Face-recognition/Face recognition/Face recognition.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "165c3108",
   "metadata": {},
   "source": [
    "# Система распознования лиц\n",
    "\n",
    "GitHub Repo - https://github.com/IgorVolochay/Face-recognition"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f73e87c4",
   "metadata": {},
   "source": [
    "## Requirements\n",
    "\n",
    "* python == 3.8.5\n",
    "* dlib == 19.23.1\n",
    "* tensorflow == 2.2.0\n",
    "* skimage == 0.19.2\n",
    "* pillow == 9.0.1\n",
    "* matplotlib == 3.5.2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "24f05414",
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "import os\n",
    "\n",
    "import dlib\n",
    "\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense, Dropout\n",
    "\n",
    "from skimage import io\n",
    "from PIL import Image, ImageFilter, ImageOps\n",
    "from matplotlib import image, pyplot"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c68df874",
   "metadata": {},
   "source": [
    "### Установка предобученных моделей распознования лиц:\n",
    "\n",
    "1. https://github.com/tzutalin/dlib-android/blob/master/data/shape_predictor_68_face_landmarks.dat\n",
    "2. https://github.com/ageitgey/face_recognition_models/blob/master/face_recognition_models/models/dlib_face_recognition_resnet_model_v1.dat"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "e33aa700",
   "metadata": {},
   "outputs": [],
   "source": [
    "sp = dlib.shape_predictor('shape_predictor_68_face_landmarks.dat')\n",
    "facerec = dlib.face_recognition_model_v1('dlib_face_recognition_resnet_model_v1.dat')\n",
    "detector = dlib.get_frontal_face_detector()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "3927fdf2",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Detection 0: Left: 70 Top: 248 Right: 869 Bottom: 1047\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "dlib.vector([-0.0902129, -0.0284728, -0.0201088, -0.0237396, 0.0195988, -0.0624656, -0.0341601, -0.0645328, 0.145692, -0.103387, 0.164995, -0.0667768, -0.19563, -0.0463215, 0.0462609, 0.0386064, -0.204915, -0.0965468, -0.0547373, -0.130571, 0.100364, 0.0206162, -0.0427951, 0.108554, -0.284338, -0.223312, -0.115767, -0.17538, 0.0108243, -0.0472676, 0.0455159, 0.0537128, -0.0945584, 0.00360177, 0.0861953, 0.0713386, 0.0278726, -0.0417318, 0.217425, -0.0155115, -0.131425, 0.0396619, 0.0762383, 0.219826, 0.112767, 0.0618684, 0.010629, -0.050927, 0.225475, -0.277578, 0.061519, 0.183117, 0.0907261, 0.0902939, 0.156285, -0.202908, 0.049825, 0.165025, -0.225213, 0.124595, -0.00210925, -0.00164449, 0.0281997, 0.0171772, 0.118259, 0.0737286, -0.16149, -0.0689448, 0.155305, -0.175737, 0.00137892, 0.170224, -0.0961124, -0.265967, -0.20611, 0.0573926, 0.432559, 0.168564, -0.122163, 0.0571624, -0.0393574, -0.0610366, 0.071344, 0.0299074, -0.113738, -0.0179229, -0.0436408, 0.0608074, 0.205917, 0.0866748, 0.0209013, 0.263666, 0.0228976, 0.0307268, 0.0403442, 0.035415, -0.105188, -0.0733519, -0.0467792, 0.022078, 0.053307, -0.108359, 0.00818931, 0.06266, -0.223834, 0.1759, -0.0215296, -0.0689591, -0.0981493, 0.0681423, -0.0973912, 0.0902572, 0.147521, -0.347928, 0.117237, 0.143078, 0.0655791, 0.166869, -0.0161545, -0.000808923, -0.00869995, -0.0707334, -0.165803, -0.0982898, 0.0219621, -0.0364028, 0.0172211, -0.0176742])"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQcAAAD8CAYAAAB6iWHJAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8qNh9FAAAACXBIWXMAAAsTAAALEwEAmpwYAAD6Q0lEQVR4nOz9ebAkSX7fiX3cPY68M9/9Xt1VXX1NXzPTMz0DYAY3BxcJ7JIEARFY0aiVIGr3D8q0OrgymlZm5JpJa9IuRdGMMhpPrHaNJARSIAhgSe4QwIDAAIOeq2e6e6a7urvuqlfvzjsOd9cfHhEZmS/f0cf0VFP9K8vKfJkRHh4e7j//Hd/f7yestXxIH9KH9CHNkvxud+BD+pA+pIeTPmQOH9KH9CHNpQ+Zw4f0IX1Ic+lD5vAhfUgf0lz6kDl8SB/ShzSXPmQOH9KH9CHNpfedOQghflwI8W0hxDUhxF95v6//IX1IH9LpSLyfOAchhAJeA/4EcBv4Y+B/Yq195X3rxIf0IX1Ip6L3W3J4AbhmrX3TWhsD/xj4mfe5Dx/Sh/QhnYK89/l6Z4Fbpb9vA58qHyCE+CXglwCU5z3fWlhACPebJZNyrMWKmZZzAUjkb6I4vvwZC0KU/p6hvFlb+lx8mpWyRPnowx/nU96n/BoWIQQ6SVAIlFIYrUnjGOUprLGkaYK1IKQgDALGUUSSpFhrqdfqGGswxmCtxVqL1hoAKSXGWMJKSLc3pFGrkCSJ64UAawxKeSgpUUqRpinGGIx1oyOkII1TlJLUajV6/T5KeRitkUoSBgFJkmKMJtWGVqvJwUGPwFdYa/GDwI28tQxHMaHv4QceIBgMRlQrAcYarLFoo2k2m4zHY9IkIQxDAJIkIQhCgsAnimOstaRpiu8HJGmCkBLP81C+R6oNvu+DEKXHMDtR5jzsw4/ndHRM0+8pzczt95r2HuxsW2tXZr9/v5nDiWSt/TvA3wFYXFu1n/u5n0VIixAWYwwAqU3nnQe4hV9+n/3dWusWYNbWLAlROq7gShKQWKtnjnW/W0rfm6Nnl0QgrEFYUEIDBpVqbBKz1mpz49VvEUdjdBJzsLdHvVqj3+/y5JNPsrW1BVIxGo0AuHnjNsvLy6ytrdEbDojjGCkl/X4fY+Hy5cvs7OywtLREpVJjdXWVl176Jv1+H6UUlx+9yt7eHnt7ewRSsby8zMsvv0wQhvi1CqNhRGI01aDC/fv3WV5qci8e8thjV7lz5w4HBz06VUU3GaEtLC43WFtbo/XU43zlyy9y7twZ7ty5y9WrF7HWsrS0wr3b9xiPh5w9e5bRaMSNm2/x5JNP0mp2uHb9Nfb29nj6Ixc5v7HBm2++yf7+Pp98/jO8/PLLnD17lv3uPrV6A4NkZ2eH/UGPK1eusLO3y8LiKuceuYwMQ2IkWmTPXAqMKM0HOxGWrZp8FqX5MY+K+VOaV7Nz7DtGeZ/F/DlbpmJOnsJckB/7T//mP7gx7/f3mzncAc6X/j6XfXckGWEAg8BimB4ca92uO/uQ5v190mCJGSlg3oM/zWQQpQl36JrGMQipDVIZrE4R45gAS393Fx3HdMIKvThidXmJTqfDm2+O2NzcZHl5mUq1xR/90R/xxBNP0G4t8MYbb/DNb36T8+fP4wlJ96BLGIZsbm+zv7dHq9nk2rVrCKH4whe+ANYjiiIee+wxvvKVr7Gzs4Pv+1Q8xe7uLp7n8eDBA4JqhUq1TqNep9Vq02g0GI1GrKys8NprrxGGIR//+LNcu3aNT3/6Ba5fv04cR9y8eRMh7vDoo4/Q6/W4dOkicZyyuNjhzt0btBYWeONrdwnrDRYX6wjl8+b122jzOs1WB+UFDEcR3/jmy1SrVT76sY/zpS99Cd/3McawurrK/v4+nh+gk4gnHnmU119/naWVZW5fv86tO3e4+pEnaK+sIkMfLSypFSivNM2n+fvk0WSbxVEbx2QjmL8BlZ+1OGauTZ1zzHyanTtCiJKkengznDe/yxviUZvmcfR+GyQ9nEHyR3BM4Y+BP2+tfXne8Ytrq/Zzv/Bn0Do5sk0nOptDgwUTKUEIQZoeljZm+pado5Ez7eSSA8zn3GXJYfb6wk4ekgcoa8AmeFGKjocwjmg1QsQoxsYpvV6PhYUFut0um5ubDMYjjDFIKRmPx8UC6XQWWVlZYTjss7/fZePMOdrtNtfeeotaWCGKItI0pTvo4/s+SZIwHMSMRiNGoxFhtcJ4PCYMQyqVAK01SZK4/ns+Wlsq1RrWJJw5c4Zr165x/vxF0jTl5ZevsbTcKsb97NkzrK0uc+/ePW7dusWVx59i6959rDboaITneQRBAL7iscce43d/9wusr6/h+4qdnR0WFtrs7vXwPI9ms4nAUK1WuXfvDo8/cpWvfe0VLlxYxfd9PM9jY2ODNE35/S++yGc+8wL1ep0//OpXGSQJS+sbvPC9nyHRKYNojAx9EmHxvACMxZrJZmKVLJ51eR0c9ZlsBkhZkjiEmGImxzGG2flhjvh+3nWPnF8zzCH/2xgzdayUslA7Z9v5p3/zH3zZWvuJQ9d6v6MyhRA/CfwNQAF/31r7Xx517OLaqv2xX/yzWKux1k49hPIDmno4QhQMwxiDUmpqUE4edHNowggU89SKnI5iDgDCAsYiLSg0AWCTCBuPSAYDkkGfTquGpy0mScE61eDevXtF/6Moot1uU6vVqNfrvPHGG+zvd7ly5RJJGrG+doZXX3ud9Y2zJEnCYDCgWq2yvb1Nr9dncXGRJEnw/ZA4jonjmLBaRWa2hkSnaK1ZX1/n5s2baAPVRp39/S7WanzfZ3V1lTQ17O/vo7XG4nbZIAgYDAasLy8xGAxYXFzkwe4eURSh0xQSTbVaBWCUxmxsbOD7Pq+88grLy4tIKfE8CSLg/v37eJ5Hp90iSSJ832fc73H+/HkajRovvfQSFy5cYGlpid3dXWqNOtVqla2tLS4+cpU3bt3CSMU4NTzzsY8yGA2ptBrYwMMg0EmKp4LJc3sHzMHOSKqHFufbEOdPVhJOpnmL/ajjjpKgf+X/+Q/nMof33eZgrf1N4Dffybll0SjfTa21BTMoXePQ8e+ECU701Jm/Z+iopo1x9gXfCqS1SAxSG6L+CJsOGR10WWzWEEZjU83du3e4dfMeZ8+edQswM76tr68zGo1QSnFwcEC73eaxxx6jUqnwhd/7HZJYs7K4wI3rbxKGoRsXk7K+tsLduzt4XjdbRJucPXuWIOhQrdYZDod0u12azSb3H2zy+uuvEwQB62ur9PpDrNXFPddqNYwxeJ7k3r17LCwsEUUR1hg8pTLDYcC9e/fcWGHRNuWxR65y8+ZNjDFom3D//l0qlQqf/exneemlrwHQ640ZjlIqlYq7VrVONzOOfuxjH8Nay717d1hZWSEIAra3t9na2iLcD3nsscd4642bLC4uIw3sdw+oNtvceus6KxvrxKMxgVfHYAmrFdJITzYQIQobX77ZzKoUh3b00ndSSmdM1pNxEpxO33+/KV8rb+uc71Bf3hMSgCwWpkJKDykn/GyiThh8qfClQhiBQuHLEKV8pPQOcXohBJ6QUy9p3bWklCA9rFBI4SFFfr3DfD6XSITzMzgJw8rJC2djUICSoAyYaMTu5l1EollpL/DEI4/zz37l99jZPqDfG7O0uEiz0aBWrSKASxcv0mo28ZTiEx//JOfOnOfJxz+CtIY3r32LWhhw7uxZHmxustxp85HHH8OXCrTBJCnPPfUIl86dIxkN+L5PfYqqrxj3+7z15uv09/ZQAvb2d2g2auzvjdjbOWB1dR0lJSsLi6wtr1CvVPn6l7/Oa6++xje+9jqteove3i6D7gEmjpBGMx70UULTqtdo1AJWFjpcOX8RrKZa8Qh8RTyKsbGhf9DntVdfodNsYVONL31q1Qq+EBid0t3fIx6NGfaGfP7zf8j29gPSNOXixYsc9Lp0+wMWV9eo1du8ceMma+vrLC10OL+xzhOPXEInY+7duc3v/97vEgjF3uYWVaXwjMGTIDEYm0JJHcXYyesYKkusOorBGIQubUxCnPgyc2fT26DS/CpsbuV5Z6eXtbaW1Bi0tcW1y6+j6KHzVnwnKN8ppuiESXASlaWVoziyRGQ2JGdQtUnEqLtHsxqSjEbomuSf/JN/wqc+/gi3b9/F931WVpZYW1lFCcloMMSTCp2kNGp1vvqVF7l71x3nB4LxeMzTTz/N3Xt3WV5e5NKlS7z++uvE0QgpJbdv7bCxscHy0gLbWz77u3vcvnmHc+fOUa9WAMl4HFGr1bj62KOcWVtHSsmNt96kWnFeivX1daqtJtXAZ2Njg9deew1pU1qNOvV6nfv37xOEHtJa0igmTVNGoxFLKxWGvS5bm/dIU/jYx57hxq07eJ7HcDhExxGiElCvVQoPjJXOlZuYBOEJhBGsLLdIkoTl5WVu3LgFEvwwJE1T4jhlf3+fX/zFX+R/+Je/hgornL94gXgc0Wm36Xa7RMMRBzu7rKysoLVG+R4GiQC0sVgsyvMmbm8hjnUZlp+053lgMnG99P1pjX7HqS/vmyfkGHrfbQ5vh5bWV+1P/MLPZq7LshdAz9gfQJhc9xOZEVJhhMlUkPTQ4KsjZoBGF9fKpZZ5Y3Qa5iCMRglDACy2mrz5ja8TpCmMh1QqtULvXuksc/bsWb70pS8hjGZlcQnP8zh79iy7e9sMh0Nu376NMYbLly/TarXo9Q4AiOOYoFqh3++zv79PHMe0G22EEIRhyM7WAy5cuIDv+wxHzh5Rr9eJ4hE6tQRBhd5oiFCS4XBYuETT1In5QjiDbhRFrKysEEUR/X6fMPQZDAZYa3nuuef41re+VeAkfN+nOxg6e0R3j3q9ThzHGOEMk/kz0VrTarUYjUbsD0ZgFNpAJVQYLBJBFGlqdc8tRKEIqwEHvQFpmiKETxAEBEHA+Y1VXn/rOqvra/THEXfv3UdLWD9zjubSAmG1woULF7BSYHEidlJSDyipFYcwNCUqexiUnSzi9G0BJOa0+3bW4TzX5oy0UP5Nn9D2rz4sNof3k8puzkOc+H3giVbk3grD669+i3YQMB50kWnCzv0HeFLw6CNX+eqLX+Vb3/oWnU6HIPAZjUZordna3qTTadFsNvmpn/opXn7ZOXV2dnZYaLsdNfQDhJLUlpZZ6iyglEInhlE0JvQD2vUanuc5JuILwqUWvYMuQiqCUHH//m0Go4h+P6bZDFlfX6fb7bK/d0ClMqBacbYJrCVNRrQaTZqNCkop2i3n4nz9tW/hS0FnaYGlhUXevP4Wa0sdRuOY5toK7Xabg4MDjBVYY+j2eigpaTUbjMZ9NlbWUGwhZcB4FBWuyzR1HoezjzxKmqbc33qA9ARXrz7GjRs3iOOUwWDA1atX2du6Ty2s4KuAZt1HJykf/eQn+fy//QN+7E/+MItLS0T9IbHVVCo1pOchlcQKCuN1HMeF3eP9pId1g37oJYcf//N/1hl9yL0NOnsv+ZWFRWacU2td2Bo0utilcsrPO43kkBsyhckW+RE7ytQYlox4EvCwdDzFwb17pL1d4v09TGrROqFSqXDv3j0ibdBxggDWV1ewWtNoNOh0Omxu3qNWqZImkqeefpQvfvGLLCwsoKOIMAwZDoc0FzuZmB2jtcazgnq9zmuvvcazzzzD17/+dR599FEMmsFgQKPRoD+Kicdjhv0BMvCpVquFRXs4HDIaRaysrXKwt0+j0WBvbw8pZYG+rNZrPPvss1hr+cpXvkKn1aRarVKv16nWa4Wh7tatm6Rpyvnz5+l2e0gpqdfrbG1tYa3l4ODAuVMbzYIpKqVYXV3lxo0bpFqwvr7Ot7/9beqdFt3BGGN9hCcg1VSqVWKd0q43GA6HPPLII7zy2uuMo5SVjTMMkxGJjkFKnnj6IzRaHZQfYIXAC3xGOnFSCXLKE2GMAXn4gZclB2kmOJt3KzkcRXPVDZsbFzP3ujD/fyg5lMV6UTIAHqOPKaWmm3i/XbVKImxmiEw1wiZsPtgkiCL6+11kHNHd76G1Jk1TB7pJUjylWFpYJPR96q02/X6f9dU1hDXUGzWU9fji73+BJ554gm9+85usLDm///LyMnGakEYxVmvOnTlDPI7o9XpcvnyZ7e0HXLlyCWs1ldDH6hCMxleQCkOzUWUUjenubXPlyhXHbKpLWAHKD6hVQqIoYn19lVqtxs7ODp1Oh8XFRTbv3WVhYYHnnnmaZrXqGJPnOVSiMQyHQx69fMkZgD0Pm0kDOk1YX1slTVNWV5bRWuMHITdv3qTTbuH7PqGSbKwsc//BNjsP7rHYbmCEIRACY+Fgv0sQBKSeRzQeI2oNQj+gu39Ao1rB2DHbDzZ5+tlnePnVb2KU4NWXX+Hc+Yt0ltw1rZI0FjvoKEYGITJzex9HU7YrcwSi6pR0lDt+Hk1vQLlb8l1d/kR6uJnDB4yEECAF0oJJNZ7RKGvobT+grXzMcIivYHFxkYODg8IlG3qKSqVC4EniccTezi5KKT7/+c/z7LNPs7v1gFs37rC03GJ3Z4snHn+UWqWOEFksBpYwDHnw4AGDfpc0Tmg2aiRJgrIB1WoFay1h6LPQbmKtJU6zOARtCKq1Av+wu71DvQ61WgOLpLO8VABqgiDg0oXz9Ho9Qj+gc+ECYRgShiHJaFiAlHK4clyvYYUp4j0WOh2SJCFJEhrNNsPhMMM5ePSGA5YWFxBCUKk4tWV7e5vlxSUXb5GmDEZD6mHMrTv3adVC4lSTRGMUlgcPHrC4uMibb75JvdUk8D20Trhz+yYYy5MfeZKbt29x+9Yt7t9/QKVS4fGPPMloMKDdbGE978Qd9jtJswbzk5jFO6WjUJ3z6CFnDrZQI8wxll2QJTCKQAiFthYh8+8mR85iF2bJGTYddsKJasZZnnCBSpNOTKPkAIS0zqFpNJiEVuBz/bW36FSrbN65RV2Ar0LSNMbqhGG/S6vVotlsZzp2xMriItsP7nH58hVeeOHH+PKX/pBOp8MTj19FSE2SpkhPsH+wDUC9Xsf3fXZ39vE9QbVSRco6AEHQxJeqkLbq1WrR11B5CCWdK0tKkiRBSsmlC2eceO8i1Ehjt7B936kelUqFpU67UDE6nQ6+7xMNB0gpne9f2oLx5YFeSeLEd8/zHDNJEuxiswB6jaJmgWQVmSU4lAtIr8JgMEBrTXcwZHdvn7AasLt3wDgZMx6nSOVg4Ts7WzzxhGMCcZTw2CNX8esNzp0X3L1/j5372/hhFaHG6HrC5u27nL1ykSSK8QQE0ie1Fi1ACImxtgA15YvXGltA5EUGqHJYGzc3joJeTyEjT1j4s6C+Q/BnYQFboKlP4mmqWBvW3dMpodQPOXM4HVk7ibG01g2skB5lDnBaJNk7oWLi4CQFZQ2eEextPqDpK+Jel06thkxTpITheEij0aDRaFCruV37/Plz3Ll7i/v37/LII5e5fPkKN25c5/z58wwGg2LCLC4uIoSg3WgCDpxkraXZbNJut/E8rwBQSSlRZCK9tegkKUR/aSy9QR/l+wS+j1IKpVwAVhRFrg3p9NtcLYgiZyz0fd9hQ7Jd3xhDrTLBk+SowzKsvRgja9na2mJpaQnf94vfG1ojMyYFliiKaNZqGOmx0GnR6/Wo1+ssdRZYWRmyd3BAb9Dn/uYOaWoYCIGVgjevfYvltVXu3H3Am9eugeehJTzY2ufixTP0hxFSeYjUcO/uXZrtFjfu3ubcpYssLC4jfX9q3szu6MJyrP1p3jnw9gKiZtspj2P5/OMCCN8L+kAyh7kLfEol0zgjxdtve/bBOq598nF5v0SaII2hEYZ0h0PG3QPsYEAlVERphPJ9as0aOtYuorLXo1IJaTRrLC0tQbvBpUuXEEJRr6+gEKytrbnFKzTCU1MI0Xq9XuA4giAoduVioVoKBJ9JnfswjmOENtQa9czA5lQGKSVKSNrNlnMlZpLABI5eQ0pJGIZukcx4g3JXprBphqYMinHSWjMcDgmDgIvnzxbowvy30Lr7soFTByqea1t7Lu6jVg2wqeGgNyAMQzqtJv1Bj1oQ0h+OuXs/YjAcEw1STBxT8QQ6HaONj/AUi60q6TiC1GBtSpQkNIM29+/eox5UWF1eQXo+qTaF19xJj7ObinAYmWLXnp5kxzGG2WOOCgfI6bgAsMN/H0Zyzj33WAl8mj4wzCH3Llg7J5cDk0FSQmKwh5BfRw3q8bEWAq0Nao7VWpSZURJjPYmXnecZy+7mHb71ykucWVrmwuoK16+/ybkz69SrNbrdLne271ANq3ieR7VWYaG9iK88Wp02GMtCvUmr3kAFIVJCmqaEviiMmEEQFLt2TnmshAnCYrJNBQaZSb6HslV+OBxQqVSKQLVc6vA8D6VUsZANtjD4SiYTPBd9c6kF4xVwXY3Ofgto+uFUX/PzVYZizdvxRejg1lpjjQDlk8qU1NN4nqJRi+iPhlRCS6vV4KA7YGmhyebWDpcvCW7cvMvSYpvd3X2sjbGxQqEg1QhrMAakp1CeYNDv0hv0ef3mdV749PewsLKCzh63tDABWGfzRZChYfPxzeZAacbl4z1vwc8ecxyVpd2jJN5iHs/O+HmqNLMs5Hh66JnDvEE5jt8ZbCYOT77JYune9rXd7uehzdFRoZB5SKRExAkIQzIYcefNGzz56KP0dnf41ivf5JlnnmJvb4/RYEir1eLc+bMZyEhw4ew5Kr7H0vmzTtWoVJEIKn6A9D2sFSRJQjVUxUQpL+IyzTM4TYKCDgfgCCFoNhtTTKS8UJVSeJ7nbAFqcv0cYZqfl6Zp4U2SXmbnADCysE9IJQrmMsucy7EyFl2oP9aIyXOwpmCOnufRbLWJkpRqENJpN2g2G0jf5+zGGe7e32R5YZH+OGJ7e4fhOCWJRwjl4QuBNpJRf4AWsLG2TrXdZn11hTjVSJH3kfmrSZhjJ2GZMR9HxzGPd31OOW/FNHyzkARP0rAfeuYwS27Aj34yGodJcJLfu0OxuwVyRJi2mCw0I1w0nrLgpSnJaMTGyjJ725vINGZ5eYFbt27i+z5Li4sgLEHgMRz2+cxnPkOzVqdWryClpFWvFQvfiekpvhfieXU86fAaZXH/KJrGgWSifymAzIqJsWueaFws9Exq8JVCp9btpBbIDXCZJCc95RiAUFORiUq5KeZ5ElG6pimi2bI3JUEI174GL2NENjNuaq2R1hSMIQgCNOBHEYGnSI112ap6A1Y6i7SbTba3txnGCZ94/mN84ff+gK2tLYSCSlijPxhjpEEoyf72Fvc27/Pg3n0uPfoIIgjpLC1ipDNuz4xstthO78acZcansTuUGfRRv8NkDs4yoqOjA06GA+T0gWMOOeYBpgda5x4GkwdD5SARAHMk6Oko9cJdJ0u3JuYsQqORwsVOKCnwDAit0eMhaTzm1puvEyqDtJaV9Q33AKXzvqyvb7C6upohIgNajVoWTZnh9QElJV4YIBEZlFhNTdQJ9xeHHrb7Wx5e+MW4ZRZ2IRBSorLgsgIOnh0rAJMdm8Vog1CAyY41ICQWjcxcloUhMs8nYCYGSYMBIdDWFkyjbLiUOUZFGuctMRYpZMHMJKqQMHzfx2Tu1SiKSJIEkya0m1XGMTSadRr1Cv3xmC/83m8z7o9YXOjQHYyxaUIt8Nnr7SFVFb/qs7qyAYFHNIyphx7SaIRSkG0OKvOIpbYYoVNnhTqJGcw+uzLTL/7OInzTJHWSmRBzw82PIykllonkeJJq84FjDt8peifWZJHzKWNJ4xjGIw42t3jr1ZfpVCtcvnyRVqvJ9TfedHkKFlqc3TjD5cuXi9wMnufhS4VUjjHobEfSMxZzKZ2MWyxAO3Gp2cxAVthdMvcl2WuWeZRzXFgyPbp021PgHAnCWKw1mMLNq3Dp9HJmpRBMpIzc/mCMKdzJIpMKrBR4wjvUH5gwJ5NPYAlWm8LuAW6MkiRxDCXra7VadRKF7yOUpD+MszyTy/j7XX74h36I2/fuc+vOPTw/oD+OkMqnWa8RpS6w7fbt2/wv/pO/xD/45f+WF77vU46Z2YnRtFiE772z622TeA+6MZvmYB59oJjDPC5ZFtfKOynkRkyRbZWH2zqqndJBh47Jj/OEQFiLTlN8JanWKjzY3uITzz5Lsr/L1TMb9A522E8SVrOApaeffpqFdoeVlZXCJSilzKDfTo+VclqclJnI6Bbk4Z3KLcRpxpZ7WHIJId9tCw/GbFYgVDZMeUBPaZwd3AGpFEqVXWuiaEcIl8DPBTUxdR3MhGlIoTBHzOpc2gHwvAlSUXgTZ74xBg0Iz0MBomT4E0KgtAYpaGUTP8lct77v088S4Lz67TeIdUKawdejfkKapgRBlf/3P/pl6rU6t67f5pIfUqnXEX7OAPMxLi2oEuNATD+DMs2zA5VzkcyqHfkY514orXWBVfCkcs/EAvLweafd3E5z3EOdz+EkOkl3ypO5ilOM13EW4ePO8ZGMhwPuXL/Jt19+hX/9m7/BhTMbvP7tV6nXqnS7Xay1PPLII1y5coWVlRUajUZhjJxcO1/8LiVdnrtCSR9PBQgURueW8kmuiXIuiaNegV9BSX+Sn8LK4rMUnpMejngded++n0kD0qk7QrkFIlThNVFKFQwQQEm/6Ef+WaBc6rYj+o5QCOk53Ep2HWMFJou+DYKgkBrCaoVKpUK1EhL4Hr6nWFtZYHllkScffRQlJLtbD/jkx5/nzp1dB+haXiAMQ7TWPPPU06RxwvXr1xn2BgwGfcgwI6fV0087b95pW/J0a/89oYebOQiRJWvJ9W1ZJG+ZfeU0xRGVdS959PFlOmTEKyXPsEa4LADGhRLbDD9gtEZpTTLsI+OIjdUl+gcuWGlra4t6rcb6+jpPPPEEtUqVWq1WXKvw82f/rJx4InI0ofAUNsugrEreguMmbL7j5As0/7vMjGbfZ8emQDtOte/sDG6tu3chDUI4m4jERwnXZyOcJIEULt5ESWcAlBapJuOd32+5v8oTeL5EKhBWghEIK5EIPCWohD6V0J9IJDkWQ7oAPSEVnh/gByEaS60SYk3KYqvBT/7Y5/j1f/5veOSRdbr7+1y5eJGLFy4QeIov/v7v0e/uEyqPu3duEgYeSdQjiYaur9luPncuZYliyjlD50/pyXllYNg8Kqc6LL4TFiucQdeayRwtP/Oj5ni5fME8T9csPeRqhdtJD92ntZS9EEcvdnPs73OveJxhyQiU9B1TiCOnZ6eGuucz7vZ57NIlvv3y13n2I48jfcnug22efPJJLly4QKVScbEMHGZCuctwdkEeeshTQK+JlbqcTPQkC/dpDFjl6wrhwqxnR2WWIed2DABjZGHEzZ+SxSLsxDcopZqaqNP9mp+T003m0nNHFdex0iBIpwbJ8zxCKUgTw8LCAs1Wh9u3b/MzP/1ZHnSH6OvXefP1V4lTw/LSggOS1eocDPskwzHRoI9f8QDniQmCgBRz4vjJWQDeOxASjlq8WmsXFFdK+2atxZrM25PZdo7ypsyqpscxiIecOcChUNTvKkmsxYn7vsFTkvFel3t377K3s01zcYHLF84zHA65dfMt/szP/AcuyjGDNh/lVs3BRjB/ARcLlQkeoEyzDGL2t1l9dFbHnWtveRs0i10oe1UKwJp1btApe0pJ355nHJNSTgU+lnX77AjyVShwRlIhMgOrEU6qy9L9eZ5LjLOyukR/PEarXXoHbbb3uoS+YtTr0e8d0AAC5TEc9Xlw/x473T1WN85SqbVYXV6EIOQ4ejveg+PoqEWrlDoCui0ht7FhkfKwixomhu7TPOuHaeXNJ2EOvU47iQ3zc+odeamZdg/toMJkE9AilUCahMAmkCastZsc7O+gTUL3YI+f+ZM/zebWfZZWFlldXSYPICuLlbnOXAbNlMX68rXnYRvmGWfnMZV53x81hmWRN+/DcceX+zkPM5HbfCRiSg2Yx7DmkSdAYZHWuByfRiLwsXiFumWcqQNfSHKjZnEdFAiNVE48r9earK2uUvMky0ttDnZ2MGlCpRKwvrKCMWCjCOKU+7fvMdg94PyZs1meiqbDswgxKXjE9DOdyxiOAB3MU1GOeuULX2fVzXIVYfL7dGbI2X7MBnPN+zxLDz1zOM6+8N3oSyEGW7DGEAQBOw82scbgZy7BS+cvUAlCVlZWuHjx4tx2gCIganYRzu7m5QWVvzzPKyQGOIzKK0+go+4FJuLrrG2ijJY8qo2yWFt+LzOy/DWbZ6Pcj9xwWXZZzjuu3M6s3aSMsSjfV/7K/65Vq7QbTdbX1mhVq3zv93wKm6SM+gOi8ZjeQZeVlSUunjtLs1qhUa3y0le/wuriAlFvUNgVyoCyk+i9kCRyklK5V+YOngVBHXWt2Q3o3w/J4SGlOI65feMmv/Ubv4GOE3QUUQ0DlhYWefrppxkNBix2FqjX68U55QWe6+izksK8hTWP8iAnYKrNchs5HWW4zA2A8/T+MiOcJxHM++7IYLQTJqLWuoj3mJ28s30oX7N8vfI45vdWPq7MJBqNBovtDosLC1QDn8cevYrvedhUs9RZoN/vc/f2HZLhmHQU0apUuXfrJv39PeI4PvZevtPkbArOQP+d3iw/cMzB2gxGW3J4lVPLz5IRTs+afeV0PCctJfIUBqxGGg1JjNApK+0WH7lwkcV6hcHBLq1GgyQdsr2zCcpy/vxZlDQI6cRaRIqQ2r1KevmUmGdcWjpfKjwhEcbliHD3lntrFFpbJl6co1Wio0TIfKEdVadh1lg4r/15zOA4VeQo8bt8jpQSD8+pD1oghUBZsEnqXjOMopAmsrRonpAoFNJm+BFhILN1SGWQKgXGrK50OLexQTQe0qpXMHpI4FtXWcvzGA2H6CRFGcVgu8vvff63ieMx0rjKYLmmcEiNUtKllyu9hJJveyHPjqFQEpdXX2SZbdMj27TWFuqWzTw4GltgTFyBpcn6OYo+cMzhu035w/AFjLt9KoHHY1evcv7sOQ72djl35gzWWs6cOVMUmJmvR+pDona+++dx+vMWbr4D1mq1KYPeSe6zoySA09xvrsK8l+R53iHpJpcg8utK6fI79Pt9Dg4OuH//vksSM+debIGuOnlK5yrM6soKly5cZH9vj7PnNqjWKlSr1SLR7Gg0Ik4juoMu58+f5f7d26RxgrSQHlOi8b1WhY+2Q3xn6QPDHE4ajPdjwPLoQyEEcRyTDocMel2+/eor1AKfM6ur9Hb3sdoQVitz9exCfCdFm2hKgsh38lmmULYfaK2n+lFWIaZQj+VdZ+bzUVJDuR/zJIZ57cz77TQ0tbCzvkxcoabIsZnHUVQqFZdHYobR5WMjxExBodz/n0G856lOvq94/NFHuHzxAsl4xGKnzde/+mU6nQ7j8dhJMULgAcPdPXbv3idUEp3GnCY48ih70XGv2XPfDyZwFH1gmIPNsA3lSTs7gHnuwu8Yl7VuISbjiOVWB5uk1KoV1pY6pCZBhj54hla7xtrqIkIaPE8hxOGJopRXGJVm77PMBGAakDQfmHSYeR4nJczT4Q/ZD4RAaQvaYLU5xAhmn8O8Phw5jKVFXV60E2MoWOve4zjG933q9TqdpcWjryFiEOnUS0hdeLcKHEmmfiph0dZgjWZ5dYkzq2uk44hoaOnuPmB9eRlprIuvsA4AF4YhL33tK/R6B3hSFRmhXP6Q+XlGymN2HCM4jt7N7/PsTaeVID8wzCGn7yYndXYHy3gw4OVvvETvYI96pUq/3y/E3dFoxKVLl4q+nqR/w/TDStN0KjfCoS7MtDWru5823n/2+kfR21FF3quUZUKIIkV9GIauliBHoTbfzXUsQRDQqNdpt9u0221++Ic/jTGGwWDgrmMmzCtNXYUtAZhS1fbce5EXTf73hT4QzKFsxPpuMgdhLfFwiB4OqVertJsNSJ3uee7cOba2tvjsZz9bQKSP63N518zJWlt4HHJjW3HtGaZQVlmO8yocRbNi9qyaUL7uaWhWfD4tzbumlLKojCWkRHmeYxAniOO5tDWrXpX7V76GFBZfClYWF6lWqywvL3P9+vUiX2betkK4bOJSoqOEB3fu4SvPSSEuuuQwgzDWSVzlOpqzxsuSNDn73bzxfb/n/geCOTws5CtFxRM0qhVGvQOwltGwT7/f59atmzz55ONuZxHzQUTFJMhiNcoLFErRhRk4ap44OI+KmITsdRSJ0vtsv6YYxZxz30tf/VSfhJhreC2YZFYDI7962S35dtXGWamq7N1qNZqMx2N6vR5nzpzh3LkzeJ6aWiBp6mqEri+vYFKNshZlsnRyx/TjKPTqd5tOGruHHD7tkqNI6QPT8QOz5G70iFasxZuTsGV2V5/dPW0eRGMtGIMSLnNQFI0IFMTjEeiYZqOO50sGwwOutM9TqYTO1mAsSkncP4EgiyYUfpF+3enehyHIZVtAOedjfowpibvzxPm5NgJjpxiEFe4mjcgmuLFI69xfSZK4EvVzpBiYziZd7ve8cc3v86jJWI7LADKXW5buD4HwMhCTyXNQWPL8FYXto5Rluyw5uGHOrmmBUsZmKwXWk0hjadTqmKjP008+yc27dzFY6rWAXn8MOOCREoo0GlOrBpAmWA+kcUlncqnGzIlllWV2W0qvNzHA6iLvxVE0f5yPPt4rPa7UTJ+Xj693QtDHO5YchBDnhRC/LYR4RQjxshDiL2ffLwoh/o0Q4vXsfSH7Xggh/qYQ4poQ4iUhxMffzvVOUimOM/gcd/xpSAgHl1UI9CjCQzLs9rBaF5Wml5aWUAga1VqBUcj7PTv5Z9s+yU5grS0SnJRhs6e1L8wDEskSym8WHzLrqZg3Vkd5NI67z+No9tnN814kRqNLhr951y2rN8eNR/l8YwytRo3VpWU8JUBYlJDUKgHdgz2sSaiFzs0pLPz+F36Pr335K/R6vUNQ8HnXLXuGvhuq8VFr4iQ19N2oFSnwn1lrPwJ8GvhPhRAfAf4K8Hlr7aPA57O/AX4CeDR7/RLwt09zkbc7kO/Un3+a830NvYMD9rd38IRL0Z6mKRfOnsNXHitLyzQbDaQQYCew6Nx3D0zZFIrdreSJgMOGPWunIdZlqaJs+Z/nBj2Oygxilsqhv8e95o3bSWN/1AKaPT/HQpSlFIOdyjVxlL5+XD+KDNqlc2vVChcvXgCdsrayTFgJGPT6vPCJj+J5LuGKJzyqoSu0u7a6Sq/XIy0ZJmf7MW9BlqWddzNPD190JgbpFJQDpY6id8wcrLX3rLVfyT73gFeBs8DPAP8oO+wfAf9B9vlngF+2jv4Q6AghNk5zrbwQrhvIo8VTON7Ic5xUMf9BZW7C1OAZg/IEyWgEpFQqIdYmVMOAJI1oVGucWzsLQmQ5GAVCSKScTMR5E1dKWYjzwky+nxdcM2toO2pilRdZjomY/b2MFC3vtvMkhtNeq8zEjqJZieOoBZLbQQ7ZIywuwnOGMeX3mLeVF/YpSyGH27Mo4Wyd1lqqVZe6L/RCPCsIg4A33ngDIQSpjugP9sAYlpptbt54k5X2kruunX62s/c6fWMTaeWo3TwIgiONl1PfSVu8Zo+dpklA1mkZKLxHBkkhxCXgY8AfAWvW2nvZT/eBtezzWeBW6bTb2XenaX/q79kJNm8yztt9jxR5S5bm2faEmKAWtx9sYeIUZeFgdw9fSOIo4sKFc3QHXdrLHSqVqtNPSwlU54VYzwYZFdfMDII2S9I62//8lU/+MgMpHYw1xmEVsohGmyZYq3HRocfbKPK/Z2M85jHYMjOeDf6aved51zpJKrFigiXIP1smO175/nMPg5TyUGBaTvMkIgfBFvi+x/n1NYYHu9SqIc1anZWVFYAin4O1TuXwpeLNN685HIU8PHZHzc+TpKt5toXZz0dJJMcx99NIf7P0rpmDEKIB/Crwv7bWdmdutCwBnra9XxJCvCiEeHE8Gr3b7r0nZIwhiiJGvT5ogxKS1aVlrDYEQjEaDFlc6iACVSz83NZwGvdevoMaYwoXmOTt2USmFpx1xX2sNgXoZ/ZVpnmTVUpZuAZnAUvvmSh8CsonUOHSFKLAPcz2J1cXZhfmSWPvciQYBIZOq8kTjz+CtGaSJFe4tHhRFDGKItCGZDSm1Wq952OhlCoqluX3cZwH6jtJ74o5CCF8HGP476y1/yz7ejNXF7L3B9n3d4DzpdPPZd9NkbX271hrP2Gt/USlWj0UKJW1O/9m7NGcOdeviiCUknvp6AfsRFsPqPoBgefhCY2wKTodU69V8HxBNfR46oknXICUnAB2yuL6zD1OXVPjPCLCTMTenFGUF/RJO0PRfhbw4wk5Jb6W+1NeQHlgl7WT4JyyepMkSSG257vxceHg5fNn+1t+L9O838r9ze0PnucVAUPltnPJZRZVOjtO5b+lFWAyNKV1cOrUxuzt7RHUQ/A0vpTUw4A0Tlx901qNYTQmiiI8KYjGydQ4yDkeAGvtVBq542ie7aVsszoNw5s3jhOajuT9jkgOwrX694BXrbX/demnfwH8hezzXwB+rfT9/1Q4+jRwUFI/3jPyhCxes5Gb5e/y+hDH00QkHfZ7xOMxSrgdNR6N6e7t4XmStbU19vb2isKws7EO+WKcp4/O0y1z63z5wR2VF+GkhzybUzDvX/79SZQvyPJ4BUFw4nmz/T6tFDRL5b7nTGneop/X/jzmlUtos8dL5fpaqVTwPIUvQMcuQ/XGxkbhJt3b2ytwKIPBgIWFhXd8b99pOg0TOY7eDc7h+4D/CPiGEOJr2Xf/R+D/AvxTIcR/DNwA/lz2228CPwlcA4bAXzzNRQ5zwKMNkeDEaQBjzdTuJbNd1JrDeQsKN/jsZBIGsFg0e3s7iChCYRhGQ0JhaTbrbKytUK9PdNO8nfIuVTaOgcVop/fnD8+VnRBgLPFoWOyUCou2pb7NFKqZlXymxgpQyqW101kF67KYau30/c7Ti2fHpLxIc2aRQ73L7cPEE1BmcvO8MUftcOW28j7k15wtQSClxNj5UuAhlUu49P+uffcZbbHW2Q4qlYBLVy6yeW+LUZywudulaaHT6bDfPaDVahGnKeMk4uDggCiKXLEf73Do/LtROcqbzDulYg4yGYe306d3zBystf+Oo1Yq/Mic4y3wn77T6+XRi+7e5nD+fLfOhCGXy3RSwUlb43orBRaJNL7rvAVEXLSRp+ESQrg6j0AgYa3dYndvl+2tTVY7TWphgKdTOq02oedTCyuFdIKxSG8i5hZMwBhMVndTybBYAJ67QcgqXuXitDEGobxitzTZw82DkQp7hpm/e6bWAgIjJAiJyOpK6CzP4PT4nqRiTWh24s4zrE4bSfNkqBMmVk5UU263/F3OTOf9nrdRMCWm+3+USzcvjmusRZeYCxikVUhfkqRD0sRQqzao1SJ2d3ep1lp0Wk16vQGh5xN6PvFgwNbdW2xcvoq1JjOWHr9THzIe52Sm790cddwpaUpSzP421k6lnT3pWT/k8OmSnki5rsN70LI4nDGp/L27piNrLXfv3mVzc5OrV68WiyNJEjzPmxIthRAoKac9INlLIvCVhycVaRIhMEgHxs8YmSlyOZTF3zKISggXHXiSWjAR5SW+H84dt9k25tkRyraK465TXrBHkbWCPNqyXHJgUrNj9vi3j5fIzzvJHpKPb37POVN241vB2ARhNSvLy5w9d4ZxNGTY7xaFfAFMlBD6AejT4QqO6sc7kTBOwzjKtqVCymKChv33ItmLE82NK+T6HjGI0xp0fKnAGHa2t3n++ed5/fXXiwQka2tr2WTyUUpgrXaF4ZzM7qpVQeGeJHMxGq3BaqSwmeG9nBx04qMvL97jDJz570e5CGd17HkqQ9l1OW8cCikny5Q9yaMwLbWUbSKzNoH8Nc9uctT4l/+eZ6ybvU7573n3mKbplJo3axeKoog40QQetOpVfCnAaJYWOmysr7G6ujq5DyHo7u5hknTqGu+Gyv0p/z3vuJOojMqEk200s/SQx1ZMyPMn9oZ5A5NvWFJOBsFaWXw3EWMndRnz4ipHOVulBD+QbN28z/rSCm+98SbrWeZibMy582dY6rSKhen7Pqjy4rXFyy26SX5DpZTz11tbAGOQgsTGSN9D+q6KtLAGkzGa8m0LIRBWZjES0+Ce3A05O8nK86Fs2S9bxWeZT55Uxkkvbtd34K6j7AeuMrkr5Ev29zQzKntK8sk/HA6oVCrFONisPVPKa+FnrszUTvqdxzLIEnPKpQLH8Cbelfw5zZs/nvDQMqJVq+Fv+Ng05vatLXZ3dvGDCt2DfZJUs37mLN3uPmTP48Zbb/LE088QYTDWqaH6iHV3Gu9OMZZHMOnTtAVM1JTsv3yTyDeu09AHQnL4ztL8HIrg7BRJklBvVPF9n1qthrAwHA5pNRoIIWi1WnPPBaZSns0Tt4Vw1aPL3DwMw0JtQKiCb1k40d+d7/xl19dpqLzjzuIZZhfx290ZT3tOtVotJKPZe87rir5dmpWCjnoG0welKAGh71Othiw0Gkhhqdfr1Go1egfd4n7SNKXZbDIcDt+VfeBhpQ8EcziN1faodGlv9zqz4nm+WAaDAcPBgDSKiccR8ThieXGJu7duT7n6ZvX2o/pxVB9nxe+ybaT82yx6sbzjH1USbdbOchTqsdyPebiIWSml7Gad96zmicizXpdZwNhxodnz+ly2H8RxPK1nz7g+5z2bXI1TSmF1ykK7xZn1Vaq1CgpLo1Yh8D1Go9HU+KVpyp07d0iSo3NKHkVHqV3fCUYzT7U4iT4QzOE7S4fFt6lFYsGkmkrgdnRPKjwpefTRR4lGYzY2NiZFZecM+FHBUIcNrBNvy3HpxubZBTiCEcwTRY+bFEctPinl29q5ZxnL7GI/meTM+zFHlhZqTjnepHyvJ107Z1ZSSipBkDF7h5h050Il8KjVaoW3KDdinl3fIAzD9ywT1sNCD7XNQQiBL7NJWVbkTsnS3I7kY0xKPi+m425yKWEyGa22eNLPEoxorE2x2uArhRSCJB6z3GnjSai3qgTVKr4f4iotubqe1roS9Yc8vfluJgXWyAzbIBBFLsmsWjUTZSdnIvMMjuWSjPm7sNNXnd4dBWk6W1fT9TenST0N5fzBwmLtZNzyvszSPDzE7AItSyCzWIUMJVA8h9w1adyw5RdxNTeLJnPVx7kQpcj6Lk1mH5mu/jSrJuX5LVRWZVMpVxNTegHGDGg2Grz++lu0mg2+9cZb1JotwsBnZWmJKNHgKSJtWOwscJCMQUqHMDWH80nK7O4MTPVh3pi5G5/F3BwacGdvyo2pYrodyvammVNzF/9JwsMHRnKYJ0Iel6nZkau1WM7wfJrdqwxhnhjpnGrRyGwNzWazMELOivOzC7ms7pQX5ux9HdefWTvCrOW+cM9hnT9ba0ypzkMZpJQvzqPiJco2hvI9nBbjP0/tmXdM+VrlOI5yct15dFyfAaxRCHyw3iHbW3m8Z59LPsZaa8IwJIqiomLZxsaGS1Ufx5PclkAURdy6devUfc4/z0sTWO7fadWM06og8xjRSfSBYQ7vnN4dkEQIl1h0PB6TpBHj8ZB6o+YqOIdhUevyqIf4blFu74RSYQvAVNmOUnbZfad023dKs9GSb99mJCevLPu0VPMNkDmTK4eyG2OmKp37vk+j0QCgXq8XG0KSJMXzXl5eRinFt7/9be7evftdC5A6ik5SI0+ih1qtyMlaDSV4bFm0zLnw5FhbGJekBBfTVN51c/nYONSggFyqnt2dPVfkna3tbfwkQSKohiGNRoNRPGKjvkKYJz9lGp1X3jXzeZ4rGsJOxMv8vPzzbBVkYaZ3m0L0PYbpeEKSWo2acVfOc+XlqkVZ2nH9kSjlkWfXO62vvYypKJCdZhrKPktCCFJEwceNzc4rVWnKRWfr/L9TzA4MxrgCx65jrhyAsa6eRVHE2IJEklqDtaCUR6o1Odgu9/QoC0bjpEM8RsOIpWYbEstYxcTjEUYqDrYGBEHA1SeeYOnCOmNjwE5Uvql7LL0XaE5x+l08p7IL+TgGmo+7MQbDYfUOPvAIyQkdJ2IdJWa+WwqCgCiK2NjYoFqtonWC70ta9RrLC4suCIujy8XNu4d5dJybcPa3MgM6SozNmWNZZM/pSJfqzNjNUy3erjU9n8hl1OfbscofpTqVfy9dDSEUs7fnzhFOtZQKkxWhlcKbc9x0+0mSMBwOAac+5AFXlUqFJEnwfZ/RaMQrr7xSlCbI7/udeMveDb0dNWTWc3EUPfSSg+v8xPImhEBJVWSHmieCvlfici5CDgYDrLWsLLWRwtJsVPDtxP1WHux8Yc7rz+yDmf3OmFyaEUW1J6+0+8Nhbn/UYhelz2XdOn+f0tFLUkX5OlIKcqNfmeYt0PKYl3f13L5RllqOk0KsdRbVsuEyF+NzBlPGj7hnhMvmjQNowUT/V9Kfki6d3GlQKAwusC4HqeXPzmQ1M8yMHQKyAsbSpQgcpqkrm5emhGFIHEWImQLJx9Fxx51GHZ06d87mVNhW3uF6eOiZA0wbFIHM0KQzMVIihCG3uLtJL8g9B+DE5rKoK4Qo6g04moieudU3t4JL6TEajkn6Xbx4QKXTwViFCD1U4KOx+KXM1lIqhCjbIITzigiKqSilzNSd6Z1GCMFoPAac1OIm5CSZbGpc9KAxBmGNy0HgeRMrfbHgLQiBQWCNndy/mFxvHqR6OnWZwdiEvMpXfszkmUzvUuXfppiBKaVEy5h8OTI2P85TLjAoLaEZ01Rj5cTuky/QMjO1VqO1QQi/GHdjJkFvOTq2bGtBSlJlsCisSbHGqRauTZdsRwjwrODM+gYH+z3ASQ/S94hGY0ap5sIjj7C5s8uZzjKjaIwnA8BJk5LZTeuwATqvewEUSM/y5lIe2xPdpDPeDSMpkLe5V6Pc5mnooWcOxU29TcOiG4Tyjv72fNC5bSNH6G0fHHBxpUO73abZbFKtVt2k9rwpX5Fl8mfhzioxnNxeUM5LYK3D/BtjCMNwalLlqc/CMMRkkGh3HQs2N6pNvBJF0Vs7uWauA1vrvCvzkqKWJ18eZn2UgS23fRxFZWkl+wKgkIYsprDDxLmBTzqAkRUT+LfWGuGpYrzyoKdp5pCFWwtDlBUYKoOeZnM1lj0/8yQYIQRSKVLjxkBrzblz5zgYjIgTzc3N2/h+gFIet27dQgZhUZ3LZNGe78IGfuR4nmTsfLuo2NPQQ88ccjrJiDLP4JLTPFvEUaqIZXoHFNayvLzMg+uKNE2p1+uF7nmSXplP3MKIpAErSeLJ5IyiCJmJqYPBgME4KvqjlMqi5gSDQZ80E1+LieLkaJIkKlyNaRpP2QzcsZkZTMgi/8Ks1DDv8yyVo0ULxjiHlCpLMhpb5NaYLOby+ORxINZa0pJLE8CmgsFgQKvVIoqignmmaVowOt9XjEZjbKoPVe/2vIndI48TmZIimI4z0VojSkZcz/NJ0wFRFBUQ+iiKUYGitbDA5s4u+uCA3iuvcOHSVSrVWnHteRLXO6UTjYczakhqjjdan4YebuZQFqfy+3QyUmng9aHJLUQuSk+SvFo7GyQEBjcplHD+cAd6EqVEs5ZqNeRLv/cVOmFANB7zyOXz1BtutygwDmWbA4DIAoKsRGb5I3IdvmwkzCf4cDgkSRJn5PK84nujU5IisYotUHij0YharYZSiv6w58KGcRM7jmM8kZWZlxJ0iihcrRT9TNCIUgTpPKbpxtJMAaDy91n7SVlFKO/OUpb0dTPZ3ZLIlbnPJanRaIQfhOgkwRq3e0vhkQpNGifs7O7hScUgGaKNW+TROEZ5kihKwLpw+DROiaIxtVoN3/eJ47hgZIVKWXI9G5ui9cT24Gws+dxxamsQeqxuLPPiS990KfWUT71SZ2drF08pzpxf48zVRxhFSZYjYtoLNc8udhp6N0zluOjanHmf1P7DzRy+y5SL/lEUIRshjaDKcDhkYaHlcjZwWIIUAFZkLkvhXgKw8lBmV2NMYew8ODggjmOWl5dJ05SDgwNqtRqj0QghBLVajTQd4fs+vV5vaufOE5JCBvISbkEmSVJIGUqpYhd1hk/X2Vz5mizmWZemmOovTAOXykbDPL6gvPOXpYJ8wuZqTfn44XBIaCy9fg8p3EJuNBoMYleirtFo0O92M5VGZOOiipwaa2srbG1tZSpYQLffo91uo7UmiqIitV0OXgJcWveZhWuti4A1ueqiLaEf0Gm1WWx3+Pb9N6g22uwf7OJX6gzGY5IkYWdnh0qtgZBqSs38INMHhjnkk63s553nMsrhw/MSeR5nHc5tDFPLXRi0SVhdXUZGvcLWUK1W8aRCSeUMT6VTFIAQWKPwMui3QJDYlDR2i2kwGBQBPEEQsL29Tb1ep1qtcv36dZIkodlscufOHdrtBcIwpN/fZjQa0e/3OXfuHLdv33ZxAJWgSA4rpaTdbpKMhlQqlcxOkUy8B/iFCiNUrpUIbKk6V76Y8512dgealRby65Y9CGXVwxkMnXSQp8DTWmOEJI7jAnEIsL29jbYGK5za9u1rryOkUxMO9nuMBmOUUgxHfZaWFhiNB0gpqVYl1958A0kWWt5z0lV1Z5tnn/oIw+GQg4MD51GIYxYWFojjmEktB2eIzPtf/BMghUSiEKnhwvoZNje3iLQlikb4Xsj58+fRmXRXSI5z6CS1GA6XBTwNFevCU1N/H32CxVh97FrI6QPDHL4bpLUmimNEli+h0WjQbDYLdWIec7IWpBZoa4iiJFuIgmFvyDiK6Ha7LC4uMhgMGI/HDIdDlpaW6Ha7jEYjkiQhjmPu3buHtYIbN24X0FzPUwwGA65duwY4kE4QeCwvLJKmKSsrK9y5eYOFZoPFlaXCc5EbzALfw9gUawPSpJRNWkxHdeZBVmXvxzw7zpT7ceY9ZxZRNHZQ5OGIfr9boAw1jkl0u1263S6ptoxGY4bjEZs7u8VCiZKJCxFNxlDGrK2t0Wq1UIFP/8YNglDS6zkVq9lqoLXmxo0bvPn6NT7xiY9z9+5dOp0OSil2dnZc0lj/ZFyABLQxNOp1eoMxly9f5stf/ybWWsbjMTdv3uTSU4+/J3aFh40+cMwhFyuNEdnnbCLnz0aWMAHaIETisi/NyTuZLxpbKtBaJMK3YISPlB6+5zEY9BmPnS57NAkMHggn2o/HY0bRkGEc8dWvfpUnH3uctbU1XnzxRVZWnBhcr9fZ2tpiOByyv7/v9Geg3+/TaDTY29sr7AxaayqVCu12G4But8v29gMqfoDC1VaoV2sMlxa4desOa2c26HRatNttxuMIz/Op+BXGUR+QheSipMLKaZ9+2Zg2i3IE5qgbJjM8UkgMSZIgrWHY6+IpxaDfZ3dnhyRNaXbavHbtDZaXl9ndO2B7e5vBaEycanZ39iexH76LY9jf32dvZ9+5eE3CW7dv4nke7XabxcVFOo06+7v7nDt3hiRO2d7eJggCxibmC7//BywtLZFm3Y+iiFgbVpcWqTeqJIlGiNwupLBpirBOKtTGuYW11lTDEGNSqtUq/fGYRMd4fpVonLDeaDCO08x9eDQA7jQSxKFzrcWKaTtG7qaU8vSQbaun63V+sG0OJZdYWWw9PRkoYuEOU47c86SP1hMXVHkH7Ha7xbVzUXtWdJtnmDPGMB6PuXPvLv3xiEcffZSbN2/yta99jQsXLvDaa6+xuLhIv9/nwYMHWGvp9/sMh+PCVZjvwHkyEasNo8GQ7v4BlUrFGTMHA3rJAUudBWe/0IZhv8vq6ipfffHLPP6RxxmNRrRaLe7fv0+tVmNxcZEkmWR4MqkBNV3nYdpgdTghzuw9565GmMZRWGPY291lPBgSRRHtxQW+8Y1vwB3F0soqN27cYGt7l9RY9g66YCWphV5/gOd5JOMxg8GIra0tlpZWHIMUIVonpFpz7949RqMR235Aq1Hj9t17BIHLr/Fge8sV+FGK8SgijmMef/xxbt++QxhUuPnWdT7+8Y+6cSierTNgKinROltEJncB9xgMBrTbbXa7XVRYYzAYEMfxlHt0dsnNekdOQ1PHHsGUv9P0cDOHU1IZEfheSndhGBIGAbvZrnnu3DmAY92Yuf48Go0YDAZcv36d2Gg2NzcZ9vpcvHiRnZ0dFhYW6PV6hd49Go3Y3NxEKZf1qN/vs7e3R7/fZ2FhgWazSW2hyquvvsry8jLLy8u88sortDtNgmqVXq9Hp9NxngitefDgAfVqjc3NTaSUtFotqoET7+/0b7G4skq9XneTWqhjB65sdT+O8riPHJthjGE4GDDqD/Ck4mBvn/v37zPqD1haW+X27dt8+ctfZml5ld1uH6Gcd6FSrXPm7Fm63S42cqpXu91G514kJL4vGfWdHWE8HhONXV7PwJMEgYfWmkajQafVxhhDt9ul0+nw8ssvc/bsWb72ta+xsbHBN155leeee8YZjK3DmkxiUibMXkpJo9HA8zzSdFSolIuLi1QqFaIowgqF9D4YEQmnkVw+UMwhjxfQKYDErUPpREJzGNjjUqKDtYJZC6XbJZ04rG2MLVUmzi34cZxQr9ZIEycF9Af7KOV2W0/IObuBRUiNNgm9/h47Ozt0u12CMARtOLdxBptqdre2GY/H7O/vAxDHDj/RqDY4ODggjeLCHdiq15A2ZfvubQaDEY899hjD4ZD7d96iUfERaUqj2eD+3gGB56PDlGa9itEJWM3dmzepeB6bd+7w0Wefw2jNwsIC4/EQKcnsJym+9JFWlCovuwwEQuD8vvmLHOJc+EWd688qTMlVmaYxSRqxvfsApKU/6JOQ0F5sE6UJd+/c4d7mXdYWF+n3R86QbCy+VPT3D9i8c5fFxUVSbRkcOCni3LklknTkwFLGVaJKksR5k4RHWPHBC/E9Rb1W5eDggMHBAdvbuzz33DPcvP4WFy6c486t61SrVe5v3uONP3iDKIr46Ec/Si10MRMpCQidSZIKaYVTv9KUs2fWuXn7PgrBYDAkNRZVrbF4dh0R+kfIqEzGjcMuznwOlRMLT503jykbW6gWR7klC3dtKa5lSjXkeJXkA8EcZhF57qbzxf3O2yx7PubpYPmu1Gw2GW5mE6QUWjz7UMoP+Xd/93cZdHtUag12trdJ05Ttzfv0+32EEIzHrtaiMQbf97l7927hkgzDsND7o+GIRMLK8jKVsMv9u5vcuL7F5SsrrK2cpdvdQ0qP3d0+QVCh2+2yv6O5eOEc/e4evoCldovXXnuN4cjZMYxNCTwfjEbiIaRE60xfFl42QV2F8PKCn73PeWOaj0kQBPT6B3iex+a9+/T3D6g3G9hU8+br17hz5w5XHr2CMdA9GKGEZX93h1inYF2p+2F/QKJjAg8gxZMGP6zQ3dunUg0YxwmLCwv4fki1VuPGzbeI45jxqF8UvvWlotGocevWLVotNw6e5xEEAQf9AVevXuXatWt0u11eeP4TeJ7ED0qLigQpHBRNKotOYjbWVugPRywuLrLd7RbP7ShvRT4mZW/ESWrGaW0SR3knjjKYT6srx7f9gZCB3mmcfA7BfaeUpime5zEcDgvQEcwPl819/tVqlZdeeomzZ8/y5JNPsLTQ5rlnnnWTvdfnkUuXCwPia6++zngwpNvtFpGfxhj29vbY29tjc3ObwWCApxR37txxCU1bIU8+tcHFS+cBS3/QxQ8Ejz9xkUazAiZmeWmBjz/3LJ4n0WnMN176Gp12k1u3btDt7mOtxqQJSoDEIqyrIJqmCcZOUIT5q+zJgMMTN/87n4y5F8L3fZrNJp1OxyEcR2OGgwEXz1/g0qVL7O/sEkURCwsLpElMp92kXqkSegaTDAg9U9T36LSbCFJe+/Zr3LvbZaHdIopG7O93eeXlN3jxxW9w714fYwyXL19mYWGBtbU1Op0Onufh+z4PHjxgNBphjCkwIOMMp3BwcOBUntGouAe38Byy0xiHxAw8P0v0owu4e65GHUc5s3m/qOwxeqf08EsO1pVuEyUvhPucp2rPU7IdJufVyB/wYZoF9RSXzN6l9dBC85kf/FFe/Df/HKWcC9OQp55z/v28BqcnJK++9E36ewd0u12GY2d3OGstrUaFhdZFxuPMtTce89yzT/Hiiy+zvtFhd+s+0XBIRUguXrqAUopRHLG5dR+lBBfPnWU46HPx7FniOCYMA/a3H/DYlUsIJfFVjb2dbT79/HPcvXmDa99+mbXlDh95+vt56aWXCAKPZr1K6DuItR9WCuCS9BT4KpMcpplA2UiZj9kso3DuUFsUKdbW4AlJYjXxcES73iAe9NncPKDTWSROnNTUaDSKZLAVIVhdXMAuWPrjiK2tLc6ur9PuHfDcc8/xxS9+kZ3NET/xJ37QeXBsymc//QLXr1/nM5/6GJV6g27f5VdAx3zfp15ACMHe3h5LS0vcv3uPRKccdLvcvn2bWq3B1u4OvV6P3d1dXnjhBV574xrnonNcrpwl9PxiJhhcGQNlINEx1956kyhK8DyHkv3o88+zu7fH0sYGJgs0K3I1zO7cJU+GmAzi5LsSqOxEMi7Ayxwxv8ttmVQjLMfmJ52lh585fBfJLf6YF198ccoinVMuLubJSXq9Hn/8x3/MYDAoDFX9fp833niDOI7p94YMh0OefPJJ1ldWWVxcZP0nVllZd/U2d3d3qQZV9vf3efTRR9Fa0+sdODQgkI6HSCm5desWtaDGx575CM1mk95w5DJTeU9w984tLlw8l3kOLHdu3UAnEQQetUqIUiILA09JdUwYVItkIMrzUHngVkZlQ2T+uTwOZaaaeywkgiRNAUOz2WRna5swDFlaWsrEfZ/d3V23A5uU/YMDhDUoAa12m6XAY2W5w9raGltbW9y/d4ePPvcMi4ttpKe4fG7D2WNaLR67ctmpnWGFS+cv4AcBo6HDjIRhyNJCh2G/x/mL59jb26PTaTEejahW6qjAp9vtsrS0xIsvvsjq6irj8ZgnP/II8WiM7x9eHu12mytXrnDQ/SZp9vx/67d+ix/96Z9iPB4jlJ+pvCdTWcX4bmQMO4kebuZgy1GZ7x/liyFNU6QwVCoVejOiY9nmYHHeiV/5lV+hElTodDq89dZbaK3dhBuOqFWqXDh3kaWlJfr9PmdWHIhneXmZRMd0u13OrK3jq4CNtXVncxCWTrtFs1Yj8DykcAartdVlTJZ8xBiD11SEnnLHP/YoWrvgqnE0xGjB2lNP4/s+g1GPNIkZDYaE9RoSSxyN8DIpYjZV26yoHMfx1HFSSsbj8RQiEjJpwjjw1DgaF6jMVrNOv9+nEvqcPbOGMYZ2q84jVy5i9ATanKfei+OYzqVzRTBV4FcyJibw/QoIQ6Vdz3ZIhwWPoyFrnQ5po8FwOCSohiwvtomspnf7gHajxceee5Y4Tnn9xg2Wl5fpdrtcuHCBzc1NFhYW+NVf/VX+w5/+mal7d14dCmkk/87zfYLAGUb9ICDV2lU/mzOfirGZMUSexj15rEoi559fNkjm17al306ih5s5lGgKLq3zOPVJ943NobBiihPn58wmvLDYQ2rFISOQ1HhW0VppsvWKZmf7PlI8gTRJkf5NCIlEcO/OXUI/YDTus7y8xLlzZ5EqoNPpUG/sU6/X2VhfJRqOWF9dYqHtdGEhDcpKlhYXiaIIrzCSSlcUWKcgFFaCMDAejZHGZAFHY6zNXG9aO/OsMFQDB9SqVaogS3kZqZJqjVQapVOETvGUj1IaKUBagRQeNs+foWweiuT6JEAqic0qUiVpghG4Nq1zCHkGYhNjSSHVCFIqoSIZJdSUIgE6K0v0R8PC9WmtBR8gi7WILUIpPCwCS6gy3ISJHaOwEqGz8G1tUIFPxQuyZ+4DmsATVDpNhPSI0wQda568eAWki8eor9QZDAYc9LqsLi6w1+sShiG9/oCvffUm3/PCA86cWXeG2iwnBxlMerHd4V64zWjkPCWt9kIWOFZDSadiGijiM46iWaxI/t3chVvyTMxpaMoels/7Q+fLiZpjs7lyHH1gmEOxS1s7pbedlubFZpyW8mIn5RDl2Ye0tbXlgqZ0TK/X4/HHHydNnU56/twZut0uC602rfPnqIYBw+EY3/dddGfggFVhBVQWSeoCpwzK8wBXAq7c59yDY4zri0ldMV4hBNFoUOzsfiWETGUI/YBazSMIQ3w/zCDZR+urs7aYk8LUy/1TSjEejtBpjDCWNIoZR0OUJxgMe2Cc1CMzlzAiLaEHU6zzV2O0JbHGXTtNSLUzEit8jExdfIg1LhekdteV1tXztKkmwWToxgq1Wo3BaIhOYnoH+6wtL9FuN9nvHqCUYn9nl3aryec+9728+OKL/MAPfJZarYqTW7McHLh6oSbVhdFPZh6neWP4Tukkb8ZJCMfcQP5u6KFnDsXuXq53OecYOZPkdfb8o/6G+YkyrANIYIzFJEkhOhbt2AkacOvBA3Z2tzh//jyvvPIKly89wp07d7h88SLLy8suxHp1hUajwe72Fp12G6kUWruIQp1aRMb0TBaOLKULd05jtwOn1uWWgMzQZEvh7MZV67ZZDL+FAt+hB67v0hqsdDttmiSuLmUp49C8nWt29zFmOmozV2tyl6tJNSr7LfcUSeuiL8fjcZHNyTEe7XZjDMK6TMBS5BmfHCOMoohEJ2xtbbG1tUUcJ6yurmKtK0/XWVpgYXERvxLSqrYyaLVFC1tEaxoMvudhTUq/30V6Pq1Wy10nC21v9xs82Nzh0SuX0Rbub+8UsTQ6Y1I2T5eXZYpyqpVTpba3t4tyfuXw+MNYyfk0D2Q2K8meZEicndfvBZLyoWcOzEZKvtvW3paoBzDBWERRVBxXhg1vb2+zvb1NFEVcuXIFKRRXr17lINNNL1y4UEyslZUVwsDDWIE2CdZYPOEhstiGFOdm09qSJJEr755qjKDI05CkE3tDEkWOyUjlQpBnfdtCEMdxlqwkRAU+oZKMxgOU9Alr9QIuXL63+T7y6XyW5QkopUTbtMBn5OeOx2PSJCGNE6JojMgyfiuEE2tFSeIZR+4erGDz3j08z+f5T33SMTSjIQtoC/0A5Xuk1hCnCb3BgHv3Njk4OODSpUsgLc1mk/5gSOBXsFbjSx+hJFanGGuziHVn3DUCWnWHfuwPRwyHt3jhhRcc49aHwXVCiKmEsn4ldLEwzYUiJZsT24+dau+KTjuP8/6+E3rXzEG4BI8vAnestX9SCHEZ+MfAEvBl4D+y1sZCiBD4ZeB5YAf4OWvt9eMbB+WVA3xw9oZZdaqY2BJr40PNzBvI8k457zjHzRWQ0B+MCap+hibMEmXg8PZ5ZOVnvud7GQwGVKtVOp0OlUqF5U6bSqVCnBnl/EC5iM3EMRlPSqwwRCZL0uJ5GKuQKiROxggZcOPmm+xubdOsN2g1nBEu8HwCobh//z5SQLNWZ5hENJo1bGoLm4HMgsqEEK5aV5IwTmKGwyEqFCwtLJMkAcMoZXG5isUgPTV3YuUu4XLgUJlRONXAoLXDSkSjMUKnxFHEsNdnOBxi0pQgCBwDEy6ZDlYBNkso43T7K1cus7KyDlYiZB0hIcBldcozQSmlXKp6KVjzPJTw2D/YxZiUt956gxu3brGxsUGnpbBRCsq5bI0sJb0VAVcunGc4jtna3mPnwRYH/QH1Ssgbb7zBzs4WTz7+RBZGrvGUwM+wEdVqjcHAxawMDlyOSZcqLstnIUCo03kgTgrGKksVR83X2bbK9E7xFe+F5PCXgVeBvNz0/xX4b6y1/1gI8f8C/mPgb2fve9baq0KIn8+O+7m3cyGXmPV4LlgkiC29O710+rijWrFlzmNBCJeRKI41/d4Ya5y4LqTbkd944w201ly/fp0zZ85gUk3voEujVqdWD0mShDBwIJxkHCGEQQow1mU40tqSWonnObXl9p077O/vs7CwwLlz53j2mY86LAKQRGMwzkWYDsc0Gi2MTkjGEdeuvc5gMGBldSnLqkyGhwgnYmtmkLLWohNNd/+AhlWoap0kivHDPDdENt62NE7Zd7kaobXBZmn0yjlsrLXFrprnnxhmqFDHWAVaW7xAZQzG2RDq9Wahtw8GI8Kgmv3ujMu+7yNUljQXl2fBCpeQt1pVpDamnhWh+cjTT6GTlC/+/u9zD8u5c2epVCp4YYDyQgeKUgotDP1+H2strWYdzpzhQhiwtbvH5oN7rK6uZoxVFmpFjm0JgoAwTul3uyy2F6hWq4fnUq5+zbg2J6X+Tq92SHF8FnKnopUhE/MDt04yQpbpXTEHIcQ54KeA/xL43wjHtn4Y+PPZIf8I+D/jmMPPZJ8B/j/A3xJCCHtanOgJVLhrcsv6CUCSfELbqYk9XZITHCeuVGoIFGFYBSQu+7VzbT7xxBNce+116pWqm4BSuTwDWWJaJTKkmkkJA480dq4/nY7AOsv79bducnBwwJUrV7h69So2q+SSJAnjKAULGrBCot3WlEk1OoNbV7l8+QpCwIMHD5C+S+riske5Y/JUc5DZWHRKrAU92aOh/GK8ciZaMNOyK9mCJ506Y9IsCjHLKpU/xjwDVV7zIa/3AKCU08eDIEAJ5xK1aKrVMCtf53ZIpSaSoNGgpCIaJ0hPIGRSuFKN8PBUSJwYfM95j+I4RipL4Hl8+oVPUfEk//RX/jHPPfcc9UYDPzSoahVjDUZIfD/IPBGOQfT6Q/r9PhcuOCDa/fv3WdtYn1j57QQG7Xkeg909quF0GL+UEsRk8csjNqa8YM9pJIDcVnP0cpmflKd8Rffd6Zfbu5Uc/gbwvwea2d9LwL61NlfUbgNns89ngVtZJ1MhxEF2/Ha5QSHELwG/BFBvNdE6KcRANykAO9NtYWAmu/RJ8FErxbTfF+XUGAA8jIkxaFIt2TvY4d7OHutLDjbrC4NA4WVuMSFEkXxkPBwRhB7aJERRFr5kweqE1OgJJsAo7ty7x7Vr1/nYJz/B+fPnqdRqmS0i88rIoMg4ba1FA9pCnnJOIp0kg6ZSa5EkCRtnLxFFkUska2xmIJNgFUYnjrmgMNYw1oZUjpCVMQ3rPCICC1aTpCnKqsxYmD+brNKDcJmHXIizJU3SqbgKk1qSKHZjIz2QIFFIbQkrWUVy66DHlSBwOrzWgEQoSWQTJB5WCjAjhzfxsrqXGqS1xCWpA3BZtnC5PoQI0CbGChhFEb/wi3+B//6//+94+umnaSKJRkPq9TqVWh0TG7SAJImxQtLuNLkUKrr9Hl5Y58HOFitiza0pK1HSz5LVRgihWFhYwCgPlxrfLUJlIVVQ0ZJYgTazFaeOnJZTOIhsPUz9NmciuzeOmevvcP99x7AsIcSfBB5Ya7/8TtuYR9bav2Ot/YS19hOVasXtYMYipYeUHkWp+qktv1QncdLOSdc59vcyQOWRRx7hR37kRxwe34KWE+azv+8wDEEQZO5InzD08X0FwtVVSJIoSz3vqn3fvHmdf/U//hs6iwt85rOfpV5vUqlUMBoSbRklMZE2JNkiGGtNZAyRBis9hnHMSBtG2hAjSaVklKRYJMY4o2UYhlSrVZeoNstBkaYGZ5cR6NRmUoUskt3mTLi80MvjZVPtJIWZGgnW2sItm1Oe5dq14WpQKN/HIpHKJ6hUCSpVrFDEqUtim2rLOEqIYkMUa6LEMEpSdg96RLEhsZJU+aTK5/7eAcb3iYBUKYbGMkgsYy2JkgRts2K6ns/dB1t87nM/xre+9e0iDiKOY/r9IZYEITW+kvhKkkQxUglarSbGaB577LHMfa6cipmFb+/u7maZqCNWV1en7nneXJoHKpt3TN5+2ZX8HgnXb5veDWbz+4CfFkJcxxkgfxj4fwAdIYpaY+eAO9nnO8B5gOz3Ns4w+d6QfTe3cjwNBgPGUeRASxk+XUpJv98vjJDl9HFOjJ/gKdxLMxgMeOmll9jb2+NPfO5zTk9VEoGaqDdSYaWiPxyx3+vSHQ1QlYDeeIjwA/b7A6RfBc8n1pZIa7rDEamFKNVTkynX1Wu1WjbhFNYIfC90GY+sZTyaGHBzW0HZCHko4axxRgmrTSFi59cqB8jliV+EUIgs65S2Ai8M8cKAWr2J8gKSjDEMxyO0NWhtSBOXY9IiSVIL0uPWvfsMI7h58wFvvnmXr3zlFfb2Rty/v89v/dZvE6WWUZoySFLSBIbjMVGiibRB+T6D/ogf/dEf5e7du4VLtd/vZ8l6nYs6h3Dnbtg0TRmNRuQ4B5cmQBWp/fr9fpG4NsgkoH+f6B2rFdba/xz4zwGEED8I/G+ttb8ghPgV4M/iGMZfAH4tO+VfZH9/Mfv9355sbxBI4QbfaJOlEIc8VLs4XVlsYUTIznyHD8oFcWWZnxBYpdBJzEuvvcqiSYltimcESIsBlldX0Sbm69/8Bh996hm80EcajUk0CusqKtmYNEl4/eWXWVpf5fzFCySpxOJhkUTWMByOQSgMisE44qsvfcMxoN6Qra0dzp8/7+IkUku9ETIe7vH41cep+pKzG2cYRWOUgX40pt2oI7AEynNWc+GiRfd6faQUSE/hWYMmxq9UsNYZCvOis9JKhBWAxdi4cM9pLKlwKpgSAqNNYc8wOsFXHjEZMlAp8AKkgUD56DSl3ag7RoUgGY1JjSGsVtDGYKwlsYJhnJAag00j3rxxg5v3brL3YMj2fpfu2FCpS4JKSBSN2eoNOXv2PC9+85vEGBYX2jz11JMMBn0a9RCbxM5tqhRBrc727i6PP/44e3t7pNbQqjdIrKW7P8APKvihm2sNvwJSoCVIT+SzzXlXRB7R6XJpWGvp9XqZ/QSMcDkxPA3xpGLyZEaX3b9lW8CM/WF2/kqyuJYZ9XnWDjHPqzGlmryNtfCdwDn8H4B/LIT468BXgb+Xff/3gP9WCHEN2AV+/jtw7XdE0k6en7VkFl1TQCy0FfiVCjIakeg0mwRQbTgIbjQa8NRTT7noN5sWKkcUR/ieszIfHBywsbHBve1tzl+4RJy4mhBRkjLWkOgETcxL33ydf/P532VzN0YCQQijGL514x6eJ7BJSs0X6GjIjbce8Lkf/iF+4zd+i/XlJV742PNYGbPfHxAqyfLCIsMoplqtIqVkpVZzmaeyylCxsSgvwIIDRc2hPAozrzYlrSvlZ0pQdZdqTmfqiBONPc+lahM4dKKvfDqdNsNeH2udHcYZTFOGwwEH3R5ShSTG8u/++EsEtSqf+t7v4xOf+R7+zt/+e6xfPEtHW67fvkMcpQivytdff5Ovvfwa0Sjl3/7Olzmz3uTX/8W/5Sd/4nt47MpjVAKFsjGexVUUkorecMDCwgLdbtc9ewRWCJIsOlQoD52maGsQgYecKdzj+z7tdpvuMCKwkmGcFEWOTBSdCJk+imbjdo6i9wK/cFp6T5iDtfZ3gN/JPr8JvDDnmDHws+/0GkdZaqWUGG0KluhKox0etNkz890QKNLZOzLYzHAoVY6PFzzx1HNc+8PfIYpjNBYvE78rlQrNWs25An2mLPe+7xONhxwc7BH6Aa9/69t83w/8MN3eEMIqUZSijSCNx3z166/w5W+8zJs3t2k1K7RbLohHhj59O+Z+t4cxlooF38LZ9ZDQC9k+2OLn/uJf5NVvfINv377Jzbfe5Oqliyw2mly/fpPO4jLWjFlfX6ebDPDDgFqryX63B4Fzc7ZarcxdVs5OlNW61AadMQglvKxoSynRTmYAd+AmhZQOulyr1eh1uwS+Rxj6NOstYuMqUi0vLDLuDbh//z7dvsvLuBdFRIlleW2Nn/rTf5r1jbNoIfnCF3+fRGoUmuv3d/jP/spf5a/+F3+NSsXQrtdAaKphFZNqdvd7dBaa/O4Xv8S1N+/z2e/9JJKIxVYTm8RgFSoLFmvVGujM4xJWQocetdahPAOfMKhglHAZ9EyOOLVZKvxqwRSTDG36bqIqhRCFHUc6aGXh6ShLAtbaKWnDuaa/c0ziA4CQdFTOhnwYrDRhHDlycJ7F9zgtJq/aPW3o9DAIPCUZ9xP6gwRtJZ7I4L5ykgA3CAIkFm2ty3ZtDBKX2r5SqdHr9Xj+U99LvzfCqoA41sRJzDD2+Ie//KvcvP2AWkPyC7/wp/nNf/0/4sWaoBpggwrj3TGvXX+DX/2nv8If/bt/y1df/H2s0VSbDbqjlEqjzae+77NEoyHPPf9xbr/xOr2tXbracO/BJu12m+3eLheuPILvO/CQ7/tuoUtXt0J6rvSeRGWLIBNjkxRVskEoAVYal5WZkgs5ey7CU8hEkmhBtVbDpinSC4hTp3aEnRavfOtl9rf22Nnfww9cJOaZC5eod5bxm22WVjrgBXjKJ440y81lGosdvvKtu/zgj/woyX/x1/i//fW/zt/6m/8NcV/TalRJhSXujfnxP/Xj/Mtf/zXu3L3NP/iHb/HpTzzJ41fP0KxWEEJRDUSB9DTC4ikFqUb6AuH5JMZlpdAGrPAcDLqErbHWFkjZPJXf2toavV7P2V2EKDC90mbuyixQTdjJfD2Kivknjl/0OUMRnB4vUZyXvZ8EjnrImcP8wTlpcMsDcOQ5x6DR5rVZrVbZ2dk5lBVp3rE6TRE4A58xBp1llHKGMNBIwqDCq6/d4Jf/4T/jzIUWP/fnfoynP/osW7tdrLXUajW8UBIbw2KnyqOXrriIxhAeOb9Gp+bClzudDvV6HWk0EkvgCR599FGihT0O9vZBGJfEdWONe1vbfPwTz9NotVBJgjTO8JarHQVyMGMMSZIUyVZt5kf0Ap9Yp4XDQggH+9aZquIJSLMQbU822d/bxVpLo9EgiiI+//nP0wgC9rf2OHv2LOcvXiLRKUsbZ4mtIvUCrFRuB5WSWqOOX60wGAyoe/Bjf+KHCIG/9Tf+awb9LjqJaKyvMIxibu7uUqtU+XN/7s+yu7nDb/yLf8VXv/IqNom5cnGder2CL0N8z3PBaBnuwnkjHCmlIPMYWClB2KxsoPs9TzE3Ho+xKmB1dZW1tTUO9CTupmAK2WebM1Lx9hDVR9kf3s65szRRn09mKA9fhokZmt31jxqot/v9PMp3v9mBk1IyGAwKl1Xe7qzbKd9JYJL4JG+30WhgrSsAOxgM+K1/+Zv83b//z/j48x/hf/lL/zOefe4ZJILV1VV2dno0W3U8TzoJwfdYW2yy0g5ZXWzSqNUJwxDf93nqqacKhphPdM/zqFQqLC0tMR6PabfbPPbYYzz97DP89m//tnNbKqc2VGoOvFV2teUowPx+5o29wDqkp85L2IO0xuUlysRvpRRSeVghiFPNv/uDP+ATL7zAxz/xPB//6Mdc8FSrxY23rjMajQo7hUUilA9C8eSTT3J3axMV+Lzw8WdZ7bRZX2kQerC6vMhiZ4Fh5FLOr662WFxcwpeKVrvBT/+pP8Fip8I3XnqDa6/fZDyOSWKdeSNs4UkqS6WzsSOz0OVcjciBUJubm2xtbU3NIWM0aI3IQs6lsUfOreNonvvzJJfoe0kPPXOYiFlploQscYE6UhfvMrMiC+mANLkvP393aEI59VuOFZMWhDR5gjOUNShhUGgkKZ5IMDrl7MVL7O8eII1wocIyq6eoJFqCEdqpJjZDDuLgxWlWYt6YlCSNiK3m7/7dv8+rr77OxY0q3/99nwAN1SCkUqmxsFDn5372Z1yMBYrFZpuVhQ6LnSYLjZBGtUbg+RitGfSGPP74k4TSc4hDBMJK4shgpOSTP/SDrJ87z0eff4Hz5y7ypRe/yic/9WmiJGFvbw/pKfwwxFqNklliWWExWoOxKDHZWfOFk8YJwlhEarBxikhNZtAVaCzGCqz0HXzd82m0WwzjiP5oyLMffY5vfP3rPNjexatVuf9gizubD/jxn/hT4J4MnhRYk2JwsQznzpzl6aeeo1Fpgk3pNKq0Qp8kGhMPBpxZX8EXlj/8wz/kF3/+5wBJu9WgXa9hkzGPXL2EF8Ab129x/dYWCYaD/gE6cx9rLEaYrO+TgjFWCHf/KAciy+aNRBAqSbPVplGroq2h1+8jSFE2xTcpPhofTQAIqbFCu9gRW5p/J4UB2EkCnfLLGFNUa8dmGdAzLFD5VWZsLiDPvaaYVNbGUfTQM4dpypOz5By+7Np8Z8ElOUzYE4IAS2gNgbVUrME3Gs9oAikYDYasrqy5nS1nWJnNwZMKg8yYUKltIYod3lrBaBjzhd/9A1aW16lUAi6fP0OrWafValGv11lc7JBGCc9/7Dk2792l3WwhhKXTCllvV1jpNHjmiavYdMzrb7zBn/v5n6VSC134uHZWM5OkjIdDDrp9fvNf/Q+MxhG9QZ/eYMAP/MAPsJgllclVpTAMHQDLTMaynNWpDM4pJlvZLJNNNCdhTHY1FfjFjlypVOgNB4RhyKe/93tpNBrcuXeX5kKbV7/1LV5+9RW2trbY3tpi0DtA5RKI1pg45ic/9zlu3rjuIOie5ypbr66yuNBySMwo4Ud/6AdZW16iXgkJPZ9K4LO+tkqr0eTSRZe89803rvO1r7zKYDRmNBoRR86zZLBFUpw8d0dRQX2K3P2EgYenJNFoRL1aodGsoYxBWYOPwbc2e6Uo6zacWQj1B4EecpvDfCqMXyXbwmxlrDJNVWAqDJiT34VOXdQfGs+6QKI0jfE9H43EoBiNRoUoaYxxYKWiD9lLeZjEIJkvog4GA37nd7/Fn/kzP0AY+Lz22mtYHdPd32N5fQNrXEBPQyr+7//VX6N7cMBf/at/jWeeucrO3hZSCt567RVskvKX/1e/xKOXzqPHA0wUsbP5gDSJSeOEZORqUixUG7Q7TbwwKHzw3W6X1LjaFWEYUq/Xs7HxJiJryXBWNjiWPTH5b+77UvHcbJy1NoVLM8+FMY5ilPLoLCzSaDQZj8ecPX+RXm/AYOyyQu/39wgPGnRW1qjXm3h+wEKzxl/6n/9Ffv03/iX9/T3a1Qo6GtOoVXjjjbf4yR//KT72/LPYJEViSD2JVhJfCfrdfT7yhEM5Xn/rJi+9dJ31lQ7NypBRbKkLSyAqVIKw6K/0fZdMAxdwpbMxUEqhraXbPSAed1labHNn+z5n11YYjAdIY0GnWJPSbDZJTYJNU1KvipFgRO7pee9plpG9F7zo4WcOwswVffLSdIcOLzGMI40y03gppLD4aYIdj1lbWyUIXCh2t9sjlZaxhcD3WFldnmozZ0hKehjpUJBkNQ7ylGc5hWHIH/3RH/LzP/8ZlleWwSi+//u/n+7+Ae0Vn/3dXfxGg09/9rP84Zf+iE67SXd3h7/+f/rf8ff+4T9CEKJI+NwP/SiXL12k3W4z3D9gd3eP8XgMWlOvVDFZUtdGo4a0EFQrhft1PI7xfQ9P+gSVicSgVMmRawyU1InyuM4bTaezT5LG5AspZzI5wrLRaDDoDuh1B4y9iHrFMSbPT2k0WiTWMO4P6I37dHf26PcGtBeX0NawvLjE6soSP/+zf5b+cMA3v/EKO3u7PPPR5/jT/2GdVr1BrFO8wMsS8rgyhgr4vu/5XvrDAd3uPucvrPHUU49wsLONES7vXpQkVOv1AszleQ7bYACkcmHXqUanMQ4VoTh37hzffus6lcBndXGJWtUnHhp8KfCkx5OPP0nF96g1W/z+V/6YHoLUOGiteX/MBe8JPeTMwQIm05Um+IV8l3J/WwzyWEYB4EnfnWdTECCtQqLxMMhkzNMXzvPoufNgNb3xkNFoxEHF5/72Dp7VpGlEEvUIgyrKyExaydo3LhW7kBajQnQSkyZO/wbBaDTizt37fOKFT3Lx4iVe/fbr1CtNFjst+n1JMhqRRDHeeMy//o1fx69UOdCWZqNGmsT85f/kL+FJhe8r9vf3SdOU3QdbjEcR+/tdLl6+hJ9FMqZJDZPGjDNwTnzQxyhnH9FYrE5ZaC0gQ2co9KVzyzppzLgMVEpmJehng37KY2zd5opBJ9q5OEvoPSk9kBJhDFIIjNaOUfVdBu44jqnVnLFzNBrhCY+gUmGhssTyikevN8DGEakx3HjjGu1Gk8SkRFHE009eJbWGSoY3SJMxwlqGUeTqhSYxoZLoJOXGjbfY399HKFCexIoGq2fO4guohArpuXiJwAsRQuD7ISjhonuFxBiJlAZPkkHGXSUwZSSNSsjjVy6Q7DygASw2mqwvLrJ34wYbVx4hGI354Wc+yq/90R8hq3UUisR6JQbx7jnFcZvgPJLOROqC7E64/EPOHMrkHoqj+RFos4N0SNTKfndl1VOEcFb1IIl4ZKFDS7gaALVqhW4SU221qCC4dvNmhjjsMI4j6qYy1b61LmJUp2kmjcoMNZiHKIesrq6hlOTrX/sGZ85fIB7FXL9+3XlBzmxABmEObJU07nHQdcVzjXC6dyUI2dvbo1qtcvbsWbYf7PD4s49z6+YdojQhGrnU+QKL50lEnmfOOkMbUqBQtFoNFwfgKzzlTY2VtRki0hgCPyjChHO1zMwZy5xyZu1epbHOJAelFJ7vE3ba7O3sAi5jt5TSWf+NJE0StLWMDw4IggrWWJLxGGUte7s7rK+v06hUsdqyv7ODF/gsLS2xtfOAarXK6uoqd2/eYtzvEUURi502tVoNrTVhNUAqwfr6OsIahE5J4hhPBXhegDaGIPARSoHIEhILSa6lKqVcRKuFUb9Pu17DM+AhCWLDmfVVlhcXqAcBF59YQmjDqNulNxrwwkee4I9v3WOYaISYdnV+t+g0Ho8PEHM4Hc3jolrrwhJtrc1AKgajDalJ+bHv+R7k3gHWWprtOkIpROgRYKh5bUy8zlu3bhEoZ4z0meRGgMxoh3CGRyKEEVjlIwPXHyV9pHKfn376aQajhIXWggv3NYZr199ibWWdrfv3aCwssrO3S7PZ5uDggE6rSRgEaM+n3Wxx+/ZtBr0+3b0uu1s7hGGVRFgqgctUpaTAClOAwcgWd2oNrVYLP/QRXmZwsxyJ98gZg1fKE1nO1l22ORy3cxWT0LrITayl2W4xHgwLI2d+fuD5JDqlXq0iUIyTMYHvk1rwPMHWvbv4vk9vMODbr7/Gx57/OJt3bmOtpbe3y87mfUySUPXDIl9EkiTUajWWVhap1ipUKgFSOG9LEsf4QYCVgqDqXLrKDzBCI6UgSXOGJ0EprHbSVeArxsMByvc5f/YsV69epRUGVD2PUHkIk1D1A/6/v/bP+eRnvpfO2iqetQXOIQdIvZ0CM98NeuiZg9GT0vATycHhwozRJYPZjD/YSiAPm87UCeuYghUC34AKwO7t0lGGaLTF1774Mr/wF/4C3Tgm8J2Y6eoxtpFS8o2XXyJNY1LPEiJQ5OnvM2OcktjUx0qJ9UBnornM9FkhLPFoTLtSZTQc4me5DS5eOIdOBRsbG2ztdvGFj9CWJx99nP6gy87ODm9tb3Pl/EXqYYCyhoVOgzgdI/0KTRVghBOb88hIrbXbEWWK0Zpmo0m1GhbYjDwgyEBmgHRuYCnznBaObAZGcl6KaUOwG/eJipfmGY91hJCTSFMBKCtIhYP/VqvVIntzzmhcKjbpGHPWdlANCKQlQmM1CDzSNKFZrfDCRz+K73kI32cYjUkELC8uuMIywnkc6vU6/X6XdqdZ4D9yVKsIBF7FMVGZPS/hB2gp0NkmYIRxbm4rUMpHiwgPwfrKBrVKhcVOi8euPsJCs4HI8msICbWwwt7d61w828Y3A2zcx4xGCK9S2COdQnw05TDpCTw6i/LNBnUWGXxIamZWJcztQPLU2sxDzxzmkbtZO+WBmKXjDGgu159FJppQQ11IknHEmdUV/uW/+HVe+P7vz3YsTRAIWnV33tr6CrV6JQvMmiwhjS1ARdJTbrGpkmEOnJgPVOoNTJIiO65Yru/7pFrj+yH7+13OntsgSRKuXbtGWFEsLS0hpXTehSxuQUpJFEXUmw28oIIxUKs6VcdkInxYqbgFkXlY8oIxwKF8AWXKJQQppouv5MCmHBxVLotXtgXNwnNl4fHIVA4oUJl5FbH8njwvcHgAY7OcmgqkxgYBGFecNw+lDsOQOHUYklarhcq8Iu4YWxQ9brfbRWV0Fzaui2elsopWVoDMvDVOL3TPzvO8qYzj7r5c6jprLVevPEKzXkMKi6c858pUHhiHSvWFxfck6WjM+aVl3uqPGL2HLs3jJLapEIPySrDTtUiOow8kc3ATbxrbP8sOnfqQDUNJ9HW7VGZvsJpFPyRMEjxtqAaKhbUN7t27x8bZs3iej7UJSlpWFpqM4jWwKZ5isgMDBVc3xiH7jMTlpSkV3hX5gpR4gVfkDUjTlEa7g9aaheU2RkugwlNPPeGQjNIQBB6echJDmroanfV6HZ2V4QsqIRYy/7w3jfizhmq1muEzDtcymF74uft1NiX9hAnnmIacfN/PEtlMVA9rbeHSFGRuZj0Z/xxD4IcBcezySWgrSCxODQKU5zIuJemQSqWCzdLV5+qStZag4iQhlCQsRZUKMakxIoRXJCm2gJCl8REU7ktrLVZKTMkQa61jMmk6kVCFdCCjNE1YWVokDDzCjNkqKxAYbrz1Fhura/T2d0jTlHajypIVXB8OXKagt0n5ZvhekBBOtzmN8PCQM4cJfmFqlxOu6vFhWK90C8CCzFKakWHbhXSqiJuuCskYzwo++pErxMMe0aBLrdbEky7rUhCGoBOsNbTbbVKjWV5ayewXgJKHjUpKgpEYZTFaTi28vI9phhcQ0hm9KihSa7JdN3CBS7jM1Gkau1DpwMNahVEpng1cijpfIoyDGQspi9TyLhErpO4y1NRkMeQqR3ncyqK9lNahAKXIxjiLv8yOyTMdle+rOF+6+h75IsoXss2SvRhrnGonRZaKzp3nhUGWNj8oQGTKToBYQlZcsVhjIE2mmJRQcoI9yWZyWWIUQmCFq2NRVEDLJAghBFLbQmUyKpd8fKxNM7E/RaQWzwrSzDujrYcRmvOXLpImEUniEdbqCFKE8en1uvT7fYyQVIIqMkqpSMlyq4K4ZRAyJZUub0Q5tWGZIeXv05Ge+bHenO9OQZmXZZorHM+oHnLmYJlky52I8RP4adkGkf2WexfLrVi3ECXepDCMlVQCxeUzZ9G330Ra8IQkjlNUILh35z6dlQVUZtGXUrk083F8iCmVRTillDPyKYXNdth54p81bhJr4XI8Cisy8FWa5WkUKD9EKB9rs3vOMmGLzKviKR+Eh5ATzIWQjmlJAaGniliHMnlZsdz8e5Xp2EKoDJJrsEbgeT46W8jM6LDzcBDzKHf55olu83PzRQ0uohUx8UYpESBKO7VJNRXlYXVSSC3WukpYJvOylOHCRWJX6QyJUzYS5SHymBiZGQmlRKCdrUhmsGIcvsYKjQVSo53KmGjC0Gdz8x7dixdotdtY4YLpxlHMYDSkWq9hhKBSrRBFI1e7QwiiXh/bcnVC5KGkLfm8yO5byULlmqb5DOEkT135uFytkPbovJPw0DMHR3nZtJysNkXZOCGEwxOUErRY4Xa7/Ayl1MRAU0gfinQ4QsaJA7hkxUskCiVcPoIk0YV4qo2dGK+O0fVy8Vp6HqYsZufn5O/GhXcjXFRj7jmwpUkOOGx/1rb0VLG7prjFbFAgDeTeBAEISagEnsoKzBbj4I7JI0TzPqU6X0yZDm4F0nPxEXGaOjCVNyn3ZoyrbwFurIWSmCwH5WFsBEV6+XIWoyLbdeZNSTIe5LJKWSeF4QBqQS1wGAMFopRcWJYkDFlSB4wq2UpKz8ZFWk6kUStdun53jJ/ZHtzGM2E02iFAM0yLVBBHYy5evMhe94CL6mK24ATDaExQqSHSmGgUu8ArocBoKqFH1fcYSjB5Il9xOA1B8T4rgeXfl934diJVn8Y1OUviBEz3Q84c7FT2acilgFlual0691lcQ1ZJSaKKfA15uXmhU7xxCknskpwoV5MySRIXZxEExBisyCYxlvt37rG0vIwnFUpNFkv5stKTTj+WDhUnlFcY8DyHFCIPepHCBdEIQeZYsW5BiyyDsTV4yhk8vSwRbG5xNiJFSImnwkK6kkJnu7JESYGSCiEmFZuknBgcp6MRyfTu7EZUDtRxUaQVPyAajhCeKNLCk1XocvfvFpjJMRU48JM1GusrFwKOu6d8EueoyiIHZR4YZwValdSAzLtklYMylyUWYYTLRl12UWfeoUJKka4ylnv2ZJuMyEKzJ8zDSueNcczGOGlOCDDSBeQVtg43dkEQ8PLLL/P8Rz+GsZZRPHIqi3A5K13MjQciRFhNqFMurq1wYzSmZ0WWhi+fQJONT0lZZBorxvaozSjLgv52yAonFUnnrjr22IecOTiy1s7cxoxoJd6BwUYKQk8RxTGkKePhCK/axAtqWJyRT1jDcDikUg3wfZ9+v8/q2trJWX+kU3lyqzhSOPXElCZZxhGEMCiZV7XOmB9uUiiTpWEDlOehbJiFgUt8r0KaukhSLJNFC5m7Th0ak7KdYDY+BSYL1QhZMJFqteqiPT2D8MSk1mXmuZi67eK7SSCc1ZbUWhQCVSoVn5fNy8n1w+nEUkzQpzabw06wmtabcy9EGXcx62GxyiOgnPB2cs5xMyYPV5/dkD3Pw/juWpubm4VBNYoi6kGDiidIjIZqTEiKZ1wymVo9ZHV5hes3b/JeICPfD3rImcP89Nw5lLeYXNZB18rBQWU/77z6jr5QfOTxJwikoj92NSk9z2NsBc32AtILkDqlUqmh04Q4ilzB2yz825qS6F+2Pdi8DmPeAwec0SZ1Hopcp8zKzgkmYuVsRS9ZmvjgjIzW88CKwpI+0eEnkZPZyM0FOEkpiyzTDgvhTcTxTOUoJ3LJvw+CgCidZEAqM5687xMjYpbrwZgiS7X0/KnnmS/SCZmStOCeqZSSGFdxSirpangw8Z54QhVqiRCiAJpN+pFlYFIOHl0GcbkxmlAudeTtTQyCToooM0IpJWtrazQajYIhCuGwLL4vCestxnsHWCsxFgKpXM7MatXZV+LD9TePozIDn+rzMerte0EfsJDto2jimjrt0V5iuLCyitQOuVcultvpdLKdaMJYGo0Gg8GISV6I+ZTvJLOkpO8UVilK7wKUzNyik6jI8qtMxlpEdm6++/m+n6VFn7hXpXCemZzytsrSRVmtyBdsXrvCzyUZbZzUIKZViKwzU4zbZAukXJA3331NqguGNCut5AvdJYfJ8QheMSYqe0mhJjk4lEJ5HjqzTZTHLndNFkxDSWczyM4Tmdj+bpaVUqqoqh3HMdYYggxnAWQGSgeCy1WRMHSh5Pv7+ye2/7DIFQ+15GAphT2XuWQuMOQTzUqQZAi+XC+fHC7EdIi2tRqhU0TcRwC+VFSrdfywjk6GxGmCUD6ucoJDEQ4HAxS2SARanuCa3FjkJqvwlAvdPuSGVeReFxfYI1y+DetBZkTNwSllaSSXQVypS4MAPG9SI0OQVcnOxkOinceB3M6QqxuTAjY5YtBmRlFjDJVKVu/RWOfVcYYErDHO8FhKIpJrrBY3tgqBLySxSQstz2RFcKy1BfDMt84moBWQuXMtppCGhDBY6SGEi4zMA6OstYT54svGQ4Qy8/iIKXsGSISXSQ1y2tg3ZewVWSo85ewis8bj2XmTP0djDPv7+yTjiF63S80PWVhYcHMqY8D1Rgvb72K1YyZeKDEDg+V4qSHvWy5TzfomCntVZgcrvj+uzayaG4DIjcJWljyB8+nhlhxOYO+FOC5sNomP57llXTuORighXeah2JIaGEcJ9WbDWd+nAolcLoalpaUTLcMusYudmojzji/vdPOwHEdFmeYp4LzM6CjFdFvMuC4PSR+l9HXl33MVpbw4ynkwyhgJz/v/tffnwbZl+V0f+Flr7eGcc+c351Q5VKayJFVpoiwhCUSBgEY00XJ3g4HoaKkxtBrbDTZ2YAt3R9B2dzjA4XaD3W2w3IIWBEaARCCFoMEgwDIICpVU1DxXZlYOL/NNdzzDHtb69R9rrb3X2fece+97WVn5kn6/iPPeufvsYe01/NZv/P4yxFqM8gFLErwt8QPLFa+UE5r5onMDR2bR20hOT8PuWvrPUL1M2++Uxintd+sgRdFJJMvxHel1qVq1ZOwcjFsKgBNtMVprjo+PO/iA6M0qisKXcnSuC1Q7PDwEZTvYvXUqQfdbVFHXQMy9XZXiIt6Nh5s5JHTWQgOHEl8E9uzzehIRTKYo85yiGGFMRuuE69evY0U6a3FRFGRZxubmJmVZdoavdW3sYwf60mlZlnUIQ6smHfQqQZzIqYEtUp736Epa+o9SoeDKGh94arCLizZ1yaaLZ6h3x8UTVY5IRvmUaFwPlx5JS/C8KN2/l/MqVx3qQ6QTvlNNuvf1Bk0fxdq/Z2xj95zEFaq08Vb/6GVQ3iAc+34VMHA6V2IB4IuSiJ8rkVmnYEN1qPGB0RwfTTGmjys5HbewmuL7vpsqxkPOHBRIxrBwrrMgzuDEYJ3GOu3dltq70QiVsH017LBgVaxbZFG6ZVxmaIRZW4dF45jNTrh07brXYZ2Ac7i2RVpLmeVJSK7CqgarGpxeFhONMV3wilMJCrHyEQvxk+5W1svESxLEUi8o1T27m4RYRAsqWyeZCFovi9OxwG5ZlhTaW9EzFNr5IrTaWVzd4JOwHNY2YB3S2q7NWZb5hWRrRDla1+CwvpSdOK/ji1dzjNa9TWTkcw7EWYTlvIxO+hMBp5C2wSC+Ypj2CVAu1B01xqtRvtaIN3x6tUp1EkbXB9Izk2ifyfP8lLSgwtikbv+h9BT7P/6mtebVr3yF3c0xG5PCp2Tg29gqhUWRFyMfZGU00ljauqIsxz6CV5YlWa8e9Z+LpHQ/mO0k2OfOKLwb6aG2OaR0nkilgpeCRFcPZ+OhxcXrj6oflLb14CH1bAZ4Y6TF4zBmuaaN5d604dad29194wKNouRQj40TqdspAmRaasRbskCr/r3SSZkm/qS7XDrhsyxbitoMNwp2hv45kTHEgr8oL4nE+9RNg3IeoBbnUFlGpjUSkqcAnHXM54vAYHJfXTskYnXvJz2GQzyeKU2WG0QpqvmC0WTsPTrJu0jSD8PdNZVwvCvYO6iUVt0z4vO6awb3T+8Z1aRVKefpNeuoY2haM5vNwhh5NUbhg/QAxtubnCzmzGYzzGSvW8neJnbazTxQmM5swzeC3pPM4etFcYI456iqinJzk+l0ijYltmm7JCet4eDggHw8OnV9avwcqgydLhqNhbpnXNr0YmhEudKqDytOQVJSb8MqlSY+Z9mj0OvHUWwuS1/hKs9ztPSxBi78H+srKOknuDdq6iWmFd2hsSDtUlxBwjy11iEnQWNdS57njLKcpmk642Ikfw+6PulUHfrF3UVDau+5QMnSjr6ufx6E1jEJ34bwbm3Lxz72MXKTsbGx1Rcs1goljizPuHbjuq8uLkJjvdS2qIOUdAG9/92kh5s5KHyRXJEuwlEp4xOO0l1HRes//Y6U3MZ24aohqxDNonVoHQrNKE023mb3uadBKYxrkLqlwVItFohYNscjzLgE1aKdhtaEWpBqyf6c7uoQGIiLz/XJQjG2QUGoMqWD2OyCeNov7mhFBw8QKyqmQfcuyZhLYHouhYpxBqKx1pFlUfQltEsHhhLCniVGZVjE9e5a3/dprIl0SW/R89G2LU7abkGnYnuOZ3ra5Gi8Xm+t83aKRC7WotECRvnsSDqjbqiBgQ8t9+1oOxuCLIVk98wxulT7sHn6uSK9GpeiXSmliFWkkFC5TAGSebj+8L8EI2O5OeGF97+faj5jlOU4k/voSFWBUtTOUmxu49QYdMnNt27TWp8c6MPC1m94NuEb91PR6jySTg0+X0J6uJkDckr0UwpvP0h2x/s1nWidMZlMyMuCfDymUjBra57d2yPLc5q6Bg2u9dmfs9mUqlowHiW7XYxTUGptclxvjQ9+9sSYqcMCiKnnZ+16KRy+iJAHNGcIBk29XLUpJYNCnFAUOa30+Q/DNGzf1lgUR1FViwB9J93iqeu6M0xGxtDhO0ifUt00TXdNNAbGdzbGUAepYwn+fYUKPFS/zurj2IfpsZXqZ/DQprMnfdZ5CyYykZOTE05OTrp38ujaM5TaAJHgzcgYb2yQlwUn84rX37xJM9ryeTrnRdk+BPRQt1CEDitgSZ+UvlrRKhfPSndXKOChQ8IWRcZxNefE1ownE6aVrwzVNg1VVTGtFjRNRdPUWNfw6muvLFWwEu2LuLSybFxLLegprkKqp6auu3QBpR6ONB06/S3P81NejHQxxL7prkUFAJKA4xgYjXMu7OI9xsMw3dpXhvKLPwZ3RcYSj0cJIh6LkkTqIk0Th+K7xfstMyeWrh3actL+WUXp+esWuRZ8kpb04xLfO7VLDD03w3c/Pj5mPB6zubkZDJ2+rmZVz7tnF0XhK4k7DzBzdHS00mvysNJDzRxgObDFmLjTOKBFKev/196NJ2JxrkakAdX6T3CL+QJYCkJoU9UIv/BL/4RJuYUuSlSeU1nv6msaHy7dzGpc5aBVXLl8nencGy6t8vB1vUfkdKBNpDgZ4uROaWgIGzKV9G9jDLkI2lqUs2QKcq38d+cotE/a0srFhEacExpazChDl6FYizbkSqHrBdQ1tA1iBduEwKe2BvGBTBpfdNfhC79EO0NEcLLiaKxXMdq6YVFVnCwq5rRY5+t3GBG0c2jnMFowxgPKRoNw2lddf1jXuUEjDfErVzGB9LfYb+kxpTxDH8YypPfzqmIgs5yB6zckqGvPAJ947HE2yhFZmFsi3sMl0uJocQ7ybELT5qgCVLDBqAtIDWfFQqw6b/i5n+vW0UOtVijVc/MUbeht31e87nlnNmN06RJWWm488WQ3AXNtOJkdoQSyXDOejNi9vMf/9Csf5dpjN9aK8OfRRa3g0Fvmh7+vcnMCpySAeC+Pm+j7LHM+h1+airZSVNWcVgmNmrK9s4mS3EfTWdeFQxfSolBUdYVrFv6Z1npfm7Po1uJQzKZewhrlPvejLizFZETb1oExhtgJrajtani/1JiZGlXjb8M+HyZvXYSWpK4BYxoulqGa5//2c7CqGg6Ojn0augOsxdm+IvnIOqz2NUEaW4ET8lHJIrrM3mFaN9fSTew8elsrTSm1q5T6GaXU55VSn1NKfa9S6pJS6u8rpb4U/t8L5yql1H+llPqyUuqTSqnvOvcBiS94CFCStMGfmuwGF3nxFsX4ylXeODyhzQquPn6jE7cXiwU5Gm36BTmdHXuEn2RyDXeg+6FUbE13u3X3WqVGpPcaZiTG49FdCaDEF2dp64q2OeZ4/yZbkw2mU0GR49qGTJkQ2CTe1OpajBJwLZmik1qccwEureXk6JAsh1dfv8VnXvoabV5gpWZR79PUi+7jAliLCqpYrEI+FN+jVJH27fDdVvVD2h/rdsXekHl6vqT3H4r/UY21NiSttY5f9+HvZnN3l8n2NqOi9PEcAjqEn0d4QKsMlpLW9rCBq8bwfnb+bwS9XcnhzwJ/V0R+t1KqACbAfwz8ooj8KaXUjwM/DvxHwA8BL4TP9wB/Lvx/JqVRc/57r7t63/BwR4kTbGjh6ice+CIvtdN89nNf5Pu+6Vky8EVgW1+aXofS69ZadJGTFeOQbIPPBaBBBRSn+6GlxRCs4+tUklULIo2xWGUr6N8xnCMuxH85Fk1Fu5gzPzpks9zh5372b/Oll29Rq5zjquG/+X/+56jFAouvv2GMIss0TbVAo7CicK1gxYvX1loy7cFuf+q/+0scHIFg+Nmf/ftUteP3//Bv5Td8z/ux4vM2HAJZjojxqFfiut0/NY6a0KcGTUNvk+iyJSXiV/QeiRg0NFQzTkknqnfPWtcD5ab9qZTqYha66wLDcigW0xlf/vLL/M9/12/3AL+2ohWLEZ9B2jQNbWPJcofFYrUAJW2We89ZVEWju9nbgpfbmdrR6KNQfduimyf0m7/g1Ht2/fCAvOaBJQel1A7wA8BPAohILSIHwA8DPxVO+yngXw/ffxj4S+LpnwO7SqnHznpGDGQZLpKhAW7t9WdxYqNpHeyfnKCLgtaB4F1bEcw0Pqt13jB3+/btzrvwILFpkdJ3GkoQQzF5nYHtPHE6zQUQsT5YqWlZzE7QCH/y//yfIK7msRs7bI5hd2fM//4P/4fcuPFEH+cQJAPnHIiP8GvbmnoxJ6Ssslgs+PxnP8uHf90HefZ9l9FYyky4vJnxP/zdf8BnP/MppscnS0bkGBsx7I9OlQqeoHSCfz3UycgYgI4xrKLh2Cy11wmNs3zsU58KKdsZGxsbnsGFILB+pSd2E+0Bda34Re2Cm/TdLm5zFr0dyeFZ4DbwF5VS3w78KvDvAtdF5GY4503gevj+BPBqcv1r4djN5BhKqR8DfgxgsrWZiNkGDwsOaYBPL/4NwD7WNNpzWo1z3m9+az5DjSa4eo51DWVZ0kznfkI4iynywDAsly9d9Vb+uiXrjFXr3W2rxNclD8aK89P/h4wDezoN21e5Wr6/tT1jsNaiQ38dHR8wPzzi9Ze+yvd973citNzZP0KM5o1b+zx2KePf+2P/AT/yI/8G165dwTlCGcA62CFaj9NgG8TCrTt3OZlNeeqZp8A6lDJkRsg3tjg62OfJq1t89Jf/GU/+nvfR1g2S+QK7Lpl2XR+p1Tv9uo3Av+fpvo/hzkOytoekT/tqiPHQbUIrdHNvGM65cvUatw5hMplA05BlGePxmLad0bQtgtAKGGW8NAqgFdoYL9GKD/i6SMCWTyA797QzaRWUwUXsd2+HOWTAdwF/REQ+qpT6s3gVom+UiCh1f0KNiPwE8BMAl69f665VmCVVId2BjEkGdf19gzXZi3MiPvx2SsvcuU49WLIDhLLsZemh37/v+34Dd/bvsLO3jRnlp2Luh7TErEJcw5BSG0PKSIZtB0IAzumoQHHOeyaahogCFd93HsrEN01DbjRtnnH18h7jvMA2wsbGVXYvz9mYfA2rLL/5Iz+IbVqODg7Z3Jog0qIUWLGI9Vb6ejEH27Az2WSy6XfNpqqpLre4puVo2nLjuWfZm2jGhWCbhnpRUYwNZH1kqM57HIbUptR2gU298a5tWw/+CswXc05OTrh+/UzB81SfxtiSlEmvG7d0fFJG21rhy5//HNfG+HiOpsGqPqnKar0Ew59lPjYm1tiQtr3vte5VELpnDNsaJeyvN70d5vAa8JqIfDT8/TN45vCWUuoxEbkZ1IZb4ffXgaeS658Mx9ZS7BCvZ7mQI6Fp2waL9Tqs8rgBUZpQSgVbwTJXjgCkThxgsWKQpiFTGUdVzdg50BliCq/hiRc9VWspVA5FwWR7h49//BM4B09u7voY/8GCXra4G6DXadPzwsnd16EhcpUqNJ/PGY/HS78r5StbHx2d+HJupkDEhtwKhXXC8XRGUWRonZGbnEt719gezTiZLchHFdvtJa5c2uPuwV2KXHH56iWvTlnry+uJj4psmoa6bRGlGG/5dozxwVt1lnHpymXu7t/jajnisWvX2ByPuHJ1D9dalFG04jAitNJHJUZyyksenh2sZpixR8qy5JOf/CSj0Yit3T0flu5O7/TD7z6nZDl2YpiKLeJxMEXEB6klqeLWeki+w+kBf/yP/1scHR1w49IVqrqhrS1NtQDn0KZ3pU6rBcqMuH3rgKmt/ZwQH7l6lkqc2pOGcRb9OX6TW3WXpWOqt0XFEPmLqGkPzBxE5E2l1KtKqRdF5AvADwKfDZ8fBf5U+P/nwiU/D/wflVI/jTdEHibqx7qHBJixJNoQhTaZzxZUIcMx+V0iIInExempV0dCpwtok2Fkwlde+Rrf8b7H/QRU+AAncb6IqghNYynLjMmopF7M+coXv8CVa5f9rhnCIxUKo7wfXSk/kaxrQDSZ8rkPek26rgtgLSljsa2vBp7KXZubm7HvT91je3s7Eb19PojWJhjH2lBcN0ObPOBYZDiTUYw3aJxmNivZu3KZvd0rHnk6TKQ8N6RlByNydVXXIYUc0IqRLij2LrHxoW/zhkqEMjO0akxmTJDMNE1jO1+/oQ+4EjQe9VnTw+es3hFFhEuXLnUS1Cn4N+nPi3kncfTTvkuxFeLxuq6Xduc00Cve8+WXX+YHfuAHUMp7g9qq7s7RAkr7YDOTZ9QnU1SWc+fowIPvxuheuZg7MT5zXT9EifdMFDSJ6tNgQp1Db9db8UeAvxI8FV8F/gDeyPnXlVJ/EHgF+DfCuX8H+J3Al4FZOPccUl1koyQD7uteivcpO/GIzwPvxKnOTHpPFAG+PsNh+corX+M73/8MdROCcrSvfWCDHl0tGooNoakqfu3jH+M7v+u7UDjEtjijMMqXrJ+dzMjGpU8LV4osM9y5c5fLO1cxOg9ejhVvqXVXA7GNk7DvgTMptUukOIdRaskyn5q+WCww1pKXI1hUOOOhy7LMMdIeQFdlOaPJBGUMbVNhHZycnJBlvu4kzDmeToFerYtGN+0gz3ImuyNELIVWiHMs3MIzeWNonPMp9lhyk+Ph4DwmZ28b98xhFfWLVnj++eeXJa20P5O+WSwWfUUr6KWC5Jz071hyb/h7jP48PDzm9u3bLBYLdna2mM/ntG3LbDbzKoMTRmWByTIwmvl8jtvOuX10gMtyyAigRGcv0pQhpJ6UBzPMptdcPDHtbTEHEfmXwIdX/PSDK84V4N95kOcMDTdRRNSZF0OXwT9jfL0sJeJEiPhouPT91aKM5ubhCaJy6uaYjbykzEfUZo61M+9vz3xQkMlH/Gvf/ev58Ie/iwiTDoHvKMjLwu+IJur8zqNHWbUkFsc2sWI36BCgo7i7wjV3ynCX2DWWfw8ShMlRucKqCttYTFYitvG3FkvrGlSm0HlGnmehEExwybmWo+MTtkYFiqxLesvKgqwcYXQoSWe92teGZC5tMpq2onE1DhUS3ArQgslzivEIbYyvLBYkLq01gluqQbG0a8b/szzkc0hX/MYpkn7nVJ91zCFRI+L9z+rf+OwoZTQhqnOUF0yKEqqWeVXTOqGxsFGOfb9kHl8yG41oyGl8fcSuEtpFaGjMju1I2wVr2EyC9C2SAPyEYxcxhj7UEZJn0VKQyoN4FTv8PMfO3iVspmgbn8EZcxDqRlHZFnGWpm2ZTqe8//3v5/DwkMn2pGsHJLaGFdb1WFdhKAZ0sRrJBB3GKlyEXHDRCV4KIbG2a/F7hRaD1iOqxjM8hYdWEw0S/OamyCFEVGpjvM3F5DgUR8cntK0Do8mznGIyJs9LtPK1JFXuixE3VU1raxYLR6ahcS1N69g/OeHytcfJi4LJeBOT52RJgNaqdx4aBuOxGE6eRoWumgQ6UdWG0sDwvqv6O0pHKULUm2++yUc+8hEmkwmLxYLpvUMmkwmXr17FCNimpRwX5OMJd+8eoEZjbh8eYMW7ytfRWe14EFp1H88gv3FqxbtC0XV5lofi7E4WwHkGId5D+KWXX+b5nU02y5KNrU2sbWhpmDdzDo+OEA3jzU2eeuopPvaxf8ELuy/4RWb6Gg/GmCUG4Dn++lakjCAVG2P1bRFBkiKu8dwhM1HaoJd2Cr/YRQSJTNBojGhGm5vYsuVw/x5fe/kVX5DXVezuXUZphVU+NtI5RyvgtGGyvUN9ckTdLlBosqL0RX2UoShHtHbmXXPK209u799DhbiIRd2Ql2OeevY58rKkKH1RXxfUoFVRoauiUIGuClTad12/pW7gZJcV6aMth3aM4RyJEmmULtpg94nXtGGD+NxnP8u3f/BDVIsZV7Z2+2xQo9ne2KF1DfNFzbSqGI83+PinPk1Tll5ZEulacuYMTaSGVS7ededHSkO4ogSq9MUklkjvOebgjYamW3Ra50nye5w80UgTRE5RkLjHdChnKUClFWVR8rmvvMSL3/OdzOcV+aSk2BhzaTLi0tUetPToaJ+6WfDJT3yC5z/w/FKbmuDvTlWFfmBVVwsiUgpJtmpheLek7tCbwdet8DiJINQgBqeW3XPxeeCTnKwzGOODmDJRtNJilTDZ3OKFb/5mDg4OEAXjrS1vyTegdI7B55joWtD4WA9VW9AaXYxRWYnOMyrXYkYFmTa01mGdcOnK1W5BZkXBZLJBZkZkhfHgvehTGlVaOzI1HA7djqnFPtqOhsvGqT7vQto+elSZ/j5rd9ahFGd098wYDPaH/9Af5HD/wNc5sQ2j3KC10GKZtQsWTcvR8Yxsc4upBZuX1IDDhcpe69swpNMeiiFjk6X+Ous+3l43iAQ9g95zzOE88tw/uDbDu3u0o/XXLOoGtajIJ1vUxwfM5x5d2FkLyoQAbe9jfumLLy0FzzjnmE6nPiBmQC7ET9wPv5bBrqfWDLqfqIrICFOA03hsqFda6bNTTVaQ5YorV32xF4dQjEfLao3WZAFWTmcFKi/QyqCzHJMVwbXcM+BiPKEoRqh57iUSa1HK94FWBb4ekEKU5j6jzlfS2niQNCQ6OderdquvSRnDUnRsAuWvteaVV17hlVde4bu+4zt7JC3nyDIfRn5wfMTB4ZTdS5eYC/zapz/JXCyODHFRhYuNWv0e6futkm6WabWicFFmcRY99CnbQ4qW42E+QUrrJs06sjrDjCbsL+ZkZcF0PmNRV772Y0BvEhF2d3f54Ac/yNHREdPptOvg0Wi0FMTT2RrCTj4sG3cWRWkjhhKnNo34/7oJPVQ/0nO01lhR1K2P54hI2saYTjVqmoY2YDjEa532hWDysmA02UDlBaic0WSDYjRmNNkA7fV/HRCXNza3ceIZUAroGp+X+u3P271SiLohDXf62A/p+8fnpP8PaQmAh9OSXFT7IiDP008/jYivOFYURQfsopTh+HjKpStXMOMJlQiv3rlNYwwtHp8ztvksdTgdu683xfuvwkIZ0sMtOSgQHbi4W14kKbnumAIXjVSpe3N1ByiJRk2HG4156Y2b7D35PrTOuHXnHmVZsrW1xSjLyEYFzrU4EZ546vHO35+rHJNnyYCf7vDoQUnb75RHOx66ptIF7pzr6mamDKdtW18eTlbbXnxhH4M4yJRgwuJ1WLLCxz44FCrzyUdZMQ44B73dw4vUBt00WCcoCrRpKXSEwfdKXNM0jCYT6rr2WJQmZ1FXflErkFCexUqL0hnWWR95Onivi+jU6WROxeuzdteUMQy9Rek9U4bYSR1KUWY5ynlQt2o2x7UL0JZFO2NiRmhd0rSWRnx0xu7lq+isoMHwyc99knJjm0PrEAyOps+60BoxD74399Bx6zeetG8jI76oDQMeesnhbM7mX/Ttc1cRYd5YPv6pT1OMSjJTsLu12+2m0MeOjEYjPvjBD/r07WRCXZQbP0jboJdC0pDsVTrysA2utR04S1wEEQGqDWJxCoBiQim6U++hFcVojBVwqM66Mx6PMcYwGo269vYuxeU2Dg2wF3GnPQgNI03XjUd6XirFpN9jf06n04DenTObTWnbhsVizvHxMU1MTjManWXUTrBZxsmioQqFj9+L9J5p9XDXGIrbkdZFkq27J4AoL1pLnjOzlmw0Ih+NfOWikHEXz93e3qZpGt5646bPwqOP8oNlfXcVXUSUPouGnH+4g65alOmOsSrZKLY5BdaJ92ysrwbm0KAzRpNNdFYkMPF9pmW8Z2QO0ZaxqCta16NUQ29TGY7FkNaJ2UO16TwJ4qxjaRp8/H9o+ykCbucLL7wQAsKgsS2tszTOkmtDE4LiGqN59a03WaCgGJ/atYc796rvb2eOfL3oPcMc1lPyCqdq/2nOesV0sBqBfGubWyfHFHs72CyjGI2XkpzquqZpGo8baLKlyd5Z55Wv0/D1GNyz77Ac8Tn87utuui43IMK8HRwcLFnfRaTDpazrmrrxKkGWZYkdBZyTpcUTMSBjDoFzvhJ3fL6vOO3vGwFn3JL648Ol/aJ+ZxaC1roHAl5FKhgsleuY2ZAiw2yahpdffpmmrtne2jpVMPhoeuTVNVGQ5Xzl1VdRWd4lXK36POz08LeQsDto132iTaFzWWJxUiNiEdXiYlJWwHhU5Kd2mDQfwwW9/VAUP/eL/4DbJwsaAUZjGhytramaBXWzoGkrrl+/zquvvxqK4DrAe0isbcE6bNNinEcEMkMommRyiFYXq2y0YldU5IjSPrZAPOy9jW4t7V2SFuvdehpm1YJ53VCMJ6gs7+5b1y1tVZNrg0Go5zO0OOr5jLZaYHKNc7bLg0hxH6OaYq0F5ahtzbye04rPoPRVuRQHx0dUbYNF0WKSmpaaxvnqUI0TLKo73grd93h+qr6l/TLsn/SYi5/E5RkZo3K+RVpCNfHYt6F4MImurrXHvvz2D307Y51TaEWhDJkSXLvw9TmyjEYpbt87pBaolCMr8uCt8UWDV0kR6Sd9h3VjDyHHKPkYpTFKnzqeftKck4vQe4I5pLRqR76IYeusgUh/L3b2OKobJIRbxwWxWHj8xKZpyPOcF198EaUUx8fHQGLVjnUPBrRqVzrPPiH0koCW02LvqncYIl5ba31E52RCnufdjm+tpaoqRqNRB9cG3qailOoK67ah7mMsNFOWJUCHMp2+R13XlGXpgWyDl0JEumIvaZ3MFF07RjqmY5i+x9ALM/QKpe+ffk/tL6tUrTi+zjlwNUpcF1eylG6vFIvFoos9ie9h8ahYohVoQ22Fw1nFv/jkJ6isC/E1p0FuH0S1PEttvCjd7zPfM8zhvA45y+U1vP6sDm6Bv/ULf5tF67q6DNGAF8Xtoig4ODhgf3+fjY2NpckeGYS4ZcPgqgKqF4Ept4qzM+4SSidzuqi2trZ8lamyZDweU4eMyjhJ8zzvg7joJ1FRFEQ4fOjDicuyPAXuqrXuGFBRFCwWC46Pj6mqaintOGVwcaGtck2vw1SA1WOZHo/XnxU/kOKB+OdbdCjIHPtQKR+MtlgscM4nqlVVtXSfKjLbFqaLiq987WVMOaKyAipDqYfbIXgWPdzMQVa7pbqfo+iYejWkAMkQt9qif+oRImgriAWkpCXHbU744uuvUAHTVpi3gioyjyVZFoBPlc3znNZZH39tnS8xRxBhw6S7yGQ9S7XQ4iWIRvn7xvtbgrgcJnlq6ErF/rj4Yhuccx1DixmIoyxnUpQo67oCu4U2GAGxIbDKaAzetYd1GJNTVVWIjahpFnNyrTD4ythlljPKCwqTMcqLrhp10zQ+U9EKbWNRaIzOuv6KpQHSd0jHemk+JMN6Sm1c4a6LDDONCen+V97o6ssbuPQinHNs7mxzND3h1p1bvHH3FrUTGjRv3jlg0RpmTcWXXnmDOydzGiXoTKFimH4sBj0oCN1XSe8nwLoNY520O3z/rye9d9naGlrHTIYd50iSUJTX4QSfxi0q45c/8UmMg/c9dp1JOQLtmVArjlu3bjHa3OLNt25RFCWTyRiRHqUn7qqxHqafrB6/ID4yZQg+2rFv50Ws7sNz08meTv64Y/ucjSZAxQuEc7LCUBQFJ0fHfZFdEWwoP5jhowS1E2rn0Ztm1YLC5P4dgdwUtFY8+Kx1HRy9RmEJadAIbQjpNsaX6wsvgRO/c3fSlVadIU+pEBOh+ziFNCZQxPdlGo7eGaZTQ60I1vWuZx8S06tnfq5EkCB/fdvaTio8PDrCWsv7v+mbuXX3gOmiYWPnEhu7l7h5+x63pzPeOj6mMjmNyrtEOO+OT2MREizU2NwVw7tuw1i3wejBPZw6P/jrPPpXjjmkdFFO6gu3xF06wypDVoz5Z5/4JM59G09fvezLrGNYTBfs7F7j0uXr/LNf/icA3Lt3j+vXrwe9fLlUOxKZUK9fw3ItRGPMkk3hrLYPd8VU749AJVEqEBH2Dw+5fPkyAPV84ReaEzJtUCYD5zg+POpUCq29YSveN1bMalxDhvKFcJUm0ya4MzU4KLIQcakybNv6+phZhjI+kcypmJ1qyYOkoLUmU8arOWWx9C7OOapgw8iybODp6CkuoiVbBqejJP33gVci+T2iPIlAj5xku6Cz8XjMr/7qr/LCi9/k7S7acHt6zMl8wSuv3sRmJZUyLJTG2eUAJZd60ZLANxNydoZv9SD7f3qPs66/H+niPcEc/IC2weB2VpMvGFQjGt+FYfACY/dZCQ1CRqMdamODf/zpT/Gvvfg8T125zO7GFgaoZnN0phmXExYzYXN7cynOvhPvQ3ajDhPAOtfF9hvojFpLpBxK+d27kwxUHNSsMw52KpVzXeGalPnEoJ1ce7j9rCzJUGxubtI0TRfG3FQ1o8IbGTuod/E5HUopRJsuRyQ3isVs7pGkgFxrNMHWosEohbOOrSgpWI8x4ZzzSVxlTt00KGU6O07sr7qu+1J9ibqU1t5MYysiRaYqxnEyPenQstK5A6fDqkV6u5CPdrVonfR1MO6KcjipOKkcM2U4aRSvH97DFDmz4xMW1qHykpmy1MpLUOft1FFScZK2KZFy/Un93wqMb1rn/RKRC3i6ojdvuS8uSu8J5nAxOjum4XTHRAYxSE5SGiWCKMUCTTYe8dmXX+HzX/oiH3j2/Tx5/So44fabr9BUU/IyR6mzO97/1guaS3TOAKeTmSAdDD0fbdsug85qTZF7sT8T6Yxq43KERjEqvEExUxpdFEjrz1FZ5j0OeUYWqntnaEQp8sR7oUwWKliBUl4ysq2vnq2ANrhQlXi8iGLkmY9YXyEc5QFlm6bp3i8aPY0xtKEo72Qy6TwEcDpDMe0b5xxbW1tLUtU6D1H6PTLV6GmJElgZvDaLqqUYj5FmTlU1fO6rL7N7+RJNuwCT0WQaq7QHmQ043+9++FI65wbM4T4a955jDqsWoVIqQG9B7IrzuLe/z2nYb6Vzj2SkMiwWVM4CKMsJX37zNp9/+WsUWpgfvkW2mPHsM+9H6wJr3dKu5ZxDRVhypNsIVtlEvCSw3k0b79ckrkBjDFVVdSpElCCUUuTadAtJKyiCZ2IULPEuWOfb1rKYzTuPjHPeqFoHgNSiKMgzjwHpE7cgL4zP7QhVnSQA9irXcnx45LNIM0M5HoGIr6BlfRxBq32QmEOxqBehjT7jUWW9RCDIKSksivcRE3Jo4I2SVOoWPY85LI1V6K/5fN65fP0zHdPpHOsUJi+59MQTHM9nOOWzS1utacXXX7WDgkpfzyjHjhGuOX6RZyml7ktleeiZQ/ryWmc4GzpCDXYQJUFdWL5u3d/dDtOpGPFEjx4cwP58nIPWiDiqtkYZRYVma2cPYwq2t7exzXzJI9AZuCTC21ta25KZYun5abu8UW2Zs6eT3DlHWZQsFguvm+c5RQCEUcDhvX2uXbuGtBYtzhvWnAeCUf5BXqyvKqS1SNtSVTW2aWiqRee7V1qzsJbWedtETD5rmgbRilzn3Nm/4w1gpo8GtW3deRt0kXN8fExZlkw2NzCtoTUaMRlG5zgcm6MNjDEcnZz4frAtgqKxbRdYlRcFeZahxMeXmIQx+A5yCRNUS2MQ+3bVuBNiUWwoqiwChS7IdIYuNWTQBDf2ZFRwPD3h3tEJ73v+BW7tH6EnYxwK6zwaOkqFQLoQINYt1AdnDum8N84bhlOyyPK0HbxnKkPbB2zGQ80chPSlJZEOvg73XsE8/IJ1pL3uVCiAE3ANjcqwYlk4zWM7lzk4OGJnq+wiB4d+92ZRMxqNyI0mVWEUYRKFc7U/yFDNMUpT15VPrRbITcZiNqfI8q4q2vTE69ouAOTO5/Mu/FlLX+Tn5PiQZlEhjU8UslWDWEtbN95boLwxT4mQiaCUYXEyZXHiQWWj6jLO/eJXSkFErNLS2Qbq2RyHUM0XtHXDZHPDl4EzOTrz6ct5plFaU+behtJ6BBtfvFibDspObKjLIHR1O/o+7EYJkT7o6SK7qDcveNtOVVUUJvNo4cqHc0f7TVM3XL16lc++9BpV3ZLFYrhKozq80ndWkRje3fXTZi19PZbKQ80cIkX9SS6g0T2or7cP0kncg/gaGI3SGCWIa6nbFqUM42ITmykaDdoYdjY3WVQVhMlZFAXinN/dgyFyKZxeAKfQ0V4RXZmD17Nti9HaLwzrg3G2JmOsbWkbr0psb2yGKD8PBTYpfYl77a1rGIF6USFthZ0vfPxC21KfTMlNhrKOHIVyAtYiKorn3lUp4vMvJCRPVU3tDZ3iw62zLEOUDxcHQaqGzBhMbqhnc5r5ArQiG5U4oyhHE+rGM+KsLDzTDXU2InPdKAqf2ITHi1TKV4yKxsolaL41cyCVzpa+xzEPRkkt0Zvkj1d13YH1AlRV5SVEaxlNJsznFdbD7XYxLcNnD+kiRsrBW6w8z2fQcuq5w+sd/Xsute8+FIuHnDmkcfQOpeMAfn04ddppfZxAlArTegDe4KSV4LR37zVOU7uGm2+9yY1LL3BwdMju9g7O2s4GMJvNKMuyM7alz9MBjj6GyqyjxWLBZDLhzp077G7u9jEUQQcnGgkl4Ci2Fm0CdmWodzk7ngbm0VAHqaNZLBDraNoaQx/iOwu5FcV4Qls1iA6h0iF8PO0vYwyjoghSk4d+czGvRPt0cZzFaZ8D0ooF1dsOOpDYzCBKPGK1EwqTeQnGOqyC+cmU7b1dqqoKJeXOrliVju9ZwUFxQUVbTWqvUGGMoitLGS8h+aI+yf3h3C0rhrV/PVLUH2zrezB6yJmDJwkirq8HEeC1hx7ApXNPD9WqCTIMIvILlsARBBOK4jislyiUBm1oneOkOkGqOQdvvcIHnnmSjfGkE3tjKHK8f7S4pyhIHrDFdKAdS1WR40KdzRiNRhwdHTEajTpLutaaarEgU5rC+LoUWil0kHokZE3auiE3Ga6pqRcLbOvtDW3TIJXDOIJdwneoE2GUZz7Ss2qYHh1RjkZIyMhMcRr8ohZaC1VTU4xLyqJgGphIu2gYb25Q2Zqm9jstFb6gzSwjK3yYdV4WmCJHBa+Btg4X4guyoGpMRmOq2dznMKCwLgDTajklFaRSw5LbUvpgJ/+37+vpdNqNSz7yNp2yLPpsSqU4mE5Z1A13jk54rGkRwOBw0Yuyds727UqRsle1K53QMripUgob279kahiKoglDVCC0oX1m6V5pG86i9wRzgL6z14YhewPFWsYwjBiMx+P/cQDjqPhdIVqIEztCWIBOaQ6PpszfeINXX32VZ59+hrLwInLMyehQidVpHIV4ns5Oi8WpRyJlNLZtKYuCtmk8QpF43bmua5yEoBrncNJimxbXWpp2znw6Q1qf86iVItdZwH8UVDDIiQi3b93i2tXLHO4fcPnyVeYnU8piRKYN86pFbB97MA2GxJhrYsoRi+MZd2/d5urlK9y5cw9XW6xytBJwDlzwOrjKh57nFVlRoDNDOS6RxYIsL1HGoMoiVIhSwd0Y1L4whkZrqmYRpKjlnJJVYfbpHMJJqKTmOsZtraWZzXw+iUBT1d6Yq3wV7c985p/wG3/jb+SN+RRdjvyskD7OZPic9P90XNfNvSX0cH/yqXlxFg1dvOmT1YDRrGvfkB5u5iBD7rrad53qnRd17Qz1xJ6j9uf0Ya6n72W0JtM5ZrLhwWhdD5jSti3Hx8dsb293SU4dvFuQIFIpYhUNcwOyLEO1LGU2KvH6MW2DOKFpWubzOZOi9JBm1lIF+La2bhiXGcZk3HrzJruXr5Eb4+ssZAVvvvE6Tz32ONY2PHH9BjeuP86nP/4Jrl27zs033mJjY4tGfJLRYjbrOirTmpOjI8AnUW2MN7l3cMjlq9eCZ6VBlDCbTTmpW3a2thGBqvGqRVs3mDzD2RaT5bSmpshH1IuKbDKCUJovzvBMaVrxklGWZ0tu4Lg7x0jLVNJZ2jHDxyM7FR3kH4Riua1FOR/WrbRhUS2YzWbMZjPyIqe2FtHaB7etkVqGlGahDo+/E7SUuJbGVp+hZg3poWYOy96K3k047OTGCkXm3Y1LzCPx4axjLiLiw88YcHyrkor3PcpTjBZslcFsb9IcZ2xdvuF3r0YC3gRcurSXMAYvNub5MgitDwnWp3aU+H4moDQb/O7Z6hatQ2vE4doGqRfUizmL6YKNjS3axYLb+/tImPzVfOZdnYeHPLZzjRNr+ZZv/hBv3HyLRQi3PplNGW9tUdcNB/f2uXbjGj/783+L559/nuPjI0wGi+q4W4B3797l+tXrPolqOqOpa6bHh95gNxpRaMXd27f8e5mMk+kRe7tbnBwfclQfcDybglbs7OwwnkxCvsYcU+SMNia0YjFZDlaTafE1MgQaJ75QsLODvIl+TqTJbmnmaDqPIo4lxtA2NSbs/jG+wQK1bSmKknlVUSvFYl7Rao11PmfGSA9es3JO0W86KWNIpdhIQ2m2K0wU53eE7U83NBcljphJ68sfDje9ocF2qEqfRQ81c0gp2gRWGpmU6XzcaUfHGPfUgLUuMKYLaU1+XxVoE6kRsKIZb+5weHjIzrUrtHVLIz4YarFYrHRtxog8LQRX5rK9xGiNVv5TBUt6Xdde3EV1BYTn0ym0Dc10imtrZtMphwf3wDqaquXg4IDDw0PatuXNmzcxxvDLb/xPGGP4pV/6Jb7pxW/m4x//OI899hiTjREbGxtcu3SFk6MjDg8PyXTGqBjx0itfY2tri6Ojo24MPvZrH+e3/ebf0r3f4eEhV0fXyPOcg4MDrl69yquvv854PCYvC15//XX++b+4yWPvexznHKPJGJzQVDUmz9jd20OXGWO1gbQWqzzid1VVXq1WEoyG0Nq681q1bY8yFcdraHNIgWPj8W5cnEevcq3tJL6IG5plma8vmo24e/c23/yt34KejJkt+upXF5VSu2k6mA9pu5f+tpZ16zadx3Hh9/c7rcY45zoVcpVn5yx6zzCHdfokLNsH0smgE0PMEB7+Iu6duJjTey5dZxUbk12++MUv8vT1q2hjaMQym826QYw67Xw+74BQImnxhsi0klMbQoqzzCcwta3twojbRYVWPtTXtTXT4xPsfI5rW+7eOaCe1xwdHdFUDa+99hpaa3Z2Nrl8+TJFUfA9H/owe3t7jEYjFlXD448/zrVr19ja3eHTn/gk89kcZ+GNt97gzp07vPTSS7RJv04mE6y1XLt2jc9+/nNsb25hreXOnTt86StfZjabcePGDT73uc+RlyV1XTPemPB9P/CbeOLmm+SF4vbdO5xMp3z8Y1/g6ad9u27dusXV61e5JFd9WndA8sq0YjazbG1tLY1L9PQA3YJetYOnlEqOcfc22seQaPFqW8zvaNuWLM8oipL9wynW+jbcqRYold23x2C4U1+EqawT/dN39PPSn/dOqCcPPXOwQWQ0KhHVxCHK12Hw5DBmmSsrpXDKIk6tTImN9/JqxQojUvhvCQjVeBeXUmBboTCG2llu7Z9wNJuzlY1x4jrcgmhbiOJuvB94aHoALcuDqoxCWoe0LYUxiNa0dQ0iZLlhPj0JqEQLWltj25Z6XjOdTnnz9Tf9LphnfPt3fjsnJye88Nyz3Lt9hxs3bnB874C8LDg8mvHss0/7iFMHd+7c4YUPvMCtN/31N5640U3C3Ut7fP7znwc8oOrBvX3KsuRrX/4qTz3xJHt7e+xdvkQR0KeLoqAYlVy6dInj42N2Ll/irbfe4onHrlI1lmeefo6vfOUrfP//4bv56b/+13jfN38z+ajkYHbM8fExr776Kt/8Hd+KyTNoFuhygliHM857NLTCOW/AcyqB0V83tixvKPGriEMH96QxGfP5HJ1ntOJoxaGto6pbjpsZx/MZrx4cU+5e9d4sK2AMIm5JKu2foZb+T9t01t/L7VzNOE5LwKvrtyxvft6e0i7VFn2vGyQvSNE1te5l10kc68haiyHrdqOeYfS7qHcrGZzOcHnW1S2w1rJYLBiNRkvuv1j45kF83XGiV8FN6JzzuI95TqsqJNOMJxNeePFFNicTmqBrX7pyBSfC1RvXGW9usDHeZFSOUSbn8OSY7e1dtMm4dOkKVTXn6pXrbG1ucvfuXaqqYjwek2WaF198EYAbjz/GZz/9GbTWPPPkUzx23TORrZ1tKtuLuRsbG7Rty8bGFnXdsnvpsq+eJT5aNEobP/RDP0Se5xzPpmxf2aMcjzojbjYqKYuSIiRDxb70agHBi3H+mA4lzjiWMbszjk8c14ODA/b29qhqrz4IGdqUlBuKxoYAjgcYv4u619/JaMuLGiIjPfTMoQ+J7V9MhbJu0WDoF/Hyzt/BxnkU2JX3XuXS6ZJw7KD+QxcPECecCzUcDJVoPvmZz/Abv/PD3s03nXb67s7OTteeKAKvEyudcz16dGhHVVUo8eJzVVWI9ffY2tqiqirMqGBva5u9vT3qRYPBx0dEMXm8tUFb1TTOkrkcK47WNYzyMQeHh2xsbLExGVHXhvG4ZDGvKcoxSnsMhei6fP755/nCl77IE088wZ07d7CLlkuXr7JYLMhGY5+h2DRcuXKF27dv46TyAVzbWzTOglZMsozDw0N2d3cZZQZRPsgrz3PqpiLLci5fvoKZZIE5ZIjy0abo6MbunA1Bgliuoxn7Lo5rmm2ZGnu11kEKVDTBcFwHfNC6rqmqGkyGQ/jVT3ySGy98EFE5pwJs1syr1HtyEQjD1B4hLIc/d+09g8GsiqEY9sv9Mp63xRyUUn8M+EP4HvsU8AeAx4CfBi4Dvwr8b0WkVkqVwF8Cfh1wF/i9IvLyWfdPu7N1/QsXSQpv/F+S3PhlAyBebViVrbnGrSPiAwjS37SJOm2UJLyfvAJcJnzhlVf58Ic+SFFmbOhNH4psMpSK2YXeW5EyiHUTJhrNYkEdEUE7IS8EqZwXJLViIx8zzoswaaEceYi1jWLE/r17bG1s4zRkozHFeIyd11TWYcWgVYHIgsY6snzEojpiNvVG1OODIy5d3qOu50iWMRpv8dWXXiXXGxzsn/DsMx+gyDLeunXT7/zHU8rJBqC4+dZtlBZ0rskk87gS4igmI/b37/L444+zbRSxLE7lWjJj2N2akI9H5GVBPilppSWflCDaR1vivIegtVjjCykbNNiQCxkkgWGf5nnOdDrtoPGWxG3nVQqV+z4vTI4ZZ7R6hG6nHC9mqHJMozQLBa1q0dIbAdNszm6eJIs4NX5Gcom0Y5KmOpEQ5HWa0mel5JAO58E5t8RQ0gqtq1zxInIuNukDMwel1BPAHwW+RUTmSqm/Dvw+4HcC/w8R+Wml1J8H/iDw58L/+yLyvFLq9wF/Gvi9F33e8sAuv5Wv03j6TVcZMc8TreJzzsJ87J7r07FoG8szzzzHrXt3eezaJVzj/eSiVYfO5FwP657GRJxHWZb5gB0nWKsoyhIbB9YJ2ShDVI4KunNWetyFPMC/2caHUFsR8vGITHk7Q9vWWIST+Qn5oWF//x57u9uIOEwGSnlXaFmWzGYzlFLs7u6gjOJkdsLR9MjHNegNsskIYwx129LYllx5T8Ph4SHXn3iS/cMDJpubONUyXUxDPYtgrC1ysjyn3JhgSu/KdFoxyo1n0DaMg3iYPksA8FUgOl3oq9W1qNJFN3QEtTXGoJyfI9EI6RQYk2FxNK7h7uER1559P08+8yxzMRc2RK7ynK1sW/J9iAh1IRXggoLAug3wvCe8XRNnBoyVh9idADeB3wL8TPj9p4B/PXz/4fA34fcfVBeQc9IU3eXw19NiU6Shu3Ol+5NlK3JMEz7Pktzfw6f7NgJNZrhzdMidk5NuAu7s7HRGyXjNEKglRlKuo6hjRwaX3quru2E0ymiUMYjxmY5ZnrN9aY98VDLa3CArCx9ohFC1DeV4hMoUJteMx2Na8Z6IKBHlueHk5KQTjZ3zLsOD4wOyMkPnftrEHBKlFCaA7Y5GI+aLBXlZcPX6NVpnvVuyqZfE58q2mLKg3JhQTMbk4xKnFbrIQxEahVKmy3uAHhfRw8stT910jIeG6XQnj+N8yvhnfGyBFYeIZVov0KOSv/wzf4PR1m4AczFr59KwLUMsz1UUa43EJKrhnE4/Z9H92hK6dz6HHpg5iMjrwH8BfA3PFA7xasSBiLThtNeAJ8L3J4BXw7VtOP/yOQ9ZOeiNeDUjdn7U71PYseHAKHGIbXFt031sU/vv7mI1Lv3EzNA6Q0lAntKCKsbMxPBn/uu/zK27h2S5L2u/t7frreJa0SaAqdAbNSXsiLEykwkVszrjm3j4tbgoGnE4jT83LzBZ4ZGOjSbLDMV4RF5kKA3FqMBkislkwsbWBF1kmDJDa8hGI0aTMbow5GXmi9A4R123mKLsIjgXTY1oxfFsSlXN2dycsL9/F62hqubM51OsbTg42ke0MKtmSKa9dJOBlZqs1LRtTTYZkW9s0Cifp1KOJ2xsbTPe2CQvx52x0veR90q1KkhJK4rORvRmEwq29GYpST7Lm0BXc8JaXNOSRLp5csKoKMnMiEtXn+Crr97kZNEG3IewkIOnZNXijfMvhrwr5RPrrFpvrYjnpRufFdd90k1uiek46expKumPeO5FmNhZv78dtWIPLw08CxwAfwP4HQ96v+S+Pwb8GMAo1CQ8j0TkXBHLCpDsNs65oBQsG2vOkhxWdqRoBMvcCT/8u38X1mQcLxr0WMhC3kQUW1cZprTWIc6fRHyGNH05fVY5mnTgJkopFD55y5gMq12QKLydwzpHlpc0VU2mDFlRdDkFVmKEnc9RyMsMWst4XDKfzxmNRrRVjcLSNLFWg+HWrVvdZB6Px1RV5QvfZAYnvoq3bSyNeCQsXx8jo26967AMksXW9na3gMrSp3Ib4zE122BwdjE9meUQ6FXjoZTCRoPuKmMcXvXIYk3Tjjk7nO2BbRd1Rb2oOT6Z8+WvvsL3/6aPsHAWtEHh0luHrNpw71VtgQBEKCARo/Tr542IZQvSO0a7w3kS8IXu/+BN47cCL4nIbRFpgL8JfD+wq/pKHk8Cr4fvrwNPAYTfd/CGySUSkZ8QkQ+LyIeL0bKVOXLOVeJjcv1KlaJ1ULeOxvbFYa2oztB5yli1RkyNpFQ8ppFQ0m3viffx8c9/GVWOOJzNuwm9WCyW4vyNMRweHiISsizD7mdYFiWHPm0fIqt9yKwyCD5LVJQBpcnykrwYofMcneeYosAZQZcGySArRhTlmHyygR5nZIUB5TwcnhZEC/NqRmNrjk4OaVyD4JGxstzgXEueG4oiw7kWcGRZkJ4yhcp8BqtowSmHU4JkvvJ0MR5hc83CNmxsbKAyQ+Mspsi7nbgVX9sT4yWPiCV5EeOtCdgVYh2utbjWeknCCTjp/p6Mxv0xBVXb+FKBiwVVU6Myw/7RlON5zT/8R78EyizXRaHfwSMji+pBqiZECWMIIHzejt79JhpEh4zKWFd0eaHHZ8eSfxYvhcZM36E0Mvwo5SXOdfR2mMPXgF+vlJoE28EPAp8F/hHwu8M5Pwr8XPj+8+Fvwu//UB5AWVr1ssDSC6dcM42IS89LaRVTOI/SZ4korGi+9PLXaNHM6sZXXKbfnSMH11pzdHTE9vZ29/cwtn74iYbLoduu07+z3NselFdNVJajshzRBmUUJjdkZe6xJUZjilFJXhTo3GdUpn3VOEsbPD9WHE3tsC04q3BiOZkeM1/MqJuK+WIW6lssW76NMUwm4+CC9DYRjI9qzPO8q7aV2hOAJTfuqvE4bzPwi6r/PaqaqZ0hLf1nEXSeIVoxrxbd/crxhDdvvcWNJ570BtAz7E9ntentke4/0n+W3jc0KzKkrzc9sFohIh9VSv0M8Gv4KnIfB34C+NvATyul/m/h2E+GS34S+MtKqS8D9/CejXNplUFnpXg5EKf8tS0OjW1DoRLlUOQEQbA/D484vb4NIS9e9QZBichUCh+tqTRzZ7k3bfgH//Qf8kPf971Mq4rRqKRUvi6F0RolsLO55QfXecv7MG+kh18XlFgvvhpfIGbo7dBaI8H+go4xAD5QKMs0OOO9GtrQYhHxKdRoh1UWMQozynFNi2DIshJdg7U+KQzj27RoF6DAietS2JXu4dVVwHgEyIoMMuPVI6OQEFMwKUqUMihjMLEflX8vUX1hH43uvlvddqrX0nif+i4o59Gvm9TWlBgjYyh7lwjlhDbYnGbTGvSIRiy1rRGjeer5b+IoIEq30qD1sncpqqRDlWfVJrWq7cPjywwlvoPynhkRP38luVal1pVobIhXLd9z9Zo5XRUspbcV5yAifxL4k4PDXwW+e8W5C+D33N/9l64/ZRdYOneluUf8ZJY0bDZ21rpnLnfmMrNJJ6jXNv0iB9A4HK1WHM0aDmcVxZamqmrGk01i+nm8T5QmjPEgNnGRQQ9uKiLkKrVTOFyAaosxF36S6ESYpKsloZTC4Q2i1jWIUrS2pbEN2IaqXkCo6jQ9PsHgXa/iWnKdY9NAnswv0LZDuvJTMtpSUL6UnM4MyggKH/jUaqHIvbtSxAcxOZEujdiJ7wvb9ghX8R2VDn3mhExpH0wFXWRkKlEshbmvGM+2bU8t4ljTM89zD6Irwslsyt3pjF/93Gd537d+B602uGDH8WPuTs2NVYwgfdZwg4vG5qGElHqjhu33/3ujsbc1nWHhBJA+OStuJkNm5IbG2AE99BGSXy+Knft2pK90wFOVRYlfoFYpnFKo0Qaf+Mzn+L7v/CCjzNDkDdJaRqPyFDDIuh0mUlwwSimcbdGZCsxOgfRemfTatq76eysfSYhIb3F3HnJvsjFiMZ1hGyErcg+C0noj4OHsCI2hbesOpr6e+yCpgyPvoo04FhjtWaOCQhWo1ksNeZ5jTEZTzX17chNcCm4pMCf2a8yGVFrTtLWvWJ6ZDnxXi9flY8yIj5dwSyrjWVTXNZPJpFs0h4eHHBwcsLu7i9El9+7dQ5U5n/n8F3ji2eeQPA9ljxQK3T3/naK3r4rcH73HU7ZXW6lXUZuEUvfkQ3StcminQPzk1Fp1iVzgDZN9ElcUMSBDQBxIzwQUBkRjqRFnggjvjV2iMmZSoZxw7/CEae1YzFtGpqQ0PTycUgonbbAlDHYhAcSDw1rnaFX0XSiyBBfAWktrazQKrYqwU0S5PBg/AXG2i+Mwqs8w9RF1GlPk1PW0SxnIgyEzN5lP/3Z5xwjzUJxmtOGzM0UpsrwIbtiw4xsJ0Y8apQyNa4I7r0FaF4yqvo2p/cg5n2FobYu0wQsQxHjPEKy/NqiQWZYtua67GTOwA2RB8jIoNseTDm7NV9wStnZ3aJylsnOKoqTOJszQ7G5ephJDRmQI7pQnoMMZFW+gXhdvk24oMVFsVbJWPN+Hdg8XrndNrJJM1tFZYdvuTLEjfbv/P6ChB+AiJMGe0H+8buzUIHBJdGBCeOBZpZkBP/VX/nvy0YgW1VmyY9bfRdoLq7Pt4nt076KiVT8WfgWvs7queK5zrcdCWCG2FkWBMhqTZ95YWZaMJmPyiY+FyMuCrZ1tTJ4xmow7/MdYx7IsS2901N69mZUFKOftCIGGlb6HuBkisoS7mR4fqg+RwcZ+itLQKkN1SqcyGkM2rhPF4XRGawz/40f/OZMbNzjGMVcS0J7PjxlIDYirjMrpu561aHuV8Z2lOH/OWg8PueTQ0yobwKkotzOufRBp0Ls5wycsNq00ggql3bLQHl8KzSnBZTmNBdGGZ77pm7h3eICx2+SbPiYgD3UeoLc7RBKRJSPX2riIwWQTaciygqZpmE0XjMZFX/G79dBu3VNC8I7gaNoGcQ6nHDrXHnEqusy08piYueHo6IiFaxGjafGL2BR5N/FbBLSEato+IjIvirD4Apx3J/XEqt/eKOxzYnxinXMtzgUbBtFWIJ3htVPjku+95NF7pVLmq00fjDT0XKEd0+MKnRXMWst4K2cqMLHap9SHfopz6EwS3dvIBqcu2w3OBhFK3Y9x4aYRnecxjo5pSj+HVj3LXYAJvSeYw1A0ux9SEqoFPQB38KJi0NOlDYwiLF4ddGelENWEZwXVwYwAw6Fu+Zm/84v84d/7ezis51wfb1LPF5hR4S3+1lIM38u2KBVrStIxEh/t0nsqvEjddousrmceJLWdomu/wDwsvunUCt92Dw8vysO+S/DHowyis8SA5dUn19aUG5s+K3Q6Z2M0phyP0YFZxkI2mF69MEphjH+nlrw3ahov3fgdywBRygmVo/BqTZQylIA2PTO0tCijUFqwro9sheBFkeBDCkbOtMYmhL7UyiOEKcXJbIGojP3pFKtzfuWTn2Fr9xK1LrBKI85hVBY8KaeTrJbnogQ0bHvKUJjalVJPRkwHUoQgvUBDA6IH5bVL73O24TL03YC5pEWIL6JWvCeYA1zcWLOKS57HUNbf2080Sdx3MfsTyfyoJi44T9q7T51jbhuuP/UUf+9//Mf89t/8G2laHxabhUkfB/l0m1dLR/HvVAx37bKKM5lMVvaJh0HrJ7YxJSJ1Z9z0zz0NgmoKH0OR4xOwNre3/KQTi4ilyExn+Oylgz6Qy4UU85iHEa3yncttYOBNF58xPk4jeiJWWdyHUoNI4poKlC62tvXVvqq65nj/kKLYYlE1/ONf+Wc8/23fwdHhYu0+MszJiC/xoPNr+fhpe0WadyNIZ2MZSo+r7p/Kyp0UeUYezyp66JnDRQwv6Tnr3Fn3LXGoiBGhuu+dx0MpfFCB34l7JuQjHUUprNJoMcyVsLd7mcrRlXzz9gDPxR0subDEeWj5uKCHbUq/p0hEp/EE0/foPRpdW8nJMnA0F9Cng46aZ+gAkov1BloRwWQ+N8M510XcxT4vQn3QNDbDhXiElFa1O322cy6RpPz/MaU9nh8Xvw62C18b5DTAbLzndFazc3WbT37q07z4oW9jf16jinLpPOccOjuN75G2USnl1U0X8SbWv8s6StWJlXM4YUSpejK0GXTXBuly6B7t+u9fBYPkKkNTumOsS60+z7ATQ6fjZrpOTIttSEnh9e3GWaws72ZWDEoEZQpanXEoitePZ/zkX/lrHFYtrXgRsmkaBIvWYG2FrzbdgOrTilPR2DlHI47GWh/+m7QtRv2lbU6h7/sdQ6PImc8qrzJhyIoyIBinUyEsRu0RuLO8pBxNuHrlOpPxJmUxJi83yMcT8tEm+WiMMhlogzE5xuRoVXjUaJWjTY5RBT6ITBGDyZbGMkRqNHbZ2Djc7YwxXYh0HKd0QcXoy44pKkdDS+MaamkQsTS25tbdu+jJhBNyysuPcdxCpTQSdmUdiu26EMthQ2h3+knnoXUV1janojKj1NMxZRfu7YL3K3xWzdfu/qJQor3aGmKzxYLCdPVjh0xlqMZ0hltROK9L9p819FAzh2WRNz2+uiPS31LdbtU190Pp7utc6ytYr22zJAOtUaagUYZrT72Pn/xLf4VZK1R108Gmt8FgyMB1VRQFdV139RTWURQxU9H7LClCRBiPxz3jkeVKXPE8kxXeIxEAWFpnu3DnKK1olYUgLg3KkOUl2uQo7StaKe0Nl9bKkmifehjSZ8YyeWjVJV5dZGzW7cxL9UyCfaV1lul8Tts4KoE//Wf+HK+88SZzJ16yewdoaU7Gdg/OWZVNueozuOpCz0+zl++HHnq1Ir5UFJ3XRcJFSqtMpfrZKrUj7fD4jOGz/W/L1ZsV0dI+rFvgA4h645gPl7ZKM9Ug4wlmY8yibcjxRsJ8gP6T7npxEaW6rtZ6Kc04Luyqqvpw6vBOp3I2EIpQVi/mBzrncK2Q5WXCYAxKG2LEgggs6oaNYuJ3MZ13/SXi3Tk6C+I7gApCqwofnZEbjZU++9HRogITUF4WB3ygUj4qsc5179m9TxCRO1fuINdgaLT2/ZEFSctHsp7MFrx1+w5NDf/glz/KN3/oW2A0oRaNVSG2JUVRUip5kfNtX6sk2WU7wWopd2gnOY+8xym5/gLt8c+5+DMeaskhpfOg1SKl4ly6UFZdmx4b1sQYTrilRJhAq6SaoQQAvihJqzWT3R3+6//mJ5gtqsT4tnoIYoht+v5nUQSwXWWsijEJEUsxjfkwxmDyArRBZzl5OeolBG08o1CGshyHDFCNQ6GzHJ3lXpXIC7wnZLl/Yh/led4VJI6f2Nau15RXCbKyWHp/Yzzgi8lP72NnIWmlxXEj06uqmqOjYxSay1ev8v5nn+f69RuUZenBXKJh9RtMAl08xTty/weUmh9y5rAcEJPqbxcRv1LVYih+pvdK7RqpCCYSLeU+ClGRg2SdJNE/qw9+AbDiveNWWhqxHpPDaWrJuPrEU9y8dZO6bXCNw9peJUjDgIdWe/ALLU56//GLNfqy4/u1rf80jYA2oA3RBKh1kKq0okV8UTDt06pj5WttMu/KtA4nQmsti6qCzKNNZaMSlWeI0ZAZWpYLpmht0NpgTIYJ7YtSTScNsQyc4hCP6ZB7RpAVOSbPUMbv6A4fZGaDfSIu4qF7zxhDUYyIqe1+s/TG46qqwMIrL32NG9eu8Zt/8LcgWY7FB7H56FQfswJe1I+Riiok1Z9H64yK8e9YB0SL/96uyKjs7BgJgNFwrkbw34upCz5a9bwArSE99GpFpLQTzrL+pmJUByE/8E/D6sCUdXaM9NgpMQ06K3octP6COJk8krNHQFAcHk353OEX+I4PvohzGc55q7vWGsmMD2AJhsdo/ffvkVj9syR+IXSHgDc2KR96bbIMRHfXemyBYCvQKmFGQXAWP2mtUWgtGPoq4EUxWqq9EfMT+oIygo7SjupVIq39+7imJSJI+5y1PjjHOR9clhpQfYJRPCeOuyMWDknVi9j3qbSQLlKllAe6bVvefPNNnnvueapFw93ZFIs3TBOqt9uuJ+n65iK0br7EsUs3KN+u5XsP5+k65nJaRfbu9aFK5SSVehMmcB/qy0PPHJZF+/7YWefD6XiHi3R2SqtVhvXP7Dl7f9ynDKkwn1tv/MsL7u4f8n2/4bt5861XeOZ9H8BRd9e04rxxyrkONWg4sMaYpcHvQUC8fV3rjCzTQRLoLf7KxLqKfQSgUgrbODKlUUH/V85hdOYlp8C4vNTiUa7btqVpLabIsCJU9YJxkXfnKd2HqUtETjX4mBAVFn3IuYjXmCBddFGQyW8xxoRw73h8qbrZCrUr/t62LYeHh7x56y2uXrnO5s4um5ubyGzfX69CKpgQYs3On1+w7A68yJqLC18rfYo5DO+/yk3pnxMkrxCo5p+7IrZnjXHhftSLh5o5CF5fj7X+hqSUiqUUlw12LKsUK+89UDWG56+SItJJm3J4253b73IQd2oB412cgkaKgm//4LdCZrhz74C6+TJPP/UExvi2KxcSprSisdZHd4pCrKaVYM0vCj/xVQgVbntPQtt6Q1+Wa7R2tKagcb5QrCELSN2mm0++v5rgufAuTS1BT0/sJ15CiGnAivG4RGufIeqzM/sxUajwtwPlU6XFCJjeDWckD56YUBezbShCfoa10sc0pOMU3G4Ob/T11dAAnfmWChhtOkOrtQ4rltliTuuEW7fu8Js/8tu4d++AVgQxUCsV6i27DtH7LKvdKtUilW5W0XDDahFMZkh3kqHLtgeZA6RlWVoIqe7KnWIw3Xkk4dxrgurOYxQPNXOIdNYC97h8vStoSKlINzx+kfuvO5YyiVVqyyqyztFaYe/yZVp3xPtfeJ7DgznHR1N297ZxribPe1E9emd0EMe1yajrljwvQRva1iLa+DWT5TgRTOnL0ulQYLapm873j9IYo8PumOxAJuxkSKdj+Cx0TZ5nPickC8WKtfIgNbLch6uqT2mtA3Zn0ldaI9Z2UGpKKaoQ0xEpprWnnqpUSkjViVSF6L06/u+yLLlz7w7z+RylFD/yIz/C4cExxhgaazkJaOExixU4hQcZ2xHHfJWU6du0fqGt2oCGNqbhuelzh4WQVs25FIZQRDAJMM156vI6esiZw+l4hZS6BRrONe50vv39WmnPYgjDY+ngXlQNEQU6M2iryUYjtrdzcjTVoiHL9RJqcay1uVjU5Nk4kVD8XpBlBUoyyIMYGvIQtPbh29AvNM9w+gVog1Dc3VHFlG9BXB9hYJ1jNB53EoeXWBR1VfnK34P8hWF/dXcK7+0jDg22aUOcR3NKiksXUzTWdlGSa8YjGmu9PcSPzeHhIa1tqeua97//BQ4ODhAXGYsPHmvbFjFZryKs2YlXvVt6LNUAVl0zZBD3QzHSMV4/VJOjxHhW7sW6tp1FD7m34nTFIFjPweNOtIpWWWmHHT38W8vFuO55Vl//u8JpQyMeR1FlCrKSja0JC6mZNfMlu0UEFL1161aQJgwYTVYWHWhpuphSL4a0grRCpjK09ujPIn73ldbDzlnV102wCKL7vo7/R+9FrK+hg2FROb8rR3cx9EwofWfnoujf20aiazSiZ0fDZq4TAF56V6vKDCrz7161DfO66gFc0w8qaAQa6zSLyrKoLLduHbC1s0c5KX2fEz02jvF4vHLcl21Ip/Es08+wAvuqxTucI+l1F51Lq1Ldh+pu2ubeo9VLbWe1aRU91JJD7MT7wWAYXt99H/AMpdZU344QZIRaA8LSeZ1x7wwmdIoGp25sbFAd3EGHRTHZ3KSZLzieTmmsZTweB8xh2NnZ6aSIqmkoyxKt+gCYtB3d7mQU2uhQRUkt7bppkwTPALVSYAVRsTitn1yxxkNcqFYcddswGo28ahHO8eMkS7ub380ab5cIunE6jIImyzPaYNtwRtHW3vaRZTGtW3eZi9Gtt9TXsOQd0kbTNg0o3btfjebatRvMZ1Www/SGzFWq4HC+3Y8x/CJ0Uclh1UK+n79XMZ91Rs519JBLDvdPqwyKkdypj49xX3WOFe+DjvUdlFvmsqt2iHV2h86IFAYnCy5GCTHueV5ycHhMOZ54T0CSKxF35MViQZkX3r0pw3vrwce7FpXSGJ0TIc7jxwVAWu8VETJRna7duGW/elyUVeUDt0YhwnI2m3W7fuqiA7qQb2MMrhWUaKqmwbaCs8GDgQ4LMQPxu36WF5RZgb+Ntx2oYNlvW9uVsYuu1FZcFyshWnVAOrNqwcl8hs59BfGNjQ2yrOgMmsM5EnfbVdGNZ0kDsGwAP08yhdVetHVqUseYOf0x6uylu+7ZQ6noLFvZQy05RPK6abv095Ak/HMWV9b4he/FUOnszkuLPj0mF+Oe56kegge6VQqqxZzSeLQoTcCcRPHUU09xdHjI/v4+V65c6fARm8bXq9wI0GyxdBqAbe1SrEEn4mpvfPR1EyyIIJouC1QUqNaDsIhzVMFQqDUo2yLkKNX3RRf+iw4JYsLGxjgY9ELx3tEGIpam9clO83lNUWQYJ9jKkWcZ0lq/kAGUonUtThwHR4dsbWwyKktwQtM2lJOxtwfQxzDooIYsGl85O6JJt20bDHJ+x1wsaubzORsbGzz77NPM5/NuBoh2vn8smCynFUcjDTp6QtYs5PMonhkBcldFqq6yqaybN0NX5EpJIZkLF23rRQznkd4TksNF8tDP44Jxt+1cn2ukw3W653li3EVpaGVOB/XSpUs888xzTKdzqsqjOxWF3/GapqFpGhaLhcckaO2SQbCLsgxLOo1YVKYHlhXVR5sivSqg/EuRGYUEBiDisyucbdA+MgtEcNZydHRElmU+FyLPAxxdS4Sqy7KYZu0rjC9m8076imMVDa6bm5s0tQeUjVLKbDbzmavSqy9RPXLOsVj4OhNRmojfAQ4O7qG1ZmNjg7qumU6np8fBRA/I2Ul0D0LnwcCdp+tfhL4RYd7vCclh2JFDJuAzA5fPHbowY2Vm/1vPHCwEv/wyQs5Q/HNOunqNq8TSdW0FehchntsfHh6SJ6K4hJ9t66iqmqtXr/Hqq6+ilKLINNViRlEUXjwOVbSHGA1dnySiY5QmIubifD7vithoAazzMQ8R9s4YZtOZL4XnQKsckYbMKB9GLBaNr66lxfdprg22blBGdYWDtdGdq9M5H34dmZjHWFi2vEcwmDi2i8WCrYDU3bhenTAeXgpg6V2iZ0IpzcHBAY89foVRucHOzl5nQ1BKdahVKohFK+NakuG7qAfqtPtxGZBlXSzNWXS/zOOU5KFW//avkLfifIsqLEsWcfcU6ZmBqGUXZwwTXkWp/rycZ/H1GdDXXvkak9G40+eBgEXp9fDJZMLt27cZjUYcH3s//NHREWXpi9seHBxw+/btrq1RYugkAOnVrighFEXBfD5fSrzqrN9tXy6+aZoebh66YKLFYpEwH98fZVn6361FnGM2m/oCPmXhCwQ5h4S2TadTnHNMF/NO+kkT5IwxmCyjtj7Eemt3hyzEZsTU8uh2jEbSmHZ+dHTU3b+ua46OjtjY2GB7Z5O6rr2tpiyXLPdpyPVZMSpvZ3eP8+ZUcNMKe86qnKGHgd4TksNFuK0PBwbwtgKLoJXuipielaqaunlO1TYM/yh6JnSW5NBdl+zmgkOJIzOK2cmUZlEh4t1/AmjtQVMFaOqWGzdueLDY2YyjwxP29vaYzWbcvXebPM+ZjEq0QD33E9+GidahLFkv/0znM5TAYjanDEV0jfbVpFqxOGfJtemwL1MjmEEQqXxbneN4fshoY+Kh6cShA8jN8fFxgKZTNI23P+jM0DQNk8mExtY4aXHtwgPsNN6YWBZbNLWvUyFWaF0IZgpRl1VTdws39qUKNpLZ3Es3zgmbm1u0bcvJyZTt7W2+6Zu+iTyb4Kwmy9RSQNVisWBjY6OzUWxsbJwZkxCvS2mld0gkJGs5hD69XonHuRiqkevmcmowPJeG3rcLJIXFtp7XjkjvCeYA9ycORbqIDQIGi/ucx5yrQkjvmVjVnpdeus3+/j57mwnWQny0UkimcQKz+Zwsyzja32c08lGPs9mMK1euYIyhrr3RTUTIyoLZbMZkMgkqhN8Z60XVRRemngVjDFVVUeZFJwZHl+RisWBzc7O7LsZPRH0+vmPbNqFoja+OXRQ+UOrg4IBLVy6T53mAcZPOmNosKpw45osFeeFtAgB13UCi2kUGFfEeY9vqaoE2pouxiB6M2WzW9dHm5ubKuaKUYmNjo+sHaW33N6kKcPbwL3k0znNn+5/1UjUzOslOdYFX3vgdLcChaFGSMqCTyXreOrgvI+o5574nmMP96n3vJK2yDqftS+MgOski2BwEYXMzhrp6XdgNdiob5sj29jZtiGs4ODhARHjs8et+53aOW7ducfXqVSaTCdPFvNshsyyjWdR88Stf5urVq+zs7HTMYTGf92G2ylf4xmTUTagCjlCUWafjt21LkWAq2rpBCyzmc4os4+BgP1ngdWdcTHdfrTXHx8dkWUaeZ9A0aAWZNlTzXl2J8GzDrNqoTkAoade2qBDKHZlDlmU8+eSTHePztpL1u3N8xng8Oj2ugzyes+be8m8SRvp01uSQ5biQCLK8iydSSBBXO7uZPg1WtMqytY4edG28J5jDRWjVIF402MSqixufIqWSgU0fk/WZhVHQ83iNCpqWrUnhi8vg0Y485kDEOQiZnMrhjGM2r9m+vMfJ0bGvFdk0GKUpi4LrV7Z5863Xce6xbkHG95nNpjxx44Z3hzaNj2FoW7QIulsEOsQKtEzGXj8vitzr6dUiQZhadAvJW/5P2N3d5eTohJ2dXeq2WbJheO8KZMHN1lqfXdpU3qvhWsv2xiZ37tzpdnljDPsHBzz22GOcnJwAHmOzLMtOesiyjHv37jEaj9FBKqnrGmst165d6yMqk139tJEuZKGKwzi4ceMarnGdqxdCxWoXw81laV1H5tXZd1yyuJPM2v553WwZ/jL4fXWeRfzq6G1kQ+Nm/52l64eqUWdLi0npF5jq7wnmMNQD7+e6IZ0nCg51smhzEOl94KdsDeGW6yLQxHkodmcXfP/3f39S2OZ0O722D+J8mLWzvkiM0l4VEOt158koY3Nzk6Iouv7Jsow33niDcTnq3J3ZKOtiAYqi6Kz2WmvykJzlWl8qr6lqj37d+vJ+EdYNwAYouvFo5O8bGIlBMZtNmdcepm53d3fJ6CYiVFXVPbcsy073XywWfVAYcHR01BkLR6MRVVX5dw6Lsmka6qbh8rVrvPXWW4zHY5588kkfjo6BWK5wBaVMQ4teWjDWWg8IE3/XpzE4U+9XfC+jevCaVfpILzmcvRLP25g84+glMQixM64vjLTKMzJ89/ulh95b8U7RN8YyHHcMH+0XF+S5IazikXsyU5BlGRubm4yD3j6bzTDG8Nprr3W+/xhRGRfXzZs3sbYXueNuHC3jKUSbiHRW/RgFWRRFV5U6Lsw0qSc+r6oqqsUC27bdDj+fz5eiKmN7tdbM53Pu3r3beU5ibkO81jlfILdtW27fvs3NmzfZ39/vAp4iA3zttdcoy5KtrS0ODg6CLSUGnJ/dt5EhOOc4Pj5mOp0uw9Ul8TKpRyHtCzgNWzj0NnwjPA5n2be+HvTQSw5pR3897Apdh3Zl3PvfVnJggg1aQTSaRUNSNDTpNNtxyXWlaQxsNtAaB3XDj//RP8qbn/4EitbnNOA6kVApD1NOCOvOtMZhUE4wophsbnN4b583b95CtCEvRxwd7uMQJiNfbq80GrENh/t3uXTpEtOTI2xbk2eaum260O2MnPiC88XUqwOuJc8Ms2oBoXCt1hoUFJlm2tY0Vc8glFJYwIqlyBQKS26gbRYcz2aIVmSm4Gtf+xrXr18nVuFqmoZ6MUMZg3NtqNg9om29hybLchaLRZCwHCcnxxwfHyNAVdXYSti6tMfWeIuNjY1OXx+CCZ+2C3lvAtKSGcHUFmkdxvRVtgQBG1O4DSqMcXT5KvF4mQKorO2zOSO0HD6lHcCpPuYhpaGksCpse+l8PLiPr4lq0SrzUHwmGjN7W8eq9z+lSojucojOoveM5LCKM98Pt77o9fETjXhDEfntkFaKN95449Tx1IWYHtMEgFdtsEgnKYA3WH7hC19ARPiXv/ZxDg8P+coXv8SdO3d48sknO9345OSki5GoqqrbBeP7RVE/iqfRzZdKOTHIKF4XJYwYRyEinTTiq1d7d958PifPc65evdq9l7WWra2tLjgrz31R3qhqjcfjTt2IUaEHBwfd++R5ztbWFlevXuXg4KCzWcS2rpPKhv0b50S89+Bs+noO65dImimZ9sfwWHzW10OySN27w3e5yLH7oXOZg1LqLyilbimlPp0cu6SU+vtKqS+F//fCcaWU+q+UUl9WSn1SKfVdyTU/Gs7/klLqRy/UuguKbWd9v0jnDEXG4fNWnZtSKnoOz00nyWI2xzYtmTEr77fWtqIN1njX3+bmJgAnJydsbnrD3qQcMSlHvPnGTbY3t7h16xaXLl3ykYZbW8By0lecxGmtCxFZUh0AqqpisVhw+/Zt7ty5c8oPH6+NjKBt2049McYwmUyoqop79+51HgulVPf7fD7vAqJikeHXX3+d4+NjNjc3l6puF0URwGMLnnnmGfI85/HHH+/cs+vGaxVTiOfPZjPKslwZ6CYC4pbve9ZcPOv34f3T48PN56w5N5Siz7r3quvvly4iOfx/gN8xOPbjwC+KyAvAL4a/AX4IeCF8fgz4c+FFLgF/Evge4LuBPxkZykVonV953QCd1ekXHYhVx1I//FntjOc34jBWqLTPpDRK2BiNPU6EClmhIbAnfrQQ6jF4FOS4U2jxmYlZlvPMc8+hMOQm4/VXXmZjvMkbX3uDb/vQh3C2Ic/gzu2b5BnYZs58eoht5hgyXCO41iK2QWyDEsv+vUOOD44R61BaQBoULYoWcTXbW2OuXN7h8OAeb735BrPpMVoJWglFblA4mqZCxBf3LYrMF+uxLdPZMaNx0WWbxpDn4+Nj8jzn+NirDEopFgvvGTk+Pubw8JCDgwM0mqZq2BhvMJlMuHbtWoiv8Ax26KFI/15HcXyLMkSRdnEIGuVyOrekCB6x2cPeiSxLdktjL3SVrVfNm7OYyKp7njc/o4SS0lB6OYspnAFc1dG5NgcR+SWl1DODwz8MfCR8/yngHwP/UTj+l8S36J8rpXaVUo+Fc/++iNwDUEr9fTzD+avnN7HfoVKuGY+n56S/pXReeOw6yQOWQ5GH94znpkYq7W+yclBu3rzp3yWeq71XIkY39ju46SaV0ioEOI28dmu81+Dxp57i5PCQwmTkeclrX3uV3b1nODo64ujkkL29PYqi4N7BfmcMVBgPCttYGufIc4/jKNaye+USi/mU46MZ5TjrJphXMWA+nzKZjNjYGHeRmDE4aR7iJ6LePp0d914A53ExNycT5iGwazwec3C8j8NX9rp58ybXHnu8C7Ta3Nzk5OQEay37hwc89dRTZFnGpe1LZFmObVUXFr5qLFfhMaRjFZmIbVqaauZrIuYp9ujyHJGIVYnyxXiUNzAPd/EhDTcLGNRFTa5JvSfptcvXP5jXLiUR6Y0P59zrQW0O10XkZvj+JnA9fH8CeDU577VwbN3xU6SU+jGl1MeUUh9rwuSD0x150ZyHdb+vum94/tIgxWes+zu9ZtUz0t1ja2trSaePkzStS5G2KU660WjkF2Duw61Fe9vAZLJB6+CrX/0qk80Nbr75JmjFtWvXePzxxzk5OeGtm2/y1s032dnapigzfMq1oswLbNNSzRcocTRVjUYxGhXYtu2yNGNehcdx8JLBYjHj+PiY+XzObDbrMjzbpgERmqZisZihtY9XyPOce/fudYv23r175FlJ2zrquqUofAzFyclJp0pEr0o+KsnKgnIy7gKzhu674fidNSbxN601GYqdjdLngaxRCxDt8SfC/xF5yiGkdStXFT3q76O6z7B25nAupuHi50nAF6W1a8UJuPX3edveChERdRa65v3f7yeAnwCY7O5IOEb6/3ARDRdrcq+l89Nj59HQkLTq+mFcg4ggSVoyCpTSaOWwtuGt115jcXyEuAWZKRDXoBS+SrUIWilsCJH0yEstWjJGZYbROdZWff6E8/Buz77/OYwx7OxucffWbYqypCxH7O8fsntpj/2DO+ztXsK6mrZuqKoTb+QsJxwfe7DVnZ0dZvMTjAIl4tUaZ2lqi8MFYBc6e0Gee29ClmXM5w3T6ZRLly6FXtDghKOjIy5dukKWw/TEuwsXsxk2JHdZ6KIfNzY2OD6Zsrm5xUsvvdTZFzY3N3n8qSdDdKXHXlDG909uMkQtY3jGT4zrGHovCIWEnWvRqsDOD9janHDctGDKcF6fTp7OH6cCelSCeIUzPZw+zhsxA/T+8vzrVUgkhIkrh9L98lPdZn5aKu7/lvjHkrSaUrx+yf2MIYZmx2su4v58UObwllLqMRG5GdSGW+H468BTyXlPhmOv06sh8fg/vujDztv9H/TYUBpIUY7XPW+oI6YWaX9SsmMA1lmcErSCalFjmxqKXryLE9Faiw0MAjSLqmI0GqFc8AbkGu3y3u6hdUgGc1x/7Aa5yWiahuODQ7761ZfZ2trgKy99mSuX9phOp4xGoyTmgE7En81m5IVfCK214Dz+AsCimkHYoT22go8zWCwWHSLUfD5nNBp1EkJUkbTWHdaDD7rKfWh3PE/58nd3797tVKA7d+5053/wgx/0bUr6tityM1hIqeQV1Zk4vqsiD5Xy/Z3nOY/feIyX7s5oEneiiHQh2x1jUeDjVhJ1Ak0sqhyDqDyDEM7CiYjnpfPGDBjK2svCvInvOtwYlxhacv8h+Ow6ySulB1Urfh6IHocfBX4uOf4jwWvx64HDoH78PeC3K6X2giHyt4djZ5KiF7VSOstVlFIqgg2DWYb3iX+nO9BF1IWz0m3F+rLtrYXaCe9//7Pd5I0Guvj9zp07SeCRJc8NbVt7H7VyiCwDmToxCBmODAK+5Nb2LqIMGzu7PPHEU4zLCb7ATcHx8ZRLu3tsb26RaUPbVIhr2dwY09YVi9mU2ckROEtVLTg+PmKxmNO0NYtqTmsb7t69G2ISxl3MQrRbtG3LbDbrVA0Rv8Peu3evS8gSEe7cudP19Xw+5+joiHv37nF4eMjJyQk7Ozs8++yzniFmGePRBsdH07D7ZRhdBsyIMEdWGKmj9OCCXWWVuhCZ8osvvtipTsPyc+mx+I4p80nVgJXj3z1XobUvMdiXpTstIaSu4XXzbXgsXrdqjg7/TtWU2Hdn2TAu4sr8q8A/A15USr2mlPqDwJ8CfptS6kvAbw1/A/wd4KvAl4H/Dvi3Q8PuAf9X4FfC5z8Nx+6LYmOHutg6F2Q8L3V1pUCl5zGU4e9Dz/c6XbCfXG0YbEXTWG9UlL5sm1KqC2He2dnhpZdeWrI490lcofJ1Ih6LKJzFV73GIzTnecGNxx/jW7/1W/nKyy8xmy6o6zqkNJ+wmM24efMmTVUxGRUYJUyPD73a0NY+vbuac3JywtHRUbfw40KbTCYBVMVnab7yyitdMFQacTmdTrt+uHPnTpdFefv2bZRSHB4fM5vNuujG6XTKW2+9xVNPPc3GxgbPPPNcWNgld+/us7d3maizx94f7nnx+UMxe9Vi832qKIqCD33oW5dwFYb9vypCcji/0vPSaM9+UunlD72Re3ivU9cOKJ2T6e6/ai6vunbo9j3rmot4K37/mp9+cMW5Avw7a+7zF4C/cN7zlq6h54AmvIMBUOqU0HaWm3EdEm/aOevEs/Rv7XzWvoh0OmRqpff36wcq0xpnoXE1hXJ827d8AFSLuIhqlCRphTTsYSyBFdXH/dN6JhGSs5RWOLFoMSCOqm3Y2N7i6OAeTz/9dMBTXFDVFaOy5ObtW2xtbVG7BhsyGltpaZuWzfEEsY6m6u0ao9GI6b6Hazs+PmZv9wrj8bgLx452hqqqurZba5lOpz4dum3Y29vj+PgYgKwoODo6wh4esnv5EnXdcO+eD49+6slneepJj/d4uH/Exnib+XTO3t5ep6ZEqUq0R9ZGqSWL+3AnTEXr+FuUzpRS1NayazSZsbRYn9firVz9+KtgSxBBpEVCDVEPDhyLy/Rzx8+HqJIEKUb1QMZWWX+dsJSP4XEgYsHbmLm5GobemF4SGKoWqa1hKGnE8YHVOKxDes+ET1uXoD31aufK8y9637Rzh9evYxQdx00mUGQYiHQTxTlH61q0GAqtMbR8//d8NwZHRqx16e8ZdeYXXngBocGHLZtTg6xV5qNetQuiatRbI1S9onWWyeY221uaar7grfoWG9sj7rz1Jjs7e/59sewf7HPt2jVmAXXKWosLDEPnhrZpmJ9MUQJt3ZBpQ1lkzKYe3MW2Hi/COZ9jMZr4PIkYAWmM8VLDeEJRGG7f3qc5POSZ557lzdu3mU6n3Lp1i8PDQz7ykY9w+epjiHjPTNs6BM14shlCkOkYaBybdWUP05T5VW7BFHOzFcdmMUK3DqVdh4a9BDKrg6HYeXAcrVUH9x9BYIfPjwZJJ6FaWYLnIFpQ4rN0RSXzbeDtEHHEWT58l1T9HVb+Ogu/snsn1RcbPosebuZwAVEJ1vuZ02PndUTKmVeRtbaLmReRZYBPuyyF9GJhmIhti5aap558EnnrJlp6INi0bUVR4Jz1SNUQVIlkV7AKEXtaYkmYpXUOg8YCxWjE008/jbiW/f19fumX/gnf9m0f9OqCs5zMphSjktl0xrjwmI1HVUVJ0b1zDHG+evUqt27d5vLly95r0VZd2vSlS5c4mfWqRFFkTKdTjDHMZjMUPrJxYzzi9TdvUtcN29vb7O7u8t3f/d089thjmHzcgcaaAEKjle7eP90dY5j0RSzuZ421MYb6ZEZmocHv9B4gdynQAScOUb3475mCAythMQ5DBnx+jIhC1HLcBM7HuZzO4rx/89+qeb9OhXoQeqiZQ6pW+AFIvAvpeWsW9FmMJXLblFKuKyJLXLm7lxMPrppAvWfO+/kR8YlSSnnwVVE+Ed+1lJlmI884dAvaEPCkddGpFUr5KedFRnA2tqdvo3Nt185lihKQ915g/L2sgtx6qLLnX3iRJ57woSV3794NOQtHKHGUo4JFU2EEJpMxjfMBRnmWoeceL7KezdnbvdzZMYpixO7uJV5//XW+9rXXuHLtMuOxT/5q25ajkylN07CztUvd1qhMsX/3HmjFiy9+gCs3HiPP8w7Xcl5XCbivDQbH0wWEUqkhHctVCyBVKVbp6Rkwmx+BNIyVUOuQgKWiSudQNgsQLqBUhpYo9Um3WZxe6Ymk4FQn/fh2O5xYlJggHQStJVR9NcG+FKM20/YOXbNDb8Rwvq/qk1XS1Dp6qJnDg9A6hjB0aQ1dPamakV6TXqeUD3WO+A5R1XGSJMI4X0vSF5t1uFYwIuzf84a5VW2CIA6HCs9+EglVNe8MltCHCqcTZHlh+BqQzoEJSNkeWFej85zxZJPFYsHjTzxFfeUKzrWcHB1w+6030UaxMd5g/949sjz3xW2N6bIjF4sFk62iA2xNEZ6effZZDo+PsFYYjzeYTqc8duMJ5nNv3Lx19x5lMea555/nfe97H+iMcqMvRde0Let2TqV8BuhoNOpE6VXuyVWULqoh/meko6Mj9vf32So3sCkKuXeVdSgrEe8yZebIxQGCuvuK9TYIlRxzQqb9huGCHUMSHIjheKd9c967D89bpTavo4eeOXTBHgl3duKWDJLrXnSVLWF4327XHhgC07+XxFcniXrhuv/jfbQIzoZciKZlYzxhMZ+yOJ7yyiuvMEqelz7Hu+e8VwMC6OxAbFT0BqfYLm88xPu+8fEP4lQXH+DEi8vgd0JdFFilMOUIA2yZjI2NDebTE8ajEePxhK+8/BKmKCgnEzSa/f19Ll++zFHAP3DOMZ3e68Bmbt26hROPCP3Ky6+yvb1L3dSU5YTL1zb4ju/6brZ2dqhbh8dUVNi2RbSirkPqt12WilIJLg2THo51qmoM1Yx0XFfNg9iP0+mUZy5f5hOf/ZLPIBXpxH9RzdI4Oeu6Z2kieKzCub7Ybdqu1CXuJV6H0iH+JZaAVn3bwlt3zCFlDKsM6+vm/nkMYZ20ldJDzxxS6havyPI+o1eHnJiEH8RglTiATmJZOdVFNca4ikgqfoQlJqEF7y0IzMHaYLVuLUZ8jr+tW0Ym57n3PcFXPv85VGMZY5BMY5zuRE0UmCgSKmjbugswiu+cMqrhpPPnJbYQL7h017ciQXdXXUSetU1XcyIrN2jI2Njaw9YNrct44dkXUTim02NcMUaUphxPeOryFV577bUAJqN4+ulnu509z3O2d3Z47rnnMUbROsEBJqRSN1aQAHwrhMCv0G4PRLt6Mcf3Gv4f1b6zrO5DQ2SMU4jU2BZt4fDWjNe++GX+wI/8Pn7+b/1tVAe8GyYAy+HaPVMOhsEoNMa/Q7CUE+XxHyMSeqeGeFuKld6z0FiHMcFdK9FYGPsj3YhkiYFGRpWqwcNFHxGxBVlCyjpP6nnPMId0Vz9Frjf6pEzCrYgbH8Y4DCfeUPyMYuyyaEdXOr57fpRANcxOpvxn/5f/lD/7f/8v+Xt/++/wm773e/n8Jz5Lrhwt3uilwsQe7mrp7gPLiTrpmKcxHGly2Cp9PC3s4sN7DaKF2glGKXRmsM67TctyjBmX3Ltzl0tXrmJFceXaDabTKUoJH/jAB3jmmWcYjXzq+P7+Pk888URY9Bqtvb6cl14NaFqHFe+2lcAOnAjWtl1Fq/iO6wzLwz4Z9stFKF1I6T2yfIRRPmL0f/i7f48bN25w881bHgxloGKCtzXF+3WGLwG08y52ETSCiC8fSGDeMXJ0ad4NNm7nXF+ZTWXh+elDliMch+8zVIvTd+/6QC2n5Z9F7ynmEP9fa4BUvgBupFjJapWIlQ7SUFxLf1vlR07rGIpISLOGsiz54uc+xY/92I/xU3/hL1LPF1y7cpVf/qf/1KsETQ20KAQn+UpXakw8GrbXT4Y+tyEyguHCWSVax0Ad0OS5AWVBa0zmK121rZCh/LG8QGzN3uVLVPMFPlzYUZZjtPb3G497V+vu7q7PKMX4/VIpTJaHsnsg2uCXC7RB/Wpt24VWn9rlkkWcVsmKOJNDJnJWfEvq0Un7Y/i8b//2b+Xm4QFvvfo6RTFhNpt5CDvj80SC1uafZ5P+tbHQrQbxzMEz4NjvEjxuMZBsWeJJDe6dtBffS2J9Uo8mtcr+kM7tdC6dYpyDTQWWU93X0cPNHEQQ68FPh5bn4YtZcUG/7g03bWD+soK7ary70CXiWvw9/u8PCZm3D3Yg41YEbIQDUxgsrq2p24b/9f/yf8Unf+XXeOLqdT7+T3+ZetFwbW+H/+DHfpR6MaUoFY1SGG1xotBiVk7YVEzsSXXRhks+boKoaTykW8pAnUjiT7e0ddshRLcCgkKbEh/o7XVtMRk4w3iz7I2yLljoRaGkCZGZmqKckOU51vlEM43CaZ8FrZSv2WBdQ+Oim9C3K9oRVjHCiH05VB1iMFQ6lsNrl+bESsYQGU8G0mK05ujOHV577S029nbZ29xl54VdPvn5z3qgHAluS6T7qj1779RSh1cdIkPzz7GIcjgrvlZpGDOlVPd+6DReIyz07j36TUvpyCA8RcPo0EDdqZKtD5brrlFuSXLsUa7OliAefpg4t5y/sMoos+yWPh0+nX5EpAtxHYbMDuPko0HIOYfYduX9mqZhPj1mdjLlP/7jf5yPf/RXuPnKq/zNv/GzqEa4srvDv/9H/k1eePYaJlNopVAmR5vzA1EikGvXFaF9WZatFLWX+mSgMqVG0GF0qFPLfZgyHujjCrTOAppUQZ4XGOPb4US66t+i+7BwrX1djlhgN94rtaGkYxXzF1I1Lg34GVKsn9nZkdxp+LTzqGkably7ggCbowmf+dSnGY1GXL16tdvJl1RPF+dDuonY5XOS8Vpyv69g+qvm3ZnSsUgHyR/77Tw6UyU/g9SDXPSNIqXUMfCFd7sd59AV4M673Ygz6FH73h497O2Dt9/Gp0Xk6vDgw61WwBdE5MPvdiPOIqXUxx7mNj5q39ujh7198M618eFXKx7RI3pE7wo9Yg6P6BE9opX0sDOHn3i3G3ABetjb+Kh9b48e9vbBO9TGh9og+Yge0SN69+hhlxwe0SN6RO8SPWIOj+gRPaKV9NAyB6XU71BKfUH50no/fv4V70gbnlJK/SOl1GeVUp9RSv274fh9lwN8h9tplFIfV0r9Qvj7WaXUR0M7/ppSqgjHy/D3l8Pvz3yD2rerlPoZpdTnlVKfU0p978PUh0qpPxbG99NKqb+qlBq9m32o3s0SlCmtisx6tz94qMivAM8BBfAJ4FvehXY8BnxX+L4FfBH4FuA/B348HP9x4E+H778T+P/io6x/PfDRb1A7/33gvwd+Ifz914HfF77/eeDfCt//beDPh++/D/hr36D2/RTwh8L3Ath9WPoQX1zpJWCc9N3/7t3sQ+AHgO8CPp0cu6/+Ai7hwZ4vAXvh+959teMbMTkeoHO+F/h7yd9/AvgTD0G7fg74bfiozcfCscfwwVoA/y3w+5Pzu/PewTY9ia9X+luAXwiT5A6QDfsSXw7ge8P3LJyn3uH27YTFpwbHH4o+pK/Gdin0yS8A/7N3uw+BZwbM4b76C/j9wH+bHF867yKfh1WtuHD5vG8UBfHxO4GPcv/lAN9J+jPAf0ifmXMZOBCRdkUbuvaF3w/D+e8kPQvcBv5iUH3+30qpDR6SPhSR14H/AvgacBPfJ7/Kw9WH8A6WoFxHDytzeKhIKbUJ/Czw74nIUfqbeLb8rviDlVK/C7glIr/6bjz/gpThReQ/JyLfCUzpq7ID73of7uELQD8LPA5scLqq/ENF36j+eliZw7qyet9wUkrleMbwV0Tkb4bDbylfBhB1sXKA7xR9P/C/UEq9DPw0XrX4s8CuUirmzaRt6NoXft8B7r6D7QO/Y70mIh8Nf/8Mnlk8LH34W4GXROS2iDTA38T368PUh3D//fW2+/FhZQ6/ArwQLMYF3vDz89/oRiifb/uTwOdE5L9MfrrfcoDvCInInxCRJ0XkGXwf/UMR+d8A/wj43WvaF9v9u8P57+gOJCJvAq8qpV4Mh34Q+CwPSR/i1Ylfr5SahPGO7Xto+nDFcy/SXw9UgnKJ3ilDz9fBIPM78d6BrwD/p3epDb8BL759EviX4fM78TrmLwJfAv4BcCmcr4D/V2jzp4APfwPb+hF6b8VzwL/AlyX8G0AZjo/C318Ovz/3DWrbdwAfC/34t/DW84emD4H/BPg88GngLwPlu9mHwF/F2z8avOT1Bx+kv4B/M7Tzy8AfuN92PAqffkSP6BGtpIdVrXhEj+gRvcv0iDk8okf0iFbSI+bwiB7RI1pJj5jDI3pEj2glPWIOj+gRPaKV9Ig5PKJH9IhW0iPm8Ige0SNaSf8/UoFMbs16WiMAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "img = io.imread(\"Men1/Men1_000.jpg\")\n",
    "pyplot.imshow(img, interpolation='nearest')\n",
    "new_img = detector(img, 1)\n",
    "for k, d in enumerate(new_img):\n",
    "    print(\"Detection {}: Left: {} Top: {} Right: {} Bottom: {}\".format(\n",
    "        k, d.left(), d.top(), d.right(), d.bottom()))\n",
    "    shape = sp(img, d)\n",
    "    \n",
    "face_descriptor1 = facerec.compute_face_descriptor(img, shape)\n",
    "face_descriptor1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9e5117e8",
   "metadata": {},
   "source": [
    "### Формирование обучающей и тестовой выборки:\n",
    "\n",
    "> **ToDo**: переписать блок для большей эффективности и лучшей читабельности"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "b88e6394",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Name: Men1\n",
      "Real prediction: [1, 0, 0, 0]\n",
      "Files on dir: 178\n",
      "Files done: 10/178\n",
      "Files done: 20/178\n",
      "Files done: 30/178\n",
      "Files done: 40/178\n",
      "Files done: 50/178\n",
      "Files done: 60/178\n",
      "Files done: 70/178\n",
      "Files done: 80/178\n",
      "Files done: 90/178\n",
      "Files done: 100/178\n",
      "Files done: 110/178\n",
      "Files done: 120/178\n",
      "Files done: 130/178\n",
      "Files done: 140/178\n",
      "Files done: 150/178\n",
      "Files done: 160/178\n",
      "Files done: 170/178\n",
      "Name: Men2\n",
      "Real prediction: [0, 1, 0, 0]\n",
      "Files on dir: 186\n",
      "Files done: 10/186\n",
      "Files done: 20/186\n",
      "Files done: 30/186\n",
      "Files done: 40/186\n",
      "Files done: 50/186\n",
      "Files done: 60/186\n",
      "Files done: 70/186\n",
      "Files done: 80/186\n",
      "Files done: 90/186\n",
      "Files done: 100/186\n",
      "Files done: 110/186\n",
      "Files done: 120/186\n",
      "Files done: 130/186\n",
      "Files done: 140/186\n",
      "Files done: 150/186\n",
      "Files done: 160/186\n",
      "Files done: 170/186\n",
      "Files done: 180/186\n",
      "Name: Men3\n",
      "Real prediction: [0, 0, 1, 0]\n",
      "Files on dir: 234\n",
      "Files done: 10/234\n",
      "Files done: 20/234\n",
      "Files done: 30/234\n",
      "Files done: 40/234\n",
      "Files done: 50/234\n",
      "Files done: 60/234\n",
      "Files done: 70/234\n",
      "Files done: 80/234\n",
      "Files done: 90/234\n",
      "Files done: 100/234\n",
      "Files done: 110/234\n",
      "Files done: 120/234\n",
      "Files done: 130/234\n",
      "Files done: 140/234\n",
      "Files done: 150/234\n",
      "Files done: 160/234\n",
      "Files done: 170/234\n",
      "Files done: 180/234\n",
      "Files done: 190/234\n",
      "Files done: 200/234\n",
      "Files done: 210/234\n",
      "Files done: 220/234\n",
      "Files done: 230/234\n",
      "Name: Female1\n",
      "Real prediction: [0, 0, 0, 1]\n",
      "Files on dir: 203\n",
      "Files done: 10/203\n",
      "Files done: 20/203\n",
      "Files done: 30/203\n",
      "Files done: 40/203\n",
      "Files done: 50/203\n",
      "Files done: 60/203\n",
      "Files done: 70/203\n",
      "Files done: 80/203\n",
      "Files done: 90/203\n",
      "Files done: 100/203\n",
      "Files done: 110/203\n",
      "Files done: 120/203\n",
      "Files done: 130/203\n",
      "Files done: 140/203\n",
      "Files done: 150/203\n",
      "Files done: 160/203\n",
      "Files done: 170/203\n",
      "Files done: 180/203\n",
      "Files done: 190/203\n",
      "Files done: 200/203\n",
      "Name: SomePerson\n",
      "Real prediction: [0, 0, 0, 0]\n",
      "Files on dir: 101\n",
      "Files done: 10/101\n",
      "Files done: 20/101\n",
      "Files done: 30/101\n",
      "Files done: 40/101\n",
      "Files done: 50/101\n",
      "Files done: 60/101\n",
      "Files done: 70/101\n",
      "Files done: 80/101\n",
      "Files done: 90/101\n",
      "Files done: 100/101\n"
     ]
    }
   ],
   "source": [
    "li_train = list()\n",
    "li_test = list()\n",
    "\n",
    "men1 =    [\"Men1\", [1, 0, 0, 0]]\n",
    "men2 =    [\"Men2\", [0, 1, 0, 0]]\n",
    "men3 =    [\"Men3\", [0, 0, 1, 0]]\n",
    "female1 = [\"Female1\", [0, 0, 0, 1]]\n",
    "\n",
    "someperson = [\"SomePerson\", [0, 0, 0, 0]]\n",
    "\n",
    "for person in [men1, men2, men3, female1, someperson]:\n",
    "    files_on_dir = os.listdir(person[0])\n",
    "    print(f\"Name: {person[0]}\\nReal prediction: {person[1]}\\nFiles on dir: {len(files_on_dir)}\")\n",
    "    \n",
    "    train_threshold = len(files_on_dir) * 0.8\n",
    "    for file_index in range(len(files_on_dir)):\n",
    "        file_name = person[0] + \"/\" + files_on_dir[file_index]\n",
    "        if \".jpg\" in file_name:\n",
    "            if file_index % 10 == 0:\n",
    "                print(f\"Files done: {file_index}/{len(files_on_dir)}\")\n",
    "                \n",
    "            raw_img = io.imread(file_name)\n",
    "            dtc_img = detector(raw_img, 1)\n",
    "            \n",
    "            for k, d in enumerate(dtc_img):\n",
    "                shape = sp(raw_img, d)\n",
    "            descriptor = facerec.compute_face_descriptor(raw_img, shape)\n",
    "            \n",
    "            if file_index <= train_threshold:\n",
    "                li_train.append([descriptor, person[1]])\n",
    "            else:\n",
    "                li_test.append([descriptor, person[1]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "72c2fd26",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "719 178\n"
     ]
    }
   ],
   "source": [
    "print(len(li_train), len(li_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "86675ec4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "719 719 178 178\n"
     ]
    }
   ],
   "source": [
    "list_with_train_inputs = list()\n",
    "list_with_train_predictions = list()\n",
    "list_with_test_inputs = list()\n",
    "list_with_test_predictions = list()\n",
    "\n",
    "random.shuffle(li_train)\n",
    "for i in range(len(li_train)):\n",
    "    list_with_train_inputs.append(list(li_train[i][0]))\n",
    "    list_with_train_predictions.append(li_train[i][1])\n",
    "    \n",
    "random.shuffle(li_test)\n",
    "for i in range(len(li_test)):\n",
    "    list_with_test_inputs.append(list(li_test[i][0]))\n",
    "    list_with_test_predictions.append(li_test[i][1])\n",
    "    \n",
    "print(len(list_with_train_inputs), len(list_with_train_predictions), len(list_with_test_inputs), len(list_with_test_predictions))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2c1969df",
   "metadata": {},
   "source": [
    "## Нейронная сеть классификатор"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "d8072a52",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0594 - accuracy: 0.8220\n",
      "Epoch 2/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0155 - accuracy: 0.9040\n",
      "Epoch 3/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0110 - accuracy: 0.9082\n",
      "Epoch 4/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0081 - accuracy: 0.9166\n",
      "Epoch 5/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0079 - accuracy: 0.9138\n",
      "Epoch 6/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0071 - accuracy: 0.9305\n",
      "Epoch 7/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0067 - accuracy: 0.9179\n",
      "Epoch 8/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0070 - accuracy: 0.9152\n",
      "Epoch 9/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0061 - accuracy: 0.9249\n",
      "Epoch 10/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0067 - accuracy: 0.9096\n",
      "Epoch 11/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0056 - accuracy: 0.9152\n",
      "Epoch 12/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0055 - accuracy: 0.9124\n",
      "Epoch 13/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0060 - accuracy: 0.9110\n",
      "Epoch 14/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0056 - accuracy: 0.9110\n",
      "Epoch 15/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0059 - accuracy: 0.9152\n",
      "Epoch 16/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0057 - accuracy: 0.9207\n",
      "Epoch 17/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0051 - accuracy: 0.9110\n",
      "Epoch 18/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0053 - accuracy: 0.9152\n",
      "Epoch 19/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0051 - accuracy: 0.9138\n",
      "Epoch 20/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0058 - accuracy: 0.9124\n",
      "Epoch 21/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0047 - accuracy: 0.9193\n",
      "Epoch 22/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0048 - accuracy: 0.9068\n",
      "Epoch 23/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0052 - accuracy: 0.9152\n",
      "Epoch 24/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0041 - accuracy: 0.9179\n",
      "Epoch 25/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0046 - accuracy: 0.9221\n",
      "Epoch 26/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0050 - accuracy: 0.9166\n",
      "Epoch 27/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0040 - accuracy: 0.9138\n",
      "Epoch 28/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0050 - accuracy: 0.9166\n",
      "Epoch 29/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0040 - accuracy: 0.9124\n",
      "Epoch 30/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0044 - accuracy: 0.9124\n",
      "Epoch 31/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0045 - accuracy: 0.9166\n",
      "Epoch 32/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0043 - accuracy: 0.9152\n",
      "Epoch 33/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0038 - accuracy: 0.9138\n",
      "Epoch 34/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0042 - accuracy: 0.9166\n",
      "Epoch 35/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0042 - accuracy: 0.9152\n",
      "Epoch 36/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0042 - accuracy: 0.9152\n",
      "Epoch 37/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0034 - accuracy: 0.9124\n",
      "Epoch 38/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0043 - accuracy: 0.9110\n",
      "Epoch 39/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0044 - accuracy: 0.9249\n",
      "Epoch 40/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0040 - accuracy: 0.9138\n",
      "Epoch 41/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0036 - accuracy: 0.9110\n",
      "Epoch 42/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0043 - accuracy: 0.9110\n",
      "Epoch 43/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0034 - accuracy: 0.9124\n",
      "Epoch 44/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0039 - accuracy: 0.9179\n",
      "Epoch 45/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0040 - accuracy: 0.9124\n",
      "Epoch 46/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0039 - accuracy: 0.9166\n",
      "Epoch 47/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0035 - accuracy: 0.9124\n",
      "Epoch 48/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0041 - accuracy: 0.9138\n",
      "Epoch 49/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0038 - accuracy: 0.9068\n",
      "Epoch 50/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0038 - accuracy: 0.9152\n",
      "Epoch 51/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0040 - accuracy: 0.9124\n",
      "Epoch 52/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0037 - accuracy: 0.9221\n",
      "Epoch 53/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0040 - accuracy: 0.9166\n",
      "Epoch 54/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0040 - accuracy: 0.9096\n",
      "Epoch 55/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0031 - accuracy: 0.9124\n",
      "Epoch 56/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0038 - accuracy: 0.9082\n",
      "Epoch 57/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0038 - accuracy: 0.9082\n",
      "Epoch 58/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0031 - accuracy: 0.9110\n",
      "Epoch 59/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0040 - accuracy: 0.9124\n",
      "Epoch 60/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0035 - accuracy: 0.9193\n",
      "Epoch 61/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0035 - accuracy: 0.9235\n",
      "Epoch 62/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0034 - accuracy: 0.9207\n",
      "Epoch 63/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0032 - accuracy: 0.9152\n",
      "Epoch 64/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0033 - accuracy: 0.9138\n",
      "Epoch 65/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0035 - accuracy: 0.9193\n",
      "Epoch 66/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0032 - accuracy: 0.9193\n",
      "Epoch 67/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0037 - accuracy: 0.9193\n",
      "Epoch 68/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0031 - accuracy: 0.9096\n",
      "Epoch 69/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0034 - accuracy: 0.9054\n",
      "Epoch 70/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0034 - accuracy: 0.9277\n",
      "Epoch 71/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0030 - accuracy: 0.9166\n",
      "Epoch 72/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0033 - accuracy: 0.9124\n",
      "Epoch 73/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0036 - accuracy: 0.9124\n",
      "Epoch 74/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0032 - accuracy: 0.9138\n",
      "Epoch 75/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0034 - accuracy: 0.9110\n",
      "Epoch 76/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0033 - accuracy: 0.9179\n",
      "Epoch 77/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0035 - accuracy: 0.9138\n",
      "Epoch 78/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0030 - accuracy: 0.9082\n",
      "Epoch 79/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0034 - accuracy: 0.9124\n",
      "Epoch 80/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0031 - accuracy: 0.9152\n",
      "Epoch 81/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0035 - accuracy: 0.9152\n",
      "Epoch 82/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9096\n",
      "Epoch 83/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0035 - accuracy: 0.9221\n",
      "Epoch 84/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9166\n",
      "Epoch 85/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0034 - accuracy: 0.9082\n",
      "Epoch 86/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0031 - accuracy: 0.9082\n",
      "Epoch 87/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0031 - accuracy: 0.9166\n",
      "Epoch 88/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0033 - accuracy: 0.9166\n",
      "Epoch 89/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0034 - accuracy: 0.9110\n",
      "Epoch 90/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9068\n",
      "Epoch 91/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9152\n",
      "Epoch 92/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0032 - accuracy: 0.9207\n",
      "Epoch 93/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0033 - accuracy: 0.9138\n",
      "Epoch 94/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9138\n",
      "Epoch 95/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0030 - accuracy: 0.9110\n",
      "Epoch 96/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0030 - accuracy: 0.9082\n",
      "Epoch 97/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9207\n",
      "Epoch 98/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0032 - accuracy: 0.9110\n",
      "Epoch 99/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9166\n",
      "Epoch 100/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0034 - accuracy: 0.9110\n",
      "Epoch 101/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9068\n",
      "Epoch 102/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0030 - accuracy: 0.9082\n",
      "Epoch 103/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9082\n",
      "Epoch 104/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9096\n",
      "Epoch 105/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9249\n",
      "Epoch 106/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0030 - accuracy: 0.9110\n",
      "Epoch 107/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9110\n",
      "Epoch 108/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9082\n",
      "Epoch 109/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0032 - accuracy: 0.9110\n",
      "Epoch 110/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9179\n",
      "Epoch 111/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0033 - accuracy: 0.9082\n",
      "Epoch 112/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0031 - accuracy: 0.9124\n",
      "Epoch 113/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9110\n",
      "Epoch 114/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0030 - accuracy: 0.9138\n",
      "Epoch 115/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9110\n",
      "Epoch 116/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9096\n",
      "Epoch 117/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0030 - accuracy: 0.9152\n",
      "Epoch 118/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0030 - accuracy: 0.9138\n",
      "Epoch 119/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9152\n",
      "Epoch 120/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9082\n",
      "Epoch 121/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9193\n",
      "Epoch 122/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0031 - accuracy: 0.9138\n",
      "Epoch 123/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9138\n",
      "Epoch 124/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9138\n",
      "Epoch 125/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9110\n",
      "Epoch 126/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9166\n",
      "Epoch 127/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9179\n",
      "Epoch 128/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9179\n",
      "Epoch 129/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9207\n",
      "Epoch 130/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9193\n",
      "Epoch 131/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0030 - accuracy: 0.9068\n",
      "Epoch 132/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9138\n",
      "Epoch 133/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9110\n",
      "Epoch 134/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9138\n",
      "Epoch 135/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9166\n",
      "Epoch 136/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9110\n",
      "Epoch 137/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9082\n",
      "Epoch 138/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9124\n",
      "Epoch 139/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9138\n",
      "Epoch 140/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9138\n",
      "Epoch 141/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9096\n",
      "Epoch 142/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9096\n",
      "Epoch 143/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9138\n",
      "Epoch 144/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9110\n",
      "Epoch 145/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9096\n",
      "Epoch 146/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0031 - accuracy: 0.9193\n",
      "Epoch 147/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9110\n",
      "Epoch 148/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9096\n",
      "Epoch 149/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9152\n",
      "Epoch 150/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9110\n",
      "Epoch 151/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9124\n",
      "Epoch 152/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9124\n",
      "Epoch 153/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9124\n",
      "Epoch 154/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9096\n",
      "Epoch 155/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9138\n",
      "Epoch 156/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9096\n",
      "Epoch 157/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9068\n",
      "Epoch 158/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9082\n",
      "Epoch 159/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9110\n",
      "Epoch 160/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9138\n",
      "Epoch 161/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9124\n",
      "Epoch 162/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9124\n",
      "Epoch 163/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9096\n",
      "Epoch 164/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9152\n",
      "Epoch 165/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9138\n",
      "Epoch 166/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9110\n",
      "Epoch 167/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9138\n",
      "Epoch 168/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9124\n",
      "Epoch 169/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9152\n",
      "Epoch 170/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9110\n",
      "Epoch 171/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9166\n",
      "Epoch 172/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9110\n",
      "Epoch 173/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9124\n",
      "Epoch 174/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9040\n",
      "Epoch 175/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9152\n",
      "Epoch 176/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9082\n",
      "Epoch 177/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9179\n",
      "Epoch 178/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9026\n",
      "Epoch 179/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9054\n",
      "Epoch 180/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9096\n",
      "Epoch 181/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0029 - accuracy: 0.9138\n",
      "Epoch 182/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9110\n",
      "Epoch 183/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9138\n",
      "Epoch 184/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9054\n",
      "Epoch 185/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9124\n",
      "Epoch 186/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9054\n",
      "Epoch 187/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9152\n",
      "Epoch 188/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9110\n",
      "Epoch 189/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9193\n",
      "Epoch 190/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9221\n",
      "Epoch 191/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9040\n",
      "Epoch 192/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9166\n",
      "Epoch 193/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9082\n",
      "Epoch 194/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9166\n",
      "Epoch 195/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9152\n",
      "Epoch 196/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9124\n",
      "Epoch 197/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9179\n",
      "Epoch 198/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9110\n",
      "Epoch 199/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9179\n",
      "Epoch 200/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9082\n",
      "Epoch 201/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9152\n",
      "Epoch 202/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9054\n",
      "Epoch 203/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9110\n",
      "Epoch 204/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9096\n",
      "Epoch 205/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9068\n",
      "Epoch 206/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9124\n",
      "Epoch 207/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9082\n",
      "Epoch 208/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9110\n",
      "Epoch 209/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9124\n",
      "Epoch 210/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9082\n",
      "Epoch 211/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9096\n",
      "Epoch 212/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9096\n",
      "Epoch 213/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9235\n",
      "Epoch 214/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9110\n",
      "Epoch 215/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9110\n",
      "Epoch 216/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9110\n",
      "Epoch 217/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9124\n",
      "Epoch 218/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9096\n",
      "Epoch 219/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9110\n",
      "Epoch 220/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9193\n",
      "Epoch 221/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9152\n",
      "Epoch 222/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9138\n",
      "Epoch 223/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9040\n",
      "Epoch 224/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9152\n",
      "Epoch 225/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9152\n",
      "Epoch 226/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9152\n",
      "Epoch 227/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9110\n",
      "Epoch 228/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9110\n",
      "Epoch 229/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9152\n",
      "Epoch 230/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9166\n",
      "Epoch 231/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9166\n",
      "Epoch 232/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9138\n",
      "Epoch 233/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9110\n",
      "Epoch 234/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9207\n",
      "Epoch 235/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9193\n",
      "Epoch 236/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9166\n",
      "Epoch 237/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9179\n",
      "Epoch 238/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9166\n",
      "Epoch 239/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9166\n",
      "Epoch 240/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.9138\n",
      "Epoch 241/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9179\n",
      "Epoch 242/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9068\n",
      "Epoch 243/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9166\n",
      "Epoch 244/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9110\n",
      "Epoch 245/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9082\n",
      "Epoch 246/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9054\n",
      "Epoch 247/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9068\n",
      "Epoch 248/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9179\n",
      "Epoch 249/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9124\n",
      "Epoch 250/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9138\n",
      "Epoch 251/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9166\n",
      "Epoch 252/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9082\n",
      "Epoch 253/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9207\n",
      "Epoch 254/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9110\n",
      "Epoch 255/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0028 - accuracy: 0.8999\n",
      "Epoch 256/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9179\n",
      "Epoch 257/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9068\n",
      "Epoch 258/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9193\n",
      "Epoch 259/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9124\n",
      "Epoch 260/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9068\n",
      "Epoch 261/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0027 - accuracy: 0.9179\n",
      "Epoch 262/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9138\n",
      "Epoch 263/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9138\n",
      "Epoch 264/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9166\n",
      "Epoch 265/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9110\n",
      "Epoch 266/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9124\n",
      "Epoch 267/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9138\n",
      "Epoch 268/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9152\n",
      "Epoch 269/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9138\n",
      "Epoch 270/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9082\n",
      "Epoch 271/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9124\n",
      "Epoch 272/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9166\n",
      "Epoch 273/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9207\n",
      "Epoch 274/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9124\n",
      "Epoch 275/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9068\n",
      "Epoch 276/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9110\n",
      "Epoch 277/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9179\n",
      "Epoch 278/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9054\n",
      "Epoch 279/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9082\n",
      "Epoch 280/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9179\n",
      "Epoch 281/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9179\n",
      "Epoch 282/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9166\n",
      "Epoch 283/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9138\n",
      "Epoch 284/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9166\n",
      "Epoch 285/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9110\n",
      "Epoch 286/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9179\n",
      "Epoch 287/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9152\n",
      "Epoch 288/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9221\n",
      "Epoch 289/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9235\n",
      "Epoch 290/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9040\n",
      "Epoch 291/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9193\n",
      "Epoch 292/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9166\n",
      "Epoch 293/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9096\n",
      "Epoch 294/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9193\n",
      "Epoch 295/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9152\n",
      "Epoch 296/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9138\n",
      "Epoch 297/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9124\n",
      "Epoch 298/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9138\n",
      "Epoch 299/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9166\n",
      "Epoch 300/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9166\n",
      "Epoch 301/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9110\n",
      "Epoch 302/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9179\n",
      "Epoch 303/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9193\n",
      "Epoch 304/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9152\n",
      "Epoch 305/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9138\n",
      "Epoch 306/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9179\n",
      "Epoch 307/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9138\n",
      "Epoch 308/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9193\n",
      "Epoch 309/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9179\n",
      "Epoch 310/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9138\n",
      "Epoch 311/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9068\n",
      "Epoch 312/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0019 - accuracy: 0.9179\n",
      "Epoch 313/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9193\n",
      "Epoch 314/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9096\n",
      "Epoch 315/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9124\n",
      "Epoch 316/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9207\n",
      "Epoch 317/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9138\n",
      "Epoch 318/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9110\n",
      "Epoch 319/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9138\n",
      "Epoch 320/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9138\n",
      "Epoch 321/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9124\n",
      "Epoch 322/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9179\n",
      "Epoch 323/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9110\n",
      "Epoch 324/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9138\n",
      "Epoch 325/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9179\n",
      "Epoch 326/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9138\n",
      "Epoch 327/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9082\n",
      "Epoch 328/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9152\n",
      "Epoch 329/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9166\n",
      "Epoch 330/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9193\n",
      "Epoch 331/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9138\n",
      "Epoch 332/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9124\n",
      "Epoch 333/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9166\n",
      "Epoch 334/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9124\n",
      "Epoch 335/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9054\n",
      "Epoch 336/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9138\n",
      "Epoch 337/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9110\n",
      "Epoch 338/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9193\n",
      "Epoch 339/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9235\n",
      "Epoch 340/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9138\n",
      "Epoch 341/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9235\n",
      "Epoch 342/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9193\n",
      "Epoch 343/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9166\n",
      "Epoch 344/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9193\n",
      "Epoch 345/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9138\n",
      "Epoch 346/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9207\n",
      "Epoch 347/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9179\n",
      "Epoch 348/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9124\n",
      "Epoch 349/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9179\n",
      "Epoch 350/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9152\n",
      "Epoch 351/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9082\n",
      "Epoch 352/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9054\n",
      "Epoch 353/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9054\n",
      "Epoch 354/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0025 - accuracy: 0.9166\n",
      "Epoch 355/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9124\n",
      "Epoch 356/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9124\n",
      "Epoch 357/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9096\n",
      "Epoch 358/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9166\n",
      "Epoch 359/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9054\n",
      "Epoch 360/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9179\n",
      "Epoch 361/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0026 - accuracy: 0.9082\n",
      "Epoch 362/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9124\n",
      "Epoch 363/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9068\n",
      "Epoch 364/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9138\n",
      "Epoch 365/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9110\n",
      "Epoch 366/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9026\n",
      "Epoch 367/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9124\n",
      "Epoch 368/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9249\n",
      "Epoch 369/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9110\n",
      "Epoch 370/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9152\n",
      "Epoch 371/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9152\n",
      "Epoch 372/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9110\n",
      "Epoch 373/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9179\n",
      "Epoch 374/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9068\n",
      "Epoch 375/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9110\n",
      "Epoch 376/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9110\n",
      "Epoch 377/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9096\n",
      "Epoch 378/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9152\n",
      "Epoch 379/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9054\n",
      "Epoch 380/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9138\n",
      "Epoch 381/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9096\n",
      "Epoch 382/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9110\n",
      "Epoch 383/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9152\n",
      "Epoch 384/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9110\n",
      "Epoch 385/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9082\n",
      "Epoch 386/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9166\n",
      "Epoch 387/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9179\n",
      "Epoch 388/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9138\n",
      "Epoch 389/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9082\n",
      "Epoch 390/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9166\n",
      "Epoch 391/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9166\n",
      "Epoch 392/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9138\n",
      "Epoch 393/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9179\n",
      "Epoch 394/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9152\n",
      "Epoch 395/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9193\n",
      "Epoch 396/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9068\n",
      "Epoch 397/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9152\n",
      "Epoch 398/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9152\n",
      "Epoch 399/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9124\n",
      "Epoch 400/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9207\n",
      "Epoch 401/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9138\n",
      "Epoch 402/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9068\n",
      "Epoch 403/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9207\n",
      "Epoch 404/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9179\n",
      "Epoch 405/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9054\n",
      "Epoch 406/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9152\n",
      "Epoch 407/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9207\n",
      "Epoch 408/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9138\n",
      "Epoch 409/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9124\n",
      "Epoch 410/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9096\n",
      "Epoch 411/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9068\n",
      "Epoch 412/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9124\n",
      "Epoch 413/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9138\n",
      "Epoch 414/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9207\n",
      "Epoch 415/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9193\n",
      "Epoch 416/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0019 - accuracy: 0.9179\n",
      "Epoch 417/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9082\n",
      "Epoch 418/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9138\n",
      "Epoch 419/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9124\n",
      "Epoch 420/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9138\n",
      "Epoch 421/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9082\n",
      "Epoch 422/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9179\n",
      "Epoch 423/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0019 - accuracy: 0.9207\n",
      "Epoch 424/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9082\n",
      "Epoch 425/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9179\n",
      "Epoch 426/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9193\n",
      "Epoch 427/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9082\n",
      "Epoch 428/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9179\n",
      "Epoch 429/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9193\n",
      "Epoch 430/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9152\n",
      "Epoch 431/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9179\n",
      "Epoch 432/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0019 - accuracy: 0.9110\n",
      "Epoch 433/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9179\n",
      "Epoch 434/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9152\n",
      "Epoch 435/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9207\n",
      "Epoch 436/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9249\n",
      "Epoch 437/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9110\n",
      "Epoch 438/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9207\n",
      "Epoch 439/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9110\n",
      "Epoch 440/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9096\n",
      "Epoch 441/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9096\n",
      "Epoch 442/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9152\n",
      "Epoch 443/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9082\n",
      "Epoch 444/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9207\n",
      "Epoch 445/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9124\n",
      "Epoch 446/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9152\n",
      "Epoch 447/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9138\n",
      "Epoch 448/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9138\n",
      "Epoch 449/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9124\n",
      "Epoch 450/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0019 - accuracy: 0.9166\n",
      "Epoch 451/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9124\n",
      "Epoch 452/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9096\n",
      "Epoch 453/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9124\n",
      "Epoch 454/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0019 - accuracy: 0.9179\n",
      "Epoch 455/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9152\n",
      "Epoch 456/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9207\n",
      "Epoch 457/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9152\n",
      "Epoch 458/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9179\n",
      "Epoch 459/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0018 - accuracy: 0.9193\n",
      "Epoch 460/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9166\n",
      "Epoch 461/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9138\n",
      "Epoch 462/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9138\n",
      "Epoch 463/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9138\n",
      "Epoch 464/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9207\n",
      "Epoch 465/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9152\n",
      "Epoch 466/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9068\n",
      "Epoch 467/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0018 - accuracy: 0.9110\n",
      "Epoch 468/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9096\n",
      "Epoch 469/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9221\n",
      "Epoch 470/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0023 - accuracy: 0.9124\n",
      "Epoch 471/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0019 - accuracy: 0.9013\n",
      "Epoch 472/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0019 - accuracy: 0.9207\n",
      "Epoch 473/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9124\n",
      "Epoch 474/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9179\n",
      "Epoch 475/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9193\n",
      "Epoch 476/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0024 - accuracy: 0.9193\n",
      "Epoch 477/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9096\n",
      "Epoch 478/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9207\n",
      "Epoch 479/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0019 - accuracy: 0.9082\n",
      "Epoch 480/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9166\n",
      "Epoch 481/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9082\n",
      "Epoch 482/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0018 - accuracy: 0.9152\n",
      "Epoch 483/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9124\n",
      "Epoch 484/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9193\n",
      "Epoch 485/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9193\n",
      "Epoch 486/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9138\n",
      "Epoch 487/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0019 - accuracy: 0.9082\n",
      "Epoch 488/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9124\n",
      "Epoch 489/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9138\n",
      "Epoch 490/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9138\n",
      "Epoch 491/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9124\n",
      "Epoch 492/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0019 - accuracy: 0.9166\n",
      "Epoch 493/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9110\n",
      "Epoch 494/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9179\n",
      "Epoch 495/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0020 - accuracy: 0.9193\n",
      "Epoch 496/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0019 - accuracy: 0.9082\n",
      "Epoch 497/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9193\n",
      "Epoch 498/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0019 - accuracy: 0.9096\n",
      "Epoch 499/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0021 - accuracy: 0.9179\n",
      "Epoch 500/500\n",
      "23/23 [==============================] - 0s 2ms/step - loss: 0.0022 - accuracy: 0.9054\n"
     ]
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(Dense(units=256, input_shape=(len(list_with_train_inputs[0]),)))\n",
    "model.add(Dropout(0.2))\n",
    "model.add(Dense(units=256, input_shape=(256,), activation=\"relu\"))\n",
    "model.add(Dropout(0.2))\n",
    "model.add(Dense(units=4, input_shape=(256,)))\n",
    "model.compile(loss=\"mean_squared_error\", metrics=['accuracy'])\n",
    "\n",
    "history = model.fit(list_with_train_inputs, list_with_train_predictions, epochs=500)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "d66ef167",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "pyplot.plot(history.history['accuracy'])\n",
    "\n",
    "pyplot.title('Средняя точность эпох')\n",
    "pyplot.ylabel('Точность')\n",
    "pyplot.xlabel('Эпоха')\n",
    "pyplot.grid()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "377390b4",
   "metadata": {},
   "source": [
    "## Тестирование точности предсказаний нейронной сетиисходя из тестовой выборки"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "910499e0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "23/23 [==============================] - 0s 1ms/step - loss: 9.0275e-04 - accuracy: 0.9166\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.0009027527994476259, 0.9165507555007935]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate(list_with_train_inputs, list_with_train_predictions)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}