Result and Debiasing.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "import gensim\n",
    "import gzip\n",
    "import numpy as np\n",
    "import os\n",
    "import pandas as pd\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "from torch.utils.data import Dataset, DataLoader\n",
    "import torch.optim as optim\n",
    "from tqdm import tqdm_notebook as tqdm\n",
    "import pickle\n",
    "import we\n",
    "\n",
    "#Reproducibility\n",
    "SEED = 42 #The answer to everything\n",
    "torch.manual_seed(SEED)\n",
    "np.random.seed(SEED)\n",
    "torch.backends.cudnn.deterministic = True\n",
    "torch.backends.cudnn.benchmark = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " Adversarial_Training.ipynb  'GANv4 - no random.ipynb'\t  rGANv1.ipynb\r\n",
      " GANv1.5_neighbour.ipynb     'Modify vectors.ipynb'\t  tests\r\n",
      " GANv1.ipynb\t\t     'Ours + Adversarial.ipynb'   we.py\r\n",
      " GANv3_learnOpposite.ipynb    __pycache__\r\n"
     ]
    }
   ],
   "source": [
    "!ls "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "EMB = '../../../Research Work/Gender_Bias/embeddings/glove' "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*** Reading data from ../../../Research Work/Gender_Bias/embeddings/glove\n",
      "Loading numpy format\n",
      "loading ...\n",
      "(322636, 300)\n",
      "322636 words of dimension 300 : the, of, and, in, ..., zuras, zuriel, zyuzin, <unk>\n",
      "322636 words of dimension 300 : the, of, and, in, ..., zuras, zuriel, zyuzin, <unk>\n"
     ]
    }
   ],
   "source": [
    "E = we.WordEmbedding(EMB)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "# E.v('test')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Helper functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "def normalize(v):\n",
    "    return v / np.linalg.norm(v)\n",
    "\n",
    "def get_max(E, g, words):\n",
    "    df = pd.DataFrame(data={\"word\": list(words)})\n",
    "    b = {}\n",
    "    for word in tqdm(words):\n",
    "        b[word] = get_gender_logit(word, E, g)\n",
    "    df['gender_score'] = df.word.map(b) \n",
    "    df.sort_values(by='gender_score', inplace=True)\n",
    "    return df\n",
    "\n",
    "def gender_subspace(E):\n",
    "        pairs = [\n",
    "              (\"woman\", \"man\"),\n",
    "              (\"her\", \"his\"),\n",
    "              (\"she\", \"he\"),\n",
    "              (\"aunt\", \"uncle\"),\n",
    "              (\"niece\", \"nephew\"),\n",
    "              (\"daughters\", \"sons\"),\n",
    "              (\"mother\", \"father\"),\n",
    "              (\"daughter\", \"son\"),\n",
    "              (\"granddaughter\", \"grandson\"),\n",
    "              (\"girl\", \"boy\"),\n",
    "              (\"stepdaughter\", \"stepson\"),\n",
    "              (\"mom\", \"dad\"), ]\n",
    "        difs = []    \n",
    "        for f, m in pairs:\n",
    "            difs.append(E.v(f)- E.v(m))\n",
    "        difs = np.array(difs)\n",
    "        #PCA\n",
    "        difs = np.cov(np.array(difs).T)\n",
    "        evals, evecs = np.linalg.eig(difs)\n",
    "        return normalize(np.real(evecs[:, np.argmax(evals)]))\n",
    "\n",
    "def gender_subspace_simple(E):\n",
    "    return normalize(E.v('she') - E.v('he'))    \n",
    "\n",
    "# def load_words():\n",
    "#     words = []\n",
    "#     with open('w2v_gnews_small.vocab', 'r') as handle:\n",
    "#         for line in handle.readlines():\n",
    "#             words.append(line.strip())\n",
    "#     return words       \n",
    "# words = load_words()\n",
    "\n",
    "file = \"limited_vocab.hila\"\n",
    "with open(file, 'rb') as handle:\n",
    "    words = pickle.load(handle)\n",
    "\n",
    "def get_gender_logit(word, E, g):\n",
    "    return E.v(word).dot(g) \n",
    "\n",
    "def get_gender(word, E, g):\n",
    "    return 1 if E.v(word).dot(g) < 0 else 0 #1 is Male, 0 is female\n",
    "\n",
    "def get_rgender(word, E, g):\n",
    "    return 0 if E.v(word).dot(g) < 0 else 1 #reversed gender\n",
    "\n",
    "def label_to_gender(label):\n",
    "    return 'M' if label else 'F'\n",
    "\n",
    "# g = gender_subspace(E)\n",
    "g = gender_subspace_simple(E)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Clean vocab"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "for w in words:\n",
    "    if w not in E.words:\n",
    "        words.remove(w)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# DataLoader"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "class gender_set(Dataset):\n",
    "    def __init__(self, words, E, g, device):\n",
    "        self.words = words\n",
    "        self.embs = torch.Tensor([E.v(w) for w in words]).to(device)\n",
    "        self.gender = torch.Tensor([get_gender(w, E, g) for w in words]).to(device)\n",
    "        self.rgender = torch.Tensor([get_rgender(w, E, g) for w in words]).to(device)\n",
    "        \n",
    "    def __len__(self):\n",
    "        return len(self.words)\n",
    "    \n",
    "    def __getitem__(self, idx):\n",
    "        return self.words[idx], self.embs[idx], self.gender[idx], self.rgender[idx]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "bs = 128 \n",
    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "train_dataset = gender_set(words, E, g, device)\n",
    "train_loader = DataLoader(train_dataset, batch_size=bs, shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "grammy tensor(0., device='cuda:0')\n",
      "thinning tensor(0., device='cuda:0')\n",
      "nfb tensor(0., device='cuda:0')\n",
      "denials tensor(1., device='cuda:0')\n",
      "frustration tensor(0., device='cuda:0')\n",
      "schooner tensor(0., device='cuda:0')\n",
      "subtraction tensor(0., device='cuda:0')\n",
      "afield tensor(1., device='cuda:0')\n",
      "foresters tensor(1., device='cuda:0')\n",
      "symmetries tensor(0., device='cuda:0')\n",
      "cinque tensor(0., device='cuda:0')\n",
      "haye tensor(1., device='cuda:0')\n",
      "hubs tensor(0., device='cuda:0')\n",
      "created tensor(1., device='cuda:0')\n",
      "fifths tensor(0., device='cuda:0')\n",
      "biochemistry tensor(1., device='cuda:0')\n",
      "huffington tensor(0., device='cuda:0')\n",
      "monroe tensor(1., device='cuda:0')\n",
      "succulent tensor(0., device='cuda:0')\n",
      "stroke tensor(1., device='cuda:0')\n",
      "organists tensor(1., device='cuda:0')\n",
      "torbay tensor(0., device='cuda:0')\n",
      "wildcat tensor(1., device='cuda:0')\n",
      "forte tensor(0., device='cuda:0')\n",
      "grandis tensor(0., device='cuda:0')\n",
      "mocks tensor(0., device='cuda:0')\n",
      "expenditures tensor(1., device='cuda:0')\n",
      "pataki tensor(1., device='cuda:0')\n",
      "romp tensor(0., device='cuda:0')\n",
      "neurons tensor(1., device='cuda:0')\n",
      "skew tensor(0., device='cuda:0')\n",
      "lindemann tensor(0., device='cuda:0')\n",
      "unprecedented tensor(1., device='cuda:0')\n",
      "dialectal tensor(0., device='cuda:0')\n",
      "spacetime tensor(1., device='cuda:0')\n",
      "enclosures tensor(1., device='cuda:0')\n",
      "briar tensor(0., device='cuda:0')\n",
      "whl tensor(1., device='cuda:0')\n",
      "blueprints tensor(1., device='cuda:0')\n",
      "folks tensor(1., device='cuda:0')\n",
      "shl tensor(1., device='cuda:0')\n",
      "presses tensor(1., device='cuda:0')\n",
      "fallacy tensor(1., device='cuda:0')\n",
      "kailash tensor(0., device='cuda:0')\n",
      "nt tensor(1., device='cuda:0')\n",
      "singaporean tensor(0., device='cuda:0')\n",
      "additional tensor(1., device='cuda:0')\n",
      "alarming tensor(0., device='cuda:0')\n",
      "extradited tensor(1., device='cuda:0')\n",
      "dupage tensor(0., device='cuda:0')\n",
      "spoon tensor(0., device='cuda:0')\n",
      "ibf tensor(1., device='cuda:0')\n",
      "centimeter tensor(1., device='cuda:0')\n",
      "flipper tensor(1., device='cuda:0')\n",
      "conroy tensor(0., device='cuda:0')\n",
      "irc tensor(1., device='cuda:0')\n",
      "south tensor(1., device='cuda:0')\n",
      "tonic tensor(0., device='cuda:0')\n",
      "menswear tensor(0., device='cuda:0')\n",
      "ent tensor(1., device='cuda:0')\n",
      "webster tensor(1., device='cuda:0')\n",
      "moorish tensor(1., device='cuda:0')\n",
      "romans tensor(1., device='cuda:0')\n",
      "necropolis tensor(1., device='cuda:0')\n",
      "gearbox tensor(1., device='cuda:0')\n",
      "scenery tensor(1., device='cuda:0')\n",
      "corrosive tensor(1., device='cuda:0')\n",
      "semester tensor(0., device='cuda:0')\n",
      "prizes tensor(0., device='cuda:0')\n",
      "rupees tensor(1., device='cuda:0')\n",
      "horst tensor(0., device='cuda:0')\n",
      "taranto tensor(0., device='cuda:0')\n",
      "turk tensor(1., device='cuda:0')\n",
      "sancho tensor(1., device='cuda:0')\n",
      "reimbursement tensor(1., device='cuda:0')\n",
      "aca tensor(0., device='cuda:0')\n",
      "bookings tensor(1., device='cuda:0')\n",
      "cdma tensor(0., device='cuda:0')\n",
      "opium tensor(1., device='cuda:0')\n",
      "killed tensor(1., device='cuda:0')\n",
      "trill tensor(0., device='cuda:0')\n",
      "sanders tensor(1., device='cuda:0')\n",
      "favoured tensor(1., device='cuda:0')\n",
      "silhouette tensor(0., device='cuda:0')\n",
      "charlie tensor(1., device='cuda:0')\n",
      "wozniak tensor(1., device='cuda:0')\n",
      "playhouse tensor(0., device='cuda:0')\n",
      "sternum tensor(0., device='cuda:0')\n",
      "stomp tensor(0., device='cuda:0')\n",
      "cow tensor(1., device='cuda:0')\n",
      "motley tensor(1., device='cuda:0')\n",
      "uncanny tensor(1., device='cuda:0')\n",
      "stripping tensor(0., device='cuda:0')\n",
      "apse tensor(0., device='cuda:0')\n",
      "sexy tensor(0., device='cuda:0')\n",
      "terrance tensor(0., device='cuda:0')\n",
      "interfered tensor(1., device='cuda:0')\n",
      "advance tensor(1., device='cuda:0')\n",
      "looks tensor(0., device='cuda:0')\n",
      "mukesh tensor(0., device='cuda:0')\n",
      "aquitaine tensor(0., device='cuda:0')\n",
      "sayings tensor(1., device='cuda:0')\n",
      "toho tensor(1., device='cuda:0')\n",
      "husayn tensor(1., device='cuda:0')\n",
      "rhee tensor(0., device='cuda:0')\n",
      "volk tensor(1., device='cuda:0')\n",
      "watercolors tensor(0., device='cuda:0')\n",
      "configured tensor(0., device='cuda:0')\n",
      "naïve tensor(0., device='cuda:0')\n",
      "lorentz tensor(1., device='cuda:0')\n",
      "rations tensor(1., device='cuda:0')\n",
      "surprising tensor(1., device='cuda:0')\n",
      "smarter tensor(1., device='cuda:0')\n",
      "carlow tensor(1., device='cuda:0')\n",
      "dispose tensor(1., device='cuda:0')\n",
      "pupation tensor(1., device='cuda:0')\n",
      "doctors tensor(0., device='cuda:0')\n",
      "respondents tensor(0., device='cuda:0')\n",
      "augustin tensor(1., device='cuda:0')\n",
      "cotta tensor(1., device='cuda:0')\n",
      "flats tensor(1., device='cuda:0')\n",
      "jointly tensor(1., device='cuda:0')\n",
      "decomposing tensor(0., device='cuda:0')\n",
      "nevada tensor(1., device='cuda:0')\n",
      "communicating tensor(0., device='cuda:0')\n",
      "occurrences tensor(1., device='cuda:0')\n",
      "walmart tensor(1., device='cuda:0')\n",
      "plunging tensor(1., device='cuda:0')\n"
     ]
    }
   ],
   "source": [
    "# test dataloader\n",
    "for i, (word, emb, gender, rgender) in enumerate(train_loader):\n",
    "    for w, gen in zip(word, gender):\n",
    "        print(w, gen.data)\n",
    "    break    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# GAN\n",
    "### Generator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Generator(nn.Module):\n",
    "    def __init__(self, nz):\n",
    "        super().__init__()\n",
    "        self.fc1 = nn.Linear(nz, 300)\n",
    "        self.act = nn.LeakyReLU()\n",
    "        self.model = nn.Sequential(\n",
    "            nn.Linear(600, 1024),\n",
    "            nn.LeakyReLU(),\n",
    "            nn.Linear(1024, 512),\n",
    "            nn.LeakyReLU(),\n",
    "            nn.Linear(512, 300),\n",
    "        )\n",
    "    \n",
    "    def forward(self, z, c):\n",
    "        x = self.act(self.fc1(z))\n",
    "        x = torch.cat([x, c], dim=1)\n",
    "        x = self.model(x)\n",
    "        return x / torch.norm(x) #return normalized embedding\n",
    "\n",
    "# netG = Generator(z_dim)\n",
    "# z = torch.randn((z_dim))\n",
    "# netG(z, torch.Tensor(E['man']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Discriminator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Discriminator(nn.Module):\n",
    "    def __init__(self):\n",
    "        super().__init__()\n",
    "        self.model = nn.Sequential(\n",
    "            nn.Linear(300, 1024),\n",
    "            nn.LeakyReLU(),\n",
    "            nn.Linear(1024, 512),\n",
    "            nn.LeakyReLU(),\n",
    "            nn.Linear(512, 256),\n",
    "            nn.LeakyReLU(),\n",
    "            nn.Linear(256, 1),\n",
    "        )\n",
    "        self.sig = nn.Sigmoid()\n",
    "    \n",
    "    def forward(self, x):\n",
    "        x = self.model(x)\n",
    "        return self.sig(x).squeeze()\n",
    "\n",
    "# d = Discriminator()\n",
    "# d.forward(torch.Tensor(E['test']))    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Hyperparameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "nz = 128\n",
    "lrd = 1e-3\n",
    "lrg = 1e-2\n",
    "beta1 = 0.5\n",
    "epochs = 1000\n",
    "\n",
    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "train_dataset = gender_set(words, E, g, device)\n",
    "train_loader = DataLoader(train_dataset, batch_size=bs, shuffle=True)\n",
    "\n",
    "netG = Generator(nz).to(device)\n",
    "netD = Discriminator().to(device)\n",
    "\n",
    "criterion1 = nn.BCELoss()\n",
    "criterion2 = nn.MSELoss()\n",
    "\n",
    "optimizerD = optim.Adam(netD.parameters(), lr=lrd, betas=(beta1, 0.999))\n",
    "optimizerG = optim.Adam(netG.parameters(), lr=lrg, betas=(beta1, 0.999))\n",
    "\n",
    "z = torch.randn(bs, nz, device=device) #Fixed noise\n",
    "\n",
    "def get_new_emb(word, netG):\n",
    "    return netG(z[0].unsqueeze(dim=0), torch.Tensor(E.v(word)).unsqueeze(dim=0).to(device))\n",
    "\n",
    "def test_generator(netG, g, device):\n",
    "    with torch.no_grad():\n",
    "        word_list = \"nanny engineer warrior boss midwife\".split()\n",
    "        for word in word_list:\n",
    "            orig = torch.Tensor(E.v(word)).to(device)\n",
    "            new = get_new_emb(word, netG).squeeze()\n",
    "            similarity = new.dot(orig)\n",
    "            bias_bef = orig.dot(torch.Tensor(g).to(device))\n",
    "            bias_aft = new.dot(torch.Tensor(g).to(device))        \n",
    "            print(f\"Word: {word} Similarity: {similarity:.5f} | Bias: {bias_bef:.5f} / {bias_aft:.5f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d05392dfe4c34630b5c36e60e5cb539c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=1000), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: [10 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.477 | errG_gender: 1.681 | errG_meaning: -0.39314\n",
      "Word: nanny Similarity: -0.80370 | Bias: 0.17054 / 0.05542\n",
      "Word: engineer Similarity: -0.90484 | Bias: -0.16721 / 0.04978\n",
      "Word: warrior Similarity: -0.87669 | Bias: 0.05222 / 0.04493\n",
      "Word: boss Similarity: -0.87149 | Bias: -0.04070 / 0.04463\n",
      "Word: midwife Similarity: -0.82738 | Bias: 0.22603 / 0.04563\n",
      "Epoch: [20 / 1000]\n",
      "errD_real: 0.020 | errD_fake: 0.359 | errG_gender: 2.781 | errG_meaning: -0.46177\n",
      "Word: nanny Similarity: -0.91903 | Bias: 0.17054 / 0.04824\n",
      "Word: engineer Similarity: -0.95964 | Bias: -0.16721 / 0.02759\n",
      "Word: warrior Similarity: -0.95735 | Bias: 0.05222 / 0.03855\n",
      "Word: boss Similarity: -0.95839 | Bias: -0.04070 / 0.03610\n",
      "Word: midwife Similarity: -0.90989 | Bias: 0.22603 / 0.04021\n",
      "Epoch: [30 / 1000]\n",
      "errD_real: 0.016 | errD_fake: 0.179 | errG_gender: 5.889 | errG_meaning: -0.47133\n",
      "Word: nanny Similarity: -0.93658 | Bias: 0.17054 / 0.04347\n",
      "Word: engineer Similarity: -0.96558 | Bias: -0.16721 / 0.02928\n",
      "Word: warrior Similarity: -0.96857 | Bias: 0.05222 / 0.03573\n",
      "Word: boss Similarity: -0.96976 | Bias: -0.04070 / 0.03031\n",
      "Word: midwife Similarity: -0.92410 | Bias: 0.22603 / 0.04076\n",
      "Epoch: [40 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.245 | errG_gender: 6.777 | errG_meaning: -0.43960\n",
      "Word: nanny Similarity: -0.93209 | Bias: 0.17054 / -0.00714\n",
      "Word: engineer Similarity: -0.95040 | Bias: -0.16721 / -0.00712\n",
      "Word: warrior Similarity: -0.96274 | Bias: 0.05222 / 0.00153\n",
      "Word: boss Similarity: -0.95866 | Bias: -0.04070 / -0.00209\n",
      "Word: midwife Similarity: -0.92693 | Bias: 0.22603 / -0.01246\n",
      "Epoch: [50 / 1000]\n",
      "errD_real: 0.004 | errD_fake: 0.049 | errG_gender: 8.443 | errG_meaning: -0.48206\n",
      "Word: nanny Similarity: -0.92879 | Bias: 0.17054 / 0.04859\n",
      "Word: engineer Similarity: -0.95858 | Bias: -0.16721 / 0.03044\n",
      "Word: warrior Similarity: -0.96159 | Bias: 0.05222 / 0.04365\n",
      "Word: boss Similarity: -0.96471 | Bias: -0.04070 / 0.04528\n",
      "Word: midwife Similarity: -0.91549 | Bias: 0.22603 / 0.04833\n",
      "Epoch: [60 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.154 | errG_gender: 12.083 | errG_meaning: -0.48303\n",
      "Word: nanny Similarity: -0.95411 | Bias: 0.17054 / 0.02479\n",
      "Word: engineer Similarity: -0.97231 | Bias: -0.16721 / 0.02329\n",
      "Word: warrior Similarity: -0.97794 | Bias: 0.05222 / 0.02407\n",
      "Word: boss Similarity: -0.97924 | Bias: -0.04070 / 0.02318\n",
      "Word: midwife Similarity: -0.93714 | Bias: 0.22603 / 0.03162\n",
      "Epoch: [70 / 1000]\n",
      "errD_real: 0.006 | errD_fake: 0.110 | errG_gender: 11.012 | errG_meaning: -0.49074\n",
      "Word: nanny Similarity: -0.95656 | Bias: 0.17054 / 0.02471\n",
      "Word: engineer Similarity: -0.97368 | Bias: -0.16721 / 0.02261\n",
      "Word: warrior Similarity: -0.98035 | Bias: 0.05222 / 0.02851\n",
      "Word: boss Similarity: -0.98254 | Bias: -0.04070 / 0.02430\n",
      "Word: midwife Similarity: -0.94013 | Bias: 0.22603 / 0.03356\n",
      "Epoch: [80 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.069 | errG_gender: 11.927 | errG_meaning: -0.49228\n",
      "Word: nanny Similarity: -0.96069 | Bias: 0.17054 / 0.01781\n",
      "Word: engineer Similarity: -0.97362 | Bias: -0.16721 / 0.01976\n",
      "Word: warrior Similarity: -0.97986 | Bias: 0.05222 / 0.02668\n",
      "Word: boss Similarity: -0.98300 | Bias: -0.04070 / 0.01775\n",
      "Word: midwife Similarity: -0.94555 | Bias: 0.22603 / 0.02336\n",
      "Epoch: [90 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.074 | errG_gender: 12.031 | errG_meaning: -0.49146\n",
      "Word: nanny Similarity: -0.96073 | Bias: 0.17054 / 0.02290\n",
      "Word: engineer Similarity: -0.97410 | Bias: -0.16721 / 0.01601\n",
      "Word: warrior Similarity: -0.98303 | Bias: 0.05222 / 0.02353\n",
      "Word: boss Similarity: -0.98480 | Bias: -0.04070 / 0.02101\n",
      "Word: midwife Similarity: -0.94855 | Bias: 0.22603 / 0.02043\n",
      "Epoch: [100 / 1000]\n",
      "errD_real: 0.001 | errD_fake: 0.123 | errG_gender: 11.945 | errG_meaning: -0.46631\n",
      "Word: nanny Similarity: -0.95400 | Bias: 0.17054 / -0.00367\n",
      "Word: engineer Similarity: -0.96374 | Bias: -0.16721 / -0.00419\n",
      "Word: warrior Similarity: -0.98054 | Bias: 0.05222 / -0.00494\n",
      "Word: boss Similarity: -0.97715 | Bias: -0.04070 / -0.00039\n",
      "Word: midwife Similarity: -0.94999 | Bias: 0.22603 / -0.01208\n",
      "Epoch: [110 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.089 | errG_gender: 13.113 | errG_meaning: -0.50312\n",
      "Word: nanny Similarity: -0.96254 | Bias: 0.17054 / 0.02958\n",
      "Word: engineer Similarity: -0.97635 | Bias: -0.16721 / 0.01786\n",
      "Word: warrior Similarity: -0.98598 | Bias: 0.05222 / 0.02513\n",
      "Word: boss Similarity: -0.98733 | Bias: -0.04070 / 0.02153\n",
      "Word: midwife Similarity: -0.94932 | Bias: 0.22603 / 0.02654\n",
      "Epoch: [120 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.162 | errG_gender: 11.279 | errG_meaning: -0.50101\n",
      "Word: nanny Similarity: -0.96621 | Bias: 0.17054 / 0.01711\n",
      "Word: engineer Similarity: -0.97722 | Bias: -0.16721 / 0.01987\n",
      "Word: warrior Similarity: -0.98873 | Bias: 0.05222 / 0.01792\n",
      "Word: boss Similarity: -0.98790 | Bias: -0.04070 / 0.01973\n",
      "Word: midwife Similarity: -0.95203 | Bias: 0.22603 / 0.02584\n",
      "Epoch: [130 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.039 | errG_gender: 14.138 | errG_meaning: -0.49631\n",
      "Word: nanny Similarity: -0.97378 | Bias: 0.17054 / -0.03241\n",
      "Word: engineer Similarity: -0.97089 | Bias: -0.16721 / -0.01510\n",
      "Word: warrior Similarity: -0.99020 | Bias: 0.05222 / -0.01816\n",
      "Word: boss Similarity: -0.98615 | Bias: -0.04070 / -0.01639\n",
      "Word: midwife Similarity: -0.96478 | Bias: 0.22603 / -0.02637\n",
      "Epoch: [140 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.018 | errG_gender: 14.728 | errG_meaning: -0.49747\n",
      "Word: nanny Similarity: -0.95513 | Bias: 0.17054 / 0.02329\n",
      "Word: engineer Similarity: -0.97014 | Bias: -0.16721 / 0.03347\n",
      "Word: warrior Similarity: -0.98087 | Bias: 0.05222 / 0.02876\n",
      "Word: boss Similarity: -0.98143 | Bias: -0.04070 / 0.03028\n",
      "Word: midwife Similarity: -0.94137 | Bias: 0.22603 / 0.03622\n",
      "Epoch: [150 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.039 | errG_gender: 14.000 | errG_meaning: -0.49934\n",
      "Word: nanny Similarity: -0.96696 | Bias: 0.17054 / -0.03561\n",
      "Word: engineer Similarity: -0.96410 | Bias: -0.16721 / -0.02213\n",
      "Word: warrior Similarity: -0.98629 | Bias: 0.05222 / -0.02426\n",
      "Word: boss Similarity: -0.98246 | Bias: -0.04070 / -0.02197\n",
      "Word: midwife Similarity: -0.95904 | Bias: 0.22603 / -0.03069\n",
      "Epoch: [160 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.019 | errG_gender: 16.917 | errG_meaning: -0.50493\n",
      "Word: nanny Similarity: -0.96991 | Bias: 0.17054 / 0.01203\n",
      "Word: engineer Similarity: -0.97782 | Bias: -0.16721 / 0.01353\n",
      "Word: warrior Similarity: -0.98919 | Bias: 0.05222 / 0.01877\n",
      "Word: boss Similarity: -0.99008 | Bias: -0.04070 / 0.01563\n",
      "Word: midwife Similarity: -0.95658 | Bias: 0.22603 / 0.01912\n",
      "Epoch: [170 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.026 | errG_gender: 14.160 | errG_meaning: -0.50734\n",
      "Word: nanny Similarity: -0.96088 | Bias: 0.17054 / 0.02533\n",
      "Word: engineer Similarity: -0.97360 | Bias: -0.16721 / 0.01972\n",
      "Word: warrior Similarity: -0.98368 | Bias: 0.05222 / 0.03305\n",
      "Word: boss Similarity: -0.98553 | Bias: -0.04070 / 0.02636\n",
      "Word: midwife Similarity: -0.94786 | Bias: 0.22603 / 0.03104\n",
      "Epoch: [180 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.035 | errG_gender: 14.203 | errG_meaning: -0.50394\n",
      "Word: nanny Similarity: -0.97135 | Bias: 0.17054 / 0.00998\n",
      "Word: engineer Similarity: -0.97922 | Bias: -0.16721 / 0.01518\n",
      "Word: warrior Similarity: -0.98948 | Bias: 0.05222 / 0.01877\n",
      "Word: boss Similarity: -0.99141 | Bias: -0.04070 / 0.01647\n",
      "Word: midwife Similarity: -0.95825 | Bias: 0.22603 / 0.01519\n",
      "Epoch: [190 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.111 | errG_gender: 14.007 | errG_meaning: -0.50040\n",
      "Word: nanny Similarity: -0.97536 | Bias: 0.17054 / -0.01283\n",
      "Word: engineer Similarity: -0.97595 | Bias: -0.16721 / -0.01042\n",
      "Word: warrior Similarity: -0.99241 | Bias: 0.05222 / -0.01365\n",
      "Word: boss Similarity: -0.99052 | Bias: -0.04070 / -0.01258\n",
      "Word: midwife Similarity: -0.96425 | Bias: 0.22603 / -0.01004\n",
      "Epoch: [200 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.197 | errG_gender: 16.622 | errG_meaning: -0.50763\n",
      "Word: nanny Similarity: -0.96172 | Bias: 0.17054 / 0.03452\n",
      "Word: engineer Similarity: -0.97586 | Bias: -0.16721 / 0.01799\n",
      "Word: warrior Similarity: -0.98703 | Bias: 0.05222 / 0.02459\n",
      "Word: boss Similarity: -0.98738 | Bias: -0.04070 / 0.02669\n",
      "Word: midwife Similarity: -0.95060 | Bias: 0.22603 / 0.02852\n",
      "Epoch: [210 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.003 | errG_gender: 16.824 | errG_meaning: -0.50272\n",
      "Word: nanny Similarity: -0.97208 | Bias: 0.17054 / -0.03294\n",
      "Word: engineer Similarity: -0.96897 | Bias: -0.16721 / -0.01948\n",
      "Word: warrior Similarity: -0.98949 | Bias: 0.05222 / -0.02570\n",
      "Word: boss Similarity: -0.98518 | Bias: -0.04070 / -0.02046\n",
      "Word: midwife Similarity: -0.96294 | Bias: 0.22603 / -0.03179\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: [220 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.102 | errG_gender: 16.276 | errG_meaning: -0.50822\n",
      "Word: nanny Similarity: -0.97229 | Bias: 0.17054 / 0.01148\n",
      "Word: engineer Similarity: -0.98055 | Bias: -0.16721 / 0.01693\n",
      "Word: warrior Similarity: -0.99134 | Bias: 0.05222 / 0.01723\n",
      "Word: boss Similarity: -0.99220 | Bias: -0.04070 / 0.01830\n",
      "Word: midwife Similarity: -0.96008 | Bias: 0.22603 / 0.01514\n",
      "Epoch: [230 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.008 | errG_gender: 15.157 | errG_meaning: -0.49117\n",
      "Word: nanny Similarity: -0.97124 | Bias: 0.17054 / -0.00814\n",
      "Word: engineer Similarity: -0.97269 | Bias: -0.16721 / -0.00358\n",
      "Word: warrior Similarity: -0.98983 | Bias: 0.05222 / -0.00331\n",
      "Word: boss Similarity: -0.98815 | Bias: -0.04070 / -0.00476\n",
      "Word: midwife Similarity: -0.95918 | Bias: 0.22603 / -0.00588\n",
      "Epoch: [240 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.102 | errG_gender: 14.341 | errG_meaning: -0.50824\n",
      "Word: nanny Similarity: -0.97272 | Bias: 0.17054 / 0.01607\n",
      "Word: engineer Similarity: -0.97960 | Bias: -0.16721 / 0.01043\n",
      "Word: warrior Similarity: -0.99226 | Bias: 0.05222 / 0.01222\n",
      "Word: boss Similarity: -0.99194 | Bias: -0.04070 / 0.01414\n",
      "Word: midwife Similarity: -0.96004 | Bias: 0.22603 / 0.01748\n",
      "Epoch: [250 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.004 | errG_gender: 16.031 | errG_meaning: -0.50997\n",
      "Word: nanny Similarity: -0.97348 | Bias: 0.17054 / 0.01574\n",
      "Word: engineer Similarity: -0.98026 | Bias: -0.16721 / 0.01515\n",
      "Word: warrior Similarity: -0.99250 | Bias: 0.05222 / 0.01432\n",
      "Word: boss Similarity: -0.99240 | Bias: -0.04070 / 0.01626\n",
      "Word: midwife Similarity: -0.96168 | Bias: 0.22603 / 0.01147\n",
      "Epoch: [260 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.004 | errG_gender: 18.461 | errG_meaning: -0.50703\n",
      "Word: nanny Similarity: -0.97747 | Bias: 0.17054 / -0.01208\n",
      "Word: engineer Similarity: -0.97668 | Bias: -0.16721 / -0.01021\n",
      "Word: warrior Similarity: -0.99328 | Bias: 0.05222 / -0.01378\n",
      "Word: boss Similarity: -0.99093 | Bias: -0.04070 / -0.01277\n",
      "Word: midwife Similarity: -0.96593 | Bias: 0.22603 / -0.01190\n",
      "Epoch: [270 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.109 | errG_gender: 20.431 | errG_meaning: -0.50856\n",
      "Word: nanny Similarity: -0.97426 | Bias: 0.17054 / 0.01313\n",
      "Word: engineer Similarity: -0.98051 | Bias: -0.16721 / 0.01189\n",
      "Word: warrior Similarity: -0.99319 | Bias: 0.05222 / 0.01183\n",
      "Word: boss Similarity: -0.99258 | Bias: -0.04070 / 0.01283\n",
      "Word: midwife Similarity: -0.96086 | Bias: 0.22603 / 0.01217\n",
      "Epoch: [280 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.037 | errG_gender: 17.521 | errG_meaning: -0.50775\n",
      "Word: nanny Similarity: -0.97638 | Bias: 0.17054 / 0.00218\n",
      "Word: engineer Similarity: -0.98092 | Bias: -0.16721 / 0.00866\n",
      "Word: warrior Similarity: -0.99239 | Bias: 0.05222 / 0.00942\n",
      "Word: boss Similarity: -0.99250 | Bias: -0.04070 / 0.00910\n",
      "Word: midwife Similarity: -0.96248 | Bias: 0.22603 / 0.00935\n",
      "Epoch: [290 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.019 | errG_gender: 16.870 | errG_meaning: -0.51149\n",
      "Word: nanny Similarity: -0.97983 | Bias: 0.17054 / -0.01811\n",
      "Word: engineer Similarity: -0.97835 | Bias: -0.16721 / -0.00749\n",
      "Word: warrior Similarity: -0.99417 | Bias: 0.05222 / -0.01287\n",
      "Word: boss Similarity: -0.99165 | Bias: -0.04070 / -0.00899\n",
      "Word: midwife Similarity: -0.96711 | Bias: 0.22603 / -0.01516\n",
      "Epoch: [300 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.116 | errG_gender: 17.288 | errG_meaning: -0.51086\n",
      "Word: nanny Similarity: -0.97538 | Bias: 0.17054 / 0.01114\n",
      "Word: engineer Similarity: -0.98161 | Bias: -0.16721 / 0.01494\n",
      "Word: warrior Similarity: -0.99234 | Bias: 0.05222 / 0.01129\n",
      "Word: boss Similarity: -0.99140 | Bias: -0.04070 / 0.01287\n",
      "Word: midwife Similarity: -0.96090 | Bias: 0.22603 / 0.01499\n",
      "Epoch: [310 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.002 | errG_gender: 17.368 | errG_meaning: -0.50648\n",
      "Word: nanny Similarity: -0.97885 | Bias: 0.17054 / -0.01622\n",
      "Word: engineer Similarity: -0.97774 | Bias: -0.16721 / -0.01197\n",
      "Word: warrior Similarity: -0.99402 | Bias: 0.05222 / -0.01360\n",
      "Word: boss Similarity: -0.99112 | Bias: -0.04070 / -0.01203\n",
      "Word: midwife Similarity: -0.96749 | Bias: 0.22603 / -0.01275\n",
      "Epoch: [320 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.014 | errG_gender: 16.117 | errG_meaning: -0.51237\n",
      "Word: nanny Similarity: -0.97887 | Bias: 0.17054 / -0.01140\n",
      "Word: engineer Similarity: -0.97694 | Bias: -0.16721 / -0.01168\n",
      "Word: warrior Similarity: -0.99309 | Bias: 0.05222 / -0.00994\n",
      "Word: boss Similarity: -0.99043 | Bias: -0.04070 / -0.00764\n",
      "Word: midwife Similarity: -0.96826 | Bias: 0.22603 / -0.01404\n",
      "Epoch: [330 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.015 | errG_gender: 18.343 | errG_meaning: -0.51014\n",
      "Word: nanny Similarity: -0.97869 | Bias: 0.17054 / -0.00922\n",
      "Word: engineer Similarity: -0.97913 | Bias: -0.16721 / -0.00714\n",
      "Word: warrior Similarity: -0.99414 | Bias: 0.05222 / -0.01017\n",
      "Word: boss Similarity: -0.99194 | Bias: -0.04070 / -0.00540\n",
      "Word: midwife Similarity: -0.96705 | Bias: 0.22603 / -0.00760\n",
      "Epoch: [340 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.118 | errG_gender: 19.269 | errG_meaning: -0.51365\n",
      "Word: nanny Similarity: -0.97439 | Bias: 0.17054 / 0.01516\n",
      "Word: engineer Similarity: -0.98212 | Bias: -0.16721 / 0.00775\n",
      "Word: warrior Similarity: -0.99327 | Bias: 0.05222 / 0.00975\n",
      "Word: boss Similarity: -0.99225 | Bias: -0.04070 / 0.01320\n",
      "Word: midwife Similarity: -0.96260 | Bias: 0.22603 / 0.01146\n",
      "Epoch: [350 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 17.537 | errG_meaning: -0.51262\n",
      "Word: nanny Similarity: -0.97551 | Bias: 0.17054 / 0.01244\n",
      "Word: engineer Similarity: -0.98128 | Bias: -0.16721 / 0.00870\n",
      "Word: warrior Similarity: -0.99336 | Bias: 0.05222 / 0.01056\n",
      "Word: boss Similarity: -0.99289 | Bias: -0.04070 / 0.01128\n",
      "Word: midwife Similarity: -0.96277 | Bias: 0.22603 / 0.01287\n",
      "Epoch: [360 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 18.243 | errG_meaning: -0.51272\n",
      "Word: nanny Similarity: -0.97956 | Bias: 0.17054 / -0.01411\n",
      "Word: engineer Similarity: -0.97867 | Bias: -0.16721 / -0.00972\n",
      "Word: warrior Similarity: -0.99457 | Bias: 0.05222 / -0.00780\n",
      "Word: boss Similarity: -0.99127 | Bias: -0.04070 / -0.00659\n",
      "Word: midwife Similarity: -0.96804 | Bias: 0.22603 / -0.01132\n",
      "Epoch: [370 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.002 | errG_gender: 19.341 | errG_meaning: -0.51325\n",
      "Word: nanny Similarity: -0.98097 | Bias: 0.17054 / -0.01580\n",
      "Word: engineer Similarity: -0.97961 | Bias: -0.16721 / -0.00807\n",
      "Word: warrior Similarity: -0.99508 | Bias: 0.05222 / -0.01297\n",
      "Word: boss Similarity: -0.99181 | Bias: -0.04070 / -0.01338\n",
      "Word: midwife Similarity: -0.96881 | Bias: 0.22603 / -0.01040\n",
      "Epoch: [380 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.013 | errG_gender: 17.155 | errG_meaning: -0.51426\n",
      "Word: nanny Similarity: -0.97568 | Bias: 0.17054 / 0.00927\n",
      "Word: engineer Similarity: -0.98205 | Bias: -0.16721 / 0.00674\n",
      "Word: warrior Similarity: -0.99383 | Bias: 0.05222 / 0.00878\n",
      "Word: boss Similarity: -0.99308 | Bias: -0.04070 / 0.01002\n",
      "Word: midwife Similarity: -0.96299 | Bias: 0.22603 / 0.01012\n",
      "Epoch: [390 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.009 | errG_gender: 19.395 | errG_meaning: -0.51416\n",
      "Word: nanny Similarity: -0.97963 | Bias: 0.17054 / -0.01107\n",
      "Word: engineer Similarity: -0.97964 | Bias: -0.16721 / -0.00853\n",
      "Word: warrior Similarity: -0.99421 | Bias: 0.05222 / -0.00766\n",
      "Word: boss Similarity: -0.99262 | Bias: -0.04070 / -0.00694\n",
      "Word: midwife Similarity: -0.96907 | Bias: 0.22603 / -0.01210\n",
      "Epoch: [400 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.001 | errG_gender: 20.360 | errG_meaning: -0.51502\n",
      "Word: nanny Similarity: -0.97756 | Bias: 0.17054 / 0.00427\n",
      "Word: engineer Similarity: -0.98204 | Bias: -0.16721 / 0.01082\n",
      "Word: warrior Similarity: -0.99392 | Bias: 0.05222 / 0.01002\n",
      "Word: boss Similarity: -0.99376 | Bias: -0.04070 / 0.00828\n",
      "Word: midwife Similarity: -0.96508 | Bias: 0.22603 / 0.00617\n",
      "Epoch: [410 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.019 | errG_gender: 18.235 | errG_meaning: -0.51656\n",
      "Word: nanny Similarity: -0.98150 | Bias: 0.17054 / -0.01571\n",
      "Word: engineer Similarity: -0.97958 | Bias: -0.16721 / -0.01029\n",
      "Word: warrior Similarity: -0.99535 | Bias: 0.05222 / -0.00757\n",
      "Word: boss Similarity: -0.99380 | Bias: -0.04070 / -0.00535\n",
      "Word: midwife Similarity: -0.97006 | Bias: 0.22603 / -0.01508\n",
      "Epoch: [420 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 18.824 | errG_meaning: -0.51733\n",
      "Word: nanny Similarity: -0.97807 | Bias: 0.17054 / 0.00540\n",
      "Word: engineer Similarity: -0.98360 | Bias: -0.16721 / 0.00961\n",
      "Word: warrior Similarity: -0.99449 | Bias: 0.05222 / 0.00968\n",
      "Word: boss Similarity: -0.99501 | Bias: -0.04070 / 0.00730\n",
      "Word: midwife Similarity: -0.96513 | Bias: 0.22603 / 0.00836\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: [430 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.109 | errG_gender: 21.304 | errG_meaning: -0.51446\n",
      "Word: nanny Similarity: -0.98050 | Bias: 0.17054 / -0.01588\n",
      "Word: engineer Similarity: -0.97941 | Bias: -0.16721 / -0.01162\n",
      "Word: warrior Similarity: -0.99512 | Bias: 0.05222 / -0.00445\n",
      "Word: boss Similarity: -0.99266 | Bias: -0.04070 / -0.00786\n",
      "Word: midwife Similarity: -0.97029 | Bias: 0.22603 / -0.01369\n",
      "Epoch: [440 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.083 | errG_gender: 18.139 | errG_meaning: -0.51560\n",
      "Word: nanny Similarity: -0.98067 | Bias: 0.17054 / -0.01079\n",
      "Word: engineer Similarity: -0.98083 | Bias: -0.16721 / -0.00433\n",
      "Word: warrior Similarity: -0.99522 | Bias: 0.05222 / -0.00317\n",
      "Word: boss Similarity: -0.99450 | Bias: -0.04070 / -0.00564\n",
      "Word: midwife Similarity: -0.97035 | Bias: 0.22603 / -0.01029\n",
      "Epoch: [450 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.008 | errG_gender: 17.905 | errG_meaning: -0.51695\n",
      "Word: nanny Similarity: -0.98067 | Bias: 0.17054 / -0.01180\n",
      "Word: engineer Similarity: -0.98007 | Bias: -0.16721 / -0.00779\n",
      "Word: warrior Similarity: -0.99580 | Bias: 0.05222 / -0.00832\n",
      "Word: boss Similarity: -0.99391 | Bias: -0.04070 / -0.00792\n",
      "Word: midwife Similarity: -0.97038 | Bias: 0.22603 / -0.01192\n",
      "Epoch: [460 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.011 | errG_gender: 20.965 | errG_meaning: -0.51727\n",
      "Word: nanny Similarity: -0.97996 | Bias: 0.17054 / -0.00796\n",
      "Word: engineer Similarity: -0.98106 | Bias: -0.16721 / -0.00612\n",
      "Word: warrior Similarity: -0.99507 | Bias: 0.05222 / -0.01397\n",
      "Word: boss Similarity: -0.99356 | Bias: -0.04070 / -0.00662\n",
      "Word: midwife Similarity: -0.97024 | Bias: 0.22603 / -0.01269\n",
      "Epoch: [470 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.017 | errG_gender: 21.594 | errG_meaning: -0.51606\n",
      "Word: nanny Similarity: -0.98046 | Bias: 0.17054 / -0.00769\n",
      "Word: engineer Similarity: -0.98103 | Bias: -0.16721 / -0.00653\n",
      "Word: warrior Similarity: -0.99594 | Bias: 0.05222 / -0.01127\n",
      "Word: boss Similarity: -0.99423 | Bias: -0.04070 / -0.00886\n",
      "Word: midwife Similarity: -0.97058 | Bias: 0.22603 / -0.00967\n",
      "Epoch: [480 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.001 | errG_gender: 18.959 | errG_meaning: -0.51791\n",
      "Word: nanny Similarity: -0.97764 | Bias: 0.17054 / 0.00816\n",
      "Word: engineer Similarity: -0.98317 | Bias: -0.16721 / 0.00573\n",
      "Word: warrior Similarity: -0.99496 | Bias: 0.05222 / 0.00646\n",
      "Word: boss Similarity: -0.99506 | Bias: -0.04070 / 0.00534\n",
      "Word: midwife Similarity: -0.96630 | Bias: 0.22603 / 0.00968\n",
      "Epoch: [490 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.004 | errG_gender: 21.511 | errG_meaning: -0.51707\n",
      "Word: nanny Similarity: -0.98090 | Bias: 0.17054 / -0.00914\n",
      "Word: engineer Similarity: -0.98102 | Bias: -0.16721 / -0.00748\n",
      "Word: warrior Similarity: -0.99571 | Bias: 0.05222 / -0.00772\n",
      "Word: boss Similarity: -0.99433 | Bias: -0.04070 / -0.01170\n",
      "Word: midwife Similarity: -0.97026 | Bias: 0.22603 / -0.00703\n",
      "Epoch: [500 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.055 | errG_gender: 18.560 | errG_meaning: -0.51804\n",
      "Word: nanny Similarity: -0.97691 | Bias: 0.17054 / 0.01150\n",
      "Word: engineer Similarity: -0.98444 | Bias: -0.16721 / 0.00816\n",
      "Word: warrior Similarity: -0.99497 | Bias: 0.05222 / 0.00877\n",
      "Word: boss Similarity: -0.99565 | Bias: -0.04070 / 0.00711\n",
      "Word: midwife Similarity: -0.96626 | Bias: 0.22603 / 0.00972\n",
      "Epoch: [510 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.049 | errG_gender: 18.816 | errG_meaning: -0.51733\n",
      "Word: nanny Similarity: -0.98039 | Bias: 0.17054 / -0.00778\n",
      "Word: engineer Similarity: -0.98212 | Bias: -0.16721 / -0.00231\n",
      "Word: warrior Similarity: -0.99574 | Bias: 0.05222 / -0.00424\n",
      "Word: boss Similarity: -0.99357 | Bias: -0.04070 / -0.01020\n",
      "Word: midwife Similarity: -0.97017 | Bias: 0.22603 / -0.00737\n",
      "Epoch: [520 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 19.958 | errG_meaning: -0.51749\n",
      "Word: nanny Similarity: -0.97932 | Bias: 0.17054 / 0.00251\n",
      "Word: engineer Similarity: -0.98346 | Bias: -0.16721 / 0.00510\n",
      "Word: warrior Similarity: -0.99459 | Bias: 0.05222 / 0.00919\n",
      "Word: boss Similarity: -0.99479 | Bias: -0.04070 / 0.00793\n",
      "Word: midwife Similarity: -0.96606 | Bias: 0.22603 / 0.00796\n",
      "Epoch: [530 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 19.852 | errG_meaning: -0.51630\n",
      "Word: nanny Similarity: -0.98151 | Bias: 0.17054 / -0.01312\n",
      "Word: engineer Similarity: -0.98160 | Bias: -0.16721 / -0.00547\n",
      "Word: warrior Similarity: -0.99548 | Bias: 0.05222 / -0.00390\n",
      "Word: boss Similarity: -0.99429 | Bias: -0.04070 / -0.00993\n",
      "Word: midwife Similarity: -0.97019 | Bias: 0.22603 / -0.00864\n",
      "Epoch: [540 / 1000]\n",
      "errD_real: 0.001 | errD_fake: 0.051 | errG_gender: 18.513 | errG_meaning: -0.51958\n",
      "Word: nanny Similarity: -0.97818 | Bias: 0.17054 / 0.00980\n",
      "Word: engineer Similarity: -0.98372 | Bias: -0.16721 / 0.00377\n",
      "Word: warrior Similarity: -0.99567 | Bias: 0.05222 / 0.00378\n",
      "Word: boss Similarity: -0.99556 | Bias: -0.04070 / 0.01270\n",
      "Word: midwife Similarity: -0.96881 | Bias: 0.22603 / 0.00508\n",
      "Epoch: [550 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.014 | errG_gender: 20.764 | errG_meaning: -0.51946\n",
      "Word: nanny Similarity: -0.97974 | Bias: 0.17054 / -0.00951\n",
      "Word: engineer Similarity: -0.98073 | Bias: -0.16721 / -0.00439\n",
      "Word: warrior Similarity: -0.99424 | Bias: 0.05222 / -0.00890\n",
      "Word: boss Similarity: -0.99348 | Bias: -0.04070 / -0.00959\n",
      "Word: midwife Similarity: -0.96841 | Bias: 0.22603 / -0.00791\n",
      "Epoch: [560 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.050 | errG_gender: 18.774 | errG_meaning: -0.51826\n",
      "Word: nanny Similarity: -0.98105 | Bias: 0.17054 / -0.00861\n",
      "Word: engineer Similarity: -0.98203 | Bias: -0.16721 / -0.00474\n",
      "Word: warrior Similarity: -0.99587 | Bias: 0.05222 / -0.00405\n",
      "Word: boss Similarity: -0.99500 | Bias: -0.04070 / -0.00814\n",
      "Word: midwife Similarity: -0.97087 | Bias: 0.22603 / -0.00695\n",
      "Epoch: [570 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.001 | errG_gender: 18.734 | errG_meaning: -0.51819\n",
      "Word: nanny Similarity: -0.98153 | Bias: 0.17054 / -0.00513\n",
      "Word: engineer Similarity: -0.98275 | Bias: -0.16721 / -0.00358\n",
      "Word: warrior Similarity: -0.99610 | Bias: 0.05222 / -0.00535\n",
      "Word: boss Similarity: -0.99507 | Bias: -0.04070 / -0.00381\n",
      "Word: midwife Similarity: -0.97116 | Bias: 0.22603 / -0.00851\n",
      "Epoch: [580 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.016 | errG_gender: 19.424 | errG_meaning: -0.51984\n",
      "Word: nanny Similarity: -0.98221 | Bias: 0.17054 / -0.00768\n",
      "Word: engineer Similarity: -0.98279 | Bias: -0.16721 / -0.00247\n",
      "Word: warrior Similarity: -0.99614 | Bias: 0.05222 / -0.00363\n",
      "Word: boss Similarity: -0.99552 | Bias: -0.04070 / 0.00142\n",
      "Word: midwife Similarity: -0.97134 | Bias: 0.22603 / -0.00687\n",
      "Epoch: [590 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.013 | errG_gender: 18.550 | errG_meaning: -0.51802\n",
      "Word: nanny Similarity: -0.97852 | Bias: 0.17054 / 0.01064\n",
      "Word: engineer Similarity: -0.98375 | Bias: -0.16721 / 0.00283\n",
      "Word: warrior Similarity: -0.99583 | Bias: 0.05222 / 0.00367\n",
      "Word: boss Similarity: -0.99544 | Bias: -0.04070 / 0.00814\n",
      "Word: midwife Similarity: -0.96874 | Bias: 0.22603 / 0.00595\n",
      "Epoch: [600 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.003 | errG_gender: 20.340 | errG_meaning: -0.51979\n",
      "Word: nanny Similarity: -0.97979 | Bias: 0.17054 / 0.00230\n",
      "Word: engineer Similarity: -0.98393 | Bias: -0.16721 / 0.00642\n",
      "Word: warrior Similarity: -0.99549 | Bias: 0.05222 / 0.00881\n",
      "Word: boss Similarity: -0.99562 | Bias: -0.04070 / 0.00450\n",
      "Word: midwife Similarity: -0.96621 | Bias: 0.22603 / 0.01125\n",
      "Epoch: [610 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.002 | errG_gender: 22.224 | errG_meaning: -0.51827\n",
      "Word: nanny Similarity: -0.97864 | Bias: 0.17054 / 0.00655\n",
      "Word: engineer Similarity: -0.98372 | Bias: -0.16721 / 0.00661\n",
      "Word: warrior Similarity: -0.99542 | Bias: 0.05222 / 0.00483\n",
      "Word: boss Similarity: -0.99530 | Bias: -0.04070 / 0.00552\n",
      "Word: midwife Similarity: -0.96681 | Bias: 0.22603 / 0.00858\n",
      "Epoch: [620 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.001 | errG_gender: 18.928 | errG_meaning: -0.51970\n",
      "Word: nanny Similarity: -0.98298 | Bias: 0.17054 / -0.01063\n",
      "Word: engineer Similarity: -0.98375 | Bias: -0.16721 / -0.00023\n",
      "Word: warrior Similarity: -0.99673 | Bias: 0.05222 / -0.00515\n",
      "Word: boss Similarity: -0.99532 | Bias: -0.04070 / -0.00907\n",
      "Word: midwife Similarity: -0.97126 | Bias: 0.22603 / -0.00615\n",
      "Epoch: [630 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.001 | errG_gender: 21.225 | errG_meaning: -0.51828\n",
      "Word: nanny Similarity: -0.98133 | Bias: 0.17054 / -0.00539\n",
      "Word: engineer Similarity: -0.98293 | Bias: -0.16721 / -0.00313\n",
      "Word: warrior Similarity: -0.99635 | Bias: 0.05222 / -0.00796\n",
      "Word: boss Similarity: -0.99514 | Bias: -0.04070 / -0.00850\n",
      "Word: midwife Similarity: -0.97116 | Bias: 0.22603 / -0.00437\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: [640 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.008 | errG_gender: 19.687 | errG_meaning: -0.52007\n",
      "Word: nanny Similarity: -0.98011 | Bias: 0.17054 / 0.00216\n",
      "Word: engineer Similarity: -0.98374 | Bias: -0.16721 / 0.00382\n",
      "Word: warrior Similarity: -0.99614 | Bias: 0.05222 / 0.00122\n",
      "Word: boss Similarity: -0.99649 | Bias: -0.04070 / 0.00479\n",
      "Word: midwife Similarity: -0.97026 | Bias: 0.22603 / 0.00083\n",
      "Epoch: [650 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.018 | errG_gender: 20.233 | errG_meaning: -0.52091\n",
      "Word: nanny Similarity: -0.97973 | Bias: 0.17054 / 0.00837\n",
      "Word: engineer Similarity: -0.98426 | Bias: -0.16721 / 0.00523\n",
      "Word: warrior Similarity: -0.99580 | Bias: 0.05222 / 0.00533\n",
      "Word: boss Similarity: -0.99531 | Bias: -0.04070 / 0.00399\n",
      "Word: midwife Similarity: -0.96795 | Bias: 0.22603 / 0.00692\n",
      "Epoch: [660 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 21.255 | errG_meaning: -0.52039\n",
      "Word: nanny Similarity: -0.98244 | Bias: 0.17054 / -0.00596\n",
      "Word: engineer Similarity: -0.98296 | Bias: -0.16721 / -0.00317\n",
      "Word: warrior Similarity: -0.99681 | Bias: 0.05222 / -0.00872\n",
      "Word: boss Similarity: -0.99543 | Bias: -0.04070 / -0.00863\n",
      "Word: midwife Similarity: -0.97202 | Bias: 0.22603 / -0.00630\n",
      "Epoch: [670 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.109 | errG_gender: 20.160 | errG_meaning: -0.51918\n",
      "Word: nanny Similarity: -0.98201 | Bias: 0.17054 / -0.01000\n",
      "Word: engineer Similarity: -0.98244 | Bias: -0.16721 / -0.00455\n",
      "Word: warrior Similarity: -0.99635 | Bias: 0.05222 / -0.00608\n",
      "Word: boss Similarity: -0.99466 | Bias: -0.04070 / -0.00732\n",
      "Word: midwife Similarity: -0.97123 | Bias: 0.22603 / -0.00682\n",
      "Epoch: [680 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 21.489 | errG_meaning: -0.52189\n",
      "Word: nanny Similarity: -0.98083 | Bias: 0.17054 / 0.00480\n",
      "Word: engineer Similarity: -0.98435 | Bias: -0.16721 / 0.00323\n",
      "Word: warrior Similarity: -0.99607 | Bias: 0.05222 / 0.00688\n",
      "Word: boss Similarity: -0.99656 | Bias: -0.04070 / 0.00075\n",
      "Word: midwife Similarity: -0.97012 | Bias: 0.22603 / 0.00270\n",
      "Epoch: [690 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 22.908 | errG_meaning: -0.52138\n",
      "Word: nanny Similarity: -0.98035 | Bias: 0.17054 / 0.00652\n",
      "Word: engineer Similarity: -0.98403 | Bias: -0.16721 / 0.00255\n",
      "Word: warrior Similarity: -0.99583 | Bias: 0.05222 / 0.00643\n",
      "Word: boss Similarity: -0.99602 | Bias: -0.04070 / 0.00271\n",
      "Word: midwife Similarity: -0.97034 | Bias: 0.22603 / 0.00214\n",
      "Epoch: [700 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 23.080 | errG_meaning: -0.52115\n",
      "Word: nanny Similarity: -0.98220 | Bias: 0.17054 / -0.00622\n",
      "Word: engineer Similarity: -0.98237 | Bias: -0.16721 / -0.00713\n",
      "Word: warrior Similarity: -0.99616 | Bias: 0.05222 / -0.00844\n",
      "Word: boss Similarity: -0.99619 | Bias: -0.04070 / -0.00374\n",
      "Word: midwife Similarity: -0.97200 | Bias: 0.22603 / -0.01235\n",
      "Epoch: [710 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 20.110 | errG_meaning: -0.51909\n",
      "Word: nanny Similarity: -0.97999 | Bias: 0.17054 / 0.00818\n",
      "Word: engineer Similarity: -0.98401 | Bias: -0.16721 / 0.00423\n",
      "Word: warrior Similarity: -0.99615 | Bias: 0.05222 / 0.00517\n",
      "Word: boss Similarity: -0.99608 | Bias: -0.04070 / 0.00512\n",
      "Word: midwife Similarity: -0.96942 | Bias: 0.22603 / 0.00518\n",
      "Epoch: [720 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.044 | errG_gender: 19.784 | errG_meaning: -0.52082\n",
      "Word: nanny Similarity: -0.98055 | Bias: 0.17054 / 0.00405\n",
      "Word: engineer Similarity: -0.98460 | Bias: -0.16721 / 0.00744\n",
      "Word: warrior Similarity: -0.99604 | Bias: 0.05222 / 0.00623\n",
      "Word: boss Similarity: -0.99611 | Bias: -0.04070 / 0.00365\n",
      "Word: midwife Similarity: -0.96868 | Bias: 0.22603 / 0.00790\n",
      "Epoch: [730 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 21.683 | errG_meaning: -0.52173\n",
      "Word: nanny Similarity: -0.98251 | Bias: 0.17054 / -0.00538\n",
      "Word: engineer Similarity: -0.98338 | Bias: -0.16721 / -0.00160\n",
      "Word: warrior Similarity: -0.99660 | Bias: 0.05222 / 0.00068\n",
      "Word: boss Similarity: -0.99660 | Bias: -0.04070 / 0.00008\n",
      "Word: midwife Similarity: -0.97144 | Bias: 0.22603 / -0.00135\n",
      "Epoch: [740 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 26.168 | errG_meaning: -0.52138\n",
      "Word: nanny Similarity: -0.98148 | Bias: 0.17054 / 0.00680\n",
      "Word: engineer Similarity: -0.98514 | Bias: -0.16721 / 0.00684\n",
      "Word: warrior Similarity: -0.99640 | Bias: 0.05222 / 0.00521\n",
      "Word: boss Similarity: -0.99684 | Bias: -0.04070 / 0.00698\n",
      "Word: midwife Similarity: -0.96894 | Bias: 0.22603 / 0.01040\n",
      "Epoch: [750 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.002 | errG_gender: 25.916 | errG_meaning: -0.52034\n",
      "Word: nanny Similarity: -0.98022 | Bias: 0.17054 / 0.00610\n",
      "Word: engineer Similarity: -0.98393 | Bias: -0.16721 / 0.00708\n",
      "Word: warrior Similarity: -0.99579 | Bias: 0.05222 / 0.00509\n",
      "Word: boss Similarity: -0.99659 | Bias: -0.04070 / 0.00342\n",
      "Word: midwife Similarity: -0.96921 | Bias: 0.22603 / 0.00518\n",
      "Epoch: [760 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.002 | errG_gender: 18.026 | errG_meaning: -0.52057\n",
      "Word: nanny Similarity: -0.98002 | Bias: 0.17054 / 0.00878\n",
      "Word: engineer Similarity: -0.98430 | Bias: -0.16721 / 0.00549\n",
      "Word: warrior Similarity: -0.99592 | Bias: 0.05222 / 0.00012\n",
      "Word: boss Similarity: -0.99530 | Bias: -0.04070 / 0.00805\n",
      "Word: midwife Similarity: -0.97055 | Bias: 0.22603 / 0.00180\n",
      "Epoch: [770 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.005 | errG_gender: 15.487 | errG_meaning: -0.52202\n",
      "Word: nanny Similarity: -0.98141 | Bias: 0.17054 / 0.00384\n",
      "Word: engineer Similarity: -0.98475 | Bias: -0.16721 / 0.00440\n",
      "Word: warrior Similarity: -0.99624 | Bias: 0.05222 / 0.00464\n",
      "Word: boss Similarity: -0.99656 | Bias: -0.04070 / 0.00397\n",
      "Word: midwife Similarity: -0.96902 | Bias: 0.22603 / 0.00792\n",
      "Epoch: [780 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.001 | errG_gender: 19.015 | errG_meaning: -0.52229\n",
      "Word: nanny Similarity: -0.98151 | Bias: 0.17054 / 0.00565\n",
      "Word: engineer Similarity: -0.98532 | Bias: -0.16721 / 0.01088\n",
      "Word: warrior Similarity: -0.99584 | Bias: 0.05222 / 0.00742\n",
      "Word: boss Similarity: -0.99669 | Bias: -0.04070 / 0.00175\n",
      "Word: midwife Similarity: -0.96913 | Bias: 0.22603 / 0.00855\n",
      "Epoch: [790 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 21.921 | errG_meaning: -0.52243\n",
      "Word: nanny Similarity: -0.98183 | Bias: 0.17054 / 0.00572\n",
      "Word: engineer Similarity: -0.98460 | Bias: -0.16721 / 0.00357\n",
      "Word: warrior Similarity: -0.99645 | Bias: 0.05222 / 0.00459\n",
      "Word: boss Similarity: -0.99684 | Bias: -0.04070 / 0.00338\n",
      "Word: midwife Similarity: -0.97067 | Bias: 0.22603 / 0.00495\n",
      "Epoch: [800 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 26.772 | errG_meaning: -0.52235\n",
      "Word: nanny Similarity: -0.98213 | Bias: 0.17054 / 0.00308\n",
      "Word: engineer Similarity: -0.98464 | Bias: -0.16721 / 0.00517\n",
      "Word: warrior Similarity: -0.99616 | Bias: 0.05222 / 0.00847\n",
      "Word: boss Similarity: -0.99683 | Bias: -0.04070 / 0.00366\n",
      "Word: midwife Similarity: -0.96920 | Bias: 0.22603 / 0.00761\n",
      "Epoch: [810 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.004 | errG_gender: 20.856 | errG_meaning: -0.52276\n",
      "Word: nanny Similarity: -0.98198 | Bias: 0.17054 / 0.00449\n",
      "Word: engineer Similarity: -0.98466 | Bias: -0.16721 / 0.00247\n",
      "Word: warrior Similarity: -0.99634 | Bias: 0.05222 / 0.00589\n",
      "Word: boss Similarity: -0.99661 | Bias: -0.04070 / 0.00693\n",
      "Word: midwife Similarity: -0.97065 | Bias: 0.22603 / 0.00413\n",
      "Epoch: [820 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.001 | errG_gender: 18.468 | errG_meaning: -0.52175\n",
      "Word: nanny Similarity: -0.98245 | Bias: 0.17054 / 0.00162\n",
      "Word: engineer Similarity: -0.98460 | Bias: -0.16721 / 0.00534\n",
      "Word: warrior Similarity: -0.99670 | Bias: 0.05222 / 0.00316\n",
      "Word: boss Similarity: -0.99646 | Bias: -0.04070 / 0.00399\n",
      "Word: midwife Similarity: -0.97128 | Bias: 0.22603 / 0.00102\n",
      "Epoch: [830 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 23.979 | errG_meaning: -0.52021\n",
      "Word: nanny Similarity: -0.98130 | Bias: 0.17054 / 0.00607\n",
      "Word: engineer Similarity: -0.98464 | Bias: -0.16721 / 0.00750\n",
      "Word: warrior Similarity: -0.99636 | Bias: 0.05222 / 0.00150\n",
      "Word: boss Similarity: -0.99644 | Bias: -0.04070 / 0.00179\n",
      "Word: midwife Similarity: -0.96976 | Bias: 0.22603 / 0.00697\n",
      "Epoch: [840 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.002 | errG_gender: 19.304 | errG_meaning: -0.52259\n",
      "Word: nanny Similarity: -0.98236 | Bias: 0.17054 / 0.00334\n",
      "Word: engineer Similarity: -0.98420 | Bias: -0.16721 / 0.00182\n",
      "Word: warrior Similarity: -0.99593 | Bias: 0.05222 / 0.00671\n",
      "Word: boss Similarity: -0.99666 | Bias: -0.04070 / 0.00280\n",
      "Word: midwife Similarity: -0.97004 | Bias: 0.22603 / 0.00486\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: [850 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 22.644 | errG_meaning: -0.52288\n",
      "Word: nanny Similarity: -0.98367 | Bias: 0.17054 / -0.00480\n",
      "Word: engineer Similarity: -0.98303 | Bias: -0.16721 / -0.00573\n",
      "Word: warrior Similarity: -0.99673 | Bias: 0.05222 / -0.00603\n",
      "Word: boss Similarity: -0.99665 | Bias: -0.04070 / -0.00261\n",
      "Word: midwife Similarity: -0.97262 | Bias: 0.22603 / -0.00574\n",
      "Epoch: [860 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.004 | errG_gender: 22.743 | errG_meaning: -0.52215\n",
      "Word: nanny Similarity: -0.98390 | Bias: 0.17054 / -0.00628\n",
      "Word: engineer Similarity: -0.98329 | Bias: -0.16721 / -0.00301\n",
      "Word: warrior Similarity: -0.99672 | Bias: 0.05222 / -0.00227\n",
      "Word: boss Similarity: -0.99606 | Bias: -0.04070 / -0.00496\n",
      "Word: midwife Similarity: -0.97290 | Bias: 0.22603 / -0.00671\n",
      "Epoch: [870 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 20.119 | errG_meaning: -0.52320\n",
      "Word: nanny Similarity: -0.98270 | Bias: 0.17054 / 0.00165\n",
      "Word: engineer Similarity: -0.98448 | Bias: -0.16721 / 0.00266\n",
      "Word: warrior Similarity: -0.99683 | Bias: 0.05222 / 0.00278\n",
      "Word: boss Similarity: -0.99676 | Bias: -0.04070 / 0.00479\n",
      "Word: midwife Similarity: -0.97167 | Bias: 0.22603 / 0.00057\n",
      "Epoch: [880 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.009 | errG_gender: 19.874 | errG_meaning: -0.52256\n",
      "Word: nanny Similarity: -0.98430 | Bias: 0.17054 / -0.00621\n",
      "Word: engineer Similarity: -0.98330 | Bias: -0.16721 / -0.00514\n",
      "Word: warrior Similarity: -0.99717 | Bias: 0.05222 / -0.00425\n",
      "Word: boss Similarity: -0.99682 | Bias: -0.04070 / -0.00387\n",
      "Word: midwife Similarity: -0.97308 | Bias: 0.22603 / -0.00623\n",
      "Epoch: [890 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.005 | errG_gender: 21.009 | errG_meaning: -0.52300\n",
      "Word: nanny Similarity: -0.98422 | Bias: 0.17054 / -0.00725\n",
      "Word: engineer Similarity: -0.98376 | Bias: -0.16721 / -0.00220\n",
      "Word: warrior Similarity: -0.99696 | Bias: 0.05222 / -0.00503\n",
      "Word: boss Similarity: -0.99658 | Bias: -0.04070 / -0.00677\n",
      "Word: midwife Similarity: -0.97303 | Bias: 0.22603 / -0.00602\n",
      "Epoch: [900 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.007 | errG_gender: 20.737 | errG_meaning: -0.52190\n",
      "Word: nanny Similarity: -0.98263 | Bias: 0.17054 / 0.00265\n",
      "Word: engineer Similarity: -0.98466 | Bias: -0.16721 / 0.00432\n",
      "Word: warrior Similarity: -0.99610 | Bias: 0.05222 / 0.00593\n",
      "Word: boss Similarity: -0.99642 | Bias: -0.04070 / 0.00508\n",
      "Word: midwife Similarity: -0.97051 | Bias: 0.22603 / 0.00556\n",
      "Epoch: [910 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.009 | errG_gender: 19.869 | errG_meaning: -0.52213\n",
      "Word: nanny Similarity: -0.98239 | Bias: 0.17054 / 0.00361\n",
      "Word: engineer Similarity: -0.98434 | Bias: -0.16721 / 0.00175\n",
      "Word: warrior Similarity: -0.99688 | Bias: 0.05222 / 0.00106\n",
      "Word: boss Similarity: -0.99711 | Bias: -0.04070 / 0.00486\n",
      "Word: midwife Similarity: -0.97056 | Bias: 0.22603 / 0.00593\n",
      "Epoch: [920 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 25.550 | errG_meaning: -0.52265\n",
      "Word: nanny Similarity: -0.98315 | Bias: 0.17054 / -0.00460\n",
      "Word: engineer Similarity: -0.98255 | Bias: -0.16721 / -0.00644\n",
      "Word: warrior Similarity: -0.99723 | Bias: 0.05222 / -0.00703\n",
      "Word: boss Similarity: -0.99615 | Bias: -0.04070 / -0.00669\n",
      "Word: midwife Similarity: -0.97324 | Bias: 0.22603 / -0.00924\n",
      "Epoch: [930 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 27.343 | errG_meaning: -0.51903\n",
      "Word: nanny Similarity: -0.98322 | Bias: 0.17054 / -0.00409\n",
      "Word: engineer Similarity: -0.98393 | Bias: -0.16721 / -0.00061\n",
      "Word: warrior Similarity: -0.99680 | Bias: 0.05222 / -0.00394\n",
      "Word: boss Similarity: -0.99662 | Bias: -0.04070 / -0.00436\n",
      "Word: midwife Similarity: -0.97270 | Bias: 0.22603 / -0.00306\n",
      "Epoch: [940 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 19.570 | errG_meaning: -0.52317\n",
      "Word: nanny Similarity: -0.98235 | Bias: 0.17054 / 0.00490\n",
      "Word: engineer Similarity: -0.98493 | Bias: -0.16721 / 0.00641\n",
      "Word: warrior Similarity: -0.99675 | Bias: 0.05222 / 0.00527\n",
      "Word: boss Similarity: -0.99712 | Bias: -0.04070 / 0.00458\n",
      "Word: midwife Similarity: -0.97088 | Bias: 0.22603 / 0.00552\n",
      "Epoch: [950 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.001 | errG_gender: 22.296 | errG_meaning: -0.52256\n",
      "Word: nanny Similarity: -0.98375 | Bias: 0.17054 / -0.00498\n",
      "Word: engineer Similarity: -0.98321 | Bias: -0.16721 / -0.00521\n",
      "Word: warrior Similarity: -0.99705 | Bias: 0.05222 / -0.00267\n",
      "Word: boss Similarity: -0.99630 | Bias: -0.04070 / -0.00366\n",
      "Word: midwife Similarity: -0.97307 | Bias: 0.22603 / -0.00481\n",
      "Epoch: [960 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 25.380 | errG_meaning: -0.52490\n",
      "Word: nanny Similarity: -0.98328 | Bias: 0.17054 / -0.00342\n",
      "Word: engineer Similarity: -0.98346 | Bias: -0.16721 / -0.00148\n",
      "Word: warrior Similarity: -0.99657 | Bias: 0.05222 / -0.00145\n",
      "Word: boss Similarity: -0.99649 | Bias: -0.04070 / -0.00016\n",
      "Word: midwife Similarity: -0.97148 | Bias: 0.22603 / -0.00250\n",
      "Epoch: [970 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.001 | errG_gender: 21.793 | errG_meaning: -0.52290\n",
      "Word: nanny Similarity: -0.98274 | Bias: 0.17054 / 0.00344\n",
      "Word: engineer Similarity: -0.98486 | Bias: -0.16721 / 0.00372\n",
      "Word: warrior Similarity: -0.99710 | Bias: 0.05222 / 0.00151\n",
      "Word: boss Similarity: -0.99741 | Bias: -0.04070 / 0.00516\n",
      "Word: midwife Similarity: -0.97145 | Bias: 0.22603 / 0.00301\n",
      "Epoch: [980 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.000 | errG_gender: 22.053 | errG_meaning: -0.52146\n",
      "Word: nanny Similarity: -0.98155 | Bias: 0.17054 / 0.00736\n",
      "Word: engineer Similarity: -0.98489 | Bias: -0.16721 / 0.00229\n",
      "Word: warrior Similarity: -0.99656 | Bias: 0.05222 / 0.00519\n",
      "Word: boss Similarity: -0.99701 | Bias: -0.04070 / 0.00651\n",
      "Word: midwife Similarity: -0.97102 | Bias: 0.22603 / 0.00501\n",
      "Epoch: [990 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.010 | errG_gender: 21.456 | errG_meaning: -0.52357\n",
      "Word: nanny Similarity: -0.98351 | Bias: 0.17054 / -0.00373\n",
      "Word: engineer Similarity: -0.98229 | Bias: -0.16721 / -0.00724\n",
      "Word: warrior Similarity: -0.99703 | Bias: 0.05222 / -0.00848\n",
      "Word: boss Similarity: -0.99624 | Bias: -0.04070 / -0.00470\n",
      "Word: midwife Similarity: -0.97292 | Bias: 0.22603 / -0.00534\n",
      "Epoch: [1000 / 1000]\n",
      "errD_real: 0.000 | errD_fake: 0.008 | errG_gender: 19.307 | errG_meaning: -0.52259\n",
      "Word: nanny Similarity: -0.98283 | Bias: 0.17054 / 0.00191\n",
      "Word: engineer Similarity: -0.98504 | Bias: -0.16721 / 0.00396\n",
      "Word: warrior Similarity: -0.99721 | Bias: 0.05222 / 0.00235\n",
      "Word: boss Similarity: -0.99726 | Bias: -0.04070 / 0.00330\n",
      "Word: midwife Similarity: -0.97140 | Bias: 0.22603 / 0.00354\n",
      "\n"
     ]
    }
   ],
   "source": [
    "G_losses = []\n",
    "G_Glosses = []\n",
    "G_Mlosses = []\n",
    "D_losses = []\n",
    "D_Flosses = []\n",
    "D_Rlosses = []\n",
    "\n",
    "for epoch in tqdm(range(epochs)):\n",
    "    for i, (word, emb, gender, rgender) in enumerate(train_loader):\n",
    "        #######################\n",
    "        # Update netD\n",
    "        #######################\n",
    "        netD.zero_grad()\n",
    "        outputR = netD(emb)\n",
    "        errD_real = criterion1(outputR, gender) #train discriminator to classify real data as Male or Female\n",
    "        errD_real.backward()\n",
    "        \n",
    "#         noise = z[:emb.shape[0], :] #Fixed noise\n",
    "        noise = torch.randn(emb.shape[0], nz, device=device) #Random noise\n",
    "        \n",
    "        fake = netG(noise, emb)\n",
    "        outputF = netD(fake.detach()) #Detach, we don;t want D to change G's weights\n",
    "        errD_fake = criterion1(outputF, gender) #train discriminator to classify generated data as Male or Female\n",
    "        errD_fake.backward(retain_graph=True)\n",
    "        errD = errD_real + errD_fake\n",
    "        \n",
    "        optimizerD.step()\n",
    "        \n",
    "        ########################\n",
    "        # Update netG\n",
    "        ########################\n",
    "        netG.zero_grad()\n",
    "#         errG_meaning = criterion2(fake, emb)\n",
    "#         M = torch.abs(fake - emb).mean()\n",
    "        M = torch.abs(torch.bmm(fake.unsqueeze(1), emb.unsqueeze(2))).mean()    \n",
    "        G = torch.abs(torch.matmul(fake, torch.Tensor(g).to(device))).mean()\n",
    "        errG_meaning = 15*G - 5*M #We want to maximize M and minimize G\n",
    "        errG_meaning.backward(retain_graph=True)\n",
    "        \n",
    "        errG_gender = criterion1(outputF, rgender) #fooling discriminator = how well has it predicted the opposite label (rgender) = how bad has it predicted \n",
    "        errG_gender.backward()\n",
    "        errG = errG_gender + errG_meaning\n",
    "        \n",
    "        optimizerG.step()\n",
    "        \n",
    "    if (epoch+1) % 10 == 0:\n",
    "        print(f\"Epoch: [{epoch+1} / {epochs}]\")\n",
    "        print(f\"errD_real: {errD_real:.3f} | errD_fake: {errD_fake:.3f} | errG_gender: {errG_gender:.3f} | errG_meaning: {errG_meaning:.5f}\")    \n",
    "        test_generator(netG, g, device)\n",
    "        \n",
    "    D_losses.append(errD.item())    \n",
    "    D_Flosses.append(errD_fake.item())    \n",
    "    D_Rlosses.append(errD_real.item())    \n",
    "\n",
    "    G_losses.append(errG.item())\n",
    "    G_Glosses.append(errG_gender.item())\n",
    "    G_Mlosses.append(errG_meaning.item())        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Models saved\n"
     ]
    }
   ],
   "source": [
    "PATH_D = \"models/D.pth\"\n",
    "PATH_G = \"models/G.pth\"\n",
    "torch.save(netG.state_dict(), PATH_G)\n",
    "torch.save(netD.state_dict(), PATH_D)\n",
    "print(\"Models saved\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,5), dpi=100)\n",
    "plt.title(\"Generator and Discriminator Loss During Training\")\n",
    "plt.plot(G_Glosses, label=\"G - gender loss\")\n",
    "plt.plot(D_Flosses, label=\"D - fake's gender loss\")\n",
    "plt.xlabel(\"iterations\")\n",
    "plt.ylabel(\"Loss\")\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Both are different losses, better plot seperately"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA00AAAHPCAYAAACP5DHWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAAPYQAAD2EBqD+naQAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAIABJREFUeJzsnXecFPX5xz/P7t5egwOk2UBBFLFhT5TYNdiJGk2iiWKisZfEEs0vFkQlRsWIxoZRrMEuFhBsIB2R3o7eywFXubL1+/vjOzP7ndnZ2Zm9vds9eN6v173udsp3vvOdmb3nmc/zPF8SQoBhGIZhGIZhGIaxx5frDjAMwzAMwzAMw+Qz7DQxDMMwDMMwDMM4wE4TwzAMwzAMwzCMA+w0MQzDMAzDMAzDOMBOE8MwDMMwDMMwjAPsNDEMwzAMwzAMwzjAThPDMAzDMAzDMIwD7DQxDMMwDMMwDMM4wE4TwzAMwzAMwzCMA+w0MQzD5BgiOp2IBBGd3gJtP0xEItvtpjnmgdr5DM5imy02Rkx+Q0RriWhUrvvR0hDRDCL6KsN9RxPRsmz3iWGYBOw0MUwbhoh6EdHzRLSciBq0nyVE9B8iOirX/csmRHQ+ET2c637kEiIarDkO+k8TEW0movFEdDsRtc91H9syRFSiOZmnt+IxdWfw1611zExQHGH9J0JEO4hoGhE9TkQ9c93HbEJEoyznm+pnVK77yjBM6xDIdQcYhskMIroQwHsAogDeATAfQBzAoQAuBXATEfUSQqzLXS+zyvkAbgHwcI77kQ88CGANgAIAewM4HcC/AfyViC4WQixQtn0UwD9buX/rABQDiGSxzR+0NsNZbNNKCYCHtL8ntuBx2jL/AzAW8qVrJwAnALgTwB1E9CchxOgWOGZfyO+21uRlAN8on3sBeATAKwAmK8tXZfGYpwLIVBX+AwDKYl8YhrHAThPDtEGI6CAAoyGN07OEEFss6/8G4Ga0vqHhGiIqFULU57gPPgBBIURTLvuRAeOEELOVz8OI6EwAXwD4jIj6CSEaAUAIEYV0rFscIgoA8AkhwgCyOqZCiHi222wt8uFezyJzhBBvqwuI6AAAEwC8QURLhRDzm3sQIiIARUKIRiFEqLnteUUIMR3AdKU/x0M6TdOt52+H2n8Px8z4hYAQIpsvKBiGsYHD8ximbXIvgFIA11odJkAaykKIEUKIDepyIjqUiD4kokottGs2EV1s2UYPARtARMOJaDsR1RPRJ0TU1XosIjqPiCZr29QR0ZdEdLhlm1FEtIuIDiKisURUB6mOgYhOIaIPiGg9EYWIaAMRPUNExer+kCoT1NAYZX0pET2t7RsionIiulszXNR+CC2c8SoiWgwgBODcVINMRIO089mstbuKiB4gIr9lu4lEtIiIDiOi77UwyU1EdK9Nm/sT0afaeFUQ0TMAClP1wS1CiO8ADAVwAIDfK8dLymkionOIaAoRVWvXpZyIHrdsU6Ttu1y7V7YQ0ceaw66Ga91NRHcS0SrI8TyMbHKalHugJxF9of29iYj063okEX2njcs6IrrS0p+knCa3405EQSJ6hIh+IqIa7RiTiegMZZsDAWzXPj6k3GcPK9ucqdzr1UQ0hoj6WY71sLbfYUT0LhFVAZjiePFc4OEed3NtbyOixdp4VZH8HrgSGaKp2YMBBCG/m/Tj2ObTUeI75kBl2VrtvhhIRLMBNAK4QVk3ymb/tN9RROTT+rFZO9/vtWuT1TwpItpK8rv1QiKaA+ngX6Otu147boX2LC0ioj/ZtGHKaSKic7XzHKScQyMRTVDHTtvWlNNE8rteENGtRHQLEa3Wjj2diI62OfaVRLRM22aBdh6cJ8UwCqw0MUzb5EIAK4UQM93uQNKRmQpgE2S4Vj2AKwB8SkSXCSE+sezyHIAqAEMAHAgZgvM8gN8obf4BwBsAxgP4G2R4000AphDRMUKItUp7AW27KQDuBtCgLb9c2+9FADsBnAjgNgD7a+sAGSqzL4BzIMNQ1PMiAJ8BOAPAfwHMAzAQwJMA9gPwF8t5namd9/MAdgBYi9QMBrALwHDt95mQb5vLANxj2bYTgK8AfAzgfQC/BvAEES0UQozT+loM4FsAPQGMALBZO58zHfrghbcAPA7glwBG2m2g3QdfAFgAGeYXAtAHwABlG7+2zVmQiuazANpDjv8RMIckXQugCDJsKQSgEqlfyPkBjIMMtbsXwFUAnieiegCPQTrSHwO4EcCbRDRdCLEmzTmnHXfI63UdZGjZSO1c/gRgPBGdKISYB+kw3QR5H36itQdtnEBEZ2t9Xw0ZIloMeZ9OJaJjLfc6AHwAYAWAv6OZYVNu73GX1/Z6yHvvQ8jrWgTgKAA/A/Bupn0UQkwn6Tifk2kbkGF4/4N83kcCKE+zfdrvKADDIO+1zyG/f/prv4ua0c9UHAngTQAvQJ7DYm35LQBmA/gUUv3/FYBXiQhCiP+6aPchyLDUJwB0hvzueQPAaS72/SPkvfoC5PN3L4CPiOgQIUQMAIjoUshnbw6A+wB0gfwu2eyifYbZcxBC8A//8E8b+oE0AAWAT2zWdYT8h6f/FCvrvoE0pgqVZQTpSC1Xlg3W2v8aACnLh0OGeXXQPreDNFhesfShO4BqdTmAUVqbw2z6XGyz7D5I46Knsux5+ZWVtO0gre3/syz/QGvjIGWZABADcJjLsbbr20uQDqc6jhO1tv+gLAsC2ALgQ2XZHdp2lyvLSiCNawHg9DT90a/N8Q7bVEOGUOmfH1bHDdKwFAC6OLRxrbbNX2zWkfb7QG2bGgBdLdvo6wbb3AP3W+7XBu06/UZZ3lfb9mFl2enWMfIw7n7IMEzrs7IVwH+VZV2sx1XWzQWwDcBeyrKjtPvpDet4A3jX5T2mn9evHbZxdY+7vLafAljkpm8prundadoWAMrs7j2b+/hAZdlabdlAm+3XAhhls3+676jukHl1n1jae0jbf1Sqc7Hpw/HWe9qyfqu2/jSbdXbfI98DWGxZNgPAV8rnc7U25wEoUJbfqy3voywbDWCZ8vlQbZst+vXQll+hLT9bWVYO+SKkRFk2UNtumd358g//7Ik/HJ7HMG2PMu33Lpt1EyHfmOs/eujTXpBqxvsA2hNRFyLqAvnWcjyAg4loP0tbrwgh1NCayZDG5wHa53MgDc//6e1pbcYAzIR8K27lResCocT8ayFIXQBMg3TojrEdATPna8ccYVn+tNbGeZblk4QQS1y0a+1be61vkyEdnUMtm+8C8LaybxjALAC9LX3dAvmWX9+uAVKlyRa7IJWUVFRrvweRzOmy4zJIFe456wrLPQEAHwkhtlu3c+BVpa1qSIOtHvLe1JeXa/3snbR3MmnHXQgR05br4Vp7QSqfswEcm+4ARLQPgKMhjexKpd0FkIb7+Ta7veSi725xe4+7ubbVAPYnohOy2D8d/Tsp0yqOa4QQ4z1sn+476izI6/yCZb+k+zpLLBVCTLIutHyPdNS+RyYB6EdEbhSv/wpzzpJeiMLN8/GuEKI21b5E1BvAIZD3tq7+Q7sOK1y0zzB7DOw0MUzbo0773c5m3Q2QzszvLcv7QBpXQ2F2qrZDhrYAQDfLPustn6u035203wdrv7+zafOXNu1FAWy0dphkjssoIqqENLq2QxoUANDB5hytHABgsxCizrJ8qbJeJV24l9q3w7U8iRoAtVrfdAPd2reNNg5FFRLjpfdlpc126cKQvNAOiXvEjvcg1cVXAWzT8hausBjZBwEoF7KIRDpcjyeAJhsHqwb2Y1cD89ilws24g4iuIaIFkLkmOyGv5QVwf48B9tdpKYAuRFRqWe5lXNwc38097ubaPgH5nM0iohUkpycYgOygfyc53X9OeB2zdN9R+risVDfSHN8qZB/b/hPRaVpOU4N23O2Qahch8RLMiXTn2Zx9bcfIYRnD7LFwThPDtDGEEDVEtAUyt8S6biZgJLWr6EbTU5DKkh3Wf5CxFNvp+Rl6m3+ADE2xYjW4Q0JWQEs0JHNnvgawF6QxtwxSddgPMpyrJV7suKpmRUQdIZ23Wsj8kFWQBvexWl+tfUs3Xi0OEe0P6QSkNHaEEI1EdCqkEngBZAjQbwB8R0S/FFqegwdcVwdD6jFqztil3ZeIfg95P30KmQdUoe13P6SD2BJ4GZes4ObaCiGWElFfyLzIcyFVxZuJ6BEhxEPN7MIRACoUZcPqzOr4Uyz3OmY5f+YsJPWfZKGQryFDo++EfHEUhgy5vBXuvuNa9PlgGMYd7DQxTNvkSwDXaUnss1xsv1r7HRFCfOO4pXv0YgAVzWjzSMjQkGuEEG/qC4nILpk8lQG2DsDZRNTe8ib+UGV9JpwOGb54qRDiB6VvvTJsT+/LEUREFnWkbzPaVNGLZDiGOGnO67faz1+J6O+QhRjOgMx9WwXgZ0RUIHaPUsa/hnwGLlXHnYiGWLZzuscA++t0KIAdomVLiru+x11cW2h9fQ/Ae0QUhCx68X9ENExkWH6fiE6CdEDVctxV2rqOWiimjlX9bSn0cekDRQUios5wp9Jkg0GQ86ldIITYpvTBGjacK9QxstIHeTxtBcO0NhyexzBtk39BJtC/RkTdbdab3iIKISog851u0PIzzBvblBJ3wXhIFebvRFSQYZv6W1BVFSDIgglW6rX1HS3Lx0K+ub7VsvwvkEbwOGSGXd+CkPNfZcpYyCqAv1baLAHw52a0qbdzJoAHII3Ddxy228tm8Tztt176/CPIogjWMdWvT1vD7lr+DMBJlu30nA7TPSZkWf95AK5R7z8iOgIyFHVstjtswdU97ubaag6DgZbrtQRybJKeYzeQnKdpFKSC8qSySn+xcqqybSm0UtytwLeQivdNluVJ93ULYnfvdYalCmiuEEKshsxdGqx9FwEAiGggEiHYDMOAlSaGaZMIIVaQnFflfwDKiegdAPMh/zH3AnAl5BtCNYfoFshy3wuJaCTkm/fukIbj/pCleL30oZaIboIsTTuHiEZDxur3hAwNmor0xskySMPqKa0QRS1kuJDdW+CftN8jiGg8gJgQYjRkKeHvATymhSXOhzRkBwH4txBilU1bbpgG+ab8DSIaAa1KG5oX1jISckzeJKLjIItC/AEJY90t5xHRoZDf4d0hi3ycA/nW+OI0asGDWgjXl9r23SAdwY1IzCf0JoCrAQwnohMhk8dLAZwNmVQ/xmN/c80XAC4F8AkRfQn5jNwI6SwYuYFaeNsSAL8houWQ5dMXCSEWQZZ5HgdgOhH9F4mS4zWQVeKay2XaNbXyBtzf426u7QQi2gr5fG4D0A/ynvzSJmfKjmO1cEcfpHN5AuQzq1cxXKBsOwEyp+a/RPQkpAPxRyS+J1oUIcQ2InoWwF1E9Blkafr+kIUzdiC1sphNvoKcBmAsEb0KGT57A2Q570xeVrUE/wdZiGUyEb0J2a+bIUumpwqlZJg9DnaaGKaNIoQYQ0RHArgL0oD6I6QRsA7SaHpJCDFf2X4JyVntH4Is2dsZMrdjLuTcQ5n04V0i2gxZIvweyLfZmyCN7Ndd7B8hoosgq4LdD5kz9AlkefH5ls0/hqx69VvIQhcEYLQQIk5ygt5HIPM3roUsUXwPZHWxjBBC7CSiC7U2HoV0oN6GfHvtpcKX2mYDEZ2lncdtkM7SO5DG+FdO+1rQr1cY0rBfCJkv8boLw/czyPLRf4RUk3ZA5m49JISo0foZI6LzIY2pKyGN4p3QnG4P/cwXRgHYG9JYHQjpLP0ech6w0y3bXgd5fZ6BLF8+BNJx+oaIztU+PwJZynoSgL+J9HNJueG3KZZPFEJscHmPp722kPMHXQXgr5AO40bI5+9Rl/38nfYThXzJsQLAvyG/b0xFB7Tn+xJIR3soZO7jvyGfpbTfD1nib5DP2fWQTv90yO/LKZDfNy2KEGIhEV0Bef7DIb8fn4XMf0qqJpoLhBAfkJxz7wHIKIZlkM/9rZD5pQzDIDHfBsMwDMMwzG6PFmJZBeAfQojHct2ffIWIlgFYIYS4KNd9YZh8gHOaGIZhGIbZLSGiYpvFd2q/J7ZiV/IWIirQKpmqy86FLHwyMSedYpg8hJUmhmEYhmF2S4hoMGQ48ljI+al+ARleOEEIMTB3PcsftFy6TyFzZLcAOAwy568CwJFKaCfD7NFwThPDMAzDMLsrCyDzr+6FnEh2G2RO0T9y2ak8YztkruINkLlwtZBO1H3sMDFMAlaaGIZhGIZhGIZhHOCcJoZhGIZhGIZhGAfYaWIYhmEYhmEYhnFgt89p0mav3xeAm0n7GIZhGIZhGIbZvWkPYLPwkKe02ztNkA7Txlx3gmEYhmEYhmGYvGF/yAmnXbEnOE11ALBhwwaUlZXlui8MwzAMwzAMw+SI2tpa9OjRA/AYhZZTp4mIbgJwE4ADtUWLATwihBinrS8C8DSA3wIoBDAewM1CiG1ej1VWVsZOE8MwDMMwDMMwnsl1IYiNAO4DcByA4wF8B2AMER2urX8GwEUALgdwGmSo3cc56CfDMAzDMAzDMHsoeTdPExFVArgHwIeQE65dKYT4UFt3KIClAE4SQsxw2V4ZgJqamhpWmhiGYRiGYRhmD6a2thYdOnQAgA5CiFq3++VaaTIgIj8R/RZAKYDpkOpTAYBv9G2EEMsArAdwUk46yTAMwzAMwzDMHkfOC0EQ0ZGQTlIRgF0ALhFCLCGiowGEhRDVll22Adjbob1CyPwnnfZZ7jLDMAzDMAzTDGKxGCKRSK67weyGFBQUwO/3Z73dnDtNAMoBHA2gA4BfA3iDiE5rRnv3A3goGx1jGIZhGIZhsocQAlu3bkV1tfWdOMNkj44dO2LvvfeGnK41O+TcaRJChAGs1D7+REQnALgDwHsAgkTU0aI2dQew1aHJYQCGK5/bg+dpYhiGYRiGyTm6w9StWzeUlJRk1ahlGCEEGhoaUFFRAQDYZ599stZ2zp0mG3yQ4XU/AYgAOAvARwBARH0B9IQM57NFCBECENI/88PIMAzDMAyTe2KxmOEwde7cOdfdYXZTiouLAQAVFRXo1q1b1kL1cj1P0zAA4yCLO7QHcCWA0wEMFELUENF/AQzXKurVAngOwHS3lfMYhmEYhmGY/EDPYSopKclxT5jdHf0ei0Qiu4fTBKAbgDcB7AOgBsACSIfpa239XwDEIZUmY3LbHPSTYRiGYRiGyQIcBcS0NC1xj+XUaRJC/CnN+iYAt2g/DMMwDMMwDMMwrU7ezNPEMAzDMAzDMEzrMnHiRBBRi1Q0PP3003HnnXdmvd1cwE4TwzAMwzAMw6Rh69atuOOOO9CnTx8UFRWhe/fuGDBgAF588UU0NDTkuntMC5PrnCaGYRiGYRiGyWtWr16NAQMGoGPHjnj88cdx5JFHorCwEAsXLsQrr7yC/fbbDxdffHGuu5kzwuEwgsFgrrvRorDSxDAMwzAMw3ji/R83YNTUNbnuRqtx8803IxAIYPbs2bjiiivQr18/9O7dG4MGDcKXX36Jiy66qNnHGDlyJHr06IGSkhJccsklGD58ODp27GjaZsyYMTj22GNRVFSE3r17Y8iQIYhGo8Z6IsKrr76KSy65BCUlJTj44IPx2WefmdoYO3YsDjnkEBQXF+OMM87A2rVrk/oyZcoUnHLKKSguLkaPHj1w++23o76+3lh/4IEHYujQobj66qtRVlaGP//5z67OsaqqCldffTU6deqEkpISnHfeeVixYoWxft26dbjooovQqVMnlJaW4vDDD8fYsWONfa+66ip07doVxcXFOPjgg/H666+7Om42YKeJYRiGYRiGcU08LnDvRwvw8OdLsLm6sVltCSHQEI7m5EcI4aqPO3fuxIQJE3DLLbegtLTUdpvmVmubOnUqbrzxRtxxxx2YN28ezjnnHDz22GOmbSZPnoyrr74ad9xxB5YsWYKXX34Zo0aNStpuyJAhuOKKK7BgwQKcf/75uOqqq1BZWQkA2LBhAy699FJcdNFFmDdvHq677jrcd999pv1XrVqFc889F5dddhkWLFiA9957D1OmTMGtt95q2u6pp55C//79MXfuXDzwwAOuznPw4MGYPXs2PvvsM0yfPh1CCJx//vlGOfpbbrkFoVAIP/zwAxYuXIgnnngC7dq1AwA88MADWLJkCcaNG4elS5fixRdfRJcuXdwPcjPh8DyGYRiGYRjGNZF43Pi7pjGCfTsWZ9xWYySGwx4cn41ueWbJIwNREkxvCq9cuRJCCPTt29e0vEuXLmhqagIgjf0nnngi474899xzOO+883D33XcDAA455BBMmzYNX3zxhbHNkCFDcN999+Gaa64BAPTu3RtDhw7Fvffei4ceesjYbvDgwfjd734HAHj88ccxYsQIzJo1C+eeey5efPFFHHTQQXj66acBAH379jWcE51hw4bhqquuMgo4HHzwwRgxYgROO+00vPjiiygqKgIAnHnmmbjrrrtcn+OKFSvw2WefYerUqTj55JMBAO+88w569OiBTz/9FJdffjnWr1+Pyy67DEceeaRxjjrr16/HMcccg+OPPx6AVLtaE1aaGIZhGIZhGNdEYwmFproh0my1qa0ya9YszJs3D4cffjhCoZDtNpMnT0a7du2Mn3feecd2u/Lycpx44ommZdbP8+fPxyOPPGJq7/rrr8eWLVtMhSiOOuoo4+/S0lKUlZWhoqICALB06VL87Gc/M7V70kknJR1n1KhRpuMMHDgQ8Xgca9YkQjJ158UtS5cuRSAQMB2/c+fO6Nu3L5YuXQoAuP322/Hoo49iwIABeOihh7BgwQJj25tuugmjR4/G0UcfjXvvvRfTpk3zdPzmwkoTwzAMwzAM45poPOE0/W7kDADA5HvPQI+9Sjy3VVzgx5JHBmatb16P7YY+ffqAiFBeXm5arqsgxcWplbbjjz8e8+bNMz537949g55Kdu3ahSFDhuDSSy9NWqerPwBQUFBgWkdEiCvqoJvj3HDDDbj99tuT1vXs2dP4O1WoYnO47rrrMHDgQHz55ZeYMGEChg0bhqeffhq33XYbzjvvPKxbtw5jx47F119/jbPOOgu33HILnnrqqaz3ww52mhiGYRiGYfKUBRurUVzgx8Hd2+e6KwbRWLIBPn3VzoycJiJyFSKXSzp37oxzzjkHzz//PG677TZPzkJxcTH69OmTdru+ffvixx9/NC2zfj722GNRXl7uqr1U9OvXL6kwxIwZM5KOs2TJkmYdJ9Wxo9EoZs6caYTn7dy5E+Xl5TjssMOM7Xr06IEbb7wRN954I+6//36MHDkSt912GwCga9euuOaaa3DNNdfglFNOwT333NNqThOH5zEMwzAMw+QhlfVhXPz8VJzzzA+57oqJWNxdAYXdiRdeeAHRaBTHH3883nvvPSxduhTl5eV4++23sWzZMvj97lSrVNx2220YO3Yshg8fjhUrVuDll1/GuHHjTAUmHnzwQbz55psYMmQIFi9ejKVLl2L06NH4xz/+4fo4N954I1asWIF77rkH5eXlePfddzFq1CjTNn/7298wbdo03HrrrZg3bx5WrFiBMWPGJBWC8MrBBx+MQYMG4frrr8eUKVMwf/58/P73v8d+++2HQYMGAQDuvPNOjB8/HmvWrMGcOXPw/fffo1+/fsb5jxkzBitXrsTixYvxxRdfGOtaA3aaGIZhGIZh8pBNVYlcoXxyVCJ51JfW4qCDDsLcuXNx9tln4/7770f//v1x/PHH47nnnsPdd9+NoUOHNqv9AQMG4KWXXsLw4cPRv39/fPXVV/jLX/5iCrsbOHAgvvjiC0yYMAEnnHACfv7zn+OZZ57BAQcc4Po4PXv2xEcffYRPP/0U/fv3x0svvYTHH3/ctM1RRx2FSZMmYfny5TjllFNwzDHH4MEHH8S+++7brHMEgNdffx3HHXccLrzwQpx00kkQQmDs2LFGSGEsFsMtt9yCfv364dxzz8UhhxyCF154AQAQDAZx//3346ijjsKpp54Kv9+P0aNHN7tPbiG35RbbKkRUBqCmpqYGZWVlue4OwzAMwzCMK+asr8KlL8hk92VDz0WRyxyclmb9zgac+uT3pmX/uuwoXHFCD8f9mpqasGbNGvTq1cvkDDD2XH/99Vi2bBkmT56c6660OZzutdraWnTo0AEAOgghat22md9BpAzDMAzDMHsoapW6aB6pOxEPRQUY9zz11FM455xzUFpainHjxuGNN94wVBYm97DTxDAMwzAMk4eoBRci0ThQmMPOKORTqODuxKxZs/Cvf/0LdXV16N27N0aMGIHrrrsu191iNNhpYhiGYRiGyUNCqtOUR+pOxKZ6HtN83n///Vx3gXGAC0EwDMMwDMPkIU3hmPF3JJY/6k7Upi/5FD7IMC0BO00MwzAMwzB5hhAC8zfWGJ/t5kbKFXYOUjgas9mSYXYf2GliGIZhGIbJM175YTVemrTK+JxPIXF2DlzYQ//ieRRqyOyetMQ9xjlNDMMwDMMwecawcctMn/MpPM+uEEQ4mt5IDQaD8Pl82Lx5M7p27YpgMGiavJVhmosQAuFwGNu3b4fP50MwGMxa2+w0MQzDMAzD5Dl2eUS5wm5yWzdOk8/nQ69evbBlyxZs3ry5JbrGMACAkpIS9OzZEz5f9oLq2GliGIZhGIbJc7yEv7U0MZvQp5DL/gWDQfTs2RPRaBSxGOdBMdnH7/cjEAhkXcVkp4lhGIZhGCbPyadCEHahgpGoeyWMiFBQUICCgoJsdmu3ZUtNI4QA9u1YnOuu7NGw08QwDMMwDJPn5H1OE6tGLUJTJIaThn0HHwHLhp6HYIBruOUKHnmGYRiGYZg8wk5VypfJbSct344566qSlrvJaWK8M39DNQAgLoCaxkiOe7Nnw0oTwzAMwzBMHlHZEE5aFskDp2RzdSOueW2W7Tp2mlqGn9YnHNSmCKt5uYSVJoZhGIZhmDxi565kp8luQtnWxq5fOvlUqGJ3Ys66auPvRnaacgo7TQzDMAzDMHlEQziatCwfJrcN+FNXI2OlqWWoVlTHhjA7TbmEnSaGYRiGYZg8IhSxyWnKg0IQTo5biJ2mFqEpmnCUGtlpyinsNDEMwzAMw+QRdnMe5UPJcSc1iZWmlqFJcaAbI8kKJNN6sNPEMAzDMAyTR+hEnWOhAAAgAElEQVQOyHEHdMJ5R+wNID/C85wco3zo3+6IWvyhMcxjnEvYaWIYhmEYhskj9FC3oN+HgF+aavkQnmengOlwIYiWQVWa7HLdMiEfVMu2CDtNDMMwDMMweYSu6AQDPhRoxRfyQcnh8LzWJ6QoTdkoOb6lphHHPPI1HhyzqNlt7Wmw08QwDMMwDJNH6A5IYcCHAp801fKh5Dg7Ta2PWggiG9XzXp28BnWhKN6cvq7Zbe1psNPEMAzDMAyTR4Q0QzkY8KEgIJWmfHBK2GlqXWJxYQrLzIbT5FQ2nnGGnSaGYRiGYZg8Qg3PCxhKU+6dEqe8pUgeKGG7G9ZwvGyE5wX9bPpnCo8cwzAMwzBMHpEIz/MjGNCcpjwoBGFVky45Zj8MOnpfAIDIffd2O6xOUjaUpgJ2mjKGR45hGIZhGCZPWLezHk9/vRyAzGkK+LTwvDwsBNGjUzFuPr0PAECw15R1mizj3ZgFpUl1mviaeYOdJoZhGIZhmDzh8pemG3/L6nl5pDRZHLeA3wfNp0Pue7f7YVWaGrOiNCVymvKhjH1bgp0mhmEYhmGYPCAeF6ioCxmfC/Os5HjIonzE4gJEsn9xVi2yTpLTlGWlKRvt7UkEct0BhmEYhmGYPZ1nv1mB/05ZbVqWb5PbWsPzGiMxaD4T4lwIIiu8Onk1ttQ04Z6BfVFRGzKty8bktqQUz2uKxNChuKDZbe4psNPEMAzDMAyTI75dug2z1lbi5Umrk9YFA6rTlHulydqHhnAUPs0KZ6Gp+YSjcTz65VIAwKdzN2Fnfdi0PhshmtEslzDfk2CniWEYhmEYJkf86Y3ZKdcVBnzwaUlD2Sg33VysSlNDOMY5TVlkY1WD8bfVYQKyM8Zq6fps5EjtSXBOE8MwDMMwTB4SDPiNHJQJS7Zh6ZbanPYnKTwvHAOBc5qyxbrKBsf12RjjqBJGyTlN3mCniWEYhmEYJg8JBnw44cBOxufZ66py2Jvk6nn1YSWniZ2mZrNuR73j+mwMcUwJz2OlyRvsNDEMwzAMw+QhwYAPfbq1x8DDu8sFOXZMrErTHWcdbIQPss/UfFIpTUYIZBYGOcJKU8ZwThPDMAzDMEweUhiQ77YDPvk71wXq9JLjj19yJC7qvw/aFxVgc3UjAHaassGW6ibb5cUFftSHY1nJaYqpOU3sNHmCnSaGYRiGYZg8JKg5TfkSAqeH5wUDPrQvkqWqfTxPU7NZtKkG01ftRH2KkuI9O5di6Zba7OQ0KeF5TRye5wl2mhiGYRiGYXJAunArv+aQJByTFu+SI+GoNLJ1Zw4AV8/LAhc+N8Vx/WH7lGHpltpmq3mhaMxUCCIb8z7tSXBOE8MwDMMwTA5IN2GtXh6aspjT0hz0nKagP2E+EitNWeeCI/fBo786wvh8+L5lAJrnNM/bUI0jH56A/05ZYyxrjOR+7q+2BCtNDMMwDMMwOSDdhLW6k5IvE8jqRnZhQHWa5O9c92134tfH74+9y4qMz726lgJontM89IslSYU86kOsNHkhp0oTEd1PRD8SUR0RVRDRp0TU17LNRCISlp+XctVnhmEYhmGYbGA1Yq0c0Fkay/mQ07RoU40xT1Rx0G8s1x06IPdK2O5CYcCHfvuU4Z6BffHc745BgVYIpDnDqzq6OtvrQpk3uAeS6/C80wD8B8DPAZwDoADABCIqtWw3EsA+ys+9rdlJhmEYhmGYbOOkNL3yh+PQbx8ZlpUPOU0f/rQRANBzrxKccOBexnJfwmfKec7V7kJhQDqlt5zRBxf131fJG8t8gIsK/EnLttXZV+tj7MlpeJ4Q4lz1MxENBlAB4DgAPyirGoQQW1uxawzDMAzDMC2KdbJYnV5dSvHLw/c2PvvyQGnSVYnBJx8Iv+IpERJ/x4WAX/nMZEaSKmRc/8zbLLZzmmpZafJCrpUmKx2035WW5VcR0Q4iWkREw4ioJFUDRFRIRGX6D4D2LdZbhmEYhmGYDElVCMJq4OqOSS7D3yo0VaJbWaFpOSmWJEfnZQer05TIacuu0lRRy0qTF/KmEAQR+QD8G8BUIcQiZdW7ANYB2AzgKABPAOgL4NIUTd0P4KEW7CrDMAzDMEyzSZXTVFpoNnC1lJacOiUVmtLUrX2Rabma08QV9LwTs5GP9PA8HX2EmzO8fhuZZGd9GOFo3FRCnklN3jhNkLlNRwD4hbpQCPGK8nEhEW0B8C0RHSSEWGXTzjAAw5XP7QFszHZnGYZhGIZhmkOqnKZLjtnf9JnyIKdpu+E0WZQm5W/2mbxj5zgXFliUJi0cMtPh/einjXh/tr0pvH1XCPt1LM6w5T2LvHCaiOh5ABcCOFUIkc7Bman97gMgyWkSQoQAGEGaRBxbyzAMwzBM/mHNaRrQpzNOPbgrfndiD9PyXOc07QpF0RCWE9t2tThNrDQ1j5A2YbCKNTxPH+FMx/euD+anXLejjp0mt+TUaSLp0TwH4BIApwsh1qTZBQCO1n5vabGOMQzDMAzDtDARi8pw7cm9cPZh3ZO2y0ZOS3PQVabSoB+lhWbTUX03zS6TN6at3IEZq3cmLU8Kz2vGPF124X+m9ezouibXStN/AFwJYBCAOiLSS8XUCCEaieggbf1YADshc5qeAfCDEGJBLjrMMAzDMAyTDayFIApS5JZkUnK8pjGCwa/PwsX998W1A3pl3Ecg4TR1sahMgNlpYqXJG1e+OtN2uTXHqDnzdFWkKSvOl8w9uc78ugmyYt5ESOVI//mNtj4M4GwAEwAsA/A0gI8AXNTaHWUYhmEYhskWkVgcf/vI/P43aJetr+Blnp6XJ63C3PXVGPL5koz6p1LbGAEAdCwJJq0zTW7rPFcvo+A0R5da0h1QlUbvx9lQ2ei4nickdk+u52lyTDgSQmyAnACXYRiGYRhmt+GD2Ruxqdps0AYD9mZRpkpTtqhtkm2VFSWbjSanKUcBekKINpXDHo3FMeLbFa63T1TP8z6+a3fUO67nCYndk2uliWEYhmEYJqdMWr4dL05c1apv3bfazJET9CfPpQNkVggiVTnzTNCVprLigqR1qquSDQO8fGsdBj7zA75atNXV9j+tq8TRj3yN92dvaP7BW4nPF2zGc9+tdL29oTR5PE5jOIZ/jV/muA2HVLqHnSaGYRiGYfZornltFp74ahkmLd/eascM+pOVkVTz5Rglpz3Yt07hX16pbYoCsFeasp3TdNM7P6F8Wx1ufPsnd9u/PQc1jRHc+2HbSXVfs6PB0/aZ5jRNXrEdO3aFHbdhp8k97DQxDMMwDLPH0hRJlHxet9ObMdscAjb5SwU2jhSgGM0epBxrOfPmYChNRTZKE5HRv2zY35uqnHNwrGTLOfzop424/X9zbUuAZ5v9O3kr8e11fBvDMTz25RK8NWMdgNT3lZc2mdxXz2MYhmEYhskZGxUjva4pe3lA6bArBZ1KaSJ4D88KR7NnDRs5TTbheYAM0RPITlGBkMewwnQltd0QicWNuYzO6tcNg47er9ltOuHzmH+lX3+3p/ripFUYOTkxi8/Pe3fG5BU7bLdlpck9rDQxDMMwDLNHEo3FccfoucbnjR5VjuawKxRNWpaqel4mOU3RePaUpjqH8Dwgs0IVKkIIvDV9LRZurPG8bzbyqH5cU6m01/JORNSjOuYzbgt3fVu1fZfp87E9O6Xcln0m97DSxDAMwzDMHsm3yyqweHOt8XlDVeuF59XbOU1p5mnyYuBmtRBEGqVJ9k9kXD1v7MKteGDM4oz2zYaTM3VVQoX5y3vz0RCO4aqfHdDsdlMR9ejpeVWarEpWj71KUm7LSpN7WGliGIZhGGaPRFdQdNLNaZNN7JSmgiwqTVktBNGoK00pwvOM/mXW/vJtdUnLAj53IWzZCM9ba8ll+79PFjW7TSd0pemCo/Zxtb3PyGlyd67WFKZDurdL2kbPc2KfyT3sNDEMwzAMs0fSGDEn/dupPy2FF6WJjPA3L4UgWiKnyT5AKZNCFSp2590+RSiglWwoJRsrW09hBBJKU4GNY9ilXfIEwl6dUqvS1HOvEnxw40l4+vL+xjI9FNRp/CaWV+D6N2djx66QuwPv5nB4HsMwDMMweyS649J//w6Yv7GmVUOV6kPJVdpSqSuZhOdFWmCepvYplKZM+qdid96pjmUlGzlNG1oxlw1IOE3WCor//s3R+HnvzknbkzG+7k7WZxnPogI/TjhwL3RvX2QsKwj4gHDMcfwGv/6jPD6AV64+PuV28bjAH9/4EQd2LsXDFx/uqo9tEVaaGIZhGIbZI9mlhefpBno2DHDXx7ZRmihFVbVMwt+yFZ43afl2VDVIp6lzabIKAqiTr2Y2gHZhiW6VpuaG59WHoqisd57LKNvo4XkBH2HE744BADz6qyPwq2P2w94dipK21+8Kt06p1QfVVaWSwsTkyQFfeqVJZ8bqnY7rf1pfhYnl2zFq2lp3HWyjsNLEMAzDMMweie646GFnXpWm78srMH7RVjx00eEoDvrT76Cgq1znHr43vlq81XFbrzktQHacpndnrsffP1kIQBr4nUrsnSbdRs/Uf7ET2NoVto6J2poVE3UiMV1pIlzcf1+ceWg3x/NNOKXuUMPzggGfoTypx9CrK9rdU0u31GLBxmrjc22Tc9iqqmoKIQznf9zCLeixVwmO2K+Dy57nN+w0MQzDMAyzR5IopS2VJq/hZddq4Uvdyorw13MO8bSv7jTddPpBOP7ATo4TnnrNaappjCQVN3BLJBbHH0f9iKN7dDQpB13bFyaFfSX6B0/9s2I3EW+qohjZZmd9cr5OxxJ3oYGZojssutqTzkH0Or7qdVLL2BcquWN7lQZR3RCxdXTPe3Zycp9jcdsJma1EYgLBAGHu+irc9M4cAMDaf17gqt/5DjtNDMMwDMPskdQbSpMenpeZ0b9+Z73nfXSVq7QwgOtO6e24rdd5kK55bZbn/uh8u7QCk1fsSJoMtUu7wtT98zUvp6kpkuw0ZXothBAYOXk1jtivA04+qEva7Rtscsv2SqGoZQsjp8llhUCvOWNqs2T6m/D5rb9AfTiKZ75ejtXb6123WdUQQdf2qe8BnXAsjmDAh6VbkisitnXYaWIYhmEYZo9Ed1zaF2YWnqcT8VipTgiB+rA01t2EoXktOT5vQ3X6jVIQiiY7EelI5NxkNn52x8z0Wny3rAKPj10GwJ3C0aBVUCwq8BnOW4srTTH7QhDpcDsmfsVTsrplR+4vQ+V8HtVLt/lqkWgcKMzu5Mr5AheCYBiGYRhmj6QuSWnKrB278DInQtG4UcCgtDB9LpRhA+dwTh0nZ8qrEpbUtq3SlFlbXsMSG7R7oN8+ZcayogJv+Wle0QtBFFgnVEqBoeS5bN/N2GmRgSanKRyNo04rL2/FybdSJ+vVnwWvLxLaAqw0MQzDMAyzR7LLMv+QqpSMmbcJG6saccsZfYxloWgMQiQb1V6LLqiV80qDbpQm7/M0ZUqqCn5OVeqomdXzQjbl0d2oVtUNyVXv3LkhCRo0xW+/jsU47ZCu+Pc3K1p8nBPhee60C69KnqrypLqediF/Zz49MWVhDKcxCSvXT/87msXJlfMFdpoYhmEYhtkj0edKKrMpOX7H6HkAgF/06YL+PTpCCIFTnvge1Y0RLHjolybHyavTpOdSlQT9KYsrqFAzlZzm4vcRHhl0RMr1icltM2u/KWIXnpd+vwtGTDF9tjoV8bhIO776BMelwQAO6d7e9bGbQ1SpnucGrzlNblQeu+IiTpUEncZEVVr1v6O5ullbEA7PYxiGYRimTTB7bSWGfrEEDWHnEshu+HLBFmytbQKgztOUbOhtrwuhtimC+nAMFXUhhKNxrLOEgHkNRVKLQLghVU7T5upGbKkxG7rNnbcobFF9bj79ICx8+JcY0Cd1UQWvOVdW7JQmN+exqTr53FVhpclFfpbuwBYH/RmVdvfK2IVb8N7sDQC8FIKQv92Or+rEpxCaPJeJjztsaKc0qX1o7j2ZL7DSxDAMwzBMm+DXL00HIBWau37Zt1ltfTRno/F353ayWpoQ5nlmAKC6MYKjHp6AooLEe2arsuQ5PE8rde52LiKr0rBmRz3uen8e5qyvRlGBD/MeTChfu9LMqZMOq+qzT4cilKQJIfSlssxdouZL7duhCJtrmhwdl3hc4H3N8VCJxIRJjWkIx9L2XQ/PKy30t4qid7NWhhvwUAhCd+ZcHiOqOPGprozqiNWHonhyfLljm07+mklpMsLzEjtEYnH4fS2bJ9YasNLEMAzDtCmaIjEs2lTTom+D2wp2YU17AisrdjW7Dd3ROf/IvU2Ttlpvq0WbagCYy2JblRHP4XlhXWlyZ0gaCohmNj/37QrMWV9t9EtVvmpTJPK7xXpPdW1flHafhGoh+7e5uhHLt7kvOa2P52XH7o+HLj5cayv19mMXbcF9Hy9MWh6OxU3q0pvT12F8momDddWyJBho8dyx8q3mMWmpkuNuCpMYjq4Anv9+pWlOLjvc5jTpz0JEidXcXUL12GliGIZh8ppV23dhxuqdxufBr8/Chc9NwQezNzrs1Xb4ZO5GzM+gRPT8DdU48uHxuP/jBXucA9lcZQNIvAk/94h9TPPauDGYG8JR05hHol7D8xJ5NK7QjXnNDrXmwqzZkZgnqqYxu06Tm/LbVoXm5H9+h18+8wMqtPBHt8c8vW9XYwJWp+uwZHOt7fJILI6mcKL/I75dgRve+snx2LrSVFzgV9QXV932REVdEwb++wfTMrc5TepWbp71aCx9IQg1p2nN9vTzjHkvBJHYfncpCsFOE8MwDJPXnPX0JPz2lRlYtV2qCzNWVwIA3p65LpfdygorttXhL+/Nx6D/TE1Z6jcVM1bvRCQm8L9ZG/CDZSLSbBGJxbGh0py/8+LEVTjjqYnYWtOEdTvrsXBjje2+jeEYznt2MoZ8vjjr/XJTPCEdep5FgY9MhqXVYLYzFutDUVMeU8RjBQQ9j8Z9eJ65L93LzOrPWmVy3dpmO03mc9GLZDj2z6Z8NQCscKkI6kpTYcAHvy99iFxBirC2aEwYhR3coobnJRQd716TECIpv0zF+hwBQIHL6nk+h/vTDjc5dqqD6HfhvLktBBHS/lYdqdlrqwzFti3DThPDMAzTJshGSFa+sX1XyPj7fY/KmWrcbqzyNjeNW259dw5O+df3mLYq4ZQ98dUyrNlRj3dnrcdpT07ERc9PwTYbReHzBZuxdEstXp+6Nuv9yoLPZJRl9vsoSWlSk971Cnsqu0IxU0heptXz2hV5y2nSu2Ut1rBWUZqs4XnFHuccsjodejl2JwiJ8DHV4XDre+jzNBUWuHNcggF78zUSixtOkIpTW3p4XnEwkKgCmIHT9MRX5Thp2Hd4a4b9y5ziguRx9Lu8kVWxyI1DpzoxqXOaEkpTQYp++Fwe1xSep/2tKpbXvTkbv31lhmMxibYAO00MwzBMm8Aa/787RKSpuTGrt3tzClXjuKWMkfGLtwEA3pmxHoA5Yb9DcUKBsHPaGm2M1+agGm1ZCc/T58rxk6k9IYCYciy7Sn27miJmp8lzeF5m1fP0MdDvm307SMVpgzL+tY2y7b5G+WxvfbOG56nX2U3/VIPd7bxN+n1VFPC5clysk8K215zPcCxuqzQ5VW8zlKZgwmGzEw7//slCPPL5kpTtvDRpFQBgyGf2yqrd+bgOz/OoNEVdVM/TRS4hBPw2ilevLqVYPORcdC4NOh53Y1UDhn+93PisX3/rddgVinpWZPMNdpoYhmGYvEX95+/2rWxbQnUsvBq3ajhfSyRaq87CgV1KAMBUcEB922+d7NXap2zkXKnGeDZzmvw+nyX8SZiuRb2N81cfjpnerrspbW3a32N4XmLyWIk+Fh20AhZ2fSkOymvideit4XneJt81h4a5PXaTjdLkNjwv6PehMCDPNRKL2zrrTm0ZOU2q06R0vLohjIc/W4x3Z67Ha1PXpM0Zi6U4aTs1MlWYoRWT0uTCEfU2T5N9QYrCgA/FQb/txMXVDWHjpc2Nb5tzxsI2SpNO1GNp/nyDnSaGYRgmb1ENZbcGRltCNfC8zmWiKwqZ7OuG5dsSypeuiKghkmruDNkEAakOr908PG6IxYWhQjQoYXJZ8JksOU2J5TI8L/G5IWSjNIWipnuzuiGCt6avdX1sr4UgrNXp9HC2Yq0Munr9w0p+EOBe7dGxGruu8scUpSmiXGu9v9ZwQpXtdSFjvqXCgE9Re1L321qqO6gpNpGofU6T0wuJxnDiWiQUs8T6+z5aaKosV90QTtmWdV8VuxcbXqvnObWvojpoamVIFfWeslO8CrUXIUbuk9ZkOBrH8Y9+g6OHTEA0FseiTeaiHPq1trsObb2K3u73H4hhGGY3JRqLY/iEclMlud2dUMSb0rRoUw3u/XC+bY5NPqIaFl4LTNWFEk5LzJSDE8XizelLsoeiMexUcqqsLN2SMIb066AqTVX1CeMxahN2oxpImYbq/eo/U3Hqv75HUyRmlOkGnI1wOyrqmpIcy4gpp8kc/mQOz7NRmiyFIADgs/mbXfenPuS15Lglp0m7WXQ1yeQ0aet09c+rnZpJGXu1f6ozGY0JzFy9E4c+MA6vTl5tu++T45cZf0unCVpbDjlNfrMyWKA5iOEUOU1OLxXqjZLjflNFOR01nw8AdtY7O02psFOaMque57xtPC6MSZsB4Pkrj7XdzpdGaSrSxtSqvu3YFUI0LhAXQEVdKOl7WT9Pu2e+rVfRY6eJYRimjfDe7A0Y8d1K/PaVGbnuSquhKhRu3rBe+NwUvD97I+75cEEL9ip7qAaq1/A8k9Kk7HvJC1NxwYgp+HZpheP+5z87Gcc9+o3xlt+KWgZdD/lS+1utKE1rdtQnOWlqeJ/XimaANL4WbqrBttoQyrfWmYxhL4b9zNU7ceJj3+Kv788zLY+ZcpoSy0VSeJ5dTlM0yQj2UupbHw/d6UmHmn8CAGE9BE8rLqBefz2/KuiidLcdXkMNAUtOU9SsMN71wXzEBfDol0tt9520fLvxd6eSoOcJZgUSKnQkFre9N1KFzAkhEhMNFwVsHTar0rZzV6ZOk53SlEn1POeBueuD+ahukPfiRzedjL57t0/RpvwthLBVE4ssSpNe5EOda2pTdSNKLPdwyAjPS3aQWkIRb03YaWIYhmkjrN/ZMhXS8hm18ICXf7jlW+3ncckG2ZxQtiFLOU1qKJMeVvfpvE2O+6/S5mb5duk22/Vz1yecJl1piphC0hLG4x2j52HEtyst/Us4G3Zv/9OhOiF+HxnqDODNCXtyfDkAYMw8sxKk51cEknKa4KJ6XjRJ7VLPNx26YVkU8Ko0mcPdEkqT3O7HtZV45huZlG+E56W5rSKxOCYs3opKTUHJRBX0KTlXqtIUjsXSVhZsr5U0v+HU3uhUGnSlNFnz5VSnye5ei8aEbT+aInGjrXaFAcN5UA9tzZ+rrE+tzjphp7K4LwSR+Dvdt8QncxPPfdAhpFmdMNdWadJCP1X17b0fN+DaUT8a22yqSnaa9Otv9z0ZYaeJYRiGaQ1STVK4O6MqTXYhYK3N89+twKEPfIXpq7ITImkOz/PqNCWMdLtcgeYUztgVimJ5ReKNsu68qsfR32br6Ma6Xf8ycTRVpykUjWekND325RLMXldlu04tOZ6U06QMp131vPqwOacJSB4PJ0Ja/wsL3BYCMBvzRniekdMkP1/+0nRjH7U4h1Oo5sjJq/Hnt37C5S9NA2CvELglLszOSSgST1JYwtE4/vDfmRjx7QoAMEJELz12fwCJ+9bJ2VOfFQElpylFIYhLXpiKXzzxneklDJB48UBkzmkyKU2Wx2hHDpQm6/2psmRzLWatqbTdz8kpU50ha44YoChNyhxcj1nUwk3VjUl5ebrSaRsmyYUgGIZhmNZgNyweBwBYvq0ON7/zEx7+bHFS8rea0xQXIitV2JrDUxOkY/DAmEVZaS/T6nlCiLQlx/0unWy7w26vC5mWN9koTVVpEuLVQhGZKE2qE9IYjpmUJreG/cjJa1KuU8PziMhU6lo1yu36vqW6ySh40KVdoexjJJZklK/ZUY+VivOpk5jM1a3SlOgbkFCaSjSD1c7hVqsbqqvX7qjHs9+sMJzSL+ZvAZBQHtXwvGN6dnTZP6V6nlJ+PRyLmwpDAMDE8gpMXrEDw79ejmgsboR5dm4XtLSV+nkwOU0iEZ4Xjgpbh3rdzgZsqw3h2W9WmNbXKVUMfcokx+p4WV9WVXrIaQpFY8ak3HYvfdznNKUuBHH+iMm44uXpqLDJ43QqnmOa3Nbmu0JXQdVraw3j21jVmBRiGo5phVtsXjZwyXGGYRimVchGye2pK3fkXSGJ579bibELt2LUtLVYYJk1XjVCH/psMU58/NvW7p4tXicMTUVThkpTKGp+g2+Xs+FU9cw8AWnyvta39Xo/VacpXQ6PqjSlCqfbWtOELxdssT33msaEcdoQjprayCRHyooxT5PPnPAuRHoHdvWOeqyrlOGynUuDhsOlj0l9KIqmSAxnPDURZw//IcmADFkq3KVDN5qtk9s6FXtQ21av8aD/TMUz3yzHA59Kx99quOtVCv84oBdeu+YEV/1Tc650oxmQLz1CFkVOdeY2VjVCCKmk6FXe3MzTZL1fAorSZFUAVV6YuAqPfJGYa0m/R8u0EEE7h836GDkVT7EyfMJynPX0JIxduMW+5LjrnKbE3+q1VF+WbKlpSjqGdT4rc5uJc7Ubaz08T51s2Po/aGuNTXheNA4hRMowybYMO00MwzBthOaG59U1RXDVqzPx21dmZDUvp7moBqU17E0Nz9tQ2Yjtde4MFrsS2M1FNVay5TS5qZ43YfFWXP/mbFO1OlVlAuzD85x87KglvCm5X/ZGvmr0pAtHU/vYaPPWGQDOHj4Jt7w7B/+btT5pneqUNUZiptyibNy/MSOnSQ6Uqua4Uf2+XyYLbRQW+Ayju6YhgvpQFMcO/RonPvaNsUZhKCgAACAASURBVK1unH+1aCv+8t48Ix/MrdOUanJb3WC1UzFUFUu9PfRx1avCqYbwyoo6o/Ladaf0QqdS+3LVVvTnTQip9uiEosk5TarTtFB7SdKpJGj0QzfSnVKhrPe7rqhE43HDkfjitl/Y7vvuzMS9pofn6ZPj2pUct+Y0eame9/IPsmLg3z5aYLzkUOfmcvsiTP3uV/tmVW6sz6ST0kTKPWX30qJYUzETTmzyWIRj8aQ52iIxOcGx3XdSPoRYNwd3EwQwDMMwWaOqPoyAn4wEaLe4DbdKhTUnxG5C0pZECGHr+KmKyYzVOxGOxnFglxIMOnq/vHHunv9uhcnYcpuLkg71mqQKPfzzW3LyyKfal+OxS44EkFxy2zY8z8EgC6epSmh9S6xfB1M56TTKmBulaZcWHjV5xXb8/ucHmNapBmBDOGZywpzui9lrK/Gvr8px//mHOvZPLTkO6IapzGdyEwa6ZocMZyvw+9CxpAA1jRHUNEZQ1RBBKBo3OfxNkRgGvz4LE8u3m9oodPkMWsPGEjlNmtJkY4ualCYb11h/7tTvlbOH/wBAOg9d2xe66pu+veyfOadJqg7mbVXHe8FGWWxkL8U5U5WNVCQpTb7EPE36fdmpNIiiAp9jKKd+jyacJjulyfwchVyGhqr3aH0oapx3u8KAcd87KUEq6qOs9k397iQyq7OA25wme5VbHxN1PierDxaLC/gsxwhF46Y51VTautLEThPDMEwr0hCO4pihXwMA1v7zAk/7NnduV/Vfm9d5bprLyB9W45XJq/H+DSehV5dS0zr1H/ak5duNEsSDjt7P06SodmpFttBzmXRaJDwvjaGuzj1lNT7s3spbjT0VUyl3m/WpwvPcGj1CCOxQwpjS5TQFbXJ7rE6TOkeUU4W3X2vFENSiCHaoOU2AOomncFXeXs/ZKvATOhQXGH3eu0NR0rafzduc5DAB3pUma05TkYPSZC4Ekdym7tzYOddd2hV6mkxaNcCtJcdTHRcAFmyUSlNnk9MErS0v4Xla9bx43FjnJ0r7oimhNBVo55F8bGsT1ry1VCzflshliwtgs1bav31RAHpxT7sCDHaYlCZlufXlR5VFaXKunpfYz+4FiJ0jaS1cEY8DUcs3SDgaty3TD/DktgzDMIwH1lcmDD+nGe/taG54nlrutTkVsjLhsbFLsb0ulFR9CbCf9FEnnYGiGw2LN9fg/o8XGstbutCg2/l10qEa/+lymuJCOpVvTV+bNGYxG6PZ2WlKHNfOgbYqQ3aFIJyo1tQWo710TpONcWcKzwtHsW5nfaI/Wt6EE+kMtObkNAEJlSIY8BtOU01jxPa42+rsJ1t27zRZqufp4XkF5pLjKuZCEDahUjEZlrUrlGzgBl32S0cN9TJVz0vjNC3ZLL0HPZ9JtpV+nibrGOv3TzSWUJr8PnLM6wOclKbENtbnyO1358qKXabPeuGUdkUJvcKu1Hcq7Bw6dSxjQphCeAFntVl1hpyUJvXesz5zMSGSHPZUZd8BntyWYRiGyZB0yoIV9Z93JlXk1H9YXiawrG4IZ7FqXXI7qZyFWFw4hsIs3FSDvg+MwzdLtuGTOc5zEmUbpze4XlCdk3SGelwIXPPaLDwwZjFmWkoM6/eSMIUVpW5LHVe7UDfdyWmv5V/oTpZd2WQ7ttSYnYR0TtOm6gYMG7fUpE7VWKrvrd2RcJpiceG6LzoBH6G6IYyRP6zGjl2J6oABSy5NXAhXz6ZeeS3oJ5QpTlPIZjw3VdlPIOw+PE/+Nia31cPzjHma4hip5c8YbZsKQSS3GY7F8ftXZ2Lx5uQ5zewcKSdUZyOcxmkKK9dNH0Pd6ZRtQWvLSWkyt6urheoLgICP0uYMpXKanJ4jt0qTdd6ueqVSn7XfbjC2VIZFPd9ING6acPqPA3o5hoCreXJ293u7wmT1rc5yX6jPoX6/RWJxU6VL6/ZtGXaaGIZhcoTXfyCqnX7b/+YiEotj9Kz1GO0yLE01Mt3mCs1aU4mjH/kaQ79IVogyQVXLGsJR+WY6xTiEorG04XlCAP/+dnnKuXhaimyFmXhVmnSWWSbv1e1UtV9Ob9ntDNumSAxz1ldh2dZaw+npWFqgrdMKQbhM5F6yxdy/dNXuZqyuxMuTVmPY2GXGMtWBqmqIYLPVEXN5D/fWwkGjcYGXf1iNx8YuxdnDJxnr/X49p0l+jgvhqAJbHeYCvw+lmvMiy44nj9EKi+qg41VpslbP052mqoYIHhtrfkbV8LxUDsj0FJU0vZaIVw1wc3iezQSnNuPToSRh3Lubp8n8WVcL1e81v99NeJ7uyOjV8+Ryt0rTzl0h/OPThVi40Vz1E0h2POu1MbXOa+QWOxVMfY4jMWGEjA46el88eNFhju2ZcppsXkDozp1+3KiNKikUpem0Q7oCkN9jqe6ftj65Lec0MQzDtAI1jRHUWOLNvTpN6j/vLxZswfEHdMLDn8vyuRf239f0BtMO9XjpQkzW72xAh5ICDPl8MQDgtalr0v4TdoNulIxfvBU3vPUT/nnpkbahZXof3bzVPbhbe8+Tzc5dX4XvllXg1jP7uJ4rR8UpJywWF/CRu3BKs9LkvK369ttaJUsfQ7VfTgajndJ049s/JeXddCoJYkNlo6I0pXeaVm3fhbs/mG9a5tYIX7k94VxsVZwkfa6b0qAfTdG4pkLGgOL0xVS6lRVitaZS6feJOn4FlvC8uHC+FtZhLfD7jPmS6kNR23tjYyqlyW3J8RQ5TSUOSlWqeZrc8NTl/T1tb1TPg/nlzMc2CrDdPWRWmpKLMVhRvzMuO3Z/o6CCqqAHlHmXUmGtnkc2x3bKaXrws8X4csEWvD3D/OJqS01jkoOhVwlVw/O8YKiNitRkKroRi3kqZa+Os2NOk9bUrqZokiMbE8LIcywyQkWFg9LE4XkMwzBMGo5+ZAJOffJ7U5hOc8LzAHMIlJvCDuo/WCelaUNlA0598nscN/Rrk+GaDfRzuEGrCHffxwtTFhdoisRcVaoiANs9zJ0CAJe8MA3PfbcSL05cZXvcdKRSwGqbIjj5n9/ijtHz0rbRFImZSqiny3FTDTnrHEn6pTU5TQ5Kk2r46ediV6hAzzUJGTlN6e/Z2Wsrk5bZhXtZw8kA4CClSIh67+kTipYUBlCkGYRulaau7ROFGexeLCRKXcvPqUow61jXSKdJGowN4fTqqIr7yW0Vhy6eMHKdcutMhrOHr5pf9OmCi/vv634HmJ26cJoXHXZOU5niNDnN0ySEwJLNtca179WlFEN/dbgR5qa+DPIRpSyeoz/jenGRvcuKtH20Y6uKrbV6nnJ9l9iENgLAKU98n+Q47NIqyrUrDOCPA3rhdyf2wD4diu07aIM11ysaixsTEwOy1Ls+tm6KeKiqmp0zYw1ZVEP/Tuy1FwD5vaMfU5/XKeqkNLXx6nnsNDEM06JEYnHMWlPpOg58d0X//z9nfSKMzHshCPNn9e2gm7CpqElpSn09Zmn5MtG48DQniRt8REn5UalC3ZpShDpZWV/ZkHGs/E+WsL6Zq3fi0Ae+wohvVyT6Z2PkpZpA8/P5m7GtNoTP5m9Oe+zl2+pM56470Ys312CmTdhUxGGOpLgl1wVwtpNNIVQO94JeCrrJg9KkhoUd0r0dgITTo7Nkc21SOBkAo/JcXVPElD+hn2/Q7zMcBavTtHxbHYZPKE9qU3WU1AqEOsk5Te4KQegEA4RS7Rijpq3FjzZOox1EXkpOJ3Jt1GvsVMVRvQ4xIbBq+y5Xz4ld9T+3/YuL1IaxbqSHbdZ3tFOabG61T+ZuwvkjJhvKzkVH7YOSYMA2PC/gSx2eV9sUgRACS7Uw1377lAFIOCZO8zQ1hGN4Z+Y67XztFaloXCTNKdeglBl/8KLDMOzSo2z7lgq9ef37c9S0tXj++5XG+kgsbryAcuM06f19adIqk8Krk6goKDfUQ/86lwZxx1kHA5D/wyJ2SlOq6nnsNDEMw6TmqQnluOLl6XhozOJcdyXv8GroW+041UBwozSpxr/dW/q3Z6zDixNXeXkp7RkimHJTggFfynGQc92kd7bXKAUCjOO47I9VsXlgzCIAwPCvEyXGm+yS2VP0y1wC2HkkrQn4c9dX4/nvVuCCEVPwm1dm4D/frzS9rVbbrrbMx6I7X+o2TgaKaQ4hhzHuqOWaRLRKa26MHtVovl0zrnZalEBr/3X0Pa0Kpz7/TDDgM4wza4jpX9+fhxHfrYSVwoDPyEOqsBiyPkrkfunG4ZSVOzB15Q77k1M7qRHw+UzOy6hpa213Kyrw4R8X9DP1y21FTLU8tHrtnJQmNffqjWlrcdbTk/B3pcJkKryUGtdRC1WkeqGgt9uc8LzHlZw3IHHtCmyUJqfqebWNEWysakRdUxQFfkKfbu20Y8M49vJtdfjbhwuwuSY5tPL/PlmEn9ZVOs7ntMCS56Q/y27LjFuxVlCcsHibaX0kFjfG1k1VPrW/izYlK2alhX5tO/lZd5rKigtMamDUUJoS5e9TztPUxsPzOKeJYZgW5eVJMgRn9I8b8M/LvL1Zaw7/+X4lpq7cgdcGn9Dqk7g6ofoHXp0mqxGhGk/uwvMS+1vD3kLRGP7xqXQY7hnYN2nfbFWL8/sIWxUjJByNJ6kQOm6VpuaoYVbFxq5Mt50ql2q8rXPUON17S7ViCX27t0e5NqeLOh/Uk+PLTW+rTU6TVWnS7iV1vFLlD/z5zdmYsCRhcDmFQKqloEPRmCulSd/m7H7dsW9HGX5kvUapDHP9mbBW39Pv3aDfZ4TTzV1fhar6MM44tBsAe8MP0JymgA/hWDzJSVbnndGNw6FfLHE+QQsBPxkGZtI6HxkObfeyIpOT4ymXzhT+lrgGTm2oOU3Pasrpe7M3pD2U2zwrFf25qaoP48nxyWofkPgOsS0E4bJ63g6L8607B7ojor8A8Gv5TKlCVGuboqjcJUPzDurazhgrVTG75D9TjeINdsxYXWlSw6zf55uqzc6W3lYmTimgOqbyd0HAfG7SadKUJhfXMJ3Drt9bRnheQyL/S1fwYkohH1aaGIZh2ihPji/HtFU7MWZe65aiTkfcJhzLLdZ/yqoxr078mQr1LZ9VXaioTRgjtv/YsjTvkY8IjWGz0WQ1ZHWaInFXOU3pSJWUDMhy6ip2hoRduexUb9NVpyXVcRdsrMaT45cZxQH275Q6r0EN81NVN2s545iN0mRXqSoai5scJsBZaeqkVDVrDMdSnrf1GIB8+69PWrpzl3mcU4WH6veeHkZnNeCDgYSqM+TzJbh21I/4Zsk2xOICJxzYybbNoOY02aEa1U7zWgEJo/Wm0w8yLS/w+1BsUxHt0mP2M3I/AKBb+0Ij9wnw5pyo6ot+DYIBn6OiIJUs14cwcBsyqKIfRw0Xs+L3E8q31uHpr5cnretgymlKrhKXsk3N6S3QxmH51jptufycKjzv4zkbsbxCbrt/p5Kk84gL4egwATIU1+oYOaF/HxS4UIHssCpwVucrHEtUsnNzDLtNurQLptxO/55uXxQwxjceV5Wm9DlNbb3kOCtNDMO0Ch1L0le5agm8ls5tadR/Gnb/QKKxOATs30ZanSzVQL921I946vL++PVx+6c8dtSh5HiFMvlmOGavrMTiIu28J+nwERlVpNLRFI1lnAunn+m0VTtw5ciZuO3MPrjrl8kKWq3F+bA7O7s+pFKaVAdwVyiKzu0Kk7a5+Pmpps8dHJ6NAzuXGEqcUxK1EZ6nTnZp2f7TuZts54VxckxLggH4SBqw1705G6u3J4dCWgkreRX6+TdGYmgIR40qc6nmbdLVsSYl0X+ZZggD0lGwGsLXvTkbfxzQK6WRrYbnWQmYnCbn83rtmhPQp1s77ApFDeUGkAa6XRW7wgKf6TnuVlaE4oKAab1b1NAs/d4r9Pscy8oX+H3wEXl+OeN1Ylsg4ejs2JVa9Y3FBX71n6m26+yUJkCG++ltvzV9bdJ+VqVJL+2u3yOpnEa12p36YsAaAueEel86UVYUQG1TNDGZcqZKk/Zb75r1f0QkGvdYCCJ5cH7WqzO+XLjFtEwffz2ktqyowLjvTNXzAumr50XaeHgeK00Mw7QKXWyMx9YgH15sqeqS2h/r/494XODs4ZNw1tOTbB0qa+EIq+H5kJaPkwpz9Tw1jEtga419GJjpeC6rlTnhI/fthFyG59mhn6ueS/ecTa6Lbf9s/ivalWe3jlE0FsfLk1Zh8opEBTqrGmS3HwB0LE5+u6ujVtdyCsG0lqIGzAU2NlQ24M735uHWd+cm7Tt7XRUeTRGSVhL0GyFsc9dXpzy+imq4lQb9hhGuqk2p7gG9z/p1V1UAQIZ42Tkbr01dk3J8nJQm1YlMF660V2kQPfYqSSq+UOAjlNiE5xUG/Kbjdm9fZFGa3IfnGTlNSISodSwtcFSaCvxk6wime/GRSfiY3qSqVHRpV4gvb/8FBp98IAD5PZPqurdXynCrxrx+C2+obMDQL5MLhyRymsx91sfFzUueTqWJPrspd+4V64uTTJQ8ILmqoLUdNTzPjWNmNzRnH9YNQy4+HB/ddHLSdjWNiYmAdac0Hk84QkZOU2z3VZrYaWIYpsVQjXQ72b81sKtQN3ttJV6YuNJz9bpMiZqcptQV76oawli7swHrKxuSEueB5AkdrWFt6cJJ7KrnhaNxnDN8Em55d46xLpXx6VYhEkLgng/m49rXZyEeF6aCCDI8z53TJOdpcu80qUaEfg5eHT27t692bVjH6M3p6zBs3DIs35aoQmVXZts6KS1gftNtRVUEnVS3WFxOyHrPh/OVZYk+pgqB1Hl1yhrb5UVBvyd1USiJ4QV+mVeih+ipuWuplSaz09SppADtlep3anieFf375k+/6GVaHvQ7hecpOU1pLKLD9pUV1qx5agFlnibTcS0KV+d2QUtOk3sTTJ0/aGOVDMXdv2OJ47UJBnzG/Emm/qa5npkoTfpzU1aUuJfPOawbDt+3A/58am8Aqatk+n1kMvJV9Uz/vvx07ibb76WA4TSZz0mfsNj6POsFH1TUvD2nfCq3DLv0SEv75ufbTZEGO6yV/ZKUptj/s3feYXJTV///SpqyvXiLe+/drI2NjW2MTS8hoYUeegKkwe8FQkIKIW8gjSQkpBJCCi+QhBZIgNDBoRowmOoG7r1u8ZaZ0e+PGWnOvbpX0szOFpvzeR4/ntVoJI1GM7rfe875HhppCt6HapLAMk18bs4wTB+aTXV1zuFeNz0vKvZ4ct3z0seTbm6riTRxTRPDMIwaWrTrN5velahufqf+5iX88LEPcf+b3VPvpEvJk4+NurSpUmrkZXtb/QfCMqpI09J1u93Gnw66upWwYmfJml34++vr8cyH27BuV4twozTNHCJNOabnjelb7j523oOudoYOXOhgTDXUUBpBSOfo9bW7POs0KSJNb0mOWgBQVar/blDbbT8BmUzZePXjnUJtGx2k5ptWWVEUUab0+R1Hu2R7XJOZMNne1IaVWxvx+ppd2mtArs2KR030JRbY1D1PxnlNqdSLKRaxQqbnhYvAyKItYhkoVbjYxSOmMHgti0eE1+ZW05T+P2XbWL8zWwvnd8wxS13TFBRJysf0JWsnnr1Gt2TqJJ1zrLLuTx+PeJD0UnWuB0f008E8kL2uZSGiijQNrSnBsBoxcgmIoiaXeioVlml4ro8+0vc7jEmDCtpHDNDUNOVgOa66dlSCLmsEQdPz0s8lSc8wwT1Pc6/QXQP7CyyaGIbpMrY3ZmeW7S41stbjd/NbsSVcTjqQHjjnm1pA87ip8JHvHzQPXCVQ5MiYKnrw1XvexPPLt2GXwlFOqGnKiBHVWFo3OA9bH/aPJevdxx1JWxBrhmGE3k4uRhD9K4uEfjyOUNMJPUMxMJOfcG7wqm0452jj7n3Y0dSm7HXkRJoeWrrBbeS6aqu3H4pfpIkKL9V5cz4/lR04/TuMu6KKmtJ4TjPj7UlvXYXTOHTTnlYcccvzOOXXL2pro+RIU8yy0Lcim97kF2lyBuxlUqpc2PQ8nQAxDOAHp2SjB3J6YNQ0ldbfadGUXbckZuWdnmeQdCjHfGBQdYnvZxOLmMr3FBSFyCfS5Ew30FRWx/nRNQ3Q/HzKA3x6zM7PpRNFp7VPdNtyOpqzDSqaLNNQnvMqRaQpXyKm4RH1NJIFpK+XfHAjTc52pPfc1JpwvwNhJjpUl7vqenEWOZ8BNYKg9woaaXLuZTeeNBHnzR6KhiFVAPTRxv2FHhVNhmFcZxjGa4ZhNBqGsdUwjAcNwxgrrVNkGMZthmHsMAyjyTCM+wzD6NtTx8wwTHioK1d3TjDRdDC/NIuwqQKtHUkc9qNncNpvXhSWL123G4t+8iye/mCL5pVpaEG+LYgmcf+0BkZVRyO/l70K0fTg0o04745XccIvFiuayNJIU/qzUd049el54cTOFmIqkUilBNFkGYZQ5O9Ha0fS19mNUlsWFwZIyZTtW0NBTz09L/R0tGReq+7TlMKelg4c/uNnsfAnzynPjRMl+so9S/G//34f723ci1WKJpJ+UVhVih/Ftfm1bU+6Z9hmxn7UlMVyKlzvSIjpeUC2LmsjcRp7e726PmprYxvaEkkx0lRBIk2WWqAA2etWTpXzFU3kmlF9FyYPrMQ73zkanz14iLtMdqVLR5q86XnxiCVEFUrjEeHYcjHHoVEGx3VxYHVxoBGE6j0FfZ6dqWmi19nXjh2X3l+ASPATTRv37MPtL6x2BZhONMlCUG5YDKR/e1T1cCojiCBilokrDh+pXF4k7aNaijTlErmlyKmDMWk7d/z3IzyRccYMF2nyLlOJcDmNr6I46tY00UwAxwiCuueNqCvDd0+ahHGZ5sFsOd45DgNwG4BDABwJIArgP4Zh0DvpTwGcCOC0zPoDANzfzcfJMEwetAlmA92nmqgY8qtbCntMb6/fgy172/DG2t2CoLjwztewalszLrxzie/r6eCVHpssmoRIk2KQK6fn+U3abdi9z+MMp+7T5L1Jao0gQoom+j46EmKzy22NbbgjUz8zVJEqQ8kl0lRTFvOkoHUkU9pzRAWorteK06CxVWM5/t6mvWhLpHv/LF3nFQFNrQlBuK7f1aKMsNAieM82woqmlO1Zl17fuujhaT5ui0A6pSyXSFNbMpntFZMZuDm9mmjqoK631uKV23H8rYtdB8eYZbo1Uc7fOtc5p3FtmZSeF4+Y2lS4IMtx0/Cm+xmG4Q4QgbQIUQk5uaapJGahpiyGoTUlGFBZhC8tHK08JhUmiTI44nNAVZHnPVAc9zyZoE8zv/Q8cRB93uyhOHRUbfr4AkSCLHjoIX/lnjfxvX+9jyffT4sBWTS57nmSMHP2KUeaVKmdfRRGEEFELUOo33JobEsIkdCyeMRz7eXrnud8cs5Pit928q5pUrxOvrzKiyKuWKe/K3FFnyYnsupYoHfnOKAr6FHRZNv2MbZt32nb9ru2bb8F4HwAQwBMBwDDMCoBXATgKtu2n7Zt+3UAFwCYYxjGIT113AzDhINaV4eZYGppT+CEX7yAHzwmdn1fs6M5p5lyoYbIJ9IUNlWA3pt2NGfrtIKK67P7UTehlY+Nig1VuleuxhUbpR4iiaQ30qRqVqqLzoQ1gmgi3eDbiaMTADz27mY3zWNwdYBoSoR3z+tTGvMMePxeS089/Rzodebc+FXRLtsWm1eqIk1NbR2COce2pjZlXxe/eqPGgLq1oog3JcbBub53t7RrP7sRdWX4xnHjtds3DCOnmfGOZFYkZ0VTenBPo2yOk55qjLpya5OblhiLmIJoiUVMxAMGnSUxb3qeTjTRGXnVx6CL5NBoQtQ0lNuXa5pK4xFELRNPXHkYXrh2oWssEQbqnOZEMJ0IpV40qWwggk0O8rMcT//vfNepW2qQ6JYFD/0eyw2L9el5cqTJaVYrrlsUkJ5nhHzr0YiJimJ1pJAK6NK4t54u/z5N6f93t3Tgzv9+JDS+9hxfF9Q0OZQXRZTW7M53YtOeVqzL1N05313HcEXVO25/orf1aarM/L8z8/90pKNPTzor2Lb9gWEYawHMBvBy9x4ewzC5oIo0dSRTSCRt5czs/W9swDsb9uKdDXtx7THp1I6l63bj07f9F6Pry/DEVYeF2i+tIfL7jQ6bKkBFwLbGNjfdKKzDkq62JJlpDPjB5kaM61cuRAqUkaYcJ+k27t6HiGlgQFUxSuMRQSRuy5h0qKJKuh4bYQ0c6OsTyRQ6NOJFLpCWeWnVDqzdGdy0FwD6lMTcjvUOKkEIeMUnFdn0PTqRNV2E7aPt3lQ7ylPvb8Vtz6xy/16ZqWcqjVnoV1mEVZmok59oCkohdWZ3U7YtXKdA+rpbv6sFc3/wjPb1i8bX402FiQUlKL2K8u6GPa44j0jpedRwxDnP1SUxwVXPwemBI/dYikVM12hChyfS5OueFxRp0okmC0CHuw3DMBCzTCGqKvdpcsRcZ9zpUjbQ0ibO4qsauB40pAqGYShFadBkUX7peeKO6HsMMiGRzwddPR4xhckPOaUxm55nKpfT44qY6vS8qjzS8yKmqY0Q02hWaTzieX/5nF8gK0xvfOQ9vLfJ68IpH18QSjt6TbSVUlEU9axnmYYyQulGmiwn0rR/i6aeTs9zMQzDBPAzAP+1bdtpNtIPQLtt23Lew5bMc6rtxA3DqHD+AShXrccwTNfTJgkEAFj0k+cw9Yb/KAeiqkHugxmHuxWKAnodVKQ8tHSD0hQBCB9posX4dHYvrCst3Q/NAU/ZNn717Cqc8IvF+NWzqwSxEaamKYiHlm7EkT99Hsff+gIAcQC+bMMe7GxuV0ZjdLVLYWuahPQ8yQiC4meAAECZ8qZjeF2p5+auSzOUGyw659W2xRS3rz+wDLc9sxI3PSpGPh2CGr3KzS+dxp/FMUvov9SZhsFOhCOR9Eaakikbjy7b7Pv6MX3LlWlGlFyO79K/vO5eZ84gqn/G/U71ecgCx4GKJjnS/LmylwAAIABJREFUtC8g4lnicc8ztaYLYk2TIl1JM4imA2FnEPza9Udgwdg6d3k8YgkDZFXdU1jcSFPKdmvtVL2hHO77QrrPjipSFjRwzcXVTz4+BzqADo40ic9T4TKsRqx71KfnSYN4hRGEqYg0WaYRGG1UEbPUphIAPOl5smjKv6Yp/bogwQQAsUjwPlQCUfVd99Q0Efc8h4hpKK8155p3tqu7F+wv9BrRhHRt0yQAZ3RyO9cB2EP+rfdfnWGYroIOklKp9KB07c4WtCdTeF/Rr0alCXT2137QNLQ1O1pw0Z9eU68XMr+6qS0bwdiu6J8UBM3jpu8nmbJxyxPLAQC3PLFcSOVSR5pyE03/fGsjAODjTC0JPS+2DbywYptyIKsTR7oIlEyTIJpS2s+wLGCwHoY5I2twwpT+OG36YE+NCH1vskkExfn7+gffEaJVb6/fgx89/qF23x9t9xdNMlv2pg0yopaJLy4cBQA4ZmI/5eBFLibXUSREmuT0vJQy0gCko3wvXbcQALRpRg659pWRHbz8tu9XzwWkhUcpEQcxywyMeKrc8+g2qD24GGnybks3YS+Igsz7rCyOCrb3MUtMz/MTOUE410hLe9L9nXRMJVQpyM4AVvXJqZouU/KJhMgDaxrRMQzDV3h7jCDIuvIEhyyanPOiizTR/aad7cT1aqRod9hI082nTNHWDdH3XhqLeMRV3pGmHNYNE2lSm4SEq2mSP8+oZSp/J5xr3qm/YiOIAmAYxi8BnADgcNu2qcjZDCBmGEaV9JK+medU3IR0mp/zz7/KlWGYLoNGVRIpsVGp6gdWFUnRpXb5IedNv7FWHbFIpGx844FluOL/3vA4zVHoIOOOxR/n3GuCRprk9DxKkxBpCjaC0FErdaB3kM/L8i2NQt2Zg0406UwJnnp/C257ZiVsOx1Vop9zh1TTRJFrT8IUL8tceOhw/PKsBsQiJqYOrsLb3znKTbehnxud/ZWPx/kc7nplbU77zlU0bdqTzvOPRUwcMqIGz199OG498yDlYL0sHk5QUptf5/OZOqjSXaaKngDAsZP6udEuXaTJaRKb68x4Qqpp8otcyCYLMrGI2DjWz3LcQeWeR88nHXgLTVVzSM+j0QM6QKWRJhtienBnIk3ONUIFo3Me/Godw4oASn7pg9I2NCJGhep7775f6bfIE2my1KLJWU7fv2l4a8/kFOGg0/W3z8/G+989BvPH1Am1UBR6fUYVzo35p+eF/yzDWY6Hu95VNU1yBDZiqYWxcx24vbo4PS9/jDS/BPAZAAtt2/5IWuV1pJOGF5HXjEXaLOIl1TZt226zbXuv8w9A+EYsDMMUFCE9zxZvgGHrJPL5kVWJmkeXbfIsa21P4q5X1uJfb29yC1dV0MH3h1sa8bcl63H7C6tzOB4imiT3PHqD/Yike6lEU1gjiNoy9c1cPi/rd+1TRpp06U9796mXX/SnJfjR4x9i8crtnmiUX3qePPjNJ01NbhRZURR1b9S0HooOzOTzkGvao0NYkwqHzXvSkSbn+IbUlCAWMZXvuyIgAuNA3fOcc1+ZGcx1JG3t7DQ99xXF3n01DKnCdSEto2WyPZZM4X8V5SFEE40SxSMmvrhwNBqGVGFcP3X2vSzE4hFLiGjRyFckINKkuybjQnpedp3ZI2rcfY3pWyZc+0Fizw95gFsUzV43fr+RuQy0HfKZvPCraQL8o5UqEeFsT44qemuaMoNyT4NcRaTJ8rrnyaLJT2T+/QuzMXN4H7ced+qgSpx7yFDPenQfEdPwnIt80h+BYEFHCeOAqLYc138WQHqiK2KZnusqYprK1zrrOZ8PN7ftHLcBOAfAWQAaDcPol/lXDAC2be8B8AcAtxiGcbhhGNMB/BHAS7ZtswkEw/RyxJqmlHADDDtQzScHWhXZuOyuN7CvPSkID+p+52eLK0dYvv7AMnzvX+8Ly/62ZB0O+f5TeG+jN+1QF2lqbksI5+TdTXvcx0rRFPKc6WbvneMYmLGAXrezRW0EoYk0NbZ2oD2R0jqxbdi1T5kiposWygNSXf2IH6rBmDNI+eN/s/Nw9DOQI3zdNfspW3E7qAZqQWlrDs4ALGnbbu2dMxufTNnaGg1qxFKuiDRNGVTlRmFyTc9zJkfc2X7T0G6jLDA9zxQiNFHLRF15HPdffijOOHiw9jWUWMQURCiNVtBrMOzMu7NNBxqtMgwDi69ZiCeunI+hNaVCaqpfT6Ug5OMIG7XK4yuVn3ue9Lf8GeSSngeI6YgUOdXTua50Qo/u1jS8oumsWUOk9fXHKbt9GoaBGz89SVgWteQaKa9Bgs4yP4hcooZhbM3D1zRlHzu/S970vIAUTDMbEd+f6WnRdBnSKXTPAthE/n2WrHMlgEcA3AfgeaTT8k7u1qNkGCYvRNEkzhqqxJBKEwSJpsbWDvzttXXY3ZI1e9DVKjW1JYQB8m4imvzT84Ktxa/5x9vYvLcVx936gkc40ZqmNTuy0STZGW5nU/Y9dMY9T057c3AiXk5T2fW79uUULWlsTeCiP72Gg777hFufQ89bImXj31JErz2hr2lqaU8KN+R8Ik1+KSFL1mRd4Wi0T05TbO1I4or/e0NYdmpA/6LOIA9KVe9BJWRUqPo0VWUGln41e3SffUpj+PxhI4S+WRFphj4XXvkobYBLB4+6gXiY9Dyhpolsh36KdGAqD8JjlimIMyqaAi3HNW89RupUZAvpypIoRmdqmzoShRkkyseha/Ab9Low5NOnSRacuUSaVNeXszl5Uqc8HlU6HspRDlt63jkGKuZ+ceZBOH5yf+F1fucrzO/TU1ctEP6W9wlAayARRC4COEy0UBlpUqZKZpc5qbyq9Dy/z9g1gmDRlD+2bRuaf3eSdVpt277Ctu0+tm2X2rZ9sm3b/nZADMP0CtrlSBOZNXz6g62eaIoN7w9qUOHotx56F9fc9za+dPebANJGDcf87AXlus1tCWEgSSNNfjNgQQ1GZY67Vdw/jXzRmVO5jxKN8HTGPU8rmjLvfVhtenC8tbEtsCicsmTNTrywYjvaEim8lXG2o6Jrz74OfP/fotNcOj1PfdyTB1W6M5BAAUUTGaSMqi8DIAoIOUXk7lfTKZoOp00fpK0LKwRhRJPOVU7GcQNrbE24YoVGmnRRNPnau+7Y8bj66LHZY7LoYDO/oUIY0VQSkLKWNoIgNU2aAT0dmMrnMxYxUa6paQqyHNddk/Q4/C3jC5OOJHdcChtp6qmaJlkYWD7XkOoz1Vu9m0pBLw/2nckcubkt3eyCsXUeseeXzhgUcT25YSCGSA27LaVo6vpIU5i6qfA1TdnHTqRJ/jijlumbrZG1HOf0PIZhGA8rtzbhN89le9QkU7Ygkn7x9Erc8PC7wmuoJnBuekHueQ9kLMlfWLEdAPC75/W1RnKkiYomvxStXISFzL+XbcIDb2xQPvfI2946K4fOuOfpBlSOeKkvL3KFFY186XAGHttJJMwZpFARuFfR7DeRSikHjt88YQIGVhVLg5rO94cBxAHDQYPTPkIdSdu9pmQRt36XJCCOG4+6cr1oymcmniIPmlRjoaC0NQfHCIJey07dR0fS1jaFlgvqASm6JA026bYbhsjeTGrobLfunMn24DKyEQQ9V/T3wm/AFo/oI020/i8XI4i4Jj1PpmCiKc9IUz4JgeqWuAGvkV7S2UiTbnVDSnfT9WlyJphM6ToWm7HmFvHRXWN/vnAmjp/SH9cfP8H7GtNwe6k55F3TpFl+xPi+nmVd1dzWiYDLEwUlMcvz2ptOnuw+dpvbsnsewzCMl9N/K3q1pGxvfvrdr64T/ral9QFx0KEyQpB/vP0GKc1tCd9Gszoc0RS2ON851rZEEpff9QbuXbJOuc4OTf8ooHPueboBFW066hRAb86k2fmh6qfUnkk7osYPqial7Qm1aBqdiQCJg/PAQ/HgV9MEiOlfjjCWP2vZvKIoamrNNABxoC1bFofBY4+sGLzEFa5bKlQDv+lDqwEAG3bvw+PvbhGeWzC2Dic3DMRlC0Z5XmcJUT+10cGfLpiJLy8Srd2BdIqfyobYQRtpChj8p2uaqPMhsc0n6/kNymOW2IiUiqapg7ICUKWPwtQ0+aVCzR+TdtMLep9ByPVQpSHty/Mxgggr2CkeI4gA97wjxtdrX6tb5kDNXyw3PU+ONInPO8dAr5lcXex019j8MXW47awGZbPuiGl4DEBkERUW3Sk5ZlI/fG62aEgRpg4xrPEJvYacmjJVjR19bW1ZDGfOzNaLsREEwzCMD/IAOiEZQQThpFN1CLUo3h9c+ebsd7NtktLzhP35zIC5tSIai1kVq7c34ap73wq9vowsmpraEq5ovOaYsaqXuAQZQURN041GbG0M7jtVrXjfzy3fiqvuXSqILpUI7Ejayrop5yZKZ29zNYIYUVuKCQMqPMvpIJamuTmfsSzi5EhizDIxob93uw61JArVL9O4NRfC2DGnG7IG36LlAdgPTpksCKllG/YIzx82pg63nD5NO8BTPZZTnOjf4/tX4Mmr5uPZqxd4BmqhapoCxEQsYgqRHPp7MHtEDYD0YNIvRc40DSE9j4qCaUP8RVMY9zy/9MVjJ/XDHy84GM/+zwLtOmHw1DRFC28E8auzG/B/F89SXhvB+wmoaZKE5Y9Pm+p7jKplC8elhRb9fusjTeLz6cdm6IbkKvJJHy6JW55JrPzd89T7j1oGTpsxWFpWOCMIZXqeLJriEeF74HXXy4im/bymKf+mAQzDMDmQSqmjJxR6Q3OiAXRmqiNpQ9YD8ajY7NLvvtYkRZoofpEmpxZLldKk47K/voEVW5tCry9Da5peX7MTp/w6G7mTu9rL6Ga1O0ikyRFCW/cGiybVIMqJEr6/OdvVYQdp/DtvdC1eWLEdiaQ60uTc1IXBeQ6GA+ccMgQ3fGqSRnBk3z9tKNqRSqEYlufGLdesRSzTLeRXQeud+lYU4V2FY6If8oBS5apmGQbiEQuN8E8NlQdglcVR31lmvyJ0WRw5UNEStUxB3EYtA6Pqy93nqEAOk55XHFCbI7+O/h5MGFCBR740F/0qi3DSL//rux0aaepbkRW6o+v1nzOgd7wLG2kyDAOHj63XPh8WeZAa3ggi/Hdq+tBq4dzkgrwb+bqk16RhiBFSVTqg/L1e+q0jXRMCKgiCapro+7cMYP6YWlQURTA508ssF3Kp7fvGceNxz2tr8eVFo0Hn6UwjdzdK+loVqrqpULbxilVUxxbGPa8kZklOheI2jpnUD4eOqs05JbK3waKJYZhuISEZQaigRgeOiKF9jToSKUAqNaE3i9aOpO8gobktqRVNqigWkL75Ohbbqn42OjojmACxp9UPH/tQeC4obUtV02TbtvveI5bpRs3CmFyoIk0O72/KCgan5mloTQmG15bihRXb081tVZGmzF1VGKjnMMCbOqjKp0jfP9Ikp4jI6XkOf71oFs75wyue5TQ9z+/c6PAYQSjed0tHMtSMtDwIqSiO+s6I+21TK5qkwnsqJKKK+hLVc7r9BkWaZHvmPpJBx6SBlcp9D6spwcc7WtzaNBpdKo1H8NAVh6I0LqYUqeZNtO55wsC965N25MPYvEffV46Sy/g8n0iKbj/emiZReAuD8xBpkZXFUTd6IVxzOvc823tclmmivCiK164/QjCgCUsup+eS+SNwyfwRAIA9LaS1hWnklTIJ6GvNVL2gCmk5LvRpijqRJnGdsnhEeF/ytktiEU/T6f2R/f8dMAyzXyBbjgevn77rteVgU/6751djm0+6WVNbhzY9TxVp+vxflmDVtmbXmrwipA10IWhLZN+33JskyIhANQudSNnue4+ahmtLHYawEbYdzelzH4+Y7sCmXeOel4005eeeN3Ww3oyADiCKIhYs00AylW2yK0eadrWoLeXnjq7FL848yHVmdKAmETU+tU86wvRpampNhHLQK5JERWVx1Dc1x69HjFYo0QGvKTbjpTPa8uw2nf3XCf0ixbXapzTmpvfGrfTzvz67Aa9+vNNjEa06XgC465JD8OiyTW59Fz0ntm2rrx+FaNIJeaFPUyfERljkgXbYSRm/AXptWUwwd8lHSDgENbel10zcEq8h1RHS445HxGaqqvQ8+TNIKd3znO3lW1OU3+dMv3OdSQ/U7d4yTc97KmRzW/q+ne+0YRgwjexEgyyI8nFt3B/gmiaGYbqFlG3nWNOUMRogjVRVTnq0HuWWJ5bj/jfVTnUA0NSW1OZUyxGoVMrG4+9uwcqtTe6NLmzD0UJA05xk0SI3C71o7nAcM7Gf+7eqSDxBxEvEMpXmDjqiESNUuoeTUpiuQ8nksCfVfZqc51UDoDCMrCvTHy+NcBCLYlc05eDgdNTEvpg+tBpTSDrPnJG1WDiuHoePrcNZpNg5u3//9yFHXVRj1aa2BGrLs4JMt025n1NlYKRJP2DU1THJKZSiaNLX9sRC1DSpBuo0kucMOI+d3B/fPnGi9r3JywdWFePieSNw0JBqd9lZs4bgoCFVmDOyVrkNlaW/bqAsNrft+gGi/LYvnjs81Ov8xq6Pf3U+Zg7r4/6dS3qsZz/S347YdaDnKCqJILX1dfaxHE1VXXNyGqXbp6mT7pyFgH7fw7aNUKETIupIU/BnqTTgUJwiemrp7xB9fZl0z+mhU93lHKBvi2GY3kYimUJrQHoeHcwmUzaSKRtbSM1NeyKFdzfucQe/7YnczCWa2xJadz050tSoSNnq3khT9jjl/VqmOEicPLBSSD9SpUF0pFLZSJNl5GRqETFNHJZxAQtDPGK5A+bbF3+EzXu8Dn3OwCeM5fgXDhvpPn7u6gV46bqFvsIgJqWFOftyjSBy6BUSj1i477I5+NYJWTvh0ngEd5x/MP54wUwMyzQKzoUw6XlNrQnUlGYjWgOqipXbkiOGQTVNspOXcBw+QskhahrC8dJUUHmgJhhBaGa+VR85rasJa+8eRnB//zOT8cDlh2oFnGo4q9t9LKQRRKGgA9S+FXF8nnwnZI6dlJ1A8ZvxrymL439Ib67ORMyCjCDaSI2mnKat2q0pRZooQkPizMPK4iguX5A9J26kSXDP83sHagoRRKTnpjM+CPpIk7emKcxnqdqe6loW6sJMeu5J2l6cI00MwzAFI2UHp+fR1LlkysaP/yPW8vzu+dU4/tbFuPLepQCAxlZ1WpWOptaE1vBBTtvb3SI6wRmG3pWuK6DpeZ5ZPEOMNEUtU5gBVNU0CZEm4p5HcRrBykRMA989aRL6h3SKo0IFAP7y8hrPOk6EQUzP827r2En90K8iKx6G1pSif6VaQLjbJgMIIeqVyj3SlD02taDIhzDueQ1DqwXDCd25p1FI00jXFvgJiJH1epEnfhbeuhEgHaWk26d1fnJaYJj0PNUgbcqgSiwcV49TGgZpjRg82ylAtEcVBdCdS/H71x2Rpuw+5o+u06Zg/vGCg3HL6dPcv4OOjI5tO3NdB6Xnfbglaxgj3weCGqrKkSbaKoKmsF5zzDj3sWs5rkkzDYtTM9cb8Is0eXu/5RdpUl3LpuYaESdPvPeoAxEWTQzDdAny/TeM5XiHFGmiJgMAcM9racc2pyns3oCmsyPqxAFiU3tC21xPHkjLdS7FUStU35xCQQcGckqhaRiSo5khDAiUNU3JlCsEK4ojHgODX5x5EI6e6G2SCKQHygOqivHi1xbi/DnDAo89ToSKDtdyPESk6ahM6uF4HxtwihhpstwB5ptrd+Oyv76OFVsbdS/VHy85tqD3FtQcVB7w0gHOpIEVuP748fjyolFCvdQAjVCkoskpxvYbHPbzcUcLIwwjlqFsdql6jWg5ro5wqURJPGLhjvMPxk9On6p4hZqbT56C8ngE1x8/PvRrZFSZU7rBpykJya7GILvw+x06fGy98P0PGrzSZwtlBFFR5C/c/Y7BXeYTaaLfX13EXOWel89AvoGkd/Y0fpGmfK5BZXNbxXZUNU3Ofh3kCcUDVDOxEQTDdAetHUn8v7+/hUXj6nFyw6CePpxOs2H3PgyoLPKdzYpFTME2O5UC9rX7p0UlU6JQCHLb27vPP9KUStmoK4+75hDvbtgjGEtQZGEiR5pUHc8LhWF4B2w0PU9OKbSkmqaoJFJUluMdKRsbdqUdtwZUFSNmmUIhbyxiCkKNEiXFv2Ec3WIRMzCtSmUVrNIitp0+3re+dVTohp50UBmPmIhmztXV/3gbAPDoO5tDbYfS2Rlr3fHJDK0pxcXz0q5bdSTSRHtD1ZbFsT1j714hiSbA3+zB7ztL36MuwhMxxZomWucXNmVSt093H3lEbiYNrMTSbx/VqYG/rYo0hRFN3WEEQR7nYmQQNHilz+drdCBvpzrXPk+q9DxyuciRJt3EF8X5XROiJHlcV2fMHIw7X/wYfSviwSt3MbrPJ98oa9iPW3et03PLRhAMw4QiFSJJ+a8vr8G/3t6Eq/6Wf7PT3sJfX16DQ29+Gjc/9gGe+WArTvn1izjr9y/jtmdW4vF3s4NROXKTtG1ln6YPNu/Fr59dhbZEUoo0pdCqGcQ7yE1JZRIpG49+ZR6+e9JEGAbw8Y4WPLR0o3Jdmra3s7kd5//xNeH54pjVZTPKKpc0WgMgixnTEAeuMUtMh5PT9QDg6Q+2ojkjQgdUFmNwnxJcfXQ2ncUyDGUTWkAc2IZpmkijOzqc9LywYqSyJBr6/MsGBMnOWFY5x+bjEveT06ZiMk3jCRgv+Ikm+lIaMWghhig0VY+65zkCqihq4d5LD/Fs+5J5w32PK1SkSXLPo5Em+pkfP7m/WJukTc9TpAjlKUo7I5gAXU2TTjRlH4f5TnQWOgiVHROz63iXBQkhv35kuUCPTxX9uf28GdrXKvs0+USaHPwaUNvwuuflOpA3DGBcvwr8+8vz8PAX5+b02q5Ad/RyhD6s22nYr4uYnqeuaZLvYd0wj9AjsGhimE7wnX++i0Nuegq7mtt919vV4v/8/sR3H3kPAPDb51bjgjtfw+trduHFVTvwo8c/xOf/8joWZ3rzyJGbZMoW6nQcjvnZC/jBYx/gd8+tlmqaEGgcEdRjKJG0UVsWx3mzh7nucut2tSjX3dnSjtXb0ja+d774sef5kmiky2aUy1WiiZyrdkmAeiJNUt+TqGV4Zva/+eA7AICqkqg7GL9k3nD3BjuirhQnTBmgPL6oJEKCyCU976ojxwAAPjtjsNIUwFYOZf2JSul5e/cF96IKQtfsFQBOmT4ID39JHFT96cKZOOcQr7MeIPaRktFZK589aygipoGTGwYKIpoOBIf0KXEfzxpRg8F9sil9Pzp1Cr52rH/qms49j34GcqSpgkSa6Nfje5+eJGxbN/ANW1fRHajd89Trmppz1VWIxgjqSJO6iN9/uxVFUSy5/ggs+85RnTo+et2qnDmPmNAX/3fJLADA4WNFU5kgIwg53fhXZzdgyqBK/PqcBu3xOLcSM8REgA7nGCYMqEB9nk1/C4nu8OX35deOgSILaq0YF377aGYAyW5Q1N0eiLBoYphOcOeLH2NrYxvuemWN73q97Qdkb2uH2wfFD6XTXMAYdum6XUJE6fmrD3cf6yIZAPDaml1CtCeoBmrL3lbsDTCCoMLNmf3UbfObD76DhT95Dn9bsk4ZPUxHmrpINClc+VJ2tgmrNz1PcjSzxJx2S2FB67Cb1GpFLBOLrz0cj391PkbUlWHm8D548qr5nvQ+WaCp6ENScmKSEYQK5/l5o+vwxjePxM2nTC6YC5mcnpeLw6IO4RwEDL4MAIeNqcP3Pj3ZXUZfEjbSNG90HerK45g7qhbj+1fg9euPxI9PnSoIanpcVDQBoiHI5EGVgYP7MJEmU3LPo9cu/dZEpfeoLWJXfKe6Q4SoUAUkdel5nUllywcjxPWj+vqEOczasrjyNygXhPQ8TZ3RnJG1ePZ/FuA3507Xvla1THZ8PG5yf/zzi3MxtEZvamIr3PPCmoq46/eu27ZwzVkaIQMAi8bVh9qe/J3UOWvS1XQpvLL5UG8b8xQKFk0MUwCCbqDdfYP1w7ZtTPnOf9Bw4xNuzdBzy7fhuvuXCWLn+geXYfQ3HvU09gzCMAy3lskwgKrS7M3Yr0ap3ZOel+3rpBokzPr+U7gmU6Oig9ZIObP7QXVSt/xnOeoV+eslMatTzR/9KNP0f2pLpISmrA6mVOwftUxhIB+19PVJMuVFUYztl03RGVVfjiXXH4GF5MYrpAJqBmxHjs+aSFDLcR10UN6nNAbDMApWPEwjFYUy78i38P/rx41D1DLwy7Oys+J+gpIO1ErjESy+9nD8+cKZANIpiqYpplHSz2ZIjSiaKH1CWMzrirw9x0gOn0aaqOiQo0WqKDOgGVz10O+lKtM6THped+Bnwe2gEnjdNXil50PlzOkwrLbUEylTpefR41bVaAah6tOUb6SptyDUEBGB47yvX53dgIvmDsc5hwzNeXuAulVFej16DtWumvJnxH2aGIbJG/rjlND0Ceou6IBr4559aGpL4HN3vIq7X12L/7y3BUBasPz15bUAgIffUtcA+eGIr3jEFH5YWzUDJyBd3EvPTSKVrYHK56YJiHVVzkC1JUA0bWlsVa4TtYJTzvJF1zT30Xc2Y8K3HvPUYcmW47GIKUSeIqbh1i/J/PDUKYHHUxKLYFB1NrVL7LfjPQeVxVFMJs1f41HTN63OMNSDUZXQy6ccSUzPK8xtjh5HLtfBpfNH4t0bjhH6XOls7wHvBEs8YnlmyOnkBl1fjjSt3tbsPg6TXqSbvZY/A11NE/3M5QkGXZRZNZDtqRl+lRGEbsJLF4HqKujudEYfqkhKd03YCTVNxbkZQagO0S89LwzqPk37t2ii4pKeE6em6bjJ/fHNEyaEfp/y+wtTK6czgpCvyd527goFiyaGKQBBvw/0B6Qlj1Shx97ZhIv/tMTj6JYLK7c24sE3N6CZ1AFFTAMPLd3gWffPL30s/J2L0DOM7KCuOGoJP+B+UR65DirYRchyAAAgAElEQVSZUhtH5ALdnpMuFCSabBvKZqxbG9tyNoI4dfqgULUCchGtc86++eA7ysGmJdWVRC1TuKHqjvPgYdU4fcbgUMeuG0CrIjej68tQQ9PzLNP3PEdNUzmYCzL+yIe4TzNXFbOG91Eul+t6/JDfWixiCsLfL001zFBDfv3sETUYUVeK2SNrhOXtOU7Q0FnksC57tE8T1RzyAD6MyYi771BnofAo0/M0X/lcU706S5iaJvW57B7ofqpLc0v1C07Py93oOdunKbssd9GU8267FHpOqMV3vrW28nnXiVOhuS25F9C5H/maPFBFE1uOM0wBCLrJ05nllrYkKnLMH//CX98AAPzsyRX4zqcm5vTajmQK33zwHbfH0fc/k62zSKRsYTbatm0s39KIGx5+T9hGayKFsszdJ6gwf8OuffhrpplpkSSaqAW5THsiJUSG2jpSbrpeGIdCFfS8O9GHfe3BpgDrM9bcZ84cgrtfTUfcNu7el/PNqbokqnTGk5Ftm4siJprbk9panHSfJiqaxOPSHWexJv1CBY3WBNU01ZbFhZqmeNT0Fci6SI3KDj6fT55GDILSBB1u+NRE2LaNE6eqzTDEmd3cBwRUhPimTobYtCxA7ro4XWAvD+RnDe+DVz7aiVOnh2tzEDp6Qj4Uen37RQV1Vv+qc9ljkSbF1aYb/I2o1dfTdAXCrL40cVFdEsWulg7MHlEDme46l/T61vVOCvNah86m5zk//UIdUI4D+d428KeHU9LJ3yPA+/5KNOKUrkUjyDRtXL4me5vgLBQsmhimAAT9ttK0tJYQg3YdTr8hytodLRhYXaz94Xzq/S2uYAKAdzfucR93JFPYsjcbVUkkbWzYvc+zjX3tyVCDfwC465W17uOiqCXcqHR1Dc6xUPc86oyXr2M03Z7zox7GFMA5BzFhVs3OWTSd3DAoVHqM3CA0HrW06XWAE2miKXOmcA3qREmxT/8e1T6yx+TvnlddGhMascYjlq841Q1eVJHF4XkMTgVDgpCpdHXlcRw3ub/2+fryInz7xAmIR6zAPjlBz/tFgMJEWUbUlmLF1ib3vemiHreeeRD+8+5mnDo9ZHQx5Lmi74/W4/l9TXOJNPXUYFU1N6M7lhnD+uCHp07pNvHk1+z1wSsOxQNvblA2nu6uc0l/V8f0LQv1mkNH1eC/K3fgrJlel0ma3ZlLet680bV4YcV217lSbJ6do2gq4Mj/t+dOx1X3LsVPPzst723ohGShIk1FmvOsM6DwF00Hpmpi0cQwBSDo54H23AlKD/Pdj7Sjf729CVf83xs4dlI//Pqc6ZrXiC+ig5z2RApbiRBLpmwhfc8h3zS5omi6HsNp3uqfnmcL6XT0OGT78rAcOyk7CHYGmGE2tSFjSx6LmLjzgoPxtfuW4eZTJgeKt+KohROn9se3TpyIbY1toQf8VJBGLTPw5m6Z/n1idIYVukJfFfRGLEa1vNuuKY0JjlmmAZx+8GDcvvgjwa3PRfP2aCTy7ksOwcNvb8SXF40OfcwO9HMyDAPl8QgaA+zpw/TaueDQ4b7P/+rsBtzw8Lu47Sy1FfLcUbVYvHI7PqWJZgHhZmh/e+50/OQ/y3H54SN91+tbUYRzZw8L3mAG3eBLvuwrS6L4zTkNiEXM0I1Wpw2uwnPLt3mWK6/1XlTT5PddDJvqWmjkiYuhNaX46hFjlOt219j1vY173cdjQ/Z++vOFs7CrpR21ZV7jHTqxonN1U/Hbc6dj6brdmDmsj2c7PZmed/TEflj2naM7JcSElMVYuKbSfnjd84LbAtBJKHpflscZ3Z2+2l2waGKYAhB0Y6JiIVfRRG/k8o/jb59fBSBtGCDj9I6Sb7BUjHQkU9hKI00pGy1t6eNbOK4eS9ftxs7m9rwtm50f4YhpoCNp+9asdCTF9Dwn0lQctQL3X1US9QzOT5o2ADeSXjFhBsXlRRE0tiawN9M0N2qZWDC2Hi9/fREA4IUV4qDvyAl98cHmvVi3Mx2ZuvLI0bh0fnogGzYyJ68btYxAkWoahuhUJs/yaW5YRTkMPmh0KUykiabktLQnUV9ehCXfOALf/ue7QvQR0I+JaUR29sgaT41OWOR+Ow9/aS4W/PhZ39cUojfQcZP7+0ar/nzhTLR0+EdtwwxyR9SV4baz9T1q8kU3+FJNFhwzyfs+VaLD4bIFI1Eat7BwXF/c/sJq3PPaOpw3e6gy6thTwy3V4ffGsV9YoQp0nxHEBYcOw4urtuPaY8aF3qdlGkrBBIjHnUukqSQWwZyRte7fnemnVehoSWeFhM5BMd9WDfLR0Gbn4n6zj+k5lBvY615zIMFGEAyTJ7nU2dDBYHOO6XlUMMg/4rqbQFsiiYbvPYGD//dJT4PYPaTRZ5sn0pRyxUppPOLO8FHRl0uq3JjMjKNz3H6uYc1tCWV6ns7Rh1Kl6IB+4aHDhdqxMKJpULXoPia/ht6cLpo7HL8/b4aQB67bx5WaWWCHCnL8EcsMFE2WaQgD1KgVrnQ+lxnbPqSYm9ZcqWqEakpjwrXoCPOIpY5E6K6Czhp/6BhWW4p5o2t91wlzfXQW0zQCxXRPmSAAnXeE8/ttKIpauHT+SIyqL8MNJ03EPZceguuPn6BMCeyp1B7V4ffGGfMBVeEbrXbX4S8a3xfv3HA0Pn+Yf/QzLIK9dp7uqQBgaQb8oY6hF372DnTyKmxarQy93pdcfwRG1avTKkXb9ux+/dKMpw2uzuuYejscaWKYPBFC0wEDHToYdCI5Kna3tCNimcLAikZQ5DQ1mk6zfEsjfvf8anxl0WgkUzZsG0jY3hqlPfuy27tj8UdC5CuRst2aq9KY5QqWq/62FBMHVOLnZ+SWj32wkyIR4ubT2JYQCuRppGkX/JvYVpbEgB0twjI5IhLGEKCuPI73N+m3QaMRjgCJRvxT1wDgK0eMRmVxBN+RDDbc4yeiyTIMoV+VCtMwhDRDVTreP74wG39fsh7LNuzBe5vSqTO5DD5OP3gwbADNbUkcOiorOFSRJrkvC41mquyRG1vVEwd+RiG5oBp0qz5/mrbXHaIpDD1ZCtDZQWKQSYxDPGLhkIxpgWrg1VPnQI5QAt1vLe7Hny6cid0t7b5NXWW6U4Tnkv4bBP0O5xIhl6E/paM1okB/DHnvtkvQR5ryO1B6vfu1ZghTL+vw+Ffn44n3NuOiuSPyOqbeDosmhskTGhUJTM8TaprUA8bmtgSmffcJmAaw+qbj3eVU5DS2iuKBzvqc8buXsbO5HcvW7xHS0hwnONX2nnx/q/BcImmjKSPqSuMRNy1i1bZmrNrWjOuPH5+Tm9n0oenZpjCiybbFY3MG3kUZ8eY3oK5URJo8oilEvx55O/JAm74P59yIPYz0+yjzcUwsjWcHBWFsotOiKftJqAa7M4b1wYxhfXDuH17xHHMY4hEL5ynqYVTiwomUHTOxHx57d7PQXLFQfZJy4dQZg/Cnlz7GonHZhruq464qjRLR1DtGSL1mjJ6n+UquqH4aeizSlIMRRE9Ae32FJZmvi04P01n3PIePtje5jxuG5Bb96E2fPSD16qKRpjxFk9B7zifFT2xu67+vsf3KhWbpBxq9Y2qNYfZDgqIBFCHSpKlpWrk1/eOessXUPxppkmfo6azPzkwN04dbGgVHvHU7xQjM3n36qI0caZLTuVo7Ur4pdpRvnzgBwzJGCGF/1On7a8o8LopYuOfS2b6vK1ekPMk3vDCRBDnNTx5Iq5qm0nNEo04yfumc9Aaos2amWKbhMbTQ3d/pTS6X9DwdKmHoREZ/dXYD3vrWURhNCsFzqb8oFBVFUTz7PwvwrRMnuMuUETLShLP3RJp610ANgJvaGPQ1zmd8XhKL4IyDB+MzBw10lw0kjZW7E1WkqTenaIVB93v9m3MKXxNXSMQ+Tfn/hpzSMAgxy8SXFo7K+bPsbaJJ16sr32ioOPHmt9/s41x7FR5ocKSJYfKENnwNGuiEEU009a4jlULcTP8o+kWadGJEEE27RNGkapB7SsMg3PfGek9Nk5wWccsTHyr3J3PYmDrBaSxs+oAgmpz0vJiFaYOrcONJE/HNh94V1r/66LHY29qB6pIY/rVsk/CcPDsZJpIgp5nJBgtUpDpRG7ofv4G3/4xvdrthGrxaprfoXpeGQ29yuUSadMQkYVhfHsewTLqQaRqolM4hFYRRKzj1sFDI38mgtMJeI5p6+gAy1FdkC/SPndQPd5w/A+P7V/i+Jt9P9uZTpgBIm7es3tbspvV2N6qvaJiayt6Mynn0a8eOUxp59CaoQOjM79ZBQ6qx7Iaj8pq8ydNfocugvw0Dq7ITC/kK+66INB3osGhimDyhN6MgUwgqmlSW3oAowtoTKfdHfs++rMjxRJpCiCa/9DwgnZJWHEv/YFL3vBJiBOHw4NKNyv3J0EanQPgZO2p6QWuaACCumG08f84wlMYjuP2F1e6yM2cOxsHD+qBvhVgsLQsgFXJ6np8RhHNc1PrVb+CtS1noX1mESQOzg9EwkTzZPc8PVR1WZ6CDj4OGVOHeS2f7pj7SmqZx/SqwbMMe7bo/OnUKrv7H2/jpZ6d2+jhlVJ8NtUnn9Lw0f/jcDGzYvQ8TB1S6ywzDwEKS6qjDzz0vDAvG1mPB2E5tolOojr8Q35meJEnSyB/9yjy8sGIbzp/jb53fG6grz4r2ztZK5Rvt7m2RJjoR1K+yCHddPKtTqYtCpMnnrer6NH0SYdHEMHlCG7sF9RGiYmBHs7dBrbwNOhu/l7jdyaJJ9wO2eW92H3JkSz7UmrKYKwaSKdt19yuLW3nP8NEaHd1xlsQsX/t1J93Q2ZYqLcw1YyDPHTWxHw4fW+9ZN4wRhNzJXq7HUQmQEjKo8ttHw5Bq/Pbc6fj8X14Xlj9/zeE5RznS6XniBzm4T4lyXUHoFSDSRGc41+1sCawVowOWG06aiGv+8TbOVDSzBIDTZgzGiVMHdKrwW4fqFFOR3FsiTeP6+UdzuppF44PFkY79tHzGRXX4XXEtdifUFnp8/4rAaGFvYc7IGvzzrfQkXU8J194nmrKPo5YhGPTkg9zPToeYnte7zkl3w6KJYfKE3ozobJ4KamKweU+r9FwSl/x5iRCNooKMGgM0tSWQTNmuCNGF1GnvpSBqS+Pu9hIp243wlMS8kaawlEozg7Jouu+y2RheW4aGG5/QbuODzY0A0j1pAPUMvJOWQLevi77Jg/tvnTABe/Z1YN3OFtz/5gYAqpomvRGE0z2dCpGgG8rRE/t5lqkG69ceMw4/eOwD7XY6krZHNB09sS+uOWYspg2qEpZHChxpoqkg25u8qZ4y9LwPqi7Gk1cd5rt+dw5SY0LqYM+KpoeuOBQvrd6BMw7umYapjLqmaf+PNO2fSpb2WiqJ95Ro6pHdavFraJ4Pqus9iHx7Qh0osGhimBxIpWx30Ejd84IiTbRXEo0CAcDfX1+PF1ZsF5ZR6+0OyU3tvjfW46RpAxCPWNpIk9zo1Y90pCnbR8lJzytT1DSFpTSuF02GkY66GIaBMX3LsHxLk/xygbFSrycHGtWhESDdj7p8kxlWW4KF4/ritmdWusvkehz5NfRvN9IUsqYpFy5bMBL7OpK49akVyufbEynPrL5hGLh8wSjPuoIRRAEiTQAwf0wdnl++DX0r1I0pKVGy/zDRvu6gqiSKhiHV0iCkZ0dIUwdXYergquAVezH75/A8i2oMqbLM35/YX0XTkJoSXH30WLQnUkK/ve6k10WaSFVTIQwZwl4Z9BL6pKfn7d+/BgzTjfz48Q8x43+fdPsedQiRJv+fnzYigtbvahFy52V3O0CMLsmi6Zp/vI2fPpEeTOuiKrv3BUcAHGrKso1J05bjTqTJEpraUobVqFPBHOQ8a/pDG7VMNxXgV2c3YOqgSvghN8jNbkctlHSDX3l5TWl6wE/r0eRIkxydUjnRdYVoAoCKIv2cVm15LPQsIa1hGxrwuYXl1jOm4bzZQ/GHzx0cuC5N++jpaI7DK19fhD98bobY66oHrNEPNHrCXr6QqH7G9/dIU0dAFkRv5orDR+HKI/0bg3clvU000fnAaAHES9gaRLmRuue4etdp6lL27184hulGfvnMSuxsbsevMpEJKpT8Ik22bQsiqLE1gTN+97L79w5FitM/l25ERzKFJR/vxB//+7Hn+d88twpH3vIcXl69Q7nPXCJNfUrjrhhIpFJZy/F4BNub1PVXc0fXYkxffaNAT6TJUEcbRtWX46EvzsWCsfr+IyPqMs5s0g8zHeTSFDTdDJwc5agpS9cvUVc7b6RJ3KnKiY4aQYSJpNRnCpynDKrEI1+a6y6X78+qqNDfvzAb91x6COrLi0LPEn7moIHoUxrDTz87FfXlRcEvCEFVSQzfPWkSJg30F7yAVzD3BuIRC4ZhCN9hVYNgJjduOX0aBlUX45bTC2/k0T0o0vMKFJ3tKZLd5FZ5INLLNFPhI00hLw36O6mKNH2SbMg5PY9hkHa0O/+Pr2LR+L74wmEjfdc1DQMrtzbhlY92usv8Ik2q5175aKdbm7RTYQzx86dWIGIa+MkTy7XbXbFVn9bWFsK22qGyOOr2REqkbDRnokslMQvbm9URq34VRbB8Bpl+6XmqmSoarakvj2NrY/qcfPWI0W6K4IQBYgFzVEjPC+6OLg/Ya8u8kabyuH9zW6URBI00+fRpcnjy/x2GNdtbMFmKsBVFLMEw5DMHDcTdr65FWTyCl1enr7WDBle5N6iwkaZjJvXH0RP79Vj/n96SAqeyZA+aQWVyY8KACiy+dmFPH0beqH7Gi3qgz1ghCUodZ/T0tlQ0+hNeCEOG4ZkJySDoxKJqcqm3pF13B5+cd8owPvzl5TV47eNduPlRffG9g2UaOOKW53DjI++5yxI+s3m6m9aOTBTHSfeT8RNM+aBLM6koirg/wK0dSbeeqiwewdkal7PSeETpRuY+H5CeJ0NrpypIitzZs4a6j4fWlOL+y+e4f8c0QkkXzaCpdrRei94QiqKm5FDkYwSRZ3peRVHUI5gAb+1ESSyCR740D9/79CTl/nPJuunJhqk0xaW3NW6ln31vGyAx3Y/Scnx/jzSxaMqb3paeJ6Q6FyAyPq5fBX5/3gwh40EFnVhU9YT6JDnqsWhiGIh9jYJQzUj7uefR1Ly7Lp7lPv5oezOuvHdpoBFCoZBtwB3Ki6Ku6KCW5iWxCE6dPggPXD4Hp04fJLxm6uCqnCJNZXH/XkZU0NFUvmopXa5hSDXZjloo6X7A6TpOah4A0JIxwzCEY5FrmmKWiamDqzCqvsw9tuKQluNBHDsp7awn1x2Nqi/HxXOH45pjxgo3zc72xOkuevOgk44ne5ugY7ofjjQxlN42j9IV1t9HTugbmGYddAl9khz1OD2PYaBvOKtCNfvk3Jia2hJ47aOdOHRUrTvgplGo2SNqMHFABd7duBdff2AZVm1r7uSRhyfdINCbbldRHHFn2fdmTAOiluEe/0FDqvEQaWp708mT0TCkGn6/2bLl+OfmDHPTGVUz+jqXPr9caZ1Q0lqOk/XryrKub3KaW3E02z9KFniGYeD+y+bAtm332EpCNrcN4vrjJ2B8/wocNcFrS379CRM8y/aXsdDMYX1w/OT+GBkyFaSrUGmioKbUzCcL1UREUWz/HhAmkvuvEURPo4qq9CRNZFJzQGWxz5qFJSgVPPYJijSxaGIYwHWMA9J2znKEgVqAq35Akikb7YkULrrzNbzy0U5cMm84vnF8eqDr3LQs04BpGqgvj+NdoMsFU2VxVHBO03UOryCRpr2t2R5NFDqYOKUhHXXym12S+2rQ/kT0mBzoObVDWhyIdUxUQGksx0m9US0RTXJqZXHMAjIfjSpylBZ91MY72LkvDKXxCM6bPSz0+vn02OgJTNPAbWc39PRhKNlfziHTPaguh/29XoMjTfnT29Lzlm9tdB/LpkVdiW5yaXR9GVZsbcIJUwd027H0NPv3rwHDFIimtmwBvirqRK23VRN37ckU5tz8tBtNueuVtcJzQDYCUigHsyDqysUeOnLKnENFURRWZmDgRJrkmiT6m+kISlp7NEHqMl/mYwShEk1UuNBGwH5EI2qxorNipSKrtjybnqeKNGX3EXzTFAwputFymYdCnYcn4RnKLzPingql/T1tkycGcsdpg3H6jEEBa3Yv63aq65+7Gp3uvvvSQ/DzM6bhqh60he9uWDQxDIDtjVkHuyaFaGrpIJGopLd3UVNrQrDnbk+k8MrqHWhLJF1B4Ayu60M0BC0E9ZJo0kaaiiPZSFNG0JRIokd14/32iRMwrl85fnTqFNxx/sGYO4p0cM+xjqVDsGQPZ5dOUwJoGoUu0kTT9urKssLVI5piudUo0ShXd85KOzd2XToiI3LYmLStPf2MeEDJUA4bU4fl3zsWnz14cE8fSsHoYMvxnLn70kNw/+VzcPqM3nUdOBN6A6u6LzUP0P9O1pbFcdK0gdr0+gMRFk0MA2BrY9YIorldIZraaSTKK5pki+9EysZnf/cyvv+v911B4ERDqktintcXkuG1pRheW4pzDxkqLNcJmbJ4tqapMSMYvZEm74/m4D4leOyr83HajMHoV1mEKw4fRfbljWr5uZO15zHlTyM8dMu6Alk6Y0wjTQcP6yOsR29IYSJH1GCjOwXMzadMwUVzh+PfX5nXbfvcn1k4rh5/vWgWFl97uLuMRRMjE4uYQjo288mjJBZBw5DqXhdlvPOCgzFvdC3+ctHMbt1vkn8nXbimifnEk0imsIP0I6LFlg4tRCg1Kp6n6XuUP720Bp89OG3b7URA8p2VMYxwzege/co8FEUtT8RGNmdwiFimZ7Avi54w9tbUMlslkGpKY27/JT9+fc50XHnvUnz3pEm+61HRRE9LGCvW4bVZU4ITpvSHZRqYnHEQWjiuHo++sxlAuMjR0Jq0QC0rinRrk7/asji+qTCIYNQYhoG5o2uFZWzHzKjIZxKHYbqaWSNqMGtETbfvlzVTFhZNzCeeXS0dwo+CMj2PRJ9UNU80fY9SWxZDIqM4nAF4PM+6l4hphEq1cPYjm1nI5gwUWeTI9uR9K4PrsA4aXIUzZw7B8NoS5fP9q4q1oumri8bgmQ+24rzZw3DIiBq8dN2iwP3pnOr8rFhvPnkyPtrRjNnkxmMYBo6b3N/9+7jJ/fHTJ5ajvCga+rO68dP+Ao/pnfBggFHBoolhsvDkUhYWTcwnlo5kClsb2zyRJVX6XUsHSc9TpO/pIk0j6src9DxnMK+LNN176SH47fOr8fQHW5XPR0wTHYp6KhmnvkeOuKhS5hxkASKv+/n5I7BmR7MgLmQMw8BNJ0/WPv/jU6fgrNtfwRdJGp/DkJoSvH79kaEsXsf0LcPyLU04pWGg8nm/FLkzNM16KaXxCJ76fwsQtYxel57BFBZOz2NUcHoew2Th38kseYkmwzCOAdBk2/bizN9XALgEwHsArrBte1fhDpFhuoYL73wNL6zYjv8nOb80tXmNCKgoUqXv6URTXXncjQ45g3lV9CIeMTFrRA3+sPgj7fFGLAMI55EAIC2eIqbhWs6qapqOz4igoEhTaTyCn59xUPidKxjdtxyvfn2RVoiE7Ylx/+WHYuXWJtcIASi821VvbsjKFA6eQWVUzBtdiyfe2yI4aTLMJxXWTFnyjTT9CMC1AGAYxmQAPwFwC4DDM/9fUJCjY5gu5IUV2wEAtz69QljepIo0UdGkTM9Ti6ZUyiZGEPqaJmfQ7zeGy6dxaixiIpE5dmoDXlUSxeJrF7qGD0E1TYWiEIKmLB7BtMFVwrKJAypw/OT+6BcijZBhHHgGlVFx1swhqC6JYcaw6p4+FIbpcXhyKUu+I6PhSEeVAOAUAI/Ytv11wzAaAPy7IEfGMN2EXCfU3JaAbdtobEugoijdi4jWNKlrntSiadW2Jqzd2QIgK3qoYYKD49LmN4jLx5ktvc/0sdHoSVHEEkSUJ9K0n0VaDKP3NlBlei8Xzh2OJ9/fiiPG9+3pQ2F6ERHLxImfoIadDOPHjGHV+MvLa3r6MHoF+YqmdgBOtfcRAP6cebwTQIXyFQzTxdz72lrc89o6/O7cGZ7GrjK2jzhpakvgh49/iN8/vxp/uWgWZo+sEUSRSiDpcuCXb2nCNx54BwCpaYp4BYmjWfxmdGaNqMHDb21UPlcas9CsOC4anaLuebJwi8j1T5pGuAxzIDFnZC1e/cYi1JZ2T+80hmGY/Y1PTR0A0zAwhaTEf1LJ1x93MYBbDMP4JoCZAP6VWT4GwPpCHBjD5Mq19y3Dm2t345Ynlgeuu0+TTgcArR1J/PrZVUikbJx3xysAgN0tORQTafCLNDn4RZounTcC1x8/Xvlclab3E62fojVNcl3V/h5pYph8qS8vCl1PxzAM80nDMAycOHUAhtaUBq98gJOvaPoigASAUwFcZtv2hszyYwE8VogDY5h8UVmCy+xSiCCnqSk1dehI2mhLJLF5z75OH1fUJ9LkaCU/0VQcs3DxvBHK4uQhfdQ23zSAVFOWFVZyXZVs091VNU0MwzBM93HStHSa4flzhvXsgTDMAUBeIyPbttcCOEGx/MpOHxHDdJIwk8a7SDNbh+G1pdiwex9aEykMqCzCxj2tANLNbJ3HnSFMpMkvPS/mGkmYnkjZlxaOgmUaOH6KaAm+rz2bNji4Oius5HTCIPc8hmEYZv/jB6dMwRkHD2FTC4YpAHlFmgzDaMi45jl/n2QYxoOGYXzfMAx1npB6O/MNw3jYMIyNhmHYhmF8Wnr+zsxy+o8jWYwvYRzaVOl2wzJNWVs7kmgk0apkysbmAogmp25IFWlySPm0B4lG0u9LbloLAJUlUfz14lk4U+pDRKNu9RVZZ7kPNjeK286hpxPDMAyzf1AUtTB7ZE1e7qsMw4jk+y36LdL1SzAMYwSAewC0ADgNwA9z2E4pgLcAXOGzzmMA+pN/Z+ZxvMwnCFkzqUwfdrV4I03DMiKoA0AAACAASURBVPm6+9qTgkNeeyJVENHkpOflW9PkRqoUoks2cnDwq92icKSJYRiGYRhGT76iaQyApZnHpwF43rbtswCcj7QFeShs237Utu3rbdt+wGe1Ntu2N5N/3DiX8cVAVgDc9Oj7mHPz09jR1Cass1sSTbGIif6V6ZqmVz7aITRz29rYhvZk5zvEZ0WP/ms3YUDWfPK82UPx6WlZ21snGqR6vSx6dHz9uHEA0m44FK5pYhiGYRiG0ZPvyMhAVnAdAeCRzON1AGo7e1ASCwzD2ApgF4CnAVxv2/YO7YEZRhwA9Y8tL/DxML0cGmn67XOrAQB/evFjXHXUWHf5nn1iel6/iiIUx9KXtNy3af2uloIclyNM/NIHrz56LIpjFk6cMgCTBlbi4bc24sGlaZtxv/S8sD2cLp47ApMGVmLqILE5rNc9j0UTwzAMwzCMQ76RpiUArjcM41wAhyFrOT4cwJZCHFiGxwCcB2ARgGsz+3rUMAy/3KHrAOwh/9gC/QBg1bYm/O21dfjy3W/ig817fddV6gdJqMhpa/0qi7S1Rut3pZ3zVK51N5400fdYKDGfnHJHppUXRXHdseMxaWC6HwI9bL9IVdhIk2kamDOyFqVSHyZZdJVweh7DMAzDMIxLvtPJXwVwF4BPA/hf27ZXZpafCuDFQhwYANi2fQ/5c5lhGG8DWAVgAYCnNC+7CcAt5O9ysHDa71n0k+fcx/98ayO+vGg0rjpyjHJdmp7nEJVEAXWVqy6J4pJ5IxBXiCIAWLczHWka0qcEH27JGihceOhwTB/aJ/R7kFPgwkDfiyNsVDVNQUW+fimBAEeaGIZhGIZh/MjXcvxtAJMVT10NIFzleX77XW0YxnYAo6ARTbZttwFwC1jCOKkx+x+3PrUCX144CpGMWKBmDypPhAgRFa+v2YU/v/QxAOCLh4/CV48YjYhl4r2N6gjWukx63pAaUTRFI0ZOQkhn1iAfP4Vevs61rErPC4o0lcX9v+r02AwjbWvOMAzDMAzDpOnUyMgwjOmGYZyT+ddg23arbdteL+cCYRjGIAA1ADZ11T6Y3keHxoSBNqhNkP5GjrjYSXoxRYm4OeXXL7rrVxZHXUGlEwrrdqbT84ZKDWTjlhk6LQ5Qi50gVFtXRY10NU03fGoiYhETvzjrIN/90PdRErV4soFhGIZhGIaQV6TJMIx6APciXWO0O7O4yjCMZwCcYdv2tpDbKUM6auQw3DCMaQB2Zv59G8B9ADYDGIm0nflKAI/nc9zM/gntNUTZ2dyOuvK05wcVVqaRNm+Y+4Nn3GWOKGhsFTV9USyb6lakS8/LRJomDKhAeVEEja3p44laZmgDBiC8WQOlutTb9kwZadJEvD43ZxjOnDkkULBRUalLU2QYhmEYhvmkkm+k6RcAygBMtG27j23bfQBMAlAB4NYctjMDwJuZf0C6FulNAN9FOs1vCoB/AlgO4A8AXgcwL5OCx3xCaNKIJmoj3pEgkSYYuPe1dcK6jiRYva1ZWE7NHWTRVF6UnlNwMucGVBVjyfVHuM9v3NOaU8PAtkTutuWzhvfB5+ePwA9PneIuU/dp0guyMBGu8qKo+5hG6BiGYRiGYZj8jSCOAXCEbdvvOwts237PMIwrAPwn7EZs234W6gwkh6PzPD7mAKKlXV0mt4MM7tuS2XVs2G40yMFJx1u9vUlYLoqmrLg4pWEQPtyyF+9syNY59a8sEgRLWyKZU8odjXKVxSNaMUgxDAPXHTdeWBaLeL8yuaQJqujs6xmGYRiGYQ5k8o00mQBUtUsdndgmwyjRiQsaEaG9lRJJ25PS155M4ap7l+LKe98Slju9mQAIluNRy/AYN/StKAIA/Pbc6ZgxtBpfXjg6J9G0d1/2mP5z5Xz8/IxpoV9LMRX1Rn4mE2EZWlMSvBLDMAzDMMwnkHxHWk8D+LlhGAOcBYZhDATw08xzDKOkI5lSGjv4RV10NU00Pa+dpL51JG3BJAIAGlsTuP/NDZ5tUKFkkmhLxDKElLea0pibvnf0xH74x2VzMKy2NNDKm7KXRJoGVBXjpGkD3b/V3nlqVFGhQgSKbvhUuufUmTOHdH5jDMMwDMMwBxD5iqYvIl2/9LFhGKsMw1gF4COkeyJ9sVAHxxxYpFI2jrjlORz2w2eQIMLpldU7MOnbj+O7D7+nfF1zW3B6HhViiVQKG3fvE9bdsrfVfXzkhL7uY2oEQYmYpmAn3q+ySLmeX8NamVwElh8q0VQIt7sFY+vxwjWH59Swl2EYhmEY5pNAXqM427bXAWgAcDyAn2X+HQfgJADfKtjRMfs1tm3jb0vW4YPN6bqgpvYE1uxowcY9rdjamI0S3fzYBwCAO/77kXI7+khTWjQt39KIR97a6C5PJG1s2iOKpm2Z/dWWxTCirtRdXqxxiotahmDy0F8jmgzDUAqnJ686zH0cs0xMGVSJb59YGDFiEYF0+3kz8PcvzC7IdgFgcJ8SoacVwzAMwzAMk78RBOx0N84nMv8AAIZhTAVwEYBLO39ozP7OY+9sxjX/eBsA8PHNxyNB6o6StK9SwHaa29WiyUnpO+qnzwvLO5IpNEvmEY5oqiyOoiyWvex1oiki9WDSRZqAdASpXUo5pOsfO7kffn6Gf5+kXKDHdQSJmjEMwzAMwzBdA08pM13Gsg17hL9p3VG70FfJXzbp0vN0dVAdyZSwLwBuZKu6JIbSOBFNUnpeWea5RePqBXOFujK9aFKZQdB6KL93d/qMQQCALy8c7bOWiMlOdwzDMAzDMN1K3pEmhglC1kJtiaz4ae1IKtf751sbMWlABUbUlbnLdOl5OtGkElmO015VSUwQOXJvpmevXoB1O1tw0JBqRBdn0wVLNLVPgKbZLBVNPqLwppOn4NL5IzGSpAwGYRWgfolhGIZhGIYJD0eamC7DkGIsNPrz2Dubccfij2DbtiAqvnz3m1j4k+eE1+nFUQK27fWd83PiqyqJCpEtOT2vtiyOg4ZUAxCFD+3hJKMyeAgbabJMA6Pqy3IycuCeSgzDMAzDMN1LTpEmwzDuD1ilqhPHwhwg7Gpux42PvId1u1qE5W1ENP3i6ZUAgPqKuFJU3PLEcoysK8VJ0wbqI02tCWGb7nIf0VRdEgX1OYhaegFCjSBoU1sZVaRJEEEF1jgsmhiGYRiGYbqXXNPz9oR4/s95HgtzgHDTo+8reyKpBM5DSzd6lgHArU+tAAB8auoAwWmP0tSeQEu7NxWPiqYxfcuwfEuT+3dLexIWqVXyi/BQcRL3iTTl0uC2ELBoYhiGYRiG6V5yEk22bV/QVQfCHDis3dmiXE5rmhzeWLMLY/qWa7e1ZW8b1mciVn+84GBc8MfX3OdsG9jZ7BVUjmjqUxrDpfNH4n/+/pb7XEnMwtCaklDvg0ahfCNNARbdcppiZ2HRxDAMwzAM072wEQRTcHSDetnRDgCStu0xjKCs3t6E9bvSPZdG1HrNElRRKGc/RRHTk3536fyRqCuP48aTJmJ4bZnntRTqnudX0xRUj1Ro34YIiyaGYRiGYZhuhUUTU3BUFuLJlK1Mz0umbF/L8SUf70JbIgXTAPpXFnue36ZJ3QPSzng0CvSdEyegrjwOADh39jC/twBASs/ziTTRnlNHjO+LUxoGCs8XWuKcNn0wfv/CaiwcV1/gLTMMwzAMwzAqWDQxBUcVaVL1TgLSdU5+kZjFK7YDAPpVFClrh/xEUyxiCq8pK4r6HbYHIT3PJ9JEHfxu/9wMz/OFjjRVlkTx0tcWcb8mhmEYhmGYboItx5mCo4octSVSykiTSkhRXv14JwBgcB91HdL2pnbta4uiluCAVxbPbY4gQl5b5BNpSnldzwUKXdMEcINbhmEYhmGY7oRFE1NwVKJJF2kC1K56MidNG6hc7p+eZ3ZONIV0z0sGqCbuRcswDMMwDLN/w6KJKTiqIEhHMqV0zwPS/ZaCOLlBI5qa9KIpHrEQi2QPpqwoV9FEjSD8Ik0smhiGYRiGYQ5kWDQxBUdZ05SwtZGmxrYO7baKoib+etEsrWhRWY7T13Ym0kSdxOM+vZgCNBMKbwXBMAzDMAzDdCdsBMEUHFW9TbtPel6jJtJ0zMR++NkZ0wTBdO+lh+DFVTuwclsT/vX2Juxq9hNcnatpSpLD5UgTwzAMwzDMJxeONDEFR6UR0ul5uYmmeNT0iJVZI2pw5ZFjUBpLL9+zz0c0RSyh3ijX9LwkEUN+kaZkgGgaojGxYBiGYRiGYfYPONLEFJyOpFccdSRTaFcsB/RGCrTHkozTN6mpTV8PFY+awrGU+ESL1MeVfW3U51h0mukvF83EU+9vxQWHDstpvwzDMAzDMEzvgkUTU3BUaXgdyRTaOtRGEDpUfZkcinzc7LLrWJg0sBJj+pZhUHVJzjbdiSAv8Qy69Lx5o+swb3RdTvtkGIZhGIZheh+cnsd0mhdXbccO4mKniii1J2xtpEmHn2iK+/RNciiKpI0gHvvKfPxB0XQ2iGQynGg695ChAIB5o2tz3gfDMAzDMAzT++FIE9Mp/r1sEy6/6w0MqCzCA1ccikfe3oT/rtzhWS8dacpNNPkJozCRpngmHS/fRrBhI00XHjoc0wZXYeKAyrz2wzAMwzAMw/RuWDQxeWPbNm57ZiUAYOOeVjz45gbc9OgHynXbEym0ZSJNX1o4CpXFUTy4dAPe2bBXu33/9LwQkaYca5hkglzxHEzTwIxhfTq1L4ZhGIZhGKb3wul5TN4sXbcb727Mih4/JzsaaRpQVYyL541AdUnMd/t+jnV+zzlU5OiWJxM20sQwDMMwDMMc2LBoYvLmvjfWC3/7Odnt60jivY17AGRd8YIiQb7ueeS1VEDVlGaFWEVx1Hf7QfSrKOrU6xmGYRiGYZgDAxZNjEAyZePuV9di5dbGwHWXb2kCAIzpWwYA2LBrn3bd3zy3Chv3tAIAJg6sAOCNFj1/9eH4zTkN7t/+RhDZ56j4qiuPu48rijonmi6ZNwKnzxiUl4kEwzAMwzAMc+DANU2MwENLN+C6+5cBAD6++XjfdZszkaV+lcVYvqUJ63a1aNd1BNa3T5yAcf3SokmONNVXxNHcXur+7ZeCR19bFDWxJ6PXasviANKCr6K4c5d3cczCD0+d2qltMAzDMAzDMPs/HGliBN7fpDdmAICXVu3AUT99Dq+s3uGKprqydHTHEUY6KooiOCdjzw14RVHUMlEWzwqdsJEm6rJHX1/ZyfQ8hmEYhmEYhgFYNDESVGjYCve4s29/Gcu3NOHM37/s1jDRlDg/+lcWI2qp0+qilgHLNATRY/lYhdPXHj2xL2KWiWmDqxAlYqqzNU0MwzAMwzAMA7BoYiSoaGpUGDs4hnIpGzmLptpy0S2PRotKM2KplIim9oS+rxN9bb/KYrz17aNw32VzBJvwshhnnzIMwzAMwzCdh0UTI0AbwW5vbPNdtzVjIR5aNJWJ65XEstGi0ozAoSl5HUm95TeNNJXFLRTHLFimIUTH8m1qyzAMwzAMwzAUFk2MQAeJ7qzZ2YK/LVmHnc3tvq+pI2LIzya8plQUTeXE3Y6m5bnHkgwXaaLRqST3VmIYhmEYhmEKDOcvMQLtRKh89Z6l2LOvA4eM6IN7Lp2tXD8WMYWUvuG1pfhwi9quXE7PKyfNZ0vj3p5Nw2pLPcscaKSJiibWTAzDMAzDMEyh4UgT47KvPSmkxO3Z1wEAeHn1Tu1ryuIRFJM0u2G1Jdp15fQ8Gmmiwuf+y+fge5+ehPmja7XbopGmEiKgZgyt1r6GYRiGYRiGYfKBI00MAOCV1Ttw9u2vIKEJ1by9fjemDKryLC+NW0Jt0qQBlZg8sBJrd7bgb0vW++6TRppoel7DkGo0DPEXPzTSFCEpgRccOhwRy8RhY/SCi2EYhmEYhmFygSNNDADgTy99rBVMAPCpX/4XrR1Jz/KyeFSINBXHLHxx4Wh8aeFod9mwmhKUxCwcNqZOeK2YnpebfqeRpqiVNXyIRUxcNHc4RtWX57Q9hmEYhmEYhtHBkSYGADCyrixwnZVbvc1ry+KWkB4XyTjWUUH063OmY0ifEo8wqggwgvCDRpeG1uhrnxiGYRiGYRims7BoYgD4u945vLdpr2dZWTwiCBinuWyZ1G9JFUmiwopGjsLy6tcXoS2REowoGIZhGIZhGKbQsGhiAPjbezu8rxBNJZIYcsRXxDJRXRLFrpYOjKpXR7HKBNe73G3v6iuKcn4NwzAMwzAMw+QKiyYGANDu00jW4c21uz3LijOpeYvG1WPJml04akI/97mXrluE9qQ6ygSIKXZsFc4wDMMwDMP0Vlg0fcKxbRurtjWjLeE1eZBZuk4vmm7/3Ay0J1OIR7L1TUVRS3C58yOfSBPDMAzD/P/27jzKrrLM9/j3SaUqlRkCmSAMIYQwBWjmMYiBBXixG7tbwbaXxttLGxrEtq8N4m0HwLFbkcuF67RspJXutrtRulUQOyKIigyiqCiCMg8JQ0aS1PzeP/auysmpqpOqpOrsM3w/a+1V57x71znPSb0rtX/1vvvdklQNhqYm95mVj3Ht9x4bdv9esyYzf8Zk7nty6Hs1tbdmo0URsU1gkiRJkhqFS443uUqBCeCkRbvz7tMXb9M2f+bWa4lGOpI0nL84eSEzJ7fyjlP226nXkSRJksaLI02qaNepbcyePmmbtkP2mMEL6zuAnQ9NHzjnYN7/uoNomRDbP1iSJEkqgCNNqmhqWwtzSkLTlLYWdpnSNvB8R5YKL2dgkiRJUi0zNDWhTZ09bO7qGdGxk9smbnMfpLaJEwauY4KdH2mSJEmSap2hqcl09vRy2BXf5cRP3EHfCNb53n/ONCK2jgS1tkygvWyFPEmSJKmRFRqaImJZRHwzIp6PiBQR55btj4i4MiJeiIgtEbEyIhYP93ravqde2UxvX2Ld5m46Kiwz/o5TFvK+sw9k2eLdt2lvnRDbBKXSUSdJkiSpERV9xjsVeAi4aJj9lwKXABcAxwGbgNsjon2Y47UdGzu6Bx539fQNe9wZB8/jglMXbTPKBHD0vrO2nZ7nMuOSJElqcIWGppTSbSmlv0spfaN8X2Rn638NfCSl9J8ppV8AbwX2AM4tP14js2bT1tDUWSE0tbZsG5a+efHJ/Pnxe/Oh1x+8zUjT5DZDkyRJkhpbLS85vhCYB6zsb0gprY+Ie4ETgH8tqrB69vKrnQOPO7uHD01tZaviLV0wk6ULlgIwyel5kiRJaiK1HJrm5V9Xl7WvLtk3SERMAkpvLDR9jOuqay9v3BqaVm/sGPa4tpbhw1B7SaCa5PQ8SZIkNbhGHCa4HFhfsj1bbDm1pXSk6Y2fu2fY41orhaZWV8+TJElS86jl0LQq/zq3rH1uyb6hfByYWbItGPvS6tfLr3aN6LhKN5x19TxJkiQ1k1o+432CLBwt72+IiBlkq+gNO0SSUupMKW3o34CN415pHSkdaaqkLw1/D6dJE725rSRJkppHodc0RcQ0YP+SpoURcQSwJqX0dERcA/xdRDxGFqKuAp4Hbql+tY2ho3v4ezOVmjtj+FXdS1chNzRJkiSp0RW9EMTRwPdLnl+df70RWAH8Pdm9nL4A7AL8EDgrpTT8CgaqqNIy4wC3XnIKe82aPOIw1D6xlgcrJUmSpJ1XaGhKKd0JDHvxTEopAR/MN42Brt7KoWnmlFamt7eO+PUmVlgwQpIkSWoEnvE2ma7tjDSV39R2KAfMdRV3SZIkNY+ip+epyrY3Pa/S/Zn6zZ3Rznffs4zp7XYfSZIkNT7PepvM9keaRjb46GiTJEmSmoXT85rMWIUmSZIkqVl4htxktrcQxEiuaZIkSZKaiaGpifT2JXr7hr9pLUCEoUmSJEkqZWhqItubmjd3xqQqVSJJkiTVDxeCaCKVQtPbT9qXd712cRWrkSRJkuqDoamJdPb2Dmo7bclsXrNkDm87cd/qFyRJkiTVAUNTk3j51U4eemb9oPYb3n5sAdVIkiRJ9cPQ1AT6+hLLP30X67d0F12KJEmSVHdcCKLBPb9uC0d/dKWBSZIkSdpBhqYG98W7H2fNpq6iy5AkSZLqltPzmtC8Ge38yVF7ctYh84suRZIkSap5hqYG19YyeDBx5uRW/vbMAwuoRpIkSao/Ts9rcC0TYlBb20R/7JIkSdJIefbc4LZ0D743k6FJkiRJGjnPnhvcxo6eQW3dvX0FVCJJkiTVJ0NTg9swxFLjL23sLKASSZIkqT4ZmhrcUCNNLxqaJEmSpBEzNDW4jZ2DR5p6+1IBlUiSJEn1ydDU4PpHmvbcZXLBlUiSJEn1ydDU4PpD0z+uOIbPvuVIAD78+oOLLEmSJEmqK97ctoGllAYWgpjePpGzl87nV1ecybRJ/tglSZKkkXKkqYF19yZ68uuXpuZBycAkSZIkjY6hqYFt6dp6Y9vJrS0FViJJkiTVL0NTA9vSnYWmiROCton+qCVJkqQd4Zl0A+sPTY4ySZIkSTvO0NTANndlK+e1txmaJEmSpB1laGpgHflI0xRDkyRJkrTDDE0NbEtXH+D0PEmSJGlnGJoaVEqJnz29FoB2Q5MkSZK0wwxNDeqWnz/Hp//7UcCRJkmSJGlnGJoa1Ofvenzgsdc0SZIkSTvO0NSgImLgsavnSZIkSTvO0NSgJmzNTE7PkyRJknaCoalBTSgZaTI0SZIkSTvO0NSgSjIT3b19xRUiSZIk1TlDU4Pa1Nkz8HhzV2+BlUiSJEn1zdDUoNZt7h54vKXb0CRJkiTtKENTA0opsW5LSWhypEmSJEnaYROLLkBj55VXO7nwpgc5eP4MevvSQPvpB80psCpJkiSpvhma6tjGjm6+dv8z/NsDzzB/5mTmz2znvifWcN8TawaO+dyfH8npB80tsEpJkiSpvhma6tgXf/A4197xOwAeXf3qoP0Ld5/KWYfOr3ZZkiRJUkPxmqY69rNn1lXcP3fGpCpVIkmSJDUuQ1MdWzJ3esX982a0V6kSSZIkqXEZmupYT8liD/2mT9o643LuTEOTJEmStLMMTXWsY4j7Lx23324DjydNbKlmOZIkSVJDMjTVsf7Q1NoSA22L5kwdeNw6IQZ9jyRJkqTRMTTVsY7uPgDmTN86De+geTO44NRF7LnLZN583N5FlSZJkiQ1jJoOTRHx4YhIZdsjRddVKzp6spGm4/abBcAeM9s557D5vO/sA/nhZaex+zRXz5MkSZJ2Vj3cp+lh4PSS5z1FFVIr7vzti1z5rV/zzJrNAJy2ZA7nH7M3B+8xg4ktWQ6OcGqeJEmSNBbqITT1pJRWFV1ELVlxw/3bPG9vbeHYhbMKqkaSJElqbDU9PS+3OCKej4jHI+KmiKh4oU5ETIqIGf0bUPlmRg2gvbUefoySJElSfar1s+17gRXAWcCFwELg7oioFIQuB9aXbM+Oc42Fa291aXFJkiRpvNR0aEop3ZZS+veU0i9SSrcDrwN2Ad5U4ds+Dsws2RaMf6XFmmxokiRJksZNPVzTNCCltC4iHgX2r3BMJ9DZ/7wZFkRwep4kSZI0furqbDsipgGLgBeKrqUoKaVBbZMmOtIkSZIkjZeaDk0R8amIODUi9o2IE4FvAL3AvxRcWmE2dAxecd1rmiRJkqTxU+vT8xaQBaTdgJeAHwLHp5ReKrSqAq3b3DWozel5kiRJ0vip6dCUUjq/6BpqzdrN3YPaHGmSJEmSxo9DFHVm7RAjTa0t/hglSZKk8eLZdp1Zu2lwaJIkSZI0fgxNdaZ8et4uU1oLqkSSJElqDoamOlO6EMSyA2bzjb86qcBqJEmSpMZX0wtBaLA1+fS8S5Yv5m/OOKDgaiRJkqTG50hTnVmXT8/b1Wl5kiRJUlUYmupM/+p5u05pK7gSSZIkqTkYmupM/0IQu041NEmSJEnVYGiqM/1Ljjs9T5IkSaoOQ1MdSSk5PU+SJEmqMkNTHVmzqYvOnj4iYPb0SUWXI0mSJDUFQ1MdeXrNZgDmzWinvbWl4GokSZKk5mBoqhMpJW5+8FkA9po1peBqJEmSpOZhaKoTK3/zIl/9ydMA7GNokiRJkqrG0FQnvv/bFwceT2lzap4kSZJULYamGnb7w6v43F2/J6VER3fvQPsf/cGeBVYlSZIkNZeJRRegofX1Jf7yKz8F4Pj9duPZtVsA+NgblnLk3rsWWZokSZLUVBxpqlHPr98y8Hj1hg6ey0PTknnTiypJkiRJakqGphr12IuvDjxetb6DVRs6AFiw6+SiSpIkSZKaktPzatTvVm8NTR/6r4cBmNzawuxp3tRWkiRJqiZHmmrUo6s3Dmo769B5TJgQBVQjSZIkNS9DU416+PkNg9pWnLhv9QuRJEmSmpzT82pQR3fvoJGmuy89jb28qa0kSZJUdYamGvTo6o309CV2ndLKO5ct4pA9ZhiYJEmSpIIYmmrQg0+tBeDQPWdy4WsWFVyNJEmS1Ny8pqkG3fP4K0B2U1tJkiRJxTI01Ziunj7ufWINACcsMjRJkiRJRTM01Ziv/uQp1m3uZs70SSzdc2bR5UiSJElNz9BUY/7tgWcAeNfyxbS2+OORJEmSiuZZeQ1Zu6mLR1ZlS42fdci8gquRJEmSBIammvKTfAGIRbOnMnv6pIKrkSRJkgSGpprR0d3Lx297BIBTD5hTcDWSJEmS+hmaasCLGzt4+w338/Sazew2tY13L19cdEmSJEmScoamGvCJ2x4ZuDfTsgNmM3NKa8EVSZIkSepnaCpYb1/iO79aNfD8yH12LbAaSZIkSeUMTQVKKfEnn/0xm7t6B9rOPGRugRVJkiRJKmdoKtBLGzv5+TPrBp7f9/7lzJneXmBFkiRJksoZmgrS2dPLQ8+u36bNZcYlSZKk2jOx6AKa0eauHpZ/+i5eWN8x0Lbf7KlERIFVSZIkSRqKI00FuO+JNdsEphP2242v/MVxBVYkSZIkaTiGpir71XPrWXHD/QPP99ltCledewh77jK5wKokSZIkzEZuDwAADNJJREFUDcfpeVX2L/c9PfD46jcdzh8fuaDAaiRJkiRtjyNNVbR6Qwe35fdkmtE+kdctnV9wRZIkSZK2x9BUJZ09vbz9hvtZs6kLgFvffQrtrS0FVyVJkiRpewxNVbJhSw8dPdlNbGdPn+Q1TJIkSVKd8JqmKpk9fRK3XnIKX/rhExy2YKbLi0uSJEl1wtBURe2tLVx02v5FlyFJkiRpFJyeJ0mSJEkVGJokSZIkqYK6CE0RcVFEPBkRHRFxb0QcW3RNkiRJkppDzYemiDgPuBq4AjgSeAi4PSLmFFqYJEmSpKZQ86EJ+BvgiymlG1JKvwYuADYD/7PYsiRJkiQ1g5oOTRHRBhwFrOxvSyn15c9PKKouSZIkSc2j1pcc3x1oAVaXta8GDhzqGyJiEjCppGn6+JQmSZIkqRnU9EjTDrocWF+yPVtsOZIkSZLqWa2HppeBXmBuWftcYNUw3/NxYGbJtmDcqpMkSZLU8Go6NKWUuoCfAsv72yJiQv78nmG+pzOltKF/AzZWpVhJkiRJDanWr2mCbLnxGyPiAeA+4K+BqcANhVYlSZIkqSnUfGhKKX0tImYDVwLzgJ8DZ6WUyheHkCRJkqQxV/OhCSCldB1wXdF1SJIkSWo+NX1NkyRJkiQVrS5GmsbChg0bii5BkiRJUoF2NBNESmmMS6ktEbEn3qtJkiRJ0lYLUkrPjfTgZghNAexBbSw9Pp0swC2gNupR7bPPaLTsMxot+4xGyz6j0aq1PjMdeD6NIgg1/PS8/B9jxClyPGX5DYCN+T2kpIrsMxot+4xGyz6j0bLPaLRqsM+MugYXgpAkSZKkCgxNkiRJklSBoam6OoEr8q/SSNhnNFr2GY2WfUajZZ/RaNV9n2n4hSAkSZIkaWc40iRJkiRJFRiaJEmSJKkCQ5MkSZIkVWBokiRJkqQKDE1VFBEXRcSTEdEREfdGxLFF16Tqi4jLI+L+iNgYES9GxC0RsaTsmPaIuD4iXomIVyPi5oiYW3bM3hHx7YjYnL/OP0REw9+wWhAR74uIFBHXlLTZZ7SNiNgzIr6a94ktEfHLiDi6ZH9ExJUR8UK+f2VELC57jVkRcVNEbIiIdRHxpYiYVv1Po/EWES0RcVVEPJH3h99HxAei5K6k9pnmFhHLIuKbEfF8/jvo3LL9Y9I/IuKwiLg7P19+JiIurcbn2x5DU5VExHnA1WTLLR4JPATcHhFzCi1MRTgVuB44HjgDaAW+GxFTS475DPB64I358XsAX+/fGREtwLeBNuBE4G3ACuDK8S9fRYqIY4C/BH5Rtss+owERsSvwI6AbOBs4GPhfwNqSwy4FLgEuAI4DNpH9XmovOeYm4BCy/6vOAZYBXxjv+lWIy4ALgYuBg/LnlwLvKjnGPtPcppKdv140zP6d7h8RMQP4LvAUcBTwt8CHI+KdY/pJdkRKya0KG3AvcF3J8wnAc8D7iq7NrfC+MRtIwLL8+UygC/jTkmMOzI85Pn9+NtALzC055gJgPdBW9GdyG7e+Mg14FDgduBO4xj7jNkxf+QRwd4X9AbwAvLekbSbQAZyfPz8o70NHlxxzFtAH7FH0Z3Qb8z7zLeBLZW03A1+1z7gN0V8ScG7J8zHpH2TBfU3p76X8/7NHiv7MjjRVQUS0kaXllf1tKaW+/PkJRdWlmjEz/7om/3oU2ehTaX95BHiarf3lBOCXKaXVJa9zOzCD7C84akzXA99OKa0sa7fPqNwfAg9ExL/nUzF/FhHvKNm/EJjHtn1mPdkf+Er7zLqU0gMl37eS7ATnuHGtXkX4MbA8Ig4AiIjDgZOB2/L99hlVMlb94wTgBymlrpJjbgeW5CPohXEue3XsDrQAq8vaV5P9NVhNKiImANcAP0op/Spvngd0pZTWlR2+Ot/Xf8xQ/YmSY9RAIuJ8sqm9xwyx2z6jcvuR/cX2auBjZP3m2ojoSindyNaf+VB9orTPvFi6M6XUExFrsM80ok+Q/RHlkYjoJTtv+d8ppZvy/fYZVTJW/WMe8MQQr9G/by0FMTRJxboeOJTsr3nSkCJiL+D/AGeklDqKrkd1YQLwQErp/fnzn0XEoWRTMm8srizVsDcBbwH+DHgYOAK4JiKez4O21NScnlcdL5NfS1DWPhdYVf1yVAsi4jqyiyBPSyk9W7JrFdAWEbuUfUtpf1nF0P0J7FON6ChgDvBgRPRERA/ZYg+X5I9XY5/Rtl4Afl3W9htg7/xx/8+80u+lVWT9bkC+2uIs7DON6B+AT6SU/jWl9MuU0lfIFpi5PN9vn1ElY9U/avZ3laGpCvJ5mT8Flve35dOylgP3FFWXipEvyXkd8AbgtSml8mHon5KteFXaX5aQnez095d7gKVlqy+eAWxg8ImS6t/3gKVkf/nt3x4gW4Wo/7F9RqV+BCwpazuAbEUqyKa/rGLbPjOD7LqC0j6zS0QcVfIaryU7d7h3HGpWsaaQXVtSqpet54r2GVUyVv3jHmBZRLSWHHMG8NuUUmFT8wBXz6vWBpxHtoLI28hWD/k82bzMuUXX5lb1vvD/gHVkIwXzSrbJJcd8luzk5jSyUYYfAz8u2d8C/JLs4sjDgTPJ5gl/rOjP51a1fnQn+ep59hm3IfrHMWRB+v3A/mRTrjYBbyk55rL899AfkoXyW4DHgfaSY24DHgSOBU4iW73xn4v+fG7j0me+DDwL/A9gX7I/7L0EfNI+45b/bKex9Q93CXhP/njvseofZItjrQL+iWyRovPy/7veWfjnL7qAZtrI7n3wFNBJlqiPK7omt0L6QRpmW1FyTDvZ9U5r8v8svg7MK3udfYBbgc35L7ZPAROL/nxuVetH5aHJPuNW3kfOIQvKHWRT895Rtj/I7tO1Kj9mJXBA2TGzgH8GNpItT/+PwLSiP5vbuPSX6WQLEz0FbAF+D3yEbZd+ts808Qa8Zpjzly+PZf8ADgPuzl/jWeCyoj97SonIi5MkSZIkDcFrmiRJkiSpAkOTJEmSJFVgaJIkSZKkCgxNkiRJklSBoUmSJEmSKjA0SZIkSVIFhiZJkiRJqsDQJEmqmoi4MyKuKbqOUhGRIuLcouuQJNUub24rSaqaiJgFdKeUNkbEk8A1KaWqhKiI+DBwbkrpiLL2ecDalFJnNeqQJNWfiUUXIElqHimlNWP9mhHRllLq2tHvTymtGst6JEmNx+l5kqSq6Z+eFxF3AvsAn8mnx6WSY06OiLsjYktEPBMR10bE1JL9T0bEByLinyJiA/CFvP2TEfFoRGyOiMcj4qqIaM33rQA+BBze/35526DpeRGxNCLuyN//lYj4QkRMK9n/5Yi4JSLeGxEv5Mdc3/9e+TF/FRGPRURHRKyOiP8Yn39RSVI1GJokSUX4Y+BZ4IPA/HwjIhYB3wFuBg4DzgNOBq4r+/73Ag8BfwBclbdtBFYABwPvBt4BvCff9zXg08DDJe/3tfKi8nB2O7AWOAZ4I3D6EO9/GrAo//q2/H1X5K9xNHBt/tmWAGcBP9j+P4kkqVY5PU+SVHUppTUR0QtsLJsedzlwU8l1To9FxCXAXRFxYUqpI2+/I6X06bLX/EjJ0ycj4lPA+cDfp5S2RMSrQM92puP9GdAOvDWltAkgIi4GvhkRl6WUVufHrQUuTin1Ao9ExLeB5cAXgb2BTcC3UkobgaeAn438X0eSVGsMTZKkWnI4cFhEvKWkLchmRiwEfpO3PVD+jRFxHnAJ2QjQNLLfcRtG+f4HAQ/1B6bcj/L3XwL0h6aH88DU7wVgaf74v8mC0uMR8R2ykbNvpJQ2j7IWSVKNcHqeJKmWTAM+DxxRsh0OLAZ+X3JcaaghIk4AbgJuBc4hm7b3UaBtnOrsLnueyH+n5qNLRwJvJgtTVwIPRcQu41SLJGmcOdIkSSpKF9BS1vYgcHBK6XejfK0TgadSSh/tb4iIfUbwfuV+A6yIiKklo00nAX3Ab0daTEqpB1gJrIyIK4B1wGuBr4/0NSRJtcORJklSUZ4ElkXEnhGxe972SeDEiLguIo6IiMUR8UcRUb4QQ7nHgL0j4vyIWJRfB/WGId5vYf66u0fEpCFe5yagA7gxIg6NiNOA/wt8peR6pooi4pyIuCR/n32At5L9vh1x6JIk1RZDkySpKB8E9iWbdvcSQErpF8CpwAHA3WQLKFwJPF/phVJK/wV8hmyVu5+TjTxdVXbYzWTXF30/f783D/E6m4EzgVnA/cB/AN8DLh7F51pHtjrgHWQjVxcAb04pPTyK15Ak1ZBIKW3/KEmSJElqUo40SZIkSVIFhiZJkiRJqsDQJEmSJEkVGJokSZIkqQJDkyRJkiRVYGiSJEmSpAoMTZIkSZJUgaFJkiRJkiowNEmSJElSBYYmSZIkSarA0CRJkiRJFRiaJEmSJKmC/w8r8OUpImy9UwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,5), dpi=100)\n",
    "plt.title(\"Generator and Discriminator Loss During Training\")\n",
    "plt.plot(G_Glosses, label=\"G - gender loss\")\n",
    "plt.xlabel(\"iterations\")\n",
    "plt.ylabel(\"Loss\")\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,5), dpi=100)\n",
    "plt.title(\"Discriminator Loss During Training\")\n",
    "plt.plot(D_Flosses, label=\"D - fake's gender loss\")\n",
    "plt.xlabel(\"iterations\")\n",
    "plt.ylabel(\"Loss\")\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "any: -0.9937635660171509\n",
      "word: -0.9972375631332397\n",
      "can: -0.9982516169548035\n",
      "go: -0.9987564086914062\n",
      "here: -0.9939884543418884\n",
      "it: -0.9957798719406128\n",
      "will: -0.9971675276756287\n",
      "have: -0.996321439743042\n",
      "similar: -0.9970186948776245\n",
      "meaning: -0.9971277713775635\n",
      "okay: -0.9871394634246826\n",
      "anything: -0.9986513257026672\n",
      "at: -0.9956119656562805\n",
      "all: -0.9928355813026428\n"
     ]
    }
   ],
   "source": [
    "wds = \"any word can go here it will have similar meaning okay anything at all\".split()\n",
    "for w in wds:\n",
    "    orig = torch.Tensor(E.v(w)).to(device)\n",
    "    new = get_new_emb(w, netG).squeeze()\n",
    "    sim = new.dot(orig)\n",
    "    print(f\"{w}: {sim}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Analogy tests"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [],
   "source": [
    "def find_closest(new, k=100, E=E):\n",
    "    dots = new.dot(E.vecs.T)\n",
    "    dd = dict(zip([abs(dot) for dot in dots], [i for i in range(len(dots))]))\n",
    "    ns=[]\n",
    "    thresh = 1.5\n",
    "    for dot in sorted(dd, reverse=True):\n",
    "        if True:\n",
    "            ns.append(E.words[int(dd[dot])])\n",
    "    return ns[:k]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "def E_deb(word):\n",
    "    return get_new_emb(word, netG).squeeze().detach().cpu().numpy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "76abeaebc05e42828f704b3ac5f68881",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=47705), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "0.05725555202486781\n",
      "0.0045502590794699775\n"
     ]
    }
   ],
   "source": [
    "before = 0\n",
    "after = 0\n",
    "for word in tqdm(words):\n",
    "    before += np.abs(E.v(word).dot(g))\n",
    "    after += np.abs(E_deb(word).dot(g))\n",
    "print(before/len(words))    \n",
    "print(after/len(words))    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### As seen above, the average direct bias has reduced significantly, this is a good thing\n",
    "## Now, Some analogies to show gender bias"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['programmer',\n",
       " 'programmers',\n",
       " 'researcher',\n",
       " 'designer',\n",
       " 'educator',\n",
       " 'computer',\n",
       " 'freelance',\n",
       " 'animator',\n",
       " 'writer',\n",
       " 'developer',\n",
       " 'editor',\n",
       " 'businesswoman',\n",
       " 'compiler',\n",
       " 'librarian',\n",
       " 'interpreter',\n",
       " 'translator',\n",
       " 'blogger',\n",
       " 'analyst',\n",
       " 'organizer',\n",
       " 'software']"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Original embeddings\n",
    "A = E.v(\"man\")\n",
    "B = E.v(\"woman\")\n",
    "C = E.v(\"programmer\")\n",
    "\n",
    "V = B + C - A\n",
    "\n",
    "find_closest(V, 20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['programmer',\n",
       " 'programmers',\n",
       " 'researcher',\n",
       " 'designer',\n",
       " 'computer',\n",
       " 'educator',\n",
       " 'animator',\n",
       " 'developer',\n",
       " 'writer',\n",
       " 'editor',\n",
       " 'freelance',\n",
       " 'compiler',\n",
       " 'librarian',\n",
       " 'interpreter',\n",
       " 'organizer',\n",
       " 'translator',\n",
       " 'analyst',\n",
       " 'software',\n",
       " 'user',\n",
       " 'author']"
      ]
     },
     "execution_count": 82,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Out debiased embeddings\n",
    "A = E_deb(\"man\")\n",
    "B = E_deb(\"woman\")\n",
    "C = E_deb(\"programmer\")\n",
    "\n",
    "V = B + C - A\n",
    "\n",
    "find_closest(V, 20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-0.9969274"
      ]
     },
     "execution_count": 70,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "E_deb(\"programmer\").dot(E.v(\"programmer\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-0.05422006"
      ]
     },
     "execution_count": 72,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "E.v(\"programmer\").dot(g)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.003285324"
      ]
     },
     "execution_count": 71,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "E_deb(\"programmer\").dot(g)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Some analogies to show meaning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['pizza',\n",
       " 'japan',\n",
       " 'sushi',\n",
       " 'domino',\n",
       " 'japanese',\n",
       " 'pasta',\n",
       " 'kfc',\n",
       " 'sandwiches',\n",
       " 'restaurant',\n",
       " 'pizzas']"
      ]
     },
     "execution_count": 76,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Original embeddings\n",
    "A = E.v(\"usa\")\n",
    "B = E.v(\"pizza\")\n",
    "C = E.v(\"japan\")\n",
    "\n",
    "V = B + C - A\n",
    "\n",
    "find_closest(V, 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['pizza',\n",
       " 'sushi',\n",
       " 'japan',\n",
       " 'pasta',\n",
       " 'domino',\n",
       " 'japanese',\n",
       " 'sandwiches',\n",
       " 'kfc',\n",
       " 'restaurant',\n",
       " 'hokkaido']"
      ]
     },
     "execution_count": 77,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Out debiased embeddings\n",
    "A = E_deb(\"usa\")\n",
    "B = E_deb(\"pizza\")\n",
    "C = E_deb(\"japan\")\n",
    "\n",
    "V = B + C - A\n",
    "\n",
    "find_closest(V, 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['sky',\n",
       " 'green',\n",
       " 'blue',\n",
       " 'bright',\n",
       " 'red',\n",
       " 'dark',\n",
       " 'star',\n",
       " 'colour',\n",
       " 'yellow',\n",
       " 'light']"
      ]
     },
     "execution_count": 78,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Original embeddings\n",
    "A = E.v(\"grass\")\n",
    "B = E.v(\"green\")\n",
    "C = E.v(\"sky\")\n",
    "\n",
    "\n",
    "V = B + C - A\n",
    "\n",
    "find_closest(V, 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['sky',\n",
       " 'green',\n",
       " 'blue',\n",
       " 'bright',\n",
       " 'red',\n",
       " 'dark',\n",
       " 'star',\n",
       " 'colour',\n",
       " 'light',\n",
       " 'yellow']"
      ]
     },
     "execution_count": 79,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Out debiased embeddings\n",
    "A = E_deb(\"grass\")\n",
    "B = E_deb(\"green\")\n",
    "C = E_deb(\"sky\")\n",
    "\n",
    "V = B + C - A\n",
    "\n",
    "find_closest(V, 10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Debias all words\n",
    ":( Cuda memory error"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "# This gives a CUDA memory error, I need a better GPU :( :( :( :( :( :( :( :( \n",
    "# z = torch.randn(3000000, nz, device=device)\n",
    "# netG(z, E.vectors)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Save limited word embeddings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "dde3eff9dbbe4182b11b93815815b06e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=47690), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "debs = {}\n",
    "for word in tqdm(words):\n",
    "    debs[word] = E_deb(word)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle\n",
    "\n",
    "with open(\"GloVe_dev.dict.pickle\", 'wb') as handle:\n",
    "    pickle.dump(debs, handle)"
   ]
  }
 ],
 "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}