pyg_torch2_sar.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "509df0a0",
   "metadata": {
    "scrolled": true
   },
   "source": [
    "!wget https://raw.githubusercontent.com/rdkit/rdkit/master/Docs/Book/data/solubility.train.sdf\n",
    "!wget https://raw.githubusercontent.com/rdkit/rdkit/master/Docs/Book/data/solubility.test.sdf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "7c763163",
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "from rdkit import Chem\n",
    "from torch_geometric.utils import smiles as pygsmi\n",
    "import torch\n",
    "import torch_geometric\n",
    "import torch.nn.functional as F\n",
    "from torch.nn import Linear\n",
    "from torch.nn import BatchNorm1d\n",
    "from torch.utils.data import Dataset\n",
    "from torch_geometric.nn import GCNConv\n",
    "from torch_geometric.nn import ChebConv\n",
    "from torch_geometric.nn import global_add_pool, global_mean_pool\n",
    "from torch_geometric.loader import DataLoader\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "ce901cce",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_mols = [m for m in Chem.SDMolSupplier('solubility.train.sdf')]\n",
    "test_mols = [m for m in Chem.SDMolSupplier('solubility.test.sdf')]\n",
    "sol_cls_dict = {'(A) low':0, '(B) medium':1, '(C) high':2}\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "849925b3",
   "metadata": {},
   "outputs": [],
   "source": [
    "n_features = 9\n",
    "# definenet\n",
    "class Net(torch.nn.Module):\n",
    "    def __init__(self):\n",
    "        super(Net, self).__init__()\n",
    "        self.conv1 = GCNConv(n_features, 128, cached=False) # if you defined cache=True, the shape of batch must be same!\n",
    "        self.bn1 = BatchNorm1d(128)\n",
    "        self.conv2 = GCNConv(128, 64, cached=False)\n",
    "        self.bn2 = BatchNorm1d(64)\n",
    "        self.fc1 = Linear(64, 64)\n",
    "        self.bn3 = BatchNorm1d(64)\n",
    "        self.fc2 = Linear(64, 64)\n",
    "        self.fc3 = Linear(64, 3)\n",
    "         \n",
    "    def forward(self, data):\n",
    "        x, edge_index = data.x, data.edge_index\n",
    "        x = F.relu(self.conv1(x, edge_index))\n",
    "        x = self.bn1(x)\n",
    "        x = F.relu(self.conv2(x, edge_index))\n",
    "        x = self.bn2(x)\n",
    "        x = global_add_pool(x, data.batch)\n",
    "        x = F.relu(self.fc1(x))\n",
    "        x = self.bn3(x)\n",
    "        x = F.relu(self.fc2(x))\n",
    "        x = F.dropout(x, p=0.2, training=self.training)\n",
    "        x = self.fc3(x)\n",
    "        x = F.log_softmax(x, dim=1)\n",
    "        return x    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "8aba9d3e",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_X = [pygsmi.from_smiles(Chem.MolToSmiles(m)) for m in train_mols]\n",
    "for i, data in enumerate(train_X):\n",
    "    y = sol_cls_dict[train_mols[i].GetProp('SOL_classification')]\n",
    "    data.y = torch.tensor([y], dtype=torch.long)\n",
    "    data.x = data.x.float()\n",
    "\n",
    "test_X = [pygsmi.from_smiles(Chem.MolToSmiles(m)) for m in test_mols]\n",
    "for i, data in enumerate(test_X):\n",
    "    y = sol_cls_dict[test_mols[i].GetProp('SOL_classification')]\n",
    "    data.y = torch.tensor([y], dtype=torch.long)\n",
    "    data.x = data.x.float()\n",
    "train_loader = DataLoader(train_X, batch_size=64, shuffle=True, drop_last=True)\n",
    "test_loader = DataLoader(test_X, batch_size=64, shuffle=True, drop_last=True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5f815e91",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "5ef858b0",
   "metadata": {},
   "outputs": [],
   "source": [
    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "model = Net().to(device)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "9ef2dcd9",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = torch_geometric.compile(model)\n",
    "optimizer = torch.optim.Adam(model.parameters(), lr=0.01)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "0b3acac8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 1, Train loss: 0.867, Train_acc: 0.522, Test_acc: 0.537\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[2023-03-28 21:48:02,019] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: 'forward' (/tmp/iwatobipen_pyg/tmp_e83a8op.py:217)\n",
      "   reasons:  self.training == True\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n",
      "[2023-03-28 21:48:02,730] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: 'forward' (/tmp/iwatobipen_pyg/tmpp4kgewmy.py:217)\n",
      "   reasons:  self.training == True\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n",
      "[2023-03-28 21:48:04,160] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: 'gcn_norm' (/home/iwatobipen/miniconda3/envs/torch2/lib/python3.10/site-packages/torch_geometric/nn/conv/gcn_conv.py:43)\n",
      "   reasons:  num_nodes == 863\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n",
      "[2023-03-28 21:48:04,164] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: 'add_remaining_self_loops' (/home/iwatobipen/miniconda3/envs/torch2/lib/python3.10/site-packages/torch_geometric/utils/loop.py:300)\n",
      "   reasons:  num_nodes == 863\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 2, Train loss: 0.734, Train_acc: 0.498, Test_acc: 0.482\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[2023-03-28 21:48:11,312] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: 'global_add_pool' (/home/iwatobipen/miniconda3/envs/torch2/lib/python3.10/site-packages/torch_geometric/nn/pool/glob.py:8)\n",
      "   reasons:  tensor 'x' size mismatch at index 0. expected 863, actual 879\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n",
      "[2023-03-28 21:48:23,685] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: 'forward' (/home/iwatobipen/miniconda3/envs/torch2/lib/python3.10/site-packages/torch_geometric/nn/dense/linear.py:127)\n",
      "   reasons:  tensor 'x' size mismatch at index 0. expected 774, actual 780\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n",
      "[2023-03-28 21:48:27,829] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: 'scatter' (/home/iwatobipen/miniconda3/envs/torch2/lib/python3.10/site-packages/torch_geometric/utils/scatter.py:23)\n",
      "   reasons:  dim == 0\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 3, Train loss: 0.608, Train_acc: 0.54, Test_acc: 0.537\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[2023-03-28 21:48:38,694] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: 'propagate' (/tmp/iwatobipen_pyg/tmp_e83a8op.py:178)\n",
      "   reasons:  tensor 'x' size mismatch at index 0. expected 774, actual 815\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n",
      "[2023-03-28 21:48:39,004] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: 'propagate' (/tmp/iwatobipen_pyg/tmpp4kgewmy.py:178)\n",
      "   reasons:  tensor 'x' size mismatch at index 0. expected 774, actual 815\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n",
      "[2023-03-28 21:48:46,226] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: 'broadcast' (/home/iwatobipen/miniconda3/envs/torch2/lib/python3.10/site-packages/torch_geometric/utils/scatter.py:18)\n",
      "   reasons:  tensor 'ref' size mismatch at index 0. expected 766, actual 922\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 4, Train loss: 0.569, Train_acc: 0.484, Test_acc: 0.463\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[2023-03-28 21:48:59,980] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: 'aggregate' (/home/iwatobipen/miniconda3/envs/torch2/lib/python3.10/site-packages/torch_geometric/nn/conv/message_passing.py:595)\n",
      "   reasons:  dim_size == 815\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n",
      "[2023-03-28 21:49:00,516] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: 'message' (/home/iwatobipen/miniconda3/envs/torch2/lib/python3.10/site-packages/torch_geometric/nn/conv/gcn_conv.py:240)\n",
      "   reasons:  tensor 'x_j' size mismatch at index 0. expected 2481, actual 2369\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 5, Train loss: 0.527, Train_acc: 0.51, Test_acc: 0.49\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[2023-03-28 21:49:14,022] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: '_collect' (/tmp/iwatobipen_pyg/tmp_e83a8op.py:115)\n",
      "   reasons:  self.training == False\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n",
      "[2023-03-28 21:49:14,124] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: '__call__' (/home/iwatobipen/miniconda3/envs/torch2/lib/python3.10/site-packages/torch_geometric/nn/aggr/base.py:86)\n",
      "   reasons:  self.training == False\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n",
      "[2023-03-28 21:49:14,244] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: '_collect' (/tmp/iwatobipen_pyg/tmpp4kgewmy.py:115)\n",
      "   reasons:  self.training == False\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 6, Train loss: 0.557, Train_acc: 0.585, Test_acc: 0.553\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[2023-03-28 21:49:26,379] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: 'forward' (/home/iwatobipen/miniconda3/envs/torch2/lib/python3.10/site-packages/torch_geometric/nn/aggr/basic.py:18)\n",
      "   reasons:  dim_size == 853\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 7, Train loss: 0.508, Train_acc: 0.726, Test_acc: 0.743\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[2023-03-28 21:49:37,808] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: 'reduce' (/home/iwatobipen/miniconda3/envs/torch2/lib/python3.10/site-packages/torch_geometric/nn/aggr/base.py:146)\n",
      "   reasons:  dim_size == 809\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n",
      "[2023-03-28 21:49:38,043] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: '_lift' (/tmp/iwatobipen_pyg/tmp_e83a8op.py:87)\n",
      "   reasons:  tensor 'src' size mismatch at index 0. expected 822, actual 836\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n",
      "[2023-03-28 21:49:38,045] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64)\n",
      "   function: '_lift' (/tmp/iwatobipen_pyg/tmpp4kgewmy.py:87)\n",
      "   reasons:  tensor 'src' size mismatch at index 0. expected 822, actual 836\n",
      "to diagnose recompilation issues, see https://pytorch.org/docs/master/dynamo/troubleshooting.html.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 8, Train loss: 0.563, Train_acc: 0.519, Test_acc: 0.494\n",
      "Epoch: 9, Train loss: 0.526, Train_acc: 0.647, Test_acc: 0.634\n",
      "Epoch: 10, Train loss: 0.485, Train_acc: 0.809, Test_acc: 0.805\n",
      "Epoch: 11, Train loss: 0.513, Train_acc: 0.542, Test_acc: 0.545\n",
      "Epoch: 12, Train loss: 0.487, Train_acc: 0.698, Test_acc: 0.689\n",
      "Epoch: 13, Train loss: 0.466, Train_acc: 0.628, Test_acc: 0.634\n",
      "Epoch: 14, Train loss: 0.468, Train_acc: 0.532, Test_acc: 0.549\n",
      "Epoch: 15, Train loss: 0.495, Train_acc: 0.617, Test_acc: 0.599\n",
      "Epoch: 16, Train loss: 0.437, Train_acc: 0.53, Test_acc: 0.525\n",
      "Epoch: 17, Train loss: 0.463, Train_acc: 0.669, Test_acc: 0.654\n",
      "Epoch: 18, Train loss: 0.505, Train_acc: 0.687, Test_acc: 0.634\n",
      "Epoch: 19, Train loss: 0.478, Train_acc: 0.715, Test_acc: 0.759\n",
      "Epoch: 20, Train loss: 0.449, Train_acc: 0.645, Test_acc: 0.65\n",
      "Epoch: 21, Train loss: 0.477, Train_acc: 0.803, Test_acc: 0.786\n",
      "Epoch: 22, Train loss: 0.438, Train_acc: 0.734, Test_acc: 0.712\n",
      "Epoch: 23, Train loss: 0.406, Train_acc: 0.773, Test_acc: 0.759\n",
      "Epoch: 24, Train loss: 0.415, Train_acc: 0.793, Test_acc: 0.759\n",
      "Epoch: 25, Train loss: 0.481, Train_acc: 0.778, Test_acc: 0.735\n",
      "Epoch: 26, Train loss: 0.476, Train_acc: 0.58, Test_acc: 0.584\n",
      "Epoch: 27, Train loss: 0.464, Train_acc: 0.573, Test_acc: 0.654\n",
      "Epoch: 28, Train loss: 0.474, Train_acc: 0.74, Test_acc: 0.724\n",
      "Epoch: 29, Train loss: 0.424, Train_acc: 0.77, Test_acc: 0.755\n",
      "Epoch: 30, Train loss: 0.422, Train_acc: 0.611, Test_acc: 0.595\n",
      "Epoch: 31, Train loss: 0.457, Train_acc: 0.751, Test_acc: 0.743\n",
      "Epoch: 32, Train loss: 0.408, Train_acc: 0.592, Test_acc: 0.564\n",
      "Epoch: 33, Train loss: 0.412, Train_acc: 0.746, Test_acc: 0.759\n",
      "Epoch: 34, Train loss: 0.446, Train_acc: 0.674, Test_acc: 0.65\n",
      "Epoch: 35, Train loss: 0.435, Train_acc: 0.674, Test_acc: 0.63\n",
      "Epoch: 36, Train loss: 0.444, Train_acc: 0.714, Test_acc: 0.712\n",
      "Epoch: 37, Train loss: 0.438, Train_acc: 0.651, Test_acc: 0.677\n",
      "Epoch: 38, Train loss: 0.446, Train_acc: 0.635, Test_acc: 0.658\n",
      "Epoch: 39, Train loss: 0.395, Train_acc: 0.789, Test_acc: 0.751\n",
      "Epoch: 40, Train loss: 0.423, Train_acc: 0.507, Test_acc: 0.553\n",
      "Epoch: 41, Train loss: 0.467, Train_acc: 0.642, Test_acc: 0.646\n",
      "Epoch: 42, Train loss: 0.417, Train_acc: 0.719, Test_acc: 0.732\n",
      "Epoch: 43, Train loss: 0.374, Train_acc: 0.836, Test_acc: 0.805\n",
      "Epoch: 44, Train loss: 0.382, Train_acc: 0.826, Test_acc: 0.79\n",
      "Epoch: 45, Train loss: 0.435, Train_acc: 0.745, Test_acc: 0.739\n",
      "Epoch: 46, Train loss: 0.401, Train_acc: 0.782, Test_acc: 0.759\n",
      "Epoch: 47, Train loss: 0.397, Train_acc: 0.566, Test_acc: 0.568\n",
      "Epoch: 48, Train loss: 0.391, Train_acc: 0.762, Test_acc: 0.728\n",
      "Epoch: 49, Train loss: 0.415, Train_acc: 0.739, Test_acc: 0.704\n",
      "Epoch: 50, Train loss: 0.378, Train_acc: 0.784, Test_acc: 0.763\n",
      "Epoch: 51, Train loss: 0.431, Train_acc: 0.818, Test_acc: 0.774\n",
      "Epoch: 52, Train loss: 0.412, Train_acc: 0.647, Test_acc: 0.65\n",
      "Epoch: 53, Train loss: 0.394, Train_acc: 0.639, Test_acc: 0.638\n",
      "Epoch: 54, Train loss: 0.385, Train_acc: 0.784, Test_acc: 0.759\n",
      "Epoch: 55, Train loss: 0.406, Train_acc: 0.751, Test_acc: 0.786\n",
      "Epoch: 56, Train loss: 0.36, Train_acc: 0.675, Test_acc: 0.681\n",
      "Epoch: 57, Train loss: 0.404, Train_acc: 0.821, Test_acc: 0.751\n",
      "Epoch: 58, Train loss: 0.378, Train_acc: 0.792, Test_acc: 0.778\n",
      "Epoch: 59, Train loss: 0.416, Train_acc: 0.508, Test_acc: 0.506\n",
      "Epoch: 60, Train loss: 0.385, Train_acc: 0.759, Test_acc: 0.743\n",
      "Epoch: 61, Train loss: 0.396, Train_acc: 0.778, Test_acc: 0.735\n",
      "Epoch: 62, Train loss: 0.388, Train_acc: 0.832, Test_acc: 0.786\n",
      "Epoch: 63, Train loss: 0.37, Train_acc: 0.85, Test_acc: 0.802\n",
      "Epoch: 64, Train loss: 0.395, Train_acc: 0.587, Test_acc: 0.591\n",
      "Epoch: 65, Train loss: 0.395, Train_acc: 0.817, Test_acc: 0.782\n",
      "Epoch: 66, Train loss: 0.378, Train_acc: 0.78, Test_acc: 0.747\n",
      "Epoch: 67, Train loss: 0.363, Train_acc: 0.72, Test_acc: 0.708\n",
      "Epoch: 68, Train loss: 0.331, Train_acc: 0.786, Test_acc: 0.763\n",
      "Epoch: 69, Train loss: 0.345, Train_acc: 0.728, Test_acc: 0.716\n",
      "Epoch: 70, Train loss: 0.336, Train_acc: 0.609, Test_acc: 0.591\n",
      "Epoch: 71, Train loss: 0.34, Train_acc: 0.684, Test_acc: 0.646\n",
      "Epoch: 72, Train loss: 0.384, Train_acc: 0.778, Test_acc: 0.782\n",
      "Epoch: 73, Train loss: 0.367, Train_acc: 0.706, Test_acc: 0.704\n",
      "Epoch: 74, Train loss: 0.414, Train_acc: 0.764, Test_acc: 0.728\n",
      "Epoch: 75, Train loss: 0.397, Train_acc: 0.798, Test_acc: 0.759\n",
      "Epoch: 76, Train loss: 0.378, Train_acc: 0.784, Test_acc: 0.72\n",
      "Epoch: 77, Train loss: 0.355, Train_acc: 0.767, Test_acc: 0.728\n",
      "Epoch: 78, Train loss: 0.405, Train_acc: 0.66, Test_acc: 0.661\n",
      "Epoch: 79, Train loss: 0.376, Train_acc: 0.555, Test_acc: 0.541\n",
      "Epoch: 80, Train loss: 0.383, Train_acc: 0.635, Test_acc: 0.619\n",
      "Epoch: 81, Train loss: 0.39, Train_acc: 0.686, Test_acc: 0.665\n",
      "Epoch: 82, Train loss: 0.398, Train_acc: 0.55, Test_acc: 0.545\n",
      "Epoch: 83, Train loss: 0.381, Train_acc: 0.675, Test_acc: 0.704\n",
      "Epoch: 84, Train loss: 0.38, Train_acc: 0.704, Test_acc: 0.704\n",
      "Epoch: 85, Train loss: 0.374, Train_acc: 0.788, Test_acc: 0.743\n",
      "Epoch: 86, Train loss: 0.391, Train_acc: 0.802, Test_acc: 0.774\n",
      "Epoch: 87, Train loss: 0.368, Train_acc: 0.732, Test_acc: 0.732\n",
      "Epoch: 88, Train loss: 0.382, Train_acc: 0.78, Test_acc: 0.743\n",
      "Epoch: 89, Train loss: 0.374, Train_acc: 0.57, Test_acc: 0.556\n",
      "Epoch: 90, Train loss: 0.35, Train_acc: 0.766, Test_acc: 0.747\n",
      "Epoch: 91, Train loss: 0.359, Train_acc: 0.78, Test_acc: 0.763\n",
      "Epoch: 92, Train loss: 0.327, Train_acc: 0.805, Test_acc: 0.755\n",
      "Epoch: 93, Train loss: 0.312, Train_acc: 0.853, Test_acc: 0.782\n",
      "Epoch: 94, Train loss: 0.389, Train_acc: 0.643, Test_acc: 0.638\n",
      "Epoch: 95, Train loss: 0.411, Train_acc: 0.829, Test_acc: 0.774\n",
      "Epoch: 96, Train loss: 0.383, Train_acc: 0.652, Test_acc: 0.642\n",
      "Epoch: 97, Train loss: 0.342, Train_acc: 0.741, Test_acc: 0.724\n",
      "Epoch: 98, Train loss: 0.373, Train_acc: 0.856, Test_acc: 0.798\n",
      "Epoch: 99, Train loss: 0.357, Train_acc: 0.736, Test_acc: 0.747\n",
      "Epoch: 100, Train loss: 0.385, Train_acc: 0.552, Test_acc: 0.514\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7fb6fb144f10>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAiMAAAGwCAYAAAB7MGXBAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAAEAAElEQVR4nOy9aZgkV3klfCJy32rvquq9W/vOIiFAQiwGCWOMBxgDtgcYMDD4w4YBxhtmvAz2N5rxwgfGgGFAlm2WwQaMsS3AYhGSEBJIQmhrLa1Wd1cvtS+5R2RmxPfj3vfGjciIzMjMyKW64zxPP1VdlZUVlRlx49zznve8immaJkKECBEiRIgQIYYEddgHECJEiBAhQoQ4uxGSkRAhQoQIESLEUBGSkRAhQoQIESLEUBGSkRAhQoQIESLEUBGSkRAhQoQIESLEUBGSkRAhQoQIESLEUBGSkRAhQoQIESLEUBEd9gH4gWEYOHXqFHK5HBRFGfbhhAgRIkSIECF8wDRNFAoF7Nq1C6rqrX9sCzJy6tQp7N27d9iHESJEiBAhQoToAgsLC9izZ4/n97cFGcnlcgDYHzM2NjbkowkRIkSIECFC+EE+n8fevXvFfdwL24KMUGlmbGwsJCMhQoQIESLENkM7i0VoYA0RIkSIECFCDBUhGQkRIkSIECFCDBUhGQkRIkSIECFCDBUhGQkRIkSIECFCDBUhGQkRIkSIECFCDBUhGQkRIkSIECFCDBUhGQkRIkSIECFCDBUhGQkRIkSIECFCDBUhGQkRIkSIECFCDBUhGQkRIkSIECFCDBUhGQkRIkSIECFCDBUhGQkRIkSIECFCDBVnNRnJV2tYWC9jq1wb9qGECBEiRIgQZy3OajLyW//4U1z3p9/D1396ctiHEiJEiBAhQpy1OKvJSC4ZAwAUtPqQjyREiBAhQoQ4e3FWk5FsIgoAKFZDMhIiRIgQIUIMC2c1GRlLMjJSCMlIiBAhQoQIMTSc1WQky8lIMSzThAgRIkSIEEPDWU1GhGekGnbThAgRIkSIEMPCWU1GyDMSlmlChAgRIsQZifWngSe/PeyjaIuzmozkQs9IiBAhQoQ4k/GVtwGf/4/A8qFhH0lLhGQEoWckRIgQIUIMESuPA9/6IFBcCf65N46xj5sLwT93gDjLyUjoGQkRIkSIEEPG7X8O/PCvgIf+IdjnNU2gusk+r24F+9wB46wmIyJnRKvDNM0hH02IECHOSJTXgc+/Hrj3pmEfSYhRBZVQqvlgn1cvAQZX/omUjCjOajJCZZpaw4RWN4Z8NCFChDgj8aNPA09+C7jn08M+khCjCMMA1p5kn9fKwT63rIaEysjoIhOPQlHY56GJNUSIEIGjrluKSL063GMJMZrYWrDOjaDPEVkNCcnI6EJVFWTj1FET+kZChAgRMB79Z6C4xD5v6MM9lhCjidUnrc9rlWCfu7JpfR6SkdFGmMIaIkSIvuGev7Y+r2vDO44Qo4vVx63PQ2Xk7EWYNRIiRIi+4MR9wMl7rf+HZCSEG1afsD4PlZGzF2EKa4gQIfqCH32KfTxwHfvYCMlICBfIZZpQGTl7kQ2zRkKECBE0CkvAw19ln1/zHvaxobPchxAhZITKCICQjIQprCHOTtQ14Pt/Bpx+cNhHcmbivpsBowbsuRrY9zzr62GpJoSM8jpQklJXgyYjYWvv9kEuLNOEOBvx+C3A9/4E+LtfADaPD/tozizUdeDez7LPn/tOIJqwvheWakLIkEs0QP/LNCOszIVkJFRGQpyNKC6zj5UN4B/+c7hjDxKHvs7aebPzwMW/AETi1vfC1zmEDCrRxLPsYz/LNEYt+OcPEGc9GckmQs9IiLMQ8iJ16n7gmx8Y2qGccbjvZvbxOW8DonFAUYAIV0dCMhJCBpGR+SvYx34qI8BIl2rOejIStvaGOCtBi9LuKwEorKzwYMBDus5WLD3CPl74CutrVKoJg89CyKAyzU5ORlopF1oB+OmX7BuJdnA+NiQjo4tsSEZCnI2gReminwde9Nvs83/5r9bArhDdobIJVNbZ55MHra8TGQkj4UPIoMCzeR9k5N6/Af7pvwA//Lj/5w+Vke2DsdAzEuJsBC1SyXHgRb8DnPszbEjXl94E6AEP6zqbsPE0+5iZBRJZ6+thmSaEE3UN2DjKPidlpKGxwXluKCzyj6f9/w5SRtLT7GNIRkYX5BkphspIiLMJtCglxwE1Arz2M+wGuvYkcPTO4R7bdsY6JyNTB+1fj3ITa1imCUFYPwKYBpAYAyYPWF/3Us9ooq/fyb61qtW9NbGffQzJyOjC8oyEBtYQZxFox5SaYB8z08DuZ7PPO9l5hbCDlJFJJxlJso9hmSYEgcyrMxcA0ZT1dU8ywks4fpVLUaJRgPHdjq+NHs56MiI8I2GZJsTZBKGMTFhfy86yj9T2eyZj8SHgK++wZPKgIJSRc+xfp/beeqiMhOBYkchIJAqoTKX39I3USvaP7UAbjuQ4kJpkn4dkZHQh54wYxugGwoQIESiEZ2TC+lp2jn2kkfdnMn78WeChfwDu+qtgn9ezTEPdNKFnJASHUEbOZx9jXB0JWhlJTTBCAoRlmlFGjntGTBMo1xpDPpoQIQYAowFoefY5LVIA84wAQGlIysjGMeDkfYP5XbQoH70j2Of1LNOEBtYQDshlGsAq5Xl5QoiM+PWMCGVkIiQj2wHJmIqoqgAIfSMhzhLIC5JMRoZZpjFNFk3/mZdZXQP9hF5kH1ceC+7vrVWB/Cn2uVMZaddNY5rAv7wXuPuTwRxLiNGGaVoZIzsuZB9jREY8lBGdyjTdKCMT/GshGRlZKIoifCNhR02IswK0IMXSVpcHMNwyzcZR9s80rBt6P6EVrM+fvj2Y59w8BsAE4jmrlZLQrkyzdhi472+Af//9UD05G5A/ybwfatTqpCETa93LM9JhmSZURrYfyDeSD8lIiLMBcluvjGEqIws/sj73u/PrBVrR+jwoMrJ+hH2cOsgi4GW0K9PQrteoAYsPB3M8IUYXVKKZOgeIcONqO2Wk0zJN6BnZfhBZI2FHTYizAW7mVcAiI3rRujkOCgv3WJ8PInRN74My4mVeBdqXaWTT4qn7gzmeEKMLKtGQXwRgSiXQQhnh16Re8jd9V950hGRke2Dks0a8EvlChOgGXspIPGstiINWR2zKyACIkKyMbDwNbC70/pxe5lWgfeiZTEZOhmTkjIfTvApIBtY2yghMf3k1YZlm+yGXGGHPyI8/C/zpAeDEgLoMQpz5cAaeERRlOKWaah5YfsT6/yBUGTKwZufZxyC6alopI+1Cz2TFZFAdRSGGhxU+k8amjHDPiFsZxjTtX/ejHnqVafyoKkNASEYw4pN7n/ouzOoWsHD3sI8kxJkCL2UEGI6J9eR9zLhK6HeZplG3SMGFP8s+BlGqEZ6Rc5q/J0LPPMo0ctDV6hN2g22IMw9uZZpWhNUZhObHN+KmjBj1wXiyukBIRjDaKaxf1k7hRft245HSyWEfSogzBdVNLESjeHHxPnz6wU/bv5fZwT4OkozIJRqg/2Ua2S9y4SvZx6dv723HaDSAzePsc9cyDXXTeJVpZJJiAqce6P5YQow2qltAkbevz5xnfV0oIy6ekW7IiKyMxNKsc4d+/wiiKzLyiU98AgcPHkQymcSVV16JO+5oLXF+/vOfxzOe8Qyk02ns3LkTb33rW7G2ttbVAfcDZGAdRc/IHUYBG5EI7itvczJSXgfuvcli6yGGh+oW7kolsWbq+Ouf/jWWShLxIGWktDK44yHzKu0M+12mIdUhkgAOvIDFcOdPWspGN9g6wTphInFgbFfz90U3jVeZxnGzCUs1Zy7IEzS2265OtkpgdZIPP9eIrIwoysj7RjomI1/60pfw3ve+Fx/84Afxk5/8BNdddx1e8YpX4Pjx466Pv/POO/HmN78Zb3vb2/DII4/gH//xH/HjH/8Yb3/723s++KCQG+GckYrJjknf7tM+7/4k8K/vA3706faPDdFfVDaxHIkAAGpGDTc/crP1vUGXaQwDOPFj9vn+a9jHfpdpyLyayALxNLD3avb/Xko1ZF6d2M+mIEvQGzo+sfkw7kkmvGfTOMs3YUfNYHH8buBTLwqus6oVnvx39vHcl9i/LgysfVBGgDOPjHz4wx/G2972Nrz97W/HxRdfjI985CPYu3cvPvlJ9+TAu+++GwcOHMB73vMeHDx4EC94wQvwzne+E/fee6/n79A0Dfl83vavnxhlz0jFZBH1te1ORkiWXHtquMcRAqhuYSVq3TC//MSXsV5dZ/8ZtIF15TEWTR/LAHs4Keh7mYaTkXiWfTz4QvaxlxtRC/Pqlx7/Ej65cT/+YmrSO/SMdsNje9jHkz/p/lhCdIZGHfj6u4HTDwAPfLG/v8s0gce/wT6/4Gft32tZpnFcE+0Ie123CAu18J9JZETXddx333244YYbbF+/4YYbcNddd7n+zDXXXIMTJ07glltugWmaWFpawpe//GW88pWv9Pw9N954I8bHx8W/vXv3dnKYHUMeljdqKHMyohmjV0LqCCQr5rd5uelMQHVLKCOqoqLaqOLvH/179j1BRgakjFCJZs+VQCLHPu+7MsLLNPT7iIwcvaN734iHeVVv6Lj54ZsBAAVVaWFg5WRk33MBKMDWcaA4wFJZtygsAj/8xPY23N7/t1arbaHP6b9rh5mKFokD57zY/j1hYPWjjLQh7BLh2FRMPL7++JlFRlZXV9FoNDA3N2f7+tzcHBYX3edJXHPNNfj85z+PN7zhDYjH45ifn8fExAQ+9rGPef6eD3zgA9ja2hL/FhYCyABogVH2jNApqBvbXBkRZGQAUd8hWqO6iRVORl53wesAAF987IvY0rakMs2AboRkXt37XFYyAfrvGXEqI7uvYlHcpRWm1HQDj4yRrx3+GpYrTGXSlBZkhJSRzKw1xXU7lGq++8fAtz4AfON3hn0k3UErALfdaP2/3+vTE99kHw+8wCLDBKGM+PCMuKknMqhEkxjDe7//3/C6f3kdno5z39KZQEYIiiPq2DTNpq8RHn30UbznPe/BH/zBH+C+++7DN7/5TTz99NP4tV/7Nc/nTyQSGBsbs/3rJ3Kj2k1jNFDhL6tujNixdQqZjIxon/tZA6lM87oLXofzJ89HqVbCFx/7ol0ZGcT7RMrI3uda5KDfZRrZMwKwQLJ9z2Ofd1uqWT/KPkplmppRw00P3yT+X1XU9qFnsSSw69ns8+0QfnaMK+IPfAFYfGi4x9INfvCXjISSapA/3d/f98S32EdniQZo3drrVAvbEXbJvPrExhMwYeLhCL+eiaiMGDoiIzMzM4hEIk0qyPLycpNaQrjxxhtx7bXX4rd+67dwxRVX4OUvfzk+8YlP4KabbsLp031+430imxhRz4heQoVPFNa3e5mGZNx6pbOL4cj3WfBbSGACg17ZxAZXRubSc/gvl/8XAMDnDn0OpXiGPaih9X8HVVoF1rmHaM9VVvprv8s0TmUE6M03YpquysgtR27ByeJJJHgUPFNGvLpp+NejSWD3lezzUVdGSmtSB5IJ3PoHQz2cjpE/Dfzwr9jnP/u/2Ee9wEL4+oHKpkXezr+h+ft0/gdhYOVrbCU1hgJvZT+m8iyfM0EZicfjuPLKK3Hrrbfavn7rrbfimmuucf2ZcrkMVbX/mghfCM0RucGMJflsmhEjI6ZWREUhMtIY8tH0CJnJ+5VCaxXgS28E/u39wKGv9+e4zjbUqlhV2HkeU2MYT4zj+v3X48DYAWxpW/iHI18HEnyX2G8TK5VodlwEpCatMk2/Q5mcnhEAOPgi9vHoHZ2PXyitcIKjAJP7AQANo4HPPPQZAMAvX/TLAABdVWC0S2CNJoDdpIzcN9ok/CRvQsjOsfbop74LHP7OcI+pE9z2P9m5tve5wDN+GUhwBb7Qp03yU98BzAYwc6F7Sm+shTLS1Nrb5hrhyshq0iLcx0yuyp0JZAQA3v/+9+Mzn/kMbrrpJhw6dAjve9/7cPz4cVF2+cAHPoA3v/nN4vGvetWr8NWvfhWf/OQnceTIEfzgBz/Ae97zHlx99dXYtculH38IoNCzSq2BemN05sBo1Q0YREbM0SJKHaMbMvLYv7FOCwC4/c9He2EeFEyThWt1+1pUrbbe2dQOKIqCiBrB2y5/GwDgbx/5W1SzPPis1G8yQiUa3kVDSoVedH98UHAjI3OXso/VLXsomh9QJ834HpEncuvxW3E0fxRj8TG8+RJrPdS8yjS0842mgLnLWEBVec0KUhtFEJk872XA1Uxdw61/wALgRh3Lh4CffI59fv0fsxyO3E72/375Rp7gLb0XvNz9+9EWcfBNnpF2BtZNAMByIiW+dKzOf+ZMISNveMMb8JGPfAQf+tCH8MxnPhO33347brnlFuzfz3YEp0+ftmWOvOUtb8GHP/xh/NVf/RUuu+wyvO51r8OFF16Ir371q8H9FT2CyjTAaHXUVCrr4nPd3AYXeCvYyIjPjpqf/l/r88UHgSdv9X7s2YJ7/hr4yOXAT7tsQaxuCfPqjvSs+PIrz3kl5jPzWKuu4Z4sKSN97qiRzavAcMs0FEoGeGeBeEGUaA4AYIrv/3nw/wAA3njxGzGVnBIPrTZ8KCOxpEWORrlUQ/kwe64CXvibzHex9LD9uh1V3PoHbATBxb/AO5hghdX1g4wYDStfxM0vAljKiB8Da7trhJORFRrQCOBobRMmMLLBk10ZWN/1rnfh6NGj0DQN9913H174wheK791888247bbbbI9/97vfjUceeQTlchmnTp3C5z73OezevbunAw8S8aiKRJS9FKPkG6lUN8Tnujk6ik3HME2YehHHo1F2Mfi52AtLTNYEgIt+nn28I1RHsMQHytGgrU5R3cJylMjIDvHlmBrDM3c8EwBwNMFvzP0s09R160ZLZCSgMk3NqGGx5N7dB6DZwAqwnTHNj/HKAvGCI2Pk+ye+jyc2nkA6msavXPwriKgRxBS24dEaHt4vauekjgryjbRLYtXLwIP/CFQ2Wj8uaBgNy2C75zlAegq47jfZ/7/7J/0nlL2gsmERg5f9kfV1IiP9aO89cS9QWWeEjc53J0gZ8dXa26abhhMOUkEBoGLU2EbkTFFGzlTkktTeO0pkZFN8rm1nZaSu4Qu5FF65dxf+MZf1R0Ye+ke2c9lzNfDKv2DR3Qv3AEfv7P/xjjJIYeo2BK9itfXOSsoIAOwfY+rm0Qhv4eqnMrL4EKuNpyaBaT6fI8bNs3qpM9K59pQt5+Iv7v0LXP/l6/GTZY/gMDdlBGDnGND5a0smTm5e/dyjTP7/pYt+CePcf5PkRKfq1aIvKyOA1FHTJvzsvr8Bvvp2VsYcJFYeY+WsWAaYvYR97er/AozvYzfzuz8x2OPpBAV+XicngOlzra8LZaQPnhFq6T3veiASdX9MK2WEyB35WnyWaVZU+yb2WCwakpFRxygGn5U166SpYRsrI3oRh2OM7D0aj/sjIyT1PuOXgNw88Ow3sf/fMeBFd9RAN9JuyYhcpkntsH2LyMgx8N27lzLitbvvBCd4iWbP1UyVAADq5IHZfudHWHsK+NizmdEZzDj6r0f+FQDw8OrD7j/j5hkBgAg7R7su00ydA9M08ej6owCAnzv4c+IhoqPGUxmRumkAy8R6+oHWHgxSyJYPdXbMvYJKNLufbcXfx5LAS3lHzZ0fGdmbnvBCZe1kvK+ekVYtvQQqU7oqI5yMpHnJz6eBddmxiT0akpHRhxUJPzottBXdajHTsI3LE3oRJd5RtRKNtL/YFx8Clh5isvmlr2Ffu/a/MlPfkduY5Hm2gpQRr/CsdqhuijKNUxk5MHYAAHCszgmPGxm5/c+AG/f2noGxzG7Y2PkM62u0GAP+SzVrh9nHp+8A9BIOrR9i4W2A+NgEInROMiIm63ZfptnQNkQrJZE7wCIjVY8W/Xqtgt+Y24GPnODlgx0XMdVBL1rj5t1A/qvNY50dswNFvYi3fvOt+MKhL/j7AeEXeY7965f9R3bseoFlj4wi6LzOOMhIv8o0m8eB5UcARQXOe6nrQ24/cTt+5Qe/iydjMQ/PCCco6Rn+/3aeEXburxjsXJ5OTgMAjsVi7HsjWO4OyQgHmVhHSRmpSK5+fQRPHt/QSygSGYn4ICOkilzws9ZOYGIfcMUvsc8HLUmPEmhX37UysikZWO3KyL6xfQCA5XoRZUVxL9M8/E9s59arsZJ29DsutL6mqlbd3G8KK70eZgM49RP84OQPxLc8yYjmVabhnpFOlJFqHiivss8nD+JYnpGCnZmdSJLKASDJiY6Ghmvr8FGjiu+nU7j5xHdRrpWZ2kBErZVvZIvIyPHOW5Il3L98P+5duhdfeMwngVjgZIQ6oQiqanXW/OjTPR1T30ATqbP287+tgdU0gRP3+VftCKSK7H2etZ458NmHPouHNp/A99IpRoadahiRjww/Zp+hZyu8g+Y584w0HotG2bXS75TjLhCSEQ5SRvKj5BmRWhxr21oZKaHIw9uWIxFA27JuCE406swvArDefxkveB/bXTzxDWDRQ4I/09GrMlLZxDKvWc+m7DvD8cS46Pw4Fos2KyN6GVjh5QC33ZtfmKYVu77jIvv3Oo2El2eiLNyDu05ZM7K8yQiVaexk5M5EFN9PJTtTRqhEk54GkmM4unUUgF0VAYBEhJGsqqK4Pn+Vk8sGDBxa56/x/OXs42oLszIpIw3dGkbZBYp8rVmtrLZ/cGXTOqbdVzV//4o3sKya9SPA4W93fUx9g5cykuNkpLTiTkgf/wbwmZ8BvvVB/7/LNIGHvsw+v8Al6AzstX9w5UH2q/k62ZQ1IsjItP3/XqDW3ho714mMHI3zUuQIlmpCMsJB82lGKfisLJmUNPe0/e0BvSiUkfWIyhwJXsFCR25jO/L0NMsvkDFzHnDJq9nnP/x4nw52xNGjZ6RSWUchwt4LZ5kGkHwjsRhblOWd7eJDzFQMuNe1/aK4xBZDRbXMqwTyjfgt00iEvXj8h2JRB4AtvU2ZJm6VaTaqG3h3qob3ze2A1kkcPZVQuHmVlBEnGUlGJTLiQiQrhvW1h1Z4rDoPUPPMGqnmrRweANjovlRTzLP5X6VaiSkzrUBKzeSBZnUBYCSPPF73/HXXx9Q3CM+I49jT05Y65kbsKBenk5LY498AFu5mXqDLX+/6kB8t/gh1niNF5ewmst9Upmk3m2YLJUVBmbeSXzXHSOOJaBR1/v1RQ0hGOEbSMyItCtt6TJ5eQklhp5qpKFiLRLyzRig/47L/yGaGOHHpq9lHihE/29BjNw3tfFNKDBnqXpEgOmpiMSbnSlk3OCV1dvSijJAqMnnQ6iAAUGvUUIl1qoxYZOSelQfEog54KCOmKXlGLGXk7tN3o64ANUVBxUu1c8PDX2EfebsmkRHy3xCIjGiq4vreVaTZUw+uckI1wcpmnmTEeQ31EJBWXLN8KW3VEeEXudr7Mc95OwCFtee38rwMAzQEMmMnI8V6GWZunv3HrVRDf4ffc79RB779h+zz5/0/wLh7nIWs5pWo08ZJ9vn1sJXM2f7vCqMBaHkscW9YNpbFgfEDSEaSqCsKTkVH08QakhGOUeymqdQtMlJTMJr1Vz/QiqJMA7TwjVTzwGOsEwLP+CX356IbaKd12zMBhmHdSLss0yzrmwCA2XjOdbilUEaS/HWWfSMyGfGKNfcD4RexSjRaQ8Mbv/FGvCxTQUFR/JMRSRn5YYRduxdPXQwA2NQ2mx9fK1vqjuQZkb0met2nKrN1wmrZvPItAICj+aMAXMo03DNS9ZhPI7f8PrTKlZF2ZGTLSUZ6UEakTKOVdsm7XuZVGVMHrc6RH3266+PqC0rNZZp/P/rveP4Xn4+vjNHAPDcy8gT76Fe1u/9v2c+kp1mJ2QM2MqJyMuKijHwjk8YLjvwt/iGXbX0MZF6VvGGqogpP2Kh21IRkhMNSRkaHjJSlRUtT3HdU2wGmVrDkR4B1c7gpI4/9G1uoZy6wchacEGO2z0IyIi9AXZ4LK9wUvSMx6fp90VETJ7nag4z08voLv4hlXv3UTz+FR9ceRV4xcTwW87/gS2WKH6SYykIttXm5hCEeT+RFESUh0zTxw1M/FA/R/f5t9/8dIzYHrgN2XADDNHA8z4hDkzISYcemKWqzH6FRR1XihYulRayUVywyUlpxb+XMn7D/v4cyTUl6rVY3j3g/0DCsbrY9Ln4RGc99J/v4wBf6N3yuG5S48iO19lI7+Pe4paKpjFzXrTwZP0RcKwC33cg+f9HvWFOBHVjIL2ChsGAdGoWUOZWRWgUPJ9g1+Ugizo7Bq+WbB+AtJ5jKSN4wWwk2JCOjC/KMjBIZqUhGN0NRUO/3zI4+oarl0VCcyoiLZ4SyJy74WSt7wokgycjCj4Hlx3p/nkFBfv+7VUa4u35Hasb1+1bwGZhlmsx+WsHaGfbw+wE0KSOPrz+Ov3n4b8S382oHyggnFwupcZyIxRCFgpftZ16jQq2AuuG4nuW2Xn6OHd48jOWKpQbofm42jTojIwBw1VsBMBKhGzqiahQ7szttD7cpI04Da73Kvi7hwdUHWSgX+Vq2HMQDsJQRCsLqRRmpWefW6tbT3g9cO8zMkdGkZbD1wjkvZkPh9OLotPmapmRgZWUawzRw/zLrDnuSCuJOZWTjaVa2BPytPXd9jJHIqXOAK9/q+bAfnGKKXJQn9FqeEScZKYnvFelc8SLsFHiWYGSbuubERiNURkYb2VH0jDjSGvVOjHUjhJJDLl/2yho5/VP2cdczvZ9MjNnuMW566yTwNz8L/P2re3ueQUK+QXerjHCjpJt5FQD25vZCgYICDKyrqrVwn34QkDu6ejGwSspI3ajjD+/6Q5vXo6iqHRtYf7CHlWaeVVcwn5kX387rjh05ddJIJRpZJgeAWsPH3/bEN9nuOT0DXPQqAFaJZm9uL6Ikt3OI0DO3Mk29KqZzEx5aeYiRpValGlIX9z2PfexFGZHWlhWn4iKDSjS7nmWFxHlBUYDnjlibr5a3yCBXRp7afEr4i04bvEzoXJ9sRLwNWc2fZmQEYHHzbt43Djr3nruTeY5EN41L/DuREGoG8Aw+o7beODvniIzY/GAhGRldjKJnpOwISNK3qTJSdMjlK24G1kbdmrsy/wx4QkQm96iMHLkNMOrshrIdpowC9jbWbpQJw8Ay89JjR9bdTJeMJrEzw3b1x2Ixq0wjl2iA7g2spVU2jRYKMHMBPn/o83hk7RHkYjlcMs1ixQuq6n9yL1dGfpBk58U1+XVEtQKyMUY2mkysLm29sl8EAHQ/f9u9N7GPz3qjuNl4ddIAVpmGeUYcRLJeRZXfhIi0NPtGXIgGqSX7r2Ef8ye6TsctSgP8VsstxgD48YvIuOKXeJvvU9asqWGCzKvxnFBZ71+yZ+Ycjsdak5F2a89t/5OR6b3PZYP4PFAzavjRIlODbzjA2n5LxEllwmMYQL0qSEiRTK5tlJHlqL2F3yrTRMVjRgkhGeHIJUbPM1JxSMx6r2rAkFB0tFguuxlY155kF2A8y6RNL8iRyb0EwT19u/X5CAYAucKmjHRx09EL1lya3F7Ph9kWLVJGiIzMXMA+dquMkCoysQ8L2hr+6id/BQD4b1f9NxwcZ+2xjIz49YwUUAPwozIjt9dUKsCJe8VMmCYy4phLU61Xcd8Sa1VN8OVQ95qsS1h/2rqxcuMqYJGRg2MHm36EAtCqqkuZpmaVaWhY4cOrD6NhNPwpI7uezebqmIZ7OccHSpIKuyqZWZvQKRlJZIFnvIF9/vg3ujq2QOHS1kvvP+HJeLw5hXXFpzKyfgT4CZtNhOv/2LvcDODBlQdRqpUwmZjEs2eZR05c4TLh4eu+RUYitq83gQysnOA6yzSL0ahtIvyoICQjHDQob5SUkYppPxZNSmTdTig6ykvL0QhLrZR391Simb+cpTh6gTwjQPcdHaYJHL3D+v+2ISOSWtBpZDlgG5K3gwKeXGDJudFmZWTf89nHbpURTkbMmQvxoR9+CNVGFVfPX43Xnv9aoWYUOirTFPDTZAJlQ8OUEsNFeg1YuMebjDgm9t63dB90Q8d8Zh77eDCZ1u68uu9m9vHcl4pJvYB3Jw1gkRHNLWdE8oxcMnMJ0tE0yvUyjmwd8SYjpik8I7XcXGsFxQeK0lqzUvNYZ7SCFePvl4wAwPT57GN5ratjCxSOwDPTNAUZuWz6MgDAE/EYUFi0l5WcZRqvktPyIUYKdz4D2GdN5625bB5IkXveruchx71BZcVkU8jkc5ATEyrh+C3TLPN5ZlSSnUhOYJwrdMcrK+4/O0SEZIRD9oyYIxK9XnEMx6ttU2WE6tFZlcnZKyQzyo710zxbYf6K1k8WlchIt6Wa9SP2MtF2JCPdlGmqW55zaWQcGD8AgMo0y2xxo1wXIiOtlJH1p1lKJXUtyODm1W/ncrj79N1IRBL4w+f/IRRFEQtyUVU7MrDexbtonjd2LlvQFu7BeJyTEWfwGRF6/rvIQHjNrmsQV9hrU2tF9Oq6tfO96ldt3zq25V2mEbNpXMmIhgrP4clEM7h05lIAvFRDJGNrwf4zlQ2gXsHtqSSu/uYv4yvjE+zr3WSNmCaK0lqzani8tytPsBttdh4Y2+n+GDekeOfWKOzGHVHwJwonsFxZRlSN4rUXvBYA8GQ8xjxZRJ5MszkrxYuw0pokdc/ceM+NeP4Xn49vHf2W7aHkF7l217W2zJ+yongoI+z8LJLY4uUhrG7CBLBiMrVLvtb3J/iMGr2F+jUkhGSEgzwjtYYJrT4CRivTRNkRAb9tyzT8xnUwyRaArYjKEmXlUs0iJyM725CRSBRQuXGu29fj6e/b/79dvDhyGFcXBtZScRFlvqua8eimARxlmtKypVpN7LPmd7RSRu75FPDDvwK++8fN3+PKyE84H33Nea8R+QdERgqddNPoFhm5dt/PsK+duA8TcdZh0k4ZoZZeRkbYQemtyMhj/8JUvdxO2wRWvaHjVImdz0TmZNgMrM73rl4RnpFkNInLZ1iXyoMrDwITvJzmJBm8HHPn2CTqRh23RbnvqRsTa62MklRO2DAN1NwG+onpxM1lqJYQZGQEboAOZeTeJdamfPnM5eJ1fzKeYCsvlWoKi4zEKtLtsh0Z4eXkhcICvvT4l6A1NPzu7b+L7x7/LgCW+PvoGlOZrtl1DRKRhL2jxoWMUDdNWQEaQEtlJK+q0DnBlKdzHyA/mJf6NUSEZIQjE7fc7yPhG6lrTQ57bUTJyAPLD+CDd37QM7mRyMjO1DTiQh2RfCOmCZx+EIuRCN6/9D2849/fYfv3J3f/ib1FU3TUdFkqkP0iQM/KyD8f/mf87x/9bxhmn0msfJxdKCPLBXYDy5kK0vKEXAeIjByPxtAor1kt17uehQfKJ/GHM1MotAoGI3PcY7c0m4O5MnJSYV8nnwiAzss0jRo2DB2P8kyU51/wWtbmWithjP9eT89IIofF0iIObx6Gqqh43s7nIc53nnoroncvb0F+9psZMeY4UTgBwzSQiWXEhFQZ9jJNczcNlWlS0RSumGGEnCkjXGUpLtlvUFzZW0gwpfAItaR2UaapF5ZQkUqjpgKsu6kY9NwTlvJzeOMw/vud/711aqsgI5sdH1vgEJ4RRkaopffKuStxzvg5iCgRFFQFS/L6RHN4ps5hk8MBb1VWkBH2vtz08E1omA2koinUzTp+8/u/iTtP3om7T98NEybOnzwfO9I7oCjWNVlSFbvyWGPb0qJ0OyipSksD6zIvx04kJhCniHsA+7N7AABHzR5CC/uEkIxwRFRltCb36iVUVDsZ8R3GNGD83aN/h68/9XX8+9F/d/0+meNysZwwU9k6ajaOAtoWvjY2hluXfoS7T99t+/elx7+EO05IHg+RNdIFOTNNNm4esBaWHsnIX97/l/jcoc/hiY0n2j+4F8gKjtnouAtopcTKYjuU1i2ZuzK7EFWj0FUFi9EI8OSt7FfufCZ+/9DN+Goui29HWhho6X0pLVvzPAC2M+YelJN8Z7Ynt0d8e4yrGb7LNFoBj8VjMBUFB8YOYEd2TngZxsuMhDSlsEoTe0kVuWz6MownxhHjr0tLMkLlxEv+g+3Lsl/ELdmWlJGK6lKmqVmtvcloEpfvYDv0w5uHUY4mrDZk2ZzKP1/gc4ZO1ApMbexCGSkVrOedbLBzajW/0PxAeu7JA+JLH/vJx/DPT/0zPvfo57x/QWqCfRwJZcQeBU9+kWfPPhvxSFwQ8Sfljhoq0cxcIBnovZQRfu7H0lgsLeJrh78GAPj4Sz+O6/dfj5pRw3u/91783SMso+baXdeKH6VSTUlR7RutWgUVRYEpnVfsMd7KyErUfTL3/nHWHHAMI3CPcyAkIxKyCcs3MnTUSmKByprso++Y6gGDdp+u8dsAinxxz8THRP3S1lHDSzRP5djE2Fed8yr8r+v+F/7Xdf8LL933UgDAt49L0z97CT5bPsRk9lgaj+y6DEej0Z7LNBQYtdGqCyEIOI+zw1LNMt8V7ogkWz4uokawL8dKJ8diMdFB8UB2AkfLjNAUnWFiMuT35dC/WJ/zjgRzbA9OcmK0W2oxzsYlZcQPGdGLyPMd/XSKqxF8RsxEgZGephRW+n8iK2r21+xmrbFxTk5rRovXlW5CFDTG0aqtF2CKB+BVprFae1PRFGbTs5hNz8IwDTyy/qi7iTV/Eg0AJ9nYSRgwcTQa60oZKXHvVsIEdjZYaXhl3WWezMZR9pEP8JPDwsSkYTeQMqIX3afhDhKSMrJUWsJCYQGqouKZs88EAJw/ycy2T8ZjlqeNzKsz57OwN6C9MhJN4m8f+VvUjTqumrsKz5l/Dv73C/83XrznxdAaGh5eY1PHr9l1jfhRIiNFpzKily3TKkfLjjNJGXFO5j4wxf6+oxGlt27EPiAkIxJE1sgIlGkMrSCk03Fw+biXeSB9RIGbAr1GtpdMtmBmE+OifrkSjVpkhO82j8SYnPjyAy/HK895JV55zivxxovfCAC4beE2q47dS/AZL9Es730O3hTbwNt3zvakjJimiSpvBfWcEhsUnAPcOizVrFSZIW822jwgzwnRURONilkuX81bo+yrZgdkhBY97hfJz5wnCNyurNXVY3lGfJZpNGsaNP0s9rLhbeMb7MbdbGBlv7cRy+CHpy2/CADEVVJGPDYjpml1MUXthM5rQB6hdeiZZWClPBJ7qcaFjGydxGI0grrkK3s6Hmsu5/hAkU+ozSgR7FDYNbi6dbT5gZt2ZeTI5hGxATm0dsjb+J8cB8B39cPOt5A8I0SkLpy8UJw/50+wm/XhWFxSRoiMXNg+54ift6uqgi8/8WUAwDuueAcAIKbG8Bcv/guhhiQjSTx7zhp7QWSkrDqVkbJtthdAwYBeBtYt21waGXunWTjgVkTFprN9ecgIyYgE0VEzAmWaqnTRTihUyx5tMtKUdgkApokCj1HOJMfdlZHTP0UDVh3znAkrZ+RZs8/CVHIKeT2PHy/yjANaELohZ5yM/HjHPtRgYikaRUNrYeaqVVtmXtSMmvCKuM5CCRKcNIklv1NlhA/J2xF3n5Mhw9ZRA6A0dRDfOvE98f0KGt47K3mh3loATj/APud+kROTTA2ZTk4LxQBgZTyADKx+yEiBERdYJR7svhJQVIwX2E3Hq0zzqFHElraFbCwrjItxniiqu5k3ATv5c6RqtmrrBZzdNC4GVqlMA0CUah5a8SAj+ZM4HrWnvB6h4YYddtSUyqx0kVWimOFEf0Uq3QBgoYSbvHTDPSNyPseGtoElr7A0NWJ1lwy7VCN109DxXzl3pfi2TRmh9YkyRmYusLr5vLrJ+Ln/99UFVBtVXD5zOZ6/8/ni2/FIHP/fS/4/vOmSN+H3n//74rwApDKNqjo8I5UmZaTY6hqpbIquOdm8CgDp1BTm6mw9PsoNtKOCkIxIoKyRUTCwyqE0Oaplj6gyQrtcV2Wkron++Gxy0vKMRO1lmlPRKHSzgUQkgV0Za7ccUSP4Gd4l8Z1jPGiqW2XEaADH7gQA3BezdhpVj/ISTBP41AuBv3qOZ8hYVSKIrmQsSOhFfDudwov27cYPk4mOlZFl/j7NegzJk2GbYwHgWzv2oiItkF7TZwFYZCTNSydUquHKyMnMBABgd86eAiu39pp+Rh/oBeQjDmUkOQbMXopxngPhZWD9YZmde8/d+VwR3U7mak8yInfZdKiM2AysTbNpNFuZBoDVUbP6IDDu0lGzdQILMbv350g61/w4Hyhy82kmksBMfAIAsFp2TO7Nn2Q+pUicdRKhOSzs0JqPUs0wyYhWtNaMjEVGrpqzBv4RGTkSi6GeP8myVUhBmDlPUkY8zv16BVuqiv9bZGWud1z+jiYPUSqawm8/57fxC+fa01lFmaaptbeEouIkIx7qoWHYlJGmFn5FwQFuNTu24VKKGyJCMiKBUliLI+AZKXP/QcoEkuTy72U4WZ9gmqZVpnErU+glITFmU9NWmSYSAYqLLLipuISn+MJ6YOwAIvzvJbxsHxt89p3j32GplN16RhYfZOmEiXHcV7Rq62WP8hL0EnPS509YOyoHqtIN2atMFRj0Ir6fTmEjEsHdqWTHysgK9xx5DcmTYZtjAeCrKnutJxITADgZaSNV41KW24BHv86IHXXS8JkZux2R9OQZMRQFZZ9lGlJGqBMHADB7kSAjzZ4RRkYO85yFZ+ywRg9Q14Hu5YeRrz+pQ6GoF0U3CbUpO2EpI2rL0DMq01w6fSlURcVyeRlL6Qn2OCIZhgHkT+EEV0bovXoqxq8b8nb4RLHKNj7ZaEqcG00prPScE/sAVWVhYcvsZk7v42PrLYZOjgIZIb9INIVNo47Dm4cBAM+ae5Z4yO7sbqQiSeiqguOVZcu8mpllf4MPZeQLY1mUzToumLwAL9r7It+H512mqVgzazg8Td5aHoDpWaYBgP1g1/Qxt1LcEBGSEQk5EXw2AsoIv7GlFBUxP/kHQ0KlXkGDl2FcyxR6kTm/AWRlA2s0wrwIh1mnxpFJtts6Z7w5Cv7q+auRi+ewVl3DT1d+2j0Z4SWa9X1X44g0mbTsVaaRv+7hK5HJSP+VkRI2+CKjKWrnZRoKQcq0D6yiG9ypaASPxWP4aXUJESWCV5/3agB0U22jjFz6GnbTXnuSjZ3nA9hOcif/nuwe248lI0mRtVAw9PbdQrpFRoQyAgDZOYw3GBlpmtzLifNynZESeagekZGa6fF76e+NJGwx38cKjNhOJ6ftx+H42wAeB+/STeMs06RjaZw3cR4A4CHw30vBZ6UVwKjhOFetXrSH3fCOmTX2ynZoYi1W2XmbiWYwwz08q84cCodf5ETxBJbLLCzsFy/4RQDAo+stZP+RICO8/Ti7A/evsEThc8bPwVRySjxEVVScP3EuAOAJ6MApPreGxiCItcf93C/pRXxujJ0D77j8HVAV/7dYq0zjbO31KNO4EXaPuTQy9vOk4aPF7kYH9AshGZEwSq29FX5jTyGKhDq6ZKQgRdS73ox1y2SYiWek1l5e7378mwCAI1y6PzjRHKgUi8Tw4j0vBgDceuxWa3fSJRn5yY4Dti+XpfHpzmN3/VyCXLrY8lBPAoNWxAZ/LV1vai1gmiZWWFRSyyh4wnRyGlk1AVNR8NHJCQDAdXuuEwSiorZQRug1yc6xMfIAcMdf8K/N40SVvU5OZYSlsHaQNeLmGQGA3DzGpLhu23nJlZElTvbn0nPiWzGuXngrI5z8OUs0LZJXCYkoe+6aoqDhIHFmrSLM6knpualU87DGk0ALp9l7zkkdZYxcPX81kpEkajBwMhrtuL23xK/hbCKHmXGm7KwYjpstPafDL3LZ9GV41ixTFloqI2l+wy8PMYVVMq+6+UUI509dBAB4MhYDjvCAxB1OMuJ+bv6ktoF8JIJd8XFcv//6jg7P6qZxKCN6SQSeiT/FGYxGqGzCALDKy5duysieKCNLp6stsmGGgJCMSCDPSL4Pyohpmnh8/XF/8jOACl8gUmrEcvm3ajkcEorSjXxL22p21OslyzMSywqmXlQVFnt85DYAwNP84nFTRgDgZfutUo3Zrr3ODY0acIx1T9wbt5/2Za/WXi2PpUgEJ6MRb2VE8oxslRb9H0830EvY4K9Txa1FtAXyeh4638zv8CglyFAUBfv5LvnONFuAX3vea8XNsuLHMxJLAhe/in3+BB+StuNCnCywfBmnZwQAssI34sPEqhWZ2RUOZSS3E1GwcDfAUT7TCjABLGvspijX1ONERrw6hepVrKsqjiU8/CIuyauEpNROrTkkfk1q2ZcNvTTF+LHCUYDiwrdOAFsnYcLKGNk/tl/87iPxztt7i1wlyiTGsYOrMato2K9lR1uvyOeYezYunLwQALBYWvRubx8JZcRq621JRmQTK82w4srIE6rR0i9V4uvBzsRkU7m5HUSZRnFTRhxlGsWjTFPdxIaqoq4oUKBYLe8S0jQkcsQ2tyEZkUDdNP1QRh5afQi/+C+/iD+66498Pb7MyUhajSHGlRFtxE4ewK6MNMwGyo4sFFMrWMpILINMLCMW3JVIBKhXYAI4Ume713PHz3X9PdfsugapaAqnS6fxqMI9PZ0YWE/ez1rh0tO4r3DU9i3nMROM6hZ+edccXrdrJzSPlsSqRIjyffeMFKQyTWfKyDI3JE40Goin23tGAGA/bwMEWHz8dXuus6bPei3IjbpFkmJp4MKfs8VoGzMX4lSRGQKdygjgaO9tl/+iFzzLNAAwxm+mgow0akBDw6aqCpOqOxnxSNKta3jH/Cx+YSqObzxtTaBt10kD2BUP5yC+qnQey90VF0+x1//Q+mMwJ3hJa/MYkD+JtYiKimJCVVTszu4WSbZPxWKdKyP8eHLJaczMMGJRUxTkZWLhKNPcv2Qll2bjWfG3e+aNjAIZ4YFn5fSUUHFcyciEREboeGfOx+cPfR7/sfooPjI54bkRqvJzP9kmy8cN5Htyi4OnUjeVfTwTWKXAs6nkFGJqrOkhiQS7Vjxb2IeEkIxIyEnD8oLG8QIzn53wWaer8G6ClBpHglz+I3byAHYyAjSbOHVtC3XFUkYURbH7RsDafIuNKiJKxHNBT0aTuG73dQCAb9d4OaQTZYSXaAr7n4/HN5iRcl+CScdlj+cpV9axEo2iEFGxWfYwsEolgC0/HSA9oKaXBLGrunVltMAKJwA7Gg0rEbMN5M6QXzj3FxBVo4JIVlXFvW4u7+hiKSAzA+y3UiZXJ/dAN3SoimrzaxCs9l4/ZRor9MyujLDnHa+zTYU4J7kHaFlarOWo7DgnDJ6ekYaGhVgUhqLgA3d8QHR3tQs8A9hNJMZb9DVHi36Vv2YxRRWdPQDboUeUCNar61ge56W1zeOsk4Z7AubT84hFYoLEPx2LMt9A1ScxbtRR5IprJj2N+NgejHG/zcq6lCgsDKz7sVxexvHCcShQRInmIl7a8OyoGQUywpWRJ+IxNMwGZlOzrucgKSMnYjGmUgA4kRrDR+//KAA+zdpDGSEyIitcfkFx8E0lGClnhGZKeYaeyYFnHsMw43yiddWra2xICMmIBKubJnhlhIyOLaOmJVT4QpyKJESZxnV41ZDRjowUpBZluthsHTUAjkywHfLe3F7EIt5R5VSq+Xb1FMva6EQZWfwpAOCB6b0wTAN7c3txIM0WonLDnYyUpGOveKgeFWnRz/czB6auYwPWTZLFivsv0yxvMTI8W280pYd6Qb65vua81wCQjJhOKZkgL6KkBlxstTBSW+98et5112af3OvfM+KmjEzU2fUiury40rLE2zNlvwgAxLmvQ4OXMlKFzm9ODbOB37z9N3H7idvbtvUSktycW3XcyOhaTzpi+pPRpFA8Dom23QWWMcK7nPaOsbZfyuY5kuQ3Qb/qSGVd+BGy6R1AJIodvLy1ssY7SfSS1U02eUCoIhdNXSRed1JxPH0jo0BGuGeElNWLpi9yfdhkchIzKjtHDsdjMGNp/I+H/4/wh7XyS1U5sUvGOicjmahcprF309AmhMhTySv0rLzWspMGAJKcjHiWI4eEkIxI6GfOCC1AfkstdOKnIkmxe9NaRXAPCUWH+dNpYi1xqTcDVUiMxNgFGZliuz55aJobrtt9HWJqDEdreSZHd5K7wo2L99fY8Tx79tlIRxk5Knu8J3Yy4t4pU5VISgUN32SzY+iWeRXwiBVvgRWuyM2aERZC5QPPnmOv0fX7rxeeBKtMo7orI9JsDtFxcvHPA1AARcUJnrLr5hcBnJHwrcs0NSml2GZgTeSAWEa0925SiY3Mq0n2vjt3jjG+m/Uq09RrZTT43/SC3S9A3ajjv37vv6JYK0KBgr25vS2PN8HJV9XxvlU4iU25kDNRqonx937zOLB1Uigj9DvJa/V0NMqIut+skfKa2HVnOLGYUR0prERskuNAakJMupVLHFZJaZSVEUaoHmuw84CO2Q3nc2/bk7EY/nl2L+5evFt8r9Kik6zMN4wpHynHTtC532RgrVlx8PN8A+WpjJTWPAPPCPHkBABA6/dgzw4RkhEJ/fSMkNHR782qzMlIOppqb6wbItopI0X+/6xiyc/OMs3TKXYReplXCdl4VkR3fzuT6qxMw81e95WZefLKuSuFkavsYQwuS2FoFY+2Xc3RFty39l69KMyrAJVpOlBGuLl2hxpv80gL85l53PlLd+JPX/in4mvCwKp6eEZoEZU7TsZ2Aa+7GfiPn8XJGjsf3PwigOwZ8aiJSyhKuTZk/gPASFBuTnTUNCkjPOekWRlhZKQGhwmbQ5cMg3/2wj/DS/a+RLQN78ruspV83JDg368advJLa0PK5edF+cPgr8XmcSB/EgsxOxnZl9uHiBJBSTHZxFm/JtbSqtV6zz0LO/iNdLXIB1k6/CJu5k9SGY7lj6HktmMfBTLClZFDvJurJRnJMVXw7lQSfxZn19kLdr8AAGXsuJ+bVV7iS7aYiu0Fz6m9elk0Acxl5qzHuK1/pRXvwDOORIqVp3UvBXBICMmIBPKM5PvgGSGlw7cywh+XiqbbhzENEU1kxBF8VuKKQlba9TWVafg4+XMn3M2rMl6898UAgHuTyc7KNLUyqoqCh0pMIbhq7iqk+Y2v7FH+KknEourVTdOGjAUGKWME8AjPaoEV7nmZ7dBYF4vEbD6GFM8o8Aw9k5URGZe+GrjstTjJb3CeZET2jLQp0xQ4uchEErZjBABk5zHRcKSwCs8IOxedizWREd2DjGjS+ZaOpfHnL/pzXLub+WEumLyg5bECzP8FAJqDRFqmx0TTz1zMTcSP8Rso1o8AhdNCGaGBhrFITBCTI/EOTKzlNSs4jpPzGR5st0IprFJb75a2JcLC5LkqU8kpQe4eX7dmGAkIMrLp77j6gdIKdACH+ZDGi6dbkBFOrr6ZzSCPBi6augjvvOKdAFr4pUzTIiM0abkDiDKNyjOEKGdHLtNwZcRzNk15tW2ZJpFk70UDo1X6D8mIhJyUM2IYwU40FJ4Rn+25FTJCxTNI8F3mKJKRpjKNo5xRpBuGai20ljISA6DgKZ6j0E4ZkX8279Vn7wW9iIcScdS5cW1Pbg9S3DtRNuos1dKBkqR6VLxyRpx/f78GgWn2Mk1F7dDAylM2d8TcQ7n8IhWzyIjpSkaorde9Zt6WjMiekTaG4AJ/7XNRl11obq45hZW/n0tcYaJdJqE9GWHHE4MCVVERj8Tx0Zd8FP/jmv+B37n6d1oeK2ApI5rDiE4bDzcyQsrI6eoaNlWVmTBNo0kZASwyfyTWQXtveVXsukldmuGbBZHCKrX1kl/EGRYGWDd311INkRFti3VcDRq1KqDlcTgeQ92sYyw+hp0twv/Ol4hWBCr+xzX/QyhHnp1k8vTlWOdkhMhgSVG4J45fS1I3DZ2zRUUFjHqzb6y0KhRnt8AzAEhInrG+lZW7QEhGJJBnxDSBkh7sBdOxgZUz1nQsgxjfzerwcPkPEVSWiPBOgSZlhN8wstJCK4LPxndi69Ufwzovh7TzjADWgllpFUfuBr2Me5PsGK6cuxKKoiDNjVxlj5JASSIaXvkwVQdJ2cov+D+mTqAXbcqI5jZwrQWW+fvkZ0heK5CB1VAU1NzIQhsyQm29e3J7XL9PC3LeK+5aQp6rjTm3hT+30/KMULmNv1fLCiMbTZ6ROCM1ugLX9FfKB0nQBFqwVtzXnv9aT3IlQ0TCm3Yy0koZycVzgnAcStFro2CTnwvy60hk/kisg+Cz8ro1+ZgT1R2UwsrzR+QyjZwv4oTwjbh11HCfAgD/nT5BgnfSPMb9QhdPXdw0M0bGuXPPhspbw9988JW4ZPoSQcQ9155ahX0PQDLROeknD5upKPYcH6mbhgysuqpAB5oJe6m9MhKXyMgoxUWEZERCMqYiyt/0oE2sVBfWGpr3qG0JZb5gpeI50XKot4vHHgJI+aCLpEkZ4RkeGanVjRj7SqOMI7suEz+f9lFnpQvWs2bqhVoZ9yXZ60i17jQ3PZYVxfXGJxOQikcWSdVBUvLOaadBwcUzYvos05imibU6+/tmHLvZTkFJooCHWuRVpgFQN+pY5N4VX8pIuzINv6ZcI9izc5hosOtFEGQysHJST5I3Ic7PrRrcM1yoTBNXOguzIohheQ5pvCpMj+4EjtSRx8ZYWyeVaKaT0zavDJF5oYz4WGcaxRVWFgBLSAZgpbDSjUq09R5oGRbW0sQaiQKJIU7u5Rkjj2bYNd+qRAMAyXga79jxXLw8uQvvev7vs6+JSH8VhmsnWdmK9e/CM5KKphw5Iux3mLWy6HiSfU5NpUzTRL28irU2nhE1kUWMnxuhMjKiUBQFY6n+dNTIseF+6nQVXntMJcaQEPLxaBmOAMszQju0pm4a/ndnJSl9hoduVeoVPLjyIAB/JRrAMWbbLxkxTdT0Eh5MMJncIiNSjdblxlqSCEjVo2236liUtgqn/R1Tp9BLtjJNQ1FQ99lKXKgVUOPnzpSPIXmtEFNjiHJloOKmjNSreCARx1ui63hi4wnbtxZLi2iYDcTVuMhLcMKeM9JCGTEMNr8GwFjCRe2RIuGFZ0QvoKwoKHAy0uQZ4eeD7pHhovPzIdHBvBEZCX4zqzjKrVVuTE9G3f08lMR6KMXWAZrW6+zeofbep+MxRgppFksLlKXpvFSGmJnkKayKwQgNV1nKuTlBNORJtwS6wR/ZPGLbcd916i686ZY34cc5IiNDiIQnZSTOXjsieK3wG6/8LP78Dd8SbboyWdQ8lBEq03QTeqYoivCNFFWrY6dar4gurrH4mLQhc2TxaHmsowFDURBRIpj0ms4dSyHByUiojIww+mVileV8P2yUyEg6MW4pIyPWigVYnhGaWdKUM8JvmBnJ0JWKpsRu9oenWUS7XzJC6klFUWD4NbDWKjiUiKGiqpiIj4tFW7T2qu7KSEm62Vc8VIiKgxBslZf8HVOn0Aq2Mg3QrMp4YZ0v/lnDQCLdHA/dKVKUl+H2+2tlfD2bwX2o4tMPftr2LfKL7Mru8hwgZrU3tomDl4fkccOlDbl5UaaxDKxFLPF6eiaWEb+LkIhSmUZhaa0OUHJqoldlxBGqRuQk6eZ9gdRRw43eJ6LNfhEAODjGlJH1SIQR19Iy2qFUZoQlpkSEUX7HDCMVRVVBZeVxTgoV3FU+gYbZwJ7sHtewsLn0HCYSE6ibdRzeYCbXe07fg/d89z14YOUB/FOadwsNRRlZRgPAEyp7DdspI26QyWLFw7xNykg3oWeAtb6VFb7ZatRR5GRVhYpUNCW1vzvKy1KJZjo17R1HH8sgzslItRN1uc8IyYgD/UphrUg75rZs1GigwuvaqcQE4pyZe4YxDRGkhJDs7iQjJd7GmHVI6VSqEYa4CZ9kRKqreqkVTdBLuI/7RZ4992xxI7SIjbs/oSyRxoqH8Zi8QJO8JJDv165Pmktj/W5/C8lalRmEpxsNlhXRI5IqhXe5kZGKkJRvP3G77VwX5lWPjBHAGQffQhnRi8jz1yOXdNkBSt00xVqRqZF6sWU6ZSzKu9YUuJdp+Hsdl9rUOwEpnE4y0q4dlMjIMaOMsqKIab0UeEZIx9LYlWF+jyOxmC9vRpGIqrSTz2bnkeQG/tWj32NfHNuFb5/8PgDgpfte6vpciqLYSjX3L92Pd3/33eIcWKTzdxhkpLSMY7EoKjCRiqawP+edlusFVVGlrBg3MlJlXW7wVrnawYqE554RKWMkE8tAURR7bLxM2Mtrbc2rAIBYSry/er8njXeAkIw4kEv0p0wj3zjbKiO1MmPGAFKpCamWHWyHTxAgzwiREWeZpshLUlk5lAqWuYrKV36VkVQ0BZWftqVG1VddHLUSHkowMvKsWct411YZkcppnmSEv5dzde5P6FtrbxEbjp2OXzKyzjtppoIiI0IZcd8dUoR2pV7BXSfvEt86wf00pKK5gco0FVVFrVXomVZ0T18VTzSHnNQhVdALXBlhx+7MGAGAOG+9rXlkuOicjCSdbcQ+QWSjCtPq3jJNVHnZKCVnpUiYSc1gNjULE8Dj8VhT4JkMmnp9JB711UZb5OZe2XuiKAqm+TW2eoIpl/rEPnx/gZERSkJ2A+WN/MtT/4J3feddqNQr4jgX+QZrOGRkFY/G2ft74eSFHQ+xI5DJ2HXIXK1sGVi7KNMAzsm9ZUbuKQeGKyK2YEC5lClljMy0mj8VSwtlxJmTNEyEZMSBsRSVaYImI9bJ27a9Vy+x1k0AqcQ44nwR00aMjNSNuhgyR7vdJmWE7/qyjrq+c2fql4woioI0V4rKgL/gL72ERX6R7pMm1lqSqOLuGZFC5rzmOFD88xxvXd6qtbiB9gBTK2KzSRnxpwytVUgZMexdDV0iyeX8iuvusCJSUQHg28e/LT5v19YLwFY6KbUiI15D8sRBTiAaSSDXkDpq2igjIs/HYwihxt/ruCO23S+IjGiyJ6WuWdJ+i3ZQKis8moi7tvUSrI4aH8qIaaLINw/OkhWF460sPwIAuCc7jmKtiNnULK7YcYXnU14yxfwt9y/fj1KthKvnr8bHfuZjAIAlk2+nhlSmeYx7xvz4RbxAZMR1cyJ5Rnot05QUnmVSK1kJuZyoeCojpVWsUieNR/oqAEBVQTZ03SNZehgIyYgD1N6brwRcppFumm3LNHpJMOxUNI04PwlHx/fMICct0g2mXC/bDLpFTkYyDildvlgmE5OYdJPaPWDvqPHhm9DLwisg74gtZcSlJGA0UJLKYhWP4WlEUma5b6HQJ0NYvrohTGwzFDzm0wkftDJC4V2uZKhWEcoIANy2cJs4H/yQkagaFc9faEXspLk0Yw7VDQBPYZ3HOO9Ay2t5QMtj0eU8IJAy0lAUNFxUJ50TlESL+UmtQJ6UqpxeW69au+kWQVl0A/1pIiHUHQo8kyFi4f2QEb2IIj/HM46W7x1RdiyrvPvp2xH2t//Mvp/x9PvIxwkAz5p9Fj72Mx/Dvtw+KFCgw8C6qg5JGVnBY1wZIUNwNyCSUXG79iRlpFsyYiMa9Yqt7Enfo4/NnpEVrPENy3SqtTcswW/9zmiCYSIkIw5YnpGAlRFpl922TKMXxUmdjqWFMlLzbosfCqiTJhVN2Zzbor3XNMUNPZtykBGpB95PvoiMtK2jpr06UNe2xI5BDrqSlRFTc1yUWoGVbzgqZsM1GI3q/XRz2zKDJbGEDb6Dzagx5IRU3IUykvJP+rzQWqqu2F63vJ7Hjxd/DMCfZwQAcnwhL9RbeEa0IluM0byrt57IYWLVLGXElYxIcey6C8nVuP8p7jJDxg9EmUbOiKlXhZKU9CjTAJYycnuOEa9sLIsJF+MukZGn4j7ISGmV7cAB5BzKJaWwrkYjqAP4XoXlw7Qq0QBMebx+//V44Z4X4uMv/TjSsTRikZi4OS5GI0MhI2ZxSZRpelFGxNRq0yUosVaxWnu79IzYIuFrVVv6KrVeW+Fojk1UeU2sczPJ1l1z1BGm62GZZmQxJoblBdxNI7XztVNGGloBGj8BU9GUaDnUoPjzSAwIREaysQwiK48LuVzkOtQ1ITFmHa2cskzu17xKsAeftVdGVkvLMBQFURO21EibGVaaQwMA0AqiVgswD4Pb7xJkhA+S21JMXwSpU2zWGBmZjGaQFCY6fyrMeoVlLEw1GkC2hbHNJywy4i5VU24FmRm/c+w7qNarWK2wzo1WnhHASlQttCpD6W08IwCQncO4o0xDCpmrgVVSPHSXTh5KTk10MN9HBiUp28s0VV83MHotqey5N7fXNbSLyMhiNIpyea31AUmBZ3SjI8xIwyzvTyaw0ahgIjHhmi8iQ1VUfPjFH8bHX/px2/tCmS6L0ehQyMip6ioKERVRJYLzJs7r+nmSREZcplYbelms292SEVt0Qb0C6FaZxqmMFJ3xBqUVQUbaKSOUlaO1STkeJEIy4kA/lBFTMqkBgN4mObMiXayMjLCTT1cVmC4th8MCtfXm9CrwyedjXGGLtFBG9JKQGDMOMiKXac4dbz+TRoa1e/CXNbLM2213KFGbxCwvGGWn8VQriGMHOPFxkTSrXPmZ5cbBvKrCKJxq/0c06sA3fgd49OvtHwtgnROhyVjGKpP4LNOsldjfP22qgSgjKf66lQ13Ex+VaX7h3F8AAHzn+HeEeTUby7qXVSRQomrRaEFGtAJLaUULMiKlsFrKCDewZpqVkagSBXks9Xrzey08I20G4nlBJLDKnpRa1Vc76M7MToxL6oXXhOCJ5ATG+PlxqtKmtbe82lQCIFAK60okgm+n2fX2kr0vaZ4B5BPUCjwUZaRRwyGDrRPnj59jI52dgkzGFZf5NNWapTJ0a2AVZRqaii0bWJvISHNrLwWeeeX4EBKCjHQw36vPCMmIA6SMBJkzUjfqtiB3vQ0brfD5JorJFrC4tFDUR6jGR50z2Sr7OMZVG9FRI+1eswn7DcimjPg0rxIoGMhvCusSX5RnFftNRFVUpHlnSNkpV+pFe5nGo+Omwk3Fc3zxNhQFpY2n2/8RR27DbQ/+LY5957+3fyyATV6ymIyPS9NffXpGeJlmKjHBvBQ9QuwO3WYlSTXuF+19EXKxHNaqa/i3p/8NAPOLtIrhBqRI+FbKj14Uu/qxmAe5kebTbOlbqOlFUVN3U0YURQGdIbrLeaXxmSoJl9h2PyCyoclkxKcyoiiKrbzgRUYAYIx7iopOtc+J8lqTOZIwM85aX1ciEXwnw8hIuxJNKxAZWYoMQRkpreAQL9FcPHNZT09F535FUZuUEdl/EUyZptzU2gvIWTz2Mo1ZljwjyTaeEb7uhcrICKMfyoiz60BrU6er8F16WlGhKAoS0pwDbYTICLX15mpsYR3nvevUUaNXN1mbJFxk4NQMovyC8DOtV4YtGMhHe+syN3DOuexW0rzkUXb222t5e5nGRRlhihf7m8fTs0ia7G/d2jza9pgeP/pdvHt+B96f1F3noDixzlWIifi45Oj3d46u6ZsA2ku3fmGRkWbCXquXUefv+XhiXExZ/ofH/wFAa/MqgfwLxRZJxfXqligHeZdp5kWZZqu6idVGBaaiIKpEm4a8EeI8XVZ3M7ByP5Acid8JhDIiDzmsa6Jzrt0NjDpVgNZkhDw3+XZ+gJK3MkIprE/GY1iORpCJZfC8nc9r/XwtQGTkdDQClAecwFpcxiHqpOki7EwGvUdVxUUZ4etDAmpLk28rEOEoUwKrNJdGtPbayjSWslEur6Gq+jSw8nXPNUl2SAjJiANWHHxwyoiz66AdGyUykuJvT0xapPQ+tY52A/KMUJ7DON85Ehkplq04alIzCPFIHH90zR/h9577e66SeSvYdw8+lBG+Q5yNNBsE01zSrjhUj1plE7qsjLgEo+mGDpN2tclJjHEJO+9jWN7x1YcBAE/EY9hae7Lt4ze4CjKVnBISsHPGiRu0hoYiv/FNuaRmdoMUGTHNZjIkz/NJRVN46X4WkCXC8dqYVwEgx8s4BQWuSagAUJSmI3sbWOcwQcpIdR1LEfZezaZnPG8WMTL2ufhVNP73xrtURpKyZ0QoI5LpsY2071cZydHOudaGjJTXUFTclZEd0xcAgDi/X7jnhV2XpwCrLLYYjTJj7SDnbJVWRAw8eW+6heimUZs9I1W+rnebQwM4SzAVm4FVlGncUopNU0xZTkdTbed8UUeYG+keFkIy4kBflBHHG+467VQC7dJT/KRWFRVRGmw0QspIgS92Wb7gj+ns7yQDa4mXB9ImXEOG/sN5/wG/fNEvd/x7idiwyOT2Nc8lbv6cd9lBp/mNpeR4T8qOJNWqS5mmKv0/kZ7GGL+ZbPnwjKxuHROfP3ziB20fv8kLfZPpGctA6qNzh6LgY6aJXG5X28f7AXV9uLU7l/m5HleiiKkxXLPrGpsXwpcywlWLgkdpDAAKQj2MevsYcjtFay8jI9RJ4z06npSRmgvJpUGViS7bNl09I3VNpHa2aweVI8zlvBwnKGCw2I6ol1etG52D0E2mZsTUWgB42b7uSzSAbGCNADAHOrl3deMIVqJsotIFkxf09Fwp2cDqUEZoVlOyy4ReQM4ZsZQR79ZeKfSsuoU1bnhqV6IBrDZ2zWdW0SAQkhEH+pEz0qSMtJlGStNQU1K4UoKMdW3Gqg8SBb6g5Hh5Zlxjx0YG1iJn6tmATzNLyvSpjPALdjbenLGRJjOmgzCWqva6dkVRYDrSCiucbEVNE7HkOMb5IrFVbmMc1ApYk57/wZUHWj/eMLDOjbKT6R1ih13xU96RMkaUnPdNuBMQGam2ICP0uqaiKbxg9wvE99t10gBAlmehtIqEJ6Ul10pNcJRplnk+x2zGu6MoTsqIS6Abxbj3TkZUK6xPHq7Wpkyzf2w/njP/HFw1d5XnRFbAKnPl27V+l9aadt2EiBrBNBh5SygR23vYDXZm2Lm3EokwWj1A38ihjccBAAfU9opBO9jLNPZ1nNb5VJfdVoD1PpTJIKuXW4ee0fpXXsMq94vMSLEJXiDfmeZ3pMYA0NVd4hOf+AQOHjyIZDKJK6+8EnfccYfnY9/ylrdAUZSmf5deemnXB91PkDJS0htoGMG00TqVEb1VfgIshp2WZFFhrPMYZT8MFLeOAgBynDSNa+zYSBkRUdNdDhbzgr2bpv3rscyd9HMuUyxTFHzWREY2AQAxvlM2FKUprVDjj0mZJhBNiC6RtvNplh4VLXgA8NDW4daPr5VF+upkZl54NjSXMokTNJdmqmEAuc7KYV5IcoWJxZrbCUmZl4TSEkmQd9W+lBEeCe+sicsQJcJWxCA9hXGTvW5b1Q3X4Dsn4vwGrLslsPZIRuxlGgo903yXaVRFxU0vvwk3vfymlp4EmtVTNHTXbByB8hordaKZjADADD+fr515Zs83cfKINRSFRZb7iKrvBqZp4s9+/Gd4563vFP/+fIVF2l8c790zRe9RRZqqS6DxDMkeunVscfD1ClNGnN00VKaRS8elFdFJ40cZoQnS23pq75e+9CW8973vxQc/+EH85Cc/wXXXXYdXvOIVOH78uOvjP/rRj+L06dPi38LCAqampvC6172u54PvB4iMAEAxoFJN1dE2qrfJoShzMiIzbJKP26kqg0Rhi73nuZkLgFimaWR7iX/Mdhmf7YWmyOQWME0TS9z8OesSkSzMsI6LssQJ1bQqTep0dCdU+e4uyTnrODeN5dt1MSw+KFzvAPBQZQlmq/wYaS7NRGZOCl5qPzjRCjxrANmAPCN8May6LMhlvtNKSzfsF+55IcbiYxhPjPv0jLQfllfgpHysRVAYFAXjvOSzVW8dBU+wPCPNi7TO1al4lzdmUkY0KfSsVisJw28q5o/ktOtGyvGbUUFVgFZx31KZxmkwB4AL9r0IAPCqS9/o67haIaJGRNBhP9t7n9p8Cn/36N/hrlN3iX9HGkxpvnK8+3wRgvCMKM2qrEVGuvMUAY4yjcMzIkLPJF+JSWUaKQrej1GdMm/8TJAfFDoubn34wx/G2972Nrz97W8HAHzkIx/Bt771LXzyk5/EjTfe2PT48fFxjI9b8vjXvvY1bGxs4K1vfWsPh90/JKIRJKIqtLqBfLWG8XTvN9Kq48LT2piGKnw3mJJkW6IltRFSRgqlZUABsnOXA2tLGKuwhE2S0It895rpMrHSC+QZqfjIGdnStqDzjhd3MsJLPo42WfLtjEWSWDOqqMFE1bGwUwt2khPFcX6T26qVWDid101j8SGxiwGATbOGE4UTTVNYBfQS1jl5mUpOiZ151ccUZ0sZCSbwDJCUESKD0o2szDt80hJJyMaz+OIrvwjDNHzFZNsGgXmSkQoQtTJJvDCemQWwiqJRw0kaktfCMB1XI4AB1Fx2jBonf90qI4JESt00cjtoKtLd8zqR5fOHCqrKvBmpCfcHltdQyrH30k0Z+e2rfxu/eMEv4pmzzwzkuOYz8zhdOt3X4DPKPppOTuO/XfXfAKMO/Ov7kK1V8YLrf7Xn57eVaZqIuAZEgFSXGSOAfWqvqZeh1MpN6hVdH6aioKyXkQGA8irWov7JSJxUOh8m+EGhI2VE13Xcd999uOGGG2xfv+GGG3DXXXd5/JQdn/3sZ/Gyl70M+/d7j3DWNA35fN72b5DIBZw10qSMtCMjfKGykxG+YxuVViytIMjG2O7nAGO7RKaD8Ixwj0W2h52CG6yUwvYJrEs88Gyq0UDcZS5Lmt9Yy44dAg1py0QSSPIyU9lBRii1NcnfmzGuPOQVA5C6PZogkZEEf80eXH3Q8+HV8pqIDJ9MTlqeDR9kZJ13NE03DCAXjDJCseYVZwql0UDZbCYjADNcHuApte1A5a6mUCcJBa52ebb10nNlLZ/MYd5R0apMQ/kLupOMmKYYVJlwURH8gJSRmqKgwf+uCr95RoCuA8WcsClLXkbRRg1GdUvEwTu7aQD2PgRFRADLxLoU6Z8yQt1cM6kZvOrcV+FViV141dYGXmKmEJt/Rs/Pb+umcSoj/JzpNmMEsN4HQ1FQbVRg6qWmnJFkJIkIV/CKtDktrWCNP65d4BlgEWrNZ1bRINARGVldXUWj0cDcnP1inpubw+LiYtufP336NL7xjW8IVcULN954o1BUxsfHsXevdxtbP0CTe4PqqKk4FoR2dTpL6rbkYCIjmpsysnEM+ObvAZvtW0oDw7G7rNkgk+cAuV0Ya9jLNEXujcn0sFNwQ5OU2QJERubqDdsOXjwXv/GVzbotar/EF7V0NIUUV3Yqjk6mKs3m4WRlnC8CW6oK5E+7H1CjDnPZ8ow8v8Le64dWH/L8GzZ5gmrUZLujlDyKvg3WiqyzZ9owgEx7Y5sfCAOtKnWFALa5NOkWQ9/awdYt4GVg5ddQzmU+i4xIbl5M7qVcklZkJKZ6kBGjzsor6L1MA1j5DlV+niURaVt+8QsiI8VWZKS8zkzZirdnJGgMIoWVpogLj8vTt7OPB68D1I5ud66wddM4PSP8xp7sUjmTnx8ASrUKtJqV20Pvq6IoyPJ7Q5E2A6U1rEY78IzQBGmfWUWDQFfvjvOiMU3T14V08803Y2JiAq9+9atbPu4DH/gAtra2xL+FhQHeZGEpI0GRkaojUMstw0CGUEak3Uq8Rf4B7v0scPfHgR99uscj7QBHbhOMPRfPAWNW9HZez8M0TZT4wpCLdreT9ILwebj0+jtBZGS24UFGeDKsc6dT4s+biaVFMFrFESJV5cpPgt/AKK57S1WBggcZWTuMgqGJDJOXlNnveWj5Ac+/YYPPlpkEM38nhDJitp1VtM47e6aiGcClvbobEEmuOuvmtQprtwaQ7uHmJt9MTTcyYpoocAUm127as9TeS2g1Xj1OyohTvq5XofM1LtHl3ybvmKucqFcCyKZwwipzKS3IiOUXiSrRrlNlO4Eta6RPZIRa9MVG7iiRkRcG8vxWmabZXE2JxN1O7AWYSZk2b6VGVZSdFMfzCjJCyoZkYPWljPA1xI8JflDoiIzMzMwgEok0qSDLy8tNaokTpmnipptuwpve9CbE461bnxKJBMbGxmz/BokxbmINqr236riJNe26HCjzhTAlLXotyQhd2FuDI23mke/ZZ4OM7RZkpGE2UKqVUOTHmunRie+EPWekNRlZ5jfj2XodcDmONCcQZcWeaVEWZCSLlMpbMh1ZJEROktygK7ppWpGRxQfFopGNZXEV2HM/tvEEah4BX0RGJrjFK8lJVVVRPEPBCGu8tXe6jYLQCTzr5rWypYy0Mpa2AZGRuqKIAEAb6lWhyo2l3JNUrSezgs8AYFqJt5xNQhN5m4x9dd1SRros06iKKrqzLGWEvX7JADvO7GWaTfcHyVHw8UxgqkwrDFwZqVWB4/ewbxwIiIxQN41bzghft5M9rncyGSGFNhNJ2DqoskTYDR0wTZjSkDx/ZIQdoz7I8Lk26IiMxONxXHnllbj11lttX7/11ltxzTXXtPzZ73//+zh8+DDe9ra3dX6UA4YVfBYUGbHL+1obBzOd1Km4TEZ4y6FbXzjdRL1KA0GjsARt5TG7fJjbiaRpIs436lv6FkqcdHkmZHYJOYHVdecsgcjIXKMBuBwHlRPKqmqLe6djz8RzVvy6V+Iiv7kJZSTSokyz+KBt0dib3YWJRgO6UcPjPA/BiXW+cE/xG2UyJhlI25xL69QV5CN7wC9oQa46yWC9ypJqgZ5aQVPRFOjWXKy63LQ0aWJvO0layhoBgNlo6+OKc4Wi1koZ6WHnS6SjSmSkwc3qAZq8qTW6pWektOqZMdIvCDISiQLt2t+7RFmY/1PAiR8xo3B2Hpg5P5Dnp44n1wRW3vqd7IGIA0Ca/45SQ0ORr/cZh7k5I9p7AdQqKJTXxOgNXwZW/vOaD9/ZoNBxmeb9738/PvOZz+Cmm27CoUOH8L73vQ/Hjx/Hr/3arwFgJZY3v/nNTT/32c9+Fs997nNx2WW9DSoaBMYCLtOQFEtJpe1MQ5RsmZKGywky4tJyCI3fRPM+psUGgae/L3amqqIySXSMpXvSfJq8lhcSYqaNybBT2Exe7QysJabizdYbQNxFGaGcEYcyUuLppunEmDASN5MRvvBxsjKW8KOMWObVqeQUlPH9uExjr9ODK+4m1k1ulJ3kbcZJPquIdWV4n0sNo4ENfsxBRcEDTs+IhzLS5qbfCoqiiHbwgnNmEABoeXH+5VxMyTbk5kTLOQDMtbnxUjKl7lSc6lWhjPRS0qB0TgqboqCsZJBkhF9vFVVF3UuBKK95zqXpF8jAuhZRUesXGSFlJJqW/CIvDGRAJGB1PDUlsJqmGI+Q7HHzlaVBoEYNJR6+l3VcTzke4EhZPKtVZlTPRdO+zs8kf881H/EAg0LHhco3vOENWFtbw4c+9CGcPn0al112GW655RbRHXP69OmmzJGtrS185StfwUc/+tFgjrrPEMqIFhQZ4ZkIDQNFVW3edckwTVTQABATJQSA+xIMd2WkoRfxeDyGCwqnEDWMQIxaLXHkNmsabyzLJF4iI406ViIxpoyYNdb665J82gvk2mm5XkGrfepSqY2BVfafyGUaow4ghkxywnLQN4UccfMhkZE4+U9U6PmTaCpGmiYjI1HJ9W5GcMXSHbgznfI0sa7zG/IEPw6SgW2x4i7Y0rdgcJPrpFfbcBeg16OuKKhpRYjbqM0z0ptUnVNj2GrUUHDLydCL9hJhyyfaiQlJGZmLT7R8OJVwmjYMDT0QMkKj2ymTgvxhyR5SO52QlchiZQ0Tbg9qMZemX5hKTiGuRKGjjqXqBtpn8XYOUkYysQzw9C3siwH5RQBnidKuCopY/x43Xxlq7zVqMBo6gFhTqTubkEzKehFreh7ADKbbeag44vznNR8m+EGhK9fUu971LrzrXe9y/d7NN9/c9LXx8XGUy6OTj9EOQUfC08Izbhg4hTYO5oaOMifxKYmMCGOdy074841V/NnunfidtQ28sbQSWNKmK0zTRkbEzSCzA1AiGDMYkdrStlDkCk+G5x4EBVVRkVLjqBg6yrUyWomSopumUQdcFl26sZYVqUxT11Dicx4yiQmre8Xh9RG7Wr5bysVzUKDAhIl84RSaKreF00B5DWuTbMGYTk0D8Rwu58qIFxnZ5DOApvgiZVsQW/iPKPBsotFANKAoePn3A4BWk8mIpYz0YuIDgFwkATTKImnVBrlM027hT89gXDL5znlM6yXQMLiaM+q+XoXOr8uelBE1BjSsc6fKyWSyhyF0TsTUGFJKFBWzjkJ1w52MyBN7Ay6jekFRFMynZnC8vIjTtXxfyAipl2klCpy8j33x4HWBPb9I0VVVNPSKKCfaYv1jPZIR/n6UFBNGowog1vQe2Sb3Fhaxxvef0y06xWQkuIqrjxAZCWfTuCDoYXl0EyO5uMmpL0Mvibp7SrqJx7yMdQCOmuz5H0zEgfzJIA7ZGyuPAfmTKPIx6kLiVSNAbqdo783reRR5PTIbMBkBZJOXt4G1XCujwN3oc2YUiDRzb9ucG1JGtKIIGkqnpkUN2BmMVuUqVZLXeFVFFYmgrvNpFhnZWM0wMjKTmgHG9wgycix/TLRFy9jgCswEf60tqVgVSZ5uoLk0041GYBkjACtlkOhdlef11CrW69ajMpLl72/RrZVdt8jIWKyNuV1VMS51c822MfeJMo1jw2DWqtD47+xleq0Y3c7PpSrPS+klKMsNlO1TbGlgtedXDALz/Ga52Ki0jqp3g2EAd38SOPWA50NIGUkXlljg2cQ+YPJAl0fbDJlk26av1yrM1Aog6VIO7gRU1i6pCgr8XpFxEBzbdN/N41L6anvzKgAkuIqrKWjbkTcohGTEBWNBh55xMjLO6+C6y4AxAb0kTmrZwOpcxGSU+MK5EIt6exWCwr1/AwAo7LwCgGNnKrX3bmlbKHLWnfFhqOoUGVI0WrRJk3k1bRjIekRtW54RKdNCy4tyQyaeFV1NFccNqsqJYUq68YqOGm0TaDjI7CLzhKwl2fNNJ6eB8b0YNwzsb7DXyk0dWeekZ5KXuxKcCOqqgkaL1mZSRqYaRmBR8ADb4VLQG4V2AQBqVatM04NnBGC1b8CKfZdRr26JzBA/u/pxSWGcS7d+HeL02jpaHnXpxtNTmYZfx0RkxTnUQ1CWG8Tr59aNBPDW3sFljBDm+TiApYgKuKlerfD0bcA3fxe45bc8HyI8IxtH2RcCLNEA9ve+Ip+btYqYMdSrKkjZRyVFFepVLmEn3WI+jaoCm8fEeAk/nTQAkOBeK82HCX5QCMmIC0gZyQdlYOUEgkbMa+3IiMvukloOay4nDqVeLkSj/TWxVvPAA58HABQOsu4p281gbJdQf/Kale6Y7QMZodknZDR1g+ikqTdcSzSA9RpXVGkqr1YQO/xMLCNqwBWzZttFUL0/IT33OK/ZbikKUFyy/zJSRngsOVNGmJfj8gpb2B5aaSYjm3z3PMWfWx6opjnajWWsy1HwAZfuUqIrRN4dyq29PSojXM0ouJCtIk+VBXySEak0M5vd1fKxMX6zcbY8anowZCQppqWy81a0gwatjPBzslAruj+gNHgDKwDM8UGJXbX3Lh9iH0veU7EFGVl5kn3h4Is6PsZWUBVVmI0rDvN2JaASJZ3TJVWx2q8drfm2Mo1NGfG31ia4Wt1QFNRbzS8aIEIy4gIr9CwgZYQvOON8gWinjNDuUj6pSRpumiVgmqIcshWJIL/5dCDH7IoHvsB8FTMXosgNq6QEAAByViT8WnkFGu28Akr+lEE5FqUWrL5d4Blg7eDrioIadW7oRbFQZ2IZpPguwhnyRSFHSWkxH+eLxlaE1XJt4GRkjZPH6dQ089pE4rhM46U2l1j4DU64JvhCYwvP0jxuNgDWCqxkN90wgGywZIRaVCs1++6wEpAyMkY7P6PZE1Pg7b4pqKJ82Qrj0vk3l2vtVIgTGXFcozqFT5m9xbYneBmI0jpFB0aPNzAnRNaI1yyr4pLV2jsgzwggZ410EXy2xqdba96KiijTbBxjXzgQnF+EQANMq3KJWFJGeiWW1rgLVfL15FwfU1QVYONYR4FngD0rR/dq/x4wQjLigsA9I3yXNc6ltpbtVHrRUkaizcpIk9+kVhYnLAAsbB0N4IhdYBhWwutz/4vYcdl2VWM7RabDqcIJ8eVMi8TLbkEXY6VFm7QVBV93besFHJ05dFFqBTG2Ox1LW8qIo/1X3EgS1kJhDz6TVCqtAKwfgQFgjb92M6kZ1vk0thtXVNnf8fDqw7YJvg2jgS1+vkzx11FVVJHnUvXa+QJY51HwU2ociAabsEktqpQkCiCw1l5AShF1CXWjqcg5n+2wc1mmPk3XG8i0IcY0QKzmyF+gkLKEovQUEEadV5YyQoQ22GDAHC9NFd3KmHWdlWkG3E0DyFkjXSgjfsgIJ18powFMnw+MBWfcJpC6VZU72eoV0U3Ty2wawD6V3JrYayeMtsh/WRnxEQUPWN4oANBCZWR0MZ4KWBkBIyNjCSazt3Iw16p5a6S4rIzwm0nN2Ymjl8QNAAAWSn3yjDz1HWD9KSAxDlzxS6LLwe4Z2S3KNKf4caQME9FYsBI0YIWVlRR4ppCKtl6PwDOAjTYXg/C4MmJW8+I1zUQzwhNSUe11bgo5klv5KGtkS3UoI0uPAADy47tR5yRmisoHE3txoa4jpkSwqW1ioWAl6eb1PAz+9tJUYMCaFOxMhZWxxuXs6R7d/W6gkK5qUxx8MGUaylEouJThCtyUOeazHXZ+8hz8xdIKPrK8AiRavxZxvqt1KiNUDrOsu90hya9jUjjFORQwGcnytaZg6s3eJV4+LEpJwINCTymsa0+xjw3ds6VdKCOGEbhfhCBSWOVOtlpFbCJ7JSPW5F7VKnU73iOhjCgqsHWiY89IRI0gyjc9Gif3w0ZIRlxAyki1ZkCv9x4KU+E7Wxqk1oqMVCT3u00ZIb9JExkpshOSY0HrT5gQ7vlr9vFZbwQSWRRqLmQkZxlYl3gIT6ZPRm3Zce412dXuGfFe7NPcWFziZESrbqAh7RpFzkiTMsL+1qRUz7WnsErKyGluXp29AABTUERXxvhexAFcHGc3kPuX7xc/tsFLErmGgZgU8OWHjKxrPLnVZ/ZAJxB1cyn3pl4riY6Tng2svKZdcJmdIc49v3J4bh43lCt4pqYDbVqBYzEiI/YTV6dMmR6XzATdyPjfRRuVXlM7ncjxmPyCqgLOnS8nIyVnR9wAQGRkKxJBxempagW9bO8U9FBHhGfENPtGRkQIoqzK1sqWgTXSW8lNLtNYyoj9/JCVEcOoYb1DzwhgrSFaC6VpkAjJiAuyCasmHIQ6UuWZFeO8o0EDbFK8jDK/gUQB2wwNunE1+U20ol0ZqZeCb9VaPQwc/jYABbiaTVwu8kwOuzKyS5RpGvxGne1xJ+kFMW3XZUYEwY9nBABS/LUtc6JRknZs6VjaPqmTyIhpoqJwMiKRBFuZ5shtwJ0fYf8e/goAYHWSlQxsO5hx5mN4gcpey3898q/iWxtEKIwGkLBuGnRTbJVAu8bJ1bSkqAQFS6q2Xnt5qnGv0n+Oq0ZFF+JOJcKcX8JDfplo0rW9W0aCP6fuUaaJ9zhDRuRUcDJCG5XAyQgnxa7zaXjHXTHK1hfnja6fyMVyyPB0jsViB2b79SP2/7v4HEzTFB6mtGH2xS8CWP6equRXM3WLjARWplEVT5OxRVgUbKmqUNP9lmkAS+UblTJNcKMizyBEIyoy8QhKegOFah3T2R7q7YYhWnXHc8z0aSpA3ai7Duyq8BPDms7BQLVsZ8thvZpn5QOO4xGFyZ/pNgPEOgF5RS54OTB1DgCIMo3tIsnttEVvA0CmT3zXumCbp2cSLGXE2zMCAOlIHKgBZa4yECFMIcIC1kgZURUpGM2KB0+lLOXBNrn31P3sn4S1sTlgxbGD4R01/6Fq4pNQcM/pe3CicAJ7cnuEMjLRMGyEKsnVsKpHa69pmljnu8SpbPB1czE8UFJG6PWLQm05jM4PspyMFBQ+mVjyaZApM+f3Bj57MTBzATB7SduHxvl7XYNTGSHPSG9khObakI+syn9PrxHiTtg8Bc4bNy8fepUA+glFUTAfTeOpegGLpSUc9PuD5BchuOzma0YNdb5ZS2fngUzwXXyAVT6vSiXEml6EERAZEWUaRYXKN7LO94jmD1VUFUtRdk6OR9MdXXcJKABM6Lq372yQCMmIB3LJmCAjvaCubYkBRmNSJLdu6O5khN/kU45FLyHIiP1mX67ayzILMd7eGxQZkdp58dx3ii+7lmliSYzL3TUAskp/TjERVuYxubdm1LBaYaUiNpfGe8FNc+mcJvWW+OKdocF0JMvKWSR6ydoJJSQyQl6H8V3Azhvsvyi3E6sTrLVxJtmsjOzKL+F5FzwPPzz9Q3zt8NfwG8/6DWzwNtZJh+8lqUQA05uMlOtlVPnufnpsn+ff3i3I+yDvDomMpAOYs5JLs9enQPNvpJyYQr0KqNaC3BaxFPDrP/I1nyTOf4+zlKpxBah3ZYQ9v2Y2gEYdVUpbDpiMCANwCzJSVEzAHKyBFQDmY2OMjPDWc1/wQUbK0qYk5cjlCBJJrp5VTAMwGoAaEes2+35Q3TQKVLibjGU16xiPCuh0Mjcj1kZLE/wgEZZpPGBljfRWptHK1gU3JrUVah4x3oKMOBb0GK9DOss0TjKyHI2iunms+wN24qdf5O28FwDnvER82dXACiCb2wVFKhNlAhwAJsOWnOpCRtYqazBhIgoFU4bR2jNCwWdERjjRyvBShF0ZoWC0gkVGJNVFGFhTY8CrP2H/99Lfxxo3i7kpI9g6gdee9xoAwNcOf40NuuOlpknDsJERkfPhEfq2zgeRpQwD6fHgg7eFiU+qm4t6fQDR5jk+ZbigqswvIKHArx1nEFRL+OyAiYkyDWzlThpsl+ihrRewumY0GHyeCZ1Dwd48qVxYUBVXMmICKPG1ZJDKCADMc9VrUd/0/0NkXiW4kRF+/iUMA1Ep6C5opDgRYPNpeHidTqqg4qvdvBXcPCPOdTamxoTx/liM/b6ZDrsWSeXT20w+HxRCMuIBq723NzJSkaZTppKTiPEFTne5gQLWhN+0o1MgQcY6Ry2bRqxPIIIsfztPrB3q6ZhtePSf2cfnvN22oLt6RgBExnYha1iLeDbAAWAyiCCUXEZ5A8AiTetV4uxVaVEXF8Fn/IZT4kQrzdsw3QysplYU5TE5BIuUEbdYdwBCrbGTEaaWoFbCS3Y8E2PxMSyVl/DD0z/ERnkFADBpKja/Q0K01rqfR2si8CzY9FWCqJtLreZE5tI9hIIRstyHU1bVplCmgslDBDvcCfpBnBNLXVFYnDiHxrs3eiUjCZpzBBOoa5bpMeCdvC0Uy0lGiouoKIoYojjInBEAmOM3zaV6BzfBDpSRtGkC7aY59wDy91RUy69G6kKyR+UMsMhIXVGEqu6mXmV5qfRYjCsjmc6yhIiMtApOHCRCMuKBsRRFwvdWpqnSDtU0ocRTSBAZ8YhCJqk75dhdUi3b6fIXJQUlgr0RdsIubDrMXr2AlJ0dF4kvGaaBEj/Opl3V2C6MS+mV2YBjrgntyjTkF5nlnTKtyQh/Lr7jpvcgw9Uo2cBKKa1a1TK5yi3Y5BnJ63kYLnkyRBJsRrNYioWfAUgUl/Hz5/w8AOCrT34VG0QqHOWuJL8pVl2mONt+T8BzaQgpFzJCiZTpANJE5RJMSUpcBawOm1wfuoTiXBmpKbC1j5KSGe9x10s3sioMoC61gwbd2iuXaSqb9m8WFsWOO6JEAk9/bYd5noK7aHiPcmgCJyOlqXMYhXIxXQplzjCBZD/LNNLmhBMg6mpLBVCWdutEc/sara1HuTLSqVGdhq9qHhvjQSMkIx4IanIvkZEkFCCaQlz0druTEZp34BycFaf0Vkctm9pRM0oMexNM/lwoBjgsjxI+pXyGYq0Ikx9H09TU3C6biTXTp4XOXqZpNrCKwDOTn+ItyQhbuMv8xlriFyeN7aYbr6koqPJFsCqREVkZoTKNTNhk0LyYpjwAKqVsncBrz38tAOB7C9/DEZ45MqHa1YakmHHiHvq2xgPXphqNwNNXAevmWZGIJ5G5dABporFIDEl+rRQqdm9BnpO8XDp4gyJdZ86ZHTSgMqH0RkYS4vkBBNiB4YRVplFhupIRa8fdS4hbN5jn3rnT8LnRK68DlXUcj0bxwvEG/vvMlCsZoestZRp9VUaszYkqyjSkaCcDKEtH1IitgSGtRBBRmxWXLN8sCWWkw7EbYt6ZV0rvgBGSEQ8ElcJa5XJ9EhFAtZIzdY92KtpdOgdnkbFOayIjlr9hb4Z1TSx0YgxrByJNkoxMJZpEJNE8wVQalgcA2R7zJrxAOwXWTdNCGaGSUSvPCP/bygabPVPkagMpJvLOsaITGdlkT+uIB09EEuLxbqUa1zINYPlGNhdw4dSFuGT6EtSNOg4VmP9n0pFdkBSR1O7eo/U8IzHTUG0twUFBSNWmTEbYDTsIMgIAOZPdJIvy+dyoocBXrZzPgKdOEI9JZRqJjNCAyl4m9gJW10xVYV6OoLIpnCDFsq4oNuJM6auUMDxovwgAzI+zjrxFFTD9TO7lfpEfT85Dh4kHE4mWnpGMYdrWq6AhyIjkV6vwEmUQZAQAMtKakvUgwLRZ2uowCp6QoBEjLYaNDhIhGfFAUGSkQmSE1+dof+tVpxNSt+MmHueDw2qKg4zwWmU6ksDe8QMAgIV6QO5o07R2IJIy4trWS5CyRjwfEwCEycsxL4Zgpa/yY2mVM0JkhEvzZU5GKFgtokZEfbXKF8FKld7X5kuI1JG8biecDaMhWnWblREysTIS8drzXmv79qTjfGirjFAUfKQ/ZDApvA8SGTFoinEw3RlZvjssyDt7rcBKDwDGOqyR+wF1uNUUxVamEcpIj2QkwcmIpigwKhsiJC5oZSQdS4tOjKLkWxPpq0PIGCHMT7Hgv7KqolDyEXzGSzRPZScAUJBbK8+I0ecyDXXXSQZW2kQGYN4GgIzktct4+O6cre3dkhG9xeTvQSIkIx4YC2hYHsn6REZifIHw6u0u88FgznjoRJxc+HZJtcQvwGwkhX3TFwMAFhDQSGi9BJAS40JGmko0QHOZpk/mODKdlj2UEWsuDX//WpVp4pZZEnoJJV6uyUgdDtTdRN1OVU40nC3YgCWRO5WRTW0TDbMBBQomnX4HqUwDAK845xW28s+kg9SJ0DGP2Tzr3Pg6HXCXBoEi8KuSL4YmKKcDIiM5/toW5K4LvSjISF88I/x1rSsKDGnHSPHtvUzsBawyTVVRUa1YXpigyYiqqOImVpDjvil9lbf+D0MZSaUmkeNrxOrmU20eDUFGjnBvhBcZIXWCeUb6X6apSBsh8m4lAzBvA1YnHwBkPZ4z43jvOgk8A6z5NF6dnYNGSEY8MBZQa29VtOqyNz7BX3KvdqpKg3aX9hMtRp4RR3m3RNJkNIW9s1cAAE5FVNQ7nfvgBrrgFdWe89CKjIztspdp+nQzJDJSVxRXM7AIPNP5zbrFDTLDiVZZUQC9IG6qGR5JDkjx51zRIjLi5p4XwWe6nYyQqXQiMdHc/jdhV0bG4mO4fv/14ttTjteRFr2qc3Ai/S4RBR9g+J0EysWoSt1dZQqcCoiA0iC8gkTq6pVNkUrpev71CHmAmC55kTQzGDJCNzJNUWyddr0+rxvGuJpWkEvClL6aYufooDNGAACKgjFegiv4iYTnZORpk900dVWB5pLAWpbTV/uaM8KIY1WVlBF+Qw/KDJxRrefJeBDVnKN9uWNlhGcF6S0mnw8SIRnxQE4oIz2WaYiMcKab4LK+5tHWRlM8nSFIcb7w1hUFhjQYrsSl5Ewsi9nJcxA3TdQVBaeXH+rpuAFYaaOJnL2tt+be1gsASI5jXMrSy/Sp318uY5UcuyTTNC3PiMZf51bKSFRSWUqrIpkyLZGRFL9ZUNx0lRt73WrE1N6bd/iCPP0iQJMyAkAYWTOGgVSHZGSdn3fTmeCj4AEgyRf7qqKwQWyGgTInJumASEKOE4OiRDZLXPEB+tOSKntCZPVS59dlvMfpx0Q6qqpimdtNpmQEjSwn7BRQyP7D01e5j2gYyggAjFEJzleZ5imUFQWnpPJzwcWPZc2lMQDp2g0a5O+pKlb6Mw3NC0rhykjnWc7Dg5WV1lYFaFZb20CkAYfKyGhjLBWQgZVc1nwRiovebncHMxkC0w5mn5Buprp08y3xsk4mnoGqqNjDu0cWVh7p6bjZQfIpto5jIS+E60KmKKJMAfRHSgeYaVRM23X4bza1Tei8fDFbJTLSwsBKnSGKAhQWGSkBkJWOnbqbqNup2sI97+UZoU4adzLClZHikvAqXDV3FX534ln4k5U1KEn7DV7szlwGyQHAWoPJx1O8jTJoUEiXqJvXpYm9AalhJE/npYTIfIWRkZSJnsOl3CA/p1xL13jXUKLHna943xQFVfId9WkZFpHw8vXByUghwc75QWeMEITqJZWqXGEYwPpTOBqzt8w6ry3AoYz00TNiL9NwZYRvCpIBmbcz0mYr49EEkJVydiajaZuR3g9EqrfHhmbQCMmIB3IBeUYqvKZIixCRkZqbacgwUAaNpbdfTHEpd0GXFmenv2Gvyi6GE35qse2gFfCXk+N48aSKI1tWdolX4BlhLGWVBjJ9IiMAkOZ98iVHnDH5RaaSU4hR3b9VHLxQRhSgcFooI3JNVixAJMsKMtK8U/YKPvNs6wWA9DRACxlXRxRFwX9K7MbLypUmZYfk4KrUWkuoNWrI8/Noeny/69/cK8j0K6TqWkWEwKUDUsNyfAdalIh7oczUhFyfli5FURDjNqmaVErV+CYh0ePOl5SRuqKgxDuy3EzQQYB2znk5i4aUkdjgJ/bKyPHXoSAlVLuicBqolfFUwv66F1waAEgZSZnGAMs03DPC12Gn169byATE6z3KSurPVLzza44ydbSQjIw2rDj4HpWRBpERtrAmRNCMizJSK4uheimHzBiVPBu6FI8t/A184dnLx9AvFBd6Om52kAXcnkphXTHxhUNfEF9uWaYBMC6F72Q77H3vBJRS61RGhF9Env/SqrWXzLCKChQWhSchLf2MCDrikiYlnyZdZHs5+EyGa+AZQVFcSzXCt+MgU62UEZr0GzFNjPeJjIhuGjLx1cqSMhKQgZXKDFIOQoGPP8j1aeYRAFChxqaMcDIS77FVXfaGbHJjqZsJOghQQm3RrLGWXgAo8rk0PM13KJ4RWKWHvGyudcM621Q9nbVfMwWXbAy7MtLHBFa5m0YoI3X+vYDIiPS+OI2q4uvSRm+mi3WWyjSa0ds9LiiEZMQDsjJiOlJPO4Fo+eILa5xLaa693XpJJDI646EVVUXCoCh5SRnhi2SGLzx706zd8bhUW+8aWkEkNd5y5Bbxt7Q0sAIYy1lTYjN9CKYSz80X9rJjYaIo+DkybyoRoEWt366MLLKPsC8I1CFCrbQiV8BFtvfqpiHPiKfRjMjIshTnT7vzJjLCyYBbyitXYCYbBlTpvQgScrmBKSNV63WLBtTay5+nIO3sqTMk16eZRwBNM7WTET0gZUT2FGxxL0cQEeJuEJOPVdVq0SdlxGM0/aCQ47+34FJusYE6aVL2c6rgkjxcpjEOA2vtVYUyQuX1ZEDkTiYgWQ+VR15/p7sox1LpX/Mo9Q4aIRnxAHXT1BomtLqPYB4PCJc1P7ko9a7mFuOtF9kJDvf2SLFjk+TjEjcNUufH3jG2E16otbnI/UArskFbYCa4W4/dyj5vlTMCYG78IOKGidl6HbFU/8o0lO5achA7oYxQTTWeaTkoza6MnBaBUDIZIUNxxdAB0xTEzK1GPMuVoUNrh2yR8C0NrABwPu+eueevmSkUkEzEHmQEzWWadW4KnOpTFDwglYlUFYZe4soIe91SsWDq5jRQ7Z7aBlY4uabOkJxLeSwoxKjjTWoZp7DBeI8yvKpYwYebAUaIuyHLu8Rsk3upm4b/PUNTRviN1K3cYgMPPDui8uOlDiEYthwYAChzYpM21ZZKaK+gzUuVZtOYplAoUwGZt+VIhIwHGZHfu5lM55sOkQZsNq8hw0BIRjyQiUfF/auXSHiaakpkhEKVvJQR2l2mXG5ytIhppIw06qBLOcNvcHunLgQAnDC1nhQdADCrW2IHBQD/dPifALRXRsYnD+JzpxfxN6eXW3ax9ApKqS05iB15RmbJZ9PmGOi11lUFtcJpNnwP9h2+ICM8GK3qkQcDAM/f9XzkYjmcKp3CjxZ/JL5OZZqZpIcycuVbgNQUsPE08Ah7rS1lxOEZkQeuObC2+TQAYNowgT6RQfn8rOoFoFYR567bHI1ucN3ERbhE01BAAzf+6EYAQJ6rCV4dBkEgzkmVvGHQOOlPBHCTI+Vlo8EUvaBSO50gha5I82nqupg1JSb2DsvAyjcKbuUWG9YOowZggb9Wl89cDgDIu2SNVPi1ko4mfU9p7gZExGuKgrpetA08TAb0esoEJOvhwZLX307begErgE93UVeHgZCMeEBVFWQTvftGRC2R71ISFDRTd2mn0ouWZ8SNjPCPNfKb1ErWjZOXQ3bvuBSqaaKiWDvxblGpbqIhXdQ/XvwxFvILbT0jGNuFi/Ua9tXrLfM9ekWGv0ZlR2uaUEaITLS5gdjahIuLlhFT9ozwv5Um91K5xk2WTUaT+Llzfg4AG3ZHaNlNAzDC8bx3sc/v+AvWSSA8I45uGjGKvpmMrBeY52RSjfdtUZa9D9Xqlk0ZSQe0K43OX4EPra4japq49dit+M6x74iddK5PYwYAIO6ijNBMqCBkeDKsbpFq2icyQsolU0Y2ReAZ1BjyDZq/NCRlhEpIHoMeBdYO43gsijoMZGIZnDd5Pvs5VWmaT1PmcQlBnX9ekEtt1VqZef34mhHU9OWMZEj1agKQ37tO59IAVlyEBhNwMcIPGiEZaYEgUliJjNBJKiJ43ZIzq3nrpHYlI7yWTYukXhLKBXlGYhP7sLPOTqyFjd46ago8ujwCBdfuuhYAU0falWkwdQ4QTQJju21j74OGGJbneC0pCn6WgoNatPUCTK0iaX5VymSwlWmo/VdVGGmkVj6PndBrzn8NAOA7x76DLW0LdaMuouBbLhxXv4N1AqwcAh6/xVsZIaXGjYxwz8xUH2/YETUilLpqrT/KCPY+Bxe+6A/w1i120/l/7/g9nOLvz1gfb6KkjOh0ozQabLAdgHgAHgsKPtzk0n4yoAhxJ0Rrr8rm4BAZMXPzOMXHBcxn+lPGa3tsabaTL7Tq5GjUgI2jInn14NhB2wBApzJSpgTWPvtgEpGEyMGu1spAvWopI0F5RiQDbs4jWdXmGemCjNAGWfMYNjpohGSkBXqeT9Ooi4RK6o6JcYnPLfXOrG61VkZEYBo7cYxqXmRiiBpjahJ7+DyWhdWHuztujiJPEM2ocXFz/efD/yyMmZ7KSGoSePt3gLf8W0+/vx3IV1NyLGhCGSFfgQ/pNM2NxSsRPlodim33b8sW0IvCgZ70eA0umboEF05eCN3Q8W9H/g0b1Q2YMKEqKiYTLUonqQlGSADg9j/z9IxQQm8VaCrHbZSZItZNu18nSPEluaoVYMitvUHuTK99D955xf+DA3oNK40y/rXGvCP9SF8lxLihVCfFra6xwXmw5/10CzKsbvE5UymP2SO9gkowefKMcL/Iem4W5XoZChTsye7py+9uhxz3VRVamSc3jwNGHUeS7Hw6Z+Icy2viSkb4XK8+l54URUGS+3wq9RJraw94+nImvcP63INopKIpKPwa7DQKHrDiIjRFAfSQjIw0SBnpOhJeL4jumCRXLugG59bbrVfWRVnEbUEnZYQySsrSaHWxi1cU7FPY7zi+/mR3x81RoDyRSAIv2fsSTCQmsFxZFq2jLW8I85cBUwd7+v3tQDugsrSglWtlkTg5R2O4fdwcqU14Ncp+Jq1EbaPVBRnhCzuRTC8yoiiKIHD/dPifhF9kMjHpOg7chue9ix3z6Qcsad1xEyTzmakANce5tMlJ5ESf0m8JVG6o1MpiOjUQoDLCkXjx7+KPdr0MAGDy98QZhR0kKAtIJ19XvcoWbADxAG50NHRxo09D8gg2z0h1S3TSLKTZ1+cz8z1PIe762HiXVwEmG8jpBuqkyUwAAA6OHxQEy5WMcIU0qJybVqBZVdVaFaiVLWUkMDJieUCyHm3KqqLi6p1XY3d2N/aPdd7CL+LglVAZGXn0rIxoBdEdQ7J6q9S7ikQuWikjOu9tL/HHR0x7DX8v3xEvFI53d9wc5A3JRlOIR+L4+XN+3vb9fu5O/YBmypQkNziZV7OxLDLUkeJjN5vmr98KH8edcSzSZFqrKApQXLZ2Qi3yDF558JWIqTE8tv4Y7jx5JwCfRrPMDDOzynB6RqTBhRVHgN46V86m+hj8BAApTvaqtaLoZFDRhzkrioIrf+5jeH1qn/hSTkqfDBpxlZQRrl42dEFGAjGwcjKyyVW4oIarOWH3jEhkhKev7s3t7cvv9YNcbhc/NsUqRTpBM2l4+uq54+falZGq5RmpGTXRfp3u47lBoPEelUYFqFVYZw3c1+1uYM8Z8V6/Pn39p/Gvr/nXrkiQ2BiHZGT0MZbq0TOiFZsYczzmHTRT4YmMMaiu0b5xvvhrvJZd4gFQGSi2XfzeFJNAT5R9zH1oAYqRzvKLgWalAIACZWjmNwLFjpelYW2ikyY96+m3cH0uTjYsMmK/QVC7alVRgOKSWHySLerTE8kJvHTfSwEAf//o3wPooLZ7zbsBmRA5/oZYLI0o31FWHZ1ZG/z8mEz0Z0gegcoN1VrZynhAxHYuBgZFwfv+w//FHE9l3b3/RcH/Do44l+CtMo2ljPQaBw9Yu+pNTnqSkf50BtGNu6SqaFQ2LDISZcv+UMkIL9NoqgrNa1je2mEYAJ7mU8jPmTjH4RmxyIhMyNN9TH0mJLmSWq07lJGABuVNJCaQiCTYpqrFOqsq7vcKP7CTEZdE8AEjJCMtEIQy4mTMcb7w6C610gqXulMeJ5dTPi7x7ICMI0567xhbZBbk0etdoMAvcAooOn/yfNFal41l+zLcqxNQ6mxZcoOLAXnpWYvt+yEj/P1ZFWQk5fi+ZGAtLrEhWWi/E6JSzTonjr5b8MZ2Ac/8T+xzx9RkAEA0gQQnI84R4Jtcrp7ooo7cCZJCqi6hzEt66S4XRj/IJnL43Gu+jr98yV/ist3P69vvialERvgmRPKMBFHWoBRmndaGgHJZnJCVy2J1XaSvHgf7u4ZJRjLxLBR+/hYKp9wftHYYp6IRVM0GYmoMu7O7PT0jlL4aM82+ZhsRRM5Og49CUIItuaVjaXz6+k/j09d/umuy0Q42MuKlTg0QIRlpAREJ323OiFZoYswJvvDoLq1UZUFG3Fv9EiQfc5JAsy0yjtCkvRPnAQA2zZrofOkGRZ6lkZXm5NDNdazPJQA/oA6ikqoKZk+dNHPpOesC8yGt0w2BlBGnZ8cysKpAYcl3jfh5O5+HXRkrHbEjo9kL3svKM7OXNLfoRhJI8kTeiiSx1ho1FHgQ2pRkgusHkiqZ+CpiPlC6jzHtAPM5vGTfS/r6O+IUTMhJXV0voy6Ukd5LKomI/foOap6JE/FIXBCfYnXTUkYa7LoYJhlRFRVZ7oErFD3IyOZx0Umzf2w/ompUIiOKKxlJG/1NXyXQGIiyofelTAMAz557Ni7fcXlgz+cEEWs9VEZGH1YkfJfKiF5wKdNQPoQLGeG7y6yH1BcT8jFbJEu8Tp9R7ItbeuIAcryjZqXSfSx8gf8eOXTnVee8Cq8+79X4tWf8WtfPGxRS/LhKqnUxuZdpfHTTcCmUDKxOaVR4RoQy4o+MqIqKV5/3avH/jlrwJg8Av/Ej4C3/2vy9SAxJKtNIkdo070Q1TYxl+k1G2HlXqVfEzaBfnSGDBJERus50KSU0CGUk6XiNkgHF57shy8/PgmZ105zgBvR9Y/s8f24QGONl50LJpUxjmkBxGU9zMnLuxLkALLVHU1VofDMGWCMh0mZ/59IQiHRUGzrqegm1Fl2Qowp5aGO9h01rUAjJSAtY3TTdkRFTyg2hmxZ1QdRcUu/IMOo1Mjqu2mvZJX4CZZw3gLFdmG4wsrNeWe/q2NGooci9GDmpBpuMJvHH1/6x7QY7LGR4u2tZUUVJhsjIfGZeKtO033lmeJub8IzE7IZRsfgoClBctLqkfMiyrz7v1VYLXqd5AGO73FNUowkk+TlUlXaIVA6aMAyofV6UU1LdXGQ8DKk7I0gIMsJ9Xbq0UAeijDiUz2QfQ7ooHK5Q3QDKaygqCtY5eR2mMgJY84UKbuGMehGolXEkzh5zzvg5AFh5mDTCgkxGZGVkAKqtWA+MGjTJu9Kvzqh+QCbWuhaSkZGGNbm3uzKNzNyFZ0QoI81kpMR3YBmPxSkhFkmujAjy4rgAxnZjipeB1rodmCcNycum+muE7BYU116WlBGbZ6QTAytfwCi3JZ1wkJGYlDNSXBbKSMqH+XBndideeQ7rrHnGzDPaPt4XInGhjMgToKnterLRABL97XaiLpBqQ5PIyPZZjL1AIxt0PhFb49dl1ET7tmwfcHbPtDJB9woRfMbXgYU4e3+mklNDN6Dn+OtQcNswFdl1/BQ/XiIjqqIiy/ODCrrVTi6UEcMcUJmGe8iMOiqcrCoA4ttIGZSJtUZ5RkNESEZaoFcDa1UiIyTzU06B5tJbX+TG1KxHy6yljHAywi/AjPMGkNmB6QZ7/vX8sa6OHVpBDMnzmo0wbJCvoyRNz3T3jPggI3H7ApZxtAeKnZCqwiitdJwr8MfX/jF+8Ms/EObinqEoSJrsGCrSFOdNfs5NNPq/Q6SbaqWhW4FTZwAZoRuKUEY42bNyN3uD85xJJfpHRrKi+4Qd+0KOGaj35IYTdiYjx4l8nicT21BcggngSJyteQfHrcyiHJWedBfPyKDKNDTF26yjwm/kyX51kvUJETWCKD+ntbBMM9rI9RgHT0FQcahiR5XgREN3OWdp4Fsm7n4TIflYIzJSo/kSDiVFVTHFfSRrvE7cMSRlZNh5Il6gnZ2uKqhpRdQaNVGm6FgZcSxgNAWZILfsVWBC46+NX9k+qkYDrycnKQFV2tXQ3z81ALmabqpVQ0eFm53T26hm7gV6T2ucjJAyEhQZcZ4zXsF5QSDHz2NBRtLsd+3LDdcvAlhdeq4m++IS1iIqCgpTQw6MH7B+jiuiBYmEkzKSGlCZJsnXlApMNigSQErpXTUbNBKUXRV204w2xlO9KSMVXktMSq1ZpIzozuRBo4ESl4UzHkqEc8dWJPLiYoCb5ruO9W6zRvSimHvjOYNmyJCTPsvaBlYqKzBhIqbGMJmc7MgzknbcENKOBU3ezVJYFTBcwxqREU2afLrJlaGJAZRp5Lo5pV96lRi3E+KcLFD7vcZJfzygVvak45zpKxnhowcK/FpeiLM1ZNh+EcDa5BTcSgTFZdFJsye7x0bgcpTCKp33orXcNAHHRqIfoBk0FVVBlW8Akn1sa+8XKIBPq4VlmpGGrIw453/4QUUnMmIZ1uJCGVHYICiCpER4TWkkwxHFf5f5bjTj0i0yxQkExZB3DKlMM6rKSCwSQ4y/LeXqls0voiqqNdfFR5nGSSqc9XRVUUXL9YbkGwg8bbQDJGHFsRPWuTdg0jCbs0kChp2MsHMyNWQfQhCIiWGWzHeliTJNMMtlwlGmSXoooUEgl5Ai4QEs8D9hFMjIGKk2EqkQKCwKMkJ+EUKOv155KeyvzEs9g2rtTfE1saoobGo1+jd9uZ+g1m8tTGAdbZBnxDCBkt75iOUql3fldscELwdoiiJ8DuwLeeZ9AJD1YPa0Y9P4jo0GxLmVdab5IrQuzQzpCFreMrCOqDICWIFvZa0gOmnm0nPsmzT8yU+ZxrGjdzP30fu4zpURufw2DNBsmKpsYOWdCVNqvDmbJOjfT2TErKPM1bp+T0wdBOI0soG332t1UkaCea+diatBjZ13gy0SHsAC38CMAhnJJZkxvtCoNn+zuIyneCfNwQn7jCsaBVCQwv7K3CuVViJApP+kgLJhqooiKeDbx7xKSKghGdkWSMUiiHJ1oBvfSJXX4WT3PI0gNxQFdU2Sxqp5lpcB71kEQj7mJKTIbwAZl8VsiodrrdW6MyaZVYmM9HkKZi/I8KyCkp4X5tVZHjVteUZ8lGkc7dRupS96Hzd4+++wd0IUx65Ji/km3yFOqP03kgqp2myIYYX9npg6CMSFZ4R1vOn89U0EREYSDsUq2ceygpxYqgNYHIHAM0KOpxHnTZe1tbgkMkacysgYV44LigHUGSGh2UhpdTBKpUhgVRSR85Pchm3twofomG81DIRkpAUURempo6bC5UfZbxCPyr3dkmqh5YVHw5OMRKlMw3ZsZT4Yymm2BIBpfkNeb7Q4yZ6+HbjnU65TM6vVTZE6OaplGgBIcZmxpBdEmWYuPQcYhuQZ8RN6lm75fwBI8fdxk7Jjhk5GeAKqlJ64ztsdJwOenOv6+0VHQQNlnnni9NpsR9D8KJ2332u8HBAPSAVz5ook+0jg5MTSE7EoTDDiPZUcfrt+lofyFcx68xpUXMKJKDu/nRNpczxqgEXCsw1dhedkpAdUNqU1vaKqlgI+xJJtt0iQDzEkI6MP8o10Ewlf5TcJ2bAm96Hb2qmqeRSpldZD6o5zNq5xElLii6Wbx2Qqy0Z0l8xG0yA1gX/+DeAbvw0sPdL0rSLfYasIfiR8kMhwQlCuFe3pq/UKAL7A+TBVNikjbmUa/hgq06SGvBMSg+pkZYQrYZMDKJek+E20AkMMKzwTyEhMXGd2MpJQgiGfSencipkmon0sK9BaUlRVcXPfN7ZvJFpQx/gaVXCZGmsWl0UA4Y6UPUk4xw3+8rC8Mo0j6LNPimCNh1BEmbRf05f7iQRfw6qO+VbDQEhG2qAXZYRuEvLsiYgaEdNW5eQ+2TPirYzwWrbZAEwTRb6eZFxSPbPZnYjz2SXU7mmDYQD5k+xz+iihwMlIRomNxMLlBYuMlCxlJDNn+UUAf2TEj2eEP0aUaYa8+JB7n8o0pmliky+Mk300RYrfTyY+00CZnyKDGN/eb9B1VuNklhKPEwF1SySkcyvZuS++I8hlmuO87DEKJRoAyPHZSQVVBcrSGmU0UKisikGCOxwzluhvykvD8kTOyIA2TqKtXVFQ4edJaoQ3bV4gA77utWEdIEIy0gZWJHwXyghfxJwJi5RXUJN7u6tbwjPi5dFIRCX5uKGz5FEAmXQzGVHS01YKa8Wlo6a6CXDPCUrNKa1FXkLKjXgdlOLHS/WK3cBK80RiaUBtf5r7UUZoR7shMkaGG/Alpuby8yyv51HnCsXkAEhBkrcOV2GyZFoA6RENyOsElJIsyjSCjASkjEjXd2qAZGQhxq6VUQg8A6xOn4KqABUp+Ky8jlV+yebiuaaONbfJvSIBeEDdXEIZURUxJM/Zsr0dQMSbsquGiZCMtIEVCd+hMlLXBGN21oTjlA8hm0urW1Zrr8fgrDiREdOAqRVR4jeArEPGBACkp635NG7KiExA3MgILyF5De0bFVDiZ7FesXtGOgg8A3x6Rjip3OTKSL9Gv/uFGFTHFxIakpc2DNG11dffz9WXiqoIYpweQMZDvyGuM/5/SjxOBETMbcpIQEFqXrCRkSQ7p0dGGZGH3pWWrW8Ul0SJZoabXGWMiVRZqUzD1cFBlQnlWVUijXkbZuzQhopGjAwTIRlpg65TWDVpYq8XGZGUkUZ1SwzVy3jcQGOi5dBApbwCU+xGXS7A9BSm+ORe16yRonTxl5oHVVG6YXbEpUeKwj9VL6Bm1KBAwUx6xirT+Fwg4mocEX45xJUoYi67YKtMQ8MPh/vaUCshKSMbVZpLM6CsBX7eVRVFmK+d4XHbEZYywsupvHstHpgyYr1GyT4vwaSy1hQFT8XYDX4U0lcBPvSOK0OF4qL1jeISVj38IoCHMsJvps6xDv2C1U2josLL66ntSEZiVlbQsBGSkTYY4ymsW50aWLW8kO+cJ6lbBG9JGhblZWClHZUOE6UyIxiqabqngMrKiHyhi1/YRhnhsmduxEOsKH78SJ2Rp6nkFCMSFHjms1NBURQh8WY8bqj0Oq/zrorkkF8b8qzQzVKQEaP/6asAkOSloIqioEzEeBsuyE7EiIwoAEwTmkHKSDAeoYT03iQDSnX1Aptyy96bUwrbnIyKMqIqKrL87y+UZDKyLMiImzJiZacoljLCr4FBKXNWN41UphnxtdINovTfCMnIyGP/FFuYnlruMC5XK3pOdo3zbAxdjjPmEntMUW2jnW0/J1oOTRS5DyRjKu4G00QOU3ww8FrxVPP3fZKRUQ48A4AMVyeeNtjxzmV44FkHUfAEKrt43VCJjBRIGfFZAuoXKNegQmREk5SRQYxR56UgU1EslW7ElTQ/IGWkxlOShTISEBmRE1f7Pc9EVVSb/ymqRq1QwBFAjncoFcqSOltcwkq0vTKiqSo07jWp8A7D9IBalmktqCsK6wZCf2P9+wUxRd6tvXrA6IqMfOITn8DBgweRTCZx5ZVX4o477mj5eE3T8MEPfhD79+9HIpHAueeei5tuuqmrAx40Lt3NFtyHT+bbPNIBrSBMfc4pnZRXoEn5EEXO8DMtQnviQhkBytwHkvaqOSsKprmUuFZebv6+XKYpNpORAk9qzA5I9uwWFLK1ztMymwPP/BMGupF6dTM5FahkbLiLDykjJLGSN2iy0RjMsDAXz4zfKcajDHGdKQrQ0KBxo3dwyoj13iQHMFxNzgnak90z1NRgJ3L8NS3IvjaprbeVMsJ+bgMNo4EKtZanBktGAMtD1s+8mH4hQWREATDkjpqOe9W+9KUv4b3vfS8+8YlP4Nprr8WnPvUpvOIVr8Cjjz6Kffvca5Gvf/3rsbS0hM9+9rM477zzsLy8jHq9u+Fzg8bFO8egKMBivorVooaZrM8FSSsID4hzgaZ5ALrUW1/UC0AUyLRYzGUyUuJKSqbFYjYVzQKoYr3S7Alpq4wYOoCEmPo5qnC6560oeOqm8U9GiIT4JSNDN7CSE57fLDd5JPakYQykTBNTY4iapgjHS5lsJ77dIYZZKgpQ19lgShVIRIMhI2o0jrhpQleUgQxXy8azAL8cRqVEQ8hFU0A9jwI/dwHwib2cjKSbyUhEjSCrxFA0ayhUNpCUArvSLo/vB2JqDCoAA1Lu0DbM2ElwYqcpClCr9H2eVSt0vHJ8+MMfxtve9ja8/e1vx8UXX4yPfOQj2Lt3Lz75yU+6Pv6b3/wmvv/97+OWW27By172Mhw4cABXX301rrnmmp4PfhDIJqI4OMNuTo+c6kAdkQ2sEacywsmIFFZV9mEYjUu17CIfzpRVvBczmk+z5jafhhOQBgCUV1nuCME0WSoigKzH0L5RgdPf0URGOijTUHmmXZmGMGwVQJjo+HtllWkG4xkB7DkZ6TOk6ityRhQFZr0qQgbjAbZuJrgkngwoSK0VcpKCN3JkhPwfcuaS1E3jVqYBLEWlqG+hzMvdEdNEfEDKiKIoIgF5U93Oygg7pzVFsdbMIaGj1UPXddx333244YYbbF+/4YYbcNddd7n+zNe//nVcddVV+NM//VPs3r0bF1xwAX7zN38TlYp3/Kymacjn87Z/w8Slu6hU08HQOd0iI86bWJwmJUqMvsgvKK9dOWAZ3xqKggInGOkWOyuKfF53m09TXMZv75jG9Xt3YQsNljtCqFVQpCj4AV3c3cI5l6fZM9JFmcajtdpJPpwkc9CgXAOql5OBdWpA3TSAvTU1cwaoIgBsnq1arcxCBhEs+SQSlxrASAG5TLNvbDQ6aQiiM0YeYd+mTAMAOX7tFfSCFXhmmFBSE/07WAdoFMUmechGPAbBDSL0jJSRIaKj1WN1dRWNRgNzc3YD1NzcHBYXXTo2ABw5cgR33nknHn74YfzTP/0TPvKRj+DLX/4yfv3Xf93z99x4440YHx8X//buHS6bv2wXW9gfOdUBGZGVEWeZhsc/y6l3Jf55poUPISYRlQ3ertvKY0LzaTYaFTQM+9Rho7SM76TTWIlG8Vg8bm/v1QrWkDyXdNdRQsYRsmV5Rvji1kGZpmPPyLCVEX68dZioG3VBRiaMwRhYATsZSXde9R1JyGRE10uSMhKcOZeCDwcxUkAOURw1ZWSMJvBKZv5qcUmYxD3JCL9G87UiyjQbxhzceQ9YCch1j3V+O4DOdVam2UbKCMHZvWGapmdkuGEYUBQFn//853H11Vfj537u5/DhD38YN998s6c68oEPfABbW1vi38LCQjeHGRgu68bEqhVQUb0MrPwEoHkARoN7NIBsi4spLi0q6+QZaWGqm8jMQzFNmLACsQiL1XURt7wcidh9I1pBjBzPjrhDPOUgI1aZpgtlJNYhGRm6MmL9fq2hYcNmYB1QmUZaQlqpdNsJcsaMXi+JJFbntN1ekDT52jCAKbNymWZU0lcJOa7eFqhkXdewVmdKblyNi4Czpp+j8k69LDoR04Y5MEUQAFKqnUhux04yWsO07aaMzMzMIBKJNKkgy8vLTWoJYefOndi9ezfGx62bxsUXXwzTNHHixAnXn0kkEhgbG7P9GyYu5crI8fWy/7wRqbXXedOiJEdKdoRWEFHwbhN4CdFYBhFea97gCamZFnXsaGaGmRnhCD7TijgKy0C8EnWSEWto3yhP7AWahwT24hnZmWGDu3Zld7l+f9SUkYTkbanUK47W3sG8b3JramoA/odBQFVURHkZRdfL0Pg1F6QykhS76QGQEX4NK1CwJztiZIQbTvOGzlpLZb9IesZzkztGUfL1Kso8cyltmsAA1yunqjXs9aAb2JQReZ7XENARGYnH47jyyitx66232r5+6623ehpSr732Wpw6dQrFolUTfOKJJ6CqKvbsGa0LwwsT6Th2T7Ab0aN+TaxaAVVK5nN6RqhOR8qIPCSvVSutqiLOF8ZNLmu2TEhNT2PKLRK+tIJjMWsXu+KijIgyzajnjEhlpFwsZ5lPSXLswFT2ny/9z/jrl/01Xn/h612/P2pkRIkmkOJkM6/nxXyOSVMFBnRscmtquo/TZwcNus3U6mVoPIk1EWCoVQKD8xkQGZnPzHtmGA0LOV5WLShgPi8p8Gzao0QDSJN7DQ1lnrmUhuprDlVQcA7KHPZ60A3snpFtREYA4P3vfz8+85nP4KabbsKhQ4fwvve9D8ePH8ev/dqvAWAllje/+c3i8b/yK7+C6elpvPWtb8Wjjz6K22+/Hb/1W7+FX/3VX0UqtX0GC122u0PfiJzA6kFGKEwJVUuJ8IqCFz/LP67zHJCWg6HkSHh5WF5pBcdi1o3DWaYxpTLNqCsjcuLirLx46dKgPJ9IRpO4dve1nnkSTa29kSGfv9GE6MpY5Cm7UdNELp4FBjRpOSl1c6W34Qh1L9B1ptXK0HhmeSLAkLsMaA5V/8+hcX7jHpUYeBmkjBRUlQ3LaxMFL36OK6IFs4YK90qlW3QW9gNNZGQbGljtnpHhkpGO3703vOENWFtbw4c+9CGcPn0al112GW655Rbs378fAHD69GkcP35cPD6bzeLWW2/Fu9/9blx11VWYnp7G61//evzJn/xJcH/FAHDprnF865El/+29etHbwEozZmhqrpYXsz3aKRFxLh9v8HbOTKudv9ewvNIKjsrKiKNMo1U3hClr1JWReGJcZF3M2chI556Rdhg1ZQSRBJKcjJwqsZTdiYYBZUAlGoCb+Lg3Or0NF2MvsPlRJvRaGRp9LUBl5C0Yx0ThKF583gWBPacXXrL3Jbj3nHvx6vNe3fff1SlEuUUiI+06aQAgl2SKaB4mymUeADlgz1LT5mQbTu0lAjUKykhX79673vUuvOtd73L93s0339z0tYsuuqiptLPdQMqI3/bempZHPeZhYKV5AJIy4puMcBf+hmIAUFsaXlmZxkUZKS7jqE0ZidrICEXNKxj9WSNKJIq0YSIfUTDLnfkApDLNGUxGonEkDUZGTpdOA+BzaQYUiQ1wEx+RkWG/HgGCRVo1oGt5NqMGQCLAHInnRMfwnNV1wGHA7gcmk5P4n9f9z77/nm5gmzNTXmdlmqgPMiIpKmWeMD1oZS7pWA+CSugdJOiYq+o284yczbiMZ408tVJERW+0eTRQ0axsj6bZNI7kTNkw2u7mT2SkKKaktljMZGWktCS+rBcXcSpq1fpXIhGYEhkp8KF9WUS2RaImvWJz8Qnri13EwbdDU87IsG++kbhQRhb5oLFBzaUhpCSfSHob7gy9QNdZTc8zCRtAPMiS5cwF9o9nKWwTeB3KSKsyjWgJVlWU+Xo1aDIib07iUEYqZt8vqEyz7XJGzmbMjiUxk03AMIFDi+1LNVUeShZRVEQd8mGCm04pTAnVLWFg9auMEFp13yCWxhR/i9fKFhlZyB+HqSiiE0JXFeRlZYTahgcQyBQEMrxNclZ+7YjlB6jsJCIJMQEVGIEasVym4cMQB9nWC9hbU9MeYXHbEXGarK0V2UKNYD0jePmNwH99EDhwbXDPuQ1BZKSqqtBLKzYD6450C8+IRGLKfK0ddJkkJZXtBjFjqB8gZWTb5oycrRAm1nalGqOBKi91JCPJpvY0MX1XIiOkdLRKYAWAhEOpaBnXriiY5jcIeT7NUe4vOJiYwgTPIFjWLE8JRTNnt4nsOGuy1+RgQgpo64MyoiiKbcEbujISjSNpsjKcVaYZXFsvYG9NTQ95inGQiBNRrxWFMpII8v2ORIHJ/cE93zaFbehdeZkbWFsHngEyGVFQpiGjAXp6/CBpIyPbM2PHbmANlZFtg8t2+Qw/K62C3lZnXRGwjHCU7Agtj7LqzzAac7xl7aZUTvFW4TUp7v0YzxzZn5oTu4+VRgWoM6tekWeY5Ia98/eJP6xn8GfLq7gqzfNBTLMvnhHAvvsaCWWEe0aWuPI16DKN/BqkR9zs3AlinIyUtDxMIiPbhJxvJ7Chd+xGXiivoCEPyfNFRlQrDn6IZCS1TckIXb91RUFDK7Z5dH8RkpEOQOFnD7dr7y2cFm29brvnhJi+y1tjqnnfykjcoYxk2kypnJbm05hc0j9WZyfdwbH9mM3MA6D2XqaeFPmNvGWGyQhhd24vfrZUhvKdD7G6c0MHyI8TsAFXfj9HQxlh72md/72DLtOkpNcgvQ0HhXmBlJGiJF2HZKQ/yPFSX6G8ho3yCgxFgQJFzNZy/RmpvLNFMQetMpr6AHkwXmqbZuzIuTNaWKbZPqBY+CeWCtDrhvcDi0ueQ/IAK9ZdgwkYBvTqJmqKv5yRhKM2mW1RVwWAKf59zayjxE+2oya7ePdPno8dPHRIDj6jORHZAe80usYNfwJkdgCLDwKf+0WgaPlj+qWMJCKJ4Zt7JQMrYfBlGuv8Tg34ZtBPuJGR2DbxUG035Pg5VCiewipPhp5KTjZ57WTICvISL+ukB6gIOn9fUh2tMDm/sM1hCsnI9sGeyRTGklHUGiaeWHKZhEsonPaMggesFkFdUYB6BUWphOI1MZYQc8iBqTaD7FLpHUjzlM716jpQ13Eswo5t/8wlwrG+HJWUkQYrMo164JnAjguAN30NSE0CJ+8FvvAG9vVIHAh4x0LzJ4auigA2AythstEAkv1vFyUkJfVs0DeDfiLBb4SFBiPmCbN5JleIYCCG3uVPSFHwsy1/JqJGkAV77BLvDEzLrf0DQFJaH5MjlmzrF1E1iiinAdp2S2A9m6EoCi7lvpGWSayFRVR42SXlMlwrLsgIgFoVJZ15UFJqvG17WFz6ftowobZrJ5Mi4deqayhsHcMav3j3T18qptzKykiRz8zJbqed7vxlwBu/yvwSy4+yr/XBUEnKyND9IoAtZ4QwyLk0AJCMy2RkYmC/t9+ICTLCVMTteavZHsjxdaZQK0hR8O2nhee4UkVrbTrVwszfByRHyT/WAxL8dQzJyDaDFQvfwsRaWGypjIicEa6MlGjonY8TOiEpI1mzxQMJ6WlM8+Cz9co6ji8/BACYMUxkk2OWgVUu05gsjC03gECmQLH72cB/+jJA5aU+lJlIERmJtEU3ZcQYtGfEksszA74Z9BNxPvSPiHkCoSrSL4zRnBlV9RUFT3B2+6V9EJggIW80U6OglHaJBFePtXp1qMcRkpEOQb6RlkmshUVUPKLgAedwoqplGPVhtpTr1mk/C2R6yqaMHN14AgCw32SkZjbFlBFWpmFJhkUeNZ9t06kzktj3XOBXvsRKFfvdhzf2AqGMjMLiI7X2EiYG3U0jqU/pASa/9htxfp0V+LWQCJfKviHH15mCqvoKPBM/5wiTTLfxzwUNeaOZHPacqh5A57reGG5r7/bsRxoiqKPm0dN5NAwTEdWFEBQXPYfkAc2TEllrWg4ZH62RCYmMZP0E7UgprGvVNazkjwIADkTYTYSUkdVIBEZxBWqjjiLv8skOeKcRGA5eB/zmk8wzEjBkA+vQ4VBGcoaJGDBYZUQq5aVaZd5sM8T5brGIBoBIUxdbiOCQS1nR7usdlGnGYmlA2synBkxGbG3+LuX47QIRfFbX2jyyvwivsA5xcCaLVCyCas3A06sefdltyjQxksU4GSny1rSsD8NoXCIjGT+97Y75NMd4ONb+BLtxTKemoYD1mW+UFgG9wOZEwFoktiWiib5Mrh0pZSRi94xMcqPyQD0j/HclDQORxJnT2ksKZJGfQklsz4TN7YAcJx75TpURx+YtnW1teg0aNjIyPnoTkf1CzKdphGRkWyGiKrhgni3Ajy26dNQYDaC4hArfSbUt0xSXrTkzPuT1uDwLxE+roWNy71EKPEvPAWCL7hRPYV2prABaURxP9gzyAAQF4RkZBVk2EoV8dk3WebbKAMs0+8b24yWlMn6lUAa24WwOL8T5NZrn10L8DPrbRg1ygBmlr7aKgrd+zjrPFdNEcsAeN3ltT22XzkMXiCnyIRnZfrhkJzvxDp12MbGWVgHTQFVtT0YaigI9fxolSl/10b0Sl8oDWdVHqSA9hSm+Y14rLeNYnRGoAzmLyc/yWv+ytgloBYuMnEGJmkFhpJQR2GOoJznpRHJwZCQyvhd/uVHB+2K7BvY7BwHKX6BrwZnvEyI4jFE3jWRgbZW+SpAN9ikT7TsLA8ZIBSD2gDhX7zWzATRqQzuO0DPSBS6aZxfPY6ddlJECK4NUeQ3RzTMim1A/9c27oEz4S18F7CE16agPMhJLYZo5CfDk5pMow4Bqmtgzca54yI70LA4VjmGlVgS0PJugiW2UMzJAnMtft4PjB4d8JAw2MmIYzCfj57wI7ADGgPfcH3jS7bBB11mZlJFtGve9HUDrzOloRGzifJERyaPky8wfMGR1dDuTkQTNSlOZbQCR4XRRhldYF7ioVZmmwEa5V+MpALp7a69EKFLGOta4MuKHjMjGyYzPUsF0nLfO8WTVXfU64rl58f0d2d3A0o+xrJrQNo+KNNjsGRTvHRRetu9l+LfX/Bv25PYM+1AAAEkppXJiwBkjAtK5dKYg5jAoJ1qkgYboDURGlqPsNc5G075a58ekbr/0EET+WCSGqBJF3ayPRqt/l0jwY68qCpt2PsDQRBlhmaYLXLSTKSMnNyvYKjtkrSIjI5UoIxxujJml3jFMqJtWWcTHzV9eJP20AgPAdHLC9v/9tTqQscxes9mdAFjWSGHtCfH1dmmwZyMURcG+sX3Dj4LnSCqWyjbVaAzUL3ImI+4gI/EwCr5vcCqwM23SV8XPSSbX9JDIIq3v21oZcXR3DgujsaJuM4ynYtg9wdjkY4sO3whXRsr84vBizHH+0k+oeZR8DskDgLg8st1nqFcuNY2o1AJ6oFYHJOe5HHxWXD/CjgVq2zTYEMNHUjI0D3ouzZmMuOPmkgjJSN/QREZ8dvHlMhIZUYbz/oyUob1LkFKvhWRke8KzVMPJyMkKMxPmy+61TJoHkFULKFFZxE/OiERu/OSSAICamRHBZwCwv1YD0lYfvxx8Vtw6zo4lrJFvC8gDuiZDZSQwNJGRbTp7ZDvAuQnz09YLODwjQ3p/RO7QIH1aAUPkjCgKUBte8FlIRrrExbxU46WMFHjkw8mNuuvPR03ezquUrNZeH2WXuORByfptZZOyRgBgv5qyDZCTlZFC4SQAa+5DiNGGPKBrsmEMtJPmTIaTjMS36VTW7YCoGkVGen1n0j6VEUlRSQ/Js/HKc16JCyYvwKXTlw7l9wcBW5lGH97k3pCMdImLeHvvo86OGt5No6msLLKy5T5AJsLJiKEaVmuvD6UjJiX9+Z6SKmWNAMCBhD0/hIblrUUi2OJD+3y1DYcYOpKSt2FiwHNpzmTEHDe35Cgk7p7ByElDFv0qI2NSFEJ6SGWSX3/mr+Mrv/CVbd15KELPQmVke4Lae59YLKAhT04tLgEAaipTRJY8yIjCyYiuKB3letjKNH6npKanhDKSMAzMOwxik4lJRKDAUBQci5Gjffsass4mpKJJTDUayJnAjmF105yBiDvIiNPQGiJY5KQODr+eEbm840dVDuGO0MC6zXFwJoNEVEWl1sDxdf4GGg2YxSVUFAVGlA1NOL3mwdgN9tJrioKy0omB1brofA+yk5SRffU61IydjETUCKb58x6JsfJMLhpe3NsBaiSOL55axJdWCmxOTUhGAkHCcS0mQnLeV+Ri1nnrl4xE1SgyPF8kvY2ViWFDGFjnrwB2XDi04wjJSJeIqAouJBMrJbGWVqGYBk5GmLpgNhI4ttKAYdjVEdM0YdRZp4qmKCjyMo2ut/dpxOMWScj4nZKansYOTkacnTSE2fgEAOBpTkbC9NVtgkgCu+oN7C1tsP+HBtZAEI85PCPb2KC4HSCXXPyWaQCrvJOeuyzoQzprQB1B+uxFwPzlQzuOkIz0AOqoEbHw3C/yWJRdWGZ9EtWaiVNb9jrcSlGDaTIysqWqMHk3zdMr7mZXGbIyksn4vGjT03hlsYRf3irgnZtbQKZ557GDD6t6mso04U5jeyDqMFaGykggiEWdysj2bd3cDpA9F37m0oif42bXtF+VOEQTSBmpNqptHtlfhGSkB1BHzSFq7+WdNIcjTFVIKewG/9SK3aF8fK0MGIyM0MhsmAoeP9V+UFE8LtVJfbrOkZ7GtGHg99Y3cKFeswWeEWYzLEWT4phz4Q57e8DpZQjft0AgX2dAWKbpN4iMxNSYTSVp+3O8vJMOy8pdQ3hGGvpQjyMkIz1AzKjh7b0GJyPH+MI1GWeTcZ9aLtp+7thaGabJyiHrfEqlasTw4EmXwXsOzKbncaGm44XlCiJ+W3uduwaXMs0OR7x51q85NsRw4VRGwtbeQBB3eEbi4c2uryAyMpOagaL4nzPz4r0vxmRiEs+YfUa/Du2Mh/CMDHlqb5hs1QMu5u29C+sVFKo1VBePYQeARV7q2JXdhScBPLXiJCMliYxwZcSI46HFzba/M5rdgX9cXIOSHG++EXn+UJztmDVOdlzKO7Nj+2z/z/r1o4QYLpxhT2GZJhDEHd0ZibBbo68gMtKJXwQA3nrZW/GWS9/SEYEJYQfNTwuVkVHH8buBT7+EfXRgIh3HznH2Rj6+WMD66WMAgHySXRjnTDK14YijTHNsvQzTYGRkgysjppHEUl7DUr5N3S45DuU/fx1489c6+ztSUraICxnZkZmz/T+Xmm56TIgRRFOZJiQjQSDmMHA7yUmIYLEny9bKA+MHOv7ZkIj0htAzMgJ4bP0x/MtT/4Jj+WOej6nddzN+vP4Iaj/5nOv3hYl1sYDKxikAQCHGjKiXzh4A4KaMlNEwmXqywee/xFW2+D10Yqv9ge+/pnPXsxT/7kpGHDuSbAcmshBDRGhg7QviMcfU3pCM9BUv3vtifPJln8RvP+e3h30oZx1Cz8gI4OMPfBy/d+fv4Yenfuj5mL9fux+/unMOn9t80PX7NMH30VNbiJWXUAOwBdY98+zdBwEAywUN+ao13ff4ehmGyU4A8oxk+U3kwRObPf1NniAyEs8C8eaFddYRhJYLycj2QJMyMpzx32canPHvSR8TtUN0j4gawQt2vwDj4fk7cIjZNEP2jJzVZGRfjvkkFgoL7g9o1PFIjeU3HNY2XB9CHTW3PLSIaXMDK9EIDJiIqTEcmJjHjhx7o6lUk6/WsF7S0TDZYkeekek0e54HT/pQRroBkRGPduCJxASisOTObGiE3B5oUkbCm2YQiKgRRKRJ1/EwdyfEGYpRMbCe1WRkb24vAOB44bj7A9aPYIErF8tGFWjUmh5yMS/TFCoadmATp6Ks/DKfmYeqqDh3B3PlH+GlmuNrLK01xmcp1Hm9cz43AYCVaUzTPUK+JxAZcemkAVjddVaVhvCFi+/2gKyMRBJAGM4VGGSal4i3T0cOEWI7ggysIRkZIoiMnCiccP2+ufgQFnhnzEpEBTabScvBmQziURXT2EJEMXEyyoypuzK7AADn7mA3dfKNHONkJJOy3+x3j00gqipYK+k4tdUHI1Gad8e0CErbIRGQkIxsE8jdNKGaFShipqUUxsMQwBBnKEJlZARAZZoThRMwTKPp+5unfyKG2C1HosDG0abHRCMqLpjLYlbZBACcSrFFa2d2JwDgHCIjy6xMc2ydfRxP228cY4msiJd/cGGzh7/KAxe+AthxEXD56zwfMsun+aZNBRFurA0x4pDLNKF5NVDYlZHwtQ1xZoI8I3WjjobRaPPo/uGsJiPz2XlElAiqjSpWyitN319YtkyrhYiKytoTrs9z0fwY5hTmKVlMMvJhKSNM3hXKyCpTRqaz9sUtG8viij0TAPrkG5m7FPj1e4BLX+35kB2T57JjCSeUbh/I71VIRgJFXPJQhd00Ic5UJKQ1RDeG11FzVpORmBrDzgxTMNxMrAtbT9v+v+pBRp61b0IoI6d5WYeUESrTHFsro94whDIym7OXQTLxDK7Yw5zkvtp7+4DZaTaxMZfbPZTfH6ILyB6RMAo+UMhkJO4MlwsR4gyBfG5r9eGVas5qMgJYvpEmMlLdwvHapu1Ly5t2ckJ4w1V78eZL2U3htMLMp6SM7J5IIRFVoTcMnNioCAPrfK5ZGbl8NyMjD5/qk4m1Dai9NxdK0tsH8k0yJCOBIq6w5TFimoiqYVh1iDMTUTWKqMLO72H6Rs5uMqIVsa/OamRNZGT5EE5E7QvQSvGU69NEIyouyZVgAjhtsjeTlBFVVXBwhpVqHj2dx2mesLpr3H7Dz8QyOH8ui4iqYLNcw1J+8CfFC/e8ENftvg5vvOSNA//dIbpENCzT9AtERkJNJMSZDlJHhhl8dnaTkb9/DfY+9i0ALmRk6WHRSZPgAUjLlVXAS7EoLGFNVaGZDShQMJ+eF986d5aVZL7/+ApME8gmopjO2GvQmVgGiWhEeEwOnW4/NC9ojCfG8YmXfQIvP/Dygf/uEF0iErM+D8lIoIiDmbiT5v/f3pmHRVXvf/x9Zp9hYJAdZBcQEMQF97Xc0uqa1s1sMW/dyltWttm+aKVlaZa/tLrtdW/mTet2WyzcUVxR3FBAZJV935eZOb8/zpzDDMzADA4M4Of1PPM8Mmf7zgE5bz7L+0N248TApi8Yn13bYiR8Fvy1nHV7B6+R4vPINbTpxnrEAABKoQPqy8yfq7YQhYZIiqfKE1KjhwRfN7I3rQQAEOimMikaAtpaaflJwBeKel+MEP0Q458jau21K1LGMKoBJEaIgY3cEGFt1pMYcQxD5yGwlRMjeTWmkZGG4rMol3C/jEb5xAMASiRis+29AIDaIhQa9ueLYnn4aEdJLfeNDnLvKEacDCPLeUfXC4W13flExLUGpWl6DD5NIycxQgxwhMgIFbA6CO9h8DcIh9rWWlQ3G7pYWBZ55RkAAI1UjSEaruW1VCwGKs0Usep1QH2JEBnhi1d5+MgIT6C7qkN1Pi9GIn25B8pFB6RpiH6ImHxGegqZoWiVFyUEMVDpC8Zn1/b/MoaBcuiN8ORTNTWGVE1VLvLBfVMCXYLhaRgaVyoWAxVmxEh9KcDqUdCurZeHL2DlCXZ3shgZiTZERi6X1aOp1XEGNEQ/wSQyQkPG7InM0GEgJzFCDHDkIsdP7qX/ZZHzEKDlUzU53HvF55FrEBb+LgFCy6vFNE1tEQCgQM4JivaRESe5BH6atrkvQW4qk6mgMpFMUKZeznIMUkmh07O4VFJndsm7LxTj6OVyGz8oMSChAtYeQ2qIjMhBbsTEwEaoGaHIiAMJnIgAPXcb8vIPc+8Vn0eeIeUS4BwATyUXGWkQiVBfmdnxHAYxUijlHgztIyNAW0cN0DFNozYaT84wTFsRq5lUTW55A/7+9Qnc+8Ux1Ddrrf6YxACFHFh7jLY0DYkRYmDjrnCHt8rboWNASIyIJQhw5WpCcotOce8ZtfUGOgdCJVVBbZiyW1LTcVgeaq4AAApEpoZnxoQaUjVSMQNfjdIkTaOSmLb5dlbEuj+Daw9uatXj4CULnT3EtYOEuml6CpmI++NCToZnxABn/fT12PXXXZgROMNhayAxAiDQfzwAIL8uj/MRKT6PPENbL+/QKtSNNFUArY2mJ8g9jDqGQS24YXt+6o5ihI+MBLipIBYxFiMjgFERq5n23oMZbTN09l4ssf5DEgMTKmDtMfhUKl87QhBEz0FiBEDAEM7kKw9aoDAFrRWZQpsuL0a8nDgTsxJxu7oRvQ64tAsFhrSORq6BysxQrSnhnnCWSzA7mjuPsRjhi1d5on3b0jTGtvBanR5JmW21InsuljjENp7oQ0gUAF9gqaACVnsideFmNCmcPB28EoIY+HRLjGzevBkhISFQKBQYPXo0EhMTLe67b98+MAzT4XXx4sVuL9reBLhHAgBKJRI0HHgHBWIR9AwDpUQBD6UHAMDTUMRa2r6I9cpJoLEShUpOQJhL0QBcR03Kq7Px3FzuWhKRBGJDLpo3POMJ81JDxACVDa2CNwnATfOtbdLCRSGBSiZGSW0zzhf0bguwXs/i7Z0XsfNcYa9el7CARAbMWQPMfA1QDnL0agYULj5xAAC17wjHLoQgrgFsjj9+//33WLFiBTZv3oxJkybh448/xty5c5GamorAwECLx6WlpcHFpS2n7enZd/7a0Mg1cBHLUaNrRn7mnyjhO2mcA8AwnOERn6Ypad/eeykBAFDgMxRoze9geGaMWGRqniQTy9CobewQSVFIxQj1VONSSR0uFNbA24XrxDmYwdWITArzgFbPIiG1GHsuliBmcO/9RZyUWY4t+zLhLJdgRpQ3pGIKrjmc8f9w9AoGJH8J+wvqWutwS9gtjl4KQQx4bH6SbNiwAffffz/+/ve/IyoqChs3bkRAQAC2bNnS6XFeXl7w8fERXmJx36pQD3AOAgDkSSXI5Ttp1AHCdi+lITLSPk2TwYmRQg0XETHXSWMJPlXTPjICmC9i5cXI5HAPXB/JrWdPL9eN8B0+tc1anMyp7NVrE0Rv4qZww/KRy+Hv7O/opRDEgMcmMdLS0oLk5GTMnj3b5P3Zs2cjKSmp02NHjhwJX19fzJgxA3v37u103+bmZtTU1Ji8epoA11AAQJ5EInTS8PUigFEBq8TIhbW+DCjgOnAK5Vy3TWeRkfbwRjPmxEikj2kRa12zFidzuYf/lDBPXDeUEyOn86tQVtd7veHGM3P2ppV2sidBEARBWIdNYqSsrAw6nQ7e3t4m73t7e6OoqMjsMb6+vvjkk0+wfft27NixA0OHDsWMGTNw4MABi9dZu3YtNBqN8AoICLC4r73ghUeeVIJ8Q2Qk0KUt7cQbnxWLJW1pmku7AbCAdywKWjihYK6TxhJ8ZMRcwatxESsAHL1cDq2eRaCbCoHuKvhoFBjm5wKWBfb1oihIK2qL1OxLo24egiAI4urpVs8aX0fBw7Jsh/d4hg4diqFDhwpfT5gwAXl5eXj33XcxdepUs8c8//zzePLJJ4Wva2pqelyQGIuRYkNbr3F4ljc+KxWLwVblgNHrhXoRhM9CYcV+AJYLWM3Be42YjYwY2nszS+vRrNUh0ShFw3N9pBfOF9Rg78US3Da650PJWp0eGcVtrrAXi2pRWN0IX42yx69NEARBDFxsiox4eHhALBZ3iIKUlJR0iJZ0xvjx45GRkWFxu1wuh4uLi8mrp+HFSK6LD/JlCpP3gLY0TYuIQQ2rBWryDZERoGXIdJQ2ctGJ7tSMtG/tBQAfFwU0yjZb+ESDv8iUsDYxcp2hbuRAeiladXqrr8vz0f5M/OPbZLRorTs2q6weLTo9nGRixAW4AujdqAxBEAQxMLFJjMhkMowePRoJCQkm7yckJGDixIlWn+fUqVPw9bX+od0b8MLjSms1WlgtJIzEpP5DLpZDYxhEViIWA+d2AI0VgFyDokHcsQqxAoPk1rdX8ukZjZkBZwzDIMoQHdl7sQSZpfUQMcDEIW1iJM7fFW5OMtQ2a3Ei27Zi0vpmLTb8mY7fzxUh2cpC1AuGFM1QH2fMMAghStUQBEEQV4vN3TRPPvkkPv30U3z++ee4cOECnnjiCeTm5mLZsmUAuBTLkiVLhP03btyIn376CRkZGTh//jyef/55bN++HcuXL7ffp7ADnipPKMRtw+z81H6QtLOBFlI1EjFw4jPuzSHTUdDIPZB91b4W01XmeGTEI7gr6i5M9DMv5PgZNV8d5gb4xfq7QqNqG4wmFjGYPpRb056LxVZfFwCOXC5HiyGaklfZYNUxFw31K5G+LsJ1D2aUWR1ZIQiCIAhz2FwzsmjRIpSXl2P16tUoLCxETEwMfvvtNwQFca2xhYWFyM1tm9/S0tKCp59+GleuXIFSqcSwYcPw66+/Yt68efb7FHZAxIjg7+yPS1WXAJimaHi8VF64VHWJi4xUGT5j2CwU1nEGYLbUiwDAGJ8xGOMzxuJ2voi11GB8Zpyi4bk+0gs7Tl7BnoslePHGaKuvvT+9Lb2SV2GlGDFERiJ9nBHjp4GHWoayuhacyKkwidgQBEEQhC10q4D14YcfxsMPP2x225dffmny9cqVK7Fy5cruXKbXMRYj5rwFjItYBcJmIuXcRwCAIYaBe/aCL2LlMS5e5ZkS7gmxiEFmaT1yyxsQ6N6xM8cc3REjaYIYcYFIxGBqhCd2nLyCfWmlJEYIgiCIbkP2mUYEOgea/TcP395bYphbA+9YsM4+OHTlEABYTLd0lwhvZ/CmrSqZGKMCO9ajaJRSxAdx7++2MlWTXVaPnPI2AZJX2djJ3hzVja24UsXtN9TggcJ7ndDAPoIgCOJqIDFihHFqxlyaRjA+4yMj4TNxufoyihuKIRPJMMp7lF3Xo5CKEeLBddqMC3GDTGL+2zUrmutk+vl0gVXn5aMibk5cN0+uFZERPioy2FUJjZKrW5ka7gkRA2SU1CHfyroTgiAIgmgPiREjuhIjJpbwABA2C0kFnPPsaO/RUErs77cx2hD1mBFluXX6L3F+EDHAqdwqZJXVd3lOXoz8NZ5LRZXWNqOpVdfpMbwTLO8MCwAalVSI1lCLL0EQBNFdSIwYwTuuMmDM14zww/JUGiDmNiBgHA4VcCmaSYMn9cianp8bhY/uHoU7x1oeQujlosDkcG5tP5660un5mlp1OJxZDgCYHzcYajlXNtRVZIOfkdO+juW6Xm7x1etZ/Hm+yOo6F4IgCKLvQ2LECH+1P/427G9YMXoFFBJFh+18zUgZq4X+1n+iGTokFyUDACb4TeiRNQ1ykuGGGF+IRJ23DN86ajAA4KdTV8CyrMX9TmRXorFVBy9nOaJ8neE/iIvm5FV0XjfCR0aG+pga0PEtvoculaNZ23l0xR7szyjFg98kY+aG/fh4fya03TB7IwiCIPoWJEaMYBgGT8Y/ifti7jO73V3pDgDQslpUNlXiZPFJNOma4Kn0RLhreG8utQOzo33gJBMjt6KhUxOz/elcBGNahCcYhkGgG9d901ndiF7PCjUjUT6mkZFoXxd4OcvR2KrDsayKq/0YXXLJYEffrNVj7e8XMf/DQzh3pbrHr0sQBEH0HCRGbEAqksJN4QYAKG0sFepFJvhNsMnsrCdQysS4IYZzjN3RSaqGrxeZZohoBBjESGdpj/zKRjS06CATi4SCWh6GYTAtos0AzRKn86pw+8eHkVpwdROYC6q5CM7IQFdolFKcL6jB/A8P4e2dF6HTW44IEQRBEH0XEiM2IrT3NpS01Yv49Uy9iK0sNKRqfjldYLYgtaCqEenFdRAxwGSDgVoAn6bppGbkgiFFE+6thkTc8UdmkuFcSYZaFHN8uPcSjmVV4OvD2dZ9GAsUVjUBAObH+WHXk9Nw43Bf6PQstuzLxBeHsjo9lpxiCYIg+iYkRmyENz5LLU9FRmUGGDA9Vi9iK+ND3eHjokBNk9as98cBQ1RkRIArXFVcWy9vktZZzcjFwjazM3NMGMKlr84VVKO6obXD9ladXiiaPXuVKZVCQ2TE11UJT2c5PrxzFF69mXOe3bgrA8U1TWaPW/2/VAx7dafVc3gIgiCI3oPEiI3wkZGfM38GAES5R2GQwvrheD2JWMRg/kjOkt5cqoZP0Uw1pFUAIGBQW5rGUuErX7wa1a6ThsfbRYEhnk5gWeBIVsfoSEpeFWqbtQCA9OLaqyp0LajmxIafpq2N+t4JwRgR4Iq6Zi3W/HahwzE/ny7A54ey0KpjTZxnCYIgiL4BiREb4dt782rzAPSdFA3PwpFcS/K+tBJU1rcI77fq9EJNxzQjMeJvECO1zVpUN3aMagBtM2mG+pgXI0DbNOHDZlI1iUYCoFXHIsNQhGorzVqdMKfHz7Wt20kkYvDGLTFgGOC/KQUma8gpr8cLO84KX18u7d61CYIgiJ6DxIiN8JERHntbwF8tQ32cEe3rglYdi1/OFIBlWVwqqcOWfZmobdbCVSXFcH9XYX+lTAwPtRyA+VRNQ4sW2eWckZqlNA0ATDSkag5d6ljEesAggsSG9uTudr8UV3NCRC4RCe6xPDGDNbh7HDes8ZX/nkOrTo8WrR6PfncKdYbPDcAqU7iBQElNU6ct3gRBEH0JEiM2wruwAoBKokKcZ5wDV2MevpD1/d2XMG7NbszcsB8bEtIBcPNkxO08SwLdLBexZhTXgWUBD7UMns5yi9ccH+oOxmANX1LbVrdR1dCCM/lVAICbhnPdPucKuidG+E4aX43CbPfS07OHws1JhoySOnx5KBvv/HERZ/KroVFK8X+LOav+rLL6Af+Q/v1sIcau2Y3N+zIdvRSCIAirIDFiI3yaBgDG+o6FVCx14GrM85cRfhCLGJTVNaOkthkyiQjjQ93wxMwIvHRjVIf9AzrxGmmzgbccFQE4c7ZoX24f4zRJUmY59CwQ7qXGTIOl/dkr3WvvFYpXNeZt9zUqKZ67IRIA8O6fafhnItdds+624RgX6gaxiEFDiw5FFopcBwoHDdGpI5ctdzcRBEH0JSSOXkB/wzhN09fqRXi8nBXYfNcopBXVYkywG0YGukIhFVvc37iItT2CDXwn9SI8E4e443xBDQ5nlmP+CC46w3fwTAn3RMxgjeGcNWjV6SE10ybcGQWGtl5f147uuDy3jfbHd8dzcSq3CgCwdGIw5gzzAQAEuqmQVVaPrNJ6i4JmIMDX5FwuvTZSUgRB9H8oMmIjg+SD4CR1AgOmz4oRAJgzzAePzQjHhCHunQoRAAgQ0jQda0Z4k7LOild5+CJW3m+EZVkkGupFpkZ4IMhNBbVcghatHpndKCTlIyN+nQgJkYjB6/NjoJCKEBfgiufnRQrbQg2GbZkDuG6EZVmkl3ACsqC6scsBiARBEH0BiozYiFgkxqbrN6G+tR4BLh0n+/ZH+DRNfrvISE1TK07mcr4cY4LdujzPmBAuFZJb0YC8iga06PS4UtUImViEcSHuEIkYRPu54FhWBc5dqeky9dMePjLi59p5VCNmsAZHnp8BpUwMuaRNiIV6OmH3RSBrAEcMSuuaUWXwemFZLvUW4d21kCQIgnAkFBnpBmN8xmB6wHRHL8Nu8Gma/MpG6I0s1RPTy6DVswj1dEJwOxt4c6jlEsT5c6mYw5nlQkvvmJBBUMo4URBrSNV0p6OmoIo3PLOcpuFxVclMhAgAhHioAQCXy6yPyvS3Ytf2bdOUqiEIoj9AYoSAr0YBsYhBi06PYqNOmD0GF9cZkV6WDu1AW6qmTGjpnRreVvQbM5iLhnRHjBSaMTyzhVBPTlBZ+4DW61ncsjkJc9470G+s5NOLa02+vlZamQmC6N+QGCEgEYsw2JD64L1G9HoW+9I4MXKdLWIkjPMbOXipXOjmmGIsRvy4yEhqYY1Ng+0aWtpM2ayJjJiDrxnJr2ywygX2clk9TudVIa24tt881NMNkRGZhPuvnWVDFIggCMJRkBghABgVsRrqRk7nV6G8vgXOcolV9SI8owIHQSYRoayuGQ0tOnio5SadOKGeaiikIjS06Gx6wPP1Imq5BC6K7rVTezrLoZZLoGeB3HLLgwF5UvKqhH/3F+fWDENkZIpheGF2WdefkyAIwtGQGCEAtNWN8F4jfIpmaoSnTS24CqkY8UFts3qmhntAZGSyJhYxgh/JeRvMzwqNDM+6C8MwCDFERy5bIYRS8tqG6lmzv6NhWRYZJZxo4tuZ+8O6CYIgSIwQANo6angXVl6MXG9DioaHt4YHgCkRHh22x3SjiJUvXu2qk6YrbKkbMY6MdKcVubcprW1GdWMrRAxwfRT3fSura0ZNk/mZQwRBEH0FEiMEAOP23kYUVTfhfEENGAaYPtSziyM7MmFImwCZFGZZjJy1SYzwbb3dj4wAQKiho6arWoqmVh0uFrYVg/aHrhS+XiTY3QkearkwcyiboiMEQfRxyGeEAAAEDGqbT7PXULga5+8Kd7XleTSWGBngirvHB8JTrYCXc0fxwBexnr9SA72eNUnjWKIrK3hrCbEyMnK+oBpaPQsRA+hZrmaEZVmzM3F6kvMF1XBRSAWx2Bl8J024Nye4Qj2cUFbXjKyyepPhiARBEH0NiowQANoiI0U1Tdh5rgiAbS29xohEDN64JRaPzww3uz3cWw2ZWITaZq3Z4Xzm4Nt6r6ZmBGjrqOmqloK3k584xAMMA9Q0aVFe32LVNXLLGzD57T14f1dGp/uV1zWjVWe5ZfinU1dw06aDuOOTI1b5nWQYnFd5kzO+Pqa/dAIRBHHtQmKEAAC4O8mglIrBssCBDM6szJaWXluQikWI9OUemOesHJpnr5oR/gFdUd+CqgbL4oKvFxkf6ib4mlibqvnfmQLkVzbivV3pQnt0exIzSjHhrT2YsX4/0opqO2zflVqMp/5zGiwLXKlqtEoI8WmacF6MeJIYIQiif0BihADAdZoEGqIjLAt4u8gxzM82u3ZbGGZI1ZyzoqOGZdk2w7OrFCNOcgl8XLjoSmfRkdP5VQCAEQGDjIperStiPZnT1oXzzA9nUNFOSORVNODR706hRatHbkUDFmw+JESjAM4w7uF/nzTxYemq7oNl2bY0jReXpqHICEEQ/QUSI4QA7zUCcF00PVkfYYstfHVjKxpaOJOyq03TAEYPaQuRjvK6ZsH8bXiABkM8eRv5rh/qLMsi2TDPx81JhtLaZjy3/YyQZmlo0eKBr0+gqqEVw/01mDjEHQ0tOiz7Nhkbd6XjVG4lHvjqBFq0esyO9sb4UM7jpStBUVzTjNomLcQiRhBPxp+zv9naEwRxbUFihBDwH9RWJHl9pHePXsvYFr6rByXfSePmJOtyArE1CJEOCx01fIpmiKcTXBRSDLEhMpJZWo+qhlYopCJ8vnQMpGIGf6YW4z8n8sGyLJ7dfhYXi2rhoZbh43tG4+v7xuJvk4IBABt3ZeC2jw6jvkWHSWHu+GDxSIR58d0/nYsRPioS5K4SZvIEuqnAMEBtsxZlddbVuxAEQTgCEiOEAJ+mkUlEmBTm3sXeV0eEtzOkYgaVDa3Ir2zsdF97GJ4ZE8pHOixERngxMiJgkFX7G8OnaIb7u2JEgCuemj0UAPDa/85j9S+p+N/pAkhEDDbfNRq+GiUkYhFevXkY3rltOGRiEXR6FiMCXPHJPfFQSMUIdueEUHa5dWIkwqvN7VYhFQs2/10dTxAE4UhIjBAC8cHcw3dujA9Usp7t+lZIxYg21I2cyKnodN8CoZPm6upFeEK7qKUQxEigK7e/ITKSW9HQafcLACQbxMhogwvtA1NCMS7EDQ0tOnxxKBsA8MrN0RgbYmqx/9f4AOx4eCKemTMUX/1tLJzk3P1vq/vovOuIn9YbYWjr5ekqJdUVJ3MrUVTd1PWOBEEQVwGJEUJguL8r9j8zHW/fOrxXrsfbxp/Irux0v0Khk8ZekZE2MaJvN6xPr2dx2iBGRga4AgB8XBRQycTQ6lnBLt8SfL3I6EDus4lFDDYsGgFnBScu/jraH/eMDzJ7bMxgDR65LgwaVdvsHV5MZJd1XveRXsJ7jDibvG9tK7M5LhbV4NYtSfjHv5JtPpYgCMIWSIwQJgS5O9mlLsMaeDGSnNOFGLFTJw3PYFclpGIGzVo9CqpNU0RZ5fWoadJCLhFhqGHAn8lMm04iDFUNLbhkmA0zymg+z2BXJb57YDxeuSkar98SY1NhcICbCmIRg8ZWHYprms3uw7IsLgmREVMx0hZZsd3O/kx+NViWixT1hqU8y7Jo0XYeeSIIYmBCYoRwGKMNaaG04lpUN1p+2F2psm/NiEQsQpC7eXGRYjA7ixmsMRkQ2FY3YvmhzhulhXo4wc1JZrItZrAG900OsVnoScUi+BvccS2llQqrm1DbrIVE1CaaeEI8rSuANUeOoc6EZdvuS0/yty+PY8q6PSitNS+6CIIYuJAYIRyGl7MCgW4qsCxwKtdydIQvYLVXZASAUaTDVFy0+Yu4mrwfakVkpH29iL3oqoiVn9Qb7OEEmcT0v3SIcGxDh5RUV2SXt6WkTnby/bEHzVodDqSXorimGf9JzuvRaxEE0fcgMUI4FD5Vc9JCqkavZ4UCSntFRgDTuhFj+OLVuPZipIt2YKDnxEhX5mUZfCdNu+JVABg8iEtJtZhJSXWFsdHayR6OjORVNILXSt8fzyNfFIK4xiAxQjgUPlVzwoIYKatvRquOBcMA3i52FCNmCjubWnW4UMjZ049sJ0aGdNHeq9XpBSHT22KkzXnVucM2sYgRUlK2pGpYlkWOUWTkVG6lzZEVWzAWPjnlDThyufMOK4IgBhYkRgiHEh/Etbim5FVBa6ZtttBgeObtrDCp4bha+BqQU7lVWPPbBSSkFuNwZjladSzcnWRCnQYPLwjK61tQ3dCxvuViUS0aW3VwUUgE4WIvgo06aszRNpPG/HW7YwtfXt+CumYtGAZQSEWobdIi00o7/O7QPgW19Xhuj12LIIi+B4kRwqGEe6nhrJCgoUWHC4UdB8YJhmd2auvlifRxhrNCgrpmLT45cBkPfH0Cf/vyOACuXqR9x4vxTJtMM6kaPkUzKmgQRCL72ujzdR85FQ0m82oALo3Fd/C076ThsabepT188aqfRok4f1cAPVs3wouRKeEeAIDfzxV1OsiQIIiBBYkRwqGIRAxGBfKpmo6h+SuGyIifnQzPeJwVUux5ajo23B6HxWMDBMt3AJg21NPsMW0D8zo+1IV6kUD7pmiAdnUfVaZ1HxeKalDXrIVKJu7QScPTncgIb7IW7KHCyEC+rqeqG6u3jmzD9W6O80O0rwtatHr8dOpKj12PIIi+BYkRwuEI5mdm6kYK7dzWa4ynsxwLR/lj7cLh2P3UdCS/NBP/Wz4Zd40zb0rW2fTenipeBbi6D96qv30642BGGQBgXIibxTSWYJxmgyU8HxkJcnfCKIMTbW9ERkI8nHDH2AAAwFYqZCWIawYSI4TD4YtYk7MrOzx8eMMzXzu29VrCXS1HrL8GYgtpllAP80WsRdVNuFLVCBHTsQvHXoRYqBs5eIkTI5PDzUdzjI/Nq2iw2lSMb+sNdlcJBm4ZJXWd+sF0l2atToj4BLmrMD9uMOQSES4W1eJ0ftdTnQmC6P+QGCEczogAV4hFDIpqmgSDMx6+HdWvByIjtmKpvZePikT5uggzZexNsNAR09bh0tSqw7EsLrXF11qYw9NZDieZGHoWyK2wLjrCi55gdyd4qOVCZIbvGLKVZq3OohDKq2iAngWcZGJ4quXQqKS4MdYXALD1mOMKWZMulSHipd/xnxPke0IQPQ2JEcLhqGQSDPNzAWBqDZ9eXCu02voPUjlkbcbwXTLZ5aaFpD2ZouEJEXxR2oRQck4lmrV6eDnLEe5luYOHYRhhZk1aUdcdMSzLCmkTvpNHSNV0Yd1vjhatHjd+cBBzNh4wK0iyhfoUJ6FweNEYLlXz8+kC1DVrbb6mPfgp5QpatHp8f5zECMC1ry//90ls+DPN0UshBiAkRog+weh2Q/OqGlrwwNcn0NSqx/hQN0GsOBI/VyXkEhFatHpcqeQiNtll9dh9sRhAD4sRIydVnkRDvcjkMI8u591EGubsXCyq6fJalQ2tqG3iBAAfEeFTNd2pGzlyuRyXSuqQVVYviEtjBOHj3laAOzbEDaEeTmho0eHXMwU2X9MenL3CrfV0fhWaWnUOWUNfIiWvCr+cKcQHey4J3jYEYS9IjBB9At5v5EROJbQ6PR797hRyyhsw2FWJzXeNtnu7bHcQG81+OXulGm/vvIjZ7x1ATnkDXBQSTBxiOVVytQQb1X3wfiyHhHqRrq87VBAjXT9EsoW2XoUwS4fveErJq7LZ/OyP80XCv88Y7PaN4bt8gj3aol8MwwjRkX8f7f1UTVOrTnC2bdWxwtyhaxm+hRwAvjiU5cCVXJtkFNdi6RfHcHaA1lGRGCH6BPH80LyiGrz83/NIzCiDUirGP5fEdxg650j4upHHtp7Cln2ZaNHpMS3CEz8+MgmezvIeu66PiwJyiQhaPYv8ykZU1rfgXAH3S2lymPViJM0aMVLW1knDE+njDKVUbLP5mV7P4s/UYuFrcwWpvNOr8fUA4LbR/pCJRTidX43T3axV6S4Xi2qhNRJdR7PKe/X6fRFjMbLj5BVU1JMPTG+yISEd+9JK8X97Mxy9lB6hZ6rtHIROp0Nra8+POifsj0YGDPFQIrOsEd8Zihbf/WscovtAesYYvqNGp2cR6KbCKzdFY0aUV5dpkqtFZIjKXCyqRVZZPepbtGBZbh6NlxU2+ZE+3H3MrWhAXbMW6k4KbYVOGqNIhUQswnB/DY5mVeBkbqVQg9IVp/IqTabwmhMVfGSkvU+Ku1qOm4b7YsepK/j6cA7W91CnkjnOGiI4DMNNLeYLha9lLhlEKMMAzVo9/n00B8uvD3fwqq4NappasftiCQAg6VI5tDo9JHZ0pO4LDAgxwrIsioqKUFVV5eilEFfBq1Pd8N8LVdhxoR7Lrw/DjcN9Hb2kDswf4YejWeWYPtQL908OEdIYvUGwe5sYySjhIhyTwyy39Brj5iSDl7McJbXNSC+uFdIu5jD2GDFmVNAgTozkVGHRmECrrvvHeS4qMnGIO5Iyy3GptM5EDDW16oSOqWD3jqZtd08Iwo5TV/C/MwV48cYoq6Jkf5wvwteHs/HeohHwcu5eF9bZK1wEZ3a0N/44X4yTuZVo0eo7TEW+luAjI3eMCcB3x/Lw9eEcPDh1SLfvCcuyPS7iBwo7zxYJxd+1zVqczq/CaENqe6DQLTGyefNmvPPOOygsLMSwYcOwceNGTJkypcvjDh06hGnTpiEmJgYpKSndubRZeCHi5eUFlUpFP+D9EJZloa6swU0sizBPNR6aGeHoJZkl3NsZ/1k20SHXDjYyL+OLVztr6W3PUB9nlNQ2I62oczFi3NZrDH+MtUWsLMsK9SJ3jw9Cdlk9CqqbcO5KNcaHugMA8isbwBraej3UHYXGyABXxAx2wbkrNdh2Ig/Lpg3p9Jp6PYvV/0vFlapG/PdUAR6YGmp2v8OZ5Xhr50WsXRBrNvrGF68uGDkYx7MrUVHfgrNXBt4DwFoaW3RC2/3jMyKw+0IJSmqb8evZAiwY6W/z+ZZ+cQyZpXX47bEpcFZI7b3cAcd/T3NuxFIxg1Ydi8SMsgH3s2izGPn++++xYsUKbN68GZMmTcLHH3+MuXPnIjU1FYGBlv9aqq6uxpIlSzBjxgwUFxdb3M9WdDqdIETc3d3tdl6i9/FXKCCViODtXAaW1QPovahDfyDEkDY5kF6K/MpGSMUMxoZY/wsp0scZiRllXdaNmEvTAMBIQ3svb36mUXb+EEkrrkVOeQPkEhGmRXjiZ39XFFQX4XRelSBGssy09RrDMAyWjA/Gyu1n8O2RHDwwJdSiKR0AHM+uEB6axjUO7fnmSDZO51XhkwOZ2HjHSJNtxsWrw/1dMTbYDTvPF+FoVsWAewBYS2ZpHVgWcFVJ4e0ix70Tg/HOH2n47GAWbhkx2KY/AOubtdiXVgqAcxCeG9v3IqB9ieKaJiRlcjVLD04NxYd7M5GYUYYVffQPtu5ic3xtw4YNuP/++/H3v/8dUVFR2LhxIwICArBly5ZOj3vooYdw5513YsKECd1erDn4GhGVyvE+FMTVwTAMNM5qMAyo9scMwe3ae0cGDrLJZI2vGzHXXstT1dAiuKzybb08Hmo5gty597bsy0R9F/4fO89xUZEp4Z5wkksEd9ozRkWsQhTGwlwdgJtXo1FKkV/ZiH1pJZ1e80ejeTZ8Kssc/KTj/emlHYYP8sWr7k4y+GoUguA7evnarRvhi5bDPNVgGAaLxwZCLhHh3JUaHM+2rd3beEYS7yBMWOZ/pwvAspx1wB2G9GhKXhVqmgbW70ibxEhLSwuSk5Mxe/Zsk/dnz56NpKQki8d98cUXyMzMxKuvvmrVdZqbm1FTU2Py6gpKzQwM6PtomRBP0wf2FCu6aIwROmqKay3OfOEfFN4ucqhkHYUO74z60f5MTFm3F5v3XbJoSsbXi8wZ5g0AiPPXADB1cW3zGLH8x4RSJsbt8Vwq4OvDORb3a2rV4dezhcLXGSV1Zj9ns1YniKDKhlak5Jk+TPni1ZjBGjBMW/Qp2dB2fi3CR5nCDOZ6bk4yLBzFfU8+O3jZpnMZd2NdrRjR6vTIq2joesd+zH9TOJ+d+SP8EOCmQqiHE3R6FoczB1aHl01ipKysDDqdDt7e3ibve3t7o6ioyOwxGRkZeO655/Cvf/0LEol1f8WtXbsWGo1GeAUEBNiyTIIYkHiqOVt3Hmv8RYwJ81JDLGJQ1dCKEqMOF2NyhJk05iMVT86KwLt/jUOwuwoV9S1YtzMNk9/eg88PZpk8+PMqGnChsAZiEYOZUdzvixiDGLlS1YjyOu765gzPzHH3+CAwDBfJaD+fh2f3hRLUNmnh46KAWMSgtklr9nNmldWbtO3uvVhqsp0vXo0dzK03ytcFzgoJ6pq1SO0kqjSQaS9GAOC+ScEAgD9Ti5Fbbr0gMJ7tlFPe0G0x0dSqw52fHsWUdXu7jJj1VzJL63D2SjXEIkb4Q4D/f5+YUdrZof2ObpVBt//r1VJVtE6nw5133olVq1YhIsL6/Nbzzz+P6upq4ZWXR3bMBMEwjJDOcFZIMNzf1abjFVKx0D5rKVXTlTiQiEW4bbQ/dj05De8tikOohxOqGlqx+pdUPP2fM2jWck6lfOHq2GA3DDJ0wLgopBhiiO7wqRreCr59W297gtydMC2C6xz69oj56MiPp/IBAAtGDRbSSRnFHetG+BQN/ytrz0XTBxlfvBpjECNiEYMxwVx0pH2Lb2ltM17+6Rz2p9v2YGBZFiU1TSipaUJFfQtqm1r7tMsrL0aGGImRcG9nTI3wBMsC3x61HLFqz2ULwx5tQadnsWJrivD92LIv0+Zz9Af4qMjUcA+4qzkfoymGoZj8xO6Bgk1ixMPDA2KxuEMUpKSkpEO0BABqa2tx4sQJLF++HBKJBBKJBKtXr8bp06chkUiwZ88es9eRy+VwcXExeRGdExwcjI0bN9rlXPv27QPDMNQq3QfhxcjEIe6dFnJaoivzM8GAzKPzGiyJWIQFI/2R8OQ0vHpzNMQiBttP5uOeT4+hor5FECN8ioYnziCgUvKqTNp627cRm2PJhCAAwLYTeWhoMU0Nldc1C0WRC0cOFmb1mKsbSTd89llR3mAYILWwBkWG6dCmxasa4ZhxfN2IkRjR61ms+P4UvjmSg/u/PI49F60vzF/7+0WMXbMbY9fsxqjXExD72p+IfHkn/vbFMYspNEeh1ekFkRrmaToDacl47nvyQ3K+IES74rIhTcOPeLBVjLAsi1X/O4+d54sgE4sgFjE4mlWB8wUDy5mUZVn8N4WrgZo/YrDw/vhQN4hFDLLLG2yKSPV1bBIjMpkMo0ePRkJCgsn7CQkJmDixY7uji4sLzp49i5SUFOG1bNkyDB06FCkpKRg3btzVrb6fM336dKxYscIu5zp+/DgefPBBu5yL6LvMj/ODu5MMSyYEd+v4SO/OxYhgQGaFOAC4qMHfJoXg86Vj4CyX4Fh2BeZ/eBAnDAP1Zg/zMdm/rYi1CnkVXFuvWi4x29bbnmkRXgh0U6GmSYuVP5wxsaX/5UwhtHoWMYNdEO7tLKQTzHXU8HNVxoe6C+JoryHM3754lYevGzmeXSFc95PEyzh0icvba/Usln17EkmZXT9YdXpWmATcPqC8N60UR/pYoWxORQNadSyUUjEGuypNtk0f6glfjQIV9S1CwXJnsCwr/Izx4jLpUplNIwY+2n8ZXx/OAcMAGxbFYW4M9zP25aFsq89hjrK65j4lBE/nVyOnvAFKqRizottEvbNCKgyuTLw0cFI1NqdpnnzySXz66af4/PPPceHCBTzxxBPIzc3FsmXLAHApliVLlnAnF4kQExNj8vLy8oJCoUBMTAycnKz7hXetwrIstFrrJpZ6enpSR9E1wOxhPkh+eRYm2Vi8ytPVjBpLhmddMS3CE9sfngj/QUrkVTSCZbmCVb92Dy8+2nAmv9pkJo01hctiEYO3bo2FRMTglzOF2JCQLmzbYeii4T0vwr24z5lhRozw7w31ccb1kV4AgL2GVE374lWemMEaqGRiVDW0Ir2kFmfyq/DuH9z02jduicGsaG+0aPV44KsTONWFD8vJ3EpUNrTCRSFB+htzkblmHi6+fgPuHMd1Svwz0baC0J6GF3Shnk4dZkRJxCJhhhDvnNwZRTVNaGjRQSxi8Je4wVDLJahsaLW6FufHU/l4e+dFAMBLN0bjpuF++NukEADAf08XCLVItrLteB7i39iFL65S0NiTnww/07OivTt0zQ3EVI3NYmTRokXYuHEjVq9ejREjRuDAgQP47bffEBTEqdzCwkLk5vb+YCselmXR0KJ1yMsWVb106VLs378f77//PhiGAcMw+PLLL8EwDP744w/Ex8dDLpcjMTERmZmZmD9/Pry9vaFWqzFmzBjs2rXL5Hzt0zQMw+DTTz/FggULoFKpEB4ejp9//rnb93X79u0YNmwY5HI5goODsX79epPtmzdvRnh4OBQKBby9vXHbbbcJ23744QfExsZCqVTC3d0dM2fORH29+SJEomeJ8uVC45dK6tDarjOkuqEVlQ1cu2BQJ90tlojwdsZPj0wSphcbh5aNry8RMSivbxEG/dkifCYO8cDahbEAgP/bewnbTuThcmkdTudVGR5wfgBgMTLS1KoTUg7h3mpBjBy8VIZmra5D8SqPVCwSPteeiyV47LtT0OpZzIv1wV3jArFp8UhMDvNAfYsO935+rNP26V0XuHTOdZFekBrSDAqpGA9MCQXDcOfP6ENTcc0VrxqzaEwARAxw5HJFl3OL+OLVQDcVlDIxxodyESdrUjW/nS3EM/85AwB4YEoI7p/MiZBRga6I89egRavv1lBFnZ7F/+29BMA6QdUb6PQsfjnDdYbdMtKvw3a+iPXQpbIB0+HVLQfWhx9+GA8//LDZbV9++WWnx7722mt47bXXunNZq2hs1SH6lT967Pydkbp6jtl2SHO8//77SE9PR0xMDFavXg0AOH/+PABg5cqVePfddxEaGgpXV1fk5+dj3rx5eOONN6BQKPDVV1/h5ptvRlpaWqdGc6tWrcK6devwzjvvYNOmTbjrrruQk5MDNzfbjJuSk5Nx++2347XXXsOiRYuQlJSEhx9+GO7u7li6dClOnDiBxx57DN988w0mTpyIiooKJCYmAuDE6eLFi7Fu3TosWLAAtbW1SExM7FPh0GuJwa5KOMnEqG/h2luNZ8zwD2lPZ7lN/iXGeKjl2PrgeKQW1AgFoMYopGJE+brg7JVqoQ3X2pQQz1/jA5Bb0YBNey7hhR1nhSjRlHAPYVjhEE/Or6aivgXldc1C8d+lkjbzLk+1HJ5quWCTfyyrokPxqjHjQtyQmFGGDX+mQ6tn4adRYO2C4WAYTkx8smQ07vnsGJJzKnHPZ0fx2+NTzNrR777ARWFmRJnW04R4OAn2858mZuHt24bbdF96isySNo8Rc/hqlLg+0gu7LpTgu6O5eOmmaIvn4utFQg21T5PCPLDrQgkOXSrr1F33u2O5ePHHs9CzXIvr83OjhG0Mw6UKV3yfgq+P5OChabZZ1O9PL0GuoaMno6QOl0rqLAqv3qK4pglldc2QiBghCmLM8MEauCgkqGnS4syV6k4dlfsL1+6gBQej0Wggk8mgUqng4+MDHx8fiMVc2+bq1asxa9YsDBkyBO7u7oiLi8NDDz2E2NhYhIeH44033kBoaGiXkY6lS5di8eLFCAsLw5o1a1BfX49jx47ZvNYNGzZgxowZePnllxEREYGlS5di+fLleOeddwAAubm5cHJywk033YSgoCCMHDkSjz32GABOjGi1WixcuBDBwcGIjY3Fww8/DLXasf/Zr1VEIgYRFlI1vBixVRy0RyoWIS7A1WKBLZ+qKavjpr52ZnhmiSdnReAvcX7Q6lmhk2XByLZIjFImRsAgQ0eNUXSEL2iN8HYWIpLTh3K/7HeeKxIiErH+HcXI2BDONVarZyFigI13jIRG1eZCq5JJ8PnSMRjq7YyyuhZ8a8YTJbusHpdK6iARMUJ3kDEPGuzrfzx1BSW1TTbckZ6DH5DX2QOaTzH9cDK/066gTENkhJ9+zU+cPpZVYfG4Lfsy8fwOTogsHhuADbeP6JAumhfrCy9nOUprm/GbkdeMNXyZxH2f+KwcX3ztSPiWdE9nOaRmBuJJxCJBhNsjVZNX0eDwKcwDYlCeMUqpGKmr5zjs2vYgPj7e5Ov6+nqsWrUKv/zyCwoKCqDVatHY2NhlOmz48La/rJycnODs7IySEtv78S9cuID58+ebvDdp0iRs3LgROp0Os2bNQlBQEEJDQ3HDDTfghhtuENJDcXFxmDFjBmJjYzFnzhzMnj0bt912GwYN6v9Kvr8S6eOMU7lVuFhUg5vj2kLAQidNN1I0thAX4Ip/GYXTOzM8swTDMFh323AUVDXiRE4l1HIJZkebFsuGe6mRW9GAjJI6wX4+rYh7sEZ4tz1Yr4/0wrYT+fghOV8oXvXTdIxoDPfXQC4RoVmrx/Lrwsxa8WuUUjw2IxyP/Psk/n0sD8uvDzf5K51P0YwJdjNrpz86yA2jAl1xMrcK3xzOwVOzh9p8b7pLYkYpwrzU8NW01fmwLNsWGelEjEyL8IKfRoGC6ib8cb7IbIoOaGvrDTVEWcK81PB2kaO4phnJOZUmtVAsy+Kt3y/i4wNcDc0/pg/ByjlDzdYXySQi3D0+CBsS0vHFoSzMH+FnVR1SZmkdDqSXgmGAf0wbgs37MrHzXBEeuS6sy2N7kpIaToh6GSJ95pgc7oHfzxUhMaMUj824uunJb++8iF/PFmL1X4bhnm4Wx18tAy4ywjAMVDKJQ172cg9tX9j7zDPPYPv27XjzzTeRmJiIlJQUxMbGoqWlcyUrlZr+smMYBnq97flFcz4yxmkWZ2dnnDx5Et999x18fX3xyiuvIC4uDlVVVRCLxUhISMDvv/+O6OhobNq0CUOHDkVWVpbN6yDsA28L376jRvAY6Uakwhbi2vmjdPd6XGokHvNH+OGVm6OhlJn+MRBmEByZxpGR4rbICM/kcE9IxQyaDVNR2xevGl/v7VuHY/l1YZ3+8p89zBveLnKU1TVjZ7u/svkUzczojlYIPHx05JsjOR1amHuKk7mVuOezY7j706Mm9viF1U2oNxScdlbbIxYxwiTnf3VSt9E+TcMwjCBAEo3+wm/R6rHyhzOCEHlhXiSevSGy09+xd44LhEwiwun8apzMreriE3N8Y4heXW+Ywi1iONO7/ErzLbNpRbVWtzBfDW2REctTp6ca0jcnc6ssuiBbg17PIimzHCwLRPo6zkZjwImR/oRMJoNO1/UPdmJiIpYuXYoFCxYgNjYWPj4+yM7O7vkFGoiOjsbBgwdN3ktKSkJERISQWpJIJJg5cybWrVuHM2fOIDs7W/CRYRgGkyZNwqpVq3Dq1CnIZDL8+OOPvbZ+whRzHTUltU04YrCX7unISJiXGiqDcHCWS+Du1HVbryXcnGR4/46RuD2+o0szX+Ng7DWSZkaMqOUSkyhH++JVY24ZORhPzxkKiZnQOY9ULMLisdyD+eukbOH96sZWHM/m2nZnRnlZPH5WtA+C3FWoamjFD8n5FvezJ6cMD+/M0nokpLb5pfDFq0Huqi7rMPhC1mNZFbhkxt+lqbVt8m+oUf0Jn6rhC5qrGlqw5POj+E9yPkQMsO7W4XhwaufTmgGuXmm+IdK38ofTePo/p/H6L6nYtDsDPyR3TB/VNWuF+3vvxGC4q+XCz4G5NuWtx3IxZ+MBLPgwCVUNPZvS4MWIl4vlyEiAmwq+GgV0ehapBd13Bk4trEFFfQvUcglGGFrvHQGJEQcSHByMo0ePIjs7G2VlZRajFmFhYdixYwdSUlJw+vRp3Hnnnd2KcHSXp556Crt378brr7+O9PR0fPXVV/i///s/PP300wCAX375BR988AFSUlKQk5ODr7/+Gnq9HkOHDsXRo0exZs0anDhxArm5udixYwdKS0sRFRXVxVWJniLSIEbyKxtR29SKsrpm3PnPoyiobsJgV6XZgjl7IhYxQoFokJVtvd2BL87lXVjrm7XIr+QehsZiBACuG9omDswVr9rKnWMDIRExOJFTKTwo9qeXQqtnEeal7jLKwHeKfHYwq8Mgv57AuHvHuLX4UhfFq8b4aBS4PpKL+Hx3rKNrdnZ5PViWcw829pXhxci5gmok51RiweYkHLlcAbVcgs+WjsHtY6wfB3KfIbqRWVqPH5Lz8dnBLKxPSMfT/zmNRZ8cQZlR6++Ok/moa9Yi1MNJWMMNBl+c9nUjdc1avGNo5U4trME9nx0TBkr2BKW1XadpgLbuuItF3RcjfERqfKib2fqU3oLEiAN5+umnIRaLER0dDU9PT4s1IO+99x4GDRqEiRMn4uabb8acOXMwatSoXlvnqFGjsG3bNmzduhUxMTF45ZVXsHr1aixduhQA4Orqih07duD6669HVFQUPvroI3z33XcYNmwYXFxccODAAcybNw8RERF46aWXsH79esydO7fX1k+Y4qqSwdvwF9eRyxW4659HcamkDr4aBf79wDiztQz2hv8LLMSj5wqZ+RqHktpmVDe0Cg9WD7UMbu2iMddFtokRc8WrtuLlosAcgxnXN0eyAQC7DfUiMzqJivDcNtofriopcsobeqWgMs1IjCTnVCI5h4vgWFO8asxdhkLW7Sfz0aI1/YPpcmlbvYixAPVyUSDCWw2WBW7/+DCyyuox2FWJ7f+YaCISrSHK1wU//GMi1i6MxbM3ROKhaaG4Y0wAXFVSnM6rwoLNh5BZyg1Q/MoQtVoyIUgoiOVN+k7kVJoUEH+8PxPl9S3wH6SEm5MMZ69U497Pj6G2hybnltQYIiOdpGkAIMqXE9WdtZJ3xUGDcVp3vYvsxYArYO1PRERE4PDhwybv8Q94Y4KDgztY5z/yyCMmX7dP25hrnbXW3n369Okdjr/11ltx6623mt1/8uTJ2Ldvn9ltUVFR2Llzp1XXJXqPSB8XFNeUYvm/T6JZq4eXsxz/fmC8zWZn3eVvk4JxpaoRDxnqI3oCtVwiFFVeKq0VOjnaR0UArobhiZkRYMF2cBntLkvGB+HXM4X46VQBnpkTKRirzYqyXC/Co5JJsGR8ED7Ycwmb9lzCDcN8OnSQ2AuWZYXo0eigQUjOqcTH+y/jkyVuXXqMtGdqhKfQKn0os8xETPD1IkPM1AhNCvNAenEddHoWIwNd8ck98UKbtq2MChzUodX1wamhWPrFceRWNGDh5iTcNykEmaX1cJKJcetof2E/P1cl4gJccTqvCn+eL8bd44NQXNMkRIteujEKgW5OuPPTI0jJq8LSL47jq/vGQt3NVnhLCGmaLu4BX/91obB7vjRNrTocz+ZM+qbYOHjT3lBkhCCuQfhUTbNWDw+1HN89OL7LYXX2xFejxId3jrJLSqQzwoxSNeaKV3kYhsHjM8OxYqb1Az27YmyIG4Z6O6OxVYfntp9BTZMWbk4yjLTSE+K+ySFQyyW4UFiDP1Otn3tjKwXVTahr1kIqZvDGLTEAgIQLxbhcWmdVJ40xYhGDOYbows6zphGdy+3aeo1ZONIfarkEC0cNxncPjO+2ELFEqKcaPz48ESMDXVHd2Ir3dnHuvbeO9oezwjQSyNvL8xGp9xLS0dSqx+igQZgzzAfRfi749v5xcFFIkJxTifu+PN7BQPBq4aMyXd0HPk2TVlTbrXTe8ewKtGj18HFRYIgVqbiehMTINciyZcugVqvNvnhbf2Jgw8+IcXeS4bsHxjn8F1FP0TYwr06Y1hvu3TuflWEY3GOYv8KLielDPa0ecOiqkuFvk4IBAO/vzrBpfost8IMDQz3UiPJ1wcwoL7As8O6faSg3eE/Y8vMxN5Z7mP+ZWmTiDprZrq3XmFh/Dc68Ohsbbh8BhZ0sEtrjrpbjuwfGC3UhAMzOeOLF1OHMchzLqsA2wxyhF+a1dfPEDNbg27+P4+YxZVVglx3Fol7PCh48nRWwAlxbvFwiQmOrTjBua88PyflY+cNps51ZvEfJ5HCPHqvdshYSI9cgq1evNhleaPzi3WCJgc2cYT5Y/9c4/Hf5JBMX1oFGuJEtPD8gb2gvft4FIwfD2SiEP9OKFI0x908OgbMQHemZ2hG+XoQXaQ9M4VJnvxkiG34ahU2OvGOD3eDmJENlQ6sw5Zhl2ba2XjOREQA9loYyRiEV48O7RuHlm6Kx7tbhZiM+IR5OiPRxhlbP4sFvTkDPctGS0UGmvjLD/V1xx1iuuDbBjmKkoqEFOj0LhuE6hDpDIhYJ3XHm6kZ0ehar/3ce207kC23MxvDFq45O0QAkRq5JvLy8EBYWZvbl5WVbwRjRPxGLGNw62h/+gwb2cEX+YXMmvwqF1VzouzfFl5NcItQkyMQiTDXjutoZxtGRjbt6JjrCR0Z4kTY2xE2InAHAEBut0SViEeYM40QX74ZaVteC2iYtGAYI7qW6JEvw3Uqddenw0ZGqhlZIRAxW3hBpdr9ZBqO9PWkldpsRwxevuqlkVnW38CnXi2bESFpRLWqauIjIpwezTNqby+qahQGFji5eBUiMEAQxgOHFCD8A0NtF3ivdQsbcNykEHmo5/hrv361Cx/sM0ZGLRbU9Eh1J5y3yDQ81hmFMCou7M6flhhhfAMAf54uh07dFRQa7KnssDWNP+FQTwHUIWaqnGhXoikEqKaoaWnEip/NpzdZibb0ID183kmqmiPVYVrnw79LaZmw/2eZbw/u6RPm6dBmB6Q1IjBAEMWBxVclMfqmbK17taQLdVTjx0ky8uSC2W8f3ZHREp2/rpDFOX80Z5oNAN1WH961l4hB3aJRSlNU140R2RQcb+L7OUG9njA91w2BXJR7txG1XIhYJ3ir2StW0GZ513tbL05nXyDGDyR5vZPjx/stCBOdgH0rRACRGCIIY4IQb/WXvCDFiD4yjI/b0HcmtaECzVg+5RIQAt7aUnVjE4MM7R+HBqaG4ZaT5OTOdIRWLhPqY388VdbCB7+swDIOtD07AgZXXdRk1mBXNpbYTUovtMo281Mq2Xp4oQ3tvfmUjaox8T1iWxTFDzc7r82Pg5iRDbkUDfj1bCJZlcdAQGekLKRqAxAhBEAMcUzHSP/4yb0/7zhp7PPQACEW94d7qDl0+sf4avDAvqttpFb5Fdue5IsHjZYiF4tW+ijWdT1PCPSGTiIShjFeLNUPyjNGopMJgR+N5U5fL6lFW1wKZRIRxoW7428RgANwU5MzSOhRWN0EmFmFscMeBj46AxAhBEAOaMKNoSH+NjADA/ZNDIZOIcLGoFtnl5ts4bYUvXu2J+zI53ANquQRFNU3CX+H9JU1jC05yCSYN4aZC2yNVY63hmTH8gDvjjho+KjIywBVyiRhLJgTDSSbGxaJavP7LBQBAfPCgDgMmHQWJkQFCcHAwNm7c6OhlEESfwzgy0p/bmDUqKeIMVvXHDQ+aq8Xc4EB7oZCKcb3BZp+3hrfU1tvf4btqzJnTldQ2Yee5QqujWbbWjADGtvBtkRFejIwzDP/TqKS4ezzne7M/nbOAn9xH6kUAEiMOZfr06VixYoVdznX8+HE8+OCDdjkXQQwkYgdr4KtRYHKYh91tu3ubeENInZ/+e7X0tPfKPKOuFJVMDB8bHrD9CX4K8+m8KiHNAgAV9S24dUsSln170upanxIrh+QZ02YL3zEyMjbEXXjv/skhJtOXp4T17FBMWyAx0odhWRZabUfXPHN4enpCpRrYnhEE0R2c5BIcWHkdvrl/rKOXctWMtaMYadHqBYt2vq3X3kyL8ILSUHMS4uHkcJfPnsLLRSF4s+y6wM0gatXp8ci/TiKvgpsUzaeqOoNlWaGA1RZLfGNbeL2eRX5lA65UNUIiYjAqyNVknbcZfG8GqaQY5udi9TV6moEnRlgWaKl3zMuGorKlS5di//79eP/998EwDBiGwZdffgmGYfDHH38gPj4ecrkciYmJyMzMxPz58+Ht7Q21Wo0xY8Zg165dJudrn6ZhGAaffvopFixYAJVKhfDwcPz8889WrU2n0+H+++9HSEgIlEolhg4divfff7/Dfp9//jmGDRsGuVwOX19fLF++XNhWVVWFBx98EN7e3lAoFIiJicEvv/xi9f0hCHsiFYsGxINwVNAgMAyQXd5gMlW2O2SX10OrZ4WBgj2BUibGdZHcX98DsV7EmNnRXPfQLsN05jd/vYDDl9t8Pk5kd+1DUtusRVMrl9LqamKvMca28DkVDUJUJGawBiqZaTTw0evDMCLAFQ9PD+sV11tr6d8xS3O0NgBr/Bxz7RcKAJl1OdH3338f6enpiImJESzYz58/DwBYuXIl3n33XYSGhsLV1RX5+fmYN28e3njjDSgUCnz11Ve4+eabkZaWhsDAQIvXWLVqFdatW4d33nkHmzZtwl133YWcnBy4uXVePa3X6+Hv749t27bBw8MDSUlJePDBB+Hr64vbb78dALBlyxY8+eSTeOuttzB37lxUV1fj0KFDwvFz585FbW0tvv32WwwZMgSpqakQi/tGoRRB9Fc0SimGejvjYlEtTmRXYl6sb7fPxXdehHure1SoPXJdGAqqmnDXOMu/qwYCM6O88c4faTh4qQxfHMrCl0nZAIA3F8TgxR/PIa24FtUNrdCoLJvu8e6rznKJTYWlvC38mfxqXCys6VAvYoyvRomfHplkwyfrHQaeGOknaDQayGQyqFQq+PhwedWLFy8C4GbHzJo1S9jX3d0dcXFxwtdvvPEGfvzxR/z8888m0Yj2LF26FIsXLwYArFmzBps2bcKxY8dwww03dLo2qVSKVatWCV+HhIQgKSkJ27ZtE8TIG2+8gaeeegqPP/64sN+YMWMAALt27cKxY8dw4cIFRERwU1BDQ3tuVDxBXEuMCXbDxaJaHM+uuCoxktFLs3qG+Wn6OvJOuwAAG2FJREFU5MPP3kR4qxHopkJuRQNW/S8VAPDEzAjcNS4InyZmIausHidzK3FdpOWRG4L7ahcD8swRaRAjF4zEyFgzYqSvMvDEiFTFRSgcdW07EB8fb/J1fX09Vq1ahV9++QUFBQXQarVobGxEbm5up+cZPny48G8nJyc4OzujpKTEqjV89NFH+PTTT5GTk4PGxka0tLRgxIgRAICSkhIUFBRgxowZZo9NSUmBv7+/IEQIgrAfY0Lc8M2RnKuuG+nJTpprEYZhMDPKG58fygLA+aw8en0YACA+aBCyyupxPLuiUzFiq+GZMXzdyIGMMlwuqwfDtBU89wcGnhhhGKtTJX0VJyfT9T/zzDP4448/8O677yIsLAxKpRK33XYbWlpaOj2PVGoaDmQYBnp918Octm3bhieeeALr16/HhAkT4OzsjHfeeQdHjx4FACiVyk6P72o7QRDdZ0zwIABAakEN6pq13e4QSudt4HuoePVaZP4IP3yRlIUoHxe8+9c4oSZjTLAb/pOc32XdCJ+msaVehIfvqEnJqwLAObP29hymq2HgiZF+hEwmg06n63K/xMRELF26FAsWLAAA1NXVITs7u8fWlZiYiIkTJ+Lhhx8W3svMzBT+7ezsjODgYOzevRvXXXddh+OHDx+O/Px8pKenU3SEIOyMr0YJ/0FK5Fc24mROZYdJwCU1Tahr1nZaMNrUqkN2OddJE95PXWn7InEBrtj71HT4aBQmzrVjDOmSlPwqNGt1kEvM14N0p62Xh/ca4elPKRpgIHbT9COCg4Nx9OhRZGdno6yszGLUIiwsDDt27EBKSgpOnz6NO++806oIR3cJCwvDiRMn8McffyA9PR0vv/wyjh8/brLPa6+9hvXr1+ODDz5ARkYGTp48iU2bNgEApk2bhqlTp+LWW29FQkICsrKy8Pvvv2Pnzp09tmaCuJYYYwi/n2iXqmlo0WL+h4cw670DOJxZbu5QAMClkjqwLNfe6dkHJrYOJII9nDpY6Ae7q+ChlqFFq8e5K9UWj20zPLP9e+KqksHXqCvKXPFqX4bEiAN5+umnIRaLER0dDU9PT4s1IO+99x4GDRqEiRMn4uabb8acOXMwatSoHlvXsmXLsHDhQixatAjjxo1DeXm5SZQEAO69915s3LgRmzdvxrBhw3DTTTchIyND2L59+3aMGTMGixcvRnR0NFauXGlVFIggiK7hxcixdmLky6RsFFY3Qadn8eh3p0wMuIxJN6oXGQgtz30dhmEQH8R7xFhO1VxNmgZoqxsB2qIx/QWGtdfEpR6kpqYGGo0G1dXVcHExNWlpampCVlYWQkJCoFAMTHe/awn6fhJE12QU12LWewegkIpw5tU5kElEqG5sxZS396CmSQuNUorqxlaMDXbDvx4YB6nY9O/Otb9fwMf7L2PJhCCsnh/joE9xbfFp4mW88esFzIzywqf3jjG7z4z1+5BZWo9//30cJnZjmu66nRexeV8mhng6YfdT069yxfahs+e3MRQZIQiC6GeEeakxSCVFU6se5wq4sP+niZdR06RFhLca2/8xEWq5BMeyK/DuH2kmx9Y2tQqtn/15Vk9/Q0it5VRCrzcfA7iaNA0AzI3xhVwiwqIxAd1bpAMhMXINsmzZMqjVarOvZcuWOXp5BEF0AcMwGB3UVjdSVteMzw5yLaVPzhqKMC811t3GtfZ/fOAy/jhfhMYWHT7an4kp6/biVG4VRAzXckr0DtF+LlBKxahqaMWl0roO25tadaht4sZ/eKq7FxWO9dcg7Y25eHDqkKtaqyOgbpprkNWrV+Ppp582u62zMBpBEH2HsSGDsOtCMY5lVaKouhkNLToM99dgzjDOlnxerC/unxyCzw5m4eltp6GQiQUfi1BPJzx3Q6RJjQHRs0jFIowMdEVSZjmOZ1d08HfhvzcyiQguymvv0XztfWICXl5e8PKybLxDEETfhze0Onq5HAcMI+GfmTPUpCD1ubmROJ1XhRM5laht1sJ/kBIrZkbglhF+kIgpMN7bxAe7ISmzHCeyK3HXuCCTbcZtvddiUTGJEYIgiH5IjJ8GCqkItc1caH98qBsmtyt6lIpF2HzXKGxISEfMYA1ujw8wGSFP9C68YZ0599y2Tpprs9WaxAhBEEQ/RCYRYUSAK45c5h5s7aMiPF4uCrx16/AO7xO9z8jAQRAxQH5lIwqrG+GraXOrFopXu9nW298hiUwQBNFPmTSEi4RcH+klFLQSfRe1XIJoP65Op701vJCm6WYnTX+HIiMEQRD9lL9PCYW7Wo4br2J6L9G7xAe54dyVGpzIrsDNcX7C+9d6moYiIwRBEP0UpUyMO8cFQqPqPwPRrnV4v5Ejlytg7DlKaRqCIAiCIHqF8aFukElESCuuRWJGmfA+L0Y8r9E0DYkRBzJ9+nSsWLHCbudbunQpbrnlFrudjyAIgrAv7mo57hoXCABYn5AuREdKr2Ji70CAxAhBEARB9CL/mD4ESqkYp/OqsOdiCbQ6PcrrWwBQmmbAwLIsGlobHPKyZebg0qVLsX//frz//vtgGAYMwyA7OxupqamYN28e1Go1vL29cc8996CsrC2U98MPPyA2NhZKpRLu7u6YOXMm6uvr8dprr+Grr77Cf//7X+F8+/bt63Idzz77LCIiIqBSqRAaGoqXX34Zra2tJvv8/PPPiI+Ph0KhgIeHBxYuXChsa25uxsqVKxEQEAC5XI7w8HB89tlnVt8HgiCIaw0vZwWWTORMzzYkpKOsrgUsC4gYwM1J5uDVOYYB103TqG3EuH+Pc8i1j955FCqpyqp933//faSnpyMmJgarV68GAOh0OkybNg0PPPAANmzYgMbGRjz77LO4/fbbsWfPHhQWFmLx4sVYt24dFixYgNraWiQmJoJlWTz99NO4cOECampq8MUXXwAA3Ny6bvVzdnbGl19+CT8/P5w9exYPPPAAnJ2dsXLlSgDAr7/+ioULF+LFF1/EN998g5aWFvz666/C8UuWLMHhw4fxwQcfIC4uDllZWSbiiSAIgujIQ1OH4NvDOThfUINvjmQDADzUcohF1577KjAAxUh/QaPRQCaTQaVSwcfHBwDwyiuvYNSoUVizZo2w3+eff46AgACkp6ejrq4OWq0WCxcuRFAQp6pjY2OFfZVKJZqbm4XzWcNLL70k/Ds4OBhPPfUUvv/+e0GMvPnmm7jjjjuwatUqYb+4uDgAQHp6OrZt24aEhATMnDkTABAaGmrrrSAIgrjmcHOS4b7JIdi05xI+OXAZwLXrMQIMQDGilChx9M6jDrv21ZCcnIy9e/dCrVZ32JaZmYnZs2djxowZiI2NxZw5czB79mzcdtttGDSo+5M3f/jhB2zcuBGXLl0SxI7xsLyUlBQ88MADZo9NSUmBWCzGtGnTun19giCIa5W/TwnFV0nZqDFM671W60WAAShGGIaxOlXS19Dr9bj55pvx9ttvd9jm6+sLsViMhIQEJCUl4c8//8SmTZvw4osv4ujRowgJCbH5ekeOHBGiHnPmzIFGo8HWrVuxfv16YR+l0rLA6mwbQRAE0TkapRQPTAnF+oR0ANduJw0wAAtY+xMymQw6nU74etSoUTh//jyCg4MRFhZm8nJycgLAia1JkyZh1apVOHXqFGQyGX788Uez5+uKQ4cOISgoCC+++CLi4+MRHh6OnJwck32GDx+O3bt3mz0+NjYWer0e+/fvt/WjEwRBEAD+NjkEgwymdSRGCIcQHByMo0ePIjs7G2VlZXjkkUdQUVGBxYsX49ixY7h8+TL+/PNP3HfffdDpdDh69CjWrFmDEydOIDc3Fzt27EBpaSmioqKE8505cwZpaWkoKyvr0BXTnrCwMOTm5mLr1q3IzMzEBx98IAgbnldffRXfffcdXn31VVy4cAFnz57FunXrhOvde++9uO+++/DTTz8hKysL+/btw7Zt23rmhhEEQQww1HIJXvvLMAS5qzB7mPX1fgMOth9QXV3NAmCrq6s7bGtsbGRTU1PZxsZGB6zs6khLS2PHjx/PKpVKFgCblZXFpqenswsWLGBdXV1ZpVLJRkZGsitWrGD1ej2bmprKzpkzh/X09GTlcjkbERHBbtq0SThfSUkJO2vWLFatVrMA2L1793a5hmeeeYZ1d3dn1Wo1u2jRIva9995jNRqNyT7bt29nR4wYwcpkMtbDw4NduHChsK2xsZF94oknWF9fX1Ymk7FhYWHs559/3u170p+/nwRBEIQpnT2/jWFY1gZzDAdRU1MDjUaD6upqk+JKAGhqakJWVhZCQkKgUFy7xT8DBfp+EgRBDBw6e34bQ2kagiAIgiAcSrfEyObNm4W/XEePHo3ExESL+x48eBCTJk2Cu7s7lEolIiMj8d5773V7wYT1rFmzBmq12uxr7ty5jl4eQRAEQQDoRmvv999/jxUrVmDz5s2YNGkSPv74Y8ydOxepqakIDAzssL+TkxOWL1+O4cOHw8nJCQcPHsRDDz0EJycnPPjgg3b5EIR5li1bhttvv93sNmrLJQiCIPoKNteMjBs3DqNGjcKWLVuE96KionDLLbdg7dq1Vp1j4cKFcHJywjfffGN2e3NzM5qbm4Wva2pqEBAQQDUj1wD0/SQIghg49EjNSEtLC5KTkzF79myT92fPno2kpCSrznHq1CkkJSV16tq5du1aaDQa4RUQENDleftBHS5hBfR9JAiCuPawSYyUlZVBp9PB29vb5H1vb28UFRV1eqy/vz/kcjni4+PxyCOP4O9//7vFfZ9//nlUV1cLr7y8PIv7SqWcWUxDQ4MNn4Toq/DfR/77ShAEQQx8umUHzzCmUwVZlu3wXnsSExNRV1eHI0eO4LnnnkNYWBgWL15sdl+5XA653DonOrFYDFdXV5SUlAAAVCpVl2sh+h4sy6KhoQElJSVwdXWFWCx29JIIgiCIXsImMeLh4QGxWNwhClJSUtIhWtIefnZKbGwsiouL8dprr1kUI7bCT6nlBQnRf3F1dbVp6jBBEATR/7FJjMhkMowePRoJCQlYsGCB8H5CQgLmz59v9XlYljUpUL1aGIaBr68vvLy8urRAJ/ouUqmUIiIEQRDXIDanaZ588kncc889iI+Px4QJE/DJJ58gNzcXy5YtA8DVe1y5cgVff/01AODDDz9EYGAgIiMjAXC+I++++y4effRRO34MDrFYTA8zgiAIguhn2CxGFi1ahPLycqxevRqFhYWIiYnBb7/9hqCgIABAYWEhcnNzhf31ej2ef/55ZGVlQSKRYMiQIXjrrbfw0EMP2e9TEARBEATRb+n3s2kIgiAIguib0GwagiAIgiD6Bd1q7e1t+OBNTU2Ng1dCEARBEIS18M/trpIw/UKM1NbWAoBVTqwEQRAEQfQtamtrodFoLG7vFzUjer0eBQUFcHZ2vipDM37GTV5eHtWe9DB0r3sPute9B93r3oPude/Rk/eaZVnU1tbCz88PIpHlypB+ERkRiUTw9/e32/lcXFzoh7uXoHvde9C97j3oXvcedK97j566151FRHiogJUgCIIgCIdCYoQgCIIgCIdyTYkRuVyOV1991eohfET3oXvde9C97j3oXvcedK97j75wr/tFAStBEARBEAOXayoyQhAEQRBE34PECEEQBEEQDoXECEEQBEEQDoXECEEQBEEQDuWaESObN29GSEgIFAoFRo8ejcTEREcvqd+zdu1ajBkzBs7OzvDy8sItt9yCtLQ0k31YlsVrr70GPz8/KJVKTJ8+HefPn3fQigcOa9euBcMwWLFihfAe3Wv7ceXKFdx9991wd3eHSqXCiBEjkJycLGyne20ftFotXnrpJYSEhECpVCI0NBSrV6+GXq8X9qF73X0OHDiAm2++GX5+fmAYBj/99JPJdmvubXNzMx599FF4eHjAyckJf/nLX5Cfn2//xbLXAFu3bmWlUin7z3/+k01NTWUff/xx1snJic3JyXH00vo1c+bMYb/44gv23LlzbEpKCnvjjTeygYGBbF1dnbDPW2+9xTo7O7Pbt29nz549yy5atIj19fVla2pqHLjy/s2xY8fY4OBgdvjw4ezjjz8uvE/32j5UVFSwQUFB7NKlS9mjR4+yWVlZ7K5du9hLly4J+9C9tg9vvPEG6+7uzv7yyy9sVlYW+5///IdVq9Xsxo0bhX3oXnef3377jX3xxRfZ7du3swDYH3/80WS7Nfd22bJl7ODBg9mEhAT25MmT7HXXXcfGxcWxWq3Wrmu9JsTI2LFj2WXLlpm8FxkZyT733HMOWtHApKSkhAXA7t+/n2VZltXr9ayPjw/71ltvCfs0NTWxGo2G/eijjxy1zH5NbW0tGx4eziYkJLDTpk0TxAjda/vx7LPPspMnT7a4ne61/bjxxhvZ++67z+S9hQsXsnfffTfLsnSv7Ul7MWLNva2qqmKlUim7detWYZ8rV66wIpGI3blzp13XN+DTNC0tLUhOTsbs2bNN3p89ezaSkpIctKqBSXV1NQDAzc0NAJCVlYWioiKTey+XyzFt2jS6993kkUcewY033oiZM2eavE/32n78/PPPiI+Px1//+ld4eXlh5MiR+Oc//ylsp3ttPyZPnozdu3cjPT0dAHD69GkcPHgQ8+bNA0D3uiex5t4mJyejtbXVZB8/Pz/ExMTY/f73i0F5V0NZWRl0Oh28vb1N3vf29kZRUZGDVjXwYFkWTz75JCZPnoyYmBgAEO6vuXufk5PT62vs72zduhUnT57E8ePHO2yje20/Ll++jC1btuDJJ5/ECy+8gGPHjuGxxx6DXC7HkiVL6F7bkWeffRbV1dWIjIyEWCyGTqfDm2++icWLFwOgn+uexJp7W1RUBJlMhkGDBnXYx97PzwEvRngYhjH5mmXZDu8R3Wf58uU4c+YMDh482GEb3furJy8vD48//jj+/PNPKBQKi/vRvb569Ho94uPjsWbNGgDAyJEjcf78eWzZsgVLliwR9qN7ffV8//33+Pbbb/Hvf/8bw4YNQ0pKClasWAE/Pz/ce++9wn50r3uO7tzbnrj/Az5N4+HhAbFY3EHFlZSUdFCERPd49NFH8fPPP2Pv3r3w9/cX3vfx8QEAuvd2IDk5GSUlJRg9ejQkEgkkEgn279+PDz74ABKJRLifdK+vHl9fX0RHR5u8FxUVhdzcXAD0c21PnnnmGTz33HO44447EBsbi3vuuQdPPPEE1q5dC4DudU9izb318fFBS0sLKisrLe5jLwa8GJHJZBg9ejQSEhJM3k9ISMDEiRMdtKqBAcuyWL58OXbs2IE9e/YgJCTEZHtISAh8fHxM7n1LSwv2799P995GZsyYgbNnzyIlJUV4xcfH46677kJKSgpCQ0PpXtuJSZMmdWhRT09PR1BQEAD6ubYnDQ0NEIlMH0NisVho7aV73XNYc29Hjx4NqVRqsk9hYSHOnTtn//tv13LYPgrf2vvZZ5+xqamp7IoVK1gnJyc2Ozvb0Uvr1/zjH/9gNRoNu2/fPrawsFB4NTQ0CPu89dZbrEajYXfs2MGePXuWXbx4MbXl2QnjbhqWpXttL44dO8ZKJBL2zTffZDMyMth//etfrEqlYr/99lthH7rX9uHee+9lBw8eLLT27tixg/Xw8GBXrlwp7EP3uvvU1tayp06dYk+dOsUCYDds2MCeOnVKsLWw5t4uW7aM9ff3Z3ft2sWePHmSvf7666m192r48MMP2aCgIFYmk7GjRo0S2k+J7gPA7OuLL74Q9tHr9eyrr77K+vj4sHK5nJ06dSp79uxZxy16ANFejNC9th//+9//2JiYGFYul7ORkZHsJ598YrKd7rV9qKmpYR9//HE2MDCQVSgUbGhoKPviiy+yzc3Nwj50r7vP3r17zf6Ovvfee1mWte7eNjY2ssuXL2fd3NxYpVLJ3nTTTWxubq7d18qwLMvaN9ZCEARBEARhPQO+ZoQgCIIgiL4NiRGCIAiCIBwKiRGCIAiCIBwKiRGCIAiCIBwKiRGCIAiCIBwKiRGCIAiCIBwKiRGCIAiCIBwKiRGCIAiCIBwKiRGCIPod+/btA8MwqKqqcvRSCIKwAyRGCIIgCIJwKCRGCIIgCIJwKCRGCIKwGZZlsW7dOoSGhkKpVCIuLg4//PADgLYUyq+//oq4uDgoFAqMGzcOZ8+eNTnH9u3bMWzYMMjlcgQHB2P9+vUm25ubm7Fy5UoEBARALpcjPDwcn332mck+ycnJiI+Ph0qlwsSJE5GWltazH5wgiB6BxAhBEDbz0ksv4YsvvsCWLVtw/vx5PPHEE7j77ruxf/9+YZ9nnnkG7777Lo4fPw4vLy/85S9/QWtrKwBORNx+++244447cPbsWbz22mt4+eWX8eWXXwrHL1myBFu3bsUHH3yACxcu4KOPPoJarTZZx4svvoj169fjxIkTkEgkuO+++3rl8xMEYWfsPgeYIIgBTV1dHatQKNikpCST9++//3528eLFwtjyrVu3CtvKy8tZpVLJfv/99yzLsuydd97Jzpo1y+T4Z555ho2OjmZZlmXT0tJYAGxCQoLZNfDX2LVrl/Der7/+ygJgGxsb7fI5CYLoPSgyQhCETaSmpqKpqQmzZs2CWq0WXl9//TUyMzOF/SZMmCD8283NDUOHDsWFCxcAABcuXMCkSZNMzjtp0iRkZGRAp9MhJSUFYrEY06ZN63Qtw4cPF/7t6+sLACgpKbnqz0gQRO8icfQCCILoX+j1egDAr7/+isGDB5tsk8vlJoKkPQzDAOBqTvh/87AsK/xbqVRatRapVNrh3Pz6CILoP1BkhCAIm4iOjoZcLkdubi7CwsJMXgEBAcJ+R44cEf5dWVmJ9PR0REZGCuc4ePCgyXmTkpIQEREBsViM2NhY6PV6kxoUgiAGLhQZIQjCJpydnfH000/jiSeegF6vx+TJk1FTU4OkpCSo1WoEBQUBAFavXg13d3d4e3vjxRdfhIeHB2655RYAwFNPPYUxY8bg9ddfx6JFi3D48GH83//9HzZv3gwACA4Oxr333ov77rsPH3zwAeLi4pCTk4OSkhLcfvvtjvroBEH0ECRGCIKwmddffx1eXl5Yu3YtLl++DFdXV4waNQovvPCCkCZ566238PjjjyMjIwNxcXH4+eefIZPJAACjRo3Ctm3b8Morr+D111+Hr68vVq9ejaVLlwrX2LJlC1544QU8/PDDKC8vR2BgIF544QVHfFyCIHoYhjVO1BIEQVwl+/btw3XXXYfKykq4uro6ejkEQfQDqGaEIAiCIAiHQmKEIAiCIAiHQmkagiAIgiAcCkVGCIIgCIJwKCRGCIIgCIJwKCRGCIIgCIJwKCRGCIIgCIJwKCRGCIIgCIJwKCRGCIIgCIJwKCRGCIIgCIJwKCRGCIIgCIJwKP8PAqNBsxsnugUAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def train(epoch):\n",
    "    model.train()\n",
    "    loss_all = 0\n",
    "    for data in train_loader:\n",
    "        data = data.to(device)\n",
    "        optimizer.zero_grad()\n",
    "        output = model(data)\n",
    "        loss = F.nll_loss(output, data.y)\n",
    "        loss.backward()\n",
    "        loss_all += loss.item() * data.num_graphs\n",
    "        optimizer.step()\n",
    "    return loss_all / len(train_X)\n",
    "def test(loader):\n",
    "    model.eval()\n",
    "    correct = 0\n",
    "    for data in loader:\n",
    "        data = data.to(device)\n",
    "        output = model(data)\n",
    "        pred = output.max(dim=1)[1]\n",
    "        correct += pred.eq(data.y).sum().item()\n",
    "    return correct / len(loader.dataset)\n",
    "hist = {\"loss\":[], \"acc\":[], \"test_acc\":[]}\n",
    "for epoch in range(1, 101):\n",
    "    train_loss = train(epoch)\n",
    "    train_acc = test(train_loader)\n",
    "    test_acc = test(test_loader)\n",
    "    hist[\"loss\"].append(train_loss)\n",
    "    hist[\"acc\"].append(train_acc)\n",
    "    hist[\"test_acc\"].append(test_acc)\n",
    "    print(f'Epoch: {epoch}, Train loss: {train_loss:.3}, Train_acc: {train_acc:.3}, Test_acc: {test_acc:.3}')\n",
    "ax = plt.subplot(1,1,1)\n",
    "ax.plot([e for e in range(1,101)], hist[\"loss\"], label=\"train_loss\")\n",
    "ax.plot([e for e in range(1,101)], hist[\"acc\"], label=\"train_acc\")\n",
    "ax.plot([e for e in range(1,101)], hist[\"test_acc\"], label=\"test_acc\")\n",
    "plt.xlabel(\"epoch\")\n",
    "ax.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "465cde87",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}