4_char-level_simpleRNN.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Preparing the Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import glob\n",
    "import string"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['./data/data-for-char-rnn/names/Scottish.txt', './data/data-for-char-rnn/names/German.txt', './data/data-for-char-rnn/names/Italian.txt', './data/data-for-char-rnn/names/Dutch.txt', './data/data-for-char-rnn/names/English.txt', './data/data-for-char-rnn/names/Portuguese.txt', './data/data-for-char-rnn/names/Chinese.txt', './data/data-for-char-rnn/names/Japanese.txt', './data/data-for-char-rnn/names/Greek.txt', './data/data-for-char-rnn/names/Vietnamese.txt', './data/data-for-char-rnn/names/Korean.txt', './data/data-for-char-rnn/names/Polish.txt', './data/data-for-char-rnn/names/Czech.txt', './data/data-for-char-rnn/names/Arabic.txt', './data/data-for-char-rnn/names/Russian.txt', './data/data-for-char-rnn/names/French.txt', './data/data-for-char-rnn/names/Irish.txt', './data/data-for-char-rnn/names/Spanish.txt']\n",
      "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ .,;'\n"
     ]
    }
   ],
   "source": [
    "path = \"./data/data-for-char-rnn/names/*.txt\"\n",
    "file_names = glob.glob(path) # find file names: ./[Language].txt\n",
    "print(file_names)\n",
    "all_letters = string.ascii_letters + \" .,;'\"\n",
    "num_letters = len(all_letters)\n",
    "print(all_letters)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import unicodedata"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def unicode2ascii(s):\n",
    "    \"\"\"Turn a Unicode string to plain ASCII\n",
    "    http://stackoverflow.com/a/518232/2809427\"\"\"\n",
    "    return ''.join(\n",
    "        c for c in unicodedata.normalize('NFD', s)\n",
    "        if unicodedata.category(c) != 'Mn'\n",
    "        and c in all_letters\n",
    "    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Slusarski\n"
     ]
    }
   ],
   "source": [
    "print(unicode2ascii('Ślusàrski'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def _load_name_data(file_name):\n",
    "    data = []\n",
    "    with open(file_name, 'r') as f:\n",
    "        for line in f:\n",
    "            data.append(unicode2ascii(line.strip()))\n",
    "    return data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['Ang', 'AuYong', 'Bai', 'Ban', 'Bao']\n"
     ]
    }
   ],
   "source": [
    "file_name = './data/data-for-char-rnn/names/Chinese.txt'\n",
    "data = _load_name_data(file_name)\n",
    "print(data[:5])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def load_data(file_names):\n",
    "    data = dict()\n",
    "    all_categories = []\n",
    "    for f in file_names:\n",
    "        category = f.split('/')[-1].split('.')[0]\n",
    "        all_categories.append(category)\n",
    "        data[category] = _load_name_data(f)\n",
    "    return all_categories, data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "18\n",
      "['Scottish', 'German', 'Italian', 'Dutch', 'English', 'Portuguese', 'Chinese', 'Japanese', 'Greek', 'Vietnamese', 'Korean', 'Polish', 'Czech', 'Arabic', 'Russian', 'French', 'Irish', 'Spanish']\n",
      "['Ang', 'AuYong', 'Bai', 'Ban', 'Bao']\n"
     ]
    }
   ],
   "source": [
    "all_categories, data = load_data(file_names)\n",
    "num_categories = len(all_categories)\n",
    "print(num_categories)\n",
    "print(all_categories)\n",
    "print(data['Chinese'][:5])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0\n"
     ]
    }
   ],
   "source": [
    "def letter2index(letter):\n",
    "    return all_letters.find(letter)\n",
    "\n",
    "print(letter2index('a'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "Columns 0 to 12 \n",
      "    1     0     0     0     0     0     0     0     0     0     0     0     0\n",
      "\n",
      "Columns 13 to 25 \n",
      "    0     0     0     0     0     0     0     0     0     0     0     0     0\n",
      "\n",
      "Columns 26 to 38 \n",
      "    0     0     0     0     0     0     0     0     0     0     0     0     0\n",
      "\n",
      "Columns 39 to 51 \n",
      "    0     0     0     0     0     0     0     0     0     0     0     0     0\n",
      "\n",
      "Columns 52 to 56 \n",
      "    0     0     0     0     0\n",
      "[torch.FloatTensor of size 1x57]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "\n",
    "def letter2tensor(letter):\n",
    "    tensor = torch.zeros(1, num_letters)\n",
    "    tensor[0][letter2index(letter)] = 1\n",
    "    return tensor\n",
    "\n",
    "print(letter2tensor('a'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([4, 1, 57])\n"
     ]
    }
   ],
   "source": [
    "def name2tensor(name):\n",
    "    # [name_len, batch_size, letters_size)\n",
    "    tensor = torch.zeros(len(name), 1, num_letters)\n",
    "    for i, letter in enumerate(name):\n",
    "        tensor[i] = letter2tensor(letter)\n",
    "    return tensor\n",
    "\n",
    "print(name2tensor('hang').size())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Define the Network: Simple RNN\n",
    "![SimpleRNN](./pictures/simpleRNN.png)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import torch.nn as nn\n",
    "from torch.autograd import Variable"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class RNN(nn.Module):\n",
    "    def __init__(self, input_size, hidden_size, output_size):\n",
    "        super(RNN, self).__init__()\n",
    "        self.hidden_size = hidden_size\n",
    "        \n",
    "        self.i2o= nn.Linear(input_size + hidden_size, output_size)\n",
    "        self.i2h = nn.Linear(input_size + hidden_size, hidden_size)\n",
    "        self.softmax = nn.LogSoftmax()\n",
    "    \n",
    "    def forward(self, x, hidden):\n",
    "        combined = torch.cat((x, hidden), 1)\n",
    "        hidden = self.i2h(combined)\n",
    "        output = self.i2o(combined)\n",
    "        return self.softmax(output), hidden\n",
    "    \n",
    "    def init_hidden(self):\n",
    "        return Variable(torch.zeros(1, self.hidden_size))\n",
    "\n",
    "hidden_size = 128\n",
    "rnn = RNN(num_letters, hidden_size, num_category)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## Training"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Preparing for Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def get_category_from_output(output):\n",
    "    top_n, top_i = output.data.topk(1)\n",
    "    category_index = top_i[0][0]\n",
    "    return all_categories[category_index], category_index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Variable containing:\n",
      "\n",
      "Columns 0 to 9 \n",
      "-2.9093 -2.8742 -2.8419 -2.9471 -2.8272 -2.8358 -2.8696 -2.8296 -2.8707 -2.9160\n",
      "\n",
      "Columns 10 to 17 \n",
      "-2.9064 -2.9008 -2.8728 -2.9201 -2.9174 -2.8841 -2.9456 -2.9735\n",
      "[torch.FloatTensor of size 1x18]\n",
      "\n",
      "('English', 4)\n"
     ]
    }
   ],
   "source": [
    "x = Variable(name2tensor('hang'))\n",
    "hidden = Variable(torch.zeros(1, hidden_size))\n",
    "\n",
    "output, next_hidden = rnn(x[0], hidden)\n",
    "print(output)\n",
    "print(get_category_from_output(output))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import random"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def random_choose(list_data):\n",
    "    return list_data[random.randint(0, len(list_data)-1)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Russian Alhimenko torch.Size([1]) torch.Size([9, 1, 57])\n",
      "French Simon torch.Size([1]) torch.Size([5, 1, 57])\n",
      "Czech Kusak torch.Size([1]) torch.Size([5, 1, 57])\n",
      "Korean Hwang torch.Size([1]) torch.Size([5, 1, 57])\n",
      "Italian Passerini torch.Size([1]) torch.Size([9, 1, 57])\n"
     ]
    }
   ],
   "source": [
    "def random_training_example(all_categories, data):\n",
    "    category = random_choose(all_categories)\n",
    "    name = random_choose(data[category])\n",
    "    category_tensor = Variable(torch.LongTensor([all_categories.index(category)]))\n",
    "    name_tensor = Variable(name2tensor(name))\n",
    "    return category, name, category_tensor, name_tensor\n",
    "\n",
    "for _ in range(5):\n",
    "    tmp_category, tmp_name, tmp_category_tensor, tmp_name_tensor = random_training_example(all_categories, data)\n",
    "    print(tmp_category, tmp_name, tmp_category_tensor.size(), tmp_name_tensor.size())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "criterion = nn.NLLLoss()\n",
    "learning_rate = 0.005"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def train(category_tensor, name_tensor):\n",
    "    rnn.zero_grad()\n",
    "    hidden = rnn.init_hidden()\n",
    "    for i in range(name_tensor.size()[0]):  # each char in name:\n",
    "        output, hidden = rnn(name_tensor[i], hidden)\n",
    "    loss = criterion(output, category_tensor)\n",
    "    loss.backward()\n",
    "    \n",
    "    # Add parameters' gradients to their values, multiplied by learning rate\n",
    "    for p in rnn.parameters():\n",
    "        p.data.add_(-learning_rate, p.grad.data)\n",
    "        \n",
    "    return output, loss.data[0]\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5000 5% (0m 21s) 2.8817 Starek / Czech ✗ (Polish)\n",
      "10000 10% (0m 44s) 1.2936 Achteren / Dutch ✓\n",
      "15000 15% (1m 22s) 0.5757 Hatoyama / Japanese ✓\n",
      "20000 20% (2m 6s) 1.1074 Ankeren / Dutch ✓\n",
      "25000 25% (2m 31s) 1.8642 Gauk / Korean ✗ (Chinese)\n",
      "30000 30% (3m 3s) 1.6130 Samaha / Japanese ✗ (Arabic)\n",
      "35000 35% (3m 32s) 2.1473 Twine / Chinese ✗ (English)\n",
      "40000 40% (3m 58s) 1.0766 Zhiznevsky / Czech ✗ (Russian)\n",
      "45000 45% (4m 30s) 4.1466 Naggia / Arabic ✗ (Italian)\n",
      "50000 50% (5m 0s) 0.4347 Ghanem / Arabic ✓\n",
      "55000 55% (5m 23s) 5.0009 Fabian / Irish ✗ (Polish)\n",
      "60000 60% (5m 51s) 0.4239 Yim / Korean ✓\n",
      "65000 65% (6m 15s) 0.5626 Kuipers / Dutch ✓\n",
      "70000 70% (6m 40s) 0.0028 Giannakopoulos / Greek ✓\n",
      "75000 75% (7m 1s) 2.5402 Granger / French ✗ (English)\n",
      "80000 80% (7m 22s) 1.8711 Fonseca / Spanish ✗ (Portuguese)\n",
      "85000 85% (7m 44s) 0.4824 Ganim / Arabic ✓\n",
      "90000 90% (8m 5s) 0.0210 Mcmillan / Scottish ✓\n",
      "95000 95% (8m 29s) 1.0116 Gouveia / Portuguese ✓\n",
      "100000 100% (8m 50s) 0.3881 Benvenuti / Italian ✓\n"
     ]
    }
   ],
   "source": [
    "import time\n",
    "import math\n",
    "\n",
    "n_iters = 100000\n",
    "print_every = 5000\n",
    "plot_every = 1000\n",
    "\n",
    "\n",
    "\n",
    "# Keep track of losses for plotting\n",
    "current_loss = 0\n",
    "all_losses = []\n",
    "\n",
    "def time_since(since):\n",
    "    now = time.time()\n",
    "    s = now - since\n",
    "    m = math.floor(s / 60)\n",
    "    s -= m * 60\n",
    "    return '%dm %ds' % (m, s)\n",
    "\n",
    "start = time.time()\n",
    "\n",
    "for iter in range(1, n_iters + 1):\n",
    "    category, line, category_tensor, name_tensor = random_training_example(all_categories, data)\n",
    "    output, loss = train(category_tensor, name_tensor)\n",
    "    current_loss += loss\n",
    "\n",
    "    # Print iter number, loss, name and guess\n",
    "    if iter % print_every == 0:\n",
    "        guess, guess_i = get_category_from_output(output)\n",
    "        correct = '✓' if guess == category else '✗ (%s)' % category\n",
    "        print('%d %d%% (%s) %.4f %s / %s %s' % (iter, iter / n_iters * 100, \\\n",
    "                                                time_since(start), loss, line, guess, correct))\n",
    "\n",
    "    # Add current loss avg to list of losses\n",
    "    if iter % plot_every == 0:\n",
    "        all_losses.append(current_loss / plot_every)\n",
    "        current_loss = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f891541b5c0>]"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD8CAYAAACMwORRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n5YE03JVSygNpuCullAf6/wEXhalBhHj7gAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f8915498d68>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.ticker as ticker\n",
    "\n",
    "plt.figure()\n",
    "plt.plot(all_losses)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evaluating the Results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAVwAAAEwCAYAAAD7IMkNAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzsnXecHVX5/9+f3VQIhBJEpAWQgIAQIPRepAgoKlUsgAoo\nVX8q2BAriJXeBb6KgCACIhAwtEiRkEJIQjUECCBFigmElN3n98c5l529O3PvzObu3tnN83695rV3\nzjxz5szde5975jlPkZnhOI7j9DwtzR6A4zjOkoIrXMdxnF7CFa7jOE4v4QrXcRynl3CF6ziO00u4\nwnUcx+klXOE6juP0Eq5wHcdxeglXuI7jOL2EK1ynXyNpZUmXSbot7m8g6UvNHpezZOIK1+nvXAGM\nBT4U958CTmraaJwlGle4Tm4kLdXsMXSDEWb2Z6AdwMwWAW3NHZKzpOIK16mLpG0lzQCeiPubSDq/\nycPKyzuSVgQMQNLWwNvNHZKzpDKg2QNw+gS/BfYEbgYws0cl7djcIeXmG4RxryPpfmAl4IDmDslZ\nUnGF6+TCzF6QlGzqE4/lZjZJ0k7AeoCAJ81sYZOH5SyhuEnBycMLkrYFTNJASd8EHm/2oPIg6UBg\nqJlNB/YHrpW0WZOH5SyhuMJ18nAMcCywKvAiMDru9wV+YGZzJG0P7AZcBlzQ5DE5Syjyig9Of0bS\nZDPbVNLpwGNm9qdKW7PH5ix5+AzXqYukMyUtG80J4yS9JulzzR5XTl6UdBFwMHCrpMH4595pEv7B\nc/Kwh5n9D9gXmAV8GPhWU0eUn4MIgQ97mtlbwAr0nbE7/QxXuE4eKt4s+wDXmVmf8WM1s3eBV4Ht\nY9Mi4OnmjchZknGF6+ThFklPAJsD4yStBLzX5DHlQtIPgZOB78SmgcAfmzciZ0nGF82cXEhaAXjb\nzNokLQ0sY2b/aUC/SwH/D1jDzL4iaV1gPTO7ZXH7jv1PATYFJlUWyiRNNbONG9G/4xTBZ7hOXaJS\n/Bod7lQfAsY0qPvLgfnANnH/ReCnDeobYIGFWUUltHfpBvbtOIVwhevk4XJgAbBt3G+kUlzHzM4E\nFsL7NlfVPqUQf45eCstJ+grwD+CSBvbvOLnx0F4nD+uY2cGSDoWgFFUV57sYLJA0lI4Z6DqEGW9D\nMLNfSfoY8D9CeO+pZnZno/p3nCK4wnXy0JNK8YfA7cDqkq4CtgMOb1DfAJjZnZL+Rfy8S1rBzN5o\n5DUcJw++aObUJc4Qvw9sANxBVIpmdk+D+l8R2JpgSnjIzF5vRL+x76OBHxG8KtrjNczM1m7UNRwn\nL65wu4GkVmBlEk8IZvZ8DflVgTWr5O/ryTE2mqJKMeYuWNfMLo9uZMPM7NkUue2AKWb2Toxe2ww4\ny8yea9C4nwa2aaQSd5zu4gq3IJKOJzwGv0KsIkCYMaW6GUn6BSGsdAYdKQ3NzD7R02NtFFm5b7N+\nNKLv6xiCe9coSR8iBExslyI7FdgE2JiwOHcZcJCZ7dSgsd8OfDouxi1RlOmHPv7ofgUYWTWeI5sx\nnmbhNtzinEhQJP/NKb9/lG/YQlATSIbCDgG2BCYCu2bIf4ro+wpgZi9JWiZDdpGZmaRPAueZ2WUN\nLvL4HeCBaMN9/39gZic08BqlI+uHHmjWk9VNwHiCl0ifyKXcE7jCLc4LFCvRMpMQ3dRnFa6Z7Zfc\nl7Q68LsapyyISjSP7+scSd8BPg/sIKmF8H41iouAu4DH6HgiWRIo2w/9UmZ2crMH0Wxc4eZE0jfi\ny5nAPZL+TucZ028yTn0XmCJpHCWZYcWMWZ+h6+Pdj3N2MRv4SI3j1b6vR5Lt+3ow8FngSDP7j6Q1\ngF/mHEceBprZN+qL9TvK9kN/i6SPm9mtzR5IM3GFm5/KI/HzcRsUt3rcHLcycRNhlj6RHF9ISecQ\nXcIIwTKjieaCNIr4vkYl+xdg3dj0OvDXnPeRh9skHQX8jc4/eDXdwmKFi5F0/kH6vwaOq6cpxQ+9\npDmEz46A70qaTwhyqXiLLNub42k2vmi2GMTH32ExdWGfQdI0M9uogPwXE7uLgFlmdn8N+aWB92Le\nhfUISve2tFpicQZ8FLCCma0TcylcaGa75R1fnbF38YygjluYpD8A6wBT6LzQ2WfsvlX/s/cxsytr\nnNPXf2RKjyvcgkj6E6HkTBswAViW4MaU+hgcFcjpBB/WIZX2ZvqBSroYOMfMHuuh/icCOwDLA/8E\nHiHYdQ9LkZ1CWIT7VyK5zGNm9tGeGFseJD0ObGBL0Jejp39kMtz/flfLnbI/4iaF4mxgZv+TdBhw\nG3AK4dE8y+54OcGN7LfALsARNCmHhaTHCI93A4AjJM0kPG5WHu+yXNsq5yV5m6BIf5risaEY/vsl\n4AIzOzMq1jTmm9mCSqSwpAEp11osJG1E1x+8WjO3acAHgZcbOY7uEp8Y5plZe9xvAYbUcnXrxg/9\nGHr2R+YCYBNJmxCyw10K/AFoiPtfX8EVbnEGShpIWAU+18wWVlbjMxhqZuMkKTrznxZngKfWukgP\n+VDu283zbiPMev4U9w8BlgL+A1wB7FclL0nbAIcBFRev1oy+75X0XWBotPt+jWBvbQjRJ3hnguK5\nFdibMOuupXBHADMkPUxn+2ezfKfHAbsDc+P+UoSIv20zzyj+Q9/TPzJJ979ze8D9r0/gCrc4FxHK\nzDwK3CdpTcLiUBbz44zkaUnHETJtDat1gZ7yoaxEb8VcCLPNbL6knQlBB7UU0O5mliwt/pikSWa2\nmdJrm51E8H/9q5lNl7Q2cHdG36cQlPJjwNEEpXhprfuQtLeZ3VbVdoyZXZgifgAhsGKymR0haWXq\nJyA/rc7xxaZgtOIQM5ubkJsbU2bWItcPvaS/ET5by9CzPzIV97/PATv2gPtfn8AVbkHM7Gzg7ETT\nc5J2qXHKiYQZyQnATwjBAqkLGgkK+VBK+jTwC+ADBPNAvRXgvwBjJH0YuJjgtfAn4OMZ8q2StjSz\nh+P1tqBjxrqoWtjM7iXMXJeK+zMJ99+F+Jh8CcVSJv5A0nwzuyuO59uEWVyawp1nZu2SFklallBu\nZ/Vancfx56aom11WtCLhhy+NdyRtZmaT4vmbA/PqDCvvD/2v6vTTKCruf1/qIfe/PoEvmuVE0ufM\n7I8Jf9xO1PDD7c61bgMOTM5q6sg/A+xnZo/nlK/MTr9NUEjnqEbp8Khgf0/4woowo/8yMB3Yx8z+\nXCW/DSFEd5iZrRHtdkeb2ddS+t6OMKOsmE/qJpeRNAK4hRABtxewPnComS1IkT0f+C7BDPL/CI/l\nU8zsiBr9bw2cQ/A1HkT4cXkn6wdMIXy44mb3fhSVmf06Q/4ZYKu80Yrx/b8GeInw/nwQONjMJtY5\n53FgOcIP/XDgTDN7KEN+LeBlM3sv7g8FVjazWXnG6OTDZ7j5qURLpYWoZv5qSRoDfI+u9thaJV6K\n+lC+klfZRhYq5Lb9Ah3218zHOzObAHxU0vC4n4y0+3PKKb8D9iT6H5vZo8rIx0BQzF+nSlnVwsxe\nl/QJQpjoROCArMWehJK/MCrGZc1sap1LnEtQ0NcRFpO+AIyqIb+ame2VZ+yRQtGKZjZB0voE9zqA\nJ9Nc7KrPiS/nEuy39biOzjbhtti2Rd5xpiHpn2a2fcIf9/1DLIF+uK5wc2JmF8WX/6j2QY2ztCyu\nIszEioSW5gqWiKYEgEckXQvcSGcFfUPGqUcQXNt+ZmbPxtnNH2pcp9ruV+k/MzLNzF5Q5xzlWcr0\n7Wp7bI1xVH9pBwFrAwcEU2XmDLTTAqSkHestQJrZM5JazawNuFzSZDoKUVbzgKSP1nOzU8FoRUm7\nmtldif9zhVGSUv+/kn5nZiclbLPV95Vlkx2QfEKIniN5AntqYmbbx79ZuTSWKFzhFuccgg9hvbYK\nr5lZoUizWs7pVSS9A94F9kh2A6QqXDObQcKmaiFt4i9qXOedxOshBG+HWjPqFxSc6C16dJxYQ/5u\nSb+MY00qny6RbN350nZzAfLdqGymSDqTsHLfZYW/G252RaMVdyLkgaj2AqncQ9r/t/LDWdQ2+5qk\nT1Q+q9GboKEpLQsuFPZL3Iabk2iX3JawAv/bxKFlgU+Z2SYZ5+0GHEpw7ckz++yxYAlJfzazg5Tu\nV1vPzJHsZzAw1sx2zjg+AjiL4MokggvTiWk2S0lp3gtmZlmZyFCYOh8GrGVmP1FIprNKZVGvSvZJ\nYOO8C5DxnDUJC1qDCOaO4cD5ZvZMilwm1qCcvouLpOWB1WuZUqLnylWEAqEimD2+UH3PizGGQmlN\n+ys+w83PIMKi0QA623H/R3A9yuIIwqLOQDqvSGcqXAr6UEq6kqDQ3or7ywO/tq65Rk+Mf7vrj1th\nKWC1rIMWkn13iSrLkK3l4ZHF+YT3clfCgtBc4DzS7Y2Fk7iY2XNx0WgVM/tRLTko7mYn6U7Comjy\n/3WNme2ZIb8i4fOwPeGz80/gx7UW3STdA3yC8HmdCLwq6X7LSORjZv8GtpY0LO7nWrAtQNG0pv0S\nV7g5Sbg6zbNQZfZ9JB0IPJ1x6hZmtl7GsSyKBktsXPnyxrG+KamLx4GZvRz/Fpp5Vc2IW4GVCIou\nS75QsmlJ+wAb0nk2Xytz2VbRy2JylH2zhr2xcBIXSfsRHskHAWtJGk1QcFn2z6Judiul/L8+kDUe\ngofCfQTXMwg/ZtcSniCyGG4hIvLLwP+Z2Q8Vkr2nUu3alsdOX5CiaU37Ja5wi3MIcGZV23cIK7pp\nPCBpg2g3zUvRYIkWScub2ZsAklYg5X+bsuj0/iFqrxgnZ8SLCF4RXfxvE+RONi3pQsKMeRdCwMMB\nQBfTQBULoz2wkm93JbIXJNMWIOvZ0U4j5He4B8DMpsSFxSzazWxRXNw6p+JmV0O+TdIaFftlNE3U\nGtMqZpb8gfuppIPr3MMASasABxG8ZOpRNINc0QoORdOa9ktc4eZE0t6EGcuqkpKBD8uS4vyfYGvC\nDOtZcuQtiBQNlvg18KCkitI/EPh5tdBirBT/1Mw+n2yQ9IfqtgRFkk1va2YbS5pqZj+S9GtCKHEt\nziakcFxZ0s8ISvr7GbLLmdlZVWM/MUO2wkIze7vKy6KWQizkZkdQgP+UdC/h87ADIWNaFndIOoQO\nF7wDgLE15AF+HGX+Gd3K1ib7KQyKu7YVreBQNK1pv8QXzXKi4Lw/mvBBTj7azwHurswuU85LXVhp\n9IKKpA3oKHlzV60ZdZwBVzMny7dTMVAisT8AmGpmG2TI/xR4wHIkm5b0LzPbStJDwKeB/wLTzezD\ndc5bH9iNoLDGWYYfcvXYY1tmkEc8fhlhkfMUwmP2CYRE5sdkyG9AcLN70MyujrPhg8ws0/MjLixu\nHXdTi3Kqcy7ZpemYxbcAc2s8kRRGBTPISZpiZqMbdf0lBVe4BYluTqLDET7TCT0+9k43s/Vz9t0t\nH8q02WatGaikWYTw1jfjvSxHSETzCvAVixFMCrHv3wWGEmyhRPkFwMVmluqXGhXF0oQZfc1k05J+\nQHCr242w8GXAJWZWL7lPzarAccb5WcJC0/jEqcsQTACZ+XYVQpK/R4eb3VjCLP+9WmPKS8LLYm0z\n+7FCmOsH07wsFuMaZwI/JYQA305YyPu6maXmkZA0A/gwkOtJrMiPapRfCfg2XW31md4o/RFXuAWR\ntBNhBXoW4UO5OvBFy65gexNwfB5/Q0mbm9nEeI0uWEaMf8oMtBV4rMYM9BLgejMbG/f3IMzkLifk\n9t0qIdsCXFrDNtcw4sLNEOscyZYmV7cqcHyyWIvgXndK4vQ5hNl5qhkovne/MLNv5hhvt9zsJF1A\n9LIws49EL4U7zCwzqivKrEtnZZXpS1yZgUr6FMEG/w3gPst2X8z1JJYy685VwUHSHYSFvm8Snga+\nSPBRX7LqnJmZbwU2wqLCeon9UcDEGvL3Eb7k4+hYwLm5QWP5Tux7EcE9bU7c/gucXuO8x1Lapsa/\nU/LIZ/S7fvy7WdqWcc4QgjK4gbDa/3WC0q11nSmEL/jk6vFnyK9JyHgGYba+TJ3+H8p5v6sk+u+y\n1ThvUvybHP+jNeS/TIhUfJOQdW0ewWxUa2zT4t9Lgb3qXSMe3x44Ir5eieDn3LDvTfX/CZjQqP77\nyuaLZsUZaGZPVnbM7KloZsjiB3k7zpopJa61cdX+6cDpkk63jMf7DF6WdDLB3QhCJNYrcXaXtto/\nSdIW1hGfn8U3CIs/aUlbjPSy6v9H+JE4J+5/lhAtdWCN6+SuCqxECR9CRYPVCFnFapXwmSzpZoLn\nyftRdlYVrGLddLOjmJcFhEXULQg/BLtE+3WXRdEqbpH0BEE5fzVeI9MkknxqIDzpDCSksUwNW1fx\nCg4Vs9vLCm6ALxH+J0sUblIoiKTfE74cFVvYYUCr1Xjkjo9r65rZP6J9sNXM5mTIZZL1xVZGYhjL\nNnOMoMORHuB+4EcEt6A1rGtE1RME+95zBAWUx9MiF5JmWJXpI62t6vg3CY/XHyOYDI4E/mRm56TI\nFi7hI+nylGar/h+nuNmJjsdts+zH68MIP3KbAVcSvSzMLNW1UNIEM9si3stWFgIsppvZhln3EM9b\ngZCroi1+7pY1s/9kyE4BNiXMvivv09Ss/7GCT+8mBNvwFYSZ9EFmlmoOk7QvwZa+OuHHdVngR1Yw\n7L2v4zPc4nwVOJaOXATjCZFPqaTMsFYlY4bVjZlShW8lXg8hKJiJpM8osbAifnxGX2mhnKkRULVQ\n/oKEkyRtbTFtoKStCKV7MrECVYHpRgkfq5G6sUquW252ZnaVQiBLxctif6ud7W22pOUIyYnulPQm\n4ccvE0lfSLxOHsqKgMv91BApVMHBzG6JL98m+FwvkfgMtyBKVKSN+63AYMuoL9XNGVahfKwp569O\neLz7TMbxUYTFi5F0Voi18hdsQvAXBRhvZo/WkM1dkFChYON6BB9NgDWAJwl26S6z6Ph+/8NyhgTH\n1fq3CD6yxxNK+Mwws8xgAEmrEd7/yuP0eELo9Owa5yTfn/usdt6CywguWFMSbaeZ2Wk57mcnQm6H\n2y0l/29CLjnbH0JQ7pPMLDUMvchTQ5S/l+D9cASwIyGx+6NZn2sFP+CzgG0IT4gPErwmZmbfbT+k\n2UbkvrYBDxFckCr7wwjuMVny/4p/J8e/FR/WWtd4hPAIP5mgbI+gxiJYyvkiKJWs448SZupbAptX\nthryJxJqXv04bo8RPC+y5B8n/pjnGGvqghM1Fp4IC5DDc/bfQoiIug64Pr6uOTbgzvieD4jb4cCd\nDXx/Zsf/wRcSbZMyZFuBJxrwuV2OoKRryXyMUIXhV8DH6sh+kGCz3yHur5G8nxT5h4DPJ97Tz1W+\nG0vS5jPcgqQ5fNdyAu/mDOsRMxuTtKGpdkWGc+h4TG4hBGjMMrO0emNImmhmm9e80c7yU4FtzOyd\nuL80wck/y753HXCCxUWlnNf4AJ1dnjLd6KKr3aYExZhc1GpUSe+i/+Oi788kwmP1Hwkz+xMJK/ZZ\n/9/croU17mkgwXOhS16Pok8N3bx+F3uwpEctw02tv+I23OJU15caQ+36UskiiUcBfzezmkUSyZmP\nNUHS5rkIuNqqkqRX8TdJXyOExybj2t/IkBedwzfbYltnoW4UJFSo3PBrQlrAVwkz28cJDvJZ3EBH\ntrXKD02n8XTXRzby37jyfnXcP5TgapdFrvcnKW/B13g/SacRXL2G15BfHpge38/KD4yZ2SczL9A5\neKaFkOozdVHOwqJau6ThVt8HursVHG6TdArBM8YIi4a3xoW9Wp+9foUr3OKcBFwn6aW4vwrhw9OJ\nuJiwmpmdB1wSF89WAjaX9JaZXV/jGp8nfEmOI/ilrk4Ie03FzK6Mbj+Y2Ws57qGSlyG52GaE6glp\nXA78S9Jf4/7+hNI41dxMSDA9vqp9B7LLb/+EEOL6DzPbVKEgZ9bMPPmeEhXQSnHs1Q70cxWi0faj\nfrKaao4k2HB/G899gGBWyCLX+6OQTeyDJJLpmNlpktoIpdyzSLoWVnIvHFLnHpIJyBcBz1kNGzQh\nxeVjCqkjM58arPsVHA6Kfys5Iyo/SIdQ+7PXv2i2TaOvbAQ/yA/G1wMJyvAuQv2rFVLk7yckfa7s\nTyF4KqxBiP2vda0Tc7aJkNnqdeANgmP8a4RV+0bf/2YEz4wTgE0zZG4BPprS/lHgbxnnPBL/Pgq0\nVF5nyOZ+TwmP6Q8SIgLPzBpz1Tmr1zi2bzPen4TMpgT76izCjDjTRpxxfgtwWI3jX0zbMmRz25WT\n35vEdW4mJCDq8r3p71vTB9BXNmBS5QNCWJV9iRAO+xNCmGy1/ISq/XMTr2tGMpGygEIiKinR9g2C\nHXOtRNvahNj/r6fIfzvx+sCqYz9PkR9CmNGfCxxNqHtVa9yZkUNkRKsRsk0Ni9e4mrCSnboI2Z33\nlGCiOJmwAPkEwf94VIbsE8DIlPYjgH/39vtDiGL8YRzXPwlrAM/VucayhAjEcwm5IESYHMwCbkqR\nX6Ob34eb8pxb9HvT37emD6CvbCRmXYQkK6cl9tPCYZ+p0VeXL29sPxT4G2GmenNiu5uUWXFUIiNS\n2lfKUNCT0l6n7ce2awkLO0cTfEB/V+c9errGsdT3g5CGsoUwa/piVCqpM5/uvKdVMpvG96wt4/jH\ngacIQSqVtu8Q7O+r9fb7Q3Cfuhf4cKJtZp1r3EQIRDiakM7xntjH6Az55GfiL/Xew4RsrpD1ot+b\n/r65DTc/rZIGWEh6shud85emvY//kvQVM7sk2SjpaLITbD9AsHWOoHN47Bwgza9zoKWk9TOz15Qe\nbqyM12n7ABtY9KuMvqP1slk9knHPXyYEYiTb0pKhV8ZwqqR/A98zs3GJ44Xf0xjosDfBVrgbQQGd\nliZrZrdKmk9Y4NmfkMNgS2BHS0+/2WPvT+TTcdx3K5R4v4bai3EQMpBVxnQp4fO0hmVnOkv2V8SO\nmjdkvej3pl+zxN3wYnA1ocTO6wSvhPHw/kJI2sru14EbJX2W8FgFwd91MGFRpQsWIs2eIziH5yHT\n8T3jmGW8TtuHjvh3LFQ0qDeek4C/KoSuVhTIGELwxqc6XazGokt0U9qIUNRwo8Sh3O9pjEQ7lDBr\nfZigrI6y6LqVhYXSRkcQFPMDhIxeWcqqx96f2OeNhPtdGvhkPP8DCtnG/mpmd9QZU5uk2TXGD7U/\nE9knZWSuS6Ho96Zf4364BYgRYKsQUulVfC5HEQIhupT1jsd3pcPFabqZ3VWj/0IlcOLqdpoCESHj\n1sAMedE1x20tearOqZcrYBc6FGXNe66FpKPN7KKU9rrvqaS7CHXF/pIxO027XjL14GCC8qq4eNV7\n/3vl/VFI03ggcLCl5PQtOqY6n4k0+cJlmrrzvemvuMJ1HMfpJWo50zuO4zgNxBXuYiKpVvE/l3f5\npl+jr8v3J1zhLj5FPzwu7/K9fY2+Lt9vcIXrOI7TS/iiWU5al1naBoxYvkt725x3aF2ma67mIS+m\nFvJlQfs8BrUM7dJuC9PlFzKfgQzu0q7W1vT+bR6D1LX/LI+fBe3vMahlSOoxa+9a9WWhzWeguo4H\nQAO6ehlm3S+ALexaxzHrfgHU0nV+sMDeY5AKjL9G/2nUHE/K/yD7/Qdra0ttz7xGhpdZrf9B2r+5\n5j0MSLmHGp8J0v7Hi95l0ICl0ofz3vwubVnjeY93WGDz6/rW1WLPXZa2/76R/j5XM3Hq/LFmttfi\nXK8o7oebkwEjlueDP8wqktCVDU7NnZkQgEWzXywk37psreRSKbQX/2Ftfzc1p3omrSuNKCS/6OXU\nai+ZtAwrli+lfe7cQvIUnHwU/R+0vVXM7TTtB6wetii1GHEmrcuvWOwCK3addNSi7cm0AiLp/KtT\njEv3+O8bbTw8do1csq2rPF3sA9sAXOE6jtNvMKC9Zj3O5tLrNlxJ35M0XdJUSVMUalgVOX+0pI8n\n9ndWqJ9V2T9GiXpOKeefplBOxHGcfoZhLLS2XFsz6NUZrqRtgH2BzSxUHh1BCGsswmhCOOStcX9n\nQi7PBwDM7MLGjNZxnL5ImWe4vW1SWAV43czmw/vVY5G0BSEt39KECgG7EcIqLyAo10WEVIT3E2pG\nDY3Jpa8GjgHaYob+4+O5cy1Udj0hHl9EKGtTSdq8gaR7CHlUf2dmZ/f0jTuO0/MYRluJHQF6W+He\nQcgE9RQhD+q1hCTR1xJiwydIWpaQ5OJEQmz2RyWtH88dBZwKjDGz4wAkDSUq2LifjC8/hZArdr5C\nmekK6xNqSi0DPCnpAjPr4iYQHbSPAmhdcbnqw47jlJD2wgU+eo9eteGa2VxCdqejCJUJriXk7XzZ\nzCZEmf/FVG7bE3KNYmZPELJojSp4yanAVXH2m1y+/buZzY8z7FcJZWHSxnuxmY0xszFprl+O45QL\nA9qwXFsz6HUvBTNrI6S+u0ehwN+xPXi5fQhZ5vcDvifpo7E96RzYhntrOE6/wWe4EUnrSVo30TSa\nUKF1lWjHRdIyMWn0eOCw2DaKYG99kpCMO+mQWb1fuVYLoUbV3YQSK8MJpVwcx+mnGLDQLNfWDHp7\nZjcMOCfaUxcBzxDMC5fH9qEE++3uwPnABXEWvAg4PNpi7wZOkTQFOJ1Qkub6WNE1GZnQCvxR0nBC\nzM7ZZvZWjiTRjuP0UayJ5oI89KrCNbOJwLYph14nlMqu5oiUPt4gVAJNsnHidbJE9/Yp559Wtb9R\ntYzjOH0Ug7by6lu3XeZlyIsL2OB7s3PLP3PsWoX6X+u0VwrJa1gx60j7m28VkgcgI19DFoteea1Y\n/y3F+i/6dJKVbyKLrFwH2fLF/D1bRxQLo22fUzA0GWgdVmxxV8sU+xxZwf9x63L5w5/1v2L/rzRC\npFl5cYXrOE4/QrTVrbPZPFzhOo7TbwiLZuVVuM3IpbCypD9JmilpoqQHJXWpWOo4jlOU4IerXFsz\n6G23MAE3AveZ2dpmtjlwCLBazvN9Ru44Tk3aTbm2ZtDbM9xdgQXJBDNm9pyZnSOpVdIvJU2ImcSO\nhvezgY2XdDMwQ9JISU9IukLSU5KukrS7pPslPS1py3jelnH2PFnSA5LWi+2HS7pB0u1R/sxefg8c\nx+khyj6NECH0AAAgAElEQVTD7e0Z44ZAVh36LwFvm9kWkgYD90u6Ix7bDNjIzJ6VNBL4MHAgcCQw\nAfgswQXsE8B3gf2BJ4AdzGyRpN2BnwOfif2NBjYlRJw9KekcM3uhekDJXApDWj1mwnHKjiHaSlw5\nrKmP6JLOIyjKBYRcCRtLOiAeHg6sG489bGbPJk591swei31MB8aZmcUgiZGJ86+MkW0GDEycP87M\n3o7nzwDWBLooXDO7GLgYYPigD5TYu89xnArNMhfkobcV7nQ6ZpmY2bExJ+4jwPPA8WY2NnmCpJ2B\nd6r6SeZCaE/st9NxTz8B7jazT8VZ8T0Z53suBcfpJxhigS2+P29P0dtz77uAIZK+mmirVJ8bC3xV\n0kAI+RMkLU6KruFApVDY4YvRj+M4fYQQ+NCSa2sGvZ2e0Qj21Z0kPSvpYeBKQnKZS4EZwCRJ04CL\nWLyZ55nA6ZImL2Y/juP0IXzRLIGZvUxwBUvju3FLcg8Jc4CZzQI2SuwfnnbMzB6kc/7c78f2K4Ar\nEufsW+gGHMcpLWaizXzRrM9jCxcVKus98tRiMecL71i12Hh2f76QfHdKbrcUjMtvm1+sDHjrsssW\n639utSm/Du09WyiwaBl2vTewvlBSvrW44mj7X7ExtbYXXAtuKTgz/FBqbv905jVGHbU3aPYqaS9C\n6a9W4FIzO6Pq+HBCkYQ1CLr0V2Z2ea0+XeE6jtNvCItmi6/WJLUC5wEfA2YDEyTdbGYzEmLHEmol\n7idpJYKL6VVmtiCr3/LOvR3HcQrSwEWzLYFnzGxmVKDXAJ9MudwyMYJ2GPAGnUt5daF0ClfS3Ph3\npKTP5pAfGRfZkDRGklfgdZwlmDZTrq0Oq9LZN392bEtyLvAR4CXgMeBEM6uZHbJ0CjfBSEIEWW7M\n7BEzO6FnhuM4TtmpRJrl2YARkh5JbEcVvNyewBTgQ4To1XNj1fFMyqxwzwB2kDRF0tfjTHa8pElx\n61I5IuZduCW+9lwKjrME0m4tuTbg9UpV7rhdnOjmRWD1xP5qdPj1VzgCuMECzwDPAuvXGluZF81O\nAb5ZcduStBTwMTN7L4brXg2MqXH+YudScBynbxGS1zRkHjkBWFfSWgRFewhdn7ifB3YDxktaGVgP\nmFmr0zIr3GoGEqbsownhuKPqyC92LoVOyWveD4hzHKesGGJhA0J740TtOEIEbCvwezObLumYePxC\nQvqAK2IOFwEnm9nrtfrtSwr368ArwCYEU8h7deQXO5dCMnnNslrBk9c4Tskxo2GBD2Z2K3BrVVsy\ntexLwB5F+iyzwp0DLJPYHw7MNrN2SV8k/OrUwnMpOM4ShxoW+NATlHnRbCrQJulRSV8Hzge+KOlR\ngmG6XtiR51JwnCUMI8xw82zNoHSKyMyGxb8LCRUikmyceH1ylJtFR/6Ee4imA8+l4DhLJp6AvB8g\niZYhQ3LLt6y4QrELFMyN8NQlWxSS/8i3niokD8Xj8jVwULH+58wpJN+63HKF5NvffbeQvM2fX18o\nQcvgwYXkNbxg7ohXXi0k3x20dMHF4NZiC1Ltz1V7UtVgwcJiY0nBaF69sjy4wnUcp98QyqSXV62V\nd2SO4ziFaV6u2zyU1tghqS1GmU2PC2f/T1Ld8UqqzqebJnNFonaa4zj9BKNQpFmvU1qFC8wzs9Fm\ntiEhRdrewA9znFdX4TqO038pc8WHMivc9zGzVwkRX8cpcLikcyvHJd0S8yicAQyNM+Or4rEvSJoa\nZ8l/SHS7Y8yxMNNnu47TPzBTqWe4fcaGa2YzY1LgD9SQOUXScWY2GkDShgR3sG3N7HVJSdeBVQgl\n2tcHbgau77nRO47TG4RFs/JW7e0zCreb7ApcV4lvNrM3EsdujLkrZ8TEE13olEthsQoIO47TO3hN\ns4YgaW1C3oNXCVnVk+9qfgfZDpJOl6kGnWQuheEtK3ouBccpOWHRzL0UFotYL+hC4NxYan0WMFpS\ni6TVCeUwKiyUVMkMdhdwoKQVYz8FoxEcx+lrFEhA3uuUeYY7VNIUQlrFRcAfgN/EY/cTkv3OAB4H\nJiXOuxiYKmmSmR0m6WfAvZLagMl4IhvH6bd4pFk3Mcu2fMdZ7mEZx04m5lmI+1cCV1bJHF61P2xx\nxuo4TnnIUSCyaZRW4ZaOwYPQOmvmFreFNYt3dqWl2MrqBj97pZD8/NHrFJIHGHD/tGIntPTszMIW\nZFafboh8UYrmRmBevRTOnWldtmD/QJiLFJAfViyXQtuKxeYmeq1mPu7OYyk49vQ+YGG7K1zHcZwe\nJ5gUXOE6juP0Cp5LIYNEvoTKdspi9DU3/v2QpMwghlj9t+CzsuM4fYGKW1ierRk0e4Y7rxIV1ihi\nnSEP1XWcJZJymxRKOTJJsyT9SNIkSY9JWj+2ryTpzphB7FJJz0kaUXXu+zNYSRtKejjOnqfGCr4A\nrZIuif3cIWloL9+i4zg9RHusa1ZvawbNVrhD1dmkcHDi2OtmthlwAfDN2PZD4K6YQex6YI06/R8D\nnBVn0WOA2bF9XeC82M9bwGcadD+O4zSR4KXQmmtrBmU2KdwQ/04EPh1fbw98CsDMbpf0Zp3+HwS+\nJ2k14AYze1oSwLNmNiXR/8i0kzvlUhhY3EXHcZzepeyBD82e4daikuugjW7+MJjZn4BPAPOAWyVV\nilIm8yhk9m9mF5vZGDMbM6i1YO0nx3GagpsUGsf9wEEAkvYAlq8lHBPezDSzs4Gb6Fz113GcfoZ7\nKdSmki+hwu1mVss17EfA1ZI+TzAX/AeoVfr1IODzkhZG2Z8DbhtwnH5Mmb0Umqpws/IlmNnIxOtH\ngJ3j7tvAnma2SNI2wBZmNj/KDYt/ZwEbxddnAGdUdf9G5XiU+VUDbsVxnBJgJha5wm0YawB/jsUk\nFwBfafJ4HMcpGWVeNOtTCtfMngY2bcq135tP2/Qne6z/1pVWKiS/aNbzxfovKA8w+4YNC8mv+unp\nheQ1eHAh+ZblhheSb3/nnULyRWl75dVC8oXvd1jxKiPt/32jvlDyGgX7b/nPa4XkbeNR+YWn3V1w\nNCnXwxWu4zhOr1FmhVvoBy6R+2CapOskFfKVkuQlzB3H6TEqfrhl9VIo+kQxz8xGm9lGBBvqMXlO\niqXNWwBXuI7j9Cj91Q93PPBhAEnfiLPeaZJOim0jJT0p6f+AacBldITyXlWdtUvSNyWdFl9vEXMf\nTJH0y0RuhMMlnZs45xZJO8fXe0h6MOZfuE7SsNh+hqQZsb9fxbaVJP1F0oS4bbcY74PjOCXBDBa1\nt+TamkG3bLiSBgB7A7dL2hw4AtiKUP32X5LuBd4k5Cz4opk9FM87sBLKK2lkjUtcDnzFzB6UVO3W\nlTaeEcD3gd3N7B1JJwPfkHQeIRR4fTMzScvFU84Cfmtm/5S0BjAW+Eixd8FxnDJSZhtuUYWbDFQY\nT5i1fhX4q5m9AyDpBmAH4GbguYqyzUtUisuY2YOx6U/AvnVO2xrYALg/5koYRAiMeBt4D7hM0i3A\nLVF+d2CDKAuwrKRhZja3aiwduRTw0F7HKTtlz6VQVOF2STaTUFpp1PLLWURnk8aQHNfPOkfAnWZ2\naPUJkrYEdiPkyD0O2DX2sbWZ1SwyZWYXE6oAs6xWWPyCS47j9DhWYoXbCEPGeGB/SUtJWprwCD8+\nQ3ahpIHx9SvAByStKGkwcRZrZm8BcyRtFeUOSZw/CxgtqUXS6sCWsf0hYDtJFZvy0pJGRTvucDO7\nFfg6sEmUvwM4vtKppIYmQXccp3mUedFssf1wzWySpCuAh2PTpWY2OcNGezEwVdIkMztM0o/jeS8C\nTyTkvgRcIqkduJdgGoCQvOZZYAbwODApjuE1SYcT8ixUvMu/T8izcJOkIYRZ8DfisROA8yRNJbwH\n95HT48JxnPJi1o9suJV8BSntvwF+U9U2i0TOgth2MnByYv9s4OyULqeb2cYACnXOHonyBhyWMYa7\ngC1SDm2ZIvs6cHCKrOM4fRrRVuIy6WUd2T6VAAvCAtxPmz0gx3H6BmbKtdVD0l7RtfUZZRS4lbRz\n1FXTo3dWTUoZ2mtm1wLXNnscvUnba8Vi1GkpWCKkva2YPLDaQcVyR6zy0DKF5F/aulZmza4sevGl\nQvKtG65XSH7hiGKeKC33Ti4kb/Pn1xdK0FZQvju0zyn2PyjMhMfyy9q8xb5co3IpSGoFzgM+RijN\nNUHSzWY2IyGzHHA+sJeZPS/pA/X6LesM13EcpzgW7Lh5tjpsCTxjZjPNbAFwDfDJKpnPEkp3PQ9g\nZnWzGfW6wpX0QUnXSPq3pImSbpV0VPSTTZO/VNIGvT1Ox3H6JgW8FEZIeiSxHZXoZlXghcT+7NiW\nZBSwvKR7oi77Qr2x9apJQcFp96/AlWZ2SGzbhFB3LBUz+3IvDc9xnD6OFVs0e93MxizG5QYAmxP8\n/IcCD0p6yMyeyjqht2e4uwALzezCSoOZPUrw2x0m6XpJT8RcCwKIvx5j4uu5kn4m6VFJD0laOban\n5kaQtJM6SrBPlrRMbP9WlJsq6Ue9/B44jtODNMik8CKwemJ/tdiWZDYw1szeiZ5P99Hh659Kbyvc\njQhlydPYFDiJEKK7NpCWUGZp4CEz24Rwc5WKD5XcCFsAnwEuje3fBI6N0XE7APMUik+uS7DRjAY2\nl7Tj4t6Y4zjloEFeChOAdSWtJWkQIQDr5iqZm4DtJQ1QSFW7FSE+IJMyeSk8bGazAWK+hpHAP6tk\nFtCRD2EiYQURMnIjEAIlfiPpKoJxe3ZUuHsAlSXmYQQFfF/1gDyXguP0LcLsdfG9FGLdxOMIia1a\ngd+b2XRJx8TjF5rZ45JuB6YC7YSgr2nZvfa+wp1OyGmQRtIHpo30sS2MwQ/VMlm5Ec6Q9Hfg44TE\nNnsSIs5ON7OL6g3Wcyk4Tt+jUZFmMSXArVVtF1bt/xL4Zd4+e9ukcBcwOLkaKGljwuP+4pCaG0HS\nOmb2mJn9gvCIsD7hF+tIdeTLXTWP/5zjOH2DBtlwe4ReneHGnLSfAn4Xc9a+R0hIc+Nidp2VG+Ek\nSbsQpvvTgdvMbL6kjxBWFAHmAp8DilUEdByndBiivcShvb1uwzWzl4CDUg5dkpA5LvF658TrYYnX\n1wPXx9epuRHM7Pjqtth+FmGhzXGcfkaZbX9lWjRzHMdZPBq0aNZTuMItQu1k650paCRq2aRYhR+b\n/nQx+W58CDWg2Mfj5e2LxcI/9ftiPufrHzejvlCCthmZ/ueptLYWy0/RuvpqheTbX3u9kLyGDi0k\nD8VzIxT9H9uiRYXkW0esmH8srzdIHZV4iusK13GcfsUSO8OVNDcrh67jOE6jMaC9fQlVuI7jOL2K\nASWe4fa4/4SkYZLGSZok6TFJn4ztIxN5Ex6PeRSWisdOjbkOpkm6uCqvwi8kPSzpKUk7xPZWSb9M\n5Ec4OravIum+SjLzhPwekh6MY7qu4pPrOE7fp8x+uL3hsPYe8Ckz24yQvObXFQUKrAecb2YfAf4H\nfC22n2tmW5jZRoQsPMky6QPMbEtC3oUfxrYvAW/HXApbAF+RtBYhX+XYmEthE2CKpBGEeme7xzE9\nQketM8dx+jqWc2sCvWFSEPDzmCCmnZBTcuV47AUzuz++/iMhgOFXwC6Svg0sBaxACFr4W5S7If6d\nSMi3ACE3wsaSKmHDwwn5ESYAv1eoFHyjmU2RtBMhQc79Ue8PAh5MHbjnUnCcPka+8jnNojcU7mHA\nSsDmZrZQ0ixgSDxW/TtjChV2zwfGmNkLkk5LyENHzoVkLgUBx5vZ2OqLR0W/D3CFpN8AbwJ3mtmh\n9QbuuRQcpw9S4m9qb5gUhgOvRmW7C7Bm4tgakraJrz9LyA5WUa6vR9tqVrKbJGOBr8aZLJJGSVpa\n0prAK2Z2CSFl42bAQ8B2kj4cZZeWNGox79FxnDJgYO3KtTWDHpvhShpAmI1eBfxN0mMEe+kTCbEn\ngWMl/R6YAVxgZu9KugSYBvyHYBaox6UE88KkaB9+Ddgf2Bn4lqSFhJwJXzCz1yQdDlwtaXA8//tA\nMS95x3FKypJpUtgQ+HfMc7BN9UFJI4FFZva56mNm9n2CEqxu3znx+nWiDdfM2oHvxi3JlXGr7ucu\nwuKa4zj9jSXNpBCT9F5NitJ0HMfpUZY0L4WYpPfCOjKzCCV3+g496bz35LOFxK2trVj/3Rh70bj5\novKjvjy5vlCC22ZnVWdKZ88PjS4kX3T8i158uZA81l5M/r3qfPqNp+jnSAXzTSx6JX/WU2sv9v6n\nd0KpAx880sxxnH5Fs4Ia8lDaTL2SVpb0J0kzFWq+PxiTlzei77mN6MdxnBLSrnxbEyilwo2eBjcC\n95nZ2ma2OaFq5mpVcj5DdxynE7J8WzMopcIFdgUWJAu2mdlzZnaOpMMl3SzpLmAcgKRvJfIo/Khy\njqTPxbwLUyRdJKmTAUrSiDhz3qe3bsxxnB4k74KZK9xObAhMqnF8M+AAM9splj1fF9gSGA1sLmnH\nWLfsYGC7mEuhjRD1BgSTBfB34FQz+3sP3YfjOL2KwqJZnq0J9IlHcknnAdsDC4DzCKG5b8TDe8St\nsuQ9jKCANwY2BybEnAlD6SgUOZAwOz7WzO6tcV3PpeA4fY0SL5qVVeFOBz5T2TGzY2OWr0di0zsJ\nWQGnm9lFyQ4kHQ9caWbfSel/ESH5zZ5ApsL1XAqO0wcp6H3Xm5TVpHAXMETSVxNtWVPMscCRlZy2\nklaV9AHCDPaA+BpJK8TcChB+A48E1o/l2h3H6Q9U/HDdpJAfMzNJ+wO/jWkaXyPMak8mmAaSsndE\ne+2D0XQwF/icmc2Q9H3gDkktwELgWOC5eF6bpEOBmyXNMbPze+v+HMfpOZrlgZCHUipcADN7meAK\nlsYVVbJnAWel9HEtcG1K+7D4dz7BrOA4Tn+hxAq3rCYFx3GcfkdpZ7hLGu29EDdflKK5BVo/sm4h\n+bbHny4kXzQ3wvdmTikkf8ZmOxWSb3vr7ULyZaRl8OD6QgnK+Dmtxk0KjuM4vYHRtLDdPNQ0KUi6\nW9KeVW0nSbpc0vV1zh0p6bONGKTjOE5u+nCk2dV0Xbg6BLjczOqVvhlJKJvjOI7Ta/TlXArXA/tI\nGgTvV2n4EPCCpGmxrVXSLxO5DI6O554B7BDzGHw95kC4QdLtkp6WdGblIpIukPSIpOlVuRBmSTo9\n9vGIpM0kjZX075jkvCLXJZdCrFX2d0mPSpom6eDYvrmke2MGsrGSVlncN9FxnBJR4hluTRuumb0h\n6WFgb+Amwuz2z3Qe7peAt81si1gj7H5JdwCnAN80s30BYh2x0cCmhFpnT0o6x8xeAL4Xr9UKjJO0\nsZlNjf0/b2ajJf2W4A62HaHQ5DTgwqpcCiL41e5IqBT8kpntE68/PBaZPAf4ZKxtdjDwM0IQhOM4\n/YE+vmhWMStUFO6Xqo7vAWwsqWJiGE5QgAtS+hpnZm8DSJpBqOD7AnBQzFswAFgF2ACoKNyb49/H\ngGFmNgeYI2m+pOXIzqUwHvi1pF8At5jZeEkbEapM3BmDJFqBzLT9nkvBcfoWzTQX5CGPwr2JEPG1\nGbCUmU2MpoUKAo43s7HJkyTtnNLX/MTrNmCApLWAbwJbmNmbkq6go1R68pz2qvPb4/hTcynEMWwG\nfBz4qaRxwF+B6WbWpahlGp5LwXH6IH3VSwHAzOYCdwO/J8x2qxkLfDU+riNplKSlgTnAMjnGsCwh\nbPftmDJx75xjT16/Sy4FSR8C3jWzPwK/JKR0fBJYSdI2UXagpA0LXs9xnBJT5kWzvH64VxNmh2mh\ntpcSPBImKTynvwbsTzAJtEl6lGB7fTOtYzN7VNJk4AmCeeH+AuPPzKUAfBj4paR2Qh6Fr5rZgmj6\nOFvScML9/46QncxxnP5AiZ9FcylcM7uR8Ohe2Z9FrLhrZu3Ad+NWza5V+1ck+tg38frwjOuOTLy+\nour85LG0XAr/Jsx+q/ucAuyYdj3Hcfo4DZy9StqLoFdagUvN7IwMuS2AB4FDzKxmfILnUnAcp3/R\nALew6DF1HsHEuQFwqKQNMuR+AdyRZ2ge2lsWVMzQrwEDC8nbwjSnkcby5ugVC8kvWzCXQlF+sd1e\nheTbb2itL5Rk1x7OpdBScDwA7W3FxPtAboSiqDEJyLcEnjGzmQCSrgE+Ccyokjse+AuwRZ5OfYbr\nOI7TlVUJa0oVZse295G0KvAp4IK8nTZd4Uqam3j9cUlPJSozOI7jFCO/SWFEjGCtbEcVvNLvgJPj\nOlYuSmNSkLQbcDawp5k9l/OcAWZWLIeg4zj9l2KLZq+b2ZiMYy8Cqyf2V4ttScYA10TvqBHAxyUt\nik4GqTR9hgsQQ3EvAfY1s3/HtpGS7or5EcZJWiO2XyHpQkn/As6MORN+L+lhSZMlfTJx/nhJk+K2\nbWzfWdI9kq6X9ISkq6I7m+M4/YHG5FKYAKwraS2FXDKH0BH1Gi5jtpaZjYweU9cDX6ulbKEcM9zB\nwI3Azmb2RKL9HELV3SslHUmY/e4fj60GbBvrkv0cuMvMjoyhvg9L+gehJPrHzOw9SesSfIkrv2ab\nAhsCLxH8frcD/tmzt+k4Tq/QALcwM1sk6TiCa2kr8Hszm15JmmVmF3an3zIo3IXAA4QcDScm2rcB\nPh1f/wE4M3HsOjOrLMfuAXxC0jfj/hBgDYIyPVfSaEIY8ajE+Q+b2WwASVMIgRtdFK7nUnCcvoVo\nmJcCZnYrcGtVW6qizYolqKYMCrcdOIiQJey7ZvbzHOe8k3gt4DNm9mRSQNJpwCvAJgTTSdL/pUtO\nh7SLeC4Fx+ljlDx5TSlsuGb2LrAPcJikSjayB+gIJT6MkP0rjbHA8RU7rKRNY/tw4OW4gvh5wmOB\n4zj9nb6aD7c3iflw9wLuk/QawaH4cknfIuRnOCLj1J8Q3DOmSmoBngX2Bc4H/iLpC8DtdJ4VO47T\nXynxDLfpCtfMhiVevwCslThcnYuhi63EzOYBR6fIPQ1snGg6ObbfA9yTkDuuWwN3HKeUlNmk0HSF\n6ziO01Bc4fYDJDRwUH7x1mLm8cIx7fmDW3qNZa+dUEhegwcXu0BbsTwBi17+TyF57VHs67DntP8V\nkh+3d5fcJzVZ9MLsQvIAGlDsHqy9mHZqHbZ0Ifm2uQUsecX+velY47wUegJXuI7j9C9KPMMthZdC\nGpLaFKr1TpN0naSajrCVnAySPiQpMydljECb1ujxOo5TDspc8aG0CheYZ2ajzWwjQkHKY+qdAGBm\nL5nZAfUlHcfpl5TYLazMCjfJeELJHCR9I856p0k6qVowOYOVtGHMsTAl5mRYN4q1SrpE0nRJd0ga\n2nu34jhOj5FX2brCTUfSAELW9cckbU7wx90K2Br4SiLQIY1jgLPMbDQhj0JlFWJd4Dwz2xB4C/hM\nT43fcZzeQ7hJobsMjXkOHgGeBy4Dtgf+ambvxGrCNwA71OjjQeC7kk4G1ow+uwDPxtpmABMJuRS6\nIOmoSq7Mhdb/MuM7Tn+kzAq3zF4K8+LM9H2KZlE0sz/FNI77ALdKOhqYSddcCqkmhU65FFpWLPHa\np+M471Pib2qZZ7hpjAf2l7SUpKUJ5S2yciwgaW1gppmdDdxE58gzx3H6IyW24ZZ5htsFM5sk6Qrg\n4dh0qZlNrnHKQcDnJS0E/gP8HFi2Z0fpOE7TKHm2sNIq3GSOhar23wC/yZI3s1nARvH1GUB1Lfk3\nKsejzK8aM2LHcUqBK1zHcZzewUN7l0BaVlyhkHz7y68U63/55Yv1/9bbheQBNGhgsWu8UywDplSs\n/8ITl6KLrAVzNdz+lVoOMl15+eBieQhWO/f1QvIAKpjrgNaCaaLnFfPWaV1pxdyyer0x6shNCo7j\nOL1BExfE8uAK13Gc/kWJFW5p3cIkfVDSNZL+LWmipFsljap/Zt1+T0sUnHQcpx9R9kizUs5wY32y\nvxLKpB8S2zYBVgaeaubYHMcpNyqY47c3KesMdxdgYbIksZk9CnwsJqKZIulFSZcDSPpcIknNRZJa\nY/tekiZJelTSuET/G0i6R9JMSSf06p05jtNzePKabrERIcdBJ8zs1BjuuzPBn/ZcSR8BDga2i8fa\nCNV/VwIuIZRQ3wQ4MNHV+sCewJbAD5WxXO65FByn7+EmhQYSzQ1/BH5jZhMlHQdsDkyIuRaGAq8S\nsondZ2bPQqgKnOjm72Y2H5gv6VWCqaJLPRPPpeA4fZASf1PLqnCnA1lJxE8DZpvZ5XFfBFvvd5JC\nkvar0X918pqyvg+O4xSkzH64ZTUp3AUMlnRUpUHSxpJ+AOwOJO2u44ADJH0gyq0gaU3gIWBHSWtV\n2ntt9I7jNI8S23BLObMzM5P0KeB3MZfte8AsYClgVeDhaD642cxOlfR94A5JLcBC4Fgzeygq7Bti\n+6vAx5pwO47j9BZetbd7mNlLhGxfeWSvBa5Nab8NuK2q7bSq/Y1wHKdfUPHDLSulVbjlw8Dy/3S2\n//eN+kIJWoYOKTaad98tJN8dbMHCQvItQ4rdQ9G4fxWM+2975dVC8kVpmTC9kPyqk4qN324bUUge\noH3XF4ud0FJsTK3LDy8k3/7GW7lli+ayyO6ovBrXFa7jOP0Kn+E6juP0BiVPXlNKLwVJ+0sySet3\n49y5Ge0/lrT74o/OcZwyo/Z8WzMopcIFDgX+Gf92IpZNL0yMUvvH4g7McZxy4wq3AJKGEcqhfwmo\nJK7ZWdJ4STcDM2LbjTGL2PSkv2489tvYPi6G+CLpCkkHxNdbSHog5lh4WNIyvXmPjuP0EEZYNMuz\nNYHSKVzgk8DtZvYU8F9Jm8f2zYATzaySovFIM9scGAOcIKmSWn5p4BEz2xC4F/hhsnNJgwguZCfG\nHAu7A/PSBtI5l8L8NBHHcUpGmXMplFHhHgpcE19fQ4dZ4eFKXoTICZIeJUSUrQ6sG9vb6fDJ/SNh\ntpxkPeBlM5sAYGb/M7NFaQMxs4vNbIyZjRmowYtzT47j9BYeaZaPGH67K/BRSQa0Et6avwPvJOR2\nJmWkAH4AAAvGSURBVMxMtzGzdyXdA2Q5gZZ4zdJxnEZS9sCHss1wDwD+YGZrmtlIM1sdeBaortY3\nHHgzKtv1CZnBKrTQkfjms4TFtyRPAqtI2gJA0jLdXYhzHKdkmKH2fFszKJvCPZRQ6SHJX+jqrXA7\nMEDS48AZBLNChXeALSVNI8yWf5w80cwWEPLnnhNNEneSPTt2HKev0SCTQixg8KSkZySdknL8MElT\nJT0WF+E3qddnqWZ2ZrZLStvZwNlVbfOBvTP6GJbRfnji9QQ6z4odx+knNMKkEKvGnEdIeDWbkG/7\nZjObkRB7FtjJzN6UtDchd/ZWtfotlcJ1HMdZLAxojLlgS+AZM5sJIOkaggfV+wrXzB5IyD8ErFav\nU1e4uREUSJ7SXjDxi1pSHSWaStFkIum+HtloUQ/fc8HELLQXu9+Wgsl3bGGx+23f/T+F5AGWv79Y\n2ue39yr4OR06tJA8b/+vgHCD7Kr5uxkh6ZHE/sWxyguENLAvJI7Npvbs9UtUZSZMwxWu4zj9igIm\nhdfNbMxiX0/ahaBwq11Qu9Dri2aS2mJ13WmS/iZpuQb2PUbS2fUlHcfprzTIS+FFgn9/hdViW+dr\nSRsDlwKfNLP/1uu0GV4K88xsdEz8/QZwbKM6NrNHzMzLnjvOkkrjyqRPANaVtFaMTj0EuDkpIGkN\n4Abg8zEyti7Ndgt7kGArqeRLuKVyQNK5kg6Pr8+QNCO6YPwqth0YZ8mPSrqvug9JW0p6UNLk6LKx\nXmw/XNINkm6X9LSkM3v3lh3H6SlC4IPl2moRo0+PA8YCjwN/NrPpko6RdEwUOxVYETg/PrU/ktHd\n+zTNhhvdLnYDLqsjtyLwKWD9WOusYoI4FdjTzF7MMEs8AexgZotiWsafA5+Jx0YDmxKq9z4p6Rwz\neyGlD8dx+hoNygRmZrcCt1a1XZh4/WXgy0X6bIbCHSppCmFm+zgh8KAWbxOKSF4WZ6+VWfD9wBWS\n/kyY1lczHLhS0rqEB4iBiWPjzOxtAEkzgDXpvCJJPHYUcBTAEJbKd3eO4zSVerPXZtI0Gy5ByYkO\nG+6iqvEMgfen9lsC1wP7EqLMMLNjgO8TDNsTE9nCKvwEuDvaivejczRZMvVXGxk/PJ2T13gwmuOU\nnsbZcHuEppkUYh6EE4AbJZ0PPAdsIGkwMJRgbvhnzI+7lJndKul+oOKIvI6Z/Qv4V4zyWL3qEsPp\nWFU8vOfvyHGc5tO8PAl5aOqimZlNBqYCh0Yb6p+BafHv5Ci2DHCLpKmERDTfiO2/jDHM04AHgEer\nuj8TOF3SZNzf2HGWHEqcgLzXFVF1rgMz2y/x+tvAt1NO2zKln0+nyN0TN8zsQWBU4tj3Y/sVwBWJ\nfvbNOXTHccqONa98Th585uc4Tv+ixItmrnBzohbRMjh/1QetsHyh/ttef6PYeFqLWYM0aFAheYC2\nOXOKnVDwg24Fcym0rvyBQvJtr7xaSL4oVjBfRtFcCi2DBtYXquLN7d8sJP/MH0YXkl/vx0VyI4AG\n5FcxWqRCfWdSXn3rCtdxnP6F2strU2h2pBnQKb9CZRvZg9fqFNHmOE4/wgiBD3m2JlCWGW7FNzcV\nSQOyCj06juNUEPXDdptJKWa4acScBzdLugsYF9u+JWlCzKnwo9g2UtLjki6RNF3SHZKGxmMflvSP\nmG9hkqR1YvfDJF0v6QlJV0lqkPHIcZymU2K3sLIo3KEJc0KyptlmwAFmtpOkPQil0Lck5ELYXNKO\nUW5d4Dwz2xB4i46cCVfF9k2AbYGXY/umwEnABsDawHY9eG+O4/QmJVa4ZTcp3GlmleX7PeJWCYgY\nRlC0zwPPmtmU2D4RGClpGWBVM/srgJm9BxAnsw+b2ey4PwUYSdfqvp1zKahYdn/HcZpAxYZbUsqi\ncLN4J/FawOlmdlFSIC6wVedGqFcHJHcuBUJhOIYPGFFew5DjOO/jXgqNYSxwZMytgKRVJWU6ZprZ\nHGC2pP2j/GBJnvLLcfo1Oc0JS7hJoS5mdoekjwAPRrPAXOBzhBlqFp8HLpL0Y2AhcGCPD9RxnOZh\neKRZParzK8S2K0jkPIhtZwFnpXSxUULmV4nXTwO7VsnOJOZbiDLHdWPIjuOUlfJaFMqhcB3HcRpF\nmf1wXeHmpd2wBQtyi7fNfae+UKf+a1lGuqLWYgnRra1Y/+Gkgh/cou7MBfvv6dwIRcffPm9ese4H\nFMuNYG3dmKoVfE/XO25mIfkvPTy5vlCCi0etnVvWGqUoXeE6juP0AmbQnR+qXqIveSl0QtLcGsce\n6O65juP0cdxLoXeo5Fwws22bPRbHcZpEiU0KfXaGWyFm/xov6WZgRmybG/+uIum+GDI8TdIOifN+\nFnMsPCRp5SYN33GcRmJAu+XbmkCfV7iRzYATzWxUVftngbExbHgToBL+uzTwUMyxcB/wlV4bqeM4\nPYiBtefbmkB/MSk8bGbPprRPAH4vaSBwYyLfwgKgkhN3IvCxtE49l4Lj9DEMXzTrBVJ9sMzsPmBH\nQrn0KyR9IR5aaB0+KDVzKZjZGDMbM4j85XUcx2kivmjWHCStyf9v725C66jCMI7/H0Ori4IVIhKa\nUgsNVVD8pK2LYnVjlUoXLtRNxZWCxZ0f4KLZqgvFRi2tBNqNlWbTgIVKkCwEq60okYraUFqMihIM\npRK4bczrYiZ0mtwk96Q3dybJ84PLzJ3z5syBwOHcM+e8AyMRcUjSzWRTD0dKbpaZLaYKPzRb1h0u\nsAN4TdJVstwLe+YON7OlrbzRayOWbIc7lX8hIgYp5EaYVnYYODzb3+bnfUDfIjbVzFolgAqnZ1yy\nHa6ZWV0e4S59EcFkrTZ/4AK1bd6UFD954be0G1y5mhYP3HTf3UnxcfZcWnziWsi2NWkrRSbHx5Pi\nYyLtPaVatTotvm3xn1FH6r9ZaW3qfXz7/EEF53o6G46tvX0qqe76qr211x2umS0fAVHSGttGlLIs\nTNJb+Rt2h/JdYFubXL9zKZitVBXeadbyEa6kR4BdwIMRUZPUDqT9NpuHcymYrWAVnsMtY4TbAYxG\nRA0gIkYj4g9JFyS9I+lHSd9K2gQg6WlJ30j6XtLAVN4DSd2SeiUNSjov6dWpGziXgtkKFZGtUmjk\nU4IyOtwvgPWSfpX0kaRHC2WXIuJeoAd4P7/2FbAtIh4AjgKvF+LvAp4AtgD78i28Rc6lYLbSeKfZ\nNRHxr6SHgO3AY8Bnkt7Miz8tHN/LzzvzmA6yqYdizoTP85FyTdLfwB3ASKG8ebkU8At/zaovFvZ2\nkxYp5aFZRPwXEYMRsQ/YCzwzVVQMy4/7gZ585PsSUHy3THGd1oycCM3MpbDKuRTMqs/pGa8nabOk\nrsKl+4GL+fmzhePX+fmtZB0mwAuJ99oA/BURh4BPyHIpmNly1qT0jJJ2SvpF0nDhV3ixXJI+yMuH\nJM3bv5SxDncNsF/SWmACGCb72b4LuE3SENnI9fk8vhs4JmkM+BLYmHCvHTiXgtmKEaRvqKlHUhvw\nIdl04whwWlJ/RPxUCHsS6Mo/W4GP8+OsypjD/Q6YsWxL2RtT342IN6bFHweO16mne9r3ewrnzqVg\nthJFNCu5+BZgOCLOA0g6Cuwmf6tMbjdwJJ+ePCVpraSOiPhztkq908zMlpUmPTRbBxT3z48wc/Ra\nL2YdUP0ONyLuLLsNc7nM2OjA5LGLdYragdGEqurH/5wYn1r/QuJ/qHu1vPZcqlh7rjSl/ua2KTX+\nn0WOfyUpfkPd6ASXGTs5EH3tDYbfIulM4fvBiDh4o22YS2U63KqLiNvrXZd0JiIebrQexzu+6m2q\nWnyKiNjZpKp+B9YXvndy7eF9Ssx1lssrdszMmuk00CVpo6TVwHNA/7SYfmBPvlphG9nGrVmnE8Aj\nXDOzGSJiQtJe4CTQBvRGxFlJL+flB4ATwFNkK63GgRfnq9cd7o1LnfNxvONbfY+lHl+KiDhB1qkW\nrx0onAezzVLPQlHhzDpmZsuJ53DNzFrEHa6ZWYu4wzUzaxF3uGZmLeIO18ysRdzhmpm1iDtcM7MW\n+R8Lh0cpaGEuQQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f891524add8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Keep track of correct guesses in a confusion matrix\n",
    "confusion = torch.zeros(num_categories, num_categories)\n",
    "n_confusion = 10000\n",
    "\n",
    "# Just return an output given a name\n",
    "def evaluate(name_tensor):\n",
    "    hidden = rnn.init_hidden()\n",
    "\n",
    "    for i in range(name_tensor.size()[0]):\n",
    "        output, hidden = rnn(name_tensor[i], hidden)\n",
    "\n",
    "    return output\n",
    "\n",
    "# Go through a bunch of examples and record which are correctly guessed\n",
    "for i in range(n_confusion):\n",
    "    category, name, category_tensor, name_tensor = random_training_example(all_categories, data)\n",
    "    output = evaluate(name_tensor)\n",
    "    guess, guess_i = get_category_from_output(output)\n",
    "    category_i = all_categories.index(category)\n",
    "    confusion[category_i][guess_i] += 1\n",
    "\n",
    "# Normalize by dividing every row by its sum\n",
    "for i in range(num_categories):\n",
    "    confusion[i] = confusion[i] / confusion[i].sum()\n",
    "\n",
    "# Set up plot\n",
    "fig = plt.figure()\n",
    "ax = fig.add_subplot(111)\n",
    "cax = ax.matshow(confusion.numpy())\n",
    "fig.colorbar(cax)\n",
    "\n",
    "# Set up axes\n",
    "ax.set_xticklabels([''] + all_categories, rotation=90)\n",
    "ax.set_yticklabels([''] + all_categories)\n",
    "\n",
    "# Force label at every tick\n",
    "ax.xaxis.set_major_locator(ticker.MultipleLocator(1))\n",
    "ax.yaxis.set_major_locator(ticker.MultipleLocator(1))\n",
    "\n",
    "# sphinx_gallery_thumbnail_number = 2\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Running on User Input"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "> Robin\n",
      "(-1.40) Irish\n",
      "(-1.93) French\n",
      "(-1.94) English\n",
      "\n",
      "> Jack Ma\n",
      "(-1.23) Spanish\n",
      "(-1.48) Greek\n",
      "(-1.69) Czech\n",
      "\n",
      "> Pony\n",
      "(-1.24) Vietnamese\n",
      "(-1.60) English\n",
      "(-2.33) Italian\n"
     ]
    }
   ],
   "source": [
    "def predict(input_name, n_predictions=3):\n",
    "    print('\\n> %s' % input_name)\n",
    "    output = evaluate(Variable(name2tensor(input_name)))\n",
    "\n",
    "    # Get top N categories\n",
    "    topv, topi = output.data.topk(n_predictions, 1, True)\n",
    "    predictions = []\n",
    "\n",
    "    for i in range(n_predictions):\n",
    "        value = topv[0][i]\n",
    "        category_index = topi[0][i]\n",
    "        print('(%.2f) %s' % (value, all_categories[category_index]))\n",
    "        predictions.append([value, all_categories[category_index]])\n",
    "\n",
    "predict('Robin')\n",
    "predict('Jack Ma')\n",
    "predict('Pony')"
   ]
  }
 ],
 "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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}