr3torch::CallbackSet$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$on_epoch_end() CallbackSetProgress$on_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-on-end-","dir":"Reference","previous_headings":"","what":"Method on_end()","title":"Progress Callback — mlr_callback_set.progress","text":"Prints time end training.","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$on_end()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Progress Callback — mlr_callback_set.progress","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"usage-6","dir":"Reference","previous_headings":"","what":"Usage","title":"Progress Callback — mlr_callback_set.progress","text":"","code":"CallbackSetProgress$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.progress.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Progress Callback — mlr_callback_set.progress","text":"","code":"task = tsk(\"iris\")  learner = lrn(\"classif.mlp\", epochs = 5, batch_size = 1,   callbacks = t_clbk(\"progress\"), validate = 0.3) learner$param_set$set_values(   measures_train = msrs(c(\"classif.acc\", \"classif.ce\")),   measures_valid = msr(\"classif.ce\") )  learner$train(task) #> Epoch 1 started (2025-02-07 09:19:01) #>  #> [Summary epoch 1] #> ------------------ #> Measures (Train): #>  * classif.acc = 0.31 #>  * classif.ce = 0.69 #> Measures (Valid): #>  * classif.ce = 0.62 #> Epoch 2 started (2025-02-07 09:19:02) #>  #> [Summary epoch 2] #> ------------------ #> Measures (Train): #>  * classif.acc = 0.31 #>  * classif.ce = 0.69 #> Measures (Valid): #>  * classif.ce = 0.56 #> Epoch 3 started (2025-02-07 09:19:02) #>  #> [Summary epoch 3] #> ------------------ #> Measures (Train): #>  * classif.acc = 0.55 #>  * classif.ce = 0.45 #> Measures (Valid): #>  * classif.ce = 0.44 #> Epoch 4 started (2025-02-07 09:19:03) #>  #> [Summary epoch 4] #> ------------------ #> Measures (Train): #>  * classif.acc = 0.57 #>  * classif.ce = 0.43 #> Measures (Valid): #>  * classif.ce = 0.49 #> Epoch 5 started (2025-02-07 09:19:03) #>  #> [Summary epoch 5] #> ------------------ #> Measures (Train): #>  * classif.acc = 0.51 #>  * classif.ce = 0.49 #> Measures (Valid): #>  * classif.ce = 0.60 #> Finished training for 5 epochs (2025-02-07 09:19:03)"},{"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_callback_set.unfreeze.html","id":null,"dir":"Reference","previous_headings":"","what":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"Unfreeze weights (parameters network) number steps epochs.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"mlr3torch::CallbackSet -> CallbackSetUnfreeze","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"mlr3torch::CallbackSet$load_state_dict() mlr3torch::CallbackSet$print() mlr3torch::CallbackSet$state_dict()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"CallbackSetUnfreeze$new() CallbackSetUnfreeze$on_begin() CallbackSetUnfreeze$on_epoch_begin() CallbackSetUnfreeze$on_batch_begin() CallbackSetUnfreeze$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"","code":"CallbackSetUnfreeze$new(starting_weights, unfreeze)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"starting_weights (Select) Select denoting weights trainable start. unfreeze (data.table) data.table column weights (list column Selects) column epoch batch. selector indicates parameters unfreeze, epoch batch column indicates .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"method-on-begin-","dir":"Reference","previous_headings":"","what":"Method on_begin()","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"Sets starting weights","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"","code":"CallbackSetUnfreeze$on_begin()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"method-on-epoch-begin-","dir":"Reference","previous_headings":"","what":"Method on_epoch_begin()","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"Unfreezes weights training correct epoch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"","code":"CallbackSetUnfreeze$on_epoch_begin()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"method-on-batch-begin-","dir":"Reference","previous_headings":"","what":"Method on_batch_begin()","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"Unfreezes weights training correct batch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"usage-3","dir":"Reference","previous_headings":"","what":"Usage","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"","code":"CallbackSetUnfreeze$on_batch_begin()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"usage-4","dir":"Reference","previous_headings":"","what":"Usage","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"","code":"CallbackSetUnfreeze$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.unfreeze.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Unfreezing Weights Callback — mlr_callback_set.unfreeze","text":"","code":"task = tsk(\"iris\") cb = t_clbk(\"unfreeze\") mlp = lrn(\"classif.mlp\", callbacks = cb,  cb.unfreeze.starting_weights = select_invert(    select_name(c(\"0.weight\", \"3.weight\", \"6.weight\", \"6.bias\"))  ),  cb.unfreeze.unfreeze = data.table(    epoch = c(2, 5),    weights = list(select_name(\"0.weight\"), select_name(c(\"3.weight\", \"6.weight\")))  ),  epochs = 6, batch_size = 150, neurons = c(1, 1, 1) )  mlp$train(task)"},{"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. last_loss (numeric(1)) loss last trainings batch. 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. device (torch::torch_device) device.","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$new(   learner,   task_train,   task_valid = NULL,   loader_train,   loader_valid = NULL,   measures_train = NULL,   measures_valid = NULL,   network,   optimizer,   loss_fn,   total_epochs,   prediction_encoder,   eval_freq = 1L,   device )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Context for Torch Learner — mlr_context_torch","text":"learner (Learner) torch learner. task_train (Task) training task. task_valid (Task NULL) validation task. loader_train (torch::dataloader) data loader training. loader_valid (torch::dataloader NULL) data loader validation. measures_train (list() Measures NULL) Measures used training. Default NULL. measures_valid (list() Measures NULL) Measures used validation. network (torch::nn_module) torch network. optimizer (torch::optimizer) optimizer. loss_fn (torch::nn_module) loss function. total_epochs (integer(1)) total number epochs learner trained . prediction_encoder (function()) learner's prediction encoder. eval_freq (integer(1)) evaluation frequency. device (character(1)) device.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Context for Torch Learner — mlr_context_torch","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Context for Torch Learner — mlr_context_torch","text":"","code":"ContextTorch$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Context for Torch Learner — mlr_context_torch","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":null,"dir":"Reference","previous_headings":"","what":"My Little Pony — mlr_learners.mlp","title":"My Little Pony — mlr_learners.mlp","text":"Fully connected feed forward network dropout activation function. features can either single lazy_tensor one numeric columns ().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"dictionary","dir":"Reference","previous_headings":"","what":"Dictionary","title":"My Little Pony — mlr_learners.mlp","text":"Learner can instantiated using sugar function lrn():","code":"lrn(\"classif.mlp\", ...) lrn(\"regr.mlp\", ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"My Little Pony — mlr_learners.mlp","text":"Supported task types: 'classif', 'regr' Predict Types: classif: 'response', 'prob' regr: 'response' Feature Types: “integer”, “numeric”, “lazy_tensor” Required Packages: mlr3, mlr3torch, torch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"My Little Pony — mlr_learners.mlp","text":"Parameters LearnerTorch, well : activation :: [nn_module] activation function. initialized nn_relu. activation_args :: named list() named list initialization arguments activation function. intialized empty list. neurons :: integer() number neurons per hidden layer. default hidden layer. Setting c(10, 20) first hidden layer 10 neurons second 20. n_layers :: integer() number layers. parameter must set neurons length 1. p :: numeric(1) dropout probability. initialized 0.5. shape :: integer() NULL input shape length 2, e.g. c(NA, 5). needs present lazy tensor input unknown shape (NULL). Otherwise input shape inferred number numeric features.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"My Little Pony — mlr_learners.mlp","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.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  #>       0.64"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":null,"dir":"Reference","previous_headings":"","what":"Tabular ResNet — mlr_learners.tab_resnet","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Tabular resnet.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"dictionary","dir":"Reference","previous_headings":"","what":"Dictionary","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Learner can instantiated using sugar function lrn():","code":"lrn(\"classif.tab_resnet\", ...) lrn(\"regr.tab_resnet\", ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Supported task types: 'classif', 'regr' Predict Types: classif: 'response', 'prob' regr: 'response' Feature Types: “integer”, “numeric” Required Packages: mlr3, mlr3torch, torch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Parameters LearnerTorch, well : n_blocks :: integer(1) number blocks. d_block :: integer(1) input output dimension block. d_hidden :: integer(1) latent dimension block. d_hidden_multiplier :: numeric(1) Alternative way specify latent dimension d_block * d_hidden_multiplier. dropout1 :: numeric(1) First dropout ratio. dropout2 :: numeric(1) Second dropout ratio.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Gorishniy Y, Rubachev , Khrulkov V, Babenko (2021). “Revisiting Deep Learning  Tabular Data.” arXiv, 2106.11959.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> LearnerTorchTabResNet","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"LearnerTorchTabResNet$new() LearnerTorchTabResNet$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"","code":"LearnerTorchTabResNet$new(   task_type,   optimizer = NULL,   loss = NULL,   callbacks = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"task_type (character(1)) task type, either \"classif\" \"regr\". optimizer (TorchOptimizer) optimizer use training. Per default, adam used. loss (TorchLoss) loss used train network. Per default, mse used regression cross_entropy classification. callbacks (list() TorchCallbacks) callbacks. Must unique ids.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"","code":"LearnerTorchTabResNet$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"","code":"# Define the Learner and set parameter values learner = lrn(\"classif.tab_resnet\") learner$param_set$set_values(   epochs = 1, batch_size = 16, device = \"cpu\",   n_blocks = 2, d_block = 10, d_hidden = 20, dropout1 = 0.3, dropout2 = 0.3 )  # Define a Task task = tsk(\"iris\")  # Create train and test set ids = partition(task)  # Train the learner on the training ids learner$train(task, row_ids = ids$train)  # Make predictions for the test rows predictions = learner$predict(task, row_ids = ids$test)  # Score the predictions predictions$score() #> classif.ce  #>       0.66"},{"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$score() #> classif.ce  #>        0.7"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":null,"dir":"Reference","previous_headings":"","what":"AlexNet Image Classifier — mlr_learners.torchvision","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"Classic image classification networks torchvision.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"Parameters LearnerTorchImage pretrained :: logical(1) Whether use pretrained model. final linear layer replaced new nn_linear number classes inferred Task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"Supported task types: \"classif\" Predict Types: \"response\" \"prob\" Feature Types: \"lazy_tensor\" Required packages: \"mlr3torch\", \"torch\", \"torchvision\"","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> mlr3torch::LearnerTorchImage -> LearnerTorchVision","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"LearnerTorchVision$new() LearnerTorchVision$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"","code":"LearnerTorchVision$new(   name,   module_generator,   label,   optimizer = NULL,   loss = NULL,   callbacks = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"name (character(1)) name network. module_generator (function(pretrained, num_classes)) Function generates network. label (character(1)) label network. #' @references Krizhevsky, Alex, Sutskever, Ilya, Hinton, E. G (2017). “Imagenet classification deep convolutional neural networks.” Communications ACM, 60(6), 84–90. Sandler, Mark, Howard, Andrew, Zhu, Menglong, Zhmoginov, Andrey, Chen, Liang-Chieh (2018). “Mobilenetv2: Inverted residuals linear bottlenecks.” Proceedings IEEE conference computer vision pattern recognition, 4510–4520. , Kaiming, Zhang, Xiangyu, Ren, Shaoqing, Sun, Jian (2016). “Deep residual learning image recognition.” Proceedings IEEE conference computer vision pattern recognition, 770–778. Simonyan, Karen, Zisserman, Andrew (2014). “deep convolutional networks large-scale image recognition.” arXiv preprint arXiv:1409.1556. optimizer (TorchOptimizer) optimizer use training. Per default, adam used. loss (TorchLoss) loss used train network. Per default, mse used regression cross_entropy classification. callbacks (list() TorchCallbacks) callbacks. Must unique ids.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"","code":"LearnerTorchVision$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":null,"dir":"Reference","previous_headings":"","what":"Base Class for Torch Learners — mlr_learners_torch","title":"Base Class for Torch Learners — mlr_learners_torch","text":"base class provides basic functionality training prediction neural network. torch learners inherit class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"validation","dir":"Reference","previous_headings":"","what":"Validation","title":"Base Class for Torch Learners — mlr_learners_torch","text":"specify validation data, can set $validate field Learner, can set : NULL: validation ratio: proportion 1 - ratio task used training ratio used validation. \"test\" means \"test\" task resampling used possible calling $train() manually. \"predefined\": use predefined $internal_valid_task mlr3::Task. validation data can also used early stopping, see description Learner's parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"saving-a-learner","dir":"Reference","previous_headings":"","what":"Saving a Learner","title":"Base Class for Torch Learners — mlr_learners_torch","text":"order save LearnerTorch later usage, necessary call $marshal() method Learner writing disk, object otherwise saved correctly. loading marshaled LearnerTorch R , need call $unmarshal() transform useable state.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"early-stopping-and-internal-tuning","dir":"Reference","previous_headings":"","what":"Early Stopping and Internal Tuning","title":"Base Class for Torch Learners — mlr_learners_torch","text":"order prevent overfitting, LearnerTorch class allows use early stopping via patience min_delta parameters, see Learner's parameters. tuning LearnerTorch also possible combine explicit tuning via mlr3tuning LearnerTorch's internal tuning epochs via early stopping. , just need include epochs = to_tune(upper = <upper>, internal = TRUE) search space, <upper> maximally allowed number epochs, configure early stopping.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"model","dir":"Reference","previous_headings":"","what":"Model","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Model list class \"learner_torch_model\" following elements: network :: trained network. optimizer :: $state_dict() optimizer used train network. loss_fn :: $state_dict() loss used train network. callbacks :: callbacks used train network. seed :: seed / used training prediction. epochs :: many epochs model trained (early stopping). task_col_info :: data.table() containing information train-task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Base Class for Torch Learners — mlr_learners_torch","text":"General: parameters optimizer, loss callbacks, prefixed \"opt.\", \"loss.\" \"cb.<callback id>.\" respectively, well : epochs :: integer(1) number epochs. device :: character(1) device. One \"auto\", \"cpu\", \"cuda\" values defined mlr_reflections$torch$devices. value initialized \"auto\", select \"cuda\" possible, try \"mps\" otherwise fall back \"cpu\". num_threads :: integer(1) number threads intraop pararallelization (device \"cpu\"). value initialized 1. num_interop_threads :: integer(1) number threads intraop interop pararallelization (device \"cpu\"). value initialized 1. Note can set session changing value within R session raise warning. seed :: integer(1) \"random\" NULL torch seed used training prediction. value initialized \"random\", means random seed sampled beginning training phase. seed (either set randomly sampled) available via $model$seed training used prediction. Note setting seed training phase mean default (.e. seed \"random\"), clones learner use different seed. set NULL, seeding done. tensor_dataset :: logical(1) | \"device\" Whether load batches beginning training stack . initialized FALSE. set \"device\", device tensors set value device, can avoid unnecessary moving tensors devices. dataset fits memory make loading batches faster. Note set datasets contain lazy_tensors random data augmentation, augmentation applied beginning training. 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. set, epoch improvement observed, can accessed via $internal_tuned_values field learner. 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. initialized TRUE, differs default (FALSE). 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. dataset must return named list : x list torch tensors input network. networks one input, names must correspond inputs network. y target tensor. .index indices batch (integer() torch_int()). Moreover, one needs pay attention respect row ids provided task. also possible overwrite private .dataloader() method. must respect dataloader parameters ParamSet. .dataloader(dataset, param_vals) (Task, list()) -> torch::dataloader Create dataloader task. Needs respect least batch_size shuffle (otherwise predictions can permuted). change predict types, possible overwrite method : .encode_prediction(predict_tensor, task) (torch_tensor, Task) -> list() Take raw predictions self$network (predict_tensor) encode format can converted valid mlr3 predictions using mlr3::as_prediction_data(). method must take self$predict_type account. possible add parameters specifying param_set construction argument, currently possible remove existing parameters, .e. listed section Parameters. None parameters provided param_set can id starts \"loss.\", \"opt.\", \"cb.\"`, preserved dynamically constructed parameters optimizer, loss function, callbacks. perform additional input checks task, private .verify_train_task(task, param_vals) .verify_predict_task(task, param_vals) can overwritten. learners construction arguments change hash learner, required implement private $.additional_phash_input().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Base Class for Torch Learners — mlr_learners_torch","text":"mlr3::Learner -> LearnerTorch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Base Class for Torch Learners — mlr_learners_torch","text":"validate construct internal validation data. parameter can either NULL, ratio $(0, 1)$, \"test\", \"predefined\". loss (TorchLoss) torch loss. optimizer (TorchOptimizer) torch optimizer. callbacks (list() TorchCallbacks) List torch callbacks. ids set names. internal_valid_scores Retrieves internal validation scores named list(). Specify $validate field measures_valid parameter configure . Returns NULL learner trained yet. internal_tuned_values early stopping activate, returns named list early-stopped epochs, otherwise empty list returned. Returns NULL learner trained yet. marshaled (logical(1)) Whether learner marshaled. network (nn_module()) Shortcut learner$model$network. param_set (ParamSet) parameter set hash (character(1)) Hash (unique identifier) object. phash (character(1)) Hash (unique identifier) partial object, excluding components varied systematically tuning (parameter values).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Base Class for Torch Learners — mlr_learners_torch","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Base Class for Torch Learners — mlr_learners_torch","text":"LearnerTorch$new() LearnerTorch$format() LearnerTorch$print() LearnerTorch$marshal() LearnerTorch$unmarshal() LearnerTorch$dataset() LearnerTorch$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$new(   id,   task_type,   param_set,   properties,   man,   label,   feature_types,   optimizer = NULL,   loss = NULL,   packages = character(),   predict_types = NULL,   callbacks = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"id (character(1)) id new object. task_type (character(1)) task type. param_set (ParamSet alist()) Either parameter set, alist() containing different values self, e.g. alist(private$.param_set1, private$.param_set2), ParamSet collection created. properties (character()) properties object. See mlr_reflections$learner_properties available values. man (character(1)) String format [pkg]::[topic] pointing manual page object. referenced help package can opened via method $help(). label (character(1)) Label new instance. feature_types (character()) feature types. See mlr_reflections$task_feature_types available values, Additionally, \"lazy_tensor\" supported. optimizer (NULL TorchOptimizer) optimizer use training. Defaults adam. loss (NULL TorchLoss) loss use training. Defaults MSE regression cross entropy classification. packages (character()) R packages object depends . predict_types (character()) predict types. See mlr_reflections$learner_predict_types available values. regression, default \"response\". classification, defaults \"response\" \"prob\". deviate defaults, necessary overwrite private $.encode_prediction() method, see section Inheriting. callbacks (list() TorchCallbacks) callbacks use training. Defaults empty list(), .e. callbacks.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-format-","dir":"Reference","previous_headings":"","what":"Method format()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Helper print outputs.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$format(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"... (ignored).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-print-","dir":"Reference","previous_headings":"","what":"Method print()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Prints object.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$print(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-2","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"... () Currently unused.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-marshal-","dir":"Reference","previous_headings":"","what":"Method marshal()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Marshal learner.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-3","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$marshal(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-3","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"... () Additional parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Base Class for Torch Learners — mlr_learners_torch","text":"self","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-unmarshal-","dir":"Reference","previous_headings":"","what":"Method unmarshal()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Unmarshal learner.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-4","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$unmarshal(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-4","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"... () Additional parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"returns-1","dir":"Reference","previous_headings":"","what":"Returns","title":"Base Class for Torch Learners — mlr_learners_torch","text":"self","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-dataset-","dir":"Reference","previous_headings":"","what":"Method dataset()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Create dataset task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-5","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$dataset(task)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-5","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"task Task task","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"returns-2","dir":"Reference","previous_headings":"","what":"Returns","title":"Base Class for Torch Learners — mlr_learners_torch","text":"dataset","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-6","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-6","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":null,"dir":"Reference","previous_headings":"","what":"Image Learner — mlr_learners_torch_image","title":"Image Learner — mlr_learners_torch_image","text":"Base Class Image Learners. features assumed single lazy_tensor column RGB format.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Image Learner — mlr_learners_torch_image","text":"Parameters include inherited LearnerTorch param_set construction argument.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Image Learner — mlr_learners_torch_image","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> LearnerTorchImage","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Image Learner — mlr_learners_torch_image","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Image Learner — mlr_learners_torch_image","text":"LearnerTorchImage$new() LearnerTorchImage$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Image Learner — mlr_learners_torch_image","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Image Learner — mlr_learners_torch_image","text":"","code":"LearnerTorchImage$new(   id,   task_type,   param_set = ps(),   label,   optimizer = NULL,   loss = NULL,   callbacks = list(),   packages = \"torchvision\",   man,   properties = NULL,   predict_types = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Image Learner — mlr_learners_torch_image","text":"id (character(1)) id new object. task_type (character(1)) task type. param_set (ParamSet) parameter set. label (character(1)) Label new instance. optimizer (TorchOptimizer) torch optimizer. loss (TorchLoss) loss use training. callbacks (list() TorchCallbacks) callbacks used training. Must unique ids. executed order provided packages (character()) R packages object depends . man (character(1)) String format [pkg]::[topic] pointing manual page object. referenced help package can opened via method $help(). properties (character()) properties object. See mlr_reflections$learner_properties available values. predict_types (character()) predict types. See mlr_reflections$learner_predict_types available values.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Image Learner — mlr_learners_torch_image","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Image Learner — mlr_learners_torch_image","text":"","code":"LearnerTorchImage$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Image Learner — mlr_learners_torch_image","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Learner Torch Model — mlr_learners_torch_model","title":"Learner Torch Model — mlr_learners_torch_model","text":"Create torch learner instantiated nn_module(). classification, output network must scores (softmax).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Learner Torch Model — mlr_learners_torch_model","text":"See LearnerTorch","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Learner Torch Model — mlr_learners_torch_model","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> LearnerTorchModel","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Learner Torch Model — mlr_learners_torch_model","text":"network_stored (nn_module NULL) network trained. calling $train(), NULL. ingress_tokens (named list() TorchIngressToken NULL) ingress tokens. Must non-NULL calling $train().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Learner Torch Model — mlr_learners_torch_model","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Learner Torch Model — mlr_learners_torch_model","text":"LearnerTorchModel$new() LearnerTorchModel$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Learner Torch Model — mlr_learners_torch_model","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Learner Torch Model — mlr_learners_torch_model","text":"","code":"LearnerTorchModel$new(   network = NULL,   ingress_tokens = NULL,   task_type,   properties = NULL,   optimizer = NULL,   loss = NULL,   callbacks = list(),   packages = character(0),   feature_types = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Learner Torch Model — mlr_learners_torch_model","text":"network (nn_module) instantiated nn_module. cloned construction. classification, outputs must scores (softmax). ingress_tokens (list TorchIngressToken()) list ingress tokens defines dataloader defined. task_type (character(1)) task type. properties (NULL character()) properties learner. Defaults available properties given task type. optimizer (TorchOptimizer) torch optimizer. loss (TorchLoss) loss use training. callbacks (list() TorchCallbacks) callbacks used training. Must unique ids. executed order provided packages (character()) R packages object depends . feature_types (NULL character()) feature types. Defaults available feature types.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Learner Torch Model — mlr_learners_torch_model","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Learner Torch Model — mlr_learners_torch_model","text":"","code":"LearnerTorchModel$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Learner Torch Model — mlr_learners_torch_model","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Learner Torch Model — mlr_learners_torch_model","text":"","code":"# We show the learner using a classification task  # The iris task has 4 features and 3 classes network = nn_linear(4, 3) task = tsk(\"iris\")  # This defines the dataloader. # It loads all 4 features, which are also numeric. # The shape is (NA, 4) because the batch dimension is generally NA ingress_tokens = list(   input = TorchIngressToken(task$feature_names, batchgetter_num, c(NA, 4)) )  # Creating the learner and setting required parameters learner = lrn(\"classif.torch_model\",   network = network,   ingress_tokens = ingress_tokens,   batch_size = 16,   epochs = 1,   device = \"cpu\" )  # A simple train-predict ids = partition(task) learner$train(task, ids$train) learner$predict(task, ids$test) #> <PredictionClassif> for 50 observations: #>  row_ids     truth   response #>        2    setosa versicolor #>        3    setosa versicolor #>        9    setosa versicolor #>      ---       ---        --- #>      134 virginica versicolor #>      138 virginica versicolor #>      140 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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"Calls nn_adaptive_avg_pool1d() training.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAdaptiveAvgPool -> PipeOpTorchAdaptiveAvgPool1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"Calls nn_adaptive_avg_pool2d() training.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAdaptiveAvgPool -> PipeOpTorchAdaptiveAvgPool2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"Calls nn_adaptive_avg_pool3d() training.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAdaptiveAvgPool -> PipeOpTorchAdaptiveAvgPool3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"Calls nn_avg_pool1d() training.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAvgPool -> PipeOpTorchAvgPool1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"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 $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"kernel_size :: (integer()) size window. Can single number vector. stride :: integer() stride window. Can single number vector. Default: kernel_size. padding :: integer() Implicit zero paddings sides input. Can single number vector. Default: 0. ceil_mode :: integer() TRUE, use ceil instead floor compute output shape. Default: FALSE. count_include_pad :: logical(1) TRUE, include zero-padding averaging calculation. Default: TRUE. divisor_override :: logical(1) specified, used divisor, otherwise size pooling region used. Default: NULL. available dimension greater 1.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAvgPool -> PipeOpTorchAvgPool2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"mlr3pipelines::PipeOp$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":"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":"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 $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"kernel_size :: (integer()) size window. Can single number vector. stride :: integer() stride window. Can single number vector. Default: kernel_size. padding :: integer() Implicit zero paddings sides input. Can single number vector. Default: 0. ceil_mode :: integer() TRUE, use ceil instead floor compute output shape. Default: FALSE. count_include_pad :: logical(1) TRUE, include zero-padding averaging calculation. Default: TRUE. divisor_override :: logical(1) specified, used divisor, otherwise size pooling region used. Default: NULL. available dimension greater 1.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAvgPool -> PipeOpTorchAvgPool3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchBatchNorm -> PipeOpTorchBatchNorm1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"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 $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"eps :: numeric(1) value added denominator numerical stability. Default: 1e-5. momentum :: numeric(1) value used running_mean running_var computation. Can set NULL cumulative moving average (.e. simple average). Default: 0.1 affine :: logical(1) boolean value set TRUE, module learnable affine parameters. Default: TRUE track_running_stats :: logical(1) boolean value set TRUE, module tracks running mean variance, set FALSE, module track statistics always uses batch statistics training eval modes. Default: TRUE","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchBatchNorm -> PipeOpTorchBatchNorm2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"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 $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"eps :: numeric(1) value added denominator numerical stability. Default: 1e-5. momentum :: numeric(1) value used running_mean running_var computation. Can set NULL cumulative moving average (.e. simple average). Default: 0.1 affine :: logical(1) boolean value set TRUE, module learnable affine parameters. Default: TRUE track_running_stats :: logical(1) boolean value set TRUE, module tracks running mean variance, set FALSE, module track statistics always uses batch statistics training eval modes. Default: TRUE","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchBatchNorm -> PipeOpTorchBatchNorm3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"mlr3pipelines::PipeOp$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 concatenating (via %>>%).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"naming","dir":"Reference","previous_headings":"","what":"Naming","title":"Block Repetition — mlr_pipeops_nn_block","text":"generated module graph, IDs modules generated prefixing IDs n_blocks layers ID PipeOpTorchBlock postfixing __<layer>.","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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Block Repetition — mlr_pipeops_nn_block","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchBlock","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Block Repetition — mlr_pipeops_nn_block","text":"block (Graph) neural network segment repeated PipeOp.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Block Repetition — mlr_pipeops_nn_block","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"Calls torch::nn_celu() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchCELU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"Calls torch::nn_conv1d() trained. paramter in_channels inferred second dimension input tensor.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConv -> PipeOpTorchConv1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"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 $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"out_channels :: integer(1) Number channels produced convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. default 1. padding :: integer() ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘TRUE’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConv -> PipeOpTorchConv2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"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 $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"out_channels :: integer(1) Number channels produced convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. default 1. padding :: integer() ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘TRUE’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConv -> PipeOpTorchConv3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"mlr3pipelines::PipeOp$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":"Applies 1D transposed convolution operator input signal composed several input planes, sometimes also called \"deconvolution\".","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"Calls nn_conv_transpose1d. parameter in_channels inferred second dimension input tensor.","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":"state","dir":"Reference","previous_headings":"","what":"State","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConvTranspose -> PipeOpTorchConvTranspose1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"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 $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"out_channels :: integer(1) Number output channels produce convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. Default: 1. padding ::  integer()` ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. output_padding ::integer() Additional size added one side output shape. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘True’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConvTranspose -> PipeOpTorchConvTranspose2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"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 $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"out_channels :: integer(1) Number output channels produce convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. Default: 1. padding ::  integer()` ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. output_padding ::integer() Additional size added one side output shape. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘True’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConvTranspose -> PipeOpTorchConvTranspose3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Dropout — mlr_pipeops_nn_dropout","text":"Calls torch::nn_dropout() trained.","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 :: logical(1) set TRUE, operation -place. Default: FALSE.","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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Dropout — mlr_pipeops_nn_dropout","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchDropout","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Dropout — mlr_pipeops_nn_dropout","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"Calls torch::nn_elu() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchELU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"Calls torch::nn_flatten() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchFlatten","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"Calls torch::nn_gelu() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchGELU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"Calls torch::nn_glu() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchGLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"Calls torch::nn_hardshrink() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchHardShrink","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"Calls torch::nn_hardsigmoid() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchHardSigmoid","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"Calls torch::nn_hardtanh() trained.","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":"state","dir":"Reference","previous_headings":"","what":"State","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"state value calculated public method $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchHardTanh","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"mlr3pipelines::PipeOp$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.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Output Head — mlr_pipeops_nn_head","text":"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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Output Head — mlr_pipeops_nn_head","text":"Calls torch::nn_linear() input output features inferred input shape / task.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Output Head — mlr_pipeops_nn_head","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchHead","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Output Head — mlr_pipeops_nn_head","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"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":"state","dir":"Reference","previous_headings":"","what":"State","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchLayerNorm","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"Calls torch::nn_leaky_relu() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchLeakyReLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Linear Layer — mlr_pipeops_nn_linear","text":"Calls torch::nn_linear() trained parameter in_features inferred second last dimension input tensor.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Linear Layer — mlr_pipeops_nn_linear","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchLinear","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Linear Layer — mlr_pipeops_nn_linear","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"Calls torch::nn_log_sigmoid() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchLogSigmoid","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"Calls torch::nn_max_pool1d() training.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMaxPool -> PipeOpTorchMaxPool1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"state","dir":"Reference","previous_headings":"","what":"State","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"kernel_size :: integer() size window. Can single number vector. stride :: (integer(1)) stride window. Can single number vector. Default: kernel_size padding :: integer() Implicit zero paddings sides input. Can single number tuple (padW,). Default: 0 dilation :: integer() Controls spacing kernel points; also known à trous algorithm. Default: 1 ceil_mode :: logical(1) True, use ceil instead floor compute output shape. Default: FALSE","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"return_indices FALSE construction, one input channel 'input' one output channel 'output'. return_indices TRUE, two output channels 'output' 'indices'. explanation see PipeOpTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMaxPool -> PipeOpTorchMaxPool2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","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":"state","dir":"Reference","previous_headings":"","what":"State","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"kernel_size :: integer() size window. Can single number vector. stride :: (integer(1)) stride window. Can single number vector. Default: kernel_size padding :: integer() Implicit zero paddings sides input. Can single number tuple (padW,). Default: 0 dilation :: integer() Controls spacing kernel points; also known à trous algorithm. Default: 1 ceil_mode :: logical(1) True, use ceil instead floor compute output shape. Default: FALSE","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"return_indices FALSE construction, one input channel 'input' one output channel 'output'. return_indices TRUE, two output channels 'output' 'indices'. explanation see PipeOpTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMaxPool -> PipeOpTorchMaxPool3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"mlr3pipelines::PipeOp$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":"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":"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":"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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"Calls nn_merge_cat() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMerge -> PipeOpTorchMergeCat","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"Calls nn_merge_prod() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMerge -> PipeOpTorchMergeProd","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"Calls nn_merge_sum() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMerge -> PipeOpTorchMergeSum","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"Calls torch::nn_prelu() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchPReLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"Calls torch::nn_relu() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchReLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"Calls torch::nn_relu6() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchReLU6","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"Calls nn_reshape() trained. internally calls torch::torch_reshape() given shape.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchReshape","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"Calls torch::nn_rrelu() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchRReLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"Calls torch::nn_selu() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSELU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"Calls torch::nn_sigmoid() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSigmoid","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Softmax — mlr_pipeops_nn_softmax","text":"Calls torch::nn_softmax() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Softmax — mlr_pipeops_nn_softmax","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSoftmax","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Softmax — mlr_pipeops_nn_softmax","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"Calls torch::nn_softplus() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSoftPlus","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"Calls torch::nn_softshrink() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSoftShrink","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"Calls torch::nn_softsign() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSoftSign","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"Calls nn_squeeze() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSqueeze","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"Calls torch::nn_tanh() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchTanh","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"Calls torch::nn_tanhshrink() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchTanhShrink","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"mlr3pipelines::PipeOp$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":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"Calls torch::nn_threshold() trained.","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchThreshold","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"mlr3pipelines::PipeOp$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":"Unsqueezes tensor calling torch::torch_unsqueeze() given dimension dim.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"nn-module","dir":"Reference","previous_headings":"","what":"nn_module","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"Calls nn_unsqueeze() trained. internally calls torch::torch_unsqueeze().","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":"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 $shapes_out().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchUnsqueeze","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"mlr3pipelines::PipeOp$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 #>  100.8678 #>  100.6844 #>   99.8373 #>   98.2156 #>   98.9629 #>  100.8301 #>  100.6073 #>   99.8778 #>  100.9331 #>   99.0387 #> [ CPUFloatType{10,1} ] #>  #> $x2 #> torch_tensor #>  100.2551 #>   99.4546 #>  100.9304 #>   99.4623 #>   99.5476 #>   99.5607 #>   99.3838 #>  100.4416 #>  100.4826 #>  100.5421 #> [ CPUFloatType{10,1} ] #>   taskout_predict_noaug = po_simple$predict(list(taskin))[[1L]] materialize(taskout_predict_noaug$data(cols = c(\"x1\", \"x2\")), rbind = TRUE) #> $x1 #> torch_tensor #>  100.8678 #>  100.6844 #>   99.8373 #>   98.2156 #>   98.9629 #>  100.8301 #>  100.6073 #>   99.8778 #>  100.9331 #>   99.0387 #> [ CPUFloatType{10,1} ] #>  #> $x2 #> torch_tensor #>  100.2551 #>   99.4546 #>  100.9304 #>   99.4623 #>   99.5476 #>   99.5607 #>   99.3838 #>  100.4416 #>  100.4826 #>  100.5421 #> [ CPUFloatType{10,1} ] #>   po_simple$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 #>  100.8678 #>  100.6844 #>   99.8373 #>   98.2156 #>   98.9629 #>  100.8301 #>  100.6073 #>   99.8778 #>  100.9331 #>   99.0387 #> [ CPUFloatType{10,1} ] #>  #> $x2 #> torch_tensor #>  100.2551 #>   99.4546 #>  100.9304 #>   99.4623 #>   99.5476 #>   99.5607 #>   99.3838 #>  100.4416 #>  100.4826 #>  100.5421 #> [ CPUFloatType{10,1} ] #>   # Creating a more complex preprocessing PipeOp PipeOpPreprocTorchPoly = R6::R6Class(\"PipeOpPreprocTorchPoly\",  inherit = PipeOpTaskPreprocTorch,  public = list(    initialize = function(id = \"preproc_poly\", param_vals = list()) {      param_set = paradox::ps(        n_degree = paradox::p_int(lower = 1L, tags = c(\"train\", \"required\"))      )      param_set$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) #> $x3 #> torch_tensor #>     1     1     1 #>     2     4     8 #>     3     9    27 #>     4    16    64 #>     5    25   125 #>     6    36   216 #>     7    49   343 #>     8    64   512 #>     9    81   729 #>    10   100  1000 #> [ CPUFloatType{10,3} ] #>"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":null,"dir":"Reference","previous_headings":"","what":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"PipeOpTorch base class PipeOps represent neural network layers Graph. training, generates PipeOpModule wraps nn_module attaches architecture, also represented Graph consisting mostly PipeOpModules PipeOpNOPs. former Graph operates ModelDescriptors, latter operates tensors. relationship PipeOpTorch PipeOpModule similar relationshop nn_module_generator (like nn_linear) nn_module (like output nn_linear(...)). crucial difference PipeOpTorch infers auxiliary parameters (like in_features nn_linear) automatically intermediate tensor shapes communicated ModelDescriptor. prediction, PipeOpTorch takes Task channel outputs new Task resulting feature union channel. one input output channel, task simply piped .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"ParamSet specified child class inheriting PipeOpTorch. Usually parameters arguments wrapped nn_module minus auxiliary parameter can automatically inferred shapes input tensors.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"inheriting","dir":"Reference","previous_headings":"","what":"Inheriting","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"inheriting class, one overload either private$.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":"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":"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":"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":"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":"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":"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":"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() | NULL | \"infer\" shape tensor, first dimension (batch) must NA. specified, lazy tensor input column needs known shape. set \"infer\", shape inferred example batch.","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$data(1, \"x\"), cache = NULL) x_batch #> torch_tensor #>  5.1000  3.5000  1.4000  0.2000 #> [ CPUFloatType{1,4} ]  # Now we try a lazy tensor with unknown shape, i.e. the shapes between the rows can differ  ds = dataset(   initialize = function() self$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\"),   cache = NULL )  x_batch2 #> torch_tensor #> (1,1,.,.) =  #>   0.0584  1.1540 -0.6838 -0.1345  0.7166 -0.1001 #>   0.1030 -1.1843  0.4625 -1.2743  0.3107  0.6001 #>  -0.4018  0.1173  0.2082  0.5757  0.4001 -0.1134 #>   0.2773 -0.1097 -0.0896 -0.3457  0.4073  0.1018 #>  -1.1756  0.0686  0.6384 -1.0616 -0.8770 -0.7203 #>   1.2721 -0.1520 -0.8474 -0.3209 -0.4132 -0.5879 #>  #> (2,1,.,.) =  #>   0.2584  0.3291  0.8919 -0.0781 -0.4938 -0.3368 #>   0.4498  0.5942 -0.1619 -1.5562  0.2340  0.4918 #>   0.6124  0.7222 -0.5876 -0.8403 -0.0998 -0.7715 #>   0.0641 -0.1189 -0.2512  0.2063  0.2547 -0.1059 #>   0.3604 -0.1723 -0.0245 -0.1846 -0.2950  0.3261 #>   1.2149  0.8136 -0.1479 -0.5586 -0.1958  0.2617 #>  #> (1,2,.,.) =  #>  -0.1901  0.2382  0.7609  0.7862  1.0801  0.0396 #>  -0.8384 -0.1858  0.2758 -0.3891 -0.2032 -1.0331 #>   0.2101 -1.2954 -0.5540 -0.9116  1.1030 -0.3491 #>   0.2743  0.2570 -0.3152  1.1351 -0.6616  0.6196 #>  -0.0025  1.0044  0.2625  1.3346  1.5293 -1.8108 #>  -0.2873  0.7334 -0.2273 -0.4304 -0.0558 -0.2041 #>  #> (2,2,.,.) =  #>  -1.0807  1.0639 -0.4367 -0.3011  0.1753 -0.9030 #>  -0.7590  0.1862  0.3582  0.6190  0.1483 -0.4890 #>  -0.1560 -0.7729  0.8068  0.1980  0.1058  0.7656 #>   0.6955  1.2231  0.6754  0.8966 -0.4761 -0.2140 #>   0.6465  0.6770 -0.0757 -1.0318  0.2228  0.0925 #> ... [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$loss #> <TorchLoss:cross_entropy> Cross Entropy #> * Generator: nn_cross_entropy_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: classif"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOp Torch Model — mlr_pipeops_torch_model","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"Builds Torch Learner ModelDescriptor trains given parameter specification. task type must specified construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"General: parameters optimizer, loss callbacks, prefixed \"opt.\", \"loss.\" \"cb.<callback id>.\" respectively, well : epochs :: integer(1) number epochs. device :: character(1) device. One \"auto\", \"cpu\", \"cuda\" values defined mlr_reflections$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. tensor_dataset :: logical(1) | \"device\" Whether load batches beginning training stack . initialized FALSE. set \"device\", device tensors set value device, can avoid unnecessary moving tensors devices. dataset fits memory make loading batches faster. Note set datasets contain lazy_tensors random data augmentation, augmentation applied beginning training. 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. set, epoch improvement observed, can accessed via $internal_tuned_values field learner. 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. initialized TRUE, differs default (FALSE). 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":"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":"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]]$weight_decay #> [1] 0 #>  #> $optimizer$param_groups[[1]]$betas #> [1] 0.900 0.999 #>  #> $optimizer$param_groups[[1]]$eps #> [1] 1e-08 #>  #> $optimizer$param_groups[[1]]$amsgrad #> [1] FALSE #>  #>  #>  #> $optimizer$state #> $optimizer$state$`1` #> $optimizer$state$`1`$exp_avg #> torch_tensor #> -0.1260 -0.0209 -0.4384 -0.3011 #> -0.1005 -0.0250 -0.1186 -0.0357 #>  0.2265  0.0458  0.5570  0.3368 #> [ CPUFloatType{3,4} ] #>  #> $optimizer$state$`1`$exp_avg_sq #> torch_tensor #> 0.01 * #>  0.0645  0.0018  0.7808  0.3683 #>   0.0702  0.0056  0.1150  0.0182 #>   0.2329  0.0109  1.2956  0.4669 #> [ CPUFloatType{3,4} ] #>  #> $optimizer$state$`1`$max_exp_avg_sq #> torch_tensor #> [ CPUFloatType{0} ] #>  #> $optimizer$state$`1`$step #> torch_tensor #>  3 #> [ CPULongType{1} ] #>  #>  #> $optimizer$state$`2` #> $optimizer$state$`2`$exp_avg #> torch_tensor #> -0.0876 #> -0.0196 #>  0.1072 #> [ CPUFloatType{3} ] #>  #> $optimizer$state$`2`$exp_avg_sq #> torch_tensor #> 0.0001 * #>  3.1150 #>  0.3254 #>  4.7708 #> [ CPUFloatType{3} ] #>  #> $optimizer$state$`2`$max_exp_avg_sq #> torch_tensor #> [ CPUFloatType{0} ] #>  #> $optimizer$state$`2`$step #> torch_tensor #>  3 #> [ CPULongType{1} ] #>  #>  #>  #>  #> $epochs #> [1] 1 #>  #> $callbacks #> named list() #>  #> $seed #> [1] 1338150431 #>  #> $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] 0.051 #>  #> $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\" #>  #> $param_vals$jit_trace #> [1] FALSE #>  #> $param_vals$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 #>  #> $param_vals$shuffle #> [1] TRUE #>  #> $param_vals$tensor_dataset #> [1] FALSE #>  #>  #> $task_hash #> [1] \"60543108f51f96bb\" #>  #> $feature_names #> [1] \"Petal.Length\" \"Petal.Width\"  \"Sepal.Length\" \"Sepal.Width\"  #>  #> $validate #> NULL #>  #> $mlr3_version #> [1] ‘0.22.1’ #>  #> $internal_tuned_values #> named list() #>  #> $data_prototype #> Empty data.table (0 rows and 5 cols): Species,Petal.Length,Petal.Width,Sepal.Length,Sepal.Width #>  #> $task_prototype #> Empty data.table (0 rows and 5 cols): Species,Petal.Length,Petal.Width,Sepal.Length,Sepal.Width #>  #> $train_task #> <TaskClassif:iris> (150 x 5): Iris Flowers #> * Target: Species #> * Properties: multiclass #> * Features (4): #>   - dbl (4): Petal.Length, Petal.Width, Sepal.Length, Sepal.Width #>  #> attr(,\"class\") #> [1] \"learner_state\" \"list\""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch Regression Model — mlr_pipeops_torch_model_regr","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"Builds torch regression model trains .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"See LearnerTorch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"one input channel \"input\" takes ModelDescriptor traing Task specified task_type prediction. output NULL training Prediction given task_type prediction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"trained LearnerTorchModel.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"LearnerTorchModel created calling model_descriptor_to_learner() provided ModelDescriptor received input channel. parameters set according parameters specified PipeOpTorchModel '$train() method called [Task][mlr3::Task] stored [ModelDescriptor`].","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"mlr3pipelines::PipeOp -> mlr3pipelines::PipeOpLearner -> mlr3torch::PipeOpTorchModel -> PipeOpTorchModelRegr","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"PipeOpTorchModelRegr$new() PipeOpTorchModelRegr$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"","code":"PipeOpTorchModelRegr$new(id = \"torch_model_regr\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"","code":"PipeOpTorchModelRegr$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"","code":"# simple linear regression  # build the model descriptor md = as_graph(po(\"torch_ingress_num\") %>>%   po(\"nn_head\") %>>%   po(\"torch_loss\", \"mse\") %>>%   po(\"torch_optimizer\", \"adam\"))$train(tsk(\"mtcars\"))[[1L]]  print(md) #> <ModelDescriptor: 2 ops> #> * Ingress:  torch_ingress_num.input: [(NA,10)] #> * Task:  mtcars [regr] #> * Callbacks:  N/A #> * Optimizer:  Adaptive Moment Estimation #> * Loss:  Mean Squared Error #> * pointer:  nn_head.output [(NA,1)]  # build the learner from the model descriptor and train it po_model = po(\"torch_model_regr\", batch_size = 20, epochs = 1) po_model$train(list(md)) #> $output #> NULL #>  po_model$state #> $model #> $network #> An `nn_module` containing 11 parameters. #>  #> ── Modules ───────────────────────────────────────────────────────────────────── #> • module_list: <nn_module_list> #11 parameters #>  #> $internal_valid_scores #> NULL #>  #> $loss_fn #> list() #>  #> $optimizer #> $optimizer$param_groups #> $optimizer$param_groups[[1]] #> $optimizer$param_groups[[1]]$params #> [1] 1 2 #>  #> $optimizer$param_groups[[1]]$lr #> [1] 0.001 #>  #> $optimizer$param_groups[[1]]$weight_decay #> [1] 0 #>  #> $optimizer$param_groups[[1]]$betas #> [1] 0.900 0.999 #>  #> $optimizer$param_groups[[1]]$eps #> [1] 1e-08 #>  #> $optimizer$param_groups[[1]]$amsgrad #> [1] FALSE #>  #>  #>  #> $optimizer$state #> $optimizer$state$`1` #> $optimizer$state$`1`$exp_avg #> torch_tensor #> -2.9507 -4.8365 -12.6544 -136.2684 -14.7779 -14.6720 -174.2433 -67.4318 -3.6596 -6.2324 #> [ CPUFloatType{1,10} ] #>  #> $optimizer$state$`1`$exp_avg_sq #> torch_tensor #> Columns 1 to 7    0.5200     2.3228    12.1758  6475.4131    13.2938    13.9329  4650.6953 #>  #> Columns 8 to 10  285.8094     0.7534     3.3957 #> [ CPUFloatType{1,10} ] #>  #> $optimizer$state$`1`$max_exp_avg_sq #> torch_tensor #> [ CPUFloatType{0} ] #>  #> $optimizer$state$`1`$step #> torch_tensor #>  2 #> [ CPULongType{1} ] #>  #>  #> $optimizer$state$`2` #> $optimizer$state$`2`$exp_avg #> torch_tensor #> -3.5004 #> [ CPUFloatType{1} ] #>  #> $optimizer$state$`2`$exp_avg_sq #> torch_tensor #>  0.7837 #> [ CPUFloatType{1} ] #>  #> $optimizer$state$`2`$max_exp_avg_sq #> torch_tensor #> [ CPUFloatType{0} ] #>  #> $optimizer$state$`2`$step #> torch_tensor #>  2 #> [ CPULongType{1} ] #>  #>  #>  #>  #> $epochs #> [1] 1 #>  #> $callbacks #> named list() #>  #> $seed #> [1] 1879465332 #>  #> $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] 0.038 #>  #> $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\" #>  #> $param_vals$jit_trace #> [1] FALSE #>  #> $param_vals$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 #>  #> $param_vals$shuffle #> [1] TRUE #>  #> $param_vals$tensor_dataset #> [1] FALSE #>  #>  #> $task_hash #> [1] \"907e8c646e24bf41\" #>  #> $feature_names #>  [1] \"am\"   \"carb\" \"cyl\"  \"disp\" \"drat\" \"gear\" \"hp\"   \"qsec\" \"vs\"   \"wt\"   #>  #> $validate #> NULL #>  #> $mlr3_version #> [1] ‘0.22.1’ #>  #> $internal_tuned_values #> named list() #>  #> $data_prototype #> Empty data.table (0 rows and 11 cols): mpg,am,carb,cyl,disp,drat... #>  #> $task_prototype #> Empty data.table (0 rows and 11 cols): mpg,am,carb,cyl,disp,drat... #>  #> $train_task #> <TaskRegr:mtcars> (32 x 11): Motor Trends #> * Target: mpg #> * Properties: - #> * Features (10): #>   - dbl (10): am, carb, cyl, disp, drat, gear, hp, qsec, vs, wt #>  #> attr(,\"class\") #> [1] \"learner_state\" \"list\""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer Configuration — mlr_pipeops_torch_optimizer","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"Configures optimizer deep learning model.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"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":"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":"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_ignite_sgd #> * Parameters: lr=0.01 #> * Packages: torch,mlr3torch"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_center_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"Center Crop Augmentation — mlr_pipeops_augment_center_crop","title":"Center Crop Augmentation — mlr_pipeops_augment_center_crop","text":"Calls torchvision::transform_center_crop, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_center_crop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Center Crop Augmentation — mlr_pipeops_augment_center_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_color_jitter.html","id":null,"dir":"Reference","previous_headings":"","what":"Color Jitter Augmentation — mlr_pipeops_augment_color_jitter","title":"Color Jitter Augmentation — mlr_pipeops_augment_color_jitter","text":"Calls torchvision::transform_color_jitter, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_color_jitter.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Color Jitter Augmentation — mlr_pipeops_augment_color_jitter","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"Crop Augmentation — mlr_pipeops_augment_crop","title":"Crop Augmentation — mlr_pipeops_augment_crop","text":"Calls torchvision::transform_crop, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_crop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Crop Augmentation — mlr_pipeops_augment_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_hflip.html","id":null,"dir":"Reference","previous_headings":"","what":"Horizontal Flip Augmentation — mlr_pipeops_augment_hflip","title":"Horizontal Flip Augmentation — mlr_pipeops_augment_hflip","text":"Calls torchvision::transform_hflip, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_hflip.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Horizontal Flip Augmentation — mlr_pipeops_augment_hflip","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_affine.html","id":null,"dir":"Reference","previous_headings":"","what":"Random Affine Augmentation — mlr_pipeops_augment_random_affine","title":"Random Affine Augmentation — mlr_pipeops_augment_random_affine","text":"Calls torchvision::transform_random_affine, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_affine.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Random Affine Augmentation — mlr_pipeops_augment_random_affine","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_choice.html","id":null,"dir":"Reference","previous_headings":"","what":"Random Choice Augmentation — mlr_pipeops_augment_random_choice","title":"Random Choice Augmentation — mlr_pipeops_augment_random_choice","text":"Calls torchvision::transform_random_choice, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_choice.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Random Choice Augmentation — mlr_pipeops_augment_random_choice","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"Random Crop Augmentation — mlr_pipeops_augment_random_crop","title":"Random Crop Augmentation — mlr_pipeops_augment_random_crop","text":"Calls torchvision::transform_random_crop, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_crop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Random Crop Augmentation — mlr_pipeops_augment_random_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_horizontal_flip.html","id":null,"dir":"Reference","previous_headings":"","what":"Random Horizontal Flip Augmentation — mlr_pipeops_augment_random_horizontal_flip","title":"Random Horizontal Flip Augmentation — mlr_pipeops_augment_random_horizontal_flip","text":"Calls torchvision::transform_random_horizontal_flip, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_horizontal_flip.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Random Horizontal Flip Augmentation — mlr_pipeops_augment_random_horizontal_flip","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_order.html","id":null,"dir":"Reference","previous_headings":"","what":"Random Order Augmentation — mlr_pipeops_augment_random_order","title":"Random Order Augmentation — mlr_pipeops_augment_random_order","text":"Calls torchvision::transform_random_order, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_order.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Random Order Augmentation — mlr_pipeops_augment_random_order","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_resized_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"Random Resized Crop Augmentation — mlr_pipeops_augment_random_resized_crop","title":"Random Resized Crop Augmentation — mlr_pipeops_augment_random_resized_crop","text":"Calls torchvision::transform_random_resized_crop, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_resized_crop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Random Resized Crop Augmentation — mlr_pipeops_augment_random_resized_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_vertical_flip.html","id":null,"dir":"Reference","previous_headings":"","what":"Random Vertical Flip Augmentation — mlr_pipeops_augment_random_vertical_flip","title":"Random Vertical Flip Augmentation — mlr_pipeops_augment_random_vertical_flip","text":"Calls torchvision::transform_random_vertical_flip, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_random_vertical_flip.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Random Vertical Flip Augmentation — mlr_pipeops_augment_random_vertical_flip","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_resized_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"Resized Crop Augmentation — mlr_pipeops_augment_resized_crop","title":"Resized Crop Augmentation — mlr_pipeops_augment_resized_crop","text":"Calls torchvision::transform_resized_crop, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_resized_crop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Resized Crop Augmentation — mlr_pipeops_augment_resized_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_rotate.html","id":null,"dir":"Reference","previous_headings":"","what":"Rotate Augmentation — mlr_pipeops_augment_rotate","title":"Rotate Augmentation — mlr_pipeops_augment_rotate","text":"Calls torchvision::transform_rotate, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_rotate.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Rotate Augmentation — mlr_pipeops_augment_rotate","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_vflip.html","id":null,"dir":"Reference","previous_headings":"","what":"Vertical Flip Augmentation — mlr_pipeops_augment_vflip","title":"Vertical Flip Augmentation — mlr_pipeops_augment_vflip","text":"Calls torchvision::transform_vflip, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopsaugment_vflip.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Vertical Flip Augmentation — mlr_pipeops_augment_vflip","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_adjust_brightness.html","id":null,"dir":"Reference","previous_headings":"","what":"Adjust Brightness Transformation — mlr_pipeops_trafo_adjust_brightness","title":"Adjust Brightness Transformation — mlr_pipeops_trafo_adjust_brightness","text":"Calls torchvision::transform_adjust_brightness, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_adjust_brightness.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Adjust Brightness Transformation — mlr_pipeops_trafo_adjust_brightness","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_adjust_gamma.html","id":null,"dir":"Reference","previous_headings":"","what":"Adjust Gamma Transformation — mlr_pipeops_trafo_adjust_gamma","title":"Adjust Gamma Transformation — mlr_pipeops_trafo_adjust_gamma","text":"Calls torchvision::transform_adjust_gamma, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_adjust_gamma.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Adjust Gamma Transformation — mlr_pipeops_trafo_adjust_gamma","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_adjust_hue.html","id":null,"dir":"Reference","previous_headings":"","what":"Adjust Hue Transformation — mlr_pipeops_trafo_adjust_hue","title":"Adjust Hue Transformation — mlr_pipeops_trafo_adjust_hue","text":"Calls torchvision::transform_adjust_hue, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_adjust_hue.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Adjust Hue Transformation — mlr_pipeops_trafo_adjust_hue","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_adjust_saturation.html","id":null,"dir":"Reference","previous_headings":"","what":"Adjust Saturation Transformation — mlr_pipeops_trafo_adjust_saturation","title":"Adjust Saturation Transformation — mlr_pipeops_trafo_adjust_saturation","text":"Calls torchvision::transform_adjust_saturation, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_adjust_saturation.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Adjust Saturation Transformation — mlr_pipeops_trafo_adjust_saturation","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_grayscale.html","id":null,"dir":"Reference","previous_headings":"","what":"Grayscale Transformation — mlr_pipeops_trafo_grayscale","title":"Grayscale Transformation — mlr_pipeops_trafo_grayscale","text":"Calls torchvision::transform_grayscale, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_grayscale.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Grayscale Transformation — mlr_pipeops_trafo_grayscale","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_normalize.html","id":null,"dir":"Reference","previous_headings":"","what":"Normalization Transformation — mlr_pipeops_trafo_normalize","title":"Normalization Transformation — mlr_pipeops_trafo_normalize","text":"Calls torchvision::transform_normalize, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_normalize.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Normalization Transformation — mlr_pipeops_trafo_normalize","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_pad.html","id":null,"dir":"Reference","previous_headings":"","what":"Padding Transformation — mlr_pipeops_trafo_pad","title":"Padding Transformation — mlr_pipeops_trafo_pad","text":"Calls torchvision::transform_pad, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_pad.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Padding Transformation — mlr_pipeops_trafo_pad","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_resize.html","id":null,"dir":"Reference","previous_headings":"","what":"Resizing Transformation — mlr_pipeops_trafo_resize","title":"Resizing Transformation — 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/mlr_pipeopstrafo_resize.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Resizing Transformation — mlr_pipeops_trafo_resize","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_rgb_to_grayscale.html","id":null,"dir":"Reference","previous_headings":"","what":"RGB to Grayscale Transformation — mlr_pipeops_trafo_rgb_to_grayscale","title":"RGB to Grayscale Transformation — mlr_pipeops_trafo_rgb_to_grayscale","text":"Calls torchvision::transform_rgb_to_grayscale, see information parameters. preprocessing applied element batch individually.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeopstrafo_rgb_to_grayscale.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"RGB to Grayscale Transformation — mlr_pipeops_trafo_rgb_to_grayscale","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_cifar.html","id":null,"dir":"Reference","previous_headings":"","what":"CIFAR Classification Tasks — mlr_tasks_cifar","title":"CIFAR Classification Tasks — mlr_tasks_cifar","text":"CIFAR-10 CIFAR-100 datasets. subset 80 million tiny images dataset noisy labels supplied student labelers, asked filter incorrectly labeled images. images datatype torch_long(). CIFAR-10 contains 10 classes. CIFAR-100 contains 100 classes, may partitioned 20 superclasses 5 classes . CIFAR-10 CIFAR-100 classes mutually exclusive. See Chapter 3.1 technical report details. data obtained torchvision::cifar10_dataset() (torchvision::cifar100_dataset()).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_cifar.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"CIFAR Classification Tasks — mlr_tasks_cifar","text":"R6::R6Class inheriting mlr3::TaskClassif.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_cifar.html","id":"construction","dir":"Reference","previous_headings":"","what":"Construction","title":"CIFAR Classification Tasks — mlr_tasks_cifar","text":"","code":"tsk(\"cifar10\") tsk(\"cifar100\")"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_cifar.html","id":"download","dir":"Reference","previous_headings":"","what":"Download","title":"CIFAR Classification Tasks — mlr_tasks_cifar","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_cifar.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"CIFAR Classification Tasks — mlr_tasks_cifar","text":"Task type: “classif” Properties: “multiclass” Missings: Target: “class” Features: “image” Data Dimension: 60000x4","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_cifar.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"CIFAR Classification Tasks — mlr_tasks_cifar","text":"Krizhevsky, Alex (2009). “Learning Multiple Layers Features Tiny Images.” Master's thesis, Department Computer Science, University Toronto.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_cifar.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"CIFAR Classification Tasks — mlr_tasks_cifar","text":"","code":"task_cifar10 = tsk(\"cifar10\") task_cifar100 = tsk(\"cifar100\") print(task_cifar10) #> <TaskClassif:cifar10> (60000 x 2): CIFAR-10 Classification #> * Target: class #> * Properties: multiclass #> * Features (1): #>   - lt (1): image print(task_cifar100) #> <TaskClassif:cifar100> (60000 x 2): CIFAR-100 Classification #> * Target: class #> * Properties: multiclass #> * Features (1): #>   - lt (1): image"},{"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_melanoma.html","id":null,"dir":"Reference","previous_headings":"","what":"Melanoma Image classification — mlr_tasks_melanoma","title":"Melanoma Image classification — mlr_tasks_melanoma","text":"Classification melanoma tumor images. data preprocessed version 2020 SIIM-ISIC challenge images reshaped size $(3, 128, 128)$. default training rows active task, test data (targets) also included. Whether observation part train test set indicated column \"test\". labels test rows, default, observations inactive, means task uses 32701 43683 observations defined underlying data backend. data backend also contains detailed diagnosis specific type tumor. Columns: outcome (factor): target variable. Whether tumor benign malignant (positive class) anatom_site_general_challenge (factor): location tumor patient's body sex (factor): sex patient age_approx (int): approximate age patient time imaging image (lazy_tensor): image (shape $(3, 128, 128)$) tumor. ee split (character): Whether observation os part train test set.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_melanoma.html","id":"source","dir":"Reference","previous_headings":"","what":"Source","title":"Melanoma Image classification — mlr_tasks_melanoma","text":"https://huggingface.co/datasets/carsonzhang/ISIC_2020_small","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_melanoma.html","id":"construction","dir":"Reference","previous_headings":"","what":"Construction","title":"Melanoma Image classification — mlr_tasks_melanoma","text":"","code":"tsk(\"melanoma\")"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_melanoma.html","id":"download","dir":"Reference","previous_headings":"","what":"Download","title":"Melanoma Image classification — mlr_tasks_melanoma","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_melanoma.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"Melanoma Image classification — mlr_tasks_melanoma","text":"Task type: “classif” Properties: “twoclass”, “groups” Missings: Target: “outcome” Features: “sex”, “anatom_site_general_challenge”, “age_approx”, “image” Data Dimension: 43683x11","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_melanoma.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Melanoma Image classification — mlr_tasks_melanoma","text":"Rotemberg, V., Kurtansky, N., Betz-Stablein, B., Caffery, L., Chousakos, E., Codella, N., Combalia, M., Dusza, S., Guitera, P., Gutman, D., Halpern, ., Helba, B., Kittler, H., Kose, K., Langer, S., Lioprys, K., Malvehy, J., Musthaq, S., Nanda, J., Reiter, O., Shih, G., Stratigos, ., Tschandl, P., Weber, J., Soyer, P. (2021). “patient-centric dataset images metadata identifying melanomas using clinical context.” Scientific Data, 8, 34. doi:10.1038/s41597-021-00815-z .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_melanoma.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Melanoma Image classification — mlr_tasks_melanoma","text":"","code":"task = tsk(\"melanoma\") task #> <TaskClassif:melanoma> (32701 x 5): Melanoma Classification #> * Target: outcome #> * Properties: twoclass, groups #> * Features (4): #>   - fct (2): anatom_site_general_challenge, sex #>   - int (1): age_approx #>   - lt (1): image #> * Groups: patient_id"},{"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 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 #> -0.2769 #> -0.2779 #> -0.1570 #> -0.2491 #> -0.3878 #> -0.1224 #> -0.4271 #> -0.2581 #> -0.3228 #> -0.0445 #> -0.3487 #> -0.6967 #> -0.0038 #> -0.5004 #> -0.4031 #> -0.4273 #> [ 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_ignite_adam #> * Parameters: lr=0.1 #> * Packages: torch,mlr3torch # get the dictionary t_opt() #> <DictionaryMlr3torchOptimizers> with 5 stored values #> Keys: adagrad, adam, adamw, rmsprop, sgd t_opts(c(\"adam\", \"sgd\")) #> $adam #> <TorchOptimizer:adam> Adaptive Moment Estimation #> * Generator: optim_ignite_adam #> * Parameters: list() #> * Packages: torch,mlr3torch #>  #> $sgd #> <TorchOptimizer:sgd> Stochastic Gradient Descent #> * Generator: optim_ignite_sgd #> * Parameters: list() #> * Packages: torch,mlr3torch #>  # get the dictionary t_opts() #> <DictionaryMlr3torchOptimizers> with 5 stored values #> Keys: adagrad, adam, adamw, rmsprop, 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)"},{"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 $x value batch corresponding names. target_batchgetter (function(data, device)) function taking arguments data, data.table containing target variable, device. must return target torch tensor selected device.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/task_dataset.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a Dataset from a Task — task_dataset","text":"torch::dataset","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/task_dataset.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a Dataset from a Task — task_dataset","text":"","code":"task = tsk(\"iris\") sepal_ingress = TorchIngressToken(   features = c(\"Sepal.Length\", \"Sepal.Width\"),   batchgetter = batchgetter_num,   shape = c(NA, 2) ) petal_ingress = TorchIngressToken(   features = c(\"Petal.Length\", \"Petal.Width\"),   batchgetter = batchgetter_num,   shape = c(NA, 2) ) ingress_tokens = list(sepal = sepal_ingress, petal = petal_ingress)  target_batchgetter = function(data) {   torch_tensor(data = data[[1L]], dtype = torch_float32())$unsqueeze(2) } dataset = task_dataset(task, ingress_tokens, target_batchgetter) 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 Descriptor — torch_callback","title":"Create a Callback Descriptor — 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 Descriptor — 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 Descriptor — 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 Descriptor — 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 Descriptor — 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 Descriptor — 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 Descriptor — 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":[]},{"path":"https://mlr3torch.mlr-org.com/dev/news/index.html","id":"breaking-changes-0-2-0","dir":"Changelog","previous_headings":"","what":"Breaking Changes","title":"mlr3torch 0.2.0","text":"Removed optimizers fast (‘ignite’) variant exists. default optimizer now AdamW instead Adam. private LearnerTorch$.dataloader() method now operates longer task dataset generated private LearnerTorch$.dataset() method. shuffle parameter model training now initialized TRUE sidestep issues data sorted.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/news/index.html","id":"performance-improvements-0-2-0","dir":"Changelog","previous_headings":"","what":"Performance Improvements","title":"mlr3torch 0.2.0","text":"Optimizers now use faster (‘ignite’) version optimizers, leads considerable speed improvements. jit_trace parameter added LearnerTorch, set TRUE can lead significant speedups. enabled ‘static’ models, see torch tutorial information. Added parameter num_interop_threads LearnerTorch. tensor_dataset parameter added, allows stack batches beginning training make loading batches afterwards faster. Use faster default image loader.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/news/index.html","id":"features-0-2-0","dir":"Changelog","previous_headings":"","what":"Features","title":"mlr3torch 0.2.0","text":"Added PipeOp adaptive average pooling. n_layers parameter added MLP learner. Added multimodal melanoma cifar{10, 100} example tasks. Added callback iteratively unfreeze parameters finetuning. Added different learning rate schedulers callbacks.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/news/index.html","id":"bug-fixes-0-2-0","dir":"Changelog","previous_headings":"","what":"Bug Fixes:","title":"mlr3torch 0.2.0","text":"Torch learners can now used AutoTuner. Early stopping now uses epochs - patience internally tuned values instead trained number epochs . dataset learner must longer return tensors specified device, allows parallel dataloading GPUs. PipeOpBlock longer create ID clashes PipeOps graph (#260).","code":""},{"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":""}]