lstm + attention.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "a9cef1b8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# !pip install build-essential\n",
    "# !pip install pydot\n",
    "# !pip install graphviz"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "c6242a18",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[nltk_data] Downloading package stopwords to\n",
      "[nltk_data]     /Users/peeti_mac/nltk_data...\n",
      "[nltk_data]   Package stopwords is already up-to-date!\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import re\n",
    "\n",
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras.utils import plot_model\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "from keras_preprocessing.sequence import pad_sequences\n",
    "from keras.preprocessing.text import Tokenizer\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Embedding,Flatten,Dense,SpatialDropout1D,LSTM\n",
    "from keras.callbacks import EarlyStopping\n",
    "\n",
    "import math\n",
    "import nltk\n",
    "nltk.download('stopwords')\n",
    "from nltk.corpus import stopwords\n",
    "stop_words = set(stopwords.words('english'))\n",
    "\n",
    "import pandas as pd\n",
    "pd.set_option('max_columns', None)\n",
    "\n",
    "from data_frame import eda_nltk, remove_tags\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "sns.set(style=\"white\")\n",
    "sns.set(style=\"whitegrid\", color_codes = True)\n",
    "\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "b443aace",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      ": reviews_Automotive_5.json.gz\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>reviewerID</th>\n",
       "      <th>asin</th>\n",
       "      <th>reviewerName</th>\n",
       "      <th>helpful</th>\n",
       "      <th>reviewText</th>\n",
       "      <th>overall</th>\n",
       "      <th>summary</th>\n",
       "      <th>unixReviewTime</th>\n",
       "      <th>reviewTime</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>A3F73SC1LY51OO</td>\n",
       "      <td>B00002243X</td>\n",
       "      <td>Alan Montgomery</td>\n",
       "      <td>[4, 4]</td>\n",
       "      <td>I needed a set of jumper cables for my new car...</td>\n",
       "      <td>5</td>\n",
       "      <td>Work Well - Should Have Bought Longer Ones</td>\n",
       "      <td>1313539200</td>\n",
       "      <td>08 17, 2011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>A20S66SKYXULG2</td>\n",
       "      <td>B00002243X</td>\n",
       "      <td>alphonse</td>\n",
       "      <td>[1, 1]</td>\n",
       "      <td>These long cables work fine for my truck, but ...</td>\n",
       "      <td>4</td>\n",
       "      <td>Okay long cables</td>\n",
       "      <td>1315094400</td>\n",
       "      <td>09 4, 2011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>A2I8LFSN2IS5EO</td>\n",
       "      <td>B00002243X</td>\n",
       "      <td>Chris</td>\n",
       "      <td>[0, 0]</td>\n",
       "      <td>Can't comment much on these since they have no...</td>\n",
       "      <td>5</td>\n",
       "      <td>Looks and feels heavy Duty</td>\n",
       "      <td>1374710400</td>\n",
       "      <td>07 25, 2013</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>A3GT2EWQSO45ZG</td>\n",
       "      <td>B00002243X</td>\n",
       "      <td>DeusEx</td>\n",
       "      <td>[19, 19]</td>\n",
       "      <td>I absolutley love Amazon!!!  For the price of ...</td>\n",
       "      <td>5</td>\n",
       "      <td>Excellent choice for Jumper Cables!!!</td>\n",
       "      <td>1292889600</td>\n",
       "      <td>12 21, 2010</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>A3ESWJPAVRPWB4</td>\n",
       "      <td>B00002243X</td>\n",
       "      <td>E. Hernandez</td>\n",
       "      <td>[0, 0]</td>\n",
       "      <td>I purchased the 12' feet long cable set and th...</td>\n",
       "      <td>5</td>\n",
       "      <td>Excellent, High Quality Starter Cables</td>\n",
       "      <td>1341360000</td>\n",
       "      <td>07 4, 2012</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       reviewerID        asin     reviewerName   helpful  \\\n",
       "0  A3F73SC1LY51OO  B00002243X  Alan Montgomery    [4, 4]   \n",
       "1  A20S66SKYXULG2  B00002243X         alphonse    [1, 1]   \n",
       "2  A2I8LFSN2IS5EO  B00002243X            Chris    [0, 0]   \n",
       "3  A3GT2EWQSO45ZG  B00002243X           DeusEx  [19, 19]   \n",
       "4  A3ESWJPAVRPWB4  B00002243X     E. Hernandez    [0, 0]   \n",
       "\n",
       "                                          reviewText  overall  \\\n",
       "0  I needed a set of jumper cables for my new car...        5   \n",
       "1  These long cables work fine for my truck, but ...        4   \n",
       "2  Can't comment much on these since they have no...        5   \n",
       "3  I absolutley love Amazon!!!  For the price of ...        5   \n",
       "4  I purchased the 12' feet long cable set and th...        5   \n",
       "\n",
       "                                      summary  unixReviewTime   reviewTime  \n",
       "0  Work Well - Should Have Bought Longer Ones      1313539200  08 17, 2011  \n",
       "1                            Okay long cables      1315094400   09 4, 2011  \n",
       "2                  Looks and feels heavy Duty      1374710400  07 25, 2013  \n",
       "3       Excellent choice for Jumper Cables!!!      1292889600  12 21, 2010  \n",
       "4      Excellent, High Quality Starter Cables      1341360000   07 4, 2012  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataframe = input(': ') #reviews_Automotive_5.json.gz\n",
    "df = eda_nltk(dataframe)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "4d6d6a39",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(20473, 10)\n"
     ]
    }
   ],
   "source": [
    "df['positive'] = df.overall.apply(lambda x: 1 if x >= 4 else 0)\n",
    "print(df.shape)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8923c202",
   "metadata": {},
   "source": [
    "# Data Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "6b703b33",
   "metadata": {},
   "outputs": [],
   "source": [
    "review_df = df[['reviewText','positive']]\n",
    "review_df['reviewText']=review_df['reviewText'].apply(lambda cw : remove_tags(cw))\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "d4665545",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.countplot(x='positive',data = review_df, palette = 'hls')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3f1cde22",
   "metadata": {},
   "source": [
    "# imbalanced data"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1c914840",
   "metadata": {},
   "source": [
    "## start random select"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "6a43ceb3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "We should select only 2578 count for both positive/ negative review\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "5156"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "max_cnt = review_df.query('positive == 0').shape[0]\n",
    "print(f'We should select only {max_cnt} count for both positive/ negative review')\n",
    "sample_df = review_df.groupby(['positive']).apply(lambda x: x.sample(max_cnt))\n",
    "sample_df.shape[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f6cb8e0f",
   "metadata": {},
   "source": [
    "### remove stopwords from the review. Stopwords are commonly used words like ‘and’, ‘the’, ‘at’ that do not add any special meaning or significance to a sentence"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "c6e7c1c0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "positive       \n",
       "0         11490    product best car mean alex customer service pe...\n",
       "          13616    wheel cleaner better stuff sold autozone types...\n",
       "          12794    one bottom step getting going buy another one ...\n",
       "          6474     kit work way thought would hard time keeping s...\n",
       "          9011     pin 10mm wide disc brake holes disk wide lock ...\n",
       "                                         ...                        \n",
       "1         5871     pt cruiser already replaced nozzles steep deal...\n",
       "          1893     bought 2003 bmw r 1150r even female understand...\n",
       "          15748    inexpensive goggle gets job done seem fog quic...\n",
       "          6036     using years never replace water pump thermosta...\n",
       "          92       rain x 800002243 glass treatment 7 oz simple s...\n",
       "Name: reviewText, Length: 5156, dtype: object"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_df['reviewText'] = sample_df['reviewText'].apply(lambda x: ' '.join([word for word in x.split() if word not in (stop_words)]))\n",
    "sample_df['reviewText']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "e5bbec76",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['product best car mean alex customer service person helpful help returning package even broke bulb said prob deduct bulb cost u still return',\n",
       " 'wheel cleaner better stuff sold autozone types stores quite good enough bmw wheels notorious brake dust use little brush get stubborn grime']"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "review_list = [i for i in sample_df['reviewText']]\n",
    "review_list[:2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "91e55e12",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['product best car mean alex customer service person helpful help returning package even broke bulb said prob deduct bulb cost u still return',\n",
       "       'wheel cleaner better stuff sold autozone types stores quite good enough bmw wheels notorious brake dust use little brush get stubborn grime',\n",
       "       'one bottom step getting going buy another one put end one keeps moving around well made good quality unit',\n",
       "       ...,\n",
       "       'inexpensive goggle gets job done seem fog quickly first put riding wind clears fog',\n",
       "       'using years never replace water pump thermostat radiator worth money',\n",
       "       'rain x 800002243 glass treatment 7 oz simple stuff works great applied correctly rain slide right'],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_df['reviewText'].values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "970f16d6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[   0,    0,    0, ..., 1235,   48,  508],\n",
       "       [   0,    0,    0, ...,  241,    9, 1106],\n",
       "       [   0,    0,    0, ...,    6,   42,   85],\n",
       "       ...,\n",
       "       [   0,    0,    0, ..., 1099, 1130, 1162],\n",
       "       [   0,    0,    0, ..., 1971,  141,  143],\n",
       "       [   0,    0,    0, ...,  267,  798,   73]], dtype=int32)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "max_features = 2000\n",
    "tokenizer = Tokenizer(num_words=max_features, split=' ')\n",
    "tokenizer.fit_on_texts(sample_df['reviewText'].values)\n",
    "X = tokenizer.texts_to_sequences(sample_df['reviewText'].values)\n",
    "X = pad_sequences(X)\n",
    "X"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "b1e8d922",
   "metadata": {},
   "outputs": [],
   "source": [
    "# t=Tokenizer()\n",
    "# t.fit_on_texts(review_list)\n",
    "# sequences =t.texts_to_sequences(review_list)\n",
    "# X = pad_sequences(sequences)\n",
    "# print(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "e1412635",
   "metadata": {},
   "outputs": [],
   "source": [
    "# print('sequences : ',sequences[:2],'\\n')\n",
    "\n",
    "\n",
    "# N = len([item for sublist in sequences[:2] for item in sublist])\n",
    "\n",
    "# print('word_index : ',{k: t.word_index[k] for k in list(t.word_index)[:N]})\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "4353b43c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# len_mat=[]\n",
    "# for i in range(len(sequences)):\n",
    "#     len_mat.append(len(sequences[i]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "a358ce3a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# len(len_mat)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "9cc59971",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average length of each review :  48.846392552366176\n",
      "Percentage of reviews with positive sentiment is 50.0%\n",
      "Percentage of reviews with negative sentiment is 50.0%\n"
     ]
    }
   ],
   "source": [
    "s = 0.0\n",
    "for i in sample_df['reviewText']:\n",
    "    word_list = i.split()\n",
    "    s = s + len(word_list)\n",
    "print(\"Average length of each review : \",s/sample_df.shape[0])\n",
    "pos = 0\n",
    "for i in range(sample_df.shape[0]):\n",
    "    if sample_df.iloc[i]['positive'] == 1:\n",
    "        pos = pos + 1\n",
    "neg = sample_df.shape[0]-pos\n",
    "print(\"Percentage of reviews with positive sentiment is \"+str(pos/sample_df.shape[0]*100)+\"%\")\n",
    "print(\"Percentage of reviews with negative sentiment is \"+str(neg/sample_df.shape[0]*100)+\"%\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a81daaf8",
   "metadata": {},
   "source": [
    "# Build LSTM Model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ea83c4a8",
   "metadata": {},
   "source": [
    "#### For the models we will use, we need to define a maximum sequence length (max number of tokens per review that we'll feed into our model). We'll cut off any tokens longer than that, and we'll pad shorter reviews with zeros at the end."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "4b635705",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " embedding (Embedding)       (None, 896, 64)           128000    \n",
      "                                                                 \n",
      " spatial_dropout1d (SpatialD  (None, 896, 64)          0         \n",
      " ropout1D)                                                       \n",
      "                                                                 \n",
      " lstm (LSTM)                 (None, 16)                5184      \n",
      "                                                                 \n",
      " dense (Dense)               (None, 1)                 17        \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 133,201\n",
      "Trainable params: 133,201\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-09-13 21:56:39.693060: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 AVX512F AVX512_VNNI FMA\n",
      "To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\n"
     ]
    }
   ],
   "source": [
    "max_features = 2000\n",
    "embed_dim = 64\n",
    "lstm_out = 16\n",
    "\n",
    "model = Sequential()\n",
    "model.add(Embedding(max_features, embed_dim,input_length = X.shape[1]))\n",
    "model.add(SpatialDropout1D(0.2))\n",
    "model.add(LSTM(lstm_out, dropout=0.1, recurrent_dropout=0.2))\n",
    "model.add(Dense(1,activation='sigmoid',trainable=True))\n",
    "model.compile(loss = 'binary_crossentropy', optimizer='adam', metrics = ['accuracy'])\n",
    "\n",
    "print(model.summary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "e7171429",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAV4AAAH3CAYAAADzODc8AAAABmJLR0QA/wD/AP+gvaeTAAAgAElEQVR4nOydeVyUVdvHfwMDMsoiYKi4kFPIooBPoCAkRuEL8hKkyUth4YoiSmpWLomm5vakjxtEYCp+Insy9zcwcQlLZAkwNwRzGVwQN3QEYYCZud4/eLkfxhmYBRzAzvfz4fPhPst1Xefc577mvq9z7vvwiIjAYDAYDL1h0N4GMBgMxt8N5ngZDAZDzzDHy2AwGHqGOV4Gg8HQM3xdK3p5eSE3N7ctbWEwGIxOxZ07d9CrVy+t6+nseAEgKioKERERrRHB+Bvw2Wef4dVXX8W0adPa25QOwdixYzFr1iy8+eab7W0KQ0du3LiBCRMm6Fy/VY5XKBTijTfeaI0Ixt8AKysr9O/fn42V/8fY2BjOzs6sPzoxxcXFrarPYrwMBoOhZ5jjZTAYDD3DHC+DwWDoGeZ4GQwGQ88wx8tgMBh6hjleBoPB0DOd3vHev3+/TeXJZDJUVFRolf/kyZM2tUFT2ktve/B3aqsuSKXS9jahQ9PR+qfTOt5vvvkGrq6umDp1apvIq6qqwooVK/Dyyy/j66+/1jg/Pz8f3bt3x9GjR9vEDk1pL73twd+prapwd3fHuHHjMG7cOFy/fl0hTyKRYN26dcjJycHmzZsRExODmTNnory8XElOXl4ePvvsM8TExGD58uUoLS3VVxM0IiMjA5GRkYiLi8PmzZtx69YthfysrCzMmDEDERER+Pzzz1FTU6NSjkQiwcaNG7Fy5Ur8+OOPAICcnBzu/0Z+++03rl8HDx6slP9cIR3x9PSk1atX61q91dTU1NCgQYMoJCSkTeRJpVJ68OAB8fl8WrFihcb5lZWVtHLlSnrw4EGb2KEp7aWXiOivv/7SqnxAQADNnz9fZ32dqa2a0LNnT/rhhx80Lh8YGKgyva6ujsLDw6mkpIRLS05OJh6PR97e3lRbW6tURyQSka2trcq89uTAgQPk6upK1dXVRER06dIlsrS0JJFIREREaWlp5OnpSXfv3iWpVEpRUVHk5eVFUqlUQc7p06fJx8eHDh48qKQjMzOTZs6cqVJ///79KSkpSWN7L126RADozp07GtdpSqe94zUxMYGtrW2byTM0NIS1tTVMTU21yjc1NcWiRYtgbW3dZrZoQnvplUqlWLhwoV51/p3aqg2xsbHw9vbGwIEDubSoqCiYm5vj9OnTmDFjhlKdfv36QSgUwtjYWJ+mqiUxMRGhoaEQCAQAAEdHR9jZ2SEtLQ1SqRSzZs1CSEgIbGxsYGhoiFWrVuHcuXPYunUrJyMvLw/+/v5Yv349QkJClHSMHDkSfD4f6enpemtXc+jd8dbX16tMr6ur4/6XyWRK+arSDA0N1ZZRp/fZus/K1CRfLpcr6W7aHrlc3qze1sSedNHbNK2l/mqO2bNn4/Lly1rXay3PtlWT/n0ebW3pXOqTgoIC7Ny5Ex9++KFSnpubG8LCwrB9+3Zs3rxZIc/AwAB8vuovBTTXR5qOZUD9ddYclpaW2LdvHxc+kEqlKC0thZeXF4qKinD9+nW89tprXPkePXrAxcUFqampABpCgREREViwYAE8PT2b1TNt2jRER0ejsrJSJzvbCr043vr6eixevBjR0dHw9fWFj48P965zVlYWIiIi4OrqitLSUgQEBMDExAT+/v6orq7G8ePH4eDgADMzM3z++ecq5R86dAj29vYwNzfH9OnTIZFI1OptJCkpCcHBwVi6dCnmzJnD1dUkv7y8HF988QX69++PU6dOAQAyMzMRFhYGLy8viEQivPXWW+jatavCBSKXy7F48WLMnDkTy5cvx7hx47B48WIsWbIEBQUFavtTF70nTpxAeHg4xowZg9zcXLi4uKBbt24YP348nj59ipSUFHh6emLixIkAgHv37mHevHlwcXHB7t27AQBfffUV9u3bh1u3bmH69OnYtm2bWltby7Nt1aR/n1db09PTYWFhwfV5e7Jlyxb4+/vD0tJSKY/H4yElJQVDhgzBvHnzcOLEiRZlxcfHIy4uDnPmzMHw4cNx6NAhAJqNZUCz60wdkydPRlFREUJCQvDo0SNMnToVMTExeO2117gJ9Ge/hjhgwABOT1xcHG7cuIHJkycjPz8fWVlZqK2tVdLj7OwMIkJiYqJW9rU5OgUoSLsYb2xsLO3atYuIiOrr68nDw4OcnJyovr6ebt26RUFBQWRlZUWbNm2ix48f05EjRwgAvfvuu/Tdd9+RWCymuLg4AkAXL17k5AYGBpJQKKS1a9dSTk4OffTRRwSAPvvsM7V6iYgSEhLI39+fZDIZERFlZGQQAC6Gqy6/oqKCduzYQQAoMzOTiIhu3rxJfn5+1LNnT1q3bh2VlZVRUlISAaDDhw8TEdHatWupb9++XDuio6NJIBDQ0aNHqby8XG1/6qK3uLiYHB0dydXVlRYsWEAZGRk0ZswYAkCffvopERG9/fbbNGLECE6PWCwmAJSYmKhgq6urqwZn/T+0Jsb7bFs16d/n1dazZ8/SqFGj6PLlyzq1pZHWxnhlMhkJBIJm45UjR44kIqLS0lKysbEha2trunr1Kpf/xhtvcP8nJibS+++/zx2npaURj8ejvXv3atTXROqvM03ZunUrGRoaUvfu3Wn9+vVc+o0bNwgA+fj4KJQfPXo0WVlZERGRk5MT9e7dmxYsWEBLly4loVBIjo6OdP36dSU9vr6+9O677yqk6TvGqxfHKxAIKDk5mVJSUiglJYVGjRpFAOjChQtERPTxxx/Tyy+/rFBHKBTSpEmTuOPr168TADpw4ACXFhgYqDS5NnDgQDI1NaW7d++2qFcsFpOVlRXt37+fqyuXy8nU1JRWrFihNr+RP//8U8EBEhFNnz6dnJ2duePq6moyMDCgzZs3ExGRl5cXjR49msvfv38/AaDTp09r1J+66g0KClIYvHK5nFxcXMjU1JTkcjmFh4crOCMianfHS6TcVnXtJGq/tmpCax1voyNauXKlyvKNjpeI6PfffycjIyMaPHgwVVZWEtF/HK9EIiETExPavXu3Qn0PDw+yt7cnIvV9re4604arV69SSEgImZmZUffu3enXX3/l8iZOnEh8Pp/WrVtHDx8+pD179tBLL71E7u7uJBaLycDAgBYtWsSVF4lExOfzKTg4WEnPuHHjSCgUKqTp2/G26rOQmmJiYoKAgADu2M/PDwBgY2MDQHXctEePHqAmGyA3PlJVVVW1qCsoKAgbN27EiRMnWtR76tQpVFRUYMCAAVw+j8dDly5dADQE6lvKb0RVvMzQ0BA8Ho87FggEsLCw4OJKQqEQhw8fhkwmg6GhITfR0bVr1xbb1lq9BgYGMDMzU2jPmDFjsHz5cpXLjzoKz7ZVXTuBzttWTWhcBtZ4/bTE66+/joSEBEybNg2RkZHYs2cPl5ebmwuJRII+ffoo1PH09ERCQgIqKirU9nVRUZHa61sTiouLMWnSJPz8888oKyvD6NGjERwcjKKiIvTv3x9bt26Fo6MjMjIyUFRUBC8vL9y/fx+zZ8/GlStXIJfLFSbb7ezsMGTIEBw7dgxyuRwGBv+JqtrY2OD69euorKxUGCP6RC+O99GjRzAzM1MZjwIaLoqmJ7c12Nvbg8/nw8jIqEW9jWsEm1uYry6/JdS1ZeHChfj111/x9ddfIyIiAqmpqfjoo4/g5uamtS5t9KqiV69e4PF47TYAdUHXsdIZ26qKRoem6QRRVFQUzp07h/j4eCxbtoxLbxzjhYWF8Pb25tL79OkDY2NjCAQCtX1dVVWl9vrWhOXLl8PHxwfW1tawtrZGRkYGvL29kZqaikWLFoHP52P+/PmYP38+ACAyMhJWVlaYMmUKtxKiurpaQaa3tzfOnDkDqVSqtIpD1U2UPtHbqoavvvpK4Xj//v24c+dOm+vJzMyEm5sbXF1dW9TbeCd7+vRpJRlyuVxtfmtwcHDABx98AHNzc6SlpSEhIQGbNm1qlUxNeXbWOS8vD0KhEKamphAIBCqfKJ5tb9MnkY7Mi9pWe3t7mJqa4t69exrX2bBhA958802sWLGCS2v8oc/KylIoKxaL4e7uzjm0lnBwcADQ+uv77Nmz6NmzJ3fs7OyM8PBwFBYWKpW9cuUK0tPT8csvv6BXr16wsLCAo6Mjzp8/r1BOKpXC1dVVyenevn0btra27bqkTi+ONyAgAGvWrMEnn3yC7OxsJCcn448//kDv3r0BADU1NUpvoUilUoVZycZlV8+uOmj6KyeRSHDy5Els3rwZ9vb2Ler18fGBs7Mz1q1bh0uXLgFoOPk1NTW4d++e2vxGGu1uateTJ0+UZlSlUilXNi4uDiKRCE5OTvDw8IBYLFYbQnkWXfQCgEgkUlg2dOTIESxZsgRAwxrPc+fO4e7du5DJZNi4cSMAKLTXysoKIpEIdXV1yM/P18pmXXm2rZq0E2j7tpaWlmLu3Llt/pq6tvB4PLi5ual0vBKJBA8ePFBK5/P52L17t0LobNCgQQgJCcHx48cV6vz2229cP6nra3XX2YYNG/Dll1+qXTY5fvx4HDhwQCGtrKwMQUFBCmnFxcUICgrCvn37MHToUC59y5Yt2Lt3L0QiEZeWnp6OpUuXKukqKytTWJrWLugUGSbtJtfKysrI19eXABAA8vDwoNu3bxMR0bFjx8je3p4MDAxo48aN9OjRI4qPjycTExPq27cv/fjjj3T79m2aPXs2ASA/Pz86c+YMETW87fLqq69SUFAQrV+/nsaOHavwxkpLeomILl68SPb29iQQCGjYsGE0d+5c6t69O7m7u9PPP/+sNl8kEtGECRMIAI0dO5ZKSkro0KFDZGtrS0ZGRhQfH0/l5eW0atUqAkBubm508uRJSkpKIh6Px9kFgIyMjGjq1Kka9aeueoODg8nW1pYmTJhAO3fupICAAFq0aBHJ5XIialgZYWNjQwKBgIYOHUoXLlwgADRixAg6fvw4ERHl5OSQQCAgBwcHKigo0Mje1kyuPdvWdevWqW0nET2XtqakpCjN6OtCW7y5tmbNGqWJv8LCQoqMjCQDAwOKi4ujK1euKNU7f/68woRTZWUlhYaGkqurK61du5ZCQ0Pp22+/JSLSaEwRNX+dyWQyMjc3JwCUnp7eYhvFYjGFh4fTO++8Q99//z3FxMTQkiVLiKhhYjQ7O5tWr15NY8eOpaysLJUykpKS6L333qNt27ZRWFgY7dy5U6mMRCIhCwsLpXP4Qq5qaKS0tJRKSkq4wd8WyOVyunbtGhUVFXHLvrTRK5PJ6OLFi3T//n0iIioqKtIqXxc2bdpEOTk5VF5eTiUlJVRQUEBHjhyhwMBAysnJabX85ggODqbAwEASi8VUWFhIYrFYqUxtbS1dvHiR60tV7a2urlZ6VbMlWruqQReeV1ubLsvSlbZwvHV1deTm5kY3b97UWv/Dhw+V0ioqKujSpUtay2qKquvs1q1btH37dvrpp580kiEWi6moqIhqamq4tPr6esrNzeWuQXVcu3atWV+wa9cuGjx4sJIveCFXNTTSv3//NpfJ4/EUHp+01WtgYABnZ2fu2MnJSat8bcnJyUFCQgJiY2PB4/EU4lqHDx9WOH5emJub4x//+IfKvMaNGBtR1V5NYn8dhbZuq1AobFsDdcTIyAjbtm3DsmXLFF6b1QQrKyulNEtLy1ZNjgGqrzMbGxvk5+fjX//6l0YyzM3NYW5urpDG5/MxbNgwje1oyR/88MMP2LNnT5tN5uuKXh0vo2Gy5urVq/Dz84Ovry88PDxQV1eH3NxcuLq64uWXX35uup88edLhPo/3vHiR2vrkyRNuksnJyYn7MXB3d4efnx++++47la8Otzf19fVITU3F6tWr23UFAdAwabpkyRLMmjWLmxCsqKjgYsJNX4vWB8zx6pnhw4fj6tWrOHbsGI4ePYpz587B2dkZMTEx6NKlC0aPHt1ifUtLS+zatUsrnVKpFDt27ECXLl1gbGyMpKQkTJ06tdnvUnRmXsS2vv3228jIyADQsNSr6V14REQErl27hpKSEs6hdBSMjIwwadKk9jYDAJCdnY3Y2FiFJ8o7d+7gl19+AdDwwSFt7qpbC49It/UyXl5eeOedd7BgwYK2tonxghEYGIghQ4ZgzZo17W1Kh6BXr17YuHEj3nvvvfY2haEjxcXFcHJywp07d9CrVy+t63faz0IyGAxGZ4U5XgaDwdAzzPEyGAyGntF5ck0mk6GgoAApKSltaA7jRaSsrAyGhoZsrPw/jW9YPvsWJqPz0NrPHeg8uTZkyBCUlpbCwsKiVQYwXnwqKirA4/FavU70ReHOnTswNzdHt27d2tsUho7U19ejrKxM58k1ne94TUxMMH/+fLaqgaEWtqpBEbaqofPTuKpBV1iMl8FgMPQMc7wMBoOhZ5jjZTAYDD3DHC+DwWDoGeZ4GQwGQ88wx8tgMBh6hjleBoPB0DPM8TIYLwAvyreHnxcdrX869fd4z507h4MHD3Kv7/Xu3Rsffvjhc/2YeHPk5eVhz5493KaVffv2hb+/v16/8cl4MXF3d+d2Vfjqq68UdliQSCSIj4+Hl5cXCgsLUVxcDB6Ph7i4OKU3qpqO0V69emHChAmws7PTa1taIiMjA6mpqbCzs8NLL72EsWPHom/fvlx+VlYWUlNTIRaLMWDAACxevFjlbigSiQTffPMNnj59ildffRXh4eHIycnB7du3ER4ezpX77bffsHnzZgANL0TExcUp5D9XdNowiHTbc+15UFlZSYaGhmRra0t1dXV61f3XX38pHN+8eZMA0ODBg6m+vl6vtnRk2mPPtWfPTUeS2xZ7rhE17LsWHh5OJSUlXFpycjLxeDzy9vam2tpapToikYhsbW1V5rUnBw4cIFdXV6quriaihj3NLC0tSSQSERFRWloaeXp60t27d0kqlVJUVBR5eXkp7f13+vRp8vHxUdj0tpHMzEyaOXOmSv363nOt04caTE1NYWVlBXt7exgZGelNr1QqxcKFCxXS+vbtCwMDA9jb24PP79QPE50aVeemI8vVldjYWHh7e2PgwIFcWlRUFMzNzXH69GnMmDFDqU6/fv0gFAphbGysT1PVkpiYiNDQUO4O1tHREXZ2dkhLS4NUKsWsWbMQEhICGxsbGBoaYtWqVTh37pzCfnN5eXnw9/fH+vXrERISoqRj5MiR4PP5SE9P11u7mqPTO16gYUPKZ51u0z2U5HK5ynpN02UymVY6Z8+ejcuXL6u0RdNBra1OVdTX16tMb2x/c23vLOjSR82dm9acb03l6ouCggLs3LlT5V5rbm5uCAsLw/bt27lH6UYMDAyavSlork80uZYaaW48qsPS0hL79u1DTU0NgIYfudLSUnh5eaGoqAjXr1/Ha6+9xpXv0aMHXFxckJqaCgCoqqpCREQEFixYAE9Pz2b1TJs2DdHR0aisrNTJzrbihXC8TcnMzERYWBi8vLwgEonw1ltvoWvXrgoD9MSJEwgPD8eYMWOQm5sLFxcXdOvWDePHj8fTp0+RkpICT09PTJw4EQBw7949zJs3Dy4uLti9eze++uor7Nu3D7du3cL06dOxbds2rWzMy8vD7NmzsXz5cvj4+GDGjBncgP7nP/8JFxcXeHh4cPtBAUBSUhJee+01HD16FPX19Vi8eDGio6Ph6+sLHx8fFBcXAwBOnjyJ8PBwDBs2DJmZmbCzs0NoaGgre1X/xMfHIy4uDnPmzMHw4cNx6NAhANDp3Kg737rKBYD09HRYWFjg1KlTeu2fLVu2wN/fX+UX33g8HlJSUjBkyBDMmzcPJ06caFFWc32tybUEoMXxqCmTJ09GUVERQkJC8OjRI0ydOhUxMTF47bXXcP/+fQBAbm6uQp0BAwZweuLi4nDjxg1MnjwZ+fn5yMrKQm1trZIeZ2dnEBESExO1sq/N0SlAQR0nxkvUEDPz9/cnooY4q5+fH/Xs2ZPWrVtHZWVllJSURADo8OHDRERUXFxMjo6O5OrqSgsWLKCMjAwaM2YMAaBPP/2UiIjefvttGjFiBKdDLBYTAEpMTCQioujoaHJ1dVWyhc/nU3h4eLO2ymQycnZ2puTkZCIi+vXXXwkAZWRkcGVCQ0PJ2tpaIX6VmZlJH330ERERxcbG0q5du4iIqL6+njw8PMjJyYnq6+vpxo0bNGrUKOrVqxfFx8fTzp07ae3atdp3ahuibYw3MTGR3n//fe44LS2NeDwe7d27l4i0PzeanG9d5BIRnT17lkaNGkWXL1/WuH2tjfHKZDISCATNxitHjhxJRESlpaVkY2ND1tbWdPXqVS7/jTfe4P5vqa81uZaIWh6P2rB161YyNDSk7t270/r167n0GzduEADy8fFRKD969GiysrIiIiInJyfq3bs3LViwgJYuXUpCoZAcHR3p+vXrSnp8fX3p3XffVUhjMd5W0rdvXwwcOBDW1taYN28et9LBwMAAf/31FwDAwcEBQqEQZmZmWL16NUaNGoW9e/fCxcUFiYmJICJ07dpVQa65uXmb2GdgYAAfHx+8+eabAMB9Wq6srIwrs3LlSlRUVGD//v1c2nfffYcZM2bg3r17+Pbbb1FVVYWdO3fi+++/h6WlJS5duoSSkhL069ePi3dHR0cjMjISn332WZvYrg9qa2sxd+5cjBkzhksLCgqCu7s79wlSbc+NJudbF7kA4OrqioyMDNjb22vUvrbg9u3bqKmpga2tbYvl+vfvj7179+LJkycIDQ3lVtw0oq6vNbmW1I1HbXjzzTfx3//935DJZFixYgUyMzMBNMSlJ06ciNzcXKxfvx4VFRXYu3cv8vPzMWDAADx58gQlJSWYNGkSVq9ejS+++AInTpzAlStXEBsbq6THxsYGZ86c0cq2tuaFc7wAYGhoCB6Pxx0LBAJYWFgoxHUMDAxgZmbGHfN4PIwZMwZVVVUoLy9/rvYlJydDJpNh4cKFSEpKAqAYGxs0aBBCQkKwcuVKAIBYLEZ5eTkcHR1RVFQEExMTBAQEwM/PD35+fvj2229RWlqKV155hWu/qalpp9zSPDc3FxKJBH369FFI9/T0xF9//YWKigqd5Lbn+W5rSktLATQ4EHW8/vrrSEhIwIULFxAZGakQo9Wkr9VdS5qMR00oLi7G+PHjsX37dmRlZaFbt24IDg7GjRs3AABbt27Fl19+iYyMDHz66aeoqKjA/fv3MWbMGFy5cgVyuVzhh8jOzg5DhgzBsWPHlOLSNjY2uH79ervGeV/IqfemA0UbevXqBR6Pp3CBtiVEhKqqKuzatQuHDh3C999/D1NTUyxdulSp7KJFi+Dp6Ym0tDRcu3YNkZGRABomER49egQzM7MXckeHW7duAQAKCwvh7e3Npffp0wfGxsYq123qyvM+38+LRoerqeOIiorCuXPnEB8fj2XLlnHpmvS1umuprcZj43yHtbU1rK2tkZGRAW9vb6SmpmLRokXg8/mYP38+5s+fDwCIjIyElZUVpkyZwo2J6upqBZne3t44c+YMpFKp0oQ3j8dDly5ddLa3tbyQd7ya8uwMbF5eHoRCIUxNTSEQCJQezQDFWV3Sctek77//HsePH0d0dDTWrVuH7t27Nytr2LBheOutt7By5UocPHiQexx0cHAA0LCQvin79+9v9T5QHQE3NzcADYvlmyIWi+Hu7g6BQKDzuWnpfAN4Luf8eWBvbw9TU1Pcu3dP4zobNmzAm2++iRUrVnBpmvS1OtpqPJ49exY9e/bkjp2dnREeHo7CwkKlsleuXEF6ejp++eUX9OrVCxYWFnB0dMT58+cVykmlUri6uio53du3b8PW1rZdl9R1escrlUpRUVGBJ0+ecGlPnjxRmtGUSqXcUpVGRCKRwhKaI0eOYMmSJQAa4krnzp3D3bt3IZPJsHHjRgDgBruVlRVEIhHq6uqQn58PAHj8+DGkUik3C9uUwsJC7N69G1ZWVgD+M0O7e/duAA13Ds8OskWLFiE7Oxuenp7ccjl7e3sEBARgzZo1+OSTT5CdnY3k5GT88ccf6N27N9fWpv3RmWgMsxw/fhwPHjzg0n/77bdWnRug5fOtq9zS0lLMnTtX5Tl/XvB4PLi5ual0vBKJRKHfGuHz+di9e7fCW2+a9LW6a0ndeNywYQO+/PJLta/sjh8/HgcOHFBIKysrQ1BQkEJacXExgoKCsG/fPgwdOpRL37JlC/bu3QuRSMSlpaenq3yaLCsrU1ia1i7oNCVHHWNVw9mzZ2natGkEgLp06UL//Oc/afPmzWRra0tGRkYUHx9P5eXltGrVKgJAbm5udPLkSSIiCg4OJltbW5owYQLt3LmTAgICaNGiRSSXy4moYXWEjY0NCQQCGjp0KF24cIEA0IgRI+j48eOUk5NDAoGAHBwcqKCggHJycigsLIwAEI/HIy8vL3rrrbdoxIgR9MorrxCfz6eVK1eSXC6n//qv/yIA5OHhQZmZmeTk5ER9+/alM2fOKLXRzc2Ne3unkbKyMvL19SUAnJzbt28TEdH+/ftpwIABBICWLl1KFRUVz/ksqEfbVQ2VlZUUGhpKrq6utHbtWgoNDaVvv/2Wy9f23BCpP9+6yk1JSVGa5VdHW7y5tmbNGqUVFoWFhRQZGUkGBgYUFxdHV65cUap3/vx5Cg4O5o5b6utDhw5pdC01Nx5lMhmZm5sTAEpPT2+xjWKxmMLDw+mdd96h77//nmJiYmjJkiVERCSXyyk7O5tWr15NY8eOpaysLJUykpKS6L333qNt27ZRWFgY7dy5U6mMRCIhCwsLpfOl71UNndrxtobg4GAKDAwksVhMhYWFJBaLlcrU1tbSxYsXSSaTERFRUVGRQn51dbXSK4uacvfuXe6if/r0qcrXnR89ekRhYWHNyigtLaWSkhIF59ER0fWV4YqKCrp06ZLKPG3PjSbnWxe5RKSwVEsT2sLx1tXVkZubG928eVMr3UREDx8+VEprqa81RdV4vHXrFm3fvp1++uknjWSIxWIqKiqimpoaLq2+vp5yc3Pp/hw8iZoAACAASURBVP37Gsm4du0ad/6eZdeuXTR48GCla0bfjveFnFzTBnNzc/zjH/9QmWdsbAxnZ2fu+NldRVsz0dN0RvrZZUyNrF+/HtHR0c3K6N+/v876OwOWlpbNTtjoem5aOt+6yhUKhc3Ke14YGRlh27ZtWLZsmcJrs5rQGO5qSkt9rSmqxqONjQ3y8/Pxr3/9SyMZ5ubmSsv4+Hy+Vh+bahpOeZYffvgBe/bs0XkCvq3o9DFeXXny5EmHjIMePHgQvXv3xhtvvIEnT55w630ZraOjnm9NePLkCQoLC1FYWKgwT+Hu7g4/Pz9899137Whd89TX1yM1NRWrV69u1xUEQMME6eLFizFr1ixuQrCiooLr16avReuDv90dr1QqxY4dO9ClSxcYGxsjKSkJU6dO7TBrXp2dneHt7Q1fX19Mnz69vc3p9HT0860Jb7/9NjIyMgA0LPVqetcdERGBa9euoaSkhHMoHQUjIyNMmjSpvc0AAGRnZyM2NlZh5cSdO3e41/JjY2P1+gnXv53j5fP5iIqKQlRUVHubohJ7e3vs3bu3vc14Yejo51sTGt/Ya472CHV0Nnx8fJTSBg0ahEGDBrWDNX/jUAODwWC0F8zxMhgMhp5hjpfBYDD0DI9It3cgvby8cO/evRd+SROj9RQXF6Nbt27o169fm8mUSCSor6/vdN9ZABreYrSzs4O1tXV7m8LQkerqavzxxx+4c+eO0t52mqCz4/3222+5j2wwGPrm1KlTKCkpwZQpU9rbFMbfmE8++YT71oc26Ox4GYz2ZM2aNTh48CCys7Pb2xQGQ2tYjJfBYDD0DHO8DAaDoWeY42UwGAw9wxwvg8Fg6BnmeBkMBkPPMMfLYDAYeoY5XgaDwdAzzPEyGAyGnmGOl8FgMPQMc7wMBoOhZ5jjZTAYDD3DHC+DwWDoGeZ4GQwGQ88wx8tgMBh6hjleBoPB0DPM8TIYDIaeYY6XwWAw9AxzvAwGg6FnmONlMBgMPcMcL4PBYOgZ5ngZDAZDzzDHy2AwGHqGOV4Gg8HQM8zxMhgMhp5hjpfBYDD0DHO8DAaDoWeY42UwGAw9wxwvg8Fg6BnmeBkMBkPPMMfLYDAYeoZHRNTeRjAYLSGTyeDo6Ig7d+5waVKpFDKZDF26dOHSDA0N8csvv2D48OHtYSaDoTHsjpfR4TE0NIS/vz9qa2vx9OlTPH36FLW1tZBKpdzx06dP0aVLF3h6era3uQyGWpjjZXQKIiIiIJPJms3v0qULJkyYAAMDNqQZHR8WamB0CogIffr0UQg3PEtBQQFee+01PVrFYOgGuz1gdAp4PB4+/PBDhZhuU/r168ecLqPTwBwvo9Pw/vvvo7a2Vim9S5cumDJlSjtYxGDoBgs1MDoV9vb2uHLlilJ6SUkJBg4c2A4WMRjaw+54GZ2KyMhIpXDD4MGDmdNldCqY42V0KiIiIlBXV8cdd+nSBZMmTWpHixgM7WGhBkanY8iQITh79iyAhkm3mzdvok+fPu1sFYOhOeyOl9HpmDhxIoyNjcHj8TB8+HDmdBmdDuZ4GZ2O8PBwSKVS8Hg8TJgwob3NYTC0hoUaGJ0SX19fZGdno7y8HNbW1u1tDoOhFXx9KXr8+DE2btyoL3WMFxxLS0sIhUJs2bKlvU1hvCAEBgbCy8tLL7r06niXLVsGT09PmJiY6EstQwX379/H7du3MWTIkPY2RWdkMhmMjIyQmZnZalmlpaWora1lS9L+xuTn56N79+4vnuNt5N///jdefvllfatlNCElJQVffPFFmzitF4E5c+ZAJBLhwIED7W0Ko53Q900Im1xjMBgMPcMcL4PBYOgZ5ngZDAZDzzDHy2AwGHqGOV4Gg8HQM8zxMhgMhp5hjpfBYDD0zN/O8X7zzTdwdXVFaGhom8irqqrCihUr0K9fP3z55Zca5+fn58PAwABHjx5tEzv0SWe2vb2RSCRYuHAhZsyYgZiYGMycORPl5eVK5fLy8vDZZ58hJiYGy5cvR2lpaTtY2zwZGRmIjIxEXFwcNm/ejFu3bimVycrKwowZMxAREYHPP/8cNTU1KmVJJBJs3LgRK1euxI8//ggAOHXqFPf/CwnpievXrxMAun79ur5UqqSmpoYGDRpEISEhbSJPKpXSgwcPiM/n04oVKzTOr6yspJUrV9KDBw/axA5t2LFjB9nZ2elcvz1t/+uvv9pc5uzZsyk0NLTN5T5LXV0dhYeHU0lJCRERJScnE4/HI29vb6qtrVUqLxKJyNbWVmVee3LgwAFydXWl6upqIiK6dOkSWVpakkgk4sqkpaWRp6cn3b17l6RSKUVFRZGXlxdJpVIFWadPnyYfHx86ePCgkp7MzEyaOXPm823M/+Pm5kYbNmzQiy4ior/dHa+JiQlsbW3bTJ6hoSGsra1hamqqVb6pqSkWLVrUKT/w0l62S6VSLFy4UK8625LY2Fh4e3tzryZHRUXB3Nwcp0+fxowZM5TK9+vXD0KhEMbGxvo2tUUSExMRGhoKgUAAAHB0dISdnR3S0tIANJynWbNmISQkBDY2NjA0NMSqVatw7tw5bN26lZOTl5cHf39/rF+/HiEhIUp6Ro4cCT6fj/T0dP00TI90KsdbX1+vlNZ0NwKZTKaUryrN0NBQbRl1ep+t+6xMTfLlcrmS7qbtkcvlzeqVSqXNG6wHdLG9aZq6PlfF7Nmzcfny5RbldlQKCgqwc+dOfPjhhwrpbm5uCAsLw/bt27F582aFPAMDA/D5zb/V31wfajqGWhrXLWFpaYl9+/ZxoQOpVIrS0lLuOwdFRUW4fv26wq7PPXr0gIuLC1JTUwE0hOAiIiKwYMECeHp6Nqtr2rRpiI6ORmVlpU62dlQ6vOOtr6/H4sWLER0dDV9fX/j4+KC4uBhZWVmIiIiAq6srSktLERAQABMTE/j7+6O6uhrHjx+Hg4MDzMzM8Pnnn6uUfejQIdjb28Pc3BzTp0+HRCJRq7eRpKQkBAcHY+nSpZgzZ45CXXX55eXl+OKLL9C/f3+cOnUKAJCZmYmwsDB4eXlBJBLhrbfeQteuXRUuVLlcjsWLF2PmzJlYvnw5xo0bh8WLF2PJkiUoKChok/5Why62nzhxAuHh4RgzZgxyc3Ph4uKCbt26Yfz48Xj69CmAhu9HeHp6YuLEiQCAe/fuYd68eXBxccHu3bvx1VdfYd++fbh16xamT5+Obdu2AQDS09NhYWHB2dJR2bJlC/z9/WFpaamQzuPxkJKSgiFDhmDevHk4ceKEWlnx8fGIi4vDnDlzMHz4cBw6dAiAZmNI3bjWhMmTJ6OoqAghISF49OgRpk6dipiYGM7R3r9/HwCQm5urUG/AgAGcrri4ONy4cQOTJ09Gfn4+srKyVO4g7ezsDCJCYmKiVjZ2ePQV09A1xhsbG0u7du0iIqL6+nry8PAgJycnEolEFBQURFZWVrRp0yZ6/PgxHTlyhADQu+++S9999x2JxWKKi4sjAHTx4kVOZmBgIAmFQlq7di3l5OTQRx99RADos88+U6u3vr6eEhISyN/fn2QyGRERZWRkEAAuhqsuv6Kignbs2EEAKDMzk4iIbt68SX5+ftSzZ09at24dlZWVUVJSEgGgw4cPExHR2rVrqW/fvpyN0dHRJBAI6OjRo1ReXq5xn7YmxquL7cXFxeTo6Eiurq60YMECysjIoDFjxhAA+vTTTznZb7/9No0YMYI7FovFBIASExO59rq6uirYc/bsWRo1ahRdvnxZp/YQPf8Yr0wmI4FAoDJeOXLkSCIiKi0tJRsbG7K2tqarV69y+W+88YZC+cTERHr//fe547S0NOLxeLR3716NxlBL41obtm7dSoaGhtS9e3dav369Qt6NGzcIAPn4+Cikjx49mqysrIiIyMnJiXr37k0LFiygpUuXklAoJEdHR5X+wdfXl959912t7NMWfcd4O7TjvXv3LgkEAkpOTqaUlBRKSUmhUaNGEQC6cOECffzxx/Tyyy8r1BEKhTRp0iQlvQcOHODSAgMDlSbXBg4cSKampmr1nj59mqysrGj//v1cXblcTqamprRixQoSi8Ut5jfy559/KjgvIqLp06eTs7Mzd1xdXU0GBga0efNmIiLy8vKi0aNHc/n79+/nbNKG1k6u6WJ7UFCQwoUol8vJxcWFTE1NSS6XExFReHi4guMlIrWOty143o630RGtXLlSKa/R8RIR/f7772RkZESDBw+myspKIlJ0vBKJhExMTGj37t0KMjw8PMje3p6IWj4P6q4nbbh69SqFhISQmZkZde/enX799VeF/IkTJxKfz6d169bRw4cPac+ePfTSSy+Ru7s7icViMjAwoEWLFnHlRSIR8fl8Cg4OVtI1btw4EgqFWtmnLfp2vHr/LKQ2FBUVwcTEBAEBAVyan58fAHBB+2fp0aMHqMmmGo2PdlVVVS3qCgoKwsaNG1FeXo7i4uJm9Z47dw4VFRUYMGAAl8fj8bgtx/Py8lrMb0RV7M7Q0BA8Ho87FggEsLCw4OJbQqEQhw8fhkwmg6GhITfp0rVr1xbb1tboYruBgQHMzMy4fB6PhzFjxmD58uUoLy9H7969n7/h7UTjUjAbG5sWy73++utISEjAtGnTEBkZiT179ijk5+bmQiKRKO0x5+npiYSEBFRUVLR4HtRdT5pSXFyMSZMm4eeff0ZZWRlGjx6N4OBgFBUVoX///gCArVu3wtHRERkZGSgqKoKXlxfu37+P2bNn48qVK5DL5QqT3HZ2dhgyZAiOHTsGuVwOA4P/REFtbGxw/fp1VFZWKoyhzkyHjvFWVVXh0aNHMDMzQ//+/RX+TExMwOPxFAZZa7C3twefz4elpWWLeh88eAAAePLkiUo5jesZm8tvCXVtWbhwIUxMTPD111/j4cOHSE1NxUcffQQ3NzetdbU1upyHXr16gcfjvTAXU3M0OjVNJoiioqIwa9Ys7N+/H8uWLVPIaxxbhYWFCul9+vSBsbExBAJBi+dB3fWkKcuXL4ePjw+sra3h4uKCjIwM8Pl8buIMaPhxnj9/Po4cOYJt27bh999/h5WVFaZMmYJXXnkFAFBdXa0g19vbG/X19SonjlXdvHRmOrTjdXBwAAB89dVXCun79+/HnTt32lRXZmYm3Nzc0KVLlxb1Ng7Q06dPK8mQy+XcnW5z+a3BwcEBH3zwAczNzZGWloaEhARs2rSpVTL1ybOz6Hl5eRAKhdxSO4FAoPLJpGm/USfcItDe3h6mpqa4d++eRuU3bNiAN998EytWrFBIb/yBzcrKUkgXi8Vwd3fnlnc1R1tdT2fPnkXPnj25Y2dnZ4SHhyv9IDRy5coVpKen45dffkGvXr1gYWEBR0dHnD9/XqGcVCqFq6ur0vK527dvw9bWtsMtq2sNHdrx2tvbIyAgAGvWrMEnn3yC7OxsJCcn448//kDv3r1RU1Oj9DaMVCpVmB1t/PV8dtVB019biUSCkydPcst5WtI7btw4ODs7Y926dbh06RKAhoFYU1ODe/fuwcfHp8X8RhrtbmrXkydPlGZ2pVIpVzYuLg4ikQhOTk7w8PCAWCxWG0J5HuhiOwCIRCKFJVBHjhzBkiVLuON+/frh3LlzuHv3LmQyGbdHX2O/WVlZQSQSoa6uDvn5+QAaHuPnzp3LzaR3RHg8Htzc3JQcr0Qi4Z6gmsLn87F7926FcBUADBo0CCEhITh+/LhCvd9++43rx5bOg7rrCWhw+l9++WWLyxXHjx+vtFtHWVkZgoKClMoWFxcjKCgI+/btw9ChQ7n0LVu2YO/evRCJRFxaeno6li5dqiSjrKxMYWnaC4G+gsm6rmooKysjX19fAkAAyMPDg27fvk3Hjh0je3t7MjAwoI0bN9KjR48oPj6eTExMqG/fvvTjjz/S7du3afbs2QSA/Pz86MyZM0TU8ObNq6++SkFBQbR+/XoaO3as0pszzeklIrp48SLZ29uTQCCgYcOG0dy5c6l79+7k7u5OP//8s9p8kUhEEyZMIAA0duxYKikpoUOHDpGtrS0ZGRlRfHw8lZeX06pVqwgAubm50cmTJykpKYl4PB5nEwAyMjKiqVOnatWnrZlc09X24OBgsrW1pQkTJtDOnTspICCAFi1axE2sETWsjrCxsSGBQEBDhw6lCxcuEAAaMWIEHT9+nHJyckggEJCDgwMVFBQQEVFKSorCrL0u6OPNtTVr1ihMDBYWFlJkZCQZGBhQXFwcXblyRanO+fPnlSabKisrKTQ0lFxdXWnt2rUUGhpK3377LRGRRuehpXEtk8nI3NycAFB6enqzbRGLxRQeHk7vvPMOff/99xQTE0NLlizh8uVyOWVnZ9Pq1atp7NixlJWVpVJOUlISvffee7Rt2zYKCwujnTt3KpWRSCRkYWHRqvOrCWxVQzOUlpZSSUmJwoXaGuRyOV27do2Kioq4ZV/a6JXJZHTx4kW6f/8+EREVFRVpla8LmzZtopycHCovL6eSkhIqKCigI0eOUGBgIOXk5Ggsp7WrGnQhODiYAgMDSSwWU2FhIYnFYpXlamtr6eLFi9w5ebbfqqurlV47bbr8Shf04Xjr6urIzc2Nbt68qVW9hw8fqkyvqKigS5cu6WxPc+P61q1btH37dvrpp5/UyhCLxVRUVEQ1NTUK6fX19ZSbm8uNfXVcu3at2Wtw165dNHjw4Da77puDrWpohsbZ0raCx+MpPcppo9fAwADOzs7csZOTk1b52pKTk4OEhATExsaCx+MpxNgOHz6scNyRMTc3xz/+8Y9m842NjVvsN1VxTKFQ2HYGPieMjIywbds2LFu2TOG1WXVYWVmpTLe0tFR6GUMbmhvXNjY2yM/Px7/+9S+1MszNzWFubq6UzufzMWzYMI1taek6/OGHH7Bnz542m0TvKHQax/t3h4hw9epV+Pn5wdfXFx4eHqirq0Nubi5cXV07/M7NT548affXnNsbd3d3+Pn54bvvvlN6dbgjUF9fj9TUVKxevbrdVxDI5XIsWbIEs2bN4iYFXySY4+0kDB8+HFevXsWxY8dw9OhRnDt3Ds7OzoiJidHozr29kEql2LFjB7p06QJjY2MkJSVh6tSpzX7b4kUnIiIC165dQ0lJSYdzKEZGRpg0aVJ7mwEAyM7ORmxsbKd5ktMW5ng7EXZ2dpgyZQqmTJnS3qZoDJ/PR1RUFKKiotrblA5DZwiNtDc+Pj7tbcJzpUMvJ2MwGIwXEeZ4GQwGQ88wx8tgMBh6hjleBoPB0DN6n1x7/fXXW/yqPuP5U1VVhadPn3b4JWj6oqKiAgBYf/yNaRwD+kLvHnDOnDno0aOHvtUymnDq1CkcPHgQX3zxRXub0iHYtWsXHj58iNjY2PY2hdFOLF++XK/69O54x40bx+4sOgDHjh3jttn5u/Pnn39CJBKx/vgb0/hBJn3BYrwMBoOhZ5jjZTAYDD3DHC+DwWDoGeZ4GQwGQ88wx8tgMBh6hjleBoPB0DN/e8fblnt1yWQytQuxny2jy27EjI7F3/07wx2FznQeOuwrZAsXLoRYLG42f8aMGXBxcdFZ/jfffIOvv/4aAwYMwMGDB3WWAzS8CbZhwwYkJydj+vTpWLx4sUZl8vPzMWzYMBw5cgSjRo1qlQ0M/SORSLBs2TI8fvwYPB4PPB4PcXFx6NWrl0K5vLw87NmzB1VVVejVqxcmTJgAOzu7drJamYyMDKSmpsLOzg4vvfQSxo4di759+yqUycrKQmpqKsRiMQYMGIDFixer3A1EIpHgm2++wdOnT/Hqq68iPDxcJ5s0kbNw4UJ069aNu95ycnJw+/ZtnXXqFX3tMaTtnmvl5eUEgN59912qqqri/m7cuEExMTH073//u1X21NTU0KBBgygkJKRVcoiIpFIpPXjwgPh8Pq1YsULjMpWVlbRy5Up68OBBq23QhvbYc42I6K+//uqQMnXZc62uro7Cw8OppKSEiIiSk5OJx+ORt7c31dbWKpUXiURka2urMq89OXDgALm6ulJ1dTUREV26dIksLS1JJBJxZdLS0sjT05Pu3r1LUqmUoqKiyMvLS2nvu9OnT5OPj4/SxrHaoomc9PR0AkAzZsxQSM/MzKSZM2dqrVPfe6512FBDz549YWBgAD6fj27dunF//fr1w/Lly1FfX98q+SYmJrC1tW0TWw0NDWFtbQ1TU1OtypiammLRokWwtrZuEzs6MlKpFAsXLuzwMjUlNjYW3t7eGDhwIAAgKioK5ubmOH36NGbMmKFUvl+/fhAKhTA2Nta3qS2SmJiI0NBQ7u7V0dERdnZ2SEtLA9DQx7NmzUJISAhsbGxgaGiIVatW4dy5cwp7x+Xl5cHf3x/r169HSEiIzvZoIufBgwf45ptv4OjoqJQ3cuRI8Pl8pKen62yDPuiwjhdo2DBSFRYWFvjggw+447q6Ou5/mUymVF5VGgCl7WeaKwegRUffWK+l7WyaKyOXy5X0Nm2PXC5vVmZ7xLRa6qOWmD17Ni5fvqyQ1rRtushVJfNZuc+DgoIC7Ny5U2nfNDc3N4SFhWH79u3YvHmzQl7jTYQqWmq7pmNB1xsRS0tL7Nu3DzU1NQAaxlRpaSm8vLwAAEVFRbh+/Tpee+01rk6PHj3g4uKC1NRUAA1htIiICCxYsACenp462aGNnI8//hjr1q2DkZGRyvxp06YhOjoalZWVOtvyvOnQjlcV58+fx969ewE0xJ0iIiLg6uqK0tJSBAQEwMTEBP7+/qiursbx48fh4OAAMzMzfP75583KPHToEOzt7WFubo7p06dDIpEAaBjMixcvRnR0NHx9feHj44Pi4mKuXlJSEoKDg7F06VLMmTOHq9eU5sqUl5fjiy++QP/+/XHq1CkAQGZmJsLCwuDl5QWRSIS33noLXbt2VbjA5XI5Fi9ejJkzZ2L58uUYN24cFi9ejCVLlqCgoKD1HdwM8fHxiIuLw5w5czB8+HAcOnQIAJCSkgJPT0/uOwf37t3DvHnz4OLigt27dwMAvvrqK+zbtw+3bt3C9OnTMX78eISHh2PMmDHIzc2Fi4sLunXrhvHjx+Pp06cayX1W5rZt2wAA6enpsLCw4Pr0ebBlyxb4+/sr7fLL4/GQkpKCIUOGYN68eThx4kSLcprrU0CzsaBufGrC5MmTUVRUhJCQEDx69AhTp05FTEwM52gbJ59zc3MV6g0YMIDTFRcXhxs3bmDy5MnIz89HVlYWamtrtbJDUzk7duyAr68v7O3tm5Xj7OwMIkJiYqLWNugNfcU0tI3xEhHx+XwaNmwYpaSkUEpKCq1fv56GDx/OxXdv3bpFQUFBZGVlRZs2baLHjx/TkSNHuNjwd999R2KxmOLi4ggAXbx4UUF+YGAgCYVCWrt2LeXk5NBHH31EAOizzz4jIqLY2FjatWsXERHV19eTh4cHOTk5UX19PSUkJJC/vz/JZDIiIsrIyCAACjHelspUVFTQjh07CABlZmYSEdHNmzfJz8+PevbsSevWraOysjJKSkoiAHT48GEiIlq7di317duX0xEdHU0CgYCOHj1K5eXlGvWrtjHexMREev/997njtLQ04vF4tHfvXiIievvtt2nEiBFcvlgsJgCUmJioYKerqysRERUXF5OjoyO5urrSggULKCMjg8aMGUMA6NNPP+XqqJPbVGYjZ8+epVGjRtHly5c1bp82MV6ZTEYCgUBlHHHkyJFERFRaWko2NjZkbW1NV69e5fLfeOMN7n91farJWGhpfGrD1q1bydDQkLp3707r169XyLtx4wYBIB8fH4X00aNHk5WVFREROTk5Ue/evWnBggW0dOlSEgqF5OjoqNW1romcq1ev0ocffsiVd3FxUYrxNuLr60vvvvuuxrr1HePt8I7Xz8+PMjMzKTMzk/73f/+X3nvvPYWJtY8//phefvllhXpCoZAmTZqkpPvAgQMK5QIDA5Um1wYOHEimpqZ09+5dEggElJyczDn+UaNGEQA6ffo0WVlZ0f79+7l6crmcTE1NOccrFovVlvnzzz8VHC8R0fTp08nZ2Zk7rq6uJgMDA9q8eTMREXl5edHo0aO5/P3793M2aYo2jlcikZCJiQnt3r1bId3Dw4Ps7e2JiCg8PFzBQRJRi46XiCgoKEjhYpbL5eTi4kKmpqYkl8s1kqvK8eqCNo630RGtXLlSKa/R8RIR/f7772RkZESDBw+myspKIvqP49WkT4laHgstjc8LFy5o3HaiBocWEhJCZmZm1L17d/r1118V8idOnEh8Pp/WrVtHDx8+pD179tBLL71E7u7uJBaLycDAgBYtWsSVF4lExOfzKTg4WGMb1MmRSqU0btw4un//PpffkuMdN24cCYVCjfXr2/F22OVkjdjY2GDkyJHcsaenJ06ePMkdq4qr9ujRA0TEHTc+ElZVVanVFxQUhI0bN+LEiRMwMTFBQEAAl+fn5wcAOHfuHCoqKhS2VefxeOjSpQt3nJeXp7aMqpifoaEheDwedywQCGBhYcHFq4RCIQ4fPgyZTAZDQ0NusqZr165q26YLubm5kEgk6NOnj0K6p6cnEhISdP6AtIGBAczMzLhjHo+HMWPGYPny5SgvL0fv3r1bZffzorS0FEDDuGyJ119/HQkJCZg2bRoiIyOxZ88eLk+TPrWysmpxLBQVFTU7PtXZ1pTi4mJMmjQJP//8M8rKyjB69GgEBwejqKgI/fv3BwBs3boVjo6OyMjIQFFREby8vHD//n3Mnj0bV65cgVwuV5iotrOzw5AhQ3Ds2DHI5fJm52qaok7OmjVrMGDAAJw5c4bLr6qqws2bN3H06FH4+PgoXAM2Nja4fv06KisrFcZZR6HTxXhfeukljBs3jjtuXD/ZVtjb24PP58PIyAiPHj2CmZkZ+vfvr/D34MEDAC2//HDr1i21ZVShri0LFy6EiYkJvv76azx8+BCpqan46KOP4ObmppUeTWlsR2FhoUJ6nz59YGxsrHItp6706tULPB6vQ14omZbhYwAAIABJREFUjTQ6NU0mbqKiojBr1izs378fy5Yt49I17dOWxkJVVVWz49PExETj9ixfvhw+Pj6wtraGi4sLMjIywOfzuYkzoOEGYf78+Thy5Ai2bduG33//HVZWVpgyZQpeeeUVAEB1dbWCXG9vb9TX12s8AaxOTnZ2NpKTkxEWFsb9lZaWIiMjA2FhYSgvL1eS+eyNTkei0zneRvbu3YsbN260udzMzEy4ubnB1dUVQMPEUFP279/PDezTp08r1W+ceW68022pjC44ODjggw8+gLm5OdLS0pCQkIBNmzbpLE8djQ49KytLIV0sFsPd3R0CgQACgUDl08Sz7Wz6FAIoz8Tn5eVBKBRyS+40kfuszOeNvb09TE1Nce/ePY3Kb9iwAW+++SZWrFjBpWnSp+pwcHAAoHp83rlzRyPbAODs2bPo2bMnd+zs7Izw8HClH4VGrly5gvT0dPzyyy/o1asXLCws4OjoiPPnzyuUk0qlcHV11Xj5nDo5P//8Mx4/fqzwN2jQIEyZMgWPHz+GUChUqHf79m3Y2tp2uOV7jXRYxyuRSCCVSlWuFLh+/TqSk5PRv39/1NTUcEthGpFKpQqzoY2/uqpkNf2FlUgkOHnyJDZv3gx7e3sEBARgzZo1+OSTT7hf3D/++APjxo2Ds7Mz1q1bh0uXLgFoGMA1NTXcBenj46O2TKPdTe168uSJ0kyuVCrlysbFxUEkEsHJyQkeHh4Qi8UahVB0ZdCgQQgJCcHx48e5O30A+O2337BkyRIADWtUz507h7t370Imk3Ff82/qnKysrCASiVBXV4f8/HwAgEgkUlhKdeTIEU6mJnJVySwtLcXcuXPb9FXwpvB4PLi5uSk5XolEotA/jfD5fOzevVsh5KRJnwItj4WWxmdjmGbDhg348ssvW7zrHD9+PA4cOKCQVlZWhqCgIKWyxcXFCAoKwr59+zB06FAufcuWLdi7dy9EIhGXlp6ejqVLl3LHmtiiiRxNKSsrU1gC1+HQVzBZ28m16OhoAkBdu3al//mf/6HIyEgaP348BQQEULdu3eif//wnHTt2jOzt7cnAwIA2btxIjx49ovj4eDIxMaG+ffvSjz/+SLdv36bZs2cTAPLz86MzZ85wOg4cOECvvvoqBQUF0fr162ns2LEKb8uUlZWRr68vASAA5OHhQbdv3yYioosXL5K9vT0JBAIaNmwYzZ07l7p3707u7u70888/qy2TkJBAEyZMIAA0duxYKikpoUOHDpGtrS0ZGRlRfHw8lZeX06pVqwgAubm50cmTJykpKYl4PB5nEwAyMjKiqVOnanwutF3VUFlZSaGhoeTq6kpr166l0NBQ+vbbb7n8mzdvko2NDQkEAho6dChduHCBANCIESPo+PHjRESUk5NDAoGAHBwcqKCggIKDg8nW1pYmTJhAO3fupICAAFq0aBE3saaJ3GdlEhGlpKQozPxrgrZvrq1Zs0ZhUq+wsJAiIyPJwMCA4uLi6MqVK0p1zp8/rzDZpK5PNRkLLY1PmUxG5ubmBIDS09ObbYtYLKbw8HB655136Pvvv6eYmBhasmQJly+Xyyk7O5tWr15NY8eOpaysLJVykpKS6L333qNt27ZRWFgY7dy5k8vT1BZ1cp6luck1iURCFhYWWo0BtqpBz8jlcrp27RoVFRVxy76epbS0lEpKShScAlHDgLp48SI301pUVKRUV5My2rBp0ybKycmh8vJyKikpoYKCAjpy5AgFBgZSTk6ORjJ0fWW4oqKCLl26pDKvtraWLl68yPWhqnZWV1dzr5kGBwdTYGAgicViKiwsJLFYrJPcpjIbabqESxO0dbx1dXXk5uZGN2/e1ErPw4cPldJa6lNNaW583rp1i7Zv304//fSTWhlisZiKioqopqZGIb2+vp5yc3MVVhO0xLVr11ReR9rY0pIcTdi1axcNHjxYqT9agjleRrNkZ2fTwIEDVQ6oOXPmaNy37fWthqY0Ot6OgC7fasjPz9fqKaM9qKuro5iYGJJIJO1til5tefvt/2PvzuOautL/gX8CYVMQUYuAC4WWiihLqxYUxVIXLENBVErFKi6lKtWx1lrRAe2Cii2tHcUfBVxgis7Uuk+lFatFK7KIWDcElxIEERXRKEuAJM/vD77cISSQhCUBPe/Xi5fm3Hufe87JzUM45y5vU35+vlrbsNPJmBYREW7dugVPT094eHhg5MiRqKurQ1ZWFpycnLrV05ufPHnSrW7j19yIESPg6emJH374Qe7S4a6gvr4eycnJ2Lhxo9Zn9jVVF6lUirVr12LJkiXc5GNXxRJvNzJ69GjcunULv/32G44fP45Lly7BwcEBoaGhMpM3XZlYLMauXbtgYGAAfX19xMXF4f3332/1PhddVVBQEP766y8UFBR0uQ+6np4e5s2bp+1qANBcXTIyMrB06VKZszS6KpZ4uxlra2ssWLAACxYs0HZV2oTP5yMkJAQhISHarkqHaH4aE6M97u7u2q6Cyrrs6WQMwzDPKpZ4GYZhNIwlXoZhGA3T+BhvZmamzJUpjObl5+dDJBIhLS1N21XpEkpKSlBeXs764znWmVd/KqLxxDtz5kxN75JpQePdrJgGrD8YTeERafguIwzTAaKionD48GFkZGRouyoMozY2xsswDKNhLPEyDMNoGEu8DMMwGsYSL8MwjIaxxMswDKNhLPEyDMNoGEu8DMMwGsYSL8MwjIaxxMswDKNhLPEyDMNoGEu8DMMwGsYSL8MwjIaxxMswDKNhLPEyDMNoGEu8DMMwGsYSL8MwjIaxxMswDKNhLPEyDMNoGEu8DMMwGsYSL8MwjIaxxMswDKNhLPEyDMNoGEu8DMMwGsYSL8MwjIaxxMswDKNhLPEyDMNoGEu8DMMwGsYSL8MwjIaxxMswDKNhLPEyDMNoGF/bFWAYZcRiMb799lsQEVd26tQplJaWYtOmTTLr+vn5wd7eXtNVZBi18Kjp0cwwXdTo0aORnZ0NPT09AAARgYigo9PwR5tEIoFUKkVpaSn69++vzaoyjFJsqIHpFmbPng0+n4/a2lrU1tairq4O9fX13GupVIrx48ezpMt0CyzxMt3CO++8A4lE0uJyPp+P4OBgDdaIYdqOJV6mW+jXrx88PT25oYXmiAhTp07VcK0Ypm1Y4mW6jTlz5kBXV1euXFdXF3/7299gamqqhVoxjPpY4mW6jWnTpin8xsvj8TB79mwt1Ihh2oYlXqbb6NmzJ3x8fMDny54Fqa+vj7feektLtWIY9bHEy3Qr7733nsz5vHp6epgxYwaMjIy0WCuGUQ9LvEy34u3tjR49enCvJRIJ3nvvPS3WiGHUxxIv063o6+tjxowZ3IUUvXr1gqenp5ZrxTDqYYmX6XZmzZoFiUQCPT09vPfee3JjvgzT1bFLhpluRyqVon///igvL0dGRgbc3Ny0XSWGUQv7xst0Ozo6Opg1axYGDhwIV1dXbVeHYdQm94137ty5SEpK0lZ9GIZhnimmpqZ4/PixTJnCwTEvLy+EhYVppFLMs+Xdd9/F3LlzMWXKlE7fV3l5Ofr169fp+2mryspKvP3220hISMDLL7+s7eowWnDmzBlER0fLlStMvBYWFnjjjTc6u07MM8jQ0BD29vbs+AG4bzkjR46Ei4uLlmvDaEPzb7qN2BgvwzCMhrHEyzAMo2Es8TIMw2gYS7wMwzAaxhIvwzCMhrHEyzAMo2HdMvE+efJErfUrKytRU1PTrn1KJBJUVFS0K0Z3o602q/v+PqvEYrG2q8Cgc96Hbpd4c3Jy0Lt3bxw/flzpuoWFhVi8eDEsLS2Rk5PTpv1VVlbiyy+/xIsvvoj/9//+X5tidCV5eXl47733IBAIWlxHm21W5/19VolEIkRHRyMzMxNbtmxBaGgoPvzwQ5SVlcmtm52djU8//RShoaH44osvUFRUpIUatyw1NRVz5sxBREQEtmzZgpKSEpnl6enpWLx4MYKCgvCPf/yj1S9IIpEI3333HdavX48ff/yxTfVRNcbq1asRGRkJAMjMzGzz/lpEzQQHB1NwcHDzYq25ceOGzOunT5/S+vXrqby8XOm2tbW19PPPPxMAOn36dJv2LxaLqby8nPh8Pn355ZdtiqENzfuNiGj37t00atQoAkD5+fktbtueNltbW9OuXbvUrS5Hnfe3oynqs/Z49OgRAaALFy6ovE1dXR0FBgZSQUEBVxYfH088Ho/GjBlDtbW1ctsIBAKysrJSuEybDh06RE5OTlRdXU1ERNeuXSMzMzMSCARERHT06FFydXWle/fukVgsppCQEHJzcyOxWCwX6+zZs+Tu7k6HDx9uc31UjZGSkkIAaPHixVxZWloaffjhh2rv8+DBg2RqaipX3qW/8YrFYqxevVqmzNjYGGvWrEHfvn2Vbq+vr48BAwa0qw66urro27cvjI2N2xVHkxT1GwAEBQUhIiJC6fbabLM6729HaqnPNG3p0qUYM2YMXnnlFa4sJCQEvXr1wtmzZ7F48WK5bQYNGgRbW1vo6+trsqpKxcbGws/Pj3s6iL29PaytrXH06FGIxWIsWbIEvr6+MDc3h66uLjZs2IBLly4hISFBJk52djYmTpyIb775Br6+vm2qi6oxysvL8f3338Pe3l6mfPz48eDz+UhJSWnT/pvr0MQrkUgAyI6JSKVSueWtbdvUsmXLcP36dblyqVQqt35LsVt6HLiqGuMqerptXV0dV5+WtlOkPX2iipb6DYDM0xuU7VdRmzWh+fvb2M+Ny1raplFb+k1Rn7W0r85y/vx5JCUlKXxwp7OzMwICArBz505s2bJFZpmOjk6L9yRuqS9U6dNG9fX1yqqukJmZGQ4cOMANH4jFYhQVFcHNzQ15eXkoLCzEa6+9xq3fr18/ODo6Ijk5mSurrKxEUFAQwsLC2nwnOnVifPzxx4iOjuZutN/UBx98gEWLFuHp06dtqkdTHZJ4U1NTERkZibCwMEyaNAmffvopTp48icDAQPj7+yMrKwuOjo7o2bMnZs2ahaqqKm7b7OxsLFu2DF988QXc3d2xePFiSKVSfP311zhw4ABKSkqwcOFC7NixA2VlZfjss88wePBgnDlzptXt2ysuLg4+Pj5Yt24dPvroI4hEIm7ZqVOnEBgYiNdffx1paWmwtraGn58fACAmJgYRERH46KOPMHr0aBw5coTbTtU+aSlGYmIiXF1dMXfuXADA/fv3sWLFCjg6OmLv3r0AoLDfmmrtF1FrbdaE5u9vWloaAgIC4ObmBoFAgAkTJqBHjx4yiUlZn7a1z1JSUmBqasodZ5qwdetWTJw4EWZmZnLLeDweEhMT4eLighUrVuDkyZOtxmrpGFKlT4GGZBseHo5FixbBw8MD7u7uyM/PV6s98+fPR15eHnx9ffHo0SO8//77CA0NxWuvvYYHDx4AALKysmS2sbGxkdlPREQEbt++jfnz5yMnJwfp6emora1Vqx6qxti1axc8PDxgZ2enMI6DgwOICLGxsWrtX6HmYw/qjvFKpVJ69dVXuXGZmzdv0uLFiyk/P5/s7e3JycmJwsLCKDU1lfz9/QkArVy5koiIJBIJOTg4UHx8PBER/f777wSAUlNTiYho0aJF5OTkxO2roqKCdu3aRQAoLS1N6fZERBcvXlR7jHfbtm00ceJEkkgkRESUmppKALjxztu3b9OkSZPIwsKCYmJiKCkpiTZt2kSxsbE0c+ZMLs7Ro0eJx+PR/v37iYhU6hNlMd5++20aN24ct1woFBIAio2N5cqa91tTJ0+eVDjGq6zNqmrPGG/z97e4uJg8PT2pf//+FB0dTaWlpRQXF0cA6JdffiEi1fq0LX128eJFmjRpEl2/fr1NbSFSf4zXyMioxXHE8ePHExFRUVERmZubU9++fenWrVvc8jfeeIP7f2vHkCp9SkS0dOlS2rNnDxER1dfX08iRI2no0KFUX1+vcvuJiBISEkhXV5d69+5N33zzDVd++/ZtAkDu7u4y67/11lvUp08f7vXQoUPJ0tKSwsLCaN26dWRra0v29vZUWFioch1UiXHr1i2aPXs299rR0VFmjLeRh4cHTZ8+XeV9tzTG2+7EKxQKic/nU2RkJPeh/eOPP4iIyNvbW6ZjpVIpOTo6krGxMUmlUiIiCgkJoZs3bxIRUVlZGQGgxMREIlKcQP7880/ug6lseyL1E69QKKQ+ffrQwYMHZeptbGwsk4RCQ0Np0KBB3C8ckUhEhoaGtHfvXpl4I0eOJDs7O+51a31SU1OjNEZgYKBMEiGidideVdusivZOrjV/fxcuXEgODg7c8urqatLR0aEtW7ZwZcqOs/b2WVupm3gB0Pr16xUua0y8RA2fLz09PRo+fDg9ffqUiP6XeFU5DpX16b1798jIyIji4+MpMTGREhMTadKkSQSArly5olrj/8+tW7fI19eXTExMqHfv3vT7779zy+bOnUt8Pp+io6Pp4cOHtG/fPnrhhRdoxIgRRNRwXOro6NCaNWu4bQQCAfH5fPLx8VFp/6rEEIvFNGPGDHrw4AG3TkuJd8aMGWRra6ty+zttcq1Xr14IDg5GeHg4nJ2dcfjwYYwdOxZAw5+1JiYm3Lo8Hg/+/v6orKzkTo2Jj4+HRCLB6tWrERcXB6D1MaXmY1nqbq9MdnY2KioqYGNjI1NvAwMDmfV0dXVhbGzMjYNmZWVBJBLJTea5urrixo0b3PmwrfXJ4cOHVYrR0VRtsyY0f391dXXB4/G410ZGRjA1NZUZZ1PlOOsuzM3Nla4zduxYbNu2DVeuXMGcOXNkhtZUOQ6V9WleXh4MDQ3h5eUFT09PeHp6Yvv27SgqKsJLL72kclvy8/Mxa9Ys7Ny5E+np6ejZsyd8fHxw+/ZtAEBCQgIiIyORmpqKlStXoqKiAg8ePIC/vz8A4ObNm5BKpbCysuJiWltbw8XFBb/99ptKQ4qqxPjqq69gY2ODCxcu4Pjx4zh+/DgqKytRXFyM48ePo7q6mtvW3NwchYWF7R7n7ZAx3u3bt+M///kPqqqqMHXqVKxatarFdS0sLMDj8bgPSlxcHJYvX45Vq1ZhzZo1au+7vds313ieobon8Tdul5ubK1M+YMAA6OvrczO7ijT2iVAobHOM9mhrmzWhaYJQR/PjrLtQ9QMdEhKCJUuW4ODBg/j888+5clWOQ2V9WllZiUePHsHExASDBw+W+TE0NFS5LY3zLn379oWjoyNSU1PB5/O5yTM+n49Vq1bh2LFj2LFjB/744w/06dMHCxYsAAAuyTdNfAAwZswY1NfXq3Rhgyoxzp07h/j4eAQEBHA/RUVFSE1NRUBAgNwv7474UtLuxPvw4UOkp6cjMDAQ+fn5eO+99/Dtt9+ivLwcgPy3z+zsbNja2sLY2Bjnz5/HokWLEB0djd69e3PrUJOnEVErz+JUZXt1NX7rO3v2rNyy1n7DOjs7A2g4IbwpoVCIESNGyCTNlvrE3d1daQwjIyNUVlYqrZs6fdDWNnclrR1nndFnncHY2Bj3799Xef3NmzfjzTffxJdffsmVqXMctmTIkCEAGiYdmzp48CDu3r2rcv0uXryI/v37c68dHBwQGBgo90sBaPhmmpKSgl9//RUWFhYAGh6ZY29vj8uXL8usKxaL4eTkpNLpc6rEOHDgAB4/fizzM2zYMCxYsACPHz+Gra0tt92dO3dgZWXV7lP32p14a2pquMcE6evr49NPPwURcX8yCgQCmVNajh07hrVr1wIAN5PfOLPZOMtcWVmJ3Nxc9OnTBwKBAHV1ddyVZ42npohEIqXbN67X9F9l3N3d4eDggOjoaFy7dg1AwwFUU1Mj86EQi8Uy3xCHDRsGX19fnDhxgvulAwCnT5/m2tuopT5RJcagQYNw6dIl3Lt3DxKJBN999x0AyNRNUb81akxQTa8QUrXNmtD0/QUavoU3n4EWi8VyVzi1dpy1pc+KioqwfPlybvZdE5ydnRX2t0gkkjkeGvH5fOzdu1dmiEiVY0hZn9rZ2cHLywtRUVH45JNPkJGRgfj4eJw7dw6WlpbYvHkzIiMjlX7jnDVrFg4dOiRTVlpaCm9vb5my/Px8eHt748CBAxg1apTMsq1bt2L//v0yV1qmpKRg3bp1AKBSXZTFUEdpaanMKXBt1nzQV93JteLiYjIyMiJvb2/617/+RYsXL6aEhAQiIvLx8SErKysKDg6mpKQk8vLyojVr1nATa1KplCZPnkwAaOTIkZSWlkZDhw6lgQMH0oULFygzM5OMjIxoyJAhdP78eRIIBBQcHEwAaNq0aZSfn9/q9sXFxTRnzhxu/cuXL6vUpqtXr5KdnR0ZGRnR66+/TsuXL6fevXvTiBEj6Oeff6aDBw+SjY0NAaB169ZRRUUFETVcdeXn50dOTk60adMm8vPzo+3bt8vEVtYnymIUFxeTubk5GRkZ0ahRo+jKlSsEgMaNG0cnTpwgIpLrt0apqank6+tLAGj27NmUkZGhcptV1Z7Jtebvb3R0NFlZWZGenh7FxMRQWVkZbdiwgQCQs7MznTp1SqU+bUufJSYmys30q0vdybWoqCi5Cb7c3FyaM2cO6ejoUEREBDeR3NTly5dlJptaO4aOHDmiUp+WlpaSh4cHAeA+X3fu3CGJREK9evUiAJSSktJqe4RCIQUGBtLUqVNp9+7dFBoaSmvXriWihs9+RkYGbdy4kaZNm0bp6ektxomLi6N3332XduzYQQEBAZSUlEREpFZdWorREkWTayKRiExNTdU6JjrtrAaxWEwikYgqKiro4sWL3JkNRA0fiClTppBQKKTc3FwSCoUKY9y7d4/7kFRVVVFdXR23rLq6WuElhKpu31YSiYSuXr3KzXTm5eWpvG1FRQVdu3ZN4TJV+6S1GLW1tXT16lWurxXVTZV+a649bW7U3rMa2kKVPm1LnzU9Xast1E28dXV15OzsTMXFxWrv6+HDh3JlrR1DqioqKqKCggLu80VEVFJSQjt37qSffvpJpRhCoZDy8vKopqaGK6uvr6esrCyZMwmU+euvv2TyS1vqoiiGqvbs2UPDhw+X6QtlOi3xtqbxA8H8z7PeJ9pMvF1NW+7VkJOTQ++//34n1qr96urqKDQ0lEQikbarotG6vP32263e40QRrdyr4cmTJ11yplybWJ90vGepT0eMGAFPT0/88MMP2q6KQvX19UhOTsbGjRu1crqhNuoilUoRHh6OJUuWcBOP7aX4Au92EovF2LVrFwwMDKCvr4+4uDi8//77Wrv2v1FpaSl3qkpLzMzMsGfPng7fd1ftk+7sWe3ToKAg/PXXXygoKOiwD3pH0dPTw7x587RdDQCaq0tGRgaWLl0qc4ZGe/GIZM+habymPTExscN2wjw/XnzxRXz22WfccfQ8e/z4MczMzHDhwgW4uLhouzqMFhw6dAhz587F48ePZcq79G0hGYZhnkUs8TIMw2gYS7wMwzAapnBy7ebNm2yMl2mTqqoqjd7DtitrvD/AkSNH8Oeff2q5Now25ObmKrzsXuHk2v79+zX+6BXm2XD//n306NGjWz0qqbNIpVIUFxfD0tKyyz2Wh9GM6upqVFdXy90rROE33unTp7NvvEybsLMa/qfxrIaUlBR2VsNzqvGshubYGC/DMIyGscTLMAyjYSzxMgzDaBhLvAzDMBrGEi/DMIyGscTLMAyjYSzxMgzDaFi3Trzq3oO1srJS7lldDNNVqfIUXabzdcb70CH34z1z5gy+/fZbiEQiWFpawsPDAwKBAGvWrIGenl5H7EJOTk4OXn/9dRw7dgyTJk1qdd3CwkJ89dVXSE5ORkpKCsaNG4czZ87g0KFDMo99NjQ0hJmZGQYNGgRPT09YW1t3St01LS8vDxs2bEBkZCRefPFFAHiu2t/diEQixMTEwM3NDbm5ucjPzwePx0NERAT3BN5G2dnZ2LdvHyorK2FhYYHg4OAu9b6lpqYiOTkZ1tbWeOGFFzBt2jQMHDiQW56eno7k5GQIhULY2NggPDy8xSchi0QifP/996iqqsLLL7+MwMBAteujaozVq1ejZ8+eCA8PR2ZmJu7cudOm/bWo+SMp1H30zy+//EL9+vWjnJwcIiK6e/cuTZ8+nQBQbW2tWo/JaM2NGzdkXj99+pTWr19P5eXlSretra2ln3/+mQDQ6dOnufKrV68SAJoyZQpdvHiRiouL6dixYzR//nwCQP/4xz86rP6a0LyPiIh2795No0aNIgByjy3pjPZr49E/itrdFeK25dE/dXV1FBgYSAUFBVxZfHw88Xg8GjNmjMLPlEAgICsrqw79vHWEQ4cOkZOTE1VXVxMR0bVr18jMzIwEAgERER09epRcXV3p3r17JBaLKSQkhNzc3BQ+K/Ds2bPk7u5Ohw8fbnN9VI2RkpJCAGQedpmWlkYffvih2vvstGeujR07lv7+97/LlInFYnJxcemwA6G+vp5mzJjRrhgXLlyQS7xVVVUEQOEzrn788UcCQPPnz1f7oZHa0FofHTlyRGHi7Yz2azrxdsSx0Vlx25J4Fy5cSP/85z/lyk1NTbn3ozmJREJjx45tV107g5eXF0VERMiUubi40LZt26i+vp5sbGxo/fr13LIHDx5Qjx49KDY2VmabrKws6tGjB2VmZra5LqrGePDgAfn6+pK9vb3cU4aXLVtGR48eVWu/nfbMNaFQiLS0NFRVVXFlurq6mDZtmsx6Te/QI5FIFMZqqXzZsmW4fv26XLlUKpXbpqUYOjryTVVU1uidd97Bu+++i507d2L37t1ceV1dHbdvVevffP3W1mttWWta6iMA6NGjh8LytrS/s3R0u1Xt7/bE7Wjnz59HUlISZs+eLbfM2dkZAQEB2LlzJ7Zs2SKzTEdHB3y+4lHDltreeBwDyttUX1+vrOoKmZmZ4cCBA9y8ilgsRlFREdzc3JCXl4fCwkK89tpr3Pr9+vWDo6MjkpOTubLKykoEBQUhLCwMrq6ubaqfZNpRAAAgAElEQVSHOjE+/vhjREdHKxwi/eCDD7Bo0SI8ffq0TfVoqt2Jd+rUqbh06RJGjx6NzMxMrnz58uXQ09PDyZMnERgYCH9/f2RlZcHR0RE9e/bErFmzuGSdnZ2NZcuW4YsvvoC7uzsWL17MHQxff/01Dhw4gJKSEixcuBA7duxAWVkZPvvsMwwePJi7BWFrMdoqNDQUAJCUlIRTp04hMDAQr7/+OtLS0mBtbQ0/Pz8AQExMDCIiIvDRRx9h9OjROHLkCBdDlfYri5GYmAhXV1fuZhv379/HihUr4OjoiL179yrso6ZaS7Cqtr+zdHS7lfW3spiA4mMOAFJSUmBqatppt73cunUrJk6cCDMzM7llPB4PiYmJcHFxwYoVK3Dy5MlWY7XUr2lpaQgICICbmxsEAgEmTJiAHj16yCX7+vp6hIeHY9GiRfDw8IC7uzvy8/PVas/8+fORl5cHX19fPHr0CO+//z5CQ0Px2muv4cGDBwCArKwsmW1sbGxk9hMREYHbt29j/vz5yMnJQXp6Ompra9Wqh6oxdu3aBQ8PD9jZ2SmM4+DgACJCbGysWvtXqPlXYHWHGsRiMb3zzjsEgHg8HgUHB1NJSQm3PD8/n+zt7cnJyYnCwsIoNTWV/P39CQCtXLmSJBIJOTg4UHx8PBER/f777wSAUlNTuRiLFi0iJycn7nVFRQXt2rWLAFBaWppKMS5evCg31FBTU9Pin9pERPfu3SMANGDAALp9+zZNmjSJLCwsKCYmhpKSkmjTpk0UGxtLM2fO5LY5evQo8Xg82r9/v0rtJyKlMYgaHi09btw47rVQKCQA3J9lzfuoqZMnTyocalCn/apSZ6ihM9qtSn8ri6koLlHDMTRp0iS6fv26Su1Td6jByMioxXHE8ePHExFRUVERmZubU9++fenWrVvc8jfeeIP7f2v9WlxcTJ6entS/f3+Kjo6m0tJSiouLIwD0yy+/cNssXbqU9uzZQ0QNwy4jR46koUOHUn19vUptaZSQkEC6urrUu3dv+uabb7jy27dvEwByd3eXWf+tt96iPn36cK+HDh1KlpaWFBYWRuvWrSNbW1uyt7enwsJCleugSoxbt27R7NmzudeOjo5yQw1ERB4eHjR9+nSV991pY7yNdu/eTVZWVgSAjI2NZT583t7eMh0slUrJ0dGRjI2NSSqVUkhICN28eZOIiMrKyggAJSYmcusr+hD8+eefXOIlIqUx2pJ4RSIRAaD+/fsTEVFoaCgNGjSIG/MUiURkaGhIe/fuldlu5MiRZGdnp1L7a2pqVIoRGBgokyyIqNMTb/P2q0KdxNtZ7VZ2vCmL2VJcdambeAHIjHk21Zh4iYj++OMP0tPTo+HDh9PTp0+J6H+JV5VjcuHCheTg4MAtq66uJh0dHdqyZQsRNfzCNTIyovj4eEpMTKTExESaNGkSAaArV66o1vj/c+vWLfL19SUTExPq3bs3/f7779yyuXPnEp/Pp+joaHr48CHt27ePXnjhBRoxYgQRNfxC1NHRoTVr1nDbCAQC4vP55OPjo9L+VYkhFotpxowZ9ODBA26dlhLvjBkzyNbWVuX2d9oYb6OgoCAUFBRg9erVqK2txYIFC5Camgqg4U9dExMTbl0ejwd/f39UVlairKwM8fHxkEgkWL16NeLi4gAoH1dqPqbVlhjK3Lx5EwAwdOhQAA1j18bGxtzjw7OysiASiTBgwACZ7VxdXXHjxg1UVFQAaL39hw8fVimGNjRvf0frrHYrO966MnNzc6XrjB07Ftu2bcOVK1cwZ84cmSE1VY5JXV1d8Hg8bpmRkRFMTU25scu8vDwYGhrCy8sLnp6e8PT0xPbt21FUVISXXnpJ5bbk5+dj1qxZ2LlzJ9LT09GzZ0/4+Pjg9u3bAICEhARERkYiNTUVK1euREVFBR48eAB/f38ADcefVCqFlZUVF9Pa2houLi747bffVBpKVCXGV199BRsbG1y4cAHHjx/H8ePHUVlZieLiYhw/flzmlEtzc3MUFha2e5y33Yn3v//9L/d/Y2NjbNiwAYcPHwYAubHGpiwsLMDj8WBiYoK4uDgsX74cq1atwpo1a9pUj46I0dzp06cBAJ6engqXl5SUAGh4vEdTAwYMgL6+fovnIwL/a79QKGxzjM6mrP0dQVPtbnq8dWWqfqBDQkKwZMkSHDx4EJ9//jlXrsox2TTpKlJZWYlHjx7BxMQEgwcPlvkxNDRUuS2N8y19+/aFo6MjUlNTwefzuckzPp+PVatW4dixY9ixYwf++OMP9OnTBwsWLAAALsk3TXwAMGbMGNTX16t0YYMqMc6dO4f4+HgEBARwP0VFRUhNTUVAQIDcL2sejwcDAwOV+0GRdifehIQEkOzTg/DWW2/B0dFR5iqx5t8+s7OzYWtri4KCAixatAjR0dHo3bs3t7x5zOavmzp//rxKMdTx+PFjrFu3DtbW1vjkk08UruPs7Ayg4STwpoRCIUaMGCGTPFpqv7u7u0oxjIyM5B4fAsjOSLenvc2p0v6O0Fntbqm/jY2NVYrZUtzOZGxsjPv376u8/ubNm/Hmm2/iyy+/5MrUOSZbMmTIEAANk4xNHTx4EHfv3lW5fhcvXkT//v251w4ODggMDJT7pQA0fDNNSUnBr7/+yl0kYmpqCnt7e1y+fFlmXbFYDCcnJ5Uep6RKjAMHDuDx48cyP8OGDcOCBQvw+PFj2NractvduXMHVlZW7X6UU7sT7927d7Fu3TqZstu3b+Pq1auYOXMmVyYQCGRObTl27BjWrl3Lzew3zm42zixXVlZyb1CfPn0gEAhQV1eHnJwcAOCSukgkUimGSCSS+RcAN7PaXF5eHjw9PWFhYYFDhw5xp2OJxWKZy5SHDRsGX19fnDhxAuXl5Vz56dOnsXbtWpmYLbVf1RiDBg3CpUuXcO/ePUgkEnz33XcAwH1QFfVRo8Yk1PxyaXXb39E6s90t9bcqMVuKW1RUhOXLl7fYb+3l7OysMPGKRCKZPmrE5/Oxd+9e2NjYcGWqHE9PnjyRm9UXi8Xc8WFnZwcvLy9ERUXhk08+QUZGBuLj43Hu3DlYWlpi8+bNiIyMVPqNc9asWTh06JBMWWlpKby9vWXK8vPz4e3tjQMHDmDUqFEyy7Zu3Yr9+/dDIBBwZSkpKVzOUaUuymKoo7S0VOYUuDZrPuir7uSan58fTZ48mSZMmEDh4eH04Ycfko2NDX311VfcOj4+PmRlZUXBwcGUlJREXl5etGbNGpJKpSSVSmny5MkEgEaOHElpaWk0dOhQGjhwIDcpkZmZSUZGRjRkyBA6f/48CQQCCg4OJgA0bdo0ys/PbzVGcXExzZkzh1v/8uXL9Mcff5Cvry8BoL59+5KXlxfNmzePgoKCaMaMGRQdHU11dXUyg+Q2NjYEgNatW0cVFRVE1HAFnZ+fHzk5OdGmTZvIz8+Ptm/fLtNHrbVf1RjFxcVkbm5ORkZGNGrUKLpy5QoBoHHjxtGJEyfk+qhRamoq187Zs2dTRkYGEZHa7VeVOpNrndVuZf2tLCaR/DFHRJSYmCg3+98adSfXoqKi5Cb0cnNzac6cOaSjo0MRERHcBHJTly9flplsaq1fjxw5QlZWVqSnp0cxMTFUVlZGGzZsIADk7OxMp06dIiKi0tJS8vDwIADc5+rOnTskkUioV69eBIBSUlJabY9QKKTAwECaOnUq7d69m0JDQ2nt2rVE1DDhmZGRQRs3bqRp06ZRenp6i3Hi4uLo3XffpR07dlBAQAAlJSUREalVl5ZitETR5JpIJCJTU1OV33+iTjyr4e7du1ylsrOz6dy5c1RZWSmzjo+PD02ZMoWEQiHl5uaSUCiUi3Pv3j3ug1FVVSX3oa+urlZ6BZWyGJ2poqKCrl27pnCZKu1XFoOo4dLnq1evkkQiISKivLw8meWq9FFna8uVax3dblX6W1lMRXGJSOYULmXUTbx1dXXk7OxMxcXFKu+j0cOHD+XKlPWrKoqKiqigoID7XBERlZSU0M6dO+mnn35SKYZQKKS8vDyqqanhyurr6ykrK0vmTAJl/vrrL+79amtdFMVQ1Z49e2j48OEyfaFMp59O1prGD8Lz6nlqvzbu1dBcV+nvtlwynJOT0+LpfV1FXV0dhYaGkkgk0nZVNFqXt99+W+6UTGU6/XSy1jx58kTtWzg+S5739mtad+7vESNGwNPTEz/88IO2q6JQfX09kpOTsXHjxnbP7HeXukilUoSHh2PJkiXcxGN7dWriFYvFSEhIgIGBAXr06IG4uLg2X5PfHT3v7de0Z6W/g4KC4O7ujoKCAm1XRY6enh7mzZuHXr16absqGqtLRkYGli5dismTJ3dYzA65H2+Lwfl8hISEICQkpDN302U97+3XtGepv5uewsRoV+Mpnx2pWz+BgmEYpjtiiZdhGEbDWOJlGIbRMB6R7HWRc+fOxa+//gp7e3tt1Ynpxv78808MGDAAL7zwQqfuRyQSob6+vkvfe0EikSAnJweOjo6ddvUf07WVl5ejpKQEjx8/limXm1ybOnUq90BEhlHXG2+8oZH9nDlzBgUFBXj77bc1sr+2mjBhgrarwGiZohsLyX3jZZjuICoqCocPH0ZGRoa2q8IwamNjvAzDMBrGEi/DMIyGscTLMAyjYSzxMgzDaBhLvAzDMBrGEi/DMIyGscTLMAyjYSzxMgzDaBhLvAzDMBrGEi/DMIyGscTLMAyjYSzxMgzDaBhLvAzDMBrGEi/DMIyGscTLMAyjYSzxMgzDaBhLvAzDMBrGEi/DMIyGscTLMAyjYSzxMgzDaBhLvAzDMBrGEi/DMIyGscTLMAyjYSzxMgzDaBhLvAzDMBrGEi/DMIyGscTLMAyjYSzxMgzDaBhLvAzDMBrGEi/DMIyG8YiItF0JhmmNRCKBvb097t69y5WJxWJIJBIYGBhwZbq6uvj1118xevRobVSTYVTGvvEyXZ6uri4mTpyI2tpaVFVVoaqqCrW1tRCLxdzrqqoqGBgYwNXVVdvVZRilWOJluoWgoCBIJJIWlxsYGCA4OBg6OuyQZro+NtTAdAtEhAEDBsgMNzR3/vx5vPbaaxqsFcO0Dft6wHQLPB4Ps2fPlhnTbWrQoEEs6TLdBku8TLcxc+ZM1NbWypUbGBhgwYIFWqgRw7QNG2pguhU7OzvcvHlTrrygoACvvPKKFmrEMOpj33iZbmXOnDlyww3Dhw9nSZfpVljiZbqVoKAg1NXVca8NDAwwb948LdaIYdTHhhqYbsfFxQUXL14E0DDpVlxcjAEDBmi5VgyjOvaNl+l25s6dC319ffB4PIwePZolXabbYYmX6XYCAwMhFovB4/EQHBys7eowjNrYUAPTLXl4eCAjIwNlZWXo27evtqvDMGrha7sC6khOTlZ4KhHz/DEzM4OtrS22bt2q7aowXUR4eDj4/O6R0rrVN94pU6YgLy8Ptra22q6KVl29ehVmZmawsrLSdlW0RiKR4NGjR+jXrx+ys7PxyiuvoHfv3tquFqMFT548wYULF1BTUwNDQ0NtV0cl3ePXQxNBQUGIiorSdjW0ys3NDVOnTkVYWJi2q9IlGBoaIioqClOmTNF2VRgtyMzM7Ha3AmWTawzDMBrGEi/DMIyGscTLMAyjYSzxMgzDaBhLvAzDMBrGEi/DMIyGscTLMAyjYd3uPN6OUlhYiK+++grJyclISUnBuHHjtF0ljcnJycHrr7+OY8eOYdKkSdqujlaIRCLExMTAzc0Nubm5yM/PB4/HQ0REBCwsLGTWzc7Oxr59+1BZWQkLCwsEBwfD2tpaSzWXl5qaiuTkZFhbW+OFF17AtGnTMHDgQG55eno6kpOTIRQKYWNjg/DwcBgZGSmMJRKJ8P3336Oqqgovv/wyAgMD21QnVeKsXr0aPXv2RHh4OM6cOYM7d+60eX/dDnUjXl5etGrVqg6JVVtbSz///DMBoNOnT6u0zY0bNzpk3+3l6upKGzdubPP2T58+pfXr11N5eXkH1ko1ndGHBgYG9Msvv6i8fl1dHQUGBlJBQQFXFh8fTzwej8aMGUO1tbVy2wgEArKyslK4TJsOHTpETk5OVF1dTURE165dIzMzMxIIBEREdPToUXJ1daV79+6RWCymkJAQcnNzI7FYLBfr7Nmz5O7uTocPH25XnVSJk5KSQgBo8eLFXFlaWhp9+OGHau8vIyODAFBNTU2b6qsNz+1Qg76+vlq3ExSLxVi9enUn1khzjI2NsWbNGo3fXKar9OHSpUsxZswYmadWhISEoFevXjh79iwWL14st82gQYNga2sLfX19TVZVqdjYWPj5+XHfYO3t7WFtbY2jR49CLBZjyZIl8PX1hbm5OXR1dbFhwwZcunQJCQkJMnGys7MxceJEfPPNN/D19W1zfVSJU15eju+//x729vYy5ePHjwefz0dKSkqb999dPLeJFwB0dBQ3XyKRAGhIFI2WLVuG69evy6zX9EkIjdsoitMVSaVSmfo1bYtUKm1xm0ZtaZuiPmxpX53l/PnzSEpKwuzZs+WWOTs7IyAgADt37sSWLVtkluno6LR4A5bW+kKVfgWA+vp6ZVVXyMzMDAcOHEBNTQ2AhmO2qKgIbm5uyMvLQ2FhoczTl/v16wdHR0ckJydzZZWVlQgKCkJYWBhcXV3bVA914nz88ceIjo6Gnp6e3LIPPvgAixYtwtOnT9tcj+7guU68zaWmpiIyMhJhYWGYNGkSPv30UwDA119/jQMHDqCkpAQLFy7E/PnzERQUBCcnJxQVFcHLywuGhoaYOHEiqqurceLECQwZMgQmJib4xz/+oeVWySorK8Nnn32GwYMH48yZM0hLS0NAQADc3NwgEAgwYcIE9OjRQyYxnTx5EoGBgfD390dWVhYcHR3Rs2dPzJo1C1VVVUhMTISrqyvmzp0LALh//z5WrFgBR0dH7N27F4B8H+7YsQMpKSkwNTXFmTNnNNb+rVu3YuLEiTAzM5NbxuPxkJiYCBcXF6xYsQInT55sNVZMTAwiIiLw0UcfYfTo0Thy5Ai3TJV+ra+vR3h4OBYtWgQPDw+4u7sjPz9frfbMnz8feXl58PX1xaNHj/D+++8jNDQUr732Gh48eAAAyMrKktnGxsZGZj8RERG4ffs25s+fj5ycHKSnpyt8mrMyqsTZtWsXPDw8YGdnpzCGg4MDiAixsbFq779b0fZYhzo6coyXiOjixYvcGK9UKqVXX32VG/u6efOmzPjTokWLyMnJiYiISkpKyNvbm/r06UP//Oc/6fHjx3Ts2DECQNOnT6cffviBhEIhRUREEAC6evVqh9WZqH1jvBUVFbRr1y4CQGlpaVRcXEyenp7Uv39/io6OptLSUoqLiyMA3Lhpfn4+2dvbk5OTE4WFhVFqair5+/sTAFq5ciUREb399ts0btw4bj9CoZAAUGxsLFfWtA+JGvp/0qRJdP369Ta1pZE6Y7xGRkYtjiOOHz+eiIiKiorI3Nyc+vbtS7du3eKWv/HGG9z/Y2NjaebMmdzro0ePEo/Ho/379xMRqdSvS5cupT179hARUX19PY0cOZKGDh1K9fX1qjeeiBISEkhXV5d69+5N33zzDVd++/ZtAkDu7u4y67/11lvUp08f7vXQoUPJ0tKSwsLCaN26dWRra0v29vZUWFioVj2Uxbl16xbNnj2bW9/R0VHmM9bIw8ODpk+frvJ+u+MYL0u8/5d4hUIh8fl8ioyMJIlEQkREf/zxB7du86Tx8ccf04svvigTz9bWlubNm8e9LiwsJAB06NChDqszUfsn1/78808u8RIRLVy4kBwcHLjl1dXVpKOjQ1u2bOHKvL29ZT7AUqmUHB0dydjYmKRSKQUGBsokXiJSmng7ijqJFwCtX79e4bLGxEvU8N7r6enR8OHD6enTp0T0v8QrEonI0NCQ9u7dK7P9yJEjyc7OjnvdWr/eu3ePjIyMKD4+nhITEykxMZEmTZpEAOjKlSsqtaXRrVu3yNfXl0xMTKh37970+++/c8vmzp1LfD6foqOj6eHDh7Rv3z564YUXaMSIEUTU8AtSR0eH1qxZw20jEAiIz+eTj4+PynVQFkcsFtOMGTPowYMH3PKWEu+MGTPI1tZW5X13x8TLhhr+T69evRAcHIzw8HA4Ozvj8OHDGDt2bIvr6+rqypX169cP1OT2xo1/zlZWVnZ8hduh+Vilrq4ueDwe99rIyAimpqYy42w6OjowMTHhXvN4PPj7+6OyshJlZWWdX+kOZG5urnSdsWPHYtu2bbhy5QrmzJkjMz6blZUFkUgkNznr6uqKGzduoKKiAkDr/ZqXlwdDQ0N4eXnB09MTnp6e2L59O4qKivDSSy+p3Jb8/HzMmjULO3fuRHp6Onr27AkfHx/cvn0bAJCQkIDIyEikpqZi5cqVqKiowIMHD+Dv7w8AuHnzJqRSqcy9na2treHi4oLffvtN5TF4ZXGioqJgY2ODCxcu4Pjx4zh+/DgqKytRXFyM48ePo7q6mtvO3NwchYWFz/Q4L0u8TWzfvh3/+c9/UFVVhalTp2LVqlUtrsvj8WQ+VN1ZW9thYWEBHo8nk5C7A1U/0CEhIViyZAkOHjyIzz//nCsvKSkBAOTm5sqsP2DAAOjr63NnGLTWr5WVlXj06BFMTEwwePBgmR91bub9xRdfwN3dHX379oWjoyNSU1PB5/O5yTM+n49Vq1bh2LFj2LFjB/744w/06dMHCxYsAAAuyTdNfAAwZswY1NfXy0wwt0ZZnIyMDMTHxyMgIID7KSoqQmpqKgICAuR+efN4PBgYGKjcD90NS7z/5+HDh0hPT0dgYCDy8/Px3nvv4dtvv0V5eTm3DnWfh3V0iuYz79nZ2bC1tYWxsTGMjIwUfrNv/o1J231obGyM+/fvq7z+5s2b8eabb+LLL7/kypydnQE0XJjQlFAoxIgRI1q8OKGpIUOGAGiYdGzq4MGDuHv3rsr1u3jxIvr378+9dnBwQGBgoNwvBaDhW2lKSgp+/fVX7iIRU1NT2Nvb4/LlyzLrisViODk5qXz6nLI4P//8Mx4/fizzM2zYMCxYsACPHz+WearMnTt3YGVl1eVO3etIz3XiFYlE3L81NTXcEx309fXx6aefgoi4P8v79OkDgUCAuro65OTkoKamhjuFp5FYLJaZxW38ttC4n66isd6N9Xry5Inc7LNYLJZrn0AgkDl16tixY1i7di2AhvNcL126hHv37kEikeC7774DAJkk17wPi4qKsHz5cm72XROcnZ0VJl6RSCTzS7YRn8/H3r17YWNjw5UNGzYMvr6+OHHihMw2p0+f5voDaL1f7ezs4OXlhaioKHzyySfcN8Jz587B0tISmzdvRmRkpNJvnLNmzcKhQ4dkykpLS+Ht7S1Tlp+fD29vbxw4cACjRo2SWbZ161bs378fAoGAK0tJScG6desAQOW6KIujqtLSUplT4J5J2h1iVk9HTq4VFxfTnDlzCABNmzaNfvnlFzIyMiJvb2/617/+RYsXL6aEhARu/czMTDIyMqIhQ4bQV199RXZ2dqSjo0PfffcdPXr0iGJiYsjQ0JAGDhxIP/74I925c4eWLVtGAMjT05MuXLjQIfUmat/kmkAgoODgYK7d0dHRZGVlRXp6ehQTE0NlZWW0YcMGAkDOzs506tQpIiLy8fEhKysrCg4OpqSkJPLy8qI1a9aQVColoob+NDc3JyMjIxo1ahRduXKFANC4cePoxIkTRCTbh+fPn6fExESZWf62UmdyLSoqSm6CLzc3l+bMmUM6OjoUERFBN2/elNvu8uXLMpNNT58+JT8/P3JycqJNmzaRn58fbd++nVt+5MgRpf1aWlpKHh4eBIAA0MiRI+nOnTskkUioV69eBIBSUlJabY9QKKTAwECaOnUq7d69m0JDQ2nt2rVE1DABmpGRQRs3bqRp06ZRenp6i3Hi4uLo3XffpR07dlBAQAAlJSUREalVl9biKKJock0kEpGpqalax0R3nFx7bhNvc2KxmEQiEVVUVNDFixe5Mxuaqq6uVnippaa196yGtvDx8aEpU6aQUCik3NxcEgqFcuvU1tbS1atXub7Ly8uTW6d5HzY9Xaut1Em8dXV15OzsTMXFxWrv5+HDh3JlFRUVdO3aNbVjNVVUVEQFBQXcLzGihlMWd+7cST/99JNKMYRCIeXl5ckkn/r6esrKypI5k0CZv/76S+7YV7cuLcVRxZ49e2j48OEyfaFMd0y8z+1NcprT1dWFrq4uDAwMFJ5cD0ClsbtnXa9evfDqq68qXKavrw8HBwfu9dChQ+XWad6Hmn5itJ6eHnbs2IHPP/9c7rJZZfr06SNXZmZm1uLxoqrBgwfLlZmbmyMnJwfffvutSjF69eqFXr16yZTx+Xy8/vrratWl6ZBKW+vSUhxV/Pvf/8a+ffuemYnrljzXY7yM6p48eYInT55ouxodYsSIEfD09MQPP/yg7aooVF9fj+TkZGzcuFHrM/uaqotUKkV4eDiWLFnCTTw+y1jiZVolFouRkJAAAwMD9OjRA3FxcV36HhSqCgoKgru7OwoKCrRdFTl6enqYN2+e3DfYZ7kuGRkZWLp0KSZPntyp++kq2FAD0yo+n4+QkBCEhIRouyodTtPDHEzL3N3dtV0FjWLfeBmGYTSMJV6GYRgNY4mXYRhGw1jiZRiG0TAeUfe5AcGbb76J8+fPt/u8ye7u4cOH4PP5MDU11XZVuoTi4mL069ePnWf9nKqtrUVZWRlqamrUusGQNnWrsxp0dXUxduxYBAQEaLsqWrV+/Xq4uLjgb3/7m7ar0iWEhITgvffeg6Ojo7arwmjBrVu3EBkZqe1qqKVbfeOdMmUKXFxcEBUVpe2qaJWbmxumTp3K3dTneWdoaIhDhw5hypQp2q4KowWZmZkYPXp0t/rGy8Z4GYZhNIwlXoZhGA1jiZdhGEbDWOJlGIbRMJZ4GYZhNIwlXoZhGA1jiZdh2rfsmGsAACAASURBVEnVJ/E+q5739rdFt7qAQh3Z2dnYt28f9+TbgQMHYuLEiS3ekf/MmTP49ttvIRKJYGlpCQ8PDwgEAqxZswYXL17Ezp07le7z888/x6VLl3Ds2DFUVlZi+PDhCA0NbXH9wsJC7imzdnZ2mD9/PrsarRsRiUSIiYmBm5sbcnNzkZ+fDx6Ph4iICO4pvo2aHo8WFhYIDg6GtbW1lmquWF5eHjZs2IDIyEi8+OKLCtcRiUT4/vvvUVVVhZdffhmBgYHIzMzEnTt3EBgYqNkKd2faffKQetR95lpxcTEBoOHDh1N9fX2L6/3yyy/Ur18/ysnJISKiu3fv0vTp0wkA1dbW0vbt2ykgIIDOnTtHN27coE8//ZQA0NGjR+nGjRv066+/0rBhw+jSpUtERHTy5EkCQPr6+nTnzp0W9xsaGkoAyMvLS61nTGnjmWs3btzosnHVeeZaR6mrq6PAwEAqKCjgyuLj44nH49GYMWOotrZWbhuBQEBWVlYKl2nb7t27adSoUQSA8vPzFa5z9uxZcnd3p8OHD8stS0tLow8//LCzq6lQd3zm2jM91DBw4EDo6OjAzs6Oe0y7IuvXr0dQUBBGjBgBALCwsMCPP/4IFxcXAA2/5ePj4zFy5Ei8/PLL6Nu3L4CGR5q//PLL8PLyQlxcHPftevjw4XjxxRdRV1fHPea8ufLycpw6dQoAYG9v36WfMSUWi7F69epuE1cTli5dijFjxuCVV17hykJCQtCrVy+cPXsWixcvlttm0KBBsLW1hb6+viarqpKgoCBERES0uDw7OxsTJ07EN998A19fX7nl48ePB5/PR0pKSmdW85nxTCdeANDR0VF6oAuFQqSlpaGqqoor09XVxbRp0wAAc+fORe/evVuNMXr0aO4hkHw+H6+++io8PT0RFxcHoVAot/62bduwcOFCbl+a0NZH9ixbtgzXr1+XK5dKpe2KrUrcruj8+fNISkrC7Nmz5ZY5OzsjICAAO3fuxJYtW2SW6ejotPgFoKX+q6ur4/6vrF/q6+uVVb1VPXr0UFheWVmJoKAghIWFwdXVtcXtP/jgAyxatAhPnz5tVz2eB8984lXF1KlTcenSJYwePRqZmZlc+fLly6Gnp4eePXsqjaGjoyN3nfinn36KJ0+e4Pvvv5cpr6mpwS+//IIZM2Z0TAOUiImJQUREBD766COMHj0aR44c4ZYlJibC1dUVc+fOBQDcv38fK1asgKOjI/bu3Yuvv/4aBw4cQElJCRYuXIgdO3bg5MmTCAwMhL+/P7KysuDo6IiePXti1qxZqKqqUhoTgMK4AJCSkgJTU1OcOXNGI33TFlu3bsXEiRMV3iWPx+MhMTERLi4uWLFiBU6ePNlqrJbem7S0NAQEBMDNzQ0CgQATJkxAjx495JJ9fX09wsPDsWjRInh4eMDd3R35+fltapeOjuJ0EBERgdu3b2P+/PnIyclBeno6amtr5dZzcHAAESE2NrZN+3+uaHusQx3qjvESEfH5fAoMDGx1HbFYTO+88w4BIB6PR8HBwVRSUtLi+ps2bSIA3JhucxUVFeTv709ERI6OjmRpaUkikYhbHhsbS1u3bqV79+4RAPr444/VapM6Y7yxsbE0c+ZM7vXRo0eJx+PR/v37ubK3336bxo0bx70WCoUEgGJjY4mIaNGiReTk5MQtz8/PJ3t7e3JycqKwsDBKTU0lf39/AkArV65UKaaiuEREFy9epEmTJtH169dVah+R5sd4jYyMWhzPHD9+PBERFRUVkbm5OfXt25du3brFLX/jjTe4/7f23hQXF5Onpyf179+foqOjqbS0lOLi4giATFuXLl1Ke/bsISKi+vp6GjlyJA0dOrTVOY2WNM5NNB/jHTp0KFlaWlJYWBitW7eObG1tyd7engoLC+VieHh40PTp09Xed3uwMd5uSldXFz/++CN2794NS0tLJCUlwd7eHomJie2OvXLlSty9e5d7lLhUKkVSUhLmz5/f7tiqWL58Ofz9/bnX3t7eGDFihMydzZr/iansibJDhgyBra0tTExMsHHjRkyaNAn79++Ho6MjYmNjQURqx2zk5OSE1NRU2NnZqbS+NtTU1MDKyqrVdQYPHoz9+/fjyZMn8PPz48b/G9XW1rb63gwcOBCvvPIK+vbtixUrVsDS0hKzZ8+Gjo4Obty4AaDhL4nt27ejsrISSUlJ2L17N8zMzHDt2rUOe3rykydPUFBQgHnz5mHjxo347LPPcPLkSdy8eRNLly6VW9/c3BwXLlzokH0/y1jibSIoKAgFBQVYvXo1amtrsWDBAqSmprYr5rvvvotBgwbh66+/hlQqxaFDhzB58uQWx9M6mkgkwoABA2TKXF1dcePGDVRUVLQ5ro6ODkxMTLjXPB4P/v7+qKysRFlZWZvjdhfm5uZK1xk7diy2bduGK1euYM6cOTJjtFlZWUrfG11dXZlJVyMjI5iamnJjqHl5eTA0NISXlxc8PT3h6emJ7du3o6ioCC+99FKHtPPmzZuQSqUyv2isra3h4uKC3377TW7c2dzcHIWFhWycV4nnOvESEZ4+fYr//ve/XJmxsTE2bNiAw4cPAwA39thWenp6+Oijj3D9+nUcOnQI27Ztw5IlS9oVU125ubkyrwcMGAB9ff0Of2KDhYUFeDyeTEJ+VqmaWEJCQrBkyRIcPHgQn3/+OVdeUlICoPX3RtmZLpWVlXj06BFMTEwwePBgmZ+Oui9tYwKvrq6WKR8zZgzq6+sVXjzB4/FgYGDQIft/Vj3XiXf37t14/PgxEhISQM3uB//WW2/B0dERNTU17d5PSEgIevfujb///e8YMmQIXnjhhXbHVEd6errMa6FQiBEjRnCJ18jISO5PYUB2Fr15/wDys+jZ2dmwtbWFsbGxSjFbitvVGRsb4/79+yqvv3nzZrz55pv48ssvuTJnZ2cAyt+b1gwZMgQAuItwGh08eBB3795VuX6tMTU1hb29PS5fvixTLhaL4eTkJHfG0J07d2BlZdUlT5nrSp7pxPv48WOIxWI8ePBAbllubi727t2LQYMG4e7du1i3bp3M8tu3b+Pq1auYOXOm3Lbl5eUA0GJSLi8vl/mGYGJigkWLFuHu3bv4+OOP5eI0/zbRkXx9fXHixAluXwBw+vRprF27lns9aNAgXLp0Cffu3YNEIuHOPW5MLn369IFAIEBdXR1ycnK47QQCgcxpUMeOHePiKovZUtyioiIsX75c4XvWVTg7OytMvCKRSKafG/H5fOzduxc2NjZc2bBhw5S+N0+ePJE7e0AsFnPHnZ2dHby8vBAVFYVPPvkEGRkZiI+Px7lz52BpaYnNmzcjMjJS5Ut6G3+RNj+ut27div3790MgEHBlKSkpcp8ZACgtLcVrr72m0v6ea1qd2lOTOmc1ZGZmUkBAAHemgpubG02YMIHGjRtHL730EvH5fFq/fj0REfn5+dHkyZNpwoQJFB4eTh9++CHZ2NjQV199JROzpqaGYmJiaNCgQQSAJk+eTCkpKTLrnDp1inx8fMjY2JiioqK4K9fu3r1Ls2bN4tb7/fffydvbmwCQhYUFbdmyhR4/fqxS29Q5q+Hp06fk5+dHTk5OtGnTJvLz86Pt27fLrFNcXEzm5uZkZGREo0aNoitXrhAAGjduHJ04cYIyMzPJyMiIhgwZQufPnyciIh8fH7KysqLg4GBKSkoiLy8vWrNmDXcFnrKYje9R87iJiYlyM/fKaPqshqioKLmzMXJzc2nOnDmko6NDERERdPPmTbntLl++TD4+Ptzr1t6bI0eOkJWVFenp6VFMTAyVlZXRhg0bCAA5OzvTqVOniIiotLSUPDw8CAABoJEjR9KdO3dIIpFQr169CIDcMapIamoq+fr6EgCaPXs2ZWRkyCyPi4ujd999l3bs2EEBAQGUlJQkF0MkEpGpqanGryLsjmc1PLOJVx13794looYDJzs7m86dO0eVlZUdvp+O0pZLhisqKujatWstLq+traWrV6+SRCIhIqK8vDyZ5dXV1SQWi7nXPj4+NGXKFBIKhZSbm0tCoVDtmIriEpHM6Veq0HTiraurI2dnZyouLlZ724cPH8qVKXtvVFFUVEQFBQUyl56XlJTQzp076aeffmpX7Kb++usv7v1sbs+ePTR8+HC1Ln/vCN0x8T6zN8lRR+MNTQwMDDBq1Cgt16ZzmJmZKTzhv5G+vj4cHBy410OHDpVZ3tKYY69evbgr9tSN2VJcW1vbFuvZFejp6WHHjh34/PPPkZCQoNa2ffr0kStT9t6oYvDgwXJl5ubmyMnJwbffftuu2E01HS5p7t///jf27dvXpS9/7yqe6TFepvM8efIET5480XY1tGbEiBHw9PTkzs/uaurr65GcnIyNGzd2+hkGUqkU4eHhWLJkyf9v716DojizPoD/BwZwUEAlhYjX4KpoIiPBeANREhDXJSJRQ4IimF0UoiS6GgrzQtTIKpauWisWi4ALSSRZA/FSCSmplUKjokTxFhENiRAuUVSUAeUyl/N+sJhynGGY4dIjcn6fnKd7Tp9+fDjT83T3tPqEH9OPCy8zikKhQEpKCqysrGBtbY3k5OQO/wZETxccHAwPD48uu1mhK1lYWGDZsmUG37jSGQUFBYiKisLs2bO7fVsvCp5qYEYRi8UIDw9HeHi4qVN5Ljzv0yJC8PDwMHUKPQ4f8TLGmMC48DLGmMC48DLGmMB63Bzv77//jvz8fFOnYVIymQy//fZbr++HViqVCleuXOmy3ydgPUtxcbGpUzCaiKjn3Cw/Z84cHDt2zNRpMMaeQ42NjT3mw7dHFV7GWiUkJODIkSMoKCgwdSqMGY3neBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBceBljTGBiUyfAWHsUCgV27twJIlK3nThxAtXV1di2bZvGugEBAXBxcRE6RcaMIqKnRzNjz6lp06ahsLAQFhYWAAAiAhHBzOzJlzalUgmVSoXq6moMGjTIlKky1i6eamA9QkhICMRiMZqbm9Hc3IyWlhbI5XL1a5VKhZkzZ3LRZT0CF17WI7zzzjtQKpVtLheLxQgNDRUwI8Y6jgsv6xFeeukleHt7q6cWnkVEmD9/vsBZMdYxXHhZj7F06VKYm5trtZubm+Mvf/kL7OzsTJAVY8bjwst6jLffflvnEa9IJEJISIgJMmKsY7jwsh6jb9++8Pf3h1iseRWkpaUl/vznP5soK8aMx4WX9ShLlizRuJ7XwsICCxcuhEQiMWFWjBmHCy/rUebOnQtra2v1a6VSiSVLlpgwI8aMx4WX9SiWlpZYuHCh+kYKW1tbeHt7mzgrxozDhZf1OIsXL4ZSqYSFhQWWLFmiNefL2POObxlmPY5KpcKgQYNw7949FBQUYOrUqaZOiTGj8BEv63HMzMywePFiDB06FFOmTDF1OowZrdce8W7cuBGbNm0ydRqM9VpSqRSXLl0ydRom0asnx6RSKXbv3m3qNAR16tQpJCQk4LvvvjN1Kp127949vPTSS52KkZ6ejkuXLvW6cWBqWVlZOHXqlKnTMJleXXj79++PWbNmmToNQT18+BBisbjX7Xdb8vPzUVZWxv0hsEuXLvXqwstzvIwxJjAuvIwxJjAuvIwxJjAuvIwxJjAuvIwxJjAuvIwxJjAuvF1AqVSitrbW1GkIRiaTmTqF54ZCoTB1CibTm/e9s3r1dbyd1dDQgF27dmHfvn1YsWIFYmNjTZ1Stzt//jwmT56MY8eOwdfX19TpmExTUxM2bdqEhw8fQiQSQSQSIS4uDo6OjhrrFRYWIisrCw0NDXB0dERoaChGjBhhoqx1Ky4uxpYtWxAfH4+RI0e2uV5TUxP+/e9/49GjR/jTn/6EIUOGoKqqCkFBQcIl+4LgI95OkEgk+OCDD3D79m1TpyIYFxcXxMfH47XXXhN826WlpYJvUxe5XI6wsDAsW7YMSUlJcHNzQ1JSEhYsWICWlhaNdSdPnoyVK1fiyJEjiImJee6KbmZmJsLCwnDgwAE0Nze3uV5BQQF8fHzg7OyM//u//0NQUBA8PT3h6OiIVatWCZjxi4ELbyeYm5vD3t4e/fr1M3UqgunXrx8++eQT2NvbC7pdhUKB9evXC7rNtkRFRWH69OkYM2YMACA8PBy2trY4c+YMIiMjtdYfNmwYnJ2dYWlpKXSq7QoODkZcXJzedQoLC+Hj44N//vOfmDdvnsaymTNnQiwWIycnpzvTfOHwVEMnKJVKmJub63zyLfDkyKj1B7uf1tLSov4jVKlUOh/g2BpboVDo/L3ZtmILQaVSgYg09ru9fXq6rXXfjPHRRx/h5s2bOnNp65Hv3eHChQvIyMhAdXW1RrtUKsWgQYOwf/9+SKVSfPjhh+plZmZmen8zuK3+MGScAJ0fC08/0eNZDQ0NCA4ORkxMTJu/BLd8+XLMmTMH165dg42NTYfz6E34iLcDkpOT4e/vjw0bNmD16tVoampSL5PL5YiNjUVERAS8vLzg4eGBkpISAE9+F2DRokWYOnUqysrK8Oabb8La2lrjCbm5ubmIj49HTEwMfH19ER0dbVBsoWzcuBHDhw9X32ff3j7l5eUhKCgIgYGBOHfuHCZMmIC+ffti8eLFePToEYAnP1QzZcoUhIWFAQBqamqwdu1aTJgwAQcPHsT27dvx7bfforKyEitWrEBaWhoAICcnB3Z2doLe879nzx74+PhgwIABGu0ikQjp6emYOHEi1q5di7y8vHZjJSYmIi4uDqtXr8a0adNw9OhRAIaNk64cC/o+uOLi4vD777/j/fffx/nz53H69GmtKYnx48eDiJCUlNSh7fdK1Ett2LCBZs6cafT79u7dSz4+PqRUKomIKDc3lwDQ5s2biYgoKiqKMjMziYhILpfTpEmTaNy4cSSXy6miooK8vb1p0KBBtGPHDqqurqbk5GQCQD/88AOpVCpyc3MjhUJBRESlpaUUGRmp3ra+2IY6dOgQ2dnZGb3frf7zn/8QAMrPzycianefSkpKyMXFhVxdXSkmJoZyc3MpMDCQANDHH3+sjvvWW2/RjBkz1K/r6uoIACUlJRERUUREBLm6umrkcvnyZfL19aWbN292eH+MGQdKpZIkEgmtXLlSa1lrjPLycnJwcCB7e3v69ddf1ctnzZqlsX5SUhK999576tfff/89iUQiys7ObrdPibpmLLTKy8sjAFRSUqK1bNy4cTR48GCKiYmhDRs2kLOzM7m4uNCtW7c01vPy8qIFCxYYvM1du3aRVCo1OtcXBR/xGkEmkyEuLg4rV65UHyX4+Pio53hramqQmpqKhoYGZGRk4MCBAxgwYACuX7+OGzduYOjQoRgzZgzs7e2xdu1aDB48GCEhITAzM8Mvv/yC+vp6XL16FQkJCVCpVBg1ahSCg4MNii0UNzc3jdft7dPYsWPh7OwMGxsbbN26Fb6+vsjOzsaECROQlJSkfmLws193bW1t283F1dUVubm5GD16dNftoB5VVVVobGyEk5NTm+sMHz4c2dnZkMlkCAgIQENDg9Y6zc3NWLNmDQIDA9Vtc+fOhbu7O2JiYtrtU6HGgkwmw40bN7Bs2TJs3boVGzduRF5eHkpLSxEVFaWxroODAy5evNhl237R8RyvEQoLC1FbW4uXX35Z3SYSiWBlZQXgyWU5ffr0gZ+fn3p564MYHRwcADw5IScSidTLJRIJ7OzsUF9fD1tbW4SGhiI2NhZff/014uPjERAQYHBsIeiaq9S3T8CTr7JPz/2JRCIEBgbis88+w+3btzF48ODuT7wLlJeXA2i/vz09PbF3714sX74cS5cuRVZWlsbyc+fOoampCUOGDNFonzJlCvbu3Yva2lq9fSrUWCgtLYVKpdL4oBkxYgQmTpyI//3vfxrzzg4ODrh16xbq6+t5ntcAXHiNUFlZCaDtGwgaGhrw4MED2NjYaM0Btnr6j0mX1NRU+Pr6Yv369Zg/fz6io6Oxbds2g2KbSnv7pIujoyNEIlGP+iNtLWqtHyj6hIeH48qVK0hMTNR60knrOCoqKsL06dPV7UOGDIGlpSUkEonePhVqLIwaNQoA8PjxY4326dOn4+LFi1AoFBpXajx9EML046kGI7Qe6Z45c0ZrmUqlwtixYwEA27dv11h26NAh/PHHH+3Gv3//Pk6fPo2goCCUlJRgyZIl2LlzJ+7du9fp2KYml8s1XhcWFsLZ2Vk9TSORSHR+LVepVOp/k4mfUjV69Gj069cPNTU1Bq2/a9cuvPHGG9i8ebNGu1QqBQCcPn1ao72urg7u7u6QSCR64wo1Fuzs7ODi4oKrV69qtCsUCri6umoU3aqqKjg5OT2Xl8w9j7jwGsHDwwPjx4/Hjh07cP36dQDA5cuX0djYiJqaGowePRp+fn5ISEjAunXrUFBQgH379uGnn35Sf52WyWRaZ4UVCgUaGxvR2NiImJgYAIClpSWio6NBRBCLxQbFFkJjYyMAaFzJoW+fWpWVlUGpVKpfHzt2DJ9++qn69bBhw3DlyhXcuXMHSqVS/Sie1iI3cOBAlJWVoaWlBefPnwfw5Kv/mjVrcPfu3S7eS91EIhGkUqlW4W1qasK9e/e01heLxTh48KDG1BQAvPLKK5g3bx6OHz+u8b6TJ0+q+0RfnxoyFnbt2oX4+HiDbutt/VB8+v+r1Z49e5CdnY2ysjJ1W05ODjZs2KCxXnV1tUluqumxTHxyz2Q6elXDtWvXaPTo0SSRSGjy5Mm0Zs0a6t+/P7m7u9N3331H1dXV5OXlRQAIAE2aNImqqqqIiOjo0aPk5OREFhYWlJiYSLdv36YtW7YQAJJKpXTgwAGSSCQ0d+5c+vzzzykyMpJSUlLU29YX21CdvaohNDSUANDbb79NN27caHefTpw4Qf7+/uTk5EShoaGUkZFBfn5+9Mknn5BKpVLHraioIAcHB5JIJPT666/Tzz//TABoxowZdPz4cTp79ixJJBIaO3YsXbhwgYiI0tPTNc70d4Sx4yAhIUHj6oqioiJaunQpmZmZUVxcHJWWlmq95+rVq+Tv76/RVl9fTwEBAeTq6krbtm2jgIAASk1NJaL2x8mJEyf0jgWlUkm2trYEgHJycvTuT25uLs2bN48AUEhICBUUFGitk5ycTO+++y6lpaXRokWLKCMjQ2N5U1MT2dnZGfX/0NuvauDC2wFKpZKuXbtGd+/eJSKi4uJirXXKy8vpxo0bGsWlPQqFgpqamqi2tpYuX76svmStK2K36mzh7Qh/f3+aM2cO1dXVUVFREdXV1elcr7m5ma5du6be72f79fHjx+pL7Vo9fclWRxg7DlpaWkgqlVJFRYVR27l//77O9traWrp+/bpRsZ7W1liorKyk/fv30zfffNPh2M/67bffdI7JzMxMevXVV40aj7298PLJtQ4wMzPD+PHj1a/HjRuntc7w4cONjtt6F5yVlZXekyYdif08sLW11boc7WmWlpZ6+1XX3Kezs3PXJWgACwsLpKWlYdOmTUhJSTH4fQMHDtTZPmDAgE6dIGtrLDg4OOD8+fPYuXNnh2M/69kpk1ZfffUVsrKyOnSStbfiOV7W7WQy2Qv1U5Lu7u7w9vbGF198YepUdJLL5fjyyy+xdevWbr3KQKVSITY2FqtWrVKf8GOG4cLLuo1CoUBKSgqsrKxgbW2N5ORkjRNsPVlwcDA8PDwEvXnFUBYWFli2bJlBN6F0RkFBAaKiojB79uxu3c6LiKcaWLcRi8UIDw9HeHi4qVPpFkJPczxvPDw8TJ1Cj8VHvIwxJjAuvIwxJjAuvIwxJrBePcd79+5dpKenmzoNQRUVFUEul/e6/W7LpUuXcOfOHe4PgRUWFr4wJ1o7QkRk4hvgTWTjxo3YvXs3+vfvb+pUBPX48WM8ePBA65exequHDx9CpVK1eZ0t6x4ymQwODg6C/5D/86JXH/FOnDgR+fn5pk5DUIcPH0ZYWJjGvfe92caNG5Gfn9/rxoGp7d69u1d/y+A5XsYYExgXXsYYExgXXsYYExgXXsYYExgXXsYYExgXXsYYExgXXsYYExgXXsa6iCHPN+vJXvT9E1KvvoHCUKdOncLhw4c1HnPdp08fDBgwAMOGDYO3tzdGjBhhwgyZKTU1NSExMRFTp05FUVERSkpKIBKJEBcXB0dHR411CwsLkZWVhYaGBjg6OiI0NPS5GzvFxcXYsmUL4uPjMXLkSHX72bNnUVVVhaCgINMl94LgI14DeHp64v3330dSUhJu3bqFiIgI/P3vf8eUKVPw448/YuTIkYiNjTV1ms+10tLSHhHTWHK5HGFhYZg3bx48PT3x4Ycfws3NDUlJSViwYAFaWlo01p88eTJWrlyJI0eOICYm5rkrupmZmQgLC8OBAwe0nnLs6ekJR0dHrFq1ykTZvTi48Bqo9ZN/6NChcHV1xdChQzF79mykpaXhv//9L/7xj3/gr3/9a6/+4Y+2KBQKrF+//rmP2RFRUVGYPn06xowZo24LDw+Hra0tzpw5g8jISK33DBs2DM7OzrC0tBQyVYMEBwcjLi6uzeUzZ86EWCxGTk6OgFm9eLjwGsjMrO2ueuedd/Duu+9i//79OHDggNZyuVyu1fb0kZBKpWozdmsh1zW/pitud+roh8pHH32EmzdvarU/vd/GxjYkZne7cOECMjIyEBISorVMKpVi0aJF2L9/P/71r39pLDMzM4NYrHuWT18/GDpmOjsurK2t9S5fvnw5IiIiUF9f36nt9GZceLvIBx98AADIyMgA8GTwx8bGIiIijlS8YQAACRNJREFUAl5eXvDw8EBJSQny8/OxaNEiTJ06FWVlZXjzzTdhbW2t9cebm5uL+Ph4xMTEwNfXF9HR0XrjdqfExETExcVh9erVmDZtGo4ePQoASE9Px5QpUxAWFgYAqKmpwdq1azFhwgQcPHgQALB9+3Z8++23qKysxIoVK5CWloa8vDwEBQUhMDAQ586dw4QJE9C3b18sXrwYjx49ajeurpgAkJOTAzs7O5w6dapb+6PVnj174OPjo/MpwSKRCOnp6Zg4cSLWrl2LvLw8vbHa6mMABo2ZrhwX+g4yAGD8+PEgIiQlJXUoPgNg4sfLm8yGDRto5syZBq/f2NhIAOhvf/ubzuV37twhADRkyBAiIoqKiqLMzEwiIpLL5TRp0iQaN24c3bp1i7y9vWnQoEG0Y8cOqq6upuTkZAJAP/zwAxERqVQqcnNzI4VCQUREpaWlFBkZqTeuXC43aD8OHTpEdnZ2Bu93UlISvffee+rX33//PYlEIsrOziYiorfeeotmzJihXl5XV0cAKCkpSd0WERFBrq6u6tclJSXk4uJCrq6uFBMTQ7m5uRQYGEgA6OOPPzYo7rMxiYguX75Mvr6+dPPmTYP3z9hx8DSJREIrV67Uuaw1Znl5OTk4OJC9vT39+uuv6uWzZs1S/7u9Pq6oqGh3zHR2XDwtLy+PAFBJSUmb63h5edGCBQuMjt1q165dJJVKO/z+no6PeLuInZ0dgCdTAjU1NUhNTUVDQwMyMjJw4MABDBgwANevX8ejR48wZswY2NvbY+3atRg8eDBCQkJgZmaGX375BQBQX1+Pq1evIiEhASqVCqNGjUJwcLDeuN3xtNvm5masWbMGgYGB6ra5c+fC3d0dMTExALS/lhryZNuxY8fC2dkZNjY22Lp1K3x9fZGdnY0JEyYgKSkJRNShuK6ursjNzcXo0aMN2b1Oa2xshJOTk951hg8fjuzsbMhkMgQEBKChoUFjuSF9PHToUL1jRuhxAQAODg64ePFit8TuDfhysi7SeoZ93LhxKC4uRp8+feDn56de7u3tDeDJgDU3N4dIJFIvk0gksLOzU8+Z2draIjQ0FLGxsfj6668RHx+PgIAA5Ofn643b1c6dO4empiatH02fMmUK9u7di9ra2g7HNjMzg42Njfq1SCRCYGAgPvvsM9y+fbvDcYVmSL97enpi7969WL58OZYuXYqsrCz1MkP6eODAgXrHTHvjrTs4ODjg1q1bqK+v1/h/ZIbhwttFTp48CeDJgG9oaMCDBw9gY2PT5vxfe1JTU+Hr64v169dj/vz5iI6OxowZM/TG7WqVlZUAnjwuaPr06er2IUOGwNLSEhKJpEu35+joCJFI1KP+kA09wRQeHo4rV64gMTERmzZtUrcb2sf6xkx74627iEQiWFlZCba9FwlPNXSBhw8fYsOGDRgxYgTWrVuHsWPHAnhyYulphw4dwh9//NFuvPv37+P06dMICgpCSUkJlixZgp07d6qPXjoa11hSqRQAcPr0aY32uro6uLu7QyKRQCKRaH19BrTPupOOJ0w9e/a9sLAQzs7O6Nevn0FxdcUUUr9+/VBTU2Pw+rt27cIbb7yBzZs3q9sM6eP2dHa8dURVVRWcnJyey0viegIuvAa6e/euzvbi4mJ4e3vD0dERhw8fhrW1NUaPHg0/Pz8kJCRg3bp1KCgowL59+/DTTz9h8ODBkMlkWhenKxQKNDY2Angyd9g6v2dpaYno6GgQEcaMGaM3bld75ZVXMG/ePBw/fhz37t1Tt588eRKffvopgCfXpF65cgV37tyBUqnE7t27AUCjIA0cOBBlZWVoaWnB+fPn1e1lZWUal08dO3bM4Li6YpaXl2PNmjVt/l91NalUqrPwNjU1afRXK7FYjIMHD+Lll19WtxnSxwD0jpn2xhvwpOjHx8cbdNtv6wdi63jUpbq6Gq+99lq7sVgbTHtuz3SMOZv9448/0rx58wgA2dvbk5+fHy1btoyCg4Np4cKFtGPHDmppadF4T3V1NXl5eREAAkCTJk2iqqoqOnr0KDk5OZGFhQUlJibS7du3acuWLQSApFIpnThxgioqKkgikdDcuXPp888/p8jISEpJSdEb11DGXtVQX19PAQEB5OrqStu2baOAgABKTU1VL6+oqCAHBweSSCT0+uuv088//0wAaMaMGXT8+HEiIjp79ixJJBIaO3YsXbhwgYiI/P39ycnJiUJDQykjI4P8/Pzok08+IZVKZVBcXTHT09M1zvQbojNXNSQkJGhdWVFUVERLly4lMzMziouLo9LSUq33Xb16lfz9/dWv2+tjQ8aMvnGhVCrJ1taWAFBOTo7efcrNzVWP9ZCQECooKNBap6mpiezs7Izq52f19qsauPB2s/Lycrpx44a6oBhCoVBQU1MT1dbW0uXLl0mpVHZJXCLjC2+r2tpaun79us5lzc3NdO3aNXWexcXFWus8fvxYfXkc0ZPCO2fOHKqrq6OioiKqq6szOu6zMYlI45ItQ3RmHLS0tJBUKqWKigqj33v//n2tNn19bKi2xkVlZSXt37+fvvnmm07FJyLKzMykV1991eix97TeXnj55Fo3Gz58uNHvMTc3h7m5OaysrNo8WdKRuJ0xYMCANnOxtLTE+PHj1a/HjRuntU5bc5W2trZwc3PrUFxdMZ2dnXXG6g4WFhZIS0vDpk2bkJKSYtR7dT1OXl8fG6qtceHg4IDz589j586dnYoPAF999RWysrIMOknMdOM5XmYSMpkMMpnM1Gl0mru7O7y9vfHFF1+YOpU2yeVyfPnll9i6dWunrkJQqVSIjY3FqlWr1Cf0WMdw4WWCUigUSElJgZWVFaytrZGcnNzjf1goODgYHh4e3XazQmdZWFhg2bJlBt2Eok9BQQGioqIwe/bsLsqs9+KpBiYosViM8PBwhIeHmzqVLiXkFIepeHh4mDqFFwYf8TLGmMC48DLGmMC48DLGmMB69RzvpUuXMGvWLFOnIagHDx5AoVD0uv1uS2VlJWQyGfeHwCorK9GvXz9Tp2Eyvbbw8h8aY6b17INAexMRkYl/aYQxxnoZnuNljDGBceFljDGBceFljDGBiQFsM3USjDHWm/w/pk+3h2qfscYAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "plot_model(model, show_shapes=True, dpi=90)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "b9fd709e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(4124, 896) (4124,)\n",
      "(1032, 896) (1032,)\n"
     ]
    }
   ],
   "source": [
    "y = sample_df['positive']\n",
    "X_train, X_test, Y_train, Y_test = train_test_split(X,y, test_size = 0.2)\n",
    "\n",
    "print(X_train.shape, Y_train.shape)\n",
    "print(X_test.shape, Y_test.shape)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "371b7181",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/6\n",
      "129/129 [==============================] - 80s 603ms/step - loss: 0.6558 - accuracy: 0.6091 - val_loss: 0.5698 - val_accuracy: 0.7326\n",
      "Epoch 2/6\n",
      "129/129 [==============================] - 77s 598ms/step - loss: 0.4764 - accuracy: 0.7859 - val_loss: 0.5393 - val_accuracy: 0.7287\n",
      "Epoch 3/6\n",
      "129/129 [==============================] - 66s 511ms/step - loss: 0.3778 - accuracy: 0.8421 - val_loss: 0.5662 - val_accuracy: 0.7364\n",
      "Epoch 4/6\n",
      "129/129 [==============================] - 63s 485ms/step - loss: 0.3176 - accuracy: 0.8792 - val_loss: 0.6319 - val_accuracy: 0.7180\n",
      "Epoch 5/6\n",
      "129/129 [==============================] - 71s 552ms/step - loss: 0.2741 - accuracy: 0.8972 - val_loss: 0.6837 - val_accuracy: 0.7190\n",
      "Epoch 5: early stopping\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7fa3e3afe790>"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "batch_size = 32\n",
    "model.fit(X_train, Y_train, \n",
    "          epochs = 6,\n",
    "          batch_size = batch_size,\n",
    "          validation_data = (X_test, Y_test),\n",
    "          callbacks = [EarlyStopping(monitor='val_accuracy', min_delta = 0.001, patience=2,verbose=2)])"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}