Exploring_Gender_Biases_in_Word2Vec.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "5XP3YOprVwyX"
   },
   "source": [
    "# Exploring Gender Biases in Word2Vec \n",
    "\n",
    "This notebook processes the data from the Gutenburg and Wikipedia datasets. Then, it trains the words agains Genism's word2Vec model, a custom skip-gram based word2vec model and our debiasing solutions. \n",
    "\n",
    "To measure biases that are hidden within these embeddings created by the models, we explore the direct bias, indirect bias, and the WEAT metric for comparision. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "h4LSSjBdfJhI",
    "outputId": "dfd47a58-6b57-4110-9160-f206ad21f54a"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Library/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/tqdm/auto.py:22: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
      "  from .autonotebook import tqdm as notebook_tqdm\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using cpu device\n"
     ]
    }
   ],
   "source": [
    "#library imports\n",
    "import string\n",
    "import numpy as np\n",
    "from numpy.linalg import norm\n",
    "import json\n",
    "import tempfile\n",
    "from sklearn.decomposition import PCA\n",
    "import torch\n",
    "from torchtext.data.utils import get_tokenizer\n",
    "\n",
    "#configure paths\n",
    "import sys\n",
    "import os\n",
    "\n",
    "module_path = os.path.abspath(os.path.join('..'))\n",
    "if module_path not in sys.path:\n",
    "    sys.path.append(module_path+\"\\\\nn_webserver\")\n",
    "\n",
    "#gutenberg data import\n",
    "from gutenberg.gutenberg_data import get_urls, read_data_from_urls\n",
    "\n",
    "#models import\n",
    "from gensim.models import Word2Vec\n",
    "from custom_word2vec import Custom_Word2Vec\n",
    "import pickle\n",
    "\n",
    "#plots\n",
    "import matplotlib.pyplot as plt "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "xhIMaI4vV8VI"
   },
   "source": [
    "## 1 - Data Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "id": "RzDntNYcmVec"
   },
   "outputs": [],
   "source": [
    "def create_tokens(data):\n",
    "    \"\"\"\n",
    "    creates tokenized list with puncuation and digits removed and lowercase words \n",
    "\n",
    "    :param data: list of strings\n",
    "    :return: a list of tokens \n",
    "    \"\"\" \n",
    "    tokens = []\n",
    "    for sentance in data:\n",
    "        sentance = sentance.translate(str.maketrans('', '', string.punctuation)) #remove punctuations\n",
    "        sentance = sentance.translate(str.maketrans('', '', string.digits)) #remove digits\n",
    "        tokenizer = get_tokenizer(\"basic_english\", language=\"en\") #remove unessasary characters, splits into spaces\n",
    "        tokens.append(tokenizer(sentance))\n",
    "  \n",
    "    return tokens\n",
    "  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "tTskJcpBpDEt"
   },
   "source": [
    "**Gutenburg Dataset**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "zsTQM81GpDEu",
    "outputId": "12fedd03-76dd-4c73-f874-c921b02c98fc"
   },
   "outputs": [],
   "source": [
    "# Gutenburg\n",
    "train_urls = get_urls('train')\n",
    "train_data_g = read_data_from_urls(train_urls)\n",
    "train_data_g=str(train_data_g).split('.')\n",
    "\n",
    "rough_guten = create_tokens(train_data_g)\n",
    "\n",
    "sentences_gutenburg = []\n",
    "for sentence in rough_guten:\n",
    "    new_sentence = []\n",
    "    for word in sentence:\n",
    "        if 'rn' in word:\n",
    "            temp = word.replace('rn', ' ')\n",
    "            while temp and temp[0] == ' ':\n",
    "                temp = temp.replace(' ', '')\n",
    "            while temp and '  ' in temp:\n",
    "                temp = temp.replace('  ', '')\n",
    "            if ' ' in temp:\n",
    "                temp = temp.split(' ')\n",
    "                for word_thing in temp:\n",
    "                    new_sentence.append(word_thing)\n",
    "            else:\n",
    "                new_sentence.append(temp)\n",
    "        else:\n",
    "            new_sentence.append(word)\n",
    "    sentences_gutenburg.append(new_sentence)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "0PcqA9wapDEv"
   },
   "source": [
    "**Wikipedia Dataset**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "QdY5fDuEpDEw",
    "outputId": "5c3c79b9-63a1-4f27-a540-fa70fc990d8d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "159\n"
     ]
    }
   ],
   "source": [
    "#Wikipedia\n",
    "f = open('wikipedia-en-1000.json')\n",
    "train_data_w = json.load(f)\n",
    "train_data_w =str(train_data_w[0]).split('.')\n",
    "\n",
    "sentences_wikipedia = create_tokens(train_data_w)\n",
    "print(len(sentences_wikipedia))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "kGQ-ic0KpDEw"
   },
   "source": [
    "## 2 - Genism Word2Vec"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "vADvrY8rDeuq"
   },
   "source": [
    "Using: Skip-gram training algorithm. \n",
    "\n",
    "Hyperparameters:\n",
    "- min_freq: choosing the top N most frequent words - for easier training, weed out words that are irrelevant to us \n",
    "- size: embedding size. We use 300 as per Bolukbasi et al.\n",
    "- window: the window used for looking at context and center words when traing the embeddings "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [],
   "source": [
    "#hyperparameters\n",
    "min_freq = 50\n",
    "size = 300\n",
    "window = 5\n",
    "epochs = 11"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Gutenburg Genism Model**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 258
    },
    "id": "cHtqmychpDEx",
    "outputId": "866140be-8705-4864-b7a7-7598b1f38101"
   },
   "outputs": [],
   "source": [
    "model_g = Word2Vec(min_count=min_freq, vector_size=size, window=window, sg=1) #sg=1 is the skip-gram training algorithm\n",
    "model_g.init_weights()\n",
    "model_g.build_vocab(sentences_gutenburg) \n",
    "model_g.train(sentences_gutenburg, total_examples=model_g.corpus_count, epochs=epochs)\n",
    "\n",
    "model_g.save(\"models/model_g\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Wikipedia Genism Model**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_w = Word2Vec(min_count=min_freq, vector_size=size, window=window, sg=1) #sg=1 is the skip-gram training algorithm\n",
    "model_w.init_weights()\n",
    "model_w.build_vocab(sentences_wikipedia) \n",
    "model_w.train(sentences_wikipedia, total_examples=model_g.corpus_count, epochs=epochs)\n",
    "\n",
    "model_w.save(\"models/model_w\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "fYDVA5_wpDEy"
   },
   "source": [
    "Now, save the embeddings to use in debiasing. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {
    "id": "e_XcKc4SpDEy"
   },
   "outputs": [],
   "source": [
    "with open('embeddings/gutenberg_embeddings.txt', 'w') as f:\n",
    "    for idx, key in enumerate(list(model_g.wv.index_to_key)):\n",
    "        embedding = ' '.join(str(v) for v in model_g.wv.get_vector(key))\n",
    "        f.write(key + \" \" + embedding+\"\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "J9hErZ1ApDEz"
   },
   "outputs": [],
   "source": [
    "with open('embeddings/wikipedia_embeddings.txt', 'w') as f:\n",
    "    for idx, key in enumerate(list(model_w.wv.index_to_key)):\n",
    "        embedding = ' '.join(str(v) for v in model_w.wv.get_vector(key))\n",
    "        f.write(key + \" \" + embedding+\"\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "9VzYjT6CqAJv"
   },
   "source": [
    "## 3 - Custom Word2Vec \n",
    "\n",
    "Using the same hyperparameters as Genism, train our own word2vec model (withought debiasing) on the data. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 366
    },
    "id": "wNHWm-S7qSiD",
    "outputId": "765dcb4b-60da-479a-db8a-b72f7493fff0"
   },
   "outputs": [],
   "source": [
    "# hyperparameters\n",
    "embedding_dim=300\n",
    "LR=0.01\n",
    "window_size=2\n",
    "EPOCHS=2\n",
    "min_freq = 2000"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Gutenburg Custom Model**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generating Skip Grams...\n",
      "...(1.5466min)\n",
      "Training. Curr Time = 16:16:00 , Estimated Finish Time = 21:10:06\n",
      "Epoch: 1, Training Loss: 8491664.738260888  (100.6639min)\n",
      "Epoch: 2, Training Loss: 8381964.123699829  (105.3134min)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "...(205.9882min)\n"
     ]
    }
   ],
   "source": [
    "model_g_custom = Custom_Word2Vec(sentences_gutenburg, embedding_dim, LR, window_size, EPOCHS, min_freq)\n",
    "model_g_custom.train()\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('models/model_g_custom', 'wb') as files:\n",
    "    pickle.dump(model_g_custom, files)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, save the embeddings to use in debiasing."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('embeddings/gutenburg_custom_embeddings.txt', 'w') as f:\n",
    "    for word in model_g_custom.corpus_vocab:\n",
    "        embedding_str = [str(x) for x in model_g_custom.embedding(word)]\n",
    "        embedding = ' '.join(embedding_str)\n",
    "        f.write(word + \" \" + embedding +\"\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.1 - Savani et. al Debiasing "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "#load\n",
    "with open('models/model_g_custom' , 'rb') as f:\n",
    "    model_g_custom = pickle.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Estimated time to complete:  5.133333333333334 h\n",
      "finding best threshhold ...\n",
      "1/4 sampled. Best objective so far: -0.30556901934939235 for threshold: 0.5918367346938775 ...(73.9692min)\n",
      "finding best threshhold ...\n",
      "2/4 sampled. Best objective so far: -0.30556901934939235 for threshold: 0.5918367346938775 ...(73.3584min)\n",
      "finding best threshhold ...\n",
      "3/4 sampled. Best objective so far: -0.30556901934939235 for threshold: 0.5918367346938775 ...(73.8759min)\n",
      "finding best threshhold ...\n",
      "4/4 sampled. Best objective so far: -0.30556901934939235 for threshold: 0.5918367346938775 ...(74.4021min)\n",
      "Updating Model with best objective function results.\n"
     ]
    }
   ],
   "source": [
    "#bias variables\n",
    "model_g_custom.equalized_pairs = [[\"monastery\", \"convent\"], [\"spokesman\", \"spokeswoman\"], [\"Catholic_priest\", \"nun\"], [\"Dad\", \"Mom\"], [\"Men\", \"Women\"], [\"councilman\", \"councilwoman\"], [\"grandpa\", \"grandma\"], [\"grandsons\", \"granddaughters\"], [\"prostate_cancer\", \"ovarian_cancer\"], [\"testosterone\", \"estrogen\"], [\"uncle\", \"aunt\"], [\"wives\", \"husbands\"], [\"Father\", \"Mother\"], [\"Grandpa\", \"Grandma\"], [\"He\", \"She\"], [\"boy\", \"girl\"], [\"boys\", \"girls\"], [\"brother\", \"sister\"], [\"brothers\", \"sisters\"], [\"businessman\", \"businesswoman\"], [\"chairman\", \"chairwoman\"], [\"colt\", \"filly\"], [\"congressman\", \"congresswoman\"], [\"dad\", \"mom\"], [\"dads\", \"moms\"], [\"dudes\", \"gals\"], [\"ex_girlfriend\", \"ex_boyfriend\"], [\"father\", \"mother\"], [\"fatherhood\", \"motherhood\"], [\"fathers\", \"mothers\"], [\"fella\", \"granny\"], [\"fraternity\", \"sorority\"], [\"gelding\", \"mare\"], [\"gentleman\", \"lady\"], [\"gentlemen\", \"ladies\"], [\"grandfather\", \"grandmother\"], [\"grandson\", \"granddaughter\"], [\"he\", \"she\"], [\"himself\", \"herself\"], [\"his\", \"her\"], [\"king\", \"queen\"], [\"kings\", \"queens\"], [\"male\", \"female\"], [\"males\", \"females\"], [\"man\", \"woman\"], [\"men\", \"women\"], [\"nephew\", \"niece\"], [\"prince\", \"princess\"], [\"schoolboy\", \"schoolgirl\"], [\"son\", \"daughter\"], [\"sons\", \"daughters\"], [\"twin_brother\", \"twin_sister\"]]\n",
    "model_g_custom.gender_pairs = [['she','he'],['her','his'],['woman','man'],['Mary','John'],['herself','himself'],['daughter','son'],['mother','father'],['gal','guy'],['girl','boy'],['female','male']]\n",
    "\n",
    "#hyperparameters\n",
    "model_g_custom.pred_threshold=0.55\n",
    "model_g_custom.epsilon=0.05\n",
    "num_trails = 4\n",
    "stddev = 0.5 #0.1\n",
    "margin = 0.01\n",
    "num_threshs = 50\n",
    "\n",
    "#run debiasing \n",
    "model_g_custom.random_debiasing(num_trails, stddev, margin, num_threshs)\n",
    "\n",
    "#save \n",
    "with open('models/model_g_custom_debiased', 'wb') as files:\n",
    "    pickle.dump(model_g_custom, files)\n",
    "\n",
    "with open('embeddings/gutenburg_custom_debiased_embeddings.txt', 'w') as f:\n",
    "    for word in model_g_custom.corpus_vocab:\n",
    "        embedding_str = [str(x) for x in model_g_custom.embedding(word)]\n",
    "        embedding = ' '.join(embedding_str)\n",
    "        f.write(word + \" \" + embedding +\"\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "L9jws4VUXiuh"
   },
   "source": [
    "## 4 - Measuring Bias"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.0 - Define Helper functions\n",
    "\n",
    "Define some useful helper functions to use in the measurement techniques"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_vocab(model, model_type, vocab):\n",
    "    '''\n",
    "    returns the vocabulary of the model\n",
    "    '''\n",
    "    if model_type == 'genism' or model_type == 'wiki':\n",
    "        return model.wv.key_to_index\n",
    "    elif model_type == 'custom':\n",
    "        return model.corpus_vocab.keys()\n",
    "    elif model_type == 'debias':\n",
    "        return vocab\n",
    "    else:\n",
    "        print(\"Model type is wrong\")\n",
    "        return []\n",
    "\n",
    "def check_w_embedding(word_lists, model, printRemoved, model_type, vocab={}):\n",
    "    '''\n",
    "    checks if the embedding exists for the given word, and if not deletes it from the lists. \n",
    "    if printRemoved = true, it also declares the words that were not in the model embeddings. \n",
    "    '''\n",
    "    new_w_lists = []\n",
    "    \n",
    "    for word_list in word_lists:\n",
    "        new_w_l = []\n",
    "        for w in word_list: \n",
    "            if w in get_vocab(model, model_type, vocab):\n",
    "                new_w_l.append(w)\n",
    "            else:\n",
    "                if printRemoved:\n",
    "                    print (\"Word\", w, \" is not in the model\")\n",
    "        \n",
    "        if (new_w_l):\n",
    "            new_w_lists.append(new_w_l)\n",
    "    \n",
    "    return new_w_lists\n",
    "\n",
    "def w_vec(word, model, model_type, vocab):\n",
    "    '''\n",
    "    returns the word embedding of the input word\n",
    "    '''\n",
    "    if model_type == 'genism' or model_type == 'wiki':\n",
    "        return model.wv[word]\n",
    "    elif model_type == 'custom':\n",
    "        return model.embedding(word)\n",
    "    elif model_type == 'debias':\n",
    "        return vocab[word]\n",
    "    else:\n",
    "        print(\"Model type is wrong\")\n",
    "        return []\n",
    "\n",
    "def cos_sim(a,b):\n",
    "    '''\n",
    "    returns the cosine similarity between 2 word embeddings \n",
    "    '''\n",
    "    return np.dot(a, b)/(norm(a)*norm(b))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As well, for the debiased embeddings, we need a way to retrieve the dat from the embedding text file. Create a function to do so."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def read_debiased_embeddings(file):\n",
    "    vocab = {}\n",
    "    with open(file, 'r') as f:\n",
    "        for line in f:\n",
    "            key, embedding = line.split(' ', 1)#split by first space\n",
    "            emb_vec= list(map(float, embedding.split(' ')[:-1])) #convert embedding to int, remove new line  \n",
    "            vocab[key] = np.array(emb_vec)\n",
    "    return vocab "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "EXVe1S3hpDEz"
   },
   "source": [
    "### 4.1 - Construct the g subspace\n",
    "First, define gendered words and pairs ground truth (taken from Bolukbasi et al.) Using PCA, define the gendered subspace g, by taking the most prominent dimension. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "id": "WGwiksMipDE0",
    "outputId": "ed4a1e49-169d-4e8a-d123-17f2774b239d"
   },
   "outputs": [],
   "source": [
    "def filter_gender_pairs(gender_pairs):\n",
    "    filtered_pairs=[]\n",
    "    for pair in gender_pairs:\n",
    "        if len(pair)==2:\n",
    "            filtered_pairs.append(pair)\n",
    "    return filtered_pairs\n",
    "\n",
    "def gender_subspace_matrix(model, gender_pairs, model_type, vocab):\n",
    "    '''\n",
    "    create matrix of the gender space using gender pairs\n",
    "        \n",
    "    '''\n",
    "    word_directions = []\n",
    "    for word_pair in gender_pairs:\n",
    "        if (len(word_pair) == 2):\n",
    "            f_word = w_vec(word_pair[0], model, model_type, vocab)\n",
    "            m_word = w_vec(word_pair[1], model, model_type, vocab)\n",
    "            dif = f_word-m_word\n",
    "            word_directions.append(dif)\n",
    "        \n",
    "        \n",
    "    return np.array(word_directions)\n",
    "\n",
    "\n",
    "def define_vector_g(model, model_type, gender_pairs, vocab={}):\n",
    "    word_directions = gender_subspace_matrix(model, gender_pairs, model_type, vocab) \n",
    "\n",
    "    #PCA of word direction\n",
    "    pca = PCA(n_components=len(gender_pairs))\n",
    "    pca.fit(word_directions)\n",
    "    \n",
    "    ratios = pca.explained_variance_ratio_\n",
    "    #print graph of PCA dimensions to determine how many dimensions should be kept\n",
    "    print(\"PCA variance ratio:\", ratios)\n",
    "    print(\"PCA singular values:\", pca.singular_values_)\n",
    "\n",
    "    #use the 1 most prominent dimension\n",
    "    pca = PCA(n_components=1)\n",
    "    pca.fit(word_directions)\n",
    "    g = pca.components_\n",
    "    \n",
    "    return g, ratios\n",
    "\n",
    "def run_gender_subspace_calcs(model, model_type, txtfiles, custom= False):\n",
    "    variance_ratios = []\n",
    "    i = 0\n",
    "    #genism model\n",
    "    if (model_type == 'genism' or model_type == 'custom'):\n",
    "        print (\"---------Original-------------\")\n",
    "        gender_pairs_new = filter_gender_pairs(check_w_embedding(gender_pairs, model, False, model_type))\n",
    "        print (\"Gender Pairs: \", gender_pairs_new)\n",
    "        g, pca_ratios = define_vector_g(model, model_type, gender_pairs_new)\n",
    "        variance_ratios.append(pca_ratios)\n",
    "    \n",
    "    #wiki model\n",
    "    if (model_type == 'wiki'):\n",
    "        print (\"---------Original-------------\")\n",
    "        vocab = read_debiased_embeddings(\"embeddings/wikipedia_embeddings.txt\")\n",
    "        gender_pairs_wiki = filter_gender_pairs(check_w_embedding(gender_pairs, model, False, 'debias', vocab))\n",
    "        print (\"Gender Pairs: \", gender_pairs_wiki)\n",
    "        g, pca_ratios_wiki = define_vector_g(model, 'debias', gender_pairs_wiki, vocab)\n",
    "        variance_ratios.append(pca_ratios_wiki)\n",
    "\n",
    "    #2016 debiasing \n",
    "    if (not custom):\n",
    "        print (\"---------2016debias-------------\")\n",
    "        vocab2016 = read_debiased_embeddings(txtfiles[i])\n",
    "        gender_pairs_2016 = filter_gender_pairs(check_w_embedding(gender_pairs, model, False, 'debias', vocab2016))\n",
    "        print (\"Gender Pairs: \", gender_pairs_2016)\n",
    "        g_2016, pca_ratios_2016 = define_vector_g(model, 'debias', gender_pairs_2016, vocab2016)\n",
    "        variance_ratios.append(pca_ratios_2016)\n",
    "        i+=1\n",
    "    \n",
    "    #2018 debiasing\n",
    "    print (\"---------2018debias-------------\")\n",
    "    vocab2018 = read_debiased_embeddings(txtfiles[i])\n",
    "    gender_pairs_2018 = filter_gender_pairs(check_w_embedding(gender_pairs, model, False, 'debias', vocab2018))\n",
    "    print (\"Gender Pairs: \", gender_pairs_2018)\n",
    "    g_2018, pca_ratios_2018 = define_vector_g(model, 'debias', gender_pairs_2018, vocab2018)\n",
    "    variance_ratios.append(pca_ratios_2018)\n",
    "    i+=1\n",
    "    \n",
    "    #2020 debiasing\n",
    "    if custom:\n",
    "        print (\"---------2020debias-------------\")\n",
    "        vocab2020 = read_debiased_embeddings(txtfiles[i])\n",
    "        gender_pairs_2020 = filter_gender_pairs(check_w_embedding(gender_pairs, model, False, 'debias', vocab2020))\n",
    "        print (\"Gender Pairs: \", gender_pairs_2020)\n",
    "        g_2020, pca_ratios_2020 = define_vector_g(model, 'debias', gender_pairs_2020, vocab2020)\n",
    "        variance_ratios.append(pca_ratios_2020)\n",
    "    \n",
    "    #combined graph:\n",
    "    colors = ['orange', 'red', 'blue']\n",
    "    if custom:\n",
    "        labels = ['Original', '2018 Debiasing', '2020 Debiasing']\n",
    "        plot_significance(variance_ratios, labels, colors, model_type, 0.05)\n",
    "        return g, g_2018, g_2020\n",
    "    else:\n",
    "        labels = ['Original', '2016 Debiasing', '2018 Debiasing']\n",
    "        plot_significance(variance_ratios, labels, colors, model_type, 0.3)\n",
    "\n",
    "        return g, g_2016, g_2018\n",
    "\n",
    "def plot_significance(variance_ratios, labels, colors, model, width):\n",
    "    for i in range(len(variance_ratios)):\n",
    "        ratio = variance_ratios[i]\n",
    "        r = np.arange(1, len(ratio)+1)\n",
    "        plt.bar(r + width*i, ratio, color=colors[i], edgecolor='white', width=width, alpha=0.5)\n",
    "        plt.xlabel(\"Matrix Variance Column\")\n",
    "        plt.ylabel(\"Variance\")\n",
    "    \n",
    "    \n",
    "    plt.title(\"Significance of each eigenvector for \"+model+\" models\") \n",
    "    plt.legend(labels)\n",
    "    plt.show()\n",
    "        \n",
    "\n",
    "#gender pairs\n",
    "gender_pairs = [['she','he'],['her','his'],['woman','man'],['Mary','John'],['herself','himself'],['daughter','son'],['mother','father'],['gal','guy'],['girl','boy'],['female','male']]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Gender Subspace for Gutenburg Models**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "---------Original-------------\n",
      "Gender Pairs:  [['she', 'he'], ['her', 'his'], ['woman', 'man'], ['herself', 'himself'], ['daughter', 'son'], ['mother', 'father'], ['girl', 'boy'], ['female', 'male']]\n",
      "PCA variance ratio: [2.2873145e-01 2.1048002e-01 1.6368133e-01 1.5716054e-01 1.3005719e-01\n",
      " 7.5725868e-02 3.4163605e-02 6.6469462e-15]\n",
      "PCA singular values: [4.0653491e+00 3.8997819e+00 3.4390182e+00 3.3698196e+00 3.0655053e+00\n",
      " 2.3391445e+00 1.5711467e+00 6.9302035e-07]\n",
      "---------2016debias-------------\n",
      "Gender Pairs:  [['she', 'he'], ['her', 'his'], ['woman', 'man'], ['herself', 'himself'], ['daughter', 'son'], ['mother', 'father'], ['girl', 'boy'], ['female', 'male']]\n",
      "PCA variance ratio: [1.00000000e+00 1.43942143e-13 1.15199127e-13 9.53962115e-14\n",
      " 8.43057607e-14 7.05038773e-14 4.84020684e-14 1.05201890e-30]\n",
      "PCA singular values: [2.59396303e-01 9.84141992e-08 8.80416912e-08 8.01178610e-08\n",
      " 7.53168950e-08 6.88763751e-08 5.70684060e-08 2.66057529e-16]\n",
      "---------2018debias-------------\n",
      "Gender Pairs:  [['she', 'he'], ['her', 'his'], ['woman', 'man'], ['herself', 'himself'], ['daughter', 'son'], ['mother', 'father'], ['girl', 'boy'], ['female', 'male']]\n",
      "PCA variance ratio: [2.29464781e-01 2.09661802e-01 1.63762456e-01 1.57819423e-01\n",
      " 1.29500481e-01 7.55221293e-02 3.42689280e-02 1.40523133e-32]\n",
      "PCA singular values: [4.06334124e+00 3.88405109e+00 3.43267312e+00 3.36981068e+00\n",
      " 3.05253721e+00 2.33110797e+00 1.57027435e+00 1.00553962e-15]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#load model\n",
    "model_g = Word2Vec.load(\"models/model_g\")\n",
    "\n",
    "#calculate g subspace\n",
    "g_gutenburg, g_gutenburg_2016, g_gutenburg_2018 = run_gender_subspace_calcs(model_g, \"genism\", [\"2016debias/2016debias_gutenberg_emb.txt\", \"2018debias/2018debias_gutenberg_emb.txt\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Gender Subspace for Wikipedia Models**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "---------Original-------------\n",
      "Gender Pairs:  [['she', 'he'], ['her', 'his'], ['woman', 'man'], ['herself', 'himself'], ['daughter', 'son'], ['mother', 'father'], ['gal', 'guy'], ['girl', 'boy'], ['female', 'male']]\n",
      "PCA variance ratio: [4.69855152e-01 2.05261717e-01 9.92370393e-02 7.70452044e-02\n",
      " 7.30826804e-02 3.46785969e-02 3.17338090e-02 9.10580083e-03\n",
      " 8.58932749e-33]\n",
      "PCA singular values: [4.42402561e+00 2.92408246e+00 2.03316458e+00 1.79146547e+00\n",
      " 1.74478882e+00 1.20189463e+00 1.14973231e+00 6.15878026e-01\n",
      " 5.98157038e-16]\n",
      "---------2016debias-------------\n",
      "Gender Pairs:  [['she', 'he'], ['her', 'his'], ['woman', 'man'], ['herself', 'himself'], ['daughter', 'son'], ['mother', 'father'], ['gal', 'guy'], ['girl', 'boy'], ['female', 'male']]\n",
      "PCA variance ratio: [9.38413057e-01 6.15869430e-02 6.30960885e-15 3.05790866e-15\n",
      " 2.70274064e-15 2.53520595e-15 2.09857069e-15 1.88229480e-15\n",
      " 4.09255393e-32]\n",
      "PCA singular values: [1.14774681e+00 2.94031186e-01 9.41131459e-08 6.55181048e-08\n",
      " 6.15958224e-08 5.96562150e-08 5.42763799e-08 5.14035240e-08\n",
      " 2.39687992e-16]\n",
      "---------2018debias-------------\n",
      "Gender Pairs:  [['she', 'he'], ['her', 'his'], ['woman', 'man'], ['herself', 'himself'], ['daughter', 'son'], ['mother', 'father'], ['gal', 'guy'], ['girl', 'boy'], ['female', 'male']]\n",
      "PCA variance ratio: [4.72228156e-01 2.05123155e-01 9.87546979e-02 7.67193925e-02\n",
      " 7.22885778e-02 3.45429825e-02 3.13888938e-02 8.95414419e-03\n",
      " 1.22786742e-32]\n",
      "PCA singular values: [4.44361099e+00 2.92864976e+00 2.03207146e+00 1.79107046e+00\n",
      " 1.73858102e+00 1.20182161e+00 1.14563982e+00 6.11888281e-01\n",
      " 7.16532398e-16]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#calculate g subspace\n",
    "g_wiki, g_wiki_2016, g_wiki_2018 = run_gender_subspace_calcs(model_g, \"wiki\", [\"2016debias/2016debias_wiki_emb.txt\", \"2018debias/2018debias_wiki_emb.txt\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Gender Subspace for Gutenburg Custom Word2vec Models**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('models/model_g_custom' , 'rb') as f:\n",
    "    model_g_custom = pickle.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "---------Original-------------\n",
      "Gender Pairs:  [['she', 'he'], ['her', 'his']]\n",
      "PCA variance ratio: [1.0000000e+00 2.7546131e-14]\n",
      "PCA singular values: [2.5387967e-01 4.2136477e-08]\n",
      "---------2018debias-------------\n",
      "Gender Pairs:  [['she', 'he'], ['her', 'his']]\n",
      "PCA variance ratio: [1.00000000e+00 5.32367293e-32]\n",
      "PCA singular values: [2.66063398e-01 6.13890364e-17]\n",
      "---------2020debias-------------\n",
      "Gender Pairs:  [['she', 'he'], ['her', 'his']]\n",
      "PCA variance ratio: [1.00000000e+00 5.01568289e-32]\n",
      "PCA singular values: [6.04471203e-01 1.35375680e-16]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "g_gutenburg_custom, g_gutenburg_2018_custom, g_gutenburg_2020_custom = run_gender_subspace_calcs(model_g_custom, \"custom\", [\"2018debias/2018debias_gutenberg_emb_custom.txt\", \"embeddings/gutenburg_custom_debiased_embeddings.txt\"], True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "7ylrwq0MXxYN"
   },
   "source": [
    "### 4.2 - Direct Bias\n",
    "\n",
    "To measure direct bias, we will be using the equation from Bolukbasi et al. shown below. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "cI6_1ww2pDE1"
   },
   "source": [
    "<img src=\"img/Direct_Bias.png\" style=\"width:300px;float:left;\"/>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "oBrC38VkpDE1"
   },
   "source": [
    "We will use the same gender neutral words that have been collected from the paper which is \n",
    "denoted by N. As well, we will utilize the paper’s gender directions (vector from one word to another) \n",
    "which were verified by crowdsourcing. The gender subspace g will be the unit vector g that captures the\n",
    "gender directions, computed by principal components in the paper. Finally, we will consider c, which is the \n",
    "strictness of bias, as a hyperparameter in our experiment. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "id": "qc-4WChgpDE1",
    "outputId": "cced117f-befb-4e48-d27c-542cd3e53a09"
   },
   "outputs": [],
   "source": [
    "def calculate_direct_metric(gendered_word_set, model, c, g, model_type, vocab={}):\n",
    "    '''\n",
    "    calculate the direct bias using Bolukbasi et al.'s equation\n",
    "    '''\n",
    "    cosine_add = 0\n",
    "    for word in gendered_word_set:\n",
    "        cosine_add += np.abs(cos_sim(w_vec(word, model, model_type, vocab),g[0]))**c\n",
    "    \n",
    "    return cosine_add/len(gendered_word_set)\n",
    "    \n",
    "\n",
    "c = 1 #hyperparameter\n",
    "\n",
    "gendered_words = 'engineer, professor, intelligent, teacher, police, nurture, captain, writer, poet, scientist, lawyer, receptionist, nurture, firefighter, doctor, nurse, manager, business, leader, follower, work, read, actress, home, sport, soccer, basketball, baseball, hockey, coach'\n",
    "gendered_words = gendered_words.split(\", \")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Direct Metric for Gutenburg Models**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Direct metric gutenburg: 0.05162829257460544\n",
      "Direct metric gutenburg 2016 debias: 3.562400578130557e-08\n",
      "Direct metric gutenburg 2018 debias: 0.05132398197252487\n"
     ]
    }
   ],
   "source": [
    "#load models\n",
    "model_g = Word2Vec.load(\"models/model_g\")\n",
    "\n",
    "#genism\n",
    "gendered_word_set = check_w_embedding([gendered_words], model_g, False, 'genism')[0]\n",
    "print(\"Direct metric gutenburg:\", calculate_direct_metric(gendered_word_set, model_g, c, g_gutenburg, 'genism'))\n",
    "\n",
    "#2016 debiasing \n",
    "vocab2016 = read_debiased_embeddings('2016debias/2016debias_gutenberg_emb.txt')\n",
    "gendered_word_set_2016 = check_w_embedding([gendered_words], model_g, False, 'debias', vocab2016)[0]\n",
    "print(\"Direct metric gutenburg 2016 debias:\", calculate_direct_metric(gendered_word_set_2016, model_g, c, g_gutenburg_2016, 'debias', vocab2016))\n",
    "\n",
    "#2018 debiasing\n",
    "vocab2018 = read_debiased_embeddings('2018debias/2018debias_gutenberg_emb.txt')\n",
    "gendered_word_set_2018 = check_w_embedding([gendered_words], model_g, False, 'debias', vocab2018)[0]\n",
    "print(\"Direct metric gutenburg 2018 debias:\", calculate_direct_metric(gendered_word_set_2018, model_g, c, g_gutenburg_2018, 'debias', vocab2018))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Direct Metric for Wikipedia Models**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Direct metric wiki: 0.06087632697703697\n",
      "Direct metric wiki 2016 debias: 0.07916512143350311\n",
      "Direct metric wiki 2018 debias: 0.060161984061677866\n"
     ]
    }
   ],
   "source": [
    "#wiki\n",
    "vocab = read_debiased_embeddings(\"embeddings/wikipedia_embeddings.txt\")\n",
    "gendered_word_set = check_w_embedding([gendered_words], model_g, False, 'debias', vocab)[0]\n",
    "print(\"Direct metric wiki:\", calculate_direct_metric(gendered_word_set, model_g, c, g_wiki, 'debias', vocab))\n",
    "\n",
    "#2016 debiasing \n",
    "vocab2016 = read_debiased_embeddings('2016debias/2016debias_wiki_emb.txt')\n",
    "gendered_word_set_2016 = check_w_embedding([gendered_words], model_g, False, 'debias', vocab2016)[0]\n",
    "print(\"Direct metric wiki 2016 debias:\", calculate_direct_metric(gendered_word_set_2016, model_g, c, g_wiki_2016, 'debias', vocab2016))\n",
    "\n",
    "#2018 debiasing\n",
    "vocab2018 = read_debiased_embeddings('2018debias/2018debias_wiki_emb.txt')\n",
    "gendered_word_set_2018 = check_w_embedding([gendered_words], model_g, False, 'debias', vocab2018)[0]\n",
    "print(\"Direct metric wiki 2018 debias:\", calculate_direct_metric(gendered_word_set_2018, model_g, c, g_wiki_2018, 'debias', vocab2018))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Direct Metric for Gutenburg Custom Word2vec Models**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "#load models\n",
    "with open('models/model_g_custom' , 'rb') as f:\n",
    "    model_g_custom = pickle.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Direct metric gutenburg: 0.19221440580995044\n",
      "Direct metric gutenburg 2018 debias: 0.32718988855668013\n",
      "Direct metric gutenburg 2020 debias: 0.3601605396326931\n"
     ]
    }
   ],
   "source": [
    "#create new word set that the model includes\n",
    "gendered_words = 'earth, get, place, heart, night, nothing, heard, thought, give, people, life, old, children, face, first, land, make, things, time, love, great'\n",
    "gendered_words = gendered_words.split(\", \")\n",
    "\n",
    "#original\n",
    "gendered_word_set = check_w_embedding([gendered_words], model_g_custom, False, 'custom')[0]\n",
    "print(\"Direct metric gutenburg:\", calculate_direct_metric(gendered_word_set, model_g_custom, c, g_gutenburg_custom, 'custom'))\n",
    "\n",
    "#2018 debiasing\n",
    "vocab2018 = read_debiased_embeddings('2018debias/2018debias_gutenberg_emb_custom.txt')\n",
    "gendered_word_set_2018 = check_w_embedding([gendered_words], model_g_custom, False, 'debias', vocab2018)[0]\n",
    "print(\"Direct metric gutenburg 2018 debias:\", calculate_direct_metric(gendered_word_set_2018, model_g_custom, c, g_gutenburg_2018_custom, 'debias', vocab2018))\n",
    "\n",
    "#2020 debiasing\n",
    "vocab2020 = read_debiased_embeddings('embeddings/gutenburg_custom_debiased_embeddings.txt')\n",
    "gendered_word_set_2020 = check_w_embedding([gendered_words], model_g_custom, False, 'debias', vocab2020)[0]\n",
    "print(\"Direct metric gutenburg 2020 debias:\", calculate_direct_metric(gendered_word_set_2020, model_g_custom, c, g_gutenburg_2020_custom, 'debias', vocab2020))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "gMIUMyqsX57J"
   },
   "source": [
    "### 4.3 - Indirect Bias\n",
    "\n",
    "To measure indirect bias, we will be using the equation from Bolukbasi et al. shown below."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "KrOSjUQspDE2"
   },
   "source": [
    "<img src=\"img/Indirect_Bias.png\" style=\"width:300px;float:left;\"/>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "6krHTKxepDE2"
   },
   "source": [
    "$w_{\\perp}$ is defined as $w_{\\perp} = w - w_{g}$ where $w_{g}=(w*g)*g$ as mentioned in the paper. We will be using the g subspace calculated above. As for the word pairs, we will run a few experiments. The first will be the most extreme words in the softball-football direction as mentioned in the paper. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "id": "kGyyr-NqpDE2",
    "outputId": "48e36e83-23b3-492e-9a74-eeac5e2bbe8c"
   },
   "outputs": [],
   "source": [
    "def calculate_indirect_metric(g, w, v):\n",
    "    wg = np.dot(np.dot(w,g[0]),g[0])\n",
    "    vg = np.dot(np.dot(v,g[0]),g[0])\n",
    "    w_norm_vec = w-wg\n",
    "    v_norm_vec = v-vg\n",
    "    w_norm = norm(w_norm_vec)\n",
    "    v_norm = norm(v_norm_vec)\n",
    "    \n",
    "    return (np.dot(w,v) - np.dot(w_norm_vec,v_norm_vec)/(w_norm*v_norm))/np.dot(w,v)\n",
    "\n",
    "def compare_indirect_metrics(words, c_word, model,g, model_type, vocab={}):\n",
    "    metrics=[]\n",
    "    for w in words:\n",
    "        metrics.append(calculate_indirect_metric(g, w_vec(c_word,model, model_type, vocab), w_vec(w,model, model_type, vocab)))\n",
    "    \n",
    "    return metrics\n",
    "\n",
    "def plot_indirect_metric(words, w, m, model, g_values, model_type, labels, txtfiles, colors):\n",
    "    fig, ax = plt.subplots(1,1) \n",
    "    \n",
    "    idx=0\n",
    "    for i in range(len(labels)):\n",
    "        b_vals_w = []\n",
    "        b_vals_m = []\n",
    "\n",
    "        #calculate indirect beta values\n",
    "        if labels[i] == \"Original\": #first iteration model\n",
    "            b_vals_w = compare_indirect_metrics(words, w, model, g_values[i], model_type)\n",
    "            b_vals_m = compare_indirect_metrics(words, m, model, g_values[i], model_type)\n",
    "        else:\n",
    "            vocab = read_debiased_embeddings(txtfiles[idx])\n",
    "            b_vals_w = compare_indirect_metrics(words, w, model, g_values[i], \"debias\", vocab)\n",
    "            b_vals_m = compare_indirect_metrics(words, m, model, g_values[i], \"debias\", vocab)\n",
    "            idx+=1\n",
    "            \n",
    "        print(\"b_vals_w:\", b_vals_w)\n",
    "        print(\"b_vals_m:\", b_vals_m)\n",
    "        #plot\n",
    "        ax.plot(range(1, len(words)+1), b_vals_w, marker='o', color = colors[i], alpha=0.8)\n",
    "        ax.plot(range(1, len(words)+1), b_vals_m, '--', marker='o', color = colors[i], alpha=0.8)\n",
    "        \n",
    "\n",
    "    ax.set_xticks(range(1, len(words_w)+1))\n",
    "    ax.set_xticklabels(words, rotation='horizontal', fontsize=10)\n",
    "    plt.ylabel('β values')\n",
    "    plt.xlabel('non-related words')\n",
    "    plt.title('Indirect metrics for '+ model_type+ \" models\")\n",
    "    \n",
    "    legend_labels =[]\n",
    "    for l in labels:\n",
    "        legend_labels.append(w + \" - \"+ l)\n",
    "        legend_labels.append(m + \" - \"+ l)\n",
    "        \n",
    "    ax.legend(legend_labels, bbox_to_anchor=(1.04,1))\n",
    "    plt.show()\n",
    "    \n",
    "#Experiment- recreate Bolukbasi et al.'s experiement but with words available in the datasetss\n",
    "words_w= ['beauty', 'fashion', 'marriage', 'charm', 'nurse'] \n",
    "w = 'clean' \n",
    "words_m = ['paper', 'money', 'office', 'police', 'strength'] \n",
    "m = 'build'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Indirect Metric for Gutenburg Models**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "b_vals_w: [0.95366323, 0.92422444, 0.9415778, 0.9205797, 1.3884088]\n",
      "b_vals_m: [0.9389032, 0.92986566, 0.939495, 0.9270366, 0.93395746]\n",
      "b_vals_w: [5.557600907791511e-08, 1.5634517056150938e-08, 9.261008223368369e-09, 3.483130576751526e-09, 5.495473793964756e-09]\n",
      "b_vals_m: [9.33836206983211e-08, 5.344228259375659e-08, 4.706881286109955e-08, 4.1290908753872074e-08, 4.330336095435692e-08]\n",
      "b_vals_w: [0.9458114033933209, 0.923822568426984, 0.9405149863167432, 0.9201988376649252, 0.9945951655339683]\n",
      "b_vals_m: [0.9378447950799984, 0.9295931689097401, 0.9392800467716164, 0.9265484751849381, 0.9335416832228952]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "b_vals_w: [0.9473365, 0.93449897, 0.93435, 0.9176478, 0.9331833]\n",
      "b_vals_m: [0.95077884, 0.9383862, 0.9414942, 0.92750376, 0.93321776]\n",
      "b_vals_w: [1.0984660422829728e-08, 9.286044492344265e-09, 4.541651333197938e-08, 2.2747489390523226e-08, 3.1110898214129605e-08]\n",
      "b_vals_m: [4.879242572408347e-08, 4.7093812382874514e-08, 8.322429296963847e-08, 6.055525881379242e-08, 6.891865381403916e-08]\n",
      "b_vals_w: [0.94717336295688, 0.9343627816570222, 0.9340751189778829, 0.9173921582455385, 0.9322388748149676]\n",
      "b_vals_m: [0.9507201941747864, 0.9380950704788638, 0.9413516064024047, 0.9274350985540647, 0.9329873093145988]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAisAAAEWCAYAAACnotfGAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/MnkTPAAAACXBIWXMAAAsTAAALEwEAmpwYAABehklEQVR4nO3deVxU5f4H8M93Fhj2RVAQREQYYADRQFJLzVLTyq5FlEuZ2l5aidntZmmSLeaWS7e0UtPMXK71S69mejXtuiVexURxR0FBUNkHZn1+f5wzNiIgKOhg3/frNS/OnPOc5zznmRnmO8/znPOQEAKMMcYYY45KcasLwBhjjDFWFw5WGGOMMebQOFhhjDHGmEPjYIUxxhhjDo2DFcYYY4w5NA5WGGOMMebQOFj5iyCi94joW3k5hIjKiUh5q8t1MxDRF0T0biPmR0S0kIiKiOj3xsq3KRDR20T01a0ux40gomwi6l2PdKFEJIhIdTPKxRi7efhD3YwQUTaAZ4UQm24kHyHEGQDujVIoO0QUCuAUALUQwtzY+ddwvOGQ6uPuutIJIV5s5EPfDaAPgGAhREUj592ohBAf3uoyMMbYjeKWFXYFudXgtnlfNFHrUVsA2dcTqPCvfsYYa7jb5kvpr4aIhhPRf4lomtwdcYqI+tttb0dEW4mojIg2AvCz23ZFczkR/UpEHxDRdgB6AGFEFEVEG4noEhEdIaLH7fZ3IaLpRHSaiErkcrgA2CYnKZa7mbrWUO73iGglEX0rl+0PItIS0T+IqICIcoior116LyL6mojyiOgsEU0mIiURRQP4AkBX+VjFcvpFRPQ5Ea0jogoAveR1k+3y/BsR7SeiUiI6QUT97Or0pFyuU0Q0tIbyPwPgK7vjTpLXP0dEx+X6+omIWtvtI4joFSI6BuBYLa/nMLk+LxLRu/ZdH0SkIKK35LJeJKIVRORb7bV8mojOENEFIhpfrb5t3X8aud4vElExEe0holZ274HJRLRDPq81RNSCiJbK9bRHbjmrqey2MoyQX78iInqRiDoT0QH5WHPt0iuI6B35fAuIaDERedltf8quLsZXO1atdVFDua75ejLGmgkhBD+ayQNANoDe8vJwACYAzwFQAngJwDkAJG/fCWAGAGcAPQCUAfhW3hYKQABQyc9/BXAGQAykrkEvADkARsjPOwG4AEAnp/9M3idIPnY3+ThX5FvLObwHoArA/XLeiyF1HY0HoJbP55Rd+h8AzAPgBqAlgN8BvGBXB/+tlv8iACUA7oIUjGvkdZPl7Uny9j7y9iAAUXL+pQAi5XSBAGJqOYcrjgvgXrl+7pDrYQ6AbXbbBYCNAHwBuNSQnw5AOaTuJScA0+TX1vZavwZgF4BgOf95AJZVey2/BOACIB6AAUC0XX3bXvcXAKwB4Cq/bgkAPO3eA8cBtJdf/0MAjgLobfc6LaylPmxl+EKu777ya/yj/JoFASgA0FNOP1I+Vhik7sjVAJZUq4se8rnOAGBuYF2oGvJ68oMf/HD8xy0vAD8a8GJdHawct9vmKv+jDgAQIv+Dd7Pb/h3qDlbS7NI+AeC3aseeB2AipC/4SgDxNZTvinxrOYf3AGy0ez5A/nJSys895Dy8AbSSv3hd7NIPBrDFrg5qClYW17Bust15zKyhXG4AigEko4aAolraK44L4GsAn9g9d4cUbITKzwWAe+vIb4LtC9futTTavdaHAdxntz1Qzl9lV+fBdtt/BzDIrr5tr/tIADsAdKihDL8CGG/3fDqA9dVep/21lN9WhiC7dRcBPGH3/F8AXpeX/wPgZbttkXbnMwHA99Vel4bWhS1YqdfryQ9+8MPxH9wN1Lzl2xaEEHp50R1AawBF4soxFaevkVeO3XJbAHfKzffFchfLUEiBkB+kX88nbqDc5+2WKwFcEEJY7J4D0nm0hdTakmdXjnmQfq3X91yqa4Mayi7X1RMAXpSP928iirrWichaw65+hRDlkL6sg+pZptb22+XX8qLd9rYAfrCrg8MALJCCOZt8u2U9ah5AvQTABgDfE9E5IvqEiNR226u/LtWfX2tQdn33v6K+5GUVpPOpXhcVaHhd3OjryRhzMBys3J7yAPgQkZvdupBr7GM//XYOgK1CCG+7h7sQ4iVI3R1VkLoL6sqjMeRAalnxsyuHpxAi5hrHq6scOai57BBCbBBC9IH0az0LUtdKfZyD9CUKAJDrvQWAs/UsUx6kbg3b/i7y/vZl7l/t9dAIIc5Wz6guQgiTEGKSEEIHqevuIQDDGpJHI7mivvBnS+B5SHXRxraBiFxxnXVxA68nY8zBcLByGxJCnAaQDmASETkR0d2QmvHray0ArTzQUS0/OhNRtBDCCmABgBlE1Jqkwa5dicgZQCEAK6SxCI1xHnkAfgEwnYg85cGV7Ymop5zkPIBgInJqQLZfAxhBRPfJ+QWRNJi4FUkDb90gBUjl8rnUxzI5z45yPXwIYLcQIrue+68CMICIusnn8h4Astv+BYAPiKgtABCRPxH9rZ55X0ZEvYgojqQrpEohdZ/U9xwb0zIAY0gaBO4Oqb6WC+ly91UAHiKiu+W6SMOV/6fqVRc3+HoyxhwMByu3ryEA7gRwCdJYk8X13VEIUQZpkOQgSL+C8wFMgTSgEQDeAPAHgD1y/lMAKOTuiw8AbJeb6bs0wnkMgzTo9BCAIkhfZoHyts0AMgHkE9GF+mQmhPgd0sDhmZAG2m6F9CtfASAV0vleAtAT0qDl+uS5CcC7kMZl5EFquRlUn33l/TMBjAbwvbx/OaQBqQY5ySwAPwH4hYjKIA0wvbO++dsJgFR/pZC6T7ZC6hq62RbIx90GaXB1FaTzt9XFK5DGWOVBes1z7fatb11c9+vJGHM8titHGGMOQm5tKAYQIYQ4dYuLwxhjtxy3rDDmAIhoABG5yt0W0yC1XGXf2lIxxphj4GCFMcfwN0hdFucAREC69JibPRljDNwNxBhjjDEHxy0rjDHGGHNozW5SNT8/PxEaGnqri8EYY83K3r17Lwgh/G91ORi7Hs0uWAkNDUV6evqtLgZjjDUrRHStu1gz5rC4G4gxxhhjDo2DFcYYY4w5NA5WGGOMMebQmt2YFcYYY7fO3r17W6pUqq8AxIJ/8LLGYwVw0Gw2P5uQkFBQfSMHK4wxxupNpVJ9FRAQEO3v71+kUCj4Rl2sUVitViosLNTl5+d/BeDh6tv/ElFxWtp6+PjcC5WqHXx87kVa2vpbXSTGGGuuYv39/Us5UGGNSaFQCH9//xJILXZXue1bVtLS1mPSpJGwWksBmFBcXIBJk0YCWIAJE/rf6uIxxlhzo+BAhTUF+X1VYyPKbd+y8tFHb8FqvQCpO0wBwASrtRBpaS8jLe1nzJu363LabduysX79Mfz3v2eQkVGAEyeKcemS4VYVnTUT3HLHGGNN67ZvWamqOgpACUAAMMp/AYslG5MmPQO1ugX+/e9noNE44+efl6Ky8iykoIZARPDwCMOHH6bB3V2DN9+cjLKyPKhULlCpXKBWuyA6uhM2bPg7lErggQemobS0DM7OztBoNHB2dkKnTlq8+25fAMB77/0Mq1XA1dUZbm4auLk5Q6v1x913hwAA9u0rgJubE9zdneHp6QxXVwUUtyCcTEtbj5kzp6Ks7BQ8PNphzJhx3ApVi7S09Xj//VFQKJzg5OQLvT4P778/CsBcrjPGbqLU1NTW7u7ulrS0tPO34vjp6emaUaNGheTn5zsJIfD4449fnDJlSp6ihn/i27Ztc12wYEGLRYsW5dSVZ6dOnaL27duXdT3lSU5ODn3ooYdKRowYUXQ9+zua2z5YuZLtTWMFQPDy0sJiMWDr1hWwWAwwmfQgUkGa3NEKq1WgrOwEUlNfBgAYDBcghAmAgBBWAALnz++Ap+e/oFRqUFFxUt4OSEGRwPr1wVi5ch2cnJyxf/9yebsCRARAAX//TkhOHgwXF2fMmvV3eT8CkQJEanTs+BCmTn0BSqUSAweOglqtgZOTC5ydNXBy0uDRR/vivff64eLFSjz55OfQaJwvP1xcnPHAAx3x0ENaXLpUhcWL/wc3N+fLD3d3Z+h0fggMdIXJZEVFBfDppxvw/vv16zazzYFJBFitgMkk/bVaAbNZqj83N4KTkwJ6vQWFhQZYLFaYzdbLf9u394Srqwr5+XqcPFkMi0XAarXCYhGwWKzo1i0Ibm5qHDlyCZmZBbBarTCbpTRWq8Ajj0TDzU2NXbvOYv/+s7BYrFfsP3p0Nzg5KbFu3VGkp5+GENJ6i8UKAPjggwcAAIsX/w//+98pWCwWWK1S/iqVEnPmPA4AmD79V+zff1I+ru3cXPDDD4tgtRphNudBCoZVAFzxwQdvAFBiwgQpUP3hh0MoLa2Cu7sGnp7So0ULV4SHe1+uS6JGepuz206vXmnYunUmhCgDkQd69hyDLVsm3Opi1cvKlfCcPh0BOTlwbtMGhrFjkZ+SgtJbXa7GVF5eTo888kj4rFmzzjz66KOlZWVligcffLD9lClT/P/xj38U2qc1mUzo0aOHvkePHvpr5Xu9gcrtqNnNupyYmCgacrt9jSYeBsNhSIGKAlKgYoWzczSqqjIupxNCwGg0obS0CuXlRpSWGlBWZkBZWRUqKgzVHlWorDRArzegstL2qILBYEBVlQEGg/QwGqtgMFTBZDLCZDKgsrIUJpMeZnMVrFYDLBYTiFRQKp0hhIDJVALAejlYAgQUCheo1R6wWq0wGs/ZbSMABGfnQHh6tgUAFBbugq3lyBYQtWzZDbGxSais1GPXrkXyfn8GS0lJg5Cc3AcnT+bhyy8nwWg8BaDqcv62VimVqi00mjZy+axyoGLB2LETMHhwEpYs2YGZMyfKx7fKZSCkpc3EgAHx+Oc/N+LLLz+66vWZNWseevWKxJQp/4elS2detX3x4m9xxx1tMH78Mvzf/31x1fY1a35AWFgLvP7619i4cfFV27dv3wAfHxc899wcbN++8qrtWVn/BREwZMgU/O9/a67YplS6IjPzFwDAo49OxKFD/5HOSg4qXF1b4uLFfRCiCLbX68+HCkFBA5GbKx0zMvJZ5ObuvyL/gAAdTpyQytyu3TBcvHgaKpUGarUz1GoNoqM7YePGfwAA+vX7BGVl5dBoNHBxkR4dO4Zj/HgpGPrgg40gIri7a+RAVIPwcD8kJAQCAM6cKYOnpzM8PNRQKm99VMStd/XXq1cafv31fUj/v5QALACsuOeedxsUsBDRXiFE4o2WJyMjIzs+Pv5CfdKuXAnP1FSEqNUQLi6wVlZCYTKBZszAmRsJWObOndti9uzZrYgI0dHRlT/++OMp+5aVzMxM5xdffDHk0qVLKo1GY/3qq69Od+rUqeq7777z+vjjjwNNJpPCx8fHvHz58pNt2rQxp6amts7JyXE6ffq087lz55xefPHF8++8885Vl8/WZubMmX7btm1z/+GHH7Jt6zIzM53vu+++yPz8/AOpqamtT5486XzmzBnnoKAgwwsvvHBh+vTprbZs2XL83Llzqscee6xdQUGBU0JCQvlvv/3muXfv3sOBgYFmV1fXTnq9ft/atWs90tLSWvv6+pqOHDniEhcXp//xxx9PKRQKvPHGG4E///yzt8FgUCQmJpYvXbr0tEKhaLYtKxkZGX7x8fGh1dff9i0rb7/9Md57bwSEKANgBqACkTfefvvjK9IREZydneDv7wT/mzTVl9VqRUVFlRwYGS8HR+XlUlBk+1tRUXU5MPozQPozOKqqMsBoNCAoqC2MRgMMhkoYDJUwmSphsZRj375fYDRWQq12h9UqfalKAYfA3r3/h/3718NqNcFiqYAUqFwuIWxfvmbzSej1lZACGNsXNmHKlHGYPt0ZFosBZnPJ5e0SwoQJr+G999Ty9nI5SPpz+2uvPQ+FQgWLxQCL5er8hw17EgqFEhaLEUIYLu9n8/DDj4JIAavVBKvVdNX2u+/uByKC1WqGEJbL223FiI7uLm+34M9AT9poNlcgOroHAMhBIsnL0vtFr78AIS5C+vK4uqm3slKP4OAhcHJygdlshJ9fNFQqFRQKJRQKBdzcvPDMM0vh5uYCP79AuLq6QwjIr40VRIRDh4rh4+OCnJyTKCo6B5OpCmazAWZzFXJze10OVqZN+xBGY9kVx09KGoAtWyYCAHS6PhDCDICgVGqgUmnQt28yvv/+RVRWmhAf/xKcnDRwdrY9nPG3v/XE66/3QHFxFd54YwVcXTVy96UG7u4a3HVXe3TuHISKChPS08/C01MDLy/p4eHhBCenmvswm7LrTKo/wGQSUKkApZJQXm7CxYtVMBotMBotMJksMBrNiItrCRcXFY4fL0JWViFMJts2C8xmC554ogNcXFTYujUbe/aclreb5VZBCyZN6g+VSoHFi/+HnTuPwmKxytul1rklS0bCYAAmT96InTsPwmQyw2g0wmCoApECr78+HGVlVfjmm3/hzJnDsFiMsFik97FCoUZS0gBUVlZh27b3If3vsv3gUgEwY+vWmQBubevKxIlodfQoNLVt/89/4FVZCYVKBVEkf2WazaBXXkHo6tUoqWkfrRZVkyah1q6c9PR0zbRp0wJ37tyZFRgYaD5//ryyeppnn3227fz580/HxcUZNm/e7PbSSy+F7Nq162ifPn3KBw0alKVQKDBjxgy/tLS0gC+//DIXAI4fP67ZsWPHkeLiYmV0dHTsuHHjCp2dnev1az4zM1Nzxx13XNFSEhMTY9Dr9YpLly4pAODYsWOa3bt3Z7m7u4u1a9d62NK99dZbrXv27Fn20Ucf5a9atcpzxYoVfjUd4/Dhwy779+8/GRoaakpISIjauHGj+/33318+bty4gmnTpuUBwMCBA9t9//33XkOGDKmxbpuz2z5Ykf75LcTMmVNRXp4Nd/dQh/kVp1Ao4OHhCg8P15tyPFvrUUXFny1HUkuS4fJj+PAekL60Ffjzy1tqyRk3bjZqbom7el31dLU34Imr0lzrGHWnq+UoNaS9kXU2c+eOAqDHny1RtpYVZ8TFdURVVSWqqipRWWmG0ViJqqpiGI16mEyVMJuNOHJkS6155+T8D507r5KfKeRWNhc4ObWAu7sLzp8vhE73KjQaF4SE3AGVSgWVSgmlUgmFQglv7xZ4992f4e7ugs6dHwSRVHdWqxVCWBAe3gaVlQKVlWY4O6thMFSgvPwiTCYDTKYqaLVtAfTAuXOlWLZs9lXle+qp19G585M4cCAPDzzw2FXbX375bUyd+ig2bz6BoUPHwclJA7Vag1OnNsjdjFZIX8JqAMCkSc/h++8fhtVqgdVqwUcfvYbk5DgsWrQH7747/fJ6q9UMq9WCr76air/9TYcPPvgFU6Z8ACEslx+AFatXf4f779fi3Xd/wPz5n1xVvq1b/w8dOwbh/ff/DytXzr3camn7W1j4FVxdNZg/fwkOHfoJf3b/So9du1JhNluQkfELSksz8GdwL+Wxbt16WK1VKC8/DIulEH+2iAKAEi+99G8AQFXVeQhRIa+XtisUKvz+ey6USpVcRzbi8v5ClNf63nEUFRVQqtWXm1oBAEolREUFrgow6mvDhg2eAwYMKAoMDDQDQKtWrSz220tKShT79u1zT0lJaW9bZzQaCQBOnTrlNHDgwODCwkK10WhUtGnT5vIVFH379i12cXERLi4uZl9fX1Nubq6qffv2JjSSfv36Fbu7u1/1z+T33393//HHH48DwGOPPVbq6elpuXpvIC4ursJWnpiYGP2JEyecAGD9+vUeM2bMCKiqqlIUFxerdDpdJVBzINic3fbBCiAFLI4QnNxqttYjZ2cn+Pq615jmhRei5G4zQGpylv7BOjtH4+OPk29WUZuNb79dhJKSYxCiBIABgAZEXvDyisCvv35Q574WiwUVFZW4dEmPkpIqXLqkR3GxHqWlVSgp0aO0VI/y8iqUlelRXq5HRUUl9PpK6PV6VFZWyoFQMS5dyoPRWHk5CLJYTJcDul9+qf34e/euw8yZk+QgyFUOhFzg4tIC3t4uSE/PxF13TYJGo0GPHoOhVquhVkvBkFKphKenJ+bN2wMhgMcfHyW3atha4kzo2TMaAODm5oSwsCgYDFUwGg2wWi/JdaWG9C/IBOkqPQsslioolSq5BUoa/+Tm5oqWLYMhhIDV+ud4sT17cpGTo8fRo+cRGBgljyWyyC0bFrz//ip8+KEK58/nwMWlNaxWMywWM4Qww2IxoXfv5yGEGUZjESyWchDZWs+kgGHixBcvt9opFOrLrX0KBUGhAPbuXQwnJ1e4uKjh6RkPZ2cXODu7QKNxgYuLC1xd3eDiooGr691wc9PA1VUDd3cXuLtr4OHx58PT0wVeXtI4Jjc3F7mrzwXOzs5yPfhACD2u/HdtAVHNn+Gbqa4WEADo0gXOBQVQu7r+GbDo9VC0bAnTsmVoklmgLRYLPDw8zFlZWYeqbxs1alTIa6+9lj906NASW9eKbZt9K4pSqYTZbL6iv3Tx4sXeH374YWsAmD9/frb9mBOdTlf122+/XfGCHDp0yMnV1dXq6+trBQA3N7crgraGqql8er2exo4d23b37t2HwsPDTampqa2rqqpuy6t8/xLBCqu/+nabMcmYMePkLo0gKJWusFj0sFqNGDNm3DX3lb7w3eHp2bhfOmazWQ6CKlFcbAuGKlFaWoniYj3KyipRXl6J0lIpAKqo0MtBUCUqK/WoqqqEwVCMCxfOwWiUuhNNJr38RV/fUiixZs0yqFRSEOTs7AInJ1doNC64suXOBNugcsAKoxEwmcphNFbhmWf+jqefroLVWmk3cP1PM2dur/HICoXUjVhefhJqtQvUaikQcHb2vNzNJQUUGri6usDVVXP54e4uBRT2gYSnpwbe3lJA4e4uBRIajQZqtbpat2bT6NlzjDxmxQz7MSs9e45p8mPfqLFjkZ+aihC9Hgr7MStjxyL/evO8//77Sx977LHw8ePH5wcEBFjOnz+vtG9d8fX1tQYHBxsXLFjgM3LkyCKr1Yrdu3e7dO3atbKsrEwZEhJiAoBFixa1aMhxhw0bVjxs2LDimrY9//zzF2fMmBH4448/egwcOLCsvLycXnnllZDRo0df8zw7d+5cvmTJEt8PPvggf/Xq1Z6lpaX1bnXS6/UKAAgICDCXlJQo1qxZ4zNgwIBmNUalvjhYYVdw5G4zRyTVy1yHqi+VSgUvLw94eXlcO3EDmEwmlJdXoqjI9pCCIOkhBUFSIKRHeXml3BKkrxYEXcKfg7D/HIht++vtrYGzs/cVA4nd3Fwuj5dxd3eBm5sURNgCClsgYWudcHHRQKPRoKZLRpujLVsmoFcvyFcDlYPIvdlcDSQPoj0zfToCcnPhHBx841cDJSYmVo0dOzave/fuUQqFQsTGxur/9a9/ZdunWbZs2cnnnnuu7ZQpUwLNZjM98sgjl7p27Vo5fvz4c4MHD27v5eVlvvvuu8vOnDnjfKPnCADu7u5i9erVx0eNGhXy+uuvq61WK1JSUi7+4x//uOYg3Y8//vjcY489FhYREdEiISGh3M/Pz+Tt7V1jV1B1fn5+lqFDhxZGR0fH+Pv7m+Pj4yuuvVfzdNtfDcQYcyw+PvdW6zpzvtx1VlS0+VYX77Z1K64GYtdWWVlJKpVKqNVqbNq0yW3UqFFta+rC+qv4y14NxBhzLDfSdcbY7eb48eNOjz/+eHur1Qq1Wi3mzZuXfavL5Ig4WGGM3VSO2HXG2K0SFxdnOHz48F+2JaW+OFhhjN10fIUeY6whbo8RaIwxxhi7bXGwwhhjjDGHxsEKY4wxxhwaByuMMcaajSNHjjhFRETENGSfpUuXer399tsBAJCamtp6woQJrRojXwBYsmSJt1ar1YWFhcVotVrdkiVLvGtL+8knn/jPnTu3zpvRbdu2zXX48OFtGloOm6CgoLi8vLzbbjzqbXdCjDHGHMfKlfCcPh0BOTlwbtPmxm8Kdz2GDh1agiaYL2fnzp0u48ePD/7ll1+ORkVFGbOyspz69u2r1Wq1hjvvvLPSPq3JZMKbb75ZeK08e/Toobe/lT+TcMsKY4yxJrFyJTxTUxFSUAC1tzfMBQVQp6YiZOVKeN5IvmazGQ8//HC7sLCwmH79+oWVlZUpgCtbFbZt2+aalJQUCQCzZ89uMWzYsJDq+fz222+ukZGRusjISN2MGTNaNrQcU6ZMCUhNTc2LiooyAkBUVJRxzJgx+R999FEAACQlJUWOHDmyTWxsbPTkyZNb2bfqbN261VWr1eqioqJ0L7zwQrCtVWft2rUevXr1CgekVqCUlJTQpKSkyODg4LjJkydfLmPv3r3bx8TERIeHh8dMmzatxpmabyccrDDGGLtugwejbfXHggXwAYCpUxFQXAxVURFU587BqagIquJiqMaPRzAAXLgAZfV963PM7OxszahRowpOnjyZ6eHhYZ06dar/9ZT9mWeeCf3000/PHDly5Lruc3L06FHNnXfeeUUrSJcuXSqOHj2qsT03Go108ODBw5MmTbpi0sdnn3223T//+c/TWVlZh5RKZa23kj9+/Lhm69atR/fs2XN42rRprQ0GAwHA0qVLszMzMw/v37//0Lx581rl5+df90zWzQEHK4wxxppEbi6clUpc8UWsVEJcvAj1jeQbEBBg7Nu3bwUAPPXUUxd37NjR4NlAL1y4oCwrK1P279+/HABGjhx58UbKVJvBgwdfqunYFRUVit69e1cAwNNPP31VGpu+ffsWu7i4iMDAQLOvr68pNzdXBQBTpkxpFRkZqUtISIjOz89XZ2ZmamrL43bAY1YYY4xdt2XLcLq2bSEhMBQUQO3qennWSuj1ULRsCRMA+PnBUtf+tak+27XtuVKpFFardKjKyspG+TE+evTooI0bN3oBQPU5eyIiIqp2797t2rVr18vjU3bv3u2q1WqrbM89PDysuAHOzs6Xgz2lUgmz2Uxr16712Lp1q0d6enqWh4eHNSkpKbKxztdR3dYnxxhj7NYZOxb5JhNIr4dCCClQMZlAY8ci/0byzcvLc9q0aZMbACxdutS3W7du5QAQHBxs3L59uysArFixwqeuPPz8/CweHh6WDRs2uAPAokWLfGtKN2fOnLNZWVmHappc8O9//3v+zJkzA48cOeIESFcUzZgxI/DNN9+s8/z8/Pwsbm5u1s2bN7sBwJIlS2o8dm2Ki4uVXl5eFg8PD+u+ffs0GRkZbg3ZvzniYIUxxliTSElB6YwZONOyJUwlJVC1bAnTjBk4c6NXA4WGhlbNmTOnZVhYWExxcbHqjTfeKASACRMmnHvzzTdDYmNjo+saB2Lz9ddfZ7/66qshUVFROiEEXSt9dd26datMS0vLHTBgQHi7du1iBgwYEP7+++/nduvWrfJa+86bNy/7xRdfbBsVFaWrqKhQeHh4WOp73OTk5BKz2UxhYWEx48aNC4qPj69oaNmbGxLimq+nQ0lMTBTp6em3uhiMMdasENFeIUTijeaTkZGRHR8ff6ExyvRXVlJSovDy8rICwNtvvx2Ql5enXrhwYc6tLtetlpGR4RcfHx9afT2PWWGMMcZushUrVnhNnz490GKxUFBQkOG7777LvtVlcmQcrDDGGGM32XPPPVf03HPPFd3qcjQXPGaFMcYYYw6NgxXGGGOMObQmDVaIqB8RHSGi40T0Vg3bQ4hoCxHtI6IDRPRAU5aHMcYYY81PkwUrRKQE8BmA/gB0AAYTka5asncArBBCdAIwCMA/m6o8jDHGGGuemrJlJQnAcSHESSGEEcD3AP5WLY0ALk9o5QXgXBOWhzHG2G3KfpLAm23Hjh0uHTt2jAoPD4/RarW6L7/88vIN6bKyspw6dOgQFRISEvvggw+GVVVVEQCsX7/eXafTRatUqoSFCxdecQO7Y8eOOd11110RYWFhMe3bt4+x3XTOXnJycmhQUFBcZGSkLjQ0NPaRRx4JPXHixDWnMbCf7NHeJ5984j937twWDT33bdu2uQ4fPrxNQ/drqKYMVoIA2F8zniuvs/cegCeJKBfAOgCjm7A8jDHGbrKVmSs9u3zVRRs0Iyiuy1ddtCszV97QjMuOyN3d3bpkyZJTx48fz/zll1+Ovf32220uXLigBIDU1NTgUaNGnT9z5sxBLy8v86xZs/wAICwszLhw4cLsAQMGXDUn0dChQ9u98cYb50+ePJn5v//973Dr1q3NNR138uTJuUeOHDl08uTJgx07dtTfd999kbZgqKHefPPNwlGjRjV4fqQePXroFy1a1OT3h7nVA2wHA1gkhAgG8ACAJUR0VZmI6HkiSiei9MLCwpteSMYYYw23MnOlZ+qG1JCCigK1t7O3uaCiQJ26ITXkRgOWuXPnttBqtbrIyEjdwIED21XfnpmZ6dy9e/eImJiY6ISEhMh9+/ZpAOC7777z6tChQ1R0dLSuW7du2pycHBUgtcqkpKSEJiUlRQYHB8dNnjy5ZUPK06FDB0NcXJwBAEJDQ02+vr7mvLw8ldVqxc6dOz1GjBhRBEiTJa5Zs8YbACIjI4133nlnpUJx5Vfe3r17NRaLBY888kgpAHh5eVmvNb+QQqHAxIkTC/z8/EyrVq3yAoDVq1d7duzYMUqn00X3798/rKSk5PKBJk2aFKDVanVxcXHRBw8edLbVga1lavr06X6xsbHRkZGRuvvvv799WVmZAgAWLFjgExERERMZGalLTEyMBIC1a9d69OrVK/xa9Thu3LjA0NDQ2ISEhMgBAwa0a2grWFPeZ+UsAPumoWB5nb1nAPQDACHETiLSAPADUGCfSAgxH8B8QLqDbVMVmDHGWP1N3DKx1dGLR2ud7fc/p/7jVWmuVKgUKlEE6ZYiZquZXln3Sujqw6tLatpH20JbNanXpPO15Zmenq6ZNm1a4M6dO7MCAwPN58+fV1ZP8+yzz7adP3/+6bi4OMPmzZvdXnrppZBdu3Yd7dOnT/mgQYOyFAoFZsyY4ZeWlhbw5Zdf5gLA8ePHNTt27DhSXFysjI6Ojh03blyh/SSC9bVlyxZXk8lEOp3OcP78eZWHh4dFrZZ6Z0JDQ43nz5+/qkvH3qFDhzSenp6Wvn37ts/JyXHu0aNH6WeffZarUl3767pDhw76w4cPa/Ly8lQffvhh4LZt2456enpax48fH/D++++3mjZtWh4AeHl5mY8ePXpo7ty5LUaPHt1my5Ytx+3zGTp0aNHYsWMvAMCrr77aevbs2X7jx48v+PjjjwN/+eWXo+3atTPZWo6qq6ked+3a5bJmzRqfQ4cOZRoMBurYsaOuU6dO+vrVqKQpg5U9ACKIqB2kIGUQgCHV0pwBcB+ARUQUDUADgJtOGGPsNlBhqlCqFeorWgWUpBQVpooav+jqY8OGDZ4DBgwoCgwMNANAq1atrphTp6SkRLFv3z73lJSU9rZ1RqORAODUqVNOAwcODC4sLFQbjUZFmzZtDLY0ffv2LXZxcREuLi5mX19fU25urqp9+/amhpTt9OnT6hEjRoR9/fXXp5TK6ztFs9lM6enp7rt37z4UERFhfOihh9rPmTPHb8yYMdec4sA2fc6vv/7qduLECU1SUlIUAJhMJkpISCi3pXv66acvAcBzzz136Z133rlqvMnevXtdJkyYEFRWVqasqKhQ9uzZswQAEhMTy4cOHRqanJxcNHTo0BpvaFdTPW7dutW9f//+xa6ursLV1VX06dOnuKH10mTBihDCTESjAGwAoASwQAiRSURpANKFED8BGAvgSyIaA2mw7XDR3CYrYoyxv6i6WkAAoMtXXZwLKgrUrmrXywGL3qRXtHRraVr22LLTTVEmi8UCDw8Pc02zJI8aNSrktddeyx86dGjJ2rVrPdLS0lrbttm3oiiVSpjN5ivGfixevNj7ww8/bA0A8+fPz+7Ro8cVLQOXLl1S9O/fP3zixIln77vvvgoAaNWqlbmsrExpMpmgVquRnZ3t1KpVK2Nd5Q8JCTFGRUVV6nQ6IwA8/PDDRbt27XKvz7n/8ccfrr17984XQuDuu+8uXbNmzama0tl3PRHRVd+5zz//fLtVq1Yd79q1a+Xs2bNbbN261QMAvvvuuzObN292++mnn7wSEhJ0e/fuvaqOr1WP16tJx6wIIdYJIbRCiPZCiA/kdRPkQAVCiENCiLuEEPFCiI5CiF+asjyMMcZunrFdx+abLCbSm/QKIQT0Jr3CZDHR2K5j8683z/vvv790zZo1Pvn5+UoAqN4N5Ovraw0ODjYuWLDABwDkcSMuAFBWVqYMCQkxAcCiRYsadOXLsGHDirOysg5lZWUdqh6oVFVV0YMPPhg+aNCgi7bxKYAUFHTp0qXMdrXPggULWjz00EPFdR2nZ8+eFaWlpcpz586pAGDLli2eOp2uzlmcrVYrJk+e3LKwsFCdnJxces8991Skp6e728ajlJaWKg4cOOBsS7948WJfAPj66699OnXqdNWMzXq9XhESEmIyGAz0/fff+9rWZ2ZmOt97770Vn3766TkfHx/zyZMn6+zSsjun8g0bNnjp9XoqKSlRbNq0ybs++9m71QNsGWOM3aZSYlJKZ9w/40xLt5amEkOJqqVbS9OM+2ecSYlJKb3ePBMTE6vGjh2b171796jIyEjdyy+/fFU3xrJly04uXLjQLzIyUhcRERHzr3/9yxsAxo8ff27w4MHtY2Jiolu0aFHjFTbXY8GCBT579uxx/+677/yioqJ0UVFRuh07drgAwPTp03PnzJkTEBISEltUVKR67bXXLgDA1q1bXVu1atVh3bp1PmPGjGkbHh4eAwAqlQoff/xx7j333KPVarU6IQRq6wJ65513giMjI3Xt2rWLTU9Pd9u8efMRjUYjWrdubZ43b172oEGDwrRarS4xMTHqjz/+uDy2qKioSKnVanX//Oc/W82ePfuqK3neeuutc0lJSdGJiYlRERERVbb1Y8aMCdZqtbqIiIiYzp07l3fp0qXOIMqmZ8+e+n79+pXodLqYe++9NyIyMrLSy8vLcu09/0TNrdclMTFRpKen3+piMMZYs0JEe4UQiTeaT0ZGRnZ8fPw1x08wZq+kpETh5eVlLSsrU3Tt2jXyiy++OH333XdfNcg2IyPDLz4+PrT6ep51mTHGGGNN6sknn2x77NgxF4PBQIMGDbpYU6BSFw5WGGOMMdakahvsW188ZoUxxhhjDo2DFcYYY4w5NA5WGGOMMebQOFhhjDHGmEPjYIUxxlizceTIEaeIiIiYhuyzdOlSr7fffjsAuHLCvhvN9/PPP/fVarU6rVar69SpU5Tt5nMAsGrVKs/Q0NDYkJCQWNuxAeDDDz/0DwkJiSWihLy8vCsuclm7dq1HVFSULjw8PKZz586RNR0zKCgoznbM9u3bx7z66qut9Xp9nXeJrevcnnjiibZ79+6tdX6n2nzyySf+c+fObdCN9W4EXw3EGGOsyazMXOk5fef0gJzSHOc2nm0MY7uOzb+Rm8Jdj6FDh5YAqHHixBsRHh5u2L59+xF/f3/LihUrPF944YW2Bw4cyDKbzRgzZkzIhg0bjoaFhZni4+Ojk5OTixMSEqp69uxZnpycXHLvvfdeEYxcuHBB+dprr4X8/PPPxyIiIoxnz56t9ft569atRwMDA80lJSWKJ598su2TTz7ZdvXq1dnXcw7Lly+/rmkP3nzzzZs6jx+3rDDGGGsSKzNXeqZuSA0pqChQezt7mwsqCtSpG1JDVmau9LyRfM1mMx5++OF2YWFhMf369QsrKytTAFKrg621Ytu2ba5JSUmRADB79uwWw4YNC6mez2+//eYaGRmpi4yM1M2YMaNlQ8vRp0+fCn9/fwsA9OrVqyI/P98JkCYSbNu2rUGn0xk1Go149NFHL61atcobAO66667KyMjIq+YH+uqrr3wffPDBooiICKN8Lte8w66Xl5f1m2++Ob1x40Zv27QD7777bqvY2NhorVarGzNmzOW5j2qrs6SkpMht27a5AsDQoUNDYmNjo8PDw2Ps93355ZeD2rdvH6PVanXPP/98MHBlC1VSUlLkSy+9FBQXFxcdGhoa+/PPP7sDQFlZmeKBBx4Ia9++fUyfPn3ad+jQIcp2rIbilhXGGGPXbfCqwW2rr+vTvk/pyE4ji6bumBpQbChWqRQqUQRpyhyz1UzjN48PTolJOXRBf0E5et3oYPt96zPBYXZ2tmbevHnZffv2rUhJSQmdOnWqf1paWp2TKtbkmWeeCZ01a9aZ/v37l7/wwgvB196jdnPmzPHr1atXCQDk5OQ4BQUFXQ5IgoODjbt3765zMsKjR49qTCYTJSUlRVZUVCheeumlglGjRl281nF9fX2tQUFBxszMTM327duVx48f1xw4cOCwEAK9e/cOX79+vXtYWJixPnU2Y8aMs61atbKYzWZ069Ytcvfu3S5t27Y1rlu3zufkyZMHFQoFLly4UON00mazmf7444/Dy5cv90pLS2vdr1+/o1OnTvX39va2nDhxInPPnj2arl27NqibzR63rDDGGGsSuaW5zkpSXjGni5KU4mLlRfWN5BsQEGDs27dvBQA89dRTF3fs2FGvWYntXbhwQVlWVqbs379/OQCMHDnymoFBbdasWePx7bff+s2aNSv3evMwm8104MAB102bNh3btGnTsalTpwbaTz5YF9u0OT///LPntm3bPHU6nS4mJkZ34sQJTVZWlgaoX5198803vjqdLlqn0+mOHTumycjI0LRo0cLi7OxsfeKJJ0K/+eYbb3d3d2v1/QAgJSWlCAC6detWkZub6wQAO3bscB88ePAlAOjcuXOVVqtt0F1r7XHLCmOMsetWV0tIiFeIoaCiQO2qdr38Bac36RUt3VqaAMDP1c9Sn5aU6oioxudKpVJYrdKhKisrG+XH+OjRo4M2btzoBQBZWVmHqm/fvXu3y8svv9z23//+97GAgAALALRp08Z49uzZyzMS5+bmXtHSUpPg4GBjixYtzJ6enlZPT0/rnXfeWZaenu7aoUMHQ137FRUVKc6dO+cUFxdXtWzZMrz++ut548aNu2LupiNHjjjVVmc2WVlZTnPnzm21d+/ew/7+/pbk5OTQqqoqhVqtxv79+w//9NNPnqtWrfL5/PPPW+7ateto9XJoNBoBSBMxWiyWOgf8Xg9uWWGMMdYkxnYdm2+ymEhv0iuEENCb9AqTxURju47Nv5F88/LynDZt2uQGAEuXLvXt1q1bOSB94W/fvt0VAFasWOFTVx5+fn4WDw8Py4YNG9wBYNGiRb41pZszZ87ZrKysQzUFKseOHXNKSUlpv2DBglP2QUXPnj0rsrOzNVlZWU5VVVW0evVq3+Tk5OK6yvPYY48V79q1y91kMqGsrEyxb98+97i4uDpnNS4pKVGMGDGibZ8+fYr9/f0t/fv3L12yZIlfSUmJAgBOnTqltg3Ura3ObIqKipQuLi5WX19fS05OjurXX3/1sh3j0qVLyieeeKLkiy++yMnKyqr3mJOuXbuWf//99z4AsHfvXs3Ro0ddrrVPbbhlhTHGWJOQr/o5M33n9IDc0lznYM/gRrkaKDQ0tGrOnDktn3/+edeIiIiqN954oxAAJkyYcO7FF18MTUtLs3Tr1q3sWvl8/fXX2c8++2woEeGee+5pcJneeeedwOLiYtXo0aPbAoBKpRIHDx48rFarMX369DP9+vXTWiwWDBky5EJiYmIVAEyePLnlnDlzAi5evKiOj4/X9erVq2T58uWn77jjjqrevXuXREVFxSgUCjz11FOFnTt3rqrpuD179tQKIchqteKBBx4onjJlyjkAePTRR0szMzM1nTt3jgIAV1dX69KlS0+pVCpRW53ZdO3atTI2Nlbfvn372MDAQGNCQkI5ABQXFysfeuihcIPBQADw/vvv59S3fsaNG1f4+OOPh7Zv3z6mffv2VeHh4VU+Pj6WhtYzAJCtr6u5SExMFOnp6be6GIwx1qwQ0V4hROKN5pORkZEdHx9/4dop2V+d2WyG0WgkV1dXkZmZ6dy3b1/tiRMnDtq6jGqSkZHhFx8fH1p9PbesMMYYY6zRlZWVKbp37x5pMplICIGZM2eeritQqQsHK4wxxhhrdD4+PtaDBw8eboy8eIAtY4wxxhwaByuMMcYYc2gcrDDGGGPMoXGwwhhjjDGHxsEKY4yxZs9+Yr2bbceOHS4dO3aMCg8Pj9Fqtbovv/zy8g3psrKynDp06BAVEhIS++CDD4ZVVVURAKxfv95dp9NFq1SqhIULF15xA7sXX3wxODw8PCYsLCxm+PDhbWx35bWXlJQUGRoaGqvVanXt2rWLGTZsWEht8/bYc3V17VTT+tdff731jz/+6NHQc1+6dKnX22+/HdDQ/RqKgxXGGGNNZuXKlZ5dunTRBgUFxXXp0kW7cuWNzbjsiNzd3a1Lliw5dfz48cxffvnl2Ntvv93GFjikpqYGjxo16vyZM2cOenl5mWfNmuUHAGFhYcaFCxdmDxgw4Io5iTZu3Oj2+++/u2dlZWUePXo0c//+/W7r1q2rMYhYvHjxyaNHjx46fPjwIWdnZ2v//v3Dr/ccPv3003MDBw685o30qhs6dGjJhx9+eEN3JK4PDlYYY4w1iZUrV3qmpqaGFBQUqL29vc0FBQXq1NTUkBsNWObOndtCq9XqIiMjdQMHDmxXfXtmZqZz9+7dI2JiYqITEhIi9+3bpwGA7777zqtDhw5R0dHRum7dumlzcnJUgNQqk5KSEpqUlBQZHBwcN3ny5JYNKU+HDh0McXFxBgAIDQ01+fr6mvPy8lRWqxU7d+70GDFiRBEgTZa4Zs0abwCIjIw03nnnnZUKxZVfw0QEg8FAVVVVVFlZqTCbzdS6dWtTXcfXaDTi888/zz137pzTzp07XQDgn//8p29cXFx0VFSUbsiQIW3NZvPl9M8880yb8PDwmK5du2rPnTunAoDk5ORQWwvPG2+8ERgbGxsdERERM3jw4La2lp3Jkye3bN++fYxWq9U99NBDYQAwe/bsFsOGDQux5TF8+PA2nTp1igoODo6z5WexWPDkk0+GtGvXLqZbt24RPXv2DK/emnQtfJ8Vxhhj12XixImtjh49qqlt+3/+8x+vyspKhUqlEkVFRQCk2YVfeeWV0NWrV5fUtI9Wq62aNGnS+dryTE9P10ybNi1w586dWYGBgebz589f1fXx7LPPtp0/f/7puLg4w+bNm91eeumlkF27dh3t06dP+aBBg7IUCgVmzJjhl5aWFvDll1/mAsDx48c1O3bsOFJcXKyMjo6OHTduXKGzs3ODb2C2ZcsWV5PJRDqdznD+/HmVh4eHRa2WJpkODQ01nj9/3qmu/Xv37l1x1113lQUGBsYDwPDhwwvvuOOOGm+7b0+lUiE6Olp/8OBBjbOzs1i1apVvenp6lrOzs3jyySdDvvjiixajRo26WFlZqUhMTKz4+uuvc954443At956q/XixYvP2Oc1bty4gmnTpuUBwMCBA9t9//33XkOGDCmZPXt2wOnTp/9wcXERtXU5nT9/Xp2enp61f/9+zSOPPBI+YsSIosWLF/vk5OQ4HT9+PPPs2bOq2NjY2OHDhzdolmsOVhhjjDWJiooKpVqtvmLAhVKpFBUVFdccW1GbDRs2eA4YMKAoMDDQDACtWrW6Yq6ZkpISxb59+9xTUlLa29YZjUYCgFOnTjkNHDgwuLCwUG00GhVt2rS5PPlg3759i11cXISLi4vZ19fXlJubq2rfvn2dLRrVnT59Wj1ixIiwr7/++pRSeX2nePDgQeejR49qcnNzDwDSPEA///yze79+/cqvta9t+pyff/7Z4+DBg67x8fHRAFBVVaVo2bKlGQAUCgWeffbZS4DU0vPoo49e1XW0fv16jxkzZgRUVVUpiouLVTqdrhJASWRkZOUjjzzS7uGHHy4eOnRocU1lePjhh4uVSiUSEhKqLl68qAaA3377zf3RRx8tUiqVCAkJMXfp0qXB3U0crDDGGLsudbWAAECXLl2cCwoK1K6urpcDFr1er2jZsqVp2bJlp5uiTBaLBR4eHuaaZkkeNWpUyGuvvZY/dOjQkrVr13qkpaW1tm2zb0VRKpUwm81kv+/ixYu9P/zww9YAMH/+/OwePXro7bdfunRJ0b9///CJEyeeve+++yoAoFWrVuaysjKlyWSCWq1Gdna2U6tWrYx1lX/58uXenTt3rvDy8rICQO/evUv++9//ul0rWDGbzThy5Ihrhw4dzm3atMkjJSXl4meffXa2rn0AqdvJnl6vp7Fjx7bdvXv3ofDwcFNqamrrqqoqBQBs2bLl2Pr16z3+7//+z2vatGmBR44cyayen/3t9Btz7kEes8IYY6xJjB07Nt9kMpFer1cIIaDX6xUmk4nGjh173QMy77///tI1a9b45OfnKwGgejeQr6+vNTg42LhgwQIfAJDHjbgAQFlZmTIkJMQEAIsWLWrRkOMOGzasOCsr61BWVtah6oFKVVUVPfjgg+GDBg26aBufAkitGF26dCmzjc9YsGBBi4ceeqi4ruOEhIQYt2/f7mEymWAwGGj79u0eOp2uzm4gg8FAo0aNCg4MDDTeeeedlf369Stdu3atz9mzZ1WAVEdHjx51stWHrTyLFi1qkZSUdEUrh16vVwBAQECAuaSkRLFmzZrL405OnDjhNGDAgLLPPvvsbHl5ubKkpKRezUd33313+Y8//uhjsViQk5Oj2r17d4OvOuJghTHGWJNISUkpnTFjxpmWLVuaSkpKVC1btjTNmDHjTEpKSun15pmYmFg1duzYvO7du0dFRkbqXn755TbV0yxbtuzkwoUL/SIjI3UREREx//rXv7wBYPz48ecGDx7cPiYmJrpFixbmqzK/TgsWLPDZs2eP+3fffecXFRWli4qK0u3YscMFAKZPn547Z86cgJCQkNiioiLVa6+9dgEAtm7d6tqqVasO69at8xkzZkzb8PDwGAAYMWJEUWhoqCEyMjJGp9PpYmJi9EOGDKlxfM+wYcPC5IHGMRUVFYr169cfB4CEhISqd9555+x9992n1Wq1unvvvVebk5OjBgAXFxfr77//7hYRERGzbds2j48++ijPPk8/Pz/L0KFDC6Ojo2N69eqljY+PrwCksUZDhgxpp9VqdbGxsbpnn322wM/Pz3J1qa729NNPFwUGBhrDw8NjnnjiiXYxMTF6b2/veu1rQ43ZTHMzJCYmivT09FtdDMYYa1aIaK8QIvFG88nIyMiOj4+/0BhlYn8dJSUlCi8vL2t+fr6yc+fO0du3b88KCQm5KmDMyMjwi4+PD62+nsesMMYYY6xJ9enTJ6K0tFRpMplo3LhxeTUFKnVp0mCFiPoBmAVACeArIcTHNaR5HMB7AASADCHEkKYsE2OMMcZurt9///3IjezfZMEKESkBfAagD4BcAHuI6CchxCG7NBEA/gHgLiFEERE16EY8jDHGGLv9NeUA2yQAx4UQJ4UQRgDfA/hbtTTPAfhMCFEEAEKIgiYsD2OMMcaaoaYMVoIA5Ng9z5XX2dMC0BLRdiLaJXcbXYWInieidCJKLywsbKLiMsYYY8wR3epLl1UAIgDcA2AwgC+JyLt6IiHEfCFEohAi0d/f/+aWkDHGGGO3VFMGK2cB2F//Hiyvs5cL4CchhEkIcQrAUUjBC2OMMXaVI0eOOEVERMQ0ZJ+lS5d6vf322wGANGnhhAkTWjVGvp9//rmvVqvVabVaXadOnaJsN58DgFWrVnmGhobGhoSExNqODQAffvihf0hISCwRJeTl5V0eN3rx4kXlvffeGx4ZGakLDw+PmTVrVo03rVMqlQlRUVG68PDwmMjISN3EiRNbWSx137Jk7dq1Hr169apxRuaePXuG1zbPT11ef/311j/++GODb+52vZryaqA9ACKIqB2kIGUQgOpX+vwIqUVlIRH5QeoWOtmEZWKMMXYTrVy50nP69OkBOTk5zm3atDGMHTs2/0ZuCnc9hg4dWgKgxhur3Yjw8HDD9u3bj/j7+1tWrFjh+cILL7Q9cOBAltlsxpgxY0I2bNhwNCwszBQfHx+dnJxcnJCQUNWzZ8/y5OTkknvvvTfSPq+pU6f6R0ZGVm7evPn4uXPnVNHR0bEvvPDCJfvb1wOAs7Oz1TaVwNmzZ1UpKSlhpaWlypkzZ567nnPYunXr8evZ79NPP72u412vJmtZEUKYAYwCsAHAYQArhBCZRJRGRA/LyTYAuEhEhwBsATBOCNGgmRgZY4w5ppUrV3qmpqaGFBQUqL29vc0FBQXq1NTUkJUrV3reSL5msxkPP/xwu7CwsJh+/fqFlZWVKQAgKCgoztZasW3bNtekpKRIAJg9e3aLYcOGhVTP57fffnONjIzURUZG6mbMmNHgq1H79OlT4e/vbwGAXr16VeTn5zsBwK+//urWtm1bg06nM2o0GvHoo49eWrVqlTcA3HXXXZWRkZFXzQ9ERCgrK1NarVaUlpYqvLy8zGq1us67tgYFBZm/+uqr7IULF7a0Wq0wm8144YUXgmNjY6O1Wq1u6tSpfra0ZWVlynvuuSc8NDQ0dsiQISG21hj7Ouvdu3f7mJiY6PDw8Jhp06b52eo6OTk5NCIiIkar1eomTZrUEgCSk5NDbbftDwoKihszZkxrnU4XrdVqdfv27dMAwLlz51TdunWLkO9c27Z169Zx9q1JDdGk91kRQqwDsK7augl2ywJAqvxgjDHWzAwePLht9XV9+vQpHTlyZNHUqVMDiouLVSqVShQVSVPmmM1mGj9+fHBKSsqhCxcuKEePHh1sv299JjjMzs7WzJs3L7tv374VKSkpoVOnTvVPS0urc1LFmjzzzDOhs2bNOtO/f//yF154Ifjae9Ruzpw5fr169SoBgJycHKegoKDLAUlwcLBx9+7d7nXt/+abbxb069cvvFWrVh0qKiqUCxYsOFmfmZt1Op3RYrHg7NmzquXLl3t7eXlZDh48eLiyspI6d+4cNWDAgFIA+OOPP9z27dt3UKvVGnv06BGxePFiH/t5jABg6dKl2a1atbKUl5dTp06ddE8++WTRsWPHnPPy8tTHjh3LBIDauoz8/PzMhw4dOvzxxx/7f/zxx62WL19++q233mrds2fPso8++ih/1apVnitWrPCrad/6uNUDbBljjN2mcnNznZVK5RWtA0qlUly8eFF9I/kGBAQY+/btWwEATz311MUdO3bUGQjU5MKFC8qysjJl//79ywFg5MiR192qv2bNGo9vv/3Wb9asWbnXm8ePP/7oFRsbW3n+/PkDv//++6GxY8eGXLp0qUHf0Zs2bfJcsWJFi6ioKF2nTp2ii4qKVIcOHdIAQFxcXIVOpzOqVCo8/vjjl3777ber6mzKlCmtIiMjdQkJCdH5+fnqzMxMTVRUlCEnJ8f56aefbrNq1SpPHx+fGgfIDBkypAgAkpKS9Dk5Oc4A8Pvvv7s//fTTlwDgscceK/X09GzQfED2rtmyQkR3AdgvhKggoicB3AFglhCiSab3Zowx1nzU1RISEhJiKCgoULu6ulpt6/R6vaJly5YmQJo0rz4tKdURUY3PlUqlsFqlQ1VWVjbKj/HRo0cHbdy40QsAbGNF7O3evdvl5Zdfbvvvf//7WEBAgAUA2rRpYzx79qyTLU1ubu4VLS01+eabb1q89dZb+QqFArGxsYY2bdoYMjIyNL169dLXtd+hQ4eclEolgoKCzEIImj59+pnk5OQrxgStXbvWo7Y6s0+zdetWj/T09CwPDw9rUlJSZGVlpcLf399y8ODBQz/88IPnF1984b98+XLflStXZlcvh21sjUqlEmazmapvv1H1eTE/B6AnongAYwGcALC4sQvCGGPs9jJ27Nh8k8lEer1eIYSAXq9XmEwmGjt2bP6N5JuXl+e0adMmNwBYunSpb7du3coBqbtl+/btrgCwYsUKn7ry8PPzs3h4eFg2bNjgDgCLFi3yrSndnDlzzmZlZR2qKVA5duyYU0pKSvsFCxac6tChg8G2vmfPnhXZ2dmarKwsp6qqKlq9erVvcnJycV3lCQoKMv7yyy+eAJCTk6M6efKkJioqqs4A59y5c6rnnnuu7YgRIwoUCgX69OlT8vnnn/sbDAYCgAMHDjiXlpYqAKkbKCsry8lisWDVqlW+3bt3L7PPq7i4WOnl5WXx8PCw7tu3T5ORkeEGAHl5eSqLxYLhw4cXf/TRR2f/+OMP17rKZK9z587lS5Ys8QWA1atXe5aWljb4qiOb+oxZMQshBBH9DcBcIcTXRPTM9R6QMcbYX4N81c+Z6dOnB+Tm5joHBwc3ytVAoaGhVXPmzGn5/PPPu0ZERFS98cYbhQAwYcKEcy+++GJoWlqapVu3bmXXyufrr7/OfvbZZ0OJCPfcc0+Dy/TOO+8EFhcXq0aPHt0WkFoVDh48eFitVmP69Oln+vXrp7VYLBgyZMiFxMTEKgCYPHlyyzlz5gRcvHhRHR8fr+vVq1fJ8uXLT3/wwQd5Q4cODdVqtTohBL333nu5gYGBV032ZzAYFFFRUTqz2UxKpVI88cQTFydOnHgeAMaMGXMhOzvbOS4uLloIQb6+vqZ169adAIDY2NiKF198MSQ7O1vTrVu30qeeeqrYPt/k5OSS+fPn+4eFhcWEhYVVxcfHVwBAdna2+plnngm1Wq0EAGlpafXu6vr444/PPfbYY2EREREtEhISyv38/Eze3t7X1RVE0hjXOhIQbQXwM4CRALoDKIA04WDc9RzwRiUmJor09PRbcWjGGGu2iGivECLxRvPJyMjIjo+Pv9AYZWK3t8rKSlKpVEKtVmPTpk1uo0aNaltTC5W9jIwMv/j4+NDq6+vTsvIEpPujjBRC5BNRCICp11VyxhhjjP0lHD9+3Onxxx9vb7VaoVarxbx587KvN69rBitygPIv/Hln2QsAfrjeAzLGGGPs9hcXF2c4fPhwnS0p9XXNAbZE9ByAVQDmyauCIN15ljHG2F+P1TZ+gbHGJL+vrDVtq8/VQK8AuAtAKQAIIY4BaPCd/hhjjN0WDhYWFnpxwMIak9VqpcLCQi8AB2vaXp8xKwYhhNF2TTYRqQDUPSqXMcbYbclsNj+bn5//VX5+fiz4xqKs8VgBHDSbzc/WtLE+wcpWInobgAsR9QHwMoA1jVhAxhhjzURCQkIBgIevmZCxRlSfqPgtAIUA/gDwAqS5ft5pykIxxhhjjNnU52ogK4Av5QdjjDHG2E1Vn7mBTqGGMSpCiLAmKRFjjDHGmJ36jFmxv+OhBkAKgBrnUGCMMcYYa2zXHLMihLho9zgrhPgUwINNXzTGGGOMsfp1A91h91QBqaWlPi0yjDHGGGM3rD5Bx3S7ZTOAbACPN0lpGGOMMcaqqc/VQL1uRkEYY4wxxmpSa7BCRKl17SiEmNH4xWGMMcYYu1JdLSseN60UjDHGGGO1qDVYEUJMupkFYYwxxhirSX2uBtIAeAZADKT7rAAAhBAjm7BcjDHGGGMA6jc30BIAAQDuB7AVQDCAsqYsFGOMMcaYTX2ClXAhxLsAKoQQ30C6IdydTVssxhhjjDFJfYIVk/y3mIhiAXgBaNl0RWKMMcYY+1N9bgo3n4h8ALwL4CcA7vIyY4wxxliTq0+wslAIYYE0XoVnWmaMMcbYTVWfbqBTRDSfiO4jImryEjHGGGOM2alPsBIFYBOAVwBkE9FcIrq7aYvFGGOMMSa5ZrAihNALIVYIIR4F0BGAJ6QuIcYYY4yxJleflhUQUU8i+ieAvZBuDMezLjPGGGPspqjPHWyzAewDsALAOCFERVMXijHGGGPMpj4tKx2EEI8IIZY1NFAhon5EdISIjhPRW3WkSyYiQUSJDcmfMcYYY7e/+oxZKb2ejIlICeAzAP0B6AAMJiJdDek8ALwGYPf1HIcxxhhjt7d6jVm5TkkAjgshTgohjAC+B/C3GtK9D2AKgKomLAtjjDHGmqmmDFaCAOTYPc+V111GRHcAaCOE+HddGRHR80SUTkTphYWFjV9SxhhjjDmsOoMVIvKyW36GiD4joleJyPVGD0xECgAzAIy9VlohxHwhRKIQItHf3/9GD80YY4yxZqTWYIWIPgRwiIhyiegzAMkAdgHQAlhSj7zPAmhj9zxYXmfjASAWwK/yFUddAPzEg2wZY4wxZq+uS5cHQAo2ogFkAPATQhQDWEJEGfXIew+ACCJqBylIGQRgiG2jEKIEgJ/tORH9CuANIUR6A8+BMcYYY7exurqBnAB4CCEyAbwjByogonAAymtlLIQwAxgFYAOAwwBWCCEyiSiNiB6+4ZIzxhhj7C+hrpaVOQBOEFEeABCRrVUkBICZiA4AgBCiQ20ZCCHWAVhXbd2EWtLeU/9iM8YYY+yvotZgRQgxl4jmA2iFpr1qiDHGGGOsVnXebl++P0pOXWkYY4wxxpoSt5gwxhhjzKFxsMIYY4wxh8bBCmOMMcYcGgcrjDHGGHNoHKwwxhhjzKFxsMIYY4wxh8bBCmOMMcYcGgcrjDHGGHNoHKwwxhhjzKFxsMIYY4wxh8bBCmOMMcYcGgcrjDHGGHNoHKwwxhhjzKFxsMIYY4wxh8bBCmOMMcYcGgcrjDHGGHNoHKwwxhhjzKFxsMIYY4wxh8bBCmOMMcYcGgcrjDHGGHNoHKwwxhhjzKFxsMIYY4wxh8bBCmOMMcYcGgcrjDHGGHNoHKwwxhhjzKFxsMIYY4wxh8bBCmOMMcYcGgcrjDHGGHNoHKwwxhhjzKFxsMIYY4wxh9akwQoR9SOiI0R0nIjeqmF7KhEdIqIDRPQfImrblOVhjDHGWPPTZMEKESkBfAagPwAdgMFEpKuWbB+ARCFEBwCrAHzSVOVhjDHGWPPUlC0rSQCOCyFOCiGMAL4H8Df7BEKILUIIvfx0F4DgJiwPY4wxxpqhpgxWggDk2D3PldfV5hkA62vaQETPE1E6EaUXFhY2YhEZY4wx5ugcYoAtET0JIBHA1Jq2CyHmCyEShRCJ/v7+N7dwjDHGGLulVE2Y91kAbeyeB8vrrkBEvQGMB9BTCGFowvIwxhhjrBlqypaVPQAiiKgdETkBGATgJ/sERNQJwDwADwshCpqwLIwxxhhrpposWBFCmAGMArABwGEAK4QQmUSURkQPy8mmAnAHsJKI9hPRT7VkxxhjjLG/qKbsBoIQYh2AddXWTbBb7t2Ux2eMMcZY8+cQA2wZY4wxxmrDwQpjjDHGHBoHK4wxxhhzaBysMMYYY8yhcbDCGGOMMYfGwQpjjDHGHBoHK4wxxhhzaBysMMYYY8yhcbDCGGOMMYfGwQpjjDHGHBoHK4wxxhhzaBysMMYYY8yhcbDCGGOMMYfGwQpjjDHGHBoHK4wxxhhzaBysMMYYY8yhcbDCGGOMMYfGwQpjjDHGHBoHK4wxxhhzaBysMMYYY8yhcbDCGGOMMYfGwQpjjDHGHBoHK4wxxhhzaBysMMYYY8yhcbDCGGOMMYfGwQpjjDHGHBoHK4wxxhhzaBysMMYYY8yhcbDCGGOMMYfGwQpjjDHGHBoHK4wxxhhzaBysMMYYY8yhcbDCGGOMMYfWpMEKEfUjoiNEdJyI3qphuzMRLZe37yai0KYoR9rWNPhM8YEqTQWfKT5I25rWFIe5bXB9NQzXV8NxnTUM1xf7q1M1VcZEpATwGYA+AHIB7CGin4QQh+ySPQOgSAgRTkSDAEwB8ERjliNtaxre3/Y+FFDASeEEvVGP97e9DwCY0HNCYx7qtsD11TBcXw3HddYwXF+MASSEaJqMiboCeE8Icb/8/B8AIIT4yC7NBjnNTiJSAcgH4C/qKFRiYqJIT0+vdzl8pvhAb9RDQMBsNQMABAQUpIDWV4sgzyCE+YTBbDVjZ87Oq/YP8QpBW++2MJgN+P3s71dtb+fTDsGewdCb9Nh7bu9V28N9wxHoEYgyQxn25++/anukXyRaurVEcVUx/jj/x1Xbdf46tHBtgYv6izhUeOiq7XGt4uCt8UZBRQGOXDhy1faOAR3h4eyBvLI8HL90/KrtCa0T4Kp2RW5pLk4VncLRS0dhFVYQCACgIAWEEFAoFAj3Cb9q/65tukKlUOFk0UmcLT171fbubbsDAI5dPIb88vwrtikVSnRr0w0AkHUhC4UVhVdsd1I64c7gOwEAmQWZuFR56YrtLmoXJLZOBAAcOH8AJVUlV2x3d3JHp8BOAIB9eftQbiy/YruXxgsdWnUAAKSfS0elqfKK7b4uvohpGQMA2J27G0aL8Yrt/m7+2H12N8wWMwT+fMva3l89QnogokUEAOC3079dVTd/1fee7T2mJCWICFZhhVVYL38mbZKCkuCscsbp4tM4U3Lmqvz/Ku+9r/d9ffkzqVKooFaqYbaY4erkiqK/F1113rUhor1CiMR678CYA2nKbqAgADl2z3PldTWmEUKYAZQAaFE9IyJ6nojSiSi9sLCw+uY6lRnKoCTlVeutwtqgfP4qaqsXi9Vyk0vSPJgsphrX8/urdrXVDddZzWqqFyUpUW4oryE1Y7enJusGakxCiPkA5gNSy0pD9vVw9oDeqIdaqYZaqQaAy79KDo863PiFbeZsLVEq5Z9vDa6v2tVVX1uGb7mFJXNc/B5rmJWHVl5VXxZhgbuz+y0sFWM3V1O2rJwF0MbuebC8rsY0cjeQF4CLjVmIMV3GwAqr1FRvFTBbzLDCijFdxjTmYW4bXF8Nw/XVcFxnDcP1xVjTtqzsARBBRO0gBSWDAAypluYnAE8D2AngMQCb6xqvcj1sA9Bm7pqJckM53J3dMabLGB6YVguur4bh+mo4rrOG4fpirAkH2AIAET0A4FMASgALhBAfEFEagHQhxE9EpAGwBEAnAJcADBJCnKwrz4YOsGWMMcYDbFnz1qRjVoQQ6wCsq7Zugt1yFYCUpiwDY4wxxpo3voMtY4wxxhwaByuMMcYYc2gcrDDGGGPMoXGwwhhjjDGH1qRXAzUFIioEcPo6d/cDcKERi3O74/pqGK6vhuM6a5gbqa+2Qgj/xiwMYzdLswtWbgQRpfOle/XH9dUwXF8Nx3XWMFxf7K+Ku4EYY4wx5tA4WGGMMcaYQ/urBSvzb3UBmhmur4bh+mo4rrOG4fpif0l/qTErjDHGGGt+/motK4wxxhhrZjhYYYwxxphD42CFsQYgoleJ6DARLSUiZyLaRET7iegJIvqKiHS3uozNARH9SkSJ8vI6IvK+xUW6KYjodSJyvQnHCSWiIXbPhxPR3KY+LmNNpUlnXb5dEJFKCGG+1eVgDuFlAL2FELlE1AUAhBAd5W3Lb1mpmjEhxAO3ugw30esAvgWgr76BiJRCCEsjHScUwBAA3zVSfozdUs22ZUX+5ZAl/8I9TESriMiViCYQ0R4iOkhE84mI5PS/EtEs+VfwQSJKkte7EdECIvqdiPYR0d/k9cOJ6Cci2gzgP7fwVK+bXR0tIqKjcl31JqLtRHSMiJKIyJeIfiSiA0S0i4g6yPu+J9fLr0R0kohetcv3Sbm+9hPRPCJSEtFIIvrULs1zRDTzFpx2oyGiVPm9clD+RfwFgDAA64no75C+dDrL9dC+WmtBPyL6HxFlENF/5HU1vtduB3V8Hu+Tz/UP+dyda9g3m4j85OVh8nsxg4iWyOv8iehf8ud6DxHddbPP73rIr/e/5XM5SEQTAbQGsIWItshpyoloOhFlAOha02fLLt0Hcl67iKiVvL69/PwPIppMROXy4T8G0F3OZ4y8rjUR/Sx/9j+5ubXB2A0SQjTLB6RfDgLAXfLzBQDeAOBrl2YJgAHy8q8AvpSXewA4KC9/COBJedkbwFEAbgCGA8i1z6+5PeQ6MgOIgxSY7pXriQD8DcCPAOYAmCinvxfAfnn5PQA7ADhDusX3RQBqANEA1gBQy+n+CWAYAHcAJ+zW7wAQd6vr4AbqLgHAH/J7wR1AJoBOALIB+Mlp7gGw1m6fXwEkAvAHkAOgnbzet6732q0+10Z8r1X/PL4j14NWXrcYwOv2dSUvZ8vvsRi5Tmz1a6u37wDcLS+HADh8q8+3nnWSbPufIz/3sn//yOsEgMfl5Ro/W3bpbP/LPgHwjry8FsBgeflFAOW1vDeHAzgpl0EDacqSNre6jvjBj/o+mm3LiixHCLFdXv4WwN0AehHRbiL6A9KXb4xd+mUAIITYBsCTpH7yvgDeIqL9kP6BaiD9QwSAjUKIS019Ek3slBDiDyGEFdIX7n+EEALSF3EopDpbAgBCiM0AWhCRp7zvv4UQBiHEBQAFAFoBuA/SF/keuc7uAxAmhCgHsBnAQ0QUBekf7h836ySbwN0AfhBCVMjnthpA93ru2wXANiHEKQCwew/V9V67HVT/PN4H6f13VF73DaQfCrW5F8BK+f1mX2+9AcyV6+0nSJ9d98YufBP4A0AfIppCRN2FECU1pLEA+Je8XONnS95mhBSYANKPjlB5uSuAlfLytbp8/iOEKBFCVAE4BKBtw06HsVunuY9ZqX6TGAHp10iiECKHiN6D9IVQV3oCkCyEOGK/gYjuBFDRuMW9JQx2y1a751ZIr7+pnvta5PQE4BshxD9qSP8VgLcBZAFYeL0Fvo3V+F67jVT/fBUDaNEI+SoAdJG/ZJsNIcRRIroDwAMAJtu6A6upEn+OU6nrs2WSf2QAf34WG6qmzzNjzUJzb1kJIaKu8vIQAP+Vly/Iv7weq5b+CQAgorsBlMi/dDYAGE10eWxLp6YvtkP5DcBQACCiewBcEEKU1pH+PwAeI6KW8j6+RNQWAIQQuwG0gfRaLGvCMt8MvwEYKI+7cAPwiLyuPnYB6EFE7QCpjuT1t/t7rfrnMR1AKBGFy+ueArC1jv03A0ghohbAFfX2C4DRtkRE1LExC91UiKg1AL0Q4lsAUwHcAaAMgEctu9T62arDLkjdTQAwyG59XcdhrNlp7pH1EQCvENECSM2anwPwAXAQQD6APdXSVxHRPkhjL0bK694H8CmAA0SkAHAKwENNX3SH8R6ABUR0ANIVCk/XlVgIcYiI3gHwi1xfJgCvQOoDB4AVADoKIYqarshNTwjxPyJaBOB3edVXQoh9cpxxrX0Lieh5AKvlOioA0Ae3/3ut+ufxVUhfpiuJSAXp8/hFbTsLITKJ6AMAW4nIAmAfpLEWrwL4TH6PqgBsgzQ+w9HFAZhKRFZIn5OXIHXb/ExE54QQvewT1+OzVZPXAXxLROMB/AzA1tV0AIBFHri7CECz/jwy1mxvt09EoZAGkMXWM/2vAN4QQqQ3Zbn+6ohoLYCZQohmeQUVuz4N/TyyxkHSPVsqhRCCiAZBGmx721xlxphNc29ZYQ5CHqz8O4AMDlQYu2kSIA0+JkhjhEbWnZyx5qnZtqwwxhhj7K+huQ+wZYwxxthtjoMVxhhjjDk0DlYYY4wx5tA4WGGsFiTNqVT9Xj3V0wyX76fRkHxDiejgjZWuQcd7j4jeuFnHY4yxxsbBCvtLk+//cSOGQ5qcziGQhD/XjLHbCv9TYw5Jbn04TERfElEmEf1CRC7yto7yTLMHiOgHIvKR1/8qz8PyO0mzTNc4l4+c7lMiSgfwGhElENFWItpLRBuIKLCGfa6azVtudUkEsJSk2W1dastLXp8h36TrlVrK9RkRPSwv/yDfXA0kzWj9gbx8xUzQdnV1hIgWQ7ohYhsiGi/XwX8BRNod41UiOiTX3ffX89owxtjNxsEKc2QRAD4TQsRAuoeE7bbiiwH8XQjRAdJkcRPt9lEJIZIg3dnTfn11TkKIRACzIc08/ZgQIgHSbMEf1JB+rhCis3zTMxcADwkhVkG6pfxQIURHSDNc15bXQgCjhRDxdZTpN/w5WWIQAJ283B3ANiJKADACwJ2QJkt8zu6W/REA/inXlR+kW693hDQvTWe7Y7wFoJNcd83hLrCMMcbBCnNop4QQ++XlvZDmmfEC4C2EsM0xU30m39X26evIe7n8NxJALICNJM10+w6A4BrS96LaZ/O2qTEv+YZ53vJs34A8y3UNfgPQnYh0kG5Xf15umekKYAfqngn6tBBil7zcXU6nl+d5+snuGAcgtQQ9CSm4Yowxh8d3sGWOrPossS4N2OfyrLJEtBBAJwDnhBAPyNttM2oTgEwhRFfUgog0qHs278tJa8pLDlauSQhxVk7bD9L8N74AHgdQLoQou8a8RPWdIfxBSMHdAADjiShOCMFBC2PMoXHLCmtW5Jmyi+zGo1xrJl8IIUYIITraBSr2jgDwJ3m2YCJSE1H1VhNbYFLTbN72s9vWmJcQohhAMUmzfQPyLNe12AWpC2sbpJaWN/DnbM/1nQl6m5zOhYg8IAUmkAfethFCbAHwdwBeANzrKAtjjDkEbllhzdHTAL6QJ3E7CWkcx3URQhjlgbKz5S4mFaSZkTPt0hQT0ZeoeTbvRXJZKiF119SW1whIs1sLAL/UUaTfAPQVQhwnotOQWld+k8tR20zQodXO6X9EtBxABqQZn23lVUKaodcLUivQbDmQYowxh8ZzAzHGGGPMoXE3EGOMMcYcGgcrjDHGGHNoHKwwxhhjzKFxsMIYY4wxh8bBCmOMMcYcGgcrjDHGGHNoHKwwxhhjzKH9PwHEi26Ooqi7AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#load models\n",
    "model_g = Word2Vec.load(\"models/model_g\")\n",
    "\n",
    "#check that all models contain the experiment words\n",
    "words_w, words_m, [w, m] = check_w_embedding([words_w, words_m, [w, m]], model_g, True, 'genism')\n",
    "\n",
    "#female correlated words\n",
    "plot_indirect_metric(words_w, w, m, model_g, [g_gutenburg, g_gutenburg_2016, g_gutenburg_2018], 'genism', ['Original', '2016 Debiasing','2018 Debiasing'], [\"2016debias/2016debias_gutenberg_emb.txt\", \"2018debias/2018debias_gutenberg_emb.txt\"], [\"Black\", \"Red\", \"Orange\"])\n",
    "\n",
    "#male correlated words\n",
    "plot_indirect_metric(words_m, w, m, model_g, [g_gutenburg, g_gutenburg_2016, g_gutenburg_2018], 'genism', ['Original', '2016 Debiasing', '2018 Debiasing'], [\"2016debias/2016debias_gutenberg_emb.txt\", \"2018debias/2018debias_gutenberg_emb.txt\"], [\"Blue\", \"Green\", \"Black\"])\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Indirect Metric for Wikipedia Models**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "b_vals_w: [0.9177660993818735, 0.9131136826158803, 0.8933444026818028, 0.9165408283011843, 0.9113092831994486]\n",
      "b_vals_m: [0.9102605897449744, 0.9063730471878055, 0.9086640325291999, 0.9085827466805928, 0.9051157100611188]\n",
      "b_vals_w: [-0.07045461146361738, -0.07011703478791004, -0.2740862614633388, -0.04918387683733056, -0.08179905446836867]\n",
      "b_vals_m: [-0.022915297929084165, -0.02568703540829599, -0.04566430692170285, -0.013686863572700096, -0.028581457005098763]\n",
      "b_vals_w: [0.9172591527180699, 0.9133586059972365, 0.8883433758321103, 0.915970275737369, 0.9111425886653836]\n",
      "b_vals_m: [0.9095022631022676, 0.9062884057890864, 0.9073636678533832, 0.9082947761641381, 0.9046372783093519]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "b_vals_w: [0.900171212079532, 0.9112466979991007, 0.9095757466580241, 0.9201485248023188, 0.9205921959504701]\n",
      "b_vals_m: [0.8868281513751737, 0.89709598280727, 0.9006454734479988, 0.9049215264656607, 0.902964799357083]\n",
      "b_vals_w: [-0.014137726149974683, 0.002430457911871474, -0.042321342000190186, 0.022591716662459585, 0.027417030713376516]\n",
      "b_vals_m: [-0.003371409415445644, 0.0007684864639188525, -0.011372901486820148, 0.009452943736959549, 0.0020927886750363835]\n",
      "b_vals_w: [0.8999307654829759, 0.9107883376751048, 0.9090135009934405, 0.9189871492854033, 0.9200029186533365]\n",
      "b_vals_m: [0.8876519416961742, 0.896979084465571, 0.9004582296997949, 0.9051296712700476, 0.9030096144199221]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#load models\n",
    "model_g = Word2Vec.load(\"models/model_g\")\n",
    "\n",
    "#check that all models contain the experiment words\n",
    "words_w, words_m, [w, m] = check_w_embedding([words_w, words_m, [w, m]], model_g, True, 'genism')\n",
    "\n",
    "#female correlated words\n",
    "plot_indirect_metric(words_w, w, m, model_g, [g_wiki, g_wiki_2016, g_wiki_2018], 'wiki', ['Original', '2016 Debiasing','2018 Debiasing'], [\"embeddings/wikipedia_embeddings.txt\", \"2016debias/2016debias_wiki_emb.txt\", \"2018debias/2018debias_wiki_emb.txt\"], [\"Black\", \"Red\", \"Orange\"])\n",
    "\n",
    "#male correlated words\n",
    "plot_indirect_metric(words_m, w, m, model_g, [g_wiki, g_wiki_2016, g_wiki_2018], 'wiki', ['Original', '2016 Debiasing', '2018 Debiasing'], [\"embeddings/wikipedia_embeddings.txt\", \"2016debias/2016debias_wiki_emb.txt\", \"2018debias/2018debias_wiki_emb.txt\"], [\"Blue\", \"Green\", \"Black\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Indirect Metric for Gutenburg Custom Word2vec Models**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "#load models\n",
    "with open('models/model_g_custom' , 'rb') as f:\n",
    "    model_g_custom = pickle.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "b_vals_w: [-0.20505245, -0.2255866, -0.20790546, -0.25365585, -0.1843143]\n",
      "b_vals_m: [-0.16277441, -0.18208489, -0.22428097, -0.15564945, -0.1061127]\n",
      "b_vals_w: [-0.3842883229751337, -0.3389887921422542, -0.20540509358898504, -0.35261335447751935, -0.32681878450665275]\n",
      "b_vals_m: [-0.23632034285137188, -0.15476884738961855, -0.1237684625561494, -0.15598221552374722, -0.12526544594993025]\n",
      "b_vals_w: [1.1652033512323063, 0.06666434427785771, 0.4342872552110165, 0.25132817796638424, -0.11211275585634929]\n",
      "b_vals_m: [-0.4234256111029687, -0.6113675927331057, 0.17572787912212173, -0.6780142571533289, -1.0238318495085044]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "b_vals_w: [-0.35634068, -0.24419995, -0.200305, -0.23023446, -0.26424268]\n",
      "b_vals_m: [-0.2786826, -0.19388175, -0.16708761, -0.14783718, -0.24191758]\n",
      "b_vals_w: [-0.522048508860144, -0.3797493213473631, -0.2173737811038662, -0.5849750567054489, -0.38954862606155033]\n",
      "b_vals_m: [-0.27376747241391697, -0.19944493287564585, -0.15606170954776383, -0.34783819970234914, -0.28622927296647965]\n",
      "b_vals_w: [0.464295449386145, 1.4701106376635986, 0.30045382435950674, 0.06344396895896799, -0.15223863353229514]\n",
      "b_vals_m: [-1.0862970364449747, 0.23991086466174685, -0.7972430729289277, -0.7073109900837466, -0.8779247318684991]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#check that all models contain the experiment words\n",
    "words_w= ['thought', 'us', 'earth', 'life', 'children'] \n",
    "w = 'love' \n",
    "words_m = ['land', 'house', 'hand', 'think', 'let'] \n",
    "m = 'made'\n",
    "words_w, words_m, [w, m] = check_w_embedding([words_w, words_m, [w, m]], model_g_custom, True, 'custom')\n",
    "\n",
    "#female correlated words\n",
    "plot_indirect_metric(words_w, w, m, model_g_custom, [g_gutenburg_custom, g_gutenburg_2018_custom, g_gutenburg_2020_custom], 'custom', ['Original','2018 Debiasing', '2020 Debiasing'], [\"2018debias/2018debias_gutenberg_emb_custom.txt\", \"embeddings/gutenburg_custom_debiased_embeddings.txt\"], [\"Black\", \"Red\", \"Orange\"])\n",
    "\n",
    "#male correlated words\n",
    "plot_indirect_metric(words_m, w, m, model_g_custom, [g_gutenburg_custom, g_gutenburg_2018_custom, g_gutenburg_2020_custom], 'custom', ['Original', '2018 Debiasing', '2020 Debiasing'], [\"2018debias/2018debias_gutenberg_emb_custom.txt\", \"embeddings/gutenburg_custom_debiased_embeddings.txt\"], [\"Blue\", \"Green\", \"Black\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "UOv0RFFOX9NP"
   },
   "source": [
    "### 4.4 - WEAT Metric\n",
    "\n",
    "To measure WEAT, we used the formula defined in Caliskan et al., which is an additional direct bias measure. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "OfIcsZSepDE3"
   },
   "source": [
    "<img src=\"img/WEAT.png\" style=\"width:300px;float:left;\"/>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {
    "id": "QNYpB3AopDE3"
   },
   "outputs": [],
   "source": [
    "def s_word_A_B(w,A,B,model, model_type, vocab):\n",
    "    mean_A_sum = 0\n",
    "    mean_B_sum = 0\n",
    "    w_v = w_vec(w,model, model_type, vocab)\n",
    "    \n",
    "    for a in A:\n",
    "        a_vec = w_vec(a,model, model_type, vocab)\n",
    "        mean_A_sum +=cos_sim(w_v,a_vec)\n",
    "    \n",
    "    for b in B:\n",
    "        b_vec = w_vec(b,model, model_type, vocab)\n",
    "        mean_B_sum +=cos_sim(w_v,b_vec)\n",
    "        \n",
    "    return mean_A_sum/len(A) - mean_B_sum/len(B)\n",
    "\n",
    "def s_X_Y_A_B(X,Y,A,B,model, model_type, vocab={}):\n",
    "    sum_X = 0\n",
    "    sum_Y = 0\n",
    "    \n",
    "    for x in X:\n",
    "        sum_X += s_word_A_B(x,A,B,model, model_type, vocab)\n",
    "        \n",
    "    for y in Y:\n",
    "        sum_Y += s_word_A_B(y,A,B,model, model_type, vocab)\n",
    "    \n",
    "    return sum_X - sum_Y"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "e16kg8Z6pDE3"
   },
   "source": [
    "Now, lets run experiments on the models <br>\n",
    "A and B are the attribute word groups (8 words each), and X,Y are the target words"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {
    "id": "z3OwU4I6pDE3",
    "outputId": "a7c7a064-6be0-4af6-ef77-d450677946e1"
   },
   "outputs": [],
   "source": [
    "#Experiment 1 - work/family\n",
    "A = ['tie', 'manage', 'work', 'paper', 'money', 'office', 'business', 'meeting']\n",
    "B = ['home', 'parents', 'children', 'family', 'sister', 'marriage', 'charm', 'clean']\n",
    "X = ['he', 'him']\n",
    "Y = ['she', 'her'] \n",
    "\n",
    "#Experiment 2 - professions\n",
    "A_2 = ['police', 'manage', 'business', 'lawyer', 'officer', 'science', 'build', 'fire']\n",
    "B_2 = ['teacher', 'clean', 'social', 'hair', 'art', 'make', 'writer', 'library'] \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**WEAT Metric for Gutenburg Models**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "id": "yRdeOsV4pDE4"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "---------Genism-------------\n",
      "WEAT Exp 1: 0.17949908575974405\n",
      "WEAT Exp 2: 0.09589393716305494\n",
      "---------2016 Debiasing-------------\n",
      "WEAT Exp 1: 0.12776186359061337\n",
      "WEAT Exp 2: 0.05573103428929263\n",
      "---------2018 Debiasing-------------\n",
      "WEAT Exp 1: 0.15221834486977004\n",
      "WEAT Exp 2: 0.09102151140409755\n"
     ]
    }
   ],
   "source": [
    "#load models\n",
    "model_g = Word2Vec.load(\"models/model_g\")\n",
    "\n",
    "print (\"---------Genism-------------\")\n",
    "A_g,B_g,X_g,Y_g,A_2_g,B_2_g = check_w_embedding([A,B,X,Y, A_2,B_2], model_g, True, 'genism')\n",
    "print (\"WEAT Exp 1:\", s_X_Y_A_B(X_g,Y_g,A_g,B_g,model_g, 'genism'))\n",
    "print (\"WEAT Exp 2:\", s_X_Y_A_B(X_g,Y_g,A_2_g,B_2_g,model_g, 'genism'))\n",
    "\n",
    "print (\"---------2016 Debiasing-------------\")\n",
    "vocab2016 = read_debiased_embeddings(\"2016debias/2016debias_gutenberg_emb.txt\")\n",
    "A_g_2016,B_g_2016,X_g_2016,Y_g_2016,A_2_g_2016,B_2_g_2016 = check_w_embedding([A,B,X,Y, A_2,B_2], model_g, True, 'debias', vocab2016)\n",
    "print (\"WEAT Exp 1:\", s_X_Y_A_B(X_g_2016,Y_g_2016,A_g_2016,B_g_2016,model_g, 'debias', vocab2016))\n",
    "print (\"WEAT Exp 2:\", s_X_Y_A_B(X_g_2016,Y_g_2016,A_2_g_2016,B_2_g_2016,model_g, 'debias', vocab2016))\n",
    "\n",
    "print (\"---------2018 Debiasing-------------\")\n",
    "vocab2018 = read_debiased_embeddings(\"2018debias/2018debias_gutenberg_emb.txt\")\n",
    "A_g_2018,B_g_2018,X_g_2018,Y_g_2018,A_2_g_2018,B_2_g_2018 = check_w_embedding([A,B,X,Y, A_2,B_2], model_g, True, 'debias', vocab2018)\n",
    "print (\"WEAT Exp 1:\", s_X_Y_A_B(X_g_2018,Y_g_2018,A_g_2018,B_g_2018,model_g, 'debias', vocab2018))\n",
    "print (\"WEAT Exp 2:\", s_X_Y_A_B(X_g_2018,Y_g_2018,A_2_g_2018,B_2_g_2018,model_g, 'debias', vocab2018))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**WEAT Metric for Wikipedia Models**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "---------Genism-------------\n",
      "WEAT Exp 1: 0.20446737622842193\n",
      "WEAT Exp 2: 0.12759239319711924\n",
      "---------2016 Debiasing-------------\n",
      "WEAT Exp 1: 0.06078536076618002\n",
      "WEAT Exp 2: 0.028830618939882563\n",
      "---------2018 Debiasing-------------\n",
      "WEAT Exp 1: 0.11953439139491051\n",
      "WEAT Exp 2: 0.05835008333097039\n"
     ]
    }
   ],
   "source": [
    "print (\"---------Genism-------------\")\n",
    "vocab = read_debiased_embeddings(\"embeddings/wikipedia_embeddings.txt\")\n",
    "A_w,B_w,X_w,Y_w,A_2_w,B_2_w = check_w_embedding([A,B,X,Y,A_2,B_2], model_g, True, 'debias', vocab)\n",
    "print (\"WEAT Exp 1:\", s_X_Y_A_B(X_w,Y_w,A_w,B_w,model_g, 'wiki'))\n",
    "print (\"WEAT Exp 2:\", s_X_Y_A_B(X_w,Y_w,A_2_w,B_2_w,model_g, 'wiki'))\n",
    "\n",
    "print (\"---------2016 Debiasing-------------\")\n",
    "vocab2016 = read_debiased_embeddings(\"2016debias/2016debias_wiki_emb.txt\")\n",
    "A_w_2016,B_w_2016,X_w_2016,Y_w_2016,A_2_w_2016,B_2_w_2016 = check_w_embedding([A,B,X,Y, A_2,B_2], model_g, True, 'debias', vocab2016)\n",
    "print (\"WEAT Exp 1:\", s_X_Y_A_B(X_w_2016,Y_w_2016,A_w_2016,B_w_2016,model_g, 'debias', vocab2016))\n",
    "print (\"WEAT Exp 2:\", s_X_Y_A_B(X_w_2016,Y_w_2016,A_2_w_2016,B_2_w_2016,model_g, 'debias', vocab2016))\n",
    "\n",
    "print (\"---------2018 Debiasing-------------\")\n",
    "vocab2018 = read_debiased_embeddings(\"2018debias/2018debias_wiki_emb.txt\")\n",
    "A_w_2018,B_w_2018,X_w_2018,Y_w_2018,A_2_w_2018,B_2_w_2018 = check_w_embedding([A,B,X,Y,A_2,B_2], model_g, True, 'debias', vocab2018)\n",
    "print (\"WEAT Exp 1:\", s_X_Y_A_B(X_w_2018,Y_w_2018,A_w_2018,B_w,model_g, 'debias', vocab2018))\n",
    "print (\"WEAT Exp 2:\", s_X_Y_A_B(X_w_2018,Y_w_2018,A_2_w_2018,B_2_w_2018,model_g, 'debias', vocab2018))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**WEAT Metric for Gutenburg Custom Word2vec Models**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#load models\n",
    "with open('models/model_g_custom' , 'rb') as f:\n",
    "    model_g_custom = pickle.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "---------Original-------------\n",
      "WEAT Exp 3: -0.06276037381030619\n",
      "---------2018 Debiasing-------------\n",
      "WEAT Exp 3: 0.00021178337204497133\n",
      "---------2020 Debiasing-------------\n",
      "WEAT Exp 3: -0.11880533997016307\n"
     ]
    }
   ],
   "source": [
    "#Experiment 3 - custom model words\n",
    "A_3 = ['land', 'house', 'hand', 'think', 'make', 'made', 'days', 'among']\n",
    "B_3 = ['us', 'earth', 'life', 'children', 'art', 'love', 'face', 'hear'] \n",
    "    \n",
    "    \n",
    "print (\"---------Original-------------\")\n",
    "X_g,Y_g,A_3_g,B_3_g = check_w_embedding([X,Y, A_3,B_3], model_g_custom, True, 'custom')\n",
    "print (\"WEAT Exp 3:\", s_X_Y_A_B(X_g,Y_g,A_3_g,B_3_g,model_g_custom, 'custom'))\n",
    "\n",
    "print (\"---------2018 Debiasing-------------\")\n",
    "vocab2018 = read_debiased_embeddings(\"2018debias/2018debias_gutenberg_emb_custom.txt\")\n",
    "X_g_2018,Y_g_2018,A_3_g_2018,B_3_g_2018 = check_w_embedding([X,Y, A_3,B_3], model_g_custom, True, 'debias', vocab2018)\n",
    "print (\"WEAT Exp 3:\", s_X_Y_A_B(X_g_2018,Y_g_2018,A_3_g_2018,B_3_g_2018,model_g_custom, 'debias', vocab2018))\n",
    "\n",
    "print (\"---------2020 Debiasing-------------\")\n",
    "vocab2020 = read_debiased_embeddings(\"embeddings/gutenburg_custom_debiased_embeddings.txt\")\n",
    "X_g_2020,Y_g_2020,A_3_g_2020,B_3_g_2020 = check_w_embedding([X,Y, A_3,B_3], model_g_custom, True, 'debias', vocab2020)\n",
    "print (\"WEAT Exp 3:\", s_X_Y_A_B(X_g_2020,Y_g_2020,A_3_g_2020,B_3_g_2020,model_g_custom, 'debias', vocab2020))"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "Word2Vec Model and Bias Measurements.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}