B_Apprendre.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/laurentperrinet/2021_UE-neurosciences-computationnelles/blob/master/B_Apprendre.ipynb)\n",
    "\n",
    "# Classification d'image : apprentissage du réseau\n",
    "\n",
    "Dans ce tutoriel nous allons étudier plus en détail comment construire un réseau simple à l'aide de pytorch : [LeNet](http://yann.lecun.com/exdb/publis/pdf/lecun-01a.pdf) créé par Yann LeCun.\n",
    "Le code présent dans ce tutoriel est basé sur un [exemple](https://github.com/pytorch/examples/blob/master/mnist/main.py) écrit en langage `pytorch`.\n",
    "\n",
    "Ce réseau est un réseau convolutif qui à pour but de faire de la classification d'image.\n",
    "\n",
    "Ce [tuto](https://www.superdatascience.com/blogs/the-ultimate-guide-to-convolutional-neural-networks-cnn) vous permettra d'aller plus en détail dans l'explication de ce réseau"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "from torchvision import datasets, transforms\n",
    "\n",
    "torch.manual_seed(1)\n",
    "use_cuda = torch.cuda.is_available()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## étape 1 : définition des image train et eval"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Il est important pour commencer de définir les images sur lequel nous allons travailler.\n",
    "Ici, nous travaillerons sur le dataset MNIST, contenant des images de chiffres écrit à la main.\n",
    "\n",
    "![MNIST](https://knowm.org/wp-content/uploads/Screen-Shot-2015-08-14-at-2.44.57-PM.png)\n",
    "\n",
    "Le réseau aura pour but de déterminer quel chiffre est écrit sur l'image. Pour cela il vas faloir qu'il apprenne à classifier corectement les images avec un dataset d'entraînement. Il pourra ensuite tester son apprentissage avec un second dataset (dit de validation).\n",
    "\n",
    "(Ces deux dataset doivent être différents.)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# taille du batch d'entrainement\n",
    "train_batch_size = 64\n",
    "# taille du batch de test\n",
    "test_batch_size = 1000\n",
    "\n",
    "# transformation des datas\n",
    "transform = transforms.Compose([transforms.ToTensor(), transforms.Normalize((0.1307,), (0.3081,))])\n",
    "\n",
    "kwargs = {'num_workers': 1, 'pin_memory': True} if use_cuda else {}\n",
    "\n",
    "\n",
    "# importation du dataset\n",
    "dataset = datasets.MNIST('images', download=False, transform=transform)\n",
    "\n",
    "# nombre d'image pour l'entraînement (80% des données totales)\n",
    "nb_train = int(0.8*len(dataset))\n",
    "# nombre d'image pour le test (le reste des données)\n",
    "nb_test  = len(dataset)-nb_train\n",
    "\n",
    "# separation des données en deux datasets\n",
    "train_dataset, test_dataset = torch.utils.data.random_split(dataset, [nb_train, nb_test])\n",
    "\n",
    "\n",
    "# Chargement des données\n",
    "train_loader = torch.utils.data.DataLoader(dataset=train_dataset, batch_size=train_batch_size, shuffle=True, **kwargs)\n",
    "test_loader = torch.utils.data.DataLoader(dataset=test_dataset, batch_size=test_batch_size,  shuffle=True, **kwargs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Le set d'entraînement contient 48000 images, dans 750 batches de 64 images\n",
      "Le set de test contient 12000 images, dans 12 batches de 1000 images\n"
     ]
    }
   ],
   "source": [
    "print(\"Le set d'entraînement contient {} images, dans {} batches de {} images\".format(len(train_loader.dataset),\n",
    "                                                                                      len(train_loader),\n",
    "                                                                                      train_batch_size))\n",
    "print(\"Le set de test contient {} images, dans {} batches de {} images\".format(len(test_loader.dataset),\n",
    "                                                                               len(test_loader),\n",
    "                                                                               test_batch_size))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## étape 2 : Réseau"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Net(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(Net, self).__init__()\n",
    "        self.conv1 = nn.Conv2d(1, 20, 5, 1) # convolution 1\n",
    "        self.conv2 = nn.Conv2d(20, 50, 5, 1) # convolution 2\n",
    "        self.fc1 = nn.Linear(4*4*50, 500) # fully connected 1\n",
    "        self.fc2 = nn.Linear(500, 10) # fully connected 2\n",
    "\n",
    "    def forward(self, x):\n",
    "        # Extraction de caractéristiques 1\n",
    "        x = F.relu(self.conv1(x)) # activation\n",
    "        x = F.max_pool2d(x, 2, 2) # pooling\n",
    "        \n",
    "        # Extraction de caractéristiques 2\n",
    "        x = F.relu(self.conv2(x)) # activation\n",
    "        x = F.max_pool2d(x, 2, 2) # pooling\n",
    "        x = x.view(-1, 4*4*50) # flattening\n",
    "        \n",
    "        # Classification\n",
    "        x = F.relu(self.fc1(x)) # activation\n",
    "        x = self.fc2(x)\n",
    "        return F.log_softmax(x, dim=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "le réseau LeNet comprend deux principales étapes :\n",
    "- l'extraction des caractéristique de l'image\n",
    "- la classification de cette image en fonction de ses caractéristiques\n",
    "\n",
    "<img src=\"https://miro.medium.com/max/3712/1*7K4ZTTfZb-hbjoADbisHAg.png\" width=\"600\"/>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### rappels sur l'extraction des caractéristique de l'image"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "pour extraire ses caractéristiques l'image vas subir une série de modification : \n",
    "\n",
    "- une **convolution** (voir [Conv2d](https://pytorch.org/docs/stable/nn.html#conv2d), [paramêtres en image](https://github.com/vdumoulin/conv_arithmetic/blob/master/README.md))\n",
    "     - une convolution est une fonction dérivée de deux fonctions données par intégration qui exprime comment la forme de l'une est modifiée par l'autre\n",
    "     - l'image est convoluée par des détecteurs de caractéristiques (appelés kernel)\n",
    "     - Attention l'entrée chaque couche de convolution doit faire la même taille que la sortie de la couche de convolution précédante\n",
    "\n",
    "- une **fonction d'activation** : fonction d'Unité de Rectification Linéaire (voir [ReLu()](https://pytorch.org/docs/stable/nn.html#relu))\n",
    "    - pour ne garder que les valeurs positives dans les carte de caratéristiques engendré par la convolution\n",
    "    - ReLU(x)=max(0,x) <img src=\"https://pytorch.org/docs/stable/_images/ReLU.png\" width=\"400\"/>\n",
    "\n",
    "\n",
    "\n",
    "- un **Pooling** : Max Pooling (voir [MaxPool2d](https://pytorch.org/docs/stable/nn.html#maxpool2d))\n",
    "    - pour ne garder que les valeurs maximales dans une zone\n",
    "    - réduit les carte de caractéristique\n",
    "<img src=\"https://upload.wikimedia.org/wikipedia/commons/e/e9/Max_pooling.png\" width=\"300\"/>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Net()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1080 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1080 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "trial = 0\n",
    "first_featuremap = 0\n",
    "second_featuremap = 0\n",
    "\n",
    "from Model_show import Plot_model\n",
    "Plot_model(dataset, model.to('cpu'), trial, first_featuremap, second_featuremap)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Classification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- **fully connected** - Linear (voir [Linear](https://pytorch.org/docs/stable/nn.html#linear))\n",
    "    - c'est un réseau dans lequel tous les neurones de sorties sont connectés à tous les neurones d'entrée\n",
    "    - Applique une transformation linéaire aux données entrantes\n",
    "<img src=\"https://jdlm.info/assets/driverless/27-fully-connected.png\" width=\"900\"/>     \n",
    "\n",
    "- log_softmax (voir [log_softmax](https://pytorch.org/docs/stable/nn.functional.html#log-softmax))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## étape 3 : Entrainement du Réseau"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- définition **Fonction loss** + trouver la bonne ! [wiki](https://fr.wikipedia.org/wiki/R%C3%A9seau_neuronal_convolutif)\n",
    "\n",
    "    La couche de perte spécifie comment l'entrainement du réseau pénalise l'écart entre le signal prévu et réel. Elle est normalement la dernière couche dans le réseau.\n",
    "\n",
    "    Diverses fonctions de perte adaptées à différentes tâches peuvent y être utilisées [loss-functions](https://pytorch.org/docs/stable/nn.functional.html#loss-functions) :\n",
    "    - La perte « Softmax »est utilisée pour prédire une seule classe parmi K classes mutuellement exclusives.\n",
    "    - La perte par entropie croisée sigmoïde est utilisée pour prédire K valeurs de probabilité indépendante dans [0,1].\n",
    "    - La perte euclidienne est utilisée pour régresser vers des valeurs réelles dans [-inf ,inf].\n",
    "\n",
    "The negative log likelihood loss -> permet de maximiser l'erreur quand le bon label a une probabilité faible et de la diminuer quand bon label a probabilité élevé !\n",
    "\n",
    "- **Backpropagation**\n",
    "    - l'erreur calculer se propage dans les première couche du réseau\n",
    "<img src=\"https://d3i71xaburhd42.cloudfront.net/db39fd79bb591b04d33207992f6ccde03cabd861/7-Figure1-1.png\" width=\"400\"/>\n",
    "\n",
    "    - Pour effectuer la backpropagation en pytorch, les étapes suivantes sont nécessaires à chaque itération de la boucle :\n",
    "\n",
    "        1. **optimizer.zero_grad()** : remet à zero les gradiants de chaque paramètre.\n",
    "        2. **loss.backward()** : calcule les gradiants pour chaque variable par backpropagation en fonction de la loss, et les stocke dans l’objet Variable\n",
    "        3. **optimizer.step()** : Modifie chaque paramètre du modèle (poids des réseaux) de manière à minimiser la loss.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train(model, device, train_loader, optimizer, epoch, log_interval):\n",
    "    model.train() # mettre en mode entrainement\n",
    "    for batch_idx, (data, target) in enumerate(train_loader):\n",
    "        data, target = data.to(device), target.to(device)\n",
    "        \n",
    "        optimizer.zero_grad() # important ! remet les gradients à 0\n",
    "        output = model(data) # calcule la prédiction\n",
    "        loss = F.nll_loss(output, target) # calcule l'erreur : The negative log likelihood loss.\n",
    "        loss.backward() # dérive le graphe\n",
    "        optimizer.step() # effectue une étape d'optimisation\n",
    "        \n",
    "        if batch_idx % log_interval == 0:\n",
    "            print('Train Epoch: {} [{}/{} ({:.0f}%)]\\tLoss: {:.6f}'.format(\n",
    "                epoch, batch_idx * len(data), len(train_loader.dataset),\n",
    "                100. * batch_idx / len(train_loader), loss.item()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Epoch: 0 [0/48000 (0%)]\tLoss: 2.341338\n",
      "Train Epoch: 0 [47936/48000 (100%)]\tLoss: 0.065021\n"
     ]
    }
   ],
   "source": [
    "# on indique au modèle si on veux travailler sur un GPU ou un CPU\n",
    "device = torch.device(\"cuda\" if use_cuda else \"cpu\")\n",
    "model = Net().to(device)\n",
    "\n",
    "# l'optimiseur permet de mettre calculer les poids du réseau\n",
    "optimizer = optim.SGD(model.parameters(), lr=0.01, momentum=0.5)\n",
    "\n",
    "train(model, device, train_loader, optimizer, 0, 749)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## étape 3 : Test du Réseau"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def test(model, device, test_loader):\n",
    "    model.eval()\n",
    "    test_loss = 0\n",
    "    correct = 0\n",
    "    with torch.no_grad():\n",
    "        for data, target in test_loader:\n",
    "            data, target = data.to(device), target.to(device)\n",
    "            \n",
    "            output = model(data)\n",
    "            test_loss += F.nll_loss(output, target, reduction='sum').item() # sum up batch loss\n",
    "            pred = output.argmax(dim=1, keepdim=True) # get the index of the max log-probability\n",
    "            correct += pred.eq(target.view_as(pred)).sum().item()\n",
    "\n",
    "    test_loss /= len(test_loader.dataset)\n",
    "\n",
    "    print('\\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\\n'.format(\n",
    "        test_loss, correct, len(test_loader.dataset), 100. * correct/len(test_loader.dataset)))\n",
    "    return(100. * correct/len(test_loader.dataset))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "on vas tester le modèle pour 15 essais"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 5400x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "target : tensor([4, 2, 3, 0, 7, 2, 8, 1, 1, 8, 6, 6, 5, 4, 2]) \n",
      "prediction : tensor([4, 2, 3, 0, 7, 2, 8, 1, 1, 8, 6, 6, 5, 4, 2]) \n",
      "number of good predictions : 15 \n",
      "total number predictions : 15\n"
     ]
    }
   ],
   "source": [
    "N_test = 15\n",
    "dict_test = dict(iter(test_loader))\n",
    "inps = list(dict_test.keys())\n",
    "inp = inps[0][:N_test]\n",
    "targ = dict_test[inps[0]][:N_test]\n",
    "\n",
    "i = inp.view(inp.shape[0], 1, inp.shape[2], inp.shape[3])\n",
    "\n",
    "import torchvision.utils as vutils\n",
    "grid_images = vutils.make_grid(i, nrow=N_test)\n",
    "images = np.transpose(grid_images, (1,2,0))\n",
    "\n",
    "plt.figure(figsize=(5*N_test, 5))\n",
    "plt.imshow(images)\n",
    "plt.xticks([]) ; plt.yticks([])\n",
    "plt.show()\n",
    "\n",
    "\n",
    "out = model(i.to(device))\n",
    "pred = out.argmax(dim=1, keepdim=True)\n",
    "print('target :', targ,\n",
    "      '\\nprediction :', pred.to('cpu').view(-1),\n",
    "      '\\nnumber of good predictions :', pred.eq(targ.to(device).view_as(pred)).sum().item(),\n",
    "      '\\ntotal number predictions :', N_test)      "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## étape 4 : évolution de l'erreur au cour de l'apprentissage"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "on entraine le modèle sur plusieurs epoch en le testant pour chaque epoch afin de voir sont évolution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# on indique au modèle si on veux travailler sur un GPU ou un CPU\n",
    "device = torch.device(\"cuda\" if use_cuda else \"cpu\")\n",
    "model = Net().to(device)\n",
    "# l'optimiseur permet de mettre calculer les poids du réseau\n",
    "optimizer = optim.SGD(model.parameters(), lr=0.01, momentum=0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Net(\n",
       "  (conv1): Conv2d(1, 20, kernel_size=(5, 5), stride=(1, 1))\n",
       "  (conv2): Conv2d(20, 50, kernel_size=(5, 5), stride=(1, 1))\n",
       "  (fc1): Linear(in_features=800, out_features=500, bias=True)\n",
       "  (fc2): Linear(in_features=500, out_features=10, bias=True)\n",
       ")"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#########################################################\n",
    "# pour remmettre a zeros les poids dans le modèle !\n",
    "#########################################################\n",
    "def weight_reset(m):\n",
    "    if isinstance(m, nn.Conv2d) or isinstance(m, nn.Linear):\n",
    "        m.reset_parameters()\n",
    "\n",
    "model.apply(weight_reset)\n",
    "#########################################################"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Epoch: 1 [0/48000 (0%)]\tLoss: 2.270189\n",
      "Train Epoch: 1 [47936/48000 (100%)]\tLoss: 0.090869\n",
      "\n",
      "Test set: Average loss: 0.1265, Accuracy: 11522/12000 (96%)\n",
      "\n",
      "Train Epoch: 2 [0/48000 (0%)]\tLoss: 0.160347\n",
      "Train Epoch: 2 [47936/48000 (100%)]\tLoss: 0.207095\n",
      "\n",
      "Test set: Average loss: 0.0783, Accuracy: 11708/12000 (98%)\n",
      "\n",
      "Train Epoch: 3 [0/48000 (0%)]\tLoss: 0.090742\n",
      "Train Epoch: 3 [47936/48000 (100%)]\tLoss: 0.094107\n",
      "\n",
      "Test set: Average loss: 0.0675, Accuracy: 11758/12000 (98%)\n",
      "\n",
      "Train Epoch: 4 [0/48000 (0%)]\tLoss: 0.043757\n",
      "Train Epoch: 4 [47936/48000 (100%)]\tLoss: 0.125298\n",
      "\n",
      "Test set: Average loss: 0.0575, Accuracy: 11796/12000 (98%)\n",
      "\n",
      "Train Epoch: 5 [0/48000 (0%)]\tLoss: 0.171810\n",
      "Train Epoch: 5 [47936/48000 (100%)]\tLoss: 0.024727\n",
      "\n",
      "Test set: Average loss: 0.0498, Accuracy: 11811/12000 (98%)\n",
      "\n",
      "Train Epoch: 6 [0/48000 (0%)]\tLoss: 0.059161\n",
      "Train Epoch: 6 [47936/48000 (100%)]\tLoss: 0.043079\n",
      "\n",
      "Test set: Average loss: 0.0433, Accuracy: 11837/12000 (99%)\n",
      "\n",
      "Train Epoch: 7 [0/48000 (0%)]\tLoss: 0.017244\n",
      "Train Epoch: 7 [47936/48000 (100%)]\tLoss: 0.026257\n",
      "\n",
      "Test set: Average loss: 0.0429, Accuracy: 11846/12000 (99%)\n",
      "\n",
      "Train Epoch: 8 [0/48000 (0%)]\tLoss: 0.008195\n",
      "Train Epoch: 8 [47936/48000 (100%)]\tLoss: 0.004662\n",
      "\n",
      "Test set: Average loss: 0.0393, Accuracy: 11841/12000 (99%)\n",
      "\n",
      "Train Epoch: 9 [0/48000 (0%)]\tLoss: 0.008630\n",
      "Train Epoch: 9 [47936/48000 (100%)]\tLoss: 0.131144\n",
      "\n",
      "Test set: Average loss: 0.0382, Accuracy: 11858/12000 (99%)\n",
      "\n",
      "Train Epoch: 10 [0/48000 (0%)]\tLoss: 0.016940\n",
      "Train Epoch: 10 [47936/48000 (100%)]\tLoss: 0.027445\n",
      "\n",
      "Test set: Average loss: 0.0420, Accuracy: 11844/12000 (99%)\n",
      "\n",
      "Train Epoch: 11 [0/48000 (0%)]\tLoss: 0.012723\n",
      "Train Epoch: 11 [47936/48000 (100%)]\tLoss: 0.009604\n",
      "\n",
      "Test set: Average loss: 0.0358, Accuracy: 11866/12000 (99%)\n",
      "\n",
      "Train Epoch: 12 [0/48000 (0%)]\tLoss: 0.003491\n",
      "Train Epoch: 12 [47936/48000 (100%)]\tLoss: 0.009266\n",
      "\n",
      "Test set: Average loss: 0.0371, Accuracy: 11860/12000 (99%)\n",
      "\n",
      "Train Epoch: 13 [0/48000 (0%)]\tLoss: 0.001251\n",
      "Train Epoch: 13 [47936/48000 (100%)]\tLoss: 0.012451\n",
      "\n",
      "Test set: Average loss: 0.0416, Accuracy: 11855/12000 (99%)\n",
      "\n",
      "Train Epoch: 14 [0/48000 (0%)]\tLoss: 0.003026\n",
      "Train Epoch: 14 [47936/48000 (100%)]\tLoss: 0.000941\n",
      "\n",
      "Test set: Average loss: 0.0341, Accuracy: 11877/12000 (99%)\n",
      "\n",
      "Train Epoch: 15 [0/48000 (0%)]\tLoss: 0.002586\n",
      "Train Epoch: 15 [47936/48000 (100%)]\tLoss: 0.000469\n",
      "\n",
      "Test set: Average loss: 0.0348, Accuracy: 11892/12000 (99%)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "epochs = 15\n",
    "log_interval = 749\n",
    "\n",
    "list_accuracy = []\n",
    "for epoch in range(1, epochs + 1):\n",
    "    train(model, device, train_loader, optimizer, epoch, log_interval)\n",
    "    accuracy = test(model, device, test_loader)\n",
    "    list_accuracy.append(accuracy)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEKCAYAAAD9xUlFAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAAAcU0lEQVR4nO3dfXRV9Z3v8feHqDxELRBEhSQCReQhSwqkLqqFsaVaBVvFqZ3bCwojS8cFWr2rqzq2td6ZTqutfbC1s0btFBWfruNDb1tRi1pFbW+1QaUSUWp5kCACGlB5huR7/zg7e0II4QA52Xn4vNbKOmfvs8/Z3yjZn7N/v/37bUUEZmZmAN2yLsDMzNoPh4KZmaUcCmZmlnIomJlZyqFgZmYph4KZmaUKFgqS5kpaL2lJo3V9JT0p6a/JY59kvST9TNJbkv4iaWyh6jIzs30r5JnCncBZTdb9M/B0RJwIPJ0sA5wNnJj8XAr8RwHrMjOzfShYKETEc0Btk9XnAnclz+8Czmu0fl7k/AnoLen4QtVmZmbNO6yN93dsRKxNnr8LHJs8HwisbrRdTbJuLU1IupTc2QTFxcXjhg8fXrhqzcw6oUWLFr0XEcc091pbh0IqIkLSAc+xERG3A7cDVFZWRlVVVavXZmbWmUlata/X2vrqo3UNzULJ4/pk/RqgrNF2pck6MzNrQ20dCr8BZiTPZwC/brT+ouQqpPHAB42amczMrI0UrPlI0v3A6UA/STXA9cCNwH9JmgWsAr6cbP4YMBl4C9gK/GOh6jIzs30rWChExFf28dKkZrYNYE6hajEzs/x4RLOZmaUcCmZmlnIomJlZyqFgZmYph4KZmaUcCmZmlnIomJlZyqFgZmYph4KZmaUcCmZmlnIomJlZyqFgZmYph4KZmaUyu/OamZm1rh07dlBTU8Pq1at5++2393q88847qaysbPEzHApmZh1AfX0969at2+cBf/Xq1bz77rt7va9fv36Ul5czdOhQioqK9rsfh4KZWTuwefNmVq5cuceBvvHzmpoadu3atcd7iouLKSsro7y8nJNPPpny8vJ0uaysjNLSUnr16nVAdTgUzMza0LZt21i6dCnV1dUsWbKE6upqqqurWbly5R7bFRUVUVpaSllZGZ/61Kf2ONg3PPbp0wdJrVqfQ8HM2szOnTt5//33KS4upri4OK/mjI5qx44dvPnmm3sc+Kurq/nb3/5G7maTcPjhhzN8+HDGjx/PrFmzOPHEEykvL6e8vJzjjjsuk/8+DgUza3W7d+/mrbfe2uuAuGzZMnbv3p1u17NnT4488sj0p7i4eI/lg1nfo0ePVv/23JJdu3axbNmy9Hds+J3feust6urqgNy3/mHDhjFmzBimT5/OqFGjGDVqFEOHDuXwww9vs1rz4VAwa8cigtraWlasWMHy5cvTx507d1JaWkppaSkDBw5MH/v169emB8S6ujqWL1++1wHxzTffZOfOnQBIYsiQIYwaNYrzzjuPgQMHsm3bNjZv3pz+bNmyZY/ldevW7bFu69atedckiZ49e9KrVy969eq1x/PGPwezvr6+nqVLl+4RdsuWLUvb+rt168bHP/5xKioquOCCCxg1ahQVFRUMGzaMI444oiD/D1qbQ8EsY9u2bWPlypV7HfhXrFjBihUr+PDDD/fYvl+/fnTv3p21a9dSX1+/x2vdu3dnwIABzQZGw+Pxxx/PYYcd2J9+fX09q1at2qsdfOnSpWzfvj3d7oQTTqCiooKzzz47/TY8YsSIA+7sbKquro6tW7fuFR5NlxsCpOFn27Zteyxv3ryZ9evX77W+8e+Qr4ag+8IXvpAe/E866SR69ux5SL9r1hwK1mlFBFu2bKG2tpba2lo2bty4x/OI2O83xKbfOLt1O/DxnnV1daxZsyY9yDc98K9du3aP7Xv27MngwYMZMmQIEydOZMiQIenyoEGDOOqoo4BcE826detYs2YNNTU11NTUpM/XrFnDSy+9xJo1a/Y64HXr1o3jjjuu2cAoLS2lf//+aQA0/Lz++uts2bIl/YzS0lJGjRrFZz7zGSoqKtKDf0Ntra2oqIijjjqqYJ9fX1+fBkXTwGgcMPX19Zx00kmMGDGC4uLigtSSNTV0eHRElZWVUVVVlXUZVmC7d+9m06ZNex3YmzvQN32t6SV8h6pHjx55NT0UFRWxevVqli9fzqpVq/aoo1u3bpSVlTF48OD0YN/48dhjj221JqCG5qfGYdFcgGzatKnZ9x933HHpt+CGb/4jR46kd+/erVKfZUPSoohodhSbQ8HahYhgzZo1LFq0iEWLFlFVVcUbb7xBbW0tH3zwQYvvPfroo+nbty99+vShb9++ez1v7rU+ffpQVFTU7DfClr4t5vvarl27KC0t3euAP2TIEMrKytpd+/KWLVvSkFi3bh0DBw5k1KhRlJSUZF2aFUBLoeDmI2tzjQOgqqoqDYL169cDuW/SI0eOZPz48fTv37/Fg33v3r0PuH28sZ49e/rAR24Q1LBhwxg2bFjWpVjGHApWUPkGwOTJkxk3bhzjxo1j9OjRh9wxaWYHx6FgraYhABof/BsHQFFRkQPArJ1zKNhB2bVrFzU1NSxevDivAKisrOTkk092AJi1cw4F20tEsGHDhhZnY2x8jXzjAKisrGTcuHEOALMOyqHQBX300Ud7HOCbHvRramr2ura9R48e6SRcZ555JmVlZZSVlVFRUcHo0aM7/IAdM8txKHRSO3fu5LnnnuPPf/7zXt/ym16T3q1bNwYMGEBZWRnjxo1j6tSpe83IWFJS0qbTJ5hZNhwKncg777zDY489xvz583nqqafYvHkzAH379qW8vJxBgwYxceLEvQ74AwYMOKTLOs2s8/CRoAOrq6vjpZdeYv78+Tz22GO88sorAJSVlTF9+nQmT57M6aefXrCpAcys83EodDC1tbUsWLCA+fPn88QTT/Dee+9RVFTEqaeeyg033MCUKVOoqKhwU4+ZHRSHQjsXESxZsoT58+czf/58/vjHP1JfX09JSQlnn302U6ZM4fOf/zx9+vTJulQz6wQyCQVJVwKXAAJ+ERE3S/oEcCvQA9gNzI6Il7KoL2tbtmzh97//fdostHr1agDGjBnDN77xDaZMmcInP/nJTn3XKjPLRpuHgqQKcoFwCrATeELSo8APgH+JiMclTU6WT2/r+rKyfPny9Gzg2WefZceOHRx55JGcccYZXH/99Zx99tkMGDAg6zLNrJPL4kxhBPBiRGwFkLQQOB8I4Ohkm48B72RQW5t67bXXmDdvHo8++ihvvPEGAMOGDWP27NlMnjyZCRMm0L1794yrNLOuJItQWAJ8V1IJsA2YDFQBVwG/k/RDoBtwanNvlnQpcClAeXl5W9TbqjZt2sT999/P3Llzqaqq4vDDD+f000/nsssuY8qUKQwdOjTrEs2sC8vkfgqSZgGzgS1ANbCDXBAsjIiHJX0ZuDQiPtfS53SU+ynU19fz7LPPMnfuXB5++GG2b9/OySefzKxZs5g2bZqnbjazNtWub7Ij6XtADXAD0DsiQrnrKT+IiKNbem97D4XVq1dz5513cscdd7BixQo+9rGPMW3aNC6++GLGjh3ry0bNLBPt7iY7kvpHxHpJ5eT6E8YDVwB/BzwLfBb4axa1HaodO3bw61//mrlz57JgwQIigkmTJvFv//ZvTJ061XMEmVm7ltU4hYeTPoVdwJyI2CTpEuCnkg4DtpP0G3QUixcvZu7cudxzzz3U1tZSVlbGddddx8yZMxk8eHDW5ZmZ5SWTUIiICc2sewEYl0E5B23jxo1pp/GiRYs44ogjmDp1KhdffDGTJk3yOAIz63A8ovkA1dfX88wzzzB37lweeeQRtm/fzujRo/nZz37GtGnT6Nu3b9YlmpkdNIdCnt5+++2003jlypX07t2bWbNmMWvWLMaMGZN1eWZmrcKhsB/PP/883/nOd3jqqacAmDRpEjfccAPnnXcePXr0yLg6M7PW5VBowYoVK5gyZQpHH300119/PTNmzGDQoEFZl2VmVjAOhX2oq6vjoosuQhJ/+MMfOOGEE7Iuycys4BwK+3DTTTfxwgsvMG/ePAeCmXUZ3bIuoD165ZVX+Pa3v80FF1zA9OnTsy7HzKzNOBSa2LZtG9OmTeOYY47h1ltv9VQUZtaluPmoiWuuuYalS5eyYMECjzkwsy7HZwqNLFiwgFtuuYWvfvWrnHHGGVmXY2bW5hwKiffff5+ZM2cycuRIbrzxxqzLMTPLhJuPgIjgsssu47333mP+/PmeydTMuiyHAnD33Xfz0EMPccMNN3jKCjPr0rp889HKlSu5/PLLmTBhAl//+tezLsfMLFNdOhQaRi0DzJs3z1Ndm1mX16Wbj374wx/y/PPPc9ddd3lOIzMzuvCZwiuvvMJ1113Hl770JS688MKsyzEzaxe6ZChs27aN6dOn069fP49aNjNrpEs2H1177bW8/vrr/O53v6OkpCTrcszM2o0ud6bw5JNP8tOf/pQrrriCM888M+tyzMzalS4VCrW1tcycOZMRI0bw/e9/P+tyzMzanS7TfNQwann9+vX89re/9ahlM7NmdJlQuOeee3jwwQf53ve+x9ixY7Mux8ysXeoSzUerVq3i8ssv59Of/jRXX3111uWYmbVbnT4UGkYtR4RHLZuZ7Uenbz760Y9+xHPPPccdd9zB4MGDsy7HzKxd69RnCq+++irf+ta3OP/885kxY0bW5ZiZtXudNhS2b9/O9OnTKSkp4bbbbvOoZTOzPHTa5qNrr72W6upqHn/8cfr165d1OWZmHUKnPFN46qmnuPnmm5kzZw5nnXVW1uWYmXUYnS4UGkYtDx8+nB/84AdZl2Nm1qF0quajiGD27NmsW7eO3/zmN/Tq1SvrkszMOpROFQr33XcfDzzwAN/97nc9atnM7CB0muajVatWMXv2bE477TSuueaarMsxM+uQMgkFSVdKWiKpWtJVjdZfIemNZH3eHQJ1dXXMmDGD+vp67r77bo9aNjM7SG3efCSpArgEOAXYCTwh6VGgDDgXGB0ROyT1z/czf/zjH7Nw4ULmzp3rUctmZocgiz6FEcCLEbEVQNJC4HygErgxInYARMT6fD5s8eLFfPOb32Tq1KnMnDmzQCWbmXUNWTQfLQEmSCqR1AuYTO4sYViy/kVJCyV9srk3S7pUUpWkqg0bNrB161bGjh3L7bff7lHLZmaHSBHR9juVZgGzgS1ANbAD+BzwDPBV4JPAA8CQaKHAysrKqKqqIiIcCGZmeZK0KCIqm3ttv2cKkr4gqVXPKCLilxExLiImAhuBZUAN8EjkvATUA3nNT+FAMDNrHfkc7P8B+KukH0ga3ho7behEllROrj/hPuD/Ap9J1g8DjgDea439mZlZfvbb0RwR0yUdDXwFuFNSAHcA90fERwe534cllQC7gDkRsUnSXGCupCXkrkqa0VLTkZmZtb68rj6KiA8lPQT0BK4CpgJfl/SziLjlQHcaEROaWbcTmH6gn2VmZq0nnz6FL0r6FfAscDhwSkScDYwGvlbY8szMrC3lc6bw98BPIuK5xisjYmtyFZGZmXUS+YTC/wbWNixI6gkcGxErI+LpQhVmZmZtL5+rjx4kd3log7pknZmZdTL5hMJhSScwkHYIH1G4kszMLCv5hMIGSV9sWJB0Lh4/YGbWKeXTp3AZcK+knwMCVgMXFbQqMzPLRD6D1/4GjJd0ZLK8ueBVmZlZJvIavCZpCjAK6NEwz1BE/GsB6zIzswzkM3jtVnLzH11BrvnoAuCEAtdlZmYZyKej+dSIuAjYGBH/AnyK3L0PzMysk8knFLYnj1slDSA3id3xhSvJzMyykk+fwm8l9QZuAl4GAvhFIYsyM7NstBgKyc11no6ITeSmu34U6BERH7RFcWZm1rZabD6KiHrg3xst73AgmJl1Xvn0KTwt6e/le16amXV6+YTCP5GbAG+HpA8lfSTpwwLXZWZmGchnRPNRbVGImZllb7+hIGlic+ub3nTHzMw6vnwuSf16o+c9gFOARcBnC1KRmZllJp/moy80XpZUBtxcqILMzCw7+XQ0N1UDjGjtQszMLHv59CncQm4UM+RC5BPkRjabmVknk0+fQlWj57uB+yPiDwWqx8zMMpRPKDwEbI+IOgBJRZJ6RcTWwpZmZmZtLa8RzUDPRss9gacKU46ZmWUpn1Do0fgWnMnzXoUryczMspJPKGyRNLZhQdI4YFvhSjIzs6zk06dwFfCgpHfI3Y7zOHK35zQzs04mn8Frf5Y0HDgpWfVmROwqbFlmZpaF/TYfSZoDFEfEkohYAhwpaXbhSzMzs7aWT5/CJcmd1wCIiI3AJQWryMzMMpNPKBQ1vsGOpCLgiMKVZGZmWcmno/kJ4AFJtyXL/wQ8XriSzMwsK/mEwjXApcBlyfJfyF2BZGZmncx+m48ioh54EVhJ7l4KnwWWHspOJV0paYmkaklXNXnta5JCUr9D2YeZmR24fZ4pSBoGfCX5eQ94ACAiPnMoO5RUQa6j+hRgJ/CEpEcj4q3kXg1nAm8fyj7MzOzgtHSm8Aa5s4JzIuLTEXELUNcK+xwBvBgRWyNiN7AQOD957SfA1fz3VN1mZtaGWgqF84G1wDOSfiFpErkRzYdqCTBBUomkXsBkoEzSucCaiFjc0pslXSqpSlLVhg0bWqEcMzNroIiWv5RLKgbOJdeM9FlgHvCriFhw0DuVZgGzgS1ANVAEjAbOjIgPJK0EKiPivZY+p7KyMqqqqlraxMzMmpC0KCIqm3stn47mLRFxX3Kv5lLgFXJXJB20iPhlRIyLiInARnLBMBhYnARCKfCyJF/lZGbWhg7oHs0RsTEibo+ISYeyU0n9k8dycs1Ud0VE/4gYFBGDyN0HemxEvHso+zEzswOTzziFQnhYUgmwC5jTeBoNMzPLTiahEBET9vP6oDYqxczMGjmg5iMzM+vcHApmZpZyKJiZWcqhYGZmKYeCmZmlHApmZpZyKJiZWcqhYGZmKYeCmZmlHApmZpZyKJiZWcqhYGZmKYeCmZmlHApmZpZyKJiZWcqhYGZmKYeCmZmlHApmZpZyKJiZWcqhYGZmKYeCmZmlHApmZpZyKJiZWcqhYGZmKYeCmZmlHApmZpZyKJiZWcqhYGZmKYeCmZmlHApmZpZyKJiZWcqhYGZmKYeCmZmlHApmZpbKJBQkXSlpiaRqSVcl626S9Iakv0j6laTeWdRmZtaVtXkoSKoALgFOAUYD50gaCjwJVETEycAy4Nq2rs3MrKvL4kxhBPBiRGyNiN3AQuD8iFiQLAP8CSjNoDYzsy4ti1BYAkyQVCKpFzAZKGuyzcXA4829WdKlkqokVW3YsKHApZqZdS1tHgoRsRT4PrAAeAJ4FahreF3SN4HdwL37eP/tEVEZEZXHHHNM4Qs2M+tCMulojohfRsS4iJgIbCTXh4CkmcA5wLSIiCxqMzPryg7LYqeS+kfEeknlwPnAeElnAVcDfxcRW7Ooy8ysq8skFICHJZUAu4A5EbFJ0s+B7sCTkgD+FBGXZVSfmVmXlEkoRMSEZtYNzaIWMzP7bx7RbGZmKYeCmZmlHApmZpZyKJiZWcqhYGZmKYeCmZmlHApmZpZyKJiZWcqhYGZmKYeCmZmlHApmZpZyKJiZWcqhYGZmKYeCmZmlHApmZpZyKJiZWcqhYGZmKYeCmZmlHApmZpZyKJiZWcqhYGZmKYeCmZmlHApmZpZyKJiZWcqhYGZmKYeCmZmlHApmZpZyKJiZWcqhYGZmKYeCmZmlHApmZpZyKJiZWcqhYGZmKYeCmZmlHApmZpbKJBQkXSlpiaRqSVcl6/pKelLSX5PHPlnUZmbWlbV5KEiqAC4BTgFGA+dIGgr8M/B0RJwIPJ0sm5lZG8riTGEE8GJEbI2I3cBC4HzgXOCuZJu7gPMyqM3MrEs7LIN9LgG+K6kE2AZMBqqAYyNibbLNu8Cxzb1Z0qXApcniZklvFrhegH7Ae22wn0Jw7dnoyLVDx67fte/fCft6QRHRBvtvslNpFjAb2AJUAzuAmRHRu9E2GyOiXfQrSKqKiMqs6zgYrj0bHbl26Nj1u/ZDk0lHc0T8MiLGRcREYCOwDFgn6XiA5HF9FrWZmXVlWV191D95LCfXn3Af8BtgRrLJDODXWdRmZtaVZdGnAPBw0qewC5gTEZsk3Qj8V9K0tAr4cka1Nef2rAs4BK49Gx25dujY9bv2Q5BJn4KZmbVPHtFsZmYph4KZmaUcCi2QVCbpGUmvJ1NyXJl1TQdCUpGkVyQ9mnUtB0pSb0kPSXpD0lJJn8q6pnxJ+l/Jv5clku6X1CPrmvZF0lxJ6yUtabSuw0w5s4/6b0r+3fxF0q8k9c6wxH1qrvZGr31NUkjq19Z1ORRathv4WkSMBMYDcySNzLimA3ElsDTrIg7ST4EnImI4uelQOsTvIWkg8FWgMiIqgCLgf2RbVYvuBM5qsq4jTTlzJ3vX/yRQEREnk7vc/dq2LipPd7J37UgqA84E3m7rgsCh0KKIWBsRLyfPPyJ3YBqYbVX5kVQKTAH+M+taDpSkjwETgV8CRMTOiNiUaVEH5jCgp6TDgF7AOxnXs08R8RxQ22R1h5lyprn6I2JBMoUOwJ+A0jYvLA/7+G8P8BPgaiCTq4AcCnmSNAgYA7yYcSn5upncP6z6jOs4GIOBDcAdSfPXf0oqzrqofETEGuCH5L7lrQU+iIgF2VZ1wPKacqaDuBh4POsi8iXpXGBNRCzOqgaHQh4kHQk8DFwVER9mXc/+SDoHWB8Ri7Ku5SAdBowF/iMixpCbDqU9N2Gkkvb3c8kF2wCgWNL0bKs6eJG7Zr1DXrcu6ZvkmoDvzbqWfEjqBXwD+HaWdTgU9kPS4eQC4d6IeCTrevJ0GvBFSSuB/wN8VtI92ZZ0QGqAmohoOCt7iFxIdASfA1ZExIaI2AU8ApyacU0HqsNPOSNpJnAOMC06zmCsj5P7MrE4+dstBV6WdFxbFuFQaIEkkWvXXhoRP866nnxFxLURURoRg8h1cv4+IjrMt9WIeBdYLemkZNUk4PUMSzoQbwPjJfVK/v1MooN0kjfSoaeckXQWuabTL0bE1qzryVdEvBYR/SNiUPK3WwOMTf4e2oxDoWWnAReS+6b9avIzOeuiuogrgHsl/QX4BPC9bMvJT3J28xDwMvAaub+xzKcu2BdJ9wP/DzhJUk0yzcyNwBmS/kruzOfGLGtsyT7q/zlwFPBk8jd7a6ZF7sM+as+cp7kwM7OUzxTMzCzlUDAzs5RDwczMUg4FMzNLORTMzCzlUDDLiKTTO+IMtta5ORTMzCzlUDDbD0nTJb2UDIS6LblPxWZJP0num/C0pGOSbT8h6U+N5vLvk6wfKukpSYslvSzp48nHH9novhH3JqOgzTLjUDBrgaQRwD8Ap0XEJ4A6YBpQDFRFxChgIXB98pZ5wDXJXP6vNVp/L/DvETGa3FxIDbOQjgGuAkYCQ8iNojfLzGFZF2DWzk0CxgF/Tr7E9yQ3QVw98ECyzT3AI8l9IHpHxMJk/V3Ag5KOAgZGxK8AImI7QPJ5L0VETbL8KjAIeKHgv5XZPjgUzFom4K6I2OPuXZKua7Ldwc4Xs6PR8zr8N2kZc/ORWcueBr4kqT+k9y8+gdzfzpeSbf4n8EJEfABslDQhWX8hsDC5a1+NpPOSz+iezJ1v1u74W4lZCyLidUnfAhZI6gbsAuaQu/HPKclr68n1O0Buqulbk4P+cuAfk/UXArdJ+tfkMy5ow1/DLG+eJdXsIEjaHBFHZl2HWWtz85GZmaV8pmBmZimfKZiZWcqhYGZmKYeCmZmlHApmZpZyKJiZWer/A/6La5AFMK63AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(np.arange(1,len(list_accuracy)+1), list_accuracy, 'k')\n",
    "plt.xlabel('epoch')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.axis([0.9,len(list_accuracy)+0.1,90,100]);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABUUAAAHaCAYAAAAnnShJAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAABhCElEQVR4nO3dd5hdVdWA8XclISEkIQkQegnSRRE7ggUUEVQE7FgQ7L2L2LF+KihYUfkEkaYgKojiRxEU7IJ0FekdAggkISEhWd8f50QuN3dm9tQ7M+f98cyTe89ZZ581c2f2MOvuEpmJJEmSJEmSJDXFhG4nIEmSJEmSJEkjyaKoJEmSJEmSpEaxKCpJkiRJkiSpUSyKSpIkSZIkSWoUi6KSJEmSJEmSGsWiqCRJkiRJkqRGsSgqNVRErBkR746IkyLiXxFxV0QsjYhs+9i/7bqDO8Qc3J3PQtJYEBFzO/Qb13c7r9EiIvbv8PX5QbfzkiRJksazSd1OQNLIiohJwGeB9wBTu5yOJEmSJEnSiLMoKjVIXRD9FfDcbuciSZIkSZLULRZFpWb5NBZEh0RETAWe1XZ4UWb+thv5SBoZEbF7+7HM/HU3cpEkSZI0cBZFpYaIiDWA93U4dSvwfeCvwD3Asrbz1wxzamPVOsAZbcduAOaOfCqSRlD7zz1AjHgWkiRJkgbFoqjUHK9i5TVErwaempn3dCEfSZIkSZKkrrAoKjXHTh2O/U9/C6KZeTBw8FAkJEmCzPwB8IMupyFJkiQ1yoRuJyBpxDy+w7HfjXgWkiRJkiRJXWZRVGqONTscu2XEs5AkSZIkSeoyi6JSc8xsP5CZi7qRiCRJkiRJUjdZFJWaY5VuJyBJkiRJkjQauNGSpFEpIgJ4LLANsD4wDVgCzAN+mpn3FbSxKvAYYCtgFg+Pln0AWADcDFwPXJuZS4b2MxjdImIS1ddmSx7++i4H7gfuAv4J/CszFw/xfbeo77sxML3lntcDF2fmiC/pEBGbU625uyGwGvAfqu+zizLzmpHOZ6SNxtdEQycipgBPALYA1gamAoupfs7/TfV9/sAI57QK8CTg0cBa9eF5wK3A7zNz/kjmI0mSpGayKCqNQxFxHvCsgrjsI2TTzLy+7ZqDgU+1xX263pW+r/vtDxzddviYzNy/JWYz4L3AK3n4j+V2fwcu7uEek+tr96P6GpT0cw9GxIVUG0+dnJkX9dD2efT+dd2k4GsKQGZGSdxQqguhLwZeDexKVQDszZKI+D1wBnBcZt42wPtuAbwTeBmwXh+xVwEnAN/OzHkDvN/OwLlth3+bmTu3xKwKvBl4O1XRvKe2rga+A3wzMx8svP9XgPe3Hf55Zu5Tcn0fbQdVsXLjtlP7ZuaP+tHOiL4mg1HSb/SzvbnAdW2Hb8jMuT3E/wB4XR9tFv3c06FPra/fn6H9HAN4EfBG4LnAlF7Cl0bEucBRVP3f8gHe8wes/HU6IDN/0BKzOXAg8Apg9V7y+R3wmcx0M0BJkiQNG6fPSxoVovJx4EqqYk1PBdHe2tgFuBQ4BngO5W/8TAF2BA4CLoyIg/p779EuIl5GNSrsx1TFkr4KogCTgV2ALwM3RsRn+3nPGRHxLeAfwLvpo/hW2xI4GLg+Ig6MiIn9uWdhXk+l+j75Gr0URGubA4cC/4iIbQtvcUyHY8+PiE6bnfXXzqxcEL0P+HnJxaP1NdHQiYgnAH+j+p54Ib0XRKFaWmU34EfApRGx4zDkFHW/ejnwJnouiK7I5znAbyPi2PqNLkmSJGnIWRSV1HURMQE4DvgsVSFuIG3sC5xJ30WuEqsOQRujQkSsFhHHAycBcwfR1CSq5QxK77sZ8BeqkZgDKaKtBnwJ+FVE9FZA6ZeIeDHwW6qpxP2xKfC7iHh8X4GZeSkrj2ReMYJ5sPbrcOykkmUORutroqFTjzj9I9V0+YHYlqoY+c4hzGki1Ujj/6HvAm271wC/qKfbS5IkSUPK6fOSRoMvAq/qcPxG4Haq9e82oFrzcaU/qiPiicCx9FzoWQRcS7VW5GKqdRNnUhUJpw4u9dErImZQTX3fqZewh6i+zvOo1lqdBaxL2QjCnu67MXAe1evVSVJNAb+FalTYhlSvbye7Ab+OiF0Hu+5hRDwHOJGVC++31h/zgTWBrTvEAKwBnBAR2xdMpf8hsH3bsf2Ab/Uz7f+KiNWAl/Rwr76uHZWviYZORLwBOBLoaWmOhVSv8Z1UI/HnAjM6xE0CvhERq2TmYUOQ2vdY+Q2Bh4Br6lwmUH2vze3h+t2AjwKfHoJcJEmSpP+yKCqNT29n5emJf+wQ97Q+2hnQGpL99FQeObrzP1Qjik7KzBtaAyNiOtUf13e3tfFtVi6ILqUqEBwD/K3TOnn1CNUtgB2APYHnURVMe9L6dV0P+Gnb+duBQa8bOYS+T88F0bOA7wJnZeb97ScjYl2q74+XUX1tevu6tF43ATiezsW3BVSv7Q8z8+a26x5NtcbnO1n5tXwa1RT2t5fk0IO1eWRBdAFwGHB8Zv6rLZfpwL7A5+rrWm1NtSZiX0sJHE+17EDr79mnRMRW7ffrh31YuYh1TWZe0NtFo/g1GSs+S7Wu7AoD6UtXGJY+NSIeS1Vw71QQvZjqe/mXrSOK69GXz6MqOHbK/5CI+FNmdvp8Sx0APLPl+eXA54Ez2jfLq0cyHwS8ocPn8bGIOK4JG59JkiRp5FgUlcahzLyy/Vi178ZKcX8akYR6t3XL4wuAvTOzvegJQGYuAP639VhEbAM8pS10EfDczPx9bzeuC6X/qj+OqYthB1CNGOwU/9+va71ZS7sHR8nXlIh4D1VBs909wGsz81e9XZ+ZtwM/A35Wf13eDjyq4NYHAk/vcPxSYK9Om8zU97sSeG9E/Ag4DZjTFvK2iDi9r7x7sU3L478De/a0q3r9fXZkRJxNNbqyfQ3Pt0TEFzJzWU83y8w7I+LXVGs6ttoP+Fh/k2+5tl2fo0QZva/JmFAX4v5bjBttfWm95uZxdJ6a/hXgoMx8qP1EZi4FTo+IXwGfZOUN9CYCx9YjoxcMML3Wguj/AB/vaSOn+uv8poj4E239PNXo5TcCHxlgHpIkSdJKXFNU0mhxIfC8ngqivditw7Ev91UQ7SQzF2TmN1p3Sx6L6lGeX+xw6jbgGf0tYtVfly9n5lv7uO9s4BMdTl1L9dpeX3CvPwHPp5rq2+6r0aki1T9XAM/sqSDalst1wOs7nNoA2LXgXp0Klq8ZyOcQEetRbT7TKqmWjejturHwmmhwDgC263D825n5wU4F0VaZuTwzD6ZaL7bdZlQbcg3WpzPzoyU722fm9+n8fd2+s70kSZI0KBZFJY0GS4EDBrg+4UYdjv1ykPmMde9h5c2iEnhNp1HEQ+j1dN7V/g31yNMimfk3qpFr7bYCnjvA3KBax/CV/Rn1lpnnAJ0K7L2t07rCaVTLQbTamGoH+f56DStPYT+/Ltz2ZrS/Jhq8d3Q4dhXwvn6281FW3iAM4K31ZkkD9XvgM/28ptPyFOtFRMlodUmSJKmIRVFJo8FPM/OyAV7baRfsldbIbIp6qvvbOpz638z8zTDfvtN9f5aZ5w2gra8D/+5wvFMBqNRPMvPyAVx3UodjT+zronozph93ONVpGnxfXtvh2DEF143210SDEBFPBx7b4dSHMnNJf9qqR3G+t8OpjYAX9T+7//psyQjRtlz+TbXMRbs+f+4kSZKkUhZFJY0GRw/i2ns7HGtfY7RJdgZmdjj+1eG8aT2Ca7MOp44cSHv1lN+jOpx69iBGrX1/gNdd2OHYFoXXdppC/5J6J/kiEfF4Vi58LQJO7uO6sfCaaHA6LR9yG3D6QBrLzN8C/yy8T4mbM/P/BnjtYH7uJEmSpD5ZFJXUbcuA8wdxfafp4J+LiE47bTfBMzsc+31mdip0DKVOu1ffQ7XL/UCd0OHYdDqPjOvLQ3TeNbzE1R2OdSo8r6TeufuqtsMzqHaSL9VpZOnPMrPjhmAtRvtrosHr9Bqf3N+RmW1OLLxPiQsGkceAf+4kSZKkEhZFJXXbPwe4lugKZ1IVvFptDFwaEQdFxNqDaHss6lQUHUzRudRTOxy7qK9NXnqTmTdSjXprN5ACzQ2Z2WmjoBKdlmPotGxDTzqNFi2aQh8Rk4B9O5wqmTo/2l8TDUK9wVWnUfF/GWTTf+pw7DH10hz9dcUg8hjsz50kSZLUK4uikrrt5sFcnJm30rlANBv4H+DWiPhtRHw8Ip4ZEVMHc78xYPMOxwZbJCmxSYdjFw9Bu53WFdx4AO20b3hULDMXdTjcvpFVb46l2uiq1a4RsX7Btc8D1mk7ditwdsG1o/010eDMonOR8OJBttvp9Z0IbDCAtgb8c0e1RES7/vzcSZIkSb2yKCqp24ZiU6T3A//o4dxEqtGTnwV+C9wXEX+OiC9GxG4RMXkI7j8q1CPHZnU4NajCc6HZHY51GlHYX7cW3qsvAx0lOmj16Mrz2g5PoNpRvi+dRpQeVzg9erS/Jhqcnr7mg3qNM3MeK4++7+1+vRnqn7sY4vYkSZLUYBZFJXXbYKbOA5CZ9wPPoJpK35dVqKacfhj4P+C2iDgiIrYebB6jwCyqInC7e0fo3u2GouDdqY2xWIDrNJq5047y/xURM+m863fJ1HnwNRnvZvVw3NdYkiRJKmBRVNK4kJl3Z+bzgL3pvCZeT9YA3gpcHhHfj4gZw5HfCOlpR/NO01CHWqdlCYbivp3aKN65fRQ5hZVHzT0mIp7QyzUvY+Xpwn/LzE6bi3XiazK+dXp9HxrMmrEtfI0lSZI07lkUlTSuZOapmfk0YGvgQOAMoK9duqEaYfl64KKIGMjaeaPBvT0cH4nNSTqNLBvIxiztOhWp7xuCdkdUZi4AftrhVG8bLnU612nTpp74moxvnV7fSRExZQja9jWWJEnSuGdRVNK4lJn/ysxDMvP5VNM+nwi8D/g5vU8v3Rw4LSI6TUMf1erd1R/scGrWCNz+3g7HZg5Bu50KuoPZvKWbOk1737feYf4RImJT4Olth5cCJ/bjfvd2OOZrUi2hMR7c28PxQb3G9drEnYrnY+k1liRJkvpkUVTSuJeZyzLzosw8PDP3AeYAzwdO6+GSJwCvG7EEh9bdHY5tMQL37VQw2XQI2t2s8F5jwbnATW3H1gZ27xD7WlbeVOaXmXlXP+43Xl6TTtPBVyok98Mag7h2NOnpaz7Y13hTOv//4Vj9uZMkSZI6sigqqXEyc0lmnpGZewG70XmH5JKdwUejSzoce+oI3PcfHY49fjAN1iPWHtfhVOmamqNKvWP8cR1OdZom32kTpv5MnYfx85p0Wv5iMMsAjNXlMR6hXpLh5g6nBvUa93D9QuCGQbYrSZIkjSoWRSU1WmaeBXyiw6kd6wJQj5cOU0qD9dsOx3bt43MZCn/scGz7iBjMeqaPp/NU7U73Gis6TaHfMyJmrXgSETtSLePQ6m7gl/2813h5TTqtZbnuINrbcRDXjjadvu7PGmSbna7/a2YuG2S7kiRJ0qhiUVSS4KQOx6ZQrUXak05rd46GtQo7FUW3AJ4zzPf9EysXiqcALx1Em51GS96WmWN2xFpm/gv4c9vhVYGXtzzvNHL0xMxc0s/bjZfX5NYOx7YdSKG/vuaFg8xnpZ/9iOjWz36nouiLImJAI2nr9W1fWXgfSZIkaUyzKCpJ0NM6jb1tttRpSu+0IchlsP4M/LPD8Y8N52jRzLwHOK/DqXdGRL9/10TEGnQuDp7S37ZGoU7T4PcDqHcOf3mH8/2dOj+eXpN/Awvajk0HnjaAtvYAthpkPqPpZ/9nwPK2Y6sBbxpge6+mWnO53Xj4uZMkSZIewaKoJMGjOhxbQs/F0hU7vS9tOzwzIoZid+8By8wEDu1wamfg3cN8+291OPZ44M0DaOvzrLwhTgLfHEBbo82PqL6/Wu0UEZsBe7LyCOV/ZOZfB3ivMf+a1N/TF3U49cb+tBMRM4BvDEFK93Y4tskQtNtvmXk98KsOpz4VEev0p6267/pSh1N/zMwLB5CeJEmSNKpZFJU0pkXEFyPi6YNs5j0djl1UF2N602kjm4GMXhtqx9J5U5RDI6LT9OciEbFeHyE/B67vcPyr/XmNIuKNwFs7nPp1Pf18TKtHcP6iw6n96DwSs9M6pKV+zvh4TX7S4djr6vVX+xQR04CT6fwGSH912lSqmz/3h3c4NhP4ef159ykiJlONBu1USO3UviRJkjTmWRSVNNbtDpwfEX+IiDfUU3yLRMSkiPgk8JYOp08oaKLT6LUD6wJD19RrT+4LPNR2ahJwTEQcVvp1ioiJEfG8iDgH+HYf911GNW23vZg8FTg9IjqtVdh+r4OA73Q4PR94R0nOY0Sn6fBvoPp+btXTjvVFxtFrcjwrr+U5gepz6HW93Ih4GnAB8Lz60AODzKXTz/2765GoIy4zz6Hz98gOwJn1COQeRcSGVKNNO30dT8/MTmsuS5IkSWPepG4nIElD5Gn1xxER8VuqjUEupFpf8x6qKa8TgVlUGw89C3gdK+/yDfAv4MiCe54C7N92bBfg4og4HrgM+A8rT7MnM/9U0P6AZeYfI+IjwCFtpwJ4L7B/RJwEnFHnOQ9YRPX1WQ94AtXXc29g7fraUwvue3ZEfIOVp+rPBE6MiLdQFQQvoNpAZyKwIbAb8HrgsT00/Z7MvK6v+48hZ1B9zVvXb9ygQ9w5mXnLYG40Hl6TzLwnIg4BPt52ajZwdkScRfX9eQ1V0XMtYGvg+cBOLfHLqUaGl/x89+QU4OC2Y9sAV0TEMVRF07tZeYkEgL9nZqdN2gbrncAzgY3bju8IXBoRx1KNlP0n1ffdGlT94Muo+sHVO7Q5j6pQL0mSJI1LFkUljTerALvWHwNxP/CazFxcEHsGcDUrF1a3AT7Xx7XDtunRCpl5aESsDXyow+lZVOtKDmRtyb58iKrgskeHczvXH/1xaGYePcicRpXMXBoRJ9B56YZWg5k632o8vCafBfaic5H2ufVHX94NnD2YJDLz8og4j5W/ZhuxctG23aZ0Xs5gUDLzvojYEziXldd9XY1qNHynEfE9mQ+8KDPvHKIUJUmSpFHH6fOS9LBbgGdl5t9Kguupya9m5Wm9o0ZmHkg1fXqwU4b7c88lVCNM/3eQTS0DPpSZnYq640FfO8rPp9pdfNDGw2tSfw7PpRrZ3F9LgTdnZqeNpwbiAKpR4KNGZl4KPB24apBN3QA8c7hHs0uSJEndZlFU0lj3P8BpDK7ot4hqx/ZtMvPi/lyYmX+hWrvv0kHcf1hl5v8Cj6ZaJ3XZAJv5LXBEP+65JDPfBLyIaspuf50PPDkzDx3AtWNCZl4EXN5LyE8yc8iK2ePhNcnMO6imiX+DDstS9OAPwFMyczBT5tvzuJ5qiYnfDlWbQyEz/wFsT9UvLuzn5YupNlV6bH/7QUmSJGksir43V5Y0HkTEr9uPZWb7pi5jVkSsSrUG5k5URcptqNbX62mZkOuAvwO/pCo+3T8EOewI7Ak8HtiKas3G6VRT+h8hM4d9+nwnEbER8EqqDX2eDHTaHCaBm4BLgLOAX2bmtYO45wSqTW5eQbWW69wOYcupCnXnAMdn5p8Her+xJCJ2o1r3sZOTM/OKYbrvmH9NImIu8GKqz2NzqvVZpwL3US1r8Qeqn+0/DHMejwVeQlUkfTTV0hQzgE4brm1aF1SHXb2Z2quo+qQd6Lxu6ELgL8DpVK/xHSORmyRJkjQaWBSVNG5FxCRgXapiwGpU09zvB+7OzAXdzG20qNccncPDX5/5wB1DOUKxwz2nUa2/OI2qAHs/cNMwbUCjAr4m419ErAOsA0yh2gRqHnBb+j+CkiRJaiiLopIkSZIkSZIaxTVFJUmSJEmSJDWKRVFJkiRJkiRJjWJRVJIkSZIkSVKjWBSVJEmSJEmS1CgWRSVJkiRJkiQ1ikVRSZIkSZIkSY1iUVSSJEmSJElSo1gUlSRJkiRJktQoFkUlSZIkSZIkNYpFUUmSJEmSJEmNYlFUkiRJkiRJUqNYFJUkSZIkSZLUKBZFJUmSJEmSJDWKRVFJkiRJkiRJjWJRVJIkSZIkSVKjWBSVJEmSJEmS1CgWRSVJkiRJkiQ1ikVRSZIkSZIkSY1iUVSSJEmSJElSo1gUlSRJkiRJktQoFkUlSZIkSZIkNYpFUUmSJEmSJEmNYlFUkiRJkiRJUqNYFJUkSZIkSZLUKBZFJUmSJEmSJDWKRVFJkiRJkiRJjWJRVJIkSZIkSVKjWBSVJEmSJEmS1CgWRSVJkiRJkiQ1ikVRSZIkSZIkSY1iUVSSJEmSJElSo1gUlSRJkiRJktQoFkUlSZIkSZIkNYpFUUmSJEmSJEmNYlFUkiRJkiRJUqNYFJUkSZIkSZLUKBZFJUmSJEmSJDWKRVFJkiRJkiRJjWJRVJIkSZIkSVKjWBSVJEmSJEmS1CgWRSVJkiRJkiQ1ikVRSZIkSZIkSY1iUVSSJEmSJElSo1gUlSRJkiRJktQoFkUlSZIkSZIkNYpFUUmSJEmSJEmNYlFUkiRJkiRJUqNYFJUkSZIkSZLUKBZFJUmSJEmSJDWKRVFJkiRJkiRJjWJRVJIkSZIkSVKjWBSVJEmSJEmS1CgWRSVJkiRJkiQ1ikVRSZIkSZIkSY1iUVSSJEmSJElSo1gUlSRJkiRJktQoFkUlSZIkSZIkNYpFUUmSJEmSJEmNYlFUkiRJkiRJUqNYFJUkSZIkSZLUKBZFJUmSJEmSJDWKRVFJkiRJkiRJjWJRVJIkSZIkSVKjWBSVJEmSJEmS1CgWRSVJkiRJkiQ1ikVRSZIkSZIkSY1iUVSSJEmSJElSo1gUlSRJkiRJktQoFkUlSZIkSZIkNYpFUY2oiPhORHxiqGP7aGduRGRETOrh/BURsfNg7yNJkiRJkqSxITKz2zlIwyoi5gLXAatk5kNdTkeSJEmSJEld5khRjZiImNjtHCRJkiRJkiSLohqUiNgmIs6LiHvraegvajn3g4g4IiJ+FRELgV3qY59riTkwIm6LiFsj4o31NPfNW67/XP1454i4OSI+EBF31tcc0NLOCyLi7xFxf0TcFBEH9+NzuD4idq0fHxwRJ0fEcRExPyIui4gtI+Ij9X1viojdWq49ICL+UcdeGxFvaWu7t89vSkQcGhE3RsQd9XIBU/v7GkiSJEmSJKl/LIpqwCJiFeAXwJnA2sC7gOMjYquWsFcBnwdmABe0Xb878H5gV2BzYOc+brkuMBPYAHgD8K2ImF2fWwjsB8wCXgC8LSL2Hthnxp7AscBs4O/A/1H9rGwAfAb4bkvsncALgdWBA4DDIuIJhZ/fF4Etge3r8xsAnxxgzpIkSZIkSSpkUVSDsQMwHfhiZi7JzN8ApwP7tsScmpm/z8zlmbm47fqXA0dn5hWZ+QBwcB/3Wwp8JjOXZuavgAXAVgCZeV5mXlbf51LgROBZA/y8zs/M/6vXHz0ZmFN/jkuBHwFzI2JWfd9fZuY1WfktVYH4GX19fhERwJuB92XmPZk5H/gC8MoB5ixJkiRJkqRCHXfjlgqtD9yUmctbjt1ANeJxhZv6uP5vhbEAd7dtlPQAVVGWiHgq1cjLxwCTgSlUBc2BuKPl8SLgrsxc1vKc+r73RsQewKeoRnxOAFYDLqtjevv85tSxF1b1UQACcN1VSZIkSZKkYeZIUQ3GrcBGEdH6fbQxcEvL8+zl+tuADVuebzSIXE4ATgM2ysyZwHeoiozDJiKmAKcAhwLrZOYs4Fct9+3t87uLqsC6bWbOqj9mZub04cxZkiRJkiRJFkU1OH+mGq15YESsEhE7U63H+aPC608CDqg3a1oN+MQgcpkB3JOZiyPiKVRrmQ63FSNS5wEP1aNGd2s53+PnV4+uPZJqDdK1ASJig4h43gjkLUmSJEmS1GgWRTVgmbmEqgi6B9XIx28D+2XmPwuvPwP4OnAucDXwp/rUgwNI5+3AZyJiPtVmRScNoI1+qdcBfXd9r/9QFWJPaznf1+f34RXHI+J+4GzqNVIlSZIkSZI0fCKzt9nN0siJiG2Ay4EpbWuHjgvj/fOTJEmSJEkaKxwpqq6KiH0iYkpEzAa+BPxiPBUMx/vnJ0mSJEmSNBZZFFW3vQW4E7gGWAa8rbvpDLnx/vlJkiRJkiSNOU6flyRJkiRJktQojhSVJEmSJEmS1CiTup1Aq1VXnZLTp03tdhqS1C9333PfXZk5p9t5SJIkSZKkMqOqKDp92lT2ev7O3U5DkvrlqONOvaHbOUiSJEmSpHJOn5ckSZIkSZLUKBZFJUmSJEmSJDWKRVFJkiRJkiRJjWJRVJIkSZIkSVKjWBSVJEkaQyJiq4i4OCLmR8S7u52PJI1nEfG5iLgrIm7vdi6SpKFlUVSSJGlsORA4NzNnZObXB9NQRJwXEW8corzGrIh4XURcGBH3R8TNEfHliJjUcn6NiPhZRCyMiBsi4lXdzFfSyIiIjYEPAI/OzHUjYv+IuGCI77F/RCyLiAV1H3RJRLyw8NqdI+Lmocynwz12iIizIuKeiJgXESdHxHot5yMivhQRd9cfX4qIGM6cJGmoWBSVJEkaWzYBruh2EgCthcMxbjXgvcBawFOB5wAfbDn/LWAJsA7wauCIiNh2hHOUNPI2Bu7OzDuHorFe+sw/ZuZ0YBbwbeBHETFrKO45BGYD3wPmUv3+mQ8c3XL+zcDewOOA7YA9gbeMaIaSNEAWRSVJksaIiPgNsAvwzXpU0ZYRMSUiDo2IGyPijoj4TkRMreNnR8Tp9eie/9SPN6zPfR54Rktb34yIuRGRbaMk/zuatB7R9PuIOCwi7gYO7u3+HfJvvf7eiLg2Inasj98UEXdGxOta4l8QEX+vR0/dFBEHt5xbkeubI+LWiLgtIj7Yw30n10sOvKt+PrHO45MAmXlEZp6fmUsy8xbgeGCnOnYa8BLgE5m5IDMvAE4DXjuQ11DS6BIRB0XENfWSJFdGxD718V2Bs4D16z7yx8B3gKfVz++t43rrg3euR59/uJ5+f3TnLCqZuRw4FpgGbNFX+318XhkRm7c8/0FEfK6H2CMi4pSW51+KiHMiIjLzjMw8OTPvz8wHgG9S94+11wFfycyb6/7zK8D+feUnSaOBRVFJkqQxIjOfDZwPvDMzp2fmVcAXgS2B7YHNgQ2AT9aXTKD6I3wTqhFPi6j+oCUzP9bW1jsL03gqcC3VqMnP93H/nq6/FFgTOAH4EfDk+trXUBVpp9exC4H9qEZPvQB4W0Ts3dbeLlTFg92AD9eFjEfIzCV125+JiG2Ag4CJdf6dPJOHR+NuCTxUf61XuARwpKg0PlxD9QbRTODTwHERsV5mng3sAdxa95GvAN5KPaozM2fV1/fVB64LrEHVD7+5t0QiYiJwALAUuKGw/aHwAeCx9RtUzwDeALwuM7NDbGv/CFVfeEnLc/tHSWOGRVFJkqQxql637c3A+zLznsycD3wBeCVAZt6dmadk5gP1uc8DzxrkbW/NzG9k5kPA4t7u34PrMvPozFwG/BjYCPhMZj6YmWdSTVPfvM7/vMy8LDOXZ+alwIkd8v90Zi7MzMuoCsD7drppZl4OfA74OdXU+NfWOTxCRLweeBJwaH1oOnB/W9h9wIxePkdJY0Q9CvLWup/5MfBv4Ckl1/bVB9eWA5+q+7hFPTS1Qz3ydDFV3/OazLyzsP1Bq0eAvhb4KnAc8K7MXGmt0ojYjqog+6GWw9Op+sQV7gOmu66opLFgvKwDJUmS1ERzqNbDvLDl78+gGgVJRKwGHAbsTrUuHMCMiJjYqSBY6KbS+/fgjpbHiwAys/3Y9Dr/p1KNknoMMBmYApzcSz43AI/t5d7HUBWGT8nMf7efrEeh/g+wa2beVR9eAKzeFro61bp6ksa4iNgPeD/VmplQ9T9rFV5e0gfOy8zFfbTzp8x8ej1K/vtUI1dPKmx/SGTmnyPiWmDt+t6PUE/FPwN4T2ae33KqvY9cHVjQwyhTSRpVHCkqSZI0dt1FVUTcNjNn1R8z6w07oJoSuRXw1MxcnWraI1R/VAO0/9G6sP53tZZj67bFtF7T1/0H6wSq9Ts3ysyZVOv5tY8+2qjl8cbArb20923gdOB5EfH01hMRsTtwJLBnPep0hauASRGxRcuxxzFKNruSNHARsQnVz/07gTXrKfGXs3I/s0J7n1nSBxYXBzNzAfA24LUR8fjC9nvyAL335Y8QEe+geuPpVuDAtnObAGcDn83MY9suvYKqT1zB/lHSmGFRVJIkaYyqN+U4EjgsItYGiIgNIuJ5dcgMqj+o742INYBPtTVxB/ColvbmAbcAr6k3I3o9sNkg7j9YM4B7MnNxRDwFeFWHmE9ExGpR7QZ/ANWU/JVExGuBJ1JtAPJu4JgVa5dGxLOpNld6SWb+pfW6zFwI/JRqPdJpEbETsBfVZiiSxrZpVEXLeQARcQDVyPSe3AFsGBGTYXj6wMy8B/hf4JOl7UfEqm0fAVwMvKruy3enl6VTImJLquVFXkM1jf7AiNh+xf2A3wDfzMzvdLj8h8D767zWp3oz7gcD/fwlaSRZFJUkSRrbPgxcDfwpIu6nGs2zVX3ucGAq1WijPwG/brv2a8BLo9qZ/uv1sTdRrRd3N9VmGX8YxP0H6+1Uxcj5VOvYrTSlE/htff9zgEPrdUkfISI2pvpa7FfvIH8C8DeqpQUAPkG1ycqv6l2lF0TEGW15TAXupFrX9G2Z6UgoaYzLzCupdkv/I1XB87HA73u55DdUoyBvj4gVS2wMRx94OPD8eg3PvtrfgOrNr9aPzYD3AHsC9wKvplpPeSURMYlqHdEvZeYl9dIiHwWOjYgpwBup3jw7uKV/XNDSxHeBXwCXUY2y/WV9TJJGvRhNS32steas3Ov5O3c7DUnql6OOO/XCzHxSt/OQpCaJiLnAdcAq9aZPkiRJUjFHikqSJEmSJElqFIuikiRJkiRJkhplUrcTkCRJkvorM6+n5x2iJUmSpF45UlSSJEmSJElSo1gUlSRJkiRJktQoTp+XJEkjauqqU3LGjGndTkNSl82fv5BFix90CYRCq01dNWfOnNHtNCR12X33zeeBRYvtO6UhYFFUkiSNqBkzpvGyvXbrdhqSuuzkU8/sdgpjysyZM3jDa1/S7TQkddn3jz2l2ylI44bT5yVJkiRJkiQ1ikVRSZIkSZIkSY1iUVSSJEmSJElSo1gUlSRJkiRJktQoFkUlSZIkSZIkNYpFUUmSJEmSJEmNYlFUkiRJkiRJUqNYFJUkSZIkSZLUKJO6nYAkSZKGRkRpXPn74hMmFDYKRGkCwMzVVy+Onb3GGsWxkyZOLIpbox9trrP2OsWxW2+1VXHsvLvuKo6dP39+cezGG21UFLfK5FWK27zhhhuLY8+/4PyiuMmTJxe3KakycUJ5/23fOT77zomFr5WkvjlSVJIkSZIkSVKjWBSVJEmSJEmS1CgWRSVJkiRJkiQ1ikVRSZIkSZIkSY1iUVSSJEmSJElSo1gUlSRJkiRJktQoFkUlSZIkSZIkNYpFUUmSJEmSJEmNYlFUkiRJkiRJUqNYFJUkSZIkSZLUKJO6nYAkSdJoNXFi+fvHkyaV/2/V5MmTi2PnrLVWcezs2bOL4jZYf4PiNrfZZpvi2DXXKLs/wLQ5jyqOHa822KLbGZRb/1GbFcdeceUVRXGTJvqniIbP7FmzymPtO8cU+077TmmoOFJUkiRJkiRJUqNYFJUkSZIkSZLUKBZFJUmSJEmSJDWKRVFJkiRJkiRJjWJRVJIkSZIkSVKjWBSVJEmSJEmS1CgWRSVJkiRJkiQ1ikVRSZIkSZIkSY1iUVSSJEmSJElSo0zqdgKSJEkjaZVVyv/3Z/311y+O3XjDjYpjZ68xuzj2eXvtUxwL0/oRq7Fkyb03F8dOnrVWYeSqxW3ec8stxbHLly8vikuyuE2NLauvvnpxrH2nhpN9p6TeWBTVqDNxQvkA5mnTuvs/MHu/aK/i2ClTphTHbrTRhsWxX//GN4vi3vrmNxe3uem2Ty2O7Y/f/d9PimOPOe64YclBkiRJkiTJ6fOSJEmSJEmSGsWiqCRJkiRJkqRGsSgqSZIkSZIkqVEsikqSJEmSJElqFIuikiRJkiRJkhrFoqgkSZIkSZKkRrEoKkmSJEmSJKlRLIpKkiRJkiRJahSLopIkSZIkSZIaZVK3E5AkSRpJEVEcO2ettYpjH/e4xxXHrrXmGsWxMK0fseq2m6+6qDj23e97b3HsdddfXxy7z957F8U95clPLm5ztalTi2Pvv//+orjly5YVt6mxxb5T/dWfvvM7R35vWHKYPHlyUZx9pzR+WBRtiHXmrF0UN2mV8m+JbR/96OLYxz7mMcWx06dNL47deOsnFceOLQuKI9/5jrcXxW205RMHmkyvFtx5bXHsJZdeOiw5SJIkSZIk9YfT5yVJkiRJkiQ1ikVRSZIkSZIkSY1iUVSSJEmSJElSo1gUlSRJkiRJktQoFkUlSZIkSZIkNYpFUUmSJEmSJDVWRHwuIu6KiNu7nctoEhE7R8TNLc+viIidu5fR0LIoKkmSJEmSpEaKiI2BDwCPzsx1I2L/iLhgiO+xf0RkRBzWdnyv+vgPhuAeB0fE0ohYEBH3RsQfIuJpg223VWZum5nnDWWbfYmIHSLirIi4JyLmRcTJEbFey/mIiC9FxN31x5ciIkratigqSZIkSZKkptoYuDsz7xyKxiJiUg+nrgFe3nb+dcBVQ3Hf2o8zczowB7gA+GlpgXAUmw18D5gLbALMB45uOf9mYG/gccB2wJ7AW0oatigqSZIkSZKkcSsiDoqIayJifkRcGRH71Md3Bc4C1q9HWP4Y+A7wtBUjLuu4KRFxaETcGBF3RMR3ImJqfW7niLg5Ij5cT78/unMW3A5cBjyvvm4NYEfgtLZcT46I2yPivoj4XURsWx+fHBEXR8S76ucTI+L3EfHJ9htl5lLgGGBdYM2IWD8iTqtHW14dEW9qud+UiDg8Im6tPw6PiCk9fB2vr79mK0amnhQRP6y/rldExJNaYp8QEX+vz50cET+OiM/10O4REXFKy/MvRcQ5ERGZeUZmnpyZ92fmA8A3gZ1aLn8d8JXMvDkzbwG+Auzf+SV4pJ6q15IkSePSsmXLimNvvvnmvoNqFzz0++LYW269pTj2ObvsUhz7jN1eUhw7llx7xZ+KY79y2GF9BwFrrLFGcZvP33334tgzzzq7OPZnvzq/OLY/ttryysK4LYvbzOVZHPvAA4uK4pb3o02NLfado0N/+s6jfvCDorj+DDgbrr5zuCxZsqQo7r777itu075zVLkGeAZVYfJlwHERsXlmnh0RewDHZeaGUE11B96YmU9vuf6LwGbA9sBS4ATgk8BH6vPrAmtQjWTsbQDiD4H9gF8CrwROBR5sizkDeD2wBPgScDywfWYuiYjXAOdHxNnAi4GJwOfbb1IXNfcHbsrMuyLid8DlwPrA1sBZEXFNZv4G+BiwQ/25ZZ3Tx4FP9PJ5rPCiOo8DgM9RFSx3iIjJwM+ArwLfphq9+SPgyz208wHg4vprfw3whvpz7vQN/0zgipbn2wKXtDy/pD7WJ4uiY9gWm21WHHvQwT1937WbOLBkNKR+XPg/JQCLFi8uC/zNucVt3n333cWx8xfML469qR//gyxJkiRJ0lDIzJNbnv44Ij4CPIWqANirevr5m4HtMvOe+tgXqAqjK4qiy4FPZWZ7gbPdz4DDImImVXH0A8Aebbke1XLvg4H/RMTMzLwvMy+vR1v+HFgbeEpmtr7j//KIeCFVQfVyYJ+I2IhqZOULMnMxVfHxf+v7/wZ4NfCuFcsHRMSnge9SVhS9IDN/VV93LPDe+vgOVDXHr9eFzZ9GxF96aiQzH4iI11IVhOfX+axUQIiI7aiK0Xu1HJ4OtL5bcR8wvR5l2uu7CE6flyRJkiRJ0rgVEfvVU8/vrafEPwZYq/DyOcBqwIUt1/+6Pr7CvLrg2KvMXEQ1SvTjwJqZ+Yjh8vWU+C/WU/3vB66vT7XmegzViNRfZea/225xUmbOysy1M/PZmXkh1ejQezKzdUTTDcAG9eP16+et59bv63Op3d7y+AFg1XrN1PWBW9qKkjf11lBm/hm4FgjgpPbzEbE5VdH0PZnZOt1mAbB6y/PVgQV9FUTBoqgkSZIkSZLGqYjYBDgSeCdVIXIW1SjKntaDaC+m3QUsAratC46zMnNmvaFRT9f05odUI0SP63DuVVSjIHcFZlJtLkRbrt8GTgeeFxFPp2+3AmtExIyWYxsDt7Sc36Tt3K0F7fbmNmCDtk2eNurtgoh4BzClvveBbec2Ac4GPpuZx7ZdegXVJksrPI5HTq/vkUVRSZIkSZIkjVfTqIqW8wAi4gCqkaI9uQPYsF4Xk8xcTlVUPSwi1q7b2CAinjfAfH4LPBf4RodzM6jWGL2banTqF1pP1lPMn0i1Xui7gWMiYjq9yMybgD8A/xMRq9ZT0N/Aw0XZE4GPR8SciFiLanp6p4Jtf/wRWAa8MyImRcReVMsVdBQRW1KtSfoa4LXAgRGxfX1uA6pp/t/MzO90uPyHwPvr12R9qoLzD0qStCgqSZIkSZKkcSkzr6TakfyPVAXPxwK97fL2G6qRhrdHxF31sQ8DVwN/qqe1nw1sNcB8MjPPWbE+aZsfUk1fvwW4EvjvjmkRsTFwOLBfZi7IzBOAvwElu0zuSzXq9FaqdU0/lZkrdjj7XN3OpcBlwEX1sQHLzCVUGzC9AbiXqth5OitvKkU93f444EuZeUm9JMBHgWPrDaPeCDwKODgiFqz4aGniu8Av6twvp1qe4LslebrRkiRJkiRJksatzPwY1S7rnc6dB2zY8nwJ8IK2mMVUhbqP9nV9D/f4AT2MXszMj7c8XsAjNxGCqlC6wppt176i5fHBvdz/ZuCFPZxbTDXq9N0dzp3HI782c3u6X2ZeT8s0/8z8G9WO9gBExJ+pipft93iItlGkmXkEcET99NP1R0f12qEH0jblvoQjRSVJkiRJkiQNmYh4VkSsW0+ffx2wHdUGVaOGI0UlSZIkSZIkDaWtqHaRn0a1q/xLM/O27qb0SBZFJUmSJEmSJA2ZzPwe8L1u59Ebp89LkiRJkiRJahRHikqSpEaZP39+cexf/vrX4tirr7m6OPbu+4tD+fQXv1Uc+563vroobuHCB4rbzFxeHPu0HXYojl28eHFx7PePPro49u9X3FgUt8HaU4vbnDF9RnHs1KmrFsdOm1wcyuO336Y4dqONNiqKu/vuTpvednbzzTcXx957371FcQ8te6i4TY0tDy5ZUhx7w403DEsO/3fWWcMSOxw2nTu3OLY/fedtt98+gGx6V+1nUub8C3rb3PuR+tN3Lp2/tDh2OAxX37lkadnPzfJ+/F6W1DuLomPYHXfeWR78UOEfgJNmDSiX0W7R3dcXxy5YsLA4ds4m2w4gm76dec7Zw9KuJEmSJElNsNrUVXPmzPI3NsejSZPKy36rTFqlOHbGjOlFcUsfKn8DdJV+5LqkH2++/fOqa+7KzDmdzlkUlSRJkiRJ0rgyc+YM3vDal3Q7ja5aa621imPXntOxbtjRs57xzKK4eXfdVX7/tcvvf8MNZTODAHZ8zl49TklwTVFJkiRJkiRJjWJRVJIkSZIkSVKjWBSVJEmSJEmS1CgWRSVJkiRJkiQ1ikVRSZIkSZIkSY1iUVSSJEmSJElSo1gUlSRJkiRJktQoFkUlSZIkSZIkNcqkbieggbt//vzi2KOP/F5R3A5P3aG4zauvvro4ds+Xv744tj8euv+WorgPHXRQcZuLFi8ujp27ySbFsS/ZZ5/iWEnS8FmeWRx7+RXlv+sWLhlINkPrm985vihujZnl74tPmTKlOPYl+7y4OHbOnDnFsT8/9dTiWLixKGr2rFnFLc6eXR671pprFce+ZJ/nF8fOmDGjOHb58uVFcddff31xm0uWlH+DL33ooaK47MfPojSerb/e+sWx/eo7T+tP3zn0hqvvvPDvFw0gm6Fz6223dvX+6r7HPHrborhXHtCfOsi0gSUzym2wxfC0u/6jNhuSdhwpKkmSJEmSJKlRLIpKkiRJkiRJahSLopIkSZIkSZIaxaKoJEmSJEmSpEaxKCpJkiRJkiSpUSyKSpIkSZIkSWoUi6KSJEmSJEmSGsWiqCRJkiRJkqRGsSgqSZIkSZIkqVEsikqSJEmSJElqlEndTkAj44I//KEo7u8XX1zc5gMPPFAcu9lmjyqOffQTdy6OPeknPymKW7R4cXGb/XH9DTcUx37l8MOHJQdJUv9MXXVqcexWW25SHHvR5eW/E4bLssK4u+9bXtzmtlvNLo7dfPPNimO32G6n4tj3vOv24tiFD3yxKG7ttdcubnPq1PLvmXl33VUcO336jOLYzOJQFi9+sDxYUtcNV995++3lfeef/vLn4thSw9V3St32uMc9rjBy2rDmMRZ8/ENvLY59+k7l/dtq/ehfeuNIUUnSsIqI6yNi1y7nMDciMiJ8M1CSJEmSZFFUkqTeRMShEfHviJgfEf+MiP3azm8fERdGxAP1v9t3KVVJkiRJUiGLopKkUa/LIzwXAnsCM4HXAV+LiB3rvCYDpwLHAbOBY4BT6+OSJEmSpFHKoqgkacRExDYRcV1E7Fs/f2FEXBwR90bEHyJiu5bY6yPiwxFxKbAwIjavp8C/LiJujIi7IuJjLfETIuKgiLgmIu6OiJMiYo2CnDaLiHsi4gn18/UjYl5E7AyQmZ/KzH9m5vLM/DNwPvC0+vKdqdbnPjwzH8zMrwMBPHsIvlySJEmSpGFiUVSSNCLqouP/Ae/KzBMj4vHAUcBbgDWB7wKnRcSUlsv2BV4AzAIeqo89HdgKeA7wyYjYpj7+LmBv4FnA+sB/gG/1lVdmXgN8GDguIlYDjgaOyczzOnwOU4EnA1fUh7YFLs18xBYol9bHJUmSJEmjlEVRSdJIeAZwGrBfZp5eH3sz8N3M/HNmLsvMY4AHgR1arvt6Zt6UmYtajn06Mxdl5iXAJcCK7R/fCnwsM2/OzAeBg4GXlky9z8wjgauBPwPrAR/rIfQ79T3/r34+HbivLeY+oHxLaUmSJEnSiHMXXknSSHgr8Nu20ZebAK+LiHe1HJtMNcpzhZs6tHV7y+MHqAqTK9r7WUQsbzm/DFinMMcjqQq3b66Lqo8QEYcAjwF2aRkZugBYvS10dWB+4T0lSZIkSV3gSFFJ0kh4K7BxRBzWcuwm4POZOavlY7XMPLElJil3E7BHW3urZuYtfV0YEdOBw4HvAwe3r0UaEZ8G9gB2y8z7W05dAWwXEdFybDsenl4vSZIkSRqFLIpKkkbCfGB34JkR8cX62JHAWyPiqVGZFhEviIiBTj3/DvD5iNgEICLmRMRehdd+DfhbZr4R+GXdFnU7HwFeBeyamXe3XXce1WjUd0fElIh4Z338NwP8HCRJkiRJI8Dp85KkEZGZ90bEc4FzI2JpZn4iIt4EfBPYAlgEXAD8boC3+BrVzu9nRsT6wJ3Aj4FTe7uoLpzuDjy2PvR+4OKIeHVmHg98AVgCXN0yIPQLmfmFzFwSEXsD/wt8EfgHsHdmLhng56ARMHHixOLYbR9dvmfW6qu3r6TQs7vvuac49rJ/9jnYud+W9x3yX9OmTSuOXWvNtfqfTIEXvuyA4tjJUyYXxV199dXFbS5atLg49sabbiyOlSSw7wT7To0tx594QllgaVw/7fS0HYtj58yZUxz789N6/bPpEZ74+CcUx5a69trrimMXLXpgSO5pUVSPsPCBofnGWqndhQuHpd0X7LFHUdy5551X3Oby7M9sXUl9ycy5LY/v4eGNkcjMXwO/7uu6+vn1VEXP1mM7tzxeDny1/mhva6VrW86dSkvhNDMXAJu3PO94Xcv5vwNP7C1GkiRJkjS6OH1ekiRJkiRJUqNYFJUkSZIkSZLUKBZFJUmSJEmSJDWKRVFJkiRJkiRJjWJRVJIkSZIkSVKjuPu8JGnETFttas6atfqQt/vQ0oeGvM3+uPOue+7KzDldTUKSJEmSVMyiqCRpxMyatTpvf8O+hdFR3O4dd95R2GJ5m/0J/fp3T7ihPFqSJEmS1G1On5ckSZIkSZLUKBZFJUmSJEmSJDWK0+clSZJ6MGPGjOLYrbfaalhymDjhz8WxF19505Df/9777i2//yWXFMfussHwfL12e9Gry+L60eZN/7qwOPYnPz2lOHbhwoXFsUuWLC2OlTS22Hf2r++8+557+pGFNP7s8NSnFsfO7kefcc9/yn+27rzzzuLYUrfeduuQt9kXi6IaEccce2xx7JZbbFEcO3O9LYvinvD4JxS3+beLyn95S5IkSZIkaexx+rwkSZIkSZKkRrEoKkmSJEmSJKlRLIpKkiRJkiRJahSLopIkSZIkSZIaxaKoJEmSJEmSpEZx93lJ0ogJggkTJhbFLli4oLjdzLK4KatOLm7zgQceKI6VJEmSJI0tFkUlaZSLiG2BvYAdgfWAdYAFwG3Av4FfAGdl5uKuJSlJkiRJ0hhiUVSSRqmI2AU4BHhiDyFbATsDbwLmR8SXga9mpkMcJUmSJEnqhWuKStIoExHTIuLHwG/ouSDabgbwWeCqiHjmsCUnSZIkSdI4MOQjRSNiLWBTYD7wr8zSld4kSRGxIdV0+O1bDt8J/Bg4DbgKuIOqCLo+8GzgpcBOdewGwFkR8ZbM/MHIZK0miohVgC8BmZkf6HY+o0Hpern9temmmxbHXnzlTUN+/39eM6849tjjjyuO3eU5zylPYsqa5bHDYKOtSt+fgje+flZx7Ek/Obk49oorryyOffDBJcWxGjn2m+rJOef+pjjWvrN/fed1119fHCuNFRdfcklx7C4bbFUcu9uLXj2QdHr1/W99sTi2Gz+v/SqKRsRewMuB6cBfgcMzc0F97jHAt4Cnt1zyn4g4DPiCxdFmW7S4fKnDQw87rDj2s18+oijube97X3GbV/Wjg7nqqquKY39+2qnFsf6wNFNEzAbOBTavDy2jmj7/hcyc3xb+IHAXcClweEQ8H/hafe1k4OiIeCgzyysUUv9MBt5L1WX5x70k9c1+U5KkUaS4KBoRhwIrKksBvBB4bUQ8FZhFNc1zzfrcCmsAnwG2Bl47BPlK0rgUEROBH/FwQXQ+8IrMPKPk+sz8VUT8ATiFavQowPci4srMvGjIE5YkaZyIiAmZubzbeUjSWDKUfWdETAa2ASYCV60YfNfHNS8DpmbmD4ciBzVT0Zqi9fp076cqeF4H/LT+d3Pgc8DBwFpUuyC/G9gD2B/4Q33NqyJi96FNXZLGlXcCu9WPlwIvAH4dEd+LiF17uzAipkfEqVTT6XcHLqhPTQWOrwuuUp8iYlnpB3A/9cD2tnMPdfezkKRyETEFOC0iPjjA6+03JTXOYPvOlnYiIj4NzAMuopqRPC8ijo6Idfu4/OvAUYO5v1Q6UvTN9b8/B16emQ9FxCTgZGBfqqkgfwWelZn/nScdEccBZwDPBV4P/HqI8pakcSMipgMfbTn03sw8PyK+Q7Wz/KsjYo/M/F2Ha6cCpwPPAp5S//tS4O/AelQj9V+H/8MwatX/w7fqMN/m9tbfz72lM8x5SNKoUf9RfwrVG5EviAgy89D+NjP0mUnS6DVEfecKRwIH8Mi+dAqwH/D8iHhVZp7TWzoDvK8ElBdFn0b1ruaHM/MhgLow+mHgn/W5j7b/wZWZyyPiQ8DFVH+sS5JW9iZg7frxxcAR9Rqhb6mPrQb8MiKem5l/WnFR/T8kP6cqhAKsCxyZmc+KiE8B36uPfxSLoqPZj3j4NRwuuwDnFcYm8EfgrD7iJgMfqeM/M+DMJKl7tuPhJWcADhngH/f2m5KaZEj6zojYhWrwXFINoDsKWEz1/61vAeZQ/Q20b2b+bEgyl9qUFkXXAxZl5r9bD2bmVRGxiGqEy997uPYyqg1B1u7hvCQ13V4tjz9Xb0z3q4j4MnBgfXw61XT652TmhfUOtifz8JR7gOt5eP3mo4FPABsBm0XEYzLz8uH8JDQufBT4JNWbodcA78/MuzsFRsQ0qj/uycxPj1iGkjREMvOv9Uayp1ItOQP9/+PeflNSowxR3wnVwBCAUzPzxS3HT4+Ib1L9rfME4EcR8erM/Mmgk5falBZFh2JIssOaJalNRKwO7Fg/fZBqyREAMvPD9WjQ99SHZgJn1muMfgTYs6Wpm4FnZ+aN9bUPRcTpwNvq8y8Aul4UjQnBlFWnFMUuWNDn+ur/NWXy5LLALG6SNddcszx4nMjML0bEz6jeqX8tsHtEvD8zj+9yapI0LDLzrMH8cW+/KamJBtt31nagnnXcof3rIuLpwIlUA0hOiIjIzJOHIH3pv4o2WgLuAKZGxGatB+vnK34Atu/h2m2p1oS4cyAJStI4tzmwSv34z5n5QOvJzHwv8J2WQ2tQreH8spZjt1EVRK9ra7t1/Z1tBptoRGwWEe+NiK/XH++LiC0G227TZebOmRnD/HFeP/L5F/B04H1USzf8MCJ+HRFzh+crIEndlZlnUf3Rvajl8CGlG4jYb0pqosH2nVRLfy3KzH/20P5iqr0STqYa0Hd8veO8NGRKR4r+mWoK5ufrhW6XR8QE4H/q8/OAL0TEMzNzyYqL6pgvU1X/LxzCvCVpvFin5fEtPcS8nerNpQPq5627yd9BVRD990pXPbK9dTqc/6+IeDbwYGb+vsO5CcBhdR7tb6Z9ud4Q6j2Zuby3e2jsqJdw+FpEnEa1AP5uwOX1WrWH+VpL6paI2AqYMQxN/wf4LPCFlmP9GTFqvylp1BqtfSd9zOPKzGURsS+wHHgFVWE0nUqvoVJaFP1fqlFJLwMeHxGXAI+jGuH0INXiuKcDf4+Ib1Otp7MJ8EbgiVTf6McNbeqSNC7Mbnn8n04BmZkR8SbgGVT9bqs39/TuKnBPy+M1+sjjbKoRpxt0OHcksD8PL4NyV/3vWlQF2rdTrS39ppWu1JhWjz7etf7++3L9sW/9/KquJiepqb7L8G9O1+qQiLgjM48tCbbflDRKjca+8xbgURGxbmbe3lNQPSjv1VR/i7ycujA6xPmqoYqKovV6Ed8D3gxsQfVH+Yo/jj+cmb+KiJOpiqZf79DErzPzp0ORsMa/W2+7rTj2B0ccUhS3/9tKR/DDlo/baVhiV1111eLYM8/ua/PSh919zz19B2k0u7fl8axe4r7KygVRgG9ExCWZeUOHc60F13s7nG+30trPEfEsqhGqSbVm0Acz85r63KOAQ4G9gddHxFGZ+ceC+2iMycwjI+KXwPeA51PNIPled7NqjnXX6XWg9yPsstN2RXHn/v7SgabTq6NP+EVx7N337NV3UO1TH/t4cewTnr57cexwmLHOZn0H1d7wjncVxx59xLeKY/924UXFsSqybX8vsN/UcPn4xz9WHPvivfYujrXvhGuuvbY4VkX66jv/DjwKeA7Q6zrMdWH0VVR/r7wMOGFIMhyjzjn3N8WxM1dfvTh2OPqBN7zjoOLYj3/orUN+/76UjhQlM98aEX+mWkB8PeBa4IjMPL0O2Z9q9NCbWtpdRDXK9MNDlbAkjTN3tDzesFNARBwCvLuH6zcGzqmXL7m17VzrqM8e333twxvrf8/OzH1aT2TmtcCLI+LXwHOpZg1YFO2niFiXaqTtcLq9XpdpwOrvrxdGxGuAw3l4Ey9JGs8S+MuALrTflNRcJX3nuVRrhu5HH0VR+G9hdN+67Ze33EcasOKiKEBmHg0c3cO5RcA7IuLDwJZUaz78c7B/hEnSOPdv4CGq/vgpETG17k8BiIjPA61DnRcAL6R6s2mP+thmwG8i4lmZ2VpkfXbL438MML8dqf5n41O9xBxMtXbajgO8R9P9iOGfzrQLcN5QNJSZx0XEmcCXgLlD0aYk9cMrGb43kl4KtE5DSuCNg53xZr8paRQYjX3nqcC3qJYc2T4zL+7rRi0jRpNqjVFpUPpVFC2RmQsA5+xIUoHMvD8i/gA8k+p/VHYHfgZQb87w0ZbwB4AXZuZv65H7p1NNNwHYCjg7InbJzLsiYiKwZ8u1vxxgiusBS6im/fXkz1TrS3cc6arxJzPv5OGNvyRpxPS27txgRMRLeXgTWXj4j/qjhqJ9+01J3TQa+87MvDUiNqbao6Dj3go9XLdijdFvs/ImsFK/+A0kSd13asvjT0RlL6oRmCs8COydmb8FqEfhvwg4vyXmMcCKxcwPoJpaD3BdZl42wNyWAQvrXXU7qs/NZ/ingEuSNOTqP+pP5OEBI0NaEJWk8Wgo+s7MvDkzb8jM+/tz78xcnpnnr/jbSBqofhdFI2KDiNgrIl4cEVsUXvP+iPhk/9OTpEY4EphXP3488FaqUaCn1MeWAi/NzEfswJWZDwAvAP5UH5oHfDAi1gY+3RL6+UHk9i9gdkSs0kfcDPrxDq8elpk7Z2YM88d5/c3L3/eSmmAoC6L2m5KawjeTNF4UT5+PiFWB7wKvaTv+O+CdmXlFL5d/CFgb+MxAkpSk8Swz59drhx5eH/oacDmwL/Bj4PiWTe06Xbs7cDJVX3sVcA6wfh3yL+AHhanMiYj2bTfn1P9uVee0kojYFJjCw4XdHgUwcULZ+3FrrbVmURzAPfeU1WOnT59W3OaSJUuLY8cTf99LaoqI2IEh+KPeflNSkwxV3znIHKYB36CatPaGkbqvxp/+jBQ9heoXfbR9PAv4S0S8bujTk6TG+CawYiToKlRrgO6amS/OzFN6vgwy877M3A24Afg18Iz61GLg1Zm5rDCHiVQbQLR+TKPq61/cy3XPrf+9uPA+Gt38fS+pKf5K9eYjDO6PevtNSU0yVH3nYKwK7F9/SANWNFI0Ivam2uU4qd4F/T7VH9u7AAdRbcRxVESsnpnfGJ5UJWn8ysxlEfEK4C/A5lTT0X8REYcAX8jM+b1dHxF7AF+vr13hTZl5YWEKfW3+cGsv51a8O/vHwntplPL3vaQmqX/3vg5YDpw3wCnze2O/KalBhqLvlEaL0unzr6P6Rf/9zHxby/HLI+KHwHFU69odHhETM/PwoU1Tksa/zPxPROwC/ALYnmrk5kHA6yPiR8BpVNPj7wSmAxtQ/dH1MmCnlqaWAG/NzOP6ce9jBpH6y+t/7xhEGxod/H0vqVHq2RT7DaIJ+01JjTMEfac0KpQWRZ9U//u59hOZeV9EvIhqhNI7gK9ExITM/OoQ5ShJjZGZN0fE04GjqYqdUK019u76oy+3AK/KzN8NU4orycwbRupeGnb+vpek/rHflKQBGOQmc6sNWSJqtNKi6BxgYWbe2OlkZibwroh4EHg/cEhE4C98Dbfz//D7orhbbu1t5u8jve0tbymOXWPDrYtjd91z3+LYddZZuzj2uBNOKIq76+67i9tUd2XmQuDl9ajRQ4AnFlw2v479Sr0rvTQQ/r4fhSLKl4DffLPNiuLuvPPO4jav+PftxbH9cdqvy36HA/zrqvLBKJ//TNl+NS959RuL2+zH3qT9VL75294velFx7NVXX1MUd+999xW3qR7Zb2rU+umpPy+Ove6G64vixnPf+ZXDDx9ALhqEg6lG2guYULgZLcAas2cXx/7mvHOLYzfccMPi2LXnPqY4tlT0I3aovnFKe6mHqKZx9iozPxgRy4EPUv3CX+4UEUkamMw8F3hSRDwG2AvYkWptsnWABcBtVNPpTwfOzMzF3cpV44a/7yWpf+w3JWlw7gAe7Oc1E4CNhiEXNUxpUfQGYOuImJuZ1/cWmJkHRkQAH6CaImLlX5IGITMvBy7vdh5qBH/fS1L/2G9K0sDcSFXYfG9mntSfCyNiLap9FqRBKR2fe1H9724lwZn5IeCrVKNfv0o1rUSSJI1u/r6XpP6x35Skgbmw/vcJA7jWN5U0JEqLoudQ/eLev7ThzPwgD//CL18cQZIkdYu/7yWpf+w3JWlgLqLqBwdSFJWGROkv4dOAZcBTI+JZpY3Xv/APHUhikiRpxPn7XpL6x35TkgZmMCNFl1FNv79h6NJRExWtKZqZ9wCrDOQG9do538B3QSWp8RJYvrxstkv0YwfG5bm8KG7ZsmXFbUZ/tj8cJ/x9L0n9Y78pSQP2W2AXgIiIzCyeEp+Z9wJzhyctNUnpRkuDkpk3jcR9JElS9/j7XpL6x35TUlNl5iKqwqjUNb4rKUmSJEmSJKlRLIpKkiRJkiRJahSLopIkSZIkSZIaZUTWFJW67drrryuO/fTnPlscu+MOTyuOfcX+7yyOfexTnlMc++ENNiiK+9BBBxW3KUkam1ZZZXJR3BMe//jiNidNuqw49pJ/3Fwc2x//unZeceyf//KXorhn7PT04jbXnvuY4tjhMnuDrYtjJ0xo4E5xkgbl7xdfXBT3jJ3+WdzmWOs7S3vO4t2A1HhPesITi2P33vdNw5hJmbtuvLKr9+/Gz5YjRSVJkiRJkiQ1ikVRSZIkSZIkSY1iUVSSJEmSJElSo1gUlSRJkiRJktQoFkUlSZIkSZIkNYq7z0uSRszNt9x+1/s/+oUbup3HMNik2wlIkiRJkspZFJUkjZjMnNPtHCRJkiRJcvq8JEmSJEmSpEaxKCpJkiRJkiSpUZw+L0mSNE5MmjSxKG7u3LnFbT5++8cXx0YUh7J06dJ+tFve8NZbbz3k9x9rli/PbqcgjUsTJpSPKZo9a3ZxrH3n6GDPOf5MiAlMnjy5KHbHHZ5W3O6ue+470JRG3EUX/Lo49qen/rw49tHbbFMUt81WZX1Lt1gUldosWLiwOPbMc84ujn3F/u8cSDp9WmODsk5m++22K27z4ksvHWg6UldExP7A0QWhz83M8h9cSZIkSdK4ZFFUkjSeLAfm9XL+wZFKRJIkSZI0elkUlSSNJzdl5txuJyFJkiRJGt3caEmSJEmSJElSo1gUlSRJkiRJktQoFkUlSZIkSZIkNYpFUUnSeDInIi6MiAURsSgiro2I4yJi524nJkmSJEkaPSyKSpLGk9WAJwBLqH7HbQq8Gjg3Io6KCDcYlCRJkiRZFJUkjQu3Ap8GHgesmplrUBVIdwLOrmMOAA7rTnqSJEmSpNHEoqgkaczLzDMz8+DMvDQzH6yPLcvMPwDPA06tQ98eEVt0LVFJkiRJ0qjgNEJJ0riWmcsj4oPAXlRvBu4JfLW7WanpJk4sf1962rRpxbHrrrNuUdwWm29W3ObTdtihOHb2BlsXxw6Xu268sijunHN/U9zmrqtOKY5dc6NHF8f2ywN3FocuWbJkeHKQxqk5a80pirPvHN99p8afVVddlW22Lvv52nXPFw1zNr373ZmnFMeeedZZxbGrz1i9OHbDDTYojp0/f0FR3Ck//1lxm91gUVSNsNmmmxbHPvMZzyyO3WqrLQeSzpBaNv+2orhLL7tsmDORRq/MvDoi7gLWAh7V7XwkSZIkSd3l9HlJkiRJkiRJjWJRVJI07kXEZlSjRAGu62YukiRJkqTusygqSRrTIiIKzh9SP10OnD7sSUmSJEmSRjWLopKksW6TiPhLRLwlIh61okgaERMiYgfgDGCfOva7mfmvrmUqSZIkSRoV3GhJkjQePLn+AHgwIuYDM4DWbU+PBt490olJkiRJkkYfi6KSpLHuDuBdwNOA7YE5wGxgMdX6oX8AjsrM33crQUmSJEnS6GJRVJI0pmXmIuCb9YckSZIkSX1yTVFJkiRJkiRJjWJRVJIkSZIkSVKjOH1ekiSpBxMnlL9/PHnK5OLYOXPmFMdutcWWxbGP2XbborjNt9uxuM3R4cHiyCv/cWVR3Gm/+EVxm/+++uri2N12fW5x7FOe/OS+g2qXX355cezy5cuLY6WxZMaMGcWx9p0wHH3nZf3oi8Za36nx54FFD3DJpZcWxV7yoQ8MczZDZ6MNNyyOXWWVVYpjr73uuoGkM6ZZFNWo058f8H322rso7nE77NSPDKb2I7b77rnnnqK45ZnDnIkkSZIkSdLY4PR5SZIkSZIkSY1iUVSSJEmSJElSo1gUlSRJkiRJktQoFkUlSZIkSZIkNYpFUUmSJEmSJEmNYlFUkiRJkiRJUqNYFJUkSZIkSZLUKBZFJUmSJEmSJDWKRVFJkiRJkiRJjTKp2wlo7Fpj9uzi2Ofs8uzi2N1f9KLyJCbOLI/tsgf/c2Nx7HEnnFgc+4c//XEg6UjSuDNhQhTFzZ5V/vtrvfXWLY5dc401i2Of8+zy34trbrRNcWz3PVAcedEFvyuOPemUnxTH/ubcc4vi7r777uI2Z82aVRx79TVXF8euMqn8f8X/dtGFxbEPLnmwOFYqNSHKx9PYd/bX8PSdPz315wPIZegsXry4OHY09J0af26/4667Pn/od2/odh7quk16OmFRVJIkSZIkSeNKZs7pdg4a3Zw+L0mSJEmSJKlRLIpKkiRJkiRJahSLopIkSZIkSZIaxaKoJEmSJEmSpEaxKCpJkiRJkiSNYhGxWkTsEREfj4ifRsQNEZH1x8F9XLtBRLw9Ik6OiKsjYlH9cV1EnBgRzx6hT2NUcfd5SZIkSZIkaXR7CvCr/l4UERsBNwDRcviB+vnc+uOVEXEU8ObMXDboTMcIR4pKkiRJkiRJo99/gHOAQ4B9gdsLrplIVQA9B3gdsEFmTgOmA9sCp9ZxrwcOHuJ8RzVHikqSJEmSJEmj2/mZuUbrgYj4YsF1/wGemJkXtR7MzOXAlRGxD9UI1N2B90bE5zNz8VAlPZo5UlSSJEmSJEkaxQY6rT0z72sviLadT+Co+ul0YJuB3GcscqRoQ8yeNasobu7cucVtvvPt7yhPYOqc8tguW3T39cWxPz7p5OLY3//xD8WxyzOLYyVprIkIJq+ySlHs7DVmF7c7Z621iuK22HyL4jZ3ff4exbFMnFUe23UPFEde9PvfFcee/stfFsf+31lnFcfeeNNNxbGzZs4sitt6q62L25wxY0Zx7MKF5V/bK//5j+LYm2++pTjW/43Q5MmTi2PtO/uj+33nkqVLi2PHktHQd0oN1zoydGLXshhhjhSVJEmSJEmSmmvn+t8lwFVdzGNEWRSVJEmSJEmSGigiNgXeWj/9cWbe3818RpJFUUmSJEmSJKlhImIqcDKwGnAXcFB3MxpZFkUlSZIkSZKkBomIScAJwBOBpcCrM/PW7mY1siyKSpIkSZIkSQ0REROB44G9gYeAV2XmmV1NqgssikqSJEmSJEkNUBdEjwNeDiwDXpOZP+luVt0xqdsJSJIkSZIkSRpeLSNEX8HDBdEfdzer7rEoKkmSJEmSJI1jdUH0BB45QvRH3c2qu5w+L0mSJEmSJI1TLSNEX061huirm14QBUeKSpIkSZIkSaNeRMwGJrYcWjHYcbWIWKvl+OLMXFBfs2IN0Vfw8KZKJ49EvqOdRdFRZvUZM4pj3/+e9xbHbrTZZmWBk2YVtzkaLLjz2qK4k08pXzP4r3/7W3Hsg0uWFMdKkioTJ0xk+vTpRbHbbL11cbtbbLZ5Udz2O+5U3CZM60ds9/3uzFOK4n76058Vt3nZFZcXx95///3FsZMmlf9vaH++DzadO7cobvPNy75fAFafsXpx7H/+c09x7L+umlccu2jR4uJYjU9R/1fCvrN/SvvOM886a5gzGf1mzpxZHDtcfecNN84vjpXGob8Dm3Q4/qH6Y4VjgP3rxzsBr6wfJ/CNiPhGL/d4T1PWGbUoKkmSJEmSJI1PrUtnrgKs00f81GHMZVSxKCpJkiRJkiSNcpk5dwDXnAeFUw0axo2WJEmSJEmSJDWKRVFJkiRJkiRJjWJRVJIkSZIkSVKjWBSVJEmSJEmS1CgWRSVJkiRJkiQ1ikVRSZIkSZIkSY1iUVSSJEmSJElSo0zqdgKSJKlZMpOlDy0tir3zznnF7d55551Fcaf+4hfFbW688UbFsQsXLiyO/fvFFxfHXvXvfxfHXvCXfxTF9ed/AB/7mE2KYzfZeOPi2E3nblocu+466xTHrr76jKK4++fPL27z31eXvwaLFy8ujs0sDpXI+r8S47XvvP2OO4pjBavPWL08dhj6zptuvqk4VpK6waLoAG2z9dbFsa942cuKYzfasrxdmNaP2LHjD+ecWhx7wo9OLIpb1I8/UCRJkiRJkjS+OX1ekiRJkiRJUqNYFJUkSZIkSZLUKBZFJUmSJEmSJDWKRVFJkiRJkiRJjWJRVJIkSZIkSVKjWBSVJEmSJEmS1CgWRSVJkiRJkiQ1ikVRSZIkSZIkSY1iUVSSJEmSJElSo0zqdgKSJKlZlj60lHnz5hXF/u7884vb/eOFVw00pcaZu8kaxbFPedKTimO33mrr4tjVV59RHDtv3l3Fsddce21R3KJFi4rbzCwOlUaF226/rav3L/05VGWtNdcqjh2uvvPmW24pjpWk8cKi6AA9fcedimM32vKJw5hJoUVlf3z+6U9/LG5y+fLlxbE/Pvnk4tgFCxcWx0qSJEmSJEn95fR5SZIkSZIkSY1iUVSSJEmSJElSo1gUlSRJkiRJktQoFkUlSZIkSZIkNYpFUUmSJEmSJEmNYlFUkiRJkiRJUqNYFJUkSZIkSZLUKBZFJUmSJEmSJDWKRVFJkiRJkiRJjTKp2wlIkqRmWbJkCTfceGNR7B8vvGqYsxn95syaWBy78UYbFcU95clPLm7zcdttVxy7dOlDxbHXXHNtcez98+8vjl2+PItjJWm4bLThhsWx/ek7r73uuoGkI0nqwKLoAB151PeHJVaSJEmSJEnS8HL6vCRJkiRJkqRGsSgqSZIkSZIkqVEsikqSJEmSJElqFIuikiRJkiRJkhrFoqgkSZIkSZKkRrEoKkmSJEmSJKlRLIpKkiRJkiRJahSLopIkSZIkSZIaxaKoJEmSJEmSpEaJzOx2Dv+11pqzcq/n79ztNCSpX4467tQLM/NJ3c5DGisiYh5wQ7fzkNR1m2TmnG4nMVbYd0qq2XdKQ2RStxOQJEnN4v/IS1L/2XdKkjS0nD4vSZIkSZIkqVEsikqSJEmSJElqFIuikiRJkiRJkhrFoqgkSZIkSZKkRrEoKkmSpAGLiDUj4oCIOC4iroyIhRHxYETcHBE/j4h9BtDmQRGRKz6GI29J6ib7TknqPneflyRJ0mDcziP/n3IxsBTYoP7YKyLOAF6amQ/01VhEbAV8ajgSlaRRxL5TkrrMkaKSJEkajEnAX4C3A5tl5tTMnA5sCny/jtkD+G5fDUXEBOAoYFXgj8OTriSNCvadktRlFkUlSZI0GM/OzKdm5hGZee2Kg5l5fWa+kYf/oH9NRGzUR1vvAnYEjgfOHJ50JWlUsO+UpC6zKCpJkqQBy8xz+wj5fsvjJ/UUFBGbAp8H7gbeNwSpSdKoZd8pSd3nmqKSJEkaTotbHk/sJe5IYBrw9sycFxHDm5UkjW72nZI0zBwpKkmSpOG0c8vjyzoFRMSbgOcAZ2fmD0ciKUka5XZueWzfKUnDwKKoJEmShkVEzAI+Uj89PzP/1SFmA+AQYBHwlpHLTpJGJ/tOSRoZTp+XJEnSkKt3Qz4WWI9qGug7ewj9LjAT+HDrZiOS1ET2nZI0chwpKkmSpOHwNeCF9eN3ZOal7QER8RrgBcDFwFdHLjVJGrXsOyVphFgUlSRJ0pCKiEN5eHTT+zLzqA4x6wCHA8uAN2XmQyOXoSSNPvadkjSynD4vSZKkIRMRXwY+UD/9YGYe3kPoF4E1gSOAf0bE9Lbzk1vaXHFuSWYuGcJ0JWlUsO+UpJHnSFFJkiQNiYg4BPhQ/fTAzPxKL+Gb1v++DZjf4eMjLbErjn15SBOWpFHAvlOSusORopIkSRq0etrnilFOB2bmId3MR5LGAvtOSeoei6KSJEkalLY/6j/YxygnADJz5z7aPBj4VB0bg0xRkkYd+05J6i6LopIkSRqwtnXw3p+Zh3UzH0kaC+w7Jan7IjO7ncN/RcQ84IZu5yFJ/bRJZs7pdhKSNNIiYmMe/n+35cC8Pi45NDMPLWz7YBztJGkcsu+UpNFhVI0UtaggSZI0pkxoe7xOH/HtuyRLUhPZd0rSKDCqRopKkiRJkiRJ0nCb0HeIJEmSJEmSJI0fFkUlSZIkSZIkNYpFUY05EbFmRBwQEcdFxJURsTAiHoyImyPi5xGxT0Ebz42IkyLihohYHBGLIuLaiDg+Ip41Ep+HJEmSJEmSusM1RTXmRMRSHrlJ2GJgGTCt5dgZwEsz84G2awM4AnhLy+FF9b9TW44dlpnvH7KkJUmSJEmSNGo4UlRj0STgL8Dbgc0yc2pmTgc2Bb5fx+wBfLfDtfvzcEH0J8CWmblaZq4GbA2cWp97X8mIU0mSJEmSJI09jhTVmBMRu2Tmub2c/w4PFz43zsybWs6dC+wMXA1sk5kPtV27CvBP4FHAjzJz3yFOX5IkSZIkSV3mSFGNOb0VRGvfb3n8pLZz69X/XtJeEK3bXgpcXD+dPqAEJUmSJEmSNKpZFNV4tLjl8cS2c9fW/z4uIia1nVsxUnT7+unfhj41SZIkSZIkdZtFUY1HO7c8vqzt3BH1v5sDJ0bE5itORMRWwElUU+evAQ4bxhwlSZIkSZLUJa4pqnElImYBV1JNkz8/M5/ZIea9wJeAyfWh1t3n7wVOBD6emfcMc7qSJEmSJEnqAkeKatyIiAnAsVQF0cXAOzvFZebhwIuBO+tDU+sPqAql04GZw5mrJEmSJEmSuseiqMaTrwEvrB+/IzMvbQ+IiNUi4sfA6cCNwG7AnPpjN6pRpq8F/hIR241I1pIkSZIkSRpRTp/XuBARhwIfqJ++rx4N2inuW8DbgX8B22fm4rbzU6l2n98SuCAznzFcOUuSJEmSJKk7HCmqMS8ivszDBdEP9lIQnQG8uX76rfaCKEBmLgK+WT99ekSsPcTpSpIkSZIkqcssimpMi4hDgA/VTw/MzK/0Er4lMKl+fE0vcf9uebzpINKTJEmSJEnSKDSp7xBpdGqbMn9gZh7SxyXLWx5v0kvcOi2P5w8kN0mSJEmSJI1ejhTVmNRWEP1gQUEU4J/AovrxGyNipTcFImIiD0+x/w/V2qOSJEmSJEkaRyyKasxpW0P0/X1Mmf+ver3Q/62fPgH4RUQ8NiIm1B/bAb8CdqxjDs/MZUOZuyRJkiRJkrrP3ec1pkTExsAN9dPlwLw+Ljk0Mw9tuX4q8FNg95aYB+t/p7QcOxF4rUVRSZIkSZKk8cc1RTXWTGh7vE5PgbXprU8yc1FEPB94CfAa4InA2kACNwF/AY7OzF8OWcaSJEmSJEkaVRwpKkmSJEmSJKlRXFNUkiRJkiRJUqNYFJUkSZIkSZLUKBZFJUmSJEmSJDWKRVFJkiRJkiRJjWJRVJIkSZIkSVKjWBSVJEmSJEmS1CgWRSVJkiRJkiQ1ikVRSZIkSZIkSY1iUVSSJEmSJElSo1gUlSRJkiRJktQoFkUlSZIkSZIkNcr/A0IVCRAqg5t3AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 1080x1080 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA9MAAAECCAYAAAD98ITmAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAABWYklEQVR4nO3de7xlc/3H8dd7xmAYZjCDXKfIJYWQWxQlkVsXKfd7qaSLpCSXSil+kZRKMeSa5JJK7hVFqSgkwrgbMxhmGMbMfH5/fL/b7LNmX9Y+Z5+z9znn/Xw81mPOXuu7vt/v3mudM/uzvjdFBGZmZmZmZmZW3ohOV8DMzMzMzMxssHEwbWZmZmZmZtYiB9NmZmZmZmZmLXIwbWZmZmZmZtYiB9NmZmZmZmZmLXIwbWZmZmZmZtYiB9NmZmZmZmZmLXIwbWZNSdpKUlRtN3W6Tjb8SNqvcB9O6nSd+ouk4wrv9bhO18nMzMx6cjBtZmZmZmZm1qKFOl0BMzOzoULS+sD7q3ZNjohJHamMmZmZ9SsH02ZmZu2zPnBs1es/AJM6UhMzMzPrV+7mbWZmZmZmZtYit0ybmZl1mYg4Djiuw9UwMzOzBtwybWZmZmZmZtYiB9NmZmZmZmZmLXIwbWZmZmZmZtYij5keJiStAqwHrAQsSbr2LwHPAQ8D/42IJ9pQzuLAZsAKwARgJDAVeAy4JSJe6msZhfLGAZsAywPjgUWAGbm8e4D7IiL6kP8ypPezHOn9vAI8DTwE/DUi5vSl/g3KXQjYFHgzsDQwK5d7W0T8r43lbAisTbpeAqYA/4yIO9tVRjtJWhbYmHQtxpPu4xdI9/DdEfFQH/NfIee/LLAM8CLp/r0P+Edf7qUm5S4KvB1YCxhHuoenkH5nHuuPMvubpNVIM1svCywFPE/6LO+OiLs7WLVhp1PXQtKSpPv6jcASudwngD9GxLT+KtfMzGzARIS3IboBo4Ejgf8AUWJ7DPg5sFMvyvogcAMp2KyX/8vAb4BN+vi+FgE+BfwNmNvkPU0BzgXeBahk/gL2BG5rkv904AJg7Rbrv1Uhn8lVxxYHvg4806Dce4Bd+/D5jQQ+T3ogUK+MycChwMg6db5pAO/jJYGvAHeVuIcfAX4EbNxC/qPye/1Xk7yfznmv3GL996v32ZEeCHwfmNmg3FuBd5Uo5yM16juql5/5ssDsQn6blzhvDHA0cH+J6/QdYOk+fpaTWvxda/m+JS1rVZ3HfiXKaXWbWCPP4wppjmux3v19LYr1m1R1bCLp/5J6/x/MA64BNujt3wVv3rx58+atG7aOV8BbP11YeCupta43X+ymt1DO2sDtvSjjLGDhXryvDwCP9/J9vbNE/m8E7mgx3znAKeTAs0QZWxXOn5z3vwV4oIVyzwBGtPj5rQL8s4Uy/khqGS/W+aYBuo8PAZ7t5fVetUT+mwAPtpjvLOBLLbyH/Wp9dvkzndpCuUc1KWcRFnwI88Fefu6HF/L5T4lzdgSeavGznA7s24fPclKLv2st37cMwmB6gK5FsX6T8v7daPxwqHp7Fdin3X83vHnz5s2bt4HaPGZ6CJK0OnAjKXCqZSYwjRQU9KWcbYG/ABs2KOf5Osf2B66VtFgL5X0NuJTUJbmWV0iBV72u12qS/ybAn0nd4Wt5IZdRNBL4LHCFpNGNymhQ9jrATcAbapRZr2v8IcAxLZSxMum+WL9OkudJrZHVtgR+Cyxatpx2kDRK0pmkBwZL1Uk2izRMYW69bJqUsRPp83h9nSTTqX0vLQp8S9JPJfXqb6ikrYGrSS3TFZWA5uU6p50gab96eUbEK8D5hd0H9KZ+pN/Pamc1SizpY8DlpOEQC1SNdJ3m1Tg2FpgkqfR9bI118lpI2hO4iNTDpmJeLrP4twXSMI1J+ffBzMxs0HEwPTSdRvpiVPEKqSvplsASEbFEREyIiMVIgcH6pC/PF5OCt6YkbQBcWShnFvATYGtg8VzOONIXq+1IXbyrvQP4Qcnyvgh8lZ4BUpCC6w8C4yNi0YhYJiJGASsCO5O65U4pkf+ywBX0DG4Abs75j4mIsRGxKCn4OoIUuFfbgfTZt2o06cvv0vn1ZaTPa/Fc5uKkse5HsODDiaMkrdGsAEkjgV+yYLD+Z+a/v3ERsUhO8yVSYAep9fbYFt9TX30POKiwbw4pqHsfMDYiFouIpUndtN9AahH7OfUf4LxG0tqkL/3Fhx+/Jn32i0bEUsDCpN4XX2fBh08HAl9u4T1VLA9cQmpJnpvf0zsrZUbEaGB14AQWfHhziqR6DxdgwaB3O0mva6VykjYG1qnaNYc0VKJe+q1JDz1GVu2OfM6WpB4oS5M+yw1Jf4uKDymOl7RHK/XsMvcBn8hb8bO6v+pYve2ZdlSiw9fizcDPSH+jXyb11tm4Umb+2/IW4If0DOYF/DT/jTIzMxtcOt007q29G+mLevU431eAt7dw/uLAXk3SLMmC3ZHvBNYokf/+pC9v1efu3OScrWqc8ySwWcn3NIrUPfRNDdL8mgW7IB5Dg3HWpJaff9Y4b5cS76dWl8cXgQ80OXdd0gOP6vO+W+IzOKJGed9s8v5eR/1xyjf14z28d43y7gXWauEe/gywbJ3jI1iwK/9c4IAm+b6RBYdOvAps1OS8/ep8hlOb/W4C29S49w9rcs4/CumPbPHz/1Hh/MsbpF2SNNdCdfpZwHZNytiE1FpZfd7zNBmPXuOznNTi71rL9y0lunk3qWOvfldosZt3B65FsX6V7cFmv6t1fica/j/gzZs3b968dePmlumhZ3169ji4LCJuKXtyRLwYEec1SfZZerZwPgi8OyLuK5H/2SzYmtesde879GxpeZY0GdJfmpWXy3w1IiZFxD21judW9h0Lu0+PiK9FRDTIdwrwHlJgX+24MvWq4cCIuKxRgoj4F/C1wu4PNzonzxR9ZGH3uRFxVJP39yQpmCu2wPcbSQsD3yrsfoB0ve8tk0e+h78XEU/XSfJ+FuzK/8WIaNiVOSLuJ30e1d3uFyJNjtaqOaQHJw1/NyPiOlJwW63h9Sa1DlYrdtmuKw9T+Ghhd6PP5SBSL5Bqe0XE1Y3KiYjbSD1HqlsolyRNjGe90w3XYgbw3ma/qxExiQV7KjW7r83MzLqOg+mhZ+nC6wfbmXke4/zpwu5PRWvLnJxCz3ptKmn9OuW9B3hbYfch0cclkAo+U3j9BCW77+b3fXhh9/qS3tliHa6NiItKpj2Lnl0zV5JUa3xkxW6kZZ4qnmfBOtcUEU/Ru2Cxt/alZ0AQwB4R0c6Avni97yDdk03lgPobhd07S6o37rqesyPi5pJpf1x4/dYmY7UvoOe46zUlbV6yrA/Sc+jGU6Qx8wuQJBb8W3BlRFxapqCI+BPw08LuAyWNKVlXy7roWpyYf0fKKN7XG7VYlpmZWcc5mB56phdeFwPRvtqWnuOK72vW8lEUaW3mXxZ2b1UnebG14j81zu2r7QuvfxYRM1s4/xekGcarva/FOpxRNmEOLP9T2L1Wg1N2Lry+qMWHH+dQcix9GxSv9zUR8dd2ZZ7Xvd2isPv7EVFrQqZ6zqBnsDqCNM66FaWvN6mrffXnvziwcr3EEfEcaQx+tbITkRXTnRv111J/E2kJpGrfK1lORfEhxhKksb3Wmm64FvNIc2aUVeyVsUZvJ/QzMzPrFP/HNfT8jdSaV7GNpG9KWqRN+W9VeP27Xubz98LrzeqkK87yen6jrsmtypN3TSjsLtWaUxERc4FfFXa/vcWq/KHF9MUeB+MapN208PryVgqKiFmkmaf7Ve7iXfzcft7mYjal59+9YMFr11BETAeuLexu5XpPJ80xULa8IK0JXm1ck9OKXb13k7R4zZSZpIks+PvWqIt38T0/S5odvbTcHfjfTfK15rrhWtzVykO6/FCwerLAEaQA3szMbNBYqNMVsPaKiKmSLiF17a34MnCwpF+QumzeHBFNZzyuoxiYLS/pkF7ks3bh9QIzDudWxNULu0uP/y7pLYXXLwN39yKf2wuv123h3Bd60Y15RuH1krUSSRrPguMoiw8yyvg7Pe+p/rAWUFwqrb+v9wM5OG7V7cBOVa9bud6PtNgSDiWvd5XrSZOlrZpfL0Fq9Z/U4Jz96Tlb/i0R8d8G6Yuf5T96+aDr9kJerXyWlnTDtZjci/Jm0HNYwZKUmI3fzMysWziYHpo+TVr2ZLWqfeOBT+ZtnqR7SGtE/wm4ISKK3ZTrWb7w+iN566viWG9YsMUY4H9tKKvaMoXXjzTo1trIA4XXS0gaFRGvljh3ei/KK66tXG9ZmeL7eykipvaivMm9OKdVxes9JyLaXW7x8+jtnALF613Mt5HpvSiv7PUGUmu2pEn0XNJsf+oE03nM7b6F3cXW7aJu+Cwt6YZrMb0X5bV0X5uZmXUbd/MegvIsxpsAF9Kzy3fFCNKaoAeT1h99VNIfJX20xJi1/vqiW6sLaq2ypre53OKavb0dG1yrNaXWA4Ja2tZtvYb+fH/tVrze/VFmf30eZa819O/1rnZ2oawtJa1WJ+27md+KDTCTNBdAI93wWVrSDddioO5rMzOzruFgeoiKiGciYg/SxDQnkrou1/uyI9JEMxcCf5P0xgZZL9zWivasg5m1SUQ8TOruXSHqT0RW3H9xRLzYLxUzMzMzGyIcTA9xEXFvRHw5It5M6uq9C2nd5puBWl2QNwBuklQcZ1vxTOH1JyNCbdgmligLmk+81KrnCq+bjUWtZ2yNfQO2PnMD/fn+2q14vfujzP76PLrhWtdSnEBsn2LvE0njgA80Oa+W4fJZDob/J4fLtTAzM+sqg+FLgrVJRDwbEVdGxJERsSWpW+0ewG2FpCsA36qTTXG87fiaqdqj1tje4oRkfVUM4FaW1Jtxe28ovJ5Rcrx0fyu+v8Uk1RqL3szENtSlmeL1XijPMN1Oxc+j1fWhK4rXu9aDn25wGT0DrZVIy9tV2x1YtOr1vRHx5xJ5D4bPsh1jcse1oR79bTBcCzMzsyHHwfQwFhEzIuJC0rJUxXVvd5W0aI3T/ll4/dZ+qRwQES8A9xd2F9cI7qt/FV6PBtbpRT4bNcm3I/JSNU8Udm/Yi6x6c06r/gMUuxb39/VeXVJvWsC78noXRcTLwAWF3cUu3cXXZVqlYcH3vGGeyKxV/flZFmdB783SSxPbUI/+NhiuhZmZ2ZDjYNoq69h+iZ7dvkcDtcZOF9fX3VpScTmjdiqulbpnL78k1hQR97Ngi+gHW8kjd5stntPuJZ364tbC6/e3crKk0cB2batNHbklv/i57dXmYm4FqpelEgt2cW4oB9/vKezuputdVAyOd5a0NICkN9MzgJpDmpSwjGLr9dLAO1upmKQ1WXD5pXZ+ltMLrye28vcjLy335hbLLK4GMBAzVA+Ga2FmZjbkOJg24LVW4GKXvlozbP+ONNNvxTjgU/1ULVhwRuG1gF3bXMZvC68PbPEBwYdJ3Wer/aZvVWqrKwuvPyqplVnZ96H3YzBbVbze20rauF2ZR8QM0nJw1T7V4gOaQ0gPmyrmAVf3tW79JSL+AdxRtWsRYM/8c7FV+jcRMaVk1vcADxX2fbrF6n228PoFFrw+ffEoPXs7jGXBNZkb+QStT45YbA0fiPkGBsO1MDMzG3IcTA8xknq1dnhugSmOpX2ymC4ingVOK+w+XtImvSk3l133y2pEXA/8tbD7R5J6OyawluL7WQn4epkTc1D63cLuf0bEH9tRsTa5mJ4TCY1lwTrXJGl54IT+qFQdPwceq64CcEGlJbVNitd7I+DQMidKWh34amH3FRFRDGS6TbF1+gBJo1iw5b/Z2tKvyT1aTi/s/qCkncqcL+ntwMeK5UfEzFrpeyMi5gJ/L+wullmTpLWAL/ai2OLfzdUl9dcqCMDguBZmZmZDkYPpoedbki7OX45KyQH46fTsjvhgXlqnlpOBR6pejwZ+L2n3VioqaSlJn2fBCdCKjqDnREJLA7dI2rRkOaMk7SvpTbWO55a7qwq7Py+p4RfpPJHXNaQJ26odX6ZeAyWPmz2psHsfSSc0epCRA+lr6b+1xRcQEbNJQw6qrQb8OXdDbUrS4pIOk7RsnSSXA3cW9p0iaY8m+a5G+jyqe2zMYWAfNvTW+cArVa/XB46h5wO0p0g9T1rxU+Dxwr4LJb2r0UmS3gb8mp7/B70AnNJi+WVcUnh9iKRiN/0eJK1D+izG9KK8f9Ozq/do4KBe5NOqwXAtzMzMhhQH00PPKGA34GZJj0j6nqRdJa1ebLWWtJKkfYHbgY8U8jm1XgER8RywMwt2n7xA0u2SPi5p7eKs2JLGS9pK0uclXQdMAf6PBWeQLZb3R+Cowu7XkQKsSyS9X9JShbJWkLSTpB+SunpOAuoFV5C+7Ba7t35b0o2Sdq7u9i1pFUmHkybM2qBwzpkRcUWj99MhJ7NgC91RpPvk/YX3t2p+kPAf5o8XLY677jcRcT4LToi3JnCXpJ9K2k7SaxNJKZko6cOSziG1bH8PqNlVPyLmkWawfqlq90jgfEmXSnp3pSUx572mpONJkzFNLGR3bEQUP9euk3uUXF7Y/ZXC63Miojjet1m+LwB703Mc+uLAdflabVb5uyNphKT1JZ1KGuO7VCG7T0TEo62UX9LP6dkzYyTw6/ww6bXVASQtJGlTSaeRflcmkh5AtHR9I2IW8PvC7tMlXSPpeEmHSjqksPVmYrRiuYPhWpiZmQ0tEeFtCG2kIDgabDNIE2693CDNlYBKlPUOUmtWvXzmkZbleY7UUlMv3bSS7+3rTd7by8A00kRqtY5v1ST/TfJnUy//6U0+t6uA0SXex1aF8yb34jpPKuSxX4lzVgUm9+L9/YU0AVn1vpv6+T4eRepy3Oh6v0ga51/v3prYpIydSAF1vfyfA2Y3OP5TYESJ97JfXz874KZW7uU6eWzb5PNcow/X6+MNrsNcUjDb6G/AsSXLKX6Wk0qet0eDsmfl+s0r7J9HGls+qbC/zO/aZtT/O1TqXgWOK6Q5rsuuRbF+pa5FIY/JzT4Hb968efPmrZs3t0wPPdHk+BjS2tCL1Dg2j9Td+4MR0SwfIrUYv5UFJ/CqEGmCsnE0ntH2n83KyuV9Ffgo8HSdJIuQuiTXGzc+r87+Sv63AZuzYBfgirHU/tzmklpCd4nUKtWVInXb34r6y93Uen9/BHYgBdkDJiJejYgDgc+QupzWshipy3+9e6vZ9f41sDULTtxUMY4U1Be9DHw5Ig6K1Mo9WFxHz+EZ1W6OiPt6m3FE/Jg0S3yt380RpJbPWtfpeVJw2q9DIyLiAuAL1P77uCipftVDHl4Cdo/US6I35f2F1Epc797tN91+LczMzIYSB9NDz5GkFqhTSUFqmW6b04GzgQ0i4tPRQlfPiHgyInYgTeJ0PgsuM1XLbFKQdgywZkQ0HL9YKO9i4PWkiYHuKnHKE6QWzrdHiUnBIi2VtQHpi/BfaRyQvQBcCLwlIj4babKjrhYRk0lrRh8B1BsTTz72aeBdkboId0REnEa63icA/ytxyoOkCcbWjYh6gWN1/reRZog/jOb301TgJ6QW3BNL1KWr5MB/Up3DZdeWbpT/VaTx7ccADzRJ/hhp6MFqEXFOX8suIyL+j7RcVKMZql8FLiL9Tl/cx/IuIn0ehwKXAfeR/ta+2uC0tuj2a2FmZjZUqEQDpA1ieSzs2qRxycsDlbF5le7edwH3thJANylPwJvytnTe5uXyppC+UN4XEa/UzaS18pYHNiaNhx6fd88gjZO+JyLKBGCN8h9P6rK5XM5/NqnF5yHgtnZ9bp0iaSPStVqe9HDtKdIySneW6Z0w0CRNJD3smEDqhTCX9FBjMnB3mQC6Sf4rMv9+WobUlXwq8F/gH934mXSrPGHbW0mf5TjSdZoK3BURd3ewapW/G1uS5l5YkrTc3/2kFvrnO1m3/tDN18LMzGwwczBtZmZmZmZm1iJ38zYzMzMzMzNrkYNpMzMzMzMzsxY5mDYzMzMzMzNrkYNpMzMzMzMzsxY5mDYzMzMzMzNrkYNpMzMzMzMzsxY5mDYzMzMzMzNrkYNpMzMzMzMzsxY5mDYzMzMzMzNrkYNpMzMzMzMzsxY5mDYzMzMzMzNrkYNpMzMzMzMzsxY5mDYzMzMzMzNrkYNpMzMzMzMzsxY5mDYzMzMzMzNrkYNpMzMzMzMzsxY5mDYzMzMzMzNrkYNpMzMzMzMzsxY5mDYzMzMzMzNrkYNpMzMzMzMzsxY5mDYzMzMzMzNrkYNpMzMzMzMzsxY5mDYzMzMzMzNrkYNpMzMzMzMzsxY5mDYzMzMzMzNrkYNpMzMzMzMzsxY5mDYzMzMzMzNrkYNpMzMzMzMzsxY5mDYzMzMzMzNrkYNpMzMzMzMzsxY5mDYzMzMzMzNrkYNpMzMzMzMzsxY5mDYzMzMzMzNrkYNpMzMzMzMzsxY1DaYl3S1pq/6vSv+RtKakOyTNkHTYQJQh6UeSvtofZXUbScdJOq8P5/9O0r7trJOZmZmZmTUmaaKkkLRQp+sCgy8uaBpMR8Q6EXHTANSlP30RuDEiloiI0yRNlrRNf5YREYdExNd7k1E/1a8r1Aq8I2L7iDhngOuxr6S/S3pB0mOSvlP9R0TS0pIuk/SipIcl7TGQ9TMzMzMza5ccX8ySNLNqW6HFPLaS9FhhX58a1Yo6ERcASPqJpP9Kmidpv7LnDZdu3qsCd7cjIyW1PrfSZXTLk59hbjHgs8B4YBPg3cAXqo7/AJgNLAfsCZwhaZ0BrqOZmZmZWbvsFBFjqrYnOl2hLnIn8EngH62cVKab92utpPnJwyWSzsvdmf8taQ1JX5b0tKRHJW1bde7+kv6T0z4o6eOFvL8o6UlJT0g6KHcxWD0fW0TSyZIekTQld5seXaeOq0m6QdIzkqZJOl/SuHzsBmBr4PT8BOZCYBXg1/n1F3O6TSX9WdJ0SXdWd22XdJOkEyTdArwEvKFQfrGMNSRNkvSNfHyr3Pp5pKSngLMljZd0VS7vWUl/kjRC0s9r1a/Ge94ldyt/QdIDkrbL+1eQdGXO83+SDq465zhJv5B0br4md0vaKB87UtIvC2V8T9JpzfItnFPridVkSdvkOh4FfCS/tzurPt+D8s8jJB2t1Br8dK7r2Hys0g1l33xfTJP0lTr1WDh/Pp/Or0dKukXSMQARcUZE/CkiZkfE48D5wNtz2sWBDwFfjYiZEXEzcCWwd62yzMzMzMyGgnrxW/5+/DtgBc1v2d6D2t/tx0r6WY7zHpf0DUkj87H9JN2c47znJD0kafuq8qvjgmZpXy/pj7mu10n6geq0kudY5zblRk1Jn8ix0KIAEfGDiLgeeLmVz6s3LdM7AT8HlgL+Cfw+57Mi8DXgx1VpnwZ2BJYE9gdOkbRBfgPbAZ8HtgFWB7YqlHMisAawfj6+InBMnToJ+BawArA2sDJwHEBEvAv4E3BofgKzO/AI85/MfEfSisBvgG8AS5NaKC+VNKGqjL2BjwFLAA9XF16jjPtq1HH5nPeqOZ/DgceACaTWz6NSVrF3sX4LvFlpY+Bc4AhgHPAOYHI+fFHOdwVgV+Cbkt5VdfrOOc04UoB4etV575O0RC5jJLAbcEHJfJuKiKuBbwIX5/e2Xo1k++Vta9JDizFVdazYAliT1Jp8jKS1a5Q1G9gL+Fo+/iVgJHBCneq9g/k9C9YA5hSu452AW6bNzMzMbCirGb9FxIvA9sATVS3bF1D7u/0kYA4phnsrsC1wUFUZmwD/JfUQ/Q7wM0mqU59GaS8A/gosQ4r9GjV8nQS8Ahwt6Y253ntFREvBc1Fvguk/RcTvI2IOcAkpGDwxIl4lBVwTlVuFI+I3EfFAJH8ArgG2zPnsBpwdEXdHxEvk4BdSV2pSwPm5iHg2ImaQ3vBHa1UoIv4XEddGxCsRMRX4LvDOFt7TXsBvI+K3ETEvIq4FbgfeV5VmUq7rnPxeWzUPODbXcRbwKvA6YNWIeDW3kkbJvA4EzsrveV5EPB4R90pamdS6emREvBwRdwA/BfapOvfm/D7nkh6KrAcQEQ+TujV8IKd7F/BSRNxaMt922RP4bkQ8GBEzgS8DH1XPrvHHR8SsiLiTFOTWCsqJiLtID0guJz0g2Tu/7x4kHQBsBJycd40BXigke570IMXMzMzMbDC6XKlX7HRJl9dK0CR+a0rScqQY6rMR8WJEPA2cQs847uGIODN/Lz+HFBMtVyfLmmklrQK8DTgm9zSt9CStKSLmkWKXw3K670TEP8u+r3p6E0xPqfp5FjCtKkCZlf8dAyBpe0m35q7B00kf7PicZgXg0aq8qn+eQBrT+vfKBQeuzvsXIGk5SRflbgQvAOdVlVPGqsCHq26u6aTWz9fVqV9vTC08+TgJ+B9wTe5C8aUW8loZeKDG/hWAysOHiodJrfoVT1X9/BKwaFWgegGwe/55D+a3SpfJt11WoGfL/8PAQvT8BSu+hzEN8juHdH1/GxH3Fw9Kej+pV8P2ETEt755JehpXbUlgBmZmZmZmg9P7I2Jc3t5fK0GT+K2MVYFRwJNVcdWPgWWr0rz2XT43qkL97/P10lbik5eq0jaM1yJiMnAjMJE0P1Kf9dsEZJIWAS4ltfYtFxHjgN+SumQDPAmsVHXKylU/TyMF5utUXfCxEVHvQ/4mEMBbImJJUktzva4C5LTVHgV+XlXWuIhYPCJObHBOq3qcHxEzIuLwiHgDqev15yW9u2RZjwKr1dj/BLB0pat2tgrweMk6XgJsJWklUgt1JZhuJd8XSQ9CgNe6i1c/BGn23p4g/RJWlzOHng9xWvFD4CrgvZK2qD6QhxqcSepS/++qQ/cBC+UuIBXr0aZJ7MzMzMzMuk2J+K3W9/hacdUrwPiquGrJiGj3cMknSfHJYlX7Vq6XGEDSDsBmwPWkhs0+68/ZvBcGFgGmAnPyYPFtq47/Athf0tr5Q3htTebcDH8mqY/+sgCSVpT03jplLUFqTXw+j38+okndptBzErHzgJ0kvTdPVLWo0kRaK9U5v88k7Shp9dyl/XlgLqkreK36Ff2M9Nm9W2nCrhUlrRURjwJ/Br6V38O6pC7hpaarz13kbwLOBh6KiP/k/a3kex+ptXsHSaOAo0n3QcUU0lCAevfehcDn8oQCY5g/DmNOmfdQTdLewIakMdiHAefkPMnjvc8HPhQRf60+L48J+RVpvPXikt4O7ELqFm9mZmZmNhQ1i9+mAMsoTw5cte+17/YR8SSpa/j/SVoyxyqrSWplCG5TeYjq7cBxShMPb0aa26smSeNJw1QPAvYlxX7vqzq+cJ6MTMCoHPM0jZX7LZjOXYIPIwXNz5G6DV9Zdfx3wGmkpvb/AbfmQ6/kf4+s7M9dt68jTTpVy/HABqSg9DekQKiRb5EGn0+X9IUcLO5CmgRsKumJyhH078OGN5Le00zgL8API+LGWvUrnpiDv/1J4w+eB/7A/Nbc3UldF54ALiON076uhXpdQJoU7oLC/lL5RsTzpGnlf0pquX6RNHFZxSX532ck1Zp6/ixS0PpH4CHSjHqfbqH+AORxFKcC+0SakfsC0i/cKTnJV4GxwG81f0bC31Vl8UlgNGkShguBT0SEW6bNzMzMbEgqEb/dS/pe/GCOU1ag9nf7fUiB+T05n1/Sc/hsu+xJaml+hjRP0sXMjyWLfgJckeeOeobUMPhTScvk49eQekZvntPOIk1Q3JDKz3nVv/KMy3cBi/SmFdLMzMzMzMyGJ0kXA/dGxLEDVWZ/trw2JekDSutJLwV8G/i1A2kzMzMzMzNrRNLbchfyEXkupF1Iq/gMmI4G08DHSd1oHyCNGf5EZ6tjZmZmZmZmg8DypPmeZpKGD3+iHctdtaJrunmbmZmZmZmZDRadbpk2MzMzMzMzG3QW6nQFylhs9KIxduwSzRMOYQstVP5SjVpoVOm0SyxRb+nunl6dU34o+6gW6jp79uzSae+974FpETGheUozMzMzM+tGyyy9VKy04gp1jz/2+BM88+xzqpugiwyKYHrs2CU4cO8PdboaHTV+/PjSaZedUD7efOeWTWd8B2DqtGnly1+2fPkPP/xI6bSbv3uXh0snNjMzMzOzrrPSiitwzZUX1T2+7c4fHcDa9I27eZuZmZmZmZm1yMG0mZmZmZmZWYscTJuZmZmZmZm1yMG0mZmZmZmZWYscTJuZDXKS1pR0h6QZkg7rdH3MWiHpG5KmSXqq03UxMzNrhYNpM7PB74vAjRGxRESc1peMJN0k6aA21WvQkrSfpLmSZlZtW1UdnyjpRkkvSbpX0jadq+3gJWkV4HDgTRGxfP7cb25zGdXX8gVJd0raseS5W0l6rJ31qVHGRElRuNe+WnV8EUln5bo/Jenz/VkfMzMrb1AsjWVmZg2tCtRfY2IASVooIsovTN/d/hIRW9Q5diHwF+B9efulpDdGxNQBq93QsArwTEQ83Y7MGtx/f4mILSSNAA4GLpK0UkRMb0e5bTKuTt2PA95I+j1fHrhR0j0RcfVAVs7MrF2mT5/Ory6/rOHxwcIt02Zmg5ikG4CtgdNzi9YauSXrZEmPSJoi6UeSRuf0S0m6StJUSc/ln1fKx04AtqzK6/SqVrOFqsp8rfU6t/rdIukUSc8AxzUqv0b9q8+fLulBSZvn/Y9KelrSvlXpd5D0z9xK96ik46qOVer6MUlPSHpS0hfqlLtw7hr/6fx6ZK7HMSU+8zWADYBjI2JWRFwK/Bv4ULNzhyNJX5L0QB6GcI+kD+T92wDXAivk++1i4EfAZvn19Jyu0f28laTHJB2Zu4mf3aguETEP+DmwOClAbZh/k/cVklavej1J0jfqpD1D0qVVr78t6XpJalYOsC/w9Yh4LiL+A5wJ7FfiPDMz62cOps3MBrGIeBfwJ+DQiBgTEfcBJwJrAOsDqwMrApUgcQQp4FiV1Co4Czg95/WVQl6HlqzGJsCDwHLACU3Kr3f+v4BlgAtIrexvy+fuRQrux+S0LwL7AOOAHYBPSHp/Ib+tSYHStsCRqtEFOyJm57y/Jmlt4EvAyFz/ircqjeW9T9JXqx4orAM8GBEzqtLemffbgh4gPaQZCxwPnCfpdRFxHbA98ES+3z4CHEJqRR4TEePy+c3up+WBpUn39McaVUTSSGB/4FXg4ZL5t8PhwFvyQ6ItgQOBfSMiqtI8nB8MnC1pfK7vUsDrSPdXhe81M7MuMay7eb/5TeX+L/ro/ge0kOvivatMl1vxjf2T7wpvWK1/MjYbpnJL18eAdSPi2bzvm6Qg9csR8QxQ3UJ2AnBjH4t9IiK+n/Ob26j8Ouc/FBFn57QXA18BvhYRrwDXSJpNCnLuiIibqs77l6QLgXcCl1ftPz4iXgT+LelsYHfgumKhEXFXbkm8HFgW2Dgi5ubDfwTeTAq41gEuBuYA3wLGAM8XsnueFIRZQURcUvXyYklfBjYGrmh2brP7OSebR+ol8EqDrDbNLd2Lk67jXhHxdMn8+ywiXpK0N/A7YAbw6YiojMWeRnp4dAfpgdIPgPOB95LuNeh5vz0PLNGuupmZWe8N62DazGwImgAsBvy9qgepSK2uSFoMOAXYDlgqH19C0siqQLJVj5Ytv44pVT/PAoiI4r4xuf6bkFoS3wwsDCwCVAdrxfo8DLylQdnnkFqjL42I+ys7I+LBqjT/lvQ14AhSMD0TWLKQz5KkIMkKJO0DfB6YmHeNAcaXPL3M/TQ1Il5uks+tecz0GOBnpJbyX5TMvy0i4jZJD5Ie3Pyiav9M4Pb8coqkQ4EnJS1Butcg3V8vV/3se83MrAu4m7eZ2dAyjRR8rhMR4/I2NiIqLVyHA2sCm0TEksA78v5KJBE9s+PF/O9iVfuWL6SpPqdZ+X11AXAlsHJEjCWNsS2OO1256udVgCca5PdD4CrgvZLqTTYG6T1WyrkbeEMOdirWy/utiqRVSWN8DwWWyV2372LBa1ZRvP/K3E/Fc+rKgesngL0lvbVk/vW8ROPfix4kfYr08OcJ0gz8dauZ/x0REc8BT5Lurwrfa2ZmXcLBtJnZEJInWDoTOEXSsgCSVpT03pxkCVLwMF3S0sCxhSymAG+oym8q8DiwV56k6wCg7viMEuX31RLAsxHxsqSNgT1qpPmqpMUkrUMaH3txrYxyt9sNSZM5HQacUxmbLWl7Scvln9cCvkrulpzHpd8BHCtp0Tyh1rpUdZ+31yxOCg6nAkjan9SroJ4pwEqSFob+uZ9yd+6fAseUzT9f5+pNpHtgj/x7sR1puEFNSpPWfYM0Tn9v4IuS1s/HNlFaK36EpGWA04CbIqLStftc4GilyQPXIs1GPqm379/MzNrHwbSZ2dBzJPA/4FZJL5DGC6+Zj50KjCa1yN0KFJfX+R6wq9JM35U1qw8mdXF+hjR++M99KL+vPkmaNGwGaZKoX9RI84dc/vXAyRFxTTGB0vrGpwL7RMTMiLiA1NX2lJzk3aQx2S8CvwV+BXyzKouPAhsBz5G6ne/qZbEWFBH3AP9HWkZsCqnL/S0NTrmB1Or6lKRpeV9/3E+nAu+TtG6J/FckPYCq3lYDPgPsBEwH9qTnuP3X5InrzgO+HRF35uEERwE/l7QI6eHV1aSu23cBr5DG+VccS5rE7WHSvX2Sl8UyM+sO6jmRZHd63fIT4sC927/iiCcg6wJRftiXRiz594jYqB9rY2aDmKSJwEPAqCG01rWZmdmQstFGG8Xtt9/e6Di33357maUDO84t02ZmZmZmZmYtcjBtZmZmZmZm1iIvjWVmZkNCREym/izRZmZmZm3llmkzMzMzMzOzFjmYNjMzMzMzM2vRsO7mvd5665VM6Rm6jz7ikNJpt3j720unXWz06N5Ux2qQNBk4KCKu62AdJuLZlLvGEkssHhOWWbpU2lmzXi6d7wszXiiddvSi5X7Hl1tu2dJ5zpg5s3TaCePHl06rkQuXSjd58mSmTZvm7uQltXQfvlz+PuwPSy+1VOm0vg/NzGxYB9Nm1j0kTQL2AGZX7R4bEXPz8XcDPwBWAW4D9ouIhwe6noPJhGWW5mtHf75U2rvuvrt0vtdef33ptG9Zp9wShJ/7zGdL53nzLY2WKe7p4AMPLJ124XErlUq30UZeoa8VrdyH9/73v/1cm8Y+/KFdS6f1fWhmQ52kNwLrACsAS+TdM4AngLsj4v5O1a1bOJg2s9dIWqjDLcrfiYijizsljQd+BRwE/Br4OnAxsOnAVs/MzMxs6JI0BjgC2BtYtUnah4FzgZMjonx3nSHEY6bNhiBJa0t6SNLu+fWOku6QNF3SnyWtW5V2sqQjJf0LeFHS6pJC0r6SHpE0TdJXqtKPkPQlSQ9IekbSLyQ17cMpaTVJz0raIL9eQdJUSVuVeEsfJD0BvSQiXgaOA9aTtFYrn4uZmZmZ1SbprcB/gaOBiaQVMhptE4GvAvdKWn/AK9wFHEybDTE5WP098OmIuDD/YTwL+DiwDPBj4EpJi1SdtjuwAzAOqLRMbwGsCbwbOEbS2nn/p4H3A+8kdft5jtT9uqGIeAA4EjhP0mLA2cA5EXFTVbJP5oD775I+VLV/HeDOqrxeBB7I+83MzMysDyQtC1wDvA54GZgEfIT0XWscsHDexuV9HyF9l3uZ9H3w95ImDHC1O87BtNnQsiVwJbBPRFyV930M+HFE3BYRcyPiHOAVenaRPi0iHo2IWVX7jo+IWRFxJymQrczYdwjwlYh4LCJeIbUS7yqp6bCRiDgT+B9pzPPrgK9UHT4NeCOwLOkp5yRJldnsxgDPF7J7nvnjd8zMzMys944kNbr8B3hzRByQewT+JyJeiIg5eXsh77skIg4kBdb/AcYDX+xg/TvCwbTZ0HII8OdCa++qwOG5i/d0SdOBlUlPESserZHXU1U/v0QKaCv5XVaV13+AucByJet4JvBm4Ps5GAcgIv4REc/kP9S/Bc4nde8GmAksWchnSdIkGGZm1sUkfSMPGXqqeerhQ9JWkh6ren13yaFPZv1hRyCAvSPiobInRcRkYB9St++d+qdq3cvBtNnQcgiwiqRTqvY9CpwQEeOqtsUi4sKqNNFCGY8C2xfyWzQiHm92Yp7U4lTgZ8BxTcZaB+kPM8DdzG8ZR9LiwGp5v5mZdSlJqwCHA2+KiOUl7Sfp5jaXsV+e6+OUwv5d8v5JbSjjOEmvSppZNf/IZn3Nt1pErFN4GN7vJE3Mn9HMqu2rVccXkXSWpBckPSWp3NT8NhitDMyMiH+0emJE/J3UwLFy22vV5RxMmw0tM4DtgHdIOjHvOxM4RNImShaXtIOk3naR/hFwgqRVASRNkLRLyXO/B9weEQcBv8l5kfPZVdKYPMHZtsBepC7rAJcBb5b0IUmLAscA/4qIe3v5HszMbGCsAjwTEU+3I7MGQ4oeAHYrHN8XuK8d5WYXR8QYYAJwM/ArSUNlre9xETEmb1+v2n8caQjWqsDWwBclbdeJClq/mwUsWmbYXpGkUcCipPHTw4qDabMhJiKmA+8Btpf09Yi4HTgYOJ00Wdj/gP36UMT3SEHuNZJmALcCmzQ7KQfc2wGfyLs+D2wgac/8+jPA48B04CTg4MoT+oiYCnwIOCG/h02Aj/bhPZiZWZtUrfAwQ9I9kj6Q928DXAuskFs8LyY9RN2s0sKb0y0i6eS8gsQUST+SNDof20rSY3nViadIEx7V8hTwb+C9+bylgc2Z/1C2UtdLcgvr85L+KGmdvH9hpVUvPp1fj5R0i6RjigVFxKvAOcDywDJ5dYor8wSa/5N0cFV5i0g6VdITeTu1MAFodd0m58+s0hL+C0nn5s/1bkkbVaXdQNI/87FLJF0s6Rt18j1D0qVVr78t6fqSDwL2Bb4eEc9FxH9ID+j3K3GeDT53k5ZNPqgX5x4IjALuamuNBoFhvc70+RdeUC5h2XQtevtmm5dOO2FC+cnxLr/yitJpN3zrBqXTlvXgg6WHWTBr1kttL3+4ioiJVT8/S1W36Ii4Gri62Xn59WTmd6+u7Nuq6ud5wHfzVsxrgXOrjl0BXFH1eiawetXrLWudV3X8OsBLYbXglVdml/59/Nxhnymd70orrlg67ciRI0ulW2/TbUrn2UraaY/cUzrt+IVLNpzN6+RS7INPK/fhpz/5qdL5XnLpL0un9X3Y7x4gTYD5FPBh0qoNq0fEdZK2B86LiJUgdckGDoqILarOP5E0dGd94FXgAlIPpC/n48sDS5NaRxs1BJ1LGrv5G9ID1ytIE25W+x1wADAb+DZpfo71I2K2pL2AP0m6jjRnx0jSQ9wecjC8H/BoREyT9EdSELEC6f+payU9EBE3kCba3DS/t8h1Opo00WYzO+d67A98g/RQfFNJC5N6bH0X+CFpnOpFwHfq5HM4cEf+7B8gBT7rR0RUxdMPSwrSw48j8vtaijRZ6J1Ved1JWtHDhp5zSCu5nJp7L55emJh2Abm34KGk+zNyHsPKsA6mzczMzKxvIuKSqpcXS/oysDFVD1Drya2jHwPWzQ+CkfRNUkBdCabnAcdWT1pZx2XAKZLGkoLqw4HtC3U9q6rs44DnJI2NiOcj4q7cuns5aWWJjSNibtXpu0nakRSI3wV8QNLKwNuBHSLiZVLQ+tNc/g3AnqSlKp/OZR5PWqKyTDB9c56QE0k/Bz6b929K+g5/WkQEqbv5X+tlEhEvSdqb9CBhRq5PZeKzacDbgDtIMzn/gPSA4b3Mn3i0ejUNr6QxREXEzyR9GNiW9IDr6KoHRY+TJqMFWAxYkTSZ7Jak+0HA1dW/X8OFg2kzMzMz6zVJ+5CG7kzMu8aQlskpYwLpy/nfq1pJRWoVrpiaA9WGImKWpN+QWn6XiYhbcst4pZ6VluYP53Ln5UPjmR8wnpPTXBoR9xeK+EVE7FW9Q9ImwLMRUb26xMNApUv2Cvl19bHq1TQaKa6qURnPugLweA6kK2qtyvGaiLhN0oOkhwS/qNo/E7g9v5wi6VDgydwyOTPvX5L5Y2G9ksbQtjNpotiDSUHy+/JWj0gruvwE+Fx/V64becy0mZmZmfVKnozyTFJXz2UiYhypJaveeNzi6hHTSBMfrVO1QsTYPNFXvXMaOZfUIn1ejWN7ALsA2wBjmR/8V9f1h8BVwHslbUFzTwBLq+eknquQWvIqx1ctHHuiRL6NPAmsWBjz3HAWZUmfAhbJZTdaC7jyWY+IiOdyWetVHV8Pr6QxZEXE7Ij4JPAm4OvAbaSHKipsM/Oxr5N+dz8VEbM7U+vOcjBtZmZmZr21OCkAmwogaX9S9896pgAr5XG/lXk4ziR1z14257GipPf2sj5/IE3C+f0ax5YgjaF+htQa/s3qg7kr9Iak8dCHAecoLelYV0Q8CvwZ+JakRSWtSxqTXAnmLyR1l50gaTxpLHitQL8VfyG1Bh4qaSGlCT43rpdY0hqkMa17AXuTZuRePx/bRNKaSitpLAOcBtwUEZWW+nNz/ZeStBapxXJSH+tvXS4i7o+IYyNi84gYS+qRsGLelswPvDbPado5Y/6g42DazMzMzHolIu4B/o8U4E0B3gLc0uCUG0gtm09Jmpb3HUlaaeJWSS8A1wFr9rI+ERHXV8ZfF5xL6mb9OHAPaTUK4LX1sE8F9omImRFxAan78yk18inandTK/QRp3PaxedJMSEHs7cC/SLON/yPv67XcAvhBUtA+nRQkX8WCk61VlhI7D/h2RNyZu64fBfw8T6T2BtIEpTNIPQpeye+n4ljSpGUPkx5UnJQnNbVhJP9OPJm3mc3PGD48ZtpsCFps9OgYN7bc/CDRUu+59mtlgc4np0ybFhHlp7Y3M7N+FxFfIc1aXevYTcBKVa9nAzsU0rxMCvCOanZ+nTImUae1NCKOrvp5Jqmbd7Vzq35epnDuR6p+Pq5B+Y8BO9Y59jKplfuwGsduoudnM7FeecXVMvKyl+tXXku6Dfh1jTLmUGi1jogzgDPyywvzVlOe9O2AvJlZgYNpsyFo3NglOHDvD5VK20ow3XOuk76nAxg5onwHmW+c/OOHm6cyMzMb2iS9E/gvacz5nsC61FkC06zd8jrulaVYhzV38zYzMzMzG1zWJK35PJ004dquEfFkR2tkQ5qkiZImSXqWNEfCVEnPSTpH0hs6Xb9Occu0mZmZmdkgEhE/IS1HZNYWeU3paRHxwRrHNieNyx9LzxF6Y0lj9neRtH1E/GVAKttFHEybmZmZmZkNb1vQc21zACSNA34JjAPmABeTJhwE2BT4CGm2719IelNhzfUhz8F0B226ySal0y61YvlJLZ99rvzwhaeffrp02rKeeLKvyyeaWTvMmDmDP91yc6m0x3zztNL5HvqFt/S2SgNu5szyk45efuWVpdJNmzateSJ7zbyYx+xXyy0/utzry99bvg99H7ZimaWXipVWXKFU2unTp5fOd8Jyr+tljeqbPHky06ZNa2V+TrP+9HFgedKM7++JiL9WHfuhpO8D1wIrAPuTllcbNhxMm5mZmdmQttKKK3DNlReVSvuryy8rne8nPndMb6tU10YbbdT2PM36YEfSWvJfLQTSAETE3yQdQ1pabkeGWTDtCcjMzMzMzMysljflf89rkKZybPB0GWoTB9NmZmZmZmZWyxhgeqNlsPKx54ClBqxWXcLBtJmZmZmZmdUyFVi0RLoRQLkJMoYQj5k2MzMzMzOz0ZL2Kex7DnidpGUi4plaJ0lanDSj9yP9XcFu42DabAhqZfbckSNHls53oZHl/mTMi3nl81zIf4bMzMzMusCSwNl1jm0FXFrn2Eak9acf6oc6dTV/izXLJK0D7AJsDrwOWA6YCTwJ3A/8Grg2Il7uWCXNzMzMzNrvEdKs3fVsQf1getf8761trdEg4GDahj1JWwMnARvWSbIm6WncwcAMSd8BvhsRLw1MDc3MzMzM+k9ETOzL6cA5wOVtqcwg4mDahq08vuMsYLcWTlsC+DpwiKQ9IuKP/VI5MzMzM7NBICIO63QdOsWzeduwJGkl4GZ6BtJPA98H3gOsSpq5cAKwHvA54JaqtCsC10rabyDqa2ZmZmZm3cUt0x10x513lk679Yprlk677c579qY6Df3sByeWTvvQ5MltL7+dJC0F3AisnnfNJXXz/mZEzCgkfwWYBvwLOFXS+4Dv5XMXBs6WNCciGi1kb9YRiy66KGuusUa5xK8+Vz7jUZ1dRvKLhx1QOu255/+8dNrll1u+VLqnpz5dOk9r0SC6D48+4pCOlv/K7Fc6Wv5gM2qR0Sz3+reUSvuJz5VL16rJ9/y1VLrZL7/YL+WbtULS6hHxvxbP2Tsiyv/HOwS0tWVa0sKS1pO0gaQxJc/5cI0p2M36haSRwEXMD6RnADtFxJdrBNILiIjfAm8Dbqja/RNJG7S9smZmTUhyDzPrON+H1q18b/bJP8rGaJLGSDoPmNS/Veo+bbnBlBxPWtT7H8DfgKmSzpbU7FH/aaRxq2YD4VBg2/zzq8AOwNWSfiJpm0Yn5j8UVwArANuRuokDjAbOz4G6mdmAkLQIcKWkL3S6LjZ8+T60buV7s8/GkHpgnidpiXqJJL0NuAPYncazgQ9J7ermfSawP2l9sYpFgH2A9+WJmq5vcL4aHDNri9xb4qiqXZ+NiD9J+hFppu49JW1fa1IxSaOBq4B3Ahvnf3cF/klaRmstYF/8YKhr5Qd7i/ZzMU956TQbCPlL4qWkB4I7SCIiTu5wtWyY8X1o3cr3ZlucBHyBFCRvKmn3iPhbdQJJRwJfA0aRGlX3K5Px7JdfbDjsYTANdehzMJ2XFTqA9CTialIw8TKwNfBx0gROv8kX4LK+lmfWBwcDy+af7wDOyGOgP573LUa6V98TEa+tk5f/IF9OCqABlgfOjIh3SjoW+EnefxQOprvZRcy/hv1la+Cmfi7DDGBd4F1Vr0/yl0XrAN+H1q18b/ZRRBwp6TrSkldvAG6WdGxEnJgbKM4jfe8RcC2wT0RM6VyNO6Md3bwPzv9eERHvi4hfRsRVEXE48BZSt++FgYsk7Vo3F7P+t0vVz9+I5LfAd6r2jyF1+94QQNIo4BLmdw0HmAzsnX8+G3g0/7yapDf3R8XNzKrl1oFdgFlVu09yd0YbSL4PrVv53myPiLiWtKrN1aTW5xMk/Yk0Me+7gDnAERHx3uEYSEN7gulNSa3SRxUPRMRDwBbAFaQLcIGkD7ehTLOWSFoS2Dy/fAX4XeVYRBxJmqG7YixwjaS3AucDO1Udewx4V0Q8ks+dQ+r+XbFD+2vfOkmMWmhUqa0Vc+fNLbURUX5DLWzDj6Txkt4maS1Jw/NDsJryl5wB+bLo+9DqGaj7UNIoSd+V9H/tzNeGroH8GzmURcTUiHgfKdYT6fv0eOBBYLOIGNa/k+0IppcHZkXEvbUO5vGDu5Ja9xYiTdTkgNoG2uqkBzoAt0XES9UHI+KzwI+qdi1Nmkiv+l59khRIP1TIu3o+gLX7WlFJq0n6rKTT8vY5SW/sa77DXURsFRHq5+2mVuokaRdJ50u6QtLR1asgSHqzpD8AU4BbgbtJEzt+xcGMVbTjy6LvQ+urAQpaFgY+mzezUhxQt4ekNYGPkhpQK3/7J9CG772DXbsmIGs4c1tEzJW0OzAP+AgpoI6I+GWbyjdrZrmqnx+vk+aTpInz9s+vq2fnnkIKpO+vcV51fsvVOP4aSe8CXomIW2ocGwGckutRfND1nTxR2mciYl6jMmxwkHQy8LnKS2BHYG9JmwDjSMuvLUPPJvmlSRN9rMX8oQY2SOQvI3VnRO2D54CvA9+s2ldqfKDvw+GnG+9DM/C92a0kHQicCiwOvAB8hTTR2IbAuZK2Az4RETM7VcdOakcw/TjwBknLR8RT9RJFxDxJe5L+Q96NHFC3oXyzMpaq+vm5WgkiIiQdDGzJ/HWoKz5Wr/cF8GzVz0s3qcd1pBbuFWscO5P0x6nypXVa/nc8KbD/JGk26oMXONMGFUnvAD6fXz5ImhBvfdJ99w3S2P3xwH3A6cD9pAc1HyN1r9pD0vkRcfWAVtz66sf0/yR41U6SNCUifl7roO/DYatr7kNJc1vMK2qcFxHRrsYh66yuuTfttSGSPyH10hRwG7B7REyW9GPgRNLD2D1Is33vUZztezhoxx+ff5JmeHs3aXxpXTmg3oN0QT4MXNCG8s3KmF7187gG6b7LgoE0wPcl3RkRD9c4Vh2oT69xvGiBrpGS3klqEQ/SHANfiIgH8rE3ACcD7wcOkHRWRPylRDnWvT6W/70c2C0i5khaiDQcZndSd8a/Ae+sXmpL0nmk8f7vIa2i0DCImT17No8++mijJK/ZeOO3la789085pXTaTbbaqXmiFp3xk7NLp535Svl8X375sVLpZs8un2cXWKfBsQG5D1tx9FFfLp12px3KT1HRH/ehtaTefeihAtZpjf5GDnd3AKvmn08EvhoRc+G1OYO+IOka0mzfq5Fm+z4uIr7Vicp2SjuC6RtJY6L3oUkwDa8F1JVFvXer7G5DPQad62+8oXTasUsuWTrtBlts15vqNHTgp75UOu3RRxzS9vLboHqGwZVqJZB0EnBYnfNXAa6X9I6IeKJwrLqVuW7vjCYOyv9eFxEfqD4QEQ8CH5R0NfO/vDqYbpG6a53pzUh/947M/yGRA5kjgXvzsaOKeeW/n0eQ/oPbuK01t6EogPoLefo+tIHR7D4M0v9p1zbJZ2Hgyzn911qtxOOPPVz6+8k3TvpR80S98NOzy62eOXXatOaJrB2a3ZvD3UTS99q9I+L6Wgki4hpJ6wLnAu8ldadvGkxPnTat4e/DYPodaEcwfQXwA2AbSetHxB3NTqhqoQ7SGGqz/nY/afr+hYCNJY2OiNcmo5B0Amlh+oqZpLGDRwLb532rATdIemdh+v/qdQz/08v6bU76fTi2QZrjSEt0bd4gjdXXTetMv440cWOPMfgRcZ+kWaSg/591zv03aUb6Zesct+71Ufrvgc6uwElVrwM4KCJ+1eAc34fDUzfdh0cBx5Ae7DwAfD4inqmVUNLipGCaiDi+bTW2btJN96alHkj7RkTDyDYipgLbS/o8PcelDwt9DqYj4glJq5DGdNYci1rnvMoY6h/SnlnFzeqKiBck/Rl4B+kP9XbAZQCSjqXn0m4vATtGxB8k3UZa+urd+diawHWSto6IaZJG0nPprN/0soqvA2aTxqPUcxvpy2vNlnUbVNrRtdHdIweZRvOK9IWkXenZElD5ktisGcz34TDUTfdhRJwo6TLgLNJkdttJ+nxENO3paENPN92bBhHR0nKvEfFdSeW73Q4RbQliI+KxiHg4Il5o8bx5EfGniPhDO+ph1sQVVT9/VckupBbfileA91fuydy9cWfgT1Vp3gxUJqvYn9QFHOChiPh3L+s2F3gxIuoOecjHZtD/XZWt/00BRktarXpnfj06v1y/zrnrkGadf7rfameDRv6SeCHzH4638iXR96G1RV/uw4j4L7AFaSKjxUizA18taWI/VdeGkT7+jbQWlemhPNS4RdiGkzOBqfnntwKHkFqdL837XgV2zWsSvibSmtQ7kNZYJefxBUnLAtVdzU7oQ93+CywlaVSTdEvQQg8Qmy+6a53pSg+EE/KSaJWl0SpPzqcC35S0cPVJOc13SF8G/t7nD8UGtTZ8SfR9aH3WjmAlku8B65Lm4tkWuEvS4ZV706xVDqT7j6QRknaWtHOn69JpHV1KII9/+T7p7+iBnayLDX0RMSOPjT417/oecBdp1tqLgfMj4qoG525HmuX2CNJSMdcDK+Qk/wUmlazKBEkPFvflf9fMdVqApNeTWoKm1jres8IwZ+6cUpVZdJFFSqXLtSiV6tU5r5bOcfarg2tq5Db5KWlFgw8Db5V0J7AeaSb5V0iTzF0F/FPSD0ljCVclTVS3IekLwXkdqLd1CUmb0vcvib4PrU/adB++JiIeIs3BczDpgc13gN3z6/vaUGUbJtp9b9oCRpNWgphHh+PJTuv0m1+UtK5uAA6mbSCcTmplfg8wijTG+SMR8cFmJ0bE88C2ksaRloLZMh96GdizslxACSNJMyTW8kHqBNO5zpBm0LVBLCKulfQT0tJEbyQFL5UnFUdGxG8lXUIKck6rkcXVnjRl2Psb6SHgnvTyS6LvQ2uDPt+HtUTEmZJ+Q1rj9n2kXhQ/6Wu+Nqz0y71pCxj282Z0Opg2G1ARMVfSR0hLIaxO6jb967ws1jcjYkaj8yVtT/pSWb0W9cERUbar4/5NjheX3apWeeDkZbGGgIg4JE9wtzdpAroHgTOqekfsB0wDDmb+3+pZpNbEIwe2ttZt8t+yfUmtAjf1oSXQ96H1Wrvuwzp5PwHsKGkvUo+yT7Qrbxv6+vPeHKpanDxsZJ3zIiLeXSP9kOVg2oadiHhO0tbAr0mT64wEvgQcIOki4EpSd7KngTGkdaS3JrXOvL0qq9nAIRFRuptjRJzTh6pX1mWf0jCVDRoRcTZwdp1js4BP5TV/1yB9Ibg3yq1jbcNA7g2zTxvy8X1ovdau+7BB/udJugb4NvV7dZktoL/vzSFoK1IrfiutzSqcV3ci3WorrrRqw/Xcr77x9haq0Fl9DqYlHdOH0xfra/lmvRERj0nagvQF8sN597LAYXlr5nFgj4j4Yz9VcQER8fBAlWXdIyJmAv/odD1sePN9aJ0UEU/TvGeXmbXHPTSfn2ckaRb+AAbsu3A3akfL9HGUfAph1k0i4kVgt9xKfRJpQp1mZuS0/5dn+TbrWguNXIilllqqVNoXZjQc4dDDvgccUDrt+uutVyrdhhuU+fVLVlll+dJp77m//LKle+6+W/NEwCVXXNs8kfW7X//mN21PO3bJsb2tjpnZYPd74L2kIT/fbdQ1XtIY4AWAiNh6YKrXndrZzXsKafbPVowAVm5jHTpuxIjyKzgsXfJLLsANN91YOu1KK61UOu2yE99cOm1ZrfQN6YanMBFxI7CRpDcDuwCbk/6QLAfMBJ4kdfu+CrjG3RvNzMyGrqOPOKR02kZdVXubdjB1cbWhIyK2l7Qf8H/AmZL2BD4WEQ/USj6gleti7QimHyEFxJ+NiF+0cqKk8aRxqWYdFxF3UX8mbTMzMzOzISsiJkn6HXAG8H7g35K+BpzUwqo1w0r5ZtT6KrMYb9CLc/1Uw8zMzMzMrAtExJS8ZOxHScMbTwD+Lultna1Zd2pHMP0PUs/e3gTTZmZmZmZm1kVyj+O1gYuAdYE/SzpV0uKdrVl36XTL9FxSN3HPUmxmZmZmZtYlIuLZiNiTNKfQ06QVb+4GduxoxbpIO8ZM/4G0Bi+SFBGlu25HxHS8ZqBZvxihcs/K5swpPwRmxMhyeca8eaXzRK1MWWdmZmZmAykifi3pD8B3gQOACzpcpa7R55bpiJgVEX/Im8dAm5mZmZmZDSER8UJEHARsS+pZbLR3aSwzMzMzMzMboiLiOuD1na5Ht2jHmGkzMzMzMzOzYcXBtJmZmZmZmb1G0u4tpl9M0s/6qz7dyt28zcyGqFfnvMqUp58ulfbf//536XyfmVG+Dv996PpS6Z6aMqV0ntf//vel045ZfEzptAstNLJUultv36l0nja4PP/C86XTHvrJT5ZO2x/34e+u/XPpPM3MeuF8Se8BDo2IlxollLQ+aQmtNwIHDkDduoaD6RI22mDD0mnfv/vB/ViTcqY9ck9Hy/csdGZmZtZNnpoybdoJJ/+47UuxnnDyj9udJcCq/ZGpWS/sC2wuafeI+GetBJI+B3wTWAR4ZiAr1w0cTJuZmZnZkBYREzpdB7NBZkfgbGAN4C+SvhwRp1QOSpoAnAO8FxBwE7B3B+rZUQ6mzczMzMzM7DUR8VtJ6wE/B94NnJy7fe8HrE8KpJcF5gLHAd8qu0zy4489zNFHHNLw+GDhYNrMzMzMzMx6iIingPdI+hJwPKkV+l5gSdJE1pOBPSLi1o5VssMcTJuZmZmZmVlNEXGipEdJrdRjSd267wTeEREtTEs69DiYNhuCnnp62rRvnXLm4OkjU54nZTEzs66RW+y+VXkdEepgdcz6haS9gdNJ8wxX7vF1ge9KOiwiZnWsch3mYNpsCPJEK2ZmZv1L0prAsZ2uh1l/kbQ4cAawJymI/gdphu+PA4cCBwBb5Nm+7+hUPTtpRKcrYGZmZmY2mEgaAZwFLAr8pcPVMWs7SRsBd5ACaYDvAptFxN0RcRiwM2kprDWBW/MSWcOOg2kzMzMzs9Z8GtgcOB+4psN1MesPtwCrAVOBHSLiCxHxauVgRFxF6up9A7Awabbv33Wkph3kYNrMzMzMrCRJrwdOILXKDcvWOBsWRgHXAutFxNW1ElRm+waOAuYA2w5c9brDoBgzPUIjWHjhhUul3XzTzUrnu81Ou/e2SgPuHzfXvIdr+tUVl5dO+6a11y6Vbu011yqdpw1dkvYDzi6R9D0RcV0/V8eaeGHGS1xzw99KpZ3XT3XYeP3VSqW7/ve/L53nyCVW6G11mphZKpUWGhT/dVovHH/MMaXT+j4c1s4EFgc+GRFTJc85ZkPSlyLiO80S5bWlT5R0A3BB/1eru/gvsZn1xjxSt596XhmoipiZmQ0USQcD7waui4hzO10fs/5SJpAupP+rpPX7qTpdy8G0mfXGoxExsdOVMDMzGyiSVgROAmaRZjM2G5YkjQK+TWqYPryyPyLKda0ZQhxMm5mZmZk192NgLHBkRDzY6cqYddDCwGdJ604f3jjp0OYJyMzMzMzMGpC0F7ADaamg73a2NmbWLdwybWZmZmZWh6TlgFOBucDBETGnszUyaz9Jc1s8JWqcFxExrOLLYfVmzaxtJkj6O7AmMBJ4Evgz8NOIuKmTFTMzM2uzE4FlgDOAeyWNKRx/bcmZqmOzI2L2ANXPrB08LX0vOJg2s95YDNgAeI60PMjr87anpLOBj/nJvZmZDRGvz/9+Im+NzMj/fo80ptRsMAngL6T1pRtZGPhyTv+1/q5UN3MwbWateAI4HvgV8N+IeEXSSGCTvH8bYH/gReDTHaulmZmZmbXiKOAYYDPgAeDzEfFMrYSSFicF00TE8QNWwy7kCcjMrLSIuCYijouIf0XEK3nf3Ij4M/Be4Iqc9JOS3tixipqZmbVJRGwVEaq3kR4mV9JW9n+2czU2a11EnAi8FbgV2Bu4R9Kena1V93MwbWZtERHzgC/klyOAnTpYHTMzMzNrQUT8F9gC+BxpSN+5kq6WNLGjFetig6Kb96KLLsraa61VKu02O+3cz7Vp7I/XXFo67TXXNhuOMN+SSyxZOu1KK65YOu2MGeXWVr/08stK52nDV0T8T9I0YDzwhk7Xx5g2Dx7uZAX+escDpdIttGT5v1tdYNVOV2AweWrKtGknnPzjjt6HZZ1w8o87XYVW+D40s7aLiAC+J+lK4ExgW+AuSccCp+TGE8sGRTBtZmati4gJna6Dme9DM7PBJyIeAraRdDDwnbztnl/f19HKdRF38zaztpG0GqlVGuChTtbFzMxsIOS5RCrjp82GlIg4E1gH+B1pJZfbgG93tFJdxMG0mZUiqeGXhHz8pPxyHnBVv1fKzMzMzPpVRDwRETsC+wAv0HyJuGHDwbSZlbWqpL9K+rikN1SCa0kjJG1KemL5gZz2x3kSCzMzMzMbAiLiPOBNwLnAH4A/drZGnecx02bWirflDeAVSTOAJYBFqtKcDRw20BUzMzMzs/4VEU8D+3e6Ht3CwbSZlTUF+DSwGbA+MAFYCniZND76z8BZEXFLpypog4uk9wAHA5sAywEBPAn8BfhJRPyhg9WzYcL3oZnZwFpxpVX5xkk/qnv86htvH8Da9I2DaTMrJSJmAafnzazX8hCBM4CPV+2elf99fd72kHRKRHx+oOtnw4PvQzMz6yuPmTYzs4G2H/MDmF8Ca0TEYhGxGLAWcEU+9jlJH6hxvlk77IfvQzMz6wMH02ZmNtD2yf/+D9g9Iu6vHMgT130YeDDv2m2A62bDh+9DMzPrEwfTZmY20F6X/70zIuYUD0bEq8Ad+eWYgaqUDTu+D83MrE8GxZjpl2a9xJ3/+leptHcecXg/16Z9Vl5ppdJpR40aVTrtgw891JvqmJkNlAeBNYH1JC1UDGQkjSJNcgcweGYhscHG96GZmfWJW6bNzGygnZH/XR24UNLqlQOS1gR+AbwBeAA4ZeCrZ8OE70MzM+sTB9NmZjagIuLXwOeA2cCuwP2SXpL0EnAvsBUp0Nk4Il7oWEVtSPN9aGZmfeVg2szMBlxEnAp8EHg67xqdN4CFSWNUxw58zWw48X1oZmZ94WDazMwGlKTFJF0MXAU8AmwLTMjbtsA9wN7AXyWt27GK2pDm+9DMzPpqUExAZmZmQ8pJpKWG/gtsGREvVx27VtLNpFmU1wB+AGw54DW04cD3oZmZ9Ylbps3MbMBIWgL4WH75g0IAA0BEzAJOzy+3kLTsQNXPhgffh2Zm1g4Ops3MbCCtwfxeUQ80SHd/1c+v77/q2DDl+9DMzPrMwbSZmQ2keVU/r9og3XJVP8/op7rY8OX70MzM+szBtJmZDaR7gVn554MkLTB3h6SRzO+C+xxpTKtZO/k+NDOzPnMwbWZmAyaPQ/1pfrkB8GtJb5E0Im/rAr8FNs9pTo2IuZ2oqw1dvg/NzKwdFBGdrkNTkqYCD3e6HtZxq0bEhE5Xwsz6RtJo4FfAdlW7X8n/LlK170Jgbwcx1h98H5qZdUaJ2G7QfOcfFMG0mZkNLZIEfAjYC9gQWBYI4Gngr8DZEfGbztXQhgPfh2Zm1hcOps3MzMzMzMxa5DHTZmZmZmZmZi1yMG1mZmZmZmbWIgfTHSBpMUnbSzpa0q8kPSwp8nZck3NXlPRJSZdI+p+kWXl7SNKFkt41QG/DzMzMzMystL7EQd1ogXUVbUBsTFpyoyWSVibNfKeq3S/l1xPz9lFJZwEf88yjZmZmZmbWRXoVB3Urt0x3znPA9cBJwO7AUyXOGUkKnK8H9gVWjIjFgTHAOsAVOd0BwHFtrq+ZmZmZmVlf9SYO6kqezbsDJI0sthpLmgysChwfEcfVOW8ssFpE/KPOcZGe9GwHzAQmRMTLbay6mZmZmZlZr/Q2DupWbpnugN52v46I5+sF0vl4AGfll2OAtXtTjpmZmZmZWbsNtWGoDqaHnuqW6JEdq4WZmZmZmdkQ5mB66Nkq/zsbuK+D9TAzMzMzMxuyHEwPIZJeDxySX14cES90sj5mZmZmZmZDlYPpIULSaOASYDFgGvClztbIzMzMzMxs6HIwPQRIWgi4ANgQeBXYMyKe6GytzMzMzMzMhi4H04OcpJHA+cD7gTnAHhFxTUcrZWZmZmZmNsQ5mB7EciB9HrAbMBfYKyJ+2dlamZmZmZmZDX0LdboC1jtVLdIfYX4gfXFna2VmZmZmZjY8OJgehHIgfQE9W6Qv6mytzMzMzMzMhg938x5kqlqkdyONkd7TgbSZmZmZmdnAcst0h0haChhZtavyYGMxSeOr9r8cETPzOZUx0h9h/mRjlwxEfc3MzMzMzPqqN3FQt1JEdLoOw5KkycCqJZKeExH75XPeAfwh738VeLbJuZ/xOGozMzMzM+sWvYmDupVbpgeX6m75o4DlmqQf3Y91MTMzMzMzG7bcMm1mZmZmZmbWIk9AZmZmZmZmZtYiB9NmZmZmZmZmLXIwbWZmZmZmZtYiB9NmZmZmZmZmLXIwbWZmZmZmZtYiB9NmZmZmZmZmLXIwbWZmZmZmZtYiB9NmZmZmZmZmLXIwbWZmZmZmZtYiB9NmZmZmZmZmLXIwbWZmZmZmZtai/wcUee8hdT7bawAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 1080x1080 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "trial = 0\n",
    "first_featuremap = 0\n",
    "second_featuremap = 0\n",
    "\n",
    "Plot_model(dataset, model.to('cpu'), trial, first_featuremap, second_featuremap)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualisation des Kernels des deux convolutions du réseau"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Les kernels utilisé par le modèle sont stockés dans :\n",
    "- `model.conv1.weight` pour le première convolution\n",
    "- `model.conv2.weight` pour la deuxième convolution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "from Model_show import transform_img"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Tous les kernel de la convolution 1 :**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([20, 1, 5, 5])"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.conv1.weight.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1080 with 20 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(4,5, figsize=(15,15))\n",
    "for num_w in range(model.conv1.weight.shape[0]) :\n",
    "    w = model.conv1.weight[num_w].to('cpu')\n",
    "    \n",
    "    a, b = num_w//5, num_w%5\n",
    "    ax[a][b].imshow(transform_img(w), cmap='magma')\n",
    "    ax[a][b].set_title(num_w+1)\n",
    "    ax[a][b].set_xticks([]) ; ax[a][b].set_yticks([])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Tous les kernel de la convolution 2 pour les cartes de caratéristique du première kernel de la convolution 1 :**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1080 with 50 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(10,5, figsize=(15,15))\n",
    "for num_w in range(model.conv2.weight.shape[0]) :\n",
    "    w = model.conv2.weight[num_w][0].view(1,5,5).to('cpu')\n",
    "    \n",
    "    a, b = num_w//5, num_w%5\n",
    "    ax[a][b].imshow(transform_img(w))\n",
    "    ax[a][b].set_title(num_w+1)\n",
    "    ax[a][b].set_xticks([]) ; ax[a][b].set_yticks([])\n",
    "plt.tight_layout(pad=0)\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}