ML-Audio-Clustering-Train.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "import cv2\n",
    "import csv\n",
    "import os\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from keras.layers import Conv2D, Conv2DTranspose\n",
    "from keras.layers import Input, UpSampling2D, MaxPooling2D\n",
    "from keras.models import Model\n",
    "from keras.models import model_from_json\n",
    "from keras.optimizers import Adam\n",
    "from keras.callbacks import ModelCheckpoint\n",
    "from keras.callbacks import EarlyStopping\n",
    "from sklearn.cluster import KMeans\n",
    "import pylab\n",
    "import librosa\n",
    "import librosa.display\n",
    "import matplotlib\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Set Root Path to Your Directory"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "ROOT_PATH = \"/Users/ridakhan/Desktop/PR3/\"\n",
    "DATA_PATH = ROOT_PATH + \"dataset/\"\n",
    "MEL_PATH = ROOT_PATH + 'audioimages/melspec/'\n",
    "WEIGHTS_PATH = ROOT_PATH + \"Models/\"\n",
    "RESULTS_PATH = ROOT_PATH + \"Results/\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Function to calculate mean squared error for evaluation purposes. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def mse(imageA, imageB):\n",
    "\n",
    "    err = np.sum((imageA.astype(\"float\") - imageB.astype(\"float\")) ** 2)\n",
    "    err /= float(imageA.shape[0] * imageA.shape[1])\n",
    "\n",
    "    return err"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Creating the mel spectrograms and saving them in directory"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Song_path: path to the audio files\n",
    "    \n",
    "Save_path: path to save the spectrograms"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_mel_spec(song_path, save_path):\n",
    "    for songname in os.listdir(song_path):\n",
    "\n",
    "        audio_path= DATA_PATH + songname #+ '.wav' #location\n",
    "        sig, fs = librosa.load(audio_path)   \n",
    "        pylab.axis('off') # no axis\n",
    "        pylab.axes([0., 0., 1., 1.], frameon=False, xticks=[], yticks=[]) # Remove the white edge\n",
    "        S = librosa.feature.melspectrogram(y=sig, sr=fs)\n",
    "        librosa.display.specshow(librosa.power_to_db(S, ref=np.max))\n",
    "        pylab.axis('off')\n",
    "        pylab.margins(0,0)\n",
    "        pylab.gca().xaxis.set_major_locator(plt.NullLocator())\n",
    "        pylab.gca().yaxis.set_major_locator(plt.NullLocator())\n",
    "        name = songname.split(\".\")[0]\n",
    "        pylab.savefig(save_path + name + \".png\", bbox_inches=None, pad_inches=0, aspect = 'auto')\n",
    "        pylab.close()\n",
    "    print(\"Saved all images at \", save_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "create_mel_spec(DATA_PATH, MEL_PATH)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Loading the images into an array"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_imgs(path):\n",
    "    dataset = []\n",
    "    for i in range(1, 672):\n",
    "        imgname = str(i) + \".png\"\n",
    "        img = cv2.imread(path + imgname, 1)\n",
    "        img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)\n",
    "        if img is not None:\n",
    "            dataset.append(img)\n",
    "    dataset = np.array(dataset)\n",
    "    return dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(671, 288, 432, 3)\n"
     ]
    }
   ],
   "source": [
    "data_set = load_imgs(MEL_PATH)\n",
    "print(data_set.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Sample Frame"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAW4AAAD8CAYAAABXe05zAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzsvcmu7MiSLba8IxmMiN2cLjPvrbpV9QpSjTQQIEDQD+hXpE/R52isXxCgN5Imb1B4ddvM0+wmOpLeaeDuZsbMvPdJo3dOgQYkMnacCAbpdLqbLVu2TOWcsdlmm2222bdj+r/2CWy22Wabbfb/z7aFe7PNNtvsG7Nt4d5ss802+8ZsW7g322yzzb4x2xbuzTbbbLNvzLaFe7PNNtvsG7Nt4d5ss802+8ZsW7g322yzzb4x2xbuzTbbbLNvzOx/7RMAAGvfZgDI6YaU/d/8rNY7dO4RAPD9+N/hH9K/AAAWBDzrTwCAoDwO+QEAMKY9Dmqg74+6XLJRClrV39cKg9H1feDsEwAgAYipVJbGnDGnWM5BKTreS7rBoR1TY6nnf1MzTDYAAK98/f9C37uoZ4Q81d9JuISPAIB/Nv8jNMrx/+P1f8e8/KX+psP/dPe/AAB+644IteL1D/EJY96Va1MOGeV9pwzmXM53zgEZ5ZoUNHrFt33Q5RyP1iJmvtZrjPU4mq47ItNOb1R5paGQ6m8apbDUz/qcsDeOfqe9P+eIUZf399ZiV8d97xSO9eOvC+Bp3MHnlbD6raMr3/Up4xLK9V1CRK/L+0op3EKsY5xxStMv7g0AWJTXg+pofoymvHcKAU61uaFw15l6P+jrCIn/7ozCUA/daeASyuunOeKxr79jgFaw/LxkDKZ8+eD48y9zgq0HHa1C/VncwrrSeS6Xh2uI6GksNW51PE4+0m+5eryUeRxDynhJt/IaiebeZ/0RUZWTMdnCoQcA7PIOs5rLZ/AH3OITAOCN/UfoXH7/h/RbHHQHAPhLfqb5qaFwQ/nuQe3Qq3JRbXwB4ClO2In5+VqfEa88DnksrxHomZrVDJNtPbfynJ/1GS8oz5NRDrE+k04NeIl/BAD8E/57/GDvAAA/hjNdq8sOoyrXOmhDc+/OWRqbUwi4xPr78PS+gYET19SevxMuGOq5aSi8qjMAoM89Xecf9X+CqvNwj0f8OfzfAIBPr/+nmGlsm8e92WabbfaNmfoatEr+h8f/lU7iz/hPAIAlnTGatwCAkGec/Y8AgIfud/g+/wcAwD+7t/hhrJ6Rz3hZyg7XaY13O/ag97ZsWoPJqM4NnC6eBwAcbUavyx+noOgzCUDM5Q85TFZnTJE3wlj/bYrseU1R4VK9o+oYIiagOvO4xYxr9YqmmHCqO/hReKn/V/qPWHLxht6rf8L//PA7AMC9A/3+pzlTVHDXKbytm/id43P8cWLPrDdAV09SAbD13N71ia7xEhREUEHXBwDXUMej/t3pjAz25Nr1h1TGu9kt8m8+dOX9g81oDuSg+R5MSdG4h8zexZJUjRvKb7XPK8XHn2LxdIFyz9p1LInv9yXwNeUM9KZdU8betuvisWj20GUcbTmDOSmEpGgM2nk5BQym/NXrDF8/oxXoWp3KNE7XyL5Tux4AOAd+3+lM59vpDKfKH5eocaqfmyIwGv7MtY5HzDw2be4ZxWOxJP5NLd4frUKdnjiHjPpokefffvO1/sP3O0Nj8DQn7OvE2llQRJEyPwtOt1lTxni09d5HRefUaUWfkXPpEhR5wloB19CupV4HQOcSU6YoJSHTHH8zGHyogfjnGSIqKdcOAEeXacwAwIp1oY1TBugcreLfXRKPXxTPRcwZJ8/n2cbj1Wd8nMo/9FrjKZTo/P94+t82j3uzzTbb7N+DfRUY91ixq4SMffWyjXJ4i78DAAx5h6n7ZwDAh/wWD67gZ/98Z/F3Y9nWXrzC01IuR0Hh6BoWyp7M2y6SN+STgq9e3b2LGHR5/yFp8pIyFKwq74esyXNxKmOpeKLP7JW/7TJc/a1Xr+n4bXeOmXf2F6/pe2dv0M/leHedhqpu4t+f/lvMFRf/Hg/4rnoI9y7h2ZfPLwlwFdN96DLeduV8R5PoOo5Ww9M5KDjFGH7buZ3maORtF7Ekvr6uXlPOClM95lz/3+uMqXpj6me+gVHte8D3Q/nNnfCcitfFnlN7PWRF5+50pntmVYav5yUcIfikcArFFZyionugxO/doiIvdhBeY86A0S0vwOfcruU3Q6T75DPI434jPF85HqNJ2NtAYy2jkXac8nd5vbeR7sFDt2CKNS+SNKxuODTPVfmbjwDO9bpfvKHrtgp4qPP/GvXKewfKNbb76NT6nrT3jeIRXpKiexwzz+cpKpiK6b7pOQrzSVO+4s4Bdy7RWLdx7YXHHbKicS/H5PNsdgqKPOqfR84tSmrHcxo4+fLXqwce+zbWoPvx4DJ+uyse7h9vDqfqtT92Ge1KDiavnhs+Pt/7KJ6JlCHOK2Gp71+jooglZGBn+JgtkrjvFEbb0TilK/6mfRUL96PmhfucS8Igq4RDPAAoSSPQTTMUhgE8id73iZJS58Ahr1Ecgo82YW/KCLaFqX3G1YU7A7iG8mWtgFxn2s5ESsjNkSdagkJXvxuzopvVacDkFtKW925RYaBzyZjr+3uXEWtiJ2WgwVcOlp6GzmiM9SbvTaIJUtId9ea7RItASUAV63XGIB6CXBeBNuGatc2rNxnX2I6jVlBJW8TbuMvFMYPhAKvKeQKAz4qOfbBJhJOaQkiFTKHoaCItVL1Oq0Rgu6qQFT08RikBs/DGq5BxiS2BqcSiAXogneYw3Kos7h/PGQ6LFY11pxP9vtMJd6ptQHwveWkqdqvnsjOJ5k+vM0EucmEfTKSFe4LBqMvK0plEzsTZM6xmVKY5rRVoDEaTaOGkTQn82Wex4PukoMELUoMJMnhBkmF/bwBX58mnmZ2At73Cm65tHHz9nRh3BdDvOmSxUTOkI52JXrdvleNocT7tWW+L401s3m96fuYOLuPPt7Yx8bqwtwmqPtv3ljesmBXuHf/dHDcrbmtCJlgtZd68rpHn5NFmzPWEl6SEI8dQTMwKmjYDYGfXm+3PbYNKNttss82+MfsqPO5bbqFlQl+peza/w6HScpwyOKF5tUyfygDDHTqJRBQnBo8W5KkalTGYRunLK0+DQvCsKDSPmcNrp3gX703Gi29JIUUJBgUO16LweNt7Mmkkw8BrgNh5i18HAIOySCIJ036/XWM5Xw79nfAELlH/amg3JU1JyymxN/KmSxQ5tO+UHxBJQGQa7/abWlxLAihZVsa6enpJkYd3DmX8gJLAbEywEvKWeeCTxljvk9MJp+pZTkmvoJUWORiVAeFNzuJeziKq6GW0kBj+eNDsRTfCY/OqrkGzZ6gzRC6PvKtTsPR+SSTyb3JUwPOtzLFyfaMNmOIvH8OYFUL10ENSSJW+GLOm3/q8dDQeRmW6f7do6LqdAnSdQ9cKqwQB28xJUZTU60yUwatItDvF8/YqkvdBeIy/HTkKu0UIiIET6VGXCKDZQpFRiaCA8hy0OQbhrZdotryW3qZS/Oy04/kEeoZCzhjqFwad6Vm9BEWJ4fL8o163xpsu0JiGGj2ViJAj53Yvs4j85sxJ9dFkehaWxOQAGSEYBUy5nY+YWFgnjX/NNo97s8022+wbs6/C426YrlEa9yi4dswJN5SdT0ERgf3oLFGS3vYZf78ryTurQF7BZLXwstkTkAmiJDC8wUTyLv5wHciDvHcB7/vyoSVpSigtSZP31mumtL16TXipUYX+BzDlxyrGU5cISvjMMVPByWMvihEWjTEVb1MpRZ7LoBUsYaTs3c1J4WhrAUzk5GdOwC2LvED9vwZjlykDU2ScvXkR1jC2B4HZtmNrxTikFzS+lJnqdhOUt5A14Z5aAbZ6iYOOFPX4rHBoCUmdMLQCoKxW5948symxF1rOgV+3ZPCUFGGRUSaRxHhwrCOTdHytMWeaG5dgyHv0iT0wmTsp3+d54sBz6VLzKJdgyVO+7wLTBIOhyEGvjhfpLI3iPAn0GlNv37lGDVMH55WiIT6vmNefbRZzwWaBMq/aeA0m41bv38XL5DbP973NuLMtAuKIVEaZU1IUIWrFkUwS15DAz6hWMrLMq/xMO76m/IOIYKPCObTz4u88dhlvOk/X93lphVUKQx2HXkRPe5vQZsotGprTIYM8+iyu0ahM82NvgKkO4NOiaPx8Unj1vB5Q0lwDPsn0+y9t87g322yzzb4x+yo87qGWGWsoXFPZBc+K+TA5J6LFTXHA01x2r9/tmZVwCozrDYYLKA42kTe0Mwm9wNiah+KFlzSYJLxJ9qDkZ5TK2BFWzhl8nxThnjEz9cm0TD0UGnS8t+x9yCIPSVUbDP+RcyZvfbQJmnZ8LrSJwsuVkYYC8FDT/wWzr3mBLoqCIU2e52gSjaVV6/LuNgrNqy04cnm9CA+318zQCJnzDDJCcSoTy2e07G0+LQ4n36idbJ1OdD+UymicipgVRQtT4mhkNJmxXp2JKXKNfJ4Q98+oTDmNdu/eOJ4PnWC4OMXe92jYM+sNU04n4cF2IqJQYDx/MBG3inE/L90qj9Ew/Fg/V66VWTODTuTtAUxJtAaEiUs6bKMyyghBY12M1O7NHIFI1E/+Dad/HqWV1zuTEevgPC8cpWWwJyxtlLRQxYVNch6W61R0bsSyiUr8Ls/Jds+mwNc3iucsZkVY879eFKa0q+8DvxtreXrQQtaB7+uLFw8mmHU0YM2Uab+Wf5Zfaba3EHkSjjSmmCnacIopvn/NvoqF+5zKorxTljjMTd8AAHplWfcjJ9wCL7jtQS4JuPL5S+CJ4UXy5Wh5AfZJE/3GiTFyihN5Ckztkjcxr/i5XAVnFd9ErYClTl4Z9HCVl6Zwbo6Jqij/4WBo0/nDBfgpngAAd2nEl6WMyd4kSkr5xA9Wp7GqPmwWxSTK4IfzGvmR2plEDyrASaSrqM67t0mEgnzNqr4XsqLrf/bMQx4NU698Uhgt09Ma/xooPOZ2LrNY9CghqjmBKiGJc9C0SexNos/LfI+CgNJE1euSRJVkNPTQtnvz7DXRHdeh/pqSStzxrOjcJWUxZo2xLtYAcAk14RoN3Usl6Ial4peTZMQlT4oglJyZ330StD6ns4CR2JloV53EGHgBGRqxzmgxl3/ukLRE2hR50Tx7fs6myJWQ1wC4+mDuxKY0iET4KRicqjN2CcBYV6W9WNznVLjcQKHjtuOHBDQFoHYuWsBbIYF0fZI491vIRO+NOVPitjdAW7pncd1y/ORzLvKnBEXJcwEKvNTm3hJLVXM5T4VTnf45sx6N08A+/O2Fe4NKNttss82+MfsqPO53tniSS0qYq/edkbCvdECtFHxV2jJKkVd+i4qSMwlc6DJF9iiMzUCrUBTenU+aim5MZo/GaQ6NT8HS6ynqlcclK/tYL0JRQkcWrTRPZhGud8wlKVm+x1oUTzNrZ4SU8aBKcdK7vsebWhU5mISfZlfPkd0ko9jz8wLCCEnhUivyRsMKf9Lb9ElR8jUKWpNPauXFt8/Ia1oVjgg4ot2nKXJ42OuMa+Dqx4PQ/vj9tYWuCm+7X6pEWp3pHk6RYTLpfVwjR1JlPFrVnoKP7IUZQUNsyaLBiPcV/7ula2VqYEhMa+x0Xs2NZvfOU8Q2R41UVewGE+nzXkQ5VhzHCY+0UOGaF6gJQpkF5CFhHKsSJa3lfW1nqAG4FkWIORCSosTxaBh6eAVHs1avE9OzdDmrPXScnEwioRzzGiJhamkinZXPylAyMQptmnKNqMdkeLDTmaiEdhUpVw83cWWvUaxxsrP8rJZnqLx+2yXcO06Gt/VlTqwmmsBQiU/rykkqKBMQ1Tnwd40GXj1HhO39W8y4qxf4ts+4+F/CLNI2j3uzzTbb7Buzr8LjblrYWgF/ymd+H6V2P+aMWNGkXmsqwLlzGce6O35ZLKluFRywvE5BURmr05loPLdoCOMqiStOjrRdUNLbnFB6mxNjl+egVzSqZ9+8dfYQaKfOjKePNuNLRc1evcJUXe53A+/mf7IaPnH5/ZuuXOtjF/Bck3fXoEltT6rR+aTIU3Qmk7c/i7LmMvZcbNS8fompFhobexcLRSPNO8eKetm8CaNZl2XJiq6pN+ylOQUq37YKiFQ2TqcHqzMVkJyDWSWOg3DHmqc/p3VSqHmnPnOp8c7yuUmlN4W1FghQ8HbGvdd5g6bnMZpInu81GIpulqQJ/z/YKGQVZESj6Xy1YgqeBasPhqxJw0R6ab3w0A82Eca8M5yQHwwnM1uCTeaDjOFjLElR6XcZu/oZJVXvWOJgMMBVtfHludeLY+7FCiO97Wvk3IG8pk5nUucziu+Hzwq36onfO+BaJ+vJsy5Ry3NcAssn3AJrlbztMxpSvaRM0dXeZvL4Uwa+VGqgAkeEg0n0nEtpi2vUwqOHkEHgPEkG66yzb9+euUzvN/LB0SbSbv9r9lUs3PfEXdY4ne4BVMF9U04vIQOpLOIHZ9DVhftoE74bitD6oDv8/lq43r3JOLQHQGV6eGQ2XYbiow3o6gIWVpxgnsQv3iLUR/hdFylxtCRNsIVPipJ6j53Ca6uuJN0IkWTKnPF+Nyj0plXG8YLUaQWfyx9/mjxiLmOQModK9x3DLCVhwmGsEQ94u6ZZLOhzUivmgxbj1GCAnYn0QEptESm+9VQnegbLfhrFbAKlhCRmZojBqkTvdzrReMxRE/vHqETaMeXvumjohGurctSJ4JdSoQgay2azV6sNgXU7WFej+5X4s9eZuPESDlmSxljft4rhtdFG6MgPY4MDrsEQBNbpiJ1piUcDVPgngZOuGWr1e+23FDIt4hKiMoorRm9RU/J8FolYlsHNzFjJDHsVFk75zCUwlKAUJzCd5nFdRHXl0XJzhinyuTnFC5UUDDsqXuiNEBV7cCxwdo3sTEyiGlMlwYRBFsnV8uZjJ5pv5LZgA78ZPHRd8k7i+s5BrYS22gZz5wI9Z/Nq0+brO9qEdx1vQPyssDbOg0t4QXMS5XPBlQM+KZJM/rJo4nf/Ndugks0222yzb8y+Co9bNilo7a5cMtQmSkHhS+l4hCkmdNU7vQSN56V4oXPSeOjaZzg5+eDyKrnU6+K5yP1sioaSSDGzlOtgInk30ltTKzVBhk2yYp2TUXzeRebv9pqTcW1TPVrg5JpXLrxKo6i12F5bdJr1OyQtr/1UCbfK8bvMYZ5SGReiO7H3bdTaq+OwMOJar7vXiWQo55gpXGxeUcxq5ckNrUmDUF872kj0uXPQrN9hOHDsBcVxFv5EzJq+2+tEUZCUMVXgSrqj4O3LJFUnquCmKBJNgs4px6Ip6t3ZQDQ+n/TKs27nK5OWRiU8VK2La7CwosqxQSvy/o024OhacosV4kLWNA+TSBZ3JmFpnPVoMVTP3emEl/osfFkcRQn3Lq8gnna8FglIWOzkDXmb9zatYMhWayArSQ8uE0z34Dgyewma5pJPrJ55EHTSKDSBfFJ4cDw2pxphdUETXdSI5hMpM5wYsiKqHVeA8jk26l27T800+Bi9Zt75hz6KiDDTuazmrQLavC0yvmWcLoL8IKsuSxKej9kUK5G5buNdz1HCLSpcZAeHX7GvYuF+Xjg13cKtO2fxtuey0qdK1nSaS2V3hot8Y+aiiiVx+DclhXO9Ke+6iA+7Cq2YgM9TgVakrOsUDS3WAChk37uJeM4n76hkWeJgPjMvudccjjYrojXldQbDB58WhdeK3Ji+LKjlmvi7e6vx/7yWY38/WHpI9iaTKI7TmTYGBWYm+J9BHE0x8pQNnXunEw6uHCcktQrHqdhGXAstVJmFiaxKuNXvSY0cmRP4rvcrXnH7zTI+rfR7WWHA7TqsZq750RkhjykEwMSiOEVDD/XOZBLsMkpRrmMRPP83XabxaNCLT27F2W7848EEKoop9QQ8Z6jM3SQax04zxl1KvPn67vqFXt9q7sIqvg6liiQAAFgTcWiSqYsjPrgPGg/1OFK4aoos1dA2hQRe5GTp/t5yQdY1rAt9tOVnsVkCVvh8s5zX978XWHZb5DKY727FZpDl4ipglkHUAqSMXy1u4QIyxosLd7u8/+oNvb5z63L9dsL9Coc3dL8lrXoW4mxzUgR1TZF1753IC8iNxqgMl/g+tDXiFtVqzP5LtkElm2222WbfmG0L92abbbbZN2ZfBVQyWtbXIAqSLg2AmzXdjneDZjUwlQnKWJKuCl6A1aLTjEmUYR9NwutSS421IbxUi89oBRxdwS1kOAdw+HdwgULgF28J63SKQ+0pai7tJi0RLijYidZGhR5W3v88czPfmBN1wvCJdcgPlvHgz4slmCBmRdh00eluEEYmvDRmJQqGNGmudADRzzIU7hyzbhr8kbPCoYa6Ep9sobjTzBKZkiZ2TBnTEqPurWct6GgIty9wRx0vnSnO7FwgeCRBIddcxM4GPM8F071GVmssEFHLESxwhFEaQGbwBS5K+KrAuFtIrxXnJQaT6BwVWFtisBGPrkBwnYm4Vf1wEzP2toyjEp1xNDKxmKyJKwjKCKzcCV2dNsbXxbEGjuJ5kLJmbZFoVvorjW3iI9NfT5UaeIkaH+fafqxLNB9iVlgEU6dBJIuAKHJmWOjLYojdkcCUTKt+JhVQ59LReXr/abFURJPBhW4SOvg16AUAgugq1Wxv00q7puH2Dy7iT1OZDyHxeXmhfiifCa2YzdNppgNazWwdJ1QAUwam0ArdEj0jSbBNnM54rLmDkIGPytFrCY06iT39in0VC3fT6ei0JoB+iiCM2yeeMDvDVLWPs8Zo+l89JlXVKRboMYp7FhqVCV9dKxKw+FRIa75tJDxKr0T826Qqmhm/FEFqiSBJvVqSor6RT7MQ67F8A3vDC7rVCm8c49ENg33xCl0V6bp3XJHXaa6QlA0S5qTxXV1MnE600PusEOsiVxKFxQYTMdTJG5KmjaxVMMrKSgtuEDAnhUZF7TVrjxhlxL0RC6ULdG+uwVJCUKkMnxqGaFafH0QbOhbXyuJuasKyr5Ex20vUdB80+MGTyTNpbSxS5rmydx5dHZeYeLHRKqNri19StBDvuwVLHd8lGpyqA2G0RaDGD8Bd1WuRi/ZlcTQ2MrEZs9TbSYSzW52hFePvbR6MRLHMGKpc8ZusCMMvMqPlHH2GkF0FVROWTe+XiWkAtEH8MAS8rdfx7B2+1F6wPwwLOQRaZWrHdg0GHyulVvLqb5GTlkZxYvxJ9Gs1iskHrJWiKVmdM9DV9eIaNdUZOJ1pw5QaLXNSlIxPmZ0ZpxP6+ps7FaDAOaCTaFCxq2P84ILoOctr1skbLORYJHq+RwuqQ4lZ0Tn/Ndugks0222yzb8y+Co+7Nf8tMqqNspVwEancRo/5OCmqhOq0wqda/DHoTFolb1xaZa8bTWdv46olVgtjExQ+T7VNmubO44NNuNdNTtZQpr4dox3fi8x6yxgvibvCt8x0qbarEUXSeF6aN59xV6mMjVFSPr8OERtytCTuIF9C+Uy/I7UupP4KS9gy26Q3aUWx4uq1SJ+fosGoAn2GICU6R6Y1uoxVRHPnuEDloZ/pOi4VSkiiItDojF31xoas8FphECcogFqtu483b/NtP+PVt5BzDd0Qdc2mVbjaCi6ugRk316gFda3JzUZq9jAFQ9cfs8Zcvbc5GSxRskq4mKthPue5p+vYOY/HXZXZnTtSpksCLgICefQJiiKQORmCkSTjxulE52l0pvGLP4MqyrlzZCQbSQwaK2obRxE852fB+shYa6U43eBDTVHg3kR8N5R7v7eeiqlelk60b4v4x3p9p2AJmnvs8qqASEretmf9eVFi3pb/T2D9mZ3heW1UJjnlS1C0viRw0c2gE3nNOxOpddkUzUrJUlrz9O9dpOf1Eg2du1almAcAjo4rWAfD13QKvI6cQ8br8rc97q9i4SZ6mFJonRKL/m570ABD/GBFGHCnM3USP9hEmK5cIA9CStNpxhy1ytTL0OmE97UCU2KRAHfkKCEqL/S3OgHnyF1TjjYQ9OA0d9KW+rvtvSUp4pjuDMMjhVdbXofEnXG0Ar4fyu+/62faRGIeVlWRhP9nRfjc+hxA53iNRkzSsOqgIql2X+ayqe1toE2t5RakGJLTmaQGtBLHgGKefOLFKUPhXENzKXnqo6aHuncBI8p137yj7168XeUo3taNYYoMxZTelXXRFXDGo0uC1sZhd6fTasEGgINbCMoADC2OctEcDC+yWcy9i9cElQw2wNdxDNFgrFDCoV9wXxc2rTKmOq8GG7AbFjpmw6d7HakSeOcC9aWck6HX0Sua2xJG4j6UmSQT7lwgvNvpvJoDDaP9slhahK5RE8QXEuvK/3aXBDatV/Ptvt6b424GzkVI7BYtQXDPc4fXet2yd2bJ01QYTjHk57Oic94b0RO0/v6ryGcEUWWZsiI++kOX8a4vv78TjthoIkvCinVA9gyNQj4hZawki3kT4XyPFXkSnxQ5cp1OuK941LPX1GfyYBV1E/prtkElm2222WbfmG0L92abbbbZN2ZfBVQiMWsn1Mmeqozc3rIi4LtekWhMFGHK0UaqYAI4rCnKf43ekyl07UR45IWKm1aAbqpv3UI46s07qmozKmMhrJXx5igrLnUiXHAi+ITLeUeAMtN/vGkK5+5cRh0O7KwizLszoIz/vfNU6jwK1TKrhJpaYtUygOlkg7huCWf4tO7Q0hgxow2ESy5pTTNr/+duQFwJdrAspJShiEVhxf0wKhH1srcRQ/39zkTEVubeBaSmoy3GVwqAAesm0LJhcYN0RhuJCVNkDcr3HrqAe1G92a6PaYr8Owfnf/G7AHA3zHAVz59mR6/3u5nGQ6lM72fR4m2wAeO41M8AY8VdrU3Q9dzf2gvOlwKzzMHS2ADAWMXSdjkQFS1qhmimYKhJbrtnIWt8WRwd413vaUwb/CNhNqcysT4AVjCUbQEPNpJA1WASvq8VylMweKr5oyUaoiz+5nDG/b585u4y4N9ejwDKMyqZcI0V5FSmjko5gap/Dyqt4Amg/JtVLLHwtlYWv+sX2KpvDwAfKoRT4KQmlKboHo82ipyKwYMu4xQzf6Y9+0B55tsQUJ03AAAgAElEQVQ5vHhLMF3GmhElIReJ4b8vhdx4dBF/0N+AOuBdnRMpC6lFp6jNl9EKO90SQZnKVvcWVPd/vXW0yH7XBxJVTyJhdt8t6B0vDodaIixpW7PATp9vA/rG4zQRu/rd56kn6tVRPMyTaH2VBNbZFssXb4lPvTOJkpedVnipVCDZ0w9g/YIpKoH9KVpMB5PowRsM62fIiSI1Gjqd6AEAmO5UfqstMoxdy3ZmCqw42B4WOYmd5l6VSjQOSFnRAj1apv2lrFabxbliwId+hq0PW4qKeNG9DYQ3vxtveK2ftyoR5l+aEZTxvga74q+3OdHrRBvZKRha0HvRjKDd3040p3I6YVfnTMoKc6NERgNrBeWy0t86G6FEb0fTxhQarmH4Q0CoeK33fJ+UzvR+CBp9HQ+tMtEKp2Bwq7K/O8c4+5dbt8KY231tMg2Fb8/z8OebMVDmxaHesyka0i3pk6bFZtW7UszZg43kWFiRTN45T5uvBhCbxLJ3hHHvDBMIFtE0IqtMc9UnzZIEgrbYVAfWKp1R5C0CJV9l3cAlsIQEsHZuKFfmPKn6SbXKOaxzKjdBlWyLulWZ1ESv0RA//rIYvISW+2Hdkp9z33/NNqhks8022+wbs6/C4247pgcXojwvCcdKkZBd298Pa4ZE6w59i1xG47OCaiGMUG67Bvamjc60gxrhgS5R047qk8YQa6WgjaTKJnWQp2jw4lnQp/3Wi2fxnmZeFCz0mgsHShszvp529CVydn5nWH1sMOwhSBYAAGITZIg2ZoJREbLGVGlsPitilcjipM4kCuU7EwlW+um2o2tt11FU29jrau8vSZOHK+GUmBWxCXTk3zz0C2w9F6UzcoMsugBtuLhlaV5oNNhXj3CJhryxJWliHRmVycv0WbGokGXaVq+5ObQs1mr//uqZtmZ0xlMVJlPIFC2EpHGdWtMPpiPOwZKXrYQHZVTGlMt9ermIiEZlHCqT5HrtcKnHBIB9fd/ahFu9f52NOFfa3y1YusdSjEsr0cavwX5RczUs1tCSbBzSKJZfFkuUwVPgJtcxA65GCXNkD1NjXSnYnsyQWZ3zNXT4dCsMk09zx9WKAM2xW2QFyE+LJTbGkjQ+LrXQDcwaOdYHam8TUSw/Lgq/qRDEaFnr+8ui4XT5h2vgzvL3WVYFM+QUk2odENGbSHNewimX4PAqGDon0RmehehYA99nUHOIc1DUsu3TbPBl/gbogL+/lv8fHfBhaINvROl3ptdHm6hcuTRSKBP6y+LwaWaVrn2dCFZMHKcT3QgvJF5Lb7sa1uhE5ciDifT+aXE0qeXCnTJTzmRvQK0YxuH+hXwz5JLea1YKlB3IBwPq25gzY3hOZdxq1/s5KnS6YdyJFpOLeJA7nYj+NYlGEaNJFA63cWj/z6qV9/KZFmnSOn5iDGQDhgaVyB6dvU44+VZqrHBAKwMHcUGXaGhxi4IvuwRDMEEImvDdSYSoa243wzyDjQQLTdGI7kdcRTlFjftKy7TIeKrYbwvXrUqcI0HGwVV4LWvaaHbOE1QSo4ZtG5POuN46uj4Ku0fmtM+zJQrgEgzBLEZw8oee75HWiXDt09QTlTAL7vsUzAoOa+PRJAIGk1aNRNonZ1EdXBYXLpFvc2ZvEs65UVqBqR7mzvJ8kJ2mXrwmiVkACKZVaXIV7ih6cF4FxPfgAmHtbzpFCzqQ8A8jUwbbOQ/CUZKqe3JDa8/gmy5RdWfK3QpObDDjSTxDTsBr0l4WVo+UTUkArrJdkuY+mqLuQSocWs0ba4EfN6hks8022+zflX0VHnfrS9dpJrA3cXkAK3bEJNoZARAJCxaKuUVFHubeRtwLr7KF5jEp2ras4dBy33m4yn6/LI4gA6cTQuT+j1J8yq888PL/mLkPokzgtGRNSZCVz84JeFMlV/7lOAnvY6RekU6xZyGLSe4dV3r1JlKS7hwMF9oETd7/oDOJX0mv/xYseTTSA5JC+9douErTcLKFBIIyJ4UGk2jcB1HUYIQW9WlxFCFYncizBkAMDGMTLpVRERMn9ZyJeLkNdazXXekbo+A2WxrjUpXLYX2zDA7xe5QiqnI+fJ29KK6xlLiK5PWsEpPeYqkRw2GccdgXjzhGRV5gP4RV85BmGsBLZWCMnSeWDQD4GnbPC2ubyGSt9Abf7ibhZfdU5NQ838HEFROIqlcFnHONCo1/IeGOBO4Vubd8L53O5HFeBWwy6CT6VYpnJmka405H1llRmZ7L0kW+/HEKZtXbVWeen81aRaJPip6bQbQLvAVLhIBHF3GsUUeBhDjKlvamRjRGJ7qvRmdiSc1J0e8eLLNKBp1oDG6Rz7NAopUlJZqpTDGTRsr/F9s87s0222yzb8y+Co+7meQVn30mWdejU4iU9AO1KCs8UvbwGvYtG7zKHXRJnMQabSD8WquMWF+f546w1k54sBoZj5X3qZHxeR7o+O2cn5eO6HuD0AFp3koGY+BTZHnK0YJ0SyS/9uQVtTzaWeB9xToPLuDH6pnNSUHF5nU58u6/G2byZs/e4S8Tqyg2udUv0VICcTCMqWrF/OfORKbg9TN5WC3pZ0VSc05c2l48sMab1uShl2tn16ITuKQXyZz2egmcGIuiQ0wSSdkpGNxXvHKOhs4tCeW7krRsPF9FnpdVrCVjVRIt2RouquFrIuwSHDf51ZycTFOHK/HUOY9ymxx549ZEep0zsMwVyzaJ6H2vc4fHyn8+HiYag5g0ee5KZQTB9V4Wvtbm0YeoKSna64g3dd62pLvVmXIXr4vj3InQtHEinyGpbTqumw63EvIpavJylQIOTVbZJsabE3ceuncB+0qv1ciCmudwXjjBJ+dkUyjUWEe5LQInYQKVSU7iFjX+UqVcPy8WL54TqG8Era89q89Lt7puzgtESvxP0RA+fQncRWpnItYt/dr9VkIhlDt4DSbjVMdPgdvWdTrjoefr+zX7Khbupj3rUwkZgHLz9wIuaayLTiv853NNMDhOPMgkxZzAUIkYQCP4lFplYok4rYmj3dmI2fOD1ELak3cYc9PmjasikpC4S3ezF28o4cgJjkxa2CchoPVlVrhU8Yy/TJbCp1uUXeEz3fyQlOCxKqGVoDGYpsVSFmygbBityORpsdzzDiwaZFQWWttcdDPHgRbxSeg8N3hEPuCD4f6GVglYKmfMkRkSbZF9N94o/Fyiwd1hos9MUzl3pTK3CItmVdDStDkmGLrWU+DN6CZYPrIV2M5ECksvwdDiXoTHeH40a4vKnVtow5CaIbO3OFZ+9673lGDUgnFjbYKpD3jwBqnJFO8DHlCy8+kVlKjEeSD++s55Woi1zqTHHYKGbsljAUNcg10VJEkHBSj3pSVhpQZ50eOoYlKxo6RlAlZJd5mMbotmzKDzOtqENxWGOLpAzlLMijaG3nAiPSaFv1z2AApLozldr4KHL+fZp8USLPNrUrznoKlV2JJYA79Dpuf5T5OBUUP9zUTjMAiYcEmsmdOZiKFBVCYJVpLjlnBZrfRUPt56Gr+HtkkpkIPpE7PlvObXD+6XbQ9/bhtUstlmm232jdlX4XG3pGLM3P1ltExBukXAe/5sS+SNNhEnuPeZPC3Z9abXCQ/Vw9vZQKG8+VlFFyXDBP3tUVCA5K5udMa+isAvUYvQnMXqu5WUaoVhgqYA3ypQmBTzL8XggcLx1K7R2RT+VKll905WRSr82614T9/1gboJpcxKhDIhaDXQ0IlBJ0rGPfQLJWivwXKTZNGxpoSITOECilB+g4LuXUDOnOhqx7CKx8IIz3cRlL5DvxAEkLOCqx7K4TiR93E+9aJ5gcI1l/F4P96IJuiWhN5w5WSDZXZGdvjh5ghagdXxhFfXjpEzMLXqRBtw33GyqlUB7pwnyt40OxrrXRcIHnk57bDfle92fUSsXtfL04iuJmXvxglTLS2XZe3ORQy7cvzLpScIpSTMBH2Q5HojUQajrD6slayTgKSk93gJVnDyWTbiy+QIGjhaTjQrnWFU6/jCE1dW/k2CBpmh8G641TELBHsplfH9rrzvdKJI5iiacstGJhoQXezjCjYBgO8Nd8BZksL7jj+7pKYICjzWce+E/G+JKMpxjMrrJtDE6QZBqD4rHAjuy6sGI63atNMMF3U64bE2e3713NDDaeDOcrXnlBgy/TXbPO7NNttss2/MvgqPu+nQGqUQKgj1h0tGbxqNJ6Orr49OUJaCJl3sBC4kWJIm4XJZEXhnEg62VqYtHXklSmWieSnFfS+tSTB1N52DJdy3U5y0PHlHAuulcpDF7+mM6vYo2zyVKr3y+sPAWOG7TlDLRhb5j5n7Wd67wJ6voCAVTYTqmQWHu45blJ1E0rMVJ8lGA7II6f14Q9+qAiM3CdgZbhf2pnpI770t4v4odCunuelCE9QatMK7qnd+6Dzpxex27GXLcW+/CwBdjlRFKXpMIERDnmKnMhSJ8rPuTG9nEjhKWcESrs2epVGZNJ99Vvi8FC++RRR3zlM7sd4GoiwWoaialBo8XPXqvLcrLeV+LJ+/x428rhj06lp0Pd+UFB0/Jk4wXqaOhLZ8NJiFVkmLOHc6UXWlUpmSpcd+IQqljKianYNhDRehW+KTRkOHJZZ9jZqiQ0n7OwVQu7C3HTe/2NtAOY2Td/TcOJNW+Yp2z56Xntu3iSYaCVgJwbXEvtOJxN0kNt0i1060KCt/82sSikoab2qOIiTG4fc20DzZucB5BgCj/2UkcAmOItgyPixq1n723gVRGGeprRvA1ZXFQ8fftM3j3myzzTb7xuyr8LgJ1zaZdu1ec4PNvVVUhOMUe16fF4OE4lF913vcO6bltB1JakdYnbCrXWT240I4o7WJjum9JlxSqUxY4/uHM+OSlx2dT6c5gz6YSPS2KRjaiZvnWTrwtK4bCZ9ag9SsmBa0YjOAaGs7k/EvdycAxYv6w+lA18eaJOzGNdYL0Eq1G5VQqhlyh/pLYKz62DFzwOiEXjB3qPMP6bOYVSeYRplas1SiYO0EDAOXvBuBjd4qVq8NR0DnU09aJcPg6R58vuy4SEgnYlXsXKDvfrqO3KJLJ44MouFCCZMIl3SKGw23prbHzgvdD6biaZWZVTJbfDmNdC6ycGa6MDumeeUxaNgm/Xq/IFaK2uvzjpgyzkUcDwUTf384YzrX4+iMudLbrlNH3rRSIK989pbofi9TTyqYLdL5YBJeqrLiJCKXazDkqd65SMUvShTCdDqTrsjRMi1PFnkdbFrJwsqG24R9B0sytHIOPXQznpfWRjDR/YuZi/Oc8qQm+HHuqNS9/fuLZ+2RXjPT6Rq1KO/naCFlkLbJaCMxs5TiXFhMCkss92DvPN37a7CrYqb2fGuhz6NE56uYFXnil2BWxYSyiExGZL9mX8XCTX3rDPDQcdjfFq3SqqhdCUuL7m0mbmfGOlRqcMDReaYdZUXQx+7ocXoqE8R7Q+GqtQm5xines3aE1FAGOKxf65ao1WSUyRRgHSZdxcKuVcbRts/whP958qKFuEpQ7aQOwr0LQjCLu36HpFcLeVtcS5jcPs19FpVIZiqV0dR1zr6jB1j2gbzUsPxVCGsZxdV2UzR4rpDFaelw9GUhGTtPibmuD7CNtxw1+n2FU/YefuYEUa7J5XfHKy1UjTbX7kELaR/6mRJghcrYFh8OnmNWeF+r43r9y7ZVn6eB27W5QAtM13s4x8d598OlvNAZ4VbD+59pV7QFdxg9un0daw3oevrHNMENYsGrc97uM7pY57DX2I2V8qkzfJ1v1kScr4Xe9rJ0BKE4nWiTldfcFusXbym839tAydw58uJ3sGnlCDVoMGcl4JFAFbkv3mDQnLxrglZHFyjp63Si1cfozFriWdFcffVudc/aOS9J0z3ZiURkOy/pABq1hiBs3XiPXcJvxwL3yarhwUSqlA2RoVilGGoqgl7lmAfnySn5fBtwq1BQp2X1Mevwv3iDDzWRHY3CWcBWkpjAW8yv2waVbLbZZpt9Y/ZVeNwtSTcaFim9RW5868CJSg1uUOoU7zxe7MIAF02MNtCO6JMmKpVbRKf2rEjR7TY58rIBEIUrRIPcoBWdcFfDzze7iby6s+/IY5G6DC2sWktcrnfYVnQjRf4TFCUrp8TNE0YbqBruEriBbulAU47zj98/wdVmqJ8/7vHlWiQ0H3cTKyRed6vigVYVqAF0rUpNc3Xefc8dXYhC5p3QX+HGtPerLvOcIHaao46YFHmMKSkS1t/tFoauZkMUOaMTXqtX2Vum2t25iSOjn1HDZCFKC6UzFCVxy3lXJUCTcK2Vcs0z/G5/WUEyjfa3v1vo+9dTh9OXElEYkwhOATjJend/w/6evxNrNWZO5b9my63On5nn4DIldEPVUOkTwszHb7BQ10ccUBLA78UYdCZSwrON0UUkqp3QopHdg5xmL9sj40HoyNAcAKgQrNeZ5nOvM44tAW0Dea3SlMrYVTjqTic6p+e5Jw+9nQdQPHcJJRxFRLGrioNt9RhMgqNqYq5c1mACw39zPOM3j68AgKfziJTL8zFFg0uN4AYT6FncW4/asApaMW1YqmcencfLwmPZIoFbtDT3HkVEPeiEN3VcpSS01M/5a7Z53Jttttlm35h9HR43Jdc48XiNCnvSc86kSHbvMu5E+6FWrrw3cdXyq+1w12DJwzz086rv37Cr1LWkCC9NkXf2EA06LwoxqudwE9hbFqW+KYOSgIOJeKleRNuTfVJIpGXMe+ag2et58ZaSLU4xVmdF+bTU93heOrT91yf2Kj9+OZCnf567+rmi19LwuVEUU9wi0+hiVvhyLh5IZ0WiNynSLZFFFZI+5eR1Ubkwn6/TieiFPmno0K4pkr71NDuYqtY/Hhby8C7Xjpo6yOYX3q9L4V9qs4PSiELRNbXXj/3MnlzixhkIfFyOCrjpgNRzjz6gG2thxw8TWovAnIHpqUYRUcHUUOqnn+7w7u0ZQPGaG34tk9E5A7drvU/ekHevZXHLkNAdK2Y7BfhrpcK5jFi9X1nIo1TGrb6+Sk+byvg93b9J4NrlMy065OI2nzQ9cz4pKt9+1/tVhNW0aabAMgkKirzWORiYquL4mgxFPaMNq/L6dh9evF09My2y64Vn2p6bUbQ/e/GWyANv+hmq9px8XjoMp1Jmf1qcyNkIvD0piqxlv8yiRtnojn6V55Ka/e3cR1HeXzxxQRNU/L1LbUoRMvDq/7bH/VUs3KSlAKHNkfjk91ZRZZZPoIljRALTKg5NnFgIO63xUJNP1iYOu32iUFsK9HddwK5WqYWgMR44vLWnEg7nrGiBmoKhB3+KhjjpnY6UWW+Jj+fFElPmwQVqwPrjbAgWKlWO5fW94x6OTmWCRzoTiZ8se/296efVItkWs8Ic4Ay33GhkM1aGXHhCdSbSYvU6c9KrTfSzZ040wL0lAaygnRZSSiGxznA4ftjPpOVxPvVom9EyWVrc+i6smvK2cNV7t9qQHyvH/Dz3eK0bllbcIHqJGj9VSVjZKNonRQnauyqt4pPGvj74RmWC1ADAT5V1IbqVaAvinc+Tha/3PmZF+iRKZWoWcb12qy45LVmrdULf4JEuIVRo5enPPbrKTtEm0UI/L5YlVkWzZaMTVZs2BsoSrLhmbqzRaZbumpIm3RKrMiXXVs00oIjJVaR7y282hgZQBK3e1GrJ3gbioO+c52cuR3KKotD7WJLGLGCTJgb21/oxEjtGMGVevMGbjqFHFnnTGOvcmCM3/L54bp4wmESVm4OJ9HyUc2BRrPY6CS0WnzQ9uxncW3VJeiU5zU25E46itsCqrXJys8022+zflX0VHvdD9ZYKNND4juwJHAUdKYF3m51h79QqTk5I/udoIyUPY9LkdY3DQolKGa56b5FSrZZcLH2mdPWuFKfIFY2HfmEZz5uhnnP3jj1YCplsIq9Bqyy874RnL+UgKz84KVIRdCpzJ/PAFWU7wxTAlBV7UuIcpS3JQCWuQDsI2Kl5p4MJ2FUPIWeFuV7T424imONSx/S+40TiErkiMCYOFadoMaAc7+3+Rp3PQ9DEq3ddRPDMYW6UN6UzbE3QjtqTl+t6VufbjR7XS+35CEUREwAKe33SdM4yBFeapYEnGJo3DzW6eZ07SqKNI88ZvxjiYvfHSJQ+JJ5POYO85vcPZ5q44arR7Sqf//4Cw4q7CFUgUSkgTI1WqGgM3ry7UTIzedBxclYEqSxXQ7BJtwvkrU83rhvYt7mpMz8fWRFV9F5QH1NmvZFOJC2dytRE4MepX4X9xzo/H4eZxk8rVr0zOmFXKz2d44SuD4bG+GXq8WlqCn4M68kmJgngyk/x3LR5fbSRIJ+b6OZ+7wIlqEPqiBJ5FJWe1iSKJiX8tIgI9rywNs0kep8encd9HY8fp14kOQOdT2lRxhxzL6CgJvP81+yrWLiXVuCgOdwpMq3l9U+TRk0c42izYGNoEnIxKq9623U06TjzW/A2LipooWPKisK2WYSRO+fpYbjcWJbUGc6CJ1FIkLIi/u9JZNJ/rBKl7/ogFjPG/jI4232NHGK9ei7dzUrhXyvu/N3AD9Ip8mKToWhjut/NxFlPmXHmJWp6OLXKeFsXKInbAWAsWzApns6jaJhc/j+YQNd8CRbXSj6QXVZCAuYKbp6mHh8eSyFRdx9Q6b5IgSfj7uhh+sZ80TC7NikAUyeFFhzq65MledO+DwS5zIslGE7i06V/J3Po22bzdphYPKhe/5vxRhtBjJpgCmMTTD2HHBTqXo/5YmjO9PtIImHQWOn+2tYzcVYItedqmBWM+HxbcJfZ0r3UxsNP1eHQGaFqV2uTaBGX5mdD+ZtmnYs4T6IXZh2Xx37CXSyL1p8ve+KhO50JJ56i5oIlzc5HWegZJmzw5NM00MawcwH3Y9mZxv1Cm0sImmAkrTI1QE5Z4aFizDkDn2rRkIT1SjPkCtHIrlRCqrf1O5WOUKc1Yc0ha9w5LrO/r5rofR+IUVVkMX5ZmGcUP+ejZTlkrbgIzqmMj42tJBg3KQN/rvfh8yJ7wXJu66/ZBpVsttlmm31jti3cm2222WbfmH0VUMnvrxx6NRPNb3DnMtGOXryiAhxgXXJO8EEwUDWEedfPFP5OwQjc2a6+24p0ZEZeq0y4Vs4KdzXMC0GTel7OikK7Y+cJR41ZE5vksYbXo2htBDB96clbul4jcMA5caPkvZW0Qo2G4l6iofc1QAyCw3GiUHe/cAPkm8iayw4tk4BcHncTlZHHpGErqHo3cqGLWSquDkWY64fdRFDNzVvWShfaHc5E7I6VpdFn+m4QBScpKuSpsoWGhDRXipVXBB9gYjz9fOmp0KV3gTD0nBQeqyrhzdsVDCJ1lj9UFkpnIlzTIa/4a4oK5wsrDLZilrt7xpqXxWD3phac9AH+zMU1zTVKvkAq7f3ltdFJFasfau7Qok2CG+pctQvdy7Boghhy5hZv1ir6TEqc69gdAlSFM15+qsVLQyDoBRemCUqMVf7tdCJ9DWmLYKT8OHX0/L7rPdzSyswjsXL2w0K/ez73dN13DzeCc6abI02V5B1BKDFrfFfvZWcinitscgmWuhY1e/GWcmXnoEXui6UcXr2F0+UYO5No3u47zwVwiQvzkhibmDQ9N4d+WUkuAK2gxq7Gz2cevwbtvHiL15rXUSidb4Aie3GN3wDG/WFo9B+Fz/Uh9YolXH1W1CfRaMbEZ1VEoYDW5Z1pXa1dWZFdbIOTid4jTSvgnPhBkhO43VD5XhC6FyFqfKmTaLRcadUwLYC7in+ae5rcD46TLZfAWixWa0pwvHpFrdxkojFmhU/UaxAk+qNVxr++3AEo3O22cF6DxcdKHywJqJZAXVMW31e8++YdYdmAo4l5HGbcH8siZy4sd9sE+l+XDg+10cC+Y7rXdXGEo94fb4TRvnzmMQpC6jRnEC1ut+dEY4qK9D5SYo42AEqa3jw/MD7pFd97oQfG4VJx2qON+Dw1aqCi8bgJfY/2Oz5pfKeLJknwhmVGTUaua0cS/Fs/Gei6CC1nSxuKEpuzNhkQPO1Uj2NHQNd7H2bDUsMuQ1eO9KefDpz02s2c4PMGfV8+c3nuaPFrTsg0O3y8FD7zyXNyrfcdVRkalXGtm9uXhYWUdibSwgOAhKXe94FyNokvB1/mHh/r+L6ZZxzqIn7xDm/rhnm7OFxqm6/T3NH9uwbmbisA9zVBG5KmvIrPirrFH6uk8Jy0EFzjBKDTmSqUH23CYz2Xk7f4Y22d9jbOOLb+pcEKGuvaWWybnRJ4e2cinG5aL6xFNAiKbsqKJJkHk/B9TV7/4WapZ2chKOBv2gaVbLbZZpt9Y/ZVeNw/Tbx/tAKWlAFdXTCvM3mkGRm2uWZ9EW1vn28NbI82khdx8o52vtFGYpsYnclLS1C0a+6cJy/7ee7xuXqqnU6Yno8ACqwgJRtbZ/bnlUodtzE72rZrOwqBYlZ4rp7v08KttACuxPQp47VCEk4ramgqGTROZ+yEIl/LoCco8jAv3uJVZNZbZPJlsXiq5/Ch9xSNdCaR1/E6dzSWnYnkCVNTgKTxUmmEEnpagiFGxxTEeD0f6DNSR0Opwjhpn78LlZIluoRrnSlkDUFT266YmU0TtWrRamEo1HuyCEnRnSj8GW3E24GbFP+laro0OuShW3BuNEnLUcTl3JME6/EwMQ1ztlgq02PYecynpqECdDWy9JPGdK2FYIabCKeo6TUSV1calwhKePo44nAs53t/f6vFSsWLbuyfvg84XysDwwh5X1EN2lhGQ+JnZRSNA+ZoqC3Z0UWKUG7RiI7loKa8g07c+FZUo2pROLZzfPy98/hUx3rvHUGMnWGJ5UGc+5I0LoElg5vdu4Bz/W5rlVeiba5QbqqFbzq/KjZiBhtXS0tdFamq2ZlI89CatKrGvtT58ZfLiGPHcsDtWF/mThT1RDgRwX1qzJpUIm8AmJLBl/kbgEomIUrbi56Tvo7sqweW+pkPA5BqOOdUxgG+NCMAACAASURBVIXoPdwhBuBFpDeJxJNukXFt2aU7Z0XYsFGZ+NoXb7lfpeiRd4vMB+10xrueQ65G3xsMV5i1W+A0lyV/WViL97HjDaucW7vJGrt6hxQUvnhBw6qfOQcOC49iY/o89TSROx3xvtKqrtFQ+GkVP5ApM0zwujiiOI6WKYzPc48zhYh1nBNXo/3btSeN8cGklWRsg2G0kIz9dN3hIZX3x85TDuFBZaL33W4ddYIfB+6Y83Tb4Wlumyo/4EZlqlLbOaYqniaLz/Xc9jbS+PnEUrwJvKm1h24SVYDackVuTAqHfTl310e4Su+zvUd+qWMT9Kqzu574Hg+Vp+5ng/Ar9QQAl7ovN+ZlSwGu4DX2+9r1RnZX0hnPlTp6XjocOi7bBgq81xZK2V9UQhMxc3d0K84rZZ6fp6Dxl6ktQrw4ZQBva7XinePK25+XjbeqSClT7HRa6Z/vK9R0ug2Y6wn1OhHkqJCpWrfR+67RUMn7YYhUWzGv6IJceq5Vpk1iipqoogqZPlNkmttaI6soe6p8zVA0J61iDrgRuaRLsET3vUZN9RsPXaZN0KmMlH+ZU5C2QSWbbbbZZt+YbQv3Zpttttk3Zl8FVPJURXreDgqHSvULmUvY71xGjXZw57Ko0MqEI5dOMK0MXGGs7Yd2JqxClqvoUNOUvzrDIjdGJxxr1ddxmLnB6tJReBSSIibHFA1Rj5zKlEGek6ZyYFu3RwWGT65RU5XhaEFZ5JyBuZUFq0xwyiVkEtq6d0G0RXMEEV1jxl3T+02MBcqmq70oWQ6ZKZhFJIh/t42ZDJ8HE3GrDBKpfd7G4p/2Ex1DgatBZVb9tHSs320ifGRqW8NjQzQsa7Bb0LVqzajpPgHA236qlyqYHAILlV1vpqiprdQPu5mYOz/NPY+NSfQ+UybX6oY0dtHgUgWeXk47PP2e22293d/o8w3mMToRdW83Mr1PKeD5dUefbzh1DJoYU7eJmRZGZQrNg6CCPt2GFbuhQT0vC383EFvDUPVjr9d63K3i1yfREk+Kv2WFpeLtz4vCteKyJwMau2tkJcYpMpT30C10b5KYezExk+vjNND7e+upE8ygEy6x5YQs4dZG6ZU+eDv3P3uuVGx5pTkpwpHfdNw5x2fuCPTb3cKl/lDQou2Vqa+XxDmbFBQpGw4m0HUMJhLbzAmNfYBlJopIHkO90v5LzYK/ioW7YdxLVDgpphQ1ClzBo9riywpcnQZ+U5hGSOAu0zErKHDrpEYN7HRaNSBovN6dDbir7y/RUBmxEg/JNVhK9i2JGwtJmmACU8c0QOX4nuh9jLEPOmOp1yoTEVYruonnwNj+YBTpnPik8WPFa0eTVj33qMFBNNTK6buBtVuWpFd9/BoOPYlFYDSRkktXIfB+dB7XSkM8iQf/+0rrcoIf7RPTniRWeJktldk7nWiBcZ4XR2tZGW+6OWorN+w8Ft9wWqbjGc0JTL90tIDM2YoHKREN8uwtUchSBo3lgwuU6G1SrjfR1fzouLVXZxK0KOV+v7/Wc8nEVX6+7GjOjJ3Hc1UkVKc9KRimrEi4P2emIYbAVEY5x77cBsJa3+xuGJv0K/i7z3NP5wlwyy3ZBswq/vc2XlZnypHEzH05P88OX0hLB/j7Kof84EquBvi56iMdGq+Bk5lX0evzaMNKSa/h1NdgKUnuU0f8cQU+t1nIRSzgpHNL3o+Gm3h8Wgza1LpFPseYFT8HwVDpfsjcsGQQEhopK5rbU2T++tPiaO4VjaSmGsjd38tcq+dmIy3WN6FieAqacgdaYyt532yzzTb792Zfhcf9YdcyysCPtyrkAva491aRyNSSFIa6O568wn8WFXD8eeCARgc0+LF2Zz7YhO8rDPLdeCU61yU4vFQq2s4GYpgAJQQtv2tWVCIpJt+8z3sXgeoVvHrOHjfvYEqKQj+fWYQ+ZFY5fOjWxTaPffUsEvCXuWkDJ3xXPa0vC3dqj0KkfRYNIT7OPWXZT8FSQdBgEr60tmQuEux0jYY8zxA0Qh2Oa7DkPbUw06iIH2uBxckzY0XoKa26qjuVqanDlDSdl1IMK1yCw/v5StfUCVpYq5B8ntfwQfPeYlbEIshgCiXATWP/NHWrIpFW0BWspnvS/t3pTGybOWpqs5cziC7kk0ZXmSxnUXg1dgtsPd/BBjr3sRMa7zbhd9890Ri8vpSxvCwdeeJWJars/e5wIR15AMS+GXqGWfo54KdaYHOLhhrYNojq2Tt8qnNJMrE6ncjrMyrTPT46ZiudAj8HVoOexUEnuvevXuEo2gu2JiijuI8f547mm1aZIiCrM1VCymYEKXM7L6uZaptQdO4BhuTuXSB2l3xuH1zGqUIlVjEzywud7Slq3GJZC/aB2SNGZWLfTFFTgd1JtKkzipt9KwEvXaKhZ25Oij6TMuh8bpELhXYmw8sH6Ffsq1i4B9NCjYL3AuvQ63nJ2Nn2QGVRYcRKgd8PwKlixj9NCq521xhNpr5ud6Lf28vcU4jYm0i0o5hYKdCJvolGJcz1AXj2luADrThsOXmDfT3/c9CET8v/t8YPSnFlqAJwrjfwLKp3Y8qEj6+Po6AVwyNt8V0SU7gehXreyZvVQtUehiVp/FBlVTu9hlPaQ/4fDmeCf5ZkaPLK/n9tA4xZ0f24RUW0td+NzJ999o4gnwfnaZOagkFjQFmdiDttVKLF8D5p6gE6Lg7PS0+fb4tMyowhzknTwn0KrKj4rgsEs1yCIVW5a+Cwuz1cD93CFblKU+XfvlvgG70uMfa+RO4w1HcBB1fO93rlBR0AzTEryrW1SdQQYrkZoiK+GQLGTlSQtmrJPfcAPT8NBOstgsL4w3jFueK9baOLWdHC9/0wi3yJpbGYoqaFcFjxonmT/LerYepqx5v1YLirlZQunZPGfX3O3vYzlcJ/vO4EjMMNGU7B0mKvVYau5ynnTal2Lb/cckpacbVyyqBNZDSZFv8PfcCHqgIIkR+4RoN9pQ8fhPTrJTjii4fM+H8ZH54zV3oWuAduhsJLXeAHnWl+Lknjz7c2Zhlvez7eskElm2222Wb/vuyr8Ljb3lLYDOW1T5xZ3VlFIWrMwNHwbtq8UKMSdoZZKK3/5GgTe2BRYwZ7061YpPTIKzvuaCLBDUbllR5w25VLSyNOmgzCC/WClfAaWihfzmVvMoamkZSUSHAoXELLpCvEeuFzAk5VXOHTFPDDrlVoGvy+iuLvpc4FePffmbgStm+hsdPMbNHgNnAFuuF9/MPA0UljAsiWVC0sPgdO1PQ6k6f6oc/kgX1ZHIXabwVMUFgzreAq4rtD0QF5nXrR1IHD1Zep55Z0Jq06tYfEVZTc6y+SKH7vO/xZaKq3SOYcOPkYMnB0HBoDRdekaVoMNlKi79NlxKHOjV7AIBqZhKq8Nwh1zF4q5AYUr7yN9LxYfD4VWMOahIfKSPn+4YTHes+ebjtKPL65u9Jxzqf1MV9vhZ3ik6YIZLqOlEhvnvq9CwimRVGaKn+14vZgTkeK5C6BE8GySGgwmVgab8ERsmxvFjMzsI4iupAaMu/HG071fn+c+hW81YTaStEZP2dN4/4UNHmnDbZ58YbEm0pEWm+wCHpC5qTtlDTBnQoZe5Fc5bWAG248zx0x2EoDBIaXWjRiBMPs02zpfi8iaX8TQlLXwMhD1MCfrn8bK9k87s0222yzb8y+Co/7oWuUtwxTgdGnhXdwrYDY1AEVc4g7zVzhneFO8L3YzUMCeda9yaL0umCmwJpneXSevLQ/33ZUGiyTE9eYybMebcZDpTLNiSlAvWG++U9zTbpFRR6xERj+FDOW6uB2OtPOu7cZl+pU3ncW31cv+E0XcKwewpwUNwUOmrySnUnk5f7p1lNk8k7g/ACXoi/J4IkSlUFQ0CweG186D/hx4sRiO5f2m7e49gOIDqUSfUbaq3eU+ANKc9/yPaaHWZNWdLnZszQr3bPOk7bI89xTomuKhri0Er/dmQyrmtYLR0mjSYSpSsXFhllrZKIvPg4zJQOB0lWmnC/LAo/jQmXr47jgVr1EZyNiPZd5sSRjel0cTlUl77ibMY6s1zLW0vbgNan8zd5SUvRunEg+98fLCIjIskWNzU6Lw75rUShj/4O4fqMy3tTnQKuitwEA/3rtaA4fTKY8kaSWDoY9z/e9p9yJzFvlzBHAzrGUqpQYyAB8apEMR+ZaZcK1p6Q44Vin31FnklLWwMr7T9VT/rxwfYLTmVxYhUzzxGeOtnY2kMTyu97j78byx5e5o8jyxTNN0Wngxzpmpb6jzUmFjzWS+jjr1Xj8sSpu7qWm9V+xr2Lhbt3cbxFE6AcyPtRIMGfgp5pHODqF0TIHt0Eot6ho4faZ2QS9yfR6MJF0LIo2By80TojJNP7mm2iIAXF0njaMow2rY/aND7pwLHYJhkSk2hU5lUlw6ugi3cBL0DQpRst645dQrrdcE+h4B5tIZCp5g2NNvj50DBN0opfiQxcwGObwtgl7DRoXwxtTC+180uiqRGanIxVuLEnju2G9CHycO9ELUOFjXfxHk7Cvx3vsIjEFUlY41EVZijvtRFFRaQ3VIJGIfdPaEMU3B51IblapTEI/shO3QsY5cMFQe5h94kXm3kVcm741FBWmyI1cWivGsSbSAnrzvPje7yeMx7rILnqlkd0+P44Let3EtTJUHbO7hwldXxNXs6HkbpNoBUoRkm3t6Y6BOO7ea+ja7u0HdSE2lFaZahRaj8fddSBZUifmyat3VIDzvp/pWbkGS4vv3+08wQRLUtjX+XyJmub5uz4QLBKFTLLVrAkiu7fv3P/L3ps12ZUkR2Oe21nuXhsKQKN7epozQ/KjaaFJMpN+vsz0JvHhkyhSnL0bDTSA2u561lz0kJGRp4bT5Cvw2c2nQuHWPVuezAgPD/eRYad21Gyt17ocCHnkpjAfBFupbcyI2zIJq8WxGxUHET7k5pZyUvQ7joIhzCp43gh8yBx+Fzxv2p1VLOA20xlOtQFZYtbnou+LasRSx8/c9VlcTgvwezEaASMT+y0HEKXK8OrPjTNUch7ncR7n8YWNzyLivqJIYT+CC3NzLZCBEGBVZPggyUcOPvM5gSzremUsK/YZ6Z9JM6aiiJKed/lmMBzdXC5ali7d9iW+WR75e9KRCuVYgvQwGi7eAXm3BvLOuigz/zNFxEoEXNFxGldw5rA2WW7y61mm1x2tYOreTdnnVm4vsaaosFCeZUzVRPFsVQysDCdFwIHoYYVUrH44LRROu8SM9ExFEyK7u6f/r5SfdHRq5qm/qQfcsGOJx2Ewz/4OiDS6xYTmNqVhLijKjHBD/P7joeKCtTGOoYpxzKpzq6rnyHw/kdMEMnQzVWhbaodllaPDBC8luGVVDEyz6ybds3enGUsmXCxabmdvJgXQ0SqOEqXMkM9wv2Ln87Kwz80eCFbbN7nweLVu+B6MVrFzTNcL/v5PpzlezGNxd71umUI5esnG1mm0o+EoUU7esUL6Z7DCFGZKBf5I54v3932bl4+5fq5YKaD485lmmp/9pupZ7uCxqZka6JE7GgHggTKpwWcYUIkwodrJDG0kmEIEjEmpUORVpPfiWaSazveiyAYo/aT9fW0s1tT30U8MOmzIcq8rAwAJXsvt/Qsz8r3cjxqziYRCymwLJxltkAJYEz14ZcK/cyP6y3GOuM/jPM7jPL6w8VlE3KlLaK6BmjblH5uAA0XBMcqa4lupWytHC90E447ecvH3WnqOjMREa1sCjEvGY+Ro82EXbYxOo2ZDgc4qxnpRZJrQtDPLyMDY3tJY1v9NkZ6RuajpgsAjnct14bKoOzKW/a7N+uTRMCEV1DzrKVwVA2ORvVNYmXi+82LkYtUwyQJaq5lGJ5B39hHgxodaW9zWMXqbRuKVtrgj7HRPx19qy8cPyNHE0ljOYvZDtsTaFD3/Pt1XIEbKyVTBWom6zjrTushZVYqmhQAX75quyBQ8pxgDrifPt7Gan1mlHNcROqeACQabxhNFqYV0qCkSeuhKpljWynKji+gNqnSvrUafuhKXPWPQbZOj3lJbbrQxxqGqSU+6NVy0rLTlYm0/8eOUImC2ilHgOCg87iOV0HqJf324BACsDyNHhNMCbdWTRrUZGdc+DIbvy9KM2FNW8Djk7tLHQeMtnX+t8nxoXK7HNE6wtrQEWCc/QOBy8rwTpdS3FdcLrJc4Nem4hjVlptTDEAQbO8yUw4eU8VqJ3Zg7rwFgrT0/p4XOGeyU7lrrPFfn2maNnaBYjOynroASc743i0m9gzMWkdeC0Qs+7kNXsijWGLJ+fWOzGN4YMi5/sgGegG0bcgf5z41zxH0e53Ee5/GFjc8i4k776rSS+u0c2I4JkwwsgTr2gXfWSgn+uZCYKN1JSLq0y7LnCOzYFxxVzooBhiKyftTczuq8xIKq8C9mLT6Q5sO6GNg4dDp8ACuYVdIzZmykZ/eaKUMiGdP2XjIzZPCSq+AJpwei208i9x/G3HQQgsg6KCGrJdbaciTVjpqbGiSyoenJGm5nnzaUDF7xZyQC20pFGmSmiCWqZKIDFhMq5aYYmC3w/anmyv9X8+aZ80nCqQvl2fVGK8dNLFU5Ykht/GPOFgBgQZ/3XvL3KOmJAgc+TwCcCaTjpnvmIZ81eSSapxKeI69rmgOVdlwL+UoEPNHz21Q949Q+iEkmEHJ01RSMQR/a8pmzi6Fnf2oK1h6xVnLWMY4KN5sjXatgiVdj3MQ42LEb/VNX4eWMVBontlxKhn/HjJkVI2eYtRlZp6dzOjdKlT1rxFwPBe5S9O+yE5MSMboFIjtnqt+TvmdjRmYPOS+YMRJpipQ5eMn4/0vp8DW5CR1Gw5F1/RdqgvMJjTQ914RZCxG4FnG0knV93sw6fH+K83qpHb4iRccQgHHIzKn0frypez5mvN5UHwvohyQxa56pEk5NhXum13puPNuOBk9DUtbMz8RP5CJsEISd//z4LBbufkItTnztaGcUf+5sXtS1AA40D1cmuqUDz7uWRi94Ej1O9JaN9CynqbVH22WvxLT47IeCF5PbqwMXpvZ9iX5MAkclT7pKOXaxrpTDBRUzCm1xT4tJOv68GHkx2Y6GYZP7IWs+7EbBnOvBC74HvQucYi0mNMb9mLuyam2haRESXrAEa6Q+Wj7HKcUpjajFkiCDXKQavXxWlEnpYipWnaxmql9jNXe6FdJzh9h0NFbjmhaYQjk0RJFb1p7v9bGpckdlMTIk0naGKXXTgl5hHF7Qd7ogcEebzmE0fN2l8lhREWmuR5b0VcLzMwtBcMdm2uBPo+aNXImAWyoAbjYN25I5m2l/unAsCxwCUC9J22Ton1nlncgr8u404wLtetZhviRefaezr6fLnHXvAx+3qB1mId6by77leVuUjr0od23FomlTTej07KcF4r5VCDSbtkPBHHslAmuM6MLjBbX/2olt3dJYVKlo6hRL93YTu7JFOfAmODjFHPTTpF9ibiwXKqXIUN66GPiZ3PcFF7Ln2vEiPuXjJ1hk6pE5+qwfcrSSnzUAlnueG8uL/otZm4XMpOcegsNgeIEuJ3pGc235HH3Iejedl/k+aZuF1ZDlkRdK8QZkZMAneS5Onsd5nMd5/Dc1PouIOykCjj53PwWACx+bIkflRoKdYFbGczquJrtzqTIF8DQpSikZuOCjrctmviIw7ccFge2JZErbktPhTdk/i1BTFCMBvJQx2jsMBk+U2glRckrL7h5NzVnBXDnsyU2mmpj/XhW5sNK47Nix1OBU6nFQ3MizKTI88qmt8LVMMqmBiz8SgVPUcWKkMHiJq5QhKId+0nwyL0f+fYou9n35zAUbSA1I1K2pLUMpxzEXvQYnGT64qjtOwaXIEM6hLdkU4GLd8L2zVmJ2Eb9zWQ8gb2H0B8VR6zDmgiSc4gYqLTyWFOHVPisCpsgJeO4YPnrJcFiGt3x2dh8NZyv6UOHiJqba3uUCoi4cqgVRUVcBlqRFNLJru3eCzRZerY+4/poKwSXgKZt09xmWM8XEDNl4lBdUiN0LaIrqbi6PHIlLDcw9XXc9oiOILTXduCC4mBsdfuJ3vJw3XMjurMIlZTGDU6ho/iwm1NJ21JxtTiHDqCGUokfP972zmiPuxbzn8zpNqHat1ZinrK4Y8/MIEi1lnOXEsKCfNOkk9c9aWZSK3tUhR6dKeFwQFPVDU6Kk+/Gy6jhTdUEwdDU4xW5YWjv+HiHCMzJDOt/BKYY2r2ct68vshwKrRLuVgd+z3VDgZFMnsmD68No4DP4/xkrOEfd5nMd5nMcXNj6LiNszpptb2AGwtnXnYnMOEAuSSVWrkHmXP4wKG9pN18oyBdAMWYR+UQ5M4ZrXPRYbKrQdNUeEQgZugghBcLRgXY6ANquGd9yistwQ0TbZxsjarNGcjm+M5d38cKy4Ldkj+9Zth5KxusfBsOqdEgF/u4wY5tKMHEH7ILj4dFX1nAlo6bEm7epmMM+ocQlLvlw10KSG9/bDhv//5frIdLXjqeRW6bKwaKguUEzob3tq9vjqaseNKJtBc7ExFoumVlpU2CxHDB1R56qBo9C+10yL09qh3VGruBwhaE7Ulw5Sxftx3JV8X50TnBlJAS72yUmRswdY22RV9dw6fxgM1y4cYYzO5/rHuuowJ7zde8ERtJCBm7YAQJocxZtF+mWAawnjth41tbz7UcA2RC0zHvJnAi3OQAYJdZw0FVEU2rUGizU1i5w0N+wImfXSs/9lYDOHY1/wnC2Nfdaole7LacwFbSDTS6f2X+OkQSb+Pfj3SQ9cCZ+L4VbxPJxp+4waeE8qhxdlh1XKhL3ElubZ0SquqyyMf2bTBsRCe4rCu4n13uNQcG3mthpZY2jwiq/Jh9zUpGQ+3zAqpg93VuF9E+tXSzMyTXj0kgv5zWC4cW0aZRd6xKbONm1pfbkqHJ/b0T7Xz/9r4xxxn8d5nMd5fGHjs4i4f4gQH/SESaJkFE1KI7FKjAxI4lmdF6wZXMqMOxXKZ7urArzzzWYDBOFqWnvoigj4xQBP0VOxCRj3cWdt9ybjhmpAeCJBmGOFxZwMbysPVSe2QIx8gEjbmlpMAcSUmTQXJGbEYNUzutw0WkjiU5XKAlLRkT1X9hNObGS+7surEzeuzBvNWYT3kqNpbTwURdzrOgs+ATFjAGLEaicRTYqSUoS5WPaYtVl4Kpn8Oi+zRrV4LhCUsoL5ssflmwgCD3vJ0bdzkr/fe4GRWBqHO8kNLQD42cyXPbM6XKtxs44Tqu1y9b8dDbMFCpXpco9NdlgvlcM21UAStQyBsUolA5DlsDlDKCrL9MVn45R/rC4c49fd3qCoCec3HsWCjqUBN/mbFE2XM4tAU0mXgSP95lDwcZ0XOH2I577vC45CZzq3c1fMLsmZCACWbxgnUW1jNRbG8ucfxkwZvKfPTJ3ab6uOj/k4FMzYWhn3TIP9NDFYTpIFL4xj2qfzkrH4Qjt8OsUGGOuzXMR1OXKWuZs0d6Uo/zCxDVwax9rkd71ml5wQMiNGTOiySngWlps5jUtioQHPnYVu6fe9U7hdR9pmP2h8OMYU62kwWA6W73vKAnunOCofJ5Tg1NCTruM/Mwv+LBbur+OzgUBUxAOiUmBS9XIhc71jepZU+jxeURrfe8n0pdFL1mroveICo1Se+bPjqHC8p8JAb7C+iA9i3AN6RqlS4yHoHMzM40LHt/bwVPICPDQKdif5O9NirbSHJ0jn09OSr/VqFd9MrT0e6IWZ0vUEsgnEx16ipxe2UJmXrJWHIuhmhOYJvZm3vOAe9hVfq/di4pou0bZZVS8tij6IifaHxnxBRcvaoTvGafL7j1dYmOfqgLuTYs2JVdXz9feThcx5yRtKZzUvysXcQdOzD9YDoOszHsUi/qxKgOps6PaaF7NxVJiTnkkIgqEB7yXDLNYphmWU9OhpQTiM+WWfakqUxQg1Jt58/GylXIa6lOPitpIe5ZgXv7bPGIchhb8w2XSKwUGV+b71Jwo45haeopXQZkhEiNxt150MfOLt7xTfWyU91us4b6UK+PBpFa+pHHBBC5EQgd+FcqLYlyCv3ubC4Ie25kV+ph0qsvA6jgY/UcDjMUnTRf75sS95IZxyofejgkDWXUlzuB01Byi1ycFBZzMk8fE0Z0W+o5Ws4FdJz/pAWgbMkpIla/NI3qSVCNw5elmMfK1SZG/Lw2iwG5L0s+eCfaUtbyK1yV60se+D5od2OLX5wab3I9Jrs95PgiRtkLhexnVkPfb4w3ZN15HJErVyPD9/bpyhkvM4j/M4jy9sfBYRd6r72QDsJgEd9UhAINv6/HJu2ZaocYKLk0oEdvcuZI4WtPAcubz7tGYa1NX69EzM/riLu6a1ilNgayUL4c9dPzlfmRXa2lyECxNTg6dDzemXnhQs3j3GqOgwmtyYoy0+UvQdIGDpfgw+6y+4kLWmpQi4Tym9CFwoedZworPl12bVMFRyd5gzpW/RDc+aEC5Ggn+kR+1IU7qX2B1qOs/xGQ0KiFFOanh5e5qz8Hw0ff33lCklQ76/nUS7m8AwFFE1pwLNiXTQNx3DBPFZpCKkhEjUUQe09Pl+yAUfPaEyDk5xFL2caHU0VnOKvVY909XSiNQ5opnVI8+ldA1AjPhXiwg1SRl4zjgvmPImnwLTBL0THJWrwiNQ4C40UKyJRrew6B4olZc5QvdesGFy7B6lwuLS4sbGlP393QpWJP0MiQVBFYtZz/cuRe0PXcXRnX9WbBTcpOSCYLrqYVQcWd+UI0etARmy28jc1PSxmT2zp5t2Hyou/MtcpFvmwv++L/ncjMyGD1FhLzd9TRvg0jWnnwNysbFUHts0ByZdtZF6nBq1HL8fnVXYJDrgpPCqZO4gBsCZatebSWNX4Pv01JesIeR8Pp/losOa3p37vszGzzYXVH9unCPu8ziP8ziPL2x8FhH3Hw6p6CZARi146Dw87SsrIxjvntrWz1TetTuXtXBtV62NCAAAIABJREFUqHCiXbNSHrOR7IeM5ehJG58toKxGs08YXo4cfEB25mgKxjGfUaUOcz6HwSusyxxp7Ca2ZwDYCgqIUV/akT90Jd536bNZI1kJYEPUMg/geyrUFNLjD8cYcX9Vj9yE5CduI4e2YMuxH+/XuaV/LNh+bK4LpkTdzFqOKmbVkLHyY8V0p0U5MIXxrsnaIFPXmWRxtSkGzAmrXJU9FzVPQ4G7bSzg7D8WXDCrzchYclWNHIFtH2ZYEA0yBMEYdwjAaZ+zhSPpVw9Ocet3VY7Y7vN5ckHJKy6kXdUdY/TOC1wmyiVFtUIE1POs9ZEiVWMc3yNtHBeC3Shx2Fb82USlHAYFv4/PQGmPYk4Rdw2091Qv6TT0Y/z9dL6tl7lAZq2C1llZMI3myfC9qY3l6E0rx9hsws/lpCnmxaxh2qoLWd1upses66MtXlMNN0CwNMJpYns2dbdRImt2LMyIPeHEx1Hja7q/82pguYMQBBeAAfA7Ojcj0zM7q5hW6EPO+JQIrA80m2D4H4lSKEXA0iQsW7PbkQvPbfOS84+Wga8vQKCy+ZrSfVyUPR+/GQo8HLO0Bb+7MmBF7+VDV+Kji3NioR3ebWPNa131rPL5ri3wFSliLrTnVvifG5/Fwl3rlEqBvRc3hcQrmiyVCngasg5JMlIoJ87PnZM86ZQILHLjgmDBJB9EXijqkReT2uSFojAOkn7/dJxxGiREln5dLltsqFC53c6462qwigV7KuXYjy91Fo4+n6MUE/F2L9it/aYc+TNvG8Oyj1P38lJ6fJc8CAMm3xmYv/71fMf3d3eqWXBqrkd8Ryl7/xfc6noiRjTSi1QWFiUVDat6xJIggcRmsFayg/ncjHx/2zF3yU075l7f7tAc47lcLBqoiUv9qaEiWj1C0YaltOcFyZQO1Yo2qVFy0fRhN+ciZG1GLspq4zGjhdMHkZkyTuEjbYIPbcULVCEdQJK+txdkoFE4tKe4wFiXTQyszTDF49Oc4YuidKiJ6y2lZysyKUMudFcTmMUBZpa7XSW9kfPbBi6v1wguL/qJQSOVn2xkOaC52DTcUanLfH939xV/dsoqeaDnFyDYsm5RjBlmsornyuAFz0klArbsASqwNLmbON1TNeF/z7TlBVqIwJtjCALHIWnQeN6Q5Ji7oefFmCE+k80nlFPZE5RugfcCa1o0L6ouF/20yd6WEx2dSmW9ESU81skKsO7YS1SIwIymEPL7kf6djpWKmY3VfI5XVc/BzeCy9s9TW3EndSUDTjbJ1npmxfzcOEMl53Ee53EeX9j4LCLub4gS1jpAiyQzGhgmAMAyh7OJe/plMeIF8Y/nWjP/0kjPafrcjNwV9dSXWBY5jElp72rZcWHMe8Ec26nKWlQrJKnIU8k0rNWyw9MupkqdVXigoqGRgXmw6e+icWrKHEpWLYuFn0wXUhOoJA0js+v2qhgYvpiq7e37kiOauVMceS7rnuGDfVNxp2NMIykyWUyKQk2V5Va15wLb0Ge6Y+4U1HhB8qMhgOGnErlD8mZzxPwynntwAgtFxTX3PKqYanKksfrOxbAUgO8CRoIb3ChYee/1bMcFzLu7JeuGzJYDG+3uDxVnVUqEZ0bFiWM77fK7HBPvXXB2MZsNaCgrsFbxPajLkSlhZXVCuSQqo/bo26QRI1DQuYyd5KjbtcjFV51/HnYSqkjdj4CjjFMXHpqonUIDp7t4Pm1r2HxCG8eRu+0Fn0N6vttTzdnmYBV3LS7NyNmjmajeAVl75DhJ4Q+jwp+b+O9aBdxM5rMS8ZibYuA5tihGNn7edyW2JLFaKMdFX+clCpoHL+pDPr7NCok/Hef8eykCww0pqo3djETFbWbcJTs3I2e/24k+ikD+Di086xZdXp1yAdyDf24OBcNVU9ikHQ16n+eSSnpJYy6AT2l+26FgXrnNv8bKjM8ygr82PouFO53zyQL3HcEXSuBEL/bggC1l8UpIXNODPViNp13ES9fG8UKoQsapQxCMO12U2eduf6w4vR0GhUAve1WPsD1JMzqRfRBNdht3XsLSDW9bwwvhyRrW074wA7exe0qHDoNhhxEPsFejEQFmIvuYXpIo/Rqv+6oMjOHVxj7DBFNjQCkdNrTgGu2Y+7trKuz61EChUKvEe/XYDsQl3wtuArqoO+apj6PCOBEVSveDRZi0Y3GocVR4IGwxBLBcaNMUEznWXFvQKvPklfTMNTdVgKIUf9wi24RMxmFfMfQxOMU1h09txZgmAGZ47LqSmxw6qzhFvao6hncObRbGSjhkpRzzimUfGA9WIvDLOxUj807CtpJ/Tpz5zmrUQ9Z/ZwjMS4Yk1vMWJvF9reT7USzss0W8eYrfefe4mPQoBGb/lL3loKS3mpteEmSwKgee+yEIxtIPfQHB0qKZF73vC64ZSQFO6W0Afr1IwYkAUeCxMZZZGu/aOuujDwUuynisF/NTFu8asoNQO2oc79d0Dv4Z9zyNF/MGe5rPx9GwhKtkjnZexEcvsk+qk1zfmbI2tqPhdvNSZebL7mmW6wIqN5G1vWG8O91DIM7h/dTfk6bh45C//0XdZUjyNGN8vrGaHXP2E83unxtnqOQ8zuM8zuMLG+eF+zzO4zzO4wsbnwVU8n8/JgZIwIwEuf8yO06QwWEEQsi4Wskpj2NnibnOLhYSgWlmxYR0f/PmyO3Wh/eG2QLBC+ikI6E8SkpvD22ZaWNBMm6uRGDGRKUdDOHslbKZ5YLUup/ts2bq+RUWE0pdwt5eVh7vKe3WAtxmq0Tglt/GasbYAgTDMkWfm08ulw2nvU9dhZeriEnvm4or9VNDX9WXnF7OtM26zEPBEElqZDr2BX7YR1hBT5oqfMgKfKeh4Hs0OsUsn+tZCySd9dqhXlPaXwD9NsMNukx6MQHlq/j529UJH34bYTITPI50XvWEzbNrK8bff/HyCdttrEX8frvGRdLpNrkN+mA1O71MtdpTKnxoS1xSu7L3uW38NBhO6ZX2rLmitMfVbZQ4CEEwfq1MAGjeShW1xQHADoqbdMZOcbv8rLLQV3Q/Wo/+p2yzl4Z3Ajc3ERMWEnj/U4QbBie5KWRTUDOJcgyFdVZn9cwwMgT30FY8f0YvOaUvlMcdMaeuisAw503V8fwcvGStl2/nJ/5+Jf2zBq40b2tjGXMXQoFeOdbCBiKEkuCRu67kiHOuLdZUC0vNNYcxW7BpGfA45DrDPd3TufLwgZq8tGMDbQATCFWzVooWHmVqebcKpcs0VmakKYdXZIf21JUTG7yRawN3bYWXE6bNI60XnZOsMW6k/08a3j+Thft/vk4aB8A7EvEZHLAdcqEyaZi4kH0ZCxlYv0CKvGjMtOXJetcZXCFzchM2JTQwxncap2P2A7RO4VrH/6ivLMY+HvjhMGPNg8FLvCRO8MWizbolk8LNcSyeuYrHc8wL9tKMeE80rKOVzE9/U+cJdLKCr3U7ZDfpNbImgg/ALonca4cDTdKFARedjk3JxUyJwPjn6CULvF8tGpxoIXp3nDNGqSac9V1f4pquO9EkV7MOCxKZ6kfNxZnTqHlxnxUj48HWKl7QhQhYkiZKObeMJ/ZbyTQ2VVvYExV/9gYFFTnVUrKeyseHJW9MSgbuFAxBcKFy/apDQcWu6D2au/MSHvqy7rgjL+HepbGs9TEvnndVpk7EdCwgdn0mwaS6Gpmut3gxQCUutI3FRwBot9kV3ofsLWmdwnJF1MuDhCCMW18orK/j7/tBY0FWZ11rcNxTB+FkDhXKQxD4nOb+4BTPVaM8P5u6HlAR3i1OFcsCt6Nmmtu66vke/ct2hYVOC6TnbsIPbc2b5+28w9V1fJ+UCTg80Ts0aCg67uYqK3cNrcYDFfsPXYmLRfy/xSTQGZzkRdFIj4o2iYTDH61EQOqinhhSyIDbMgtnXdMc1hP7utNouCdAK4dVmaWRE2VRiCgxDMTNs5/4jaZ5WCjP7+XajFibRFvOfHQpAte5XBBcqKyVhBb/8dJ9hkrO4zzO4zy+sPFZRNzJhksiskYA4K4TLL2phcDKJAogMKfP/2I24NtVjKI6qzna3A0FwxRTXYrHtspWVd8D64u4mxeF5Qq+HSXrZFRri8UtdTMdBo4mKy+YXG9tlkl9tew4YlIqMI0upZ9G+meO2jepWWbmOJp+tTix9sluXLPT/abI+gizIou3V8oxHTCxO4DIKkkNA1IGFDVd3yAxUIRnlGPWhdEOm0X8HjHJXpT0fK1+0uGWuvnmmwF6mxUB6yqew3ZfcxRqjOVGlGZbYDPv8nenZ29yZFRdecbK5AzQK/r5foAl2VPZeSxfU2pZbrF7iuHsvqkYxtrMWywvYsSUzAqAqIfxu/uLeA4TF/ta22x7lgxdZYbCBqswU/Hn+dpi/0gdklYzVDKMKkMrp4ozk/Jo0e/i70+HkmmKprSY3+RIPkXlshQQyZT3waO7o4zp0UPo3CyWGm3QmtyR1/VZxF95NjtOw3mJi6SGOSrsU4Qp/XNtl4ncaxqDU9xcszaOmRBT0+E36sTMiWNfoP0p3utF2WNzGd+5pek5+najZDVI0Qeex8HbZ9mDmTT4JPnd1moUhLe9nEiwpoyqUo6pfiernxmvTDs9EyX0zXVuXPv4uOR5kK4tfffUlCJF2VU58vO+ujzha4LMHvZzvh+7rmQ0AAC/0xsBfqcFArSYSEn+lfFZLNzTNvbUOdm7gAOleKsi29W/rHK7rkfGIkPIGsrfbnaMmTVdgTotaKeaF/d21BC7rMU8p5dTGc/Qx+mhQE2deqljEIg3P32/8xKWBJyqcuQNoFpZzEN8YbYfY+pX1fnBfv+44XOfmzHzZPtigrlKvu5KTviqfcHCOldlz2mT0Y7Fj6QKz3jS5ZpS2t4D+/g7pTxzlB/2c158AHD6V2oLTb+/2Rxzu/oxHn/9pseaIIOnHzM0YVTWI19d9ewKE3yLYcjay4nzLGT0bgQA2wCBvDZ1CPlvA3D6FDdVIYDVN4RT33oMbRZwSoti0xYQW+qam4/clg6AucV3XcUbuwsSilLg1DHng2AIZ1YNqGiBkSqLZdUT/vNi2aGkdnbbCzw+RIzUzDxkmZ7FRNQbgLmM1yq0gD/RJvgRaPeU7teBOcTHbX6hjXGMpxvj8HIZd7XHpsYVHaNejazf3Tea7qPgTX3awdj2ht+tRTmwN6dEwCPRRt83M6bdro1lXnTnNO4Jqrksc6fgti9xM0sBARjOkdKzZve+qVBsU1u+4AW6KCwHGdVsxJxgIVM4HH9K9Z4cWCT+tRSBpRlO1nAL+9Eq9qV8WXe8UBrpebNvmlzHUdJnISwvebH2yNISJni+1qocORDyXmCkz0+F3ITIkGeUcKY6gpMTymXg8/y5cYZKzuM8zuM8vrDxWUTcvzsknYBYfASAxz5gR+Xlp17glyvSxghgxkjrCkjEPFqKwGnQ8e6SIyolA8MT82LEa2JU9KPG0zFG3ItyYN2LEHLTzTgGDKTvsD3l1L9zmguVftIYsGwrXFAnp7uXDCukjs6Xs5bTqmZSyNz2Je6oAPY4aG4OGLzAkYJgCcGiVVJkr81iNFyoWc5z40wYcuTysJux0UAIgrv/Dm05Ycd4fDxlsZwXFCUtZj1H0dv9jBt5UnH00x/nHNk7L5kxIgS4AWnXVBzVKenxaR+j9kI54Kd4fd4J7kQUAtid4n2/WLYMBwy9xrGJx79valzt4r2+vjryNX06zXHlW76Oh6fIrpgdLN/70lhm37iJx2BnFa4WWZsCiPCBRX5WaZ44JzFfE3S1nDgAXXioefxuufX46jZCebKcdImeAhcnhQjAPUFaZWAX+/1DiYrgrfrWs5lEddNB1vFvd7+VOB4o83p95MxH/cljT79f3A5sDDIQkhCZMnScpuJMyyjHDUbreccR49Ox5k7kSrlJETB7Iy7MiM1EEjfNq1I5hglCyM5KerLyKBFYOyUEgcUqZ7dJrvd0KLEkj9FyafFL+QgA+O3ba1wRRJIEvfi+ArhvZlxMvak7Zktp4dHRe/NjP5toHiku2G/mHbPNCjg8UZY7eskFdiM9Fy39foaby7i+2FFhe8oQ1Ad6L2YTBstjX/J7PPW3PVnN0tU/N84R93mcx3mcxxc2PouIu5vAOacxRZseigQMGmfx45FwwEn0symAucpR6JF280ZJ7G2iWGUc/Dt5muzENRcwAUy0G7KGgRQBW4rWZzpbF41ePjMJOE1ofw1zRiUrqiUO52NXctv60Srsafd/HCXz1Gcq4GDzfrojeP5TJyBI/6F1CjvakecqGyb87tMlFhMvvg8p09AO70jf4RerA8tmOi+Yx927fE1STBTjJlKx277gCCHJtz50WUL3RZ2V2PZjgSOotV14jtQPo2a+6i8vtxztPrQz3OiI0S43HRabGD0enkp4crqv5wNaKvjOtOWC0nZbM63wRQD+RHZQpcoO4KdRc9S4oXMCgHHCJ56bkaOkhJtu5rngJSe+pvU8Z2lKe8a11VzkcGiib2H3AYr6BmQBbmH3o4DrUw9BpoBtXuaoM3hAryV/Rt1QfQEdzA8Ry7atRPL6KEvLlMSPf1zg+qsYBc4uYiT59GGGh0PMrgrtONq1VjH1UQITbnU+Lze5XxITz8lFNsv883bNWeDUs1GqgAU9JzeZ41J6Lsg/HWvOvKQKXCcZBs0YvVS5UHg9azlL8C09m3Lk5ycRmBarlYOiOdNYzZKxRnosqfiolWMsu+kNR9AhCMybbMGW6k2HIRuiaOmfYfKvrmK2tT9UPN+u5i1rytSPy2cep8lMpbH/eTz9WSzcSb416iDEk37oBeZJ4lIbXhCWBiyBeluNeE0T4zAa1tSduluUynNxYjsp/L1eHnHzMj4U7wQe7uJb5YJgvu7oFEMflbaYk5Tq1fqEB5L/vGtqrk5vyp4rxoXyXIlORdOnwTBr4bocJhuHZpGZk5O4pJf6D0eFi4J0k0uP1mVp202CD7wEJlrCx4lX4i9o0am1ZenLXVfyAl0oz5BHpSQ3StRm5BdGiMAF19te4Ym0hxPs4ILgl/1y1vIi0D1s8JbSw1o5bpT6an7Cm9tcub9/iLDJy80B1Yy4saOEmcXvuXiTHXD0Epi9ii/D0x8HOJe59wkqWa07fEf39fvtilN5I8Een51V/By+XZzw+jJWa62VuN9TMVFmOdZEcUn3AQB07TH/miR/jYB9igfqPwpoMv811wL+RIVVC8iSXkgtoKkSOzwCZpEXRttQUXYeoJbEKnny6PZ0E7zA+EhsqGuJYkkL20HyfapXI+pNfK67TzUOBPelYuqxKxgqWs06htQObckb6b7Pn9F/0TCSlhUbgJXJGiPsG+kk3/dpgPTt1RObU9tR8vkMg2boxkOwR+vN5ojZivoVvMAjaavP+pHhuVg8psI+QZLWZ0EqKcIzQa1mUgCsWChqYDaUlCHL/1a5ONp2hg21R5d9Ma/WDU4E53RWs5CYVBmOMsYBtA9vm4rZOtu+5DVrbUY2Va6kxu9P/zGr5AyVnMd5nMd5fGHjs4i495T5rAxwQx1zcw18Ra3nnZMccb+qcppUKcccWykCp8JGeqb0dE4xV7JQHjfUAl3WI4pLou70Aa/XMZLzLrYhA4DtgJFkK83MwRM8oavsY1koxwW51mpuRX9Rt8/ODSCIxWc4IimGVdIzraqxmV8KVGgoslhojxmdV+ez47USgelZL6sGl8TF9kFwh6QUIbfFB8GdZlMa22AVn+fmsoGZZV9IN+YIZBPIRWZIRbqcldTVyDSzTTlwkccoz+46lxcnzF5Qan4UKPc5spZkVjM8gqlzro2RdryoLHta1iP2u+xykhx5isJyJ56QgTnKEd4q+fNJMbKbFImLwuGXv4hFrwQPHLdlbmmeOT6+WQByTa+Pz1G5LANsgvVkjprlPCBQWuVPHuOWsoVBQofkCg9uhe+fJAqbo+nEeb74quX52X/wKK4oKh49qwYCDo6gEikDq0SG5GhTZmf77almOKhQjiWNlcwRow8CczKHuG4KNBTZHiZKeCnrAqIsaYIqZtpy8VOI2BkJAEVlGUaq/aS4e7JMUXVOYGjzdUylhrl1vhHcGr8jad1tVzI3f/QSDcF+M2359z4IfvZh1Hz+r2/23FsQJpQ+rT1TeZ2T+ES02KYpWD2zUA6PTzFjk5OCq3eC53/ndC7QysCRc4JJgChdEXIS9lfHOeI+j/M4j/P4wsZnEXF/MydMUIA7BZ8G4GU1FWaPnymVR0uYdecUC8+0VjN2+2ZxZEGp01Cw1q8PAk+kWXwlAypLwvMrAbunTsESaFOXmo7CPwDQngxHAmmHTSN1ko1eMh6opM9FiHpCk6IdXIiA40iNOcpxJPA4GBaiKmUuVH7oNF4TPezGjCyopUTAN+uI0Zal5ehqe6y58Dg3lqOSQ1/wZ7TPXZ8uCG5UEjJbZTWHgumAzksuBKV72o8aM4ryR5u78O7ais0KlnUWm3dWsnv57mkGQ1mSbQTXAswC7HwenOBIFRJs51XfeCgTI+vjtmQX+XePKyxO5BM6Gv794CTPm0I6zozmesT7x0gpvVo0TJvsu4Rt5rrI03HGWcnX+gmSMO/hIfD9GjvJTTHLsge70reAjjVTuBMwUCZXzLKJQPA52yuvAxz16ZRXHiV1OtoTIKnI6R1gya+1eK0g6fnFiJ+Kr0fFGjTXm4ngVUj1AYlDm6O9BVHqhAismz56iXrSJJfuR6T6JW13xxTAw1DwfV/POji676NVjAG3p4Kzs1NTcESvtYf32cgk2bQ1TQFLP/vJXN11FbZJS2ZinvLU0zzsC8wn+iFpVNpxdro91Uwxvn+a4/oi3ic/cWQ3xuX7MSqeP7ttwZ95vTxylL1vKqYVP3UlDlR7ckGgpvO56wv2nW2cwFxl3ZLVpJP4r41zxH0e53Ee5/GFjc8i4k5j8GDmxIfG4arI+0pPG5ALJTra/SsZmAq3KQaOBE6jYY2B3ipmiYiQyYTWSsbY3CHg7W83ACIGm4awAeWcIoSHmqPNYcJKUNIzGb/w4hlNkKvgFB0s6x6fiLXwsZkxjfB9V8DQri1FtjfajRJ3Xfy+r2eeo3ktpu72Co+EsZeDw5pwwJeXh2xmu59xM1BjdVYtHDUU4X9fbQ5sktydDDe9mMJxM0xSCQRyxL2ZddmcV2e5ANMXHIHtmoqbMMzKsZVWUVjUxNQZOwlP3VfF0kMtqBYwCxjuSeNjW7BdGaRHeR0/3+wD06r2g+HW56tFgw/U7PP2VOOSWD6dU0xrnDDwEALYsTudbz9m09epJMD+rsKMVBGHzqBIWjeD4nvnWoGR8O76lYdaEl56rVC01BZ/F/gzqg4sNaxelBAHauM/OOiXZKR8NYP9c8yw/GOAIIaVrCTKv4vzQL09IRC/VJUn2B9z1ApE7D3h9heLTNcz2nFEDIBla/tR8xxYVAOKMbenp3tSVyNUnyhyBb9zVT3ycetqZPy4gOVoejHvUc7ycc3kHJLT1FRB8KcPq6zAaEZWg0xqjU1bcMQ9pctu+4LfuctiRN2RhK6Xz+zK0gg+14n6QWfMOgjczuPnnM+NduVE80grhx1Zs/WT+QZkCuXruuN38ffHml2ZAOBh+AKsy16QAFHvBD528YR/s5ZshdRYgXKCTiQo4WAFLgsqXojAN+FxKLhTaWlGCLLTqs04cZYGCxYBwBXxNU3ls3WZzxrKq3WL8JRodJl4XhaW4YZ+0PjTLqbdLkhcEFTA2ieDYW73Zdmjp0VzYzxzui+LkYuTYyjxLcFIpQxoXXaBTunZwlhOtd8d57ina/3FeofVOh7/ctWwtsgvr7acGk+dwU9dgTUVqazNllt1PbLuxTAofplTUTPSwOL9uro+YqPiZ/d9yS8UkHnRdlDsLamMx5h0XlYjNNFkvQNs6iZUYHrdXPYob2iRuygw/ED6Mtrzvb6qW9x+RbrUKqf7p1FzkfrVxYGLTvFa8sYj8i0BANxv51xYevVmD7Oi85oLyCrR+wIC0b3cbmAoqP5aQH8X8ZFgPcITLZJSQGgS4LpV8LQQmksBuSBO8m6Ab6nguVTwpGssv62g/zGu7nXxAe5Am8SThV7HG2j++xuEO5rPv9H4mro3k/Z5/yl2ZgKxKJxGc5+Xg6lVV2ksL35Keby8jhvH8VDxonUxa5hC+maygCntcf11fNFUCeaaS+Nhbkig6sFBEiwqjIDdJeojIMmbs7uTrLlytW6w3SeLPAGVoBBa8JXyeJHelWJkQkAzFHjskidsLnDerrO3JZDhUV04nktTCGzKO9feMyxbz0a24hsOuai9nAhUdU5xr8WyGHAaEn23xBV9bmlGrM2ZDnge53Ee5/Hf1PgsIm52TZmkCpUCElJSq4CKUsKX1YAiuW5rj28WMbJorWathLmXTPcCwD/PypFlPvXM40hKc/VyRH0Zd83ytYI/xR28/UlAz5LsKCB1jCalCuhPFEWfSo7YZBnwctbw+aToPhXmLgsLJPOG0eDXtNM7L3DfkStM1XKxczdqblwZg3jm8LNi1x3LGg1Tg14hYpMDEKlUSYZVqgBL0WZVjSgnspkpde07jYvreB2q9DARbcCwFeg/aP7bdMyavmPoNH56iBmHkR4zOq+icFz0W35tOYvxPWA8SbNeCqg1FXB2I05/TpQ6oCZlw9l3NcKJrm9mUPyKrqPpcCVjVLf6tYeoyOj4o8WbX20BAC/2+6ykVzrUIsEcmhX/AHBDSxr1ZmAaKGSAviVVupcLpM6S0I6MuQR7wuq7GPmK724AS7DCbz9h/EAFuAsFsST6XG8hCGoKfQAowlSv5lBEB/S7HmGI3+9+dw/13WX8zDdr+H99iseqJfxDfMZqXgAVRe53DUf3ya+muA6Yj/Fc7EmwMfNUNbBcOrwo4vxsDyY3uZxKjmzremAKXvCCn/HYFCzjK3WEjACgedAMieh5gKA0Wt9K2I80n44BNicBKNcMRhOtAAAgAElEQVSU5a49Du+pq1PlufV0nGFHGiKpGHixabjjcegVVJJALk9MfTTGYX5J7+coWVZ4OGpWr5xdDFx0NjOH2cvUNCXw8JGynnpkmCyeG2WH8wGC0B0lMqQrROAsZbCKKcP/sD7gepnhoPbhP16azxH3eZzHeZzHFzY+i4g7FYjWxuG2ynvJ19QCfbIKTxT1BACvKZLVMnARYlP0eHUVsbemKbiQlqJeIGK3OdoTMBQV1L8QULdxBw3NCJmaWx5GjspXrwZUN7RrGsHKd+rPLXSddlNgeJtE/DvUs4F/DwDjoNgOalX1HCH4ILBwE887wsBe1R230H/sCvyPt/cAou5vagt2Xk6wumMujkz0fEMQjNsFL5ieGIJAdUHnXgCOtKAXbyyEoYi3kEzH699rvP5VjMIURUL+1GPcx5939xVHEPdNjQ2dl9KezSnMnUf9mjs7GNuUM8UFNVlJLH5D5z54CPIhhQ8QNVET71qor2IqMPsvHsP/RcW+nUd5G7OXotbwO2oCWgfMEe/r4c8K5QXhoouBwxc/gguFqfCpAXiKJNUc8NQwFIY9JGkyiEUB0DmaX22AX38Vv1Br4F+/jz9Lwc046s0KmJO65J8fWcMEAGPWcu2AOs5heSUBNfGX/CnOc/ntFcw/Gv4eR+fW/u8PKGJQDqEEAhWLOPIeyPeSxnCkcy9zg9FwzPNKac/ZmA+Co8qpM+zUss15iSVF7hGXjt8zuxqZ7ggA3Z/Ss8n3pv/g0R+SFRgwUIYaAtgMZGg115XUpNkuRbKXIvA5jseS9YM2i5YL8N4LSKq1+xFwVNcaB8W637IAXJ8pqoJWy/prgQvianonYGb0zAxwuqMi8txifkWWfgeF9kRYdm3RURPS9w8bvheLYnxGPbwqJ2nHXxnniPs8zuM8zuMLG59FxJ1MO4302BJ+vbO5zX1lLEZPLAflcEnUnUI5VrEDwAyC9UWLF2vSxW0kTru4C25uWlRfUSPIKWD2khgpv7oB6hgBiU9bdP9nbHvuDzpHFxKwVHyWJsB8RedzO8IRO8VcClwP8R+2n7iTE36mx+zeo43HfkdmwW12Yam048hhKpzlA5452syIIbFc9NxCq9Tk+8vsHNM8GXbycaPAaR/vR1E6HD5Q88LNgOIlZTU2RLwVAEJu7V68yFkBMyqcR0mPYC1bdjIpC8vRzfylRUHC4q6XGB9TBAjo29w2Hjq6X1clJOHH4nYJENbrftg+4+/5O2ptLxWqdfz+x+8r3FCdQb0o+TranwSzErbbGksb51C9GZhtIRVgbuhn0hcQxkPQd6i1ytddaghiNMVMgMK3eQW8/RR/bgYEYoxAS6ireKP8U4vhnyI2bTswUwYA5ILw+R8bpg+KuYGo8n1yn8j8Wj9B/G20vZffBPT/RzyuNDHSBgBRSTiq2YBwciHBNRqlPRsz9yeF7TZrsqf37zgU3N5eacd1DiEC12OsldwKX5uRo0pjPNv5LX7p+f6JQkLTvR7ejkwFDU4wBVBO2u618c/cZU5EMZwVI67W8Z07EHPqsK/YBUnIgDnh4aPNGtmLckDxSHPmac6KmbPZgJoya1kLBJqrzSfF71+5tFBE+lCTyF2VYLqj7SWq2ySD4NAc4/lutzW7K63KgQWwPMBsr8Ou+E9b3j+LhTsJnY9eMtWvtWCKnBLZzV3LwBoDPmRYYVEOvIDZQcH0SW8gH0eoAN8SnasQkF9RK5sUCH/4GD9TG4YB3v/LGq9uopKd0ICn2kG306hGMmqo478BoD8ElEtKm1TA4ZGoR0R/M6XjczwcszfiY19mj0fhuStr9JK5xz5EkwAAsIcl37PVqsuqb6eSaXeLV0eYV5Rqvx0zBLD2cCN1PXaaedTDXmIgS7MQJOpLOudXiuEB12eZ0vEjXecsdznqGYBYC0Q58VIUBVBQVhis52dSvCkizADAP7S8aIlKQ1wSftCP8B9JWrNz6N7S8ysDzGWS7oxKfACw7Ht0H+KfmmPHanunXcFc5J+Oc1x2tHDvLK6v4vcXM8c5qJiTo7cdoBhi8KB1CuHYsyeker0EXtAFtj1T8YCJVKv1wMzwz+nFHI4aZkVUv1rA3tG7sBcYf4yfWf2D5Q3LPQ1QlwSPHAbIHyJ8JmYFzDUZjHwfgEkhsviGvCN/zLBh0iQ5PJXQtGEWVaa2CptXjkUxsLrkohx4gS6UYyrovithkpVbNWC2iM/eWYnlgjZ8n+eNdw7la+L8v9Jo/0Rz8pTpouXScnGwO2qGborKcgf0TGdu+KzOC3Qy3HATY5R21EyAGBqVF9BFl7n3VjL0V9ZZM8c4B9vS+tJK3lxOx5JlaJeXPasffvywQt91fD6po7ixmoM0Iz1Ti3ur+ee5GZnW+3PjDJWcx3mcx3l8YeOziLi/J8uso5WsVdJ7sDKeEYIj8VoZnCzpC4uAX5PLu9aeoRIhA1SdiyNJa8N2Ep4KNbNfq0zn+uER/W9jOK1vFAvV/83fP+DxLVlxvQfqK0q/moAf/xgjrJubA0xF9MGDQbmmokkdeNdXXTYTHSmL+Ok4nxg2ZM3u/VCwUti71uAbKnCujWU96cuJa/f90xxffRWzgtl6xNsf4nk9/KnCak+i9aPEYRuv6WbTYP6SIugfcwFKiIDTgYwXlj0r9bmdRfOOGkpuAgSx2GyS1G7A279tctebkDk69w3472Sdi2Vh9LB/jNGpMAL6iu51qRE+UHfgYYAw8YvcwaN8QfPjk8DwKUEult3RzQqsznf4UDC1y1rFkdfLecOO9mVlWZcEAPSOzl+TDnTr0X/KBUv5i1zEnhZNw+8ozC8UU/EAIBCcIy8qiDI11/So/iEWVuvrBX8Whw7BUXr9/QHhHRVFX2+4+Am3Z1qhmkvYP1KhcqEg1xQJX3ZZffDJQ5x6vvcA0P2UjYPHUUPrTI1ckm1Y1xim/ZnKs6P9rqmY/loWlk2lfYiGt0AsenOzigqsNx4y0gapgP5HotcdJUbSGxkGxQ1rF6bB7A1Bf4sRp4/ZuCI1hf3+wyVWbbImJA3ytkJPRg4SmYr30Jd4Iij2f7p5ZAf5vtMc5UsV2OrN94Ca0Rp0I+GbnL7bff7bpCNv+3zdNzcHLkh2vcGC3tnNvMWRmtg+nuZ4SwYn3yyPTOv1reBz/rlxjrjP4zzO4zy+sPFZRNw/UiHD+sx6uikDqL6CgxVMqWudYkeNi8IypjuOikn5QG6tLVYer2+pUHnICm2piSGNRPXpvveof0mNI681NuSw6gaJfku0qcozLuqdwIf3semkLCxWiqKhtcAtubXcE1l/d6pxIHxwacasGaIcq4cVyuOWiq+lDLinCOFkFePgVxVQUtPGq5d7pq61HyVHkutXHWPPx/uKFdoO7w0qKlTOrwZ4ynCECigJ+zYzzwXXx7c1FmuSCbDgrT4VZCCB4weKLFrDzUZR1yEep771EGWiDwZW+BseXcTFAei5QGiogHrXwhMtTr8sIVZUKwAwvo/3t3wh0H2I37n/WKGmqGf5S4eE76rGY0aR1OlQckR27EtsySh6g5bxZmclY+JqnTFeTbID+mWJZz3xhDt3/3XLkZn+bsn/HfY9Zwuhs4ChguDfvgBI5wR9bt7xhx7hQBSyj4Ebn9yHI+RFjHjl379CsY6FzeafDjxv5cGi+JYMev9hA/n7mBJtf1dwFlRdEfVTx0wtjRRlm8pxAXdqgKwKz5odx8HwLRiswp7m80XVc/bmvISkYmM5G+Ap0m4fFbqGqHlfdTCXdM/WHr6h40lMGm0CAn2PXkssJFExLbD7Y1IEHLmpJjUPoQUbCFflyObUrZO4o+t4bCteL5qu4Hb95W3OZsetYIkD1zjoK6I7PmVFR2Ny4TFKJtDnJ/Zji1nHxs99p7MZ+UTbaNeXz6Ls5Ez1c+OzWLiJNBChESqAL3Vgp+PdCFyShdd+zD2RS53lKX0QvDidjiWeqDq+WbVYf5cnhX6dfdISH9Y/9LzIyTKg+QMVWX4pUVBq3v6YU39ZBlQkIdk85a4y7wROdyQ401n0lCq1vCi7ZxPqp218yTunuOvzNGq2V3ocNJMovm8V/ter+LczbZnz6UaJnnQ9Tvs8AcejfOa+noTihQw4HeNCuNq00GSTdnis+G9tJ2H3eeJtH+K9NHvLnadCkxDWp5ohqs7qvDieSi7EFqsWxTrBHQ7bD3ERqucjJHHphRbwZLApCgH9OgmXBF7QxaJAINF9dWkwv6AF7597FCvqfLsHV/x1mVN8HwQWy+edrACw2PSwJOiTFhUAXMQe98jDTTjllYb7iTQ45gLmNxGiCtbD/iH+kX5dQ7y5iJ8fbZ5v95/gnii9nspkeLD/pHcS7ZbgjLcSV7+MHyxfriG+uYr37zjA3VNB7ilg+IE2/FUJ/Yu4WC0PB96Y0vwVAvj272MV+fC+wCeyj7tYtqjmZLdX+SxV20hebCrlsCEhs37U2NL5Dk5xh7LRjhdRO0jUt2lBclkHZJUhM3VleKP0jcPm1/Ecmh8Fjj+QmNRV7rg1G8E+muqD52Mlc4PDaFhLZ1YPLIj2ohpY4OkvR9LmEZ8Cd0Lq2meITwq0fyJ20yhZKM05we+N0h4jXV/XK6zWZDrSGmy32fQjMUneLI8sS22kZzZZbewzHaG/Ns5QyXmcx3mcxxc2Po+ImzbB0QMDhZh3vcATZS0zLZDSXzNx2jYy4IF0CpZmwFrkiOrjKUfcnpqQ/IhMq/rjI8afsnFAUkYrlxblC4oWHhzIVYosrOLPuvXMi/ZOsDqYEFmA346KzWxTZ9duLJmOtHSKnbDv+wLVQJHyROR+prOU60Uhnpm27imCLie8UCECmz00TcGRpZKBZUqV9LB0XDuoZw7eaejKs8VU0xXsbl+Ujq2k0hAiMK3xNBqckmC8F/hFFaO6YMFZjJCC08Zi7lB+kwrKgG9I3e02U6HCacD4Pj7XyKOOv+/+NKB8Q/zyv5OwD7m7MaXRh2PJ1LUpD36zyVKmzb5gKlg9H9DuU9Sdn2/iDwcPhJ7S9c7C7SbFqh9yaJ5gIXE1Bwh+8n9+xPCO5psGJH1GLwMXf83MMywjy4DS5nQ5RZv2n95D/w8v42e+vYI/xKKoqgOHYf0/71H9Y4z0q//lEiPpmaQO2LGRDEEIGTgjXLncrXd8KJlmN+0fKI1ls1sAuJnFe3nX1Kx65/3Uekvh4fc0l9oCr76jrCMjEgiNY2ql3QaG4YpVwNvfRhhydhq4mIiHCSyCaOQLRE43AJysyRaBMrC643Xd8s+bsudIebXoONNoDgU+fIrHvL44wdB7HhDYwq89ZXMRYxyKiZ1iev/HUXGGYwoHR92pSmeZ2dNg2MDh3/ZzvKGu6nUQz7LCvzY+i4XbTdaOOXFmR5+lQF3IDBMhJKcJvz1UeEk3rdaWF5DVrMN1gi+84MW3vBUsUjR+dGgfybFjnyvlzkqsL0iApwG3365uOjx+ICd4JyFP+aWdOuJ8OMz5fLJmN8E8XcGL70xLrKnS3FiFH4iB8qoa0dG53/cKUtAiG8Ca3YOXWNJL0o4GW2pCWhQjX8d9M8OMNpRSW8xpUlineFIPg2K98Wo2YvcUF8yH3Yw1xFeLyct8KDEQjLOgdFkIcBNU5xR+dbHl36eNTpYZWhmPEuWSRIp+ZSAuSJ6zGaFI5tMfBoSGhJds4M5qe+d44TYbAUcMEPPdEhqUlv5o2eFcFw7DXfzO7/dL3oCWIQtqAcDdY0yxF9XAcFuCedavOoZe5Lrkjf/wTz3+7ftbABFnTRvjN/9bA/2ruGiGXYv+X0gE7VFBEd5dXnleOO0uLgQAMP+mZy47fMB4R/NnLiBnBOf84OAO7+P3/HcbqNsYoBx+1yLQczWVg6aNRL9ZQl2QqNHbeI7lxuPue5qn1fhM1jTJ7O6PFbeKz4oRy3l83s5JvH2imo5yoMeKSjvcUxB1vWywvsiiZkmQaVYPE7hRwPyXXGfq/+khfv9Ew1zPPM/nPzxcYK5HPlZiZBnpua8jBUI/NhW7Z9XHDDsMTuGpz9h4Yq/0g+YApu811nStUuZ6TLBgj9PtqWL3HAC4v4/zJwTBev5aezzdZx/UBG1GeCm7diXt/VoFPNL93o2a63g/N85QyXmcx3mcxxc2zgv3eZzHeZzHFzY+C6iEMkicbAB81g4mMxDshgASjsNa+0kzTsCa2B0PXclp0FdeMhvjeCoxI/NYcy3gdkmYAqzqdb0+oXhBGgpa4Pjb+PP7n9acQn1zucXly5gGff+nS06ln9qaoYdFOeCyppQyCP7ba9JSWJQD/r+HmEZ3VrGmthTR1QYADlbhnirTT4PAivC+UgauBVxVPbfHKunxguhc/aBREcQRTsC/bGNK+6vlkWGCYeL88m67gtzS9d1s8epvieI4Ak8/xrT38jcDt227fxOYz7PrDAC0jXlmlpzgkaKweCCbtvluQElUtHoVuPFJ1AboqMX7h5YbRHwbmI7oesHYopl5ppYVr2X0ugMQdh3ki3gs0+zhyEBXlQ7FY7zul3WLXcJmH4EXROe0nX5Gw0rty8lQGT663QBAGBw8FV6Cz3ILy7rH5QtyCXof4E8P/LfDIZ77/I3jaxr32RTY9ZmVsPuDZuy2aw0iATKyb5IWNADc/YHqN8ctSwl4p/HTXeS63l4e0P+/JH1wt0XxdWbLAMDxY8G49f12/szRyXvC8CfzV0nPjkFyoi392/2C+4Je1xlSm897lBfZJDnhtW1nYAj+WdUTKYPWon+KX3T3YYElwXPDY25zv6o6Ps+qHPn9Wy56xpvvCKa8LEZsCbffjJrn57vTDGOCIduaWRzX84ZrQwD4+6QMWBCsJ+cKN0V8gMunjpvVgGyZ9uf7rPY3r3uWD6jnI2rC0J8eZ7xO2Uk963FQvMb9Z0bBwGeycJOeD2Y18KFLHYTAhqimxaQe1k24j0s9sVQSAVc0eebFgCfq9T8NBsWGCj7rEr6lhbUFgiUs7XIiXHTyjBUu6x6bWxIykpm7vCgHfKSOp1XZY0V4r/cSW/KZW8061IS/H475Ib8iAaQQBItiKhFQTRbuE52XEtkEwgbwdbcTvQMlAk+Q7anmn69nLZsqAMC7faQeauknOLvFzSa+PM5JnEg8KATg6m/iNenXs8hBBrC67zGStdy0YPWCrml0knHRbZMlXtuDgTREVbsBGyag0rB/IpsxnYWrWJgJgFkECJoTp8eCNwbvBlSv4ueHdyNrxOhvV3D/T+Qwuz5quQDRtqqgO/7UVvj0Ay1+ZY/bC1rErcR8TZspFSSHo4QmI4dwGjF8oA3rlIXBjHYoLnMB8/B9fq0Snh9sLjDev1vw99fVyCJMx6aamE9YpqV5J5k26MdcuLp/t8BAx1rOO1ws0twCU/kefqhxMSbLNPB17ruMs/7qddxodJFx2dJYlMk1fdFz0a3pimzi0dT4Zh6POStG+BDn/tBrVC1tyI3EctHzcWcXqUArcPrn5DeqsPybeHO+Wu/x+H0MGpJMLADcbI6MSVub6YkhAMtNDlzS7xJ63jmNDdWSXtQddx+7INih3hjHUq7NMc+xejlCUJAhKwnzgp7B4FBS52b3pHGk93vwCn+m9+zvypH58adDidUl2QhenWB2cYOtRZ7nN1XH7+vgFFsv/tw4QyXncR7ncR5f2PgsIu7XdVIHFLijNGVuAl6QjdOrOmBLu+3aeHaC73zedwIEpz7zYsCKdtnTYLJB6UPP6aofBbpjjPzMamAdguMnw/ZKm9smd6+1uTni9rsjrvsYpb378wYtaQ8Y4xiiCT5rpCQX9m9utkzdS6a7AHBRdigonNd9CZ2iygB8oCq/EvH+ANEFPtGIlpVj2uE7okACwNXqhHURI62uNdwYU9cjOqJPNUN2c5cqsM5CUVnoC4puPrYcCZevAEnQg6YGwdnJYkGaKD++37BexawYcft1ZDYUl7nRxB0CxkdSDdQDR6FSAYq+U84VtCco6EIjvCcHdSuxeZlT8h1BWuXSw//X+DyqvymyCe2fPNuS+TswEyFFjGmkSK4sLQ5Pz01aV5cdm0qMP1lWi7v++x7Lu3guH9+uULzN9lUJRlImZEikzRZet98doTf0LKVgKOpim69Nlpky6A6WQyw3CG7sUCZg7OhaRwlNb/PpVHBEqqRnXZ3Xv4iMn8vvOiy38Zn1p/+fvTfbkSTJssSOiO6qtpub7+HhERkRGblUV2Vn79PDIcHBEDOfQoA/wM/gK3+AjwQIAiSGM8AUZnqaXUtWZVZlZGYsGYtv5u62m6rpKsIHuXrVC8yqeSIQSZg8eVjYoioqKnrvueeeY3FWUCQWQ0V5bnPWOk987HTMjTMarbBeNlLKdQfv/u4SO4V5z+/PdnEvNdd+93gFt0OZ19TmzmXZteHmxBIJKs7C7HsednyzWBavFPxWkzXWjVLJ2uXMMs0czl66NC9paSMjaKfjZWzbJ0SjLHp3xBuPIaqok/P3pWsHNhlxqLyAc2guvjtUqBLSPNopuUuy4+ZMg0wzB5MLcz92/ea62p7GATnEC6HRogxrnbqY3ZWoVn86pn4vNu6gdmmWAkPyoUvKhiaYVvIPKIP1xuZLxfiVgMaAOKVB0GBgvV7SSJFONXLqCPR6CgGJ68RjGytajNLSfBHdVsl4ItDgn/5+AYsWoH/Z0Ipsq0JE6U6lBKaJgVPqDWO6DJHRwyUp7Tu+kRXzn/eDDXBH0nFVa20L4JggmbZb8AIsKwmLdAIe9+cs+vPyps+Uoq6XNVzr3OVFvdNOEMdmMU7WIXY7ZvOLwgrFjflsNpMsnOUcOXAqMwe1dOpiEmBDnZNSaIxTM1+dKkd4bc6jq1IW7skWkjHrdGMjopu6xkQBk5bWD4tiXICaJRFEBZw+pa5tC5rwWPMQNue0+ZVC68Acb5E2ziPdIEOVUCenVfE8VUrwWgFslsitb8DVzIOwzbz7BxLCpU38puJ6zMHDJVMWs5UFSXeVHWl+MFlBw6N2BpKFzCAFQOJTTj9qxKQAph46rg3k1M331RWWzwjKe6qhUoLJvpX8wIiiRuI42bgMeaTEUXcHBdydWoPbbNj1GNyj9P7Gxvdzs8iPWjE/CKaTCBEJn/1sNOGah9dTcKkFunOdMzwi78DrtqtYJ1ynFayI1vZBCDUjemlewSHZgHa6wuS1WUOVkvzASlK3kXi1Syzp3q1rBa5VcZ/D0Gloe3llYUP3wdDLuDaltdXI7G5s5oWXpWTXncWli15G8GFPsMxDNCqYUotboKz3I6FZCne+8ZGdm2vca22Yb361iljGolKS6YwCmmHRPza2UMl2bMd2bMePbGw37u3Yju3Yjh/ZeC+gktrpJlMC1wQHte4cWa4ErijjcaUFghyRKYEzEnu5F+a4WBlAOp7ZGBLGvVMkiGOiRrUy1tS2e40tkRdVmM9rM1TNtktFbjGWtlgF3FWH50Br16SIveEG5+eEm6Qurgln7gcp+iQoVTvXWEKz+M08cdG+w/q4oW4qCaBNqW3dSQWY9v4+qQZ2wwYzc5yKq+nWnfbep0e3TOe6mbbQuqOh3KXXVxuP07a97gqztTn2VpZxWu91FSvQFZcFdy5Or81cv5530SYlM1sqZDRHjqc4dVWVQL4ifei80QBv9Ro8URWG0VPPQf23kID/kGoRkwKr1+Zgwt0SNrGOpq891kS2LI3pa7Kwin2GlPb6a3RqEbLcYakES2hmFyhdoa481EwFy1NwDKsS5aKhI+rSYP4AgErg+mvzffM4wOGeYbUEPcHWaVbfhvvUzK/wbKBNcJhtAeoO/asukqw2TJeTey122LEfDRBODI61+V7Av0dt/I8LJGdN52TN+rEsxVS3ej3MX7qM/eeZzbRNpYHByqzZqJ3h0yPzO2Up2YQ6iV2ukUhLs975euwyVJNXEtczEq4qEnR3CQaRYKqmsDULu1WrNVMui8sE7kOCPoYW7DNaE0ozlrwpbIYW91oxT10NWYZ+jgNa72/mHb7PrDuSEY5UDGsCxmINMN22IUGlYb9gnXchc5bFEHda7uObpibWDzZIy2bjqmHcu3NfC2EBQFZZmNE69K0SKa3PwKqwe0dz/4fGe7Fx18XGTSUYy55mwA4VZ1yp4ROOG1oKLVog98KEMaJ+kPKm2O0n8HrEI83AlkPSBkTdUbxpFo50NHYPalqaZswvXnlME2pVKV7dGg72ME/ZIT7crbCXmc9e3nQYt9Z33LBrR/h3qzZKwtXmhQ1/U1+0iguPe34G16qP3UVKr1caf+CvWd5p2a4lYX2rwr0ds2m0dgs2Qwh6c7x5aY59POvggBZ7+w7XVErN9L2Lyy72d01xSaSNZVOydllnpB6WUFygigsHCR1XL2z0H7y+QkFuXpupj/6+oVI5XaCk+847tNjBffXbAim1gXcOMijS+IBssOfJ6wCjJ2aTGT5JEZ+RDVZLwa/M+28WEfYHK74eNc6vtcAJGXAIoblgvE5d1jGxqTCuCgFBN6l94ECRdyYUWMmyXGmMnprP7fdLxr6F40AO6SIMW4Cka3azRPXMWOVVi5Lpg36nhGxq1ozti5dTuLumsGh/vMM6JKv/MEd6XitZ2ghPKMh4bvPGLS3NuH24Q/S3HUn2ZkD/JEUfZmO9et5qXNArwdIBRWLxpum6TQt5mjqYJeb8LKl4zXx4csNSp5alkZO/pRtVSGeN2UKtFCik5k3fdhXi/7t+qDgcLCkt2HpumbtMxdwUDq9bp5aBSF1W4Ou6OfcwKC2YjptXFr9n/4M1Yw/pRP6Bfk+9PlUJRHvmuOKx3RyvB0yvzYMvzl0+FiE1SqoLVHf8Oz2n5L/nuccWhFIY2VkAWJU2v+ePjS1Ush3bsR3b8SMb70XEfZ3VT9UGIpECmOZNI0odiftSYZ/oNY5UXMVtRRk8Ug+L7t1JwzLA7zdMAWaYzAU8ItRDADZpA8cXFjdzaNWk8r3DFMd5I8YznVCnnr9i+CWYFw1u+WwAACAASURBVMxW2BTNU7N+Cr9YtbCobZnciivHoV3iPjUypJXFVD8ACOn3j4Kco4yWl2N0j9JoD7DahB1VQHJu/v7+mz4GHfOdrWHGBgtXcYiYIBghNLqk8S0dDUmRmZuVLDak7rhup5nD2tuDoQlFfL/AdNk0s9QRRLefIDigqGTHhUt2Xv4kRWnE6qDLhlYoIgeamjaclkJ4Qo0wLQfVxFzX5VuHDQAsoZHemGzI7TYO4OtrF0HbvL8Xpfy3FTRR3e0s4qiqrCz0B2aeusMNC+fXwmTZyoIHUn+0BOwjwmekgJqYuXPbgBwRDNINmAGCnS5fR/XsAvkrs27dE5fNiDErG4ghtiCp2cgfNqbKugTya4KO3AmsUwObtP95F/kzE4lXsxJWl4w5OgWWNyY6FZXRHK+/BwCEbyG4T6JqKw1nzxzLgb1GNmnWXp15OmEFtaqFyZoouNVO4RDMdDbr4n67tu0DJDG2hNAICM6RfR8uOdTLtoQu6T7bKF63VqA5Q12vfPRH5tqs5y4kavYScNQz96LjVE1jDmVUllRMwXUtxfBhpQUWmXm95eSsilgmjWFCMFIQpFEvA8nzHq882AR92m6jnumgZBryMnOZqdKLUv7dt6uWYYvRYPMXJfDVzKyRB60NWpQZ3GQWbov/N23x7ngvNu7vFuZEnnYFQjqipGw6Kqe5QEQyZMdhipO+Scm0EugNzYX1+ordrAHjhg0A3k4js1mtFHvxQWpUlL7LECz9mqc2liR63u6kWC/Md3adFMM9s1ldX7aZ4je/CTA8MccwGMa4uanNEWyoWmSeaD5pJbmdVUBjRZu8k3sNtFna2KGNPrAq2PSk6bsFHu6aHc92KhDNGcGTxsmkvEwQHpkvOsSCF5cqBLoH5gRHyxRzokr1opQXrLABr6IbSQveTKpSYrluIJr6wbQbmBvH7WR84ywSHwdD8gD1NPtSyqgCNnTACgwHCEfAGpn51VnFDufh3w0bDHiyhiBKVxhn2EGDadYjnUg2UnDCHIuxOd6ylOyt2LtfwaZNPM9SxmyrUjIVdPe0+W6LprW7ryGC2plBsm+krhS37lu7kYFCABN93MGv9bcXfH7ep2Zt6HXO7iH2vodQUmfhquF6S1/AHpJS4axiQ4erX7g48Mwcyye7cJ5QN+i/TyAuqEPYuiM94JXs+Vi7+2xeFvCPmg26pAejPbRR0cMzW1gIBuRAvpDUgm/gpNrv0Q0qVlmMnAKrG/JE/ShH64SMO0IHyAmqiAu+F3WqmNcrPMFmC3cDp3YnZUeeqJNjTU7pJ/0FQ5iqEoy511RbzykZvrvdBFwrGnkZfJaK0Lh3au4ndwDYu+bY1apA+o66ZhOJgj7rOBU/DKNOzvzy5dTH3shcj366wfOx6dnc3IE7lBaYUg3rJnOxJthpxys5eHsb+3jUJgjTKfD7xR0v0h8YW6hkO7ZjO7bjRzbei4j7gLSGk6oRWHEcsON7ZGuOuBeFw0WF9jCDRWmNe+xyBFTdZLD6d5j/td5ApKFJnF5XjfO4cxKxHkf5VnLhryolF+/yxEJ0aN4z0mu8fGPso9pBhnTWpJQOPdFbTo60+sPpPQ4zBKSRnVeS7cpCN+fCWbFsI7nT3eVwWiUbYZ29jP361CpHeUPCUmUjUO/1NXesrccuQmJ+9KINO0svEx8dSvftQMEhTRcrKFCQ8Ptq6fI5FXdgHE7FBXiOXKvCcmVe1xrofUBwR2RD1wo6UsFum4mvFgVHnjqt4D6kQl47AGYm+sifzWGPKOrpCchZLcJk/0FWUE7MnEWjhikwTQLOENzXCwTHJFj0IGUYolj+YaNDDTvV1mVaAfyOUkGtmmq/9SFhba4NjCm96IbAmtK3eMEsEZ1WqMbk+N51m6JlvwX3KX1NO2xYJVIAFqX+V1Pon1PkrgVmvzIHP2jNIcglvv/3FSY/N/N9cxOx5sn1pI1BQVkpQYnJ2oMqKCMdGpYJYK6XQx2dq7ENNyM4pZA4J8GytLRYqGxfrxrYxMvRHtHcVAKaPiuHIXQtKpZXDFXCknxfJr/LkFNmVJUSOdn/tUcZspX5rfkk5Pu+qpoC4mIR4oaYXHWWsWNVuKUmNqWB+9SpeHowY1hTWMbH0hyjx5okOsvh3ycihC9QEhw3/76pGleFhNshHZciQ3FHlKrWkYmcEiuK1pNKciekbym+p32pENUa8Xc0xjeVjb7b6Jj80NhG3NuxHduxHT+y8V5E3BuqPDpSYEIFyawSbApcNHAijsOEn7bS0lzc0lkFTTiqfb/FuBpcC4LctXVWwu4TZ3ej4JyYyFMc9oCL2rmlYs0DrQQ6BxQ1+kB6Q09oR+GYaHdlKVnjQwjNRbJvFh1+siYUKd+PNiwF2vebJ+qbRYeftqmSjHsp3eiTWEJx27HTFdwyreNGQwFSoLww0d5mLNnRRivBhceonWFvSZixm8MJG4Ngh/q27Z5AtWnU6+rIVmtgc03WbJQhLBMfe0S56/cSNmm2HQUZ1BFNCESEZV+vkL8yEZCz57BprgwlRJeMfV/dcuQpI8kYrHAEHJq360kLLlEyvbBkVbvi3IJD0cog3PBxTq9D7LXIYDYCilszr96uhhUQD70vIWvlwoq4eLWWLswaA0WS1mG7iY4nK6BF53e7groxEa5KFex7JiK2Px4B+6RZ1+sa/jZgONwZ/VaaGd4ZAFxNoRdU0EpLzqQGhwlyyoZmP0/R/QlxpCMHQZfcne5Q2tpBxpmaSJvbfXpt5qudpszp3lxKrpF0DjKsxybK1Bp4QPWV352P2NnJC0skK5IorSSKpLa3UfB3qbg9STj61mnV8PMdAWtI7uihxru35kbe21lhQ2p79rxiCmOeNpkUAK5p7HdW0K/Mtay1Pi5WLXRIm8esAYpkE4fXjx1pFFNa72XKzkPWrsdF8nJSIKN1cj7pslJnuCkQkJNWZ5CijClDryR2iWTQ8nPcOzJztncTcY/JMGjkaePcwY6qW+Sbc8uVhcj60xH3e7FxB6Rf0HE0egSVLEuJ9I7ju3snN2An81QiGNHmEDl848O2gPpCh25T5S8EZJ9usDyF6JvFqydrbL4ym0l3t2GerG48LC/N4nL9EtfkRTccrBENc35Prd18c9PmCvNpK7nju0c/ryTaZCHmuSVuF+bBIQX4IdV1SrYzikubK83z3MUxcXOza8BDLbva6JqouIC9T0I4mwzn35uboePlfFz7u0s8PDYynpvYRTIz5+e3CqxJT6F9r0Tw2Gxg3rLA/DmlfInLEqj1OOrM+cEVrz0MB2Yeo/0SoEWZ/34Ga0hMFkfCHtKyE4ItvOReBL1p3NxrxkZ6rvgYOwc5ggfmBju258jX1HjkadYhny5DHPXMQ3UUrtk2KwgKtusa7CZY3Jq10lEpF+9kywIqupnpoa5mKUpiQkChcZ+PPCAmSMR3gIRYFJ7NfHSJEuKYNuujPei2uXnFYgl8/cp85eUSigrputDAD5AJ8mvN9mnuiQt3Zdbz1W9cJP9oXh+eJHCIxLLTjblg1+lt+Ppcz8x6aHkZ6+7cTFssS2zbCvKKhJ8OBLqkQ3K3aPpkNMWYCmdi3Kx3S+qmkO9t2CJQ55ofhnIYQCbkS7nIUZyR0NVC8jFUpeRreXXbwZFtAqqol+P6gpzrWxlrDrkdhU6bPkvrrSs0a2R7fomzsZmYQklcvjJz8OBgiu7TmqnTwDZ6U3KRXNiGbQUAD44mzK6qjxMwbJd23/zWauZxQdJ1Sw7ifK8AiJDmWhUf5yzzmYW1346ZkXKVBHhOD6//Dj88tlDJdmzHdmzHj2y8FxH3yCdJRUtz5JlXDXd7VQpYNVVZC6afJYmL4NJEgN6nATQ53ahZCuvUdJcZe2+ivDkWQyjOxwOAIBR1tWa5znLddMoF7QKzGxOVS0uzEeh65aMsqEtr47CBblw4mBHtp+UUjfIXQR9SAAnxoB27YmecvLJgUeRSKMnnqtAEYL5VodM1kYV3AC6mlOMN4tfm73jlYXDPRKrrcdNd9mzexZBSx26UYnBKEZaXYXxuIpDe4wJhh4q7szsXRwMXt2QO65To9k0quJhRx5ytGErJShsJdTy66xIsmSeB4rxJda0WRTeqZBoWXx8Yqp0cmkwiarvAr81vXrxo40ATFc4Cxrfm2KN1zlS+oF1gMSEzi0GKFtHGprMIMxLRv0sjtTzNsr/LLyuExD236u5oKaAJybD6FuSIeIKbHIgo+l4kzXxJAUGcdfloBESUEX1/DlzXNNY7qVipoDIqyJ+2DA8cgLpZI39BkF0lUNJP6Dc5nIGJtwaHCVMfp+8CDE/N+9t7GTCmr88tLhrWxgxlbjGdLYgKvL0w98pub431nDK2XgqLpH29vuQCeNfecMY7WwVcgD8dzBASVOO07sgXRJKhB51VHNmquGL5ACdUGHTMsY9ftbjYfTBa8jSpSjD/+ey2gS36iw1apES4t1/PLzCfmvU5XYV3FDslyxtIS7OsA6B4qdbqj/X3uLvUva00Vt8T9Bm7COtsJHFYWjeICizISEUIYyDCx097wNmqzTDOTeaiQ2qQYg3c0Pp8l7iMNvyxsY24t2M7tmM7fmTjvYi4a6bYZSHvkNbBzTiVaqLv29RHl8SWAj/nSMR+t+bIxR460BNqpggcxi2BJtoRAKpvjYhO+rbippDN0mHhnEpJ1nlYrzwWkJEAVvR0LJRkLPs29SHp+PNKsnDUIXVNCWimUq1inwuSk9RDQkW0RWFhlzC+tJLYoLFpe3FhKIg/GY6ZDlhMNVqPiIKUp0jOiFqZuBxp7FcpH8s/vNvHX5bmvFutDDu7JmNJLwFrThFTV3DRMLsWrPdRFBZTn3yyZXs77uFoaCKdYTfGipp15uMQI+rcdI59xnHLRcOvUwka/0nHgs4IWwwcjmb11QKCsEJLaiwvKaMZZtilZp83V3301+a37EgzBryeuxwdWlJxAWidufBrDFYVXLvYP1xy96GmkKe4VaxvIwMLekG4thQAUQPFbptFVPRiY4rdgImmf2GwbJ1VjH3DlhC1H5/WsAn/hxS8btUk44YZ6WmmZ6pKYHlp3m/ZCu0dczzlRmJ1QU0vUnPdZbNxWC+6dUJNLu80F3Ol0OhTV+0i9rnA99Wlg6ejKQDg4NM7GQUawaS7ehqLVdB0qbbvNvdULCBldZ2mphFIaPIMTZeStWG6/Q1mFC3vPNzwZzdnAhFF1o5T4ow6Duexj5D0wZd1gbqyWJ9k0E6YsliWEj6tJSesoCnT0oXmWpGIXMjYrOfkXMDyas0aiRllmd/Nevjz4zEdS8WZ397OCh7RfWeLRsff9wp8uG/qSneb2Q6DlKPvQknMSQzsUSvF4L8gMrWNuLdjO7ZjO35k472IuGuqX2BpLKnpJrA0Nz4EduOC3nJKVgEMooJdriHB9D7YkiMjvcobXYisZOaJjnOsvyPK1Gc+5IGJuvx1huSfDCvhl78/YinIpLQZf98PNmzlNFmFeBubanehBUfXQMMWq+UdI7vA67V5746XwyVNhtCuOHoZuDm3yDtSc6Yx9DI8PDQRkC4pcgXgP7jTGFAW7CSzOGuw4w9GMxwSy+Bi1cJvxjsAgMfpEg8+NoC29E10bYbGZkrt+L5ibDTq5chI6c2hSGSvu+b/9/sV9KX5Dj8qmJ1T3WQcIth9iXJSZ0Z3Ik+A8V2AKIEA5OkA0b80cxacLxB/aSKUPLa4an96MGNsdjBI0X9iIr/Fy6ZJp+8kWFLW4VoVxiQ7ehLMcPrZnH4UkCMTPYk2RUZqgvyW6JBZxc7Vsu0B9XukhL4ya0Z8sMeuwOqbK6bCibYLEVFE7NusFCiCBgctXy9Qraih7Ekbrb8d0n8o6LcmYstfxWy8nK1sZpvYvkJ53cj41tlh4BSMtXrEfnBbGl5NeSslO9pYUvEcjYKUG15uv/FY76TGcwHDjKqVBUM/R/ghZaRdD5rYI/aTRpIWSkM9M41EalHCHhEjZV0yc6j3uEBAGV42k2gdm2MP3RL6BTWrHBdoT8ku76yH11cGo49L897D1pprMf5QIR6b33lxMcQOuWQ5HTADC6DrCUBvCkhy5ml/7qJ4TeymCbhJzrMU3t2aiD9yC6b3TaYRm2WHboE9UhytCsHaP1Jonte9KObsRWuBxx1q/LFLZpj8sfFebNw1BTBTgmGTHU8jo4M/S4AnxNcehQlfFMvRLLvotgV3PxYXGZx9sxBk12U7KJ1X3KlXvI3hj8xnhS2hLk16JA86CH9i0pzH1xOWcpVCMy97t7dGe2AWTqUkHHqQ1KIyABB4BY5JU6VO18tS8kOnrCRSWgi+VcKVzYWq/5zmDtu6AcByYTaK1ska9g5t2LbE5hnRoXLJDvUAcE3dY4PWhoWUsjsO8Z7d8IPtvoTba46/84j4yh0bORUWrVDATswc15ZS7f4d5/ccnC7nGwtRI6MCtaE52gsgSTsbloSgGwaRb2h1ADBZQX5+Sq+HwIyuTVog/JBukl82lmOOo9jfc/nWZaqm7VSIVx5fg8OB+Z6LaYddtA+LJVRKhTQfKF6Zm82qfR0PQgR7BN8tMi5u60oBU/IObPsQD2kxrRLouZnrapIx3766SqFyEqV61GLNE4Quw0L2fh92ze9uR41Od7cN8fg+AMA7/h7q356ZOb610R0oPk7nyFzM9c8LTtkHw5ipa4t3NfWzxPATuk63Cutr83r/oCk8pqnDRcLrWZvtvColmWrn2BUcgiSSjQu1NOtQDoOmMzTwoN+ah7AIXchHpttU9ldQN2b+isTi7+9WOZwD6hVIKqYM2iObexHKFeAOSPxtvkFX0+ZKhX/PLZkCuXzlYLUxa2CRu/j06Ziu792AwW8gu8CBrKmBSQGLuivtWCGgHozD1pr7AzaFzXOd39Hizkob67mZV88vcUYb/XgToEsQZmCX7EdZKYk1iV7llWSp5p/hh8cWKtmO7diO7fiRjfci4v5uZZ4fxyGwT0WKVAlE9IS9HwnseeS83tlwt5+wwdoKKDXKK4oE9l2jSgYAns30LJ0UUKu6SAZ4J/Rkv9ggfkdNQJ8vOVoYfTrDu5+b39oJN+hGTaNCPaTQ3DzQP9pwEWlyE7HgvKdrhxXBxctCNQpui9xFm57mubJwuakLn4Ij8X6QGsF3GJlU0aMuw8kG/gmlYX0f5bmJ9h61JnhEh7mZN8L6kZ+zE/v+/RX8x40qmgyoGBZK0xkIQC1SrC7IsWYvZ3fyuungdhyxPOZOESOszX/DEvZDomSOOsDlnI434TYx4dtG5wMA+q0GW/rkAbCgjoXfv24acyIX1gcGPtixppj+sjGB6O+Y846XLr75xkS/D48nCFvmeBazgN9z6k8hiQIXrz10/FodzzEaGgCqW2qocfIGHhkGEANKI1p+c7y9NnBrzk/frhv9lUwDVHB1PtttVAMDH4gJUsvLJrLeZEBBzWJnt9AFwSz3R8AuwSb39hH8hYlUy3iO6dcksysTyIiUEAcJGxlsYpeVEFs75pzShc2SqnYEdsjJ1yVnTFXVZG6BU3DTiNaCC59aC1wsyNjXzZHdUNFylEMQ9AilmKaLTQFBzXBqkaKkOS5zl6Ni+9sKo48J0hloTF6Ydd7dpNAEOd6cRRgdm3vB9UqsSR+nFVEWXElMqLntMm4MeXf8lE0dgrbD+4Lwbei60PxgF4Iaq/TrCWTbzG/wSGHfM7+5uGgglttFxHo4vWjDblTPpn3EdTPeaI69rvnsNPMYfk1LCzF1VIZ2iR2fzM6dEn7a/MYPjW3EvR3bsR3b8SMb70XE/RHpDnhSY0wgfl4By0aXDalnnjHXsxZOWqagFuw3mh1wLVgU6Yhh1LS8R15jGSUTprnZbWMbBRiNBotaW5e/ytGi4hYqcDEjDHL0jgiP2gjGdTeZg3VuopuonbFlWnuTYTwlbW56IvdbCdMBCyW5oaDn5ZjQE3ZR2LghvLbSYO3e09yBs0dWT77NLeGy7ZrzpWHVGi0y4yadKqvYXslxSizpeKUHLtZaroVqQhhl12NqXnGWon1ovnJza8GmxopalbFSkrOI6TLEOjbncfR4yRimLJVRzQMgfQd6RRFNUkDUTSx52TSrLFbQz875nIrXhCXbCeyHVDzsB3BIl/ryqoPDQ4Nft/sZe2eWhcV+gEoLxEsqgO2n2O01rfuqZrsd2pyp2RRlq0UGHVPTRlhygRFZYbIEAHgz5nOC0pzVOQ9bEI/JmNK1gYQKCm+vOYtAXqGaUaQ4a5pS7LaA1ad1cP49rGPCiT97BHxk8O5okWL2f5GK4/MKXpeoZZnL2V68cbHzEZkXkEtJ9nuNy6+IcncUs5lGvUYA41VZFM39VxfmAOOrCQBZabHpQTfaNESBUkGvKYK1JeSpyRb09ZIjW+QVJN3T0TDHgW2uX5FbuH1m1lDYzhHROSUzh30eZ5sAHrWlR90cmvp0FitzXLdJwPftbrDhxKhuvgEAKN3UV2yrKYxfzUzdAYAcRUxyKN5mWF+btXEzbyGhe7rjZfy9ZWkx9i3R6I+MZy10iMJ8ECV8bZLSxjw1v3WTebhHtmqBU2IQNiSHHxrvxca9Q7oJSSVxVnsQWgKnETFJbMX6HRHxHgHjCiKpa80ZWRAkF6rHK74BBcCdbWqRM3PCPwA2F40vX82AuLruoE2LKwgL9Hrms/N5iFZMUExL4/rCLPyLdQu7NOFvL/roULX7rmgMc7tzm91wzpIQ7bqoISTWtKGHlkKboJi2UyGvhaUshfgFOep8Uv4BJ1i9pVZHITitz681chK/mU0jLMiUdNSO2YgiHtuQrmFDWH3bFHIBwLdRnZuNrdqADSFm05DPhaVc7QplTp18nYTT6DIRsIi7rRYFrL2mc7Le2OQwANK8nqQ7cMOKdWSQlxC2md/L34U4EGTUcC9C+3Nz41lfLnDxzhR/Qj9ns948tbGamfdEYc76Hetbl82F89RmM2Q1yaCvyF3pQ7MLibxqNu6dqIFHbAvqmyvznq7PEIuapY3ca79l4A8A+tk58tdkQPCk3TwAnAqSiuoyso0xMAzrieWIrQzpb8w5+cW3EKeGFST7AXZPDdMoX0m8+MZskA8fTNEi4axoniG/Je70suZzCzi35kZb3XjonZpj3I3WuHhhIA7LUrzBdNop8nlEl8PiNVxWkvnSnl+yIJN9KDloQOgCdN8Ix0L5jtbVSjGrRCsFn8TfoqDE8q2ZmyKzGOIrCgvXS2JvKYn5OuDfrWGTzcJc39sk4GDKtSpouofeLNvsA+l/P4N3R0K3hkerq3XTaR3YkHVgMyxw9WszN0lpN70bScDF/kJJDAOzfn56cI2CoBKtgbcL89lKC+Zoaw3cErSy45bNvWVVuFiZAOUD/PDYQiXbsR3bsR0/svFeRNy1D6MUQB30SaFBjVVYlxJ91zyNbuMQ3tw8nTujlDUGtAJAkZzOdfNEKtecnulUsRJcdgPcjs0T/PjPVnAemSfio/srXPyC6Fle42U4tBuluXCVc0pUKIFZWru1l7iOzXs8WUER1FNzMvPSYnjEERrXxJlVuulC2wsSDCiFvEl9jsq70QbesOET1wG9WuWsuVAtFCrKsNYTj1PHqzjEDdmVzTMXT6gjLtorkROF2ZVlI0yn7uo4AG/emELebyY9PCb+ekjHtcxdLO5wV1uk1KYqAXuvVmuUDO2IyOFsQbR9o/kBGChhteHXa0lT0Q3gfGIU9g6dGZbPyTBhs4ZzSFHVjkb6Pb0uNFtuOWmF6RVdDz9DSHM/XwYYOuY8bKdCck383KxiX0a7Lla1XOhJrd5XNfxu2UAAKEqoMdEIf3YP2DcRMW5n0K9Ml6rYa8Ojjko9ixvZYVs2+ie+A9Q+jEnBWaPsB5AXxN3+eg3rLR37wza8j80aLn+9xoA6ICE1Z3zhgcLyrfmemira360QGLQFVVYwJVQ64E7a9cJDZ2CuZb6xcEnrOqskRrUKn581EXe7hEP1SBQVG06IomKKpb3j8rqSgYDw6hVXYjWmzGiQwQ3MRXh33m+6NKWCX98Ldsn339m4i1PPZJw1LDYMUlyRuUJoC/QpCu4GKcZkImJF5toCABzLFM1hbOjqojBKxa/rtML9E+qj0ALn5ybD+/2sxxBHz8sZFt31S85Y0szBSddkuYuNzzBLpQXrGDlSsZ5JpSTuytn80HgvNu550WjSLimjLhTg12pL0HWhHl0/Y86usNHgvq5sFkXbYlnJapyiWpHnXgp4I9pEV4J1t53jEKJHoklKo0PiT1cX9Uo0KfjZ0qQvzlpxU81+uGEPx8tNgJggj65TIqKFVvvcuVaFtLT4tTmlUpkS2CXWTFrZfNEcqbAhBsjFrIPOvMYNK8i8dmjRsPqEybUB/ZrwTKnRJhbMLPXQJRyuUBJL8u4Lb3NEx8TX7jtNw8xtDtml5ohNwfjmvTBlGcr6nFeFg5j+/6ubIUaE9Y6iBBFtxNZhG4LS6+JtCueYmh0mjccjFilQV/lLBfFnD8zrnguMDb5rP1Twr81N+s1vdhB+TdrmPY0uyeV+dzPA0Sbma1azFSbrkCUGIqdgTnxvkMAOqMZy2MBtzD6alCxDe9crEp7dwAHdEPLvnjbnck4KT7M1RL9pKqoZNGKvMRHGIuZagFoUDd/90Gc4Ra8yDlDOn3cQkLvNrrOG/dB8V+u/GcH/xjTpTL92eRPzXYXWrpmn69dm09JfAtGAcPgWsLoinHXewqBtNiFxt519EWDomes63gS82XhuCY9gzvk4xMAh2YFSQxAjrBon7Eals4o9RoVrcX9F+BML6tdmrbx+NcRu32z0B6MFFgszf+vM5XU42fjcAzEvHMQvSbhN1vIGmgOVx17GvROr1MM++Tpm1wIqJcNtH6wdL3ybaz0qrvj19Fzh7MwEMN1o0/Dk3YL9JI1AnJm38+su92NMUp+dr86TgIMxpQXf62ll8fGnpc2B0R8bW6hkO7ZjO7bjJtGmZAAAIABJREFURza2G/d2bMd2bMePbLwXUEnd2l7pxg3HEo1WSaEF6n+Ffs7NH6oAFr+vK+8V2k/pE1KwPkl5USK+oVTKq7B+Y95ze9vCwQmxD16XkJeU1o88dH5KqfFvVvh3354AAHa9jNOzVEmEBH8MgopTnE1lYUwwTlJJdGu9b0qH7kcJ6xpYUjNjJFMSLn1Hx82ZrrcobE634tLBampSsrZM4datzl0bFbE3rI7NBshKCYaUCiWR0W8N3Lxp0U1tRDDHo9YlhH2HClOb3OZg7WNb6gY26Zp09mYd4paq/beZw/WKvVaM9Tekp3Izgz2sJfYARSwNkTXt/CJyIAmuwsMjNsrF1S3rdMCS8D8yqfPxfI7zawMTyEWAmHD2oZ/ijPRgOlmOk5EB8fPcxpSYCP1Wgg3RsBazECOyNCvGFRyaG01t8JuxRERWaBK4YzztAjXkcbALkD47Xl9CvTZYKGPXALIvxgx3uI8iiH36bDeCJEszGXiATxQ1z2W8W5yP4RBmfIwlNmShd/6rEIeFOT/nrw5h//U9AMBOf4zpP5r5u/yqg6Mjs87rBqQ3bwfoZAYO2HsSo/eAJALOlkg3zZag72CuIbVpH0qNC8KP792RY5BSMftIeBY3t2DVGEKXswqS5tU+aNTz1CKDt2PW2Gi5ZoXJwM/hEnvJKjRyWlt9P8OarvfIT1k3qIbFLtcRDgg6W+YuntcQp9Q4pXtxv7dmVksxLlmFMH8VQ3rEjFoBNjVHOS2w28980oNH5973MjbQnucujsi6LC0tTLKmiSatCILTaCBUu8K7DUGOpYUrkqgwlGBzrp/jh8d7sXGHVACUopFv7UpwcbLSwKU2JzJetLgbzu0CDrktlytAxbTxJDnKJelL5GbDBowlWe3Ofm9nAdTSl2PBOKecp/COzbS07iv87dpQvv7pYg8ZUZN8S2FFBYbZxueLeBikCCzSG1ASY9JLyOl3ChVhhyRbKy0wThuBqJg2qknuMoya3ilSlMrFfdqIvbUNd0A4WVKhWtC5bgourikl8KtrUyQbeTn2SBPBurP5rmctuGdEwzqukBJ12vI01LU5pzKT8AgfF0JjQoXYDi1iRyp4VPSTduNaXVQW/BGdiAKyC7oBPSB9Q3PtaPgfED4/DIE96rS8mUE/N/Ne3aSIX5uXvb6C99RsyoN/YSH62mzoX/xmH3stc/O2WxkOyCdqtgjZdCNqZ9iQlsXNosU3e5bbbEbguBVC2oDdfXIvf2S4yADMg62+IP2o6Wa8nQFXZrMGwMJS1dmSDSG8//rEdFjWo34iyztJb5bf6ZwcM41OjVcsgWoPLdhrWs8LjdtvzOaw2x7D+mTffOWTPQzKSwDA8t83HP7armzUW+NmbuYx/crB4X2z+fvdEuva79GqkJI1XOAVvCkqbQIpAJitQw5EAq8AaB1k32dwdshLNLKhSGbX2XO4DqXjnDVwrl81D7goyhjDn61DtvrrtxIsCbe+3QS4pWt5GKSsUVKPXEmWSfak4mBixyuZT11lgKR5tEIBtaIAJgOETfZju5JFv4qlQEbfGd/R+wnsErsd8+BvxQEmJIt70GrqN98uW819oQSW1GlsSY0BBQWRVTGufZ05eLn+0yJTW6hkO7ZjO7bjRzbei4i7Hq7UkASJ3KTN67kCnnaoOy/zkL41DQ6n8QyD+2TndWxxeqsyDUUsM1WIWmUTZS7ZXEAIzd2PtquYNpZkDvpjA5v0Tkr0j8zfT5MFrpOQjrNisvz365BVA4/DDUe2lRa4Tw1EterXu8THtE7xvJwr97PcwhXptRwEwI7bPJ1Tita7TsU2XH6/RE4BXr62YJG6Yho7rDExW4d4nZjfatsV2nTecWHjgpTxPKnQI0U3ZxpzI830OmAKo+tUrNECgOfvjBoEvo/9O5K0JQbUIOW5JTe2yEAieU5pd9koGLb2mmYqRD5Hreqy0YuxAgfewqT6r5/1cKpMdOg9CeGemijwk9U1/ulb09751JlicNgI/78amyh+J9sgouit0oI1JQD2QEA4LNgUN3ndHJpF18MKKvh/TSqAngs8M2/Scc4NOHq2YbaE9cEQ2KeoXCtgQ/N4NkH+bM7fL5wmukrOCG6QDbOjSC0EQ6I4DiSCB+bYB4ixuDbX8lf/5xCfrUzK5Hy2C3lq4Jd7n43x4hfUTCQbpkjN3Fitfbx6aY7x/vGM4ZTpdYSrufnurp/xNf521UKNqF0nAXapw2+9ctHumfNze4rZMWqTo6LgU18rWBRcC9tEsYAx9Kgj/aBXIpAm8nQnFV5emvNwpOLMdlPZzGy52PgMSdRR8Mhv1utFEvLx3mY203vzpcRmSlnVQcPgcA9tVDNqCHwN3I7NOl+lHt/zR62YO6C/nPbwszvSrHUm7qeNYcJBkGOX5J5vNj4zXq4zBwGjDZrpgF2nwjL/0zH1e7FxT3JarBAguBZ5BRwRDNZzNZLa4VxJHAcmNZmtQ0x/Z1bC4e4C4U7dCQUWk5lNQ4YGFqnH/NbNrc1tpWGYszzlZh3iLfkwfpjOsHNkfmt3uMI7EoTxrRIuUYwOg8apotISSS01KhVGhHfVF2RVSu6KjMtm8RW6oT5uSo1bYlS3bYVl2WD+05k5V9tZwm0R9uZXjEXaToWLqdmovlm08ZggpXlhY5oTZz1MsUOdW0oL7kw7e9Nr5F7dEnUy5nklb2xJ4rKjdkQp4ZO2YhePeebiNeGfkVMAX5vPdfsbFj0aL1p4+tRwm60AjfznIjFdawBk5EBdGN6rWjdt4O0gw2ZKfqP/kKPzkTmw8KHAT+NrAMBkEUJekfSBAEMor+Y9PJRms9zpx0gSczzLxGc96nRmwyad8SwmAbJSYvRT6vSMbMCjtHy6YuxWOBYLKYl+AHlKnZNRwN2g1W/PkZ2RsFMbEISjqo02CwBGajQ4oJQ9Aaw2QXOoEL8lF5dYI7hPtYMjCSHMGp6tQ3z/K4Ob31uN4f+EcN1Pd/Bh3zz4Vl+ZDSlLbIYPB14Mmzbo8VUH956Y4919sEaX5Gnjlcdt7k/7C5zdEUa6qxtdu+qUeYqgR92gDuDum2uWXVTIJkQBPAVa/4zkC6RA+IX53eWli+598kf9oMRjaeCwVxcD1qnfCxPs0bKJCwd92qjrbkmtwdTPVEn4dK86UuP11MzRJ1HGm/jyzEXnmLp8VxXjENLSGI7M+oniHDPqiygrC8d75nhH3TXeTOpef/C9FRc2B2n3og0/hKUAw6WXG5f3i0K5fJyLwsKOv4VKtmM7tmM7/n81thv3dmzHdmzHj2y8F1BJnRQ4soEMIhts7CkAhhhCq8KY2m/3ohgH++SOYmnuAMtzizFM16lYL/oPflMAuycmNXcGAqkpwmOQ29gQTvXL8Q4eUzV92E4wotbZSeqhJ6izTjfdUt+vfa6473kFfjcx+NwOdQ0+aac4J/rPupT81FQaiKnaftpRCOlcbaGwLMhFQ2q4VHVeL31YtUqfpTjtV1ow3CGgcUytuFllsaMGAO7i0lrA82rGCPCKDFhPOiuu7EejnHHfF+MhdkqTmt/bMen3ZuOgICjKKWzsU9V+XbjIlyQLUFicKh4Nl4x9OychtxerWcrQQ/4q4XZ26yCER6nwQG74GqeZg/JLk3IOf1Zh+DekE/52ji9+bdgVbbfAyT3TaXnXhFYrgb0nJgXeiWPcnlM7d24DJBrY7VNXrTISCmYyjMIdAIh+CE3Hq65jZo+Ig16jr/39JbJf3fC8u4dkRJwro9VNo5YdqMYaHrEEhQvUXnnZdUPNW049XJ6bNbl/uGQXp4cfTnDzxkAYZ8+7OAaJUv1MQj4yb+oOzTmv/9OCLeiCXom9AUGPZz42E2KSDEtWuqzKApK6bVt+jseOOeDFxsMl1X3aTglNtRPHqdh1J2gX0Npc4/W1i+49oh4+GjaWZgB8UuRTP7/G6sysVb9TINwx63N3HePtrOlkru2/el6GTouwdTJ4fj7tMVRiCY2AIIg9P2UWjNYCgnSxo2GOihDPxTuHzYWL0sLeI+rCPcmBZ+Y9t4uIz6+7m+KRY2ozvzsfoUuwYc/LmdkSOQW6pNkfFw7vL76lcL4hMTwBfE5rbuQB86KFPzXei4271LWWxx9u1rWlWWgrTMl5fVXaGLG6lmBTBacvYC0Ii5zaiKiQZzmaaWA7YcmC8If358wJ1qscWULYW5jjxDKL/qQPfHdrNt+rJMQR4aV/dnTNNLN00uMCzdBP8YI4zavSQptw87pNXGmDcwOAI4AV4de5AmpZj4FbspTromouz16w4cKRP1Rc2FlOGtw+Ly1uRS+0YE2Eae4yJaqvJCKi9x3vLFCQ3MA6c/HxgWktDzs5/CPzncK1AHqofFKN8fodPYyoYNceZVjdmI3ky0kfg5rv214hCMzfd2Uto0EG74RaoNMS5VsqJEqw27n318Om2He7RjE2x2u3BXqRucbxWYXvz82x+C8miJ4SPv8kwl+0Tcv56qWFiqRJo+MKrfukInchmDZpRcDovrmu+VJivSBJ0RPCnTs2KlLV01kFedrcULVfpvXnJ0zj069uUJyZc5KRBfeUHhalYpqgtRc2WiVSwCF/xuxNjuSWsPVCor1Phd5dgYCKevK7DN8Tpjp56eNjZbB9p6Uw2CU+88zD7Ixom5slor+l492r2+MtuL812PHqnQ03MMcyfJRCUV0vnUneuFUl0SMdFD8sWAJZX4O1Snb6MQq6Rz2/RLRPEyzRqP0VFtyPafO9vw9MzX2mrxYQx+Zahv/tIZwvTBQ1/jLE4NjM5cGnCQakaPj61ZA34EpJxp5bFGw8HU2Z8z1eRzgZLPi4xhODq2slUNJ7RMI2ochzm4O+44+W8D4zx6XzCiOSyOhdTHH5bYvPqX9gXv/Mu8SLd4aCawmNhx1TAHatiutcWoPVRAUCJPSQ2FQSkW3WwW5vzXvKHxtbqGQ7tmM7tuNHNt6LiPvzfkPf4W7CqlHOKhXwlMRvTrpLjjBb7RSbuTmFzRxcJbZsBbdTUwMFV5srJfDxRyZCCX4SMW0re75BQRXgdeJhNKInZUvhER3Dl+Mdfsp3DnJmWszWIQtHjbox2hRZjtcRdujJWkM13y1b+AtysI6cgsn6L9aNcE/HKbBP6miv4wjvEvPd08xDcGkipkMsWBBqvGjz7x+019gjJstelGCRNk0TNe2o7ZRseqqUgKqFl9ycbauijx3W9U6fbWBT30h4JLAfm+P/6u0eAODp/i0L3B8EGeumtzspO6x30wzDfUo5nzS2cjrO2WlbtFyIXYrGVhsUzwzEkV4By6mJqIKwQOcxfedPJT7ZNzDE8q2L4jeUOv8sh/PIRKT9owLxr02UvXztoH1ozs9paSzOzO92DzO45CRuL0pYTi0wRCwAX8E+MuGu6IfGQKGe1CFNzO2CWSVaaciI1C67LjQ17FQ3GTeFOUnFjSj5rYZFOlRWBLQiyi6GDrQidcDIAcgN3p8t0DmjrKNwmMnRUwn8PkECH1ZQZBLy/a+66C/Nmtj5lwSjHQ/g/h1Zgn1xibe/oKamIma2kqokJDFs3KBE3SpW5hIp0UzXmYd3pLZ3eLjk+S2WAvEVQZWtkjXa9z7PIe6RsYRW0FcmEi5erODUZr0P9uB8bqide7hAdil4XsOPzTF/9GCB5Lm5F7PEht8xv6vLGk7yEZAdnZVouJQdtnYLaEVdsoXkte94FSLKsNqtHNWMdNM/GwEBQTjPxpAjM9fe/T7uPzT3wfrXKWecvZMMn/bN/rK48ODUNFJHYT4xn91vx7x32NLHEemmLwqbDRlcrwTu6K/90HgvNm6mxammtd2RGg4aHLDmRwph4AzAOJzUm4OqBOKYIJGjmG+GYi1wNFzwZ2uPRUiB7Hdm8idvQ8i6bb2wMbmlhaxjxjrvrWP88sbwXV27Yprg6cMJzt6YjeL59YCF1PNKsprggDbzJ5014+FGtpZ45AKYUUof2uUf4NQdSksXhY3HBEPcXrcwo01/VTj4YGgwx25/g4JaZbsfNNzU+88d/Oc35oa53DTdmloLbmUeHa8RfGS+U7Q9KKLmWQFw+bWZj+HhGp0Dcy6PK5Nq/+Z8l3nsHSdnruvZuIf7Rwb7O/xpzApxouU2NLq2d+dvH/qC2tNfxbDa9ED52IZ3a67B4pWL6dfm9Z2/BtynZqPvR2u8/YXZQK7/nY1Hf2m+x37YQfgpwWG/SXH2zLzfc0uEBLmUKVCj/8ITLCVQ0/Ws017TLZkWjXONFCj/41sAgEo1nAOHz8kamnnUeQUQhm+NPFhD+h6lUS0aJyZd09lVo1KXnxdcC3C7zfWoNsDB0Kzb23nE94KQDR9dpxWsHXOdT386x+vfmvV59r+YY/zJP38H+6lZy/ajHu4V5iH59os28jFJGpzO+BrgpqH6hWEOizDj0M0xILd6t6sgicLmQCOlzj9dCgx/Sg+j0y7QJh7f1ZTdZVQKnP9v5j2Hf/8a8oE5NuezA9hEu5v+pxxd2uSsoQ9v12yu1XnFG3bd8dg/2EDR/bSTJrglE4igl7NkrOM1dYPWxxZErURqS7in1MG70wGuze/H3ynIl2beW//Chjg274nSErOfU/B4U/ADoHuY4eqluW/2Hq6xPzLXL76wkVFH9ak7w9Xc7BGz3G5a6uchyzZ8iB8eW6hkO7ZjO7bjRzbei4j7kLqKpNA4oyr1q9jFiFINBbCedJI7nHZ7u0CVkE5HAXRrzec1kI+pgaSnAJBl08cWBFWv09+v8YsvTEo28DO0qOAZuQXWtSbCuwBPT006/sGjW/QuTITwzaSPT6koM3q6wYOOeSoP3iT45tIUJ3y7YnGmGpoAHFxR5T2trLrmh5FXwJH0tM1d9Kgy7UiN+2SxdRikfN7S09jLzNO/KpomGmlpdB8SS8QVUMQ86Rzn+BttCj6vb/qsibC/u0SLmjnso6jRf45zFrmvsoojuVcvh3j4wETRO4/NXPz97juM3zXdZaNu4+VYF7egBetSC9eCmpFJAozBAGC6JcszEvrqWqzNUS0KLloG0xzvXptIx/nNEiExiqy2xPGnZj7OftfGuy+o2ahawDmhaOu+gHNLTVaZwxE3lEA5pm5a22i5A4D9E8NMgWsDE/PdatFAesXrTVPgbAn2qtNpieqaNNHbFuQuVRU3BUoSMrOPQjhkUQalOaIvXi2w+I7YGD0F1PaML9YMjTldgeGHVGieLDB7ZtbN63cDZu7sD1bo3qdjcAVOPzPrc/3WvHf8Ww9tEu5qfWxBRtS1e7rE+rbJyGpRLFUJfEeNXZ+6N6wVVK0CbkSTPmAfUporBYaPzJ/VRME6pHP98AQoadKygudM2I2Oyvg/W+i9NTo1wZ/3IEbms71PJ7j6BcFbuwmcO6SL2uCkzoIBMGtGCo2IipZVJpGszPkNHqassy4kWDtGZ5XRoaFRvjDZiBAal2cmYzv4DzMEHxC7yGuYJ9PXPjbEPGkPMj6n5aUHL2is+8I2QV1Ll7uqQ0tx9ttupThWfzqm3kbc27Ed27EdP7IhtNb/5Xf9fzxe/uv/XgMmYvuaTDUDq8IDot9VWuCSJA+PwwRHuwazbh8W3A6tFeDukIpb2ERsQoIjT+dxF9W1+c53P/cQUdTVPswbw9gMTKX6eryDfaLl3D+ZQtZ487WP1xMTgXywO0X/EUU3nsTyObnRT1rYoyLnhrD315M+dii7sKTiCGlTOIjv8KwjitC1Frim897xN3hwf8rnVA//SEAOCbfPK44c0pcZfvuViRo/fnDNSn3xhY3fvjGFxc8eXqLzl+azctSEMHoSQy1I12NVYf6SHH4mHebH1kVe79iCSsx3T791+Zy693LGSO2nw6ZVHGCLMvgOW5QVL1as2bG5ELCJ5unuSuY8p1fAem6OdxH7mNLcfHRyzXZl2UJiNTfXr6okU+Sko6EpM9ksHZREuewfp3BPqHg9LRgblk8p4k4LqDOKulwLyRdrvgaSpn157jbSqPseF1/VZINyRrozy8Z9yfn8wGizAEBRQo8X9Hcjk6oWWeNItCyRvCXXma6CPTT/IdtOEynGBaa/NVHmL9/u45NdE1H3hhu4XcpeaNq/ez7iQnsv2mB4Spz1AweauNCr54LndHnpstQqAOb4A8B0ZaLTfivB4b+iA/ZsoKJw3ZIQLZqowyEwo0zx2ZgznSoDJKn4rq5cvLsxRfij4RLDT6jIOPJRkYDR+pWRJAaAsFswle/60mRv7VbKdN3zeYfvp3aQMX/fOfQgKQvUWdkYBJ/usKSwfneL4gVJL6RASbj9q+8HfP69KGW5jEXsMxFBQmNBpAhbahy2zXl7XslFy3Tj4KuxydCf9OdswK2VgEMIw9H/+j//YO/7e7Fx/x9/8z9qwGzWr+PG6uk0Mguq1BIFVab/5sk5PKqex2Mb7fsEDTgC9h6l45ELkDegOps33nKlwuarJp3yH9NiVBoV+QpafYf5tstnwMszUyiRQuP0wNzA0WHJBaXZmc9V891PUgiXdEmeC8QrEs6hAtImcVkec9iNmei/TjzcEERUKsGCUI5UzMXejWIcfEibhmv8+wBAdLxm0Q0jIKUC5v++YoGellvg3rFJl92uQjxuELLhvyEmRztoJEXTRkO5eL7A1/9o5sBzSi6K1jfDwXCF7ikVyGyB+Yta+7xkloP/z3abjWqyYgsvNV5h/HPzstZmEwUA78RpYJtSQRE/X7iSi0ib5zluzkiatLT5homcglk2V0nI5e2em+OUtLm9oMB6SVKuTtVsDnsBxAmJQtXejxdzdpxX5wvMfkXQkqv4Yeg+akEQ60MtUqgJPchbFkCa7KLtQXSoaLlMUV3U11JCral4d68FUWu43za0gmpW4iXpkDhSwSeti1Y3Q/tT+v5ByBt/+S7G7DtifsQev7/2aVSVYGaW1sCU4LtHpxOERn4eKlNcKK1WCme/N+vEthRWxFZqeRkHP0oJDD6q7yGXOe5i1DIbNgDEG+gzCj4sydo0628UF9XDfsFMsavbDsu67j+N4Zw0Gt6bZySGdRaiN6J9gvTAr6477E/puiXW1DwUBRkO/p601Q/a7CcpbGnuHcDcB3QPqbMZS/pm38UoqeUgmbmYL82xTDY+y7Q+7C6xs2POKVm7eHlrgrtKCyyo8Phxf4FB31zbJHExWZvv6fopM2FWsc+EjY//7f/0gxv3FirZju3Yju34kY33ojjZp6LiunDQdpp0cU5PskIJ7kLUCgxrBL2SObP2Sce4ZAOAYzPfVrgWRzHFmxVK6pBsf96kftmLBMtzE+FFgxzeIck93tc4yUyU9vqmj+fnJnL40Lphnnh7J8UF0QHffWHj6CnxtI+BmAoVG2pJD6OcU/R17HGk0w9S7JMLy5tFB+eksCfudF85TqOS550G7DAiBhHQonPJS+ibNX/2wxNTWN3ELr58YeCRT+5fI9qr3RbAUYdUmnntcG2oG5O+T79zkZEqmyg1O2bXsgPfvOrgr2LzOwcfrtE+JD5zBfh/S/3YlmQ4QHR8YE2O5S9TSMIbhp8UsO5R9F8qFK/NeSxeuchItnN4ksD/xETZ0X/VRkAmu5N/yPH8asjXs6ZTDrwMo5Y5v/E6wvnEfP/D4wn2PjFRmi41mxmIyAVic36aTI/FYY9fS1/mCKhL2zuxIbuUHSoNdb3m7xNhE2WzYcKmgK7hg1LBPqYOvkpBWCQFPE6QX1HUnEhYLsF9NhBS1JxkDiyCAd6d9+FRV+nByQLRX5nzc362i52eiWzz/1jhZkFqfpTJWVJx4bzVzliy4MsXe/goNdey+wQQPnWGKmBIkeR8GmJ/x8AHs3mIODZre/d0DVXLtxYZ3A9porohZy9IcgivzrbW2HxPJii5jdm8liPYIBqZNfToaIpiTvKtv29hZ26OwT+QcHokO3xd4svvDKxVwxGhVyChbukgKHBADkBFaqFakAlFK4M8oPUW+UBQw40l1NsxvceFnqd0PYDJhZnHbj/BYZ8ICUsHX70zapBZaXPmIIRmavMkd9kgeJ07iAjmSTIH3VoGWkl2rAKA5wQZf4wfHtuIezu2Yzu240c23guM+9t/9T9oAJhnHt7GtVmBgkOFLkto7FMX4tPPJ02TwrXiRhvvswFjp/pmzY0ScidivDB+ViK4TxGQAltiXY/brNl9to4wJGrgx59c85O9mGr87pmJWteFg4D0Rz48voXbMn9PLlpMJbx/fwY7pOIjaUhfjLvY6ZpjKQqLPRArLTlKzCvJugauVfFxPX1yDf+gLkrZLJSPooJOKMqNc8bk8hcx6ynEuYvv6Ak+8jL0qSFopxuje2j+dk89SKKo6UWK7DtznNevIpzNzWfPEp+9MWvhrOKOTdST3gJHD0gX4tiC3DffJ6RoIk8AOXVF3nwTYPdjE3HYhwGqmTmW2VcW/BZdP0djPTHRUFlK9htt/dTl2oUax1yY+/W7PS6g3oti9EjcJy8tXKybAuzTfaPL0j3MuDgpj7rQ1DWn8zoyc1G+MZFc/FqgRTQ3GVrI3zXFNf+EIq3Igd6QbknZUP3yK8VhkrsjYY0oSwocLuQJz+Y1jKipXeirBdb/ZK5HunawXJnP1p2xgBEY++hDUzD2T+2mSP2q+H/a+5LmSNLzvCf3zKrK2gs70OgV083ZSYqUSMqUGLIVti9e/oF9cfjki0+O8MU+OGxHWBdb4aN9cPhkRdhhRShkiZI4NDkcDjmc6enpnm4AjR0FFGqv3L/Ph+/NNzHS2OFjYyLfy2DQhazML7/lXZ73eXC8Tw1iNAdMTfI7u1+fou0TnG1a5ZpKt7KAX6OxXo55nQ13HZwP1TimQucC3GvLA4a5OQ0B+466R32jyQXo9MWE7zedSia0MiyB8zN1bz886+Ldlvrcvc1L5iJaDG08P1NzPhYG3r6r4K12XSCeECCARA9eTny0aQ2nUmf4a70SokbPZFcyVB7Re9/uFOM+mDLsUwYp5p8O9J4SAAAgAElEQVSpeTC8qOBT4i2KhIa3e6r4W2+EWFDU8flli3Vp207EY3MeWRhRlPT11owjg1nkYEod28vXpM5MU+DDM1W0/Ds/+xdljru00kor7atgr4TH/d73/qkEAFMXDEhPhYaten4y2ax+8e67p7B6BNcRgNErQP85o5xm6iq3BgDTAPMfqXyfs1Ew3Z38wsOQPN6N5REEITyennbYY+t5Iba36W9XARJEx2TPwJ8+V4rad2ozbFFbrmEKTEhNZBw42N5SnmUObZueO9i/UN6PoUluD2+4EanOABfTKi4IubFemzMdZKUaw79LCirLHnNXaK5ZMM1ds+DDEf7LX9xV17cyrhHk1wWArc0hKvcpj7lRV7AoANnpHBnB2NI5cHmsPKyzaRUj8iJyj61iZPAJuhRlOjZ8lYdcXp0y6sLc8BhlIGcx4gOCzrV0zgcnpwkENZxYLY25PEQI7D9VY7aILc7TSqlhc4tQIstAMlKf7+/XMCSY4CiyuU7SsWMWcDV0gXPi2LizNEStQ4pAiQbvNkU1DfLmxzHnnQHAJPBBPAJHQEbPZU4SGaSMyJECnMtGp6bkzgCF3skjEMMAMrq+kAUlrJAMm0w+GyKbFo1mF4fqfcwiGxEhaMLMxIyQC6YucZdysDU/xIw89BqpF43HHg4nysNdpAZ7p5YusEF0vXFsYnZNzHplSXnBpicwv1Ke/m6/aOZqeCFL6zUfZtBbxJMziPhdQifBZQAiEEyDkA0T9J+o9fr5ZYuVYxxdsiL6hj9jKterhcdr59bKEE6VGvUI7nnVr/D7BYBLqpGsegFTTty7f4nKrxGHbiYUrTAAzdI5YorPMmgEU9RtILxQ7/vxiyVeB0075mh5nlqMfpulRiGirUl4BL/sRw461GDnW0VTziSxsUZ1rpob44zqEt/8s3/7pR73K1GczMMLU2iIaAMVKEJBSxdo0OS63Kugeql+rr+hM2wMtskwHrSqXHDKTqbw7hUY4uGHajBHc5e5NOpvGhAL4v0dB4ypnicmPn2uCg935wPU7xDJ1O0Mv1vdBwCcHtXx033FA7JZnbMm4zyx8MvPVdFkgw6gdmeOVeos7E+qfEAIqaHeVhPHq8SYH6mi3sG0hgc0cZu9AEadCpJrTSDI2+oyIMfJeg7kiTosTp7XcY+oOMPMwJzGeM0LuJPTqkloHhWLBguIKYX+4wxGizrPqgJdIqmuzwOmUn0yVhvS4cLCDm0IW/6MF5RVEzCXChgmHwrDFDqdqSISiA5zvT6dMeDiUiIN1Ni4LcF4+MPTJp4M1Sbu6gKHj6no9jTlTdmyMhhaLlUlMKUQNREOL7D15hQP1lWq5GpYhUkF8evag9k58WgEEsmMKGaF0jMFAKsukVF/QHy5YMy1c98r5Nhsk7sDMY+Yy1tcLJiqFgIQtG9ngYZgSptWJS1I0zyJcEjvSWjcQbtcmzE2e9x38dGpmqvT2MDHxKvTm4ZM45t3Fvq3plgL1Zj2d6v4+VmPx3QlVb9vdAJ4AUH9Mh0HJ6TdWZ/DqxIZmZXgkIrUhi6xcp/4VKaACKgb2tOh1/MmCcl6jppVdKxOTywcEBf8rcYEd4l/6GhUR58Oj5N+Bw8btI4qAUMbnx73+H1vUIql2S40R59etVgHtV0JeEyjqQnnSF1Ps3VWedcsjSG9hgeYPSJEE4A5JwisP0MwUvNwklg4WKj1l0ngdlWtywp1lOaW73G2Lvh9TBOLD5KqmWCPUln+IvkCf/6XWZkqKa200kq7YfZKpEr+0zv/XALAViWCb6nTMcoMHBJcrmpmeL2rvOPe5gz2EoWzNQPJqToFna93GNIjh3NkdJoaHZfTCtHHY24+aX/HhvaQ1AIcG3ip+BH6fzDlrqvBtIIzaozJpIYtSgNsbQ5hkeq8bgPzE/X5g5MWBlEBM2za6vTNna66E3HXWd5xBgBhaqBLdKyel3DxZxC6HE5V7QQ9Uirp/LbLTSFYxKzEIucxd5f1P3b5ORwnxf4lpWh0iVtdEjr9tmRYoZjGSC8KjzPn4RAJWCHGcxOY1PWVd4N+1u9iSMTzFUNwx97G91P2PMUwQHpGHuwCrDYyGzioNiltUpFFkXDVB0xKrUxDRJ9QUUsA6UIN5v5+mylFT0ILL4mb4mE9xddprvjVEAmlEl4MClWUhh2zd2bbGZo7xNS37EGG6sGTMyq+ToBwUnjBhqfeu1kFrB3lJWoNr6Dmi1NOXYnBAqMPib1SaJiTknmcGljdUikJ3QASgqiarrjWwavBahZr8+AT9V0/6XfwLrFB3rp9xeka6KpYDwAXL2vYH6nPJ0LDKXmt75CgwK2tK7jUGAoDSC7V91wc1nBMjJZbrTFc4tfQNMme6nXx7U5jweycs5mD9e9QA852C5Ign/JqwV3MMpXcxSwFuLN2vgv84Se3AABvdUZYXVb3mRcmAeD80McT4kvRAWxShFWvhJCUnsiFi3VNMmvidObiaKqeqesFuH2PUp/bFrIhPZ9VZCM0W4eYE/Q4LrpNpxcOLkZqXV4HDZhGxp7yYWCzB91zUvb0Y2HgCTV8eYbEt3pDvs9TiljGicmcTQ034ka33/iLf10WJ0srrbTSvgr2Snjc53//H0hAwb2G1AJq6QIV8k6F1JgPoL0dwn5NnaDaznpR5DkfQo6ohdUyimaSRhViVzUVpCcB7B+ogh221oCYcuJ7RxDPiAD9l1mRW7Qks4ydXdbxgryRRGh4m7y69Z0JzBblQCOJZEh81LtNPpVb1ASi6xLjOUlKVUJuxRUAc0cYuoRGRY2jUZ0bc7rLc/jfIXidZzHcUfPdgq9iuEB2SrCxQ4nBmTrNe5uza/zkiiUNUBJoOY+0uE7u35eIpqSL10gxHyiP7ReHy3iUc2BQm7FmSHY2x32Pvfy1N2YFB7dnQlJ+MDlPEY8Lf8H/FjUbPdwoinf9K8gDNb7ZIIQIiGtjDuh56k8veKwHBx5+dqLyu2Gmcy4UUEx5+djrlHeMQrMQkPAjZkjUKzoyynXG6ushEh0WiQtYS0bBGrjW46gA/QFkPseOF6wVqVka0iEVaDs6Fzzj/ZB5L+KFAYMk+tyegO7kvego3seVxsx0i9jidmjPStGuq7lV60TcOKZXDJZbm+9rLKd1kvPeODFD0lbuz2B2CMpo65AL0ofcLThx5hMbZq752oq5OJllGmoUMdm+gOETF8trraIZLhUqCgHVUYiHXASCWQnjvsCPf7Ghrik1rJDnubU2RG2LoJKezs0z45cFd4ppCLR6agzCGcm+pfoX1lbevOS5CTy63+qbFWaslNOI9wu5SLi+IRPJBcnp0MHLKxW1DmOL9VxXOxOYVhEZjAic8MmgjQU1221UQswJZimkhg7BcStmygXXeWyxkELbX2AwUWv3jf/170qPu7TSSivtq2CvhMe99zf/kQRULjimnKSUGhp+IZjJntwPJLN6wbOBStGqijFVkzs+0MxlpUbInipv2vjOA8gNhQDR5nPgfdWTPv7hBHGgTsTJ1IVJnlm1FrEck+5qCFUaHEdHTXxKVfDNaoCHd0gO7bauGNsAZP0I54/VfeaKIaYpOCeYe9UAcHHhMxFQsxMwuiJLNJz1lad179sTmN9SEEQ4NjClZx0vCu+7ajN6Y/bjKV4eKARIsxpg+XWq8tdN9qq+wKJo6/xzepEpjmkAesNEcqyu/8c/vcWiw2+0VB6yXQvQu6/uxfB1hMeEtJiZ8G8ToU/N4GfNxiksEtzV3twGunkjUQJ8rhRlwj85ZqSF1dSgV4jo5yTD2Z56ryu3p7DaRTNVRlJj4yMHT6+1v9eoRuBZKUPaKisZQw9FDBg0hXKPEQCiC8plu4DzOkV4b98FGjSvpjPg8b561ifDQkFFquYSQEmR6U6OSJGqEQkKTZNHC5p97TOpZHkzTQO7VVbPwOKFGtePPl9hKFrPDdGj2sg4dNEmJsvlezOOAjXXYBKrnGHwsl/DhCBy09jCCl3j1tdGsLZpbaUC2TDnLAeTVgWBzagWAJiPCBa3HcG+Rw1OlsHPiorNHrcchxBTEvTuVaARfDf8aITZmbr+46MlFs2tOzFHF7cfDfn9iKBYO6dPavDrykPO15ZhSY4WRhceCwfrmsTyHRXB2isGNJvGqOUBVNsQ44jZLmUm+d2kQ4EXT9RcHQQuK10NAhe3mmpeddZmAOXZRaZhNFBjeTCsc9Nb1Y6ZYkDXJOff63b8BQX6K0LU/bX3/tWryw6Yd04CioEOKAoNABBnBuqEZ179QcHkJRcxa/3paw2gRYsqiICxmlxyGkH7NSUAJNstaJcq1Jc/eYrJzwiCtyxAKCh88mQZcQ5JlBruL6mYubUR8mJIx4JD3cvTKk6nasKuN6dMqm40dA6Tpydqck+mLuoEnTPtaykZRyCgBTCf2zwR3XrKE1B3AOcRbSBOgeKUqSgKY0KyHFR8mGB+UbCs5RDA6xueecvn1Eo6yGBvEcXro1WgTovw+ALhe4q7IZ1p6BNfwxl1IZq6wCYVvTr3QhiNYpPOUziaqUHMCIP+ziqws63u3fOgjanw+Nk+sl1ijqtaBe47ziCoozJ4KfHhMyV+YesZurRR9XpTVDdpDAwgG1OKYarj4FAVtE7mFazRBrWxOoTbIVX4exUuXqf7Ey7UWS0K+99dBr5G6TXDAF4eq6H+6JDpW+GYKtwGkA0i3hCyYQZzWX3G2G4VvQXNOlCln02joBE1r6FzbYsx3drBCcSHu+oz8wTDj0hbceLygX8xruEiVOmBRWbgfkON6637Q6Y7zg+CdCi46DY693BJYbltZlwYrG4X+4K57HEBPN4NMTggVfVGxJw9AGBSl7H1qMs0vnI4L+hqhSwAAe0m8PIEABD+ydEXqIpj4idZTGwu1J8tKnjQVkW93saMO1WlAOITUnwndsDxwIPfpHXmCoSTYlzr5EyY2wUTY85SmN+jjOnQftRlzcns8Rkmj2lehSYmlKoZhi4GtMm2nQjrTbWRdG8toFHmLzjXWQ5RSA1VrwAt5Ev3clKFS+lgTZPc13Hrv/9+mSoprbTSSvsq2CvhcZ/9vX8oASAMLZh06lhWxgWITOicNln663bBmTwNCp7nOIW8JLc5FQUP8rv3IVuqqKC9PELyR08BAKcfV9AmwdHab7WZF2L+oytMLtQ1B5MqM/W1nBi3l9SJ768URPLxVEccqO+azxzu2Npsj+F31alvUzZAMzQOwdNAR0bFizQxUG0Th/M1h3V0VkAL650I1V9XaROt5jALYO4JAQCSDPpdYuSLEogjBRuLDhIWhzgf+7izqaKOyobkhhL7ga9SAYBKB1wo71f+cg+jn1BnnSsYDpdD+uZXDvORH141sE1QJ8vJYLlU0Noxob+7rf5gc1WlRQDg012IffKyrxVc00HRUWZtVPj3s08zvHip3v3JwmMuGwngNjU5ra6PGaqp6cV9Ts8dZmbsT2rYXlPfW7stoNeURxbtJ7C65DV+R8HTcGu9YLd7ug/ZpznmmADdVzYIWebLfNACHqpnlctLkA3qznMcwPySfjchuFkMaVp0TmYZtDkxPe4dACTnBb9SNF8N50gPlGfd/6WNwyv1XUcLj1MMTSvFMhX7Or6a781eAGeJvl8DQkX7gSQ00B+QMHRjgfpGDom81lHacjnKXTxNMSEemfZmCGs5j5IkjE2K2OKM56j+jbuQK+qLtash5PvPAACXP0yQJOr67Vsh9GLaI6OszPTM5vdnWYLnXHMr4s/mTWMQErPn6sfFzEa1XnCo5FFzNhfQPSrqJ0VKRO840L+mCqVo1YETVXSOfniA0b56VreafAGquBiSUMpZCzXi8DE0ydzcjp9x6u86/HM8LIr5tWrEzySEhoggtnf/8N+XHndppZVW2lfBXgmPe/9vqeJkzQ+5WSWJDS42ZKnOjRqVHZPVRqBryvMBoFXsgkP57gqwoWBb0rKgvTwCAMR/+AzRpfoC/wdt4N3X1GeWesBcHe3a3iHkU5XHnL8/YxjWs1GDT7m1yoI5qjdaYz7Rs0RHRFy7p0Mfy9Te3sg9767kIlY2lTCJcyW9yHB1SCod9RjVO+p7NNfgwmA01gvFle1CCQSuqSCBALDeK6CBv9pH+Fh9fzTSYToUUYxsHktnFbDeUcVa3N0oPL+9I4h95ZVrtoFkV13nR+9t4F5HedTdTcrlO8ojAxRHxB5Bph4sDdB7hxpbvnsXWCcY3WwOfPQCgGrMkQl5Y1UL2o66F3l7E6AoCQC0AWHzjs+AfVUIjj4e4eKFGoc/O1yBRcXeNztDhsglicFUAmZVIguI6THUYZAMnRSqxpA/i/c75G3dpv8GIRchZZRCzqim0qkAO+ozcmMVsk1R4HXPOk2BiPKZ0ylAbeBaEABXKhpCnBYRyGDKPCfXoXN6y2bWR5g6tC55s36lgMMmKXChvO/4yRCHH6l5+3JSx96cRHYJtrZZWTAbXau7KLxHvSjUxqNChPr0rA55rUhob9h8LznMU6ZSNbtBwW71BjH/rdSAr22rz7Rb0C7pXT494EL64mcT/NefqmjP0SW+3qO6kh8wj4xZVVBWQDEFzuiZpqGDO9tqrjpLlJvuWNz1lg0TROf0apYB656KSsQ4RLxPc6Olw1gmWOqb24BP43vSR/ahKpjHJyk+e6yi2SA1sUEFyXoj4PHLYh2Tq0I27wXRM6xW58yrY3clYtqDdEsimqo94PzCR43gz66XoLZCRfXf+8+vbnHyCRUn65UQFk2uNNVZ2sv1EsREUO64KZpvE5fAml8gKl6/xQgFadvQaMFg7xDRH+8DAOyHDWhvqV1Rrq5AeiR1FgTQjile7F8V4WqcMjY83ZtyZf140OCCxGVk4e2u2swqTsLhnGkK/jnnVYgzA8uEK/aaMSzicTcaBqdq0oFgwQSzgQILbWgQUzXRjZ4LLQ/nuu0C/5xlwIF6juB/7OLwGfEpRDYXPpZaU1gO6Q1+vwrtNUKq6Bqwp/42F1cA1MYdfapWzE8/XOcuyTcJx16vh3xYWVaGkAjs1389gvE9VRRGq84blXx8CLmgDalTKRb1+hqkT8VlXWc1cG0yhravDl4EYXG4RHGhvn4+Q7inJv3FyxoCIlvqzytMZr9ZLyhF6/WQN6XaUgzvXeqA3OoWqJGxurY8uOTimrbVAe4qbS/R6wFV6n6VsrjfYAHtTB0uOD6DeKZ2DTlPoFVJj/DekhL7ANR7vxzzc+cUvZCykDEbBkgOSMDieZWxv0uPQpg5QmfpmpBInAJD0ms9myE+ogLmS3W/Z5MaDkkisGZm3BHcbc5he0X3bK6IDoDD+P6khhalXlbuz7gPAKngoqy+4gNLdPBurUI6aq1oZ33gc+UUyVQwnW38bIqrXbXh/eq0x9TBVSvFiGiSdzpDVImSNY4MOC6BGFKd53Nu86mD7m01h61lk3lIoj6RxQHQayZ0eh/aw3VgqZM/KHCo0lJi9wKgwyV4nuD55woPfxU5zDdyETqM6e41rhU5Uaz7MLJQI3BFmhmoN9T4uS3B6C0RSj6YDKfoXK7+/pdv3GWqpLTSSivthtkr4XH/8rf+iQRUF5JNJ2kcmlycrFUj/v104jID3dp3E+jfJmb7agWYUiXj6ALxxwO+vv1dUkC9tVZ4p6MJ5K8UxCo7nAFENzn+VINJklH+GyZ3VwHgYlR6HmC2p86866KgrpnhmPhVlt0Iy1V1PznvwOGiggcE02pWA5yOlHdXtRNsbhN/gQEMT9U1jkd1LNE1VndmsO9QOLezCvjk7Y1nwJn6WzFYILtUHlv/UxeDqfp8mJkYkuey4gXcTdi6F8GgkFamErpH+NLNJtClotpwCnGiPEI5TzB/psbm+Fj9+8XCQ0xdiOPExCPCd+/8zgz6jhKewDyCoGKq3vKAh6rwJ9stlVoAgCiCdkLe6a/2WcjB6hUq8pdPHC74Vl5zlFAyCB6ZpximEZID5dHEVxqmQ/WZ87GPgLvXgJ1VgoVKoPM2da1uNRjWxwW1lgd87XZxv9QtqS2CAhr45AS6T89RsYHca9Y1yNxb7/qF0MfJUM05AJqrM1TSetQGGvReB9OCg2Yc4uTPiA9n2MDzad5dLJmu915zzCyK/g44bQG7qHbnkY4YRIjO1LVH5x4m1GUYpwY+Jd6N7986QWNVjfX8wmbxBkMXyOh9z2MLdcInL23P4O6oeatvd4BNSo0lKRf4ZH+ixgEA1rrAiKKg8zFHzmIYIj5Ra72/V8NLEvEYJRbuUGTgmClHVZ6VcodyjoOOUpOpkXteiF6TsNtuCrelPuNsWdDfpn3Bc4EJFfv3LwuI8YpfQDjHC4ghCUIcBRjuqvd9PKjj/YFaCw/8EMvkfQeJCYPSd9PY4j1rubpAvabG9XJc5Xte3Sgiwqt+hUXFX//j33t1UyXPf/cfSwBoNAIOe+YTGwGR49T9EHGsXo7rJWi+qf7OuN8tsNtnV8gOKOQUEnpLvTjt4foXwl/5VGFHRz8KWB+u866AZqnJePWBxMWVCj+DxGT+583tEewlwiXrGmRCJFPVok16+NzBFak/H8+qzE2dhzW/umrhYVPdY7MaYJFTVs5q0KhPemdlgIzC+NnCRZ8OggcrAyz9XdpM2z7kS7UYsvMFc4xDB/Mgx88DTE7V9Vv3E4iQ6GwPHOg0oUw7g3+Xfr7TAO4qjDSqHjAnoO/eaaGGbRvFZkKNFLNnkpuXpjMHY8IS3701YB5y974L/SFde22pyMuOp8ChSivIwVxpNAIKKcSfWSD8uTqYDp818JTyhpexiW8Q2dLG6hCVVdpo/aKpyGjZnIKKXsasuJImOudsJzOXq/9eN+P8rf4GpaJWukV65moE+YTSNkIWRF+erTYokPoSbZA5gZe64DUVIPNaoHsNdQEUdLIykypXC4W4CR6rd/DBx2t4Tjzyli55bvWcAokDgHmy73SHaG1QLpdw1rqrf+Ee0gv1t/GVhiyhVvzQhEvKTsfHDfzsQqUhO07COqimLhi9sr0zKjRG1zrAkFJNJ6MCz7+zpvDbADBfAE/VWCa74yLN4hlc95BhMS7pWGDRp3k2cbFLDXCLzMBby4qiN29cOx7V4egFhcRSS92LU0lR2SCc/utdvi9xOuFcvd5yod0lh6NR48MlT3mpmxGQUU5EJZCN1X0GFzquBkR8Nq3ho5H6+a3mnGmHw9RkDnVDk7iiRqiLyMY3ltRzmKbgVN7t//kfylRJaaWVVtpXwcqNu7TSSivthtkroYCT2/UKseulDA3UDYF6W4UydlNCX6KUgWtD7lEIo2kwdqirwLWKXHaaMfxs9t4Y7z1WIXAq29ik/LH7bAinQzzMsYsRIUYSoePZRKVN3u930KFwdNUL0CSuguXNaSFY7CXYqKvwfTUZc5trLiBsaBIzkmVqVgNElHPdm7toU4trEFiwSJHl2aiBKYVVb/kJxKkK2+TuFfSeCtPN17rMQw5dY7krPA8QEwJEhDH0qhrMxkbEPB3ujgt9g0LXlq9QGwDw4gTJM/Uc2VQiHKrzfXBZw9IqVeu5yUVHbVmlTeobMdoXakwrdzSYr1HI2awBLr2Pw1PIfRUSimkM3Ve/1xpF14Uczrl1X0YZK+bEmYGEQnNLk/hkqPKff3LWRuuxGrN1L8Z2g1rwe/MCqhVpqHTVGOsVIOqr67jVhOGXxnIN2gZ1S+UcOKd9Zv6Drn1RcYkac+TlFHKSk59IyIyalE4DZgrUfYPTcVrdKWT2qjY0t4C36suU1rMKwV95NoZJv359qw+TFJIuI5sVfp7PHZbKyqQGIdX9/2JUg/xc/e1mRc3f9UqAdYKz1RoRcv/N7RQMf9k8RjQg0eXtIZY6Kt1w3G/glwNV04mEhjtVNd/WphMYH1OoPy+aYrROFWhfQwvtUV1g95IlwsJj4MdPVCqt5cSok9rVUmeGyhK9Mxf8/qQM8U5LzcM4NHme500rw8hm+TPfzOBQ2qhtzhGdE38OrVMA0BtOsQ6uQyxfHCM7UuOUDRO8+FA9t6kLbgisLaUsZ+f1BBop/b4a4Tal8iYLl7ngJ5HNbfHTwMGAeGeELPaJrjPHMLgG+f0SeyVy3P/tm/9MdU5mOsO37vpzpHnRK7aYOH3nW1cwV0kdfRDC3CJMXePag04DiL7KW04/yfB4T20gGiTuUPdjeyeGeZu4P3wX4lgt9tM/13kCnExreEqFoHmqg7iZ0HMErmJ1b4kAHvhqYlSNDA9I66/ixRhRLjIhWKNtZkzuY2qCD6aLwOUN+lZ1wUWNMDWwSzS33+xd4c53Kbe4VVcq4ICCVOUkPkGC9KV67qc/LYQGEqmhQfClzfoE7Q6R0H/TKYQUBgGyMUHarimyR33gxa6CSn1w1cBZqJ5lhboi23aGh/TM7fqCKS4bb2kwNujdpALZCXUcCsn5TDHLkE2LPGYOLdOXaqrIBwCaBnmmFs/hH2noE69GLAzWWNxfOExgHwkNWyQAcBmb2J0R3r4i8BYtmNXGjMn3Vx7MYG1QkbPuFCIIpEEIXYPmECxvniIdUO60osFYor/zHWi1a/n5/CA1jcKBMAvI518x0/jy3y/o8PrkJcS42AwFvad0KhEOqWt36mBMRcb+wsPZNQX43IbUndgwBcYEVZ0lGh5RH8Jmdc6df5aVMWeOVckgUzVeaayziMY8cDCjjefO5gD+O1QfaLh/ZT0CSkYw15yUqUR0pJ7jbM/HH7xUxcyjBfB6Q82JmilwSPWTW5UUa7mzVJ2j0SgI6PKalE8cILoumH9lGLksbeZbCe4/VIeLvaQXe0elqIXIcYD0mHo6bB2SKIUXxxp+9FQ5fWehxZj4y8hAz1HPcbs25zXdqS2YRyYIbCxyHU0zxYLGLEwNHJJQy97cwjda6nsbdowj+v3f/um/LHPcpZVWWmlfBXslPO5c5X2RGnhKsCNbB+4SbGaSGKhT+uDX3z2CvUGeWbfCAPn0eIHFUXE45c0DjScIaJcAAAxaSURBVOUQ7l3yBNYbSkgYUMgJYmVDEEG+UKD72U9mMKjLUGRg0do0MbgJSNOAIFT3cDipY0yeXyx09Okzri7RIpmv3BtcdkOEmfpsIjRWWzd1wWmTaWyhTl5PDiECgM2tESqvUUhtG18Qs5XktMoMiCcE1ZrYjJw4uGqw555IDTu++t77awN4DepurIJls3TfuMZwl2C8q8ZvNPEwpXBumlCKAxITEjb9YOjgjYa692+uXqCaC6d2U04nGS0T+pJ6B1qrWnDNVBzAouczjULkIoqBI+UlLX58xTwPebcjACQLg71ATSuaRU5HPuapxeOdK74fBSYeEUvjg/aQRXR1C8w5krMD6lWd01LaUr24X9sqmmiAQqk9zQpukygBQoIGJhmLWcsohYyvKbvnEY6Q3OwjBRgdE59lCK5Iwd0WzCqp6SjevdT45yQyEAbqWWdh4e3tUQR2EZmYE2qhZwtsUnPIPDXxAcEn/8bKBEvEpvh81EBE0e+KF6JJqQzbTBkaWPdDdL9P76BV4QhWTFPIPNIQYOFgzTNZZu/T9zr4BaFE+qGO7WrKw/EZiSe7hsR2Rf3e0iVOaP117Qx3aB3lTTGGLmBSE08qdEzJw12pz7D6lnomc73KUYwYF/eoGRqnunRP5whh/lxieKHmQZwaDEccRQ6LJIwSk9Fh09SAS3TOS07Cnb0tJ+KmorYTwSY4Z5wa3I19fUv+9p//m1dX5T0PeX0rwTsULoxjCyktwLqV8eanW+C28S/QcHY1+G/QpKjZaOTUp36T8bMQkjvKxLNzpEcq3Fqc6oip+++z01U0SCuu5sRMr+g4KSo1CiMrAnWoSddeFNzEugkmnIpCkzfOK0qZnC0qnKOtGBlDrGaZhjVXLequEzMEcHfuwqUX3mnMoX2uvl9mGtx1yvWumtfa/i3kmWJ/GDBZU+Uoxk6ux2cIbn+3uxJmlzbglleE+4auNh0AmjNHlfLNbi1BLlWomTklLZDO1bV/IzSYZ7y2JTiVoDdqQN6W79rFRqVramMGgLMhp7eyUYrpXjGJc+qDP392CzWa6F03QK9BbfeGYIUW2035+Sq1mFMiuiGQUj44jgz+vFtL4SxTJ27L5lw7j4VXdA9CSOCKYG7DhYJiAsjGAvMzavG2CiiXpkkmElMbaw4nlbC9otsvmpt8j/mcUYo9RFMw9fCM5spFZOOM5upFpCFf1beqGYfsvpmyon3FTDm/mq+hTAI+vb+OkzBULUqK7aDlhVxrqVsJzin1cjCv4CmlIc5CA2821Nx4zRAY/VitJ687ZzInvWWr1AkA1JzisAtiiDHpjRoCUxqnuiXRJghu24lwh3QpK2YKnzDjmi4Z5xwkJqd38jTkZ+MaLimVaenAPXIgUqFj8T519q6PYbfzFnkTmkdr6HraS9cgx9TlOJ6gkZKSlVlADzUdwDVu/ZyPOwl1LGZq7iSJgYCcht2Jj1XqPD1aVHAcqM8sOQm67LABe7RnfBtfbmWqpLTSSivthtkr4XFvUqOKoUkcUzgnAGxSqKZpkrujsgAAKXCbHRPe1ymMdcwvhJx5KCpPxshIYfz8sYezsSpkfDpew2vELXB3dVAQWkkNJ7my+7yK3TkVEjJgiQpyt6shunRq6prkJh2/HvJ1qn4E01M/5/qMt5IhkxvplsQbRAc5nznsqTY7AQzq3Hw0cLgZp9JI4G6TZ9arMI/FXzZxTUghb8BZLGw45I25lYQbY8yuDa1J3lCSQRwSkuQqQTygYt/UxOWAtPZCBzV61gaF11U/Yu/DdARsn669VoGWIySEZM9F7g0QE3HWom/CconYvlqkfMwq4G8TOf5M4uyFemc6gB9dqndjaB7qfeKmkYU32bEz3PELrpW8EaNZCWHbBQ+HQ2G3uyxgrhPfR6PokuX7PRgiPqQi05XBVLVmRTJRvl7VUNsihMS5Dh2FZqlJ3B9OV4NGepLhsWDdRsvOODT22imLXKQjCUHBiBkI7k49j0wQZQ3CTKJHzuFVrOOcGrpezizEFPrfq+sIaQ5l9EVLLmcYcRLYaFjquW3jWqFYF6gRN0grspjrpmbHHN5ruoRJnX/d+yGsOzSOVbuIcnWNu1qva6LGfcGkX1U7BvXcIJQadkmXdZYaqBhFyiPvMmz4AacRGwCLJizTc67UZ/zvui44UnY7AmaDoo+eVzR82de2wTBhyuTsIkR4SpqQVw4+OFIgh4aVsKJNzYlZUd60Mzg+dWZaGWxC3BhOkf7YjIYwqW77YKZhMSYRj1Rn/hWvmeK18f97ay497tJKK620G2avhMe9vj7mn13Ky6apjmbzmuYktbxbXQ1Go8C9ZsfqdIzPBFMkJpGBEbWefzpcwsFCeSKeIRGK3PsAzojPYGniouarE7Rhx8yP4Ogyr1Vhfw4cU2Hs2bSCOMsFFsB5xqZd5LpalsBdKgLmXp9pCIYs1WoRZlNSyxY6XLNoWWai9anHSveaIRmylB3OEPdzbUedc7ci05jf5XzexAk93+7MxIK8kdf8BDsUaSwfTuF6pBdpCZaPMl3JhUrLzWATDvZP+z2M4vxZ6d91VUgGgGUnRY/ydI9OLlGpq3b2JDTYs7bqhR5gZSnl79F0sJyWCAFBBSKZgj2athMjyhz+/gkN2SQBqtTCfRVr2J2r95dJgOQWYesaclnIqinx3a7KVddPYmycqkjDcGYQBJnLW7+VeATd70rG9y4FuJApE8k/u71CFiwLNC5uh7sGTMoZe8tAu3dNjOOSinGJUmUHAJEZXGQ1rQwPiep0mi7hjAqPW1VZSF9FGlL6n6NFiIcUPUSZhqM59UbQABiaDpfy5+ehhhckQ+dbgJXD2ZwKPJp7UWLif1+q4uGGl6BJMng1K0GDPM/wIoZmUs2hFSEbUV/ElWRJMSk1mBSxGF6hu2pZGeb0nuYpcEQ1k6uKze9skmhY91Q03r3KWEG9Ygq8GQ5x3S4DD3WqU1XsmD3ZdCEgIpKssyNIKo7GlwKC7jGNilqg6cgv9GjkXu5//LwCj8DbTUdDi9ZC1VRrAABsXTCgousGzOkipIZOj+psmcYMqKYpOJrTzGK/+7/ZK7FxX56T5p2VYUDoh1z3EVAPNaRkffbJDAkVUTKhIUhIf1JqCAkLXXNirqTvzm20KfVQMQR8muiT1MAZVab9UQM+cRXEwsCIFu80AYeZti6xRi8xEkCXoqxRrGGS5AumeOnjCHi7ozaQvBA0iDXcreaYzwXzFDyZOnmfBr7Xm6BDaZi9iY/399Xme+tAYOMT9fLvdYZMGZsXswBVTc8JrS4jGz5tuHXLYHxpPzLhUDrqk5GPh3W1iedCswCR9FCB1DEyDqHv1VI8If2+fFyqpkSD0kOHgYmfEB/IB8Mt7Phq8dzxZ6x56QxSBET9ehW4WFBF3jEEF8CuYgP3CFHUtCN4hBaomgneaqrrjBIDVbqvRGrYn6t3v+QW9yMA7FHhL8qACWkJBhlwRpwqg8jhMWz7c1xQKi3XPlXzKkcC6bigewwyg9MzYabjbeIp7zTmGIwJQxy6fHguORFM2qg2xhMO5V0vwYIci4NhnYuKADAmNMSmP+M0Qd1M8S3SyzwKLIzogHk2TtAgQql7vsdUJJGQ8Mx8rhBaSoI3yigrwviOAxCQBSeBiy4VAI1rxbePxzbmKaVkphl+e1XN8d+UGmqE0jgd+pw+rFzjUMn7IwBgEVucYrPNDK8Tuun5zIKkjXDFFbwWl1yJQUyFW83AmJ5791KDratDJT9QPp1UcBaqNN0gkvheTz3sO90rTrl+/nOfCdEqbsz3kqYGC5ZfLjxOEXlWggY909sdBwPa4INMgvqq4JsCJ1Q4/sVAQCev5I7v4U6V+gMAvEWokkzq+NNzlYY8WgC/SYf5G50hF5RzMMBftjJVUlpppZV2w+yVwHGXVlpppZX2/2+lx11aaaWVdsOs3LhLK6200m6YlRt3aaWVVtoNs3LjLq200kq7YVZu3KWVVlppN8zKjbu00kor7YZZuXGXVlpppd0wKzfu0korrbQbZuXGXVpppZV2w6zcuEsrrbTSbpiVG3dppZVW2g2zcuMurbTSSrthVm7cpZVWWmk3zMqNu7TSSivthlm5cZdWWmml3TArN+7SSiuttBtm5cZdWmmllXbDrNy4SyuttNJumJUbd2mllVbaDbNy4y6ttNJKu2FWbtyllVZaaTfMyo27tNJKK+2GWblxl1ZaaaXdMPs/FjyfeQsA93UAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(data_set[0])\n",
    "plt.axis(\"off\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Splitting the images into blocks of size (144, 144, 3) to use as input to the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(4026, 144, 144, 3)\n"
     ]
    }
   ],
   "source": [
    "# Define the window size\n",
    "block_size = 144\n",
    "\n",
    "w, h = data_set[0].shape[0], data_set[0].shape[1]\n",
    "block_data_set = []\n",
    "\n",
    "for img in data_set:\n",
    "    for r in range(0, w, block_size): # 144\n",
    "        for c in range(0, h, block_size): # 288\n",
    "            block_data_set.append(img[r:r+block_size,c:c+block_size]) \n",
    "\n",
    "block_data_set = np.array(block_data_set)\n",
    "print(block_data_set.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "PR3_60\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         (None, 144, 144, 3)       0         \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 144, 144, 32)      896       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 72, 72, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 72, 72, 32)        9248      \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 36, 36, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 36, 36, 16)        4624      \n",
      "_________________________________________________________________\n",
      "max_pooling2d_3 (MaxPooling2 (None, 18, 18, 16)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_4 (Conv2D)            (None, 18, 18, 1)         145       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_4 (MaxPooling2 (None, 9, 9, 1)           0         \n",
      "_________________________________________________________________\n",
      "conv2d_5 (Conv2D)            (None, 9, 9, 16)          160       \n",
      "_________________________________________________________________\n",
      "up_sampling2d_1 (UpSampling2 (None, 18, 18, 16)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_6 (Conv2D)            (None, 18, 18, 16)        2320      \n",
      "_________________________________________________________________\n",
      "up_sampling2d_2 (UpSampling2 (None, 36, 36, 16)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_7 (Conv2D)            (None, 36, 36, 32)        4640      \n",
      "_________________________________________________________________\n",
      "up_sampling2d_3 (UpSampling2 (None, 72, 72, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_8 (Conv2D)            (None, 72, 72, 32)        9248      \n",
      "_________________________________________________________________\n",
      "up_sampling2d_4 (UpSampling2 (None, 144, 144, 32)      0         \n",
      "_________________________________________________________________\n",
      "conv2d_9 (Conv2D)            (None, 144, 144, 3)       867       \n",
      "=================================================================\n",
      "Total params: 32,148\n",
      "Trainable params: 32,148\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "MODEL_INDEX = 60\n",
    "modelname = \"PR3_\" + str(MODEL_INDEX)\n",
    "print(modelname)\n",
    "\n",
    "input_img = Input(shape=(144, 144, 3))\n",
    "\n",
    "x = Conv2D(32, (3, 3), activation='relu', padding='same')(input_img)\n",
    "x = MaxPooling2D((2, 2), padding='same')(x)\n",
    "x = Conv2D(32, (3, 3), activation='relu', padding='same')(x)\n",
    "x = MaxPooling2D((2, 2), padding='same')(x)\n",
    "x = Conv2D(16, (3, 3), activation='relu', padding='same')(x)\n",
    "x = MaxPooling2D((2, 2), padding='same')(x)\n",
    "x = Conv2D(1, (3, 3), activation='relu', padding='same')(x)\n",
    "encoded = MaxPooling2D((2, 2), padding='same')(x)\n",
    "\n",
    "x = Conv2D(16, (3, 3), activation='relu', padding='same')(encoded)\n",
    "x = UpSampling2D((2, 2))(x)\n",
    "x = Conv2D(16, (3, 3), activation='relu', padding='same')(x)\n",
    "x = UpSampling2D((2, 2))(x)\n",
    "x = Conv2D(32, (3, 3), activation='relu', padding='same')(x)\n",
    "x = UpSampling2D((2, 2))(x)\n",
    "x = Conv2D(32, (3, 3), activation='relu', padding='same')(x)\n",
    "x = UpSampling2D((2, 2))(x)\n",
    "decoded = Conv2D(3, (3, 3), activation='relu', padding='same')(x)\n",
    "\n",
    "extraction_model = Model(input_img, decoded)\n",
    "extraction_model.compile(optimizer='adam', loss='mse')\n",
    "\n",
    "extraction_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "PR3_60\n",
      "Model: \"functional_3\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_2 (InputLayer)         [(None, 144, 144, 3)]     0         \n",
      "_________________________________________________________________\n",
      "conv2d_9 (Conv2D)            (None, 144, 144, 32)      896       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_4 (MaxPooling2 (None, 72, 72, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_10 (Conv2D)           (None, 72, 72, 32)        9248      \n",
      "_________________________________________________________________\n",
      "max_pooling2d_5 (MaxPooling2 (None, 36, 36, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_11 (Conv2D)           (None, 36, 36, 16)        4624      \n",
      "_________________________________________________________________\n",
      "max_pooling2d_6 (MaxPooling2 (None, 18, 18, 16)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_12 (Conv2D)           (None, 18, 18, 1)         145       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_7 (MaxPooling2 (None, 9, 9, 1)           0         \n",
      "_________________________________________________________________\n",
      "conv2d_13 (Conv2D)           (None, 9, 9, 16)          160       \n",
      "_________________________________________________________________\n",
      "up_sampling2d_4 (UpSampling2 (None, 18, 18, 16)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_14 (Conv2D)           (None, 18, 18, 16)        2320      \n",
      "_________________________________________________________________\n",
      "up_sampling2d_5 (UpSampling2 (None, 36, 36, 16)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_15 (Conv2D)           (None, 36, 36, 32)        4640      \n",
      "_________________________________________________________________\n",
      "up_sampling2d_6 (UpSampling2 (None, 72, 72, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_16 (Conv2D)           (None, 72, 72, 32)        9248      \n",
      "_________________________________________________________________\n",
      "up_sampling2d_7 (UpSampling2 (None, 144, 144, 32)      0         \n",
      "_________________________________________________________________\n",
      "conv2d_17 (Conv2D)           (None, 144, 144, 3)       867       \n",
      "=================================================================\n",
      "Total params: 32,148\n",
      "Trainable params: 32,148\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Epoch 1/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 969.6721\n",
      "Epoch 2/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 565.6660\n",
      "Epoch 3/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 503.3285\n",
      "Epoch 4/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 458.1660\n",
      "Epoch 5/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 458.5461\n",
      "Epoch 6/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 420.0312\n",
      "Epoch 7/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 415.1882\n",
      "Epoch 8/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 408.9406\n",
      "Epoch 9/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 416.8250\n",
      "Epoch 10/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 401.1034\n",
      "Epoch 11/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 407.3689\n",
      "Epoch 12/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 378.3725\n",
      "Epoch 13/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 381.7477\n",
      "Epoch 14/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 373.1960\n",
      "Epoch 15/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 366.4481\n",
      "Epoch 16/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 355.3190\n",
      "Epoch 17/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 339.5967\n",
      "Epoch 18/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 334.0274\n",
      "Epoch 19/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 323.0029\n",
      "Epoch 20/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 312.8232\n",
      "Epoch 21/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 313.5172\n",
      "Epoch 22/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 311.6402\n",
      "Epoch 23/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 313.6391\n",
      "Epoch 24/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 308.8254\n",
      "Epoch 25/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 298.7511\n",
      "Epoch 26/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 304.3110\n",
      "Epoch 27/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 301.0438\n",
      "Epoch 28/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 297.5313\n",
      "Epoch 29/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 294.2930\n",
      "Epoch 30/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 298.0006\n",
      "Epoch 31/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 289.4968\n",
      "Epoch 32/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 286.1351\n",
      "Epoch 33/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 283.0552\n",
      "Epoch 34/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 283.1088\n",
      "Epoch 35/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 281.2906\n",
      "Epoch 36/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 282.1829\n",
      "Epoch 37/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 279.1983\n",
      "Epoch 38/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 287.5917\n",
      "Epoch 39/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 272.9728\n",
      "Epoch 40/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 274.2317\n",
      "Epoch 41/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 291.5550\n",
      "Epoch 42/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 275.1032\n",
      "Epoch 43/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 275.8950\n",
      "Epoch 44/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 273.6194\n",
      "Epoch 45/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 275.0654\n",
      "Epoch 46/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 269.8291\n",
      "Epoch 47/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 271.3097\n",
      "Epoch 48/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 271.6294\n",
      "Epoch 49/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 271.3098\n",
      "Epoch 50/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 266.8923\n",
      "Epoch 51/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 273.4346\n",
      "Epoch 52/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 267.8590\n",
      "Epoch 53/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 267.5688\n",
      "Epoch 54/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 270.0528\n",
      "Epoch 55/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 272.7704\n",
      "Epoch 56/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 265.1891\n",
      "Epoch 57/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 267.2220\n",
      "Epoch 58/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 265.7175\n",
      "Epoch 59/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 265.6469\n",
      "Epoch 60/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 265.2745\n",
      "Epoch 61/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 267.9415\n",
      "Epoch 62/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 269.2947\n",
      "Epoch 63/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 265.5148\n",
      "Epoch 64/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 264.0240\n",
      "Epoch 65/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 265.3611\n",
      "Epoch 66/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 260.5739\n",
      "Epoch 67/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 260.9236\n",
      "Epoch 68/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 267.2196\n",
      "Epoch 69/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 265.3335\n",
      "Epoch 70/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 262.8551\n",
      "Epoch 71/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 263.7414\n",
      "Epoch 72/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 260.1729\n",
      "Epoch 73/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 263.3624\n",
      "Epoch 74/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 259.9866\n",
      "Epoch 75/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 261.6360\n",
      "Epoch 76/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 259.4509\n",
      "Epoch 77/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 263.7151\n",
      "Epoch 78/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 257.9455\n",
      "Epoch 79/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 260.8941\n",
      "Epoch 80/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 256.0585\n",
      "Epoch 81/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 264.3340\n",
      "Epoch 82/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 258.4882\n",
      "Epoch 83/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 256.7939\n",
      "Epoch 84/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 259.0901\n",
      "Epoch 85/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 264.2135\n",
      "Epoch 86/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 257.2832\n",
      "Epoch 87/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 256.7628\n",
      "Epoch 88/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 258.5800\n",
      "Epoch 89/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 254.9159\n",
      "Epoch 90/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 254.7529\n",
      "Epoch 91/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 258.3952\n",
      "Epoch 92/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 256.7316\n",
      "Epoch 93/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 257.6335\n",
      "Epoch 94/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 257.5360\n",
      "Epoch 95/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 253.4142\n",
      "Epoch 96/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 255.0045\n",
      "Epoch 97/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 253.9819\n",
      "Epoch 98/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 254.3834\n",
      "Epoch 99/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 258.9078\n",
      "Epoch 100/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 254.3526\n",
      "Epoch 101/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 255.6107\n",
      "Epoch 102/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 252.9263\n",
      "Epoch 103/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 254.6637\n",
      "Epoch 104/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 253.0140\n",
      "Epoch 105/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 255.1613\n",
      "Epoch 106/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 255.9498\n",
      "Epoch 107/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 254.2308\n",
      "Epoch 108/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 251.1354\n",
      "Epoch 109/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 251.9611\n",
      "Epoch 110/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 255.1731\n",
      "Epoch 111/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 253.0172\n",
      "Epoch 112/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 250.4789\n",
      "Epoch 113/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 253.3048\n",
      "Epoch 114/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 254.0508\n",
      "Epoch 115/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 254.4327\n",
      "Epoch 116/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 251.5789\n",
      "Epoch 117/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 252.3724\n",
      "Epoch 118/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 250.0203\n",
      "Epoch 119/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 251.5867\n",
      "Epoch 120/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 253.5316\n",
      "Epoch 121/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 250.5659\n",
      "Epoch 122/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 249.8738\n",
      "Epoch 123/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 249.5955\n",
      "Epoch 124/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 250.0762\n",
      "Epoch 125/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 252.0884\n",
      "Epoch 126/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 250.4843\n",
      "Epoch 127/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 249.1406\n",
      "Epoch 128/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 249.0301\n",
      "Epoch 129/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 247.1101\n",
      "Epoch 130/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 247.7309\n",
      "Epoch 131/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 250.5032\n",
      "Epoch 132/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 251.0792\n",
      "Epoch 133/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 252.4151\n",
      "Epoch 134/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 247.6461\n",
      "Epoch 135/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 250.4793\n",
      "Epoch 136/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 248.6740\n",
      "Epoch 137/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 247.1644\n",
      "Epoch 138/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 248.0695\n",
      "Epoch 139/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 248.7240\n",
      "Epoch 140/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 249.1341\n",
      "Epoch 141/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 249.1618\n",
      "Epoch 142/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 248.6900\n",
      "Epoch 143/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 248.6023\n",
      "Epoch 144/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 255.9443\n",
      "Epoch 145/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 248.0437\n",
      "Epoch 146/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 246.5400\n",
      "Epoch 147/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 245.7702\n",
      "Epoch 148/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 247.9877\n",
      "Epoch 149/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 245.5012\n",
      "Epoch 150/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 248.5879\n",
      "Epoch 151/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 249.8610\n",
      "Epoch 152/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 246.3896\n",
      "Epoch 153/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 247.1502\n",
      "Epoch 154/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 249.4695\n",
      "Epoch 155/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 245.5445\n",
      "Epoch 156/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 245.3395\n",
      "Epoch 157/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 246.8200\n",
      "Epoch 158/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 244.8130\n",
      "Epoch 159/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 246.9720\n",
      "Epoch 160/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 246.7967\n",
      "Epoch 161/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 244.6973\n",
      "Epoch 162/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 247.4181\n",
      "Epoch 163/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 244.7034\n",
      "Epoch 164/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 245.5902\n",
      "Epoch 165/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 245.5496\n",
      "Epoch 166/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 246.7022\n",
      "Epoch 167/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 243.4110\n",
      "Epoch 168/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 246.3056\n",
      "Epoch 169/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 243.8266\n",
      "Epoch 170/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 243.6567\n",
      "Epoch 171/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 246.7129\n",
      "Epoch 172/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 243.8829\n",
      "Epoch 173/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 245.7393\n",
      "Epoch 174/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 244.7716\n",
      "Epoch 175/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 247.8971\n",
      "Epoch 176/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 244.7583\n",
      "Epoch 177/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 241.5911\n",
      "Epoch 178/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 245.4186\n",
      "Epoch 179/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 243.0693\n",
      "Epoch 180/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 243.9389\n",
      "Epoch 181/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 243.3615\n",
      "Epoch 182/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 241.6456\n",
      "Epoch 183/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 241.5433\n",
      "Epoch 184/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 243.3721\n",
      "Epoch 185/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 245.7297\n",
      "Epoch 186/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 243.9993\n",
      "Epoch 187/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 241.9610\n",
      "Epoch 188/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 244.8376\n",
      "Epoch 189/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 244.3368\n",
      "Epoch 190/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 240.8236\n",
      "Epoch 191/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 244.0897\n",
      "Epoch 192/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 242.0925\n",
      "Epoch 193/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 241.6271\n",
      "Epoch 194/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 242.7032\n",
      "Epoch 195/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 241.9483\n",
      "Epoch 196/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 241.0083\n",
      "Epoch 197/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 242.0132\n",
      "Epoch 198/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 242.6820\n",
      "Epoch 199/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 241.4738\n",
      "Epoch 200/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 240.2033\n",
      "Epoch 201/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 243.6227\n",
      "Epoch 202/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 240.1339\n",
      "Epoch 203/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 240.0487\n",
      "Epoch 204/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 240.3190\n",
      "Epoch 205/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 242.2791\n",
      "Epoch 206/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 242.5768\n",
      "Epoch 207/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 240.0705\n",
      "Epoch 208/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 243.0084\n",
      "Epoch 209/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 242.0941\n",
      "Epoch 210/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 240.4524\n",
      "Epoch 211/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 241.4943\n",
      "Epoch 212/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 240.7077\n",
      "Epoch 213/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 240.1633\n",
      "Epoch 214/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 238.2788\n",
      "Epoch 215/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 239.9368\n",
      "Epoch 216/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 241.8946\n",
      "Epoch 217/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 239.6402\n",
      "Epoch 218/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 240.4831\n",
      "Epoch 219/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 243.2972\n",
      "Epoch 220/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 239.1234\n",
      "Epoch 221/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 240.7524\n",
      "Epoch 222/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 241.6155\n",
      "Epoch 223/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 238.8169\n",
      "Epoch 224/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 238.6102\n",
      "Epoch 225/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 239.4580\n",
      "Epoch 226/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 238.6840\n",
      "Epoch 227/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 241.9776\n",
      "Epoch 228/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 237.5007\n",
      "Epoch 229/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 239.3472\n",
      "Epoch 230/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 240.2312\n",
      "Epoch 231/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 237.9582\n",
      "Epoch 232/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 240.1086\n",
      "Epoch 233/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 238.7331\n",
      "Epoch 234/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 240.6746\n",
      "Epoch 235/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 238.3787\n",
      "Epoch 236/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 239.3406\n",
      "Epoch 237/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 243.5210\n",
      "Epoch 238/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 237.6020\n",
      "Epoch 239/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 237.4650\n",
      "Epoch 240/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 239.2625\n",
      "Epoch 241/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 240.6905\n",
      "Epoch 242/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 236.6201\n",
      "Epoch 243/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 241.4246\n",
      "Epoch 244/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 239.0501\n",
      "Epoch 245/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 236.6989\n",
      "Epoch 246/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 240.1033\n",
      "Epoch 247/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 237.0515\n",
      "Epoch 248/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 236.5037\n",
      "Epoch 249/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 237.6982\n",
      "Epoch 250/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 238.2301\n",
      "Epoch 251/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 236.5782\n",
      "Epoch 252/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 236.6457\n",
      "Epoch 253/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 240.2385\n",
      "Epoch 254/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 236.9663\n",
      "Epoch 255/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 237.5737\n",
      "Epoch 256/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 235.8069\n",
      "Epoch 257/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 241.3255\n",
      "Epoch 258/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 238.4302\n",
      "Epoch 259/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 240.7158\n",
      "Epoch 260/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 236.6577\n",
      "Epoch 261/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 236.6547\n",
      "Epoch 262/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 236.9681\n",
      "Epoch 263/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 236.5363\n",
      "Epoch 264/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 237.5842\n",
      "Epoch 265/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 236.1134\n",
      "Epoch 266/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 235.5474\n",
      "Epoch 267/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 238.6774\n",
      "Epoch 268/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 235.3751\n",
      "Epoch 269/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 238.0874\n",
      "Epoch 270/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 234.3075\n",
      "Epoch 271/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 236.2804\n",
      "Epoch 272/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 238.3350\n",
      "Epoch 273/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 235.6055\n",
      "Epoch 274/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 239.7922\n",
      "Epoch 275/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 238.6589\n",
      "Epoch 276/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 236.5212\n",
      "Epoch 277/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 234.2353\n",
      "Epoch 278/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 235.0096\n",
      "Epoch 279/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 234.5622\n",
      "Epoch 280/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 240.0413\n",
      "Epoch 281/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 234.9432\n",
      "Epoch 282/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 234.6775\n",
      "Epoch 283/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 235.5709\n",
      "Epoch 284/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 235.0451\n",
      "Epoch 285/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 236.5394\n",
      "Epoch 286/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 234.4927\n",
      "Epoch 287/500\n",
      "336/336 [==============================] - 1s 4ms/step - loss: 235.2635\n",
      "Epoch 288/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 234.8190\n",
      "Epoch 289/500\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 234.3367\n",
      "Epoch 290/500\n",
      "327/336 [============================>.] - ETA: 0s - loss: 236.5731Restoring model weights from the end of the best epoch.\n",
      "336/336 [==============================] - 1s 3ms/step - loss: 234.9301\n",
      "Epoch 00290: early stopping\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x7f5638629a10>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_json = extraction_model.to_json()\n",
    "\n",
    "with open(ROOT_PATH + 'Models/' + modelname + '.json', \"w\") as json_file:\n",
    "    json_file.write(model_json)\n",
    "\n",
    "earlystop = EarlyStopping(monitor='loss', \n",
    "                          min_delta=0.09, \n",
    "                          patience=20, \n",
    "                          verbose=2, \n",
    "                          mode='auto', \n",
    "                          baseline=None, \n",
    "                          restore_best_weights=True)  \n",
    "\n",
    "checkpointer = ModelCheckpoint(filepath = ROOT_PATH + 'Models/' + modelname + '.hdf5', \n",
    "                               monitor='loss',\n",
    "                               save_best_only=True)\n",
    "\n",
    "callbacks_list = [earlystop, checkpointer]\n",
    "\n",
    "\n",
    "extraction_model.fit(block_data_set, block_data_set,\n",
    "                     batch_size=4, \n",
    "                     epochs=500, \n",
    "                     callbacks=callbacks_list)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.16"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}