{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Multi-label classification\n",
    "Previously, we classified things into 2 classes: dogs and cats. But what if our image belongs to multiple classes?\n",
    "\n",
    "The process is largely the same in terms of fast.ai, but there are a few differences.\n",
    "\n",
    "## 1. Softmax --> Sigmoid\n",
    "Instead of using a `softmax` activation function to evaluate our classes, we'll use a __sigmoid__ function. Softmax wants to pick one thing -- remember, the $e^{activation}$ means differences between numbers will be greatly accentuated. Sigmoid works better than softmax because it will normalize between 0-1, but it will still allow us to normalize towards multiple classes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2b50231390>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# The sigmoid function aka logistic function\n",
    "# Sigmoid will be close to 0 when -ve, and close to 1 when +ve\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "\n",
    "def sigmoid(z):\n",
    "    # Apply sigmoid activation function\n",
    "    return 1/(1+np.exp(-z))\n",
    "\n",
    "test_input = np.arange(-6, 6, 0.01)\n",
    "plt.plot(test_input, sigmoid(test_input), linewidth=2)\n",
    "plt.grid(1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Read from CSV\n",
    "In our previous function, we were able to use a keras-style set up -- put each class in a folder with that class name. Since each image can belong to multiple classes, we'd have to copy each image to multiple folders, which isn't really practical or optimal. So instead we must read our data from a .csv file.\n",
    "\n",
    "## Note on one-hot encoding\n",
    "In single classification, our data might be one-hot encoded. Imagine we have a bunch of data like this:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([['Dog', 'Cat', 'Frog', 'Bird'],\n",
       "       ['0', '0', '1', '0']], dtype='<U4')"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# The second row indicates which class our data is -- there is a 1 in the column that represents the data\n",
    "data = np.array([[\"Dog\", \"Cat\", \"Frog\", \"Bird\"],\n",
    "                 [0, 0, 1, 0]])\n",
    "data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Obviously it would be a lot more efficient to just enumerate the data:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([['Dog', 'Cat', 'Frog', 'Bird'],\n",
       "       ['0', '1', '2', '3']], dtype='<U4')"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data = np.array([[\"Dog\", \"Cat\", \"Frog\", \"Bird\"],\n",
    "                 [0, 1, 2, 3]])\n",
    "data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So if our data is tagged with `2` we know it's a frog, etc. Note that even if our data is tagged like this, `PyTorch` will turn it into one-hot encoded vectors.\n",
    "\n",
    "More info: https://hackernoon.com/what-is-one-hot-encoding-why-and-when-do-you-have-to-use-it-e3c6186d008f"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Doing the training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# First import our data\n",
    "%reload_ext autoreload\n",
    "%autoreload 2\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from fastai.conv_learner import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "PATH = 'data/planet/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "__MACOSX  models  test-jpg  tmp  train-jpg  train_v2.csv\r\n"
     ]
    }
   ],
   "source": [
    "!ls {PATH}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Recall from previous lessons that we had our confusion matrix showing false negatives, false positives, etc.\n",
    "# f2 is a function defined by the contest to measure how good our function is. You can always define your own evaluation function.\n",
    "\n",
    "from planet import f2\n",
    "\n",
    "metrics=[f2]\n",
    "f_model = resnet34"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "label_csv = f'{PATH}train_v2.csv'\n",
    "data_size = len(list(open(label_csv)))-1\n",
    "\n",
    "# Generate some data to use as validation data by taking some random rows (20%)\n",
    "val_idxs = get_cv_idxs(data_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Set up our get_data function\n",
    "# This sets up our transforms and returns an image classifier which we can use to browse data, etc.\n",
    "# Notice this time we can set top-down transforms since we're using satellite data\n",
    "def get_data(image_size):\n",
    "    transforms = tfms_from_model(f_model, image_size, aug_tfms=transforms_top_down, max_zoom=1.05)\n",
    "    return ImageClassifierData.from_csv(PATH, \"train-jpg\", label_csv, tfms=transforms, \n",
    "                                        suffix=\".jpg\", val_idxs=val_idxs, test_name=\"test-jpg\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Now set up our data using the function we just defined\n",
    "data = get_data(256)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we will explore our data some more. There are a few concepts from `PyTorch` that it's worth knowing about:\n",
    "\n",
    "* `Data set` (ds) = a single object (e.g., a single image)\n",
    "* `Data loader` (dl) = a single mini-batch (e.g., of images)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# So this line of code will get a single mini-batch, convert it to an iterator, and then get the next item from that iterator\n",
    "images,labels = next(iter(data.val_dl))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\n",
       "    1     0     0  ...      0     0     1\n",
       "    0     0     0  ...      0     0     0\n",
       "    1     0     0  ...      0     0     0\n",
       "       ...          ⋱          ...       \n",
       "    1     0     0  ...      0     0     1\n",
       "    0     0     0  ...      0     0     0\n",
       "    0     0     0  ...      0     0     1\n",
       "[torch.cuda.FloatTensor of size 64x17 (GPU 0)]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# We can see that we have a list of vectors with a 1 where they should have that classification label\n",
    "labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([64, 17])"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Note too the shape\n",
    "# The batch size is 64, and we have 17 different labels\n",
    "labels.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQYAAAD8CAYAAACVSwr3AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4xLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvAOZPmwAAIABJREFUeJzsvcuubNmSpvWZ2RhzuvvaESfOyawOlwYNXgE69EpI0KJbxQNUiwfgWapBG4l+Ski8QbVBApWQUGWSKZ3ME7Fj77Xc5xzDzGjY8LUj65DohJQhIqU9pLgsX77c5xwXu/z/bzYlM/k6vo6v4+v46dD/vy/g6/g6vo5f3/hqGL6Or+Pr+KPx1TB8HV/H1/FH46th+Dq+jq/jj8ZXw/B1fB1fxx+Nr4bh6/g6vo4/Gr+YYRCR/0pE/ncR+bci8t//Ut/zdXwdX8c//pBfQscgIgb8H8B/Cfwl8G+Af5mZ/9s/+pd9HV/H1/GPPn6piOE/A/5tZv6fmXkC/yPw3/xC3/V1fB1fxz/yaL/Q5/6HwL/7yc9/Cfzn/9Cbt33L64cbIoqp/L3fpQhEIiKICEGQkWQGrRkigooCAiSRiU8ngZ8GQyKs9+VPXlMUYbrj4YTH+jvhGUkpICKg9Q2qCiLk+kaATP/ysSJfriUS1nXXbeX75yuCaP0Okoysz1OBTEQVMpnhKLbuzRExVEBNSIRM1rUmGQEpa65AVCBZ3+nrYnVdc/2o9uX99S/IqM+b7qgqKgJSr2ctCh4TEUMQlLpPVaGWQkmPmk+p361LJNNJdF1n1NyoIpp41JzV/UjNR9S8qQgqQkDd9/v9rvUBRsRas5pDEtQUaxCp6/0BCFE3hOrz+2teIvJ9GUVkec41H5mo1HzIurQ5J5mBoojVHhPkfR0zYMas1/69fSPrk7WWF0mBFJIg0tcv6n3C2r8iiOaXz8s1X1nr/XxPZhLuzDHXuVK+/8P3f5uZ/4w/YfxShkH+X177ezmLiPwr4F8BXG5X/ov/+p/Tm3K7dMTqwAI4wvEYNG3YZsycpDsS8PKdsTWhNQM10mGcwR8+PnCvvRHhmCZdhet+wzlJoFnjuu106fz+99/z8fMrn358I0NwV45wjKCL0hH0pnRRLtcrNCFt3WQkj3kinog6iIF0ksTGgVvDWuNmdbzPCM6h2A6mDelCRhJz1oazxtY6+9Y4z4MfP37iun+DNOMx3mit00y57LV0w5MZg5HBhuJviWpieyC65mAqyCAJyA6+QQZOsH1oNIPeBDMj3TgeB+Enbw6XrdH3DTE47yfudaiP8w21K6bGFid927hcdvpVcJ0cnw8+/+EVuiLbjrIh3vj08a8ZvhNsbG2w7zv7fqVvJ69vg89vB+c5QKGJYa54Dr657Nx6ZwY8CEYm8zjweyDbxgj4dH+tv1EhPfE42a/f8tvfCu5X5nkyzjciOucZkAeXS+PDNzf6vgPweD0wVdqmmNVhRxzPyXEErXV6a5gZ4c4fvv+B4zy4bTt971y3RpdG3gQ/En9zvv/4EdOGbRdckwwHTyRBd7h8cyHPCacjGEly5uDl5VsmwHFgmVjf1v5w+nWj750AzjlRhK11mjUEYRzOpx9e+au/+RusGft143/5n/7i//pTD/AvZRj+EviPf/LzfwT83z99Q2b+a+BfA3z3Z79LlbKODrQUCqYQJJ0ck7c86WGICs02tm1DRfGcxHQyHV+HOjGQgEw0E4149wiZiZmxbca+Gw0hs7wbLkgYBrSfJFkBNBdClOmOIUiAa+JAJKDlXSSXT5MEaU/rgQOI4Rguk6YNcHIkkSxPCJZBdsUJIhzcyxu3RtMb4ScTmKPui3BCZ3lm23E7ywDULBACWEDUdSrrZ75EKAF4KjoBD+IMYsL1ZcOagCRRH4FoAMpl31FroIJ4og20CdaMEUKK1+G2ZO/JZqAJbx/h4ZPAiK6YdZBGDMfn5DyC4wxQ8JYYRqPe1/YOMbGZZILtjXNMXIRQ2DB2U8KUqUEkuJSP3rvQxdDsHGcDf8UMTAxBUZKIoOModWhl+bIkSYFdFAcGTkTUYcWADbEN0ogpZIPMwD2YI7ET2ibsZkxJ5orGPIJmicXOOZ05B2YDazub3ri2vRwJg9MnKk5LaB5ggezQuqFdIQQV4z7vdDrhgqRCKMd0zuP8WQf4lzIM/wb4T0XkPwH+CvgXwH/7D71ZRdhaA5L3qLyBmdYPXTjur8SZ9K2x7x/YbzuZyXkqc07cT5IBgGAYsjywIBiK4WFIGFtrXKyxi8FM5j2Yd9DYEDXMoLUKV2UFy34H7zDuTgdUwRVmT/aYYDu4gYNpHZC+BVNieT9QrAyOHdxu38Kc5OnMTFwUsdpmfh51WN2x205sCumIB/E6ySvEhwYDxKEPAxVmHtBA1GgqNOuAgiqP+ycyE7eaV9EyQrosW+ogTOlN2K/JHFSE5YNzBOGDMt0Vy92uO0jHdeDHiQkVTkcn7m9sCr/5D/6MD/uVS7+wZUNCyTn4m+/feDsnxsCkDv6hcJCckRxzAIm7oHQu0jl3pbkyPRlzom3HWiM+OBdRrkP4/aty0mmi7BbInDVHR3B7UWRvWCq4EQT7ZWe/NqwB4ZDOdEMBHWtvBkgXdm2ckmyAujMSHp4IxgdTVBocZTj8ltgZ+COZj4D9gm2N620jzRkO4xTck90aew7cJ8OV0E7XjgncY5JAIExTxKQMsCpYgjhmigL3zyd+Jj/6G5dt57Lv7L/d+fP5W+7HSQz/WQf4FzEMmTlF5L8D/mfAgP8hM//Xf/APJDEttyn5TCKC1LLorTd23zAcBZxk4ogbOZX0Rvoks7I29aj8V5SfQhZzDEzLCc+RHDHx4ZyH47O8vxAQSqKVZ5pgoszTWUEHvlLVaCBhgBOVaVfEEFQEYeBSf+QrRKzI4QvWkQo+g5mOWkIz8Fw5gNKakuL4cOJwBGhamMDMxGeQdeG0F4FW/k+AHOW1PEDMEHdYOI4aqBtMxSOYavU7EbZdUa25jlw3HIrZyooTtBJqQGsuCDwCi4GKISb0a0O3Amc8JkSyXzovLx0xMBkoXp7ZHYugZ1RksjZO4So704PX42T4IBNubeEhPBH0rHld6X1DmNIAwXEODwRneDD9rD11abTNECtswTOZKxo0gJHQhT4E6UrqirRS0QhiRYemkDEhjDQBrRTICOamuBtTjGMGeDKpw44KTsNtgoFYBzrTlcGgG6gZve30dGwzTBWxQdozKhm1JHOSI9lkp9NoZmiD/VshPgvzZ570XypiIDP/AviLP/X9hR8JtpCI1CRwNjNaM279ClmgTEzjmEablTrg1L6IAiElE8UQa4itUNAHcT5ghboMGGMy7iePcTLcOdNRN1wMZmANGlag2i6YUSH5gHCtqIRKbWbbCY4FbiWoMzSYNMDQBK3jQ6Awg0ypkDodPyescDzU0OFYOGLKzJPpBSbtl429V2TyeEyOtwMCVIXr3rBrK6DVASbuyZzO7brj00AV25QmBRQeb+CuuECq0NTYtA4CEU+4FFXQZpBewJgZSN0XGIkyiZW7XPEIPh+Dc05a3lGvtCgluV164UIz0GyQxvk4mMcEz5orW4BoBmbgOTgPmD5ppsgTYDuS2QIPpWnQzGhWxqJrr2uTyXAnPRjnIB1a22i90zqkxgI+K03KhAwv3HMIuryyGQt4rSVuUS5hkjQCtENr6AaaZUj16mVYgHNMUp0gy4CZrPzOsGZsDqQyJgwH2wt03brRUawJLkmiRJRDeYLMrdV6bVpgq4ojAr0beTW8/zwC8hczDD9/RB3mL1g/KQEEZoLZFdw5fTK9M6ehcSzHeyJxAA7ZUG9Is4XgC8Eob50HcGHGJGYiHoy1eWc6M8sjqy+nLmCahFUO3TowYE4FFUyEpkqzjljHc4LPAv+sgsAnTq4kncRZG2EqKUpGou7IOZANZCsgNQgswNyZFAbiQN+V1o15JOfdebufkElvAg9n32pjvF+/OGnF4IAWQm3CZoIM5RFv5SkTSGULQ9TQyDJgaxOLCmZCppLuhAiqtqIjI9bKTRLzB47zNgeWSouORi1Puxy07UKzCqv9NGYYx31yHsmYSYSSVuDpyKS18ujTnRkTkw1Vxd3hhDPgJDCF3cA6TGCnEHqzKPR+BsxZEUG7YW1FdWulipCoPRAJEoKIEh6ECs2MSdQ7UumZHJqc4TRJ0hx6gybYNKw5aspFCgn34UQ4qVkRnAjNHdMdmtOm19xO4zEMF0etYUGlMJY4g3SFWKmmUjhMV9SEk5NQxTWwFLoatP3vI/9/wvhVGAYJYUyAyW/USITwpIdz5oPLfiNvwafXB488kdixqZxy5+Rg142b3hCCGVk5o+205mw28bzwBpxn0g0s3/AhPPLGaY3szhiNmYPmFXk012KLQuAUfDMkks0CbpUCiJaBuNgLn+OV8PkeSRitgCEc4gBPfFKvtw3ZG+M4GMMJMbhcSQt6g35MXJPYCyK8ZCdaMC152Y1+afz+npwn1IkOBo1v5IULP8Kx83h0cKf1xuW6I62AOTEFqYOMwVs6yIaJ0/NAD4P9O2xzPn28I3pF9p3EOY7AqDC7JRijDJ13Lr0wlBiNh74yDuM8BtMPPIx9e+F2+4ZIRY/J8BM1Q6478xh8/Pha1kwb2hroBHH2HnycJ7dt53K9klYH1w/BMR7iyJxoBphxStQcS2O/gTnM/EAykc0hd0Ym29URvSHuRRNWXliMgU02hSbCFOdNne/0G/r2gTwfjHkU6Np3zMHOA7tcsOY0LWzjSCVjss3JYzhiybZ9wDNxHnW90pmXwbd2wa7fcljw6fXgcb8zHwfMb9m2mqcJlRKN4JyTrlZgvBlsRbnqCXvbsWHw5rgM+tYJb9x9/qwz+aswDAhrw0ZZcCtAEjXcncmJpxHSYQEw1u7MuaB2krAnzwuIEtqKspd1SNvOy7WRHJhvlR/6JD3BQXOu0Lmwglyf8ySR0yuCcVtByuLtzRTnju6VUtCiaCgS97MiEAoQ1A3wCikPDeYcQNTikoxx8ji8UOWVZTuJSLAZWBoOWNZnZ1S2aqZghqlBfCC84SEQBcOiRk7H08FBmoE2QLF00EC6Ib0jTdEMxJ22mCEQ/MloVFDN6/xc1KAp7aUhEhw5Oc4TSI6j0hixjoow5uTTpx+5Umh/YQeBGMzTEVFQqzXUp+6k9CM+JxFGejEkUAwRBJ6CPrUaUqDwmg7uDj0BApV6bVshtwl4JI7jGURQaWCPWmcp4FnV6E/dQhRImJnlMNRQaag6M7ywHgNmhfIBpBhBYkTt8yxAs0DcZLeOOjizIiIfzDgLO6Dgnbk+1p/ak7Uquj5LntoUKTo6F3cSTFRe6Oa4/QrAx587UoVtb6gmWC1sYASNOR/k6ZCG5M6l7+x7su/Bx9cDSSMBTycNwGjygKZsFuyWhBvOyfE4K7T3F3I64/zI/TgYhxHzrS5GqJC+L7ZxU7QJ8joLcUawXeib0prRMI63H+mXC5etoXsjpkNMTutwArN0D94hNogH+DwIgq5C67Vhxn0yIskX2Og0NlIHs53ssrHrzpECc/J4fGb6G6Kw907rO1tL4tg5xmDOA9HA9EK34Hh74D5IVaQ1bIOmG017aTJ0gV+meNTh600BxUncHU8wKQxkE0diwARjK+PkwhzOH374DNH4zbcXVJPPx8n902fGefDthyvaN1AwgeYFCqvOongxdEG5qsuIxax5ogBCFehbQIProxNRBsINmimbrZRCoJGMLP/hWLFdDUKFGYP0R9Hd2GJz4J6DIwdXu/Kh7VgzMLgfr2tuE2utTGYvZvw+HnQajIbg0CdpG56tTq92miqZTtDKCDns1xvHPDnGwf2YPM5JRFHqAswJOZfmoVXm9ozaFEEc5IQFZXHLcliuwmkFRndVZIml/tTxqzAMkJXDavHRT5gkA2I4aQ1ykuGICbBBXleEkWgsHsPjXamoahWqJqQ37o+D7394cL1ttAym8/5PAIjSrUQ+rZW4xUQw+aKCy6z0oZnQTWgqGMJIIx9ep1sEVYheB8vHOlReFlub0nujeR3ATGG6MlceK03ePUsDCGUgzKWSLFlEFAhIUaOyDgEIHk76QcSjMACK7Smio0ArTSFDSBW074TU5zGVJPAGWFmwwAmHyEmGEGaoJO4Hl9bYLjuSG+Oc3B9vfPx08vZ6cts7e79yjDvjPvEzMBSVDbG97hFBdLE/WtGJrntO9fdrbX2n2bYUmBNVaA2yJde97umMJL3irKJ+kn0XWlKMk9d+eqoDT5wWgUTpKGt1pCjVNV+RS0thQTTFH/UZGU6EsKw+Iuf669ofpqBaOgUp8SkiRYNXHNLITIRgaOIjOKIoUNTYtsYzWMsEz+C5RCKArqg2Si2qM9CVJl5l0qTjapideAw8hak/76j/KgxD+afEpEJlIoqKiTo8gqM6QSeJ4WkML/mv0TDVQvyXDDRwIgdEsRvHGby+HXz+fJBq7DJIFbL8CWig1um7Yl1pvcK74gzKOvetEV4HcDOlm2GikIpYw4+5QrcCENXa2nQOzJIkeyJi9Ca4gg+IRWJO4t0VCEtCnU5KkGG4SYWeIhUqSxmxxRgSeNGoDpFJqlDLW5swl4RWtQDIklxD6w3P2ti6Ep9UClBcysmn2CrSaCw6MJTvrt/wZ7/7HXE0fv+HH/j92xuf/3DgOGxgauRQ/CxbtolhsiFSykvLRHOpTLatwvMsqfJTKh40zGre3lneqP0iFOruiymafAmzJVcILmtOJhhSLFMXzjgBK/o6BRMjzWA6MhsSFbW6B2cG5jWXPKXlBEuGthinZeRsMSq0ohH95ElNL7UyIlYS8Zg8/CTm4DyTcECKQrVlJGU5mzooS1AdQghMXdS4w5Xaj3tzeuuwlTrz09udGMbPlDH8WgxDLpWiMJdWvkR/TmkTKydE+7si0nNUiMUX7x4FhXF3x2Oic9LcOefG2/0kQnjcJ9ID672ALtto9obIxvUi2GYVvj4cXLEULBPZN3xCl2AXpenStQdYM87zJN3BoLnSRoKWQo9mmNfBNgfZslIjq5BZrPLPcF2SXiMiS2GXQDYqlAqs1eE0a2irU/I0LSNmKeAQhA3RkvniR20wWyxjE1SK17dF7aUopo1mikqUelJ10Y9AlqZGqSjp2/0Dv3v5DX/+zW/5/vEgjiAewDRa7xCJH2NpUxrqk2SucDAqpPZY65a0vQ5S5Wu5aFLKa4qTOO6lYUkgJyCCo0X/6Qq3hfK2WtGgJRWNRSUoNCnJ85jQGx5a2gOpCDMtsdmIEHIoZ3N0BB2lxPGy8vtYRmopH91Rk3c8KUOIOfAxv3BTVjGRSyuoJIN5DvwIfBS+xUrrrK+aCAUt0GrRtxRYvVa9HE4g7CWy2pS2N7bdMOscRzBikPPXoXz8eaOI49IIaOVNeB2qTXZUz8phZa9ESybEWdZdbVE/JXVJ4GRnRslsbZaKb85ObxtjvLLrwJpicsHUCS35ct8E3RRTw+VLuQ4I2ooiarBSiCVjXspn1yIIFMADk6hcUltdu4JElMwpJpG9cmnzOuySuDe2lSwuYooZYNmwhBYlczZVVCqkDyYhBXhGHCgbQG0SsaXoewO9YlCCLV2eKJ1GkpZEE7T1Ej9xALHEWUW1VnpX3r0h3D7stIvhTP7w49/x8dNHHn5CM3pzwp1xDgShW2fEwZgOLUhzIueSHpXNawuYTSnrkFHrkipETnyR2RFl/N0FTDmzPKizwMrCWotROoFFQ47lbM0UutAepVkIzaVHK23HXKIinUVJhjtzOpaULkH0iUcvYn0r4/BOnbNYNcPHwMeApSFYWjcUIVJLEDUf5AiYSUjth5KRylpHfqqHW59fCGdF115pc6xUvG1oE9oeYFoRWs4KT3/G+FUYBhMtXnrhr1MENNFwhp0Qwm47+y6ITcZwzpHItQCVtIKD+7zwdk4+Pl6RmEjqyuEP5jiJ+8F+Mc7Yidmx5txaclwvlYPOKLVkF9rLBjEJP8mVAtBnKS4xthSMCVYerO8N9bUBTJgKN6w2dT5KZNWEieCZSHuj206XnfTJjIT9At3Rs/E4TsZ5IB2+ue0VoYQvXv/Ch6sTx+CYwnkqb+n0OLluwrZv9G6oVZ58xE6ooq2zv+xYKufnAzjJ776hzVkqUlFsS0S+5TF+5O0x4QCmMDFaTF5a4zffXfnNy+94fT3467/7S/7dX/01nz8NjqF0MQjDXjZOn7y+Hnx6HNAH1724+OSO+8BP6LohTfju9h2f7m8cOWn7xiYC7si+8bi/MV04Vy3Dtj1FTINHAq5VgdoFccWncRfhzMnpQUP55mJYK3rZHXK7Mn0UjqEKbeIaOCecTp6T04rVsUfj3gZvnNxa48N2K/X7LDjaro1vXnbgxHRWHYUqlx3oyhm2ajIuaMA54XE+OOJ1AdGsituKUHElFHbLql+JVffhhacIr9AbkrlojuQIKVXlCTmd41DcChQ95sFxyh8fvP+P8aswDMkSlBQnhGgh0oZVWauUhytvZ2RXAmE8JmaV26Un5+ncj4O388RS66DOKgEmk4YXmvwMZx1Egr03aMrhzjMV23oVVHVtYMGZA3BUg2fBF5SAapOByIZ2KX5ME10hnj9OfB5YN/r1Btb49P0P7Nf9Xcr1/O/a3pyZ72XKhFZ5dqv/N2+4l+ioWyc8cZ0r382qsBMDL5wGKTpyv9xWGgHMUZWgHpi0qigdRX/JJrglh09mkedV4ESwb73wFxGOz8kf7q/89Y9/y8fXgXpfSd+i6TxgE6wJ1pVMA0msVd1KKpw6cLwUhHFCTjQcnaDaSghmtoA+LX2lVfFcehJesqpcmyhGLiS5PG0eEBFc9uS6daRtHJlMHzAqYtAlLxcSxVd9zcpqnmX4a22qnElQ2yjloi8sZMP6ZIws6TzOt7eOyYXMHT9fQYNYjoOsIiw/A7EdkwlZrQXMjLY1tBma9+fGqD+LKFAUx6IMCRgWVYciHpwzMUrmPwjejjtz+D9NgRN8SS8zlzKtF27QrJOeqK4VLzi78t8nSkvgczKOg8fbgxheqrUsCi2zRD16vbFfjYy18hlkFlOgsmoJcolc2BeoWXo+D0eXzqIwAX3uwZIZyxJE9QIR5RR8sSQmVXptvRPNaLq96xv4kk0/s1ecqm+AAsyetQ1YokOK/HieBrwYknZhepSaT6tGIjzqd2Jcm5SCU32FrKwYVXGHMQok1WmLBitlXn1RSdFbu6AC5xz83duP/O3rZ3748QA6tnXEi8sf2hCrlKxvcFGYXkdLU7FFSSIrRRepwxqOuBMzCQukGT4HoEgEunQFWON0J31ALyAzMkp74rFSi5I0mtX8MaN0H73oypnLSJmucL0mVTfBZtK9FKnFmUJmMGbSCazNldoa3Z5OYhDh+AQ1ZdtqHsPBeysGyZ1OoDExT5pvuCm2FfWpposRW8bpKGeYldUsWj5xAVAsdCkfG3OWMCut1KeSXkzH3ZlH4vPnmYZfkWEoXse9ADFQpBmbNnxUAU5GSYo9EveB6ESsrb9LzlkVeV32ytW8iigylrql74XCRxJn4l6NQTpBqmC6gM+5HK00cCldekwwQ1rhBSllVAQIOqmjSpJ1UaYo9+lcrNH7BdurBiMdrpcXRj5+cu9ZEmhW9BDVUyGWfNlMkVYY/ZN+NBc8T2YONrtwu3zgbQywxXHIQvyl0cTwBZaN4WhCa63wCoPoq+lJmyDJTV7ALqTNipaAmUUFjjkYPvj44yfuj0Lwt32jtVoDRk27Z+DTS3OwOHQNf/fUVfhVPDsrAmtm6KLvgiVXXr0GCuKHCGNqY8TAI9nWGqsWPlXRZ5bqU4LWq/T9OAYhA1VjuzRcqL4KvRU2QeA5GFHg7LYZjDLU04S3wwkvJaT5CWw064VHxGR41azIMhYmkLkwIt2RXFhJjJJmI6juJGdVfa7qSXm6yDmLfhd9ctEEiUfgarSUVfBdUPPdJ8w7N4NYzJWiMAwfk3m+u7E/afw6DMMyZg7ELBoyWhkL01ac9pzkLNQ/fZIetEsUp75yYNcOm9G1ozEIPypnzFlUZjRcdyImI8eSMCepDcNpqnhm1dlHXZjP0u+nAKb03nlWbUk86wC1aiPEKxWizHumg+zYtlfTlEomae3COY5Vi7CoxkV/5drkKVWb0Tb90hMhhVO8Fk0KqJvhCxBtXHbltM9kFpvTaEvLUGmSBKQ6DeMixt4aIYOtg6oRJmybs4sCjU9vZxnIzEpNEMYZvB1vfP92IrNj0dn3hvZEI6E14rWilSOy1JhC/c4Tn2exRzELfRcr1Ws2uinRpbjHZ9HWmuvwWfuk1ElV+CQwYlZUZUrbSkpPVFGUdEFsAdJRtPTmpSA1ldKTtEoL3Ac4BZBSlKN5I1BEgzkconp3pK8o76mMxRdDZlX0p636LXhyRrDrBlZalumz/mEjesdyVn2MlZIzPSpKWrqXL6H0wugp3U45n8VKRXJOB51sOK0Z2oRMq8LCJ7P0M8avwzBQqWEkDI9C4NcCkhXqMkufYE98fFF3ZAmP3OtQWVtt2Z6Euwsxq2IOnTzOQc6Bj6M2m1MLLsblekFGdYjSrNUYfnLMgbTqKNS1lcA1JloljDRZzWEWzZpZEuISzlS+Xj5x1mvuPBmUZ71lxQxVf28LXRfKOGgv4REOpzunO2032tixw5nu3Oed/cPOOU9MlK0ZTTsRMHJAFtbiq0Y7u5CbMeaBaYG7yEbjwPPE1fHHQF2I0/EZhBtzBq+vEz+tKEcBfRoga0QXbDyYJ/iZWK9D9kycpldqNqIA40orhPNwNpGVIgUhRjMpDQK860DCiyFpUvM0Y76XeW/dmFulIoSTshWjYGDXpd8Qw+9J4UOVSiQJM3CfMM9KcXQZLBFQRT0ICSI2nFbeeDWtibRKF+1ZBliS8CAIEpW+urQJZ1ZPCbpXUd5clINXAdX0SYxZ+M+uyGobV5tllXkva5HAfIpNUlCHI4r9ilQGFdWaQfunqHwUwB5OVVI5aCenwl3wLbBpHNlxTzYKJcvVfCV84tM5jsl5DtKD49TSKrTALhumDZ/K/fhcdcY79E3Zu3E+kuOArQnDDvbN2Pedt3NyjomS7C2YbavFGdBMiDAiqwfB5TabXz4QAAAgAElEQVSx1iE706sGwueEpoRVGlOqRcXHyY/3H/jmw7dA4K5MOrM54q/YEVz275h7x6WRPeE8sNaxtmPinOOAbdLX3hxj8joO+hF82K4gS5qVVRFoujP14HGefP70AIT5rfLhqticWLsi2jjOyadDOMf3TDt4/WHQ9guwYVnqwk/j4A+fXvn2w3fIqM32aU5erLpe9Zy8WMd9oFuF62QjWyK3iZ7BgZLSkBD8QYkN/Czwd2ZVL3oy3h40SQ4/aQ6gzOakDm65MfY797tXOI/RRuFP0SkO1H0dwsZ1MTWZwZzB7dJIvzPu90L3Fwn4Yf+mcnN1UrW+9zw45BN77qQabI2+73gEr/6JbdsgNkRbKXOlIiOfSrrw9vKJD+PCdf+G9ttJzM7jmLwed26t4Q7HObmP8Q6oNmtcW37hON0xnG4w24n4xjzgrifWgm98Z3ry5q+8SgHyZsIOoPmzD/qvwjCQ4FGW11EIx/PBjAl8W5JdRjVxSQNfzVykauxzBOKxPEpwprN7eSOA0wczqj6/iyPSVp2DIFGqQ9FguNBK11TNQxEiSmTSoIppFvZZTWttORTHZPV6knrdA1QmKomILxlrpUJijdBAs/pEWlTapO2GsoBAq3ZgbYXikRPmZEapF/e9mpCcu1epN8k5lNMnkU6mInK+97PYXvZ37+wBMX3RbYq7kD45j5PjPrl/dgbOOY/iwrP6G54OEYpqL+Wh5vJy9d30mjNVw3o1ttEuEEpYlBZkRX/FGVZkhBjTbFUeVtqICGJG7K16NqhXiXpOiIGrExzs2lBg+skjA7yUiA0lmzyj8FKRJ+8pgOgXacAikIq6ZDGA6GrSI0g29r5X6CIOfiJRLJR7I1zY+trIURGrRylIowmdVqK1p/pKtHQHAcf9qNTZq8FQRlTBHZR6kqenL+VEwcVf8ALlSwmBUE2F/MlzOUu8JT951582fhWGIUnCB6vbXvUHcIdI2kXf0XvwqotILfoojmquMWvVSxj1TCuq+LmC+UFocpMLTaqacWsXNjOGTt78tYqwZquwcxWrPKdfUXZ7AjqsKs7qHxFacqRtq5576ax6DiXjgUrQtLozeVSjkqcCJ0oDi1hWX0PpBVqOT2xqdBpdq6OxRInAAqU3rbRHjXE44QcxE/fJeR7MWeyAmRG70d3YX14ogIZqlOpJnI5eqgrVPRjz5PX+4NPHrKY1LUiZtGp1TM7CarbVOk4JRjjnHDRRYFvKolIX8tSiWjE1YQUQelbYTFZxUiHuQnqlTHWIS1wmUmlbNcYsrEPc1z5vXG1/3y9jnHQROr1A681I9epilVnViWKLl4VSYNYOiQhywNDSqris8Fur3+bed0ZWIxnxgcRAZCexasu5U3O09DBpEJtAgy0attIWZzXo8YpoPn++s2kHQPInADQlt89VJvCFk8slOIt/T2iVRXdre8dYKkte4KX8UzQMGRTrWtkTIUyvw2OM6tq0POXT/EsGU7yAoCyLoJrgsGl1ys3lAroZjQqn6iUpcMsqig0KGBKsIgaH5xKIJiFZdRGUfFekilaqmajQUPbeKpdOsGGYOI29MGN5UpCFJFSYW94jTVGppbdFMxXAUvMwpjDE6Ko0UUyStronYY7uG31Wd2zGQZ6TeTruRX01k6rnX4VEntVHKlNKbm0gKrXBVTkzeTtPRs6qFgxWPYDj4w5Mmvl7PQlkNXvxjmkZtgLigHVgxVgGPRljluHOmjsH0itrl6dDtcJDJJUeyjEnOZ4NVhoWDdiqr2VbYKNbLZwp2hpmHd0Kx1kyFjxK+VgV/UUJeybDozQQ8wkEO26JGPiKGra2ERgRZ6lJvUq5g4RVnZnkitYCRBZWVLhDqZ39/dDOcM4xedwPctdlaCl61ARstZFfWIKrr3LvqM+rk7PcX43qbioFaqdDZjW8+VJl9yePX4VhqF75dYMFIZYdzDDcDyySpJf+vFwe7rMMQ5SIZWFfaCgXislwKx7/IhuC8JgDn8VPmw/C5mr7FYyRdAm8rc5FQomAcollAEhm1ibYVs2BaadLp2/23hNR1RegdymiLavF96SQeSNQH+8cPq1KfrsHksn1wwsyVyfhY5a72HrJuFe4eIyT4yyQTvYNLDEfJRFfoWOm4aGYw9t5Mr02mVqBdapC9a3eQAzTE6yRdizO0VY6V1GRjEGzar2v2t+BYHsCaBVrIzR2K4YpvOavKUgYb+OtwFfp1QNClPDBxlNd2kCSnKVpkBHkqGeFGILF6hoNKBuu9+qOJrp6I7CQyUTVaC2qOlG++NaniOmZek4vGlAdrNs6YOvtImCdfU9mTHyuGgkCWXFlpDMnX56j0ZJmy5AHpGn1d4gT88Tn4BgnD5/MLIdl78+MEFIUl1I5PnPXkl9VkrBhVK/pXGRlOR5jSV1zWUKerNlTpvWnj1+FYahCqB2RJ6MPq+YaZoXnng0RZTIhqnryrD4llYAIaKumHZpSYWurAqe9VWux1/PBjMIj1Ef1i4xZz3Xw+pufPojF8qlIWJW8kRX+Z4XJZp1mwmYNayW5FapXpHWQXurGiEROZeiqAJxJbwdTy0tDbaLrviMIl90Yj5P75wfneS9OXpMhja7VvPW4O+f04vulIhT0xtYGbL7y0Cq08QhilMeu8najt9Up2pOGoKk0Ok0bu1UVouTJdDgRNpR2BrptmPX33gzWeinvkPd92KyxtYo+ZlTLtaZL8ThWJyEritRWRWOJM9bEaR2+jPKSTRoupTJsTTENMt7A4cgHWha2nMOijV1WiwJTtjTACpB7PrBlRZ8+kznLODSEy74jMatN3/LHYsq2Nc4zK8qTp6Kkkt8QiFk6E6dqHTZtmBZK8KyjmbNaCh7n5JizUjAT0oSwMkZZuGvpTmhlovJLypCrnLZSi8VYUA7VWGX370ZgYXFLzPdzxq/DMEiwd8dsp8lldcMJJII8V7NUGQWkrNs+MY63WdRVk4oWmhIJn86P7HLjxS5su0E7eYwBh9C0KBxmSeCMxg7Iyv+fwKI1fbfGhi98AGQG1bdjYxe4NMH2B9J2xGyV3Bp9F4av5w64gG/47Ogc7N8GoRv3Y3LcT3KcXML4zT/7hnbb0bcf+STOPGE8DL0b0zfm1vD4npeXl6X8DDazAuuOk8sHo327kw7HCecIZiwyNCa9GbtUc9vdAomD2PbqZ0mwy8l1nvhlB78Rjzd+xDnSMYHbHPUwoH7hx7fveWkf2HoDdp6WOF057KC3YM47j9NpoYSX5qFbGQfTqFRg63Ct5zxUYddAZmlQSEcvL/z2cuMYpee+XF+wnry+/oAdznEm2EQs2PYkzarobFYxmZnQ+sZ7GZhWNat2Q36EOILzLP9bS7Wi0eMstkAH2jqtOTtOdGegDEt0elWO2s7Woqo404kpDKsmstKCmzQ++uSHe5IjeH0kx+ngg5ddGYvwcq+0ZJKcCL9pybbULo4zxKuTH5eVYFS08ZQ6dIxzOKm2ygdKGZlZ0czPGb8Ow6DGdr1Wzf/xqBZeKzcQTcZb8vZa9QaxK7et8WKN71+SHMUCRJ6oNHYR3H6HNCc7YEaLK7jyo51c04itwRYcfqCzQEX6gfeGvRh6UZgH+64kjccJSC+8QrUikpnMODkS9rzxzf6CNGXmwOeosJRqvOICns7e4MN2Y7QT1Ua35MA4qIYtRnIVp3/zgupGzs+QzuvbiYRjQ2E2Bg5X4zyT9Dc2g23fsKynEUXTBX4PzJP0WZ2qptC2K/3SqpoRuNnqvpSCzJ14Pflx/AgK29a4aOOiiXvwNiYWb5zxmZd2xfTAJPjt9aXEPB54TpyTcxhTN2i+emokcx/E4RxLCNbP6rJlwAeEOSuknn3ncQmOY3JlYHbhhb7UjVXU1sTwfeKPe9W05MZ+3bEmhCYjJnsCKdjOyiJ0ocqNOAdtT3resJjM+QlIbrc/J7aDezqf3h6Ew2Xb2UPwfqHvN1CvLk7TuW7GfQQnStfCne5n8DgGe2vcdmOGcLOG9ODhgeTEsvqHJFWnskdiEZyZ1XtDFGwvGnwchI8SfykwE5sdaEwzjI6dymmTSVGehiK98xaTDWVv+886k78Kw1AA0Wq0khUqVb89LdZBlgTZIb0e6eZd6Ajneh5AWjVFSW1M3grFnx3PyoT9rMak5y603uitoVHtvLs0ur1w5MlO0NJJCvxCKvT2pcGXFjwppHoUYjVVmRyrfX2sOoCqiLNchdshaFZO3Vu1fzMN2s7Svj8xjQLMrAnX285w5fXNGTM5ZrBv9fUVHD+fcFVofrdrbaio+aznRgJNkFG9Iwjen78hIhznUX0vU3kcvro0s3owllAiKbWfuKJBsRzXvSIrbKViRRVmODRWhaGtEFlW6bkwbCv2SNezLsPRCMacjNQKxkWKiXCv/pPbBnMQkqRqRW9RWJPpXvSebTTdC2N4Fhut50yuKaJgV62aCAFPqecxnCcWSsOIOPA4mTJJ8Wq+oiUMe4qXSp4fbFoUdpiyWcfPwhoklW/2K81gemJdOY+BJtwuF0zaApadOZVTCnRuUUVbIkq3ErzV/l3KW6xoVCmcKd+5iqwephk0NU6P1WG92K4Czf8JRgxk9dGvYG8ZBXnPkFabq1V6umiYprrqI5zpAeZVjCL1sI7apKBawNr0yglz0wqd10MinMHturNtF16PZDeQJWcN8p0uypyrQWlbDy+tqKY0/Y2RAw3FUguVsESz2APJ5Pkw2eofASy58GaguxBR0mcBnNL4f/h/qHt7XtuyLr/rN8aYc6199r23nu4Hi8YigYAYAouEBIRESgQpAZK/AsREfAWckSBBYkGEQJbI+QKOkOWgjY3bz/NU1T17r7XmHIPgP/e5ZbXVXeUGq3pJt+q+nLPPfplrzjH+4//y6Q4M/p/fPzjHxTinWiN8vT/IqGQZxvTWmaWSduZco0JNHfzwFTP3Cl6TBd15ncxR1Gg8n5PjKOZwOSff1ih5LE1JOh2nE7y9rbl+wXkIC6iF1ufL4i58haag8WUY5heta1tzAi+DUTyusT6jhAhtRGWrVG9SI1KkaZqSq2SOLnyutyRehqclf4Nhui28fNnh698KbRSjgnO+yxsi5SydeTIZTBcHoDmEp7gu6ObzxXA1lDXS16s5ZzEu+SN82jfwyTkP7rcltUc40r5tnNfkh/eT80iuFac4Kb1XaJKW82KamJm5qO1mjoeTvWRDZ6JpMwErmikjZGlcF06mackvuX4VG0MhmXG+xjCu3v41DfAA24Cz1oknJumP1+B4aJTZ+pS11dY56CvySxtGtsC907c3qePKF6FEp8DW5fO42w4uddpZSQ3hDlmwqWNZixzhCdZQvsTOkU8CYQryrgzMJ86SfcNi0Uw5SXujeRAkkaqUwto6/WUGGt7Yh9NbqFKZtubxRpu1OAmqMIbEhJzj4hzJdcnZ6SVVj7APl2FtKfIjjCUFnjOZ17IXGyw7OG3G8tFJrDdab8IVIkj7iW7gNVc3XwlNarvmyzdzwfxtkcGyNEVYGS+MOdUf18KMEItw8zs2BTgLStV78DG27ixJfn2QoOaZUhs2sRnbfEF3ixGTIqqd07jqJBl4dHxr1DVkFhRKQGsEzZyjiqOuj02hkoU+QYZxPt4ZQ3yZZkAeTCuGybRGrWJgCMCWpi0Y8+Jakgu3BAt6C7amSuwl/V6DMv0KCHHRtYENif+sxDS9am0MBZHB9PHXs2L4Nkj6yax1UdCWmzjbJpc/M/ndzYTjODSym4HMXIzond3uXPnOOQcXF9acrcHdb+yUStXl1L9hWIdhFy125rLjIqGmzgnHsbc1CvP15NbQ/VtU2z//ityM6EAa01X2asYN5MS6dHE5xYevWi80QFCgqVexpO/BPhRJP+aTluJ+iDofZDmRk6MuznFxnXNVYCZBWWkMpiLnFXmvX1v0D5ZnuE6c4MWMDJGqVogs3bRDbs7jXJOPUiuUseq9ucDddbrZCtCt0pCpLcZoLnbowtuE9vNtqOYhZmr0Rh1fqTVqZevrfUEigPU56CYTQ3NMhQK1TRMCBQ+tknpVbtfXi+OUDifXSDqrYAylQrkvSz7JxcnBUAFGpaziLtaok+J4yJk82o6Z8bieWidb8DwuzBoWwXGcvD+fPJ7HUgufhG+Er4CkLfAeyuVEIPwMF1uTlKUc7UN7oe27lniPpQdZbSCvTZ0VpvPzr1/FxuBm3LdOoQCUkRNhs8ltD3oaW9uZn4KZRg6Yz0mcE5R/yuxwOjRLeJO+PWrIQrw5vYdi6eYBuQEmt2QTj6xKs/LH46Dm4NPbG/O8qFmSJidcWRxX0UPOu4RD12Pe2/6NQ+JygjreT8DEYchzZVUEzcTIlGFAfsShJX1RmW9c18G4ZNn+5Tc3KosfrnfG9YS6rRRraL6zt2Dvg+OajFVyK/rAXjKeD28EGYEoraqAx7kt7Uni3Xj77DwPGGNyHiVT3NKCKyvOGnAmeVyypGtOdIMr16gsZQazrgnSQBzFdcL97U1jZbT5qOKY9G1jzIdk1i67OzkrXRzXBbXGlUgDMFeP0rb8oFI/VrkzGVQkvn/HkqqImlbCT+aE9x9PqkNrCp1hHIz3g+NTcG+NPTot4ZnJYwzqfAC7KhSW6XAl413VZ/MbPVYWZjMqBkHRBjwfF22xmx/vgx++nhzvB3kqKPjLvnNrajGvAIsi/WLHmUM29GaG5yQGXLuqlYhFUkuldvUIRk+sgq0Cz8ajBsIa/hq6RJvJTONlggtrJrysq+SKLD+DM03y64noHg6iFUuzQAxGOjPWDuwbGxtexjku3q93dnORjyZMH8QJnZ0fz688l5DrFs5sos+aiceflZyy0OE6B+MmvvvMnWBfeRhzAT+d8/hR3zu1Dbgr0bpFk7zXUKvSjZmSheUMog0yh5SQ0WhbJ/nKuAbhU+agYYvZq3yLrU/GXMQu7Yqqtq2Wf+U3ZmG82I5ZPK/F5HRFp9FE3LJy6pxkieMgPkZSpzNfKcuhqsRtEdOmWI2SDb6qfVGRr5mcV9Hf5BZt8HEKGivIg4tiKqJuzeonMLtj88XsXeEvZgo9bjtMZ9RKJzdFtRXJuPQ+nOfA13qSHBqqOfEGdr1hDCqfMAazdljaBjd5KIxC4b/hRObHa7uAsUhc3325rYnHlH/k3rDrguvA2DRxQ7hUxIbbCZUig3ljj4AFOMLShcTORxAyCNyu10/XzfJyCbcFSs6F75hOIUZO4Tu/UHr9q9gYkmLUic1giyYASmuUe3POUpx8w2gVHMCThLbz9llvlpmL6DKM7/aB909c2871/lT4ynajOtzyTSKcSrbYae0LA+PdTo73KecflDN5/3LnOk8eP36l2caoyfP9YKyT+i3vmDtzfIXZORDDbl8OTfNM3q8HSXHbN+69YXVxNVU2RSPYVvf3KqUPjifUALuMSXEeJ0EJxJrCLrb4jlGDcSr5ONq2WiClQ70Ss0ZBHcvluBKa+uBxGHMMRiR5PeGCyBsbwd4LUh10WpGcWo+HUWcw3Mm3kzfuUMHxdSogJTaokzbkVHWRzOckz4lVsHdlYs4DxngQUXy+34nY+cOfndistekl1ne23snFUmVO2tbYuqQxV8Ctd44ojufvmEfxb/4b/zZbL97fv/L99we3t8CPJ9l3hhtxXOyux/m9D7rvZJxUm8zbHQPut4MtVhtEMc6D9+vg8/3G2+7MKZVvzoNxHjxsY9sa7yfcdrjdG+22URwkHeyOje9h6x+U6mCy70HbPnF8fZC9GLFiD4BmG/smx+doq7uMIkO6DK7EdkX5zWsw56FDsNpihcfalCe91ApVHb/onvxVbAxmxmZKJ3ILeuxAV4bD7kSp9Ld0Rk3OOMAn2z4XFrBOs5nUdVL+Rn5IztRfK1R0USUNyvxjE80BGHiltAXWICU0olzVypBRiW0NMzkYz2syHpOzGScvXrqKWTNbbsPxgQjP0kzgmxrudQC8qKsv4dEKflmnXIF65aZkhGlwcHzQx5XaZWyRnNeiya/X2RyaCzBLkOKOiZ36utgd6yGnYnsRYRavrjr66VKRyqZf/Xoj6dmw5deAq1mvOXX4F2JaUutladvLceHx7R2YU6PVi7Gs51QiewqA6DTe1wTgA7fwJKwIH9RX8FPsT6ykKYjENjgP6CPYU47Z6XpNw4IeUxZq6QyfH9ENabFk37k0HakRrQd7fyObErctGn4akNq8bt9CYqA4r8E4JnUtB6pEgPih5G1CRLp2dmZN5lT6uK+xbrKW6sI/eOGt9a1eEGCvz22JNjWpSsnXayr9236hgAp+LRsDS547k246/cwNz+L0gaWxzU1v7tDcfWay75+pNK4TOCU/nqQkvi8wrOnTqlmcj4vYNGeOJkv4ctYcH8ik9U4PmGOSLsvtFo3394uMRts2WpxgcvU53ie2GdftVBJZU3titsQ8cyzjkA+K/TeItRZCXS+KrRAi+Tm95DwmJltrxL5DaQw55yWRGIv7vfT6vDgFJRFQCEPjsr5qIcmtc6pP3aNpAwr1+3OldNVHtkNiIZyld1vCMdi3fTEK5T/hL5MVUfC0ghfAaR8vOMk5cG8fyVNjGlnFwVjelU26ClA8QNsZ05ZwTHmbbuAuD8d8lMaim5y15iVXrrCkHhPDuK3JyszX6FnIv3gai7OxMiGIpo15Jtc1yCFQb4tGeCNsKPYzFL48HkP+GzfXhKJcpq1TWaHyxFX5djwOjucl+8Id2u5Ydp5PeY+3FivzQ5Op8Rrhr7EwrE3fWXaDa/C8dFfWhW3lmIwhx6t9b0sk98vQx1/FxgB6sWXJ3pLP9w0mvL8f/Ok//GfYMjS1SnIW58pR/LTtjDnxHBylBTFxDoo2FWVWu4DIPE6uOZjvzrYH3uX3Zw3snALZ4qJv8mp4vP9IzZNoG82Md38n6sbOpxUKAyM749o4/tH3+Ejubzv9rTF70jbpJ/oIWn0T1eDJG/BAKVtjyLVnpoFtOPNDImQhrv3n2HhrG+MO/+yfDebzoL42wjZyDp7zoeStCb4F26Lr5izm6ZQHnz7fwIvrPJVKFEHbds3Aa2PMweN5cLwfPN9FVmqbUr765myb8entbd0cQbQTPHmcF888JCZ7uUBNnVqvGkiZvULwzTbCO7dNoY9Xyk2ru7G3nbd9WxT0BzMPJqql27YTsXENRPZ6OMccuCdt39nvO/M8GMupmSoGg9oaY5zkIcm0l8hmX740TtYJbEoiCze6N8nXj4PjeXDNC4uOtZ3H+0m0J7Zpw2q5UefLEu5d9njNITvbJlOdao0M53oe/PCHHzgOWa/d28Y+nbzvHM8f9EQSKJNOYoq7sN46RS2+DhdXS1FMskqalY0l3pprzK4s17aLL2O/jMbwK9kYCoy+LNM7Xo3rGrw/Dh5PjRt7Gx+CEPfGZoqYdxNX4cMjD2c6+HUtx6uuFqAKWlCXFqxyFA0i8WW02tpLmZjrrFz4xUL4vSaRJRsvX6i5GeOEH98vMp0tgX2wjeTmjVrKvtaCaLEm8cbLNaRqymquHGgfeg3zoLWSQ3QaM4yZzvdb5zhOziE/zGRCDcYwdgxvsViWL6GdMwmuOagxmMepamIHs9QYcAbzgvG8OJ7SCDhF85IZbUtaD/q2Nmh3aggknNeUDXqTvZ2xjjPXFMDCiBRgCbUqhVguyzBSXpe7qdBQ52IigmUxSoaobYGBIrQV7o05c4m9YlnAvfI1oWxyyw3fulqCExGmgD4FGo6ULZ4o27AtJE/GLfXBx3B7tTECjBUfsFFh+N6hNrwNtrAPj01mLQ+NwpcF/jn1/dGWenM6hMhsIXRajM4prDA0YdWhuaZvwMcaTZU/H+S5ouRXYU7rjUKycpvIJvEXXL+KjUFjJGVVjmlcF3w9Tn7//Y/MWYSJROKpSFBcCyfNyGXv7e66CUuRZTMLG7XkqE7Mwrt8Cr3x4b6MidUWDlvfFC9WMvZ0GlQsK3aBhNYnsW2qNtZ826IxRnEdyiyY1xD+sd8hi/Y6Nrv8DV7GRZAKmnmlqK5dqFyx8NGN5k6v4ppglXzX78w4+B0KODJboSgE4ZMXiQfdm8vfQHkPYxwwJt6C1o1oq+1Agqx5DZ2shHAC5nKoknuz8e3hxwo2ynMtWktYaeWwqAalrcI0hgd0kimYT5Uc+RKuKZ1qTnET4hWYc+hnxppGvHjd/RZqOwLKnas0QqQ2CqHzn22KDq7sWUATCwyO8+QKvhkCIWR/iDT54fgUi1YdwAzU/08XnXoxxiw2uTpT0mm48kfChH3MdOpa3LYN2ptj7cXgHGzbhpUs/19+oO4ihpV9wxVEgvqAMXSwLN5SJdRM+YFWFws2peRcLpm/6J78K20MZvYPgB9QZzmq6m+Z2W+B/xH4t4B/APznVfW7v+hxiuJKKceiJkcO3o8n3z++Lq644RlYzeW4k5hNoqWYgq4bHAoJs0V0YTlAl6k0jgjGpnHHi+hTvCA/2LwzzVdmZlD+IhohooRr8XtbCzWBmmy9cdoaoVZycVFV7H3HSs46L3QoQwqHYC28Sp2aU22GxoLybMC6VMhLFFMF3203jhb8zlIeDAZbGc0aPaQAHbnKTySNbt0ZLu6bmzbFvlqdcR28XK3lTyAPRUE0bS0qW7/fyOw4jVkPRBiybyO0EKjGXBURPwFYP/47ZBKTXTjBqo/dxF0ZM6CJFeoe2LHyJNZjkXIGv+1ONsc340wjx8SHY9P0Pjsc7cTZmbUqg6b2zEMJ0YMUwDyFw5gZV8nhfq47zlgqxQJM1PCqhSGMQXERkfA4qU1v8BZBf1Pr2KLzfmoDzdJUtu36HGpAjqLtgZXJGr5KNoVopFv+Ahk/9mSRtSrJ146BcBorGJeqTPNFyT5PjOID7/2Z1y+HK//89R9V1b9XVX9r/fm/Bv5eVf07wN9bf/4Lr8K5jgMfD65Z/NN/+nu+//2P9LarL6530huxf+H26c7bW+fWlmFsyt+vlYJjZ/48OtwAACAASURBVF6088ADef7lweQJW7FXSlm3CO5BYufgLPDWl9ejSt7qy2j2Oph1EfXEOeluy22nZMc+4MpJlHPk4DEm57NzfDXO85Q2wYNRanf2Al+UXLMBpozmxPWzzpO2N/atc2/GTiPZllhqct27Np1t0lvShkpF2+SbyAr6tX6jthsT53yeuDlf7p/4dL9z/3Tj/vnOftux3KmZjOuSa/sUKWrQ2T3wDWabXHOQ7WLExeN6crbGY0Witdud6jdw4Qe5TfZN1m9HXovd5zKWrZRL1bjUsN0cNsB2bTI+iW6UN57ToO/s/a7NlWJuwdVE4Y7YOY+NyOIWTvbGEV8Z+T1+XrzFH9FPMVDbvROhFuGB4dmpY0pzUQ/wdzwmn5tK/Ucvjj5kA2ifOTiVcxlN9HN/gg1s28mb1sicxTg1rbLcoW3MvXj6xRN9/d5FnrJpHMeJezKG8xyQddLsIBB3oppQxWZaIWclGZPbtiT+buzWeFst6rzg9oKYbchIpw6u8QPn+cMvuqn//2gl/lPgP1y//++B/wP4r/6ibzAQOFJGucC4cybHKF5GJLQ1m09FsteEycAwPEzpvm1begCVxjlkrjlWSXvmN6szXDTaLBl1MGHkUycEYidK9ShEt1kj7MWce4l1REdt3pmr9vTUeNDKlMgM0JQR/eIq6GSYi8qyKNPre3zldn6cEaYTMkzJxjbfMU96Ln/A3pYCUH0z/rIZW9erHPJl9R4pYxCWVdh1MI/BuA6Zk7jaLmNgb477yslIlMicF3M4zLH6iaF2aqiMBhmZFKjN+5iTrs95JT/JHBdtBgnmHS/Z60XK+WgOJTudTKJkxNJgeSYmJxOzm07vnMxLPytaF/O0nRqJVFNWCMqxTIOz1K51NO1wpCXpt07HeBs31iJjzCcZT7b9E+XrfZ8SruWp99dMojRFTYiaVeu1zUtRilvf5Ct56nlsrTHzYrhs8StlKBzLYcvcRT5Dh6ezbOOYumFeASMW5Ah8OjXe1XZXkBZ4pAxw1Bv+7OuvujEU8L+Zoqb/u6r6O8CfVNU/Aqiqf2Rm//q/6BvN7G8Dfxvg/vnO+RiEbzBPjuMrz+fJYzr3gC023rYk58HjOjgegzGS/dPOWwu2W2e7K9sgr2SMd+aRvD8v3s/81kd3uLdYPbcz3JgRcEyoA2OAwbRc3HYTZz7h/uUTuBb/MQfGEBrcIVvjug44B75ilasZ862R+1rNBOXGQLTqaw5qGmNsyjPgZO86cfcVga76OSB2qoZGUOOd/bs/5rfjByqK67YzmfjxlWNOtqnNAVNqVpjhrlPFrPBto+2dLRrnUWR0Hnny43kyMtlvnXt3Zg2236wR4VrgP/7+FBOxjC9tgx7YnsQ08oI6J2PTKJV9bb7PgAHuRfWi3RbeYALPahZM8UOaSTW7u8BPsnhfXhK0xraFaOXlHOdk+OC7HZ6zeMzBvTtvW2Btp6JD/YgROsVzLr2AiEAed3qL5bc5sBiYD+J2cauEw7FovMfBo76ynY39bQrhdwW6ZDeeK/5t33YipJ0p4LzGwjXULnYLut+gYI4VkBPwOJJjnlSJ+XrrG7etc9t30lJjyE0sz1s2KuD9+MMybOmYdTwMr4FN47I7w+Ts1NKJKe3GL6Uy/FU3hv+gqv503fz/u5n9/Z/7jWsT+TsAv/0bv61raqZvIWTVrdj8JbcRT3zMsdyFJ0bytmmxvO2ioHqY4tznTsYiFcSgctJKPgJ5HiJ/rHFOvnZgik92Y9jkIDnqIGhstq0yuGPAcO3w9aKiOty60euFa6gnHPaKj1+aCC7IWMDoJAeyk1uVkqFqJlyLoz6Es5qRnHNwXaei57bGfr+L5+DGHIukBR8qPJHFTCCj+U8qCCcWL18dfxfiUeBVQhLeNO+Prk0h05mRXM+LwqjmZENYzRJnscBjajJ7MUPMnDLVXPMlyy6ofEl/bHEuIGylcdmL6IPSq8vxQ5TrYzEeK7VwLY3jOLhy8U3eHIvFxTgHOx1w8nyoIprSoYRv7Ov5p0wi6RYaqw5TiO+8NEnZYWvBp+0LHkL85YAdOqy9sJHq6Zc6rdIZV36QkJStcTFysLWdiEYyOM4TeT4Lu2iuKqG3wJtjTCl5kdpzR6Y/709Z7nkp0ijdoF/QnXEeiG3f0JwqpfL1f4Wy66r60/X/f2Jmfxf494F/bGZ/c1ULfxP4J3/p46AA2MlJW8STHgFdN06GM5ARawKxNbkd3Tf2bmxvQdtUUlUU5BtEElFE153aMjkSvv/dpNIUOTYW229NRd7e7hx5cc5DHg8uP7/mrtP3lUxUk0rZnTWcvW3cAKxxxuRIGO60rvLXXGhgvliMOJbyCigTgBre2LYmK7LWtWnUWKy4B2M8OI93lYQJ26eda05sYQNZO9EPWq7TQbT8Ze9lXPMiLGEP3dQkVSfXccA1NSo06NQyfA1hMctFZlDUkFw8ItBMbWK5ovqWJ8AcDrd1g+BMn2AyuZ0DfNiarkk7UFWqZHxN3dZ47psiIdckweSduNbM7ppenIfk5bEyG0ZOrhSXwTaZ9Kj9HFChfBCc1g5t8IttkYjANC64xsk5U+LI3tjenC9753hf7KL5MRRYTCN0CL0Yo8uuPhfvoEWjUpZxsxoztbvNMcWRoIjFmvVl9e7LdwIk9DO35VSuH5kFdi1GWRjTRXaYx8HUGI9wY9veYNPI/5dc/9Ibg5l9Aryqfli//0+A/wb4X4D/Avhv1///57/0sQAolcslnwPbICKpSyi1WIDgrdFaY+vF/TuN3XpfcWNrF6eGeAABbd+Q5XZR1+DL7BwT5qyVQCTOQAeyFzkMpuFT0xBiIcMGmNFCbkIFtFn0Jd/7LjrWgkcvPJPhLPnsy1diUinqsm4u5SuE0k1o3mg3AWQ0pVjm0Ggvp1MjmdekomHniX/3GZ8X9YevzBzM2NncF1AFiyMlo5Fw5qkWQyYg0DJpOZmPH/Aq9t7AG82diI41Z5babGVWTGoMCbZWBaIcjxIu0vxj+mMe1FwGMkjIxSJcjXAFz/jaGJhL2NWknHU0CSr13TaUSh3myvOk1iiwURZKpK4QbmFORq3PW9wJf2EILtZkRNCjk1xkPpR1iUr0mcF046zl2GQbWzRu3fHePpyla75MYzQVUBC7Jju5xqkvdunMtTEs3OG8DmbV+lzk0NCW4O2n2TDphUIPtBkaCDcr6L5BzRX9B9OLY0yYRmVnmMyRy9Y8x/45+unPuv4qFcOfAH9XZAsa8D9U1f9qZv8n8D+Z2X8J/EPgP/tLH8nkdzhSfgXp4r17M4xJa3pj3dT33faNtx3aXT2+IQVgpObvYx44Jl5/AwQtYQ7ffbfxGJPrnDD0cQ2MlsEIAXyO4dWUeoUGxy+PgTA9lyopFGI6l022BRZdob8TJqQAm7XzfZwAzrI8S2kuNM0MYjMiirTBtMGktCPVhl0DH5MzgzgHbAF5kbwrQbmBRxNfAxFk3L6tibbSuzrObg3JGZwcT8w67W0D0YVWdWTMecBw2aKnpirh+rCHN16pT0qKVRo0Y3ldzteYbb34En+/TVMGJC+AstZN3T76YJ20Il9xJv7W1Scb9AW9hEntuN8DH7JGm6GgFy9VZBXvRDpEY9Qm2nNreDhH7jAfqwcb4nlUUtaZdgDKBbnZm+TPpfFq5hKWlUx2FXYswhGs1HCf5BxcUxyd47qE/0ZyXAfXnNz3jd47Np0wZZYq32aZ0q8NZVZxUXhObCZjQFOQqohg+qmc75qQbXun7NLnwUNs2PFiQfz86196Y6iq/wv4d/8Ff/9nwH/8Sx4rc3JME198Huztt+yb0/0HfjeeXOMkorjdOvu+s+87rcPeuzTwq7zPeZLXxFd8uaxdFCySBV/MeRD0pnJttpXjsOhMyYGVQLKZcj7a+o57YzfnqBO7TjYMWtI2seWcd87PfyTy1RiiY3enUo495GSckwwtju4XnS4hV6kP9hW4ag51TDZ8OR0Vlz25fXFunz/z/fGgfGccGgF+d/8T+nzncR78pjXer688M6B2go0eCfPkgXHf/ohoHZ8Xf+NLh/tv+L//8fccp3Oci4W3SfB8PpwfR9HsEjfLN448eI6D+TTCdzgvwiefPt/h+eTx/g5M/Nq4OMmRjCfUA3xr+L0YnLTY5YV5SVXovdG/K/J7OI+iYtDMsNo5/Ae8Bb1pU5vDyZSQqh1JvTlXGGTjvuzgxvRlqrKWdxYdlVLRB9kfbJeMaJwuwV03zhzE+9p4bkENODjBJ3c+MTbgSvwKtthkjJNf1WJuznw/iAkWnR+Pg27F57eNyY12/0z7g3Nc35PtUDtx3th2GfDgMvBt4UJGcjBC6l6Vp00Hl18kU9qRaEwLcog9OnOwxx2/QsQzH7Brk/9zXkJ/yfWrYD66G/s9GLOTpSn6lc70oscNt43+BvtubK+NoSFKdE3dfCsVuAiIk0B/DorLWHZXsHVjluNR5DJvXecksf2WffWH312Tl5eq0obF19P3OLQN2xtt2+n7oDb/EEO15sSmzAkztPhglckr22C8uCmLyWmLvTjg7AUp49Eo1ner8d79O44++ToOriMZ7yd5Sip59iL8i3r+LMiTPGxxPwfYY4F+jWh/xHf3T/zxpz/w3i4eTVXIdg/+8F58PSdzrHGb1xKnafFOEn9cgmx9wy7DhrO3z/hb5/HDQ5qEN4jPpYDhLFoW989/QtnkmN9zxcm+fWZ/uzGeRU0R1JirdQM8G21MZXY04UEzBcKNDdpLMQtc5ZDGRKPmbcrd6qzlb2AQOWlzAkFFcC0P0ToVm+cLaL5tNyI3KTDboOa1Kj2YLZmWnCnvD/OOn53JxewX0eC+3SmDqwV9iw/h1Ke4YW0XAFwFregh0pmHcjZsUbOP5zvzWKrVvpZALf1FFTEGWw65ed135n3nfAzpKJJFUG/r2PtX10r8f3YZKzC2FFlvTCTVMynoXD2km+zHzGWG4YDnqi1FkF+PqL7XECXA1dIyEMgVqZ/qAbnoxGam8WQsTm8PPrLNqkiTfHqcol4Lc17eDa2p52RJflfS8NabyEaLWDiYSs1C1QFelKdm+zhRL3ExH//VkHwBHQ5Ryq5sM/HlHeguKOw4T6J9XjLzi+JiZpDV5BTlmvycl/M4LrqfzNsl38FxrhN249aC2S5+sFNU6ZEMW28iUCmhzzSHbrQ09dkRtN452wO/EK5gwt2XkJi3voNLZsx09nDaNI5THhnFxHKSDmXGwLErmT3BlG9J2ksYSWyrKaqJzSF/zRT5zPYvWkPoUFBvIwzCaK9d/4ORXqM462LzWDF3iiishPM82bf7Gh8vCLkgyujZSGvQdmzhSvdo+rxt4C/afYPWNr65ASYE9GbL+GbhAbEAojzxJs1GflsYjHMuCMkX8c3xjzV2LnDXsfAPPsTL6u3nXr+OjcFWtkOsbMiQVZW7i+JZ9c2zblFos5KG5M1MVQrqTYVbJ4tW/NMbDHgxftxkJh7wEevmea2kJgXOMFd/PaHemm5CW4QoBR1iwzE2BbWmcAxf/f0rnj1chCJIpsmzIFZg6uI9aSQ1V2WiT5uiqHxlSgnpz7xITzEiTdyMgwuGcXwd7J9Wy7u491J8C/MoVL4fx8Hv7EeOIWT+ZHLOwufAY7K5UXvxfigqrs65EpBWCGwiEC5slfW6geI1fNv1ec0rlVQ+Wf+uPrq1xr1JKdkz4FSuJiaClxMaf5pj3rjGQawkrRziPphLKFbxDbSbKy16vsYBHwtsfexqyDUy5YUKrF/5Ep0BvuG2fXxuiTgZCtlljXolIc8czEjMfhRGtDVa2+jtxqiLa556XvPCN53d8doU/Bv25WGKOHxR/HnJwB3Ll7djqiK6UjqhEBlKkzUZ5tJZTk+GrRCml8fDL7l+FRsDALFAqG6aF68knzxkACsgbdm18UpNPMTeQws0+QZqVb3YCfkhNNH3LEaZOJU/MUi1FZ9mmvWHy0xjqoTPXZ7+SXJcyTkvtTI5sNmXklDjwRZBDz1HmXz6txwAE3PPXvqCj+e14O1cxEe+pQbAen5AceopEhDge+p0vGAcg9g1vmK6UH4r0uYHH6ESns+DkcnXK5hXY2WicRXE48TvE9sm2y5B27Jh1c00hZk44mT4TPxNr6em2oDmGtE6/MRs1bDNhTHQ2GMTpTwlZGrV8C5wOZZD9TTAOz/MJ+3SFETg3hpDDhc5LDTuu4C0paFwX95YK4pvPYcX7J+8nMn5WBxlSsbqrROtqXItHVCx35jzeE0r1zobXHVAhT73dmfrOxYObeBjwDw4h0MmfRWjNr+lVM2h9DLWeltYN5RerxPgYtvmWKXNBRXGMNdokm/7njfpRQynr3UN8M256Oddv4qNoUq4MA7WDbxEnPFOtCfRoG8yCN66GGMRrjcph/r3Eptx8jKwsHXmiuqaqcrjp6m/vgJDbcl8r1SrYWHL9NPwrhrNWnyYufpRMBQg0nB6SkFrHrQebE2eBeY6YQQUOTAVu0cjpyobLz5GfBjiYczXhpFkQPuJZNaigRenT7Kthe2Onx3yCSSZoSeU+tqFzULJA+J5nDxzctXO82HMoWmL11wzeDB32lvAYYxD1QrECnoxZjPdNK/nHkHNg7wutlR3W1vJaTqlP/DdOcYTeAHGwbESG/YJsTdVWXO5ObviAD1X+M+QuKi6Xp64UfVRFFb5irqHNJUH7klLpTu57nL5ReYSIfHtPbRWxOIDwBRdeV4UnW3fxehczlxVks9nTmZckF9o/sYencsurvrKGIPjOLmGWIi+96WCrJWF4RJS8a2YsfWMREMvXmOtSXFdUGMQsTCrTKb8hpVP4iE8wtba/al5zk+OmZ9z/So2BoD3x1dib8S8c2vOvncex6R1o8Vks51uTuckbGqs028cq9u3LMnihqmsXhWDU3Rgt8VHwLjcSJNk9x5Bj6b8gy4WnBO09tAG5B23nRbOOU+OOPi8N+7zN4znYByDi5PPt08LFF2ThZyIGigOg9rJTniDfTJqmcTOk5rvmF/cPv1GnAM3Rh5wwV4b19oXWgaRnchJ3E+dIj8UHHA+E+tGn9DDIbRJniOZR9I/BTUvvh+D2UTmOQ9jtInnxkYn60cZwvY3widbQf7xJ/Z70X6c/PBn70Bxa9IltB6kT05OtpDn5dfjR+q42Ped3m9MhVTge2NrjXO8c4wkzAnb8djxgLetLXXlgW+BRUB2rsfBp95o+07NYjxOuCZ2ewM/lUC1GzA5Hw9677RQu/icRmdyPA+e5wN88untO9r9MxaTPicdKV6nGaNPrvcDroOKIQ4LDkdyjknEd8SuDMt5NNr2W97fBsf7V+62sjHqE5Og5uQcgx+PgvPi7csbG855HUQUW2zE2HnPE8tgDqVY92bUcXC+v+PtO9KD4zqYnPQdTjrjKZ2HJt3CfZSAEZzRwAeVqqSaqdcqrz9/0/0F169jYzAgnApnlPgMI09FsC9DkwjZcfUe9HWyPJb99usl2xIjldni1H+08MoNWCeLQN+kmchH4bnAGVOZ/xMIkNVTzuVgLPuHZbSxSRorY88mdll8w7g8Ll4N7sc8P1nPaAhbsYDYhDm9CvSEmpNMUaojbWnyjVr9orGp1I2CprHUeQy2mbit7EJc1OHlC2AEnimfxSpmylp8DMXBGUU0Z2uBRXKbHSp51rWS1R2siaTuU4YjLaihaDZGEmlIUWHs+xtRgzlPlcOj5DPpXb1xKJIu5xKTCVUhF6sUUnhT/QvK4DCsNVaS3U8X0vp0plqZ9TMNlftKuS5eZnr5er9RWeU2l0eiysBrQp0SlfVNyyS62qLmxf28sU1jzMEYB3EFK9geO4sYhodEVCDehuhRF6RxVbEve7sXq5MpL8z0AZdDzoXfCB6+VrQCCzsyWKSoXPkYq8Ir48XslHP5z79+JRuD4a1j0ahlUqp2PNm3RveVNRk/AVLM8apvncGr2nUBUobO/tdN/ELGJ9LeE9ImbK4WZQLTjCpFuTlCemWACmeNxe5fgaOOvBLc2PeN1kP5CqhtKTTrfHlBvQhSrwUJiXtfN/irxUwsp8xmpjZIKT35iOfLOkXUyWBOo04UeGqHWIpT48XVkOtWWyQcEbdUg8uEVe+fqpZJi6LHRu/CWVoLcp7K05iX5NKm8WfsEG+dbe+cczCHiGHRmqjS7ty2zozgHMjNekxmOtvKSZg2l++BMVbITlVIWbk2UmuOjxdkXB+HiFtQTdqAOVQxUHKZFgCvFvKVJxlto3UICzITcwF4aj9MbYlYSB+LaOb6/muQWWptswMda9DC+DLvTN743dffKafyOsAmNhucRptSbG7RqNThM2FFIE51RyVy0rmYkjEhK7iu82Om4q5BGW5cq016gd2EUQoExRYnRjjMimRwSPvWjv6c61exMRjQfMMQj6Gl0dkwL7Ztx22BgaCo8bN0Uq6K4HUlOv2n6/TVyPInE9xXExcygm0LCXbX18i9CY3S6ie1iMvUE0On10+48iIttUUzmKvOWNF2KGH4FfdGXmQlxYXRcUsxHq0tI40ia+DZPkZsE7VBH9P6VKVRA8ZjcDyePN+/8jy+0uYu/HJqDDbRvD1dpabSsIZaG0LKy4JRk+lDkXnmhCnAJVoQZ0Gd5DwJG+Arkas50ToRG2NckPK1jH0nUn6brxv5lYKl6L2GWaOQKG4wCGtcdIpUSwjYimBbJFg+EjcVMiJ5czgTZYbKR0NWf2HfJjGzJmnGtjU5jvuKtjVVqHMqCSvFZBb9uAIyFi/lYuYFdTDs0zKQEj/GSPYe5B5s5yc8BHcbchCrmngMmt9E0WaxXlP1TGZAG8IBLuRwvY4xs2COCfPEvGhNo0xf61nAaS2XrlfwzFxOVzoODXF2DPvnsLWfc/0qNobXTRYmFqTRaSEKqqzFXl+okzenU8t/0V0ipFeCs970opO4aVT4Gv2R9rGwohk9lvcgmga0NdPW7FpaeBDIAwpYIVnuvys0lpT19yzcPmYi2PI81OvTLCxrDcnqJ/3eq8ox3dDyhNz0KC9OsxuezlxoPA7XLK4xOc6L87oYx0EnyLyw7At8Wz2yGUyRlAYySIloNN+Yl9oxTc+08JiXWrLQz35pRFuI64GHREmnXs84TiyNiI6H07rs/o/zWK9VJx5hxPo3hb9cTJTzIO/HdVuJdIBZEv1bqxgk1dZ8ci4B1JoSvOJBVXoFnoohLJPWovWNtq+3f+aHJeCs5MpahCA412TldWe8lJ6OUbFyR65BTXlAWgto8Om+AwN3bdznZQwLstXS+Qy1Yvr4NDG6Cm+TYn64c7dVsHgPOGSg46ZqliiF/XhoGjNT1ZPQd6rEVZE3hMNyrLZ8TfF+/vWr2Bhknq9h41UPeq0xW0CNC2rlFYYqAVuBICOXZTvGKL1PI4txKTX4p4KiRByEufwe20vMAx+mmi90Ibw+ounnMvB8sRPdv0mFHY3HqnQ65IqzMwBzchx863H0PKmVVWb24fdfKK1J542yLF9VMwbT/DV4gyZEejpc4czWoXUiblx1EgydKqWK4GUQQ13q4deI30IxlIcnUbmSvEJEmio8DVpb51zHfGfvEqcNc/K4mPPkuiZznIuXIYylrwnLmSd+voq0ggbbW9cpfR6q+NYJXFyKrYcPibbVGot603Ny5LQdwDjJS1rJXK2AU3p/Z5OAq07MjX15ObjJlD9T1KBM2bONU5hIRGfmoeqj+gdJLrzh1pQLUkldlzwVHFoziMHtrRiXUdVWtogzvGEWnLxLDmVBvbI8c3yEE58MjnF9tLnmTS0ZF+cYNBfQWKkKNdrCrHKtLne8jDmKGfmhUP7YDKqUKv4Lrl/FxhDmfN4/4w7XH/6MycZ1aSO4v8nWS71+QiY18tspBitXUn4r10ge86IyhNOu2bDwmJT3vmtebqHR3LgGXMWt3+QlGQKrFI6q6iC88N7Z9zdauExLESdia0DUErC9IL/gGuND1PZi6rklZkWwWHk1mOPUXNxuQOc8H8hcLYlwLgpLo0/jQcFZ1DXxLdm/2/nN/GM+5Re+Xj/QYteCsAvJfHSU1vyiqchoUkoCRzx4jK8ILusLCBXQu9mNrItGcVlwbRsZixqdk2wb5wXjmEzrtLZjJjCQObEdYnPiXIKqNqg2ub01hhdxvLF/Xe3QkfogVzVXTMYoGEUMmG/QX/qAZTt/fP8jYzQC4SV1XCQn6U52VouQ9Ag+v+3aYMekzlzmtZ16H8zHYByXWJv3jc3uqKlFG/ScCojBGEfAPrBtarrUdvZoxCgub2Q1rtMJP7ke30tF2Y3723e8LX+FGUq/9kvHQBzwB7s4jndaJZsl3r6Q0Tn8B77Og2Y7RdCrs+G0L0YeSR3fHMiRFQjjHOLTtHVCjUkk9PnXcGMokm07sQl//K/9DS3nfKxeymlbf3lg4NtGVTKuQdWE9hk61Hww5pMxk3hc8LZTtx1WHJwTHLedrYrNBWiC/AJGK6bDyUGvho+N8gk2ac3YS67QsJx5YDHh4AUs2ss6a7J8FwbRllW8ifgk3URg3hk2YdF4bYrSm/VgmtHv0HmDEnfTzosq+Qwwpa7zaMznwfH1YI7J1eFzvHG0xAM2F4j6Po33Kr6bSbgzYuDznbgczxvfffrESJe/BAdMw/zOrR+YfeH3778j/UnfG19PofydBo/J4xzULG7fTWY8eM7GPjoxG9Ymew9ISchtg/alcZWcm+YcSms6L64r2XfwT2/0pt4hrTH2zhxPog5y2xh7U6+djfROIIyFuIkafZ48vw6O7QG96NHwPTjGDxr1+ga4BF7nk7d2g+7kFVwzyHPSvxTUTUrSm9H6Z+J58LuvX/nUB8EbkTIAskxyOBGTmI20pZWYulHTT4pis++4txvZDgqjsbE9J5kHP+QPnN8jFuN9o701cWayw1l0u8HCC/ZNE6Pzx3fm8dBjv93xzTlIuN9pX99hFuM4IZxujqUzj7+G7IHd+AAAIABJREFUGwNmWGzqdf0b3mBmtDA5Bq9/+2jKWe1E6QQQTbfBhK/1AJLdPohuOLBTVAWZQVkIbwDMGrGCGBKRgKbp18ftv2y9fzLF5NUxJic/KdzWlSjiGC10RGaaWdicJO8f6LnKmlheEKxpxppnVLFCE6ASj43M5HEcPI6Lc6YQ6+UTEiHq7pyTOZcwacLlF5IonVyux2s2uKYxSpJejOXH4ETzZfC+gmBTLVhNOUPni85tBXNZ2Jval2wqf5mmqiVEVNpYoUH5wlPUElWAbW1NnNZfvsZxxHp+EKnAIWrwPAfnMfBth1j2ZZYEHW8N6yWH5CkPBlK093L19M13ia2ay4d/6vXk8jxgjaXlu+i0toEPsFcgjGj7rYBc2IyjtmgBrm6Gu1F+MmnCJ2CN2Cd4MmbDmgR33oxJ57iSOS8ec5JhtN4UfWDGmUoljwrxGBaoOM6T5GKLhsKCTQDuFstx65ehDL+OjYHF0mrGxsutRptD74qwz/yJ1sE0IkgGR0qYNC9p1a8hb/1bTo3/wsV2W+XhsaYL+ZOJhrlQcF+70lhGHaCev9Y3lE0BmIvz8JohzwTqJxbn6/teGw8lwOsjFjLB6kFG4G3Tgkz9jJws0VWuL9UM0tKwVGjNQPb6j3My1kbS2pKtV1up06k2aCACTVt+lqSATpcp6pyyyzcgF6Ldwtgj6LbpPp1Do7jeFolOeVnVdBN5Ou6igXcLnq4wiHlpg7O2AMAU6FhTN/iye1RV1/uLvsfrTTAC0jgTsCSuwmvICHZMjjl5a42X0C3KCCts6U3GFGU7KhT06g7rhnTbpRptkBuynnf/CSniY0AKrpwL4zUZWuPoAZsnuTn9Nf9aaLlF0KzLNsMGV13k7FiEsB5juSoFW+tUd9yWvmUcnJciCDyCfZMhEabkshdeFL6MgNbkxhEQ78gTRIE1vmz2PsgeP+v6lWwML+Ta5WzjGiU2kENTMzlGrwViVYvqq1tHAJFJRLVsh4JUkpLzYQnmzpps6G6qxXoPkdJhhoDAnAyrJaIR4DmXr6BHfBxmqhoEfK55w58HeQrhIsDLVVpj1jdIhZe85jLli8CRL+aFAFmv15xVf1elkWyi7E3LFLkrDV8aiTlLYTamm4I1zlNHpscPD5GN1pN/CXS7SeNx7zexQtH7d7NVQ6Te6zRtll6NfcmLzQKbEr0/j0E047b3xZcoRibnlZKML7EY5sRm0KBMAJ+jHA2F+hZzJFcN8TYAIoit1udwYesDKVdrV1Nfk+UcjyJb8uaBddfYEjgv5U1QrIpj1YDumK+Z09RYNZFGQ2IrbeCQK0xZcvSPT95E/HIrtiZdxJgXaUZb0j6LRrRk3ycx4FqBOzlES8+lPI0W7A4txZ2ua/KY70Dw1oWhhfuqFJRLKlDSgKDG9THa/CXXr2NjWM/ZkHgqfFFmkRFKhHOO1623TmpLdu+EJxlBVWCeXPWUa+7r+F7jTk91auuA/X+pe3te27Ymv+tXVWPMufY6L/fepm2rbSM5ISOHEIkIhOQIJBIMQnLCB8AZqVMkJCQHCJwgyCAgQ0JERHwCAmRsjN328/RzXvZec45RVQQ15jpHpnHfi5rW7fXouffcvfdZe6015xij6l//l2+/tCbeXO1JBd+Us7RgdYBdsUQUQ00JJJTUUoOW6eJlwKWXYhyZ8u1yxLICA9KC5IZEkj4RL9F2KKQpOXVBhuUwUa+1/jRnuQFtfWfuSeYocU0ERKlBPWqUJ1ZReoLV9CDXVMLBtYDJe9s4pQQ69hSn1RvY2wvNdkT7xR9a4bBF/KFNRANrjX1v6F4enTbLWv04T+6y02VDSI54ZY7Guao7vBwzhAJ8U5JQUFupV0kpGHHc4bi2AFW2vWIKI0uJWa3mtsDRJFcID8BxHoQb225sZoht5DyZc5RwbtSGiOk338k1ypx+cs6TKVRq9joC4np94bW5r/JTV8Rgs3Ln7l3AOnkOXGZ1l1ob6LbdeX9P5msB0NNHHQTo8mjYisAUc1n3lwo1LnKctdIUrbtF8oo3TFJyWQSCL5XsL3n8KjYGFWG3rbQJ7VrPtXwbW7H84sAvinONqjnzIEblT7yewZsnh0ze394j+wvTjEmwrTj1jSrbVS/n5JpuhBeRQLOs3UgvEJIiLkySfdmCR1y7ry/du4D0Gg9ShKqrm3M/n+/xe0wCUSK/FmSx6LiRkKMIVl+P8xk9B+uEylmkI3dEdzp3dv1ahiECwYbH5C2+QChdd+59Q5qQPRifiwEow/Fz4Daw9p5uCwdIp8tG24z3987Hl/e8aOPd/c67l/cMBsfrJ45jcqZx8pVGYG7cP27cf2rEdD7/9mCeb7y//wjNcH3gpuh2Q887wslG0q00Ckmr8n1FQLRN2NqGtMYxhbfHiXqRpUUC0xWxp23xGRyNNyxA2gvOCXOWZub+schxt+XsrLrYkAJzMivym20rL4wxnWMOxMq1SjTws9ilt9sLPgNVRwWadWqJHswUJlaAszq7lGAMhbYbXXdeX4se37aNDaALmxhd4chX5mtVL5mKto1uG1/OVxqtMkq8KtnM5Hb/gMmdfe+IDuY8GefE2oH2OxmVWTJ1jbf95JuN+M97/Co2BoBmsS7GAgwvElDWqZBPvvwCDFFeaMw5Gccob8AZ7GK8s0azFVB7AX/AKYLKXFZa10lcPpMS37UAqxRbkRJcJ+h3538952LVaMsq92VRjq9tbdFU6ynkeQqCkL44+VaVCgl5jtU/Z40OWUEyKzTHA/Y9iuw0A5+Dmr0ZaCP15JSB6cYmRqNMQUIGTaoLJiaZ52pnZIGUo/pSgl0aH1rjY9+RrBatd0UfkOPAfZI00kbRwn1hQiZlGns6k4OtCaTxpq8MM7rt7LaDlUYgKH5GtQtBehn1KLb8ORrq1X5wVh9tlqVlYFnAIbT3t6e227Z99doQqYwZaK/F2aRyK9OdOYPmRljF8UkqMoBRvAZvSfSiJl+v6WY7j/FW956U8L+UnUVKu0p1IZHlY5laQJktTopko2zdY42tnVe8MIImiJfQpu7KYMwT3Vo5pq+7ziOrXVhwyGUp0LdG6/W9HKWzSZd6E8wCTn/JevxFP/3/06MIMFS8vFgZhC6wcW4nWyixw5SgcxZ92t7xOD7D3tFoyOnEnLVrmyCcNYemgW6EFXMMqgW4xp9lD1zxYsNrOiCbPQtGy7zgh2IHJrBorUVV8UL2z2p5xLjgShStNKxRWAErAMZ8o7MXK9GrLwbnnCfn6cj2jqt1kKye18LBYpmDvnJrN7LfefXO63mS44HPg1u8Lx5FM1ydhrGzEzaYBK/niUTn3u81ziPY2wtdg13hvgm/9/4nfvh4I+Ij9n/9A+LLF/rZ+NI/kvmgifFxGmdMTkkyH4xzZ4hgP+78eN4rEKXDzp34Cq9vb1g/mXoy5UGXHWUrz0dJ+o97zeC9dA1zOn/0209EFsAn7R2yv6C7YzqZbwfnnNxHGQC7wdt09t4JE17HpPfPiL3jvm/A0id4IuGc3FFOdm4g8LBHjRrzRvaTR4KcsJvx8u6FPzp+W6aqUVMMa0GT8pc840HOQd/Bbu8BQ8crwh2dibYGLVB5Q9kJOiMPXA4MYTRlu3cqGzeZA3xWLOI8Jtrfse+N3j/iM+hyMDAGE8PpNMb4xBEvvLeGb5OKQUvaW7UQIX+GobZ/Wo9F8yb0Omul+PWUlNVI9khMiyleEQaT6bVTz+nMWcEy1pK2vSzb9jqN8/olYgTz+kphAXmJmlg57LB2jO9e4Rpj5OotY9miX383WSPFCyxZf78tdpUs1uLzua8wklgo/yK7SNG0nXha20/AozYWz2Bf245nMSYXBEltU33pK+oocQzLsqxvW1niz9mYWh6KNRYdC71O0OWl6FqBJqt3r+GKIzGxnIgksld5V4GxwXnM2nyDNV2o9ypZaHodmwmzL2ZnjdVqtxXmWJOgSMStRqTpRXGX6/1NNAOLoFsvUPUaKPskvryWYSuJOdxaYSz2/KyXK5IY4/RFdFthP1qMRV8UautS12Clj2+tqNqhSwi3tByV5m0gKz9ypZZH5nPqNccb4aOmVHEQzLL814lkZ6fTMaY1ZjinOZOE2w7cEN0wnQijRrGpq3KqzyPipVSmMfnqVU2ISYXoNAjszyfzESj8YFEZyxOxRpbTi2neyUqEprIMJ5NzJu7J2+Ec5yQyuG3GvusaMUWV8wkRq2PPePYWCwrkyW9eVf+3seOaDKwRG3ybKuR3vAUL1i5z8Y1Zo7GKM4/Lven5G+uHPAulZ5WfxUFg7ZJXAqKWyIkCGE06l+HKWAupkrETYS/PP0rHoTFJjExhu+0sUITznHBNcVazVKrJ6sHf3NlnIudXxjwqxCVLb6KrbdAVU++jcZ6TsLMMTr2CcEUu3gZAibU0QaKtKU3jW461c85Rn4s4zINzOpGBqizHgYAcqCfNhJdtJ6KMbEpcJIxzwIS2NTZrvPSVNZIsg5ks7UMIMU6kxbJYFBIDCVKLUmybEssez5hsHUbU9Xlu5qkrmareowcw/Sm7Xx8r5zHwCd2sRGxzMqNyWY1YNO+GuSy7x8TCS22ce2V0nF9BT6x9KAYoIFoEunRlnEWoOqys8zbbamRqpcuY35n9/JzHr2RjWAKQoMAoAV1y52/lRI2IhpcEesbJHK3Q6rPGWdZ0GYQUWSWl8gae/o/r5q7Z+1r8V5NGRXmxbuRveMCaWWdxCmobuSbaq/Kw/+ecQ6HaFCi6NLFGpbGeRyvNIJf/ACwilxRaTi5hWPEErsmErnPoTF8ORFVRxfoulIjs8ob2cFRqVi4o3ECtFIXn6XXjyzrxQpiZPM6T16Phb2+8+VHsUAosVNswK65EZSIq4zFxdXaMLlIIrdacWHxtmEpND7Twg/rYr34syZbP9s7D1/NXvJsGKyD3ynMwWjNUlLc4wISQjdnq+ohsbL0X6csLVLzMeitEyEgOLIVLPCLrdSDx9MmI9XcmySYKTRhBGRCv12+tKp9U8JzkTNwLiryqyjkFzwIUE8pvZAplyjgQK1Nft+UwmVGRobOmHvN0znGCGNvu7O8qkFmkQ5aSc46JS3EyxArLMLNnZoX8ebR2E4TWqu83WQKoJmwiaHbecvLlOBg+eXs446x57xHJHCcw2W7G+/vOh/sG9gIXoTApFk2CT3uSeeC7KQElw72+V4El8s0TMJ51Rd1cQn3wWJGoWifDF+fgQk8LSEIFbcW8Y0aRhRiI1vqpQqTkvxVynTXOCin7M3QRML6N5S7S0YhSmJLCoHHKQS5NeKYRbrjWrITZaCZgK2bOy8Dk7cujyDtbqc5Odz4fD0zhL/zwkb/sPzEzkT984/FPvixVaVRr4FJgF1rS83VDhl5o22JIXvyTlZSFwKkHRNBcIJywk9vtBVXjOAfzFDbpbHclzqiWSvw5gvYGaU77DMepDBF0t8Ip8hXL5PX8Ac3B8CU6komaVaDR7QYTDp+VUI5ibLgdNRYcZTlfbYLyeIzFB6n7QZLly1GkLB+lkPScQFndl6z8ZEDdByacTIafgNJUeZE73r9w+OA4ir+w0Xgx40u+lYjQjcYdtRut7TgHlbKhKAfOgeNo3umhqGttdGUsstysrwPw5z1+HRuDCGZbVQhacXMVviLEORl5kWMLqCo2ycaYb8w56Q1ue+flvgJDvdSCkQX2XZBBIer6hAXqEVxKJ5EaW8q1INe3V9d4FROgdW43rpOl3ImLwORrzJhgFSsv6+9cvzNhUYxAVMiounv4mpAsI5jn2DOKmpTLN+KCRrmqnmJWMDmWtHudgLqmLggjA3W76Bjocl9yjnX6tGViWh2Hi2K2s+/veHc7uPegyxdEl6zdtsUOdaQbfa/WIocvSTKrhavXJ6si1AU1ZKywVSq2eIxJ70FIY0w4ZxHLGjzNtHPtiqrPTw99MfzNGWOiXuSgzGQ25xiO4SD7t2tdwAqus9zH40TN2GwvjKE+cZgldutmWGu8jgE9V6VTZChdmExSXJfiT1xTiwrydZ9FvweKMVm0eMMxTrCdZp25iHnhcF5zCa+8T1B6D9qmSK8wHWFgeRJMkIbajskGo0a7005OM27bXkQr+3O4MSCV8lyHzMUFkAKJ5sG8CmhZsmZraG54HnjUPLhtO327IaK4H4jK4vovkpJXNibsXDdHsRgKgMqrVF+bw9Ozf+FjIfXT1dddF79YZ2WAEqRXGnJeN2C9uedzoPoMIiwrcikbewUWlhEepdZcfHdyLdSFd7VrU8h8gqhX5pav21r4Lmk7amMs/tPa3rRGq2VZd4JuBfAsr8hcKUdvUznT6sZTo2+trhUUWDfr6LReFR9UOS1nEW+gVQ4EK1kcyiIfr35eoogrGOc46IvWfHq5GfWLN6Q19hXJ+lzkoq8r8q7IO3GcdBcyy/Bl0CpnAl8kLVmn5xol8iiGIQPBaGa0JjxClqqyKs6mrZysLmYXT8orLOfuOhjy6b8h1z9i+WvkhhLgy2qttO0oVLyhUJVJJh6BL8+Mc058WqlfuyN2Lvm3olQ8oUmC7IjuiBr+KPEfXpc0xZafxJ9DgpMQ9F7oKens1mimhDu2vfBhwOeH8zqMt3EyTifmgb+9se/Bjz9tvP+wU77azg/3H5g+8fOs+a2ulJ+zc87XuqBWc20VqVIeIBrIBJkMBMIqE1NgpBGijAgyl5d5r4BdC2GgTKm1LzLxCHY6Mh1d4bQuQm5JO0rCK93orViOY84SWHFVULV55Jy8PqKSmS0Q7szVJo0cHKczT7AQbhG8mfAmsJnwrjVa65xN+TLeeGvCS4MuSZ6T83DmlwceH5i6EUxaBJzw2j7TjsD9a/W/4xORbfkefiA4sNiR3srZel5ZoTs6JrFZAWausO0VpTYOfjs+l3OWbrg6hzhG48PtHZs1RiF42Ob0bvBwdrvzyivOXNWY43qwyYZ8NvbzRKaRx85oJ6M5PQ1cmO6IHey3O9bek3IyeEVnqSlb2zFaeSm40m/CcHhoIlOYb5NtwP333/H18+8w22m2IUjJp1Pp0XmzA5+9QGo5q8Vggm8c8cZGR0O4idHazknyKYLb/MSHn/4ybQz8+A3z9SsjNug3bt0YHBzHZB7GzWBvFQzknnjc2cPQND71ROQTL/0d2OQ8v/IIaL//l0oj9PXrL1qTv4qNAaoU86zQEvdArSA6p+bvX+PgyLkQ8hrd2WZFXmnbsoablTR9lXHruat6fhqarRP9oj1dbNYqCy8zFs8s9OcCL2Vf9OgLCLxedz4jAa6epYrcy6bsuVtc+FuBRXzXTmRymfgmlGfj4k9Ell9FwQwG84BU8qzwlRAh9noXeVRk20urzM2+twLX5gOjLb8FI6VwgukT6zunTixeabOmQ1/OV9434+DAcNq+sb809HcBo9KzwlY1U5guEZRdPkmoF96w7bSupHZGODEqB0JEF6fDnm3c9TArf4HibxTuJLZhMbAs2zbTyq90baAH3pTcykDVrYhHAbj35Qta24klTDc0OyYDbGVZPG2TBGslrrJFEvKIwgTOyb7tmHSE9mQi+vLBlEV5Cgmkr6pkQS0vKGQwGEtFqWwpaCSqyRgDIdj2nds94EjOkcSoZC33ui+rpc1l6lUVUGiBtXX/KWxHMUjzVszLmAgNfbZTP+/xq9kYRKPkyA7h/alZ0KG8jcHX44RpmG+Yn7gP9v097973FXK7Ywjhk9NrZJcaCFGOPZm4Jps1fP0PH7gqfZlva6sWoZbwG5nnupFrEyqHoQoRrJ65xmtTCqnXKBKN4GgLbDpp9feQRVKazqsHbK3ITqY1m0+IVk6CyCx5cxS7TVsBstYg962Ifm/LhCVmPa+C90ZPuO3Gdt9orTHHiU14aQ3dthIrRXlUWg/+hR/+Kp/yE6/+lXfHjYjOP5VP2HHn3Acf32+83N7x0w8/8ff+0W94HAdzBvt9iaKoRKZSTOsaxAhMeLl9QAmOY3B8eePz10Lgwxp36+ztBSE4YjwnO2bw/mXHKUOT7Mb0V844itNilTQaUOArg20LzIxDJvusNk4cRhsgGzFPzjjohy3EXtnazmyKdEE3JaJ0MMfXYggK1croihs0A5sbuOFeKt45A7MDTHl/+4FDDo75WklYYjRbYCudt8fg9XzjQ39hMyVmIuv+8DhpzdhunTmD4+1gPk4iBh0rxaiCacM2qyiEWWKssElwsEeC7rANRDu7vaOZ4MfJTGf+eTRquVxorkkXakSWSalMYc4g5mo1fc3ERdlE2WSp28PXDLIcemUdwTXJLlTJNIjVj1dSVaG3sujRi45EnfpjCa46inEApC70o/p7IxFJPGtaMLhUd+W4XNra6vEzg+HrZocat13yclcQwyyLlRklnYUsM9fi5pIJlq1e5WbIbNio0Yl6Iqa83ktD4Dlh1KbY2r3m5Fb24wol934xXBt9GKez+uTg9OCYjnVleuV5Nna23njMYPpg5z3oylWMkmfH2rzAEWnF+POT9LNs7vJbn6tSVOqLM5JIgXKRT4+BoKLojuML53kyXbHW6E2hzxKDxVmRggLJqM+uIBSyJYQyB6QXaLFZeVd4xHdpYFHu4H6FJJeW5sIyBEFdON5qmuApy+sioFGCvt0qDnAUhd1T2EWRLXgdMEWwacQQgmDOk+M4sW2nLXzhPAbH22COiVBGyM0MnbMmJGaYtsKZrKq1U4o0YRm4K/it3veW2N4W1+eoSdIvePyKNobiL4RLGWXO4PTJHJPjPIk5iViYsQRosPdOb+W/6z5QXXbaS85rUmnRElX21pz8qtm1UqGWFiFVVkDuJd+qMarKhuhG+nmt8SpLpUpdVdaosizGz7kugIG3vUZKCSOTYzoxC+RrS8gl17SD+t2X+SspkI6a4lJ6Aib0tzdCe21uZrCVZl9T2EleWxnSujuZUdOGNZNfHLgq0VtHWuNzHjSkRrk6SZKmjVece+xMz7KWM/jwbmfS8Cx+yPqQrn+sKQvI/kK3Blsj3iZzrs1w78Cgy+VdsLgFUqzNsbwQROv1+nS6llVdRdnBkAUfuz/j7ZCGinBF6SFGanADUo2pBlRwkJnSenB+qUb1kiezwON9q+mYUtdhRtGU4wgexyhz4YVLoVxudIUnXkPtVErjUxvfkVcbq8RIBmOF0wQvrPsni4sTvuzaNuHlZa9K6HDGXKM5LTWqLjC5uDG1UZ3Mwk4iYBcayZwDz7k27J//+FVsDFBlqGrRd2cUV/wYR2UnzkD84vMll97p9tLYXgSYRQaKMnQlWE1/Db0llWvDVL3GfAUtS8C0qji6BBoNtPG0d1cpqvIK8bhwgStAWKw8EKofXeGwCM2NvBJugXDnGEXjftda4Q/qz3CZkY6HgNoaZ+a39luVOIFZrkSpML3jE1JWeKlSdvCH10xeKwqeHNh8I6VD36FvZR4aQsPRdnLzkvfOdiAStLYzlh/m6ZNuB7onP3x8B35HzsmYj+sTJNKQWp/1NduX16ZzTuc8iiW49U6409VQKdSnxm0FzkY4Y1V+QTDd6Wbs/SNv/saZc0nda2H4WRSVPcv6LzhKOl6usTCT1pJ2A7/CcgQmx5UCSNVmWlMSWGPIqjJzLXbPKLHZqjBMwCxgRfEpUbL2iCcP5/IMzUr3KTRLCh+LgBRF+471ztvrY0kAjJf9DkxGTO4vOxiMrOdxSusRrWZmGV5eI2mcPnll0t0rQ2IY+wlxvGGbsvX7L1qPv5qN4Tp4nCppz1G25JOGODRXUisnQVGQzvbS2HoyvBiFkpW4NBEkIOxbptRFUFpdfN0Ei1k4s1iSZeDSqiaWhkjpKkScbRbCKFLKOKQ4CIhi2atiyErGdozMRqg825PwXLjJ0nykF0GIxHPpPqJGtVODlKQpRUhIKRalJydKujIPysGaMlnFQPqOfZmFmNx2sgV+HOT5GfReoKU28tK2Z2KavOheyV5ZLk+73XAblRxNMGVgm/Ph3Q1/fcHl5Lf+KAAMLVUgVEsTBbwqUdLm42Qexfpse3lOm1xuSE4y1+YpS2fgRcyi6MiYsO8/sLlx+CtSttK4G+EN27SUk6r0tSl069Cc8/GK2k7fJywn6UEQMbgYtlGkT4iqUo5JidoWcSl1VY5GlSxaI3PVKhfKeVo5Fj2/CJI1hi6Lf1ZFBKmTkjYp0hpmG9qEx+MLzTbu+46+bKAneTzovZPqqJVew6OYqt5WqC91sIwoRubBrJF83GjnC66CzQPbXuj7nzL4KCL/BfBvAf84M//l9bXfA/4b4K8B/zvw72Tmb6WYQf8p8G8Cr8C/n5n/65/0O5J1wRC6Jef5wMdA2XhH42u+MlMKSNomXeAlG8QDn8K2v8cFHuNEZ9K3B0ZfYbDOtCCts6fCWKccFdvliyjUEGJb82yEOYtU1JbVfDUsgXNWoI0aGg11I9rkfCsuxNYNlVvZ04dyxuCMk+N0bGhZq6+8jOptvwl2A/Ac7Os486Bm47MSrdOg20sZwB5nTV8oktBuOzLh/u6F86xF1bMTM/h8nPzwrkrpLr38ANRxBOTGoZPDBumtXLPjQSv6DX50Rg7e9533W9J+70fuHwa/9/oT/+TTP+APP5U+gRlM2dGevDsU3aw0FA5yf2GG4mK820o4JvOgU8zBcwaP4xMvvWGWvJ6vHAdsvNA/wvH64Bgn85xsm7C1na/HK69ff8Nf3H7imAfuyg8f3qFeeZ3HDOzDu3JYOjqmIHISBqE3ggcvP/5Q2Q5HxfOhgT++4h9faLcdc2hzgJ0cD+OdGfStYuDmScaAlw2L5PPxlZI83QDh9JpA3FtnTjhdeczyEDW2cirjtSY1+x2l8eUtOeMoH4+oXdYXIH/bG/d3O5nK109fabcb93cfeRcTn5/4R+ONPpVsxcvBTrCB6Q8VQhOvP29HWI+fUzH8l8B/Bvzd7772t4D/MTP/toj8rfXf/zHwbwD/0vr/vwL85+vff+IjvNxpjmNyHKPcvnISAAAgAElEQVRoz+G8MjjCGa1OexPYbzWJcKvMCfHFioskpSzYilpSIOFiByzKzwHr8lR3GcwFzbR4oVIKc32lJJ/BwG0r9mSsGiSMkAKlIjqfHp9wOntX+vYAMTxfmAljQCxarogsYVQt/tJu1IRD694kL+atQrqUHnTRszWT0QK7JdsUjlf48njj9fWN+67cf/yIiDNHkHGgVpbtQXIcR01TdmPbhW3v3PsHvj4O/BOcXlXa8RX8Ybgd7NpoY+doL7y7v+OHdx/4CwafbxvH+Uf849/+IV/nG2obm2/4AWMf6AhkKD60FpjVIeyeWJtgjVje/sJEpU5hNcfSK1FswKdPg5ZJk5P7Pti1EafR7SM/fXzHl2MWMzRrtNfNSAMzpWvDNTlyIpRY6Y6gAce9tBSnlzWVSk0qXqPIVfP1UXiQV7jt1hrH8UCixG7gTxpnalT2A6sQk8Jk1Mr6TqL4nS1LZo0pucxh08H/6AvdFGkvRbuWxHLy9vqVcxpvr+XGtXVh35Xeha1phRcjbK2xax0tbhsWE85Abjtxq9G7v/0p+zFk5v8sIn/tn/nyXwf+tfXn/wr4n6iN4a8DfzeLAva/iMiPIvIHmfkP//m/BHKWB+BxOudcRqaeK0pe6RY04GbGfdvYXzqPOWAt4GWXsKLAC0S7MoCWugDnghourJnn9+p11IiQLCyjxs+xBC7UdGFdfl+T6xpClAajuOzC1ssIpdKWKr39yq2R1ddkQGg99xJYPnHRpxo5ilegIYSA5QrCsUA3WTtHMs8gZWAY76TSouKYa0JSn4lYgVLTJ80B2SrAdmucXnF0ohCZnMfEJXnjYFq1KXMDs87HF6Fbo29Vvrsnx0w6js66Ie/7HaZznqN64rAKr1muyfRcORIVIlz8AS+QNK7PuEr21iDmjmylYo2Ax3TSnC5wjvqcM8uApWCdb7yRmgCVNHoTo0kJ1cb4Cj5xn7g7TRuaVi7Mpy/m5rf705mlAF50+iv4thSiCzemNnqQZ/sQT9ora1T6XdBQOk4r5mZCa5VsZQ6WWoAjlM/oAmOnavF3Wo12U8sX8rZb3YutIVHTFYUCwU8vt6pf8Pj/ijH8pWuxZ+Y/FJG/uL7+V4D/47uf+/vra//8jYFkevkqHNMZs3jikVlS1AY0ZRNh7xt722hNkVlSWV2qua4UAlaDwzVu9EUd1lUFlKjoUkBWUG31bErJmTNi0aUXUSaD5PU7XpMC8znglHhh01qQeze6rXHoURRn96oKTL7ZvsW1OSQryYoFSNaJtlTCBZTqdaEqcsyqLi5fv4RLISpAaxXfN1fiUVHCS3+SWYScsveotxHIxe25hgvM0+ukfTvRDtKN8wTtrzUz1h0whgQpVY1klO5APGnRcB+EP8r9WjfQXv8HtAm0S06dpNYouVorIbxGuXvfuN87j9fCMDzAjwI0wwPdamOuzykWVXgtpkWDF0lebGlvbOEGYlgcxYa1SoTKtZkMd5qDtl5JZb2VIew81pVbFWOhiXWXRCJ2uWNct/QyL2Z5aaCo3NZd1yriDy8DFVmTjlVVBXaFsyFmvMj1eU7OA/b3JX1HioovHW73jdmCJo0cA9MSrWUGc058fCPl/ZzHnzb4+MexKP7YVyQifxP4mwA//PieMSenl/X7jGuEVzdPTynyUduw/QZNmPmtNBJbS1tqVBQX0kwswVENiyo1ohZzApNr8SxS0xL6uBc5KlZVUXmZ9buqXqhTLq6LqYPbvmFq7KJ0et3E4WWVPkupVzaFa0dYtnPfzFqqlRCy4tXW5gA1bbgGHNNLCZihjCz3YlPDemN/UawXs7K1Mn5BC7v0WYNRj+KHnOHYUVmXx1lo+XUFK8fB4Vj6DSk243G88doqKdqiCDa97xUOxUSlDETK1QhGznptlkXSIcgVLwgVGpseK1B2OV2z8hi6sS1K997P2jTDSqi0rM1y1sKu+2kJ5ta1jQBPr/Esyz5+3ReRSWs7vW/L7dpRbdhCn8SLxSitpgesY6ZyMypxXT2JJWKJ+ObOcTWIZE01XGt0HKKodghB8/LCWFtJbxWcunCnutitXM2aYGac51hVtCLaKbp8gdkp8PLSmTtEbMyjeBBIJ87BHI6PPxs/hn90tQgi8gfAP15f//vAv/jdz/1V4P/8454gM/8O8HcA/uCv/IUsm+5cpiTUKvh2r9bNtSu6NVKd6W/Vp4nhYkvMFDUeYuPaCIoeG+uSL5ARvmszHKM/d7Q6VesX2yr5AiizWBCpjUXFll2c4nKw33Y2oYAe32E99zOb/srGlKpuNPk2dw+ep42utiOzbu7MxY4s8mTRgunMIURUaGvbGvd9p7XERxSJKuHaxopNushIskZoZxGEWlSVNmfdpihlC3+FpEoQrciSkYMxlDcVdpmoGPfbC1/HmrNbh21npjMiKq59RQHUhMfXpGaDWbJkjyJhxVgj4K6r9C969BiQTNzrz3OuaYyU0W7r6+9FXQuWCC6T1U4WN6b+O/Aol+2+Nbo1hjmWdfKTsLeOuIDHcmyqjeYKPVJbHJZ14WrqPIhl//XcX5f0PAXOVZW1p/IWuA4bPWFli1TPMiEaSauxpymW6/pTm7161UW+0skc2PcNy/KNPLQVl0w6+Xgjzsn0X7Yx6J/8I3/s478H/sb6898A/rvvvv7vST3+VeB3fyK+ANSseFl5Ua3zVbGJFsOwWWBdy9uAJPwgYlabvTTz9UyT0iA2yt70oiTVxZf1U5cJR9UFsS5XPmv8fL6ydbF9Qni1LVIKtkraLouubWu0XoViZl8I4nN1w5NzyXf/5tumcLUWFK7gq2Lw9TWKHLl0GTXzvjaTbsa+bUBnDDiO5LzMkqcyJkTWnL2s2GozOI6Tx7Gi4nxx40RILc5/WjI7+CbFD5bazI5I3KWyJ/ZegbHNoG1Iu9VmSOBaAONUcHWcUVfAS+DErKCgdBhnvddlbFwmKyRvb86o4oVjCocXWJzywHmrYFmTZbeuTwPfWDqVoHgA5wjGOfDjxB9vzFmfb33e6/4bzrY1ulR6uc/l3BzlCiWSmKy7SkoXkkvqTfB094t1Ya7LP1NX2NAEznVQ1Bhax7l2mhrg4uUETWp9aT0HUTdM7999dme9xvCk9UbfNvZNaa0jrQNGTidrHv8nL8PvHj9nXPlfU0Dj74vI3wf+E+BvA/+tiPyHwN8D/u314/8DNar836hx5X/wc16EJ3w+neHBfF2BIcsb4C6K7Tsvt6q+Y5yAsPGRMxObScRnzmeQRENJLCbllNDW7Lg+nEHZeCO6sgwruHbGZI6Ni3ngKyRmZVOXok1l2WopZwRxOEqwyw5cwGCSUVZeXz6/EVmnhLX3RZqJyTwKBalReC4QKsGKX/8inVNOTp+ICnctxp60gOioTs4B+7tBv3f29p69OTM6X9/O9Z6D8zgxM172/ekeHmMW6DaCN230Xv6H85jkHJgkH/b34Ma4nZADi4lOeHQhp3M/g6/tD/m43/iww2Nz3n+djNfB69dPvEqjq/Fh25/NZUR5Ve5NF303cbeVUu18uDXAiSlYvoAEJw/eZuN9EzYSdjDdsEdjjoPXOdiH8/L+Tr99wMfBlImZobGReWIrsm1y4pK0vdHtTubkD3/zFRlSYLMEYcq+v8c/nvg8UYeIRZhyod06utKuRcCasN+KnTh+V9cKq01wLrFYVyWiNsSeHXxy5oNYbdV2+xHzMsiZVq2mWaBW4HGeB+dxYNL58f6e1upAeDuS1+PgmAMVo79UWFNq0FsrAhYn3N/TOZhvj5+zFJ+PnzOV+Hf/X771r/8xP5vAf/SLXgGF/D+O4EQ4jlF6/3uj9RKNyIqBN1nlYlAjoJUcres5qvG8zvoLp+A7uB/8sluL61Th6QcbXv1+LHKRQKkXKbTeA8wTZDD9Wagi24ZELuDompMIIwv4tOULaAugfoxBNOeFjlFoPW0Fx3hDX4QWrTb5rPK14AKFNpBMXvaG6lYAa1doQTuV8/NZHgipC80uzODCyJ+sHgRCECvaeKRgsezc17c33Yp05JPwLCMUKWLSblcxndg+y1BHBrl8FJpe8XvVzEmAutZ40J3MYhZtzRAmuwknyQgYPmn4QpgMP6Pox+saqiVEOU5cmhLIwiZEiPhml+c+l0Br+XMkjDzxobifbFmj66JUGLYFEoJtWy1MvxLOypINX1GBVq2OsSHeGONAu5TZjbDcwqsYaLZMcf2qVKsKKOCw4/OoGtZBmi3eDIwIfBYGZr3aq8xkzEnOgFSSzkkwvFi72op0JVbxQR7O7N+thZ/5+FUwHzPgPIJDjTnOSvNJwawjG0g45MoetEIBNWR1D1ersMpxd1ZabD2uTQFAqwS+arxcyPHKPSp2olamomQUICnFg4jpxSRzUC1VYXV3yey1QawunVDBVRlS/gEVERGkFIr+5pNNG0Mmy8iahlVqtZ9s0p5fezpRLyKWqUKMwio0QSpzUU1r/OkTYWNvJbOePvAxics/Elkq0XWUqyzNh5Xwy5PLkq0cmsvoAzFCJpPkkUFa4QScTjsmIyfejLF3tnmuTlCeAC7UZRru+JjI8qFsrRamWmlVhGCOwYxinbbmzKgmULVamUqmKhp8u11NoC/QTteqW2xHqful0Z4hQzOCGQtYbBtqfXEPHAlfZCgrwikwMpiLMk1oZZxkjU+XSzuXGuIyE73aCMiykvOadl2S6RRFfS38K5d18TrSoxCwsKfk2rRGoJ5LBk7lhEaDU8pqQB0Mfzpf6wqAbZt9M0D/mY9fxcawRgSl6x9eev7RiqsMsNh9VxUg6yu5iv1ggUsZRI5vMyPNbzMRyWIPrv/8vg+8QMDKraxwF4tvC90BGxMXoS0H5OrD61cdK9fSVrVSaIITqpgsL0RbVuVZpWBLJdI5srIfmlT24JGD7rPAnzWyCi0lYKTSspSnZx7MGMBWwh5X0qva6s3ofWMM5+FjnXS2FvzljsUFvRTAdQXKzqjMSwV/c1BBZKO1TrYDn84ZgZ8PyMYN4yY3TnmAJtpX1SZR1y+qvGXRLmImHrPs0DUw2UGNQW16WxZn4PD6uz+1kpPL4hUYk6HxHD3bond7OpJXFlhhKkRZutd+0bBlxSZEESA2QBuhbd2ACaO4BGpRBrS6KPAr7CYsiKnrHlwCq1mVl7SqWgLwkeQKI3YrIpNp0LtUojilhjxeX7FW1UpVeFEg8RyE34ksdataMXGmOxPHEHqzJYpTxumkCzpn8VN2SmQnpVa17c9mKvGn/hCsFm4W53weATZpe1s+j3WvQZX0NZKKuhl5pveRKHNJrr/BLbn2iCq/6pulvYiLt7DAKiGRCI4RiMeTS2DzwE3ZqR041mRC9Xre74gAq7TdpCOUQ9BVVoKxS42iTgpfsKCcl1VQSp5d8pwathUnvrI2OsoBDM8yaQXwtVl5cRwKpBU6QoTRFni3bAUJqwmJXUi4UXHtgIVBrEi4dDSySFtmpDVGAqnMt0/E/oK9/IjxAcbE/JXbSNyueYg/uSJ1weoINhNMlzXdWsCDpKuVQGl9ip5GI7GX/Vl35KmYO72D0mmtJPbp/gTYas+L53ZPCNLW+6zhJZIPhi8lok5Ksn0WkE3iIlyKS9IL+JW9wLz6dEjVVQElff9GciJK+Oe+3reXFT2Neo+9JieiyVu+obyUz6gAetnM6Jq+JNZqujPnWCG8+uSeqAg3lMMnczgaTtdeuNbKAqWxks1+/uNXsTFcCkbNimifCTmC9mWy7Ybts/rAFHIW+2BSffSSIVHngIK0bwBsrqn2VTjERaRhtRull4i5btDWVjmWfD1HfX1WyaicYEbsZQGuakiXKjsZWFlqALWzpyS23EvTE5c6uYTGrsULUAC3RWmo2XOXxlijzYY+DadVBGNVHrFmZr5RQ7DanIyiaTuBRn1Oe2/ovjH9IL16UYTvnLDnBWAUN2xmYQCrLJ2r3fLZqq+WotzFlDVCDc7DebwO/Jw0t5rzK9heV0aDWghzneDthum1qWlBHhefg5KFb1RitlOTAKXX9ZRiDu43pa2NIWWW12QoFivOXhxXB2dxXABdn2sYuVuNfLPuppD5xIi+m10Xz8IDbUZePqMrFrBJ5dilHkVKQspbo8giS5SXiBRJKi78YW0+GpWF6VTbZjimCVJt4OCxTGI2MpQ5FY/KSEXBpfwqCIMxGI8ytlWHpAhbWE2wxH7ZAPJXsTFAmXgGn6HfOF4H/njD1Ln/8ML2Ya9+8XUS5hWkQWILoAtdkuw4KxOSFyD45thhRAg64UwpY49euZAmzmtWibx1JeyEnIjupf2LBxkHxs5uyuxGdqNtwr4cem72Ewdv+AHKg9Y6rb3nsx1IHIuyp8QQrDvtfWO6ss9AvUpYUyO0MXkljpPHZuQOd+n4dNQa77SchnLekPG7KpcJmk8kqtxs+0bGIFJ42e/se6M15+EH+XogjySPYnBYe8F1r0N1JjGjFKDxhrtj3TmP5HwT3vLk9+/vwJLJwc4dYufTl4NP//R3qDck7zzmgYzg/tOdfG9YZkUMPoLjeGAhlVS16olaO063vfwVSER29l7Eqq+fHZmDrlHCNBe26LgeWH9wf/97HK+JPyZUFnaliVvw+KPP3F4+8K7tpAm5FYndptP2j4zXwdvrG34cNBPe9XfsHz+UqtQnp4+qCDywwzjmZ8YsxY3pAz8OrN25v+u03Pl8vnHMB5s0tnc/YeMNOd+YlK/GPI3Zkm4lGXwcYNtLBdJ6lUqCVctF0sW4cWOLDTrIizIz4PWByoZuO5gwvx4cXw4eTG7ng9mTQaNj3Em89dL6/ILHr2JjEEo+3xLaImyYKq3bd1Zr389hY5VzC5MO+FaAXrHgF6izaLLVjVSozGo1MitlutKEEl3hIwunpvb1Yh2sAoOVWvaN2sxqJVbJyno9qYVXaObqXmK9REM9OdOvKn79hqQDUxWfs9Kbg6IgF2ZdTM1ULj8CC+BpPSbceyOk4dkwGr3tRT822Pp75NaZeZBM2BVr5elwJW0VQ7NQeVmj36YOVlySsCIaGIpZkcsOn3w9TzKMgXKqsbEyGcY32fl6umWlX4Rzwp/szolfTLJieM2iyL96Yo9gmrH1Oi0DaNuONfB5yZEXeUi/3QmmBWomUeEtpmtqIGgT0ifjHMxzVNZnJFnRVHVoUPgqsxANWQG1F5NSKIs2x2l9VWJWJDdrAvQyA1qv50o0Cc9KlxpADDa7DinQLCxMNThTGGvYbhdKjT15JosSV7fdwidUo1qMXLhcq8SM/Ge8Nf+kx69iYyjiTImNDi/EWE2R3svUY2VFXvBAUuV66BrZrMYua7JTaH34czO46kO1pBKoY9HOBLJhVniDrdGnSG1UuUoxT7sA42K6Za4swGBqnXIZ3wYgefXz+QyAYyWUcNmhXdT1tm4Io7qbZsI8v9GyWDQsIRg6kanVZwbFdFu2eL0Ze1Nc1yQgFsFFK6zlZTdetp3HPshUuu04yu++lm2Yey2w8Kwbe0XIdZmonZDOFGfTRhNl2+8Mr5738FHvVRopy8o+YR/rPWixKLEK34kMGH7loawNu1drsrIZppZbw8DRlCIgrYmJqK3hYy3uXK9dCvCpbEeS3hqtSQmgpgKO7la07BUZkB74rETT6UGkL2q3Fl4za2g6h9doVxNDaVbJXvOcZeTSi5fSEKQXCVS0FXg5qtefWZMwn1LkpANmJB8+1rr2FTx02b56CicHddhtqHY0lLkInuWAVZ9Tiizrt3ouy5V+1l6QVH5hrMSvY2PIzCcw5qPAxUJpW3kD5toVL8uzKKnrM78hqX+vvVmun13dRCHruk4+K1ApvgGRtqLodS18w5YYqX5GseLNay16CSoBino5F2NTvyscHMpSDtaq57qa9ZyxbNYEkBrq6Ur7FpKedTrFAtEmyVG20DXfdghV+jKUlSaMOEB6VUgW4MW38Dh42Y1mHb01IoQ93nEM8PM3pdeIXP029N5RvejIhf6PmEyClybcWmezDu5oVkr5091otQjzUmlR3hD1Ouo6pWeBvR5PfsZts2tExGBxTFqiUWIqFS/rPlkkwSykXy4AFa1NyWrMGMDWNqxZbRyjJPTWDFptCs2ErTdyBqJWeZLnALNKTM/S2mg4OSdZzTvIVhb/dJJZbAovroat+1AZNS6UundN4Jh1NecMfMI8YaRAF0KtHLmoCq7unIpitDRqeJ1VQW5VBVRJm8+RtpLoXmQ4o86iNLlOqV+0Jn8VGwNAz6oUWonO2Lpxs16L9gIPVdeb/OO3v7g2h+UspEoh0gusKcCuRjgXfq1PYEoWKFfiILNvpCZZIZpXGXxNPwQWbbkW4gVAXY+6lNeERLiidBrGC7WgxKRuuFUiu1i5BknNJCKKXATJIYVVjPAFt3JVl2BwrFFbdQ9ZeQ9emo19/0DvHV+xctt4YYxBZtHCr95I0ZUqvngRanxvJNpMedlaWdfPyyexM85krI/eY4F6vQDhecylmF2vN2oH1qjqTiPKQzJWDmNU5Mok2S2f9GilNgZb4TC1167MEKv3nqZr+CFsFNEnfda6yFwj7Rrv7d2I+47S1sZVlZMIFdQiLMu1uqajhBprM9orIcvK2HeGcrlAcQa0o+5dMyaGao2Gcq4TPutQUUqYxaoCrrslKUKfa0Oko2xrcjcrzzOrhD0pLY5QtgTWNtq63cnknGe1JvHLSoZfxcYQBG9MZivX4hmONNj3Bo2ys1rmqc3r2C4T0b46BnkuwLQgj5pWxJqf61LtcU7QuUxBX7i4ewNIds4lVqi1P4BSWabXptOtSsTZAA2aFxPSZZALZS+bd2VXeLVvOIaWIBAFel/SAxXShAF4DHJW6extmbVawznJWfz4wye4FTf/AkEMiCj7O9vZt42tQ7dybKptoUxMRZTtuHMeB2/+GXjw0/tbbUhZFnnnDL7+0QOfyb0ptnU+fHjhx0YpR60j0vh0PPjNp0/89tNXfBrgWINtv4DFQt9zOufpzHlZv1Mz9U0rj2UGYyRvn77SVjv2nDMJ3PdyzYqo66HS2J8zf19yZqE1ZdtA2oqBy8QxWhbr0FoSrTbYfBsERtvgti9U/ygHb+c6eFb7KoLsnVvr+FvH4ytOWQOknVhLZN851gKcp5Q/RZ60d3fstnMs5enG4sQ0gV1pL8JLAufJdawh5T9OGGIbL1Ij66vNwJU5Ji1qIuGUtuL+smG7gJxoN2gV+/f6+gbWafrLlvqvYmOAXLkEsHWFdDZTmq3Kjaf9xTXNRrlcFRbjEXhG0V++fZTRS6wf0JiEHkgayL7i8IqDkFSCMyGL/DSWeKX+fi73aNEFLCK0LEeGhVqswnnhBkb1MUFJc6/XrtVrVy9bhq1BWYuVqMeXiaw9jxChCqUoHBDW66jUtVVCZmC0517RRRGzb3kNsuTJ40YifJUqzf5v6t4dVrKt2+/6jTHmXKuqdnd/D3x52ZZIIDAJEbEzJBJEBgkBgQkgIyICyXLGI0FCsgVCBIAIESERCQgREAASkiUIrvHjXt/zne7eVbXWmmMMgjFXdftx7fOZi3Vu6WvpO/3Yu3atteYc8/+8LvUWjyhN4b7vbM+dfQvWtwu2BMvauV6UxQX3ukEzgv0YPLcN8QXTwLrVLqyBjCnOctAZhmoKrWWFzFCZEDlgeDCkPp+VCcIpuEJb6hxWLdIQWGlStp2MQb/e5jGmzE1FiE5IWMq+bihiXgrOqWqTqRPwKfcmS5YW40DSXg5bPCo9HMP0ClZtZx5UZZ6VAc+lnI+eyu4yj00lW2fqOYpmLLBYZndoy4PtOQi8NjGpidWZkv+pbTmnDFJ4bIMFo5mUJZ4yeS2Lse2/QUq5VVkMe0CLyQT99NfPYmFo0rBwnltwW1esCUszPobwZiCyc/zmxrAHtirLekU0uY9n4QWZFdKqAEY0nR+iv8Q/ZkJbb8R4I3Ce+pg3g4BfSLzOc7YAwtid8bRpohm1c7dWqL3HVKtVIa0n5FGafG1XRBx8Y10v3PeNr8fAD1htQW1BmrBwA5KmZdiN9Kp0H8nt7RO2gMfGsQ12r7iyMZ2RixjWK4KuwLmkZRWeigSiC1ibV/fsT7DKhxDhaOUSxZO+NURbcfsM7hvlA3HnyHdu66/58OGNt2VhHcnzsXN/7nx5/6soya01vhwBamQY4xH0641lTYgd3858A/AI3gSMBdYL3gzjifHg634wzEizqv1rOtu5E7YDY8VynatwsMcGmbR2wcJoXOeDCZtnKQEzsbcVFkNj0M0Kw9gHe7tzs5Wvz8Hn40GgXK3x3J/cdEVb41gUX1u9323nehMe9wvbs5ydt76wLDf2PWgHZJTH49OStOuVbAbivC2d4/HgGO+lgmXB4sx/6LSrlErycBgV+6Ys7DyqLWxpLFSSt8fgw2WdPadHReVJwtpIMa6XT982jabcrlrBys8/+szH/99fxTLUr1WC22q8XRq/+tj4xeUTx/gDfv/5tWi2bJRMZUe1v6LkMUr8M2mIM9zopDlTecloT6KHSYRCTQWdNkGt4Llv7EeBfuRBn3kPLwwhS1VoWWfECQO8fh4kaKvNY0nZekcaB07a+qLpgJdRBwxsplh7VfXtDrRiKyznQmWlumPuMpUAVIKv15c9MQMVMhSLHZOGNqNL+QPicJ7bToQzhrEdFY+GLogtKBuNMrETzjhKSJQM1naDloy+8/Vxr8nKDaejxyimqFXfpnbK2yKUN8AK2Tep1Oipf5rinyrbFau/f17gGrOl8GJRTDvq80pqzZChvc7bI9GjavhyONHq85QJ0rk6LY10Q8LoLKgIa2vsVrPHkcVSVJu1v+6W1g1yWrB17sKaJZRjekuYlHWO6oqgwmGqHvAkp7NqGRmT2ixPzQnCQyliT8wici/2blSWo0qQZuWZyUnb4jhLNacZ5TnSrJbs3w57/JksDFm0TZCsC1wujV9/uvAnPl75dPkVP3z5DX9tf8ddS19roHQAACAASURBVD2mztJHnQ2h7hqrZqNT4JWTu4wX6PKt7zdfIE/947q9gq61q+1j8Nh2xjEZxhaQRU8VDUEJ/yfW2ZqQQyCm9HXSIctq2F7Hhn2UXXxEcVnjJZdlZrkoqQIhxRIMJnqd9NvUy7syKBXkbgW9i8a3I0P0l+7iPNLMAxBE0lqh8oXnLuxt8GM+8b3KVL4+Du6PAeikxSf6nTDmZ1LsxMFtfcPSiV35jXwuO/loaCpoVdaJnaBa4UMmdfzIYvFQqkGqztUKqiU9bjoLUipS3lTnzzRboQSaV6FQZMxFAVJm4K4H5oZLhe5YVoFRESWFT8HZ+yA0jG5KX3oZzKIkkDlmwILqiwHpTehtASoc0P07JcvZIUJtFH5UM5ZTuFCKklobUzEYyfCdC1e6SRndPKuVPKZoTbM2uTPzIaokR1Qrro4q3fFtkF4AsInQWh3LrVcMwR9LE9UkcUCCtS/88u3C7/ziA7/z8cJtfWPbV/x4Z4yaKxorkis+mYNTQTPtM5xZjVAoe8ALLyiH3pkdXbFiEUWXqjmD4IgDHwO8KtD7bDIS6oYRS8h9jiEDWxdeQXPn5CBTVJFKHlPYYCWuEjEGW11oL4ouQqooxLIMRe5wOOoVXNL0NAhthFgF2MxBwaZRRqMxV4tvAgHqP8cMAhlxVCgsiYuybcL2HHz9uvP+vrEdjroiGEHH06r9OZ33bVCBeM7aLmQ3lrUwjj2CkgPVDtkCVuYb1BnflloLYszko0lV1oWarJDVomA2ESQPpAt4sSVoVcctbmgEG0E5ZQOfoiNrNhOynBmcCKn4iEKBLUkrg9mR1YTVmqINVu30QwkfjKMEQ2aGi0JUWFDXJNIYI2e3aYK2l0BLOR9iebE+NTH0E2KEGdbqGdVhqg0RZehBTK9KWE0G6Bycpu9cZwtYKeO/6XjCgxh3tJXl3bQq7SLiRUX/1NfPYmEgg2WBy2J8euv8+tNHfnn7yFWUmzlrdsxXvjx/j+3z73H7+Cvy15+4rA3NqFTgmTGQIyrX4KSARCbdW0pEfw0ZMoUwlJtt6hcyhHb+viXZEu+w6B2VjupZ3FEcQ1B03kx5o+mMSkd4PEH2xk2vrJeF5dq4XBrb8879OSBOSfDEQZbaMY896TZYe3CZFW1BMGRhaQsixqqgPvMoTCpyTos6a01fUWSelbjdWiNH8ti+lkpQShq+Pd/5+nnjxx/u3O8VcPvp4y9oWknQ4cr9s7ONivE3FYhGLFvFvy+C9DeQvY5B7Y63W4GgezFGM9QNBxar6jT1Kl/Zo3b1mmZmA/aE6jRKvRjrBBCxKTwCHQlHYgN23xj7g5SNfr3Sr2tVC54PLsUrHvuOeLAEXNZeDn5T1JaKdFPlaLUQHgy2qVxVAppiOuv1VGqniTrjq0q1SevAfIMYJbbyTlgDCUwaTRM/xizjZS6EnUxj96o3rG7QipdyKLxFJl1ObRLOc4LapdsQlOv6RjRhPza6Gm1p1YkZey1Ofs7LP+31s1gYBGWxXjd0rwd6P5z3w7kfn/nx+RWACKvKunCkOd2q6q3UopUzUIhuPbJnAMf5Xep/3xiO+jvfwIHtGK9V/vXPdJ4a5uZXMYbCy90kFVCqWsk7TTsm5ZBTkqUJXNaS6l4KEHy/b/jG650yR36b37UlNKsehNBGxii5ttd5GaU6t79JJuqnMsd0oalhUj+ZuzN8p0kVpbRW3v5GpUaf/13BI4AJy7WKd4TBiEFsyT6OQrddiSHsyrSmB2adlkJblLUrR9zBhSOXysUURbqxAC2UpUCU6oWcN6yZvKaHcJ2qzzp+OAZR5qomzOAe/6ZNMSVaidCYwJvNmTEPJ6exzs1heNGiphX+FzGtLInNbLmEYovmdTk/56WV0nSSHqVhwEgagwHpJWPP776GeIHravjuNd2441Lgdbdrea7CeYzB5gdM1b50ZrK1QRyVwRlO0msCmT4TgD5pYhuN9GTkwdif2FJq03X9Y8hKACyyYOcHTzKOwWM4X/XBD4+vuFyAFSLL0NQ6atBP3G2GoJwS5Zjn1+oYqB1XTvs2vHZpPYEC4B538Aae8yE1tHJ/5mg+SyLPB1IKJPCUVx+t2fn1FWsOS0maaIYuxhDn8eNBH7MiTYpyrRKS+rdLZImUFiGGgdcxSzInEFbn20n4TelPwinKoTIXzizDI5yezqIL1gKVRkvjyIPWLvTFsbbRekIX2uq0DmMf1Rc5kjEGvdfP5a6MIaTU0c5UMVtYzFi6EPfPEMaOgjRsEboJvc40tNkAFimMyFrc50dakvNaHKqPU0+k+JtELE8ZelG0pw44TepCOUgmeBnGQhNfnCFVN1jPriNakW7DAMqf0ufiVF+pppM2r3lrUq3YURuMmGIzX3T32XGSgnqbb/kEAVdqX89S7Xq+jgUinSQZkRzuHJ4VwY+w9I61lVMXkl59JYFOjV/M+zixtRZXbOPYB8ezyoTXduPWLix/1BV1/zBeZsrlVh/i2j6yb8Jd7gwGjz/4PTZv/Pj8kQN4+/SB621hOwYfrjfUOhq8Cmois6rDJ0BjBVkz85jKXZmVuaBkiaCsjp7jx3rE4lAeByDFF9/kgmpJaNtF0UHx9NNjzwBtUhfnppitxB582A1fK2o2IvFtJ56OjuA3+6CJs1ilCxXfLRCO94MxGuKtslPaFSRZcW79Nkfw2fkYOYFKSNm5fvxEa8E+DjZXQo1lWdhGMnAWWWnaqzF8CNe3X/P1fudgA1u5vf2SdRGO3Wl0tn1nq7gj+liQFrRb0Frj2Kvr0X3QEJoYjYZzrR7RMVO2vDrBQpJuK9IqpGaMO06ytpK+Qy0M4U5KZVKYwvo0rNcUY306SEZhHW1pdDGGXUhJXK2auzC+ft6J3fl4q8fU04sKbAveG++/GVUdTwnjPHO2ZTnqQR+tmsF68HYtJahR5/3hAqOwDLPkfYs6Rq719bQ1ltbpYjz9zoiOawmz2oAujcuy8oWDj/aJbXzm4XfGSG52YW0LNvrsShnggUdNwAcP4kgsjHVZkcXYfKNpQ9oV9QdL83Jn9guN/l3o0U97/SwWBhXhzZZSKKZz7M7mG7tvfAlhIEQzTIJuymrCdYKKGtXD8HdAK+dk8B3/QJbH4aStQsoRvQ3wo9TDCTWeSVSkVuWYY9O1dkqZqiOkgkVyWauHwW2aJBy1mOo4BRuoFk0WBPthqNTZlFMOKd/eq4l8m0rmG69uRWVQjdqPsZXfQqkpRr9NLScRgdZyKID7rO4TI6TG8IjEnwdywBILpo1FpFD8LFS/KIyp37UqeQ2Zgst0dt+IcaAomySDA4teeF+Al6gPWyErM7dAslkLZyfm8z0GeR7bzp9efB745DTMEuEERemSUZQqgfrEVbw6Tn2GCltT1umv6CY8t6NYn8xXRKCKEGO8xE6nWsxN2a3mxxGV2rTvFajb2lJZIozyuVj9AKEVKCxZwjU7GSQCmxRjuNB75+7v7LnRtdOtcJZDBspS97Fk1d3FvLFtIVt1TIx5tNmOgzHAMhBRWqvCXNN5P/9xpCtNlY+XG46z7clzP7gfT7bx4CELiaGLoyEsfZbM2jTjWO2aOaGCUxv5tzxX81OpG8m//dmcUn2vX2PMs2lEjYkmWOuYNdQSOV1SVouZWanPWA3dQFILmDcBMcZ4fqdtqJSoY+oZWpsjuNZx51wLTM+ReXLdUljJq3Bmegn23Yvq02rgklZdFfq9vfYcN8kCtFIZckDW+XfbDh733zD2AWoVv26jNCV5EiyFo6sWYyAmqCS9Fd3XrBSHHuVUHK5ceIOW+DjIzUmvbMIoNJQxfB6byvPAi1maP7ue+En9rZjvoSQh9eYj66FOP4jU6VMrNeNxOPuRXMrhhTRjaYX8u1fz2Pt9m0Grk3XW6cEYe3031VIqWq33zxws3ghP9t05NseHY7KAKKtlhbAKYIZjs408axKJwDRpVopOpByync79uLP5jkpjaStJsrHRM6ZgrXReMX1mMdW5miX9zuFszwOToKWwXozlcsG6chwOPv4we9Ef+vpZLAyqFRH/fN75a4+vPH1KStrKzQ3ags/dxrThXdkUxKOERVGWVUJm4WjOXfg86U3TSvBd7Nh8uSLPgWyB7/GiMdeL0taV9dpo3Vg0CqQTr8JUUVbrLFZn7bhW+GYoM+/fOLwCXvf3weEHu1eX5eVmXIWXf0HmzrRanXHD6gy5Z6I5aJlzm1a24wCUt8uHOSgkpkXAuki59Ry2Y1a6e52rj6NiyB4DwjfuY/D585PPf+OdjMbNfoH1kiz7Fujm3I+jroOV5PayOhUvmchF+KQ33sbK1/Wdpw+6XVjbjS032BL8QJsiq9RneBiHOp71wLemtGYl/jot73X1ai7Lir9bbBbQZCJnmc6IkppbDTNhnQM4hpey8bHjl05rCwtCVyVt4V2C+wgew7lKmwuNT+22sss6OzYLLEoS3w+ej43BUnV/JqxLI7qB7hw2WJYVRpCHlwdkKYwl8WIGwlmkla5AltcEdow7PArpDALXQaOxcmFoLaBmRV+Wccrw+7M0FrVlICbcvLQisuSrsGc/nC9fn5goS/tjCD5GBmN/cOxbqetIZPYtng9N0VhO08rPb1Q2pGa5+U4wUTBSy3swT4+80KuE6zRgRIH8jJh89iktnbtTW6pIxTpoq5W/OiSrCs2azjRhIA2dhSThg2M7OJ7B7lYmoRHsR3JQxpdLh843HASDnEUmOrP5jqmMG3keLazAOqoBqheqxSsGJEsKnVMsVNqN+hrFqwuHQ+zOsSfv28H7l40fvtzp7UpfYbWzR1LKLEYjpEDxvjbWVpZmwRnPA9+csSfHMf0kc1sfPkp/QRmGzrjBjKCS7e3lQFcSRCsEhgL1zsGsdFn153KOd1r4EOFkOCbV91FfewYEZ90XI4o1ILXauEzKMbtXf8baDMmssp35MfqEHQWQrOIZDUWHglbZbdNOFfJmYTUaePQpQvpOAktZzoViX8TBWkO1gvvO9uJFb3MS3IqUVKNZY59Z5BEzZXKi4GH2CicSa2UlvxlNG2kDsVJtHiPIKFGVn2EmP/H1s1gYKtBkYM1Y1jYLUst0UuEqxTbgZYrJqZK0LKViZEmazgQf+QYwcI7S5yPfRKth2Sljzpjt2u6sbaLiULx6rzOw2ClfKprPTEqroPG6uEIrodIRHM+d+8PZwgoc9JmdSH3trjWOy1T75WRLzrhyqNTrCl7N6nSob1M7ViY+R141rRwHNbIXxXX+2KHUgkmdCyIrRLbi24Ixgvd942rVnqyr0C4Lt7cLzXfensX0qJ4TQyu/Rhjvj419O3jfNnQYC4akMThKBTpH8/N4FeEc05quXVHJuu5ZnZCvSS6k1H5R94UAoSd3FBPVdzx3Ir3q5LKueZsZDyVGrH/bRRjEFKBpKUBHUYu9ManFKVlWm6HA9eGdxzpVo9tKW3yKrOY9G1UcqyHsx8QAQma3w2STtkH2mhzlOO8jrZwHq+6OLIQR2b7VE9Ccbzbxs+agTqlTW1u/F5OhUMgOeSZRjaL3l6Uk0mdq2U99/SwWBiShK0tXLirVtbc7jKSpTwNRnXuZZ24INOOklOcOM1E3+cO/1V7cZeULVilTIZDuyCIz6abKWlQFqSTY+dKJR82sB6n3gz4pf7iSe33NfcD92F5uR5ETS5gEea+e7QryiAL1zp9l3gSZJYU9d+KUCh8Rd45t7i59oTdlkYWj7VPZSfH5eqZfA1Ri0pwf6mMXaLrQbeG2rHy4KLcPN37VP7Gw8fi8Fc6hFVWu1tgO56BUgQ/fecQDo9NZCInaQb0hHXQBqsyKY4mZ0NTKFqzOGfd9yhGUibGcN/rU5CRML0Ki6QQDj1EOVTriJ81Y13FM+7whNJ0JTORsFQuWdJJB13q4xjmRmRJTcpzzeiOln1BrtHWHSRVWCnTpQMQavh31b0yQXhOFRsBxgK2kO2Ovo5idIEoWTf8ohTUmRsuapoaNWlDRuUjWByFW+ErObk3XoyhkG0XP9l44BMyukI4zyFlx8FNfP4uFIUXRXsWydh/YKNxBrsJ+d1acq8GTmhaMRrc3Ip5V+IFUE3Y3tAk+Rq2aJKDT7CLgyWLOsMaW8L4NHu9Pcnd6WzBX2jrQ7ogk3UpQE5Hc9A2Pg9G8kOhWlWojBo1b8e5Zo/fdn3y+PxgRVcnW5mKlAtnp2RjiWEusjbrwh6Jr/WLrSDzBj5ntd60HSQ5W/8SWO6GDVRqrNqxB2MaHWzlHPTaI0ukzn7/ASN+Io1yVieMa6CrYpzdut40PqXxqxp/5x/4kP379wuV33vjdv/l7/PD5M50V1SvddlZzxsfG4w8S7juxPCY4bMhD8VsAnW0/sITWe4WLWLKsCyMHwwdNYLEVw/Btf3Vv2NlyZRCHc+HC0DGPiFa1hCag16qRh+l+aIgJvYGssDdn11FmrhHIs9NotOvHWZi7ldbDKuylLdDTCHaSjcRoutK1wZI01+qhJJB2YKswvHHkxv3YuK0r63oFFXy/E3qQbwqfN/Zd+fRLhbHz/hisa+fXv1zZ6PRdgRveZnu5wTUbT5HaEeWJqSNyA2/oMKQ38pI8LVlNeLPOzp0LKy5BWAGXNq3/449j5qMkaFa240WctGRzJ46yAFuLilKPRKjUYD2YbUFBqM0at1MJIyfsU8PDiWgR+FEagPvd+fL5yfb5yWJK12I7utmkqazO/7NZNKd2RuQUwAhgmCaLNe67E3twbAdj38k4KCGeglj1KerEC5Yp0/1DXqmB5RSspOHmJe1WYTivEKu076YlKAlxEzS1DEVTLJZMtkIran7zKrJNr0X3Arz1xqePH/jFLz6yflhRHvz45W/y+csf8MPnH/mK0S4fCJJu8Njq3GymXNcL2mfEuQH7ey1kWd2eCohmAZdH5WdKO8VYQHpJuucx0c8INxXWtWH5JGwUU5eN0PKt5OzxnNYEjnR8DEYGrlnJTqosBB7n2DdZAptjqHxjJr7dMI1XL6AI0kuuLUQBi+74Oblm4ZY2W7Vsjj6qQJbE3i1IjclSQKbhaWyhRAp7BjvOYCdxGsoRAmEz0HaZLE0DkvBahL8bK6mQyeXFTL5ClKOOa+f3/qmvn8XC4BF8/vJk1c6Hq9IX4fMD7s8CkTSNa7uy6gER5XAja3o3LcWbltDEsnzKrhRIpfPMKICUPv/YnMfnje2HJ4/7Rvt0AVX6stC7oatAW1jWWf3GBMGYOvnKrJ/4Q6nx3t+/sN8H27Zz359kwNtqsCi2NlprLFJeimyB7lOoMx9qmWRKHHBEKdwaKyHOncGajSWMYSXkSpNqkNaZZyHVf71Ixy0JA5cZq0+yHcl9cz5/eXJ/f+AMmhhvv/jI8nahXZ3r2ycutzf27cHxeOf3P3/my7bxdRvsj40njlu5Go/3qsNDgw+/upAC2zMYvnNmN5qXh8OWhrWGreD7TpMyIB1SrSAtvVgAvukX6qRWbsotV8Y4cRlDxGh6B4f3+VnhQm512Fvtgl4aj/W9Hugs12t4HSt2cz5lL0zHJjYl5ZVxfYA0kNJimA1aG1zbG7tdefqDbd9rgexFpR47rJdO70qzOupIVKpC18Z+UQYPtn0rsRHKkfDDttPZOHIaoDJoGKtX/sKP+xeaXMBKHek+2EcV5jYMHQI28HYQlxvaLuzPDWuTZldh33bGcXZh/PTXz2JhyEzGOOhW5h8X49KnAGccLK2x9j73nkkbmuHs377IeegXq4ryzFeFpUi+sjAH1VdgppV640W/HbFzcENUaGo0lbItKxDyUuLJecaM6gg0tCrl98GYu1VqlZtq60gzrAmtnWdcYeRsTzrfu5w7TO0+wyuFOnw+Y17vWyNoa7wWkUqejplePPX5MLX3tSHOakU2P9i3ZwV27BttNS7XTjSpCW0cPLeN5dnZdMO3AbmzLsrbZYVj48tzr+zGHEDDLFkaLFo2432qxIzBfNJqwpI6h4spbdnKHnxU/rGygEKTfWY5tpoeqHLa52NwxJPIUfhMu9BsQaUWc4nHi8kZWRuAmrE0RbWxqlWT9LePuvKNjpxJ3YX76LSuy57fbqVzqkutPEipyUdMpyNWJzgITBYpRnCQBSJCqW+t0dtSQbOAWpCSZCRIn/L7WvSMyogQKfYrfHDs5QAaY7CPQFoldQmFvVRx8Hy9GtWYkEvMUOPf7pn8WSwMAixSnvha841VyxQT60BbKc1rLM4ZhFYij1MBpypEL3+6+vzgswI0JlVRMnqHTCnVXVtgEQ7f8C0YV8eivkY7P8mYM/t8owlkOO4lg0aK7kqoI4xUd4EgHGqVtiR1VJH55XDFpAJdT1pO6kvVKJhSqc2TjLRp9w4tBaTFBBin9GeGe1VEuQcjnC2qwm6aN0vLEQMTZ12Edm0s15XkYIsN3x+M8cbwwdAkW4GGXTrrImwa5HjiJB7OeqsHbG2NxTrHgE4UAyBP0FapWpbQCkDNCKw1Ikd1WMwpzOnYGWk3H2yQEnIdyRiG56hJJUvZ43kgESxRrU2B4vEgNBiTYn2TxoJV4QuBYnSp6yQzzQo5w2RlXt/pxJqPqWKkC1tA9Hral6Xhoxiy3Ue5VoLyqySMmGcbPa9NHVF37ROJnot7QNMVlwQZlR8BpfVQsLTaNPcDz4HnQbjSL9eaGKggGaSMfxXNfYr/a8GK2YnBH0uMQUow0hZhZNGHGUkXgbWszeOo6DWfqUU1opZDMLKYC4kk29x6J6nz7eOoxcTEykAzd1uEmcI8nWsVwlgaAi+133BegFiJqZKQCorpWfmTrTXUDRi0USahMZIuoN2Q0JkilZh8p758fQZwPsWGlI4BB1EWdOLkg/SlUHsv9aJIiaYalX/g4exHpS0dXlw7s3tiWQ1uS1WzXzuixvasyaQWqcJNRo7yYqCo1O6EzwW7GUnRtQvVxfgKm4lauFQ7WCNbXata1ypwpKLQq5lpDeeMRA8qZWjkrHabXSLWrcJmI0CrwzMiCmAdB3hNHDZX7YjvaFBucw+uzAiTCn0RK1Viis4UaKuaehKxMx26LOZCxxH2cNQds2TVxhBj38asoauzf6bgDpXXCGfwzPBRC78tVdjcHaVataC0FjJv2ySIqadpprUoOERFh9N6MQ1dzsbW6hJhL1JWT4EfvPpCEJlg+U9//WwWBltBW/K+1bgqDk1mD0AK+z44xoH7PgFB5dI/ka4gPsMtmMNB7aMi341Y1B/23mCUx6BaqU7PfSJRjcmVQl3ocwSMLBpTSpZQKTopMwykxEa9n6KTYHMgnBFOaKvgk7aUJFud0sTZKcz8W14R1M2pJW6RmA8fg5AowVJMwY6UTNvng13FtoXAe0R9NvP3l6Ux1oZkxcS1bsSABytOo4JyFsDYt5192xD16V0YkA9MnLZ2zIx9N9AkXTikMh98bAx/YryR0ok2HYfyTWrmE3foTbGsTKYRwaDjURXw9bDVrt264vs7yVYnJ0oiP3KFbPh4sBgzk2Na6d2waHhaGY/ynJmsAk9ESmN8Wuex2SruIF5Aqc3bTIxM44iDFoXxNTGIxr4NiDNOntd7PjcmlWoXP/LJQgNpuE3tQ5bIyTlY5Cw9nhobrI5NWYuzzwXElkbvrSL6yBc1fTZiETtyvb7k5JFFUWN1pPltXj+LhYFMPDvQOcadxz7IFEQSfzcOOfBjw/cDV+V6XflwXUl/Z+w7WKO1pc59m+PasThY0ghpDBOOHJg3YjjmFXixrEY4SBN6q1XVfeD3QNcFj8G6BpduXAkiB7e1UhCP4cgemBjd4MP6xsZgj7mzZ3Df3zGg2YLLIJtPmnElWmfzJ2PfWVuhz55RQQvUjS2paC9un1gwX9gedRZXESQNi6pVb6sw3Oom0s6iYG0vocsQWgbXvrJtXousVwjI28eFT3sgduH2JvyjfeX3tr8OcaudzpRmC/J2A+6Qge+JZfLcnLsHvn9m3zYqpGhFzQkfxDNYm3FZlcu109eFwzc0hMhrpRhTU1Qcgo7En8FBkE2QxbhJw1GEBbqQXdi2J4/3B8uy4rngM8PSZuK1WiPpfNkfqFW7WSW0H1wxLtbIWCE3RMa3egFPthysdqn4tgiOcbCPQaXOXPA9awIyR64Bd8XTuVIMmWstC7sUq9E0yj3qjX4kHPW82wLLmkjvxOdiFdKM8GPmWjbEHPeDwagMjbbWpHAexSRYlgrzdYG0FVlLaFceFbB1fQWO/zavn8XCkAnbXllIR9Y5K+cZbZedI62KRSmw67I0rpcL78eTsNpdcqogJfOMUGK2LL5sE0GyrA1vOs+kBrGze3VQLKalfmMpLbqBRiX2ilXMGhTw6DEIUcig0zmOQquPw0s0BSxa3eihTqrStPz3tDoqlWJrqh8VOOZoyDkMxlQ7Zk1FAkszRIVjlMgHETwqhu1vB5gyi03JTJpUg9UxBvu2Q6tj1q01Oiu9Kde2YDjPseFHm+f4soObLYgOMpzwndwGRwo7hm9ObIO2rLRlrYJcgqbKdW3cbsay9sKK5i7tXnRnfQ4OYUU1bhXpnk1o3sgPHRUnrdqiThq6a8PCsDbKyGVOtqopVAlMg7FF9XrUgaU2GsqIJuovgC4maQWCTu9F4QQxMx2N3lspJOfvucDuiZjMuroa/0zOE+HEkKaYLjLoqjRdkJ6lqm3KDmzbA9VB6wumRYcOoizdrY7YNRaedOdRgqgmaG91bBvF0EmUIliYEOQUzv22r7/vOiIi/5mI/A0R+d+++71/T0T+ioj8r/PXv/jdn/07IvKXReT/FJF/4ae8iRoxK/aKODuelHStM3LOM5ZKMRTrQltWamksFVjsTmy1m3nGVMvJDMycjr1Wo2mVkzTWtbFeG9ohLAiNyl2kkqHmZIl4Kd/EFjKMzBK6RCojjecI3p93vj6f3B97ZR9klbOYadmcJcrMoqWTKH1+XbKzUm0uEwwfs+/gZADOXzutzwDVqLiuEcHhzu6DfRuVplsNEwAAIABJREFUVXki0zInCynxT6OzaGfRhUV7ybubsna4tDJxWcL2OHg+ntUVN6qirQUsNMSNPBTXhqfgg9m8fMH6irWVjBJbmAptVZZVS3INpB341KoUqyCoV41cvkb+elh9wDgS9JvLtW53mZoBuF0a61LAdSVs1cQWvr/cjqaNtiy0pdOWVhmSkogVSFkluyUt1NbKiDfH9Kh7muvSqPSlkpKfMnpRpTeBszBXZyygJz5yOjgLVLVrst6EZS1/Q1Jht9vY2OOYR5npt3ls+D5m6pYiTefTWmY4h/JMaAGv+wjGUUfN8z7SmQsK/Narw0+ZGP5z4D8G/ou/7ff/o8z897//DRH5M8C/AvyzwD8J/Pci8s9knuLWP/zl7qTK3OWom86zWoQCfHrM++TEXax2rpjS0KSwAhNGZ8qZbQJI0AN0laKKEsygdyHD2LyEQ095YmKoOV13QqpY9BjVA1FilKKAjlG7u7jxGA+ezwfPR7AdUuO8CMvSp+qodvzi5au2zmOUXFmBFELOVX4OEpEv1Vq8dh2tC6alDMUhvFKt2QRGAbJnrZpZI1IQcfbNi4ptC5XIbLQu06AWrH2w2AUQ9i0Yzw1rIH6UfboaWMk9GVuwXzr7MeAYWC+wES0UnCxi+TRQFTZyFHYjTlLHJp9+CLw0ESJU1LlbBfam8HgO1vmwqmg1iOuoCECSy/qxNpZgsgpeOEsmZgVqquqs6D2l1wo25vae5ZnJEpEtrRyUOCUlzpxy9g7HNlkwKZzW68FrysvAVYBfziZr0CzhXEilY0trkLO35ChWaYijS1QcWUz/zj5KvL6Uclatpq1AKh5gtpg5FVJ0PHbCnKalrxYp4Lz9g4wL/ISFITP/BxH5p37i1/uXgP86Mzfg/xKRvwz888D/+Pf8HiTOwMMYVHjo4YWoms9C9kVpXWi9sseP3Mg469sFwUv3sBjNIWSyABOIQZOuxWII1DGhVSPg6v3FG69LY20L3WogVAzxZNunfz/B48BH0KMEMmPbGPuU3Xp5KWqXo3Id5szqGa9MQNcZQFdKpBnxliA6k4FPREsoIlBnurNjZvTWypjkzn4MPEt6jAUm/RUtrlY74WCvZCMDQgrLKDyMlrD02pnCrXocx8B7MQiHb2xxZ7ixD9j2wWaVSCVHtUoHSRxBuBdobMKyFJK+e5KMKoltswDXyyDkJG6JZjU5qTbC4RhwHMmxPWi20KUWvNmWS1uqAVs6E7DMWTpb9wFnarWU3+BUOJbmIGgrdKuF5gifhi1h1QJga4GZi4za9Ed4MSvCDIdV2tSQDFWIwgT2I8hDylkzU8Qc8MVp2RCHPJLx2HGzymVcG7oo+fTq7zzt5emY9WIWrCY0VGoTRclIxu4c2462hHXhW11ZAZ023ce/zev/C8bwb4nIvwb8L8C/nZk/AH8S+J+++zu/O3/v73iJyJ8D/hzAx1+8leGHMxJVgJKQHmF08VoU1k5flIzBNgakTf6hhCaZxU2fQS7FI8824IhCk8XLOixlLTYzLuNCI1l7cF2uvF0v9KVx7Eyl2cb9vjNGxZaklzsvs26MfU/c66NUm1yyGmLxXfNwVj8lhvnCWWJCvpi4Of9VJgLznNgEVGrsDFWOcZTtuq24J4dvc/dx3uzGq69v3gsRRcOiVtmNh8+U4hL2XFpyZeWyCm1dYQMnilVYaio5VBhR4SM56+jj8SyGRIxjBOE7SKLWsGtnXZX10iojwmtKazjavoX7C7NIRwyXVqN9TuOXD+LYOWLj0ihKSGVGugW9t1nbt5XF2LMyKGdgCrNX5CyzrWYKmSEvhrVAzhaxyGrDVntRiMNLjXiMAxC2x5nJWeI6s0aoYDKQbnCUHHvfD45tR2g0WSvM1Yot8gPQajQfrUx77INLW1iWhYUqJCoAgbJ2zzGytSrGrev/TVAfWUeaul9sNpfXtFDHaV6f9W/z+gddGP4T4M9T996fB/4D4F//Q77/33Wtysy/CPxFgD/1p38nL83YpsXw2AO/H4QnBxu6Nn59/UDT6gYY24Efg92E0Xvp0Ef9OEafyK5SMd4TgtQVScezMh6aJTIKuf34cQUtyUjXYG0bl9uNbXH8sbE9N+6bcX/fMQ+u60qTxngE7/vvc5iVNfb0RHTBFuW69Ipwk1oEPGvMlKdzebtht0puynAignW9sfTO9vnrDGFpFdeehUpLKK13lkVpHVJWEK3AVpz0B9Z/RV8XAOI4cIWB1RbcgMWw3VmOo1q3Tflwa/zq9o+gtvE1B7/6ZHxer0g/+PHz4Mt7km4cvbFvB812YiimC9mElAe9VbuTZcyyE+NmC/sItmPwnKfJT1SFnxgscuHuB27Bx/YBW2JaqSuqPoB1S3KF5/N9ogBC6wurGZbK5s4xGmPcWSnwLtvAuSB7kkbF4NFZ+xUa7LHTUnm/P7G8ILEwdCNjw33h85fP3Lc7bWlcLzc0k+fXd1yutAWaVCis9RW5FG0ZsbM/B8c9iAHIA7l01vWCtcZjc57PjbaAfVwwN+RuGBttCMue6Bssv1jhAstzw72xWKNbLyPUNiA2bv3C8/gK3ri2T6QZv9l/5No6H7WVMGpUrYGve6Vkj38IOobM/Ovn/xeRvwT8d/M/fxf409/91T8F/D9/v69nKnxcG7knW9SK32wywmkYMtFhXn57TyXHUSNiOLOCZ2oQWsVvU1aJb/aZ77HWyTNLZQOgTshW2mSbOnqUzIWMRDnqX0cQUcq0KsTNUtfNbyZWZqcmirYzhn4KWINqwRKfAPPpJJw/W5SARznzkKHevU1wUtBFawfKBAtkofoRcgqhouLOalofeHjlK9Zpm3WpFuVFCoy1TDSq/m6xFfrCp9tHkhvb8QPpg+FFey4EQ41hbxzbj5gFIm02iVllRYjNhpyGm7x4dPPzsz/IHHOcOcB3SEfadWINgU9VZIrWGWdGv53MgTADVzwq4DVmprcaZrNxKhPPB6CE77jMtC9TIne2aOxZU3dlmBTAvG8bPsY09lH6gAxEliqjnfjOHCw4PbzzrZX/Skp3ggapyTiOOsLKzOY49hlL59UiNsHizLquE4Pl7XJlQfB98Bg7hx+FQ12WiadZKUw1adJqI4li55JiZxYXRlQa2G/z+gdaGETkn8jMvzr/818GTsbivwX+SxH5Dynw8Z8G/ue/75sw48PtyiMfPALQrHpxA7zNs1WNsHgQXgGbvseUPE+a8nyAvhufBF4l2K4xH05FRQlLCtMtkGyIlzy2LaBJUK1NzAIaiZMWrfOfiGPm1VE1uXCZLs1mBXTKFL14GKnlpQ890FHTS4EdJ8/shE9x1vxsKvCrFoUA6PM95BRntaTNiHU5ZI76hbJ7FShyLj2aTlejr42mApq0x05TWES5NGVtFz6tF/ZjZX9v86aFwLCxoZ44C7sP1lk+PLaSBB80pDX6NLaFAp1KQqKWuMhOEIRTxqdRwJ1KEAQ+jhJLZZS1mhKGFcpe1CMqxQwcQXVb1wLclpKk50x4GhT4R37rG0EqDm54KRVz9pIsYogpz/dHZTkiaCh5OEck12UhWi3kojLFaTmPsLVAoFnsQYNCGiHiwPfCmtbe0UxiBrZY3fwQ1cw9b8/JdwrrorDBOJzH/sQ1uK3XasmWNheGWgC6WfWrHnUUHIyii0evTdT/iBcGEfmvgD8L/AkR+V3g3wX+rIj8c9S6/38D/wZAZv7vIvLfAP8Hxa/9mz+FkTBtfLhd+IPHo0rOLCrZN6Vs0FLoe0S8WIucdJZapTebWV0UM87wUA1m8tEEDWEG2ZTqTWEW4VYYB3altQut3RCetaNlgWaVMl8N01UYkq+E45DqG4TCDLMFaTOQg6I1oeS6p+w4vELOXaacV0u+nB7IvNHPHfKMaksAXavY1IvtUKsHjyj7rrnPZqoS9rVWQpt9cySD3jrXVunDjs8OCaMtwtJk4giDcTwYo7HQeOsLB40v2xeObXBsxdZgVfLqRTjU8a1r4QPT/psy36PMyS+UETOCzesahgSslOBpJH7UkYtJaSYVP1/nLuUMffUj8BzFGHSwVldIoBa0mNc2GjqLb1RBQxFzus8wHpnHNoHf3x91T86gnpi0sPUL1Vo1uy0USvhe192PWWRklfgUx0y29kBTWaRN013R4Mh0czaIoS+NRJ64KcqIo8DRKXVuS+dyvYHN0BuBszKw96BJsIXgOCG1ke6yzqyKP2LbdWb+q3+X3/5P/x5//y8Af+G3eRNmxnVdWKweklRF5m5rc1Xfhs8jRJCvp4RiE5pii2G9il04vXRzUuClaygwrvTkk9aLqm4BfQWH9LYSY3+hwsp8qEvMXqIcnQBmm5HhM7ehVimdx5qpW/dv7zlDgQXVb+AjaUWh5cF5YvruE+V7EtqHw1FgpkotfDF9BU8CS6d5pVc30WrrMuNhT4QSSDWTyXk7dl1Yriv9ckGX5NiTr487X5/KyOBiQr9eeDfj936A5+Zs9w1dGiYd04a1MlbJqSTbx4wmu04yYO6yUsDxmHpHSYg8cOpoNkJnxmuUnsXqsw4tIO20qMf5WeakGq36Hl4nxcl46FQMdosCRpW5CRwVoiuXKRuXimzLov/sNFZpsTIjg+jnDVc0akTyLWy4DFV4Xf4mwpiJXj5KvyJ9Hg19JlEhiFb9QAgEJXVH66HPJgyco9cCc20rrTVar2NUZZFIGeXSZ0+lcY+kac6GcngfwWJJkz+GeQwqwlWMlWSt+E58wvTjqJF4HMExVWfFU0PvNSXY2kvDPxWPVTlSL8+5ezCJbipcs/L8asiOma/QZ8uiWTn6MtoEBpOMvc6a373n1htrN5ot3zIdhZmNUF+T9FoM0kvmrdO1JzYtMHV/mcnk96sU5fwz5rs+jwO+1QU2YSLsU/cRsEuyxBRlTcbT5yFENDC04ssj2LYHHg5vH7F+YV0/0NvOY3eeZ1mNPLktC0LjwDmyVwgKTpNyvEYaYqMalygwN22gadg8052tEOWLNU4/vIqVHgPYs1yhh4OHUtxFToFK/ZznIg/VFSo1q5coKKaNGeaDW2IjkwZtlsXYpHDD2O3J1YwX9BG1KIidyM7M3sCwGHM2qNaxcj0ymY9JJ/sGMaPe0ZcOIx1YFdfSJ5S7JileKGvjoxSq5yZUORBWi0kUQN29oQix75zNW8EE4z0RXakckkETed1fOwMT6C8/x097/SwWBuaojCcLK6JGxCDCeH9+IVJ4bkW17ccoz4Qnv/OP37CmLNdGv6yAkmOnWhKp5GCA5CWO9iyR0fnYJYH7mc5bNuLkwQE8UnhkEmwM35jVhjMcdOG6XvhwW4jckNngXEq5+fPoDqGEKh5luS0l245ZnV/jGDUxh85lqTQdf/uicP7/cT9gVey21hndkz2DrQW3pYp62eDw4JEbLkko6FbBqLHAY3vwww8/YAS/WAftduHTWmrQ47nzdg1+aReOLxvoL7m78vXxV1glGFfFr19ZviqxJ88UpFVEvY/CfS6/XJHLStMbMpzwrX6pc9MPs79GWZeiDhHl/v6FCMG32rm1N3Q9dZBlQ/csbUFrsFwrc8H3ibm4s4+yVx8DIrRSn9oyPRRJ6ChOfzj7sfPp0ug9EU16wJ7O2mcEvCsmWr0R1ji2eugww3xOnmQJlqTjYqDBIoFJ1IInJVP+sLyx5c723FlMq78jwQ8IDtzr32YADfpSksyqDuiEO35sHATmK4M7sqy4GMex43v5bxzlozphVo7LCG48MVNGP1OCf9rrZ7EwZECOnTGEz9uzdOxikDuyNvbnkyHOobCFVyKXNvq1c/1wK9GTD+y78VyS2klSCnwR5dIWdMyLACW5HldyPICDYcqBsGUiFtzeOu4Hn/cdsQY22AfcVmV9a9gqbBmwUyO7aFGnkpOarIOgNGMxm4BW3YRh5U5kjopjCmrw8jWMAJ8RY2u/c2Rj+JWxvtdo6iV6SilRjx3gN2E9NgbBA+UIymQF/ObLHYayyIFzsMfg2hcW/yW+X3kfg1xuLKvyq/4n8eNOfnjjhy93xvtnbup8/PiJ8cNG3mGTZ2E/eRBHAaaLVbHtzW9YGu/P/5e6d1eWLlv2u36ZOeacVbUufdm7z0U6BAJCCBMsXFxw5GHrHTDQI+DKIgIPeTwDQQQmGMglCCRCCOnsI07v7v4uq6rmHCMzMXLU+vqEkDgdgdF77ejo3t3rW6tuc8zM//WNJavte9jKftw4Lk6zU2VMeGCuxN657408DhgHzYzWNpDGVXaa6YyBa4U7RDC80Vbj8+5s7lxQvO0cdyP3BWRU43M2dh+MKW3oOdi5s+gTPXutjttCz6R/PrBYSco1uY8DayvShOANblbu2EVZlpI8RhhnS+y84bcqNrbzynX/gNFY9JlmRrjiF5s1foJ6ogyu6TT3Up+30vFwlLr2+fLIEB1F3S5bZaGqoeOg9+B21HTzcmr4IvRuLGrluaActotcOC/nX3RN/joOhnTcr8BOW6pAw48g4uAkpTEICnyUqcP/oh1OyKxgFpiInc6/x2N7mL8nMWulshxe6PfhRXtqrw7E1vEZByeeWAaLJuEr6jrvfhP5nB7+1ChU+dGNMGXYOvslBeqDIHNWEUf1IcqKKU7R6kP0usPXbj0q91K0DpoYtRYgRaNNxkKomLnsRfM9DEpjwOFgAZ/vHXEl21KiG1aIxghn5ME4dsaiOAdPLxu+Cr//4Ucy+0wqOlMj/E6z4DaKVv1CAAuPsogDL1/FXM6UmPtSVrRUekXmV847Fsl9v84yV2ZyV7UnaVIrE6UHEaO8Ldk5jo7lDVNDtFEVgQ9BV7LYBZPt8eaTvX6WeilEZSkVmI9RTFckti1fwNSg/DZqmFa6dZu7vOljlis6uE0jnHvCkZhXgpSGcBw7Yc6qNqvpnDFKkKdtxRZm+3VJ9TUAyy85GTqzF+YU0gPEk+6VWpUYx/zms65/tY1MqoGryR9gHkNSfQeelIuvVcegk6zLQozghtdq0P1nPQteYE77mUJBCwwquGqO9FJU3ziq43B4cO/O7TYYd8foSOsgZ3rzqnhvxfu3pmytccxPSg4nZkXa/HUcUr6GoCi1eizy5cOTFWJrWW1BKpOEfGS3xWPPnP+3C8Go18BgaI3Smb2APndue0WZM2lHtJBxIar7YST7SPoBDPj46UBSyNPKadtQ2XCHzz14uw2u1x20sOvz5Rlbgs9vV5YlUa2DtB8zdyFmnU/WRCLTOKQIiDFssHIic2GPjkvSJOeBRvH3OiYzUoDucexs1mhtwbQQBo9Baw1GBdAMBqqCqZUqcgzUO7ouZfcOJX2WtMyw1hCKTXCt1ylBo5H0QvUjOPYKWM2AthR4PfYCugmv9HA11m0tiTx118aA3JHo2LoyejD2g9GhsdJawwzu41Y0cRNiH/Q+yhsEtO1UWMMEku2R7pRUt6YUkC2TmmTmQ2o+4JYpt45k9aS1+vR5Jj3qPUnmPfQXfP0qDgagSkCpD4ohhBpY+czdSu/+yDJ0rycsk3rUKaKJh+gk4QET5twrMmDfB3vu7Efnet+5vg3GPtha0pYopL2XMnKTL4xH2wThA5W8WORPtTdX3kPmUbJmtUkgVBE91moaeLAoomRxq/ReJqr0CmNRreQenzRWPmpSsjIWfH4gkGAMuH4e9TutWrEevzqstPZHBMcOY1e8w36tHMZVBs3a9G84P11vXMx4XbcS+rRWNKcJXz99x7H/wOe3O2/7zqfrlU/3xPvKQ8NdcSM5tfxWaVWtItBHVOntINm0OhOcpNqva9J6OFIaytYWztuGUBmVmRUMk1GsR/hAjmJ9Osnug4VWzsttw7Lay4XqBdn7Z9y28pCkVibGQ3KcMcvCgzE6o5d8fTEjBxzRyUmbSpSvQld7F8ko+Z756Uy36B6MY+AOT8uGrfYeXY9Ajsq67EdZ9nWZqc/zM+tRTWcPZvYYUU7UyYDVhJgTnk9WrfyHksCV4sPnDaXjdB/Vnualb/glX7+Og0GSbTO2TYi3z8Q4k1Tgh2fxtDBPynkoLNomVy/vAqEHY/iveEZyJjHdnX2/sx/B9d7Z3zp+BJxKgGQ72FwHMp1lmcEnj11+WnUfIoOc/xPPentzKinn+vAuNntkG2adYE6lksUMXlDJufVUIpNSgYsyJ4nDSzuRUTV3+x78+HGvD60J61LlsLYZbnVXHC5k15o+hqA+7xw+OI695AYMPu8Hbyt83ht6wCIFpro7l/bEplYs0XBux0EPJey9YA+hWBhMsKVs3IuV6AbfGeNWDkO5YNnQVutVZVDyvu2tNFZtJXunrM09O5s0ZKMayoeX1HenTESiiC3YYsgC4l7BNrkQsvD5+nu2JXjdzijKoc4xW7/MlNAkwhlTQSk66wNK3ohTnR8F6Bczkl74A6uwSNnqhYWb9wqG9Upz1tNWGBdRXZICx37QZ1SbbY22LZXt4cmeoANElXU1ctbMjXTQMhBW4U6yWMMpB+k6P45mMmMIx/sNNhI0sty38a/w4P/Gr1/FwbC0xr/3Z/8ur69/RP7v/5if3pK91xvytu8ld/YKMw1GFaSeYN021nVFFyOJao1ykOFfwk7mm9+789NPnf1e7cje5x3YO7e7VU/Fdie8cbtBs8G2lR/e40vFXGutJoN0xn7Dd9isrK6lwKs6sZJP30GLb2aW3jgOR5XFhEgVvE7xwvCSMDsb4UIfC3sEb+MDeTiMBprsx+CnD1dAWJfG5byxbCuLJv1xZ/Cs56eGtuT5+UJIgZxv9ys2YFng8vo1H3alfRzcOfjqyXjZTviS7J/fOMaV2965Xo2MwWkztu3E9W2i6wLaAl2qXnBdGvvbDu3gNqp9cdOVthoNJai1y0dwXKtA18M5XRoizt4Hx+gcE3BjDNb1CcM4TBkjqhnaVzZ5Lt8FjvlbJUyfDOsCtyT1DKJYGjN3BVOjmWEN7vtR0nTXwihQwlr91RZ8rW6RoQtdg5Sd4R0fSaNxbkbLmhI+787qGy80DgvaE6xSJi1bKlhnH4naiWVb3zssPXbe9jvXW8cCYlvR5xPtudiW26deNOXa8AbrpmzN2C3pwKbBJZPDkqs4mV65H9a4WJkG+zjY+/6LrslfxcGgwNenBbjwejFu9x1PJ/xe2YZRIMtB0nFcYGkV3FlKs0csJhNjYOYxFoJdPYMFNjL572YLGR23Cew5zMJsoG7yZQWGThC+T9FMK5mtF49MgFzki/76IUxIqV7FkFopmHeiLPBNZp4As2Yvkgp4GUmjvBTHgLsn1/1WwFkvCvsYneylszeUJksp69BKHprhNllIV9GkWyu607PMOAc0bXgf3Ag+SuUXbBK8bM+gii6vYB9BAo8rOe6onurOPxOdQwpjaKIz1l84PjtE0t2m7XypbAYJRndSC0/qo/b7QXJ6be8YDgqWOlvFV8QbplpFu7KXDvKIWWg7NQUOsqyzSKZUlC/bmUMq4CbUSQ4gWCZOcD98dpWWtoSAOMZkhyb66EKIE8OR85Q5I3hWm3lk9TaENSwbbWsIHTsVCG2HQlSuA9poulXgjBYuILoSeSsMaU6GUC1SIkrMYlqbbeDFej8k88FCUpFFZTo7ZLCY1aHdDAgiBsP/AHslAkHsO14253n9yG35AKdP7ARb69gefP/94O3zTh/O07Ky2MJCsqoipzOeB9w6OeC8Xng7diLB2hMtjPv+ifvnP2fbTjwkLEMXfIWIUV6H/YbbBVkq4mz0wUEnR2e0E8Ioae22EuvCKiWmCgOZNePhlGRXBbEVxRE6jcByIVF26Yy4zQ+IEu5wOMJGb058uPP5WtmXrIbKGfeBZHK7Xlnbhd+8bIQr0hbWtrJuRUvuEZXqlDWqP7CW8CSGwB5IT9KKJTBbufpBjsEaK5eE1wNOqxLbznapQJAe8NW3v8Ejufe3Krgdg733sgrbQuslLBKS2/Uz2/m54uJHcL+OapXC4BiMfee+d/Ze+uDcn/AIPBKacT5tbLaw+cKneKNprRrVMt0hBujBSrksEyFHlos0HJbBQUN14/PHN+4RyLayLSCLc2nP/PT2A+x72ZgvF7Z2Zr+/MYYRsrFtlQ+KG+sFFj+TSy9JMskSB96DYwhL37kC6Ma2KU9t4Lky7MQmlZuRrpXW3ZQRB0fvle/AwcqBu7IP5RTByRPDOZ2UdEpMt600da5edoFLKBGDGzsRjdSl/BIu9AzMdtbtWxa/4fnhF12Tv46DIZy36w/c9p232wf2fsVj1B4VTo4qiDmvC21Rns4nXp7ORKt9FO+MKC/AQ4sm1likAMr9trNf74i2KZgpqu+hS3+wCMSYfwkyTVHK8kWQPCW1IjLTk0vCLcy1dCL1pXEt/CFUZ7h4rRFJTlPX43EwpxreVX1XF64H3O4lf8aAXrV1Plq5DjdDG/PO8OjY/PKaFpH2RRyVAtbq96kqQkOssV/nChTwdjpY1p2nk5K2crr8lsvb4Onykadtpx8VcithZB/FzCRzXTJ8Cmv0kanxAMwi6BmVUO1BRjCm8KvSiVq5WCkZNAbStMZ5kZoAppRYEhYTcjZMlbS5ciFm8fj0MeiMmp+S8ayioFDBA27a6Z60Voe6rpAaxWB5kKPyHYKKs/cBvjkSNtukgi4Opqxy5nZcSa+xzkU5xlbO1P6Zwai7/sQAkoFIsC4QXWna6nEAthgpylBYtxUYNelOIfnwWr3L/Rvv9LyqvOt2cn6mwpkTCH+VwvxrfP0qDoYxBr/7i7/g837n0/UnPEryu4jRMA4/EGLKcJPVlK0JulQDkvdBRplntCmYvhe4xO7sbzdubzfQRsrDdVj1aCV5sAlcOhXZXS+izO4oVUV9pzooq6GqTZlpdVNmgaGTGooogVPMsVVnSeMgcT8wrySliAqR8dnO/OAr757c+uB6H6h4xZ1FUaGLTVfdpKVSy0syekKTL4eYyHsycD52K36WEkVD7cQxd89Dk8/3QbQbL09CW4Ovz89cLleeTmebdkdMAAAgAElEQVROm6F+1ONoEyuhgEcRQ6zco0PBpFVI74NpykkrH0FF4tShoG15f057LzxH2rSWr3OnBxiJewcvqfDWDF+U6DHZG+NRS1/mrSkUov55qktqBXA4UjluV0bvhVdtRmtl7NKlYfcgekcsUSs3pmC4HFiuJZ9OZcigmbHqwk0akveykdtK5PN0Vt65ebLaQrNlUsnBssCmxb6ta0Xi+ygjWPkkgtO20WzhOCrT4ggn1FiqQYTUR+TNUuAjTGn3dNf6BEup4KBf8vWrOBhu+8E/+T9/oG3Cb75+ZZv9Bp3gn//4kd/13zPiSmQWj+4HyWA7PZXD0Qdmwvb0QttW2rHxw4ef+P7Hz9yvztiDMRTzgV2EOGrczpy+hlaHiI+KWTMvmFDnxRdZtlY1Y1sfBbiTwpIyu1QCtBUjoDXlhE59RRihzkGFu5qvdATzmVKVEyFJplz3I2Ncib6jttKWS6UFLw2TmgOiT+GRKsOcq3cusRXnrfpIgKhpRMtjcb3vxf2jnLeNzYS0BYnCHt6uBzcPfrvBq8AuPzCuP5HjSo+d69GRbLRh6GWryAGnhEIBOkrQLZcNG0Wjejh9H4xxcBeQlyfWbeO0nVi3wbHvjH0n28yJWCrJyX3g3mvHRmoUT9iAy9IYGDcZ5N0RF/ww9jGIiRcsNKSNcrk+8ivmEJMJjIOtw8vzieenFSU4upeE2m44yUJDV6ANzusT9M7er+zDIRuLbWjCnTfWtZcXJjYkjT72Es95ggVHv6MD2ulMWxrKYPSD1Yxzu7BuJalPSULgCNh0Hnpu4IOFWkXwwo9CsuoQpfI1RKupCo3JtBg6diDI9gd4MCCgF+HpcuZ0ElbKuBMqvI3kw4cPZU5pDUTYThfOry81QUg1O7XZRdlMyeEc9xLt3O8BXqOtmVamgvsUKoEupc3PKIsPPkNgJpijUq6ktZW+YVmExuPCK8SxCln1nZ+cmprJcyVB2ZCBGsO1/HWezHBR+OKjzfc1ZVmMtjTWrVqjrDXGUYCljOkWNLClJNyeJcoSHsnQWYBqFBjnw/Ejyg06XUke5X32rEZvdcf3DR/Jbb8zxpjPccFzIEPgEJan+dHRCktF44vCdAaFPBKKZSpsEmFE0kJpKeViJBg+sLbSVMvRmDHvdmDZQLxG7ACstBAx96a2wNqUdakYP8gJyBaYG1kr3ZeoG4DAtpWNA9tWcml0P+r98iDFycVxrA6m0cE75/NCqBNaZifTlbBioWIclEu2xHn9eq3VLmDbFvYxGO60pAKHifq8SU5/zcyC1S+CpHAnRoHTZm1+j3KPg8lz12xU42StcBMHt2ZINoyS6D+m4L/u16/iYAiSMIV1wbSzSpV0yLbyXQS//36lHOsPiWglIJs1sJh1YzUy9yPY33b2/Sj3tSfZS3Ryeq4sg+xB38uh+Rg4Ca+LOVdiykxrQ65H2KwhbeZJZl0HMkVLD/V1Unu0SqVViwlkHUDh05UHlRg8O8nmZwdRnW1BwnI6c/bEV2ddNrbzNvn+ykogapfRNqPkNBEU90QrYX3eGWd2AEFIiaGWZqxLTR+YvseNE8V5Lw+E3hYODaQZl23j9bSxyp2Dcrhu1Tder549XoeiZbxXVJ1sBZKZwpg77hg1JmvUVFZP3qhEScOiMI86SaMs9QEqhrd6T3LKf01qbdR5gcTDbi+KSNKPqMCYMVeo2UGCJ1xO6IDQhbsbcRwoybo0+sUIrTg7P0ox677XjSmlWA2xCp+Nh5qyfrSsU6x0CDmbbRddiKbsDjl6vV5NsGUpY1hOXEygMS3cKJqdJFmWKfHP2UAWHZnMSmWKPhamSsPGmK/mSoujFlz9AzwYPODzEZwOWGOwNkHPyXZufDPOPG0LhOM96b2z31b2/UD0VKPXlIoe7mQ4n97ujF49hQ3Y+4FH0s7PjI93YoLaNbrNSz8D1gpJySiDkjwEpeVyfe8hUP+CQjDTiSJLwzALl98bk/0RbpqGRuleXaaceRqSf3afB+D8vNUdZTinbWVpC94rgZpWPEfaBEEx8MRdyFbPyR5wqsT7hw6dY6jCuqy0Za0phGDMiHSTmeOwrbRtJQ3aajydGy+nhaU5R+8MArxUndW1WC/QQxgd1/pAaxQOKybYUs+tdnyvYhSpKkCzmsGMh/tiHjki2NLQXo1fbloO1hl+alJCLqfoUp/Ur0jWqN0n1lPdAqjPkuMIIlaIcjYeuxI3ZVmCr142Qjtphu8rYxxkVIpSH0cJjDCwxGOvQGCHsO39PY9hyGEgA9ajMj+mCc/HoEtgrTouxl4ZHDavfQutizq/0O8staYOp3I03Ov10vrcaHqtq+/IyhQ8Ye/x8b9QEf3rOBgE+Pz2keg3nv7kleWysZ42VjNWfWaxJ8RekPYDjY1mB8qGRAeO+kiFIW+DfR/cB+zD2A9lH6PEN00Yt4orHzmVbghLW8uU050uCzJ2YhzY5RuaLPXmiMFa5qsn27AtuFFhtToaaiXGEaTi2jPxqOzJKt+tjkfxgsC2ZWH3QJYVsRWPA09HF2FZLvTbG0+vF9alfqbvFxYdwJWcyR7BRqrjcS0OnIoKdx7lqwDJIcGwxC5L0YVjCqqysgVutzdEn2h6hXbh/NT47cu3fG0NFuPEv83+dGJ7+ci6w4c4+O3LV7hJAZ4C5xT8XtjGcjb8snK97xz9xnlZuKhw94WjS52qVhTvjN6i2caSVT87SV/MyvzzvJ5we6qSlVElskEDP0gfbBflvFZg6uefbugY6Aa+Cb7Acj5x91tJ36MhqvQYHG8/sa4b+FtF3q3Q2souOxLGyY3w4LMMXKsjRHIwZGXXBVNYdTBicFNhZSM4MIK1OdjKiFG6lO82fFTnZQuv8X4vA59Y1HrBgtk6R70dvD96g1DZGK7crjv7vrPq6V0200SxteEeDD/KLm5VZdeUyuj0oB9/iDqGDAaJLDoNToPNnZVSyy0sWBoWClFTwAxrfw/pNCnDSmSDuE0h0Tx15xQVGuWWa422BeCz36FGgS/3LKu4L+1TEJWINDS9dvJ4KBmZjso6j5N8xxkCZtJT/vy/1tZCMRIPI0zOv9cdIHhaTpgZVVCc7MEMLC17d7qSLhWvJlI5gjW81L6rdWcWLfxDKMGQmr1jJpGB5s6pFa7RluD1bPzm+cLr6cTW4FBwDjI6IkZbEvrB7XZjfX6lU4pQjxrhm5a1PLDJAChdmMXDtV51r6yJzrQjqUIasdhUqz78E1WwMrKyFvyhL5fZ9KQKHpU9iYMu2LoURUxy5AA1og/GGDSieinkfdarafCRFVgoMT6U3gPf58EvwXoytk3x24DVq4U9YRyOGlyWkrgzRV+JziCvWuQU0IgSQnmQk2KuIBjH5FTMjvicbCClMXLU59yDGCUTT4T1ZNXwTsm585hrRRrOdP1m0bXvaWf5ByiJJoXGymV94dvnb3g5L1y2Z7a2cdPPVH5Cr9TjXIh03N+438uyWrVzwto2RIytVVS6t4A+xSSjTDfH7qytcX4641J6fEM4bWcMr3j2trFYom3MvEKlWeUl3Psd8RO6LKg6Ijs+Snf/nhccdSg5Y0op9WcnALW7Dmekc4RwZIFi28wZXJZ1jp6J55iERdlkFhrZk6MHGisteVfq9Rz0feBHRaVZq6SotRmSV9QWwAg15IB0YXuumPHLKfiz7y78nT/9M/7Gb/8G+/jE7/7Z79jv/5IfP35i3I2//e/8KX98/YbrW+OH6yf2rNj+WApHqOqHZCSEK3uHYY7kQEVYVuN22xmuqAdqyiKVxredldE70evgXZYTJsbdneh7HdCqbFpq154HPYR7d1YbiAW6LBNiLJ2EiXF9e+P29sbz+YROsYAB27pWJ597tXUpoFWi+9Fv7PtbHa5mNF0QbXQLJAJGlR2RwvO2cd6MoU7ENunZAPq0Ord6ziOQOApYrKOaxEgT2qTBzYIk6AIjjR4Hz0sxdH4cjP0og9+WFYbrXm4Bj3LOek2YYs6IJG1UMlTIe3jRX/frV3EwqAhNSzXw7devvFzKYRfeMcvSEVgl6vjkokcM4FTgi1boaVtXQpNl3fFw1r149TF58D2S2Hdsu0xaqkJRtwanRclwbMZGyixCUXl0QCagjPDKUygfXAFQCPF+KEycCKZ6ahKHM96NiEK75/fmz3ROJg+MafoEUiB8og8C0kq9OO9yqlOSbJUwLEPJQV2YlOQWShhlNU5BDLRPULUJrVUD17IaW2uzDk4LDxk7x/hI+A1L4/XbE1+9PPHjR+cv/9kPkIKFsWnFjkHQ0+uuLLyX05CdNp2LqkvpDab+Iyb7gxcboZIzi7Nwj6TKYmAKyqabMERLXzCUGFQZboCEzobu8lXEmHFpqu9yd5FSEmbTGeNf4FxKvX7uE5tIUJ9aFxXOl5V+eNGaAttmrKsRaSyNMtz5dAA/bggIu3cMOK0/vzhjvrcbqscUibXSR+TgyI75ZCym3kEQLAO/HxNLmfUDqow+8HQ2uZAxJ+F8oA4z1v8XfP06DgaF0yZsq/D8tHE6LfTeufed1RrbsnFeFsbS8FF1935Aa2daE5Y1q7S0bUROF6Tke1KyQo1+vdiH/FnYqKixrRWSGszmqFY6fNWKI1dVIjuWX8Lci0xm0mbTKfg4HuZqUMhiAUQTZaxrM0qxWZVrfHHwZZQAK476b5Naxa1WBrVSAHqtQe9WZ+ZorA1GNSz1afrKh/OzbTCj0chRj2mpMFLRVtqRFPbe8b6jmbyeviKaY3HiJs758kTcN97aT/TojKi7nllp8sekPzdptWJ5xbGpFysSETTTUjNSa1Y9pqhIt3koFJZch4qaI01+hp7NMHmlXpM6Acs7oq3oTgVHZ8aC07QEWI/0cDGhqTKsuAz8EVAPn9+cY6ea0IkCpR1Wg7U1blGYIpZsa3VqRq/JQiVQKmPD++M4S3YdbKI8rcv8N8wojsBSIXZyMgk+g2ar7ay8JaFtitRK6Zk+Xw5jBsnUAZrv81Ki1Npr1VzEzyN1/jpfv46DQeD1tPDN05nFGmSd5B6VTGSLsmyG7QI3pbuwH192ppwfeBefFWz77P6rYtBH8Z0Bsm6oFSNggLSpjzBjsamLsPkBwtCsDoF6k1cgmfTAFCtq1ao97kbws7fgwTaUG/O98CKVUEN99nrn42JIkHinP+u5ldZBtD7YyAO1z3cRU1FWjktD1wdFWinQkZVElVbja4009n4XDIRFVshgP+DHtyvfPN1YzPn68lI06OrcP31kX0+M+8o4/m+0PMjVxt3ki3JzBLrOlGVLGBUxJqlkGm0O0vU1AZEBh3faqcJMoCa88Lph0FZyHoYeimjpLl3LDu5qLMCyrLNtbOZDzhVtWaeTcYb5Vhp5HfH+CPiYQqRxVHyutBkEfPeS5ztIU1YrZWYVzggegWgycilg9FGmozNrQQYyiilhFE3t89MlaFGPMyDH8oCoxK10I8Q5jr1+zqg/E2YVMxhRALeXTLz0DApyvBsJmTiNkYiMX3RN/joOBoTXbeObyxMtGrEnfhj4gtteLsGWpCVDKw5tj8EYO1UB1RFx0o19Pzj6wXDHvWK3ik2bdtt1nXfJmZwsgC51gdiCSsfsS69DXegVR05YnWJatGZ1Z/JXzuL8K8+LKluZF3PyICmlrNg/O0zqH8sPsE7aD2YKNI9mCWZqEO9NWybvwm3CqkjGm1YEHrwH4oZPqbcVHFaTVKPPMJgtGrcBH247n4+d51U4Nee01OtyRfjJnYhB7x9oWXSk0ktUowV0PZ6vSUXG5+TfM6pfw2IgD4gw6tmnCyMdq3osIGdOgYOuMMN1352HUVFoKVotWVG4Q1sbuOBjL3FQVMN1aw3RElcZQFPEs7o5Hk5KT/CK3LMGqxnjgD0Gxy3pt2S3WmVaNcbRo/otThaEb3MofHgTVly9ymCGlM/lGLgIaTXBqCnkwHODSMRKuGRhkI1dlHHsSCaZjaT+rC0LIyprs7oslWVC5+hRMq4Htz79KrNQ5a/99es4GJpxuXzLqT1R7QgyNegXPt5W9uv/xr4fhHwD+j0RAz821lMjTBjZWCz5uH/ip58cSaPpgWtFsacYbdlpu3MJoVPJyjmS1oX16cx2Vmg7lhstVmQZZBNcKmbs5BuHHlyWM4LRb52RA8RYzzrH0cnbS3UaNNnwCMa4zZ6ZOkL6/hkfymhTwxBJLkZsK63viCzoFESIGmHB8PkZFoBpoOJnFlxpPAEflkBGIHljxJ2MxklfiHGt1coEti8qy0suHPsnrt0Qq27FDx++R08XruNEO694g+3pO7YPO3++/zm3/cY1dvraEDvxup7owBFOP5J9/AB55kkWuFj1hDiYd3abnRJsdXB4Z/Qq37ZuxEi6OmFFATZ3bvuOWVakWxjjOHBzpJ14fXnF2BBT1u1E32/QO+SCbnBZnthvN9iDTaqz1I+d/rpwHMDSamIccB9B+Gd03ZClrOKNg2jwcb+i28FpO9NOZ2zA/bZX6e1qPMutVJntmX04/bpX+IxcOC4Dbnd6Tnxo38l1JdYTJ1bu4xO4stmGSeMwZ2hlXtrF0O7se/AWxuYNp1yfSy/tSFiUbJJiqmzWBcJcMYbVOvoLvn4VB0OpBv+SsTiq/z7s/5IP1498vK+83e7cAr5+3eBj56qn2c3nXN8+4mkc1/ohb0fnfj3YbOdp+5qvzi88tYN9dIYvaHyGEI79YB+DzZSn5w2xRhLlkbBAViV3h4Pab1uSWrXzziyg0ZinfBC6ofu11JPTU6HLieOY4x45gcS6I7oo1g6wSZVSg0hmVZUxg20fJiibB4oktCZkzi7PUAb+6D0hNLG5bZzNIBqpStOBXU4VMOtJDse7M6LDaaMfTlLg1fV+5+OPH7ksG7Ymv7288kev3/DN64Xn3zT+6HhGxp/w+59+4BZBj4rVZ2noZWHZhPuHrysD4umOqZBh7MBHn07Q5QSx4L3X674osZSmYTlqDSmDW+PzzVg2wexSi1B+JLkjrph8IC8bHFl9lNc7cd/h8IpdX+b4rI5qoy0b66JE28jc8WGkF0GtACboWr6J4ohhszPr6cISQh/KHcHGXv2Th7Otlwoszh2hVhoEdqvckGMBxPB14T5K6dlUaR602yDPwKF1EzKq63KA71IfF5kippkr6t45Rqumc/wdYN4EQFnY6A+ZeQjbspJZgb+/5OtXcTBkRtle7MLhO/eR3I/gevvEDx8/ch/JshiLdZrNPR/Yj0oTvr9VdNt1BMONDUeyoWlodLLvZG60BhGtWq5HgNSHZVkbZlm8djOwCi3V2hZA4zHIzyG4RtrqrTNiPPwOVXFek8EjBTpLaYcVQRFfFo8qSPH5vcWLx9x/q8xq4iJSzsqc8MDMeqlHkV8eV7pPNKVUjJu1mRcYrNpwYvozCsxzF+IQ8qAOm4QcyU+S3KyzXJYqeJUbbdl4fdl4ebqQ3wz++Lvv2LOUh0urnIkjjOHG7fa7mno4YWFVaOs7gxsxrEJystil6IqZEEvDQ+mujDw41AvIG4GtCbnO5zbRWp4xvdZaltB7BbI8Ip5rIGpThXpUAnNLdAVtSd+tmA2tSP+SsyYRRoYjD0DSGs0aWyq7OMLDyFfUs80e1CP9faUrJYdMGbIw/CCkSm0fwprUSrOKGZKLTset1SQLgAtiBVjKBDarByVLP0J5S6vIuDQQ0pbCH9KB0pUMAbc/QIwBgmavnPRbrsdP3Bw8NsxvXO833KtRSChuOuYt8tZ3Yjj3IypRCMOs0XSFgOM4uN6uXPcrKfHIZq1jZa5hNYdPi/TkyGkV2hoPOkgKB0FKjGXYFD5pJTyNUbRhswKHkPrgRO3DmTP6vGaNAp3ykTRd6LU4HBOW2qzShOQx8cPsx/zi7XhgfVMaAZQ1ZJs1ayLCamtNHxZMC0WJi2T+oEx6H/Qx0VUpMZRH0A2e5c6HcUflhrXkvPwpzV65fLXyH/xZY6Sze+dtvHGMzu0IrmPwXZzQXPlphzy8inX8IHLQZn5SkoWwa8Wzt+k58QwOr1Qnk4J7/ThIG2AUqJoKrYEuMCM00pOuBTaXuaKmNKEyDkytdB1W72v3ioWd6i8eiG/vjkYBeu9Mc3u0hcc7u8T0S4gGaZ0x8YzHsV96qTo07r2EcuvDUNHq3ehJgbHNkCakzcHy8YO8kakzaDhYQnHKxVkgOJUbEvVeRsT8GfFekhU6Pzu/UBP96zgYBJZ2Qdl4u/8Ler5geuK0rGxt41jioZPDxCvXYChjzFBVtMBFbaQvCCtjDO794NPnO/vo2Hbn1LZSLkqJoqwBUqizCMjyhdrESpH3RSn3ZeR8V/NPujD6ROa13tjMiiOvC/exSoBNjr9IT2PGPQE1DXgc0Iw12xexmv6VGWXW88WMk3/oJerAifmxLDtuue3MgtRqaEqpp6RUcIkAt/3KGKPYBxPUbVJpTt+TY3TQpK0fOOlKa2Cr8NvvXsgY7H7j49tBdGWP5B6wP32DufB//PAXfOTOGCXE6dFYdbogxdEGMot6lvQpiC4NhUbDdKU14xh1ETeCzIWRnUKKIHv95Vl4S0hhM2mBjl5aFKsbhsoCUUDmGJ8I2VBb6o4dNqe2Wn9MGyEGmYg7h1VyNvM6tGkdrXyPPvsnCyiu978A7mr2jrLNT8YpW+U56HDEDVt1MmXw7tA1YCiETL+LsmgtCyW7qBi9Ap3K4YlRbVmA+fTwUJ/3P0wdg5RAZMQbx3XQ1lEMwenCb74KWnvjd7+/1/eyQw7GsZF9wUcjcvL5ovReo3KPK7f7zlt3XISzFQ2lqmxbxYSdGqxbvQm2fuG08Zog1LIOhiwR0cM9KDDLUesN0jbbCCfizrwjTgq8GA15KN4KpRApVyTzA5VQzkmBDHtvz86Id7FNj7pzRVaf5txfeFBfj7r3QZBSyUCm4HbQk2ktjwIrphSg5MYBUZqAdHjUNt3C0UVpXfl46/zT45/xetn46vWF19OfIuOOhNP0wuncaO1MtI3Dd7ZUur3x57bzyZ0+DM3GEWCzV0INhNJP9D7mFpd1gY+5kV2UfltpWZZl9/IFqAxC64j9QlzPi3WyO8BcIVpJ2qkMRU+n37PSpx9/KrMyetJBN2RdaynwqCSmNFpbKeNbZXGJ1AQzhrJIe3d6eAZ9DJiN6Ktsk2KsR2lU8A7dS0C3FZ1azMjsjJgrbMllaj1pBlPmVCG4Ws7KyMIuitAd8zpRNGXKx6Ts67/g69dxMJixD+d2/YFvnr5GCDJ30GeW12e+3j7wj//J7+nXN8g3Ninu/3oTRgy2PJHXC06njxvHJ8e3oFvZbjVW3JXcjNPlwklAlywb9QZPL6X224+DLaFh1WchO6KNxsLojkkjPSpkS4KOQdUZcoo27bbQs+bN61p0lHKZAOEnRBfMntjvNxCrycGLArW2QsJhR9GPGIvWY3Mvc1TsRvcrF3+paDbdIZQ1z3Q6++HARrN6DgMnhjLcK8RFrN712Qq1ni5sSeUwZDLGwYikTwvqORdUg913Pq6wDzgOWL97mjJeYbHB2Bynswzn9atveNKV/3D9O/zx07f8i+0v+d33H1huB99/f8WzsZyM5WWjbQ33QbMNonPsiY/Goo3tJPgO7J0bV4bCpvXcbg79qqQpp3Nwtifuc/dWX2lDiMuF7sGSiVmScaXfD47bYFtP9DA2hUWdewUgsMoULUnSu7N49V106Yz4xNhXNOC81sV460LTA3v6lrHv7NcbxzhKlxA73q/85jd/E9kWPv7wL0ucdj4TQaVpbztfn18Yx+CIAghFZ2eKzIazdKYDrzQrTjFLW90E87Pg950kGW+l70gxmiy8RAMdhPwBgo9QEVrrtnB6fiLjxugd952nVpPyaQtOq9DjiSoy2zAZbLZiudEd7mMwjhKBiilre9fPYpvR1qpyQR+CkDqBGXVnfuQlZMTMDZR5aFedGFNE8gCLbO4O+kAu4oFh1PGtc+mQPKilfgFvhNxowQQGqX3YA81AQvG9k1OEpAu0c8XcIa06IzXpWWYkD0fTiSj+cpZH81c8+EHVuInU84paFZxiXfRh4xuPvfVnQi2pqUeycL2jJfcO9/3K88sLT+cnWjp3ZqCawdE/kdxwVp4vK3/y1St5OMYbP8m9Volwonu1UCMcNvM2NGsclro3H0HpRhQwI1v1ebLv6JiS3xi4CbaeQZYC8dzgCDKSLM0cnkU/u2olK7UHva+YBGlWr4HWWvhFQTZVLVPrkFDdEVJ+FGsOLZCj7u3mNoVzoIugSxB01umC9BilAt2ETCuT1Jza5uUAUYU4hQXNfxCK4RCvoS95V7BGrHWQ7FrW8mVOoyJUC7b8oivyV3EwZAYZnWZPXM7PHAM8b4TvLFxY2xMvT41vX4RtPeGyMgxyb4hsuBt737nuHd8rAn3RhW1ttHkAsCpta7RR5h1Rph9CZ1xOXTzMg0DCKjma2tPcjwLJbJ0XfBTbkA8gb+oYeFipAvHqm5D3y6xR4fCfUJ+dig/gSxKmfVonuFUAo08F4/SUWhlmBhXw4VKlKJIHSaOCXfVd5PiQypoWt60ihGaZbMJZNoURuNTdJ+Bdoi1LXSCDRBxWjH2AWvLj5zeW7cI3l2fs9Az7UXu+JPcxuPbOUFi2M6+vZag6Ujmf3+her1L2QbLAOgd6tfrl8+44MbVyHRqwNmgFyFViCcVSRVW+rRPzGcMZE+OxKCGUm1SYas78Hq/aPCaMZFYs0jEPxaTwg7AopsAp9WFI9YNKAX22gq3KoJNTJl/hNFYK0KakJPt+Q2xBCPbbjkeFEVX4Sq2vGQ9Z3YyvUkNCphhN6rFqBd1YggypkNcm9FgK2xj1nDQCa9XpGggef4CrRERwTE67LRuZnU5DI+jhIBe+/foZHZ17b+xi7Nz4/vtkYO8XJIeTx0FftSSyrXG2hljCQpWSSgF+0qgDYsachTJR34pIW2A2QSWupX+vq2aCPZnTy1Sg2eMxRCtq4oIAACAASURBVMajq7Xak0RRVuDh/ANPfe/jhUKRZX74BNi2CgcNNdw6ex6lBHzYg8XoTLhx/pyRo8DOPibQXgi8Sz2mtlQDslFryZCyL7VValfPmtpIRTXeE7DLLZjVUdnqse44P74FbbmyLY0XTZoB7og5m1+45pWPx15sSCTeQJ8ar69P0+W646Pq28IeWQxSu7M5lej00I68UzOzvetBzUk1Vfe6A2uW4cqPQdydbDZfb0N6lfv0HnVoROV2FkwzU78xiH0eOlm/R+bd2kvLUl2AiU8KYZ0hwP1+r2RpJlX6gKqlelmvt53L9oy7c79+ItJZn1Yyyv+S74a7Cn17eD8eCVzyyA/Fq1E7lfJeldw9yXK6TiWqTprac8yu1V92Tf5/Hgwi8m8B/xD4E+p1/G8y8x+IyLfAfwf8LeCfAv95Zv4oVbL3D4D/DLgCfy8z/9G/6XdkJn2GfxqVmbDICUw59AoGf/Tt17wuJ8Z+4nOHz+ONH3/853weO8ckAjeWYhnEERUWq/VBNMnF68O+8mVOtochZ2YnqSKR71V35VgtIVMLq1E2wSVmJFrMxKZiEXyCXp6TaqfXh0M3SrjcAS/mJJmzRSHVKtDeY6LmhSDVZwmPKdYnem3vhwyT/AsgxpgpQxRtGrVKJUFuC+9KKKk7dqhjshVz0cq+q1mBMrUEVXLyI2Ko6LskxLl68v3nD2Te+be+fabZVgW0kix7uQH7pyvX/TPsRUG2LfjmmyeuvXN9g/1ajdyRiXqVuZoqSIXOqCWRo1q6Zcaij4pzFxNY2uPIeKdsdY4ZGUnMwsYIrTQpd7LHTJe2KU+njEjCe7hNndpTampaE4wHNJvmNirrAK338YD9emcS2LAKIll2aKpH0nvApjOKvqT6Mf0qj3tEzPq8B/MhMdvRyZK4a9Za3EpJKj6f69RWjCPQJdBFK/NRJ8M0qfNf8vXXmRgG8F9k5j8SkRfgfxGR/x74e8D/kJn/lYj8feDvA/8l8J8Cf3v+9R8D//X8+7/+SyqO/Mido18hoxx7bUPsxMB4urxyaRdyGKe3g/WtY3Su908c0VjjzGoLuSSj3WmL0Frx12GVabAAovMu8nj11Wm0uVhHFcqkETF4uNXQymsM7NEm93jgE923L6P9A2dIQXIGyr5/q0+b9YpyPH5CUaBatWKmxvVWkd8IcxXa3n/diKxpYAqjmtRNzSct1r3wDYssb8jckyWifN3Mu60VgKtZj8AkK9WZan2uuIG5V1cRRtmas1iFQfJ5vxJx5fXJeN3ObNuCrAuks3pjEyO8KuZXDNWN/emgDcM0EHbud6XIgD6xI0itDs8QcDnQLPET8N4PyRnUGr7fYYJ27yyQNVgbmW8zXDWItAKr5sX0rg9B6MCqdajoA5YaID71C1ay88iBVkfQtGsXjTP6mFkJjVxnyI/Px5kC5qwPJ2c4lT881wByWs6nO1Kp6YTaMlWi1roSRpCpeGp9/xxqnCBGuTF1s+n5WSonc37OlP+f6crM/B3wu/nPn0TkfwX+JvB3gf9kftt/C/yP1MHwd4F/mJkJ/E8i8rWI/On8Of/vvwPhfgg/fDho8oll4T3z4Prjlev99/zlp52nl694eWl8dVrZtmcu938B39+4srOvo3bMk3F52rg8ndkuGyZSXK7BRmN36Fl1dSOK8nl+XmlLo9vBykoL440dZ9AwVllwSzKCPgphaGqsUj2WIkYdNUWU1QWkcARqb2geIGeEQHJDjjuH1NCoWW9YE2M9b+jaeFodzY7EQBCGr/i44uPKkI3msMwFetOFJDgSyIbTKYdIJQQtUvy3pZbxSmtfbWZsbIxbLQxNa1cFYT+0xttVUXe0B9mVT8Cyl1091FjEOE7OP/mLv+Qr3fnuN1/x1XcbbE+wN37z/Fv85WCMzsdPN35662z0yn24bCymfLw69zdHDsPWUx2G2lATtF0Y5oz9I7JssJ3IPFd3334gfdBXQ9YVMWEcB3FMiXas+FHdI1jV16cErRU2IF4MzTHhjYUCbNtpQXJBjjpl1KBZI1YY+51FFFvqshGpifKeJajSppMaNTL3EisZ3D8Nnm3j/vlgv+9YlDjrdofz0on2WslbdqfsqEqOhmyjsAoRKuZ1BRq361HrDyBSVYjbArmUbb2mjEFkB+uVwP0LUYNf9N0i8reA/wj4n4E/flzsmfk7Efmj+W1/E/i/fvbH/vn8d//ag4FMrvsHPBeQwdo2mjaEO3/x6RO3jx+47Xf0OXg9/THn5TfY+Atc3tjHD2T8KSYfkeUF25ynS/UwxMx1tGxoNuJk6FMjPv7Adb+T48x2+n+oe5td2bIkz+tnZmvv7e7n3BsZkZnVWUDTjdRvwZBnYIgQQuoJDBjzBIx4gJYYIiEkkGDAE/AASAgxabXURdNdlV8Rce857r73WmbGwJb7jcoqqjIGLUV6KhWZ5944x8/2vW2Z/e3/0cB2Oo0tFEmHDNY4VSXWzpF98o+loOamUz9fF38Mw2ertsyLGgRDFPhQ868C6ESbTw82TBFkLNk9WMJZXWjrCr1YntYUzBjXFbsP8tZxtfIrnO11SJFfh8NqQY/OqlvpHUzxcTDGUQy7pTwm1onnvS87RbBS2rJi5qS/c+hgcaGP5GCamY6BG5zWjQ2DceB3x9eVQ658f3Puv95o7Z3mxmovbO3E0jrHCps71/fP9CMZvHI5v7BtV75tV/bfGvfjjQrpUboJfbxxupxILRVCDOE2rvR9sOCMOXmVu3TZyQ0VjlFZjbrsvLx+QHDebrcyh3XBQ7ClQNs2qip0Kwq5XKcnKLCsxro2RFZYb6zW6HGwHwetrZzaCcJRMV62Vyo0puzVVJwmK8LGth7VgkVnaYK3Mzkf+LE0PtmV7bSgeWLcdhijyFQ0Duqe1KiOIHyaw4yEA3IUcU1eFFsEm2No3Y3GmcbDYvfHvP7owiAir8D/BPxXmflJ/v8jr/62P/gb0IeI/FPgnwJ8/fOP7FQi8bf7jh2j1ogonz5/W5TjBL/eGZ+/41he6TbYOTg4GNlp6c9mKWG2ZtXJyQ/egu4DTermE3jYuxuBbFqIvlMrpHjw1gOZKyvNmiuZ7ENJf1J8H7Sacq3WGXkGPB2dC4ySLCFTSFnRF7dfZ8zaUgGmVhOrKOBjMo++XNov4OVjTZpFmjLBXEvqPbMqUmrbMByyUzTeqFUnrVpwZH47pcJyE9KsdAs9aDLYWlF4aiwxtFWKFGLce2f0G3YdmN2wMM4fnNfLxrkZH14+0pYXzn3hu9sn3lIQg+Yb5/iG34137HCuvYxLew+OYQgbwpkHmFcYVGJaXp+t1TUyqTbbas+BhVaMJUGOQXrdQ6KGZVZOxcSDzA1Ep5WnoTlp6hITfBSIyiQR2gT3NoQNj4MIR6M/+WY2aW4k1Xl5aSxk/nlhFF7o0q10LUurniW0skLFtOwGHrjQBEScRH3eBw2e9tIIFcD7oPA/HsXK9fxxCMMfWRhEZKGKwn+fmf/z/PJfPUYEEflz4Nfz6/8K+Ic/+Nf/PeBf/+H3zMx/BvwzgH//H/95WaoE9Pu91oCT+nt8vhHLxsKJ493pbUc/BL2tHFkuvYd3zEdZZac8HaALorPn85RaVXahMRYDW0iZXo2qk6cuqNc6K30ikEDObEGZT9ATHQ8vZyV+8AGGggk+qrVrVms0gWpPRxUUzdr9P2jWJRIql4SaMKYQqx/PDMKYLeTw4jzUXVbvcWlCtCJaqQg5Wd0Vk6lfnH80eYRiM5ON6vd5UG0XyMbhwhEVZ78RXLZzwRQJffRKlmiN0Maxd/YxJvMuUUrMtsuZX1wuXOyFl/Mr9hWwCWsMMGHJJE5wDIfW8atz7ZX2nVRLjT18s0atBLU8PcOM9gDvZqI5VEuvGYys6Lqj74w+iik4dSKtaVHl3WBU0XGZvI7pvCvC1OAUHvOkzptNuXu5YcHkIUzXsMcmI6fazaN0MhU0Uw93xVE44xaw6sylqPdXoqd5s2Bl9jtZqek/uM8aPFuBoj3WfZo8DyKhBHr+I3uGP2YrIcB/B/zfmfnf/uCP/lfgPwX+m/nP/+UHX/8vReR/oEDH7/8ufOHx0okc+YyrL6ruVEFabQ16d/ox2BwWNZa20qyhvdaHQWU4dK/w0SD5a9dDihCj7cxmdfVHDlwdU+PYD2yi8jFVKDJlr+NxnIbwkEXHRL9b1uK/2voqCqQ+0CPEdFqW1YOs5qytVqVP3klWW1g6h6I0l01kxdZ79+oSxKZ1Yz7NWIQp4NmEZX0E99b7CKVOlaiHKx7WR1oPXt39+VyV1h6+bosYnWMf9PtRaVeXF5SGpHE/rqUMNDixgmnxC0ToPZEQ9utOaEf94FjLK3PZhG/aV/zcKm+RHHy+XPn2+08ghlKJVOsRjKy8iMHAs1Zvml6dhq1YMwh75n9GxBSeVfdlVnrH3os3sGopOatJN+6xkyrzehhhpd/IKUIpE59JrvZk9B0xm8VhPomUk7QLX7xRTCvnw6tjeZpQm07Ae9LuY9LkJ/GMx8/LJOejnO7lEua1qnUvFmfdM3/YiNczUJ1zrZ3ThUj5A9D873/9MR3Dfwj8J8D/KSL/x/zaf00VhP9RRP5z4C+A/3j+2f9GrSr/ObWu/M/+vh8gBBYd1HCKHFLtvZa8Tp2Mdw4LxrKidNbofNxe+LCd2Y+6FJHMgJEqLBmNiHz4o/BwXi3wrc1INimCTxz4/XgsnCrpSYv0ZFomJJrzpoF6mB6n7mzFJ1nx0VpMMQ5f9spMtpoqp5Pw2IuWICpmHBxAEDbVf+6MY8p8KeDQqdUic81umk/t/roseEw1nWbddMaTJxHxWG7OohCUvDfnX5+jVB7g7wd5HUQfxcTMh1FeXdCICiQePjC82H4mjKij7Oidt7cD7YPr2hG78quXD3xsZy7bhXXdiBg03fjd5fe8njb664lrd677YL8P3t8PBsoxkiOSh4xtsROKoTbqRHRgtusm9QyO3ACtSSyUCiWeW4O5YZHnA1NxfnhOU94ZaCtzk5NFLVekAnxDZzmuzE2RHbT4KOjjk4pitD5TuuYWK4tpCQpnoTWZaQRBqW7qO4cX+S69OqhqAQPWKlzMewCokGOjvv4AtmsPVuxP/3FEhj9mK/G/87fjBgD/0d/y9xP4L37Uu0BItRK6aOVG2lLyX103rC30IxitcxdhL/oP//DPfsl3n79lHytXB0QZY3Dca2b3pbFoHeUiMfMjvbQJ7pwK1+PT5533/RPaLtWmQ8WsTzCnzdO1vFGOMs8IMBaEBdUaU2r+jSem0dbSecs84Se5ePo/TMHN3Je7J0i5Q3+/JzkGHh1RWBfDxLEIBg16sg8np6GtSZQjGlEdQzp9FElIpVrWzFY5EBQPgPGwd1OEyu+wWajux8H97oy3N2w4r+3C1jb67WAMp0XRyrSGXvajY2POIgYSKyaVunQLiBFs+xWWK7fbJ5al0Tblsmy8tBdWGv/4H/0TLlpJWAfBfT94//TOv/j1t/yLf/Mv8dF5H87ugDVevGz5fnY5laN0q7xLy8KLqriv5ckohrSgbT/wxAw4s1YVaY5bYK5EzGyRWTMfdV89uLycSKvCfPjOuF8LX7Dksp2K9eozpzSqjV+B2KqwDC8/SDFhnRjSetowKwPXcMio7NZVjEOc+zjoR6JZdnC2WQHgMQvLg7k0q6G5FVFN6zbPadwh/xY6hn/7L6nAGNSmAcac3XMwOBB7QZtBHPiAHlf2ZadN/YM8KuX8h/t0bX7899kxAM1Ko+BFg21aJ5B7FD142juP8EJ9KVvxtur0T8gnWPiY4sqROuayuT4BITCpfMfwAsaqrag5tphu9baU6Utg5f/w8DtMH0jTUvqNhGPUDpuHXyL1iT9ACoqcpfPEk8m3D5KMNjucCrAlJlirCpKV+v1AtOe8HgTaDF0abVmrULljntyOG03hLBtCtUqFCxkXaxSRaxBp+KiHydLZWyHq4+joKrRlIVfnci4z3m1RVlXWlmza+JQnfv2XKwnslJmMZmOb32/M7sizVqyeExcK0LUukmp9NqLMMbDGUy2Et0DiBPE5KvIgM0BYEZkCZds+4hMMHqM8R90dSeVlNTRmMM6jqZx9grUypp1tzSw8Wupa4MgyX4kQRhbAa7Kidqv3N8dk0/JtKM7z5Lw/vRzrAHzgCkza1NPV/E+yMGQQ2dEHhJdK0w3lQvhOunFZGzLguifN3mh8w4dL8vOPf8Zf/vbf8P31jd2/Zlu+Zzt9XevCMRgrJYsWwSOxCGgJi02IbOHSjCuNz58/sW6v6HLi7fPv2FJo2wlOxm13llU4r+t8iKZE2ByGk7I8NwX10FYISQJuj8jR0mY0tIQ1x4BRgiFbjdOysKbwvtzwe5AH2Fr2+Ulx/eU4sGXl9HJC6CwaNFswWRAdxHC2pbEtZQbyILy9Hfu8uv4caSTkuVpNdYSVBaPJgQDXHGzbhdN2KQ2CdzLgOALNKjS3w+H9yrouNKtU7lve0Qw+rsVeFaOuO0azBiwgydjv3PZPxPnEbr9AXr5CT69srbYgMs5887ry1S/+Hy5H4+Xd+e67O5+vV0Z7xS6vBFlrv1Ru98GWtWk5Fri0DT96qSVx+t7Jo8hq+mKonmowi14nepYmJW0wotdhoScup8Z37zuv79/Dy2stKcYbQWe5fGS9HezHlXXb2Gzjdgw+vV85n0+8nM6Qxn7cCmSUFTxRDbamjJNhd6fHJCntd8KEozmaSgydmpKKNlhYcK7laN4KhD48WTmxyUq2HVenW1Y6my/sNtjzTzCibp5vdcBPzwCJmpCIRnhyMOAYDHM2f2OPVj7gJ0PPG+39gnBG7FpUUa8TzEe1VLTaKDiJaoE71Wz77AQKJMoA7/kEeoYG0by+lwvxg1lOrNKty3W6BDtfFrMPp4BpEQ9l7cWs41Mnkz/oaPSxZPCi7x4xoEP2Ez6C7oMx5ipNZ4ZBZcwTNPzw6fHw5XiIR7YdICmQM4uBOeu6zUZHKSl5sLNz584j1hcmkeaJ0GtZvqlgWtsOH0Akoj5RkMRaUC4F5aplCLcsJ6RmjcVeWNhoJgwGfezsx+zaIhgKpt/xj/7szzkGfLp2zqfPLN9/5tNu+FjQ1ut37BOyESk6dIfgTvTpialzUzRtu2Qa6HwRHRYImDbxmrocRXyaDNbruCFjATVaFIO2PEGrsyz9TJQHw9owKRr78BtOsLYzkTZxmRpvdSSzDfrClZFamYpSI+JSoGVONqjoVipNT5LpSpWd47iDLZgay/wlDjoobG35UU/kT6QwMMk+xUqMuZuROf9mdo6jdv/GtPSOK9vlhfMH5XIxTk3pAaalPNMB2YOhRR1eUlhMcSnKr0hV216yqcq97FOkQ+AjqO49kMNL4jsfCKJIMVrR0pg1TKTWoQnPTPSo5+3pvkWVCxctqbfzSLP7AWE1CVe6O0c/kAX8SIYHezpxtLK9bznNOua0lIG2rQxQpERfJRufq7eC1p/rq6odtdp96Hei+laGlzGsyVJt6WwxDHmCrUcvExK03IjoEy9BwAarBuc5R7dQFqt4vSNvCKOYl+3ERqJ6cHgj9iBi51gFmwKoJsavvv4VDnx9HLy+nFjOjeu/KpVm90RHaS4eDlzh4CN482sJm5wyQpGSnqNasmnzyVcARKnIuAk8RhXDYpFXQb8dO8t6p+lS7EIpz4vyn5thQFK41enUqmD2g8PvmBlbWxkDenZ6hYuydScG9YFEFs9ismkfYTvSStEZVkrXpoL3ZHhtn8wWiFttTWiz2GtFMNpBk5XV/gTVlQBPI4G5Ey5xUtIWJ2UnsxGyTvnwgtrKx1Pjlx9e+fX5yrcmRB9oJG2Sr9yBnqTUDdsq632u5Uptd0yrN7Na76jAw0wxQhieyIinNL9Q+QqheRiJusnztH/Q31Pmz/cgVOe0Vw9ZUpH2PnLKih8fxCPJav40zSd4VHvvKPl3tqkDWSfnv1aYyzKDUR9r0ZwjM1Kg4IMYI3ONGsY+3Y0NeOiSZVfa0QjK8yGz6N7yONWyiiYzKMdTZrdV+NAYg9GE09IIL/fjCnP9wSqNCtklAndhiKDjYPSDfTTaacNsw/IDhLAtC9tlLafkcfAXf/Utd9/Zd9AO2QX72VLriD2IobzvO+qKZsNc0QmOEo5PebdqCbQESBWaaVWWGTEfVOp1oxE6CnsRwTSfIqcQJXWOZOk0M05t5bgPRj8gpNanWaE3eJ93ks6bJCczsm4gbbVWrc6w/rdXrZ9ZGw4KzWbaVTjQsHYpLVDEPPQGpQtxftAa/VGvn0RhEGrNhQjqJejJeVDlslaBkAW00ZqWY7RAsw98vSx8tQ5k/QtufbCmIC9baeY98FG7b88o7cC6TOVscBzJtTtNFGxhOydblBbXpi+/R5CHw+upzF7n5uSxqMlMrkdnFf1CQlMl1Tj6ZD4+EFAtezkXYBTg6cBiUEEpyYigiXJezpX1eLYiAnkDPXOz62T6beXEPHffMeG5ek91XSPmjQRzCZZPgZRS6/stXmqE253r3tmvO+9vOyOEtgiRtS4F2NZiTARMqnaxBUefs9vEWA7vtG3j3ldidzKvNINVhW01hJWdznu/IzEwN9rFMO40GbQhtKOxmKJ5hth51Vdelhdelo2P24WlAXbgt0E/kkjhLB+R1qpLioHdCpqPLCermTULCZezTXJSpY2lFMK/mbHnHDOtFQ9gH5xo2OVnDB0MdYaUmK15xcDpHC9D5mpRgCia3aYfGN25XXfUgqUVD6apwEM2vR/VlW0rbVvYFI5dWFBCFhzHj2AcA1sbL6eNy0tj+M7b7Z22bbyeX4ge3HznPjoM2G4NZ7Bz/VHP5E+iMPw1Z1wVyqBiPkxjEnF4YKyDHsHehaVfufc3gnfWFpy3efQu7ZkAndNVOAjGUuBjHWxl3XXffbr3UsakPPZUX4C7DJC9bLp9M4Ryk5ag8IoR5DJngrl5qFM6v8zoPIqEPuPqHyYqMf0cnnOjJyaGtDPaJi4CLGH0dpoOS32+sapG5T+ZQENktvwkquUP6T7KYXnMkwfDJKZWPzgO53Y/uN93+qQ8I0upGUe13GFLFRYeEvE5Jo2YCHyhKRXqWmS1Yxz4GCxiuBn7GCxWFL0UaD44WePc9+ooWoNcyA699/paDG5HJzjoh3MdzmorL4vS+s7A2TOr0o06hWv6rvsqpcx5MJm0Zvgir6w56rFiHh748OK4CBC1ZmxS17cs3wFrFVEgFfhSjFAp0pxTPoxZY+axFzHPTsa2NMwqKqAPZZuu1Tm7hgmTIFmcDZHq1GTqYtwTc8fHqF81FJUGUuY9texaWCbF+yh7MvIHw+of8/pJFAYo7gYy2zIpBFtQchyAk1YhpSHO3pV7JFtz9r4TOKe28LK94m0nVZ9z4UQ0aysx8nlquge9J/1waGUvZo88CCCyHq6I4lXIrWOh+LlNqLRIKR6QXvmQPJmMWl59NbNMA5L5i0784fjD/dEEPmvbVvkQNuOmMnyuooS2nCmZ8c5jTyksIMK+748fUt/nURDyUSRnIGrWmBZSIbYRwb13juPgOA5EV6TVyrFYxhVkQ3zxDihbu/rdeZB1pq/FQBlk6Sy8lwErRriRfrCtRe5wjJaTlR2fZnLzuZhiIwkPjtWRW3VX++h071zHztKMDym0EewqhGd1Cl0mfTimLWqND9pgsdoIWSr7E/WZ9Trrnz0eYbAUNJ3UerI9VrkOKZg1FlZExjMVDBdcIbozoqz/RJTRy6l8kVLPpgt+CP1Itm1C08FjnTUt28qT4rl/VmWaBSCy4yO4e3XRoidClH3UTWTZOLFxiPOuOxJKe+Sm/pGvn0RhyEz8NpClTR173b2JsO9OW9Zqp0aiB9zOB+hO7sKg883HFZV/wG9uVz4dK28cvN2+J+7Oh/YV7XxmtINFnPOyEVH750xhjMCPG+t65qzV5netzcSIAVtDtkoncpTr2x1bnXUzkgu0E+cPRSQyy/JzYO7EZ95EmGGLsbaaSbkm93Xl9v6J2A9e9hfQM+MctAHtIgynvBX2YIsgshFhxLjSVkWXjbHX6GHasaX2+NJilp2NzEHkIPOE5MGWWb6HRyU9pVYrvvtgH15GqpetChkDXRuL5zQEGRx7aR9zGhJ4C2JNPtgU9cpg4FwwfBz0Y0NZOOJOH1fe5c6SwbGvbGdlPTXa1ugSDIx1nAg29q1jp2CLFf/dlXsI23CWfiVjkP1gvQik0bafkbHT+w3//fd88obayqYrV3akKW2D8+nCsmzs14PPn658WIyxeUXUSQUFRTp73FlMsNXKE5JC9NMGXAbbUEwWooGPHR2Ntl2QozJWTSnzlT1rNHTnvgdb23jVDR1VuMpWrgMreb8CiT5FahCt0XQpGzdJlmXqQ+435Dqwy8JyaYitFV84ajOyrqA+6OOdm+4s9jW6Z/mc/IjXT6IwwCSZPHkMEC5PSqekT6PWyVgKgRD6fUdRPl7OrG1hO7/w7fXgX/72nWUUX10E8nDCB12gH3266ATuWpRVDEnBYy05dRSFzijilbGg3AFHLGhWfgYmMdcfxdcnq9UL/DnXkvbYKBbhkmTPwXWHfVRn5GL0CPIY9CG1ovIsg0KCsPIFzAjCO7gVkDq7CNQQbdAHOR4Ua392CgT0GHgMhuckSdnU4AhthtRMJu8EUIv6W71tAWMuXygzFTQ7fRtkblwmGKa0ev+exRx9bD/cCV2eFGZ3kLp8SK9rvkjACBatkWpX+NTf2LRxicaiymLGSVfEndgOLgi2bLx9l+yHczsGu3jhGVbdxtgn3+QAoqzsSFDPYrtl+W2wMH+ZIoGFUitMA/ETgZQYSpW0eiIrBUqn1X9M5yYtINNmZoRWgnuMv06l99soZqJSpjlReJCnoByIZGWpjFHdszssK1irxPR6esouIB6Adz1T1R9RXar+QYf697x+EoVBENpcJe2RHB6MEUQoPa0SOCSvbwAAIABJREFUlvyGZLWFORpdywNhUeP1qws/W8/8coff/PY7fvPrb3Ez7ms9EPf7zuGdOyfiU30iOShr8pG0ZaWJkey4F46xbEVt3lqyLg5eFFXbVi7bymltsDhhB90VZt6BezAySltgVXCalxIvox7Azw7vn264GCqNzz05Pr1j56K7riTsx4xNF662Ma47ft2R1VhDeD1vaCajD/ZjMNiRfXDeKh0rgrIk32+M4aQFXbzeWxQFerWNW94mglGOQqqKm9XefZqllmXRinqBfFoIB5FGHJVIRSuXpCErJxNucWfvwXkru/djV3w/WC4rQTlR+fDJ1HM++8Fqyro6bYNlLxUskwtwNGNR42U58bIYkd/z7XHj6MY3y4a1jW/9M6dP73zed0YkbcqXv/+uc3u7sujBasaqwn04G1pQijiulRVhYogH0qttz0kz8GHgA2nGstWD02zjlgf9/Z0+x0vLwqDW1p5JVvuHYlRex5hr4wJhEtivg7bOrkWLjVlF01kW5XIWjqPh92nbH4ZttRP1YK7IA2IUjjZjCovmviGSZBtPHO2Pff00CoOUWrJKRAXWjrkVD07VWsG0A6s5Lxx6whhOSBmgnmLj0ozTqebcns49BqM7YwQM2I+jlnNzDoVEtAAh8pgmqKBtLZKLSfEGdKkV1rqwtgWbZi2hiceBSPk+jlG03FqbF6lk3mLYVGwuY0GYXVAqfQySzqWdQYXrfsdGn/mJjb07fh01btFQaeh5I+j0sfN+vXHtoyxn7YQtpS70hH3v7HvQNgVRbKocIOgc83CJJ5fiyySqjPHFfk5Ciq8/wdk+Le8zlRCd3QQFVs7vLWK4rDhKUjmaaoLKY74v78MKWXD2xZ/vTUaFujaDJq1yPKWxLCuntvISnaMJ8f2OpUA6tgSnEwTKMYJ7TyTLEdsfLlWTpzB22KSAP1kmxjU7pFokCTImL2VMQFoHqBBRGJaUDJc4BrfhLFou1msrVadHdW22aXWMT4njBKJVsbMirZVqs9Rbz62SNrDWWLxWlqZR9yKUuEp8bpiqKwmkxnAcpboaphxclz9RHsPM/0Sn7kCEcsZ9ZB7wIKJAitJTyL1m2s7OSYUPuvLV64lvvnqlyYH3O0ceBR6iaKci1gs6rO7AKaNRK4SenNZtS6U4LVpqu8WqqldOwPMtfSEMmeFRDD4fgYnTbCu0eaZglYZLYIfh5YIsEYzotZu4OjGc3q91Imk9RH0flcHgjg1wK3czPDi6c793rvcDb8qLB9omUSmCniU8CnfUFhQl0qmsZFBbCGagLzU1tAnE7ceB6kzNYobV5A8/Na/ewaYYiSR8MNwrPxFlZK3vAkokJ4po1JYmpYpz1tg05qioDw24ARuoFYU4GWXS0pSvzi8zvt5JTw7fEQu21fCe+BH46KjW79aa0ZZieeYzPrDVeySqYNhcsxQp4amInbARGgJRzs4RjtmYo6QTuxefhSwJttQBF+nYaj8omrUleTAd22kpDs0fGB9FMqXdzNi9UfiRJJKlkH3qZVSndsKxUcSm0ESjUsukgcmfIPMxZ49QuQ2wWhlseMJg1Gn1+I9Qpw/lVRhA5FHGJc04Lxv/7i+/4aLvRHf2OOrEskaz5LgnZrWyylkYBrWLyGgYBSCtk2RicwNqj/+/ZPERdMAsMOl1UgTOcDh6qQ9DFNVq7GK+9zGCz7ed675zflCLpNCT9/ed4/OOj4NtM07bWqutPko2awq503vyflUkD47Dq6BFI4nKmpz32JGBq5Ct0dOLEUrgERNKrQDduaupApcBbk9/jGolHs5AddKnBcpamI90aGsd7aNDjmmkWoy9GNMh2cs1Cx7gexb5TKlT16v7Ci+iepI1Lzdh0ZqNRh4cudOlsa7CC8b+1Ua/d7gmH84XaEnjwOOg3QcZFTtvy1Igog92L7duWCsVurhOJc+f2xfNWsCaAmt5X8jw6XHRSJzUQFsDWWl5g0hGPwprYKkTXsBCiKjQ2ckYq4414NTKTughhnoE7tVBuTLC6R44O1i5frUHlb/uKqDeP5E02tTQVvcVOSr1nfVHPZM/kcIAQypGXERoqoQVXuJ+Q55Hx2Nt0wisAjsw3IIhglPJyb/4+VdIh7f3K9/6lVsT0hutObd7sBRWRHo+oipLpYeSdIRgk4qkM6m8g3IHrlBd14FLIti0gnsUhqBnKTM9av9s1iYRqQxUjhF87ju3/T6j6pU0IUzYP++8H++0mCf3/HRynsAFHnSC5Hqnio8nSmO1FbEr3etULT5BkKbY1qAb/OCGqlFX6haSJy+TjArdTXgKrR4zhms9sJDIqA5BtR6W5/Eag4yGpmGp0AN6R0Np8iXfQKO+Z8y8xkihT4ysTYjTiVpnxpmUkqaPOLhl4yKCWGc7n0tsdAhtPWMBZjtDrrx9uuFM49620NpKd+hDQEcVxygHq0nhnKw6ipMghi5MGzbQ3InmZYvw4Li1UxnHyEFmJZ8PH8VhWWvbgU/8jzoEY5LMyLqnJLyAZU0euaLl59nokYwoY2IXUKmx4wmQMkexuYc2aaiOYnZSXgxFtPoTHCVMlIWN8KS7MKLQbhzEX+bMD2h1FoyDcLivkxUoDRzefEe3xha/5PV14/Vnwktv7Pd3RnPu78aKM3ynO6gFrkmf6yRGB4MxnV0svZRwCa3DdlpYL9tkKRa3QEQIDdzfIWBBcDf6bcfvzsfta9rLwmh7dRkO3376luVI+kWwBqtuU+KttPML9+sVDSOHMbSKFgkyhOturCdDJLm+O2OvHE1Z4Of6gdveGQ5tvSGA98bttrNusGoDSlLtY+DAOVuxBI0Z9tPITE6bsh9rza1YyZqpQBhVmUYjMA7Brzs14wxK61ZciBENYa/Ry430Mm0VBfdR5X7dcAv62Ivlp8IhdQ6uBMorcC3w8u7IMWhp6Olr+ihp/seP3/Dhxfj86TdYWxBO9D34/PHMdXQ4ggUje3A9nI5g445++BnS91o7qoFeWOUbjvye3d9ZuHBprywC3++/x06/oFmw6hWV0rBYvHG5wLe/P2jN0FazvnuBpy/LCrrR8xP7FeI4MNk4b2fa6wG248cr2J2UYxKnDB9OM7juN7rvnG1FtLP7ldQP2JHVwSDEAtkSvJPqLEegruTiGAPBOCT/5oP3d7x+EoWh8CerG7zo8xxZu3yXPk+ZmUkpUpU4YDqmEjjXcXCMT3y6vwPGd5++4y+/+z33/eBl3Xg5f2QY/L+//Q0PLYIkWFSQ6213PBfMYTmAe5T3ng4SWHTwtsNnP7Gd1zJPsYLtYigjnGM4PQcPD8nWNr47vsXff4N5g1joGXzsr+ipnHuaCrDj5nBy1OByeQGHo3f8vcxEbTXaoozJ8V/UyJXKr5g+hu9c6QMsDS9L1FIKrmDrgS4roxcprN86ixqnX65lTmKKWaOP4PZ2LW3ltiCPRKZw4qgC5QEs04dSqLAapWZvkiYVh1dioBXMy5rejdAiVcdkVNb1DYZMctaoe0Gy9Aw3v8KLgjS61Uj09vnKz/aobC8RvoorH7aV/+Cbn7OtZ26786vzZz58gF///sr/9c+/5z0PVJKMg1PubOvXLA1SjMwz0RqDRv/0ezQFW8vGzvc3YuxcOMNp1BgqK5G12bhd3wkOtovS2orZicFaK0d1htwwDY79YOyN1mBdE7Rz9J3P+8q2fEJJwg3NldWEZVEOTzrF4WycWeNE953hN8SMthXvQZn8FD1XFJ19Bm6ILJieCjDO/W88d3/X6ydUGKLsd2YbW7gD0+uPabevc96Uv0HwrL87jT8D3o47O44sC9t5Y7ts9LcrS6ui4CmoFcvSGaXq7F+YfbLoZL+VCYlqcrgT9wG6FFhU8DDhC86B52QWPsw+snHdO7f3d9Z9KzR8ha9PH8lzRaSJJuSY1FcAxdoyzVoG/TiqW9Di8+eDejvBQG1UKIkyi0HN1TkSsWljbjlbyemMHHWtmjx0DgkTC0GYXpaTpPUYMYgJOdTPjsksFbO5T58em7XC4KFkFC0BkUy77uThfF3UdJ6GNTo7xMfkImCKR+C+TgVrEav2MXgfndUMbzvaE9PG6/IVYnC+NLAzX7WP7Huyjt9xj7LzdyBiobmhZmWEg+A0sg/e7gebLWxrTa77xB+2ZUHF5+/EJMg5vZcG47S00rZYAbW7ZykiswDDMRJ3ZVsNm/4clYcZsBohJf9PP4p6rsoYpf5NmVZ9WclcHIO2KrTCwGQ44a0Kux01utgCtiGDCkv+G/6Qf/frJ1EYAJxKe87H+opEtSzPCTiyIluhikNFv81wTxLUnjyB3oNAWNoGamyXM9aE23hnWXWyT2ewbcVaE+Hkfse1EpFtWTAz1lZOxMMEC2EcyTDnkMJByKS1cl9q8UAq63v2XTnuwX5zvDuLJbEq7bSSp1Immj5oykH52QsyhBy1uRjUrj96FhOy6dycGGYPjUI95KENkfIbeqilhIHpDLqNgCgxmS0Gpux7GXi0tC8zf0Vwl8ZoboIeSdaOPDaNiNhUARoR8+sYbuVQ3QAVQ5m+Bw94X4wh5WoUnhOgXYoMNn9kaRuARTgGrJpYBhLgI3j3QW9CE7jvB5/j4CIbKRvLtrGeTvzChf3SER30XqtfmRDoPjoXMba1ODTD61rEkcR2kFTMHtQhoCuUu0T98ooj4SxexCW12lqYPcqnkzlJdK0OC0mpwmKKRq1+m8PUbYMU1d2BwaV0M9LKh8EpIDIG24MLMQs6RuEsAHIr3IIz2Ir6tfZh8Se6lShJcfWpdaMXucajfuFwxyWxHrBIjR0x2SdEXTy+3LhNGqcl0Sgz18yO+43WznSn/p2HS2z9i8Q4qLyEEhB98Ukr/X4x9oL9dhAjWJqVI9MlJx2auvlbEgM+33b8MDI2BoK2KMOMZWYzFnxJWpF+tB8QMHan91EZhxQ70SPqdDF7shmNah70QWmRoil/IebMnd/sAoo3W6s7mRuuY++AEuoT8CokTsxgFAoqTbFpj6b5yCgYmFQ30CciXsV87jukPtlqv56I3rQ1B8kkuk/8QVBbS6AZk10ptcIThN0PskP5KNauvuMcKD9rxjEci+SzHgQ7LzS2deUX7cL19I5etNzGvYiNiwiffHaiodiieBZHQA0Ux2KU65SuRXO2gDFtAal71NSnjV51O9VcJjklnEEB4swU8RK51cltM4NUR0nmsZlXMSPluifqhrWNmF6gRxZ57rToBCjLTZw2r3dmfU7W0FgQKUzp4Xr9Y14/icLwPKHKT7sIKzK/vEzfhFH+ApKKhj9l0TwR3aJPR5TEOrXq9nBn3Bw1B3tkKk9izVxWWxgWSU/QLFNXm0j94aXUvR93kjLZYMDYF2KpIFehz5OkaLzeg3E49/0+6csnchvImpya4t7RTpGVbKmuZbaBmcmx92IrjlLYqcpcKJYbkbtOjwR4uMDUKTuLqU+2ns/WP2fR8GLnba3WvY5PjwFmmrbMm9pBFnIawYhavU+RpwjLW5/RaaAzaPbhL9h0relgNn9V3GoVt029UEyFoGdRq9ppqVZ6FkMysWyIK8rAR+eWSuWYa22w0jiyPiBdhfsYcL9W7J8rW7uwbRs//3iGo7wbgl7Gr6Mwkx4Dz2KYomBbTq6HkQOaORKdHFIr4yeXRordqlFrxCoV83bW50Mbj6KsSbNOGfQZIsHSFFuMGzeEC6tdyic0B90HLSuVa2gNc84gmJsuIMe0xlmMmSCEylYmv3rUgTC5Jz8y0/YnUhiAuto65856OMuntdY5vlc1NkrboFIiq5wnqVPKS4+KaY9I9nDuR629zBxtKzYea/58OhmpM+fMmRilFT2eLmVAinO/37A2I8q92H2SxWN4fxtsWuZ0Y0+Om3O/Bbvf2fJMs5XeHGvBakKMOlGq/rUvrj3GnEsfZrCVgyGqpMz8i4RIrRWaPq8c5V+ppVeQmAuCCRziSKu5zNJoaqSWW5XVL8/j1ovwKkCTr/CQdddJb/V7A24z1CeyKNHhqPoX5uhUkWYWbvNQOq5e3zO8yFxk1mrt5VxmruTsbgBJmtfatr5euhHVNhGTxu47aWVCd+DI2CtAF+H2+oq2xq/OJ2yD34+DTwyubefMhg/n8KBlwmbVQa6OeytrQaeKfu70o01nLJsFoFKjwvYiOz1Qr2LmzRO91po5+SNVLb16CaU2O9bI+IRwptmJJidivOO+Y8zA5UdJyA5RlOcifiSeM68yqG57vaByR9gnN0IfQ8ePev0kCoNQ61tPBwlsLf9/NWh7wtbYbJ8j7Fp/JoEfMdur0lUpg9BgvN+4vXfePu/ch0MzTlvj48sLbTTc73Uiz7jz3RKa8osPr3RLEOe+D66jxocmyu1eANTSXuhjlC4ihGMExjuyXMhx4KMYaSGBqHBqyrk1ul0Y4txwzq0xnvDpYBUHdVor43PRwLYTtMRylElNShm2DpBWp7gNphKpcA/xSjsa3Rn7weHB2oR1NU7rhqxGuHCMxEfh3QNmNN6MTJdG2+ok/OKwMNfmMioMpycSC8UarD2/AMhWprHHbCuiOo5yZK7ZZfcyxF3PH0jdud7uxD3YblG8/1VqpPCiu3uDxZXQAea0MJZFYK3EplMZtOND2bNz6J2DwSWcd+5oJP/gF7/iw+vBL/ud37xd+c3nnff79xzHjZ1iEupYaAjYV5h0jJUYypF3zBa+sZXRNgY76kFj4QKMpTAZp5S5LkZbVvLzJw6CiJVFApUP5DKIVPqRGJ22rHQdcJw409l4x5sxZmT6DeesJ5o39uNKurCxzsg7CJsiqRGFK6xtOjxtICdAGMvOcfMSkf2I10+iMEzi6YxV26vdigoBOy2lJxBtlYkwfQ5E68abn0Z9OBq4dHjYdX0p3IhacdJZMXmDcdC9REvLkixL47RtwOBwyux0tuJhoG41z82WrV5JSkxvgodjlEMYpOJxx1kJW9A20zhTOfY7LhsPn9bQquvDIVw5hk++fM3Zy8Sm8hGfjDDGnHEyJ8W/zEh7ryQvRrLyXPSwWSMR9jE49r1yJ7R4BLVBKXZebWBqlFua8Yhaj5wxKLNNUQTRYpCWpVnOa1C4Tzy8BR7GKI8rJsaRiU5nqTa58KM7WBBatGhyjkwj0KW6KROjtbmdGEUTHnagCRYVVgSg7uzS6Z93tm1he1lwSQ4JLqfkZzT2cWXc7owwYghqHRdjW8+Y1GBQQT3GuijrAixB9IJqZjOLmZEpqNuzkELMyIC6+ZbV8HSOsT9DfR7y/ujOaT1zUmEh8KP8II6AdUq/84HVDJknaEysQ5/PDWqkVACQMIlVc9ujzPX+j3j9NApD5lzxFXmYLBsyvNG2urkbBpEVTEMpCIt9yNME5RFS6o/e2GTGgtdHlq2xDzjoDO/sh5Gr8uG11js+TVCHByrGeVWY67j984GPPi36ovjns+U2o9rg+d8yNNEnYmwNaIKkkSM5DhgcZFuwVgy7CssJRg/2vR4ME4jVKmsCMNf5zUoPEDM3Asp8ZBzOPasoNC+Aq8mCaptjAHhP9r3Tj86yLbxetsJFqG1GhExQQJ/y6BilPJRWa0nR+Vk9Lrxq5UZGgXc1jgABrlU4QyqRKagCqkqd0GJ0he6DcixSqkyW+9Y+at1nlNS9iPG1Lg0GZolb4JMRm1GrwT0cH05XY22O60FwoM05XZT2aYN8x4+ylqd1ojVO8hU68afwgdqormebo1T65IqALobZCU+hWQE9auXG5EetMc0d12TIJKKlTVPhUvESsC4bKPR+EP2YAVMFAJdnZ3xJPFNIKRB3Xr6iaVtR7svdLyp6D3vg1PzIuvDTKAzF06/dmEmQ88QIiuQkc2+vPIpHGWtqEyyF7LXVGAkjKz/CoT4oVcrYUwlN7r7jWvr6mU1NW41mC/0ovrsmLGsryrWVX8LbIsWFHwORwPLxbqrthNlMeBYyHUGTtZSBVo6/lrMjyI1jvGOqCBtmJdH1cefYnf2odeMiReetGb+Slmk2u4OYu5xCyQ3h6IPR6/2vYmzNWFslecXkLjy2G2URBpr1YDwyPYxK804P9smHcC9A0GwuaaR+4nOO9scMOxO/Ix9yADSqyTDmYSe1etBYqOyUCsKN2SHV6k4gBpEdHwdjroBzIgteP6pm+0OIrdiGk3E2rf8LQ+k9sWWUpfwEX9tSnpo7rUaA1Oo6UkBXwg+yZ4FPGujWYD1hnlMdWjeXqJUtvijWxjPv1D0ZPcgRmBv3dpDNK4ErKocST4Z31mVBrDEkip4e8ZS2Z8BIL/OieYXLf6PWlGr1DIhVa1yf34MuXcQ/x3EN4ke2DD+NwpCTIjt5BQ8wTZM5w+dcgded+Nh1W2s0lJReq6fu5Tg8orTqM+zh0cxGlAnD2hrn02Vaxc0ZXjqYsigszIj3h3eiO9tpw0ewH2X+OXcKPOk4qfP9USCZZykyrU1WYaLphFWYLUedsLVarF34GGW8GllEHqf213WAS62ftJpDoh6ymdX9LIY5qqhWOrdiqjRpXH3U5oFyCmoUVyD88R3q/9f2oLY2w3MaxBRQqpGEFugYUUXlyU2Yn4mSjIeKEGbJqlu2ikhgkdgSleycikRda6xiCmu79uWcS527QGxGCzw+0aSpTaw+eXqDxuP+SRjOtm3YogxTepZb9eXc2Lcz6TM6jgVl4drfwN9L25GN5oE5WCquglpii5eGRKGtgS1CamNE4L3MWgrDKu5H+ChXqGbT2S+fMXZokZQOkkODNGbMvda62gMNn11NXeeQwmmmL0C1zDml8zqt7bJGjof2QtY/pAT+3a+fSGGoLANRWLRm4XjwE4by8FeXxw5MCjlXrQcvEKwST2AkfuTMmazZq/j+geSgqdNOFxLH/fuiHY+NzIMhK2cRTibEuXDmPEr08nJa2feDq9aoUjd8JTxvPk9akck5qPWdlldLvX8ru/aq8rBOG3sfUSQjKeMSvP69nABUeKUp1/cohqXkAweoNSZzk0Gr7xlReQstBYtlRv3V6aEC69ZYlipYIwu8xOpBNVWi198VlOJ/PSz3edqYh0+KYlSRqqS8nH+zCk0gM4atTjmdrTR0UpJhj9PRylxWiydQcfZH8QmkVnsPTgPUny/z2sqiYFVEvdfPFZNqpa9OqHDZNsQE94XbfuWQg5cXYb9ccP9M7uC5Fea0v6NcEVakaa0cM9n3K86J8E5I4mp4JFrYdnFTjlrZhpeoL2a830WLJi7A0OIkmKwYjVhgzdrGqEG0KgqayjiO6bdQHB6WsqwvyKH4EZKlBA2vgya0YnHp1XVaU3S6Pf2Y10+iMDyhbxY8GslOad9aPQg4mg+7sOPLfnwB1YYYNU/ed26f71xvV1xPwEYLZ7OkNWXfD2xx2A8WMU4fP5J98L7fsA02Onre0NNWFmPp6EV4+aCM68L318G+G9d35/Zph5eVyzcb2i54f8NcUD9zjMHhb3QP7Cs4fVi57xCjs23JxZ1DwEdy3PpsmwuMu3xY4buD3BLUkMXpwGZn1qXT9xfo7/jlhKSSt0/sw4l24sPayO1Cz53gzpWdscN9Bz4umCjCQEzJy8aijYjB/bhhoVwuZ5oZ72+fiITL6YU9B8FApLHame4Hu++V7C29bkBvrFv5AYyRE2iN4kFYUb1NtTISIlDatN1TdNuKoNWMNhKOXqG6cXAYnF4W2lZcjgTICg/yY+eIwXG60XxhHY2r7VgT1lzoJLZdWO3Gy7ZxeT2xtk/03+zsn+6kCX/+73xE/uqN62+F4/9r73xiJc+uu/4599z7+1W997pnxgkgEyJwUFiYTbCsKJJRlkC8MezCgmSBFBaJBBIsDNlkC4IgISGkREQKCIiQAGEhIREQEhsScCLHf7BCHGIJE8sGgj0z/ap+v98957A4t6o7090zPbg9742oI/W81zVV/c771a/uPfec7581gI2r64l5umaepqyMts4bh43bVXn0xleo+6v0t6greKOWPehGLQ2pwGostsKszFLZK7x29RraoOvCupQ0L1KQ6sylsY7+RO2VJQoWRtjGcruy2JZiNSWRlW2a8aFs5co4ImVPzFVZ+QbR99ArGkatQi+Av08t6k6NrGwgDmy4c/bJ8EiAj0ecLcz7QMptMdYMSOnz1dApEInxCw6sWii33dCeQKQTjmGKkt170fNEYfRuKLUgVZl3sHOltkCls9HxDZZjIWZJ7H+XYawKQaPVHFmeMAke6ci8LBt6kvCKxBw4Bl5wD2RMIfKIItCdiPyghacwSInsxWiq0BKqhNxg5TbNcJ0UVgFCVrxXOvaY1evn/wxZscThrz1HakUU751SglazMhN3SnQkDHEFzTfCp4Bp7OoKbT3xXMaZGBmq1aTaN4m6pJ2qQNjsOIY9NjYDJ2s9TWUkADF6bEM1f4x4l0opQzVKgpPsf8vSKc/f551HEW2UstBq9hR2c+NqB0gZ6tBZCZ4afVIKpRjLslDrnObLpJZoa6NaEDnZX57Vl7SmfZ/UQqcnRLWPJrXH2Wk8ZzrjyHTq2J7QSHrCyuioFHLqIzXHk41yrtBsNHrCa1Z6KkgMnIOfOzMvHPdiYUhBjDFuOQ2x0xCA09TF8CF2YZxo690SJhoyfC9V0zvJoBpMldwdFTaEvsG2GWYFPY3TACdt7UMT4148VW9yolDQKRWe96Vx9Tr0JbDuEJ1lWdJsdis5jRjvqZTKPCuFNoRn+2hOwrpuSJsJK3jnLLAWnvwFnQqxyeNWsuXRJMQ4Ea7Ch8TX+DBm765ishL0LDEt6dKmAdvgXRjsRNDIRqkoaZBaUjC3rzYUqSrW09ykaUVLo1uKkeoQVRBIjodEfkJEBttvSzk+L2fXcS+Q5sU2HLhLkrZKThqWnhbxDNSqxEnYpeDruAlKekOekOEJRQ5cAtcgqiMuzGVgewhqqQk7JxGCtTZabUTdIJTdvrG7KXQN1sVRqURkA7GWbOyJpyntXGvyF9aOlkDrhuqcsP0hNkOMpmS5Hv2PAAAV40lEQVRLDUdtwsqCWIVwNsvNTc4LZsBpHHnuzJzgSCcx4ok0MVjx6GhlmCOn0HCPLUFiLtDzelZlOFLlIEvfj81HkdzdQjxXvtOFcqAqHid+gGe14Lnar3WcxTVnua2mdVuvE1OrzFNK/q6kp+C2ZVOwdD9DYCkDuqpCqT0Zd2ro4K+nOEvQ9hNXZc+DBzvoycXoY1fv24Yck6QSkp+RJsrUWtrc9fy5pzFf4hNmwLJ77WNnkgLRUOkp0T6ql0EUwEiOPpowcRtywzHopzodsoFnIKTFnpPFR13ycCYSbFpGY4rRBxGg0ntn3Wzkl78bFIoPNKUlflFV8W3swiWIKGBjfNmzQSeSpDgvnvJ83nNy0NdRJijYiR8ggych6fQlms1ahzUqx0cbMhV0LkjR3DmPRnFJcJo6EsYUp8ogd+FaOlPboS0xLa1V9vuZ/bKwYum2Nc/sr4OFlW59+EgGqKViV62p2oWzrR2TnoxFFa7mSsHopo+brVLQIrSiNE08QbdjtqgtFxA7V7LZZI3zzCH7Aqf7X6Kilj2T8QBRNorWxFCEIBI5zRheJMmqzbGxEVjPJnl7EkzyAnEvFoZT5CV5vBudQNAwHISiIO4DLa+skm+ECEhVwoL9XDlczdT9nlobDEaaW0pwu6RqkUaWu05e+KKKtp7yhVMZ2IixEBkgwjQ3Hj64pjFxtTeWxVm24PXlmItEBFJzlCSSjMKwgQXI+xcrgk5JbkncQ37oKzLEYw0sO/U5pyyPmYYSVFWcSl/9zDXwbcPdmHeVqV4lPDyOdFvpkQDnusGE4S2rqm55c15NCeIJk8FfSLp5LjQNBFZL3oaFD7m7mqO5GkjkzV9O4KdOHm0SOZLgL/z8biIpmHl26IqgiDH5YGky+AfD9dwMtvV15jpTy47ahjfj7QoW3O5n1GEarNwIG9xHpUpnaprSbuq0UtnvJm72lTdXxTYGsK2wX5yugt3aIEPlbi4lJfCnqbLcJvNRVYFBBde0VfRy+h2F2pR5i1SxMkmJPc+FjFNztmT1KwyVpdMH4LQokAhTBuYDTXCftGlsMOCe2JNh5D1G/AbIcALLaQsMxep3EfdiYcipRCIWpShyQvdInq9x8P5YfsxKYhkIY57KKKU7XjvX15rAmRlMC2xG2VIurEuna0V7Yes5EfAteNCEWYLiEzM75noFjTzPD6WibVuZ55kPvPoacZOEpNcfLXzj0S3yRuV/L6+DpffhVJN67LblliwpLX+7bPS+sZsMtyO9pwtUU2GaJlQcFWcJ8jpoRaeGzlDGdGG1jVA4HJN9aWPxYekcdvDwlRwR3taKHTfEjJlR/ptB2xEqebTYHL1+BSHYbCFso7a0a1+3Tm0ktHozoijzzQ5fne1oxNUoaUn16KKB20IXYzlk1aGREG053bgCXZMsZcURCUyMGgZFaZLn+x7Csa8sfcOWleubG3b7PfubHYiwrEsChmqqUkmArcHtmys2GzY7WwvQNxH9AKcKYqozD/Y7YmkEG7evdyYtPJwrMStyKLw+gFQ6FUSdzRbAuXm4p9b0mSzTnqtd+mRu6y1bd+aaPI5aCyHKXMEGr6Iuc/JWilF1ohUo0ulxgNhnJUUerwaiKftNcQDGGJr0Td3LRKWx2sJit1QR5lK51h3dDXZ5ZO7HwNyY9hUq9He5MpR3eoKIfLeI/HsR+aKIfEFE/uJ4/KdE5H+IyGfGn48/8Zq/KiJfEpFfF5E/+U4/I0jwTQ62EtE1ujOIOxJ5xLBIRZtOokPtCfGJZFVago/YwDfCt2y+EHk00GCmUBHUsmHUu9M96AGlG8WTDCUKVCe0E7ri2SFEVYfisKA1YblXU6Gp0Yoza2EeWpHEqeGjhKUX4/GwsG1LlvmuuCfcW6WlJd3oOKMMbYfRvR/leu+d1bIJZZZEslx8Ui6/6CNKO0BVSrlCuWJmz4mIk5IRhayOHWEeP8gg1jw7a6onE2Na4qmZUEuDSL+K6JaYiU7S393G81d637C+jR1Q8opLQ0vD4hHmB2y4gVl3+iZ0UaLUlFgTxSR5GIRRaxoJuwXrurEsR5bY2DCuStqqiKXl27I6fcvRb7d00jZnICWTaKakxuY6PnSu4KMrW1VpVZlqVo0ePdmwub6n1V3Nv/Qu3B4XDsdDAqTsdFCUrCQGbb94JTw1KhIUpaM/kwsoJwzMCeQVed1CJqK0UUnltRBbKCjhsIbRI8fKKjUZx6oYOtypfDhd+6gkXjxepGLowF+OiF8VkQfAr4jIL47/97cj4m8++WQR+TDww8AfBX4/8G9F5I9ExHMzS8PXpK7OcYUwj6l7H25NBrWlHqHDYVuJuEVjToSbGhvpsjR5pVy/Sjdn7UtWZpK2Yq/dfAe3h28idWbr0A+3vLkdmeeHXFOJaZ+gkTHikDKMWTCYH7AcD6mGJLAsadZ6tZ95xIH9dWNbCp05z9x9xWLHJIbFGyyHI9tqzFPl6kaww4RxC+GYNKzWlOvQbK5SZ8oMaKdvTucIGqjMWF/Q2qm1EVFYbWPbBdcS3B4rU525Uqj7JW/2PvNI4c3Dgf76m1Q5st9fMV9dscQblNVTq0CvCWA9LqjDYjNGJ4oN78oFLca8h+nUkteCHDfiEJgUSt1jt4/QlkeyE0LUcVxhWmdWW9LbUpLuIWHobcBsWCNNcdZATJhuZubrlNELMSxWNjpVK5M0bB7iOZGU9rkWhGDpnevdKxxL8M0VbiZ4KBtl9wp6/Q3idwo6Oz12NAlee7UylT19eYQWRaMOVe2GjwW76I5aUyNh6cY3D0e4fURjZtvtaZOnqhQT3+jfoMiUzcn6KKu/qhQWAiO0YvIQtluu59E0NKcU4xbjdgmUlTJNTAUmspq2orgvtBLs2kRtlTrvsofVYUfFZeOoSzaXBzbMdi9ZqCUivgp8dXz/hoh8Efiut3nJJ4BfiIgF+C0R+RLw/cB/fP5L5DGMU1OxKEcvaQhaigw9AE+gDOCR51qpkpJakqjF0IJtWf5BznlTHk6zwsgJXZa3CmXLzrph2JaLQSkO4+fVU3PLV/CNA/lPL2tPWXeHSW/YTdkAs57Tk4iC1HxXzIXVE3jkBO5Tci1agqhUk779+HIkr0CQ82g2zt3YQEvJ39OGub0kFNYCYl3RgRHQAjqnScu2pNNTiXRoqgJa0o371K8+Wc0VFVyC2le0O+6eOINSkCl3QiGGAW+a5Z7He0NVWgpnR3AZAL0oDjPIpojHeXIiAVPTBC3VLBZ9ymba1FJcxoeql5sR3egdEKGsOmzfhvhNSX4MIiwRPFqN69qZNeiRRLMSheV4oG8biFInZZon2s1EfyV/j+6wWr7HW46S0Ol6oBaN2BwOa+Jfa8K43fJo5AGtJprX7MjUxrGY9JmISFdqBFSCkEgbvHxKbk5liMMXIVTZSt4hXqCV0RYevdZT4RwUikKrwVV1Ns+j3uQgZ+Lfi8W76jGIyB8C/hjwy8DHgJ8QkR8BPk1WFf+HXDR+6YmXfYVnLCQi8mPAjwG8+tqD0ZmPYcqRyMUIoW8L2tpozg1AByBeAU9ZdVFaaWmv5nC0jXLqVQRgQtgAjowRqE6p/DwtgDudzjxPlJYmtJNWahGuphjnxSAUFoPuzmFNKqt4oUw1O9RhHGyln5STVIiSEmkROYcu4RRr0ALdVSZNsozCGbRRTguXD4RjnGENeDGUFAlZi6WCdBFOAibrsqIly2GdhblNzG1m6bfMmiVsk0KtnrPt4YLsJ4elSP1ILekOnv2Bgp5VmBhEp4Qm0x3rI8EqidIVoUhONrQNp29lOHclAclWSVKYFDQEE8vdrZMLjQ2dws1IdZfEJ9j4ebbmjX6UE06BZBmKUhJ9xWbGm8vKXo/M6lyZIbGlR6QxKOsrhYm5FnZXDXvV6O7cbp3lNheFZc3XtDmr14gjvhocV3SuyJQfzhMUXwvU2nBf2fptojE0r61Z6k0UscSgDAxO+l6O1nuQDEyyUgClu5w3xWwO56QlVbeG52uBqpVdg+oxJAqFVt6tq8QL9BhOISI3wD8D/lJEvA78PeAPA99HVhR/6/TUZ7w8nnog4mci4qMR8dHrm/0ZWuu+pXW7RaLjhtKw6DR0A/KmA825+3h+8YSLSiQTs6E0KpWKkrNsLG8uiiAVWi1MVQHHIicS1LyBRZRpUua5sds35l1lnpPwYlbYtmBZguNi9MMtKk6toDUQjZSFB6qXHDu5UDAKPT+AJdC5Mu8mpqqUcf7P3TVHl4WCeHb7xU9n0byhRDVhxRjIGI+WeoaAxzhNV5UE8xSlSmWS0R8p4wg3mJAuYzI6ztqqhT5V+tSwqdI1sQa9Zw+BKNmXHWfZzXIkm31jRU4VnZBCtzLGlqTQLEZuy/1cr+QN7nmblxKpSHUSfBkkKB8SbTFwGMtiZ0FbH4uER8LPPYxt3TgsB263Ww79lm4bAlztrqiam4tt67CPc1oTypBKc/KI4gZucvadzCqtkyoMmkxfN9YtsRDdgjrNCUWOYDGn9230ZQATNAqTAxv01VnHV9vAV3KBXLOHc6J69y1p/aJQWsmjiZRzxVcGh2IuyoNWuZ4bdZ4TxFW/DZqPItLIReEfRcQ/Hx/srz3x/38W+Ffjr18BvvuJl/8B4Lff7t8Pzx1gqhPOkGDvTpgytZsE7vSUXYiQ86hXVMfQIPUJnFx1a19RD5omKaUbHNZg8YUDBbdsELWpcXWT7fJinb6uWBM6Rp0KFimosgwre+rEvHWWvnA8HHl0WAiEG1N03lHazE6VaQvcO8ubaZt2JkJVJarw5nqg+cZ09TAXG1vSW3OaqBg6XJ7MI1WOejY9ldyFUphwjAoNppJMygfTDcfwdO0uKfixHDdi2TBTItKsRBuUOsA1ZkPeMr0M3KFHwWrFl9uBisyOdg/SPGcq9GXLBYqCeVZj7kbtxlTGQtgHLq9whu66F2yT8f6mezg12O1H01UAgtI8d70NNhO8D8k8L5TSmOdx79hAp2ph3YTeYdJUxp5ixlbh1gtuC8fFebjv7OrE1f6K6dHCsjxi7c5KY2pQ55nWk2fS1qCWbNr2HtjihM1Q5ySB3STWIKyzxnIWLrbo1OkarYWpHdCWMvzb7YKtQfQAWYkS7F65wSgctzyGCoZ7grhkKlChnMBJJSglsFISIj1GPRIwT84kQmsVNaWYsuH41T4h4e0lLwySWOW/D3wxIn76icc/OPoPAH8G+Pz4/lPAPxaRnyabj98L/Ke3+xlJux5w1nFuTb/ZZJgFBrqlXLoIWnIHYtiGew/WyGZLuFCLDtl0ZSCG2SLOpbqPczuk01CJZA5WVzBBqqKSrDcfOxAYUoLalKkJu7kQoyH0oO1JSdUEA61ubNvGsq0JtSbJRCJJFtpWZ+srdXaYh8lYDCHSIYTSI7BIfQgJpXhyQpImPXoBLmm+QqotmCjb2tEINk5GveBFs+suMrrrkjddT7xAGf+eAlh6L7oKxVOpO/OTrHl1lLluqQAtQClD7s4wh6nVpInHYxyfWR4RTuatqVOa9DYJiPVx5ZLdVyc8vTlFSo6r+5hcaaHWfO+n0ctAlXVNYFfUwLWzArsKdU39jd4d4oBcb7TxXiRkPXIn59SPyuPS1CpX+yzm1y0wX2CNXPCko6KEChZrVihJXqCbYbYiJZjmAecviZOorY6zYs1m8qT0eFwRCKksZgFzrXgbDNlCVpWqTyB9UhnL3ZkV5gr7eQfWsZ5z7NKUq6bsvg3syo8Bfw74nIh8Zjz214A/KyLfN973LwN/ASAiviAi/xT4L+RE48ffbiJxjrEoJio2ew24sa1J4okCWmZqq4OUY/QF1EvapPcl4aZeuN6/klBbcVQL7vl8p+JlSfFSH+ICVSkdNIx+2xO/70avsETNctSUh1c1TUtnoe0n9ttEm3In37Udj5ZbJHp+2Eqn9wQYiQSlFOoAJ6VM28Zx7Y/FRU8Y7xODsJBMOYzunepkQ8rzRk6YsSOR0O7ThtIpWeb3DZOWHTBpRFFgHXQBobTEfkAZ7riPz3+SlXri/ltLiXhRPALvKWsWDMUmGM2+vMbFs5GWylmS6NJx/g3yA9c0x4XhgXtJSzmg6TQWER+Lg+PiqAvL6imZ13MipAPpikb6jJRClMKyApEytxpOCrd15ipYP2DW2bVHXF91HkyvDKPi1M/sYcO/ZB3QrLQp2M8pH98Mjo8W0lF4NIwnUhA2dDQA83URK2bbsBVw/HbB8MRqVKW0KZ3VGukRugmxJRYjAU9OdCglKcVShEYeb02yj1YLQyZ+TLYFJi3sdhPW86jqmzEXaEWYTqSUFwyJJ7AAdxUi8j+BR8D/uutcXiC+k/dHnvD+yfWS58uPZ+X6ByPi97zIi+/FwgAgIp+OiI/edR7vFO+XPOH9k+slz5cf32qu766+uMQlLvH/RVwWhktc4hJPxX1aGH7mrhN4wXi/5Anvn1wveb78+JZyvTc9hktc4hL3J+5TxXCJS1zinsSdLwwi8qcGPftLIvLJu87nrSEiXxaRzw1q+afHYx8QkV8Ukd8YX1+7g7x+TkS+LiKff+KxZ+YlGX9nXOPPishH7kGuL422/xLzfJ7EwL26ru+FFAIRcWd/SGzGbwLfQ1ILfg348F3m9Iwcvwx851se+xvAJ8f3nwT++h3k9YPAR4DPv1NewMeBf03imH4A+OV7kOtPAX/lGc/98LgPZuBD4/7Q9yjPDwIfGd8/AP7ryOdeXde3yfOlXdO7rhi+H/hSRPy3iFiBXyBp2/c9PgH8/Pj+54E//V4nEBH/Afidtzz8vLw+AfyDyPgl4FUR+eB7k+lzc31enGn7EfFbwIm2/22PiPhqRPzq+P4N4CQxcK+u69vk+bx419f0rheG7wL++xN/fyZF+44jgH8jIr8yqOIAvy8GT2R8/b13lt3vjufldV+v80+MEvznnjiO3Ytc3yIxcG+v61vyhJd0Te96YXghivYdx8ci4iPADwE/LiI/eNcJ/T/EfbzO3xJt/9sZz5AYeO5Tn/HYe5bry5ZCeDLuemF41xTt9zoi4rfH168D/4Iswb52KhnH16/fXYa/K56X1727zhHxtYiwiHDgZ3lc2t5prs+SGOAeXtfnSSG8rGt61wvDfwa+V0Q+JCITqRX5qTvO6Rwicj10LhGRa+BPkPTyTwE/Op72o8C/vJsMn4rn5fUp4EdGF/0HgG/GY8r8ncRbzuJvpe3/sIjMIvIhXoC2/xJzeqbEAPfsuj4vz5d6Td+LLuo7dFg/TnZVfxP4ybvO5y25fQ/Zzf014Aun/IDvAP4d8Bvj6wfuILd/QpaLG7kj/Pnn5UWWkn93XOPPAR+9B7n+w5HLZ8eN+8Ennv+TI9dfB37oPczzj5Ml9meBz4w/H79v1/Vt8nxp1/SCfLzEJS7xVNz1UeISl7jEPYzLwnCJS1ziqbgsDJe4xCWeisvCcIlLXOKpuCwMl7jEJZ6Ky8JwiUtc4qm4LAyXuMQlnorLwnCJS1ziqfi/0Cu9P6VaXGcAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2aca269c18>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Let's take a look at the first image\n",
    "# Note, because the image is just a matrix of numbers, we can *1.4 to make it brighter\n",
    "plt.imshow(data.val_ds.denorm(to_np(images))[0]*1.4);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('agriculture', 1.0),\n",
       " ('artisinal_mine', 0.0),\n",
       " ('bare_ground', 0.0),\n",
       " ('blooming', 0.0),\n",
       " ('blow_down', 0.0),\n",
       " ('clear', 1.0),\n",
       " ('cloudy', 0.0),\n",
       " ('conventional_mine', 0.0),\n",
       " ('cultivation', 0.0),\n",
       " ('habitation', 0.0),\n",
       " ('haze', 0.0),\n",
       " ('partly_cloudy', 0.0),\n",
       " ('primary', 1.0),\n",
       " ('road', 0.0),\n",
       " ('selective_logging', 0.0),\n",
       " ('slash_burn', 0.0),\n",
       " ('water', 1.0)]"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# And to see which labels this has\n",
    "list(zip(data.classes, labels[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Start off training on small images\n",
    "image_size=64"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "data = get_data(image_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "31148b41e48344949d8ce54289e7534e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=6), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "# This goes through and resizes all of our images for efficiencies sake?\n",
    "data = data.resize(int(image_size*1.3), \"tmp\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Now we want to find the learning rate\n",
    "# First set up our model\n",
    "learn = ConvLearner.pretrained(f_model, data, metrics=metrics)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d10efd9527314edaa84cd57252fe5ae9",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=1), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss   f2                          \n",
      "    0      0.220537   0.422463   0.789931  \n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7ff03f8582e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learning_rate_finder = learn.lr_find()\n",
    "learn.sched.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Note that the learning rate is decreasing fastest at around 0.2\n",
    "learning_rate = 0.2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2beac4b5ac8c4f17b50484ebc403462d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=7), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss   f2                          \n",
      "    0      0.14736    0.13469    0.879414  \n",
      "    1      0.142267   0.129391   0.885923                    \n",
      "    2      0.138263   0.127576   0.887258                    \n",
      "    3      0.140281   0.126938   0.888762                    \n",
      "    4      0.137914   0.124001   0.890997                    \n",
      "    5      0.131529   0.123408   0.893647                    \n",
      "    6      0.130181   0.12329    0.891949                    \n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.12329047, 0.8919489753927757]"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Do some training for the final layers\n",
    "# We now know that when we train, what we are doing is setting the kernel (image feature filter) values\n",
    "# and the numbers of the weight matrix in the fully connected layer\n",
    "learn.fit(learning_rate, 3, cycle_len=1, cycle_mult=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# The satellite images are quite different from the image-net images that this network was trained on\n",
    "# Therefore, we use slightly higher learning rates than our cats/dogs examples to train the network\n",
    "differential_learning_rates = np.array([learning_rate/9, learning_rate/3, learning_rate])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2da22cfecd9144f2880ebf591ef70933",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=7), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss   f2                          \n",
      "    0      0.120932   0.108382   0.909156  \n",
      "    1      0.115432   0.105932   0.909191                    \n",
      "    2      0.105167   0.10103    0.915758                    \n",
      "    3      0.111772   0.102488   0.91311                     \n",
      "    4      0.105998   0.100651   0.915267                    \n",
      "    5      0.101719   0.09802    0.916691                     \n",
      "    6      0.098998   0.097133   0.917914                     \n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.09713308, 0.91791370766977]"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Unfreeze the previous layers so we can train them with our differential learning rates\n",
    "learn.unfreeze()\n",
    "learn.fit(differential_learning_rates, 3, cycle_len=1, cycle_mult=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Save the weights\n",
    "learn.save(f'{image_size}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2aca256278>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Take a look -- we can see the loss decreasing as we train\n",
    "learn.sched.plot_loss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Now we train again with slightly larger images\n",
    "# Still not 100% sure what the reason for this is -- just simulating different data?\n",
    "image_size = 128"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "edcf55e4065249f0b9f96d90c1f671a1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=13), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss   f2                           \n",
      "    0      0.098714   0.096537   0.917149  \n",
      "    1      0.100112   0.095297   0.919537                     \n",
      "    2      0.095763   0.094022   0.920467                     \n",
      "    3      0.095169   0.093924   0.919777                     \n",
      "    4      0.098707   0.09404    0.921207                     \n",
      "    5      0.097631   0.093178   0.921026                     \n",
      "    6      0.095902   0.093244   0.920144                     \n",
      "    7      0.093888   0.092771   0.921678                     \n",
      "    8      0.092554   0.092677   0.921674                     \n",
      "    9      0.09418    0.091928   0.922913                     \n",
      "    10     0.092501   0.091704   0.922293                     \n",
      "    11     0.091666   0.091564   0.922321                     \n",
      "    12     0.090078   0.091956   0.921488                     \n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.09195557, 0.9214877960864891]"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data = get_data(image_size)\n",
    "learn.set_data(data)\n",
    "learn.freeze()\n",
    "learn.fit(learning_rate, 3, cycle_len=1, cycle_mult=3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "54cb7bc1cf5044f2b38a4b0a6f3fb2e6",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=13), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss   f2                           \n",
      "    0      0.092793   0.087805   0.927363  \n",
      "    1      0.09623    0.088546   0.925771                     \n",
      "    2      0.091078   0.086233   0.92801                      \n",
      "    3      0.087241   0.085489   0.9284                       \n",
      "    4      0.095144   0.089547   0.926124                     \n",
      "    5      0.090785   0.087691   0.926577                     \n",
      "    6      0.087424   0.086907   0.926908                     \n",
      "    7      0.086251   0.086164   0.926724                     \n",
      "    8      0.082321   0.085769   0.92639                      \n",
      "    9      0.080621   0.084496   0.929468                     \n",
      "    10     0.080943   0.084145   0.929588                     \n",
      "    11     0.075996   0.084212   0.929369                     \n",
      "    12     0.077381   0.084321   0.9293                       \n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.08432135, 0.9292995924747179]"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Now tweak the initial layers again\n",
    "learn.unfreeze()\n",
    "learn.fit(differential_learning_rates, 3, cycle_len=1, cycle_mult=3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "learn.save(f'{image_size}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Now do it AGAIN, with even bigger image sizes\n",
    "image_size = 256"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2de642c9cf5848e383a4ec40af258159",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=7), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss   f2                           \n",
      "    0      0.091138   0.090089   0.924436  \n",
      "    1      0.089965   0.089359   0.925116                     \n",
      "    2      0.08823    0.089175   0.926036                     \n",
      "    3      0.088119   0.088418   0.926243                     \n",
      "    4      0.085408   0.088381   0.927063                     \n",
      "    5      0.085668   0.088337   0.926008                     \n",
      "    6      0.086202   0.087519   0.927309                     \n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.0875194, 0.9273087637207408]"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data = get_data(image_size)\n",
    "learn.set_data(data)\n",
    "learn.freeze()\n",
    "learn.fit(learning_rate, 3, cycle_len=1, cycle_mult=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4ab85ff4d87247f28d6ebc2922ef7309",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=21), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss   f2                           \n",
      "    0      0.085767   0.083354   0.932566  \n",
      "    1      0.08356    0.081909   0.93292                      \n",
      "    2      0.078167   0.080999   0.932661                     \n",
      "    3      0.086357   0.083911   0.930388                     \n",
      "    4      0.083617   0.083095   0.93072                      \n",
      "    5      0.081474   0.083832   0.927965                     \n",
      "    6      0.076187   0.08217    0.93121                      \n",
      "    7      0.075701   0.081892   0.932569                     \n",
      " 34%|███▍      | 174/506 [03:31<06:43,  1.22s/it, loss=0.0751]"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                   Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-36-b7dd3dc6cefc>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0mlearn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0munfreeze\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0mlearn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdifferential_learning_rates\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m3\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcycle_len\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m3\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcycle_mult\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m~/fastai/courses/dl1/fastai/learner.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, lrs, n_cycle, wds, **kwargs)\u001b[0m\n\u001b[1;32m    207\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msched\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    208\u001b[0m         \u001b[0mlayer_opt\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget_layer_opt\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlrs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mwds\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 209\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit_gen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmodel\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlayer_opt\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mn_cycle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    210\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    211\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mwarm_up\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlr\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mwds\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/fastai/courses/dl1/fastai/learner.py\u001b[0m in \u001b[0;36mfit_gen\u001b[0;34m(self, model, data, layer_opt, n_cycle, cycle_len, cycle_mult, cycle_save_name, use_clr, metrics, callbacks, use_wd_sched, norm_wds, wds_sched_mult, **kwargs)\u001b[0m\n\u001b[1;32m    154\u001b[0m         \u001b[0mn_epoch\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msum_geom\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mcycle_len\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mcycle_len\u001b[0m \u001b[0;32melse\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcycle_mult\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mn_cycle\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    155\u001b[0m         return fit(model, data, n_epoch, layer_opt.opt, self.crit,\n\u001b[0;32m--> 156\u001b[0;31m             metrics=metrics, callbacks=callbacks, reg_fn=self.reg_fn, clip=self.clip, **kwargs)\n\u001b[0m\u001b[1;32m    157\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    158\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mget_layer_groups\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmodels\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget_layer_groups\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/fastai/courses/dl1/fastai/model.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(model, data, epochs, opt, crit, metrics, callbacks, **kwargs)\u001b[0m\n\u001b[1;32m     94\u001b[0m             \u001b[0mbatch_num\u001b[0m \u001b[0;34m+=\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     95\u001b[0m             \u001b[0;32mfor\u001b[0m \u001b[0mcb\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mcallbacks\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mcb\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mon_batch_begin\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 96\u001b[0;31m             \u001b[0mloss\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mstepper\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstep\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mV\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mV\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0my\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     97\u001b[0m             \u001b[0mavg_loss\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mavg_loss\u001b[0m \u001b[0;34m*\u001b[0m \u001b[0mavg_mom\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mloss\u001b[0m \u001b[0;34m*\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0mavg_mom\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     98\u001b[0m             \u001b[0mdebias_loss\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mavg_loss\u001b[0m \u001b[0;34m/\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0;36m1\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0mavg_mom\u001b[0m\u001b[0;34m**\u001b[0m\u001b[0mbatch_num\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/fastai/courses/dl1/fastai/model.py\u001b[0m in \u001b[0;36mstep\u001b[0;34m(self, xs, y)\u001b[0m\n\u001b[1;32m     47\u001b[0m             \u001b[0mnn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mutils\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mclip_grad_norm\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrainable_params_\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mm\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mclip\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     48\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mopt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstep\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 49\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mraw_loss\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     50\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     51\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mevaluate\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mxs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "# (Interrupted the kernel here as it started running for 21 cycles for some reason. fasi.ai bug?)\n",
    "learn.unfreeze()\n",
    "learn.fit(differential_learning_rates, 3, cycle_len=3, cycle_mult=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "learn.save(f'{image_size}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                                                              \r"
     ]
    }
   ],
   "source": [
    "# Finallly, get our predictions using test time augmentation\n",
    "# Reminder: TTA = Test Time Augmentation, applying transforms to the test data and taking an average\n",
    "multi_label_classification_predictions,target_values = learn.TTA()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "predictions = np.mean(multi_label_classification_predictions, 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.76759, 0.0016 , 0.02022, 0.00089, 0.00083, 0.99982, 0.00006, 0.00021, 0.43698, 0.02853, 0.00011,\n",
       "        0.00003, 0.99958, 0.22834, 0.00342, 0.01352, 0.73973],\n",
       "       [0.18095, 0.00006, 0.00077, 0.00014, 0.00077, 0.00035, 0.00022, 0.00009, 0.02922, 0.00091, 0.00005,\n",
       "        0.99785, 0.99984, 0.01146, 0.0006 , 0.001  , 0.05649],\n",
       "       [0.99405, 0.00013, 0.01334, 0.00012, 0.00006, 0.99976, 0.00001, 0.00005, 0.09454, 0.1095 , 0.00048,\n",
       "        0.00003, 0.99309, 0.9827 , 0.00023, 0.00075, 0.07534],\n",
       "       [0.01294, 0.9995 , 0.04651, 0.00125, 0.0003 , 0.99966, 0.00012, 0.03368, 0.00482, 0.02961, 0.00009,\n",
       "        0.00026, 0.96083, 0.30619, 0.00502, 0.00017, 0.99083],\n",
       "       [0.99758, 0.00004, 0.00231, 0.00024, 0.00009, 0.99609, 0.00003, 0.00016, 0.11167, 0.79664, 0.00222,\n",
       "        0.00135, 0.97242, 0.99619, 0.00044, 0.00086, 0.0222 ]], dtype=float32)"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Predictions contains the likelihood for each of our 17 classes\n",
    "predictions[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1., 0., 0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 1., 0., 0., 0., 1.],\n",
       "       [0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 1., 1., 0., 0., 0., 0.],\n",
       "       [1., 0., 0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 1., 1., 0., 0., 0.],\n",
       "       [0., 1., 0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 1., 0., 0., 0., 1.],\n",
       "       [1., 0., 0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 1., 1., 0., 0., 0.]], dtype=float32)"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Target values contains the actual correct classes for each image\n",
    "target_values[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9308049768554749"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "f2(predictions, target_values)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Note on structured/unstructured data\n",
    "In the next lesson, we will look at processing structured and unstructed data.\n",
    "\n",
    "These are not 100% standardized terms, but generally we think of it as:\n",
    "\n",
    "1. __Unstructured data__: Every \"piece\" of the data objects we're processing is the same. For example, in an image, everything is a pixel.\n",
    "\n",
    "2. __Structured data__: The parts of the data objects are different. For example, if we are processing sales reports, one column could contain profit, another weather data, another quarterly dates, etc."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}