_CLF_W2V_MLP.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "### On top of the Embeddedings gained from Bertweet, I added simple Neural network to "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "from torch.utils.data import TensorDataset, DataLoader\n",
    "import torchvision\n",
    "import torchvision.transforms as transforms\n",
    "from torch.utils.data import TensorDataset, DataLoader, RandomSampler, SequentialSampler\n",
    "from torch.optim import lr_scheduler\n",
    "\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.metrics import f1_score\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No GPU available, using the CPU instead.\n"
     ]
    }
   ],
   "source": [
    "if torch.cuda.is_available():       \n",
    "    device = torch.device(\"cuda\")\n",
    "    print(f'There are {torch.cuda.device_count()} GPU(s) available.')\n",
    "    print('Device name:', torch.cuda.get_device_name(0))\n",
    "\n",
    "else:\n",
    "    print('No GPU available, using the CPU instead.')\n",
    "    device = torch.device(\"cpu\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "pheme = pd.read_csv('./data/_PHEME_text_AVGw2v.csv').drop(['token'],axis=1)\n",
    "pheme_y = pd.read_csv('./data/_PHEME_target.csv').target\n",
    "pheme_event = pd.read_csv('./data/_PHEME_text.csv').Event\n",
    "\n",
    "ext = pd.read_csv('./data/_PHEMEext_text_AVGw2v.csv').drop(['token'],axis=1)\n",
    "ext_y = pd.read_csv('./data/_PHEMEext_text.csv').target\n",
    "ext_event = pd.read_csv('./data/_PHEMEext_text.csv').Event\n",
    "\n",
    "rhi = pd.read_csv('./data/_RHI_text_AVGw2v.csv').drop(['token'],axis=1)\n",
    "rhi_y = pd.read_csv('./data/_RHI_target.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "### Data with BERTweet Embedding consists of 797 Dimensions: which are 767 Embeddings and 30 additional features\n",
    "\n",
    "### Other Dataset consists of 328 dimensions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Print to Test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'data' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-7-a9fb23772048>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"Shape of the Train and test data of Averaged Word2Vec: {}/{}\"\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mformat\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtest_data\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0;31m# print(\"Shape of the Train and test data of Doc2vec: {}/{}\".format(train_doc.shape, test_doc.shape))\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0;31m# print(\"\\nShape of the validation data of Avg: {}\".format(valid_avg.shape))\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0;31m# print(\"Shape of the validation data of Doc2vec: {}\".format(valid_doc.shape))\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0;31m# print(\"\\nShape of the data w/ BERTweet: {}\".format(df_bertweet.shape))\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mNameError\u001b[0m: name 'data' is not defined"
     ]
    }
   ],
   "source": [
    "print(\"Shape of the Train and test data of Averaged Word2Vec: {}/{}\".format(data.shape, test_data.shape))\n",
    "# print(\"Shape of the Train and test data of Doc2vec: {}/{}\".format(train_doc.shape, test_doc.shape))\n",
    "# print(\"\\nShape of the validation data of Avg: {}\".format(valid_avg.shape))\n",
    "# print(\"Shape of the validation data of Doc2vec: {}\".format(valid_doc.shape))\n",
    "# print(\"\\nShape of the data w/ BERTweet: {}\".format(df_bertweet.shape))\n",
    "# print(\"\\nShape of the validation data w/ BERTweet: {}\".format(df_valid_bertweet.shape))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(5802, 1) (5227, 1)\n"
     ]
    }
   ],
   "source": [
    "# Dropping Target values from the dataset\n",
    "# train_y = train_avg.isRumor\n",
    "# test_y = test_avg.isRumor\n",
    "# valid_y = valid_avg.isRumor\n",
    "# df_bertweet_y = df_bertweet.isRumor\n",
    "# df_valid_bertweet_y = df_valid_bertweet.isRumor\n",
    "# for dataset in [train_avg, test_avg, train_doc, test_doc, valid_avg, valid_doc, df_bertweet, df_valid_bertweet]:\n",
    "#     dataset.drop(['isRumor'], axis=1, inplace=True)\n",
    "\n",
    "# print(train.shape, test_y.shape, valid_y.shape, df_bertweet_y.shape, df_valid_bertweet_y.shape)\n",
    "print(y.shape, test_y.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Base dataset are the baseline feature set to be inputted to the model\n",
    "# # Here, 4 features are dropped for their lack of predictive power\n",
    "\n",
    "# train_avg_base = train_avg.drop(['hasURL', 'hasUserURL', 'isNotOnlyText', 'char_count'],axis=1)\n",
    "# test_avg_base = test_avg.drop(['hasURL', 'hasUserURL', 'isNotOnlyText', 'char_count'],axis=1)\n",
    "# valid_avg_base = valid_avg.drop(['hasURL', 'hasUserURL', 'isNotOnlyText', 'char_count'],axis=1)\n",
    "# train_doc_base = train_doc.drop(['hasURL', 'hasUserURL', 'isNotOnlyText', 'char_count'],axis=1)\n",
    "# test_doc_base = test_doc.drop(['hasURL', 'hasUserURL', 'isNotOnlyText', 'char_count'],axis=1)\n",
    "# valid_doc_base = valid_doc.drop(['hasURL', 'hasUserURL', 'isNotOnlyText', 'char_count'],axis=1)\n",
    "# bertweet_base = df_bertweet.drop(['hasURL', 'hasUserURL', 'isNotOnlyText', 'char_count'],axis=1)\n",
    "# bertweet_valid_base = df_valid_bertweet.drop(['hasURL', 'hasUserURL', 'isNotOnlyText', 'char_count'],axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Torch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## BERTCLF 노트북을 참고한 방법 - Train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_inputs = torch.tensor(pheme.values).float()\n",
    "val_inputs = torch.tensor(ext.values).float()\n",
    "\n",
    "# Convert other data types to torch.Tensor\n",
    "train_labels = torch.tensor(pheme_y.values).float()\n",
    "val_labels = torch.tensor(ext_y.values).float()\n",
    "\n",
    "# For fine-tuning BERT, the authors recommend a batch size of 16 or 32.\n",
    "batch_size = 16\n",
    "\n",
    "# Create the DataLoader for our training set\n",
    "train_data = TensorDataset(train_inputs, train_labels)\n",
    "train_sampler = RandomSampler(train_data)\n",
    "train_dataloader = DataLoader(train_data, sampler=train_sampler, batch_size=batch_size)\n",
    "\n",
    "# Create the DataLoader for our validation set\n",
    "val_data = TensorDataset(val_inputs, val_labels)\n",
    "val_sampler = SequentialSampler(val_data)\n",
    "val_dataloader = DataLoader(val_data, sampler=val_sampler, batch_size=batch_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Length of the Inputs are:\n",
      "Train\ttorch.Size([5802, 200])\n",
      "Val\ttorch.Size([485, 200])\n"
     ]
    }
   ],
   "source": [
    "print(\"Length of the Inputs are:\\nTrain\\t%s\\nVal\\t%s\" %(train_inputs.size(), val_inputs.size()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "\n",
    "class AvgW2VNet(torch.nn.Module):\n",
    "  def __init__(self, freeze_model=False):\n",
    "    \"\"\"\n",
    "    In the constructor we construct three nn.Linear instances that we will use\n",
    "    in the forward pass.\n",
    "    \"\"\"\n",
    "    super(AvgW2VNet, self).__init__()\n",
    "\n",
    "    D_in, H, D_out = 200, 16, 1\n",
    "\n",
    "    self.classifier = nn.Sequential(\n",
    "        nn.Linear(D_in, H),\n",
    "        nn.BatchNorm1d(H),\n",
    "        nn.ReLU(),\n",
    "        nn.Dropout(0.2),\n",
    "        nn.Linear(H, D_out),\n",
    "        nn.Sigmoid()\n",
    "    )\n",
    "\n",
    "    # if freeze_bert:\n",
    "    #     for param in self.bert.parameters():\n",
    "    #         param.requires_grad = False\n",
    "\n",
    "  def forward(self, input):\n",
    "    result = self.classifier(input)\n",
    "    return result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# from torch.nn.optim import Adam\n",
    "\n",
    "def initialize_model(epochs=300):\n",
    "    \"\"\"Initialize the Bert Classifier, the optimizer and the learning rate scheduler.\n",
    "    \"\"\"\n",
    "    # Instantiate Bert Classifier\n",
    "    model = AvgW2VNet()\n",
    "\n",
    "    # Tell PyTorch to run the model on GPU\n",
    "    model.to(device)\n",
    "\n",
    "    # Create the optimizer\n",
    "    optimizer = optim.Adam(model.parameters(), lr=0.1, weight_decay=0.1)\n",
    "\n",
    "    criterion = nn.BCELoss()\n",
    "    # loss = nn.CrossEntropyLoss()\n",
    "\n",
    "    # Total number of training steps\n",
    "    total_steps = len(train_dataloader) * epochs\n",
    "\n",
    "    # Set up the learning rate scheduler\n",
    "    # scheduler = get_linear_schedule_with_warmup(optimizer,\n",
    "    #                                             num_warmup_steps=0, # Default value\n",
    "    #                                             num_training_steps=total_steps)\n",
    "    scheduler = lr_scheduler.StepLR(optimizer, step_size=0.1, gamma= 0.99)       \n",
    "\n",
    "    return model, optimizer, criterion, scheduler\n",
    "    # return bert_classifier, optimizer, scheduler\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "import time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'set_seed' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-13-01d45872d52b>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mset_seed\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m42\u001b[0m\u001b[0;34m)\u001b[0m    \u001b[0;31m# Set seed for reproducibility\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moptimizer\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcriterion\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mscheduler\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0minitialize_model\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mepochs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m100\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0mcriterion\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mBCELoss\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0;31m# criterion = nn.BCEWithLogitsLoss()\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0;31m# criterion = nn.CrossEntropyLoss()\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mNameError\u001b[0m: name 'set_seed' is not defined"
     ]
    }
   ],
   "source": [
    "set_seed(42)    # Set seed for reproducibility\n",
    "model, optimizer, criterion, scheduler = initialize_model(epochs=100)\n",
    "criterion = nn.BCELoss()\n",
    "# criterion = nn.BCEWithLogitsLoss()\n",
    "# criterion = nn.CrossEntropyLoss()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 356,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Start training...\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "   1    |    -    |   0.639528   |  0.610964  |   25.60   |   0.53   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "   2    |    -    |   0.647421   |  0.771703  |   25.60   |   0.48   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "   3    |    -    |   0.647474   |  0.784059  |   25.60   |   0.48   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "   4    |    -    |   0.648490   |  0.875346  |   25.60   |   0.48   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "   5    |    -    |   0.648081   |  0.878954  |   25.60   |   0.49   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "   6    |    -    |   0.645979   |  0.753151  |   25.60   |   0.59   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "   7    |    -    |   0.643810   |  0.814137  |   25.60   |   0.48   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "   8    |    -    |   0.646240   |  0.859356  |   25.60   |   0.49   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "   9    |    -    |   0.643474   |  0.869719  |   25.60   |   0.53   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  10    |    -    |   0.647404   |  0.886819  |   25.60   |   0.53   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  11    |    -    |   0.643991   |  0.793549  |   25.60   |   0.54   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  12    |    -    |   0.643471   |  0.747277  |   25.60   |   0.48   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  13    |    -    |   0.640912   |  1.118812  |   25.60   |   0.57   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  14    |    -    |   0.646596   |  0.849899  |   25.60   |   0.55   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  15    |    -    |   0.647087   |  0.913701  |   25.60   |   0.54   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  16    |    -    |   0.646527   |  0.790857  |   25.60   |   0.48   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  17    |    -    |   0.646991   |  0.815333  |   25.60   |   0.49   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  18    |    -    |   0.646528   |  0.818208  |   25.60   |   0.55   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  19    |    -    |   0.646458   |  0.827993  |   25.60   |   0.49   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  20    |    -    |   0.647680   |  0.818778  |   25.60   |   0.48   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      "Training complete!\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": "<Figure size 1296x720 with 1 Axes>"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": "<Figure size 1296x720 with 1 Axes>"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "train(model, train_dataloader, val_dataloader, epochs=20, evaluation=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.save(model.state_dict(), './MLP_W2V_to_Sigmoid_ord.pt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "PATH = './MLP_W2V_to_Sigmoid_ord.pt'\n",
    "# bn_state_dict = torch.load('./BERT_raw_to_fine_tune_ord.pt')\n",
    "# bert_classifier.load_state_dict(bn_state_dict)\n",
    "model.load_state_dict(torch.load(PATH))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 370,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import accuracy_score, roc_curve, auc\n",
    "\n",
    "def evaluate_roc(probs, y_true):\n",
    "    \"\"\"\n",
    "    - Print AUC and accuracy on the test set\n",
    "    - Plot ROC\n",
    "    @params    probs (np.array): an array of predicted probabilities with shape (len(y_true), 2)\n",
    "    @params    y_true (np.array): an array of the true values with shape (len(y_true),)\n",
    "    \"\"\"\n",
    "    preds = probs[:, 1]\n",
    "    fpr, tpr, threshold = roc_curve(y_true, preds)\n",
    "    roc_auc = auc(fpr, tpr)\n",
    "    print(f'AUC: {roc_auc:.4f}')\n",
    "       \n",
    "    # Get accuracy over the test set\n",
    "    y_pred = np.where(preds >= 0.5, 1, 0)\n",
    "    accuracy = accuracy_score(y_true, y_pred)\n",
    "    print(f'Accuracy: {accuracy*100:.2f}%')\n",
    "    \n",
    "    # Plot ROC AUC\n",
    "    plt.title('Receiver Operating Characteristic')\n",
    "    plt.plot(fpr, tpr, 'b', label = 'AUC = %0.2f' % roc_auc)\n",
    "    plt.legend(loc = 'lower right')\n",
    "    plt.plot([0, 1], [0, 1],'r--')\n",
    "    plt.xlim([0, 1])\n",
    "    plt.ylim([0, 1])\n",
    "    plt.ylabel('True Positive Rate')\n",
    "    plt.xlabel('False Positive Rate')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 388,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch.nn.functional as F\n",
    "\n",
    "def model_predict(model, test_dataloader):\n",
    "    \"\"\"Perform a forward pass on the trained BERT model to predict probabilities\n",
    "    on the test set.\n",
    "    \"\"\"\n",
    "    # Put the model into the evaluation mode. The dropout layers are disabled during\n",
    "    # the test time.\n",
    "    model.eval()\n",
    "\n",
    "    all_logits = []\n",
    "\n",
    "    # For each batch in our test set...\n",
    "    for batch in test_dataloader:\n",
    "        # Load batch to GPU\n",
    "        b_input,b_labels = tuple(t.to(device) for t in batch)\n",
    "        # print(b_input)\n",
    "\n",
    "        # Compute logits\n",
    "        with torch.no_grad():\n",
    "            logits = model(b_input)\n",
    "        all_logits.append(logits)\n",
    "    \n",
    "    # Concatenate logits from each batch\n",
    "    all_logits = torch.cat(all_logits, dim=0)\n",
    "\n",
    "    # Apply softmax to calculate probabilities\n",
    "    probs = F.softmax(all_logits, dim=1).cpu()\n",
    "\n",
    "    return probs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 425,
   "metadata": {},
   "outputs": [
    {
     "ename": "IndexError",
     "evalue": "index 1 is out of bounds for axis 1 with size 1",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mIndexError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-425-6a157c3629fe>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m     10\u001b[0m \u001b[0mtarget\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtarget\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0munsqueeze\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdim\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnumpy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     11\u001b[0m \u001b[0;31m# print(target)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 12\u001b[0;31m \u001b[0mevaluate_roc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mprobs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtarget\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m<ipython-input-370-bde1be800687>\u001b[0m in \u001b[0;36mevaluate_roc\u001b[0;34m(probs, y_true)\u001b[0m\n\u001b[1;32m      8\u001b[0m     \u001b[0;34m@\u001b[0m\u001b[0mparams\u001b[0m    \u001b[0my_true\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0man\u001b[0m \u001b[0marray\u001b[0m \u001b[0mof\u001b[0m \u001b[0mthe\u001b[0m \u001b[0mtrue\u001b[0m \u001b[0mvalues\u001b[0m \u001b[0;32mwith\u001b[0m \u001b[0mshape\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0my_true\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      9\u001b[0m     \"\"\"\n\u001b[0;32m---> 10\u001b[0;31m     \u001b[0mpreds\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mprobs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     11\u001b[0m     \u001b[0mfpr\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtpr\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mthreshold\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mroc_curve\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0my_true\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mpreds\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     12\u001b[0m     \u001b[0mroc_auc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mauc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfpr\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtpr\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mIndexError\u001b[0m: index 1 is out of bounds for axis 1 with size 1"
     ]
    }
   ],
   "source": [
    "# Compute predicted probabilities on the test set\n",
    "probs = model_predict(model, val_dataloader)\n",
    "probs = probs.numpy()\n",
    "# print(probs)\n",
    "# print(type(probs))\n",
    "\n",
    "# # Evaluate the Bert classifier\n",
    "# print(len(probs))\n",
    "target = torch.tensor(ext_y)\n",
    "target = target.unsqueeze(dim=1).numpy()\n",
    "# print(target)\n",
    "# evaluate_roc(probs, target)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.eval()\n",
    "correct = 0\n",
    "total = 0\n",
    "test_loss = 0\n",
    "outputs_list = []\n",
    "y_list = []\n",
    "\n",
    "with torch.no_grad():\n",
    "    val_loss = 0\n",
    "\n",
    "    for i, data in enumerate(test_loader):\n",
    "        x, y = data\n",
    "        x, y = x.float(), y.long()\n",
    "        outputs = task1_model(x)\n",
    "        loss = criterion(outputs, y.unsqueeze(1).float())\n",
    "        _, predicted = torch.max(outputs.data, 1)\n",
    "        outputs_list.append(predicted[:])\n",
    "        total += y.size(0)\n",
    "        correct += (predicted == y).double().sum().item()\n",
    "        val_loss += loss.item()\n",
    "        y_list.append(y)\n",
    "\n",
    "print('Accuracy of the test dataset is: %d %%' % (100 * correct / total))\n",
    "print(\"Loss of validation set: {:.5f}\".format((val_loss / test_size)))\n",
    "acc = (100 * correct / total)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## adf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "task1_model.eval()\n",
    "correct = 0\n",
    "total = 0\n",
    "test_loss = 0\n",
    "outputs_list = []\n",
    "y_list = []\n",
    "\n",
    "with torch.no_grad():\n",
    "    val_loss = 0\n",
    "\n",
    "    for i, data in enumerate(task1_test_dataloader):\n",
    "        x, y = data\n",
    "        x, y = x.float(), y.long()\n",
    "        outputs = task1_model(x)\n",
    "        loss = criterion(outputs, y.unsqueeze(1).float())\n",
    "        _, predicted = torch.max(outputs.data, 1)\n",
    "        outputs_list.append(predicted[:])\n",
    "        total += y.size(0)\n",
    "        correct += (predicted == y).double().sum().item()\n",
    "        val_loss += loss.item()\n",
    "        y_list.append(y)\n",
    "\n",
    "print('Accuracy of the test dataset is: %d %%' % (100 * correct / total))\n",
    "print(\"Loss of validation set: {:.5f}\".format((val_loss / test_size)))\n",
    "acc = (100 * correct / total)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<bound method Module.parameters of FC_net(\n",
      "  (fc1): Linear(in_features=792, out_features=130, bias=True)\n",
      "  (fc2): Linear(in_features=130, out_features=60, bias=True)\n",
      "  (fc3): Linear(in_features=60, out_features=1, bias=True)\n",
      "  (drop_2): Dropout(p=0.2, inplace=False)\n",
      ")>\n"
     ]
    }
   ],
   "source": [
    "print(task1_model.parameters)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch.optim as optim\n",
    "\n",
    "criterion = nn.BCELoss()\n",
    "# optimizer = optim.SGD(task1_model.parameters(), lr=0.001, momentum=0.9)\n",
    "optimizer = optim.Adam(task1_model.parameters(), lr=0.01)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_loss = []\n",
    "train_accuracy = []\n",
    "\n",
    "prev_loss = 10\n",
    "PATH = \"./state_dict_BERT_fc.pt\"\n",
    "best_acc = 10.0\n",
    "num_epochs = 10\n",
    "\n",
    "val_corrects_list = []\n",
    "val_loss_list = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 0/9\n",
      "----------\n",
      "Train) Loss: 0.3190 Acc: 0.0042\n",
      "Epoch 1/9\n",
      "----------\n",
      "Train) Loss: 5.4705 Acc: 0.6325\n",
      "Epoch 2/9\n",
      "----------\n",
      "Train) Loss: 5.7149 Acc: 0.6568\n",
      "Epoch 3/9\n",
      "----------\n",
      "Train) Loss: 5.7213 Acc: 0.6568\n",
      "Epoch 4/9\n",
      "----------\n",
      "Train) Loss: 5.7149 Acc: 0.6568\n",
      "Epoch 5/9\n",
      "----------\n",
      "Train) Loss: 5.7213 Acc: 0.6568\n",
      "Epoch 6/9\n",
      "----------\n",
      "Train) Loss: 5.7149 Acc: 0.6568\n",
      "Epoch 7/9\n",
      "----------\n",
      "Train) Loss: 5.7149 Acc: 0.6568\n",
      "Epoch 8/9\n",
      "----------\n",
      "Train) Loss: 5.7213 Acc: 0.6568\n",
      "Epoch 9/9\n",
      "----------\n",
      "Train) Loss: 5.7213 Acc: 0.6568\n"
     ]
    }
   ],
   "source": [
    "for epoch in range(num_epochs):  # loop over the dataset multiple times\n",
    "    print('Epoch {}/{}'.format(epoch, num_epochs - 1))\n",
    "    print('-' * 10)\n",
    "\n",
    "    running_loss = 0.0\n",
    "    running_corrects = 0\n",
    "    task1_model.train()  # Set model to training mode\n",
    "    for i, data in enumerate(task1_train_dataloader, 0):\n",
    "        # get the inputs; data is a list of [inputs, labels]\n",
    "        inputs, labels = data\n",
    "        # inputs, labels = inputs.float(), labels.long()\n",
    "\n",
    "        # zero the parameter gradients\n",
    "        optimizer.zero_grad()\n",
    "\n",
    "        # forward + backward + optimize\n",
    "        outputs = task1_model(inputs)\n",
    "\n",
    "        labels = labels.unsqueeze(1).float()\n",
    "        loss = criterion(outputs, labels)\n",
    "        loss.backward()\n",
    "        optimizer.step()    \n",
    "\n",
    "        running_loss += loss.item()\n",
    "        running_corrects += torch.sum(outputs == labels.data)\n",
    "        # print(running_corrects)\n",
    "\n",
    "    epoch_loss = running_loss / train_size\n",
    "    epoch_acc = running_corrects.double() / train_size\n",
    "    train_loss.append(epoch_loss)\n",
    "    train_accuracy.append(epoch_acc)\n",
    "\n",
    "    print('Train) Loss: {:.4f} Acc: {:.4f}'.format(epoch_loss, epoch_acc))\n",
    "\n",
    "    # if epoch_loss < best_acc:\n",
    "    #     # print(\"prev_loss: {:.5f}\".format(prev_loss))\n",
    "    #     # print(\"loss: {:.5f}\".format(loss))\n",
    "    #     print(\"Saving the best model w/ loss {:.4f}\".format(epoch_loss))\n",
    "    #     torch.save(task1_model.state_dict(),PATH)\n",
    "    #     best_acc = epoch_loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy of the test dataset is: 68 %\n",
      "Loss of validation set: 5.20883\n"
     ]
    }
   ],
   "source": [
    "task1_model.eval()\n",
    "correct = 0\n",
    "total = 0\n",
    "test_loss = 0\n",
    "outputs_list = []\n",
    "y_list = []\n",
    "\n",
    "with torch.no_grad():\n",
    "    val_loss = 0\n",
    "\n",
    "    for i, data in enumerate(task1_test_dataloader):\n",
    "        x, y = data\n",
    "        x, y = x.float(), y.long()\n",
    "        outputs = task1_model(x)\n",
    "        loss = criterion(outputs, y.unsqueeze(1).float())\n",
    "        _, predicted = torch.max(outputs.data, 1)\n",
    "        outputs_list.append(predicted[:])\n",
    "        total += y.size(0)\n",
    "        correct += (predicted == y).double().sum().item()\n",
    "        val_loss += loss.item()\n",
    "        y_list.append(y)\n",
    "\n",
    "print('Accuracy of the test dataset is: %d %%' % (100 * correct / total))\n",
    "print(\"Loss of validation set: {:.5f}\".format((val_loss / test_size)))\n",
    "acc = (100 * correct / total)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# W2V to MLP"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import random\n",
    "import time\n",
    "\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "from torch.utils.data import TensorDataset, DataLoader\n",
    "import torchvision\n",
    "import torchvision.transforms as transforms\n",
    "from torch.utils.data import TensorDataset, DataLoader, RandomSampler, SequentialSampler\n",
    "from torch.optim import lr_scheduler\n",
    "\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.metrics import f1_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No GPU available, using the CPU instead.\n"
     ]
    }
   ],
   "source": [
    "if torch.cuda.is_available():       \n",
    "    device = torch.device(\"cuda\")\n",
    "    print(f'There are {torch.cuda.device_count()} GPU(s) available.')\n",
    "    print('Device name:', torch.cuda.get_device_name(0))\n",
    "\n",
    "else:\n",
    "    print('No GPU available, using the CPU instead.')\n",
    "    device = torch.device(\"cpu\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "pheme = pd.read_csv('./data/_PHEME_text_AVGw2v.csv').drop(['token'],axis=1)\n",
    "pheme_y = pd.read_csv('./data/_PHEME_target.csv').target\n",
    "pheme_event = pd.read_csv('./data/_PHEME_text.csv').Event\n",
    "\n",
    "ext = pd.read_csv('./data/_PHEMEext_text_AVGw2v.csv').drop(['token'],axis=1)\n",
    "ext_y = pd.read_csv('./data/_PHEMEext_text.csv').target\n",
    "ext_event = pd.read_csv('./data/_PHEMEext_text.csv').Event\n",
    "\n",
    "rhi = pd.read_csv('./data/_RHI_text_AVGw2v.csv').drop(['token'],axis=1)\n",
    "rhi_y = pd.read_csv('./data/_RHI_target.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Length of the Inputs are:\n",
      "Train\ttorch.Size([5802, 200])\n",
      "Val\ttorch.Size([485, 200])\n"
     ]
    }
   ],
   "source": [
    "train_inputs = torch.tensor(pheme.values).float()\n",
    "val_inputs = torch.tensor(ext.values).float()\n",
    "\n",
    "# Convert other data types to torch.Tensor\n",
    "train_labels = torch.tensor(pheme_y.values).float()\n",
    "val_labels = torch.tensor(ext_y.values).float()\n",
    "\n",
    "# For fine-tuning BERT, the authors recommend a batch size of 16 or 32.\n",
    "batch_size = 32\n",
    "\n",
    "# Create the DataLoader for our training set\n",
    "train_data = TensorDataset(train_inputs, train_labels)\n",
    "train_sampler = RandomSampler(train_data)\n",
    "train_dataloader = DataLoader(train_data, sampler=train_sampler, batch_size=batch_size)\n",
    "\n",
    "# Create the DataLoader for our validation set\n",
    "val_data = TensorDataset(val_inputs, val_labels)\n",
    "val_sampler = SequentialSampler(val_data)\n",
    "val_dataloader = DataLoader(val_data, sampler=val_sampler, batch_size=batch_size)\n",
    "\n",
    "print(\"Length of the Inputs are:\\nTrain\\t%s\\nVal\\t%s\" %(train_inputs.size(), val_inputs.size()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "class AvgW2VNet(torch.nn.Module):\n",
    "  def __init__(self, freeze_model=False):\n",
    "    \"\"\"\n",
    "    In the constructor we construct three nn.Linear instances that we will use\n",
    "    in the forward pass.\n",
    "    \"\"\"\n",
    "    super(AvgW2VNet, self).__init__()\n",
    "\n",
    "    D_in, H, D_out = 200, 16, 1\n",
    "\n",
    "    self.classifier = nn.Sequential(\n",
    "        nn.Linear(D_in, H),\n",
    "        nn.BatchNorm1d(H),\n",
    "        nn.ReLU(),\n",
    "        nn.Dropout(0.2),\n",
    "        nn.Linear(H, D_out),\n",
    "        nn.Sigmoid()\n",
    "    )\n",
    "\n",
    "    # if freeze_bert:\n",
    "    #     for param in self.bert.parameters():\n",
    "    #         param.requires_grad = False\n",
    "\n",
    "  def forward(self, input):\n",
    "    result = self.classifier(input)\n",
    "    return result\n",
    "\n",
    "def initialize_model(epochs=300):\n",
    "    \"\"\"Initialize the Bert Classifier, the optimizer and the learning rate scheduler.\n",
    "    \"\"\"\n",
    "    # Instantiate Bert Classifier\n",
    "    model = AvgW2VNet()\n",
    "\n",
    "    # Tell PyTorch to run the model on GPU\n",
    "    model.to(device)\n",
    "\n",
    "    # Create the optimizer\n",
    "    optimizer = optim.Adam(model.parameters(), lr=0.1, weight_decay=0.1)\n",
    "\n",
    "    criterion = nn.BCELoss()\n",
    "    # loss = nn.CrossEntropyLoss()\n",
    "\n",
    "    # Total number of training steps\n",
    "    total_steps = len(train_dataloader) * epochs\n",
    "\n",
    "    # Set up the learning rate scheduler\n",
    "    # scheduler = get_linear_schedule_with_warmup(optimizer,\n",
    "    #                                             num_warmup_steps=0, # Default value\n",
    "    #                                             num_training_steps=total_steps)\n",
    "    scheduler = lr_scheduler.StepLR(optimizer, step_size=0.1, gamma= 0.99)       \n",
    "\n",
    "    return model, optimizer, criterion, scheduler\n",
    "    # return bert_classifier, optimizer, scheduler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "def set_seed(seed_value=42):\n",
    "    \"\"\"Set seed for reproducibility.\n",
    "    \"\"\"\n",
    "    random.seed(seed_value)\n",
    "    np.random.seed(seed_value)\n",
    "    torch.manual_seed(seed_value)\n",
    "    torch.cuda.manual_seed_all(seed_value)\n",
    "\n",
    "def train(model, train_dataloader, val_dataloader=None, epochs=4, evaluation=False):\n",
    "    \"\"\"Train the BertClassifier model.\n",
    "    \"\"\"\n",
    "    # Start training loop\n",
    "    print(\"Start training...\\n\")\n",
    "    val_loss_list = []\n",
    "    val_acc_list = []\n",
    "    for epoch_i in range(epochs):\n",
    "        # =======================================\n",
    "        #               Training\n",
    "        # =======================================\n",
    "        # Print the header of the result table\n",
    "        print(f\"{'Epoch':^7} | {'Batch':^7} | {'Train Loss':^12} | {'Val Loss':^10} | {'Val Acc':^9} | {'Elapsed':^9}\")\n",
    "        # print(\"-\"*70)\n",
    "\n",
    "        # Measure the elapsed time of each epoch\n",
    "        t0_epoch, t0_batch = time.time(), time.time()\n",
    "\n",
    "        # Reset tracking variables at the beginning of each epoch\n",
    "        total_loss, batch_loss, batch_counts = 0, 0, 0\n",
    "\n",
    "        # Put the model into the training mode\n",
    "        model.train()\n",
    "\n",
    "        # For each batch of training data...\n",
    "        for step, batch in enumerate(train_dataloader):\n",
    "            batch_counts +=1\n",
    "            # Load batch to GPU\n",
    "            b_input, b_labels = tuple(t.to(device) for t in batch)\n",
    "\n",
    "            # Zero out any previously calculated gradients\n",
    "            model.zero_grad()\n",
    "\n",
    "            # Perform a forward pass. This will return logits.\n",
    "            logits = model(b_input)\n",
    "\n",
    "            # Compute loss and accumulate the loss values\n",
    "            # torch.argmax(logits(i)) for logit, i in enumerate(logits)\n",
    "            \n",
    "            logits = logits.flatten()\n",
    "            loss = criterion(logits, b_labels)\n",
    "            batch_loss += loss.item()\n",
    "            total_loss += loss.item()\n",
    "\n",
    "            # Perform a backward pass to calculate gradients\n",
    "            loss.backward()\n",
    "\n",
    "            # Clip the norm of the gradients to 1.0 to prevent \"exploding gradients\"\n",
    "            torch.nn.utils.clip_grad_norm_(model.parameters(), 1.0)\n",
    "\n",
    "            # Update parameters and the learning rate\n",
    "            optimizer.step()\n",
    "            scheduler.step()\n",
    "\n",
    "            # Print the loss values and time elapsed for every 20 batches\n",
    "            if (step == len(train_dataloader) - 1):\n",
    "                # Calculate time elapsed for 20 batches\n",
    "                time_elapsed = time.time() - t0_batch\n",
    "\n",
    "                # Print training results\n",
    "                # print(f\"{epoch_i + 1:^7} | {step:^7} | {batch_loss / batch_counts:^12.6f} | {'-':^10} | {'-':^9} | {time_elapsed:^9.2f}\")\n",
    "\n",
    "                # Reset batch tracking variables\n",
    "                batch_loss, batch_counts = 0, 0\n",
    "                t0_batch = time.time()\n",
    "\n",
    "        # Calculate the average loss over the entire training data\n",
    "        avg_train_loss = total_loss / len(train_dataloader)\n",
    "\n",
    "        print(\"-\"*70)\n",
    "        # =======================================\n",
    "        #               Evaluation\n",
    "        # =======================================\n",
    "        if evaluation == True:\n",
    "            # After the completion of each training epoch, measure the model's performance on our validation set.\n",
    "            val_loss, val_accuracy = evaluate(model, val_dataloader)\n",
    "            val_loss_list.append(val_loss)\n",
    "            val_acc_list.append(val_accuracy)\n",
    "\n",
    "            # Print performance over the entire training data\n",
    "            time_elapsed = time.time() - t0_epoch\n",
    "            # print(val_loss_list)\n",
    "            print(f\"{epoch_i + 1:^7} | {'-':^7} | {avg_train_loss:^12.6f} | {val_loss:^10.6f} | {val_accuracy:^9.2f} | {time_elapsed:^9.2f}\")\n",
    "            print(\"-\"*70)\n",
    "        print(\"\\n\")\n",
    "\n",
    "    fig = plt.figure(figsize=(18,10))\n",
    "    plt.plot(np.arange(0,epochs),val_loss_list)\n",
    "    fig2 = plt.figure(figsize=(18,10))\n",
    "    plt.plot(np.arange(0,epochs),val_acc_list)\n",
    "    print(\"Training complete!\")\n",
    "\n",
    "def evaluate(model, val_dataloader):\n",
    "    \"\"\"After the completion of each training epoch, measure the model's performance\n",
    "    on our validation set.\n",
    "    \"\"\"\n",
    "    # Put the model into the evaluation mode. The dropout layers are disabled during\n",
    "    # the test time.\n",
    "    model.eval()\n",
    "\n",
    "    # Tracking variables\n",
    "    val_accuracy = []\n",
    "    val_loss = []\n",
    "\n",
    "    # For each batch in our validation set...\n",
    "    for batch in val_dataloader:\n",
    "        # Load batch to GPU\n",
    "        b_input, b_labels = tuple(t.to(device) for t in batch)\n",
    "\n",
    "        # Compute logits\n",
    "        with torch.no_grad():\n",
    "            logits = model(b_input)\n",
    "            # logits = logits.flatten()\n",
    "\n",
    "        # Compute loss\n",
    "        loss = criterion(logits.flatten(), b_labels)\n",
    "        val_loss.append(loss.item())\n",
    "\n",
    "        # Get the predictions\n",
    "        # print(logits.flatten())\n",
    "        preds = logits >= 0.5\n",
    "        # print(preds.flatten())\n",
    "\n",
    "        # Calculate the accuracy rate\n",
    "        accuracy = (preds == b_labels).cpu().numpy().mean() * 100\n",
    "        val_accuracy.append(accuracy)\n",
    "\n",
    "    # Compute the average accuracy and loss over the validation set.\n",
    "    val_loss = np.mean(val_loss)\n",
    "    val_accuracy = np.mean(val_accuracy)\n",
    "\n",
    "    return val_loss, val_accuracy\n",
    "\n",
    "def evaluate_roc(probs, y_true):\n",
    "    \"\"\"\n",
    "    - Print AUC and accuracy on the test set\n",
    "    - Plot ROC\n",
    "    @params    probs (np.array): an array of predicted probabilities with shape (len(y_true), 2)\n",
    "    @params    y_true (np.array): an array of the true values with shape (len(y_true),)\n",
    "    \"\"\"\n",
    "    preds = probs[:, 1]\n",
    "    fpr, tpr, threshold = roc_curve(y_true, preds)\n",
    "    roc_auc = auc(fpr, tpr)\n",
    "    print(f'AUC: {roc_auc:.4f}')\n",
    "       \n",
    "    # Get accuracy over the test set\n",
    "    y_pred = np.where(preds >= 0.5, 1, 0)\n",
    "    accuracy = accuracy_score(y_true, y_pred)\n",
    "    print(f'Accuracy: {accuracy*100:.2f}%')\n",
    "    \n",
    "    # Plot ROC AUC\n",
    "    plt.title('Receiver Operating Characteristic')\n",
    "    plt.plot(fpr, tpr, 'b', label = 'AUC = %0.2f' % roc_auc)\n",
    "    plt.legend(loc = 'lower right')\n",
    "    plt.plot([0, 1], [0, 1],'r--')\n",
    "    plt.xlim([0, 1])\n",
    "    plt.ylim([0, 1])\n",
    "    plt.ylabel('True Positive Rate')\n",
    "    plt.xlabel('False Positive Rate')\n",
    "    plt.show()\n",
    "\n",
    "def model_predict(model, test_dataloader):\n",
    "    \"\"\"Perform a forward pass on the trained BERT model to predict probabilities\n",
    "    on the test set.\n",
    "    \"\"\"\n",
    "    # Put the model into the evaluation mode. The dropout layers are disabled during\n",
    "    # the test time.\n",
    "    model.eval()\n",
    "\n",
    "    all_logits = []\n",
    "\n",
    "    # For each batch in our test set...\n",
    "    for batch in test_dataloader:\n",
    "        # Load batch to GPU\n",
    "        b_input,b_labels = tuple(t.to(device) for t in batch)\n",
    "        # print(b_input)\n",
    "\n",
    "        # Compute logits\n",
    "        with torch.no_grad():\n",
    "            logits = model(b_input)\n",
    "        all_logits.append(logits)\n",
    "    \n",
    "    # Concatenate logits from each batch\n",
    "    all_logits = torch.cat(all_logits, dim=0)\n",
    "\n",
    "    # Apply softmax to calculate probabilities\n",
    "    probs = F.softmax(all_logits, dim=1).cpu()\n",
    "\n",
    "    return probs\n",
    "\n",
    "def testing_process(bert_classifier, X_val, y_val):\n",
    "    #  Run `preprocessing_for_bert` on the test set\n",
    "    print('Tokenizing data...')\n",
    "    test_inputs, test_masks = preprocessing_for_bert(X_val)\n",
    "\n",
    "    # Create the DataLoader for our test set\n",
    "    test_dataset = TensorDataset(test_inputs)\n",
    "    test_sampler = SequentialSampler(test_dataset)\n",
    "    test_dataloader = DataLoader(test_dataset, sampler=test_sampler, batch_size=32)\n",
    "\n",
    "    # Compute predicted probabilities on the test set\n",
    "    probs = bert_predict(bert_classifier, test_dataloader)\n",
    "\n",
    "    # Get predictions from the probabilities\n",
    "    threshold = 0.5\n",
    "    preds = np.where(probs[:, 1] > threshold, 1, 0)\n",
    "\n",
    "    # Number of tweets predicted non-negative\n",
    "    print(\"Number of tweets predicted as Rumor: \", preds.sum())\n",
    "\n",
    "    preds = np.argmax(probs, axis = 1)\n",
    "    print('Accuracy Score:\\t',accuracy_score(y_val, preds))\n",
    "    print('Precision Score:\\t', str(precision_score(y_val,preds)))\n",
    "    print('Recall Score:\\t\\t' + str(recall_score(y_val,preds)))\n",
    "    print('F1 Score:\\t',f1_score(y_val, preds, zero_division=1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Start training...\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "   1    |    -    |   0.625833   |  0.818905  |   27.93   |   0.24   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "   2    |    -    |   0.630503   |  0.726403  |   27.93   |   0.20   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "   3    |    -    |   0.637972   |  0.785952  |   27.93   |   0.30   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "   4    |    -    |   0.637503   |  0.876912  |   27.93   |   0.20   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "   5    |    -    |   0.636795   |  0.795632  |   27.93   |   0.31   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "   6    |    -    |   0.641695   |  0.793410  |   27.93   |   0.23   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "   7    |    -    |   0.647039   |  0.824728  |   27.93   |   0.31   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "   8    |    -    |   0.647247   |  0.829863  |   27.93   |   0.28   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "   9    |    -    |   0.647049   |  0.755864  |   27.93   |   0.26   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  10    |    -    |   0.647271   |  0.806786  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  11    |    -    |   0.645872   |  0.821857  |   27.93   |   0.24   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  12    |    -    |   0.647203   |  0.808231  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  13    |    -    |   0.647042   |  0.870187  |   27.93   |   0.28   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  14    |    -    |   0.646081   |  0.731604  |   27.93   |   0.24   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  15    |    -    |   0.641666   |  0.756556  |   27.93   |   0.23   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  16    |    -    |   0.643175   |  0.927687  |   27.93   |   0.24   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  17    |    -    |   0.642479   |  0.799591  |   27.93   |   0.23   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  18    |    -    |   0.637692   |  1.269078  |   27.93   |   0.23   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  19    |    -    |   0.635175   |  0.864799  |   27.93   |   0.23   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  20    |    -    |   0.640331   |  0.981267  |   27.93   |   0.26   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  21    |    -    |   0.638440   |  0.762799  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  22    |    -    |   0.637942   |  0.839429  |   27.93   |   0.23   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  23    |    -    |   0.640615   |  0.852383  |   27.93   |   0.23   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  24    |    -    |   0.633888   |  0.755869  |   27.93   |   0.26   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  25    |    -    |   0.643349   |  0.824528  |   27.93   |   0.25   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  26    |    -    |   0.644395   |  0.913133  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  27    |    -    |   0.646439   |  0.875998  |   27.93   |   0.24   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  28    |    -    |   0.644989   |  0.790695  |   27.93   |   0.28   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  29    |    -    |   0.643697   |  0.890981  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  30    |    -    |   0.646246   |  0.892282  |   27.93   |   0.25   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  31    |    -    |   0.645971   |  0.773487  |   27.93   |   0.24   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  32    |    -    |   0.646809   |  0.832520  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  33    |    -    |   0.646401   |  0.834457  |   27.93   |   0.23   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  34    |    -    |   0.641048   |  0.848086  |   27.93   |   0.23   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  35    |    -    |   0.645153   |  0.808226  |   27.93   |   0.26   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  36    |    -    |   0.645408   |  0.847751  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  37    |    -    |   0.639216   |  0.789688  |   27.93   |   0.23   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  38    |    -    |   0.635907   |  0.798173  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  39    |    -    |   0.641000   |  0.878758  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  40    |    -    |   0.645996   |  0.792904  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  41    |    -    |   0.645962   |  0.839677  |   27.93   |   0.26   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  42    |    -    |   0.643605   |  0.831533  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  43    |    -    |   0.640593   |  0.733573  |   27.93   |   0.23   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  44    |    -    |   0.644848   |  0.912575  |   27.93   |   0.26   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  45    |    -    |   0.641433   |  0.888075  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  46    |    -    |   0.640401   |  0.823264  |   27.93   |   0.27   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  47    |    -    |   0.646238   |  0.806412  |   27.93   |   0.25   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  48    |    -    |   0.646889   |  0.770704  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  49    |    -    |   0.642800   |  1.617158  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  50    |    -    |   0.642829   |  0.790340  |   27.93   |   0.23   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  51    |    -    |   0.642838   |  0.826624  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  52    |    -    |   0.643718   |  0.785409  |   27.93   |   0.23   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  53    |    -    |   0.640354   |  0.756534  |   27.93   |   0.28   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  54    |    -    |   0.642836   |  0.793468  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  55    |    -    |   0.636111   |  0.855268  |   27.93   |   0.24   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  56    |    -    |   0.637412   |  0.786505  |   27.93   |   0.25   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  57    |    -    |   0.642381   |  0.815196  |   27.93   |   0.29   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  58    |    -    |   0.636727   |  0.839657  |   27.93   |   0.23   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  59    |    -    |   0.640781   |  0.807747  |   27.93   |   0.26   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  60    |    -    |   0.646833   |  0.863636  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  61    |    -    |   0.644212   |  0.856328  |   27.93   |   0.23   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  62    |    -    |   0.644620   |  0.805490  |   27.93   |   0.27   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  63    |    -    |   0.642586   |  0.772529  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  64    |    -    |   0.638425   |  0.833255  |   27.93   |   0.26   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  65    |    -    |   0.643744   |  0.986252  |   27.93   |   0.21   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  66    |    -    |   0.645873   |  0.873477  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  67    |    -    |   0.640759   |  0.943265  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  68    |    -    |   0.640340   |  0.726604  |   27.93   |   0.26   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  69    |    -    |   0.639568   |  0.865714  |   27.93   |   0.26   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  70    |    -    |   0.637699   |  0.748586  |   28.92   |   0.26   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  71    |    -    |   0.645709   |  0.797234  |   27.93   |   0.25   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  72    |    -    |   0.643618   |  0.875587  |   27.93   |   0.26   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  73    |    -    |   0.645396   |  0.960840  |   27.93   |   0.25   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  74    |    -    |   0.636591   |  0.655081  |   75.39   |   0.26   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  75    |    -    |   0.639425   |  0.828418  |   27.93   |   0.27   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  76    |    -    |   0.639558   |  0.853486  |   27.93   |   0.23   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  77    |    -    |   0.645732   |  0.808325  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  78    |    -    |   0.643891   |  0.830487  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  79    |    -    |   0.646469   |  0.783216  |   27.93   |   0.24   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  80    |    -    |   0.646667   |  0.833110  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  81    |    -    |   0.644853   |  1.089943  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  82    |    -    |   0.645923   |  0.809933  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  83    |    -    |   0.638074   |  0.747266  |   27.93   |   0.23   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  84    |    -    |   0.641662   |  0.741405  |   27.93   |   0.23   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  85    |    -    |   0.640662   |  0.809182  |   27.93   |   0.29   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  86    |    -    |   0.642800   |  0.798641  |   27.93   |   0.21   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  87    |    -    |   0.646119   |  0.780812  |   27.93   |   0.19   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  88    |    -    |   0.646703   |  0.873645  |   27.93   |   0.24   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  89    |    -    |   0.643728   |  0.761585  |   27.93   |   0.19   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  90    |    -    |   0.641387   |  0.798949  |   27.93   |   0.19   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  91    |    -    |   0.643310   |  0.819039  |   27.93   |   0.20   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  92    |    -    |   0.646521   |  0.785057  |   27.93   |   0.19   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  93    |    -    |   0.643852   |  0.805425  |   27.93   |   0.21   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  94    |    -    |   0.643555   |  0.775573  |   27.93   |   0.20   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  95    |    -    |   0.643410   |  0.808128  |   32.58   |   0.20   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  96    |    -    |   0.641841   |  0.815496  |   27.93   |   0.19   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  97    |    -    |   0.646069   |  0.846401  |   27.93   |   0.20   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  98    |    -    |   0.642153   |  0.804605  |   27.93   |   0.20   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  99    |    -    |   0.644219   |  0.852138  |   27.93   |   0.22   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
      "----------------------------------------------------------------------\n",
      "  100   |    -    |   0.639209   |  0.956005  |   27.93   |   0.27   \n",
      "----------------------------------------------------------------------\n",
      "\n",
      "\n",
      "Training complete!\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABBEAAAI/CAYAAAAhjUEXAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAAC8e0lEQVR4nOzdd3wkB33///dsUy+ndlW66qu2784+d2NsqkMAE2pIIRAIISGQfEOSb+o3JN98Q34phJCEJECAAKEZCB0bDO79bF9vvn6STvW0qrva2Z35/TE7q7bSzhZpV9Lr+XjkEU5aSXNn7e7Mez7FsG1bAAAAAAAAmfiKfQAAAAAAAGBxIEQAAAAAAACeECIAAAAAAABPCBEAAAAAAIAnhAgAAAAAAMATQgQAAAAAAOBJoFg/uKmpyd6wYUOxfjwAAAAAAJjFc88912fbdvP0jxctRNiwYYP2799frB8PAAAAAABmYRjGhXQfp50BAAAAAAB4QogAAAAAAAA8IUQAAAAAAACeECIAAAAAAABPCBEAAAAAAIAnhAgAAAAAAMATQgQAAAAAAOAJIQIAAAAAAPCEEAEAAAAAAHhCiAAAAAAAADwhRAAAAAAAAJ4QIgAAAAAAAE8IEQAAAAAAgCeECAAAAAAAwBNCBAAAAAAA4AkhAgAAAAAA8IQQAQAAAAAAeEKIAAAAAAAAPCFEAAAAAAAAnhAiAAAAAAAATwgRAAAAAACAJ4QIAAAAAADAE0IEAAAgSRoZj+tdn31GHeFIsQ8FAACUKEIEAAAgSTrXO6oHT/bq4KVwsQ8FAACUKEIEAAAgSYolLEmSmfz/AAAA0xEiAAAASVI8FSLYRT4SAABQqggRAACAJCluOeEBlQgAAGA2hAgAAEDSRDtDnBABAADMImOIYBjGZwzD6DEM48gcj7nTMIwDhmEcNQzj4cIeIgAAWAjxZBtDjHYGAAAwCy+VCJ+TdPdsnzQMo17SJyS93rbtXZLeUpAjAwAAC8qkEgEAAGSQMUSwbfsRSVfmeMgvSPqmbdsXk4/vKdCxAQCABWSynQEAAGRQiJkIWyWtMAzjIcMwnjMM4x0F+J4AAGCBmbQzAACADAIF+h7XS3q5pApJTxqG8ZRt26emP9AwjPdKeq8ktbW1FeBHAwCAQonTzgAAADIoRCVCu6T7bdsetW27T9Ijknane6Bt25+0bXufbdv7mpubC/CjAQBAoZiseAQAABkUIkT4tqTbDcMIGIZRKekmSccL8H0BAMACMuPuTATaGQAAQHoZ2xkMw/iypDslNRmG0S7pzyUFJcm27X+3bfu4YRj3STokyZL0adu2Z10HCQAASlPcYrAiAACYW8YQwbbtt3t4zN9J+ruCHBEAACgKtwIhTiUCAACYRSHaGQAAwBLAikcAAJAJIQIAAJA0KUSwqEQAAADpESIAAABJE20M7oBFAACA6QgRAACApImZCLQzAACA2RAiAAAASbQzAACAzAgRAACApEkrHmlnAAAAsyBEAAAAkqRYPLni0SJEAAAA6REiAAAASRPhQSxBOwMAAEiPEAEAAEiamIkQZ7AiAACYBSECAACQxHYGAACQGSECAACQNFGBYNLOAAAAZkGIAAAAJFGJAAAAMiNEAAAAkibCA0IEAAAwG0IEAAAgafJgRdoZAABAeoQIAABAkhS3nPAgRiUCAACYBSECAACQJMXiVCIAAIC5ESIAAABJE5UIzEQAAACzIUQAAACSJlY8xi1btk01AgAAmIkQAQAASJpY8Tj9fwMAALgIEQAAgKSpbQy0NAAAgHQIEQAAgKSpwQHDFQEAQDqECAAAQJITHIT8zqkBax4BAEA6hAgAAECSExxUhPySpLhFiAAAAGYiRAAAAJKcrQyVyRDBjNPOAAAAZiJEAAAAsixbCctOVSKYVCIAAIA0CBEAAEAqNEhVIjATAQAApEGIAAAAUtsYKoOBKX8GAACYjBABAACkKg/cdga2MwAAgHQIEQAAgEy3EiE1WJEQAQAAzESIAAAAZlQixC3aGQAAwEyECAAAYGImAu0MAABgDoQIAABg0nYGBisCAIDZESIAAIBUO0N5kBWPAABgdoQIAAAgVXlQFSJEAAAAsyNEAAAAqRkIqe0MtDMAAIA0CBEAAECqEqEiNROBSgQAADATIQIAAEi1L1TSzgAAAOZAiAAAAFKhQUVqxSPtDAAAYCZCBAAAkGpnqExuZ6CdAQAApEOIAAAAZlQi0M4AAADSIUQAAAAyrWQlAtsZAADAHAgRAACAzLhTeRD0+xTwGVQiAACAtAgRAACA4tZEiBD0+xS3qEQAAAAzESIAAIDUNoaA31DAbygWpxIBAADMRIgAAABS2xhCfp9Cfh/tDAAAIC1CBAAAkAoNAn6fAn4jtfIRAABgMkIEAACQ2sYQ8BkKUokAAABmQYgAAABSlQfuYEWTwYoAACANQgQAACAzYclnSH6foaDfSK18BAAAmIwQAQAAyLQsBf3OaYGz4pEQAQAAzESIAAAAZMbtVIgQ8PtSKx8BAAAmI0QAAACKW5aCfkOSFPIbqZWPAAAAkxEiAAAAmQlLAbcSwcd2BgAAkB4hAgAAkJmwFfQ5lQjBAO0MAAAgPUIEAACgeMJSMJAcrOijnQEAAKRHiAAAAGQmbAXcSgQ/7QwAACA9QgQAACAzYU3azmAoTjsDAABIgxABAABMCRFCfp9iVCIAAIA0CBEAAIDilp1a8Rj0+6hEAAAAaREiAAAAxeKTVjz6DWYiAACAtAgRAADAjEoEQgQAAJAOIQIAAHBWPCYrEYJ+QybtDAAAIA1CBAAAoFjCVsDnhghUIgAAgPQIEQAAgOIJS6GA084Q8PsUt2zZNtUIAABgKkIEAAAgM2GlKhFCydkItDQAAIDpCBEAAIDMhJ2aieBuaYhbtDQAAICpCBEAAIDMhDVlO4MkmXEqEQAAwFSECAAAQHHLViAZIqTaGahEAAAA0xAiAACAZCXC1HYGNjQAAIDpCBEAAMCUEMH9/3EGKwIAgGkIEQAAgOIJe9JMBOf/x6hEAAAA0xAiAACwzNm27cxE8E2tRKCdAQAATEeIAADAMmcm2xZCgeRMBJ9TiUA7AwAAmI4QAQCAZc6tOHDDg2AyTKCdAQAATEeIAADAMudWHLhbGUIMVgQAALMgRAAAYJkzLafiIJQcqOhWJDATAQAATEeIAADAMpdqZ3BXPAYYrAgAANIjRAAAYJlz2xbcrQxBnxsi0M4AAACmIkQAAGCZcwcoBv3uYEXaGQAAQHqECAAALHPTKxECPtoZAABAeoQIAAAsc9NXPLrbGWhnAAAA0xEiAACwzJmpdoZkJUKyrSFOJQIAAJiGEAEAgGUubk0brOinnQEAAKRHiAAAwDJnxt0Vj7QzAACAuREiAACwzJnTKhHcMIFKBAAAMB0hAgAAy5xbiZBa8ZgME9w2BwAAABchAgAAy1zcmjpY0Q0TYnEqEQAAwFSECAAALHOxhNvO4IQHhmEo4DNoZwAAADMQIgAAsMy5qxwDvonTgoDfoJ0BAADMQIgAAMAyF3crEQITpwVBv492BgAAMAMhAgAAy1wsWYkQ9BmpjwX9vtSsBAAAABchAgAAy5zbzuAOVnT+tyEzTjsDAACYihABAIBlzky2MwT8UysRTCoRAADANBlDBMMwPmMYRo9hGEcyPO4GwzDihmG8uXCHBwAA5ptppatE8KXCBQAAAJeXSoTPSbp7rgcYhuGX9P9J+lEBjgkAACwgt21hejtDnBWPAABgmowhgm3bj0i6kuFhH5D0DUk9hTgoAACwcOKWJcOQ/JMGKwZ8PpmECAAAYJq8ZyIYhrFW0s9J+rf8DwcAACw0M2FPqUKQnHWPMdoZAADANIUYrPgxSf/btu2MtysMw3ivYRj7DcPY39vbW4AfDQAA8mUmrCnrHSVn3SPtDAAAYLpAAb7HPklfMQxDkpokvcYwjLht29+a/kDbtj8p6ZOStG/fPm5vAABQAuIJS8HAtEoEP+0MAABgprxDBNu2N7r/2zCMz0n6XroAAQAAlKZYwlbANzVECPgNRUzyfgAAMFXGEMEwjC9LulNSk2EY7ZL+XFJQkmzb/vd5PToAADDv4glLIf/UdoYQlQgAACCNjCGCbdtv9/rNbNt+Z15HAwAAFpyZsBSYPljR71OcwYoAAGCaQgxWBAAAi5hp2QpMq0QI+A0qEQAAwAyECAAALHNm3FJoWiVCyO9TjBABAABMQ4gAAMAyF5+lEoF2BgAAMB0hAgAAy5yZsBRMMxOBdgYAADAdIQIAAMucmbAU9BEiAACAzAgRAABY5uIJW8HA1HaGoN+QSTsDAACYhhABAIBlzkxYCqSpRIhbVCIAAICpCBEAAFjmzISt4IzBij6ZCVu2TTUCAACYQIgAAMAyl26wYigZKsQtQgQAADCBEAEAgGXOWfE49ZTA/TPDFQEAwGSECAAALHNOJcL0wYrJECFOJQIAAJhAiAAAwDKXfsWjEyqYDFcEAACTECIAALDMpV/xSDsDAACYiRABAIBlLpZmxWPAlxysmKCdAQAATCBEAABgmYunWfEYCjinCDEqEQAAwCSECAAALHPpVjy6f6YSAQAATEaIAADAMmbbdvoVj8l2BmYiAACAyQgRAABYxuKWU2kQmr7iMcBgRQAAMBMhAgAAy5gbEkyvRHBXPpq0MwAAgEkIEQAAWMbckGDmTATaGQAAwEyECAAALGNuSDB9O4NbmUCIAAAAJiNEAABgGXO3LwR8U08JQn7aGQAAwEyECAAALGOzVyI4f45TiQAAACYhRAAAYBmbCBGmz0Rw/hwjRAAAAJMQIgAAsIy5Kx6nhwhuO0OcdgYAADAJIQIAAMtYLO6ueEzfzsBgRQAAMBkhAgAAy5hbiRCapZ3BtKhEAAAAEwgRAABYxtxKg+mVCO6gRTNOJQIAAJhAiAAAwDKWabAi7QwAAGAyQgQAAJYxM+EOVpxlxSPtDAAAYBJCBAAAlrG4287gm1aJkPxzjHYGAAAwCSECAADL2EQlwtRTAp/PUMBnKG4RIgAAgAmECAAALGMTMxGMGZ8L+I1UyAAAACARIgAAsKy5lQbTKxHcjzFYEQAATEaIAADAMmbGnUqD6SseJUIEAAAwEyECAADLmJmsRAilrUQwUiEDAACARIgAAMCyZia3LwTShAgBny8VMgAAAEiECAAALGtxa/Z2hlDAx2BFAAAwBSECAADLmBsSpGtnCPgMxZmJAAAAJiFEAABgGXMHJwZ8DFYEAACZESIAALCMxROWDEPypwsRaGcAAADTECIAALCMxRK2gj6fDCNNiOAzqEQAAABTECIAALCMxROWgmmGKkpOO0OcSgQAADAJIQIAAMuYmbDSrneUnI0NMSoRAADAJIQIAAAsY6Zlz1qJEGKwIgAAmIYQAQCAZcxpZ5i9EoF2BgAAMBkhAgAAy5iZsBWYYyYClQgAAGAyQgQAAJYxc45KhKDfJ9MiRAAAABMIEQAAWMbMhKWgb7YQwZAZp50BAABMIEQAAGAZiydsBQNzrHikEgEAAExCiAAAwDIWS1gKzFqJ4FMsTogAAAAmECIAALCMxROzr3gM+g3FLdoZAADABEIEAACWsbg114pHtjMAAICpCBEAAFjGYglbgbm2MyRs2TbVCAAAwEGIAADAMhZPWArN1s7gcz5OSwMAAHARIgAAsIyZcw1WDPhSjwEAAJAIEQAAWNacFY+ztzNIkpmgEgEAADgIEQAAWMZiCSvVtjCdu7WBSgQAAOAiRAAAYBmLJ2wFZl3x6Es9BgAAQCJEAABgWTMTc6x49FGJAAAApiJEAABgGZsrRAgxWBEAAExDiAAAwDIWt+zU7IPp3K0NDFYEAAAuQgSgRI2Ox3XLR36ix0/3FftQACxhZsJSYJZKBAYrAgCA6QgRgBLVNzKuy4NRnewaLvahAFiibNuWmbBnbWeYWPFIiAAAAByECECJiprOSftYLF7kIwGwVMUtp01h9hWPtDMAAICpCBGAEhUxE5KksViiyEcCYKlyVzdmameIU4kAAACSCBGAEhWJESIAmF+xZDgw62DFZLgQI0QAAABJhAhAiYqmKhFoZwAwP+KpEGGWFY/Jj8dpZwAAAEmECECJop0BwHxLzUSYJUQIsJ0BAABMQ4gAlCi3EiFCiABgnsTiTjgQmKWdIUg7AwAAmIYQAShRbiXCKO0MAOaJW4kQyjhYkXYGAADgIEQASpRbgUAlAoD54rYpZKpEoJ0BAAC4CBGAEhVNVSIQIgCYH6kQwZdhJoJFJQIAAHAQIgAlKsJMBADzzEy2KYQC6SsR3DYHM04lAgAAcBAiACUqajon7ax4BDBf4hkqEdx2hrhFiAAAAByECECJitDOAGCeuZUImVc80s4AAAAchAhAiYomw4NY3FKCfmQA88CdiRCcbbCij8GKAABgKkIEoES5lQgSLQ0A5ofbpjBbJYLPZ8jvMwgRAABACiECUKKmhgi0NAAovFjcqXKabcWj5FQp0M4AAABchAhAiYoSIgCYZ5kqESSnpYFKBAAA4CJEAEpUxJw4aR8dp50BQOFNzESYI0QIECIAAIAJhAhAiYrGEqoK+SVNbW0AgEJx2xQCvrnbGeK0MwAAgCRCBKBERcyEGqpDkmhnADA/3HAgFJj9dCDg8ylGJQIAAEgiRABKVMRMqKGqTJI0RjsDgHngtinMVYkQCvioRAAAACmECECJipoJNVZRiQBg/qRmIsxZicCKRwAAMIEQAShRU0IEZiIAmAfuTISgb47Bin4fKx4BAEAKIQJQgsyEJTNhT8xEoJ0BwDyIu+0M/rkHK1KJAAAAXIQIQAmKJisPGippZwAwf7zMRHAqEQgRAACAgxABKEFR0zlhrywLqDzoY8UjgHlhWraCfkOGMXuIEGDFIwAAmIQQAShBbiVCRdCvylBAo7QzAJgH8YSloH/uU4GgnxWPAABgAiECUIIiU0IEvyK0MwCYB2bCnrOVQZJCfp/iFiECAABwECIAJcgNDcqDPlWG/BqNUYkAoPDMhKXQHOsdJaedwYzTzgAAAByECEAJmlyJUBEKMFgRwLwwE5YCc6x3lJKDFalEAAAASYQIQAlyZyKUh/yqop0BwDyJJ+w51ztKbGcAAABTZQwRDMP4jGEYPYZhHJnl879oGMYhwzAOG4bxhGEYuwt/mMDyEp02E2GUEAHAPIglLIUyDlaknQEAAEzwUonwOUl3z/H5c5Jeatv2NZL+r6RPFuC4gGUtMm07Q4SZCADmgZdKhACDFQEAwCSBTA+wbfsRwzA2zPH5Jyb98SlJ6wpwXMCyFok5J+zlyUoEZiIAmA9xK/OKx5Dfp1icEAEAADgKPRPh3ZJ+WODvCSw7UwcrEiIAmB+xhK1AhhAh4DMUt2hnAAAAjoyVCF4ZhnGXnBDh9jke815J75Wktra2Qv1oYMmZGKzoU1UooLFYXLZtyzDmLjsGgGzEE5ZCmQYrBhisCAAAJhSkEsEwjGslfVrSPbZt98/2ONu2P2nb9j7btvc1NzcX4kcDS1LUTMhnOGXEFSG/LFsap5wYQIF5XvGYsGXbVCMAAIAChAiGYbRJ+qakX7Zt+1T+hwQgEkuoIuiXYRiqDPkliZYGAAVnelnx6HM+T0sDAACQPLQzGIbxZUl3SmoyDKNd0p9LCkqSbdv/Lun/SGqU9IlkqXXctu1983XAwHIQMRMqDzrhQVXIeZqOxeJqqAoV87AALDGmlxWPAefz8YSt5MsSAABYxrxsZ3h7hs+/R9J7CnZEAKaECBVUIgCYJ55WPCYrEWIJSxUiRQAAYLkr9HYGAAUwblqp8KCqjBABwPwwvax4TFYiMFwRAABIhAhASYqYzkwESaoITrQzAEAhmYnMIYI7eDGeYCYCAAAgRABKkjtYUdLEYMVxKhEAFJYz5yDDYMXk56lEAAAAEiECUJIiZkJlQefpmWpnMAkRABSWmbAUoJ0BAABkgRABKEHRye0Mye0MEdoZABSYmbBTKxxn47YzmLQzAAAAESIAJSlqJlKDFSuTYcIo7QwACszLTATaGQAAwGSECEAJmjxYsTLZzhChnQFAgTkrHjOFCLQzAACACYQIQAmKxBIqT4YIIb9Pfp+h0XHaGQAUjm3bMi1LoYyDFZPbGSzaGQAAACECUJKippUKEQzDUGXQr7EYlQgACidh2bJtZaxECLjtDHEqEQAAACECUHISlq1Ywkq1M0hOS0OEEAFAAbmVBZlnIjifj9HOAAAARIgAlJxocvZBRWji6VkZCmiU7QwACsgNBYIZ2xmcz8fZzgAAAESIAJQcd4DilEqEEJUIAArLDQUCGVY8MlgRAABMRogAlBg3LCibFiIwEwFAIbmhQDDgcTsDgxUBAIAIEYCSE01TiVARCmiMdgYABZQKEXyZQgQGKwIAgAmECECJiZrOifrkEKGKSgQABWYm2xmCAa8rHgkRAAAAIQJQclIzEUKTKxEIEQAUVjxZiRDIUIngrniMMVgRAACIEAEoOW6IUD5jJgLtDAAKJ1WJkGHFY8itRGCwIgAAECECUHLcwYrlwYmnZ1UoQCUCgIIyPa54DLCdAQAATEKIAJSY9IMV/RqPW0owHR1AgbgzDgIZKhFSgxVpZwAAACJEAEpONM1MhKpQQJJoaQBQMLG4286QYbCij0oEAAAwgRABKDGRWSoRpIlWBwDIl1uJkGkmgs9nyO8zCBEAAIAkQgSg5Mw2WFGSRgkRABTIxEyEzKcCQb+hOO0MAABAhAhAyYkmg4KywMTTs5J2BgAF5s44CPjmbmeQnJaGGJUIAABAhAhAyYmYCVUE/TKMiRN7txKBDQ0ACsWtLAgFPFQiBHxUIgAAAEmECEDJiZrWlKGKEiECgMJz2xm8VCIEmIkAAACSCBGAEuNWIkzmtjNEaGcAUCDZzUSgnQEAADgIEYASEzETKgtOfWqmBiuOU4kAoDDcmQgMVgQAANkgRABKTDSWphKhLNnOYBIiACgMd8VjwO9hsKLfRzsDAACQRIgAlBzaGQAshFjceztDwO9LVS4AAIDljRABKDFRMzFjsKIbKtDOAKBQ4pbbzpC5EiHkZ7AiAABwECIAJSZiWiqfVong9xkqD/oUoZ0BQIHEsxys6LY/AACA5Y0QASgxUTMxI0SQnJaGMdoZABRILNme4GnFo9+QGaedAQAAECIAJScSS6giOPOpWRH0a4x2BgAFEk9YCvgMGYbHwYpUIgAAABEiACUn3WBFSaoq82ssRogAoDDMhOWplUFiOwMAAJhAiACUmKiZUHloZohQEQpolHYGAAViJmxP6x0lZ/gi7QwAAEAiRABKimXZGo9baSsRKoN+RahEAFAgZsJSyGMlQoB2BgAAkESIAJSQaNwJCdINVqSdAUAhxbOoRAjRzgAAAJIIEYAS4lYapKtEqGA7A4ACMq1sZiIYiidoZwAAAIQIQEmJxp07fWkHK4aoRABQOGbC9hwiBKhEAAAASYQIQAlxKxHSD1ZkJgKAwnFXPHrhtDNQiQAAAAgRgJISNZMhQmDmU7My5NdoLC7b5kQeQP6yWfEY8BlUIgAAAEmECEBJiSRDhIo0lQiVoYAsWxqPcyIPIH9OO4PHFY8BHzMRAACAJEIEoKTMNVixMhksMBcBQCFkU4kQ9BmKJSwqoQAAACECUEpS7QxzhghsaACQv2xWPLphQ9wiRAAAYLkjRABKSKZ2BkkMVwRQENmseAy4IQItDQAALHuECEAJ8VKJMEqIAKAAsmpnSFYsxBiuCADAskeIAJSQuWciOJUItDMAKIR4wva+4jHgViIQIgAAsNwRIgAlJGI6J+hzDVaknQFAIcQSloJp1smmE/A5jzNpZwAAYNkjRABKiNvOUJbmxJ52BgCFFE/YCnqsRHDbGUwqEQAAWPYIEYASEjUTKg/65EtzYl9Z5g5WpJ0BQP6ym4ngViIQIgAAsNwRIgAlJGIm0g5VlKTK5MdHx6lEAJA/M2Gnti5kwopHAADgIkQASkgklkg7D0GaWPvoroEEgHzELUshv7d2hoC7nSFOJQKAhReLW7rr7x/S/Ue7in0oAESIAJSUiDl7iFAW8MnvM9jOAKAgzLjluRIhRDsDgCIaGIvpXN+ojnYOFftQAIgQASgpUdOatZ3BMAxVBv20MwAoCNOyUxUGmbiPo50BQDGEx0xJ0lDELPKRAJAIEYCS4g5WnE1lmZ8VjwAKwkxYqQqDTFKDFWlnAFAE4bGYJEIEoFQQIgAlJGImUrMP0qkMBTTGTAQAeUpYtmxbCviyDBGoRABQBOFkeDBIiACUBEIEoITMNVhRkiqCfo2NMxMBQH7c2QbBgLd2hmCynYFKBADFMOi2M0QJEYBSQIgAlJDoHCseJamqzK8x2hkA5CkVImRZiRC3CBEALLxwxG1n4EYKUAoIEYASEp1jO4MkVdDOAKAA4gmnLSHocbCi+7hYgnYGAAvPHaxIOwNQGggRgBISyVCJUEk7A4ACcCsRvK54ZLAigGJyZyLQzgCUBkIEoIRkHKxIOwOAAnAHJHqtRAjQzgCgiNyZCGOxRCoEBVA8hAhAibBtW1HTmrsSIeTXWIxKBAD5cSsKgp4rEWhnAFA87kwEiTWPQCkgRABKxHjypH6umQiVoQCVCADy5lYUeG1nCLmVCNwBBFAEk2chDEW5mQIUGyECUCIiyXCgIjj707Iy5Nd43FKCXe0A8hCLO68hoSzbGSgjBlAM4TFTKyqDkhiuCJQCQgSgRESSWxcytTNIoqUBQF5SlQieVzw6YYNJOwOAIhgcM9XaUCmJdgagFBAiACXCDRHmHKwYCjiPpaUBQB7cioJgwGOI4KMSAUBxmAlLw+PxVIhAJQJQfIQIQIlwgwFvlQiECABy51YUBH3e2hl8PkN+n6E4lQgAFphbebDerURgzSNQdIQIQIkYj7szETKHCKO0MwDIgxsGeB2sKEkBn0ElAoAFF06GCG1UIgAlgxABKBGRWHI7A+0MAOZZqp3B42BFydnQECNEALDAwmNOaLCyrlwhv09DEW6kAMVGiACUiNRgxYCXSgRCBAC5mwgRsqhE8NPOAGDhDUZikqQVlSHVVgRoZwBKACECUCImBivO/rR0qxQitDMAyENqJkIWIULQ76OdAcCCcysR6iuCqq0I0s4AlABCBKBERD0MVqxKtjMwWBFAPlIrHrNoZ3BCBCoRACwsNzSorwyqtjzIikegBBAiACUimtVgRUIEALmLxZ0QIZRVJQKDFQEsvPCYKcOQasqdSgRCBKD4CBGAEuFpxWOZO1iRdgYAuYtb7naG7CoR3AoGAFgogxFTteVB+X2G6iqCGopyDgQUGyECUCJSgxXnCBHcKgXaGQDkI56sKAj4shms6FMsTjsDgIUVHoupriIoSaotD1CJAJQAQgSgRETMhEIBn/y+2e8M+n2GygI+QgQAeYklZxtk084Q8htUIgBYcOGIqfpKJ0SoSw5WtG0CTaCYCBGAEhGNJeach+CqKgtojHYGAHlIVSJk0c4QYDsDgCIIj5kTlQgVQcUtO1W9CaA4CBGAEhE1LU8hQkXQTyUCgLy4YUB2Kx4NmbQzAFhggxFT9ZUhSVJteTD1MQDFQ4gAlIiImVB5MPNTsjLk19g4IQKA3LmrGoPZrniknQHAAguPxVRfMdHOIElDESoygWIiRABKhBMiZK5EqCwLaIwyPgB5MBOWAj5DhpFliEA7A4AFZFl2shLBbWdwtlRRiQAUFyECUCKiZkIVIQ8hQtCvsXESeAC5i1t2VvMQJKdqIZ6gnQHAwhkej8uyJyoQJioRCBGAYiJEAEpExPNgRWYiAMiPmbCymocgJVc8UokAYAG5YcH0mQhDUUIEoJgIEYASEY17CxEqQgGmEgPISy4hQsjvoxIBwIIKjyVDhGmVCLQzAMVFiACUiEjM40yEoF+jtDMAyEM8YSvgy66dIeAzmIkAYEGFIzFJUl1yJkJNuTMTgcGKQHERIgAlImpaHgcr+hWhnQFAHmI5VCIEAwxWBLCwplciBPw+VYX8VCIARUaIAJSIiJlQRcjjikczIdumrBhAbuIJO6v1jpIU9Bmp1ZAAsBDCybDArUSQnJYGZiIAxUWIAJQIr4MVK0MBJSxb43HuCALITS4zEVjxCGChDY4l2xkqJkKE2oog2xmAIiNEAEqAbdueBytWJtdA0tIAIFdmwlYgh3YGBisCWEjhMVOVIb/KAhPnR7UVQdoZgCIjRABKwHjckm1LZVmECKMxhgoByE3cshTKoZ0hlrBopQKwYMIRMzUPwVVbHtRQlHMgoJgIEYASEE2ubPS64lGiEgFA7syElX0lQvLxCYsQAcDCCI+ZqqsMTflYbUWAdgagyAgRgBIQcUOEUOYQoSr5mDFCBAA5MnNZ8ZgMERiuCGChDKWpRKhjJgJQdIQIQAmIms6wMm+VCLQzAMiPmbAUCmRbieCEDqbFcEUACyMciam+cmY7w/B4nKoooIgIEYAS4LYmlAczPyWraGcAkKd4DpUIbjuDyWYYAAskPGbOCBHcTQ3DrHkEioYQASgBbjtDeRaDFWlnAJCrXFc8Ol/L3T8A88+2bYUjpmqnD1ZM/nkoQkUmUCwZzyAMw/iMYRg9hmEcmeXzhmEYHzcM47RhGIcMw7iu8IcJLG3ZDVZ0QwTePAHkJpcQIeC2MySoRAAw/6KmpVjcUn3F1MGKbiUCax6B4vFyBvE5SXfP8fmfkXRV8v/eK+nf8j8sYHlxWxO8DVZ02hmoRACQq7hlp2YceBVKVSIQIgCYf+FITJLSzERwzoOGaGcAiiZjiGDb9iOSrszxkHskfd52PCWp3jCM1YU6QGA5iMZzqUQgRACQGzOe+4rHOMPMACyA8JgTEkzfzlBLJQJQdIWYibBW0qVJf25PfgyARxODFTOHCGUBn3wG7QwAcmfmUIngtjPEGKwIYAG4IULdLIMVWfMIFM+CDlY0DOO9hmHsNwxjf29v70L+aKCkRbMYrGgYhqpCASoRAOQsl5kIISoRACygQbedYdpMBCoRgOIrRIjQIal10p/XJT82g23bn7Rte59t2/uam5sL8KOBpcHdzuBlJoL7uLFxQgQAuXFWPDJYEUDpSrUzTKtEqAr55fcZzEQAiqgQIcJ3JL0juaXhZkmDtm1fLsD3BZaNSMw5KS8PeHtKVpUFNGYSIgDITSxhKRjIrp0hyGBFAAvIrTSYHiIYhqHa8gArHoEiCmR6gGEYX5Z0p6QmwzDaJf25pKAk2bb975J+IOk1kk5LGpP0rvk6WGCpisYTCvl9ngedVQT9ijATAUCO4glLwSwrEYKpSgTaGQDMv3DEVMjvSzt0urYiSDsDUEQZQwTbtt+e4fO2pPcX7IiAZSgSS6gs6P2EvjLk1yjtDABykLBsWbaynomQqkRgsCKABRAeM1VbEZRhzKyaqqsI0s4AFNGCDlYEkF7UTHha7+iqpJ0BQI7cdoRAttsZfO5gRUIEAPNvMBKb0crgqi2nEgEoJkIEoAREzITnoYqSVEk7A4AcudsVsl3xGErOUIjRzgCPfnysW//0wIvFPgwsUuExU/UV6UOEuoogKx6BIiJEAEpAJJZlJQLtDABy5LYj5NrOEGewIjy6d/8l/etDp5VgLShyEB4zZ69EqAhoKMrNFKBYCBGAEhCNWyrLqp3Bn1oL6UUklqDsD4AkybTcdoZsVzyynQHZ6RyMKBa31D4wVuxDwSI0GDFVVxFK+zkGKwLFRYgAlIBoLKGKrAYrBjQ67j2B//B3juqdn30ml0MDsMS42xVCWbYzsJ0B2eoMRyVJZ3pHinwkWIzCY3PPRIjFLUWZDwUUBSECUAIi2Q5WDPk1Hrc8l4gebA/rfN9orocHYAlx2xEC2a549FGJAO8isYSujMYkSad7CBGQHTNhaTSWmHUmQm3y48xFAIqDEAEoAVkPVkw+1ktLg2XZOtc3qnDEpJcZQKqSIBjIMkQIuDMRqERAZp2DkdT/JkRAttxWhdkqEercEIE1j0BRECIAJSASS6g8i0qEilBAkjTmoaWhIxzReNySbUsDY7zZAsudW0kQ9GW74tHdzkAYicw6w06IUB70ESIga+Hk+Upd5SwzEcqd8yDmIgDFQYgAlIDxeHYhQlWyEmEslrkSYXIvqltaCmD5cisJsh2sGGSwIrLghgg3b2rUmd5R2TYVLPBuMOKcr8y14lGShiJsaACKgRABKAG5rHiUvIUIZ3snZiH0j45nf3AAlhS3kiCY5WBFv8+Q32fQzgBPOsJR+Qzpts1NGoyY6hshxIZ3qUqETDMRaGcAioIQASgy27azHqyYameIZU7gz/ZRiQBgQjwVImR/ChDwGVQiwJPOcEQra8u1bVWNJOYiIDtuiDDXdgaJdgagWAgRgCKLJSxZtrIarJhVO0PPqFobKiQRIgCYNFgxhxAh5Pex4hGedIYjWlNfoS0t1ZKk06x5RBbC7mDFillmIlQ4N1PYzgAUByECUGRR07mrl91gRTdE8FaJcH3bCklSP+WkwLJnWskVj1m2M7hfQyUCvHBDhNV15aoM+XWGSgRkYXAsJsOQapIDFKcrC/hVHvRRiQAUCSECUGTR5JrG8qD3p2Nlqp1h7kqE4aip7qFxXbWyRvWVQSoRAKRmGoRyqEQI+n2KW4QImJtl2eocjGpNfbkMw9Dm5uopQ36BTMIRU3UVQfnm2CJTVxFksCJQJIQIQJFFkkFANjMRvLYznOtzhipubq5WQ1WIEAFAqpIgl0qEoN+nWJx2BsytfzSmWNzS2nqnlW5LSzUzEebwhSfP68fHuot9GCUlPGbOupnBVVseZLAiUCSECECRRczsQwSv7QzunZ/NzVVqrAqxnQHARIjgy6USgXYGZOaud1xTNxEiXB6MamScu8bTWZatv73vpL7yzMViH0pJCUdM1VWmn4fgqqsI0s4AFAkhAlBkbohQnsVgRa/tDGd7R+X3GWprrKQSAYCkicGKubQzBGhngAepECFZibC52RmuyFyEmc72jWp4PK7hKAHLZIMRD5UIFVQiYHE4eCmsi/1jxT6MgiJEAIosmkMlgt9nqCzgS7VCzOZM74haV1SoLOBXQ1UZIQKA1IpH2hkwXzqSIcLkdgZJzEVI4+ClsCRxMTzN4Fhs1vWOrtryAJUIWBT+8JuH9effOVLswygoQgSgyCYGK3oPESSpMuTXaIZ2hrO9o6k7QI1VIQ2MmbIsLgCA5cxtR8htxaNBJQIy6gxHVRXyp9bwrW+sVMBnMBchjYPtYUmiEmEad7DiXBisiMUgnrB0pmdEW1fVFPtQCooQASiySMw5Ic+mEkFyWhrmamdIWLbO9o1qU3OVJKmhKqSEZZPaA8uc284QzGnFo4+ZCMjIXe9oGM7vWNDv0/rGSkKENNxKhGEqEVKs5LmK13YGbo6glJ3vH1UsYWnbSkIEAAWUy2BFyalEmKudoTMcUSxuTVQiVDsDivppaQCWtXwqEZzBipywY26dg5HUPATXlpZqnaadYYrxeELHLg/JZ0gj43HZNs8tyanKsG15Gqxo29JIhqpMoJhOdjmve1sJEQAU0sRgxeyejk47w+whgtt7uikZIjRUOW/GzEUAlrd48q5drjMRqERAJh0D6UOEi/1j/P5McvzysMyErd2t9bJszfmevpyEI855ipcVj5I0RIUlStjJ7mH5jInZMEsFIQJQZOM5z0QIKDJH+n6md1SSs95RmhwisOYRWM5SlQg5rXj0KU4lAuYQNRPqH41pbX35lI9vbq5W3LJ1oX+0SEdWeg4l5yG8ZEuTJFoaXOEx598h42DF5MwN2jRRyk51DWtDU1XW5/mljhABKDK3JSGXdobR8dnvWpztHVFdRTAVHjRWlUminQFY7syEJb/PkM+Xw0wEn8GdZMxp+npHl3sXLpe5CFEzob/47lH1Di+tEPzApbCaa8pSA9cYrugIR7yGCG4lAv9uKF0nu4eX3DwEiRABKLqImVDAZ2Tdn1xZFki1QqRzpndEm5qrUoOtVlQ5b7ZXRggRgOUsnrAVyCFAkKRgwKcYIQLm0BmOSpoZIrjzeXIJER451avPPn5eD5/qzf8AS8jBS2HtXlevmmRZPpUIjvCYc55SVzH3TAS3nYFKBJSqqJnQ+f7RJTcPQSJEAIouYiayrkKQpMqgX2NztDNMXu8oSWUBv2rKAlQiAMtcLGEplMNQRUkK+gzaGTAntxJh7bQQoaosoDV15alWu2w8caZf0tJqxxuKmjrTO6o9rXWqKQ8kP8YddWkiFMhUieCugBwifEGJOt0zItuWti+x9Y4SIQJQdFEzofJQ9iFCRcivsVnaGYajpnqGx1PrHV0N1SEGKwLLXDxh5zRUUWKwIjLrCEdkGNKquvIZn9vcUp1TJcJTZ50QYSmF4IfbByVJu1vrVZsMEWhncAwmZyLUeVjxKDFYEaXrRNewJKValpYSQgSgyKKmpfJg9k/FqjK/xsxE2pVQZ1NDFadOgm2sIkQAljszYeW03lFy2hlY8Yi5dIYjWllTnvZ3bHNztc70jsiyvP8O9Y2Mp07El1I73oFLYUnStWtpZ5guHDFVXRbI+DpVUxaQYRAioHSd6h5WKODT+obKYh9KwREiAEUWieXYzhAKKGHZafuT3fWOm6dXIlSVqW9k6ZSDAsiembBzDxEYrIgMOgcjWlM/swpBcoYrjsUSujwU9fz93CqEUMC3pELwg5fC2tRUpbrKYKqdgUoER3jMzFiFIEk+n6GasgBtIChZJ7uGtaW5WoEc33NL2dL7GwGLTK4zEdyvSdfScLZ3VH6fobaGqSEClQgA4paVVztDnBABc+gMR2cMVXS5GxrOZNHS8MSZflWXBbRv/Qr1LaH3r4PtYe1urZfkvJ/7fQaVCEmDkZinEEFyWhoYrIhSdap7eEnOQ5AIEYCii5iJnHbHVpUlQ4Q0GxrO9o2oraFSocDUp3hDdUgDY7G0LRAAlod82hkCftoZMDvbttURjswYqujKZUPDk2f6ddPGBrXUlC2ZwYpdg1F1D41r97o6SZJhGKopD1CJkBQeMzMOVXTVVQRpZ0BJGhwzdXkwuiTnIUiECEDBHbgU1n88fMbz46NmQhU5DVZ0yh/HxmeedJzpGZ3RyiA5lQhmwqb0D1jGzDxWPIb8hkzLIohEWv2jMcXi1qyVCE3VIdVVBHW611uI0BmO6FzfqG7Z3KjG6rIlMxMhNQ8hWYkgiRBhknDEe4hQW04lAkrTqR5nlsu2JbjeUSJEAAruq89e1N/cd0LRNBUC6UTNhMoDua14lKSx2NSfk7Bsnesf1aZpQxUlqaHK2blMSwOwfJkJa0aVklcBv0+27bzOANO56x1nCxEMw9CWLDY0PJlc7Xjr5iY1VIU0Gkt4fm8tZQfbwwr4DO1cXZv6WE1ZkHaGJGcmQsjTY+sqgqx4REk6uYQ3M0iECEDBdQ1GZdtS+8CYp8dHcqxEqCxLHyJ0DEQUi1tpKxEmQoSlURIKIHvxPCoR3DYIWhqQzkSIkH6woiRtaa7WWY+VCE+c6deKyqC2r6pR4xIKwQ9eCmvH6toprYw15QwIlJyWmMFIzHslQkVAQxH+3VB6TnUPq6YsoDVp1t0uBYQIQIF1DTkX6Of7PIYIMSunmQiVbjtDbOqb55k+5+QsXSVCY1WZJKl/iZSEAsheLJ8Vj8mBjKbFcEXM1BF2ti7MNhNBcoYr9o3EFB6b+33Itm09dbZfN29qlM9npELwxf7+ZVm2DrcPandr3ZSP15QHaWeQc2PETNiq9zpYkXYGlKgTXcPauqpGhpFbaF/qCBGAAutJrq463z/q6fHRHLczVIXSVyK4U683p2tnqF46d3IA5CaeV4iQrESIEyJgps5wRJUh/5yT9Te3OFVymVoaLl4ZU0c4ols3N0qSGquTIfgir6Q72zeq4fG4dq+rn/Lx2vIA7QxSKhDIZrBixEwoxmsSSoht2zrVPaytS3QegkSIABTUeDyh/uQF+sUr3ioRomZC5cHsn4puC0RkWohwtm9U9ZXB1F2bydxy0H5CBGDZilt2Xise3e8BTNcZjmhNfcWcd962NDsn1ZlChCeS8xBu2dwkSUumneFgcqjinklDFSUGK7rCY06I4HUmQm0ysGIuAkpJ7/C4wmPmkl3vKBEiAAXVMzRxh+R8f+YQwUxYilt2TpUIbjvD6PR2hp4RbWqaOQ9BksqDflWG/Iv+JAxA7mLxfFY8GqnvAUznhghzWbuiQmUBn6cQoaWmLDXfZ6lU0h1sD6u6LDCj5dBpZzCX/eaTcMT57ztXNctk7uNY84hScrI7OVSRSgQAXnQnWxnqKoK64KGdIZKcMp3TYMVZ2hnO9o2mbWVwNVSFFv1J2GS2bevyYKTYhwEsGnHLTs02yFaISgTMoSMc1do5hipKkt9naFNztc7MMVzRtm09eaZPt25uTFU11JQFFPQb6lvkMxEOXgrrmrV18k8bblpTHpBlS6Oxxb99Ih+DY9m1M9RWODdUGEqJUpLazLBy9vPxxY4QASigrmSIcMOGBrUPRGQm5r5bF02eLOQyWLEs4JPPmDpYcShqqnd4PO1QRVdjVWhJtTM8dLJXt/7NT/X02f5iHwqQcrpnWHd/7BGd6Boq9qHMYOYxE8GtRMj02oblJ2om1DcyrjV1c1ciSNLm5iqdniNEeLFnRH0jMd2abGWQnPWQTgi+eGcijMcTOnZ5SLuntTJITiWCpGU/FyGcw0wESQxXREk52TWspuqy1CyXpYgQASigrkEnRLh5U4MSlq2OgbnvkKcqEXIIEQzDUGUoMKUS4WyvU/2Qbr2ja7GfhE33zPkrsm3pn396utiHAqS8cDGsE13D+q0vvTBjg0qxOSse8xysSIiAadz3v0ztDJKzoaF9IKKomf6u+xOn+yRJtySHKroaq8oWdSXd8cvDMhO29kzbzCA5lQiSlv1cBHcmQr3XmQjltDOg9JzqHl7S8xAkQgSgoLqHoioL+HRtcupypg0NUdM5Ec+lEkFyWhoiU0KE2dc7uhqqynRlkZeDTna4fVCS9NjpPr1wcaDIRwM4eoadoO5M74j+8rvHinw0U8USlkKBXAcrupUItDNgqs6wE5p7DRFsW7O2NDxxpl/rVlSotaFyyscbqxd3JZ07VDF9JYIbIizvi+FwJKZQwOd54HQtlQgoMZZl61T3yJKehyARIgAF1T00rlV15drQ5Jz4XMgwXHFiJkJuT8XKkH9K/+SZ3hEFfIbWN1bO+jXuSdhSGN5k27YOtYd1z541qq8M6l8fpBoBpaFnKKra8oB+46Wb9ZVnL+m7BzuLfUgp8YRFJQIKriMZIqz1GCJI0pnemUF7wrL19LkrqdWOky32mT4HL4XVUlOmVbUz50a47QzLvbd/cMxUfUVwzg0fk9WxnQEl5tLAmCJmQttWLd15CBIhAlBQXUNRrawtV3N1mSpD/oyVCJE8ZiJIzoaGyKRS6bO9o2prqJyz37mhKqTxuDVjIONidPHKmIaicd28qVG/ettGPXC8R0c7B4t9WIC6h8a1srZc/+uVW3VdW73++JuHddHDxpaFEE/kv+KREAHTdYajMgxpZV3mHuANjVXyGenXPB6/PKTBiDllHoKroSqk/kVcSXegPaxr19WnvUCupZ1BktPO4HUeguTMhwr5fVQioGRMDFWkEgGAR93JEMEwDLU1VGa8aIjmMRNBcioRxqZVIszVyiA5J2HS4l+TJUmHkq0M16yt06/cukE1ZQF94sEzRT4qQOoejqqltkxBv0//9PN7JUP6wFdeKImL71jCSm1ZyBbtDJhNZzii5uoylQUyv5+VB/1qbajUmTQhwhNn0s9DkJzBwCPjcY3HF18IPhgxdbZ3NO08BInBiq5wJOZ5HoLkzIeqrQhqKLK8wxeUjlPLYL2jRIgAFIxt2+oajGpVrXMXZkNjVeZKhDxWPLpf57YzJCxb5/vG5hyqKDknYZIWdV+p60jHoEIBn7aurFFdRVDvuHW9fnDksk73DBf70LDM9QyNa2WNU7Lc2lCp/+9N1+rgpbD+/kcni3xkznrGfCsR4iUQhqC0dIQjnuYhuLY0V6etRHjiTL82N1dpZZqSf3fS+WIMwd35PenmIUgMVnQNRuKqy6ISQXLWPNLOgFJxsntErQ0VqioLFPtQ5hUhAlAggxFT43ErdeKzvqlSl65ElJhjn7pbiVDu4c5NOlWT2hnaB8YUS1ja7LkSYfFvaDjUPqgdq2sVCjgvZb9620aVB/z6xENUI6B4bNtWz3BULZMugl5zzWr94k1t+o+Hz+rhU71FOzbLspWw7NxXPPpoZ0B6neGIp3kIri0t1TrXNzrlPdJMWHrm3JW0rQzSxPvXYmxpONgeliRdu7Y+7ecrQ375fcayr0QYHIupviLLEKE8yHYGlIyTXUPatsSrECRCBKBguoac9Var6pwLhw2NVYolLF0enH3NY76VCJPbGdz1jpsyViI4d3IW40nYZJZl60jHoK5dO1Ea2lhdpl+8qU3fPtBZMv3nWH4GxkyZCVstNVN7w//stTu1bWWNPvS1A+oZjhbl2EzLufjPNURwtzrQzoDJbNtOViLMrB6YzebmasUSli5dmXitPtQe1lgskXaoorS4K+kOXAprU1PVrHfZDcNQdVlg2VcihCNmaliiV3UVhAgoDbG4pbO9o0u+lUEiRAAKxt2R7U5ddjckzLWhId/BihWTQgR3VVbGSoTqpTET4Xz/qIbH47pm3dT+0l+7Y5P8hqF/e5hqBBSHGxBML8cuD/r1z7+wVyPjcf3uVw/KmqNKab64F//BHNsZqERAOldGYxqPW1m1M2xObmiY3NLwxOl+SdJNm9KHCIu5ku5Qe3jWVgZXTfnyDhHG4wmNxRJZDVaUnDWPDFZEKTjXN6q4ZWvbKkIEAB71DDknNe6Fw4ZGpyJgrrkI+Q5WrCoLaCzZznCmd1QrKoNaUTX3QKKqkF+hgG9R3smZ7HDHxFDFyVbWluutN6zT15+7NGcVSCHEE5b6R8Z1umdE+89f0QPHunXv/kv69KNn9dVnL87rz0bp6k69FsycUr91ZY0+/Lpdeux0n/79kYUPutxZBjmveAwQImCmzrATnGXbziBJp3snQoQnz/Zrx+raVFgwnTsTYbFV0nUNRtU9NK7d69IPVXTVlAeXdTuDGwTUVXofrChJdRWBZb8aE6XhRNeQJC2LEGFpT3wAFpDbzuCGCKtqyxUK+OauRDAT8vuMnO8KVgT9ipqWEpatM70jGasQJKdksmmRr8mSnHkIZQGfrmqZ+Xf+9Ts26yvPXNInHzmrP3/droL+3C89fVH//vAZDYzFMt4xuv2q5qxOqrE0dA+lr0Rwve2GVj16uk//8KNTumljo65fv2LBji3fSoSgj3YGzNQRdgLbbCoR6iqCaq4pS21oiJoJ7b8woF++ef2sX1NbHlDQbyy6SroDl8KSZh+q6KopX94Xw4NjToiQ60wE27bTrs8EFsqp7mEFfIY2NWU+H1/sCBGAAukaiqqxKpQa8ufzGVrfUKnzfXNVIlgqD/hyftOrKnMqGCJmQmd7R/Wy7c2evq6hOrQoy0EnO9w+qF1rahVI09vd2lCpn9u7Vl9+5qJ+884taq7JvLfcq28+3y4zYelN161TfWVQKypDqq8Mqr4ypPoK588v9gzr3f+1X2d6RggRlqHeYee5NdvvnWEY+sgbr9HBS2F98Msv6Ae//ZKse4Bz5VYQ5DoTwf06KhEwWWcyRMj29W5Lc3WqEuH5iwOKxa1Z5yFIznNnRWVo0YUIB9vDCvoN7VhdO+fjasuDqUBmOQonKxGybWeoqwgqbtkaiyWW/ER8lLaTXSPa2FSVuhZYypb+3xBYIN2DU6exS9L6xipdvDJ3JUKuQxUlqSLkvFl2DUbVNzKuTR4qESSpoaps0Z2ETZawbB3pHNS16+pnfcxv3LlZsbil/3zsXMF+rm3bOtE1rFfuXKkPv36XfucVW/Urt27QPXvW6qVbm7W7tV5tjZWpFotzcwRIWLq6h6KqqwjOOeuktjyof377XnWEI/rvpy8s2LHFkxUE6cI3L9x2hjiVCJikMxxRRdCf9cXf5pYqne4ZkW3bevJMv/w+QzdubJjzaxqqQupbZJV0By+FtWN1bcb5R7XlgWXdzhBOVSJk185QmwxhWfOIYjvVPbwsWhkkQgSgYLqGolo1rQd6Q2OlzvePyrbTn3BHY4mchypKUmXya492OvMBvLQzSM6E68U8E+Fs74jGYokZ8xAm29RcrZ+9do2+8OR5hccK83dtH4hoZDye8Q2iuaZM1WUBne2duQMdS1/3UDTtPITp9rat0K41tXr45MKtfIylKhFyHaxoTPk+gCR1DjqbGbKtqtvSXK3haFy9w+N64ky/rllbp5ryuYOIxkVWSWdZtg61D2r3HKG3a7kPVnTfq7MerJj8nWG4IoppdDyui1fGlsV6R4kQASiY7qFoar2ja31jpaKmpZ7h9Cc8ETOR81BFaaKd4XC7EyJkWu/oaqhafOWgk7lDFa/NMKTq/Xdt1mgsoc8+fr4gP/dE17AkafuquUtSDcPQpuYqnaUSYVnqGR5XS423VXd3bG3WcxcGNDK+MBcO8TxXPLpfRyUCJusIR7Oah+Da0uKcbB9sH9TBS+E5WxlcjYusku5s34hGxuMZ368kZ7DiyHh81hsPS93EYMXs2xkkaSiyfAMYFN+LyfkuW6lEAOBVLG6pbyQ2Y5DaendDwywXkxEzv0oEt53hcMegAj5DbQ2Vnr6uoSqksVgitR1isTnUPqjKkD9j+8b2VbV61c6V+twT5wtSInrisvepuxubqnS2lxBhOeoZGleLh0oESbrjqmbFLaeUeyGYcXewYm5v/36fIZ/BTARM1RmO5DT/xd3Q8JVnLipu2bp1c1PGr2lYZJV0By45ofeeDEMVJacSIZHs7V+OBiOm/D5DNVnONaitCKS+HiiWU8kbTVQiAPCsd8SpNFg1LURw1zzOtqEhmmclQmXIbWcYUltjpecLg8bk+qzFdCI22eEOZ6ii35e5dPa3XrZFgxFTX3wq/5WLJ7qH1dZQqWoPJzibmqrVORhZtEFNqegdHtcXn7qwaO7MWZatnuHorJsZprt+/QpVhvx65NTCtDSYyUqEQI7tDJITQLjfBxiPJ9Q7PJ5TJcLKWqf166cnexT0G542lTRWhTQcjWs8vjheWw9eCqu6LOBpZpHbyrFcWxrCY6ZqywNZt8VMVCIQIqB4TnYPqzzo83xDb7EjRChRUTOhf3rgRY0uUIkr8tM1mFzpNq2dYU19uQI+Q+f7Z6tEsFSex2BFN0QYGY97nocgKbWD+8oiG04lOXvuj3YO6pq19Z4ef+26et2xtVmffvSsInne3TlxeUjbPZapbWqukm0zXDFfX3jqgv70W0d0umdxzJcYGIvJTNhq8bgRJBTw6ZZNjXrkxYUJEdw2hKAv97f/oN+XqmgA3Pe/XEIEwzC0OflaubdthadBww3VzvvXwOjiuGA82B7WNWvrPIXeNeVOQL1chyuGI6bqK7MbqihNzERgsCKK6WTXsLaurJHPw3N9KSBEKFEPn+rVPz5wSj880lXsQ4EH7l746ZUIAb9PrQ2Vs1cixBKqCOb+NKwMTdwR9zoPQXIGU0lS/yIaTuU63TuiqGl56i91/dZdW9Q/GtN3D3Xm/HOjZkLn+ka1PcOKLpf734OWhvw8f2FAktMzvRi480+8ViJIzlyEC/1jujBL2FhIZp6DFd2vpZ0BLncl4Zp677/zk21OtjR4mYcgTa6kK/33r6iZ0PHLQ7q21dv7lRsiDC3bSoRYTutu3X832hlQTCe7nRBhuSBEKFFHkoPjXrg4UOQjgRepSoQ0Fw7rkxsa0sl3sGLlpLs22VUiOHdJF9NwKteh5MXkNVmECDdsWKH6yqCeO5/78+nF7hFZtjxXImxsckKEc32L4w56KUpYtg5cCkuSDrWHi3osXrmBopftDK47tjZL0oK0NLgX/7mueHS/Nk47A5I6w87vfC4zEaSJuQhe5iFIUmP14nn/OtIxKDNh67q2zG0a0uR2huV5MTwYMbPezCA5r0nVZYElP1jxgWPdevsnn1LCohKs1FwZjal3eHzZzEOQCBFKlhsiPH8xXNwDWSLm+w25eyiqUMCnFWne/DY0Vuli/1janu58BytODRG8VyKk2hkWwUnYdIfbB1VdFtDGRu9/X8MwtHtdvQ7mcSF6vMsZqug1RKgMBbS6rpxKhDy82DOskfG4fIbTV7wY9Aw5d0e9bmeQnFWwrQ0VevhU33wdVoqZbGcI5REihPw+xWhnQFJnshJh+nYir+7Zs1bvv2uzrmur9/T4xfT+tT9ZSeVl1oMk1abaGZb2xfBswmOm6nOoRJCcf7ulXonwgyOX9eTZfrUPpK9uRfGc6k4OVVwmmxkkQoSSZNu2Dnc4Fywnu4YWbPXXUnWkY1B7/vLH81rV0ZXcC59uGND6xkoNj8fTnvBE8w4RJrUzNHmvRKgtDyjoNxblYMVDHYO6em1t1j1nu1vrdap7WGOx3J5PJy47A3PWZxFebGqu0hlmIuTs+QthSdKrdq7S8cvDi2KQWs+wc1e22eNMBMkJue64qllPnulTLD6/d/jjiUIMVjSoREBKZzii5poylQVyey9bW1+h33/1ds/VMW47Q98imOnz3IUBbWyqUlO1t9cDBivGcpqJIEm1FcElPxPhaPLaYLHMCFpOTnYRIqAE9AyPq29kXHdua5ZlL54y3lJ1/9EuJSxbRzuH5u1ndA1GZ8xDcLkbGs6nmYsQNROeBknNxu8zVBbwqaEqpBVV3t94DcPQisrQohusaCYsp790XX3WX7untU6WLR3pyO334GT3kLatrPE0HMu1salK53pHFs1mgVLz/MUBNVSF9Po9axRLWDpxebjYh5RR99C46iuDWYeDd2xt1mgsoefnuYUtlpqJkF87AzMR4OoIR3Iaqpir2vKg/D5DV0p8JoJt23r+woDnVgZpeQ9WTFi2hqLxnGYiSMkQYQlXIkTNhE73OuEBIULpOdk9rLqKoOehyksBIUIJOpzs+f6lm9ZLkl6gpSEvD57skSRdmsfyr+6h2Ve6tTU6q16mD00zE5bMhJ3XTATJaWnIppXBtdh2bUtOuVgsbumatd7nIbjc4CGXsnjbtnX88rC2r/I2VNG1qalaQ9H4ovt3LhXPXxzQ3tb61H71fNpRFkr3UDSnk4hbNzcq4DPmfS5CajtDviseEwRjcHSGI1qb41DFXPh8hhqqQjm3M3z60bP6+E9eLPBRzXSub1T9ozHt2+A9RKgM+eX3GcuyEsENAHKZiSA5ax6XcjvDia7h1CwEQoTSc6prWNtW1mS9nnQxI0QoQUc6B2UY0i2bG7WpuYrhinnoGYqm7jy3X4nMy8+wbVvdQ+OzViKsW1EhnzGzEiFqOqXZ+YYIG5qqdP36hqy/rrE6VPJ3cqZzA7ZcQoSm6jKtra/QgRwuRHtHxnVlNKbtq7MrU2NDQ+7CYzGd7R3VdetXaHVduZqqy3TwUulvaOgZHs9qM4Orpjyo69pWzPuqR7cNIZ/BimxnWHj/9cR5/e5XD+ips/0lVdlk27Y6w1GtqVu4SgTJaWnoz7GS7pvPd+irz14q8BHN9FyW8xAkp0qwuiyw5Mvy0xnMM0SoLQ8u6fDlaKfz/rduRUWqIgGlwbZtneweXlatDJIUyPwQLLQjHYPa3FytqrKArmtboZ+e6JFt28sq3SqUh5J39VbXlc9bJcJQNK6ImZh1qFRZwK819RUzKhGipnMSXp7HikdJ+vr7bs3p6xqqynR4IJzXz15ohzoGVVMe0PpkdUe29rTW51SJ4JbRZ/sG4c6pONc3ohs3Zh/0LGduBdbetnoZhqE9rXWLohKhZyiqzc3epsxPd8fWJv39j06pb2Tccw91tmIFqkSIU4mwYMyEpY/++JQGI6a++UKHtrRU6xdvatMbr1uXc+l3oYTHTEXMxIK2M0j5VdJdGhjTcDSuSCy/dsJMnrswoNrygLZksTlJcloalvLF8GzCbohQketMhKU9WPFIx5BqywN66dZmfedgJ9cFJeTyYFTD0bi2LrMQgUqEEnSkY0hXr3HKpq9rW6ErozFdvMIk1lw8eKJHq2rLdee2lnn7N5xY6Tb73ccNjVWzViLkM1hRcuYiZNOn72pchO0Mh9sHde26upzfOHe31ql9IKK+kewqME6kNjNk186wdkWFQgEflQg5eP7igHyGtDvZhnLtunqd6R0p6V5hy7LVMzyulizWO072kqucVY+Pn56/LQ3uYMWgL4+ZCD4jNVsB8+/ps1c0GDH1Tz+/R3/35mtVXRbQX3z3mG766wf0+/ce1IFL4aJVJ3QkNzMUI0TIpZ1hcMxMXaCfnef1u/svDOj69SuyHgJcUx4s6de5+RIec/571uYYjNVVBDUyHk+9xi01xzoHtWtNna5qqdZwNK7e4cVVSbqUnXQ3Myyj9Y4SIULJ6R0eV9dQVFcny7X3JlcezfewraXITFh69MU+3bW9WW0NlQqPmfPyxtw1mDlEWN9YOaMSIeK2M8zjnZC5NFSFNByNz/s0+EIZjyd0omtI16ytz/l7uBek2Q4rPXF5WCtry1Krxbzy+wxtaKzUGUKErD1/cUDbV9WqqswpmNvdWi/blg53lG5Lw8BYTHHL1socBytdvbZOKyqDenge5yK4bQjBQB4rHgMMVlxI9x29rIqgX6/auUpv2deqb73/Nn3vA7fr5/au0/cPX9Yb/vVxvfafH9OXnr6o0QXe5uSud1y7wCGC086Q/UXU5IrE+Qx3w2Mxne4ZyaqVwVVTHtDQMqxEKEQ7g7Q0N1uYCUvHu4Z19dpabWlxLlSZi1A6TnURIqAEHEmeILshwtaVNaoK+RmumIP95wc0Mh7Xndta1NrgnOBcmoe5CF3JSoTZZiJITiVCeMzU4NhEiBGJFWYmQq7cC+KBscVRjXCya1hmwta167Kfh+C6em2dfIZ0IMve+hNd2Q9VdG1qqp73O15LTcKydeBiWNetr0997Nrka+Kh9tINEbqHnIuaXGYiSE7odPtVzXr0xb55u7PsDkQM5FC95FrM7QxPnulP7fNeDCzL1v1Hu3XntuYpgfPVa+v0kTdeo6f/+OX6v2+4WgnL1h//z2G9+mOPLOgq1M5UJcLCDVaUpMbqMg1F41mHWZeuLEyI4N74yWVeUe1ybWcYc9sZcq9EkLQk50mc7hlRLG7p6rV12tLitMcwF6F0nOwe1qractXlGIAtVoQIJcYNEXYl2xn8PkO7W+upRMjBQyd7FPQbum1Lk9oanB76+ZiL0J2sRJirhNnt4b9wZeKkJVKgwYq5mti1vThK4g7lMVTRVVUW0NaVNVlVIpgJS6d7RrIequja2Fyli/1jS7bEcj6c6h7WaCwxZTXaiqqQ1jdW5jTTYqF0D2d+Lcjkjqua1Ds8ruPztM7SLMSKR9/iHKwYT1j69S/s14e/c7TYh+LZC5cG1Ds8rruvXpX28zXlQf3yzev1w99+iT761t1qH4jooZPzO5xzso5wJLVmeCGlQvAsWxrcc4D6yuC8hrv7zw/I7zNSm2WyMd/tDFEzoc8+fq7k5ge4IUI+Kx4laSiy9AIYd0X5rjW1WllbpuqyAJUIJeRk1/Cym4cgESKUnMMdg9rUVKWa8okX0evaVuj45eHUnWt48+DJHt20sVHVZQG1rkiGCPMwF6FrKKoVGfbCb2hypvRPnovgzkQoK3IlQq5rsmbz8Kle3fl3D+qbz7cX9Pse6RjUisqg1q3Ir2x29zpnuKLXO73n+kYVS1jakXMlQpXilq1LA/OzHWQpckPT6fvVr11XX9KVCL3JSoSWmtzvyt6x1ZmLMF9bGuIJWz5DOc1RcQUXaTvD8xfDGorGtf/CQOr1t9T98HCXgn5Dd21vmfNxhmHo9bvXqKk6pO8c6Fygo5M6w1Gtra9Y8AFvEyF4liHClYhqywOpGSvz5bkLA9q1pjandsX5Hqz4Nz88ob/47jF98pEz8/YzchGOxFRTFsh5c4wbPpRaOFIIRzoGVRH0a2NTtQzD0OaWakKEEhGLOzeatq3MboDqUkCIUGKOdg5p17Q7rXvb6pWw7Kz7uJez9oExneoe0Z3bnBPy+sqgqssCap+HC7nuocwr3dxKiAt9E5UIhVrxmKvG6sKGCOPxhP7qe8f0K595Ruf7x/ST4z0F+b6uQ+2DumZdfd4nq7tb6zUwZnpubTl+2bkDkOvqnk3JydxnKT307PkLYTUkKw8m272uTh3hSMkOlHKHrOZTibCytlzbV9XokXmai2BaVl7rHSUp6DNSbRGLyU9POK9JsbiVWr9Xymzb1n1Hu3T7lqZUv/dcAn6fXnvtGj1wvHvBBvN1hCMLPlRRyj0EvzQwptaGSm1qqtK53tF5aRsyE5YOtodzmocgOSHCyHh8Xo7twRM9+twT51Ue9Olr+9tLKgwcHDPzKgevrXDm5yzFdoZjnUPasbomFf5uaSZEKBU/PdGt8bilWzfntpVpMSNEKCFXRmPqCEd0zdqpdzz3Ju/GvVDCZbyl5sFkOad798YwDK1bUTEvGxq6h6Kzrnd0lQf9WlVbPqUSofiDFZ0LnVx3bU92pndEb/zEE/r0Y+f0jlvW665tzTqWvPguhKiZ0Knu4RnPjVy4MxUOeAzlTnQNK+AztDnLNV2uTckqlHN9xRuuaNu2Pvv4Of3hNw7JsvI7MY0nLH3sgVPzUtXjeuHigK5rmxkY7U6WBpdqoNo9HFV9ZVBlgfye03dsbdb+8wMaixX+bqQZtxXKN0Tw+xZle85DJ3u0e12dAj5jXjdgFMrRziG1D0RmbWVI5/V71mg8bun+o93zeGQTOsORBZ+HIE2E4P2j2QWKl66MqXVFpTY3V2k0lkjNMSmkY51DipqW9uUwD0Fy2hkSlq2xAlef9g6P6/e/flDbV9Xoo2/do97h8YKH/fkIR8ychypKE4MVl1olgmXZOto5mJqVJklbWqrVMzy+JAOTxebe/e1qqSnTS64iREARpYYqrplaidBQFdKGxko9vwjunJSKh070qC15t8HV2lA5b+0Mcw1VdE3f0BCJOSfhxapEqK8IymfkV4lg27a+8sxFvfbjj6kzHNGn3rFPf3nP1drbtkLn+0cLNin8+OUhxS07r80Mrm2ralQW8HnurT9xeUhbWqoVynGa/YqqkFZUBou2oWEwYuq9X3hOf/HdY/rKs5fynvx/39EufeyBF/Wfj50r0BFONTAa09m+0VR4OtmuNbXy+4ySnYvQMzSulXm0MrjuuKpZsYSlp872F+CopopblgL+/Kp5An6fYousEqEzHNGJrmH97LWrtae1Xo+fKfy/baHdf7RLPkN6xY6Vnr9mb2u92hoq9e0DHfN4ZI7xeEI9w+NFqURoTIbg2bx/WZat9oGIWhsq5rVCbP8Fd6hi7pUIUmG3DNi2rd//+kENR+P6+Nv36lU7V2p1Xbm+/MzFgv2MfA1GTNVX5D5bIzVYcYmFCBeujGk0lkjNSpM0MVyRaoScfeuFDt3zL4/l1drWMxzVQ6d69cbr1uVd4bcYLb+/cQlzV5dNb2eQnN7gF4q4C3oxiZoJPX6mT3dta55yJ7N1RaXaByIF/Tc0E5b6RsbV4iFE2NBYlbYSoTxYnKehz2eooSqk/hxDhMExU+//0vP6w28e1t62ev3wt+/QK3c6J7s7VtfKtp27+IXgPjfy2czgCvp9unptnecL0ZNdw9qe58CcTc3VRWlnONw+qNf+86N68ESP/vRnd2hVbbk+9ejZnL+fbdv61CPO1//4WPe8vB69cCn9PARJqgwFdFVLtQ6U6FyE7uHxvFoZXPs2rFB50KdHThX+brmZsPIaqihJIf/iG6z44Ennjutd21p065YmHW4Pl/wdyx8e6dKNGxvUWO39d8owDN2zZ40eP92nnuSgz/nSPejcxV/o9Y6Sc8Ho9xlZVdL1joxrPG6ptaEyVVk2H3MRnrtwRWvrKzJWKM6mJrWqsHC/n5974rweOtmrP/3ZHdq6skYBv09v2deqR17sndeqsmyEx2J5tTNUhvzy+4wld3d+YuD61EoEiRAhV+GxmP7iu0d1sH0wrxky33qhQwnL1puvX1fAo1s8CBFKyNHOQbU1VKadTLu3rV69w+Pz0tO/1Dx97oqipjVjEFVbQ4UiZiLrQUxz6R0el23Pvd7Rtb6pUn0j4xpJ3p2PpkKE4lQiSE6Vy5Usy0El6dnzV/Qz//SIfnS0W39w9zZ94d03TTlh2plMzAvV0nCofVBN1SGtzvGkbLrd6+p1pHMw44XQ4JipzsGotq/Or41iY1OVzi5gO4Nt2/rCk+f1pn97QomEra+97xa95yWb9K7bNuiJM/2pk5JsPXPuig62D2pvW706wpGCtqy4nr8QTm6lSR8Y7Wmt16H20gxUe4aieQ1VdJUH/bp5U+O8zEUwE7aCeQxVlJxKhMXWzvDgiV6tW1GhLS3Vum1zoyxbenoeKj0K5XTPsE73jOhnrl6d9dfes2eNLFv6/qHL83BkEzqS6x2LESL4fIZWVAazCsHdi+XWFZVaWVumqpC/4BVitm3ruQsD2rchtyoEaaISYahAlQjHLw/pIz84oVfsaNEv3bw+9fG33dAqSbp3/6WC/Jx8DUbMnDczSE6AVlcRLPlwMFtHO4cU9BvaunLiZkbrigqF/D6dIUTIycceeFGDEVNr6sr12SfO53Q+Ydu27t3frr1t9alQZ7khRCghhzsGZ11f55b2suoxswdP9Kg86NPNmxqnfLx1HtY8diUHqa2qy3ynaEOj01pxMVmNEDUTMgypLMcy+UJwQgTvJ2G2betjD5zS2/7jSQUDPn39N27Vb965Zcak9zV15aotD6SGEubrcLvz3CjUBPDdrXWKmlbGffEnupzjz78SoUq9w+MLMuxsZDyuD37lgP7s20d165ZGff+DL0nd1f/5G9tUFfLr0zlWI3zq0bNqqArp4z+/V4bhVCMU2vMXB7RjdY0qQ4G0n9/dWq/wmJnTfJMHjnXP20A9y7LVOzyulQWoRJCkl1zVrLN9owW/S2gmLAXzfM0J+n2LarDieDyhx0/36a5tLTIMQ3va6lUe9OmJeW5piCcsfe9QZ06bldyZBq/a5b2VwbWlpUa71tTqW/O8paEzGSIUo51Byj4Ed9/7WxucbRIbmwsf7rYPRNQ9NJ5zK4Mk1abaGfJ/v4iaCX3wyy+orjKo/+9N1055D11bX6E7tzbrq/svFT0UtG1b4TFT9XmECJLzb7fUVjwe7RzU1pU1U1oqA36fNjZVUYmQg5Ndw/rCUxf0Cze16YMvv0rHLw/pmXNXsv4+B9sH9WLPiN5yfes8HOXiQIhQIgaT0+J3zTI4bvuqGlUE/XrhYnhhD2yRsW1bD57s0a2bm2bc4U+FCAU8Ke8edEKETNsZJKUmzbtzESKxhCqC/gVfjTVZY1VZVndynj53RR974EW99to1+v4HXzLrDmzDMLRzTa2OdeYfIozF4nqxZ1jXrEv/s3LhHvfBS3PfkXfbMXbkWYmwqclJqed7uOKJriG9/l8e0/cPder3X71Nn/mVG7Ri0v72uoqg3nZDm7536LIuD2ZX1XS6Z1gPHO/RL9+8Xq0NlbqubUXBQ4R4wtLBS+G0rQwut6XlYJYtDe0DY/q1L+zXm/7tCb3xE4/rviOXlchzyORkV8Ziilu2p9cCL1661RnSVOhVj/GErUCelQghvyHTskqyGiSdp89eUcRM6K7tzraesoBfN2xo0BNn5ne44j8+cEq/9aUX9Pc/Opn11953pEt7Wuu1ui63C/R79qzRwUthnZ/H1xw3RMi1bD9fjVVlWYXg7kaedcmVz5uaCt9m9lye8xCkye0M+V8M//UPjuvFnhH9w1t2p22LefuNbeoeGk8Noy6W0VhCccvOa7CiJNUusUoE27adrW1rZp6DbGmp1mm2PmXFtm395feOqrosoA+9cpvu2bNW9ZVBfe6J81l/r68/d0nlQZ9euzv7arGlghChRBzpdE6IZ6tECPh9unZdnV6gEmFO5/pGdaF/THclVztOtm6FczJW0BDBrUTwFCI4lQjuXISImSjaUEVXtpUIT5+9IsOQ/u8brlZ1Wfo7xa6dq+t0omso7wu145eHZNnStbM8N3LR1lCp+spgxin/J7qGVF8ZVEtNfneXNzc7/+3PzuNwxXv3X9Ib/vVxDUfj+tKv3az337VFvjQXi++6bYNsSZ97/HxW3//Tj55TWcCnd9zilMO+cudKHe0cSpU0F8LJ7mGNxhJzhghbV9aoPOh9MKbr3v3tkqTfe9VW9Y3E9L4vPq+X/cND+sKT53O6Uzxdar1jnr8rrs3N1VpTV17wloZYAWYiBPw+2bYKGsLMpwdP9qgs4NMtmyamZ9+2pUmnukfmbW7AYy/26RMPnVFdRVCff/J8VgFi+8CYDncMZrWVYbrX7V4jw5C+PY/VCJ2DETVVlxWtJa+hOruZPpeujKm5ZuJ4NzdXqyMcyWuw2nTPXRhQVciv7atyD54LNVjxgWPd+vyTF/Se2zfqjq0zz4kk6WXbW9RSU1b0AYvhMee/Yz6DFSUnKF9KMxEuD0Z1ZTQ2ZTODa3NLtS5dGSvo7+9Sd//Rbj1+ul+/+8qtWlEVUkXIr5+/oU33H+3K6lwmaib0nQOdunvXKk/rd5cqQoQScXiWzQyTXbd+hY52DvGCMQc3Tb9zW8uMz1WGAmqqDqXuRhRC19C4Qn5famf1XKrLAmqqLpuoRDATRZ2HIDkhQnjM9FzKuP/CFW1bWeOpb3HH6hpFTUvn+/O7cD6UvON8TQGGKroMw9DudfU6kOFC9PhlZ6hivtUibY2V8hnKq3Q2YdnqHorqwKWw7jtyWZ97/Jw+8sPj+p2vvKCf+8Tj+v2vH9J1bSv0gw++ZEYrz2StDZX6matX6UtPX/RcLtszHNU3n+/Qm69fl7qb5Q7RfKCA1QjPJyut5goRgn6fdq2py2rNo2XZ+vpz7bp9S5N+62VX6cHfu1Of+MXrVF8Zcto+/uYn+uiPTqp3OPd1bz3JVXFehqx6YRiG7tjarCdO9xd0iGG8ACGC+/XxRRIiPHSyV7dsbpyyTve25E7vJ+ehpaF3eFz/62sHtLm5Wt/9rdsV9Pv0Nz887vnr7zvSJUm6e1fuIcLqugrdtLFB3z7YMW8VIx3hqNYWYb2jq7EqlNVgxUsDY2pdMVHZsam5SrZd2Aqx/RcGtLdtxYwWv2wUYrBiz1BUf/CNQ9q5ula/f/e2WR8X8Pv01n2teuhkT6qypBjCY87fNZ/BipKz5nEpVSIcTVZzzlaJYBX493cpi5oJ/b8fHNPWldX6xZvaUh//5eSNkS88ecHz9/rRsW4NReN68zJuZZAIEUrGkY5Bra2vmFJ6PN3e1nrFLTvnoWjLwUMne7SlpTrVujDduhWVBZ2J0D0UVUttmeeLzA2NlamL6nHTKtpmBpe7a3tgLPObbjxh6fkLA7phg7fd16nhinm2NBxuH1RLTVnBysRdu9fV6VT3sMZi6e/2WJatU93Ded1RcpUF/Fq3ojLr0lnbtvWBL7+gWz7yE2390x/qpr/+id7wr4/rfV98Xh/+7jF99rHzeu7igAI+Q7//amfAZbOHO+G/9pJNGh6P66vPehuo9YUnL8i0LL379o2pj21urtam5qqCtjS8cGFATdUhtTbMXcK9e129DncMeg6/njjTr45wRG/Z57zh+32GXnPNan3rN2/Vve+7Rfs2NOifHzyt2/6/n+qPvnkop+oK9452oWYiSNIdW5s1PB4v6ErLuGXnveIxmPz62CIYrniub1Tn+kZ117RgeeeaWtVVBPX46cK2NFiWrQ/de1BDEVP/8gt71dZYqd+8c7PuP9rteWXn/Ue7tH1VjTZMWlGci3v2rNXZ3tHUhUihnesb0doVxZmHIDkh+GDE9ByyXboSUdukc4NNBa4QG46aOtk1lFcrgyRVhfzyGblXIri/g2OxuD7+9j0qC8x9s+JtN7TKlvS1Ig5Y7Eq2hq6ozK8SobYiuKRmIhzpGJRhpG+p3NLMhoZs/Odj53TpSkR//rpdU9Yxrq2v0Kt3rdJXnr3ouSrx3v2XtLa+Qrdunv2GzXJAiFAijswxVNHlDlcsxlwEy7J14FJY3z3YWbJ9sKPjcT199opetn1mFYKrraGwIULXYNRTK0Pq5zdW6sLkdoZQ8SsRJG+7tk90OaXmXqdOX9VSo6DfyHuC/6GOwYKsdpxud2u9LFs60pH++C4NjGksltCO1fkNVXRtaq7K+mT1RNewvnuwU1taqvUbL92sv3rD1fr0O/bpex+4Xc/96St08q/u1qN/8DLd+75b9f67Zg64nM3u1nrduKFBn338fMYL8bFYXF946oJeuWNlare665U7V+qps/0Fu/Pz/EXnLl6mUM4djPmix5Onr+2/pLqKoF61c+qQOsMwdMOGBn3qHfv0wO++VG++fp2+8XyHPvS1A1kfe3eyEsFLiOPVbZub5DNU0JaGWLyAlQiLYLjigycmVjtO5vcZumVTox4/3V/Q97RPPXpWj5zq1Z+9dmcqgHzPSzZpTV25/ur7x2RlqN7oGY5q/4WBvFoZXK+5erWCfkPfeqEj7+813dneEV26Epmz6mm+NVa5IXjm9y8zYenyYGTKDYaNTW6IUJiLsAOXwrJs5bWZQXJel6rLAjlXInzm8XN69MU+/dlrd2pLS+b3r9aGSr3kqmZ99dlLWbcoff7J8/rc4+dyOs7JfnikSzVlgbzf62srAkuqneFo55A2NVWlHTS8qblKhkGI4EXXYFT/+uBpvXrXSt22pWnG59956waFx0x9+0Dm18rLgxE9drpPb7pubdqW0eWEEKEEDEVNne8f09WzDFV0NdeUqbWhYsE2NAyMxvTtAx36X189oH3/7wG94V8f1we+/MK8TTfP1xNn+hVLWLozzTwEV2tDhTrD0YJNIu4eimZ1h3xDY5UuD0YVNROpwYrF5IYI/R4mXD973ple67USIRTwaXNzdV4bGkbG4zrTO6Jr1tbn/D1mc21yUONsd3mPX3aGKhaiEkFyTljP9Y1mdcHywLFuGYb00bfu0e+9ept+6eb1esXOlbp6bZ0aq71XwKTza3dsUkc4oh8kS6dnc+/+doXHTP36SzfN+Nyrdq5S3LL10MmenI/D1T8yrvP9Y3O2Mrh2Z/hvN9ngmKn7jnbpDXvWzNk+tLm5Wn/9c9foF25s06H2wYwXe9N1D0W1ojKY8a5fNuoqg9rTWq+HXyzc3fK4ZacqCXLlVjIUss1ivjx4skebm6vU1jizOu3WLY3qCEdy2vSRzgsXB/R395/Uz1y9akq5bHnQrz+4e7uOdAzpfzJc0P/4WLdsWwUJEeoqg7pzW4u+e6iz4PMrfjpLOLOQ3NYqLyH45XBUlu2sd3RVhgJaW1+hMwUKEfafH5DP0KwDh7NRWxHMqRLhxe5h/e19J/XKnSv1Cze2Zf6CpF+4sVWXB6N6+JT31/JvH+jQ//n2Uf31D09o0EM142wisYTuO3JZr7lmdd4tnnUVQcXi1pJp+z3aOZh2HoLkvK60rqhkuKIHf/PD44pbtv7kNTvTfv7GjQ3asbpWn/Ow7vGbz3fItqU3Xb9uPg51USFEKAFHk3dCZ3uhmOy6thXzVolgWbYOXgrrnx54UT/3icd13V/9WL/9lQN6+FSv7riqSR954zWSnAn9peinJ3pUXRbQvvWzX+S2rqhUwrJ1eTD/YVq2basryxDB3dBw8cpYScxEaKzyfhK2//yA1tZXZLXOK98NDUc7BmXbmpdKhOaaMq2tr9CBWXrrT3QNyTA0ZTdzPjY1VytiJlJrQb144Hi39rTWF/Tutuvl21u0qalKn3707KxvmgnL1qcfO6vr2up1fZrn1d7WejVVl+lHBWhpeCE1D6E+42PXN1aqriKogx7mInz7YIdicSvVypDJjtU1Goslsr6w7BkeL3jLjeS0NBxqD2sgiwFyczELOBMhFp87RPjPx87pU4/ktk60EDJVp92anItQiFWPQ1FTH/zKC1pZW66/eeO1MwK+1+9eo93r6vR395+ctYVKcuYhbGyq0rYCve7cs2eNuofG9bTHVgqvfnqiR1tXzt46uBBSIbiHuQhuBeK6aa1Smwq45vG5CwPatqo2NdMgHzXlQQ3lECI88mKfYglLf3nPrqxC5pfvWKmm6jJ96WlvLQ1HOgb1B18/pM3NVYrFLX3nUO4DPH98vFujsYTesHdtzt/D5Q65G1oCcxH6R8Z1eTCadh6Ca0tLtc5QiTCn5y5c0bcOdOrXXrIxbZgsOdU/77p1g050Deups7Nf49i2rXv3X9KNGxtSw9KXM0KEEnA0uZnBS4iwt7VeXUPRgg/A+cHhy7rh/z2ge/71cX3sJ6dk2dJvv/wqfev9t+nZP3mFPvbze/X2G9u0dWV1TvtU55ttO3dDb9/SNGWX7nSpNY8FaGkYHo9rLJbQqjrvF3gbki86F/qdibrFDhG8tjPYtq1nz1/Jukxz5+pa9QyPq28kt4F1qYGjBdzMMNme1vpZ72afuDysDY1VBWs52dyUXf9t91BUB9sH9Yod2e+J98LnM/Srt2/UofbBWZ/T9x3p0qUrEb33jplVCO73eMWOFj18slfj8fzu/DyfnO1wrYdVnoZh6Np1dRlXdEpOK8PO1bWef4fcypNsK2h6hqIFG6o42R1bm2Xb0mMF6t03E7YCvnxDBOfiZK7Bip3hiP7mh8f1d/fnN7AyH2512mx3yzc3V2llbVnecxFs29YfffOwOsNRffzte9MOh/P5DP3Za3eqayiqT84SrAyOmXryTL9evWtVwVb/vmLHSlWF/AXd0jAUNfXMuSt62fb5eW3yqjFVSechREiGgpMrESRpU5PTZpZvS0vCsvXCxQHty3MegqumPLd2hp6hqEIBX1ZtlpITDL513zr99ER3aj7BbPpGxvXez+9XY1VIX/31W7R9VY3uzWOewrde6NDqunLdtNFbleNcapNDn5fCcEV3lslcA9e3tFTrbN/ootmUs9Asy9aHv3NMK2vL9Jt3bpnzsa/fs0YrKoP63BOzt+c8d2FA5/vH9BaqECQRIpSEwx2DWl1XrqY0O3ynu2594eciDI6Z+pP/OayW2nJ97G17tP9PXqFvv/82/c4rtmpPa/2UPusbNzZo//krBWsHmEvXYFSne4Y9PfZk97AuD0ZTe8Bn455AFGLNY8+QO0gtu3YGSbrQP6poCax4XJE82c10J+fSlYh6hse1z2Mrg2vn6twuyFzPnLui1oaKebkTLzm99e0DEfWnCTlOdjubGQplY3N2/bc/Oe6Ulb5y5/ydqL/punVqqArpU4/OvKixbVuffOSMNjRW6pU7Zy+tfuXOlRoZj8+Z3nvx/MUB7Vhd6zm02b2uXie7h+cchHS0c1BHOob0thu8T1DetqpGPiP739nuofGCrXecbPe6etVVOHusR8fzHxjmVCLkO1jRl/pes/m3h87Itp3hi196ujjr41LVabO8bhmGods2N+nJM/1Zt69M9tVnL+n7hy7rQ6/aOudQvX0bGvSz16zWfzx8Nu2F2gPHuxW37IK0MrjKg369+upV+sGRywUr8X7sxT7FLXvO+UMLIRWCewipLw2Mye8ztLpu6vv1puZqjYzH8w66TnQNaTSWyHuooqu2PJBTO4NTHZlbq9vbbmiVZWvOQCAWt/SbX3xe/aMxffId+9RUXaa37GvVofZBnezydr42Wf/IuB4+1at79hSmv9zdHLUU5iK4IcLOuSoRmqsVi1sFXV2+lHz9uXYd7hjUH/3MDlVlWEteHvTr7Te26cfHumf997x3f7sqQ3695prV83G4iw4hQgk40jGoXXMkjZPtWF2rsoCvoHMRPvaTUxqMmProW3frDXvXpvoM07lxY6NGY4m8h+Vl8uSZft39T4/o7o89Ome5tevBE7OvdpxsdX25/D6jIGseuwadk45sEv+6yqDqK4M63z/qDFYscogQ8PtUXxnMWIkwMQ8huxMkd6JwLi0NCcvWU2f7U6vY5oPbW++ukXSNxeI63z9asHkIkvN7Uhnyey6dfeB4t9oaKnVVS3XmB+eoIuTXL928Xg8c75nRF/zs+QEdbB/Uu1+yac6BjbdtaVJF0K8fH5t7tsJc4glLBy8NemplcO1urVfCsnXs8uzVCPfub1fI79M9e9Z4/r7lQb82NlXp2GXvJ8SWZat3ZLygmxlcfp+hv3j9Lr1wcUC/9J9P59V7LBV2xeNsIcLlwYi++uwlvWVfq+7c1qwvPHUh70qVbHmtTrt1S5P6R2M62Z39BZAkneoe1oe/e1S3b2nS++7YnPHx//vu7UpYtv7+RydnfO6+o11aXVeuawtceXXPnrUajsb10MnCDOj8yfEe1VUEs3q+zof6ypAMw1s73qUrEa2pL58ylV1yZqFIyruv3J0VVagQoaY8qOHx7J/r3UNRrazJrSJqfWOVbt/SpK/MMWDxL793VM+cv6K/ffO1qequN+xZo4DPyKka4XuHLith2fq5ArQySE74ImlJbGg40jmodSsqVD/HxorNLWxomM1Q1NTf3n9C17XVez4H+KWb18swDH3xqZnrHsdicX3vUKdec83qjIHEckGI4JFt2/qlTz+tj//kxYJ+35HxuM72jWbczOAK+n26dl2dXihQiHC6Z0RfePKCfv7GtrQrZKZzy83ms6Xhq89e1C//59Nqqi7TXdtb9FffP65f/8Jzc5anPXiyR7vW1GasCgj6fVpdV16Qdga3t31VXXZv2Osbq3Shf8wZrFjk7QySczcn00nY/gtXVFMe0FYPk54nW1EV0uq68pwqEY50DGooGtetaSbpFsrVa+vkM5yp2pOd6h6RbUvbC7SZQXLuem5s8rahYSwW12On+/SKHSsLVtY8m3fcsl6hgE//+djUEr5PPnJWKyqDevN1c5ftlQf9umNrkx441pNzSfCJrmFFzESq0sqL3ck5GQdmaWmImgn9zwsdetWulXOehKWzY3VtVr+z/aMxJSx7XmYiSNIb9q7VJ37xeh3tGNLbPvlkap1kLsxE4VY8mrNsZ/iPh8/Ksm395p2b9au3bVTfyLi+f+hyXj8zW16r09wVXbm0NETNhH7rS8+ruiygj75tt6c7qW2NlXrXbRv0jefbp6xrHh2P65FTvXr1rlUFn/h92+ZGNVWH9J2D+W9psJKDVO/c1jzjgnyh+X2GGipD3toZBsZmtDJIhVvz+NyFAa2sLdO6Aq28rMmxEqF7aFwrszwnmeztN7apIxzRoy/ODJy+9PRFffGpi/r1l27SPXsmLvobq8v08h0t+p8XOrIetvo/L3Ro5+pabStQ1V/dEmpnONY5NOc8BMlpZ5DyD8HSMROWPvS1g/rXB08X/HsvhH/+yYvqH43pw6/3Ph9kTX2F7t61Sl9+5uKM2TU/PNyl0ViCVoZJCBE8MgxD4UgsdUe2UI51Dsm2pWvWeb/jeV3bCh3pGCrInZ2/+v4xVQT9+tArt3p6/Mracm1orJyX4YoJy9b/+/4x/e9vHNYtmxv1zd+8VZ/85ev1Z6/dqZ+e6NFr//lRHW6fecEwOGbquQsDnqdEt66oLEjpV3cO7QyStKGxUuf7RxU1LZUFi/8UbKwKZdzOsP+80+uZy8ntztW1OVWuPH7GOam/ZR5XiFWVBXRVS82MAX0nkse7o4CVCJJTOnu2L/Ob/WMv9ikWt/SKnfNfLtxUXaY3XbdW33iuPdXWcbpnRA8c79Yv37LBU9D1qp2r1DUUTc2wyJYbinrZzOBqqS3X6rpyHZpluOIDx7s1GDH1Vo8DFSfbsbpWHeGI5xNR97VgPtoZXHdfvUqfeecNutA/prf++5NqzzEINROWQgVb8TjzgqF7KKovPXNRb7puXXJ1XJO2tFTrM4+fW9D1wO72gEzVaWvqK7SxqSqn4Yp/+b1jOtU9oo++dY9asrj7+5t3bdGKypD+6vvHUv8mD53s1XjcKmgrgyvg9+m1167RA8d78i7zPtgeVv9orOitDK6GqpC3wYpXImlDhFW15aoI+vMOEfafH9D16zOvp/XKDRGyec7Ytp1XJYLktKc1VoX05WemtiA9e/6K/vw7R/TSrc36g1dvn/F1b93Xqv7RWOp558XZ3hEduBQuWBWCNDETYbG3MwxHTZ3rG51zHoLkhCbNNWUFr0SwbVv/+xuH9I3n2/WPPz6l8wUaPrpQzvSO6LOPn9dbrl/nac7SZO+8bYOGonF964Wpc2S+/ly71jdW6sYCzO5YKop/BbOI7FpdpyMdgwU9EXLvRGR6oZhsb1u9Ygkr1S+VqwdP9uihk7364MuvmrOFYbobNzbo2fNX8uohnW5kPK73fn6/PvXoOf3KLev12XfeoNryoAzD0Ltv36iv/votiidsvenfntAXnrow5b/Bo6d7lbDsjHecXK0NFbo0UIh2hqjqKoJZD0dc31iljoGIYgmr6O0MUuZKhIHRmF7sGcl6HoJr55panekdzbof94nT/dq2smbe5iG4drfW6eCl8JTfqRNdw6oK+Qt2V8m1salK7QORjAHgA8e7VVMe8LxOM1/vvn2TxuOWvpAs4fvPx86qLODTO25Z7+nrX7a9RX6foR/nuKXh+YthNVVnfxdv97rZB2N+9dlLWltfkXYndCbuLI8THsMvt596PgYrTnb7VU364ntu0pXRmN7y70/mtJoubuVfieAOZoylCRH+/eEzSli23n+XM8TKMAy967YNOtIxpP0LuB74oRO9nqrTJKca4emz/VndRb3/aJe+9PRFve+lm3XHVm/vPa66iqD+1yuu0lNnr6SeM/cd7VJjVWjenvP37FmjWNzS/RlWumby0xM98hnSS7P8O88XL5V0kVhCfSPjam2Y+fri8yUrxDyEu7PpGoyqIxxJu8EmVzXlQSUsW5Es3jdzGfY8XSjg05v3rdMDx3tSc586wxH9xhef09r6Cn385/embW976dZmNdeU6d797Z5/1rcOdMownIF2heJuZ8i37avY3BXTXgYCb2muLniI8Lf3n9Q3n+/Qu27boIDf0Ed/fKqg33++9AxF9fGfvKhf+NRTKg/69Xuv3pb199i3foV2ranV556YCL4vXRnTk2f79ebr1s17dehiQoiQhavX1mpgzCzIekDXkY5BtdSUZXXyubct/+GKZsLSX33vmDY2VelXbt2Q1dfeuLFR4TFTLxboRat9YExv/rcn9NCpXv3fe3bpL+65ekaZ5PXrV+j7H3yJbt3SqD/71hH99lcOaCQ5ZOzBE72qrwxqT6u3u5itKyrVOzw+50A2L7qGollPQJacSgQ3fymNEKFszjs5bq9nrie3O1bXKmHZerHb++9L1Ezo2fNXdOuW+atCcO1urdfAmDllTsbxy0Pauqqm4GXFm5urZNvOdo7ZJCxbPzneo7u2teTdu+7VlpZqvWx7i77w5AW1D4zpG8936E3Xr/M07FVy2lb2rV+hHx3NNUQY0HVt9Vm/OV/bWqfz/WMKj039/e0IR/TY6T696fp1c85zmM2OLAeC5lqVlIvr16/QV957i8yEpbf++5NTSuK9MOP5z0QIBZLbGaa1M/QMR/Wlpy/q5/aunbJK641716muIqjPTGuZmS+DY6aeu+i9Ou22LU0ajSVmzEaZzXDU1P/59hHtWF2rD73KWxXfdG+/sU1bWqr1kR+e0Mh4XD893q1X7VqZ0++rF3ta69XWUJn3loafHO/RvvUNWbcIzZfG6syVdG7VzmzrKDe3VOcUyLnc98hCbWaQnEoEKbve/lyGPafz8ze0KWHZuve5dkXNhH79C88palr61Dv2pd08IjnVLm/cu1YPnuzxNKTStm1964UO3ba5qaCvm6GATxVB/6KvRHBf1zO1M0gTax4LdYPzv544r3976Ix+4aY2/Z/X7tS7btuo7xzszGtd93yybVtPnOnTb/73c7r1b36qj/74lK5qqdFn3nlDVhViLsMw9M5bN+hU94ieTFaoff25dhmG9EZaGaYgRMjCrmQimO1J21yOdA5mvb5uZW251tZX5DVc8YtPXdCZ3lH9yWt2zDl0Kp2JuQj5751+7sIVveFfH1dHOKLPvesG/fItG2Z9bENVSJ/5lRv0+6/epu8d6tTr/+UxHb88pIdP9eiOq5o9n3y5JxK5lgO7uoeiOfUerp90cl0KMxEaq0IaGIvNWlny7IUrCiVnceTCvas71wC86Z6/OKDxuDWvQxVd7nDFA8myeNu2k5sZCtvKIEmbmpz+xbk2NBy45JQLv2IetzKk82sv2aT+0Zje+dlnZSYsvfv2jVl9/St3rtTJ7mFdnCMgSadvZFwX+seymofg2jPLYMxvPNcu21bOvYsra8u0ojKYuhuUSfeQc9LcnEVFVz52rqnV1379FpUFfHr7p57S/iza7Ewr/xDBrUSYfuf+kw+fVdyy9Vt3TV2lVRFypl7ff7Qr79ddLx550a1O8xYi3LKpUYYhPeFxLsI//OiUeobH9ZE3XpPzv2XA79OfvGaHzvWN6v3//bxGYwm9elfhWxlchmHonj1r9MSZvtTFZra6BqM6dnlIL9tRGq0MktRYVZaxEsGdgbQuTTuD5Kx5bB+I5Ly9Yv+FKyoP+uacop+tmuQd9WzWPLrDnvO9KN/YVKVbNjXqK89e1B9+45AOdwzqH9+2R1etnHtuwVv2rVPCcsKBTJ6/GNbFK2N6QwFbGVy1FYFFPxPhaOeQmj3eYNzSUq3h8bh6CrBK94eHL+vD3z2qV+5cqf97z9UyDEPvu2OzassDaYfBFtNgxNRnHjunV3z0Yf3Cp57W46f79c5bN+jB37tTX3zPTXm1Hbxu9xo1VIX02SfOy7Jsff25dt22uUlr6wtbnbrYESJkYceqWvkM6UiB0rixWFyne0ayDhEkp6XhQI6VCAOjMX3sgRf1kqua9PIcTgbWrajQ6rryvOcifOuFDr39k0+rqiyg//nN2/SSqzKXR/p8ht5/1xb993tu1nA0rtf982PqG4l5bmWQJkKEfIcrdg1GtSqHaezrk2seJWXdCjEfGqpCsmwpPMub7v7zA7pmXV3Ox9rWUKmqkN/zBZnktDL4fYZu2jT/5fzbVtWoLOBLlcV3D40rPGZqRwGHKro2NDm/e3NtaHjgeLcCPmPBy4Vv3tSgq9fW6nTPiF6xY2VqarlXr0qugfxRllsa3IqqbOYhuK5OBluT5yJYlq17n7uk27Y0znrnMRPDMLRjda1OdHl7re8ZjqqhKpR1IJuPTc3Vuvc3blVzdZl+6T+f1sOnvE3eNxO2Anne7U63naF3eFxffPqC7tmzRhuaqmZ8zTtucaZef/7JmVOvC+3Bkz3J6rR6T49fURXSztW1qTksczl4Kaz/evK83nHzes/ffzZ3bmvWS65q0sOnelVTFtCt8xya3rNnjSzbmYifC7ffvVTmIUjO+9fAmDnn2mm3yixdO4PkDFfMVCE2l+cuDGj3uvqCVo6lKhGyGK5YyIqot9/UpktXIvrWgU596JVbPa0a3tJSo71t9fra/ksZ74p/64UOlQd9evWuwoflKypDqWB3vo2Mx/Vrn9+vLz51YdaNFrk42jnoqQpBmjRcMc/q4GfOXdFvf/WArmtboX9++0TbSl1lUO+7c7N+eqKn4HPhcnGya1h/8PWDuumvH9Bffu+YasqD+vu37NbTf/xy/elrd2pjmvefbJUH/fqFG9v0wPFuff25dnWEI3rLPqoQpiNEyEJFyK/NzdU61lmYSoTjl4dl2dLVOaTXe9tWqCMcSb1pZOMfHzil4aipP/3ZnTn19hiGoRs3Nujpc1dyLp/61wdP63e+ekB72+r1rd+8LfUi6NUtmxv1gw++RDdubFB1WUAv3er9pMY9kchnzWM8YalvZDyndobGqpCqk+thSiFEaKxO7tpOUxIaNRM61B7WvixXO07m8xnavro2q1K4x8/06dp1dam7MfMp6Pfp6rV1qRDhePLCcT4qEWrKg2qpKZtziNcDx7p106aG1JTphWIYhn7jpVvkM6T3vTTzqrrp2hortW1lTdZzEZ6/OKCAz8ip0qW2PKjNzVVTNjQ8dbZfl65EchqoONmO1bU62T3s6cSwe2h8XocqzmZtfYW+9r5btKmpWu/5r2d135G5Lw5t21bCsgvWzjB5O8OnHz2rWNyaUYXgWlNfobuvdqZej47P3/o1y7L18MlevXSr9+o0yWlpeP5CeM42t3jC0h9987Baaspy6rWdzjAM/cnP7pDPkF6+o2XeQ6gtLTXataZW3z6Q25aGn57o1roVFfO6djZb7vvXwBw98JeujKk86Ju1UsgNTOeqEJvNWCyuo51Deb1HpuOuKsyqEsHdGFWAEOHVu1aqtaFCr9u9Rr/1svTP6XTecn2rXuwZ0cE5WoNicUvfO9SpV+5cNS/v8TdscOZ2LcRa2UdO9erHx7r1p986otf982MFuciOmgm92DPieVZaIUKEU93Des9/PavWFRX69Dv2zTg3fdetG9VcU6a/ve/Egg7Ina5nOKo3fuJxfffgZf3c3rX63gdu17fef5vefP26gp9P/9LN6+UzDP3pt4+opiwwr5Vii5WndyzDMO42DOOkYRinDcP4wzSfbzMM40HDMF4wDOOQYRivKfyhloar19bpSEdhKhHctohrcjh5dvczZ7vq8VT3sP776Yv6xZvW57VS58aNDeodHtf5HJL7gdGY/vHHp3T3rlX6wrtv0oqq3Horm2vK9MV336Qn/uhlasjiezRXl6k86MtrQ0PfSEyWrZzaGQzDSLU0lMZMBOffLt1chEPtgzITtm7Ic2DUzuTKPC9vPsNRU4faBxeklcG1e129jnQOKp6wdCJZMVGolVPTbWqumvVk9UL/qF5MVgIUw89eu1rP/Mkrct51/qpdK/Xs+Ssa8LByzfX8hQHtXFOb8wnA7nX1Otg+MRjza/svqaY8/zf8HatrFTUtnfMwlbpnODrvQxVn01Rdpi+/92ZdvbZOv/u1g6kNG+m4F/3BAg1WdCsR+kfG9fknL+j1u9do0xwVLL9620YNR+P65vPeh69l61DHYE7bA27Z3KhYwtL+C7NfBHz28fM6dnlIf/H6XQW7+Nm+qlb//Z6b9Uev2VGQ75fJG/as1cH2QZ3q9l4ZJjkXNo+f7tfLt7eU1GAx9/1rrpaGSwNjWreictbjdu9c5jIX4eClQSUsO+fXzNlMtDNkNxOhtjxQkDbJsoBfD/zuS/XPb9+b1X/v1+5erfKgT/fuvzTrYx451auBMVM/t7dwAxUnu3Nbs8ZiCT17bv4HuT52uk/VZQF9/O17FR5zBt7+9ldeUFces9NOdjnhtddKhJaaMtWUB3IOETrDEf3KZ55RedCv//rVG9Oek1eE/Prgy7bo2fMDeuikt6q3yR441q23/seTObdSuf7xxy9qPG7p+x+8XR9547U5VXJ7taquXHdfvUqxuKXX7l5TEjf9Sk3GEMEwDL+kf5X0M5J2Snq7YRg7pz3sTyV9zbbtvZJ+XtInCn2gpWLXmlp1DUU9DY7J5EjHoBqrQjmlxjvX1Crk9+n5LFoabNvW//3eMVWF/PpfHlc6ziafuQg/OHJZccvWB16+Je87Lz6fkZrG65VhGFq3ojKvdoZ8E/8NyZaGUgoR0p2Eual6vidIO1bXang8rnYPWzGePntFCctekKGKrt2tdYqalk51j+hE15DW1lfMWyXAxqbqWdsZHjjulAsXK0SQ5HmYYjqv3LlSli3Pa77iCUuH2gdzamVw7W6tV+/wuLqGohqMmPrhkS7dsyf/N3y3ncXLcMWeoXGtLEIlgquuIqi/e/O1ipgJferR2YcXuhf9+VYiBAPuikcnlPjUo+cUjSf0Wy+7as6vu66tXrvX1emzj58v6HafyR5Mbg+4w0N73GQ3bmhQwGfo8dPp39PaB8b00R+f0it2tBT8jtQtmxsXZCinJL3xurWqLgvoH7Lsb37ybL8iZsLznImFkgrB5xiu6Kx3nL2XuaosoNV15TmteXwuGTrl8xqWTk2qEsF7iNA1FC3o71FZIPvX0NryoO7etUrfOdg564yJ/znQocaqkKcW1lzcsrlRIb9PD570vm4yV4+f7tPNmxr1+t1r9MCHXqoPvmyLfnikSy/7h4f0iYdO51QNcSRZ7ez1AtkwDG1pyW1Dw+CYqXd+9hmNROP63LtunHVuiCS97YY2tTZU6G/vP5nV6/dzFwb0/i89r2fOXdFf/+B41sfoerF7WF999qJ+6eb1c4bVhfTrd2xSbXlAv3hT24L8vMXGy5nEjZJO27Z91rbtmKSvSLpn2mNsSW5kVicpv/G/JWxXsrzoaAFaGg53OEMVc0n1ywJ+Xb22NqtKhJ+e6NGjL/bpd16xNas79+lsbq5WY1Uop7kI336hU1taqlMD94qhdUVFXu0Mbsqc6xt2qhIhVPyOosYq5+KnP02IsP/8FV3VUp1ztYjLHTjlZS3p42f6VBbwFfykbC5ub/PB9rBOXB7W9nmqQpCcDQ3hMTPt3foHjnVr28qanHv5i+2atXVaVVvuuaXhRNewImZCe5OVVblw2yAOXgrrOwc7NR639LZ9+b/hb2mpVsBnZAwREpat3pHxBbsInM2Wlhq97to1+vyT52etRnAv+qdvv8lWMNkmEEtYujIa0+efPK/XXbsmY1uaYRj61ds36mzfqB5+Mfu7WV48eLJHe9tWZP2aVVUW0N62ej2RZi6Cbdv6P98+KsOQ/iI5bGyxaqwu0/teukn3H+3OaiDnT4/3qCLo182bFi7c9cINPTNVImR6Td3UXKUzHqqOpnvuwoCuaqku+LaKXAYrdg+Na1UO1ZGF9tZ9rRqOxnX/0ZnzcYaipn58rFuv271m3rYPVYYCumlTgx6a5xDh0pUxXegf023JGx6VoYB+91Xb9JPffalu39Kkv73vpF71j4/oJ8e7s2oBONo5pNryQFYrj7c0V+t0lpU0UTOhX/vCfp3rG9V//PL1GQeDhgI+feiV23T88pC+d9jbXJWzvSN6z389q1V15frlm9frWwc69dTZ3Iay/80PT6gqFNAHXz53WF1I166r16EPv3peKx4WMy/P4LWSJtcltSc/NtmHJf2SYRjtkn4g6QMFOboSlM3F0FxSPU9rc7+Qvq5thQ61DyoWz7zbOha39FffP65NzVX6ZY+73+fizkV4JssQoSMc0TPnr+ie3WuKeiLW2lCpS1fGcu7tyneAkVuJUArlUSuqnJOV6SdhlmVr/4UB7SvA3vJtK2vkM6RjHu7qPnG6XzdsaFjQf5u2hkrVVwb17PkrOtM7Mm+tDJJzsippxl7ywTFTz5y/olfsLK07fdkwDEOv2Nmih0/1epp07m6YyScw2rG6VkG/oYPtg7p3/yVtX1WT1+uqqyzgzMDJFCL0j44rYdlqyWHIaqF98OVXKWom9MlHzqb9fCxZiRDKs53BPfmPJyz952NnFTET+oDHvumfuXq1WmrK5mXdY+/wuA61D+qubbnd4bx1c5MOdwzO2DH/wyNd+umJHv3uK7cuiencv3r7RrXUlOkjP/TW32zbtn56oke3X9VUEu9Zk83Vjic5r6vD0bjaMoUITdU625vdmjzLsvXchYGCtzJIUlXIL5+RXSVC91A0p5V2hXbzpkatW1Ghe/fPbFu673CXYnFrXrYyTHbnthad6R3Nq201EzdwvH3L1NbL1oZKffId+/T5X71RAZ+hd//Xfr3rc896PpajHYPatSa7G4xbWqrVOzzueSuFbdv60L0H9cy5K/qHt+7RrVu8tY++fvcabV9Vo3/40ckZ23mm6xsZ1zs/+6wMw9B/vetG/fFrdmjdigr9n28fyfi10z15pl8/OdGj37hrc943QVE4hYoB3y7pc7Ztr5P0GklfMAxjxvc2DOO9hmHsNwxjf2/v/NyFmG91FUG1NVTmXYlwItnzdE0e6dbethUaj1ueJoh//snzOtc3qj/72Z0FS39v3Nig9oGIOsLe7+h/96BTpHLPnvl9A8mkraFSw+PxnNcAdQ1FFfQbaszxxezua1bpQ6/cOi/D+7JVFvCrpjwwI0Q41TOs4WhcNxRgYFRFyK+NTVUZL8h6h8d1snt4QVsZJOfi99p19brvSJfilq3t81gl4655PDOtdPahUz1KWHZRWxkK4ZU7VyliJvT4HOvyBiOm/uWnL+off3xKa+rKs7rjMl150K/tq2r1nQOdOtQ+qLfuay1YQLljdU3GrSI9ySngpXDyvqWlWq/fvUaff/KC+tJUI8Qt58Qt30qEQDKE6B0Z1389cUGvuWZ1xvVvrlDAp3fcsl6PvtinF7Psy8/EvfN457bcgrjbtjTJtqWnJrXpDUVNffg7R7VrTa3eeeuGQhxm0VWGAvrdV27VcxcGdP/RzFVDp7pH1BGOlNRWBteKypAMI30lnSRdvDL3ekfX5uYqDUfj6p1jpsh0p3tHNBSNz0uIYBiGqssCnisRLMtWz/C4VtUVP8z0+Qy96bp1evxM34yVrv/zQoc2NlVpd44ro726Mxkkzmc1wmOn+9VSUzZrBdYdW5t13+/coT/92R3af35Ab/73JzIO7zQTlo53DXueh+DKdrjiY6f79P1Dl/X7r96m1+/2PpvC5zP0e6/apgv9Y/raHHMvxmJxvftzz6pnOKpP/8o+bWiqUkXIrz9/3S6d6h7Rfz1x3vPPtCxbf/2D41pTV65fvS271dOYX17OJDokTR5zvS75scneLelrkmTb9pOSyiXNiLVs2/6kbdv7bNve19y8sOvLCunqtbV5D1c83JFdz1M6162vlyT97tcO6oNffkF/e98Jfenpi3rkVK/O9I6k7gT2j4zrn37yol66tbmg/YzuDtZns6hG+NYLHdrbVq+2xuKWa7snFLm2NHQPOom/L8dVabXlQX3g5VdlNT18PjVWhWachD173rlLfEMBKhEkaeeauowbGtxkfyGHKrr2rKvTWHIy+455rERYt6JCQb8xY2Dfj491q6m6TLvX1c/bz14IN29yNqaka2noGYrqIz88rtv+5qf6+x+d0p7Wen3yHfvyvujf3VqnjnBEQb9R0DtcO1Y7M3DmGhTZM+xWJRX/5F2SPvDyqzQeT1+NYMaT7QwFWvH4309f1Mh4XB/MMAthurff2KZQwKfPZnEi6cVDJ3vVUlOW9Qm4a09rvSqCfj0xKQD7+/tPqm9kXB954zV5hy+l5M3Xr9OWlmr97X0nMt4V/MkJ57l8V47hzHzy+wzVVwTTbheSJlY5z7be0bUptaHBe0vD/Ueccn2vd3GzVVMe9FyJ0JesiCp2W5Xrzdevk21L33hu4nKhMxzRU+f69YY9a+e9EnVTU5XaGipzGgLohWXZeuJ0n27f0jTn3yXo9+k9L9mkr//GLTITtn7+k0/NOcDzTO+IYnEr62sDN0Q44zFE+Jefntaq2nK95yXZX5S/fEeLrmur18d/8mLaisN4wtIHv/yCDncM6uM/v3dKpeErdrToZdtb9I8/PuV5u9x3DnbqcMegfu/V20quEmq58/KO+KykqwzD2GgYRkjO4MTvTHvMRUkvlyTDMHbICREWZ6mBB7vW1OnilbGc72JLTrlSfWUwr9LI1XUV+sDLtqi5ukwvXBrQJx85qz/+n8N6x2ee0cv/4WFt/7P7dMP/e0Cv/5fHNRZL6M9eW9jpz9tX1aqmPOB5LsLJrmGd6BrWPVmknvMlteYxx+GK3cPRkrloKISGqtCMk7D9569oZW1ZXneJJ9u5ulYd4cicz5snTvertjxQlP6z3cm5CCG/ryB7hmcT8PvU1lA55Y5ELG7p4ZO9esWOlpyDqVJRFvDrzm3NeuB4T2r40sX+Mf3x/xzW7X/7oD71yFndtb1FP/jgS/TZd91YkP/W1yaDl1ftXFXQUscdyYqUuSpo3H3kxdrOMN3m5mrds2etPv/k+RkDgM1kJUK+A23dECE8Zupnrl6VdftPY3WZfm7PWn3z+XaFx7xv8pjLeDyhR17s1V3bct8eEAr4dMPGBj1+xqlEeOHigL7w1AW945YNqd+xpSLg9+kP796us32j+uqzs99RlJx5CFevrS2Jfvt0nPev9L9Hbgm5l5kIkvcQwbZtffOFDt28qWHeWlxqygMa8hgiuBVRpRIitDZU6tbNjfr685dS7wPfOdgp25beME9bGSYzDEN3bmvWE2f6PbXWZetE17D6R2O6zWOAtH1Vrb78azcrYdl6+yefmrVi4GjyBmW2Qei6FZUKBXye5iLsP39FT5+7ol+7Y1NOwzMNw9D/vnu7uofGZ1QU2LatD3/3qB443qMPv36XXjVtCK1hGPrz1+2Uadn6f9/PPGQxaib0d/ef1K41tXpDkSuYMVPGMwnbtuOSfkvS/ZKOy9nCcNQwjL80DOP1yYd9SNKvGYZxUNKXJb3TLuYi0XnmPrmz2Xs/3eGOQV2T41DFyT70qm368ntv1qN/8DKd/Kuf0RN/+DJ97ddv0UffulsfeuVWvWxbizY0VeoP796uLS2Fvbvq9xm6YUODnva4oeE7Bzvk9xn62WtLIURwKxFyCxG6BqMle0KVi4aqshk9pfvPO/MQClkaLs19Qfb4GWfScTEqNNyLhKtWVs/7HcdNzdVTTlb///buOzyuu8wX+Pc3TaNpkqaoS7aKiyTHLcaxExO3NEIKaUDgEm4WyNICy0IukF0IsA/LXeBhQ1/6UpPdFHA25FJsxyFxEieOHVe5yE19VK1Rm5Fm5nf/mDmyLKscaWZ0zoy/n+fhIZbbL09GZ868532/72tnetAfCqf9KIPi+toCdA2E8MQbTfjEY/ux6ZvP48m9zbhrdSl2fnoTvnvvqhlDnGZjfaUHTqsJ9yUh72U8pYgwXZaH8jRlqh30WnhwSzVGwlH86IVTF319LFjRkNjr22gQUL5F5xpydf+GhQiORvHYa9N/gFXr6X0t6A+GccuKooT+nGuqPGjoiLXvf/7pQyhwWvGZG5ck5Yx6s7UmH2sXuvHo9pMYDE3+YbV3cAT7GnuxRYddCAqPIwtdU2QiNPUOISfbPOMWp+KcbFjNhhnbzRX7Gs/jTNcg7lxdOuvzquWymlWPMyQa9pwK96wpRVPP8NiDpj/sb8Hq8lws8KSuSD/epiU+DI9GZp3dpYYyrqe2iADE1kY/9sA6RKXEvT95FQ0dl45zHW7tg9VsmPX2AaNBoNJrVzXO8L3nG+C2W3Dv2rIZf+1Urqr0YONiH36w69RFD4Z++MIp/ObVRvz9xkrct37hpL93gceOD2+swjMHWicNsh3vly+fRcv5YTx8c03aP2DJRKruJKSUz0kpF0spq6SUX41/7YtSymfi/3xUSnmNlHKFlHKllPIvqTy01hLd0BAKR3DC3z/25ySL0SBQnJuNtRVu3Lm6FA9uXYR/u3s5fvvBdfjQtZVJ/bsUayvcON05OOPKSykltr3ZimuqvfBpuApN4bKakZNtnnsnQkD7NPZk8kx4ktNyPpZ18ZYkznrWzlB8a+weQnPv8KzelJPJ58zC4gIH1qRgvnWiSq8d57qHEIk/odle74fVbNDs3z3ZNi3Jh8kg8NmnDmFHvR8femslXvrsZnztziuwMAVdHmVuGw596UZcleTkeJ8zC15H1rS5CB39IXjsloSf7idTpc+Bd6wqwW/2nBsbtwDGr3hM/GbMkWXCjXUFY4WW2Vpa6MLVVR786pWzsw7ZmigcieIHuxqwojTnkpCz2VK+Bz/+u3041t6PL99eB0eWKaE/U6+EEPjczUvRNRDCT16cPIzzhROdiEpgi44LnBPfv8Zr6hmecZQBiM16L/TYp201H++pfc2wmg24+YrEilbTcVpNqscZ/P2JrZ1OhZvqiuDMMuGJN5pQ3xbAsfZ+3JHiQMXx1ld6YTEZUjLS8FJDF6p89lk/TFpc4MRjH1oHKYF3/3jPJbkwR1oDqClyzelBipo1j4db+rDreCc+sKECNkti17WHblyCvuFR/CQ+OveH/S34+p+O47YVxfjsjUun/b0f3VQVD1k8MuX1v3dwBN97vgGbl/gy5t4o0+jnrieN+JxZKHRZ57yh4UT7AEYjiYUq6sVYLsIMq6L2NfaiuXdYF6MMijJ3NhrnkIkwEApjIBTW1Zt1otwOC3qHRsaSqZXVX8nYzKDId1rjH8gm/77ZreQhzHOo4nhPfeRqPPz25I79TKbSZ8dIJIqW3mFIKbG93o8N1V5kWzJj3i8n24zP31yDh25cgpc/txWfv7lGN+3+s1VT5Jw2vLYjENRFYXSiT2xZhNGIxI9euPDh8EIRIfG3/t9+cB2+fveKhP6M+6+pQFtfcNJVcLOx7c1WNPUM48EtixLunKotciHXZsb+xvO4vrYAN05ox800q8vzcPMVhfjx305fVHBS7DjWAa/DguU6vl+ZdpyhdwhlM4QqKqryHTitYs1jcDSCZw+04qa6wpQWmJxWE/pD6joR/H1BGATgdegnuT7bYsQtK4rx3KE2/PrVczAZBG6Zx07UbEtsJemuE8kNVxwJR/HamZ45FywXFTjx+APrIARw709eHSskRKMSR1sDWDbHB4zV+Q409Q5NO77x/ecb4LSakrKlbVlJDt6+vAg/330G295swUNPHsC6Sje+cc/yGbsGrGYjvnRrHRo6BvCL3ZNv6vnuzgYMhsL4/M2pvyejuWERYY7qil043DK3ToSDLecBIClryLR2RUkOss3GGdvFtr3ZiiyTATfU6edpRrnbhuY5jDPosW0wUR67BaMROTZ/ufdsLxxZJixNcsBgTZFzytbw3Q1dyHdmoWqWbXzJ5LSa5zQjOFtKq+KprgEc9/ejuXc4Y0YZFB/YUIGPba5Gjm36NmK9qy1y4aR/YMqnJXrtSlroteMdK0vwm1fPoSM+cjEaH2dIRhHhitIc5GQn9t92y9J8LPDY8LOXzsx53W4kKvH9XQ2oKXJha03iLfcGg8A11V7YLEZ8+ba6hP+8dPDQjUsxEo7iOztOXvT1cCSKF453YNMSfWe1eOyxIrjS2aWIRiWae4dnzENQVHntaOoZQig8/Qz9jvoOBIJh3HVl6kYZgNkFK/oDIXgdWboL/7xnTSmCo1H8bk8jNi3JR948r+fbtNiH052DaOxO3qrH/Y29GB6NJPR0vDrfgccfWAeDEHj3j1/F8fZ+NPYMYSAUnnMwbHW+A1JOnevR0NGPPx1px/vXL5xxvEetT1+/GKFwFJ98/E1UeO340fvWqL6Huq62AFuX5uPR7SfR1nfxA71z3YP49atn8c41ZViscvMPzT99XW3SSF1JDk51DmB4ZPaBLX870YlCl3XGvcXpwGw04MoFedOGK45GovjjwTZcV1sAZ5IuXMlQlmdDc+/wWOiPWsoMtB4/OMyVEkanPM15/WwPVpXnJv2GpLZ48g9k0ajEK6e6cc0MSceZQgluPN05iO3xLQZbkvABiJKvpsiFkUh0yhuzDh2HrH5iazXCUYn/iHcjhCPKikd9fI8ZDQIffGsl9jeex7MH2+b0Zzx3qA2nOwfx4JbqpF07vnxbHZ75+AYUpygwT28qvHa856pyPPZa00Xt/G+c60UgGMZWHa52HM/jyIKUuCSks3MghJFwFGUqw4ErfQ5EJWb8wPn0vmYUuqy4OsVbhJRxBjUFtvZAUJf3JKvKclEVD62cz1EGhbKRLJndCLsbumAQwLqqxLomq3yxQoLJKPCen7yKp/c1A5j71raxNY9TjOT84PlTsJqM+LsNyVuTWOlz4H9fvRBl7mz84v61sy4sP3JrHcKThCx+/U/HYTIY8I/XL07aWSn5WESYo2XFLkQlUD9Nm+tkgqMRvHiyC1tr5p4grTdrK9w41h5A39DkbXe7G7rQPTiiq1EGACh12zASiaJjhjyHiZROhMwKVlSKCCH0DY/iuL8/aasdx6uNfyCbOHd63B9LOr46wTfldOGxW+CymnCmawB/re/AyrJc5Dsz5/WUSabb0BCJSnT2h3T7326Bx447V5Xgt3ti3QgjScxESJb3rC3HshIXvvLs0VlvPIpGJb63swHV+Q7clMSxA69j6t3vmeoTWxfBajLgG386Pva1ncc6YDYKbFik73lk5f1r4ppiJTi5VOUDG2VDw3S5CJ39Iew60Yl3rCpJeQCw02pGJCoxrGK7gF+nRQQhYoXCBR5bUjqFZqvCa8cCT3JXPb7U0IXlpblJeZpf6XPg8QfWw2w04Ds7G2AyCCwqmNu1p8Jrh0EADf5LM3wau4ew7UAr3nNVeVI3GAHAP7+9Brs+s3lOW0rKPTZ8dFMVnj3YNhZWua+xF3881IYPXVuZtmOQlwsWEeaoLl4pPDLLkYZXT3djaCSSUa3LayvckBLYe27yboRtb7bCZTVh4xLfPJ9sesrTidmGK+oxwChRHnvsSWr3QCyJW0pgzcLkBwzWFk0erjiXpON0JoRApc+BPad7cKApNntN+lTps8NiNExaROgeCCEqodtOBAB4cMsihKMSP9h1amw7QzLGGZLFaBD42h3L0T0Qwjf/fHzm3zDOX+v9OO7vx8c2V+m63T4deB1Z+PuNVfjTkXa8EX8v33msA1dVeHTVQTgZj1JEmLChQXlvV5uJMDZmNs2ax2cOtCISlbhrdeqfqjutsbwFNSMNsSKCPq9D964txwsPbYbVrE3mz+Yl+Xj5VFdSVj0GgqM40NyXcIDreBVeOx5/YB0KXVasKMud80hllsmIcrdt0k6E//jbKRiFwAMpCFkXQiRUUPvwxiqUu2344rbDGAlH8a9/rI9dj1IUCE/Jo587iTRTnGNFns2Mwy2z60TYUd+BbLMR6zPoievKslxYjIZJRxqGRyL485F2vH150bzMms/GXNc8+vuCcFlNGROCB8SCFYHYOMPesz0wGQRWluUm/e+p8NqRZbr0A9nLp7pR4bVfNu3DQOzD6cl4knImFRUzjdlowKICx6RZHkoXk56flpR7bLhrdQl+91ojmuMfqhJd8ZhsV5Tm4L71C/GbPeewv7FX1e+RMtaFsMBjw606WBucCT741gr4nFn42nPH0Ng9hJMdA2Pt4Ho2/v1rvKZ4cHKpynEGR5YJBa6sKUeXgNgow/LSHCyahzntC0WE6Tt0gqMR9A6NZtSDjWTauMSH4Gh02rFbtfac7kEkKpP+wGOh147tn96In963JqE/Z7INDe19QTy5txl3rynVZbeK1WzEl26rxanOQXzgl69j77le/OP1i2HP0K04mURfdxJpRAiBZSU5ONKmvhNBSokd9X5sWOTVrCKbClazESvKcia9QG+v92NoJILbVsz/LNxMSnKzIcSFGw212gPBjBplAMY9yRkcwetne1FXkpPw+p/JmIwGLCm8OFxxNBLFntPdl80og6IynotQ5s7G4jm2L9L8qClyTbrmMV3yUR7csgjRqMSP4qu4LCb9PbX/9A2Lke/MwsO/PzyW3TCdXSc6cailDx/dVKW7MLl0ZbOY8KnrFmPvuV780x8OAYDu8xCAi8fxxmvqGUK+M2tW91uVXgdOd00+znCsPYAjrQHcOU+z/Uq7fGCGTgRlxbber0NaWV/pQZbJgF3HE89F2N3QBavZgNULchM/2ASOLFPCwZNV+Q6c6Rq86Br647+dRkRKfGRjVaJHTJktSwtwXU0BXjzZhep8B965JrWhpZQcfOdNQF1xDo6392MkrG7H9dG2AFr7grguAwPU1la4cbilD4Ohi9/str3ZikKXdWwVpJ5YzUYUOK1onGUnQrtO09gTYTUbYbMY0d4XxIGm83jLguSPMihqi1w42hoYC4s62HwegyOJJR2nI6V1duvSgozJR8lUNUUudA2Exm7WFf5AvBNBhysexytz23D3laVoi+e56K0TAYjNf3/p1jrUtwXwi91np/21Ukp8d8dJlORm445VvNlMpneuKUWVz44XT3ah0mfHwnixU8/cttgHr65JxhnUbmZQVPrsONUxMGmY4dP7WmAyCNy2cn6KCGrHGdqVYmaGPdxIFmu8+zcZuQi7G7qwtsKju85aRbXPgdGIHLuv7R4I4XevncPtK4pn/b0w3x65tTaWj3NbHQvDaYL/lRJQV+zCaETixCQhJpPZUR+rgqZDe+Bsra3wIBKV2DeuFfX80AheONGBW1cUpTyAaK7K3Nmzz0To02eAUaLcdgv+drIToXAUa1IQqqioLXahd2h07APY7oZuCBF7WnA5WVmWC68jS5PEapqdmviq04ljOB3xfBSfzosIAPCxzdUwxa/DZpM+3/pvWlaILUvz8a2/nkDL+ak7xF453Y19jefx4Y2VsOj03yVdmYwGfPampQCALUvS417FZDQg12aedJxB7WYGRZXPgUAwfElIYzgSxe/3t2Dz0vykB9NNRcmimGmc4UJHlP6vQ1rZtNiHM12DONc99ajKTPyBIE52DGBDtX7vVcY2NMRHGn6++wxC4Sg+ulm/XQiKMrcNzz74Vlx9mT1QSmd8902AsoZlYkjcVHbU+zM2hf3KBXkwGgReGzfS8NyhdoxGJG6fp6r9XJTl2dA8i06ESFSicyCUkbOHHrsF5+KrrVIRqqhQ0u6PxkeBXj7Vhdoi17zvj9ZacW429v7zdViRguwJSq6pNjTEdrNbdBVUOJUytw33xFtErTr94C2EwJdvqwMAPLLtyJS/7rs7GpDvzMI9a8rm62iXletrC/CNu5fjgY3pE2zmtlsuKiKMRqJo6xueUycCgEtyEV5q6EJnfwh3rZ6/zhdXtspOhL7MC3tOtk3xglgi3QhKAHSqV3smomrcmse+4VH86uVzuKmuENX5qc/woMuPPu8k0sQCtw2OLBMOt86ci9ARCOJAc19GjjIAsVmuZcWui3IRtr3ZgiqfHXXFLg1PNr1Stw1tgaDqkZSugRAiUZmRbYPK05VKrx1eR+qeaCyNP9U92hrA8EgE+86dv+xGGSi95NktKHRZL+1ECAThS6Oi8MM31+BH77sSnhR+fyeqzG3DP1y3CNvr/fjzkfZLfn7v2R68crobD1xbmVHZQnoihMA9a8rS6oGHx25B97hMhLbzQUSl+s0Miqr4mNnpCQn3T+1rQa7NjM1L52/LlNpOhI7+ECwmA3Ky9b1FQ0sLvXZUeO14PoFchJcaupBnM49tmdIjl9WMAlcWGjoG8OtXzqI/FMbHNldrfSzKUCwiJMBgEKgtcuGwijWPO4/FLlxbMziFfW2FG282nUdwNILW88PYc6YHt68s0fW8d1leNqQEWqdpnR1PaRvMxIq/O77mMZVdCEDsxmiBx4b6tn7sPdeDkUg0o7aVUGaqKXLiWPvFo2sd/aG0aiF2Ws24sa5Q62PM6O82VGBpoRNfeuYIBibk7Hx3ZwM8dgvec1W5RqcjPZrYiaCMKZa6ZzfOUJybDYvJgFPjigiB4Cj+cqQdt60ontdZeLvFCINQ14lQ6LLq+l5LDzYu9uGVU91zWvUopcTuhi5cXe3V/TrZ6nwHDrf04ee7z2LTEt9Y1zRRsrGIkKC6klhqdyR6aQjPeNvr/SjJzR57CpuJ1lZ4MBKO4mBzH/7nQCsA4PaV+l69Va6seVSZi5DJbYOe+JqsVOYhKGoKXTjaFsDuhm6YDAJr5+HvJEpETZELDR0DCIUv3ID6A0EUpNHT2nRhNhrw1TuuQHsgiG/95cTY1w82n8cLJzrxgbdWpGR7DKUvjyML3eOCFZXVzeWzHGcwGgQqvfaLxhmeO9iGUDg6r6MMQKwjxJFlmrGI4A8E06qYqZVNS3wIhaN45XT3rH/vqc4B+AMhbEiDrslqnwMn/APoGRzBx9mFQCnEIkKClhXnYHg0gjNTrAQCYjt8X2rowtaa/IyuFL8l/gR7z+lubHuzFSvLcrHAo+9kZ2VeUu2GhrEAo5zMe8P2xVuc16RwM4OittiFs92D2F7vx6ryXO4DJt2rKXIhHJVjgVWRqETXQAj5vHlPiSsX5OHeteX4z5fPjHX7fXdnA3KyzXjfugUan470xmO3oHdoBNH4A52m3iGYDAJFObPrRABiuQinuy4UEZ7eFxvNXF46/090nVYzAiqCFTMx7DnZ1lV6YDUb8MIcchF2N8QKD2lRRIjnIqyrdM/LQyG6fLGIkKC6kths1OGWqcMVdzd0ITgaxXUZPMoAALk2C5YWOvHkvmYcbQvovgsBiO1VNhsFmnrUjTO0B4IwGQS89sz74HDn6hJ8+90rx1YPplJNkQtSxhKE9RxSRKS4EK4YG2noHgghKoF83rynzGdvXAq33YKHf38IR1r78Nejftx/zcKxWXEihdtuQVQC54djH7ibeoZRnJs9p81QlV4HGnuGMBKOorF7CK+d7cFdV5Zq8hDIaTUhMDx1J4KUEv4MXDudClazEesrPdg1h1yElxq6UObO1v2aRABYVR4LOv/k1sVaH4UyHIsICar2OZBlMuDINOGK2+s7YLcYcVVl5lcE11a4ca57CAYBvH15kdbHmZHRIFCSq37NY3tfCPnOLN3PxM2Fx5E1b5s0aseFbTJUkdJBhdcOq9kwFq6orCgtSIP1jukqx2bGF26pxcHmPtz/i9fhyDLh/qsrtD4W6ZASDNwTD1ds6h1C2SzzEBSVPjsiUYnGnkE8vb8ZQkCzVbwuq3naYMVAMIzh0UhGjlimwqYl+TjbPYQzXepXPYYjUbx6qjstuhCA2Oa4g4/cwKwpSjkWERJkMhqwtMg1ZSdCNCqx85gf1y72zWsgj1bWVsQKJddUe9Mm2bnMrX7NY1PPEJ88JkFxjhU52WZkm41YyRWHlAaMBoElBc5xRYTYaBOvB6l124pivHWRFx39Idy3fgFybOxCoEspG4W64rkITT3Ds97MoFA2NDR0DOLpfS24pso7p7GIZHBap89E6BgbseR1SI3NY6se1XcjHGzpQ38onFYPPDgiSvOBRYQkqCt24XBrH6S8NFzxcGsf/IFQRm9lGG99pQfOLBPem0bJ2aV5NjT1zjzO8MeDbXjtbM/YmxDNnRACGxZ5cUNdASw63VlPNFFNkQv1bQFIKdHRH+9EYCZCSgkh8LU7r8B96xfggWsrtT4O6dSFToQRDI9E0DUQmnPreaUvluX0xN4mNPYM4c7V2nQhAPEiQmjqToR2pYjAjihVyj02VHrt2DWLXITdJ7sAgKOXRBPw7j0JlhXnoD8YnnSufnt9B4QANi+Zv93CWvI4snDgkRtw0zL9jzIoyt029AyOXLJKbLzm3iF87umDWFmWi49urprH02Wu779nNR5910qtj0GkWk2RC71Do/AHQvAHghDiwhNQSp3SPBu+cvsy5NosWh+FdMoTLyJ0D46gWVnvmDe37gGn1QyfMws7jnXAZjHipmXarUV1Ws3TdiIoY1WF7ERQbeMSH1453Y3hEXWrHl9q6EJdsWusUEVEMSwiJMGyeLjiZLkIO+r9WF2eB89ldKOZbnkBytxk0xQjDZGoxKf+601ICXzn3atgNvLbJlkyeVsJZZ4L4YoBdPQH4bFbeD0g0oE8pRNhYGRs21IiIXiV3lg3wtuWFWm6TlQZZ5is0xUYtzGKY1WqbV6Sj5FwFK+qWPU4NBLGvsbetBplIJovvPtJgsUFThgNAocnFBHa+oZxpDWQ8VsZ0p0yNzlVEeF7Oxvw+tle/Ms76lDu0X8yLxGlxtIiJwDgaFsA/kAobXJfiDKd2WiAy2pCz2Bo7L18rpkIAFAVX5N315XajTIAsU6ESFRieHTyp+b+QBA52WZYzZmfuZUsayvcyDYbVeUivH62F6MRySIC0SSYvJEEVrMRi/Idl4Qr7qiPXaCuq+EMvZ4pTysmy0XYe7YH395xAnesKsEdq0rn+2hEpCMuqxkludljnQjMQyDSD68jC12DIzAZDcg2G+F1zL39/JblRQiORLCuQtuEe6c1dpveHwxP2hHR3sfr0GxZzUZcXeXBE280Y3g0grddUYRrqryT5jPtbuiCxWjAWxbmaXBSIn1jESFJlpXkYNfxDkgpx1q0t9f7Ue62oTpe0SZ9yrOZYbcYL+lECARH8cnH30RJXja+cnudRqcjIj1RwhUDwTDqinK0Pg4RxbntFvQMjGA0HEVpXnZC43JXV3l1EaR3oYgwOunIgr8/xFGGOfjCLbV4dPsJPHeoHf+9txlOqwnX1xTgpmWFuHaxb6yz46WTXVi9IFfTkRYiveJ3RZLUFbvw5BvN6Ihf0IdGwnj5VDfee1U55751TggRW/PYe6GIIKXEP/3+MNoDQTzx4fVwWrlWjIiA2iIndh7zQ4KbGYj0xG234Fz3EM4PjyaUh6Anrvi9R2CKcEV/XxCL8rUvdqSbhV47Hn33KoTCEexu6MJzh9rx16N+PL2/BXaLEZuX5mPjYh+OtgXwmRsWa31cIl1iESFJlpXEnkgdbulDgcuKF092YSQcZR5Cmihz29DYfaGI8NS+FvzPgVZ85obFWF3ONjYiiqkpciEazzjL5xNAIt3wOCzY19iL0GgUazOk/Xz8OMNEkahE50AIhbwOzVmWyYgtSwuwZWkBRiNRvHKqG//vcDv+cqQdzx5sAwBczTwEokmxiJAkNUUuCAEcaQ1ga00BdtT74bSasLbCrfXRSIWyPBt2N3RBSolz3UP44rbDuKrCjY9sqtb6aESkI8qGBgDI5252It1w2y3oGhgBkNhmBj1RuiD7g6OX/Fz3QAiRqGRHVJKYjQZcu9iHaxf78C+31+H1s71o7BnEqrJcrY9GpEssIiSJI8uECo8dh1v6EI1K7DzWiY2LfVz/lSbK3NkYGonAHwjhE4/vh9lowL+/ayWMabaukohSq9xtg91ixOBIhLPIRDrisV/4MF2awGYGPZmuE8EfCAHgesdUMBkNWF/lwfoqbYM1ifSMn3CTqK4kB0daAzjQfB5dAyGOMqQRZRXUQ08ewMHmPvzfO69AcW62xqciIr0xGASWFMZWPfLmnUg/POO2MZS5M+P9e3yw4kTtgSAAXoeISBssIiTRsmIXWs4P44k3mmE0CGxa4tP6SKSS0vr44sku3Lu2DG+7okjjExGRXtUWu2A0iIRWyBFRcrnt44sImdGJYLeYIMRUnQixIkJhDosIRDT/OM6QRHXFsXDFJ/Y24coFeci18QYzXShPLSp9dnzhllqNT0NEevbhjVXYUO2FieNqRLqhFBFybeaxrQbpzmAQcGSZpiwiGATgsfNek4jmH4sISVRXHAvcGo1IXFeTr/FpaDZsFhO+9c4VWF2ex33ARDSt0jxbxsxcE2UKryOWiVCWYd+bLqsZgUnGGfyBIHzOLBYziUgT/LSURHl2C0pys9Fyfph5CGnoztWlWh+BiIiI5iAv3v2ZKXkICqd18k6E9kCIeQhEpBmWL5NsbYUbSwudqPQ5tD4KERER0WXBYjKg0mfHygxbyRcrIlzaidARCLKIQESaYSdCkn31jmUYjUitj0FERER0Wdn+qY0QGbaZ2Wk1j4UojtceCGLNwjwNTkRExE6EpLNZTMjJzoxAHyIiIqJ0YTAIiAyrIkw2zhAcjeD80CgK2YlARBphEYGIiIiISIcmG2foCIQAAPksIhCRRlhEICIiIiLSIafVjP5gGFJeGJVtj483sBOBiLTCIgIRERERkQ45rSaEoxLB0ejY15SMBAYrEpFWWEQgIiIiItIhpzWWszV+pMHPTgQi0hiLCEREREREOuSyxhapBcaFK/oDQWSZDHBlc8kaEWmDRQQiIiIiIh1yxTsRAuM6EdoDIRTmWDNuEwURpQ8WEYiIiIiIdMgZ70Ton9CJUODkKAMRaYdFBCIiIiIiHZoqE6Egh0UEItIOiwhERERERDo0sRNBSgl/IIhCV5aWxyKiyxyLCEREREREOnShiBDrRAgMhxEcjXK9IxFpikUEIiIiIiIdsltMEOJCJ4K/P7bekUUEItISiwhERERERDpkMAg4skxjRYT2PhYRiEh7LCIQEREREemUy2oeW/HoD8SKCIUsIhCRhlhEICIiIiLSKaf1QieCUkTIZ7AiEWmIRQQiIiIiIp2KFRGUToQQcm1mWM1GjU9FRJczFhGIiIiIiHTKaTVfyEQIBFHg5CgDEWmLRQQiIiIiIp0aP87QEQiiIIdFBCLSFosIREREREQ6NX6cIdaJwDwEItIWiwhERERERDqljDNEohKd/SEUshOBiDTGIgIRERERkU45rSaEoxLNvUOISiCf6x2JSGMsIhARERER6ZTTagYAnPQPAAAKWUQgIo2xiEBEREREpFMuqwkA0NAZKyIUuJiJQETaYhGBiIiIiEinnPEiAjsRiEgvWEQgIiIiItIpZZyhoXMARoOAx8FOBCLSFosIREREREQ6pXQiNPj74XNkwWgQGp+IiC53LCIQEREREemU0okwOBJhHgIR6QKLCEREREREOqV0IgBAAfMQiEgHWEQgIiIiItIph8UEEZ9gYBGBiPSARQQiIiIiIp0yGAQcllg3QmEOiwhEpD0WEYiIiIiIdEwZach3MhOBiLTHIgIRERERkY4p4YrsRCAiPWARgYiIiIhIx5ROBGYiEJEesIhARERERKRjLCIQkZ6wiEBEREREpGNOqxnZZiNc49Y9EhFphVciIiIiIiIdu6GuAF5HFoSy65GISEMsIhARERER6dgty4txy/JirY9BRASA4wxEREREREREpBKLCERERERERESkCosIRERERERERKQKiwhEREREREREpAqLCERERERERESkCosIRERERERERKQKiwhEREREREREpAqLCERERERERESkCosIRERERERERKQKiwhEREREREREpAqLCERERERERESkCosIRERERERERKQKiwhEREREREREpAqLCERERERERESkCosIRERERERERKQKiwhEREREREREpAqLCERERERERESkCosIRERERERERKQKiwhEREREREREpAqLCERERERERESkCosIRERERERERKQKiwhEREREREREpAqLCERERERERESkipBSavMXC9EJ4Jwmf3livAC6tD4EUYrw9U2Zjq9xymR8fVOm42ucMpkeX98LpJS+iV/UrIiQroQQe6WUa7Q+B1Eq8PVNmY6vccpkfH1TpuNrnDJZOr2+Oc5ARERERERERKqwiEBEREREREREqrCIMHs/1voARCnE1zdlOr7GKZPx9U2Zjq9xymRp8/pmJgIRERERERERqcJOBCIiIiIiIiJShUUElYQQNwkhjgshGoQQn9P6PESJEkKUCSGeF0IcFUIcEUJ8Mv51txDir0KIk/H/z9P6rERzJYQwCiH2CyGejf+4QgixJ34t/y8hhEXrMxLNlRAiVwjxpBDimBCiXgixntdwyhRCiE/F708OCyEeE0JYeQ2ndCaE+LkQokMIcXjc1ya9ZouY78Rf6weFEKu1O/mlWERQQQhhBPB9AG8DUAvgXiFErbanIkpYGMCnpZS1ANYB+Fj8df05ADuklIsA7Ij/mChdfRJA/bgf/xuAf5dSVgPoBfABTU5FlBzfBvAnKeVSACsQe63zGk5pTwhRAuATANZIKZcBMAJ4N3gNp/T2nwBumvC1qa7ZbwOwKP6/BwD8cJ7OqAqLCOqsBdAgpTwtpRwB8DiA2zU+E1FCpJRtUsp98X/uR+zmswSx1/Yv47/slwDeockBiRIkhCgF8HYAP43/WADYAuDJ+C/h65vSlhAiB8C1AH4GAFLKESnlefAaTpnDBCBbCGECYAPQBl7DKY1JKf8GoGfCl6e6Zt8O4Fcy5lUAuUKIonk5qAosIqhTAqBp3I+b418jyghCiIUAVgHYA6BAStkW/6l2AAVanYsoQY8C+D8AovEfewCcl1KG4z/mtZzSWQWATgC/iI/s/FQIYQev4ZQBpJQtAL4JoBGx4kEfgDfAazhlnqmu2br+/MkiAtFlTgjhAPAUgH+QUgbG/5yMrW/hChdKO0KIWwB0SCnf0PosRCliArAawA+llKsADGLC6AKv4ZSu4nPhtyNWLCsGYMelbeBEGSWdrtksIqjTAqBs3I9L418jSmtCCDNiBYTfSimfjn/Zr7RLxf+/Q6vzESXgGgC3CSHOIjaCtgWx+fHceGsswGs5pbdmAM1Syj3xHz+JWFGB13DKBNcBOCOl7JRSjgJ4GrHrOq/hlGmmumbr+vMniwjqvA5gUTwR1oJYsMszGp+JKCHx+fCfAaiXUn5r3E89A+D98X9+P4Bt8302okRJKT8vpSyVUi5E7Jq9U0r5XgDPA7g7/sv4+qa0JaVsB9AkhFgS/9JWAEfBazhlhkYA64QQtvj9ivL65jWcMs1U1+xnANwX39KwDkDfuLEHzYlY1wTNRAhxM2LztUYAP5dSflXbExElRgixAcCLAA7hwsz4w4jlIvw3gHIA5wC8U0o5MQSGKG0IITYB+IyU8hYhRCVinQluAPsB/C8pZUjD4xHNmRBiJWLBoRYApwHcj9gDIl7DKe0JIb4M4F2IbZPaD+CDiM2E8xpOaUkI8RiATQC8APwAHgHwB0xyzY4Xz76H2BjPEID7pZR7NTj2pFhEICIiIiIiIiJVOM5ARERERERERKqwiEBEREREREREqrCIQERERERERESqsIhARERERERERKqwiEBEREREREREqrCIQERERERERESqsIhARERERERERKqwiEBEREREREREqvx/V2m27hlCIWUAAAAASUVORK5CYII=\n",
      "text/plain": "<Figure size 1296x720 with 1 Axes>"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": "<Figure size 1296x720 with 1 Axes>"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "set_seed(42)    # Set seed for reproducibility\n",
    "model, optimizer, criterion, scheduler = initialize_model(epochs=100)\n",
    "criterion = nn.BCELoss()\n",
    "# criterion = nn.BCEWithLogitsLoss()\n",
    "# criterion = nn.CrossEntropyLoss()\n",
    "train(model, train_dataloader, val_dataloader, epochs=100, evaluation=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.save(model.state_dict(), './Model/MLP_W2V_to_Sigmoid_ord.pt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "PATH = './Model/MLP_W2V_to_Sigmoid_ord.pt'\n",
    "model.load_state_dict(torch.load(PATH))\n",
    "testing_process(model, X_val, y_val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Compute predicted probabilities on the test set\n",
    "probs = model_predict(model, val_dataloader)\n",
    "probs = probs.numpy()\n",
    "\n",
    "# # Evaluate the Bert classifier\n",
    "target = torch.tensor(ext_y)\n",
    "target = target.unsqueeze(dim=1).numpy()\n",
    "evaluate_roc(probs, target)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "task1_model.eval()\n",
    "correct = 0\n",
    "total = 0\n",
    "test_loss = 0\n",
    "outputs_list = []\n",
    "y_list = []\n",
    "\n",
    "with torch.no_grad():\n",
    "    val_loss = 0\n",
    "\n",
    "    for i, data in enumerate(task1_test_dataloader):\n",
    "        x, y = data\n",
    "        x, y = x.float(), y.long()\n",
    "        outputs = task1_model(x)\n",
    "        loss = criterion(outputs, y.unsqueeze(1).float())\n",
    "        _, predicted = torch.max(outputs.data, 1)\n",
    "        outputs_list.append(predicted[:])\n",
    "        total += y.size(0)\n",
    "        correct += (predicted == y).double().sum().item()\n",
    "        val_loss += loss.item()\n",
    "        y_list.append(y)\n",
    "\n",
    "print('Accuracy of the test dataset is: %d %%' % (100 * correct / total))\n",
    "print(\"Loss of validation set: {:.5f}\".format((val_loss / test_size)))\n",
    "acc = (100 * correct / total)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Using the BERT 최종 > MLP"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import re\n",
    "from tqdm import tqdm\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import time\n",
    "\n",
    "%matplotlib inline\n",
    "from sklearn.metrics import accuracy_score, roc_curve, auc, classification_report, f1_score\n",
    "\n",
    "from torch.utils.data import TensorDataset, DataLoader, RandomSampler, SequentialSampler\n",
    "import torch.nn.functional as F\n",
    "\n",
    "import time\n",
    "\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "\n",
    "import random\n",
    "import time\n",
    "\n",
    "from transformers import BertTokenizer\n",
    "from sklearn.metrics import accuracy_score, roc_curve, auc\n",
    "from torch.utils.data import TensorDataset, DataLoader, RandomSampler, SequentialSampler"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "rosetta",
   "language": "python",
   "name": "rosetta"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  },
  "orig_nbformat": 2
 },
 "nbformat": 4,
 "nbformat_minor": 2
}