[homework]simple_embeddings.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "BIWqBuEa6j0b"
   },
   "source": [
    "## Задача поиска схожих по смыслу предложений"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "NUkwMPLA6j0g"
   },
   "source": [
    "Мы будем ранжировать вопросы [StackOverflow](https://stackoverflow.com) на основе семантического векторного представления "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "uS9FwWNd5a3S"
   },
   "source": [
    "## Задача ранжирования(Learning to Rank)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "wdwY9-f75a3T"
   },
   "source": [
    "* $X$ - множество объектов\n",
    "* $X^l = \\{x_1, x_2, ..., x_l\\}$ - обучающая выборка\n",
    "<br>На обучающей выборке задан порядок между некоторыми элементами, то есть нам известно, что некий объект выборки более релевантный для нас, чем другой:\n",
    "* $i \\prec j$ - порядок пары индексов объектов на выборке $X^l$ c индексами $i$ и $j$\n",
    "### Задача:\n",
    "построить ранжирующую функцию $a$ : $X \\rightarrow R$ такую, что\n",
    "$$i \\prec j \\Rightarrow a(x_i) < a(x_j)$$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "WG2IGBsh5a3U"
   },
   "source": [
    "<img src=\"https://d25skit2l41vkl.cloudfront.net/wp-content/uploads/2016/12/Featured-Image.jpg\" width=500, height=450>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "MQk_rolFwT_h"
   },
   "source": [
    "### Embeddings"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "xUe1PGXn6j0l"
   },
   "source": [
    "Будем использовать предобученные векторные представления слов на постах Stack Overflow.<br>\n",
    "[A word2vec model trained on Stack Overflow posts](https://github.com/vefstathiou/SO_word2vec)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import math\n",
    "import re\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from tqdm.notebook import tqdm\n",
    "import matplotlib.pyplot as plt\n",
    "from IPython.display import clear_output\n",
    "from tqdm.notebook import tqdm\n",
    "from sklearn.metrics.pairwise import cosine_similarity\n",
    "from copy import deepcopy\n",
    "from gensim.models import Word2Vec\n",
    "from gensim.models.keyedvectors import KeyedVectors\n",
    "import spacy\n",
    "import pymorphy2\n",
    "from nltk.tokenize import WordPunctTokenizer\n",
    "import nltk\n",
    "from nltk.tokenize import word_tokenize\n",
    "from nltk.corpus import stopwords"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "id": "mYkI54Y-rk7a"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--2022-03-14 12:49:10--  https://zenodo.org/record/1199620/files/SO_vectors_200.bin\n",
      "Resolving zenodo.org (zenodo.org)... 137.138.76.77\n",
      "Connecting to zenodo.org (zenodo.org)|137.138.76.77|:443... connected.\n",
      "HTTP request sent, awaiting response... 200 OK\n",
      "Length: 1453905423 (1.4G) [application/octet-stream]\n",
      "Saving to: ‘SO_vectors_200.bin’\n",
      "\n",
      "SO_vectors_200.bin  100%[===================>]   1.35G  6.78MB/s    in 4m 46s  \n",
      "\n",
      "2022-03-14 12:53:58 (4.85 MB/s) - ‘SO_vectors_200.bin’ saved [1453905423/1453905423]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "!wget https://zenodo.org/record/1199620/files/SO_vectors_200.bin"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "id": "O8YJTOYv6j0s"
   },
   "outputs": [],
   "source": [
    "wv_embeddings = KeyedVectors.load_word2vec_format(\"SO_vectors_200.bin\", binary=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "aIcT_g-C6j1E"
   },
   "source": [
    "#### Как пользоваться этими векторами?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "DWO5SPDY6j1G"
   },
   "source": [
    "Посмотрим на примере одного слова, что из себя представляет embedding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "id": "KeSBlQfk6j1J",
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "float32 (200,)\n"
     ]
    }
   ],
   "source": [
    "my_dog = 'dog'\n",
    "if my_dog in wv_embeddings:\n",
    "    print(wv_embeddings[my_dog].dtype, wv_embeddings[my_dog].shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "id": "T4Eq-D1qxpMJ"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Num of words: 1787145\n"
     ]
    }
   ],
   "source": [
    "print(f\"Num of words: {len(wv_embeddings.index_to_key)}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ZT6NTCys6j1Q"
   },
   "source": [
    "Найдем наиболее близкие слова к слову `dog`:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "n08z2PjMwC5o"
   },
   "source": [
    "#### Вопрос 1:\n",
    "* Входит ли слов `cat` топ-5 близких слов к слову `dog`? Какое место? "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Слово \"cat\" НЕ входит в топ-5 близких к слову \"dog\"\n",
      "Однако, на 4 позиции в топ-5 находится слово \"cats\"\n"
     ]
    }
   ],
   "source": [
    "MY_DOG = 'dog'\n",
    "MY_CAT = 'cat'\n",
    "TOP = 5\n",
    "COUNT = 0\n",
    "idxs = []\n",
    "\n",
    "for word in wv_embeddings.most_similar(positive=[MY_DOG], topn = TOP):\n",
    "    if MY_CAT == word[0]:\n",
    "        print(f'Слово {MY_CAT} входит в топ-{TOP} к слову {MY_DOG} и находится на \\\n",
    "{wv_embeddings.most_similar(positive=[MY_DOG], topn = TOP).index(word) + 1} месте')\n",
    "        break\n",
    "    elif MY_CAT in word[0]:\n",
    "        COUNT += 1\n",
    "        idxs.append(wv_embeddings.most_similar(positive=[MY_DOG], topn = TOP).index(word) + 1)\n",
    "    else:\n",
    "        COUNT += 1\n",
    "\n",
    "if COUNT == TOP:\n",
    "    print(f'Слово \"{MY_CAT}\" НЕ входит в топ-{TOP} близких к слову \"{MY_DOG}\"')\n",
    "if len(idxs) != 0:\n",
    "    print(f'Однако, на {idxs[0]} позиции в топ-{TOP} находится слово \\\n",
    "\"{wv_embeddings.most_similar(positive=[MY_DOG], topn = TOP)[idxs[0]-1][0]}\"')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ai48-5vv6j1d"
   },
   "source": [
    "### Векторные представления текста\n",
    "\n",
    "Перейдем от векторных представлений отдельных слов к векторным представлениям вопросов, как к **среднему** векторов всех слов в вопросе. Если для какого-то слова нет предобученного вектора, то его нужно пропустить. Если вопрос не содержит ни одного известного слова, то нужно вернуть нулевой вектор."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "id": "EhNuxBJd6j1f"
   },
   "outputs": [],
   "source": [
    "# Токенайзер, основанный на регулярных выражения, возвращающий \n",
    "# слова без знаков пунктуации и без применения нормализации.\n",
    "\n",
    "class MyTokenizer:\n",
    "    def __init__(self):\n",
    "        pass\n",
    "    def tokenize(self, text):\n",
    "        return re.findall('\\w+', text)\n",
    "    \n",
    "tokenizer = MyTokenizer()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "id": "YHcvu6186j1m"
   },
   "outputs": [],
   "source": [
    "def question_to_vec(question, embeddings, tokenizer, dim=200):\n",
    "    \"\"\"\n",
    "        question: строка\n",
    "        embeddings: наше векторное представление\n",
    "        dim: размер любого вектора в нашем представлении\n",
    "        \n",
    "        return: векторное представление для вопроса\n",
    "    \"\"\"\n",
    "    embed_list = []\n",
    "    for word in tokenizer.tokenize(question.lower()):\n",
    "        try:\n",
    "            embed_list.append(embeddings[word])\n",
    "        except (KeyError):\n",
    "            continue\n",
    "    return np.mean(embed_list, axis=0) if len(embed_list) > 0 else np.zeros(dim, dtype='float32')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "u5Q_4j7r6j1u"
   },
   "source": [
    "Теперь у нас есть метод для создания векторного представления любого предложения."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "EsJSNkhm6j1y"
   },
   "source": [
    "#### Вопрос 2:\n",
    "* Какая третья(с индексом 2) компонента вектора предложения `I love neural networks` (округлите до 2 знаков после запятой)?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "text = 'I love neural networks'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 243,
   "metadata": {
    "id": "a62r11cT6j10",
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-1.29"
      ]
     },
     "execution_count": 243,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Токенайзер на регулярных выражениях\n",
    "tokenizer = MyTokenizer()\n",
    "round(question_to_vec(text, wv_embeddings, tokenizer)[2],2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 244,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['I', 'love', 'neural', 'networks']"
      ]
     },
     "execution_count": 244,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "MyTokenizer().tokenize(text) #результаты токенизации классом MyTokenizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-1.41"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Токенайзер Spacy помимо слов токенизирует знаки пунктуации и применяет нормализацию (лемматизацию)\n",
    "nlp = spacy.load('en_core_web_sm')\n",
    "\n",
    "class SpacyTokenizer:\n",
    "    def __init__(self):\n",
    "        pass\n",
    "    def tokenize(self, text):\n",
    "        spacy_results = nlp(text)\n",
    "        return [token.lemma_ for token in spacy_results]\n",
    "    \n",
    "spacy_tokenizer = SpacyTokenizer()\n",
    "round(question_to_vec(text, wv_embeddings, spacy_tokenizer)[2],2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 246,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['I', 'love', 'neural', 'network']"
      ]
     },
     "execution_count": 246,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "SpacyTokenizer().tokenize(text) #результаты токенизации классом SpacyTokenizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-1.29"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Токенайзер pymorphy2\n",
    "class Pymorphy2Tokenizer:\n",
    "    def __init__(self):\n",
    "        pass\n",
    "    def tokenize(self, text):\n",
    "        morph = pymorphy2.MorphAnalyzer()\n",
    "        pymorphy_results = list(map(lambda x: morph.parse(x), text.split(' ')))\n",
    "        return [res[0].normal_form for res in pymorphy_results] \n",
    "    \n",
    "pymorphy2_tokenizer = Pymorphy2Tokenizer()\n",
    "round(question_to_vec(text, wv_embeddings, pymorphy2_tokenizer)[2],2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['i', 'love', 'neural', 'networks']"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Pymorphy2Tokenizer().tokenize(text) #результаты токенизации классом Pymorphy2Tokenizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-1.29"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Токенайзер WordPunctTokenizer помимо слов токенизирует знаки пунктуации, но НЕ применяет нормализацию.\n",
    "tk = WordPunctTokenizer()\n",
    "round(question_to_vec(text, wv_embeddings, tk)[2],2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['I', 'love', 'neural', 'networks']"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tk.tokenize(text) #результаты токенизации классом WordPunctTokenizer"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Y60z4t6W6j16"
   },
   "source": [
    "### Оценка близости текстов\n",
    "\n",
    "Представим, что мы используем идеальные векторные представления слов. Тогда косинусное расстояние между дублирующими предложениями должно быть меньше, чем между случайно взятыми предложениями. \n",
    "\n",
    "Сгенерируем для каждого из $N$ вопросов $R$ случайных отрицательных примеров и примешаем к ним также настоящие дубликаты. Для каждого вопроса будем ранжировать с помощью нашей модели $R + 1$ примеров и смотреть на позицию дубликата. Мы хотим, чтобы дубликат был первым в ранжированном списке.\n",
    "\n",
    "#### Hits@K\n",
    "Первой простой метрикой будет количество корректных попаданий для какого-то $K$:\n",
    "$$ \\text{Hits@K} = \\frac{1}{N}\\sum_{i=1}^N \\, [rank\\_q_i^{'} \\le K],$$\n",
    "* $\\begin{equation*}\n",
    "[x < 0 ] \\equiv \n",
    " \\begin{cases}\n",
    "   1, &x < 0\\\\\n",
    "   0, &x \\geq 0\n",
    " \\end{cases}\n",
    "\\end{equation*}$ - индикаторная функция\n",
    "* $q_i$ - $i$-ый вопрос\n",
    "* $q_i^{'}$ - его дубликат\n",
    "* $rank\\_q_i^{'}$ - позиция дубликата в ранжированном списке ближайших предложений для вопроса $q_i$.\n",
    "\n",
    "#### DCG@K\n",
    "Второй метрикой будет упрощенная DCG метрика, учитывающая порядок элементов в списке путем домножения релевантности элемента на вес равный обратному логарифму номера позиции::\n",
    "$$ \\text{DCG@K} = \\frac{1}{N} \\sum_{i=1}^N\\frac{1}{\\log_2(1+rank\\_q_i^{'})}\\cdot[rank\\_q_i^{'} \\le K],$$\n",
    "С такой метрикой модель штрафуется за большой ранк корректного ответа"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "eHCnH-jw6j18"
   },
   "source": [
    "#### Вопрос 3:\n",
    "* Максимум `Hits@47 - DCG@1`?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ответ: Максимальное значение РАЗНИЦЫ двух функций достигается при максимальном значении Уменьшаемого и минимальном значении Вычитаемого. Максимально возможное значение Hits@47 == 1, в том случае, когда дубликаты вопросов расположены среди первых 47 вариантов ответов, в число которых могут входить и отрицательные примеры. При этом минимальное значение DCG@1 == 0, в том случае, когда позиции дубликатов в ранжированном списке ближайших вопросов для вопроса 𝑞𝑖 не расположены на 1-ом месте. Соответственно, максимум Hits@47 - DCG@1 = 1 - 0 = 1. Что характерно для ситуации, когда дубликаты расположены на позициях с 2 по 47."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Hits@47 - DCG@1 = 1.0\n",
      "\n",
      "Hits@47 = 1.0, DCG@1 = 0.0\n"
     ]
    }
   ],
   "source": [
    "# Проверим наши выводы численными методами на искусственном примере\n",
    "from ampligraph.evaluation.metrics import hits_at_n_score\n",
    "\n",
    "# пусть N = 8 - количество исходных вопросов 𝑞𝑖 \n",
    "# n (aka K) - длина рассматриваемого списка любых вариантов ответов перемешанные отрицательные примеры и дубликаты\n",
    "K1 = 47\n",
    "K2 = 1\n",
    "rankings = np.array([3, 12, 6, 14, 35, 16, 42, 47]) # 𝑟𝑎𝑛𝑘_𝑞′ - позиции дубликатов в ранжированном списке ближайших вопросов для вопроса 𝑞𝑖\n",
    "\n",
    "Hits = hits_at_n_score(rankings, n = K1)\n",
    "DCG = round(sum([np.zeros(1)[0] if rank > K2 else math.log2(1 + rank)**(-1) for rank in rankings]) / rankings.shape[0],2)\n",
    "\n",
    "print(f'\\nHits@{K1} - DCG@{K2} = {Hits - DCG}\\n\\nHits@{K1} = {Hits}, DCG@{K2} = {DCG}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "_tFemBkP6j1-"
   },
   "source": [
    "<img src='https://hsto.org/files/1c5/edf/dee/1c5edfdeebce4b71a86bdf986d9f88f2.jpg' width=400, height=200>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "0sUSxk866j1_"
   },
   "source": [
    "#### Пример оценок\n",
    "\n",
    "Вычислим описанные выше метрики для игрушечного примера. \n",
    "Пусть\n",
    "* $N = 1$, $R = 3$\n",
    "* <font color='green'>\"Что такое python?\"</font> - вопрос $q_1$\n",
    "* <font color='red'>\"Что такое язык python?\"</font> - его дубликат $q_i^{'}$\n",
    "\n",
    "Пусть модель выдала следующий ранжированный список кандидатов:\n",
    "\n",
    "1. \"Как изучить с++?\"\n",
    "2. <font color='red'>\"Что такое язык python?\"</font>\n",
    "3. \"Хочу учить Java\"\n",
    "4. \"Не понимаю Tensorflow\"\n",
    "\n",
    "$\\Rightarrow rank\\_q_i^{'} = 2$\n",
    "\n",
    "Вычислим метрику *Hits@K* для *K = 1, 4*:\n",
    "\n",
    "- [K = 1] $\\text{Hits@1} =  [rank\\_q_i^{'} \\le 1)] = 0$\n",
    "- [K = 4] $\\text{Hits@4} =  [rank\\_q_i^{'} \\le 4] = 1$\n",
    "\n",
    "Вычислим метрику *DCG@K* для *K = 1, 4*:\n",
    "- [K = 1] $\\text{DCG@1} = \\frac{1}{\\log_2(1+2)}\\cdot[2 \\le 1] = 0$\n",
    "- [K = 4] $\\text{DCG@4} = \\frac{1}{\\log_2(1+2)}\\cdot[2 \\le 4] = \\frac{1}{\\log_2{3}}$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "B4L6HJJC6j2B"
   },
   "source": [
    "#### Вопрос 4:\n",
    "* Вычислите `DCG@10`, если $rank\\_q_i^{'} = 9$(округлите до одного знака после запятой)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "DCG@10 = 0.3\n",
      "\n"
     ]
    }
   ],
   "source": [
    "K = 10\n",
    "\n",
    "rankings = np.array([9]) # 𝑟𝑎𝑛𝑘_𝑞′ - позиции дубликатов в ранжированном списке ближайших вопросов для вопроса 𝑞𝑖\n",
    "\n",
    "DCG = round(sum([np.zeros(1)[0] if rank > K else math.log2(1 + rank)**(-1) for rank in rankings]) / rankings.shape[0],2)\n",
    "\n",
    "print(f'\\nDCG@{K} = {DCG}\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "J5xWOORI6j2F"
   },
   "source": [
    "### HITS\\_COUNT и DCG\\_SCORE"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "I1q9WQOx6j2H"
   },
   "source": [
    "Каждая функция имеет два аргумента: $dup\\_ranks$ и $k$. $dup\\_ranks$ является списком, который содержит рейтинги дубликатов(их позиции в ранжированном списке). Например, $dup\\_ranks = [2]$ для примера, описанного выше."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "id": "F5VwySUB6j2J"
   },
   "outputs": [],
   "source": [
    "def hits_count(dup_ranks, k):\n",
    "    \"\"\"\n",
    "        dup_ranks: list индексов дубликатов\n",
    "        result: вернуть  Hits@k\n",
    "    \"\"\"\n",
    "    hits_value = round(sum([np.zeros(1)[0] if rank > k else np.ones(1)[0] for rank in dup_ranks]) / len(dup_ranks),3)\n",
    "    return hits_value"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "id": "82hQaxCH6j2R"
   },
   "outputs": [],
   "source": [
    "def dcg_score(dup_ranks, k):\n",
    "    \"\"\"\n",
    "        dup_ranks: list индексов дубликатов\n",
    "        result: вернуть DCG@k\n",
    "    \"\"\"\n",
    "    dcg_value = round(sum([np.zeros(1)[0] if rank > k else math.log2(1 + rank)**(-1) for rank in dup_ranks]) / len(dup_ranks),3)\n",
    "    return dcg_value"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "PcwHeXN26j2Y"
   },
   "source": [
    "Протестируем функции. Пусть $N = 1$, то есть один эксперимент. Будем искать копию вопроса и оценивать метрики."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "id": "gLa_Wqfh6j2m"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ваш ответ HIT: [0.0, 1.0, 1.0, 1.0]\n",
      "Ваш ответ DCG: [0.0, 0.631, 0.631, 0.631]\n"
     ]
    }
   ],
   "source": [
    "copy_answers = [\"How does the catch keyword determine the type of exception that was thrown\",]\n",
    "\n",
    "# наши кандидаты\n",
    "candidates_ranking = [[\"How Can I Make These Links Rotate in PHP\",\n",
    "                       \"How does the catch keyword determine the type of exception that was thrown\",\n",
    "                       \"NSLog array description not memory address\",\n",
    "                       \"PECL_HTTP not recognised php ubuntu\"],]\n",
    "# dup_ranks — позиции наших копий, так как эксперимент один, то этот массив длины 1\n",
    "dup_ranks = [2]\n",
    "\n",
    "# вычисляем метрику для разных k\n",
    "print('Ваш ответ HIT:', [hits_count(dup_ranks, k) for k in range(1, 5)])\n",
    "print('Ваш ответ DCG:', [round(dcg_score(dup_ranks, k), 5) for k in range(1, 5)])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "MoHC3YoQ6j2t"
   },
   "source": [
    "У вас должно получиться"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "id": "B0NFWq4f6j2u",
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>HITS</th>\n",
       "      <td>0</td>\n",
       "      <td>1.00000</td>\n",
       "      <td>1.00000</td>\n",
       "      <td>1.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>DCG</th>\n",
       "      <td>0</td>\n",
       "      <td>0.63093</td>\n",
       "      <td>0.63093</td>\n",
       "      <td>0.63093</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      1        2        3        4\n",
       "HITS  0  1.00000  1.00000  1.00000\n",
       "DCG   0  0.63093  0.63093  0.63093"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# correct_answers - метрика для разных k\n",
    "correct_answers = pd.DataFrame([[0, 1, 1, 1], [0, 1 / (np.log2(3)), 1 / (np.log2(3)), 1 / (np.log2(3))]],\n",
    "                               index=['HITS', 'DCG'], columns=range(1,5))\n",
    "correct_answers"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "tHZqgDTo6j0i"
   },
   "source": [
    "### Данные\n",
    "[arxiv link](https://drive.google.com/file/d/1QqT4D0EoqJTy7v9VrNCYD-m964XZFR7_/edit)\n",
    "\n",
    "`train.tsv` - выборка для обучения.<br> В каждой строке через табуляцию записаны: **<вопрос>, <похожий вопрос>**\n",
    "\n",
    "`validation.tsv` - тестовая выборка.<br> В каждой строке через табуляцию записаны: **<вопрос>, <похожий вопрос>, <отрицательный пример 1>, <отрицательный пример 2>, ...**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 28,
     "status": "ok",
     "timestamp": 1628256058346,
     "user": {
      "displayName": "Deep Learning School",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14GhNf0RkP5WvkU5MixKfC1Sv3mb-9QWgAbC6VcfQvA=s64",
      "userId": "16549096980415837553"
     },
     "user_tz": -180
    },
    "id": "jKVK2lDGvrIe",
    "outputId": "51944c9b-d6e8-41af-bec4-bb35fba5d51b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "unzip:  cannot find or open stackoverflow_similar_questions.zip, stackoverflow_similar_questions.zip.zip or stackoverflow_similar_questions.zip.ZIP.\r\n"
     ]
    }
   ],
   "source": [
    "!unzip stackoverflow_similar_questions.zip"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "hil2UsUG6j22"
   },
   "source": [
    "Считайте данные."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "id": "B4EBho8s6j26"
   },
   "outputs": [],
   "source": [
    "def read_corpus(filename):\n",
    "    data = []\n",
    "    for line in open(filename, encoding='utf-8'):\n",
    "        data.append(line.split('\\t'))\n",
    "    return data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "kkTxY3Mk9_nG"
   },
   "source": [
    "Нам понадобиться только файл validation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 274
    },
    "executionInfo": {
     "elapsed": 21,
     "status": "error",
     "timestamp": 1628256058355,
     "user": {
      "displayName": "Deep Learning School",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14GhNf0RkP5WvkU5MixKfC1Sv3mb-9QWgAbC6VcfQvA=s64",
      "userId": "16549096980415837553"
     },
     "user_tz": -180
    },
    "id": "PTVB9Tnp6j29",
    "outputId": "9c55c802-3d82-471d-eab2-195dabf5026c"
   },
   "outputs": [],
   "source": [
    "validation_data = read_corpus('./data/validation.tsv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "bTHfL-9y6j3F"
   },
   "source": [
    "Кол-во строк"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "id": "z6ubXhIe6j3H",
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3760"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(validation_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "kaOQblBy6j3M"
   },
   "source": [
    "Размер нескольких первых строк"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "id": "yRx6e-Pe6j3M"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 1001\n",
      "2 1001\n",
      "3 1001\n",
      "4 1001\n",
      "5 1001\n"
     ]
    }
   ],
   "source": [
    "for i in range(5):\n",
    "    print(i + 1, len(validation_data[i]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ySQQp0oQt1Ep"
   },
   "source": [
    "### Ранжирование без обучения"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "iElEDhj-6j3R"
   },
   "source": [
    "Реализуйте функцию ранжирования кандидатов на основе косинусного расстояния. Функция должна по списку кандидатов вернуть отсортированный список пар (позиция в исходном списке кандидатов, кандидат). При этом позиция кандидата в полученном списке является его рейтингом (первый - лучший). Например, если исходный список кандидатов был [a, b, c], и самый похожий на исходный вопрос среди них - c, затем a, и в конце b, то функция должна вернуть список **[(2, c), (0, a), (1, b)]**."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "id": "1yP8wJWj6j3X"
   },
   "outputs": [],
   "source": [
    "def rank_candidates(question, candidates, embeddings, tokenizer, dim=200):\n",
    "    \"\"\"\n",
    "        question: строка\n",
    "        candidates: массив строк(кандидатов) [a, b, c]\n",
    "        result: пары (начальная позиция, кандидат) [(2, c), (0, a), (1, b)]\n",
    "    \"\"\"\n",
    "    result = []\n",
    "    q_vec = question_to_vec(question, embeddings, tokenizer)\n",
    "    \n",
    "    for idx, elem in enumerate(candidates):\n",
    "        candidate_vec = question_to_vec(elem, embeddings, tokenizer)\n",
    "        result.append((idx, elem, cosine_similarity(q_vec.reshape(1, -1), candidate_vec.reshape(1, -1))[0][0]))\n",
    "        \n",
    "    result.sort(key=lambda x: x[2], reverse=True)\n",
    "    \n",
    "    return [elem[:-1] for elem in deepcopy(result)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "TnBszTb76j3c"
   },
   "source": [
    "Протестируйте работу функции на примерах ниже. Пусть $N=2$, то есть два эксперимента"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "id": "xvQgtP176j3h"
   },
   "outputs": [],
   "source": [
    "questions = ['converting string to list', 'Sending array via Ajax fails'] \n",
    "\n",
    "candidates = [['Convert Google results object (pure js) to Python object', # первый эксперимент\n",
    "               'C# create cookie from string and send it',\n",
    "               'How to use jQuery AJAX for an outside domain?'],\n",
    "              \n",
    "              ['Getting all list items of an unordered list in PHP',      # второй эксперимент\n",
    "               'WPF- How to update the changes in list item of a list',\n",
    "               'select2 not displaying search results']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 328,
   "metadata": {
    "id": "bPj1JGFi6j3m"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Tokenizer</th>\n",
       "      <th>Эксперимент_1</th>\n",
       "      <th>Эксперимент_2</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>MyTokenizer</td>\n",
       "      <td>102</td>\n",
       "      <td>021</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>SpacyTokenizer</td>\n",
       "      <td>102</td>\n",
       "      <td>021</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Pymorphy2Tokenizer</td>\n",
       "      <td>012</td>\n",
       "      <td>012</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>WordPunctTokenizer</td>\n",
       "      <td>102</td>\n",
       "      <td>021</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            Tokenizer Эксперимент_1 Эксперимент_2\n",
       "0         MyTokenizer           102           021\n",
       "1      SpacyTokenizer           102           021\n",
       "2  Pymorphy2Tokenizer           012           012\n",
       "3  WordPunctTokenizer           102           021"
      ]
     },
     "execution_count": 328,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tokenizers = [MyTokenizer(), SpacyTokenizer(), Pymorphy2Tokenizer(), WordPunctTokenizer()]\n",
    "rez_list = []\n",
    "\n",
    "for tokenizer in tokenizers:\n",
    "    for question, q_candidates in zip(questions, candidates):\n",
    "        ranks = rank_candidates(question, q_candidates, wv_embeddings, tokenizer)\n",
    "        rez_list.append(\"\".join([str(i) for i in [elem[0] for elem in ranks]]))\n",
    "        #print(f'{ranks}')\n",
    "        #print()\n",
    "\n",
    "df = pd.DataFrame()\n",
    "df['Tokenizer'] = [str(i).split('.')[1].split(' ')[0] for i in tokenizers][:-1] + [[str(i).partition('(')[0] for i in tokenizers][-1]]\n",
    "df['Эксперимент_1'] = rez_list[::2]\n",
    "df['Эксперимент_2'] = rez_list[1::2]\n",
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "jm4cidj56j3q"
   },
   "source": [
    "Для первого экперимента вы можете полностью сравнить ваши ответы и правильные ответы. Но для второго эксперимента два ответа на кандидаты будут <b>скрыты</b>(*)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "0LeKMIsn6j3s"
   },
   "outputs": [],
   "source": [
    "# должно вывести\n",
    "results = [[(1, 'C# create cookie from string and send it'),\n",
    "            (0, 'Convert Google results object (pure js) to Python object'),\n",
    "            (2, 'How to use jQuery AJAX for an outside domain?')],\n",
    "           [(*, 'Getting all list items of an unordered list in PHP'), #скрыт\n",
    "            (*, 'select2 not displaying search results'), #скрыт\n",
    "            (*, 'WPF- How to update the changes in list item of a list')]] #скрыт"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "t1ttnIBe6j3x"
   },
   "source": [
    "Последовательность начальных индексов вы должны получить `для эксперимента 1`  1, 0, 2."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "5WQgYDWd6j3y"
   },
   "source": [
    "#### Вопрос 5:\n",
    "* Какую последовательность начальных индексов вы получили `для эксперимента 2`(перечисление без запятой и пробелов, например, `102` для первого эксперимента?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 329,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[1m 021\n"
     ]
    }
   ],
   "source": [
    "# для токенайзеров - MyTokenizer(), SpacyTokenizer(), WordPunctTokenizer()\n",
    "rez_2 = ''\n",
    "for rank in ranks:\n",
    "    rez_2 += str(rank[0])\n",
    "print('\\033[1m',rez_2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "fPllOY-Y6j30"
   },
   "source": [
    "Теперь мы можем оценить качество нашего метода. Запустите следующие два блока кода для получения результата. Обратите внимание, что вычисление расстояния между векторами занимает некоторое время (примерно 10 минут). Можете взять для validation 1000 примеров."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 330,
   "metadata": {
    "id": "nu7K4mis6j32"
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "30e232d224b2485c886b42088f54e505",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/3760 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "tokenizer = MyTokenizer()\n",
    "\n",
    "wv_ranking = []\n",
    "max_validation_examples = 1000\n",
    "for i, line in enumerate(tqdm(validation_data)):\n",
    "    if i == max_validation_examples:\n",
    "        break\n",
    "    q, *ex = line\n",
    "    ranks = rank_candidates(q, ex, wv_embeddings, tokenizer)\n",
    "    wv_ranking.append([r[0] for r in ranks].index(0) + 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 331,
   "metadata": {
    "id": "gDtS520v6j35",
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4b101d1bfd894a76bba939b28214b660",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.410 | Hits@   1: 0.410\n",
      "DCG@   5: 0.500 | Hits@   5: 0.580\n",
      "DCG@  10: 0.521 | Hits@  10: 0.645\n",
      "DCG@ 100: 0.568 | Hits@ 100: 0.875\n",
      "DCG@ 500: 0.581 | Hits@ 500: 0.973\n",
      "DCG@1000: 0.583 | Hits@1000: 1.000\n"
     ]
    }
   ],
   "source": [
    "for k in tqdm([1, 5, 10, 100, 500, 1000]):\n",
    "    print(\"DCG@%4d: %.3f | Hits@%4d: %.3f\" % (k, dcg_score(wv_ranking, k), k, hits_count(wv_ranking, k)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 332,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0aab7d285e234affb21f79e61687be35",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/3760 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "tokenizer = SpacyTokenizer()\n",
    "\n",
    "wv_ranking = []\n",
    "max_validation_examples = 1000\n",
    "for i, line in enumerate(tqdm(validation_data)):\n",
    "    if i == max_validation_examples:\n",
    "        break\n",
    "    q, *ex = line\n",
    "    ranks = rank_candidates(q, ex, wv_embeddings, tokenizer)\n",
    "    wv_ranking.append([r[0] for r in ranks].index(0) + 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 333,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "22281ff9d2d541fa9d4923827e1919b1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.392 | Hits@   1: 0.392\n",
      "DCG@   5: 0.482 | Hits@   5: 0.561\n",
      "DCG@  10: 0.503 | Hits@  10: 0.627\n",
      "DCG@ 100: 0.549 | Hits@ 100: 0.850\n",
      "DCG@ 500: 0.565 | Hits@ 500: 0.967\n",
      "DCG@1000: 0.568 | Hits@1000: 1.000\n"
     ]
    }
   ],
   "source": [
    "for k in tqdm([1, 5, 10, 100, 500, 1000]):\n",
    "    print(\"DCG@%4d: %.3f | Hits@%4d: %.3f\" % (k, dcg_score(wv_ranking, k), k, hits_count(wv_ranking, k)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1000\n"
     ]
    }
   ],
   "source": [
    "tokenizer = WordPunctTokenizer()\n",
    "\n",
    "wv_ranking = []\n",
    "max_validation_examples = 1000\n",
    "for i, line in enumerate(tqdm(validation_data)):\n",
    "    print(i)\n",
    "    clear_output(wait=True)\n",
    "    if i == max_validation_examples:\n",
    "        break\n",
    "    q, *ex = line\n",
    "    ranks = rank_candidates(q, ex, wv_embeddings, tokenizer)\n",
    "    wv_ranking.append([r[0] for r in ranks].index(0) + 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "fd70a0ba37af421496fc57d88b63ebf0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.410 | Hits@   1: 0.410\n",
      "DCG@   5: 0.500 | Hits@   5: 0.580\n",
      "DCG@  10: 0.521 | Hits@  10: 0.645\n",
      "DCG@ 100: 0.568 | Hits@ 100: 0.875\n",
      "DCG@ 500: 0.581 | Hits@ 500: 0.973\n",
      "DCG@1000: 0.583 | Hits@1000: 1.000\n"
     ]
    }
   ],
   "source": [
    "for k in tqdm([1, 5, 10, 100, 500, 1000]):\n",
    "    print(\"DCG@%4d: %.3f | Hits@%4d: %.3f\" % (k, dcg_score(wv_ranking, k), k, hits_count(wv_ranking, k)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "LL6_Rjg3InL8"
   },
   "source": [
    "### Эмбеддинги, обученные на корпусе похожих вопросов"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "id": "iNvbpR5gJIPz"
   },
   "outputs": [],
   "source": [
    "train_data = read_corpus('./data/train.tsv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Nr281ZyEJfjT"
   },
   "source": [
    "Улучшите качество модели.<br>Склеим вопросы в пары и обучим на них модель Word2Vec из gensim. Выберите размер window. Объясните свой выбор."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "import nltk\n",
    "from nltk.tokenize import word_tokenize\n",
    "from nltk.corpus import stopwords\n",
    "stopWords = set(stopwords.words('english'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Эмбеддинги без нормализации (лемматизации)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Подготовим препроцессинг данных для обучения эмбеддингов. Используем слова в нижнем регистре, со знаками пунктуации и без использования нормализации. Кроме того, отфильтруем слова согласно стоп-листа."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "def preproc_nltk(text):\n",
    "    return ' '.join([word for word in word_tokenize(text.lower()) if word not in stopWords])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "id": "f6Y46SSQMTL0"
   },
   "outputs": [],
   "source": [
    "couple_list = [couple[0] + str(' ') + couple[1] for couple in train_data]\n",
    "words = [preproc_nltk(question).split() for question in couple_list]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### with Skip-Gram"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "id": "QuJzAM0cI-UH"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "min_count = 1, window = 1 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a1debbbbf4d64dbf98a244fbc061f50c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.412 | Hits@   1: 0.412\n",
      "DCG@   5: 0.503 | Hits@   5: 0.584\n",
      "DCG@  10: 0.522 | Hits@  10: 0.642\n",
      "DCG@ 100: 0.563 | Hits@ 100: 0.842\n",
      "DCG@ 500: 0.578 | Hits@ 500: 0.956\n",
      "DCG@1000: 0.582 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 1, window = 2 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e2bc8ee15f9f454ead37cc159bdf1d7c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.446 | Hits@   1: 0.446\n",
      "DCG@   5: 0.530 | Hits@   5: 0.604\n",
      "DCG@  10: 0.553 | Hits@  10: 0.673\n",
      "DCG@ 100: 0.592 | Hits@ 100: 0.861\n",
      "DCG@ 500: 0.606 | Hits@ 500: 0.967\n",
      "DCG@1000: 0.609 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 1, window = 3 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "83b1179ffe7947429da31e4c19cec27a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.463 | Hits@   1: 0.463\n",
      "DCG@   5: 0.549 | Hits@   5: 0.623\n",
      "DCG@  10: 0.568 | Hits@  10: 0.680\n",
      "DCG@ 100: 0.608 | Hits@ 100: 0.870\n",
      "DCG@ 500: 0.621 | Hits@ 500: 0.970\n",
      "DCG@1000: 0.624 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 1, window = 4 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6b27d1936e024876a011178e78ccc16f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.482 | Hits@   1: 0.482\n",
      "DCG@   5: 0.559 | Hits@   5: 0.628\n",
      "DCG@  10: 0.582 | Hits@  10: 0.700\n",
      "DCG@ 100: 0.621 | Hits@ 100: 0.887\n",
      "DCG@ 500: 0.632 | Hits@ 500: 0.973\n",
      "DCG@1000: 0.635 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 1, window = 5 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c93a93b85d3244a48073ef8922624b50",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.494 | Hits@   1: 0.494\n",
      "DCG@   5: 0.575 | Hits@   5: 0.645\n",
      "DCG@  10: 0.593 | Hits@  10: 0.701\n",
      "DCG@ 100: 0.633 | Hits@ 100: 0.893\n",
      "DCG@ 500: 0.644 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.646 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 1, window = 6 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "99ac9ea5b9774ceba320ea6b61b144f4",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.498 | Hits@   1: 0.498\n",
      "DCG@   5: 0.577 | Hits@   5: 0.646\n",
      "DCG@  10: 0.597 | Hits@  10: 0.709\n",
      "DCG@ 100: 0.636 | Hits@ 100: 0.898\n",
      "DCG@ 500: 0.647 | Hits@ 500: 0.975\n",
      "DCG@1000: 0.649 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 1, window = 7 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "04f48d222b4f41d9b699542ee660def1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.509 | Hits@   1: 0.509\n",
      "DCG@   5: 0.587 | Hits@   5: 0.656\n",
      "DCG@  10: 0.608 | Hits@  10: 0.724\n",
      "DCG@ 100: 0.644 | Hits@ 100: 0.896\n",
      "DCG@ 500: 0.655 | Hits@ 500: 0.979\n",
      "DCG@1000: 0.657 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 1, window = 8 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6c507f761fd74b71aa2b5ff2bd026001",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.507 | Hits@   1: 0.507\n",
      "DCG@   5: 0.589 | Hits@   5: 0.661\n",
      "DCG@  10: 0.610 | Hits@  10: 0.725\n",
      "DCG@ 100: 0.647 | Hits@ 100: 0.904\n",
      "DCG@ 500: 0.657 | Hits@ 500: 0.982\n",
      "DCG@1000: 0.659 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 1, window = 9 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "68a913994a4d45f3aa73c0c91bf69f38",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.512 | Hits@   1: 0.512\n",
      "DCG@   5: 0.596 | Hits@   5: 0.670\n",
      "DCG@  10: 0.615 | Hits@  10: 0.730\n",
      "DCG@ 100: 0.652 | Hits@ 100: 0.907\n",
      "DCG@ 500: 0.662 | Hits@ 500: 0.983\n",
      "DCG@1000: 0.664 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 1, window = 10 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3554bee37bef49828e48ec2d71608a4b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.511 | Hits@   1: 0.511\n",
      "DCG@   5: 0.597 | Hits@   5: 0.672\n",
      "DCG@  10: 0.613 | Hits@  10: 0.723\n",
      "DCG@ 100: 0.651 | Hits@ 100: 0.902\n",
      "DCG@ 500: 0.662 | Hits@ 500: 0.983\n",
      "DCG@1000: 0.664 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 1 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "348834c2c97d4cdeb799330cc11bf4fc",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.423 | Hits@   1: 0.423\n",
      "DCG@   5: 0.516 | Hits@   5: 0.597\n",
      "DCG@  10: 0.534 | Hits@  10: 0.655\n",
      "DCG@ 100: 0.576 | Hits@ 100: 0.860\n",
      "DCG@ 500: 0.590 | Hits@ 500: 0.964\n",
      "DCG@1000: 0.594 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 2 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "cc4e4598de9a4b96bd1ed36790d42e28",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.457 | Hits@   1: 0.457\n",
      "DCG@   5: 0.548 | Hits@   5: 0.624\n",
      "DCG@  10: 0.570 | Hits@  10: 0.695\n",
      "DCG@ 100: 0.608 | Hits@ 100: 0.881\n",
      "DCG@ 500: 0.620 | Hits@ 500: 0.970\n",
      "DCG@1000: 0.623 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 3 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2873746ce96341f69f850f83e0415ecd",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.478 | Hits@   1: 0.478\n",
      "DCG@   5: 0.563 | Hits@   5: 0.636\n",
      "DCG@  10: 0.585 | Hits@  10: 0.703\n",
      "DCG@ 100: 0.624 | Hits@ 100: 0.892\n",
      "DCG@ 500: 0.635 | Hits@ 500: 0.977\n",
      "DCG@1000: 0.638 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 4 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "77594273d59d4e25acbfc99f0de068cd",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.492 | Hits@   1: 0.492\n",
      "DCG@   5: 0.575 | Hits@   5: 0.646\n",
      "DCG@  10: 0.597 | Hits@  10: 0.714\n",
      "DCG@ 100: 0.635 | Hits@ 100: 0.901\n",
      "DCG@ 500: 0.646 | Hits@ 500: 0.981\n",
      "DCG@1000: 0.648 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 5 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "89b777085e18480fbf4eea8961e873a8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.499 | Hits@   1: 0.499\n",
      "DCG@   5: 0.584 | Hits@   5: 0.658\n",
      "DCG@  10: 0.604 | Hits@  10: 0.722\n",
      "DCG@ 100: 0.641 | Hits@ 100: 0.901\n",
      "DCG@ 500: 0.652 | Hits@ 500: 0.982\n",
      "DCG@1000: 0.654 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 6 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d765ae27d27f462c89420c61b5fbb47d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.507 | Hits@   1: 0.507\n",
      "DCG@   5: 0.588 | Hits@   5: 0.657\n",
      "DCG@  10: 0.608 | Hits@  10: 0.722\n",
      "DCG@ 100: 0.648 | Hits@ 100: 0.908\n",
      "DCG@ 500: 0.658 | Hits@ 500: 0.984\n",
      "DCG@1000: 0.659 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 7 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "8fe49774760849a9ba3c0aa96eecc2ae",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.519 | Hits@   1: 0.519\n",
      "DCG@   5: 0.599 | Hits@   5: 0.670\n",
      "DCG@  10: 0.619 | Hits@  10: 0.733\n",
      "DCG@ 100: 0.657 | Hits@ 100: 0.914\n",
      "DCG@ 500: 0.666 | Hits@ 500: 0.984\n",
      "DCG@1000: 0.668 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 8 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "46a07282fc0e4c5c8c19d2777549fc9c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.512 | Hits@   1: 0.512\n",
      "DCG@   5: 0.597 | Hits@   5: 0.670\n",
      "DCG@  10: 0.616 | Hits@  10: 0.731\n",
      "DCG@ 100: 0.654 | Hits@ 100: 0.907\n",
      "DCG@ 500: 0.664 | Hits@ 500: 0.982\n",
      "DCG@1000: 0.665 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 9 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "95d08797b2fc4128b7b5d654a9a5fc8f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.520 | Hits@   1: 0.520\n",
      "DCG@   5: 0.604 | Hits@   5: 0.679\n",
      "DCG@  10: 0.621 | Hits@  10: 0.734\n",
      "DCG@ 100: 0.659 | Hits@ 100: 0.914\n",
      "DCG@ 500: 0.669 | Hits@ 500: 0.988\n",
      "DCG@1000: 0.670 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 10 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6f0002bba1864cb18e8a58db6b16b553",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.517 | Hits@   1: 0.517\n",
      "DCG@   5: 0.603 | Hits@   5: 0.677\n",
      "DCG@  10: 0.623 | Hits@  10: 0.740\n",
      "DCG@ 100: 0.660 | Hits@ 100: 0.915\n",
      "DCG@ 500: 0.669 | Hits@ 500: 0.985\n",
      "DCG@1000: 0.671 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 1 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "54fae733f9d04d14872de5b5dd980164",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.434 | Hits@   1: 0.434\n",
      "DCG@   5: 0.527 | Hits@   5: 0.609\n",
      "DCG@  10: 0.545 | Hits@  10: 0.664\n",
      "DCG@ 100: 0.586 | Hits@ 100: 0.864\n",
      "DCG@ 500: 0.600 | Hits@ 500: 0.967\n",
      "DCG@1000: 0.603 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 2 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b8229c5aa59d448ea6972ae41627157f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.469 | Hits@   1: 0.469\n",
      "DCG@   5: 0.557 | Hits@   5: 0.633\n",
      "DCG@  10: 0.575 | Hits@  10: 0.689\n",
      "DCG@ 100: 0.617 | Hits@ 100: 0.891\n",
      "DCG@ 500: 0.628 | Hits@ 500: 0.975\n",
      "DCG@1000: 0.630 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 3 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2d4cbf3c97e64172b5aa25d2a7d207c3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.482 | Hits@   1: 0.482\n",
      "DCG@   5: 0.571 | Hits@   5: 0.646\n",
      "DCG@  10: 0.590 | Hits@  10: 0.704\n",
      "DCG@ 100: 0.630 | Hits@ 100: 0.896\n",
      "DCG@ 500: 0.640 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.643 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 4 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4120e61fece540fe8e01956a34cc32a8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.499 | Hits@   1: 0.499\n",
      "DCG@   5: 0.583 | Hits@   5: 0.652\n",
      "DCG@  10: 0.602 | Hits@  10: 0.713\n",
      "DCG@ 100: 0.641 | Hits@ 100: 0.899\n",
      "DCG@ 500: 0.652 | Hits@ 500: 0.981\n",
      "DCG@1000: 0.654 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 5 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "16cb1a4f2bec401d87d07b7d31e1555f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.504 | Hits@   1: 0.504\n",
      "DCG@   5: 0.588 | Hits@   5: 0.660\n",
      "DCG@  10: 0.608 | Hits@  10: 0.721\n",
      "DCG@ 100: 0.647 | Hits@ 100: 0.907\n",
      "DCG@ 500: 0.657 | Hits@ 500: 0.982\n",
      "DCG@1000: 0.659 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 6 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1e7ca2cf4e2b4512a6a31da825bfe906",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.515 | Hits@   1: 0.515\n",
      "DCG@   5: 0.596 | Hits@   5: 0.667\n",
      "DCG@  10: 0.615 | Hits@  10: 0.725\n",
      "DCG@ 100: 0.654 | Hits@ 100: 0.909\n",
      "DCG@ 500: 0.663 | Hits@ 500: 0.981\n",
      "DCG@1000: 0.665 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 7 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6e4377151e574f80a5e51725c4fc6fe0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.519 | Hits@   1: 0.519\n",
      "DCG@   5: 0.599 | Hits@   5: 0.669\n",
      "DCG@  10: 0.621 | Hits@  10: 0.736\n",
      "DCG@ 100: 0.657 | Hits@ 100: 0.909\n",
      "DCG@ 500: 0.667 | Hits@ 500: 0.984\n",
      "DCG@1000: 0.669 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 8 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1e90758706d94024865ec9e3653b2aa7",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.519 | Hits@   1: 0.519\n",
      "DCG@   5: 0.605 | Hits@   5: 0.679\n",
      "DCG@  10: 0.623 | Hits@  10: 0.737\n",
      "DCG@ 100: 0.661 | Hits@ 100: 0.916\n",
      "DCG@ 500: 0.670 | Hits@ 500: 0.985\n",
      "DCG@1000: 0.671 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 9 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "8a2b5a7fb9a94ede89676f203edf2b50",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.519 | Hits@   1: 0.519\n",
      "DCG@   5: 0.605 | Hits@   5: 0.678\n",
      "DCG@  10: 0.625 | Hits@  10: 0.738\n",
      "DCG@ 100: 0.662 | Hits@ 100: 0.912\n",
      "DCG@ 500: 0.672 | Hits@ 500: 0.986\n",
      "DCG@1000: 0.673 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 10 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2e646c2715654483ad402f4b6b27192b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.523 | Hits@   1: 0.523\n",
      "DCG@   5: 0.608 | Hits@   5: 0.680\n",
      "DCG@  10: 0.629 | Hits@  10: 0.745\n",
      "DCG@ 100: 0.665 | Hits@ 100: 0.915\n",
      "DCG@ 500: 0.674 | Hits@ 500: 0.986\n",
      "DCG@1000: 0.675 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 1 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "91fbf15f33f143489448fc7d67e2eb50",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.445 | Hits@   1: 0.445\n",
      "DCG@   5: 0.537 | Hits@   5: 0.618\n",
      "DCG@  10: 0.554 | Hits@  10: 0.671\n",
      "DCG@ 100: 0.595 | Hits@ 100: 0.871\n",
      "DCG@ 500: 0.607 | Hits@ 500: 0.968\n",
      "DCG@1000: 0.611 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 2 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "123a910a87754761ae7423baabb5c3c1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.468 | Hits@   1: 0.468\n",
      "DCG@   5: 0.556 | Hits@   5: 0.630\n",
      "DCG@  10: 0.579 | Hits@  10: 0.702\n",
      "DCG@ 100: 0.618 | Hits@ 100: 0.889\n",
      "DCG@ 500: 0.629 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.632 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 3 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a86e08af29d945ec920ad8075eeed383",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.488 | Hits@   1: 0.488\n",
      "DCG@   5: 0.576 | Hits@   5: 0.648\n",
      "DCG@  10: 0.596 | Hits@  10: 0.710\n",
      "DCG@ 100: 0.635 | Hits@ 100: 0.899\n",
      "DCG@ 500: 0.646 | Hits@ 500: 0.980\n",
      "DCG@1000: 0.648 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 4 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "8e8d829186e447b8b686ee74ea278062",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.489 | Hits@   1: 0.489\n",
      "DCG@   5: 0.580 | Hits@   5: 0.656\n",
      "DCG@  10: 0.602 | Hits@  10: 0.724\n",
      "DCG@ 100: 0.639 | Hits@ 100: 0.902\n",
      "DCG@ 500: 0.649 | Hits@ 500: 0.981\n",
      "DCG@1000: 0.651 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 5 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1ee44d798b7445b78d337f4143918d42",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.508 | Hits@   1: 0.508\n",
      "DCG@   5: 0.594 | Hits@   5: 0.668\n",
      "DCG@  10: 0.615 | Hits@  10: 0.732\n",
      "DCG@ 100: 0.652 | Hits@ 100: 0.911\n",
      "DCG@ 500: 0.661 | Hits@ 500: 0.982\n",
      "DCG@1000: 0.663 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 6 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "058383d00bbe4976b7a1176492f20aeb",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.521 | Hits@   1: 0.521\n",
      "DCG@   5: 0.599 | Hits@   5: 0.668\n",
      "DCG@  10: 0.618 | Hits@  10: 0.728\n",
      "DCG@ 100: 0.657 | Hits@ 100: 0.912\n",
      "DCG@ 500: 0.667 | Hits@ 500: 0.986\n",
      "DCG@1000: 0.668 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 7 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3ac3df5ef80247049613668f10419bfe",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.529 | Hits@   1: 0.529\n",
      "DCG@   5: 0.610 | Hits@   5: 0.681\n",
      "DCG@  10: 0.626 | Hits@  10: 0.732\n",
      "DCG@ 100: 0.665 | Hits@ 100: 0.916\n",
      "DCG@ 500: 0.675 | Hits@ 500: 0.987\n",
      "DCG@1000: 0.676 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 8 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c76b4267568449179402c227ede98eb4",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.525 | Hits@   1: 0.525\n",
      "DCG@   5: 0.608 | Hits@   5: 0.679\n",
      "DCG@  10: 0.628 | Hits@  10: 0.741\n",
      "DCG@ 100: 0.664 | Hits@ 100: 0.913\n",
      "DCG@ 500: 0.674 | Hits@ 500: 0.989\n",
      "DCG@1000: 0.675 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 9 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6cd564fbf1b34551b1239ccc99db239a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.527 | Hits@   1: 0.527\n",
      "DCG@   5: 0.611 | Hits@   5: 0.687\n",
      "DCG@  10: 0.629 | Hits@  10: 0.743\n",
      "DCG@ 100: 0.666 | Hits@ 100: 0.918\n",
      "DCG@ 500: 0.675 | Hits@ 500: 0.985\n",
      "DCG@1000: 0.676 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 10 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5c8f3846bfa44e3993f6f0a668deb5dd",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.529 | Hits@   1: 0.529\n",
      "DCG@   5: 0.614 | Hits@   5: 0.686\n",
      "DCG@  10: 0.635 | Hits@  10: 0.751\n",
      "DCG@ 100: 0.670 | Hits@ 100: 0.920\n",
      "DCG@ 500: 0.679 | Hits@ 500: 0.988\n",
      "DCG@1000: 0.681 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 1 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "83bd8c6312ab4e778105883d96299af5",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.433 | Hits@   1: 0.433\n",
      "DCG@   5: 0.530 | Hits@   5: 0.612\n",
      "DCG@  10: 0.547 | Hits@  10: 0.667\n",
      "DCG@ 100: 0.590 | Hits@ 100: 0.873\n",
      "DCG@ 500: 0.603 | Hits@ 500: 0.972\n",
      "DCG@1000: 0.606 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 2 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9f80c086b8b14626a83f971b3a4aac69",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.465 | Hits@   1: 0.465\n",
      "DCG@   5: 0.557 | Hits@   5: 0.635\n",
      "DCG@  10: 0.576 | Hits@  10: 0.692\n",
      "DCG@ 100: 0.618 | Hits@ 100: 0.896\n",
      "DCG@ 500: 0.629 | Hits@ 500: 0.979\n",
      "DCG@1000: 0.631 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 3 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "81f100b3d65042edb584a7d2895e8ecc",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.481 | Hits@   1: 0.481\n",
      "DCG@   5: 0.572 | Hits@   5: 0.649\n",
      "DCG@  10: 0.591 | Hits@  10: 0.706\n",
      "DCG@ 100: 0.631 | Hits@ 100: 0.897\n",
      "DCG@ 500: 0.642 | Hits@ 500: 0.981\n",
      "DCG@1000: 0.644 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 4 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2f8fe01b95184cc3a9730f3db2068a50",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.507 | Hits@   1: 0.507\n",
      "DCG@   5: 0.591 | Hits@   5: 0.663\n",
      "DCG@  10: 0.609 | Hits@  10: 0.721\n",
      "DCG@ 100: 0.648 | Hits@ 100: 0.907\n",
      "DCG@ 500: 0.658 | Hits@ 500: 0.983\n",
      "DCG@1000: 0.660 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 5 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ec5f8bff41944eeca5aa52a1fbc1c4ed",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.508 | Hits@   1: 0.508\n",
      "DCG@   5: 0.595 | Hits@   5: 0.668\n",
      "DCG@  10: 0.614 | Hits@  10: 0.727\n",
      "DCG@ 100: 0.652 | Hits@ 100: 0.910\n",
      "DCG@ 500: 0.662 | Hits@ 500: 0.980\n",
      "DCG@1000: 0.664 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 6 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b997153de9784d6fbe88cf583b1a10c3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.522 | Hits@   1: 0.522\n",
      "DCG@   5: 0.604 | Hits@   5: 0.678\n",
      "DCG@  10: 0.620 | Hits@  10: 0.729\n",
      "DCG@ 100: 0.658 | Hits@ 100: 0.907\n",
      "DCG@ 500: 0.668 | Hits@ 500: 0.984\n",
      "DCG@1000: 0.670 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 7 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9485b1410d364e649f0c032f861e4718",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.522 | Hits@   1: 0.522\n",
      "DCG@   5: 0.603 | Hits@   5: 0.675\n",
      "DCG@  10: 0.623 | Hits@  10: 0.735\n",
      "DCG@ 100: 0.660 | Hits@ 100: 0.908\n",
      "DCG@ 500: 0.670 | Hits@ 500: 0.984\n",
      "DCG@1000: 0.671 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 8 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "661d7fa584b24907bfcc784863bef059",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.520 | Hits@   1: 0.520\n",
      "DCG@   5: 0.605 | Hits@   5: 0.680\n",
      "DCG@  10: 0.625 | Hits@  10: 0.739\n",
      "DCG@ 100: 0.662 | Hits@ 100: 0.916\n",
      "DCG@ 500: 0.672 | Hits@ 500: 0.989\n",
      "DCG@1000: 0.673 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 9 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "8c0675cdf1904ba8938eed1ed495d316",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.528 | Hits@   1: 0.528\n",
      "DCG@   5: 0.612 | Hits@   5: 0.685\n",
      "DCG@  10: 0.632 | Hits@  10: 0.746\n",
      "DCG@ 100: 0.667 | Hits@ 100: 0.913\n",
      "DCG@ 500: 0.677 | Hits@ 500: 0.988\n",
      "DCG@1000: 0.678 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 10 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "99e7b01fb4d1409db4076046da874564",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.527 | Hits@   1: 0.527\n",
      "DCG@   5: 0.611 | Hits@   5: 0.683\n",
      "DCG@  10: 0.632 | Hits@  10: 0.746\n",
      "DCG@ 100: 0.667 | Hits@ 100: 0.915\n",
      "DCG@ 500: 0.677 | Hits@ 500: 0.988\n",
      "DCG@1000: 0.678 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 1 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "edb2aada8ecc45b39df934e563c61f86",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.430 | Hits@   1: 0.430\n",
      "DCG@   5: 0.531 | Hits@   5: 0.618\n",
      "DCG@  10: 0.547 | Hits@  10: 0.667\n",
      "DCG@ 100: 0.590 | Hits@ 100: 0.872\n",
      "DCG@ 500: 0.603 | Hits@ 500: 0.971\n",
      "DCG@1000: 0.606 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 2 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "8da471ddfe164a0f9d91d983ccae880d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.474 | Hits@   1: 0.474\n",
      "DCG@   5: 0.563 | Hits@   5: 0.638\n",
      "DCG@  10: 0.581 | Hits@  10: 0.694\n",
      "DCG@ 100: 0.623 | Hits@ 100: 0.897\n",
      "DCG@ 500: 0.634 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.636 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 3 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a08a408af08740b99b375be156f7a565",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.484 | Hits@   1: 0.484\n",
      "DCG@   5: 0.577 | Hits@   5: 0.655\n",
      "DCG@  10: 0.593 | Hits@  10: 0.703\n",
      "DCG@ 100: 0.634 | Hits@ 100: 0.897\n",
      "DCG@ 500: 0.645 | Hits@ 500: 0.981\n",
      "DCG@1000: 0.647 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 4 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2477cf40aacc43319d249adaf534ed75",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.497 | Hits@   1: 0.497\n",
      "DCG@   5: 0.586 | Hits@   5: 0.660\n",
      "DCG@  10: 0.604 | Hits@  10: 0.717\n",
      "DCG@ 100: 0.645 | Hits@ 100: 0.910\n",
      "DCG@ 500: 0.654 | Hits@ 500: 0.982\n",
      "DCG@1000: 0.656 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 5 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b8bcdcd98a054eeeb5351a51680ab01e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.514 | Hits@   1: 0.514\n",
      "DCG@   5: 0.597 | Hits@   5: 0.671\n",
      "DCG@  10: 0.615 | Hits@  10: 0.727\n",
      "DCG@ 100: 0.654 | Hits@ 100: 0.910\n",
      "DCG@ 500: 0.663 | Hits@ 500: 0.980\n",
      "DCG@1000: 0.665 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 6 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ccbfa9bb67524c18821fe249cc0e7a54",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.521 | Hits@   1: 0.521\n",
      "DCG@   5: 0.604 | Hits@   5: 0.680\n",
      "DCG@  10: 0.621 | Hits@  10: 0.732\n",
      "DCG@ 100: 0.659 | Hits@ 100: 0.916\n",
      "DCG@ 500: 0.668 | Hits@ 500: 0.985\n",
      "DCG@1000: 0.670 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 7 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "8a0fe7aef2e2453ebb8d63819d5696a7",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.523 | Hits@   1: 0.523\n",
      "DCG@   5: 0.605 | Hits@   5: 0.676\n",
      "DCG@  10: 0.626 | Hits@  10: 0.739\n",
      "DCG@ 100: 0.663 | Hits@ 100: 0.918\n",
      "DCG@ 500: 0.672 | Hits@ 500: 0.987\n",
      "DCG@1000: 0.674 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 8 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6decef83efe7457f8b1cf098be0d07fc",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.527 | Hits@   1: 0.527\n",
      "DCG@   5: 0.611 | Hits@   5: 0.682\n",
      "DCG@  10: 0.627 | Hits@  10: 0.734\n",
      "DCG@ 100: 0.666 | Hits@ 100: 0.917\n",
      "DCG@ 500: 0.675 | Hits@ 500: 0.986\n",
      "DCG@1000: 0.677 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 9 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f463bbb4b5734f979faf5774f8503a6d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.527 | Hits@   1: 0.527\n",
      "DCG@   5: 0.611 | Hits@   5: 0.684\n",
      "DCG@  10: 0.630 | Hits@  10: 0.743\n",
      "DCG@ 100: 0.667 | Hits@ 100: 0.915\n",
      "DCG@ 500: 0.676 | Hits@ 500: 0.986\n",
      "DCG@1000: 0.677 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 10 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a29f84bc1a41431eaffe3a2cf344e503",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.528 | Hits@   1: 0.528\n",
      "DCG@   5: 0.615 | Hits@   5: 0.688\n",
      "DCG@  10: 0.632 | Hits@  10: 0.743\n",
      "DCG@ 100: 0.669 | Hits@ 100: 0.917\n",
      "DCG@ 500: 0.679 | Hits@ 500: 0.988\n",
      "DCG@1000: 0.680 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 1 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e964133da7a44bb5ae078b31daf237f0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.446 | Hits@   1: 0.446\n",
      "DCG@   5: 0.539 | Hits@   5: 0.620\n",
      "DCG@  10: 0.554 | Hits@  10: 0.667\n",
      "DCG@ 100: 0.596 | Hits@ 100: 0.874\n",
      "DCG@ 500: 0.609 | Hits@ 500: 0.968\n",
      "DCG@1000: 0.612 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 2 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "10887c16b49a4a22b445a60c3e7a71c0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.474 | Hits@   1: 0.474\n",
      "DCG@   5: 0.564 | Hits@   5: 0.638\n",
      "DCG@  10: 0.582 | Hits@  10: 0.696\n",
      "DCG@ 100: 0.624 | Hits@ 100: 0.895\n",
      "DCG@ 500: 0.634 | Hits@ 500: 0.979\n",
      "DCG@1000: 0.637 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 3 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4518fe365b0a466aa68c72de8c4c3dc1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.485 | Hits@   1: 0.485\n",
      "DCG@   5: 0.576 | Hits@   5: 0.653\n",
      "DCG@  10: 0.593 | Hits@  10: 0.707\n",
      "DCG@ 100: 0.635 | Hits@ 100: 0.905\n",
      "DCG@ 500: 0.645 | Hits@ 500: 0.981\n",
      "DCG@1000: 0.647 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 4 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c081e4aa01814ceca303cf6473e06d86",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.503 | Hits@   1: 0.503\n",
      "DCG@   5: 0.588 | Hits@   5: 0.658\n",
      "DCG@  10: 0.607 | Hits@  10: 0.718\n",
      "DCG@ 100: 0.647 | Hits@ 100: 0.906\n",
      "DCG@ 500: 0.656 | Hits@ 500: 0.980\n",
      "DCG@1000: 0.659 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 5 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "57f3020fb53e4adf90f1d6d85ff7706c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.510 | Hits@   1: 0.510\n",
      "DCG@   5: 0.596 | Hits@   5: 0.667\n",
      "DCG@  10: 0.613 | Hits@  10: 0.722\n",
      "DCG@ 100: 0.654 | Hits@ 100: 0.912\n",
      "DCG@ 500: 0.663 | Hits@ 500: 0.984\n",
      "DCG@1000: 0.665 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 6 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "bd46eb8da58f446abfac155b31f8f85a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.514 | Hits@   1: 0.514\n",
      "DCG@   5: 0.601 | Hits@   5: 0.674\n",
      "DCG@  10: 0.619 | Hits@  10: 0.731\n",
      "DCG@ 100: 0.657 | Hits@ 100: 0.913\n",
      "DCG@ 500: 0.667 | Hits@ 500: 0.984\n",
      "DCG@1000: 0.668 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 7 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "cfdf5d6096504039b548e71e214c8485",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.523 | Hits@   1: 0.523\n",
      "DCG@   5: 0.605 | Hits@   5: 0.675\n",
      "DCG@  10: 0.626 | Hits@  10: 0.740\n",
      "DCG@ 100: 0.663 | Hits@ 100: 0.918\n",
      "DCG@ 500: 0.672 | Hits@ 500: 0.985\n",
      "DCG@1000: 0.673 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 8 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2408c1f8dbd949ecac08d625197a2569",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.521 | Hits@   1: 0.521\n",
      "DCG@   5: 0.610 | Hits@   5: 0.687\n",
      "DCG@  10: 0.627 | Hits@  10: 0.740\n",
      "DCG@ 100: 0.665 | Hits@ 100: 0.918\n",
      "DCG@ 500: 0.674 | Hits@ 500: 0.987\n",
      "DCG@1000: 0.675 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 9 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "96acb5529eef4dcc88e03c126558bbf1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.521 | Hits@   1: 0.521\n",
      "DCG@   5: 0.610 | Hits@   5: 0.688\n",
      "DCG@  10: 0.628 | Hits@  10: 0.743\n",
      "DCG@ 100: 0.664 | Hits@ 100: 0.914\n",
      "DCG@ 500: 0.674 | Hits@ 500: 0.987\n",
      "DCG@1000: 0.675 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 10 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "544fb59ac2d24eedbf5d5076683d716b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.526 | Hits@   1: 0.526\n",
      "DCG@   5: 0.611 | Hits@   5: 0.686\n",
      "DCG@  10: 0.630 | Hits@  10: 0.745\n",
      "DCG@ 100: 0.666 | Hits@ 100: 0.914\n",
      "DCG@ 500: 0.676 | Hits@ 500: 0.987\n",
      "DCG@1000: 0.677 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 1 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "805ca754aba84de896035deeca1bb486",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.439 | Hits@   1: 0.439\n",
      "DCG@   5: 0.535 | Hits@   5: 0.616\n",
      "DCG@  10: 0.553 | Hits@  10: 0.672\n",
      "DCG@ 100: 0.594 | Hits@ 100: 0.873\n",
      "DCG@ 500: 0.607 | Hits@ 500: 0.973\n",
      "DCG@1000: 0.610 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 2 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "cca7e4d3f47947e98d6a626a7b1279b8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.466 | Hits@   1: 0.466\n",
      "DCG@   5: 0.559 | Hits@   5: 0.635\n",
      "DCG@  10: 0.580 | Hits@  10: 0.700\n",
      "DCG@ 100: 0.620 | Hits@ 100: 0.897\n",
      "DCG@ 500: 0.631 | Hits@ 500: 0.977\n",
      "DCG@1000: 0.633 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 3 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4806e970ea8a441ea21ed5b663c9b011",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.491 | Hits@   1: 0.491\n",
      "DCG@   5: 0.582 | Hits@   5: 0.660\n",
      "DCG@  10: 0.599 | Hits@  10: 0.711\n",
      "DCG@ 100: 0.638 | Hits@ 100: 0.896\n",
      "DCG@ 500: 0.649 | Hits@ 500: 0.980\n",
      "DCG@1000: 0.651 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 4 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "8f4c653709d849af935443c403701dd8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.503 | Hits@   1: 0.503\n",
      "DCG@   5: 0.591 | Hits@   5: 0.668\n",
      "DCG@  10: 0.608 | Hits@  10: 0.718\n",
      "DCG@ 100: 0.648 | Hits@ 100: 0.905\n",
      "DCG@ 500: 0.657 | Hits@ 500: 0.981\n",
      "DCG@1000: 0.659 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 5 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e3fd2d967212444381afb1079d6c776f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.511 | Hits@   1: 0.511\n",
      "DCG@   5: 0.598 | Hits@   5: 0.672\n",
      "DCG@  10: 0.618 | Hits@  10: 0.732\n",
      "DCG@ 100: 0.654 | Hits@ 100: 0.908\n",
      "DCG@ 500: 0.664 | Hits@ 500: 0.982\n",
      "DCG@1000: 0.666 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 6 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "75955f3172544c4181ef18fb497abd09",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.517 | Hits@   1: 0.517\n",
      "DCG@   5: 0.601 | Hits@   5: 0.676\n",
      "DCG@  10: 0.619 | Hits@  10: 0.733\n",
      "DCG@ 100: 0.658 | Hits@ 100: 0.914\n",
      "DCG@ 500: 0.667 | Hits@ 500: 0.984\n",
      "DCG@1000: 0.668 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 7 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3eae3522cf5d4c27b3439990e3543051",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.517 | Hits@   1: 0.517\n",
      "DCG@   5: 0.606 | Hits@   5: 0.680\n",
      "DCG@  10: 0.626 | Hits@  10: 0.743\n",
      "DCG@ 100: 0.662 | Hits@ 100: 0.917\n",
      "DCG@ 500: 0.671 | Hits@ 500: 0.984\n",
      "DCG@1000: 0.672 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 8 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4db22483c9054f61b509166475696cb7",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.529 | Hits@   1: 0.529\n",
      "DCG@   5: 0.613 | Hits@   5: 0.688\n",
      "DCG@  10: 0.629 | Hits@  10: 0.740\n",
      "DCG@ 100: 0.667 | Hits@ 100: 0.919\n",
      "DCG@ 500: 0.676 | Hits@ 500: 0.984\n",
      "DCG@1000: 0.677 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 9 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4e910d241abe4ddb8fe9c9f54ca5cf1e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.533 | Hits@   1: 0.533\n",
      "DCG@   5: 0.615 | Hits@   5: 0.686\n",
      "DCG@  10: 0.634 | Hits@  10: 0.744\n",
      "DCG@ 100: 0.670 | Hits@ 100: 0.917\n",
      "DCG@ 500: 0.679 | Hits@ 500: 0.985\n",
      "DCG@1000: 0.681 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 10 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2defa60b96ac49aa9328e96e22088b14",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.529 | Hits@   1: 0.529\n",
      "DCG@   5: 0.616 | Hits@   5: 0.691\n",
      "DCG@  10: 0.634 | Hits@  10: 0.747\n",
      "DCG@ 100: 0.670 | Hits@ 100: 0.917\n",
      "DCG@ 500: 0.679 | Hits@ 500: 0.987\n",
      "DCG@1000: 0.681 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 1 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7d988e5ae16043a3828bf7269bba798b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.448 | Hits@   1: 0.448\n",
      "DCG@   5: 0.539 | Hits@   5: 0.616\n",
      "DCG@  10: 0.556 | Hits@  10: 0.668\n",
      "DCG@ 100: 0.598 | Hits@ 100: 0.873\n",
      "DCG@ 500: 0.611 | Hits@ 500: 0.971\n",
      "DCG@1000: 0.614 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 2 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f9ee55e7dd6949d4bbca72c4c9f3b15f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.472 | Hits@   1: 0.472\n",
      "DCG@   5: 0.566 | Hits@   5: 0.645\n",
      "DCG@  10: 0.584 | Hits@  10: 0.701\n",
      "DCG@ 100: 0.624 | Hits@ 100: 0.894\n",
      "DCG@ 500: 0.635 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.638 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 3 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "939166d11a454f6b887ae5404576ab33",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.488 | Hits@   1: 0.488\n",
      "DCG@   5: 0.578 | Hits@   5: 0.656\n",
      "DCG@  10: 0.594 | Hits@  10: 0.706\n",
      "DCG@ 100: 0.635 | Hits@ 100: 0.896\n",
      "DCG@ 500: 0.646 | Hits@ 500: 0.979\n",
      "DCG@1000: 0.648 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 4 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0cf9bbf37eb24addbd6cb0719c821168",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.497 | Hits@   1: 0.497\n",
      "DCG@   5: 0.587 | Hits@   5: 0.663\n",
      "DCG@  10: 0.604 | Hits@  10: 0.716\n",
      "DCG@ 100: 0.643 | Hits@ 100: 0.901\n",
      "DCG@ 500: 0.654 | Hits@ 500: 0.981\n",
      "DCG@1000: 0.656 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 5 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2ae3cae61fc24fca9afb20551994fa9d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.513 | Hits@   1: 0.513\n",
      "DCG@   5: 0.598 | Hits@   5: 0.673\n",
      "DCG@  10: 0.617 | Hits@  10: 0.730\n",
      "DCG@ 100: 0.654 | Hits@ 100: 0.911\n",
      "DCG@ 500: 0.664 | Hits@ 500: 0.982\n",
      "DCG@1000: 0.666 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 6 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c92ffc5a288941dab8ff0fa0770c8730",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.518 | Hits@   1: 0.518\n",
      "DCG@   5: 0.602 | Hits@   5: 0.675\n",
      "DCG@  10: 0.619 | Hits@  10: 0.731\n",
      "DCG@ 100: 0.659 | Hits@ 100: 0.919\n",
      "DCG@ 500: 0.668 | Hits@ 500: 0.982\n",
      "DCG@1000: 0.669 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 7 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "68b50fce4523426c8718b8543049c048",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.520 | Hits@   1: 0.520\n",
      "DCG@   5: 0.607 | Hits@   5: 0.681\n",
      "DCG@  10: 0.624 | Hits@  10: 0.735\n",
      "DCG@ 100: 0.662 | Hits@ 100: 0.914\n",
      "DCG@ 500: 0.671 | Hits@ 500: 0.984\n",
      "DCG@1000: 0.673 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 8 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ddebf808e3f44bec9a381367514d1095",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.527 | Hits@   1: 0.527\n",
      "DCG@   5: 0.610 | Hits@   5: 0.680\n",
      "DCG@  10: 0.631 | Hits@  10: 0.744\n",
      "DCG@ 100: 0.667 | Hits@ 100: 0.915\n",
      "DCG@ 500: 0.676 | Hits@ 500: 0.984\n",
      "DCG@1000: 0.677 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 9 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "505943654ca542c2aaf039343a248440",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.533 | Hits@   1: 0.533\n",
      "DCG@   5: 0.614 | Hits@   5: 0.684\n",
      "DCG@  10: 0.633 | Hits@  10: 0.744\n",
      "DCG@ 100: 0.670 | Hits@ 100: 0.921\n",
      "DCG@ 500: 0.678 | Hits@ 500: 0.984\n",
      "DCG@1000: 0.680 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 10 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "05b5aabaf916474a9f2261816c490e5a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.534 | Hits@   1: 0.534\n",
      "DCG@   5: 0.616 | Hits@   5: 0.685\n",
      "DCG@  10: 0.634 | Hits@  10: 0.743\n",
      "DCG@ 100: 0.670 | Hits@ 100: 0.912\n",
      "DCG@ 500: 0.680 | Hits@ 500: 0.986\n",
      "DCG@1000: 0.682 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 1 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "efc1884d4759418bb3586bba671edbe0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.442 | Hits@   1: 0.442\n",
      "DCG@   5: 0.538 | Hits@   5: 0.618\n",
      "DCG@  10: 0.554 | Hits@  10: 0.670\n",
      "DCG@ 100: 0.597 | Hits@ 100: 0.873\n",
      "DCG@ 500: 0.610 | Hits@ 500: 0.974\n",
      "DCG@1000: 0.613 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 2 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "098ebfa9fc2f472a879afd2ffa73d3fe",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.479 | Hits@   1: 0.479\n",
      "DCG@   5: 0.570 | Hits@   5: 0.652\n",
      "DCG@  10: 0.586 | Hits@  10: 0.700\n",
      "DCG@ 100: 0.626 | Hits@ 100: 0.892\n",
      "DCG@ 500: 0.637 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.639 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 3 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "242653f7030942a19cf87f2dd711f6f1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.489 | Hits@   1: 0.489\n",
      "DCG@   5: 0.580 | Hits@   5: 0.657\n",
      "DCG@  10: 0.600 | Hits@  10: 0.717\n",
      "DCG@ 100: 0.638 | Hits@ 100: 0.898\n",
      "DCG@ 500: 0.648 | Hits@ 500: 0.977\n",
      "DCG@1000: 0.650 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 4 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ede9fecba52444b4ae0e21904e7150c3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.501 | Hits@   1: 0.501\n",
      "DCG@   5: 0.587 | Hits@   5: 0.658\n",
      "DCG@  10: 0.605 | Hits@  10: 0.715\n",
      "DCG@ 100: 0.645 | Hits@ 100: 0.903\n",
      "DCG@ 500: 0.656 | Hits@ 500: 0.982\n",
      "DCG@1000: 0.658 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 5 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "651e2eb8d60147fd81a92775ff12e524",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.513 | Hits@   1: 0.513\n",
      "DCG@   5: 0.595 | Hits@   5: 0.665\n",
      "DCG@  10: 0.617 | Hits@  10: 0.731\n",
      "DCG@ 100: 0.654 | Hits@ 100: 0.909\n",
      "DCG@ 500: 0.664 | Hits@ 500: 0.983\n",
      "DCG@1000: 0.666 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 6 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ec8f7b4a1e9f4463808bd74abad90d38",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.513 | Hits@   1: 0.513\n",
      "DCG@   5: 0.599 | Hits@   5: 0.674\n",
      "DCG@  10: 0.619 | Hits@  10: 0.734\n",
      "DCG@ 100: 0.656 | Hits@ 100: 0.913\n",
      "DCG@ 500: 0.666 | Hits@ 500: 0.984\n",
      "DCG@1000: 0.667 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 7 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b2427e50a1c14848a9dbbd4c542a6ba6",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.523 | Hits@   1: 0.523\n",
      "DCG@   5: 0.606 | Hits@   5: 0.677\n",
      "DCG@  10: 0.625 | Hits@  10: 0.735\n",
      "DCG@ 100: 0.662 | Hits@ 100: 0.910\n",
      "DCG@ 500: 0.672 | Hits@ 500: 0.985\n",
      "DCG@1000: 0.674 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 8 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "dd811560e8f249758ba0136ff7ed8e11",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.529 | Hits@   1: 0.529\n",
      "DCG@   5: 0.610 | Hits@   5: 0.679\n",
      "DCG@  10: 0.629 | Hits@  10: 0.740\n",
      "DCG@ 100: 0.666 | Hits@ 100: 0.915\n",
      "DCG@ 500: 0.675 | Hits@ 500: 0.986\n",
      "DCG@1000: 0.677 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 9 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b7a56fc52b614cc8bde90aaa973fd1b9",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.526 | Hits@   1: 0.526\n",
      "DCG@   5: 0.613 | Hits@   5: 0.687\n",
      "DCG@  10: 0.632 | Hits@  10: 0.746\n",
      "DCG@ 100: 0.667 | Hits@ 100: 0.914\n",
      "DCG@ 500: 0.676 | Hits@ 500: 0.985\n",
      "DCG@1000: 0.678 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 10 with Skip-Gram\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1c57ecf03fce4eb289d76c7244427ff6",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DCG@   1: 0.533 | Hits@   1: 0.533\n",
      "DCG@   5: 0.618 | Hits@   5: 0.690\n",
      "DCG@  10: 0.637 | Hits@  10: 0.750\n",
      "DCG@ 100: 0.672 | Hits@ 100: 0.920\n",
      "DCG@ 500: 0.681 | Hits@ 500: 0.986\n",
      "DCG@1000: 0.682 | Hits@1000: 1.000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "dcg_history = []\n",
    "hits_history = []\n",
    "for min_count in range(1,11):\n",
    "    for window in range(1,11):\n",
    "        embeddings_trained = Word2Vec(words,                    # data for model to train on\n",
    "                                      vector_size=200,          # embedding vector size\n",
    "                                      min_count=min_count,      # Ignores all words with total frequency lower than this. Consider words that occured at least 5 times\n",
    "                                      window=window,            # Maximum distance between the current and predicted word within a sentence\n",
    "                                      workers=4,                # Use these many worker threads to train the model\n",
    "                                      sg=1).wv                  # sg=0 - CBOW, sg=1 - Skip-Gram\n",
    "        wv_ranking = []\n",
    "        max_validation_examples = 1000\n",
    "        tokenizer = MyTokenizer()\n",
    "        for i, line in enumerate(validation_data):\n",
    "            if i == max_validation_examples:\n",
    "                break\n",
    "            q, *ex = line\n",
    "            ranks = rank_candidates(q, ex, embeddings_trained, tokenizer)\n",
    "            wv_ranking.append([r[0] for r in ranks].index(0) + 1)\n",
    "        print(f'min_count = {min_count}, window = {window} with Skip-Gram')\n",
    "        for k in tqdm([1, 5, 10, 100, 500, 1000]):\n",
    "            print(\"DCG@%4d: %.3f | Hits@%4d: %.3f\" % (k, dcg_score(wv_ranking, k), k, hits_count(wv_ranking, k)))\n",
    "            dcg_history.append(dcg_score(wv_ranking, k))\n",
    "            hits_history.append(hits_count(wv_ranking, k))\n",
    "        print()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>DCG@1</th>\n",
       "      <th>DCG@5</th>\n",
       "      <th>DCG@10</th>\n",
       "      <th>DCG@100</th>\n",
       "      <th>DCG@500</th>\n",
       "      <th>DCG@1000</th>\n",
       "      <th>Hits@1</th>\n",
       "      <th>Hits@5</th>\n",
       "      <th>Hits@10</th>\n",
       "      <th>Hits@100</th>\n",
       "      <th>Hits@500</th>\n",
       "      <th>Hits@1000</th>\n",
       "      <th>SUM</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Min_1/Win_1</th>\n",
       "      <td>0.412</td>\n",
       "      <td>0.503</td>\n",
       "      <td>0.522</td>\n",
       "      <td>0.563</td>\n",
       "      <td>0.578</td>\n",
       "      <td>0.582</td>\n",
       "      <td>0.412</td>\n",
       "      <td>0.584</td>\n",
       "      <td>0.642</td>\n",
       "      <td>0.842</td>\n",
       "      <td>0.956</td>\n",
       "      <td>1.0</td>\n",
       "      <td>7.596</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Min_1/Win_2</th>\n",
       "      <td>0.446</td>\n",
       "      <td>0.530</td>\n",
       "      <td>0.553</td>\n",
       "      <td>0.592</td>\n",
       "      <td>0.606</td>\n",
       "      <td>0.609</td>\n",
       "      <td>0.446</td>\n",
       "      <td>0.604</td>\n",
       "      <td>0.673</td>\n",
       "      <td>0.861</td>\n",
       "      <td>0.967</td>\n",
       "      <td>1.0</td>\n",
       "      <td>7.887</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Min_1/Win_3</th>\n",
       "      <td>0.463</td>\n",
       "      <td>0.549</td>\n",
       "      <td>0.568</td>\n",
       "      <td>0.608</td>\n",
       "      <td>0.621</td>\n",
       "      <td>0.624</td>\n",
       "      <td>0.463</td>\n",
       "      <td>0.623</td>\n",
       "      <td>0.680</td>\n",
       "      <td>0.870</td>\n",
       "      <td>0.970</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.039</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Min_1/Win_4</th>\n",
       "      <td>0.482</td>\n",
       "      <td>0.559</td>\n",
       "      <td>0.582</td>\n",
       "      <td>0.621</td>\n",
       "      <td>0.632</td>\n",
       "      <td>0.635</td>\n",
       "      <td>0.482</td>\n",
       "      <td>0.628</td>\n",
       "      <td>0.700</td>\n",
       "      <td>0.887</td>\n",
       "      <td>0.973</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.181</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Min_1/Win_5</th>\n",
       "      <td>0.494</td>\n",
       "      <td>0.575</td>\n",
       "      <td>0.593</td>\n",
       "      <td>0.633</td>\n",
       "      <td>0.644</td>\n",
       "      <td>0.646</td>\n",
       "      <td>0.494</td>\n",
       "      <td>0.645</td>\n",
       "      <td>0.701</td>\n",
       "      <td>0.893</td>\n",
       "      <td>0.976</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.294</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Min_10/Win_6</th>\n",
       "      <td>0.513</td>\n",
       "      <td>0.599</td>\n",
       "      <td>0.619</td>\n",
       "      <td>0.656</td>\n",
       "      <td>0.666</td>\n",
       "      <td>0.667</td>\n",
       "      <td>0.513</td>\n",
       "      <td>0.674</td>\n",
       "      <td>0.734</td>\n",
       "      <td>0.913</td>\n",
       "      <td>0.984</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.538</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Min_10/Win_7</th>\n",
       "      <td>0.523</td>\n",
       "      <td>0.606</td>\n",
       "      <td>0.625</td>\n",
       "      <td>0.662</td>\n",
       "      <td>0.672</td>\n",
       "      <td>0.674</td>\n",
       "      <td>0.523</td>\n",
       "      <td>0.677</td>\n",
       "      <td>0.735</td>\n",
       "      <td>0.910</td>\n",
       "      <td>0.985</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.592</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Min_10/Win_8</th>\n",
       "      <td>0.529</td>\n",
       "      <td>0.610</td>\n",
       "      <td>0.629</td>\n",
       "      <td>0.666</td>\n",
       "      <td>0.675</td>\n",
       "      <td>0.677</td>\n",
       "      <td>0.529</td>\n",
       "      <td>0.679</td>\n",
       "      <td>0.740</td>\n",
       "      <td>0.915</td>\n",
       "      <td>0.986</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.635</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Min_10/Win_9</th>\n",
       "      <td>0.526</td>\n",
       "      <td>0.613</td>\n",
       "      <td>0.632</td>\n",
       "      <td>0.667</td>\n",
       "      <td>0.676</td>\n",
       "      <td>0.678</td>\n",
       "      <td>0.526</td>\n",
       "      <td>0.687</td>\n",
       "      <td>0.746</td>\n",
       "      <td>0.914</td>\n",
       "      <td>0.985</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.650</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Min_10/Win_10</th>\n",
       "      <td>0.533</td>\n",
       "      <td>0.618</td>\n",
       "      <td>0.637</td>\n",
       "      <td>0.672</td>\n",
       "      <td>0.681</td>\n",
       "      <td>0.682</td>\n",
       "      <td>0.533</td>\n",
       "      <td>0.690</td>\n",
       "      <td>0.750</td>\n",
       "      <td>0.920</td>\n",
       "      <td>0.986</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.702</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>100 rows × 13 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "               DCG@1  DCG@5  DCG@10  DCG@100  DCG@500  DCG@1000  Hits@1  \\\n",
       "Min_1/Win_1    0.412  0.503   0.522    0.563    0.578     0.582   0.412   \n",
       "Min_1/Win_2    0.446  0.530   0.553    0.592    0.606     0.609   0.446   \n",
       "Min_1/Win_3    0.463  0.549   0.568    0.608    0.621     0.624   0.463   \n",
       "Min_1/Win_4    0.482  0.559   0.582    0.621    0.632     0.635   0.482   \n",
       "Min_1/Win_5    0.494  0.575   0.593    0.633    0.644     0.646   0.494   \n",
       "...              ...    ...     ...      ...      ...       ...     ...   \n",
       "Min_10/Win_6   0.513  0.599   0.619    0.656    0.666     0.667   0.513   \n",
       "Min_10/Win_7   0.523  0.606   0.625    0.662    0.672     0.674   0.523   \n",
       "Min_10/Win_8   0.529  0.610   0.629    0.666    0.675     0.677   0.529   \n",
       "Min_10/Win_9   0.526  0.613   0.632    0.667    0.676     0.678   0.526   \n",
       "Min_10/Win_10  0.533  0.618   0.637    0.672    0.681     0.682   0.533   \n",
       "\n",
       "               Hits@5  Hits@10  Hits@100  Hits@500  Hits@1000    SUM  \n",
       "Min_1/Win_1     0.584    0.642     0.842     0.956        1.0  7.596  \n",
       "Min_1/Win_2     0.604    0.673     0.861     0.967        1.0  7.887  \n",
       "Min_1/Win_3     0.623    0.680     0.870     0.970        1.0  8.039  \n",
       "Min_1/Win_4     0.628    0.700     0.887     0.973        1.0  8.181  \n",
       "Min_1/Win_5     0.645    0.701     0.893     0.976        1.0  8.294  \n",
       "...               ...      ...       ...       ...        ...    ...  \n",
       "Min_10/Win_6    0.674    0.734     0.913     0.984        1.0  8.538  \n",
       "Min_10/Win_7    0.677    0.735     0.910     0.985        1.0  8.592  \n",
       "Min_10/Win_8    0.679    0.740     0.915     0.986        1.0  8.635  \n",
       "Min_10/Win_9    0.687    0.746     0.914     0.985        1.0  8.650  \n",
       "Min_10/Win_10   0.690    0.750     0.920     0.986        1.0  8.702  \n",
       "\n",
       "[100 rows x 13 columns]"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "columns = ['DCG@1','DCG@5','DCG@10','DCG@100','DCG@500','DCG@1000',\n",
    "           'Hits@1','Hits@5','Hits@10','Hits@100','Hits@500','Hits@1000']\n",
    "row_list = []\n",
    "for i in range(1,11):\n",
    "    for j in range(1,11):\n",
    "        row_list.append('Min_'+str(i)+'/Win_'+str(j))\n",
    "\n",
    "df = pd.DataFrame()\n",
    "m = n = 0\n",
    "for row in row_list:\n",
    "    for dcg in columns[:6]:\n",
    "        df.loc[row,dcg] = dcg_history[m]\n",
    "        m+=1\n",
    "    for hits in columns[6:]:\n",
    "        df.loc[row,hits] = hits_history[n]\n",
    "        n+=1\n",
    "\n",
    "df['SUM'] = df.sum(axis=1)\n",
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>DCG@1</th>\n",
       "      <th>DCG@5</th>\n",
       "      <th>DCG@10</th>\n",
       "      <th>DCG@100</th>\n",
       "      <th>DCG@500</th>\n",
       "      <th>DCG@1000</th>\n",
       "      <th>Hits@1</th>\n",
       "      <th>Hits@5</th>\n",
       "      <th>Hits@10</th>\n",
       "      <th>Hits@100</th>\n",
       "      <th>Hits@500</th>\n",
       "      <th>Hits@1000</th>\n",
       "      <th>SUM</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Min_9/Win_10</th>\n",
       "      <td>0.534</td>\n",
       "      <td>0.616</td>\n",
       "      <td>0.634</td>\n",
       "      <td>0.67</td>\n",
       "      <td>0.68</td>\n",
       "      <td>0.682</td>\n",
       "      <td>0.534</td>\n",
       "      <td>0.685</td>\n",
       "      <td>0.743</td>\n",
       "      <td>0.912</td>\n",
       "      <td>0.986</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.676</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              DCG@1  DCG@5  DCG@10  DCG@100  DCG@500  DCG@1000  Hits@1  \\\n",
       "Min_9/Win_10  0.534  0.616   0.634     0.67     0.68     0.682   0.534   \n",
       "\n",
       "              Hits@5  Hits@10  Hits@100  Hits@500  Hits@1000    SUM  \n",
       "Min_9/Win_10   0.685    0.743     0.912     0.986        1.0  8.676  "
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Строка с максимальным значением Hits@1\n",
    "df[df['Hits@1']==df['Hits@1'].max()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>DCG@1</th>\n",
       "      <th>DCG@5</th>\n",
       "      <th>DCG@10</th>\n",
       "      <th>DCG@100</th>\n",
       "      <th>DCG@500</th>\n",
       "      <th>DCG@1000</th>\n",
       "      <th>Hits@1</th>\n",
       "      <th>Hits@5</th>\n",
       "      <th>Hits@10</th>\n",
       "      <th>Hits@100</th>\n",
       "      <th>Hits@500</th>\n",
       "      <th>Hits@1000</th>\n",
       "      <th>SUM</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Min_10/Win_10</th>\n",
       "      <td>0.533</td>\n",
       "      <td>0.618</td>\n",
       "      <td>0.637</td>\n",
       "      <td>0.672</td>\n",
       "      <td>0.681</td>\n",
       "      <td>0.682</td>\n",
       "      <td>0.533</td>\n",
       "      <td>0.69</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.92</td>\n",
       "      <td>0.986</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.702</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               DCG@1  DCG@5  DCG@10  DCG@100  DCG@500  DCG@1000  Hits@1  \\\n",
       "Min_10/Win_10  0.533  0.618   0.637    0.672    0.681     0.682   0.533   \n",
       "\n",
       "               Hits@5  Hits@10  Hits@100  Hits@500  Hits@1000    SUM  \n",
       "Min_10/Win_10    0.69     0.75      0.92     0.986        1.0  8.702  "
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Строка с максимальным значением суммы всех метрик (DCG@1 + DCG@5 + .... + Hits@500 + Hits@1000)\n",
    "df[df['SUM']==df['SUM'].max()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 270,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(15, 5))\n",
    "fig.suptitle('DCG@1 / Hits@1 with Skip-Gram', fontsize=20, fontfamily = 'Verdana', fontstyle = 'normal', color = 'midnightblue')\n",
    "ax.plot(df.index, df[['DCG@1','Hits@1']])\n",
    "\n",
    "plt.xticks(np.arange(0, 100, 4),rotation=45)\n",
    "\n",
    "plt.legend(df[['DCG@1','Hits@1']],loc = 'upper left', shadow = False,\n",
    "          facecolor = 'yellow', edgecolor = 'r', framealpha = .1, fontsize=16)\n",
    "plt.tick_params(axis = 'both', which = 'major', direction = 'inout', color = 'm',\n",
    "               labelcolor = 'b', bottom = True, left = True, labelbottom = True, labelleft = True)\n",
    "plt.tick_params(axis = 'both', which = 'minor', direction = 'out', color = 'm',\n",
    "               labelcolor = 'r', bottom = True, left = True)\n",
    "plt.grid(which='major', color = 'm', linestyle = '--')\n",
    "\n",
    "plt.minorticks_on()\n",
    "plt.grid(which='minor', color = 'm', linestyle = ':')\n",
    "\n",
    "ax.set_xlabel('', fontsize = 20, color = 'midnightblue', fontstyle = 'oblique')\n",
    "ax.set_ylabel('', fontsize = 20, color = 'midnightblue', fontstyle = 'oblique')\n",
    "plt.setp(ax, xlabel = 'MIN_COUNT / WINDOW', ylabel = 'VALUES')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### with CBOW"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 272,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "min_count = 1, window = 1 with CBOW\n",
      "DCG@   1: 0.358 | Hits@   1: 0.358\n",
      "DCG@   5: 0.447 | Hits@   5: 0.525\n",
      "DCG@  10: 0.476 | Hits@  10: 0.614\n",
      "DCG@ 100: 0.518 | Hits@ 100: 0.824\n",
      "DCG@ 500: 0.534 | Hits@ 500: 0.949\n",
      "DCG@1000: 0.540 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 1, window = 2 with CBOW\n",
      "DCG@   1: 0.367 | Hits@   1: 0.367\n",
      "DCG@   5: 0.464 | Hits@   5: 0.550\n",
      "DCG@  10: 0.485 | Hits@  10: 0.617\n",
      "DCG@ 100: 0.532 | Hits@ 100: 0.850\n",
      "DCG@ 500: 0.547 | Hits@ 500: 0.968\n",
      "DCG@1000: 0.551 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 1, window = 3 with CBOW\n",
      "DCG@   1: 0.377 | Hits@   1: 0.377\n",
      "DCG@   5: 0.470 | Hits@   5: 0.554\n",
      "DCG@  10: 0.494 | Hits@  10: 0.630\n",
      "DCG@ 100: 0.541 | Hits@ 100: 0.859\n",
      "DCG@ 500: 0.556 | Hits@ 500: 0.974\n",
      "DCG@1000: 0.559 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 1, window = 4 with CBOW\n",
      "DCG@   1: 0.385 | Hits@   1: 0.385\n",
      "DCG@   5: 0.475 | Hits@   5: 0.554\n",
      "DCG@  10: 0.500 | Hits@  10: 0.634\n",
      "DCG@ 100: 0.548 | Hits@ 100: 0.866\n",
      "DCG@ 500: 0.563 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.565 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 1, window = 5 with CBOW\n",
      "DCG@   1: 0.393 | Hits@   1: 0.393\n",
      "DCG@   5: 0.479 | Hits@   5: 0.555\n",
      "DCG@  10: 0.509 | Hits@  10: 0.648\n",
      "DCG@ 100: 0.554 | Hits@ 100: 0.866\n",
      "DCG@ 500: 0.568 | Hits@ 500: 0.977\n",
      "DCG@1000: 0.571 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 1, window = 6 with CBOW\n",
      "DCG@   1: 0.395 | Hits@   1: 0.395\n",
      "DCG@   5: 0.485 | Hits@   5: 0.565\n",
      "DCG@  10: 0.511 | Hits@  10: 0.646\n",
      "DCG@ 100: 0.559 | Hits@ 100: 0.879\n",
      "DCG@ 500: 0.572 | Hits@ 500: 0.977\n",
      "DCG@1000: 0.575 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 1, window = 7 with CBOW\n",
      "DCG@   1: 0.396 | Hits@   1: 0.396\n",
      "DCG@   5: 0.493 | Hits@   5: 0.578\n",
      "DCG@  10: 0.517 | Hits@  10: 0.652\n",
      "DCG@ 100: 0.564 | Hits@ 100: 0.880\n",
      "DCG@ 500: 0.576 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.579 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 1, window = 8 with CBOW\n",
      "DCG@   1: 0.397 | Hits@   1: 0.397\n",
      "DCG@   5: 0.493 | Hits@   5: 0.578\n",
      "DCG@  10: 0.517 | Hits@  10: 0.651\n",
      "DCG@ 100: 0.564 | Hits@ 100: 0.880\n",
      "DCG@ 500: 0.577 | Hits@ 500: 0.977\n",
      "DCG@1000: 0.579 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 1, window = 9 with CBOW\n",
      "DCG@   1: 0.405 | Hits@   1: 0.405\n",
      "DCG@   5: 0.493 | Hits@   5: 0.572\n",
      "DCG@  10: 0.520 | Hits@  10: 0.654\n",
      "DCG@ 100: 0.567 | Hits@ 100: 0.879\n",
      "DCG@ 500: 0.580 | Hits@ 500: 0.981\n",
      "DCG@1000: 0.582 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 1, window = 10 with CBOW\n",
      "DCG@   1: 0.410 | Hits@   1: 0.410\n",
      "DCG@   5: 0.500 | Hits@   5: 0.578\n",
      "DCG@  10: 0.523 | Hits@  10: 0.648\n",
      "DCG@ 100: 0.573 | Hits@ 100: 0.886\n",
      "DCG@ 500: 0.584 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.587 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 1 with CBOW\n",
      "DCG@   1: 0.381 | Hits@   1: 0.381\n",
      "DCG@   5: 0.463 | Hits@   5: 0.539\n",
      "DCG@  10: 0.491 | Hits@  10: 0.626\n",
      "DCG@ 100: 0.534 | Hits@ 100: 0.836\n",
      "DCG@ 500: 0.550 | Hits@ 500: 0.963\n",
      "DCG@1000: 0.554 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 2 with CBOW\n",
      "DCG@   1: 0.379 | Hits@   1: 0.379\n",
      "DCG@   5: 0.472 | Hits@   5: 0.554\n",
      "DCG@  10: 0.498 | Hits@  10: 0.635\n",
      "DCG@ 100: 0.544 | Hits@ 100: 0.861\n",
      "DCG@ 500: 0.558 | Hits@ 500: 0.973\n",
      "DCG@1000: 0.561 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 3 with CBOW\n",
      "DCG@   1: 0.382 | Hits@   1: 0.382\n",
      "DCG@   5: 0.477 | Hits@   5: 0.560\n",
      "DCG@  10: 0.502 | Hits@  10: 0.637\n",
      "DCG@ 100: 0.549 | Hits@ 100: 0.869\n",
      "DCG@ 500: 0.563 | Hits@ 500: 0.975\n",
      "DCG@1000: 0.566 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 4 with CBOW\n",
      "DCG@   1: 0.394 | Hits@   1: 0.394\n",
      "DCG@   5: 0.484 | Hits@   5: 0.566\n",
      "DCG@  10: 0.510 | Hits@  10: 0.647\n",
      "DCG@ 100: 0.556 | Hits@ 100: 0.870\n",
      "DCG@ 500: 0.570 | Hits@ 500: 0.972\n",
      "DCG@1000: 0.573 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 5 with CBOW\n",
      "DCG@   1: 0.400 | Hits@   1: 0.400\n",
      "DCG@   5: 0.491 | Hits@   5: 0.570\n",
      "DCG@  10: 0.516 | Hits@  10: 0.649\n",
      "DCG@ 100: 0.563 | Hits@ 100: 0.880\n",
      "DCG@ 500: 0.576 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.579 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 6 with CBOW\n",
      "DCG@   1: 0.401 | Hits@   1: 0.401\n",
      "DCG@   5: 0.494 | Hits@   5: 0.575\n",
      "DCG@  10: 0.520 | Hits@  10: 0.656\n",
      "DCG@ 100: 0.565 | Hits@ 100: 0.879\n",
      "DCG@ 500: 0.578 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.581 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 7 with CBOW\n",
      "DCG@   1: 0.398 | Hits@   1: 0.398\n",
      "DCG@   5: 0.496 | Hits@   5: 0.578\n",
      "DCG@  10: 0.522 | Hits@  10: 0.658\n",
      "DCG@ 100: 0.568 | Hits@ 100: 0.882\n",
      "DCG@ 500: 0.580 | Hits@ 500: 0.979\n",
      "DCG@1000: 0.583 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 8 with CBOW\n",
      "DCG@   1: 0.403 | Hits@   1: 0.403\n",
      "DCG@   5: 0.498 | Hits@   5: 0.580\n",
      "DCG@  10: 0.523 | Hits@  10: 0.658\n",
      "DCG@ 100: 0.570 | Hits@ 100: 0.885\n",
      "DCG@ 500: 0.582 | Hits@ 500: 0.977\n",
      "DCG@1000: 0.585 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 9 with CBOW\n",
      "DCG@   1: 0.412 | Hits@   1: 0.412\n",
      "DCG@   5: 0.504 | Hits@   5: 0.587\n",
      "DCG@  10: 0.529 | Hits@  10: 0.664\n",
      "DCG@ 100: 0.574 | Hits@ 100: 0.884\n",
      "DCG@ 500: 0.586 | Hits@ 500: 0.977\n",
      "DCG@1000: 0.588 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 2, window = 10 with CBOW\n",
      "DCG@   1: 0.406 | Hits@   1: 0.406\n",
      "DCG@   5: 0.502 | Hits@   5: 0.584\n",
      "DCG@  10: 0.528 | Hits@  10: 0.661\n",
      "DCG@ 100: 0.574 | Hits@ 100: 0.882\n",
      "DCG@ 500: 0.586 | Hits@ 500: 0.979\n",
      "DCG@1000: 0.589 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 1 with CBOW\n",
      "DCG@   1: 0.382 | Hits@   1: 0.382\n",
      "DCG@   5: 0.467 | Hits@   5: 0.546\n",
      "DCG@  10: 0.495 | Hits@  10: 0.631\n",
      "DCG@ 100: 0.538 | Hits@ 100: 0.843\n",
      "DCG@ 500: 0.553 | Hits@ 500: 0.959\n",
      "DCG@1000: 0.558 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 2 with CBOW\n",
      "DCG@   1: 0.380 | Hits@   1: 0.380\n",
      "DCG@   5: 0.472 | Hits@   5: 0.555\n",
      "DCG@  10: 0.498 | Hits@  10: 0.636\n",
      "DCG@ 100: 0.545 | Hits@ 100: 0.864\n",
      "DCG@ 500: 0.558 | Hits@ 500: 0.969\n",
      "DCG@1000: 0.562 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 3 with CBOW\n",
      "DCG@   1: 0.387 | Hits@   1: 0.387\n",
      "DCG@   5: 0.481 | Hits@   5: 0.562\n",
      "DCG@  10: 0.507 | Hits@  10: 0.643\n",
      "DCG@ 100: 0.553 | Hits@ 100: 0.872\n",
      "DCG@ 500: 0.567 | Hits@ 500: 0.975\n",
      "DCG@1000: 0.569 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 4 with CBOW\n",
      "DCG@   1: 0.389 | Hits@   1: 0.389\n",
      "DCG@   5: 0.483 | Hits@   5: 0.567\n",
      "DCG@  10: 0.510 | Hits@  10: 0.651\n",
      "DCG@ 100: 0.556 | Hits@ 100: 0.872\n",
      "DCG@ 500: 0.569 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.572 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 5 with CBOW\n",
      "DCG@   1: 0.399 | Hits@   1: 0.399\n",
      "DCG@   5: 0.495 | Hits@   5: 0.580\n",
      "DCG@  10: 0.518 | Hits@  10: 0.653\n",
      "DCG@ 100: 0.564 | Hits@ 100: 0.875\n",
      "DCG@ 500: 0.577 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.580 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 6 with CBOW\n",
      "DCG@   1: 0.405 | Hits@   1: 0.405\n",
      "DCG@   5: 0.496 | Hits@   5: 0.575\n",
      "DCG@  10: 0.520 | Hits@  10: 0.650\n",
      "DCG@ 100: 0.568 | Hits@ 100: 0.881\n",
      "DCG@ 500: 0.580 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.583 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 7 with CBOW\n",
      "DCG@   1: 0.408 | Hits@   1: 0.408\n",
      "DCG@   5: 0.501 | Hits@   5: 0.583\n",
      "DCG@  10: 0.524 | Hits@  10: 0.655\n",
      "DCG@ 100: 0.572 | Hits@ 100: 0.883\n",
      "DCG@ 500: 0.584 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.586 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 8 with CBOW\n",
      "DCG@   1: 0.406 | Hits@   1: 0.406\n",
      "DCG@   5: 0.500 | Hits@   5: 0.582\n",
      "DCG@  10: 0.528 | Hits@  10: 0.668\n",
      "DCG@ 100: 0.571 | Hits@ 100: 0.882\n",
      "DCG@ 500: 0.584 | Hits@ 500: 0.977\n",
      "DCG@1000: 0.586 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 9 with CBOW\n",
      "DCG@   1: 0.408 | Hits@   1: 0.408\n",
      "DCG@   5: 0.501 | Hits@   5: 0.581\n",
      "DCG@  10: 0.528 | Hits@  10: 0.661\n",
      "DCG@ 100: 0.573 | Hits@ 100: 0.881\n",
      "DCG@ 500: 0.586 | Hits@ 500: 0.980\n",
      "DCG@1000: 0.588 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 3, window = 10 with CBOW\n",
      "DCG@   1: 0.409 | Hits@   1: 0.409\n",
      "DCG@   5: 0.505 | Hits@   5: 0.588\n",
      "DCG@  10: 0.529 | Hits@  10: 0.659\n",
      "DCG@ 100: 0.576 | Hits@ 100: 0.885\n",
      "DCG@ 500: 0.588 | Hits@ 500: 0.980\n",
      "DCG@1000: 0.590 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 1 with CBOW\n",
      "DCG@   1: 0.384 | Hits@   1: 0.384\n",
      "DCG@   5: 0.471 | Hits@   5: 0.550\n",
      "DCG@  10: 0.499 | Hits@  10: 0.634\n",
      "DCG@ 100: 0.542 | Hits@ 100: 0.847\n",
      "DCG@ 500: 0.557 | Hits@ 500: 0.964\n",
      "DCG@1000: 0.561 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 2 with CBOW\n",
      "DCG@   1: 0.385 | Hits@   1: 0.385\n",
      "DCG@   5: 0.479 | Hits@   5: 0.563\n",
      "DCG@  10: 0.502 | Hits@  10: 0.635\n",
      "DCG@ 100: 0.549 | Hits@ 100: 0.869\n",
      "DCG@ 500: 0.563 | Hits@ 500: 0.970\n",
      "DCG@1000: 0.566 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 3 with CBOW\n",
      "DCG@   1: 0.389 | Hits@   1: 0.389\n",
      "DCG@   5: 0.480 | Hits@   5: 0.561\n",
      "DCG@  10: 0.506 | Hits@  10: 0.641\n",
      "DCG@ 100: 0.553 | Hits@ 100: 0.868\n",
      "DCG@ 500: 0.567 | Hits@ 500: 0.980\n",
      "DCG@1000: 0.570 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 4 with CBOW\n",
      "DCG@   1: 0.394 | Hits@   1: 0.394\n",
      "DCG@   5: 0.487 | Hits@   5: 0.570\n",
      "DCG@  10: 0.512 | Hits@  10: 0.647\n",
      "DCG@ 100: 0.558 | Hits@ 100: 0.869\n",
      "DCG@ 500: 0.572 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.574 | Hits@1000: 1.000\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "min_count = 4, window = 5 with CBOW\n",
      "DCG@   1: 0.398 | Hits@   1: 0.398\n",
      "DCG@   5: 0.493 | Hits@   5: 0.576\n",
      "DCG@  10: 0.518 | Hits@  10: 0.655\n",
      "DCG@ 100: 0.564 | Hits@ 100: 0.881\n",
      "DCG@ 500: 0.577 | Hits@ 500: 0.977\n",
      "DCG@1000: 0.579 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 6 with CBOW\n",
      "DCG@   1: 0.407 | Hits@   1: 0.407\n",
      "DCG@   5: 0.498 | Hits@   5: 0.577\n",
      "DCG@  10: 0.524 | Hits@  10: 0.657\n",
      "DCG@ 100: 0.570 | Hits@ 100: 0.879\n",
      "DCG@ 500: 0.583 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.585 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 7 with CBOW\n",
      "DCG@   1: 0.397 | Hits@   1: 0.397\n",
      "DCG@   5: 0.499 | Hits@   5: 0.589\n",
      "DCG@  10: 0.523 | Hits@  10: 0.662\n",
      "DCG@ 100: 0.568 | Hits@ 100: 0.883\n",
      "DCG@ 500: 0.580 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.583 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 8 with CBOW\n",
      "DCG@   1: 0.409 | Hits@   1: 0.409\n",
      "DCG@   5: 0.502 | Hits@   5: 0.585\n",
      "DCG@  10: 0.525 | Hits@  10: 0.654\n",
      "DCG@ 100: 0.573 | Hits@ 100: 0.882\n",
      "DCG@ 500: 0.585 | Hits@ 500: 0.977\n",
      "DCG@1000: 0.587 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 9 with CBOW\n",
      "DCG@   1: 0.410 | Hits@   1: 0.410\n",
      "DCG@   5: 0.507 | Hits@   5: 0.591\n",
      "DCG@  10: 0.530 | Hits@  10: 0.661\n",
      "DCG@ 100: 0.576 | Hits@ 100: 0.885\n",
      "DCG@ 500: 0.588 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.591 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 4, window = 10 with CBOW\n",
      "DCG@   1: 0.416 | Hits@   1: 0.416\n",
      "DCG@   5: 0.509 | Hits@   5: 0.591\n",
      "DCG@  10: 0.533 | Hits@  10: 0.662\n",
      "DCG@ 100: 0.579 | Hits@ 100: 0.886\n",
      "DCG@ 500: 0.591 | Hits@ 500: 0.975\n",
      "DCG@1000: 0.594 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 1 with CBOW\n",
      "DCG@   1: 0.392 | Hits@   1: 0.392\n",
      "DCG@   5: 0.478 | Hits@   5: 0.557\n",
      "DCG@  10: 0.506 | Hits@  10: 0.644\n",
      "DCG@ 100: 0.548 | Hits@ 100: 0.852\n",
      "DCG@ 500: 0.562 | Hits@ 500: 0.964\n",
      "DCG@1000: 0.566 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 2 with CBOW\n",
      "DCG@   1: 0.386 | Hits@   1: 0.386\n",
      "DCG@   5: 0.478 | Hits@   5: 0.559\n",
      "DCG@  10: 0.504 | Hits@  10: 0.640\n",
      "DCG@ 100: 0.550 | Hits@ 100: 0.867\n",
      "DCG@ 500: 0.563 | Hits@ 500: 0.971\n",
      "DCG@1000: 0.567 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 3 with CBOW\n",
      "DCG@   1: 0.384 | Hits@   1: 0.384\n",
      "DCG@   5: 0.483 | Hits@   5: 0.567\n",
      "DCG@  10: 0.507 | Hits@  10: 0.643\n",
      "DCG@ 100: 0.553 | Hits@ 100: 0.865\n",
      "DCG@ 500: 0.568 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.570 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 4 with CBOW\n",
      "DCG@   1: 0.396 | Hits@   1: 0.396\n",
      "DCG@   5: 0.490 | Hits@   5: 0.572\n",
      "DCG@  10: 0.513 | Hits@  10: 0.645\n",
      "DCG@ 100: 0.560 | Hits@ 100: 0.872\n",
      "DCG@ 500: 0.573 | Hits@ 500: 0.974\n",
      "DCG@1000: 0.576 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 5 with CBOW\n",
      "DCG@   1: 0.399 | Hits@   1: 0.399\n",
      "DCG@   5: 0.494 | Hits@   5: 0.577\n",
      "DCG@  10: 0.517 | Hits@  10: 0.650\n",
      "DCG@ 100: 0.564 | Hits@ 100: 0.876\n",
      "DCG@ 500: 0.578 | Hits@ 500: 0.979\n",
      "DCG@1000: 0.580 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 6 with CBOW\n",
      "DCG@   1: 0.405 | Hits@   1: 0.405\n",
      "DCG@   5: 0.497 | Hits@   5: 0.578\n",
      "DCG@  10: 0.524 | Hits@  10: 0.659\n",
      "DCG@ 100: 0.569 | Hits@ 100: 0.880\n",
      "DCG@ 500: 0.582 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.584 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 7 with CBOW\n",
      "DCG@   1: 0.410 | Hits@   1: 0.410\n",
      "DCG@   5: 0.502 | Hits@   5: 0.582\n",
      "DCG@  10: 0.527 | Hits@  10: 0.658\n",
      "DCG@ 100: 0.573 | Hits@ 100: 0.880\n",
      "DCG@ 500: 0.586 | Hits@ 500: 0.979\n",
      "DCG@1000: 0.588 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 8 with CBOW\n",
      "DCG@   1: 0.407 | Hits@   1: 0.407\n",
      "DCG@   5: 0.503 | Hits@   5: 0.585\n",
      "DCG@  10: 0.528 | Hits@  10: 0.664\n",
      "DCG@ 100: 0.573 | Hits@ 100: 0.882\n",
      "DCG@ 500: 0.586 | Hits@ 500: 0.977\n",
      "DCG@1000: 0.588 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 9 with CBOW\n",
      "DCG@   1: 0.411 | Hits@   1: 0.411\n",
      "DCG@   5: 0.505 | Hits@   5: 0.588\n",
      "DCG@  10: 0.530 | Hits@  10: 0.664\n",
      "DCG@ 100: 0.575 | Hits@ 100: 0.879\n",
      "DCG@ 500: 0.588 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.590 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 5, window = 10 with CBOW\n",
      "DCG@   1: 0.414 | Hits@   1: 0.414\n",
      "DCG@   5: 0.509 | Hits@   5: 0.591\n",
      "DCG@  10: 0.535 | Hits@  10: 0.671\n",
      "DCG@ 100: 0.579 | Hits@ 100: 0.885\n",
      "DCG@ 500: 0.591 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.593 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 1 with CBOW\n",
      "DCG@   1: 0.385 | Hits@   1: 0.385\n",
      "DCG@   5: 0.476 | Hits@   5: 0.558\n",
      "DCG@  10: 0.504 | Hits@  10: 0.645\n",
      "DCG@ 100: 0.545 | Hits@ 100: 0.849\n",
      "DCG@ 500: 0.560 | Hits@ 500: 0.965\n",
      "DCG@1000: 0.564 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 2 with CBOW\n",
      "DCG@   1: 0.389 | Hits@   1: 0.389\n",
      "DCG@   5: 0.480 | Hits@   5: 0.562\n",
      "DCG@  10: 0.503 | Hits@  10: 0.634\n",
      "DCG@ 100: 0.551 | Hits@ 100: 0.867\n",
      "DCG@ 500: 0.564 | Hits@ 500: 0.970\n",
      "DCG@1000: 0.568 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 3 with CBOW\n",
      "DCG@   1: 0.384 | Hits@   1: 0.384\n",
      "DCG@   5: 0.481 | Hits@   5: 0.566\n",
      "DCG@  10: 0.508 | Hits@  10: 0.648\n",
      "DCG@ 100: 0.552 | Hits@ 100: 0.864\n",
      "DCG@ 500: 0.567 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.569 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 4 with CBOW\n",
      "DCG@   1: 0.392 | Hits@   1: 0.392\n",
      "DCG@   5: 0.488 | Hits@   5: 0.574\n",
      "DCG@  10: 0.511 | Hits@  10: 0.645\n",
      "DCG@ 100: 0.558 | Hits@ 100: 0.872\n",
      "DCG@ 500: 0.572 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.574 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 5 with CBOW\n",
      "DCG@   1: 0.404 | Hits@   1: 0.404\n",
      "DCG@   5: 0.495 | Hits@   5: 0.578\n",
      "DCG@  10: 0.519 | Hits@  10: 0.652\n",
      "DCG@ 100: 0.565 | Hits@ 100: 0.872\n",
      "DCG@ 500: 0.579 | Hits@ 500: 0.979\n",
      "DCG@1000: 0.581 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 6 with CBOW\n",
      "DCG@   1: 0.401 | Hits@   1: 0.401\n",
      "DCG@   5: 0.496 | Hits@   5: 0.575\n",
      "DCG@  10: 0.523 | Hits@  10: 0.659\n",
      "DCG@ 100: 0.569 | Hits@ 100: 0.882\n",
      "DCG@ 500: 0.581 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.584 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 7 with CBOW\n",
      "DCG@   1: 0.411 | Hits@   1: 0.411\n",
      "DCG@   5: 0.505 | Hits@   5: 0.587\n",
      "DCG@  10: 0.528 | Hits@  10: 0.659\n",
      "DCG@ 100: 0.573 | Hits@ 100: 0.875\n",
      "DCG@ 500: 0.586 | Hits@ 500: 0.975\n",
      "DCG@1000: 0.589 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 8 with CBOW\n",
      "DCG@   1: 0.406 | Hits@   1: 0.406\n",
      "DCG@   5: 0.502 | Hits@   5: 0.584\n",
      "DCG@  10: 0.525 | Hits@  10: 0.654\n",
      "DCG@ 100: 0.573 | Hits@ 100: 0.884\n",
      "DCG@ 500: 0.585 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.588 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 9 with CBOW\n",
      "DCG@   1: 0.416 | Hits@   1: 0.416\n",
      "DCG@   5: 0.510 | Hits@   5: 0.593\n",
      "DCG@  10: 0.534 | Hits@  10: 0.665\n",
      "DCG@ 100: 0.579 | Hits@ 100: 0.885\n",
      "DCG@ 500: 0.591 | Hits@ 500: 0.981\n",
      "DCG@1000: 0.593 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 6, window = 10 with CBOW\n",
      "DCG@   1: 0.417 | Hits@   1: 0.417\n",
      "DCG@   5: 0.510 | Hits@   5: 0.592\n",
      "DCG@  10: 0.533 | Hits@  10: 0.662\n",
      "DCG@ 100: 0.579 | Hits@ 100: 0.884\n",
      "DCG@ 500: 0.592 | Hits@ 500: 0.981\n",
      "DCG@1000: 0.594 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 1 with CBOW\n",
      "DCG@   1: 0.391 | Hits@   1: 0.391\n",
      "DCG@   5: 0.476 | Hits@   5: 0.553\n",
      "DCG@  10: 0.504 | Hits@  10: 0.639\n",
      "DCG@ 100: 0.547 | Hits@ 100: 0.850\n",
      "DCG@ 500: 0.562 | Hits@ 500: 0.964\n",
      "DCG@1000: 0.565 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 2 with CBOW\n",
      "DCG@   1: 0.383 | Hits@   1: 0.383\n",
      "DCG@   5: 0.478 | Hits@   5: 0.563\n",
      "DCG@  10: 0.504 | Hits@  10: 0.643\n",
      "DCG@ 100: 0.549 | Hits@ 100: 0.871\n",
      "DCG@ 500: 0.562 | Hits@ 500: 0.972\n",
      "DCG@1000: 0.565 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 3 with CBOW\n",
      "DCG@   1: 0.385 | Hits@   1: 0.385\n",
      "DCG@   5: 0.481 | Hits@   5: 0.565\n",
      "DCG@  10: 0.506 | Hits@  10: 0.645\n",
      "DCG@ 100: 0.552 | Hits@ 100: 0.866\n",
      "DCG@ 500: 0.566 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.569 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 4 with CBOW\n",
      "DCG@   1: 0.396 | Hits@   1: 0.396\n",
      "DCG@   5: 0.492 | Hits@   5: 0.576\n",
      "DCG@  10: 0.515 | Hits@  10: 0.648\n",
      "DCG@ 100: 0.561 | Hits@ 100: 0.874\n",
      "DCG@ 500: 0.575 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.577 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 5 with CBOW\n",
      "DCG@   1: 0.399 | Hits@   1: 0.399\n",
      "DCG@   5: 0.494 | Hits@   5: 0.575\n",
      "DCG@  10: 0.517 | Hits@  10: 0.647\n",
      "DCG@ 100: 0.564 | Hits@ 100: 0.877\n",
      "DCG@ 500: 0.577 | Hits@ 500: 0.977\n",
      "DCG@1000: 0.580 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 6 with CBOW\n",
      "DCG@   1: 0.397 | Hits@   1: 0.397\n",
      "DCG@   5: 0.496 | Hits@   5: 0.582\n",
      "DCG@  10: 0.520 | Hits@  10: 0.659\n",
      "DCG@ 100: 0.565 | Hits@ 100: 0.878\n",
      "DCG@ 500: 0.578 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.581 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 7 with CBOW\n",
      "DCG@   1: 0.403 | Hits@   1: 0.403\n",
      "DCG@   5: 0.501 | Hits@   5: 0.587\n",
      "DCG@  10: 0.524 | Hits@  10: 0.659\n",
      "DCG@ 100: 0.571 | Hits@ 100: 0.883\n",
      "DCG@ 500: 0.583 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.585 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 8 with CBOW\n",
      "DCG@   1: 0.411 | Hits@   1: 0.411\n",
      "DCG@   5: 0.506 | Hits@   5: 0.589\n",
      "DCG@  10: 0.531 | Hits@  10: 0.667\n",
      "DCG@ 100: 0.575 | Hits@ 100: 0.881\n",
      "DCG@ 500: 0.588 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.590 | Hits@1000: 1.000\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "min_count = 7, window = 9 with CBOW\n",
      "DCG@   1: 0.421 | Hits@   1: 0.421\n",
      "DCG@   5: 0.509 | Hits@   5: 0.587\n",
      "DCG@  10: 0.536 | Hits@  10: 0.669\n",
      "DCG@ 100: 0.580 | Hits@ 100: 0.884\n",
      "DCG@ 500: 0.592 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.595 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 7, window = 10 with CBOW\n",
      "DCG@   1: 0.412 | Hits@   1: 0.412\n",
      "DCG@   5: 0.507 | Hits@   5: 0.587\n",
      "DCG@  10: 0.532 | Hits@  10: 0.663\n",
      "DCG@ 100: 0.577 | Hits@ 100: 0.882\n",
      "DCG@ 500: 0.590 | Hits@ 500: 0.979\n",
      "DCG@1000: 0.592 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 1 with CBOW\n",
      "DCG@   1: 0.388 | Hits@   1: 0.388\n",
      "DCG@   5: 0.477 | Hits@   5: 0.558\n",
      "DCG@  10: 0.506 | Hits@  10: 0.647\n",
      "DCG@ 100: 0.547 | Hits@ 100: 0.852\n",
      "DCG@ 500: 0.562 | Hits@ 500: 0.964\n",
      "DCG@1000: 0.565 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 2 with CBOW\n",
      "DCG@   1: 0.389 | Hits@   1: 0.389\n",
      "DCG@   5: 0.483 | Hits@   5: 0.566\n",
      "DCG@  10: 0.507 | Hits@  10: 0.641\n",
      "DCG@ 100: 0.553 | Hits@ 100: 0.868\n",
      "DCG@ 500: 0.566 | Hits@ 500: 0.973\n",
      "DCG@1000: 0.569 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 3 with CBOW\n",
      "DCG@   1: 0.387 | Hits@   1: 0.387\n",
      "DCG@   5: 0.483 | Hits@   5: 0.568\n",
      "DCG@  10: 0.507 | Hits@  10: 0.643\n",
      "DCG@ 100: 0.552 | Hits@ 100: 0.863\n",
      "DCG@ 500: 0.567 | Hits@ 500: 0.975\n",
      "DCG@1000: 0.570 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 4 with CBOW\n",
      "DCG@   1: 0.397 | Hits@   1: 0.397\n",
      "DCG@   5: 0.491 | Hits@   5: 0.573\n",
      "DCG@  10: 0.514 | Hits@  10: 0.645\n",
      "DCG@ 100: 0.560 | Hits@ 100: 0.867\n",
      "DCG@ 500: 0.575 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.577 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 5 with CBOW\n",
      "DCG@   1: 0.399 | Hits@   1: 0.399\n",
      "DCG@   5: 0.495 | Hits@   5: 0.579\n",
      "DCG@  10: 0.517 | Hits@  10: 0.647\n",
      "DCG@ 100: 0.564 | Hits@ 100: 0.874\n",
      "DCG@ 500: 0.578 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.580 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 6 with CBOW\n",
      "DCG@   1: 0.400 | Hits@   1: 0.400\n",
      "DCG@   5: 0.498 | Hits@   5: 0.584\n",
      "DCG@  10: 0.519 | Hits@  10: 0.651\n",
      "DCG@ 100: 0.566 | Hits@ 100: 0.874\n",
      "DCG@ 500: 0.580 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.582 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 7 with CBOW\n",
      "DCG@   1: 0.406 | Hits@   1: 0.406\n",
      "DCG@   5: 0.502 | Hits@   5: 0.583\n",
      "DCG@  10: 0.526 | Hits@  10: 0.658\n",
      "DCG@ 100: 0.572 | Hits@ 100: 0.882\n",
      "DCG@ 500: 0.585 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.587 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 8 with CBOW\n",
      "DCG@   1: 0.405 | Hits@   1: 0.405\n",
      "DCG@   5: 0.502 | Hits@   5: 0.584\n",
      "DCG@  10: 0.528 | Hits@  10: 0.665\n",
      "DCG@ 100: 0.573 | Hits@ 100: 0.881\n",
      "DCG@ 500: 0.586 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.588 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 9 with CBOW\n",
      "DCG@   1: 0.418 | Hits@   1: 0.418\n",
      "DCG@   5: 0.508 | Hits@   5: 0.588\n",
      "DCG@  10: 0.532 | Hits@  10: 0.664\n",
      "DCG@ 100: 0.577 | Hits@ 100: 0.878\n",
      "DCG@ 500: 0.590 | Hits@ 500: 0.980\n",
      "DCG@1000: 0.592 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 8, window = 10 with CBOW\n",
      "DCG@   1: 0.416 | Hits@   1: 0.416\n",
      "DCG@   5: 0.510 | Hits@   5: 0.593\n",
      "DCG@  10: 0.534 | Hits@  10: 0.667\n",
      "DCG@ 100: 0.578 | Hits@ 100: 0.881\n",
      "DCG@ 500: 0.591 | Hits@ 500: 0.980\n",
      "DCG@1000: 0.593 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 1 with CBOW\n",
      "DCG@   1: 0.393 | Hits@   1: 0.393\n",
      "DCG@   5: 0.481 | Hits@   5: 0.563\n",
      "DCG@  10: 0.508 | Hits@  10: 0.646\n",
      "DCG@ 100: 0.550 | Hits@ 100: 0.854\n",
      "DCG@ 500: 0.564 | Hits@ 500: 0.965\n",
      "DCG@1000: 0.568 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 2 with CBOW\n",
      "DCG@   1: 0.391 | Hits@   1: 0.391\n",
      "DCG@   5: 0.480 | Hits@   5: 0.559\n",
      "DCG@  10: 0.507 | Hits@  10: 0.642\n",
      "DCG@ 100: 0.553 | Hits@ 100: 0.868\n",
      "DCG@ 500: 0.566 | Hits@ 500: 0.972\n",
      "DCG@1000: 0.569 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 3 with CBOW\n",
      "DCG@   1: 0.385 | Hits@   1: 0.385\n",
      "DCG@   5: 0.482 | Hits@   5: 0.567\n",
      "DCG@  10: 0.507 | Hits@  10: 0.644\n",
      "DCG@ 100: 0.553 | Hits@ 100: 0.870\n",
      "DCG@ 500: 0.567 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.569 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 4 with CBOW\n",
      "DCG@   1: 0.396 | Hits@   1: 0.396\n",
      "DCG@   5: 0.490 | Hits@   5: 0.573\n",
      "DCG@  10: 0.513 | Hits@  10: 0.644\n",
      "DCG@ 100: 0.561 | Hits@ 100: 0.873\n",
      "DCG@ 500: 0.574 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.577 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 5 with CBOW\n",
      "DCG@   1: 0.401 | Hits@   1: 0.401\n",
      "DCG@   5: 0.493 | Hits@   5: 0.574\n",
      "DCG@  10: 0.517 | Hits@  10: 0.652\n",
      "DCG@ 100: 0.564 | Hits@ 100: 0.876\n",
      "DCG@ 500: 0.577 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.579 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 6 with CBOW\n",
      "DCG@   1: 0.401 | Hits@   1: 0.401\n",
      "DCG@   5: 0.497 | Hits@   5: 0.581\n",
      "DCG@  10: 0.519 | Hits@  10: 0.651\n",
      "DCG@ 100: 0.567 | Hits@ 100: 0.879\n",
      "DCG@ 500: 0.580 | Hits@ 500: 0.977\n",
      "DCG@1000: 0.582 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 7 with CBOW\n",
      "DCG@   1: 0.413 | Hits@   1: 0.413\n",
      "DCG@   5: 0.503 | Hits@   5: 0.582\n",
      "DCG@  10: 0.530 | Hits@  10: 0.667\n",
      "DCG@ 100: 0.575 | Hits@ 100: 0.883\n",
      "DCG@ 500: 0.587 | Hits@ 500: 0.975\n",
      "DCG@1000: 0.589 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 8 with CBOW\n",
      "DCG@   1: 0.410 | Hits@   1: 0.410\n",
      "DCG@   5: 0.504 | Hits@   5: 0.586\n",
      "DCG@  10: 0.529 | Hits@  10: 0.664\n",
      "DCG@ 100: 0.573 | Hits@ 100: 0.876\n",
      "DCG@ 500: 0.586 | Hits@ 500: 0.979\n",
      "DCG@1000: 0.589 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 9 with CBOW\n",
      "DCG@   1: 0.409 | Hits@   1: 0.409\n",
      "DCG@   5: 0.504 | Hits@   5: 0.585\n",
      "DCG@  10: 0.530 | Hits@  10: 0.665\n",
      "DCG@ 100: 0.575 | Hits@ 100: 0.880\n",
      "DCG@ 500: 0.587 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.590 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 9, window = 10 with CBOW\n",
      "DCG@   1: 0.417 | Hits@   1: 0.417\n",
      "DCG@   5: 0.511 | Hits@   5: 0.594\n",
      "DCG@  10: 0.533 | Hits@  10: 0.662\n",
      "DCG@ 100: 0.580 | Hits@ 100: 0.883\n",
      "DCG@ 500: 0.592 | Hits@ 500: 0.979\n",
      "DCG@1000: 0.594 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 1 with CBOW\n",
      "DCG@   1: 0.392 | Hits@   1: 0.392\n",
      "DCG@   5: 0.481 | Hits@   5: 0.563\n",
      "DCG@  10: 0.508 | Hits@  10: 0.647\n",
      "DCG@ 100: 0.550 | Hits@ 100: 0.851\n",
      "DCG@ 500: 0.564 | Hits@ 500: 0.963\n",
      "DCG@1000: 0.568 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 2 with CBOW\n",
      "DCG@   1: 0.389 | Hits@   1: 0.389\n",
      "DCG@   5: 0.477 | Hits@   5: 0.556\n",
      "DCG@  10: 0.504 | Hits@  10: 0.637\n",
      "DCG@ 100: 0.550 | Hits@ 100: 0.864\n",
      "DCG@ 500: 0.564 | Hits@ 500: 0.972\n",
      "DCG@1000: 0.567 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 3 with CBOW\n",
      "DCG@   1: 0.389 | Hits@   1: 0.389\n",
      "DCG@   5: 0.481 | Hits@   5: 0.563\n",
      "DCG@  10: 0.507 | Hits@  10: 0.642\n",
      "DCG@ 100: 0.552 | Hits@ 100: 0.862\n",
      "DCG@ 500: 0.567 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.570 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 4 with CBOW\n",
      "DCG@   1: 0.396 | Hits@   1: 0.396\n",
      "DCG@   5: 0.488 | Hits@   5: 0.570\n",
      "DCG@  10: 0.514 | Hits@  10: 0.650\n",
      "DCG@ 100: 0.559 | Hits@ 100: 0.867\n",
      "DCG@ 500: 0.573 | Hits@ 500: 0.976\n",
      "DCG@1000: 0.576 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 5 with CBOW\n",
      "DCG@   1: 0.395 | Hits@   1: 0.395\n",
      "DCG@   5: 0.492 | Hits@   5: 0.578\n",
      "DCG@  10: 0.515 | Hits@  10: 0.648\n",
      "DCG@ 100: 0.563 | Hits@ 100: 0.880\n",
      "DCG@ 500: 0.575 | Hits@ 500: 0.974\n",
      "DCG@1000: 0.578 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 6 with CBOW\n",
      "DCG@   1: 0.406 | Hits@   1: 0.406\n",
      "DCG@   5: 0.499 | Hits@   5: 0.582\n",
      "DCG@  10: 0.522 | Hits@  10: 0.652\n",
      "DCG@ 100: 0.569 | Hits@ 100: 0.880\n",
      "DCG@ 500: 0.582 | Hits@ 500: 0.973\n",
      "DCG@1000: 0.584 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 7 with CBOW\n",
      "DCG@   1: 0.412 | Hits@   1: 0.412\n",
      "DCG@   5: 0.501 | Hits@   5: 0.577\n",
      "DCG@  10: 0.526 | Hits@  10: 0.653\n",
      "DCG@ 100: 0.573 | Hits@ 100: 0.880\n",
      "DCG@ 500: 0.586 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.588 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 8 with CBOW\n",
      "DCG@   1: 0.405 | Hits@   1: 0.405\n",
      "DCG@   5: 0.503 | Hits@   5: 0.586\n",
      "DCG@  10: 0.527 | Hits@  10: 0.661\n",
      "DCG@ 100: 0.573 | Hits@ 100: 0.879\n",
      "DCG@ 500: 0.586 | Hits@ 500: 0.978\n",
      "DCG@1000: 0.588 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 9 with CBOW\n",
      "DCG@   1: 0.415 | Hits@   1: 0.415\n",
      "DCG@   5: 0.509 | Hits@   5: 0.592\n",
      "DCG@  10: 0.533 | Hits@  10: 0.665\n",
      "DCG@ 100: 0.578 | Hits@ 100: 0.881\n",
      "DCG@ 500: 0.591 | Hits@ 500: 0.980\n",
      "DCG@1000: 0.593 | Hits@1000: 1.000\n",
      "\n",
      "min_count = 10, window = 10 with CBOW\n",
      "DCG@   1: 0.416 | Hits@   1: 0.416\n",
      "DCG@   5: 0.513 | Hits@   5: 0.597\n",
      "DCG@  10: 0.535 | Hits@  10: 0.666\n",
      "DCG@ 100: 0.581 | Hits@ 100: 0.886\n",
      "DCG@ 500: 0.593 | Hits@ 500: 0.979\n",
      "DCG@1000: 0.595 | Hits@1000: 1.000\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>DCG@1</th>\n",
       "      <th>DCG@5</th>\n",
       "      <th>DCG@10</th>\n",
       "      <th>DCG@100</th>\n",
       "      <th>DCG@500</th>\n",
       "      <th>DCG@1000</th>\n",
       "      <th>Hits@1</th>\n",
       "      <th>Hits@5</th>\n",
       "      <th>Hits@10</th>\n",
       "      <th>Hits@100</th>\n",
       "      <th>Hits@500</th>\n",
       "      <th>Hits@1000</th>\n",
       "      <th>SUM</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Min_1/Win_1</th>\n",
       "      <td>0.358</td>\n",
       "      <td>0.447</td>\n",
       "      <td>0.476</td>\n",
       "      <td>0.518</td>\n",
       "      <td>0.534</td>\n",
       "      <td>0.540</td>\n",
       "      <td>0.358</td>\n",
       "      <td>0.525</td>\n",
       "      <td>0.614</td>\n",
       "      <td>0.824</td>\n",
       "      <td>0.949</td>\n",
       "      <td>1.0</td>\n",
       "      <td>7.143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Min_1/Win_2</th>\n",
       "      <td>0.367</td>\n",
       "      <td>0.464</td>\n",
       "      <td>0.485</td>\n",
       "      <td>0.532</td>\n",
       "      <td>0.547</td>\n",
       "      <td>0.551</td>\n",
       "      <td>0.367</td>\n",
       "      <td>0.550</td>\n",
       "      <td>0.617</td>\n",
       "      <td>0.850</td>\n",
       "      <td>0.968</td>\n",
       "      <td>1.0</td>\n",
       "      <td>7.298</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Min_1/Win_3</th>\n",
       "      <td>0.377</td>\n",
       "      <td>0.470</td>\n",
       "      <td>0.494</td>\n",
       "      <td>0.541</td>\n",
       "      <td>0.556</td>\n",
       "      <td>0.559</td>\n",
       "      <td>0.377</td>\n",
       "      <td>0.554</td>\n",
       "      <td>0.630</td>\n",
       "      <td>0.859</td>\n",
       "      <td>0.974</td>\n",
       "      <td>1.0</td>\n",
       "      <td>7.391</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Min_1/Win_4</th>\n",
       "      <td>0.385</td>\n",
       "      <td>0.475</td>\n",
       "      <td>0.500</td>\n",
       "      <td>0.548</td>\n",
       "      <td>0.563</td>\n",
       "      <td>0.565</td>\n",
       "      <td>0.385</td>\n",
       "      <td>0.554</td>\n",
       "      <td>0.634</td>\n",
       "      <td>0.866</td>\n",
       "      <td>0.976</td>\n",
       "      <td>1.0</td>\n",
       "      <td>7.451</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Min_1/Win_5</th>\n",
       "      <td>0.393</td>\n",
       "      <td>0.479</td>\n",
       "      <td>0.509</td>\n",
       "      <td>0.554</td>\n",
       "      <td>0.568</td>\n",
       "      <td>0.571</td>\n",
       "      <td>0.393</td>\n",
       "      <td>0.555</td>\n",
       "      <td>0.648</td>\n",
       "      <td>0.866</td>\n",
       "      <td>0.977</td>\n",
       "      <td>1.0</td>\n",
       "      <td>7.513</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Min_10/Win_6</th>\n",
       "      <td>0.406</td>\n",
       "      <td>0.499</td>\n",
       "      <td>0.522</td>\n",
       "      <td>0.569</td>\n",
       "      <td>0.582</td>\n",
       "      <td>0.584</td>\n",
       "      <td>0.406</td>\n",
       "      <td>0.582</td>\n",
       "      <td>0.652</td>\n",
       "      <td>0.880</td>\n",
       "      <td>0.973</td>\n",
       "      <td>1.0</td>\n",
       "      <td>7.655</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Min_10/Win_7</th>\n",
       "      <td>0.412</td>\n",
       "      <td>0.501</td>\n",
       "      <td>0.526</td>\n",
       "      <td>0.573</td>\n",
       "      <td>0.586</td>\n",
       "      <td>0.588</td>\n",
       "      <td>0.412</td>\n",
       "      <td>0.577</td>\n",
       "      <td>0.653</td>\n",
       "      <td>0.880</td>\n",
       "      <td>0.978</td>\n",
       "      <td>1.0</td>\n",
       "      <td>7.686</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Min_10/Win_8</th>\n",
       "      <td>0.405</td>\n",
       "      <td>0.503</td>\n",
       "      <td>0.527</td>\n",
       "      <td>0.573</td>\n",
       "      <td>0.586</td>\n",
       "      <td>0.588</td>\n",
       "      <td>0.405</td>\n",
       "      <td>0.586</td>\n",
       "      <td>0.661</td>\n",
       "      <td>0.879</td>\n",
       "      <td>0.978</td>\n",
       "      <td>1.0</td>\n",
       "      <td>7.691</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Min_10/Win_9</th>\n",
       "      <td>0.415</td>\n",
       "      <td>0.509</td>\n",
       "      <td>0.533</td>\n",
       "      <td>0.578</td>\n",
       "      <td>0.591</td>\n",
       "      <td>0.593</td>\n",
       "      <td>0.415</td>\n",
       "      <td>0.592</td>\n",
       "      <td>0.665</td>\n",
       "      <td>0.881</td>\n",
       "      <td>0.980</td>\n",
       "      <td>1.0</td>\n",
       "      <td>7.752</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Min_10/Win_10</th>\n",
       "      <td>0.416</td>\n",
       "      <td>0.513</td>\n",
       "      <td>0.535</td>\n",
       "      <td>0.581</td>\n",
       "      <td>0.593</td>\n",
       "      <td>0.595</td>\n",
       "      <td>0.416</td>\n",
       "      <td>0.597</td>\n",
       "      <td>0.666</td>\n",
       "      <td>0.886</td>\n",
       "      <td>0.979</td>\n",
       "      <td>1.0</td>\n",
       "      <td>7.777</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>100 rows × 13 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "               DCG@1  DCG@5  DCG@10  DCG@100  DCG@500  DCG@1000  Hits@1  \\\n",
       "Min_1/Win_1    0.358  0.447   0.476    0.518    0.534     0.540   0.358   \n",
       "Min_1/Win_2    0.367  0.464   0.485    0.532    0.547     0.551   0.367   \n",
       "Min_1/Win_3    0.377  0.470   0.494    0.541    0.556     0.559   0.377   \n",
       "Min_1/Win_4    0.385  0.475   0.500    0.548    0.563     0.565   0.385   \n",
       "Min_1/Win_5    0.393  0.479   0.509    0.554    0.568     0.571   0.393   \n",
       "...              ...    ...     ...      ...      ...       ...     ...   \n",
       "Min_10/Win_6   0.406  0.499   0.522    0.569    0.582     0.584   0.406   \n",
       "Min_10/Win_7   0.412  0.501   0.526    0.573    0.586     0.588   0.412   \n",
       "Min_10/Win_8   0.405  0.503   0.527    0.573    0.586     0.588   0.405   \n",
       "Min_10/Win_9   0.415  0.509   0.533    0.578    0.591     0.593   0.415   \n",
       "Min_10/Win_10  0.416  0.513   0.535    0.581    0.593     0.595   0.416   \n",
       "\n",
       "               Hits@5  Hits@10  Hits@100  Hits@500  Hits@1000    SUM  \n",
       "Min_1/Win_1     0.525    0.614     0.824     0.949        1.0  7.143  \n",
       "Min_1/Win_2     0.550    0.617     0.850     0.968        1.0  7.298  \n",
       "Min_1/Win_3     0.554    0.630     0.859     0.974        1.0  7.391  \n",
       "Min_1/Win_4     0.554    0.634     0.866     0.976        1.0  7.451  \n",
       "Min_1/Win_5     0.555    0.648     0.866     0.977        1.0  7.513  \n",
       "...               ...      ...       ...       ...        ...    ...  \n",
       "Min_10/Win_6    0.582    0.652     0.880     0.973        1.0  7.655  \n",
       "Min_10/Win_7    0.577    0.653     0.880     0.978        1.0  7.686  \n",
       "Min_10/Win_8    0.586    0.661     0.879     0.978        1.0  7.691  \n",
       "Min_10/Win_9    0.592    0.665     0.881     0.980        1.0  7.752  \n",
       "Min_10/Win_10   0.597    0.666     0.886     0.979        1.0  7.777  \n",
       "\n",
       "[100 rows x 13 columns]"
      ]
     },
     "execution_count": 272,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dcg_history = []\n",
    "hits_history = []\n",
    "for min_count in range(1,11):\n",
    "    for window in range(1,11):\n",
    "        embeddings_trained = Word2Vec(words,                    # data for model to train on\n",
    "                                      vector_size=200,          # embedding vector size\n",
    "                                      min_count=min_count,      # Ignores all words with total frequency lower than this. Consider words that occured at least 5 times\n",
    "                                      window=window,            # Maximum distance between the current and predicted word within a sentence\n",
    "                                      workers=4,                # Use these many worker threads to train the model\n",
    "                                      sg=0).wv                  # sg=0 - CBOW, sg=1 - Skip-Gram\n",
    "        wv_ranking = []\n",
    "        max_validation_examples = 1000\n",
    "        tokenizer = MyTokenizer()\n",
    "        for i, line in enumerate(validation_data):\n",
    "            if i == max_validation_examples:\n",
    "                break\n",
    "            q, *ex = line\n",
    "            ranks = rank_candidates(q, ex, embeddings_trained, tokenizer)\n",
    "            wv_ranking.append([r[0] for r in ranks].index(0) + 1)\n",
    "        print(f'min_count = {min_count}, window = {window} with CBOW')\n",
    "        for k in [1, 5, 10, 100, 500, 1000]:\n",
    "            print(\"DCG@%4d: %.3f | Hits@%4d: %.3f\" % (k, dcg_score(wv_ranking, k), k, hits_count(wv_ranking, k)))\n",
    "            dcg_history.append(dcg_score(wv_ranking, k))\n",
    "            hits_history.append(hits_count(wv_ranking, k))\n",
    "        print()\n",
    "\n",
    "\n",
    "df2 = pd.DataFrame()\n",
    "i = m = n = 0\n",
    "for row in row_list:\n",
    "    for dcg in columns[:6]:\n",
    "        df2.loc[row,dcg] = dcg_history[m]\n",
    "        m+=1\n",
    "    for hits in columns[6:]:\n",
    "        df2.loc[row,hits] = hits_history[n]\n",
    "        n+=1\n",
    "    i+=6\n",
    "df2['SUM'] = df2.sum(axis=1)\n",
    "\n",
    "df2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 277,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA4UAAAGXCAYAAAAEUfUZAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/MnkTPAAAACXBIWXMAAAsTAAALEwEAmpwYAAEAAElEQVR4nOxdd3gTR/p+5V7lisFywab3ToBUUi+kN9KOJCS5S/sluTh36Xc2pJCEFF9Iu5AC6ZQUQgqBFEoINi64997Wtix3SZZlSfP7YyUhS1skrWTJsO/z6DHo3fne+ebb2Z3R7M4nIYRAhAgRIkSIECFChAgRIkScnvDxdAVEiBAhQoQIESJEiBAhQoTnIE4KRYgQIUKECBEiRIgQIeI0hjgpFCFChAgRIkSIECFChIjTGOKkUIQIESJEiBAhQoQIESJOY4iTQhEiRIgQIUKECBEiRIg4jSFOCkWIECFChAgRIkSIECHiNIafpysgQoQIESJOPchkaUcAzAMgo6hMjafrI8K1kMnSNgDIAPB3isr8wMPVESFChAgRAiFOCkWIEHHaQSZLs07QqgEwAKAewDEAn1JUZiGPjVAA6wFcDWAhgCgAQwBqARwB8BVFZf7JUM4XwHUAbgBwBoA40NdiBYAyAAcB7KKozDoe/TMB/BXA+QASAAQA6AZQBOBnANspKnOQy4aFrZlG7R0UlbnOnjI89mYAOAfAm/ZMCGWytBQADQD+pKjMszmOawWQQFGZEovvVoNusyaKykyx+P55AM8AuJOiMrc74wdHPeYDuB3AxQCSAYSCbvsKAL8A+JCiMrvstDUBQBWAci7f3QmL/pBKUZmNxu8uAu3LxxSVuX4M6+JQv5LJ0rYDuMPKjA6AEnR//hLAForK1LLoLQJwP4DVABKNX7cBOATgbYrKLLI6Pg5AB4AWisqczGDvSgB7AXRSVOYkBv5iAAcAfERRmXczNoIIESJEeADi46MiRIg4XaEE8Jrx8wGAwwAmAHgUQIFMlvaWTJbmz1RQJktbCaASwFsAZoMePL8LYA+AIACPADgqk6U9Z1VuOoA8ALtATygKAHwI4DPQk7llADYBqJHJ0tayaMtksrSfAfwJevDcDGCn0YejAGYB2AKgWSZLu5XNeZksbapMlva2TJb2DYB8AL5sxzqBu4x/x2IFqQV0DLe6W0gmS4uQydI+Bx2rhwH0AvgGwHugJ6YyAC8CaJLJ0h7lsBMlk6W9KZOl7QR9HkW5u+48MPWDfk9Wwtl+ZcRunPTjHdD9eTqAVwB8yqL3POg++HcAcgCfGD9yAH8DfR14wbIMRWXKQcc/WSZLm8Zg9kLj34kyWdpcDv4AU51EiBAhwlMQVwpFiBBxuqKfojL/Zf2lTJa2CvQg//9AD/Kvs+JnA9gPIAT0BPJNisrUWR2zCMALAMItvksCvfogA/AygGcpKlNtVc4XwEUA/g0ggqFus4w2/AE8CHq1YYjhuJWgB8Ofy2RpsRSVuYXB/yQADzB8LwgyWZof6JWbXIrKLHa1fWsYV1Rt4uhqyGRpE0GvVCUBSAcdd5tJlEyWNgf0xPA1mSwtkaIymSaHEaDj5xVg6gdjDWf7lQW2UlTmr1bHzwA9yVwrk6XdYxkvmSztEdAryc0ArqOozHyrsstBT/iflsnSFBSVmWlBHwCwCHRfrbWqxwUAKND9/ALQK/DWPAHwG4MPIkSIEOExiJNCESJEiLAARWVmyWRpKwBkAbhWJku7i6IyP7I45A0AUgBPWQ0ULW0UArjc+FikCVtADxQ3UVTmMyzl9KAHxvtlsrRRk0KZLC0MwE+gV3MuoqjMFpks7WqZLC0DwBwA7aBXS/4JemXkfABfAHhdJks7ZD1Bo6jMQwAkRturQa90uQKXA5gE+n0zt8O4WlMD4DBFZa42fncIwHnGQ7bJZGnbLIqkUlRmo0yWZlp5ugnATNBPzjSCXoH9gKIysyw0fAF8ByASwAqKyiyRydLOlsnSXgS9utsD4AfQjzxWUlTmapks7TUAj8pkab9SVOZPlnU2PqJpavsU0I/OOuP7TNCTnoMUlXmBFZcP+p3OOKvJ0GugJ11/oajMA8bvRj2Wa/G+IADcIZOlWT6eafM4rkyWdh3oHxjOAD2uyALwT75HsK3gbL9iBUVlVstkaf2g42Z+ZFwmS4sG8BwALYDLKSqzlKFsrkyWdgWAXAAbZbK0bRSV2Wek9wN4HPSq3/8s7E4A3eZPAviPkX/Tgo8AsARAAUVlKuzxQYQIESLGCuLjoyJEiBBhBePq28PG/z5k+t44GL0Y9Pt/jANXKzuNxnKJAK4xlnvezjpYr0I9BXqydZlxQngHgG+N330JegXxehiv68ZVlvUAugA8bY+mi3A3ABWAHWOoaY2dALKN//4ZJx8rfA1Av0yW5gPgR9Arej4APgb92OAggDsBPGFl7y7QE57rjBPCCwH8DnoC8DXoyfqZACZalHkMQCHoVUW3gKIyq0C//7bCuEILwHyeLgH9nulVVsVWgZ4MHeUwfQz0DwwAvdJl2X7WE6jnjccSANtBTwgvAP3DRpg9fjjbr+ywuxr0hPArisocsKCuBxAG4BumCaGFThHoPhaO0U8MHAWgBnC+TJYmsfj+AtCT/QOgV5XPM55rJpwH+jHtX+ypvwgRIkSMJcSVQhEiRIhgxh+gJ1SLZLK0CcaNQ84xcr9RVOawA7ZWG//+yvS4Jx+MK1X3AHiLojLrZLK0YACvA6gDvXLVY3GseUWEojLVMlnae6BXrPwpKnPEUW0H6xkP4DLQG/UM8B3PgBSZLO1VDt7mkVomUFTmuzJZWgKAlQB2MqxsXQB6AL8P9EqR5SrSdJxcZTThAYze4OQt0O+kLqaozCaLso0WdTDIZGlvgF6pjKeozHZ76u4EfgO96c0SADnG764DoDfW8XoY36mTydICjMcdt3502RIUlXlAJkszAFgLII/n8dI20D9UnDB9IZOlfQZ6E6TbQL8TyAdn+5Ul7pHJ0i41/tsfwAzQj3fuxMl3XE1YZdKzw+6vAG4EcDaAjwCAojK1MlnaYQBrACwGYPL9Apzc7OlX0H1hCej3iE08IE4KRYgQ4YUQJ4UiRIgQwQCKyiQyWVoz6M1nEkFPEOONdBNrQWbIjH8bbQhZ2jIANzOU2UVRmaZB/lIAsaAHuAC9KhUNIMNyQsiCg6AfBZwCepdLd2I96JUQZzeYSQD9+Ku7EWn8O2A5IQQAisqsAf04KgDzbpOLYFzhlcnSpoLezOc1ywkhC0yP5C4A/XivO2CaFJ6N0ZPCw6DPt1tlsrQwispUgj6PAkGvcroK71pOCI34AfSkcKadNpztV5Zg3JgJ9Ar9kEyW9iBFZaqM35l2BW21w67pGOudRPeDnhReiNGTwkPGa4dpwnkhRk8Kh8C9SitChAgRHoE4KRQhQoQI98M08WDa4XMemCdClTg5yE8y/jVN6kyPKVpvcsGEDuPfaDuOFYq7QL9TZ5OKw07YlZLCSduWOAygD8BNxvfAvgf97lgxQxoPb2970+TjbNDvj04CvRL2IOhJ4V0ArgD9OK9phcyVk0ImdBv/xrhZxxIXmzaaMT7SGQx6Mv4P0D9WxAMwrSRKmAywwDp9jQmm3UMvBPCKTJaWDGAagP8avy8BvYvpBQBetnjf8BcBq6EiRIgQ4TaI7xSKECFCBAOMA8tk439NKxhy498UB82ZVommWhMUlbmdojIlpg+AjQzlTYNY0wDVVI9IO7RNExK7chY6C+P7W9NAp9jwalBUZjfoRxZ/AnAu6PfYjgJQyGRpu2WyNJnF4V7d9hSV2QZ6wnqW8atrjX+/Af0IYy/oR0gBelKoxsn3Ld0F02PK9k6+nO1XjKCoTEJRmWqKyswGcCvo9yL/Yty9FAA6jX/t+YHBlLuww/JLisqsAJ0O5RzjY7mmR0N/N9XB+O+zjfz5OPm+oQgRIkR4HcRJoQgRIkQw40zQj44es3hE85jx70XG9/rsxWHj3wtlsjSm7fT5QBn/TjH+rQBgwMkJAADA+B6dNWaA3vjF3Y+O3g16MvCxm3VcAorKLKWozMtB5whcDXo3yUIAN4DeadR8qPGvqe2bQE/yrrbcZEQmS4sCvTpliRnGv9aPV7oavwGIM6ZguA7AUYrK7DS+Q7oXwBrj+brKyDEmcvcgnO1XvDBOzkw775riYdpZ9gLbEjYw5RVkWv3+BXQKjVVGW+3GyaIJvxn5lRDfJxQhQoSXQ5wUihAhQoQVjAPTN4z/Ne+GSFGZ1aAHlNGgdwPlsxNjLNcCehfMCNB51hxFPuiJ3ZVGe22gdxy9USZL2yOTpT0mk6W9C6Caoex1AHa7c5MZ41b71wPYa9yQxxtgynEXwHUQRWUqKSrzMEVlvgJ6cF8HYJlF7CjQj4qa2n4EdHqRlQAOymRpT8hkaa8bj4mzMn8d6HfMKLgXpkdIrwQ9wf3KgvsKQCiA+0CvjNn76Khd7ecKONuvHIDpnUVTGoivQfenG4z5Edl05oE+rwdBr7xaY7/x74WgJ33WaV1MeRNNGxt1gn6sVIQIESK8DuKkUIQIESIsYNz45Q/Qm3J8QlGZX1kd8jAADYBnZLK0Z2SytEAGG9NksrTPMTpB+T9A57N7SCZLe8e4smRdzhf06uQoGN9B2gF6F1HTgPhBAHtATwQ2g95pcR3oSc2tMllan0yW1gd6MPqcne47i7+CXiVzdoMZd6DN+PdCa0ImS1snk6XdxxA7KeiJuxr0pMGEj0Hn6zOtND0H+jHZMwG8BHqjoCdw8nFBU9vfj7FJB3IQ9MrxU6DfW/3agjsAOrflv43/t3dSaGq/s4w5Hd0NZ/sVJ2SytFtA7ybbCrpfw5gj8D+gJ7w/yWRpSxnKLQOdtsQfwH8oKrOXwfyvoNv9LtAT7lGTQmPqjHrQ/WM66N2H2d5RFCFChAiPQtxoRoQIEacrIizSHwSCnowtBv2IGQGdk806Xx0oKjNPJku7BnRi+OcB/EMmSzsI+p0jKYCFxo8P6CThpnLVxvx2O0FPFm6XydL+AD1o9AW9Q+kq0LuMjuDkZh0mbAC9w+K3Mlna5cZE2tcaH0cNoajMTgCQydKOYfRjjGqKypRb2YJMlpYEOnE7cPJdx9kyWZop/UApRWX+bNtsjLgb9PtV3vS+1A+gJ3c3ymRpkaAf+zQ9XhkLegV4k4xOdN8C+h3BS4zcvygqU2NhKxPA3wF8J5OlnU9RmR0A/iaTpT0MehLZaUxBYcppZ4KWaZVQJkuTgk4xAtCPrwJAgkXbt1BU5k7rcmygqMxemSytAPQPGccsNY3pE74H/YNBP+x8lJWiMmuMNhcDOC6TpWWDXnF7x4Hzwm44268sYJmSwhd0uy4ylhkAcKvlajlFZWYak9g/AyDX2G9Mj5kuAD3hB4DnKSrT9NSAdZ17ZLK0fADLjV8xTbh/xclYi4+OihAhwmshTgpFiBBxuiIMJ3f91IJ+RKwR9GOjH1JUJutjXhSVuV8mS5sJ+pG8NaBX6aSgB5+1oCeUO6y36qeozEKZLG0+6Lxn14IexJ8HeqDbCXqzkx9BJ9XusSrbKpOlrQW9U2aBTJb2OIA9xt0yBy2O64R9mArgFavvlhg/AL06xjv4N27esQTAsxSVabBT2+2gqMx24yT8ZdCPeq4CnWpiBHTePh/Qq6zngZ7ItYNOHfA6RWVar/ioZLK0K0EnJC+SydKeBPC5Mdef2uK4bthO5pkQDdu2T7H47jBOph+xF7+BPp+sV7YBOrn8OgCHKSpT74DNK0Gfy+eDTnvRBDr3oVvgbL8ywjIlBQEdlybQu4G+anzk2lrvPzJZ2teg81CuBr1LKUC/R7oNwJsUlVnIU+39oCeFTRSVWc/A/4aTk8JfGXgRIkSI8ApICBGfZBAhQoSI8QLjY22fgc4BpwC9AQYF+h2wCNCrYV9SVOaWMarPW6BXPqfYkbdvXEMmS5sGuu1XgJ6o/AmgGfRjj1LQE+3jFJX5uMcqKUKECBEiRDgBcaVQhAgRIsYRjI/ZzQG9Aca1oFfBLgb97tMg6PyGdWNRF+O7Zn8F/a7UKT0hBACKyqwFsNL4mOKNoNNAnAMgCPQKWi2Acs/VUIQIESJEiHAO4kqhCBEiRIhwCjJZ2q0APgdwI0Vl7vZ0fUSIECFChAgRzkGcFIoQIUKECBEiRIgQIULEaQwxJYUIESJEiBAhQoQIESJEnMYQJ4UiRIgQIUKECBEiRIgQcRpDnBSKECFChAgRIkSIECFCxGkMcVIoQoQIESJEiBAhQoQIEacxxEmhCBEiRIgQIUKECBEiRJzGECeFIkSIECFChAgRIkSIEHEaQ5wUihAhQoQIESJEiBAhQsRpDHFSKEKECBEiRIgQIUKECBGnMcRJoQgRIkSIECFChAgRIkScxhAnhSJEiBAhQoQIESJEiBBxGkOcFIoQIUKECBEiRIgQIULEaQxxUihChAgRIkSIECFChAgRpzHESaEIESJEiBAhQoQIESJEnMYQJ4UiRIgQIUKECBEiRIgQcRpDnBSKECFChAgRIkSIECFCxGkMcVIoQoQIESJEiBAhQoQIEacxxEmhCBEiRIgQIUKECBEiRJzG8PN0BcYKsbGxJCUlhZEb6RqB/wR/1rJcvJCyom3vsu1JbdH22Nr2pLZo+9Sx7Ult0fbY2vaktmj71LHtSW3R9tja9qQ2X9n8/HwFIWSCDUEIOaU/B3HwyoM4uHXuhLmk93Av6fqui3Tu6CT9x/tJyxstZKh5iFQ9UEX0I3pSfkc5IYSQ8nXGv3eUE/2InlQ9UEWGmodIyxstpP94P+nc0Um6vusivYd7SeOLjURVrSINGxrIyOAIqbqvihBCSNlfywghhJRcW0IIIaTmkRoyLB8mTS83kcHiQdK+vZ107+8m7Z+1k/bt7WSweJA0vdxEhuXDpOaRmlFlTbaq7qsiI4MjpGFDA1FVq0jji42sPtU9WUfU9WqnfGp7r430HO5h9ansr2VkqHnIKZ9MZdl8anuvjXTu7nTKp7on60hfdh+rT6pqFalJq2H1iRBCKu6uYPSpe383aX6tmdUny3gx+dR7uJe0fdDG6JN+RE9Kbihh9WmoeYjUb6hn9alhQwNRViid8ql9ezvpPtDN6pMpXs74VH5HORlqHmL1qeWNFtK1t8spn2oeqSEDBQOsPg0WD5K6p+pYfSKEkLJ1ZYw+dX3XRVreamH1yTLWTD71H+8nLW+1MPo0MjhCytaVsfo0LB8mdU/VsfrU9HITGSgYcMqnzh2dpGtvF6tP5euM8XLCp6r7qkb1a2ufml5uIt0Hup3yqeq+KqKsULL61H+8n9RvqHfKp84dnaTtgzZWnyxjzeTTYPEgaX6tmdGnYfkwqbi7gtWnkcERUpNWw+pTyxstpC+7j9EnQggpuaGE1aeu77pI5+5OVp+sr8OO+FTzSA1jrC2vET2He5zyqfyOcqKuV7P6xHfP5fKpfXs7af+sndUnplg7cs8VMo7gu+cKGUfw3XOFjCP47rlCxhF891wh4wi+e647xxF891wh4wi+e66QcQTfPVfIOILvnitkHMF3z3XnOILvnitkHMF3z+UbRwDIIwxzJo9P2sbqs3TpUsKGljdaWDk+XkhZQghpernJbba5eC5dobaF+CzU9njVFmPt2np5s7YYa9fWi0/bne3tSW0x1q637a3aYqxdb9tT7e1JbTHWjum6W9uTttkmheI7hQCkK6VO80LKAoB8l9xttrl4Ll2htoX4LNT2eNUWY+3aenmzthhr19aLT9ud7e1JbTHWrrftrdpirF1v21Pt7UltMdaO6bpb25O22SBOCgFoGjRO80LK8kGobU9pC9EVanu8aouxdtz2eNUWY+247dOxvT2pLcb69NEWY336aIux9i5tT9pmgzgpBOATzN0MXLyQsgCQkp7iNttcPJeuUNtCfBZqe7xqi7F2bb28WVuMtWvrxaftzvb2pLYYa9fb9lZtMdaut+2p9vakthhrx3Tdre1J26zlnCp1isEvknsTVi5eSFkA8AliD4FQ21w8l65Q20J8Fmp7vGqLsXZtvbxZW4y1a+vFp+3O9vakthhr19v2Vm0x1q637an29qS2GGvHdN2t7UnbbBAnhQDUlWqneSFlAaD+6Xq32ebiuXSF2hbis1Db41VbjLVr6+XN2mKsXVsvPm13trcntcVYu962t2qLsXa9bU+1tye1xVg7putubU/aZsNpk6eQC5HnR9L/GBoCBjoBvXYUHzV3CKCUjGW5OHv4QN8mgGJ+IVSobS6eS1eobSE+C7U9ivcNAKQTgeBg1uOtYT4XHOTs4YWU9VZtIbpCbY9XbTHWjts+Hdvbk9pirE8fbTHWp4+2GGvv0vakbTac8iuFhySHrjwkObRV06xB35E+KPYqIN8px0DOAFq3tELTokHVXVUwDKpQe/9BIAqo2UQBsjDUbG6HIS4Eta90QkP8QO0bwGAngSJXg54SLfqb9WjcqsCQxhctX/ZCFxGEui2dgCwM1UYbVeltgCwMDe91QRsYgLY9/VD1SyA/qkZvzQiCzoiF/Kgaqn4J2vb0QxsYgIb3ukaVNdmq29IJXUQQWr7sxZDGF41bFehv1qOnRAtFrgaDnQTUvgFoiB+atnej69gQaja307682D7KJ8SHs/rU8bsS7fsGWX2q3kRBka1h9al1Rx+rT9WbKETemMjqU8fvSrR91e+UT03bu9Hxi5L2qcMAxR/d6Pm6CH2/tIPaSkFdo4ZfuB90Sh2q768GAJSvKx/1t/L2Smi7tGje3AxliRIdH3eg50APeg70oP7xeihLlGje3Axtlxa1abWjypZdXwYAqL6/GjqlDo0bG6GuUYPaSqHvSB9CpofYnHv1T9XDoDOgbC1dtuK2Cvrv+goYdAbUP1UPTYsGtQ/XYiBnAPKdcij2KtB3pM/sU+PGRnRs62D1ycffh9Wnjo870JrZyupT+bpyyL+Qs/rUmN7I6lPF+grEXhnL6lPrlla0bG5xyqfatFq0v9/O6pOyRAn/KH9WnwCg4tYKRp8UexVoeLqB1ScAKLuhjNWngZwBBKcGM/qkU+pQcWsFq0/aLi1qHqhh9al5czPa329n9UkikbD6JN8pR8vmFlafKm6rgPwLOatPDU83sPpUfX81Yq+MZfWpeXMzWjNbnfKp+v5qdGzrYPVpIGcAARMCnPJJvlOOxvRGVp8s+zWTT8oSJYKSghh90nZpUXl7JatPOqUO1fdUs/rUuqUV1LsUo08AQHSE1SfFXgWaNzWz+mTq12w+1T9ez+pTbVotY6wtrxFtb7Y55VPF+gp0ftrJ6lPfkT4Exgc65VPHxx1oeraJ1SemWFtf9wLjA7nHESw+GXQGVN1VxeoTtZVC25ttjD4BANESVp96DvSg6dkmVp+s28Xap7pH61h9qn+qHjGXxbD6VP9UPah3Kad8MtWLzaeeAz0ISgpyyidqK4XmTc2sPvHdcyPPjWT1SV2jRvU91U6PI/juuaZYO3vPlX8hd3ocEX1JtNP3XCHjCMVeBYJTg53yiW8cwXfPjboginMcUfNAjdPjCL57Lt84gu+ea9kHHB1HsEFCp6s49bFs2TKSl5fHyOmUOvipWoEoAAEBtrxKB79Q5kVVLs4eXtutRUCMraYrbHPxXLpCbQvxWahtG16rBXoBTEyh/9ulRcAEDm2lDn5hLNocnD08l7ZQ21y8EJ/5eCE+C7U9XrXFWLu2Xnza7mxvT2qLsXa9bW/VFmPtetueam9PaouxdkzX3dqetC2RSPIJIcusvz/lVwrtQf1j9fQjo/7+jHzTc02sZbk4e/iKv1a4zTYXz6Ur1LYQn4XatuH9/Uc9Dly8ppizfP1j7M+Xc3H28FzaQm1z8UJ85uOF+CzU9njVFmPt2nrxabuzvT2pLcba9ba9UVs7rMGxmw64xbY9ZU/HWLuzb3lSW4y1Y7ru1vakbTZ4fKVQIsGlAN4A4AvgA0LwEstx1wP4CsByQpAnkeBiAC8BCACgBfAYIfidTYdrpRAAQFUDsjCn/XAWRZcWYeHPC08bXY9oU0pANgMAkLcsD8vybH4cGRN4Svt09Pl01T4dfT5dtU9Hn09H7exP/oP5dVsheawaIWERY6oNnH7t7UldUfv0irUn4ZUrhRIJfAG8DWANgDkAbpFIMIfhuHAA/wBw3OJrBYArCcF8AHcA+NTZepiexWVD9YPVTnH28AHx7MvWQm1z8Vy6Qm0L8VmobT4+MDGQk+c6F/jOEz6eS1uobS5eiM98vBCfhdoer9pirF1bLz5td7a3J7XFWLvetjdqB3XkIlSiQWPJMbdoi7F2THc8a4uxdkzX3dqetM0Gj64USiRYBWADIfiL8f9PAQAheNHquP8C+AXAYwD+RQjyrHgJgG4A8YRgmEnLW1cKRYwBLFYKRYgQIUKEiPEC+YZUxKEH2VMexsrbn/N0dUSIEHEKwCtXCgEkAGix+H+r8TszJBIsAZBECH7ksHM9gBNsE0I+eHKlsOJO9vfrhNjevn0nJJIE8yc0dBpSUlbg2mvvxq5de1G+3tZnhaIHTz31IubOPR8h/lMREjIV8+dfiCef3IT29s5RxxbcU4SXXnoLy5dfBql0JgICUpCcvBxr196Ddy77CFw/NnD5zORXaWkl7r33cSxdeikCfCdDIklgKcnfZiXXlHDy7vxVh0vbnb8YCfGZjxfis1Db41VbjLVr68Wn7e7VIzHWrrXtre3tCW1FRwvi0AMACOw84RZtMdaO6Y5nbTHWjum6W9sbVwq9Ok+hRAIfAK8DWM9xzFwALwO4hIk/JDl0D4B7VAEqHAk9Av9Yf6Q+m4qQ2SGofoCeQESeEwmDToeSS4sAAD4hvpj/zTzUPFJjTgA5VD+E3oO96NrdBQBIuD8B/hP9MVQ7hKJLixB5biQS/pFg3tLaL9oPc7+YC6InKDLanfnhTHR/1w3FXgUAICktCeoatZmPvjgaE9dPNG/EYnrMsuLOCmjb6c1SZn8+G53bO9HzC32j6D3cC6IhaMmk59axV8Ui5uoYNL9G//+/F2dgxbNLUfnvarTJ2/FHdi5u2ft/WBoyD2+2bUSQbyBSNqSgNK8S1/3r7yAA/n7FLTjryTPQuqUVlQN12PG/PaiqqsOzkjQY1Hp0ahR4RP4c2j5txw3xl2H99Osw5f5UVOfXY/dXP+D/FP9BysYknHfTmeZtfMMWhWHqS1Px64UH8cmJr5H9dQHa9J0waA2QBcZhddwqPPnRgwhvDTW3afxd8QieHoxvb/8R39Xux8LJs7HsjIXIys5H0aVFjHGa8c4MUB9So+OEYdQ8R68SD7cNQ6fUoXB1IQDAP9YfC39eiLKbyzBUOwQA0LRoIP9SDvkuOQAgJT0FPkE+UFeqkbcsD7FXxkL2gMz8gnJgYiDm75kPvVKPvGW0zoJ9C0C9Q0HxPR3rKZumQFWqMvNxN8Yh7pY4lF5bCgAInkbnUiy6tAgjihEAwKJDi9D8QrM51n1H+qBt16L5FXo74Um3T0L0mmiU30J3/or1FZi9fTYKziuAXqUHACzJXoKBrAGz7ox3ZkBdoUbrm60AANk9MiQ/kWzmpculmPHuDOSvyAfRE0h8JVh6fCmq76/GQO4AAGDWtlkYyBoAtZUCAHR83GHTn6a8MgUnVp6AukKNgvMKsPjwYlSsr4CqVAUAmPPlHPTs6zG3afJjyQiID0Dto7XmvjDjfzPM9WKK07xv56F5c7NNnEzJaLWUFtourU2cSq4pwXAr/fuRtktrEyeDxmCuF1Oc5u6YixHFiLlu1nGa9vo0KIuVZt46TqHzQgHAJk71j9Wj748+AMBAzoBNnKSrpKi8k05xUH1/NWOc+v/sN+taxynxoUQkPpxo5i3jBAC+ob6scer4pAMAIN8pt4lT8jPJKFxdCHWF2vy+sHWc5F/KzW1qHafYK2MxLXOauV5McVqwbwEaNzbaxKnx2UY6jh1aaFo0NnEy9ae8ZXmMcdK2a831YorT7O2zMdwybK6bdZxmvDMDyqKTsbaOk3Q5nY+Vqz8pS5Q2cbLsT7VptYxx6j96MtbWcUp+LBmye2Rm3jJOAHt/srzuKfYqbOJkuu6pK9QouaaEMU7UO5S5Ta3jFHdjHKa8OMVcL+s4AXR/qn+q3iZOpuueVq6FukZtEydTf8pblscYJ3WF2lwvpjjNeHcGhmqGkLcsj/W6pyw4GWvrONHjCANnf1LXqG3iZNmf6p+qHxWnnvlZuGQyQOljkThYhtxlOZj/7QKb+9Ok2yeZ68V2f2KKk6k/9RzosYmT6bqnrlCj7OYyxjg1v9BsblPrOE26fRJSMlLM9bKOE0D3p9q0Wps4ma57I10jUJYobeJk6k/5K/JZ70+mejHFacorU8w8U5wMGgNat7Syxin64minxxGm/tS4sZExTn2H+jjHEXE3xnGOI9jiZM84Ql2hZh1H1D9Wb24zd4wjRrpHMJAzwDiOAOD0OCL5mWQzzxQng9bAOY6IvTLW6XGEqT81b252ahwRe2Us5zhizmdzBI0jGEEI8dgHIKsAst/i/08B5CmL/0cARAGQRuNHAxAKIMuMfCJAqgFyFp/W0qVLCRtqHqkhpK2KENLG+KlPP+oUZw9f+Jef3GJ727bXCQBSU2N7zFdfbSUSSMiDD95JCGkjIyNNZNasaWTq1BTS2VlkY3tkpIns3bvN/P/zzltFooIjSG3tn4zan6zLJBUVh22+37p1MwkKCiLLoxeQd999kRw+/DU5dOgr8v77r5ALLzybRERIyc6d79r4pde3mP/9f2ffTujT1oE2aasyxzp3aS7reWA+F5zg7OG5tIXa5uKF+MzHC/FZqO3xqi3G2rX14tN2Z3t7UluMtette5v2sW1PEJIhJT/f9wAhGVLS3sxsQ4y1a3l39i1PaouxdkzX3dqetA0gjzDNy5i+HKsPQPwAUg+QVIAEAKQIIHM5jj9kMSGMNB5/nT1aXJPCYfkw56RwWNHgFCeUF1KWa1JISBu56rJLSGBgIFGpasmuXf8jAMju3e/x2s7K2ksAkM3P/duher355vMkLCyUfPPNB6y2d+36HwkJCSZffbqV1faTj/4f56SQ0bbFpHBYPsx6HvDxQsqKtr3Ltie1Rdunjm1Paou2x9a2J7TzN19OWjbMJOVHfickQ0ryfvzI5drjrU3Gu21Paou2x9a2J7X5yrJNCj36TiEh0AF4EMB+ABUAdhGCMokEz0okuIqn+IMApgFIl0hQaPzEOVOPjm0dnHzXzi6nOHv4zu2drJxQ21w4M2QJhoeHkZdXhF9++QO+vr647LILeW3/9ttRAMBK3SK761VVVYt//vNZ7NnzIa69dg2jzzqdDmvXXon3338F99z/OAYGBhltq8vVnH7xtQn1DsXJc50LfOcJH8+lLdQ2Fy/EZz5eiM9CbY9XbTHWrq0Xn7Y729uT2mKsXW/b27QnqavQGToT/r8lQEP8MdJ0nKGkGGtX8+7sW57UFmPtmK67tT1pmw0ef6eQEPwE4Cer79JZjl1t8e/nATzvijpEr4kGYHtibPy+FeXUEAzDevi8V8NYlouzh1c1KBH6noa17LzvW5FxZSIjH3l+JKtdPkgb6Pea2ts70dJCYcKEGISEBPPabm1tBwDMvXbW6LoaDDAYDACAsHPCYDAY4OND/+bw0ktv49Zbr8GFF54DADj+TT5u+vr/kJ9fgtTUJPz3vxuxZs06NDRk49Zbr8Ubmz/Al1/uwb333majHzg5iNMvvjZRfK9ASkYKK0+fC45z9vBc2kJtc/FCfObjhfgs1PZ41RZj7dp68Wm7s709qS3G2vW2vUm7v7sTMiJHU9zN6NsxgJFrpyOyp8jl2mKsHdMdz9pirB3Tdbe2J22zwdO7j3oFlCeUnLzpJU5HOXt4IWVVJSqnbRPj7qASicQp26YXfU144IGn4O8/Gf7+kxG5YA7S018xc99//wvuvPMmAIBaPYQH8v+D6OhI/Pjjx3j66Ydw771PjLJ1/swzcfhwNqPuiHyEu14C2gTgPhf4zhM+XkhZb9UWoivU9njVFmPtuO3Tsb09qS3G2vPaLeX0PTAshd41vj96IVK0NdAO2/6ILMZa1LaHF2PtXdqetM0Gj68UegPYkqmbVuh6D/ci6rwoxmO4OKE8X1n/if6sHB+0FxiAE0B8/EQkJcnwyy9HoFYPmVcL2WwnJsYDADqhQBJOrmA+/fTD+NvfbgUALF9+mfn7np5edHf3Yv58emXxhx9+wUiADrt2vWfW8vX1xbp1D5nLTEyagBPFpYz6vqHcv2PwtcmUTVM4ebZzgY+zh+fSFmqbixfiMx8vxGehtserthhr19aLT9ud7e1JbTHWrrftTdrKBnrXwaQ5K2DYFIBG/QoEdX6JmvIcTF98rsu0xVg7pjuetcVYO6brbm1P2maDuFLoYRANez4/d+LAn4cRFBSEpUsX4KKLzoZer8e+fb/zlrvggrMAAPuzDo/6Pjk5AcuWLcSyZQtHfa/T0audQUGBAIDGxlZMTUgZ9ajqkiXzR5VpV3QiNta5pW8+GDQGt9j1Zu3T0efTVft09Pl01T4dfT6dtP27StCOCYiMnQSDxgDZPPr1i56qP8esDsD4bO/m6kJUN/8w5rqugKh9euh6K075SeEhyaErD0kObdU0a9B3pA+KvQrId8oxkDOA1i2t0LRo0LqlFQadATWP0O/+1Txk/PtIDQw6A9o/bIemTQPqQwqDBYNQfKdAz/4e9Gf3o3tPN4bqh9DyWgt0Kh3qnqwDcDKJeutrdH6QhowGaLu1aHunDaoKFeS75Og93IuG5xog3yWHqkKFtnfaoO3WoiGjYVRZk626J+ugU+nQ8loLhuqH0L2nG/3Z/ejZ3wPFdwoMFgyC+pCCpk0DxQ/dAIDG5xttfNq96wf8nHUIf1t3C/q+7MXFqedgWkIKHnvoOdTta0DHZx1Ql6rNPlU9XoUff/wV1Q9WY9WqZVgWvwCvf7IVhx46yugTAKjK1VBVqKDZpUZQUCAOpdE3MsN+HZob2lD5f5Vmn+oq6TpqmjVo+6QNX//4E1bFL7HxqWlTE/TKk4/UWsepaVMT1JVq2zjl9oPaSkFdo0b1fdXQKXWovp9uU1OCT9PfltdaoO3SonlzM5QlSnR83IGeAz3oOdCDrt1dUJYo0by5GdourTkHo6ls4wbaj+r7aY3GjY1Q16hBbaXQd6QPtY/W2px79U/Vw6A7mc+m4jY6R2XF+goYdAbUP1UPTYsGHZ90YCBnAPKdcij2KtB3pM/sU+PGRmgaNKw+Vf29itWnjo9pu2w+la8rh7Zdy+pT90/drD5VrK9A47ONrD61bmlF/5/9TvlUm1YLdZWa1SdliRLVD1Sz+gQAzS81M/qk2KuAYo+C1SfLWDP5NJAzgJpHahh90il1aH6pmdUnbZcW7R+2s/rUvLkZ6io1q0+Vd1ey+iTfKUf/n/2sPlXcRudDZfNJsUfB6lP1/dVofLaR1afmzc0YyBlwyqfq+6uhadCw+jSQM4DqB6ud8km+U47un7pZfbKMNZNPyhIlav5Rw+iTtkuLFmO+WCafdEodqHcpVp9at7RCVapi9AkAKu+qZPVJsVeBvkPsPpn6NZtPXbu7WH2qTatljLXlNUJZoHTKp4r1FRhuHWb36Ugfah6qccqnjo870PtLL6tPTLG2vu7VPFTDO45g8smgM4w6jularixQjvIpTlmFFu1UOtZ3VmJiwhTISRR8WvJsrnu9v/Sy+mTdLtY+yb+Us/pU/1Q9GjY2sPpU/1Q9VKUqu32y7gvadi1jnEznXs0/apzyqfPLZ7Cy9SEcydzk1D23/t/1rD6pa9Sg3qWcHkfw3XMtc9E6c8/VtmudHkc0ZDQ4fc8VMo5Q7FWg5pEap3ziG0fw3XMb/tPAOY5o/7Dd6XEE3z2XbxzBd8+17AOOjiNYwbQl6an44UpJMVg8yJmSQlle7RRnD8+Vp1CIbVNKit273yNZWXvJ4cNfk08/3UJuuukq4uPjQ1bFLCFDQ3Xm48vKDhKZbBKRySaRl19+hvzw0cfkt992ktdfzyCzZk0j11xzqfnYlpZcMj0llURGRpCnn36I7Nv3GTly5Bvy9dfvk3vvXUcAkJdeetp8/F/+sppkZDxKCGkj7e0FJMg3kPzzn/eSrq4SUlr6O1m4cA4BQA4c+JJcc82lZPHceWR4uMFcXqWqJbt3v0d2736PXH3JX8x+7d79HsnN/Ym/TRzIUzhYPOgUZw/PpS3UNhcvxGc+XojPQm2PV20x1q6tF5+2O9vbk9pirF1v21u0B/q6CcmQkmMfPT5Kl05RMcOl2qdirOs3LiD69AhCMqTk+DdvOFzenX3Lk9qnYqyF8p6MtSdtwxtTUngLevb1cPJ9B/uc4uzhY6+KdZttAFi79l6sWnUV/vKXv+KZZ17G8LAWO3a8g10b30VQ0MmdPOfMmYGiol9w223XY/v2Xbjh3ntw+eV3YNu2XbjuujV4772XzccmJsrw/UMf4bHH7sPPPx/C2rX34sILb8Ijj2Sgu7sX2+57DU888X/m4//xj7uRmfk+GhqaMWlSHLbc9Sy2bv0cEybMx9Kla8yb0Fx55XpERITjs/VvICDg5PPQcrkCa9fei7Vr78V3B/ab/Vq79l689dY2h9ok7sY4Tp7rXOA7T/h4Lm2htrl4IT7z8UJ8Fmp7vGqLsXZtvfi03dnentQWY+16296i3VJOp54Imbx0lK520hIkkg70drW7TPtUi7VBr0e8vg1/4goUBy3FssJ05H73jkPa7uxbntQ+1WLtCt6TsfakbTZICPHMO21jjWXLlpG8vDxGTtulRcBIIyALY+a7tQiIYX5pk4uzh9e0aRCUwJxmQahtLp5LV6htJu722x9GTk4hfvrpU8gCJ8I3zhc1NQ1ISJiEiAgpysurkZqahODgYNe3N6UEZDMAAJoWDYKSOPzu0iJgAos2B2cPz6Ut1DYXL8RnPl6Iz0Jtj1dtMdaurReftjvb25Pap1qsDXo9Cn/5FBJDCBZfdqNT2qdKrLO/eA4rq1+F4r5SxE5KMuuWZ+3DnP03o+jc97Dwgptdon2q9ev2pirEbzsDWdP+g8XX3YfaNy7HbE0RCpZvxrIr7rFL2519y5PazrY3MRiQ+81/MXnedZg4K8Xl9eLSdoVtb421J21LJJJ8Qsgy6+/FlUIAzZs4nq8F0LalzSnOHr7q7iq32ebiuXSF2mbi3n//FSxbthALF16M/zvnKZSUVCIpSQYfHx+cOFGCzz77GvPmXYiGhma3tnfptcy7mprAdS7wnSd8PJe2UNtcvBCf+XghPgu1PV61xVi7tl582u5sb09qnyqxJgYDCvZ/jKYXFmNJ9j8wIfvfTmufKrH27SyGHNGInZQ0Sjdl/pnQER+o64+zlnVU+1Tr14rGMgCAtmgCgkLCMOWhvagKnIdFuU8g/6dtvOX5dPnKCuW9sV83lOfijNKNKP34XbfUi0vbFba9NdaetM0GcaXQBKqadaXQnSi6tAgLf17If+Apovvtt/uw8a7XUDJQZU52HxAQgIsuOhtPP/0wzjpruetFLVYK85blYVmezY8jYwJPaZ+OPp+u2qejz6er9nj3mRgMKPp9J8KyXsE0fR1aJDKo/CKRoG1A+MYOt2o7i7HSbnx2HvoCE7Doif02urXPLcGQnxTznzrk9npYa481nNHO/nITVla9DMV9JYidlAwAUA70ovXNyzBVW4XSs7Zg8SXrXK7rKnij9vEdL2JF5UvIkt2BVfdsGVNtd8OT7e1JiCuFHDDt2sMG0+6fjnL28EGp7MvWQm1z8Vy6Qm1zcddeuwY7b34bvb3lKCs7iIqKw+jvr8CPP35qnhC6s72DpwVz8lznAt95wsdzaQu1zcUL8ZmPF+KzUNvjVVuMtWvrxaftzvb2pPZ4jXXZulIUH/oaNZtWYNHR+xBkUCF38YuIf7oIvbJzES4Zwoh22CntUyHWamU/kvStGIqdx6jbHTkfKZpKGPR6m7LOaJ9q/VrSXYtBEozOfw2YvwuTRiHhwR9R7z8Nc/98GIW/7fBY3+LjvbFf+7ceAwDo6tl/rBkPfctRXXdre9I2G8SVQhM8tFIoYgxgsVIoQoQIx9FFNaL253ew4o4X4ePr6+nqiBin6JG3ofP9GzB7pBztmIDWBQ9h0RX3wT+AzmN7fOfLWFGxadQqz+mGytxfMevH61Fw5tuMK1q5e97C8sJn0HTT75g8e6kHaujdKHnxfATpBzH937bjvf5eBeRvX4rJIw2oXfM55qy81AM1HF8gBgN6nk1BDPpREHo2Fj/2o6erNC5QU/gHlL+8hKTb3vXKa9lpu1JoT57CnLk5nHkKC1YXsOYpLL68mDNPYd4K+sLElqew6NIi1jyFprJseQqLLy9mzVPYtKkJVQ9UsfqUvzKf1aeOzzpQvr6c1afqB6tR/WA1q08lN5ay+lT9YDXKbi1j9anjsw6U3lzqlE9Nm5pQcVcFZ57C7GnZnPmFjs8+zppfKP/MfM48hVmpWbSPLLl4cpfksuYXyp6aDYA9T2HesjzO/EKlN5Wy+pSVksWZM6n48mLO/ELl68pZfSpYXcCZM6no0iLO/EJFlxY55VNtWi1KbyjlzC90fOZxzvxCx2ccZ80vdOLsE5x5Ck2xZssvlLswlzW/0PEZx1l90nZpkbs4lzO/UOkNpaw+HUs+xpkzqejSIs48heXryhl9qv7yRaxqfg/lnx1lzZlUdGkRZ57C4suLnfKp+v5qlN5Uypmn8Pjs4w77ZOpPBasLOPMUmmLNlv8uZ34Oa06/47OZY23qT7kLcjnzFJZcW8Ka0+9YwjHOPIWFFxdy5ik0fZh8yj8znzNPIVOsLa8RxVcVM/pU9N7bmD1Sjt9GHkLUY8UI23sJ/AMCzT4NZEno+rxRwprTL2dujlM+dXzcgYILCzjzFFrH2vq6lzM3h3ccwZbTL2duDmeewuKritG4sRG91bkAgPjZK0/2a9kxs08xCSsBAG3HDpr7U8GFBZx5Ci3bxdqn/BX5nHkKCy8p5MxTWHJtCa9PbPfc8nXlnHkKc+bnOOxT7HALuvWJKLy40ManQGUYNMr/YRj+kH/xAatPBecVcOYpzF2Q6/Q4gu+ea4q1s/fc8nXlTo8jCi8qtPHp6L17EYN+AIB/X7dbxhGKvQrkLsx1yie+cQTfPbfgggLOcUTu4lyHxhH5mfuQ//waTN9zBaYoC3DiiR+dHkecOPsEZ55Cyz7g6DiCDeJKoQniO4Wnrrb4TuFp5/Ppqu0OXWIwoP25WZCRTpRfupP11/XTsb09qT0efc7NXIuU/lxMMCZttkbpH99h3m+3o/wvOzBn1RqXarsCY6Gd899bMK3vKKLSmyDx8bHRNej1GHwuCVXRF+CMhz9za12stccajmpr1EoEvJyI45Pvwaq7NrMe1/jsfPQGJ7Oueo0nn92tfXz3q1hR9hzqfFPhR3SYnM69MYsrtccCrtJtqshH948bsUR5GAMIQVnybZh3/ZMIj4h2QS1dj9N2pdAemH4ZYYNppcxRzh5eSFlv1RaiK9S2UG2uc4HvPOHjhZT1Vm0hukJtj1ft8RbrupJjkJFOAMBwf6dbtL3N59Nd2119a8JgJZp1U1nLBUdMAABoBhQu1XUVPxbaMYOVaA2aYZ4QWsPH1xdNQbMR21fiEu1T6TpK1ZfBR0IQMHEGp22VXxQCnDzH2LRdybvLtjO6vs3HoEAkFJELEK7vd0u9+ODN7Z318E/Ie/16JO24EDMGc5CdeDfIw8VYdddmhEdEu7V/uKNfiyuFAHRKHfwG6llXCnUqHfxC/RzmhPKibReVtVgp1Cl18AvjKM/BCykr2vYu257UHm+2s97/B1a1bQcAHJ/7H6xY+69xUe+xsO1J7fFme0g1iIDNSTiecBfOvOd1xnIdLbWY9OFS5MzLwBk3POoV9R5Lbb2fFj4vJiIvYd2oXR6ty2Z9+C+c0fwBhv7ZgDBplMfr7S22T+zbhiXHH0HttT8hZeoK1rL5r12DOGUVkjIqxrzefLw32SYGA7qenYqW8IXQRqTijJZtkKR3M75X7k31HivbWe8/gjNat2MYAShKuAmzr3sGkbGTxkRbqG1xpZADra+1cvLt/2t3irOHb3uDPa+eUNtcPJeuUNtCfBZqm49vfoE7dwvXucB3nvDxXNpCbXPxQnzm44X4LNT2eNUeT7EmBgMSqQMo96d3QjQou9yiPV7b25Pa461fN1XkwFdCoG1lXymMiJ4IANCrup3SHu+xbq7Ig79Ej8CkxZy6oVNWwldC0FhyVLD2qXQdHe6k8y/Lps7jtK0LioHUwL7q5c6+xcd7U7+mGisQhx7oks6EJCQGvhKCwT7mFVZv71vO6HKV1QypsKz1ExSS5Rh64ARW3fOmzYSQT9udfvGVZYM4KQQQd2scJx97baxTnD1835E+t9nm4rl0hdoW4rNQ23x8zy89nDzXucB3nvDxXNpCbXPxQnzm44X4LNT2eNUeT7FurMhFEqEwOP0aDJJgSNTsg/XTsb09qT3e+nV/Hf2kTsql57GWCw4NxxAJgGSo1ynt8R7rnlp6k5mJM1dy6k5ecA4AYLA2S7D2qXQd9eutRydiEBIWwWnbEBKLCIkK2mGNw7ps2q7ivalfU4W/AgAmzb8QfmH02Gqghzkthbf3LWd0ucqafsDRz7kZMRMTndJ2p198ZdkgTgoB9B3s4+T7j7H/osTF2cMLKcvFb9++EyFTp6G2tsGG0+l0WLT/MmzY8Jr5WIkkAY2NLeZj/vOvzfj996NOaQvxeXBQiUcf3IDVq2+AVDoTEkkCDh06ZrdtIdoA97nAd57w8ULKequ2EF2htser9niKdUf2bhiIBFPPvRkDkMJvmH2wLkTbm3wWtd3TtyQdReiFFP7FkZxlBxAOXw33QM0RXVfybtfuKMIAQiFLmcl5bETMRLRIZAjqLBCsfSpdR6WqRnQFJvHa9gmj313tU3A/WeSItit5d9l2WLfpT/RCiuSZixEgpdtM1St3eb344I3t3VtH/4ATLOfuq+7sH+7o1+KkEEDIrBBOniu5JV/iSz4+ZUOK22zbi8svvwhZWXsRH3/yl4XXvt2K33//0yltEzc0NIQ33vgA55xzLaKi5sDffzImTz4Dr/p/gLy8Isay3d292HF0L/z8fHHxxec6pGsPP+31aZw817nAd57w8VzaQm1z8UJ85uOF+CzU9njVHk+xjm/bj4rAeYidlIRBnwgEDrMP1k/H9ubSLvp9B5p6H3eb9njr1zEDFWgJmo7QOdy7fCslUvhr+5zS9lSsXaUd1V+B5sDpNpvMMOl2SOcjSV0OYjB4Xaztte3K9iYGA+J1rVCFpfDa9pfSjykPKCiHdflsC+W9qV8n9p9AQ+hCSHx8EBJFt5mmn/kVAm/vW87ocpZtL0IfwpC4YI7T2u70i68sG075SaE9eQpbt7Ry5ils/7CdNaefYq+CM09h8yv06htbTr++g32sOf1MZdly+in2KljzFCp+oB/zany+kdEnANAN6EB9SCGoNQDTOidDdUhlzlMIAAPZA6x5CvX9elaf+v/sx0/P/4ppqWfhzRc/wl9Wn4c3Lt+Aw4e/Qtqsu6FVjeCslVfjySc3ofaJ2lE+Bf7hh5IXfsGux97FrQuvBgAofuw25ynU9eo48xQONQxx5ilse7ONM79Qy6strPmFur7p4sxT2JBuzMXIkoun+/tu1vxCjRvpOLHl4un8tJMzv9Bw+zCrT80vNXPmTFIWsvtUvq4cuj4dq089+3s4cyZp27Wc+YWc9ak2rRaaJg1nfiHqPYozv1Dzi82sOZO6v+/mzFNo6qNsPnV93cWaX6j5xWZWn7RdWrR/1M6ZX0jTpGH1qem5Js48hVw+VdxWAV2f7mRetaf2IcXQDLn+fAzkDECli0DwcC9rziRtu5YzT6Gy0Dmfqu+vxnD7MKdP7R+12+UT07nXs7+HM0+hqV+z5b+T75Db+FT6Yg4mH/knVg19jxHtMGtOP+p/FGec1DVq1jyFDf9p4OxP/X/2c+Yp1PXpWH3q+qaLM08hU6wtrxGqctUon1o/b0ayrgm9w9M5fapYXwGVRIoAZS/rNaLz006nfOr4uAO9B3s58xRax9r6utf5aSfvOIItp5/lcUzXvf7SXkweacCgdI7Nda/hmQYbn7ShixCLPlR/lY/eg72ceQot28XaJ/lOOec1Yrh1mDNPobpGzeqTqlzFmadQ16fjzFMo3yG326eu5lZIoYIuMJVuzz/7WX0KktI/gKt7Oxh9UpWoOO9P1P8op8cRfPdcU6ydvefq+nROjyM0DRqzT8fu3o94dKG7fS4AoPfjEfrvsQaXjyMUexXo+rrLKZ/4xhF891x1lZpzHNH+UTujT5FdZWgJnI6Wl1udvufyjSO6v+/mzFNo2QccHUewghByWnyWLl1K2NC5o5OQtipCSBvjp2tPkVOcPXzhX35yi+1t214nAEhNzVEbbmSkiQAgGRmPjjq2oSGbENJGANh8TMfm5PxILrroHBIVHkGCgoJIamoyuf/+20fZP/DK5yQ4OIj85z+PEJ2umdHn4uJfSXJyAnnssftZ/frlly8JAHLw4G5hbdJWZY517tJc1vPAfC44wdnDc2kLtc3FC/GZjxfis1Db41V7vMT62LYnCMmQks7Wevr/z64lnRkpbtEer+3Npp3/yhWEZEgJyZASRUeLW7THU7+uLjhCSIaU5P/0Ea/tnI1XkKaNs53S9kSsXWX7xHv7CcmQktzv37NLt6bwD/r4H7Z6Vawdse3K9i7P/pmQDCkp/H0Xr+3mmmK67b57x2FdPttCeW/p1zl73iYkQ0pqi7MIIYSolQOEZEjJsY//7fJ6WWu72rar23tYM0SG06PJsf894NZ4uNM2gDzCMFc65VcK7UFQahAnH5gc6BRnDy+krD229Xo9dDrdqI9er+csk5W1FwCwfv2NyMrai6ysvfjb326BUqnCX/7yV/j6+uJ/z76Iffs+RXr6I9DpTtobGRnBPW88icceux/PPvsYfBm2LtYTPebNm4VffvkS7777Cf78M9duv9zZ3gD3ucB3nvDxQsp6q7YQXaG2x6v2eIl1XMt+VPrNRlxCKgDAEBKNSDIAYjC4XNtbfHaFdsH+j7FEeQS1vvQum0qW3frcoe0q267uW70WG6jw2R4JikS4YcAluq7m3WlbjWIAQNyMM+zSmzx7OYZIAHRNuV4Va0dsu7K9lW2VAIDYyfN4bUtj4gEAukHm9+Mc1XY17y7bjuiSxqMYQChSZtNZC8ybQLFsNuatPrtDu6XqBAIkOgQkLnZrPDzRr8VJIejHJBmxbxOw7XYE/Px3YNvtjB8uzh5+1nkvcpfdt4m13oMnBnl9mzXrPPj7Tx71CQqawllm5cqlAICEhElYuXIpVq5cisREGSora9Hb24fNm5/B2eHLsXr1mVi//iZs3brZXPaLL74FtEB6ehoAoLe3DzfddB/Cw2dg0qRF2Lz5HTzYmIGPP96FGTOm4sEH78Rbb22z2y8+n/n4SbfbbhlsCdZzgYezh+fSFmqbixfiMx8vxGehtser9niIdVt9Gabq69GXusbMadURCJDooBzsc7n2eG1va+3+7k4kZaWj1ncqlKseAwAMcSTJ9oZYO1PWUW3SXoQBhECWMpPX9siQFFKihIHlx0tvibWrbZP2IqhIEBKnzrdL1z8gEA0BMxDVU+RVsXbEtivbW9dVDS3xw6Tk6by2pRHR0BJfECVz33Rn3+LjndVWDvSibNM5KP39J6frZakd31eA+pAF8PU7meuun2MTKG/uW65ub9MuwXEzV7i177nTNhvESSH40xj4R7MngOTi7OH9Ip23HX1pNCcPAN9++yFyc38a9cnO/p63HBOmT09FZGQE7r33CexXHUFLi22+we+//wW33Xa9eYVw/fo0FBaW4ZNP3sBHH72Gr776EScaS83HX375hTh8ONtuv/h85uXXcPPuTIfBpe3O9CNCfObjhfgs1PZ41R4PsW45ugMAMPnsm81c8BQZAGCgm3nHvtOxva21qz95GBFkEJKr30JwJL0xw/Ag++Y83hBrZ8o6qh3VX46WAHoDFd77bWIcZz40b4m1q23H+NSgOWAqY2JwNt2BmIVIHamF9PJQp7VPletoUH8DKN948ySGy7bExwd9kgj4Djl+jvHZFso7q91UcgxztcWYrEmHWsm8C7u97a2gmpBEKGhko1OjKH0j4M+yA7U39y1Xt7eBKoSSBCMhdY5br7PutM0GcVIIgHqHeQcqrHkauPMTtHS/ANz5CeOHi7OHL/3uMe6ya55mrXfnx528vs2bNxPLli0c9Vm6dIFT7RQRIcXBg7sgk03Cw4//B8nJZ2DevAvw9dc/mo+prm5AQjs9GJLLFdi79wA+/vi/uPbaNbjssgvxzTfvQ6sdMR8fFxcDhWL0oInLLz6f+fjyW8o5edZzgYezh+fSFmqbixfiMx8vxGehtser9niIdXTzz6jxm474ySe329aU0Y+jDLLkqTod29tSu+jgbizv/xl5ibdj6oIzERIRAwAYUbJPCr0h1s6UdUR7RDtMb6ASNccu25o6+jxjy4fmDbF2tW29TofJunr0R811SDcg5QwESHQo+eAXp7VPletotKYZvUHJdtvuN0QgcJj5UUh39i0+3lltZSv9Y3sCOlH8yWNO1cuk3VhwAAAQM/eC0RojUgSP9Dll21uv4c60d0RfBZoDp8HH19et11l32maDOCkEkPJcCief9HiSU5w9vJCyQmw7i0WL5uHrr99Hd18ZsrL2YurUybjxxvtQWko/z6/T6ZB4SwIAmPMeLlp08kaXmChDVECE+f9tbR2IjR39S42n2hvgPhf4zhM+XkhZb9UWoivU9njV9vZYtzdVYYauGorkNaO4+OtmAAA0fc69h3Mqt/dgfw8mHn4STT5JWHIb/ch/WCS9w6FOxT4p9Fa/Xdm3WqoLESgZgV/iYrtsR19Iv8Oq6mPe+t5eXXfw7rLdWluMEMkwfGWLHNJMnH8eAMB3frXT2qfCdVQ3okW8noImYqrdtodCYhA84lzeVa88z7oqMUiCcTzmGpzRuQuVub86rauvPwolCUbqvNErhbrIaITqmVchvbVvuVpbN6LF5JE6DBh/wHHnddYT13BxUgig6m9VnHzdv+qc4uzhuXKJCLUtBP6+/hga0rDyTU82YeXKpXjuucdgMBhQUUGnh0hMjEf2q3kAgLg4evm6sbHVXG5wUIkB3cn3/j777BtcfPE5o2y7s71D53E/ZsN1LvCdJ3w8lzZX2eNv3YVDT//daW0hPvPxQnwWanu8agu17e5YNx3dCQBIOvOmUVzvVi0AYGSAeVJ4Ora3Sbv8kzTEkW4MXfYGAoPoa3p4JL1SSIb63KI9Xvq1oiYHADBhxgq7bPd9Rb9LqGF5F9PTsXaH7a5quo1iWTaZYdONS0hFJ2JASnKc1j4VrqOdLTUIkOjhN2EaI88EZX84wvR9DuvaY9sT/TpsoA6UfzICCv4Pckksgvc9As2QyqF6mbQn9uajPnge/PwDRvEqeSikhPldNW/tW1x8U0U+YtbcgfYm+8/h1tpiBEu08EtYJLhu7jyP+Mqy4ZSfFNqTpzAwPpAzT2FAXABrnsLwZeGceQqJjgBgz1MYe20sa55CU1m2PIXhy8JdkqdQ00RP/voO95nzFM6eOQ3ffbofP39/EHvu+BEU1YH3rvgYV121Hpsu+C9ar5bjg1s+w6P/2IiwoFAsjJ0D+S45VqUswcG+Y5DvkiNWHYUFybPwwL1P4/DDR9HRIcetSx+AzqBHxfvVSE9/BV99/gOe/tfDZp86PutA0eRKbP/3Duz/4BAA4IfMX/DFe99i+x07MPXVqZx5CpMeS+LMUxicGsyZX8g/yp81v1DU+VGceQrN8WLJxTPhugmseWtMsM7FU/NkNeZ2/YQVwT+g62gHay6e6W9Nd8qnjo87kPTPJM78QrO3z3bKp4r1FZi9fTZnfqGEBxKc8qk2rRZTX5nKmV8odE4oZ55C32Bf1vxCMWtiOPMLwbgHBlt+oZg1Maz5hXyDfVl90nZpETonlDO/0NRXpjrlk3ynHAkPJHDmKZy9fTbCK35EnU8qtB+GjPIp/lz6V3hNcwdjzqTZ22dz5ilM+meSUz5V31+N6W9N58wDFb4knNMnrtxWE66bwJmn0NSv2fLf9S8qworuPTgWcj0iy+aafTL0AUoSDImmjzWnX3BqMGduq9TnUllz+gHgzNcVf1c8Z55Cy3hZ+xR1fhRnnkKmWFteIyY/M9ns00hFPlQkCBG9k3l9qlhfgWmP0q83DLV3MOYgizgzwimfOj7uwMRbJ3LmKbSOtfV1L+LMCN5xBFtOv8D4QNa8akNFOdAQf+h3RTLen0z1YvKpSxKPiKAuzjyFlu3iiE/1T9Vj5gczOfMUpj6XyporbvIzkznzFM7ePpszTyHXPdfSp6q3jgAAhgvjzP0p/q54zlxxWkMkIg39DvskdBzBd891dhwxUduELlUiFn90Bsq6/oXJhhYcy0gbdY0IXxLOmdNv+lvTceL//kSKoQXyrnk21z3/KZMghQq9Bd0uHUfw3XOFjCO47rmNn76FVN9WVG7fbvc4ovp/++nr0/f0DrZC7rlCxhF891y+cQQrmPJUnIofrjyF5evKOfMUVj94yCnOHr742v1usS0kTyEhbeTL698kS5bMJ4GBgeZjKysPkxtvvJKkpCSRAF9/EhsbTdasuYBkZ39vLtfZWURC/YPJ11+/TwhpI2VlB8n06anmfId//et1ZHH0XCKRSMjq1atIaenvNvVLCJ/ImCtx8uRE59rEIk/hiXNPsJ4H5nPBCc4enkubrWx9WY4531nBgc+d0hbiMx8vxGehtsertlDb7ox19p2/0bmoPnrchitbV0qG0mNI1rv3u1x7vLa3WjlAmv89nbRumE7n8bIC9Z8pJOf1tW7RHi/9uvz5laT8+VX22749m5AMKcn6NMNhbXfG2p3apS+cRcqfYR+jcOnmv3IFafz3LKe1T4XraNbnzxKSISU9cspu2788+gghGVKiGuxzSNce22Pdr/sUHcY+k27mc16/kYykR5KawqN21+vEuSdI/r5thGRISUXOLzb8gUeeJiRDSrram+2qlyP8WPdrg15P2jZMIyRDSopePN/usllv/52o02OJbmREcN3ceR7xlQVLnkKPT9bG6sM1KSSEcE4K3fnhSl4/XnU//PA1EhoaQn7++XNCSBsxGFpJbe2fhKJOEELayI/nfES6u0vHzl8Hkte7E85oH/8qk5AMKRlOjyY5r98wZrqugqg9vnSzv9xESIaUNFbkM/LtGVNIzus3ukVbCDylnfXOvYRkSEnJ0b2MfO2zi0jBy39xi/Z4aG/dyAhRpU8gWW/dbbdtg15PtOlR5Nh7DwnSdgfcoa3X6chA+kSSveV2p3Szt9xOejISXF4ve7TdDXu1s7fcTvoy4olBr7fb9vFv3iAkQ0ra6iuc1nUHnNGuyN5PSIaUFP620/xdn6KDdGUkk9pnFxHtsMZu7ay37ibq9FgyrBmy4fN+/ICQDClpKHd9+4x1m1cXHCEkQ0o6/pNCNOkxjD/qMaHshbNIxXMr3Fw794NtUnjKPz5qD8xLuSwwPa7oKGcPL6Sst2qfU7wUL7zwBK68cj3WrXsIv/9+FJGRUkRGSlFTU48/unJxzjnXYfdu5tQYnmpvgPtc4DtP+Hinyrbmoh+hyNWeh5n9R6EdZn7P0y3advBCdIXaHq/aQm27Uzu49Ac0+iRh8qwljGWVvhEI0LJvnOKs9nhs7/rS41jesQOH+y/BvLOuZDxGNRyCwBH2fFHe6jcXd/ytO6E7/9922W6tK6U3UDG+f2NPvSrvqkK/JBw+Gsc3AfHmvsXGtzWUI1wyhIGGVKd0DSGxiCBK6HU6p+p2KlxHwwYb0OGXBImPDyPPhIGD9COFg92O79LobefZQAu98+iEKQvNfETMRDSveh5T9fXI/2Kj3boTevJQFzQHAYG2yc/7f6IfbVX12r5X7o19i4tX5OyGjvggv+EuBEpGUJOzn7esQa9H8nAt+iNnu6Ru7jyPnG1vCT1h9BwkElwK4A0AvgA+IAQvsRx3PYCvACwnBHkSCWJM/wewnRA8yKWzbNkykpeXx8gZdAb4yGsBWRg778c8f+bihPLj3XZhYSmee+4N7Nv3u3nTGolEgiWL5+Efj/wN69ZdD4lE4v56U0pANsNlfjlT1lnbDc8uwGBAHPTL/obFR+9F0XkfYOH5a72+3t5u25Pa3mq7u7MVke/MQ07y3Vh192uMZUtfvRhBugHM+Heu19TbU7ZPvHolpg/mQv9gESInTGQse2LzFYjWNCElvcTldfNUm3VsmIpJUKBl3VEkTbNNtG5ZPu/797As/3E0rD2A1Lkr7K5X86YF6AlOwZLHfnBZvd3NO1s295s3sLw4HQ1rf0Xq3OUO2z6+40WsqHwJ3feXIWZi4pjV25tsyzekoiliOZan7bLbdlXeQcz84RoUnvMeFl148yhuvJ1H2e/cgwWdexCU3g4QySj+xKtXYd7gn2i/+QCSpi/mtN3X1QnpWzNxPOVerLrzZRu+tigL0769FCdWvoEll64XXG97eVfbJgYD2p6bg96ASZj6f3vh89oUFE68FisfeJ+zbEtNEZI+Pxe5C57F8uv+4XV+OVJWIpHkE0KWWX/v0ZVCiQS+AN4GsAbAHAC3SCSYw3BcOIB/ADhu8bUGwH8A/EtoPRr/08jJt2xucYqzh296rslttrl4Ll2htk2cKX1Fb285amqOoqTkN3R3l2L3Fe/itttuYJwQ2mvbWb7+sXpOnutc4DtP+Hgubaayg/09mKxvhipuMUJ+XYBBEozhom8c1hbiMx8vxGehtsertlDb7op17ZGd8JUQTFxxI2tZbUAU6459p1N7N1bkYdHgHyhNvBmKTYO2hYxQD4Qg1MDOeyrWfDwb1yNvwyTQu4K2/r6Vt7yutQAa4o+kGYsdqpfaLxKBLPnQPNW33Kbd9Cd6IIXh0yindP3CJwBgzx/KV7fxfh1VDfYhDj3QR01j5Nmg/JpeEBnpt81t7M6+xcc7ox0yUAfKLxE+vr42fPK6t6GSBEPz9QOo+3ctp+0Tr3wFHwmBdOZ5jPzgl/TOKCODtjsDe2XfYuEbynORSNqhnnYFqAw5qoMXQab4k7dsp3GX4OjpK1xSN3eeR3xl2eDpx0fPAFBLCOoJgRbADgBXMxz3HICXQU8EAQCEQEUIjlp+5yxkD8g4+Yl3MP8KzMfZww8cZ3+0SKhtLp5LV6htay4wMBDTpqVi3rxZiIqK5NTWjWjd2t59f/Rx8lznAt95wsdzaTOVbSo5Ch8JQeiUlZj8f1NQFXEWZvQdwYh22CFtIT7z8UJ8Fmp7vGoLte2uWIfU/oAWSTxS5zCvVsgekEEXHINIA3OeqvHY3sRgQOzfoll5tvKKnzZBgwDMuuYJTm2fuBiEE6VTdePjPdGvW8uzAQBdhihMp/YyXossy4f3lqHZP3XU1vb21GvYPwIhOub7hKf6lru0EwYK0Bi6CAn/x7zKx6cbGEHnw1T12E5u7KnbeL+OtteXAQACJs1wyHbqenqVWzdo+yikO/sWH++MdpymEX2hUxj52ElJqFvyb8zUVaJr+hectonmOIaJP6YuZp4UTv07vSOpQWU7KfTGvsXGdx7fBQORYOo5N6Hvjz6ok89DsqENVKNtGgfLsrqWAmiJH5JnLmbkHa2bO88jvrJs8PSkMAGA5dJOq/E7MyQSLAGQRAh+dFclFN8y50Myoedn9ndouDh7eCFlvVXbWd1hzRC6t6xG/q5Mt9TLHnCdC3znCR/vaFllLT0Am7zgXCi+VcB33rWIhBKV2T+7XdteXoiuUNvjVVuobVdry9sacPytOzFHU4B6rB71Xo51WRISgxDJMDRq9omOI9r2cPbwzpbNfv8hqD5Ziv4e9kTp1uWbqwuxeOB3FMnWImpCPKe2ti8EQZIR1vby1vOMjVM15QMAchvvRCz6UHpoN2t5+v2bGvRGzLHh+Oo1EhCJcJYfH/jKupN3te2O5hrIiBzaxJVOa4dE0j+EavqZ84fy1W28X0f7Wumt+KOSZjPybFDvH4aaBEKidrwO3nSeDfb3YBIUGImZwWp76RX3oCh4BZY1b8Hxt+5CF9XIaDshvBR1gbMQFMycM3DgRyUGSTAk6m7eejnKCynrKB/fth8VgfMQOymJ/v9S+p3wllzbfS4sy4b1lqHJPxX+AYEuqZs7zyNn29vPqVJjBIkEPgBeB7DeWRuHJIfuAXCPKkCFI6FH4B/rj9RnUxEyOwTVD9A5ZkKmh8BwjQ9K/lJIi4b4Yv4381DzSA3UlWoY1AZEnR+F3oO96NpNDx4S7k+A/0R/dH7cia7dXYg8NxIJ/0hA2fX0r1Z+0X6Y+8Vc9P5ysszMD2ei+7tuKPbSwUpKS8KEtRNQdGkRACD64mhMXD8RFX+lXxD1CfGF7G4ZKu6sgLadThw9+/PZ6NzeiZ5femBQGxA8LRhEQ9CSSc+tY6+KRczVMai6uwoGtQGDeYOY+e5MlN1aBl0P/SL63K/nImRmiFk3ZUMKRjpH0PZuGwBgwtoJCJgYYOZDZoVg+n+no+S6UhjU9OMDqc+loiGjwbzqN+XFKRiqGUL7R+0wqA3wCfRB+NJwc26XsEVhmPrSVBRfXgyD2oDiy4ux4McFqHuyDspCesA0clcBlqEb9b31KLq0CPF3xSN4ejDqn6IfK5CukCLmihhzvazjBAAJDyaA+pA6Gaf7ZPCXaFHzHP0+aejcUOiUOhSuLgQA+Mf6Y+HPC1F2cxmGaodgUBno3JFfyiHfRd9kU9JT4BPkA+pdCh2fdCD2yljIHpCheE0xACAwMRDz98xH19dd6PiEfoRnwb4FoN6hoPiejvWUTVMQd3Mc8pbR9Yi7MQ5xt8Sh9Fr6BXH/KH8kPpyIokuLMKIYoe3ekI8mgwxdlzTBoDIg+c3zoSKB6Pn2M+Q9moBJt09C9JpolN9SDoPKgMETg5i9fTYKziuAXkXHaUn2EoRMDzHrznhnBtQVarS+2QoAkN0jg3+sv5mXLpdixrszkL8iH0RPIPGVYPqbdL7AgVw61rO2zcJA1gCorRQMKgP8IvxG9afIcyIx5ZUpOLHyBAwqAwrOK8Diw4tRsb4CqlI6me6cL+egZ1+PuU2TH0tGQHwAah+tNfeFqL9EmetlHScASMlIQfPmZps41T9Nny/hS8Kh7dLaxKnkmhIMtw7DoDIg7pY4mzgZNAZzvazjFDwtGHN3zIX8S7k51osOLULzC83o+YX+QWLa69MQe22sue6WcQKAgEkBSHw40SZO9Y/Vo++PPhhUBkhXSm3iJF0lReWdlTCoDFBXqG3itPT4UgRNDjLrWsYJABIfSoRvuK+Z9z13GMNTt2GxfA+WwICjyksw46qnGOPU8UkHDCoDtHeHAwCyrjyI8P54RF8cjeRnklG4uhAGlQFFlxbZxGnet/Po9trWgY5POmziFHtlLMJXhpvrZR0nAEh9IRWNGxtt4tT4bCN93q6SQtOisYlT0aVFGG4aRscnHTZxSnwpHnOpbyGVqPDHC39H6uS3RsUpdF4oZm+fjY7tHeZYL8legtaPN2KCnz9GvrgKA4sHEHN5jLnulnECAO2ZwUA0kHPlEYT0x5njZOpPBpUBkedH2sTJ1J8MKgM0DRpzfwIA31BfLD68GAGTAsy6lnECgOTHkiEJlJh5yziZ+lPqs6mMcZLvksOgMiAoJcgmTgFBJWjVT0RU/sWQp2yDT8HHKHlnoTlOpuuefIcc1T/k4IIzh6DVzx513QuZffJ6ZBkn03Vv6qtTof41BFLfQeQuy8H012dA265F8yt0bq3I8yKhrlHbxKngvAKMyEfQ8UnHqP4EnLzumfq1dZxM1z1qK81bx8nUn6IvjjbX3TJOgHEc8YDBJk6m/mRQGRC9JnpUnNS3HsYkAIOfpYCqp6Bt19rEaeHPC+EfdfIabRknAIj4Jz1AlX9dibz0PMb70+SnJ4/qT5b3J4PKgJBZIaP6k+V1z6AyoOzmMps4mfqTYo8CHZ90YNrr00bFadLtkxAgO3mOWsbJdN2bvmU6atNqbeJkuu5FXxwNZYnSJk75K/KhH9Cj8/NOjNxbDQORoOPuEXTr8k5e94yxto6T6f5EvUtBckME9PV0v7O87iU/mYzWLa2j+pPl/UnIOKL+6XoYVAboenWM4wjf0JPXaKZxhEFrGNWfBhdXYi6A3v1RKNtXhqRHkxjjNPDn48ha8j+ciT3Qv7cXv/eugV/ROky5fi6i10SjYP1xrLyoHof6bsYcgPH+1PNzD3CDFD79dJ5CV40jACDmcjpPIds4ouvrLqfGEcnPJJt5/1h/RP5XjxRDC/bVX4ahZXlIyUiB6lsl2vWxkOTuhyJwPeM4omhNIWZcXoMi9dmYDtg1jmh8thEGlQEGjcGpcYRuUMc6jgidF4qEBxKcHkewgmlL0rH6AGQVQPZb/P8pgDxl8f8IgCgA0mj8aABCAWSZxTHrAfIWnxZXSorOHZ2EdDQQMtxAmFIadO0pYvyej7OH79xZ4DbbXDyXrlDbzvpcsWUFIa+GkcLXznddvYYb6Nga0b693Sb+lujc0ekUZw/PpW1d1qDXE0VGknn7fxOf98pVRJGRZM6RY4+2EJ/5eCE+C7U9XrWF2hYa6+7OVpL17v1EnR5LRtIjyfHMm0lbQyWv7c4dneTE/k8JyZCSmsI/nNJ2hrOHd6a9c797l85RteEc+u/Br3jLt9aVk5H0SJL1zj12aR95/U3OLdzdHWtneTaudcN0kv/KlaR9ezs5tvUfRJceQdqbaxjL5/34EeO5Yk+9sj5NJyRDSgb7e2x4T/Utd2gf/++tpD8jnuh1OqdjrR3W0LlFP/yXU3Ub79fR3NeuI1TGVKe0K59bToo3rXZI117bzvKOah//ZgshGVLSXF3Ia7tzRydprSsnOZk3EV16BFGlTyDH/vcg6ZFTpOj33YRkSEnx4T2c5aueW0aKXrzAIZ/s4ceqXx/76HFCMqSks7V+lG72G+vIYPpEm/QdprJt9RWEZEhJ9q5XXFY3d55HfGXhpSkpcgFMl0iQKpEgAMDNAPaaSELQTwhiCUEKIUgBkA3gKkLAvI2ok/AJ9gGkE4GeYUCrBax2ZPUJYm8mLs4evv2jdrfZ5uK5dIXadsbn9qYqzNIa3w3geAfH7noRQseyZ5iOrRGmX0xYywdz+MXB2cNzaVuXbW+qRgz6YUhYNpqfcw1i0I/KnAN2awvxmY8X4rNQ285qE4MB/d2dXhNrV/Jcuv09Xahq3ICgtxdjefsXKI1cjY7b/sAZj3wJWcpMXts+wT4INr6/pO61fX/JG2PNVTa49HO0SiYhet5naPJJRNyhJ6AcsE2DYFm+7fvnoYcvpl39tF3a/sH05iFDA7aPW3HVzR5+rPt1f08XEkgnhifMR+ubrUi+8F74SggafrXdsc8n2Afa1hPQEl8kzVzicL18QmMAAAM9to9EeqpvuUM7vu8E6kMWwMfX1+lY+wcEYoCEwofjMchT7RpuWTZC1YiuoGSntIcCohGi63NI117bzvKOauvlldASP8SnzOa17RPsg4Qps7H8kR2g1h1GecQ5WEF9ioC3FiHk6CaMEF9MXbKas/yQfySCGTaB8ra+xcbHte5Hpf8cxCWkjtL1n3kJwiRDqMn/nbFsRxX9Ok/U1OWMvDN1c+d5xFeWDR59fJQQ6CQSPAhgP+iUFB8RgjKJBM8CyCPk5ASRCRIJGgFIAQRIJLgGwCWEoNzRevhF+gHBwQCSgd5OQK8axftqhui0Bgzg4uzhfXrVbrPNxXPpCrXtjM/t32xDnBxo9UtEsGHQNW3iGwBEJBtjax/8Itm7BBdnD+9IWarsCGQAYmaeOYqfdc61GDr+OJQnvgLOvMwt2o7wQnSF2nZGuzxrH3DwBczRlqB4zleIxcVjpu0q285q139wO85SZ6FAuhqxV2zA8pmLHLLtF+mH0Gj6BxbtAPv7S1zlneHs4R0t21JbgrnaYmSlPojZE6TouPS/SPpxLXI/eRQrHtzGWL69qQqLe/bhxISrsUI22S7t4IhooAMYZpkUeut5xthmFdmIABCWshTDABKmzEZp4CKkNH0Ng34TfHx9R5UPLS9Ds18KpgWFOFyvgOFYAICqtxMw/mBhDzzVt5yxraCakEQotMluEqzdL5HCT8P+Pv2pdA23LEsMBsTr2lAWvcgpbW1AFMLVtpuL2KPtTt6RssF9tWjzTUSqcTMne+ORNH0hkh79Gk0V+ej+cSOWKA+jRDsb88MiOMtrA6IQN9TAWy9HeSFl7eVba0sxVd+A7CmP2hwzbcVl0B3zQX/JPmDVGpuywy0FGCG+SJ49OouDO8dPnhibeXqlEITgJ0IwgxBMJQQvGL9LZ5oQEoLVlquExhXEaEIQRggSnZkQAjC/j4bgYGBiCp3TzuKj6kyw+c4ezh4+5u5VbrPNxXPpCrXtqM+6CSlI6j+Ckugz0RUzF4EBI66p18QUmwmh7B7uHZnM54KDnD08l7Z1WV1TDoZIAFKMO0Ga+NDwSFSGnYFUxUEY9Hq7tIX4zMcL8VmobUe0K3N/RcmLqzFn/81IGqbfN+hrYE/uOpaxdiXPpZusLsNxn4ux9J97MJlhQshnW12phjSGtq8btN2YxVtibU/Z1t+2Qkd8MP3ie6CuVGPWGRcjZ+JarFB8g3KrjZxM5Zv3bgIBMPnqZ+zWlnTRuW9HVMwDdnfFWqhtJk7ZQN96E2avMGsPL7gN8ehC2dHvRh2rKlciSVONHuksp+oVFEGnWRjqd+w8c2ffcrV2YwH9tEfM3At4y/PFepBEsKbw4LM9Xq7hTGW7O1oQJhkCYqYx8nza+pBYRJF+EIPBbl17bTvLO6o9QdOAntBUu2wzcZNnL8WSf+1F862H4SN7g1dbHxSNCGK7M7A39S02vuXYTgDA5LNP5qU06UojY1AdMAcTOo8ylg3tLkOzX7LNJjzuHD+50zYbPD4p9AZEnh/pNC+kLEBvkOAu21w8l65Q2476XHroK0xAL8ji26ALjEQY2FcC3dnefOXHMtZRvcVoCJxp3srdktfPugpx6EG1xWMO4yXWrrRtj3ZNwREUvXQRZv14PRKG65A9/VEMrj8EAPCL8cx55ol+3dvVTj+OPGk2I2+P7cjzIyGNjMEI8QVh2JLc07G2t+yIdhjT2/eiJHQlYmWTzfyC218FJZmI8P1po3YLjTw/EvK2BixW/ICCmMsxKWn0AJRLO+5s+rE2vdr2sVSmujnCj3W/9ussQQdiER2XYNaed+Gt6EMYRnK3jzp2ZH4vojAIEr/IqXqFRNKTQi1DPjRP9S1Xa+vrj0JJgpE6byVveb5YawKjEMoxKRzP13Cusp0N9MYdITLbHx/s0ZaEToC/RI+BvtEr+e7sW3y8I9oatRLxBjm0UdPtss3FJc9YhMmr5/Jqk5Boxh2ovalvsfExzT+j2m8G4ieffPrAUncg4TxM09dB0dEyqiwxGJCoqUJ3uO3905PzByG22SBOCgHIv+B+FIqLF1IWAOcuQEJtc/Gcuw8JtO2wzwWfoAtRmH/+jSBBkQiXDEGv07m8XozaDpQfq1gPa9RIHalDf8xCRn7muWuhJX7oy//aLm2virWd5YnBgIrPsm1+wbVXu6HsOCo+uwbTv7sSkzUVyJryMAL/WYKVf83ABOO7BMNN7O/Vnmr9ur2O3rFX15bktG35F3JIfHzQLwmH75Dt45DjoW8BQMnBXYhFHyRL7xjFh4RFoOeCV5BEKBR+8vio8vV7XoAPDEi6cvQqIZ+2eq8OBiIBGeqzq26O8GPdr+NUlWgPmTFKOzAoBJVxl2Pe4FH0yNvMxzbuPwwAiJyyzMaOPfUKj6IfU9YpbSeFnupbrtae2JuP+uB55h/+hMR6SCNFuKHPqbqN53umkqL52MlzGHk+bb/wOABAv6JtFOfOvsXHO6LdVlsMHwlBwKQ5jLyj9bJH2zeUfrS73yovpjf1LSa+vakKM3TV6E6+lFU3ZhH92Ghjzg+jysqpBkRjAGTSApfWzZ3nEV9ZNoiTQgCJ/2RPGsvHCynLB6G2PaXtiK68rQHzVdmolV0FP/8ASIIjAQCDfcwvzbvTZ77yY9XeDaVZCJDoEJRyBiMfHhGN8pBlSOn81TxpGg+xdqT88R0vYKnkWmRve8zhsqV/fIf4XZdjZlAZsibfB9+0Eqy6/TmEhkcCMG7MgFAEx7E/XuGtfctZ7cEW+hf1aX89x2nbJm7QJxL+w47nA/WGvgUAvoWfQI5ozDv3Oht+3jlXIyf6Sixv/wLVJ+iJTfBtBIvke3Ai6lLIUm1XJLiQ/K/JGJSEwGeIeaXQW88za0450ItEPQXNBNtB0aTz70GARI/qAyc3nAlMqIOO+GDynDNsjrenXtKoCfRkWu3YeTZe7pk98jakGFqgil/hEm1J/EREkEHWH9E8cQ23x7bQ9iaKGmiIPyYm2j4+ao92oDHHo4ph4yy+su7k7S3b21QCAIhJnW+XbVfE2i+cXsVXWk0KvaVvsfFNR+lHR5POvIn1uKnzz0QPpEDtr6PKtlccBwBEMPzI5cn5gztifcpPCg9JDl15SHJoq6ZZg74jfVDsVUC+U46BnAG0bmmFpkWDwnMLYdAZULGefseo4jbj3/UVMOgMKDy3EJoWDVq3tGIgZwDynXIo9irQd6QPJZeXQF2jRuPGRuiUOlTfT+dYKV9XDgDIX0In+61Nq4W2S4vmzc1QlijR8XEHeg70IGBSADo+7oCyRInmzc3QdmnNuf1MZU22qu+vhk6pQ+PGRqhr1Ci5vITVp/qn6lH3zzpWn2AAq0/UVgrVf69m9al8XTnqH6tn9an85nJWn8rXlUO6XGq2VbIlE74SgoC2a6CuUWO4mf7VtG1fncM+1T9Vj5r7a1h9UteoIZFIWH0CgIIzCxh96jnQg9KrSll9soyXdZyorRT6jvQhYEIAo08GnQH5S/PNPvVVHwMAyGafg/qn6qFp0aD40uJRPimlf8EkdKFs/69o3NiI2kdqWX0iWsLqU8fHHai6q4rVJ1Os2XwqX1vO6lPF+gpIl0sZ42Q69ypvr7SJU+2RPCys2oI+EoZVLR8ga9sTNj7VptWi5sEaG5/yPtqNqb/+DZSPDEV/bMfSK55D54YeG58GJFIYGjsYfVLsVaDs2jJWnyxjzeTTQM4A/KP8bc49U386seIEY5xM/anooiLGOJnOvZoHaxjjBACGIQOjT7raMjrH5VNaVp8qbqtA/WP1rD6VXVtGn3uIQKCqx+a6Z9mvrX1q3tyMqruqnPKp+v5q1D5Syxgn07nnE+Rjl08nnjyCeepcVEjWQF2ggXynHOVry0fFKaznX+iWRMF3z4PQDmtQ+loG/KHDSMk6G5+UJUr4Sf0YfdJ2aVFwZgGUkjAY2hQ2PumUOhSdX8TqU+uWVtTcX8PoEwDoB/WMcTL1p4q/Vtice5bXiPrH6hnjpCxRovSq0lE+1f76B70iEbsQtWm1o2KtfiEElX6zEV+9EyMDWjRubERAUxGakQhNns4hnyrWV6Dun3Vo+k8zBhAKfUunzbXcN9TXKZ86Pu5A+c3ljHGy957rG+rrknFE1Y8/AgD8R5bbdc/VD+hZfeo50ANleSD8JXq0ZzczXsst28XRcUT40nCn77lCxhE9B3rgJ/VjvefWP1YP/5YatPsmoPbBOpvrXsVfK1h9Mt1zQ4yTQk1fxyifQmeHco4jis4vcnocwXfPNcXannuu6kQhdMQHwx+Hmc+9+sfqWe+51uMI6/tT2KIwVp9M99yhHHoDk87f61w2jlDsVcA/yp/1nsvlE9s4wvqeG9m4DzX6FCROmzfKp9B5oeY4yT/rQo3/ckzpPw5qWxuUJUoUXVQEdV0eDESC5Dln2LTLiRUnWH2S75Sj8vZKp8cRpnsuk0/V91eP6teOjiNYwZSn4lT8cOUpFOEZ6HU60rZhOinZdK75u4JfvyQkQ0qq8g96rmIeRt6r15D2jCmcx/R1y4k2PYoc+9+DY1SrsYFepyOlL5xNBtInkvbmGpLz+lo6B9fH/+YtW579M1GlTyANG+cSRUcL57GVz51Bijed56Jaez+KN60m1c+55hqY9+rVpHnDLJfYGmsc+/AxQjKkdM4pDpiuQ9lv3klU6RNI7mvXO61Z/ewSUvjSRU6X9wZkff4cIRlS0tXWyMgf/+YNQjKkpDz7Z0IIIfKMyXTfFYDmDbNI3qvXCLLhrch6626iSp9AhjVDLrGX+907o3LVnS5o3jCL5L9yhdPl5W0NdD/fudl1lRpDnNh8GWncOGdMNRsr8gnJkJLcH7aOqa4QdLbW0+OIjx7nPdbUl6oLjpi/K3j5L6Rx41x3VnHMAS/NU+gVMM2wneGFlAWA/BX5brPNxXPpCrVtr8/lx36AjHRCM/+vZi4oLBoAoGHZwt2d7c1XfqxiHa8sBRU2l5UHgIjoCagIXoykjgMgBoPXx9re8rnfZGKuthgVC55Az1NaLHnoC+RJL8Kq+i3I/nwja9nKvN+Q/NPtUPjGIuyenxAzMZFTe8g/EkEsm38w1ctR3l39uqbwD5SsK3JYd+JwI/pCp7gk1rrAaESQfru1HbHtLG9Pe+t1OqQ2f4OSwCWjHgNlsr3owpuRJ70IKxRfIwhaTLj8305ra/zCETRiu1sfm7a9/Fj2a9/OYigQiVhjKg5r7bkX3Q4lCYby2EdQUE2YgF7oJ9o+aupIvdS+UgRobfuop+6ZrtSe0JOH+qDZCAgMsqs8X6z79tO5lVU9zI9BjuU13BHbQrSL1xUi3tCB4YipTmtHxEwCABiUo3e5dWff4uMd0Y4ZakBPcIrdtl0R63BjWiK91SZQ3tK3mFD06lYAgOzMW3h1U1dcCQBQFP5oti1TV6ErjPnVAU/OH4TYZoM4KQQw5zPbl5Tt5YWUBQCiJ26zzcVz6Qq1ba/Pwznb0IcwzLvw5KQwWEonLdYqmSeF7mxvvvJjEWtFRwtkRA7tpCWMvCU00y5HIulAfWm218fanvIdLbWYW/oqSgMXYfm1/8Ccz+bA188Pix76EifCzsPKmteR/eUmm7LVJw4j4fu/otcnCsF/+wmxk5J5tUcCoyD1HXTaL0/065aaIkzfcwWGrj3okO5AXzfi0ANdzEyXxNoQEgspVBjRDvNqO2rbWd6e9i47+h0moQvahevssj31trcgRzTyIy9hTeFhj7bWP4LOvcpRN766O6or1LY1FztYibbgGazaoeGRKIu5GPP6fkdjPp3SQ8rw/o0j9Rryj0SwznYy7al7pqu0+3u6kKprxOCklXaX54v15L/T75QN9TNvLjFW13BHbQvRjkwH/CQG+E2Yzsjbox0QGIR+hMJHPXpS6M6+xcfbq60d1kCmb4cmcjoj70y97NE2ve9rUHXbcHxlndUWals24Q80+iQzXsetdWMmJqLWdyoi244AAOJeDafvnyw/cnly/iDENhvESSE8O9OX+ErcZpuL59IVatsen3u72jF/4A9Uxl02Ku9LaBT9ErNexbyK48725is/FrFuLfkDABA5/Uxe29POvQl6IoE8Z7dXx9oe28RgQOfn98EHBkTf/B4kPj5mzs8/APMf3o2C0LOxsuplHN/1irlsbdFRTNp7CwZ9pPC76wdMkKXYpW3KtcS2MYM39uv2Unrjk+HcYw7pUrWFAIBg2RyXxNonjHn3OW/vW7q87ehFOOZdcAsjb42oCfEI+1chgvLZVwnt0dYFRiDUwJz+ZDxcw4dUg0jWN0MdM49TO/qcvyNYooXsxGvG929W2BzjSL1GAiIRpredFHrqnukq7YYTv8JHQhAxa7Xd5flirdiiAQCMDDBPCsfiGu6MbSHaVVvo66E0kTnNjr3a/T6R8NOM3tDInX2Lj7dXm6ovhZ/EAH+rNEPujrWffwAGJKHwsdqB2hv6FhMUHS2YNVyK9oRL7NbtmnQOpmsrMNDXjcLn6ZTp4anMP3KdaiuFHn/Xb6w+4juF3oWsz58lJENK6kuzR32vGVLRz35ve8JDNfMsjr33ENGmR5Eh1aBdx5dsOnfM3ylwB3L2vE1IhpR+d4kFw5ohUvDSJYRkSMnxr14ntcVZpDdDRtozppK2hkqH9I59/G9CMqRENdgntOpjhuwtt9HvOjy7xKFyOV//l5AMKWmpKXFJPfJ+/Iix73ozFB0tZDg9imS9c++Yax9770GiTY8iBr1+zLVdgcrc3wjJkJIT+z/lPM6g15PaZxcSkiElTRtnC9bNeudeokqfINiOtyHrnXuJJj2GDKmVLrM5pBqkr5/bnnKZTW9H1qfphGRISV+3XJCdsufPJKUvnO2iWo0dTNfh2qI/x1y7aeNskvfKVWOu6wyyd73i8P2qLGsffc37eTs9Fs2Qkv5ehRtrOfaA+E4hO0w7JTnDCykLwLzTkztsc/FcukJt85Wtur8Sk2p3ospvJlLnjv41OTAoBEMkABKWvF7ubG++8mMRa6miEI3+UxAUEmaXbdWUyzDZ0Irsf/3AyPPpctm2h3dFeys6mjGj4HlU+s3G8hufZLUdEBiE2f/4FsVBy7GseCMmfHUthhEI/e3fQ5Yy09o0p7ZvKP2Ycn93B2u9+OrNBXf069i+YgDAZF0jtMMau3X18ioME3/Ep8xySayDIuIA2G7j7s19q+aX9xEg0SN+9d/HXFsSHAV/iR5qle2q13i4hvfV5wEA4medvFYzaUt8fKCYTm/3TmmZ3/NypF7mJNlDqlG8p+6ZrtKO6c5DXeCsUU/I8JXni3XrMx1Qk0CAIX8on21vvWeWZ/+M7Id+ZuV1FRXogRQR0RMEaWsCohCqG/1kkjv7Fh9vr7a2oxwGIkHCtAWMvDP1sldb5RuBwJE+h2x7aiwcUvsjmg3xSJm93G7d6UvOh5IEY6TqF/jWFqJVEg9pZIzL6+bO84ivLBtO+UmhPSkpiIFwbiVNDIQ1fUNAYgBnSgpNCz14Y912+Tf27bFNZU22rLfzDUgM4NxKOunxJFafuvZ0caakmHDzBM6tpJOfTmb1SbpKypmSorH9T6QYWtA/+1bGLYoHEAbS0+OwT/VP1WPiHRM5t5JWfKfg3EpaP6Rn3Uo6aEoQZ0oKU7zYtl3u+YXZJ4POgOH2Yeh1OqQMV6Encr7NFsX+E/wZfZqYeBkAwDCnmNWnrm+6OLfHjrkmhnMr6eSnk1l9ClsSxpmSYiB3gHMr6ahLo9C49T4EES0Ms19F50dyc39KeDjBxqe6vzVixsN7UGRYhCFJMFr930eEKolx22XF9wpWn/zD6UcgB9rbGbeSDpkTwpmSYrh1mNWngZwBdO/vZt1KeqR3hDFOpv7kG+5rE6eGD6uRomtEE5EhQKJD7jM/MdqQfyW3iVNQVzVaJQno/roHUZdGcaakSH46mdWnkDkhUOxVgLTTP1iomqhR14iB3AFWn5o3NyPmmhjW7bHj74nnTEmR8HACZ0oKxY8KVp+SnkxEfPUuVPrPhv6TKJvrXtiSMNYtvy37NZNPyhIlun/uZk1JoR/SwyckCgCde9X6Wi7xl3CmpJh4x0TW9A3y3XLOlBQR50dwpqRIfjqZ1aegKUFmn/SV+egl4Rj+wt/cn5hiXX1/NaaftR49JBy6pHOc8qlifQWSHk9C/VP1IIgEAFB/1I+67nX/1O2UTx0fd0C6SsqZkoLvntv9U7egcUR3RTumjNSiJ3iJQ/dc+S45q0+me24fIuDTr2C8llu2i6PjiP6cfqfvuc6OI2q/LsLMfTdjWfQtyN58Cwo3/WHjU3RwGyhdAqNP1FYKEedHcKakGG6nr+FKRRgiDP2jfOr7s487tZU/d2orrnEE3z3XFGu+e65/RxXaEQd0+43qT8lPJ7Pec9nGEab+1J/dz5mSIv6eeDRvbsaQJALBmj6XjSMUexXo3t/Nes/l8sk0jmDyqWZDJWZpilA3vGrUKymWPvUf67eJ0+AhFSp9FiOpJwuJvrXoCJ7Jen8a6R3hTEnBdc/lG0eY7rls9yfLfm3POEJMSWHn46NNLzdxLrNy8ULKEkJI7tJct9nm4rl0hdrmK3vk0WuIMj2OKAd6Gfm69HnkxMtrXF4vQoT57e5Y15dm01s9f/eO3bZHtMOEZEjJwYz/c0qXy7Y9vND2PrhxC/248Hbbx564bBv0elL3Yo3T2hXZ+wnJkJKig18x8t7Wr0uP/nDy8doMKTn+Vabdum0bppHcV691WtuaU3S00Nu4f7mJV9tR287yXNpHn/2CbrNv3hhz7aaXm0j+vm2EZEhJXYnt40vj4Rpe8+xiUrxptd3aI9phl8Ta1G61xVl2a7vznukK7aLfdxOSISUlR/Y4VN6eWFc9t4wUvXiBw3Xzxntm9o4XCcmQkj8evZIMp0eT4fQokr3ldtLRUms+pis9iRzPvFmw9rEPHiX69Aj6XmqEO/sWH2+vdv3GBfTrFA7YdlWsj//3FtKZkeKQbU+MhWuLs+jx0YbXHdbN3vUqIRlS3pRYnpw/CLEN8fFRdkSviXaaF1IWAGZtY97m1hW2uXguXaG2ubjB/h4sDfsTZdEXITQ8kvEYjV84Ahh2nRNaL0CY3+6OdVfFnwCASXPOsdu2n38A+hCGwDD2nTQ9FWs+7f7uTswzvIJa36lYdkuGQ7YlPj6Iu3yS09ohUfQjkNqBLkbe2/r1QF02AGDGRXdikASDtDOnpbDWVSv7ISNyjERPd1rbmouMmWTcfW70luTe2rf8Ar6GkgRj3sV3jLl29JpoBBjT7Az1Kxh5Z7XHol8Pa9RI1jVCGT1vFM/Z3v4BLol1YDj9WKD1jpqeume6QltVcxgjxBdTFq92qLw9sR7yj0TICPMGbe66hgu1zcaH1f2EJp8kLFr/BXr/noOC2CuxWPE9It9fgey3/4amqkLESvphiJ4mWNsnbAJ8JAR93Scfh3dn3+Lj7dHWjWiRoG/FUKSt/2MRa31QDCKtNmrzdN9iQk/9CQDApBVnOKybfMYV5n+HpTBvMiOkbkLLCrXNBnFSCEB5gnlnOHt4IWUBYCCLefLjCttcPJeuUNtcXPmBbQiRDCPirLtZjxkyhCFExzzJcWd785V3e6xbc9GLcCRMsd1KmDOWPhHwUzG/S8Kny2ebjxfic/UnDyGCKCG5+i34BwS6tF582tJoekKpU9oO1N2t7YztoI58tEriER2XgCbJVET2Me8sZq3bVku/hxgYP9tpbWvO188P/ZIw+KhHt5039q3+XgUWqA+jLPYShIRFjKm2qWywlH5UeVjZw8g7qz0W/bql6gQCJHoEJI1OkTMWsTb9cDNslbPWU/dMV2hHyXNR5z+D8VwUGmttQDTC9Lb5Q/lse9s9s0fehlnDxWiXXYKBrAFMTJyKFQ99DMWdWSiKvgTL5F8j4YvzAQCBk2zfJXdU2y+czrs32HPy/XJ39i0+3h7t9sYKBEh08Jtou/PqWMRaEhqNAIkOKmW/DecObWdt69tLoSV+CGtPdlg3fvJMNPkkAgCSWHZSFlI3oWWF2maDOCkEEBAf4DQvpCwAUFspt9nm4rl0hdrm4kJqf0CjToYZS1azHjPiH4EQlrxe7mxvvvLujnXcQAmag2dD4mPbLblsq3wjESzpc0qXzzYf76zP8rYGLO/fj+OhN2DqgjMZj3Fne0ujJkBPJCBq5sm0N/VrYjAgSV2GjnB6taYveBaSR+qhG9Hy6vY1lQAAYlIWOKXNxg34RMB/ePQkxxv7VtWhLxAs0SLqrLvGXNtUNiSC3qBAr7KdFHr7NbynNhcAMGnW6F/axyLW4ZH0pND6hxtP3TOFahtitJg6Uo3eOOYNL4TGWh8cg0jSz5hmxx3XcFfYZuJrj+yEr4QgbuVNo7RlKTNxxj++QOcdR1EQeTG6SBQS550tWNu8cVZPu5lzZ9/i4+3RVjTQ1/WI5HmMvLP1stdvn1DjO/kWq6vedM80IaSvCi1+yQhJCGXk+XSpydegTDIXkbHsTyV5cv4gxDYbxEmhiDHDiHYYUzVlaFQuZJz4mKDzlSKcOPcrx3iFUtOLZH0r1HFL+A+2giYgCqGGPtdXyo2oP/IlACAy6TaP6Pv4+qIf4fAZsh2oexs6W+sQiz7oE4yPsITNR7BEi1bjKiAXdJ2VGCG+kE2Z69I6qXwjEajtc6lNdyCw+ntQiMP0Red6rA5hkcbcq2rvP9esQahCDCAEshTmXHDuhDSGXsGxfkx5vKK18xj8JXqETj/PPQIhMQiSjDDucjueEFT7A1ol8Uidwzx5TpgyF8vTdsH3rBOj8tI6izDjUyOa/k6eI70HwxT9pIhs2kKP6AdI6Wuaste72yxeU4eesOlOl191xwuIX8W+s/upCHFSCEDbbvuLu728kLIAkPhQottsc/FcukJts3H1xX8iRDIM30T2X/cAQK+XIkQyzLjtvjvbm6+8O7Vb64/BR0IQNnWl4/UKjIaUsA8EPBFrPm1pw09o9ElGrJb9gu3uWA9AarPaNRbajtpuKz0CAIieeRYAIAz0r8Nd1Tm8ukF9NWjzlZkfz3VVrOlt3Ps4tZ217QzPpD3Q143Z6nzU+Z3H+SOUu2MdFh4JHfEBYUiz4+3X8Kj+crQETLNpv7GIdUBgEJQkGJKh0e/JeeqeKVRb03EUeiJB6pILHC5vT3v7htGrN/0K2zQ77riGu8K2Nd/f3YnZQ4Voib8YEh+fMYm1NFYGANANnHx31Z19i4+3R9uvpxodiEWYNMoh266KdVCE7fu+3nTPBIDernZMQC/0E+Z47DrKx3vSNhvESSGAsCVhTvNCygJAyOwQt9nm4rl0hdpm43rLDwIAkhdeyGnbL4Z+3Gqg13YTEHe2N195d2obfItgIBJMXsC8osFl2xAcgwgMwqDXO6zLZ5uPd8ZnRUcLZg2Xoj3hEo+1NwCo/SJYV7u8qV+PNOVAQ/yRMod+hC/1rGVQk0Do2wp4dWOHGtETkuq0Nhs3EhSNcMPo95e8rW9VH9mFAIke0vnXCbItNNYSHx8MSsLgo+lzqbaz7U0MBmR/+QKGJrOvwIUtCcOIdhiTRxowGGW7yjxWsR7wCYfv8OhJoafumUK1YwwFqPefhvAIls1VBMbaX0o/BqnstZ0Uuvoa7irb1nzVkV3wl+gRu3wtr7arYh0RHUe/SqA8Od5w5z2Tj7dHO1JVD3lQisO2XRXrsEh6Fd9yozZvumcCQFt1HgAgNGmBx8bCfLwnbbPhlJ8U2pOnsPbhWs78QrUP17Lm9Gvc0MiZp7DyjkoA7PmFyteVs+ZMMpVlyy/UuKGRM79Q9w/drD6VXF3Cmaew89NOzvxCPft6WH1qfaOV0afg9uNo0svQ9ZSK1SdqK4Whan8AAPV9nUM+1T9VD/mXcs78QqXXlHLmF6p+oJo1v1DT802ceQpN8WLLxVN+azlrzqSg1hNo8U1E20NyG580LRo0/LuB1aehlmD4SQwoejiX0aeSK0s4cyZ1fNTBmV+oZ18Pq08tr7Vw5imsfqDaJk5Fb34AHwmBf/dFoLZSrDmTFHsUnDmTunZ3sfqkLFGi9LpSVp8AQDkYhhBdP2N+oeaXmjnzFJpizZZfqOzmMtb8QlX3VLH6pO3Sov6J+lE+RcgLUS+ZhuGqETRvbkbPN71oNKRA2ltuY6P48mJznHpLuyAztEMlmWK+RlBbKc48hT37elh9an6p2eyTpi8MEWQQ9RvqzP2p+oFqzjyFHR91sOZM6trdxZmnULFHwZmnsPSGUhuffCv3otMQjYiK2aw+yXfK0fJaC2eeQuvrsPV1r+ymMtY8hdUP0DYGDaHw0w7YXPfqHq3jzFMo/1LOmtOv+LJizjyFbW+3MfrUUluMlVWboT74d5SuK2H0qen5JlTsPopAyQi0yhk21z2mWFteIzo/7XTKp4r1Fej+odvsk9IQDn9Vz6hrRNnaMtZ7bs++Hs48ha1vtHLmKeS755atLXNqHDE0OIhpuip0SZc6dc8tXlPM6pPpnqstDwIAdPxQbXPds2wXR8cRVfdXOX3PdXQcEVC+F+1kAqI756DnQA/Kbipjvedy+URtpdD2dhtnnsLKO+lYV62vRq8kAro6yuxT5d8qOccRdY/WOT2O4LvnmmLNds9teq0RCboWDPqlMl7Le/b1sOb04xpHNG5sRNW9VXbdc30V4QCAoXrKJeMIxV4Fym4uY73ncvnUuqWVcRzRX0bvPBqXtJjznlv19yrOcUT9E/Wc44iqe6o48xRy3XP5xhGW91yma7llH+AbR4h5Ch3IUzgsH2bl+HghZQnhzs0i1DYXz5ePxtVtohsZIQPpE0n2G+t4tfO//5KQDCmpOH7ApfUiRJjf7tI26PWkJyOBM+cSZyy/e5eQDClprCxwSNce23y8Mz6XbDqXNG+YRQx6vcf6FiGEZL92K+nKSB5zbUdsD2uGiCY9hmS9c+8oPvvNO4kyPY7odTpW3boSY97LH7Y6pc3FmfIl9na1M2oLse0Mb6092N9Dt9tbd7v1OsqkzVSWziF3vku1nW3v/J8/Meffyt7xEmvZnG/fZL2ujFWsi148n1Q9t8xubW+INRPKjv1ESIaUFBz43Kny9rR3a10pIRlSOm4O2PaWe2Z/r4IMp0eNutaNVazrN84flRvZnfdMPp5Pu7GgmM67uvs1h227KtYGvZ6MpEeSY1sfttv2WI+Fj//3FtKTkcg7zvBkrD1pG2KeQnY0b+KYNfPwQsoCQOQ5kW6zzcVz6Qq1zcQ1lGUjXDIE39SzebVVh+m/w0rbnSHd2d585d2lTTVWIAqDIKaNRBy0HWjcPU3dx/zS91jHmku7t6sdszTFaDW+M+KpvgUAakWoTa6lsdB2xHZj2XEESkYQkLJiFC9JWIRQiQZt9aWsur3GnUejJy9wSpuL8zPmkOvvPvmomjf1rao/vkagZATSJde79TrKpM1UVuMXgSCGNDueuIZr2+lfqUv1czCv/HW0N1UxltW3FUJNApE41XaHw7GKtTYgCiH60e9Le+qeKUS7v/IQDESC1KUXO1XenvaWxsQDAPRK29cuXHkNd6VtS776yG4ESPSIXHaDXdqujLXKPwrBFjke3XnP5OP5tOu/OAoAkCbZ9ks+266KtcTHB30S6aiN2rzlnmm2N1iDtsApvOMMT8bak7ZZwTRTPBU/XCuFnoR+RH9a6GZ9tpGQDCnpaKnl1W6uLqRXOL57x+X18FR7c2mbVvpqi7OcsltTeJSQDCm9AuCA7ljAWvv4V5mEZEhJTeEfY65tjazPNhCSISV9PV1jrm0vsr54gZAMKWlvrhn1fW1xls0qoLXusQ8eJbr0CDKkVrqkLpYoPvwNIRlSUpa1j1F7rGGtnf/KlaQrI5noRkbGXJsJua9eS1o2zBxzXca6vHYdac+YStoaKokqfQIpevECYtDb2ip/fhUpf36lS7UdRfZbd5H+jHiPaDPBWe3iTeeR2mcXulXXoNeT4fRocux/Dzit46y2K3Di5TWkMyNl1NMPY6Wd9+o1o/qnN59jWZ+m0/ctRceYa1uiYeM8kr/5co9o80E3MkJU6RNI1tt/H1Pd8QSIK4XsMD2L6wwvpCwAnFh5wm22uXguXaG2mbjAtmy0SSZiYuJUXu2OF+gNLHSqXhvOne3NV95d2j5VP6CHSDF5Fns6Ci7b4TH0ltojFrun2aNrj20+3lGfA2t+ACWZiKnzzxRsW2isB3N86b/dthszeEu/9mvPhxzRmJgwZRSfPHMxtMQPupbRm81Y6gb2VoPyiUdQcOiosvZqc3Ehxo0Ghi22cfeWvjWkGsSswWzUxayGr5+fW6+j1tpsZfWBEQgjtiuFnriGR6oaIA9KQd+/9SiZnYYFmjzk7X1n1DEl64oxWVuLgcg5LtXm46x5Q3A0pFCNysnpqXums9qD/T2YqSlB8yB3+gChsTat3vgypNlx5TXclbZNvGqwD7NVOWiYcAF8fH3t0nZlrEeCYhBhkdbJnfdMPp5PW1daAgUiEWFM2eKIbVfGWuUXgaCRPrttj+VYmGqsoHe6nzSXt7wnY+1J22wQJ4UA5nzGfOOzhxdSlg9CbXtK25oz6PVIVRehLWKpXdqLttKPypEh20mhO33mK+8ObUVHC+Yrj6E6/kr4+bMnG+WyHRlLPzZkUNk+NmQP3Hn+W6K/V4HZQyfQPPEi8xb3nupbABB39QwAgJLhsVtv6VuTBkvRGjp3VEqAOZ/NgX9AIJr8UxHWW8ZqJ1rdAEVwitPaXFy48VE1y93n+DBWsa48+g1CJMMIXXSdS2y7ItaGoEiEE5XNDsFjfZ4Z9Hok6FqgjpiGOZ/NwfK1j6PCfy5mFm6CgmoyHyf9jwEhkmFIZItcpm0PZ837hNI7Uff3MP/gJcS2M7wzZauO7EaARIf4G9a5XXvQNxIBDGl2XHUNd7S8ve1d+cfXCJKMIHzJDZzHO2PbHp6ExiJcMoRhjXrMtR3FhLBWdASmOGXblbEe9h+dlshb7pkA0FWbDwCITF085tqO8J60zQZxUgjPzvR9Q31ZOXf+isClK9S2NddUmYdIKIHJZ9qlXXNXHZ2fimELd3e2N195d2jXHtgKf4keuvy/OG07MCiEbi+17TuYbLr22ubjHfHZlCIgyuKdEU/1LQDo3akDAGj6bSc23tCve+RtSCQd0E5awsxHzEHycM2odyJNuiPaYSToKWgipzmlzceZfqU2WLy/5C19S1+2F70Ix6yVa1xi2xWxlgRHwldCoBzsc5m2M+3d0VyDYIkWPnEzUb6uHD6+vghd+w4CiRYtnz9gPpeq3v4ZABA7/QyXadvDWfO+YfSkUNl7clLoqXums9q+lXshRzR0WyY5rW1ve6v9Ike9G2ePbW+4Z0rKv0M3IjBz+eh3Lscq1r5hcQCAPkU7r66rta3BpU0MBiSMNEMVPoX1mLGKtXVaIm+4Z5ow3FYKA5EgccZi3vKejLU3rhR6/F2/sfp46zuFpwOyd7xISIaUtNVX2F2mPWMqyXl9rRtr5XkY9HrSvGEWKXv+TMG2WjdMJ7mvXeeCWrkPJ16+lLRnTGF8f8kTaK0rp3dx+2aLp6vCiIJf6V14y479xMhn73qVtV81VuTTOxHuedtt9RtIn0iy3v6b2+w7A82QigymTyTH/3uLp6syCse/ecPha6A7UPgb887Oxz7+NyEZUpL340eEEEKy3rmXDKXHkBEt9w527gbTu6vjCcqBXjKUHkOy37xzTPRyX7uOtG6YPiZaroJaOUBU6RNI9pbbPVaHE/s/HbN33YWgo6WWc9fgscSx9x8huvQImx2wvQEnNl9GmjfM8nQ1vBo4Xd8ptCdPYd6SPM48hXlL8lhz+hWcV8CZp/D4jOMA2PMLFV5cyJozyVTWZMs6x0vBeQXc+YXuZc8vlDM/hzNPYfkt5Zz5harvr2b1qfiK4lE++TUdQ4chBvGTZ6B8XTkq1lew+kRtpVBydQlUCIXvQK9DPtU/VY+K2yo48wvlLsrlzC+UuyCXNb9Q4QWFnHkKTfFiy8VTcEHBqDgV//w9kggF5ZxbcHzmcVafNC0anDj7BGd+oQGdFH49XYw+5czJ4cyZVHZjGWd+oer7q1l9Kl5TzJmnsGJ9BSpuq4ByoBdzVLlonHA+Gp5pNJ97pdeVsvpU+bdKzpxJFesrOPML5S3L48wvVL+WAgAoT7TY5BcquriIM0+hKdZs+YUKzitgzS+UMzeH1Sdtlxb5K/OhLFFiIPsP6IgPIrvnjOpPFesrUJtWi+hpywEAHVXZJ8/BmcehU+rQ8MWfoJ1MGnWNKL2ulDNPYfX91aw+FV1cNDoPlCQC6JCb+5Nlv7b2qXlzM8puLGPNmWTyialdqu+vRuXfKjnzFOavyIemRYOsZz9GmGQIg7Xn2OWTfKccxWuKOfMUWl+Hra97J84+wZqnMHcBnTu0/zd6Ba7mpaJR14j85fmceQorbqtgzemXPS2bM09hydUlNj51/UzXR/l2CKrvP5lrMLrgVtT4TkPK8QxQx+oR3FKCJp9UDBxUMt6fmGJteY0ov6XcKZ8q1leg6t4qs09D2X4AgJ7jTeZzL//MfNZ7rqVPTPen4iuKOfMU8t1z88/Md2gcUfnHtwiSjCBs0XWc4wi+e2721GxWnyzvuVoSiQjDgM11z7JdHB1HlN9e7vQ9155xRNaz2xAiGYa273yba/mJs0+w3nO5fDKNI7jyFFrec4ON70irFO2of6oepWtLOccR+cvznR5H8N1zTbFm8qnpV/rega7J7LmB77fNDWzvOKLsr2V233O1ilD4SghafqgVPI5Q7FWg4LwCp3xiGkfEKmshD5hq1z237OYyznFE/sp8znFEztwczjyFXPdcvnGE9T3X+rpn2QfYxhFinkInVgpHBrl3qOPihZQlhDtHilDbXDxfbhZXtYlBryeKjKRRq1j2aJe+cA7jznfubG++8q7Wzn3tetKfMYmolQOCbRe8eAmpfXaRXbqO2hbSJibt3B+2EpIhJeXZP7vMttBYa/uHyXB6FDn2vwfHVNte28WbziM1zy5m5YfUSqJNjxqVK8qke+yjxwnJkBLVYJ9T2vZwlc+dQYo3rbbRdoVtR3mTds7rN5L+jHgyrBlymW1XxNqUp67kyB6XaTvT3jmZNxF5xmRGvq4km2jTo0jua9eR/vRJnCs3YxXr9qZqejX/q0y7tL0h1pbIe/VqoshIIrqRkTGJtanfa4ZUdtv29D0z97XrSE9GAtEOaxzSdmWsW2pKjDke3+LVdbW2Nbi0TflhFR0tTtl2ZaxNu6Y3VxfaZXusxsLKgV6iT48gxz58zK7ynoy1J23jdF0ptAetr7U6zQspywehtj2lbck11xQjBv0wJJ/lkPawvxTBetvd+tzpM195V2r393RhXv8hVMReiuDQcMG21epwhOv77NZ3xLaQNjHBp3wvuhCFmcsucpltobFuy6TQJ4mAr8b2XUxP9y29TodUTSW6I+ezlg8KDkWzXzJCu203m/HvqQEliUNIWITD2vZyQ/6RCLHYaIAP7o61dliDmf1/oCribAQEBrnMtitiHSyNpuuo7GHk3altiQhlHToDJzPyU+atQF7ynVg28CukEjUk8dy7ZTqqbQ9nzUuj6Xe9DCrm96WF2HaGd6SsRq3E7IFjqDXugjsW2j5hdP5Q07tx9tj25D2z4dVazOr/EzVR58I/INBluo7yERNkAAC90r4NjTx1nkkUVegjYYg21tdR266MdYCUPtdUxvd9PX3PNP+76gR8JARBiQsYeXdqO8p70jYb/JwqdYoh7tY4p3khZQFgzpfsOwQJtc3Fc+kKtW3JdRT/hskA4hde6JC2/CcpwlS2k0J3tjdfeTauoew4uvdvhjZhKVLwtF3alQc+wArJCGLO+Ruvrj08YmIR2UMnYrfcqdJa1xnbQs7/OV/OgVrZj1nK4yiZcDkm+I5+qdtTfctUvuNLKfyHbTdm8HS/bqkpRIpkCD5JyznLd4fPxrS+o+a4m3Sj1PXoCpwM66GDK2OtDYxG+FC1+f/u6Fv28nO+nIPKrB+xACr4z7/WpbZdEeuQiFgAwIjVpHAsr+HEYIBspAVlkZex2l667gU0vnwAKYZmRE2zPfec1baXs+aDQ8IxTPwBi020PHXPdFS74ugeLJYMI2Th9YK17W1v00Bd2dOJiYlT7bLtiXumCf1zcjG9fAiBC65l5Mcq1mHhkRgm/iAqBa+uq7UtUXzwK4Rf9TJqnieM/GxdG1r9UhHpw76eM1axDo6k/23aqM3T90wT+pvoR/Tjpi22q7ynYu1p22zw+EqhRIJLJRJUSSSolUjwJMdx10skIBIJlll895SxXJVEAu7tGznQd7DPaV5IWQDo2We7fbSrbHPxXLpCbVtyvi3HoEAkEqeeXPGwR9sQSG/h7sp62attL9dUeQL5r16N1N2XYNnAr1je8irqSrJ5tYnBgAnVO1DjOw3TFp5lV735eK0yAgESnc3uhpa6ztoWcv737OtB5dFv6RQBi69zqW1XxFrtHzkq19JYaNtju6uCfidw4pyzOcuTSQsQjQHIqQazrl6nQ6KuFUMR0zjLOlo3a04fHI1IMmDesdKVfctRvmdfDzRF30BJgjHrrKtcatsVsZZGGVe8hvoYeWe0HW1vOdWAMMkQEDeL1XZAYBDI9R/gqM9VSJ27wmXa9nLWvMTHB/2ScPhoTv5w46l7pqPa+tI96EMYZq28VLC2ve0dGEG/G6e2SrMj9Bpuj7YztkcqvsMAQjD7zCsd1nZlrE05Hv2Gunl1Xa1ticA/NyNW3wm1fxTjpyF4HhQBtzit7cpYhxmvaSODXXbZHquxMOkohZoEIn7yLLvKeyrWnrbNBo9OCiUS+AJ4G8AaAHMA3CKRwGbaLpEgHMA/ABy3+G4OgJsBzAVwKYB3jPYcRsisEKd5IWUBoOMT28TZrrLNxXPpCrVt4ojBgOSBE2gKWzRq9coebRIciUDJCDRqpcvqZa82H9dSW4K8129A4pcXYNZgNrIS7kTruj8xgDCQPQ+MSrTMpF1bdBRTDI3omXmz3fXm4/2Mj1kNdNvm3BuLWLOh45MOkLLv0AMpZp1h+7uNp/qWqfywfyRC9f2MnLu07bFNWnPRj9BRP6YwlY+YSq/mtFccN+t2NFchUDIC34mzndK2l5OExo76IcIVfctZvu2zVkzvPYJK6ZkICg51qW1XxDooOBRa4gdYTQrH8hour6N/QQ9LnMdpO3XuCsxb/SZn3tSxjLXSRwp/bZ9d2t4QawAY1qgxs/8oqiNPPhY5FrEOjaLTXgz3yxl5R3Ud0XaUG9EOY47hmM3j3vZquzrWdI7Hbl5dd2gD9BNHM3VVyKVuwMInfmH9LFh1m9Parox1RAx9rumVCrtsj9VYOKy/Gi3+qfCxeCrJU2NhPt6Tttng6ZXCMwDUEoJ6QqAFsAPA1QzHPQfgZQAai++uBrCDEAwTggYAtUZ7DkPXp3OaF1KWD0Jte0rbxFGNFYhDD3RJZzqs7RMcBQAY7FO4rF72arOhvbkSOf+9BfGfnou5/YeRI/srtA8WYtXf/4vEafNQMeFJTNPXIfeLjZwaPX+8///sXXl4G9X1PSPvm7zEcSLFduLE2fcQSCh7WVNKCy17gUJ/hTYU2piW0kBrZSlbWNKyk1IghQIBWiDsW0kCxI73xPsSx6vsyLIsy5Ks/f3+kCVsaWak2Txyovt9+iC6uue8+868eW88M+/CShKw+KJfhN3uUP4Y4n1nyWzoY/0dH2whx78z1o5FpoNoyzqHdpEp19jyxbsSs5DmMU0qdzjY04216ExcPGFio4vPX3wq3ITCaFeV/7uB9iMAgLT8Zby4w/XFjL2/ZBoM75iTUuuh/CpkwoSYZcHTRyScRymFAiYqFTEBjypP5nFm6fW+e6oqXBkSW8rzKNecR2PTae/mi4HN1R9ubNPBvUijJj4WORncvoW6a2SA1i82rxDsppIPkU5ZELeM/tFRIbx8/Na4LCTT1HicDG4AOL7/BThILFKaN0jGLabWSclpsI17tFvuORMYq+PoaMewcv6kc/Pxy4nNZHJfFM4C0D3u3z1j3/mNorAGQB4h+JBrbLjmGfXw9guJBYD8u/Mlw2bzs/EKxfb5eg9/CQCYseL8Cf5wuGNTxy5yjANBPr7tCpebzio/egmFrT/AyqHPUTHjKlg2VuH0Xz2NzOkq/28WTL8SVSlnYU378+hqqaHltowYsWzwM9Rnfh/KjGlhtzuUP47yYo0G/IXYxysEW9Dxf30DUigbklYGPzoqGFsErT1JWUgnZrhdriCfVNyhYs2mIeS7O2GZvprWPz4+OTUd3TG5SBqs8/OOar1bU6sLV3Hm5qJHvNJ7d3rE0O/nFgubqz9t1UFYSQIWnRl8nEl5HgXC19qiSEOsY5jRz5Wba38r9M0YghJZObNCYkt1Hg0HO9Bvj89Asvu7P9zINWdy4bYf8T4Wueh7PxSFO9z+VmZOh4so/O/GhYMth9bH6g8h7usdsJBELDqT7j5AaG6xtbYnZCFtbOMsKedMOr9t1IJFuo9QqzwLC3/DvsGTlPM1l7y9j3Yr/Y92R8JaeKCvExkwg+QsDTt+srWOFGwmi+iNZigKCgCPA7iZL8Y+at9tAG6zxFtwIOUA4rLjULCtAMmLk9Fyu3eThOT5ycj+aTaq1nv/2h6TEoPV+1ej8eZGWOos8Fg8SF2TCsPHBv+t5vy78xGvikfHlg50PdKFrAuzkH9fPmrOrQEAxGXHYeUnK9H3Qh+6HvHWBFn2zjLoXtdB96Z34T6neA5sPTZUrK0AAGRflg317Woc2XDE346ca3JQe3kt7D12AMCKj1dA+4wW+vf18Fg8iM2MhcfmQce2DgBAztU5yLkuB3VX1MFj8WDgnQEsfWMpDl9yGE69EwCwat8q6N/X+9tV+HghHH0O/79n3jQT8ep4f7tSlqVg8cuLUX1ONdwWNwBg/hPz0VbUBuPXRgDAgmcWwNpoRc+TPfBYPHANu0CGvsaQJxW6G91wnNqCBc8uQOW6SriGXeh+vBunHDoFLRtbYCr3TviLXloEU4kJPTt7MLjMCiwFhmp7MXgTBQDIOCsD06+a7m9XoE4AkP+nfPQ80ROkU9td3hozyYuSkXVpVpBO9dfWY7RtFB6LB8rvKYN0Sq5+Hlr3DHR8+xhmn7sEypgZ/nYk5CZg+bvL0ftML0ao38B2ThWsb/4aR12vYugDb//MfWAuRttHUXHnU/jBHBucI5fD1m1D3RXexXxcZhxyrskJ0qnr/i4YPjfAY/EgXhUfpFPWhiw0XNeA4UQ3cAHgMOkm6LSmdA0G/jPgjxmvEwCob1MjJjXGn4vyVKVfJ+ImoGIozH9yPq1O2l1aeCweeGyeCeMp46wMzH1kLqrWV4Gc9QGGlSlYdPqlE3Ra8voSGD42QPusFl2PdAXplHVhFjIvzvS3K1AnAJijmYOuHV0TdFIkKtB+bzsAIHVFKjK+n+EfTz6dfOPJY/GA3JMJBUXw9fe/QpI1E3MfmOsdT2Pjevx4AoCkwiQsfWMpev7W4+/T8Tr5xpOl0eJv+3idACB+ZjxyrskJ0qn97nYYvzZCpz6EOWsIPKZlfgz1bWooT1ei6ZYmeCweGPcZ/TrZzpqLQlctAOD4a8fhLjwCXUomkrsToH1TC+0ubz3G3DtzQTzEjzleJ994mvfIPFqd+v/VD4/FO8n4dDKorcAawDrQj4q1FXCb3eh7qS9IJ995r/+lfnQ90hWkU/Zl2UhbnxY0nsaf9wruL0DH1g7o39f7x5PvvOeGCwsu/RqNSetAzmmeoNPhSw7D3mlH1yNdtDo5+hx+rQN18p33uh7u8ms9XiffeDIfNvv943UCgMTZici5JgeV6yqReH4yFKPeBZRvPHksHiQvSfaPJ59OvvHksXhgKjUF6bR6/2r0vfzd3DJeJ8B73nManf4+zbowC2kp7eixz8LRtRX+eZBOJ92bOngsHiiSFEE6+eYnt9mN468fD9LJNz/p3tCh65GuCToB3vkpeXGyv13jdfKd9+Y9Og/tm9v9OjmvUiLNbfLHpH8vHalrUoN0qj6nGk6dE12PdNHqZG20oushr5aBOvnOex3bvccCFUPRz0/lJn+fj9cJ+G4dUf69Q1i0YT9qrafhjMTksNYRbXe1wWPxwFxjpl1HaHdpWdcRjuMOf/8UXJoGhVk/YTzNvnc273WE2+xmXEd03d8F/bvetUQ464jkP42i78X7sD75IEZIEqpn3oWZ9/bRriMAIOPcDCQvSQ7SqXJdJdwmd8h1RNcjXUE6+c57vnHvG0/m/hSkJw6jfG0Z8u+azbqOSFmawmsd4RtPHosH1iarfzwNLP4EG+aZkXDaLei597u5ZbxOgPe8Z+v+bs1INz/l3ZXHex3hNrsx+PFg0HrPN54MnxjQ9UiXX6cRZxqoPh20u7SC1hEAkHVRFhILEmnXER6LB73P9Aat9wLXEc6f1yMHwNCbmah4tMK/LvdpTbeOyP9TPus6Im1NWsh1RMb3M4J08o0nj8UDW6eN1zrC2mZlXEekLEvBrNtnMa4jPBYPEgsSg9Z74897tEZXp2KyPgA5HSCfjvv3ZoBsHvfvdIDoAdIx9rEBRAuQtTS//RQgpzNxsdUp7P47c82XUH4hsYSw10gRis3mD1WbRYw+6d1SSKp2/IAXd0vVfkI0SlL92b9Fa1e43IF2vKedEI2SfLV1U1ixZe88RYhGSUpe+2sQd+P2daRj61Licbs5tTuUv+XxBi/n7j8H+SZDazqz20bJcPFMUrbzGtGxw/GHk3f5+88TolGSjsbKSeMOFfvV1rsI0SiJUd8fVnzJq1sJ0SjJQF8XKT+lnLRsP4UceeAcXtxc9Og91uStIfffvxNChB1nQtpVX/IxIRolqfjgH6Jjh+MPV+uahy4Mqjs5Wedwj9tNjBoVKX3ixrCwpTiPhosd6D+463fEWZxB3C5XSO5I0PrI/v8SolGSqk9fEY2bS38f27osaN6VW+vu1lpS/thPias4nZiLc8jBXb8jRn1/RK2PSv71F0I0SmI2DUk6Z9L5ax84m/RumU/cLpds8zUh3LU+8sA5pHH7OsHtCsUdLnbJ7j/Tzp1yrYVD+eXERoTWKSwHMJ+iUEBRiId345i9PichGCYE2YRgDiGYA6AUwI8IQcXY766lKCRQFAoAzAdQxqcR2Vdk8/YLiQ1lQrHl4s6+Ihv9Xa1QEx3ss07nxZ2c7n0c0mEeDPLxbVe43IF27Os3AAC557Pv+uWLXfujjTiSeCpWNP0d2mPf/UXGOL0Ni1yN6Jt3VVDZCKF55f5k9oTn+7mYVMd/48H3oaSsiF9xuejY4fhDWfYV2Ujw1Voy6oJ8UnGHik2Pq0c3pUb6tBlhxacVeDdk7m0shQduzHJ2waIM3nk0HG4uemRke99f8gS8vyQGNhe/qeo/sJM4LDzrp6Jjh+MPN9YZp0SyZ4TRLyX3oK4X6bCAZC8MC1uK82i42IF+KmUaYikPRobZdwnkg83VH06stea/sJBELD7z8knnBgBLbAYSHUOMfjF5Q8W71htQ9rfrMPOVs4Lew0+fNiOi1kf+Go8Dod+RFpO7t70ey+w16Jz9E9p3yMXkFltrR3ymf6O2SFiPxuobcRzTgubOSDrOIgWbyWS9KCQELgB3APgUQCOANwlBPUVhG0XhRyFi6wG8CaABwCcAfkMI3HzaoX1Gy9svJBbwPtYjFTabn41XKLb2GS16ar4AAGQvPY8Xd+pYDRyPdSjIx7dd4XIHWmr7h+hQ5EPxAfsg88VSCgVyrn8WBBQG3/i1f8v+uFM+gYPEYMGFv+Tc7lD+vuf6vYXYR4MvCqXWmsnste9hhCRh0feYh7JcY8sXn5Q+sdbSZHCzxRKPB3m2BvQrg3cdZYrPX7oeAGDtrETMBTYkU3ZQOYvoQkXVOjk1HaMk3v+HCCHHGd92jVpGME/3OWo9a5CqzBQVO1x/uFq7EjKQSsyMfq7cXPq7/2gNACBl1jJaPxdertxcsQP9MSnePxCOGI6H5JZK67bD36Jj2zKMXHSI1u+LdbtcmG/Yhybl6UhMThWtbVz62x6fgRS3MWxsqbQu/+/fMevDc1nfw4+k9ZGvnMeIQSvpnBno7/riebgJhcKLfg1AvvmaD7crIRPpZFhwu0Jxh4udZW5Ff+JcTvGTqXUkYTMa3e3DE/HD9vio2+lm9IXyC4klhBDniFMybDY/G69QbLfTTQ797XoyrJlJXM5gnnC43S4XcRenk5J/FInWrnC5x9tAXxdxF6eTgy/8njN36Z6HvY/Xvb2T2EYtxKCZRSofuYxXu8Pxt25bTWoeuiDIJ7XWTNa+dQWpeTC4PWJgh+MPJ+++rlavRm89NmncbLG+RzJL9zzMKb5ryyJSueNSUvXpHkI0Su8jlRy5Q/npfH2aeaTs8asJIcKOM77tKnnmNkI0SlL71fuiY4frD1frgy/8nhCNcsI5cbLO4aVvPEiIRkl0vcfCwhb7PMoFO9Bf86X3mG4q/zIktxRa222j5OjWFYRolMRenOltD0Ns3TcfEKJRksqPXhS1bZy0fuImMqRRh40thdZl7zxJ3MXp5MgD50445sTkFlvr8a+sSDlnjvc7HXYyoMkn1Q9f7PfJNV/z4T74zz8QolESh90m+1rYYbcRe3EmOfjcHZziJ0vrSMNGhD4+GhHW/Mtm3n4hsQD8LypLgc3mZ+MVit38y2aojFVoT1qBmNjgvYzC4VbExMBMJYOyDQX5+LYrXO7xdvTrPVBQBDPXX8WZ+9Sf/h718cux+MhDqP7PI8jECOJOu4VXu8PxW+MykESzdbvUWtOZzWpGnrsLuu45omOH6w8n7/Qs71+H3QG79ck1rrX1BwAA0xaewSlel7oQM63NOP52CQBANY9+BzuxtTbHKBHv8D7WJ+Q449OupoovcVr/Hhyadjlsf5gpKjYXf7haU0kZACaW2Zm0c/hAM0ZIErJn5tP7OfBy5uaIHehPVHqf0LCZdCG5pdC68jUN5no6UL7qfnS48rFo/+2o3f9f2lhz9dsYJfFYdFbwLriTpbUnORtKYplQL3cyta54/3mcUv1n1CeugqflMUxXz5GEW2yt06Z572A6TMclnTPH+2u/ehPZMAKrb/L75Jiv+XIrUrxjc9igk30t3NN2BPGUG3GqpbR+PrzhcvP1y4nNaHRXioEflWrTDJVq09xwfhupH7Y7hXJaqJdcpyLvQG+Hd9OTV4oFcfdsmU/KH/uJmE3jnPeRB84hXVsWBW0ME651t9YSa3E2IRol6f3zPP9mCVJY+WM/Ib1bCoO/l+EYa678ihCNknx4HfMdL6kt3LwtxdNJyTO/koU70EqevpVYi7OJ02HnFjf2gn3p3ReSQU0uL24+dvjB75Pm7WsJIZN7nNlGLaRj61LSp5lLTMZB2c6jhISfd9m7TxOiUZLu1tpJ5SWEkLr7zyJN208ThZcrt1DrOVpHiEbp3cBrkrmPNZQTe3EmqXj0x4QQQg6c9Tlp27aSjBZPI7Vfvzfht26Xi+g0s0nljktFbweXnEtff4AQjZLo+9k3mpCCu+LDF4irOJ3U3X8msZpNk8ot1EYtI4RolOTgy5snjbf6oYuITjN7wvl+KpzLfFbx4QuEaJTkWIPwNgvNu3zvc4RolKS9rnRSeaeqQeCdwtsBtKrVRZfROdXqolVqddEVanVRcLXkKWCNNzby9guJBbxbGUuFzeZn4xWKXf3QHgBA1pLg9wm5cFsVaYgLqOslZX8Hxhv1/VhsO4we1YWgFApe3LmFy3B4/m8AAPWmixhfJBdDa1fiNKTTFGKXUmsm31BbOQAgzcZ+SpBrbI2PN1FKxNgMtD4puJliiceDHP0hHHUXIjYunlN8yhzvZjOrkyrRFz+bM3c4fjqfIz4TqWPvLwk5zri2q+rV+zDb0w3dOQ8jLT1LtvMoEL7WcWO1V62mQVo/V24u/T3D0Ynh1LmMfi68XLm5Ygf6UzN9d/MHQ3KLqbXb5YL9P7fDQqWg4ManAQDK5OnI/NWH6I9RYe7nv0RDycf+3++/49+YjiF4FtO/Qz1ZWsemeTdM8dUPDYUtltbVn+7GykN/QEv8Esy5830kpaRNqfVRYnIqLCQRlEUv6Zzp8x/vOYrl1kNom/XjCed7OeZrvtz+jdqGdLJr7eyrg4PEYFZh8FMycq2FQ/nlxGaycC8KzwIwCODjQIdaXfQvAJUA3gZwWK0u+kytLmJ/c3MSbR+177J91L5dti4bjAeM0O/VQ7dHB1OZCT1P9MDWbUNCbgI8Lg8ab/Z2oq8zG29uhMflQUJuAmzdNvQ80QNTmQm6PTro9+phPGBE+lnpsLZa0bG1Ay6zCy0bvTVWGm7w1hLxXpADbUVtcAw40LWjC+ZaM/p398PwmQF5d+Whf3c/zLVmdO3ogmPAgbaitgmxPqyWjS1wmV3o2NoBa6sV6WelM+bUvrkdC19ayJhT2uo0xpy0u7SYXTybMaeGGxqw+JXFjDmlqGphIYnIoVYG5dRwQwNWfrKSMSftLi1Ut6qg36vHqCcVCY7wc2rf3I6CBwoYc7K2WpF+ejpjToD3BOHLqeG9VxFLeZCU/AMYPjMg84JMWp0CtQ7MaZb5F/h25hYsOusPtDp5XB5g7FqRLidbtw1pp6Yx5tSxtQPzn50PS2cSUigbDt9YOaFdKYtTaHUyfGZA/+5+5P0xjzEnn9Z0OhkPGDH9qum0OVHaGgyRNJz59iWMOfU80YNZd85izYlJp7aiNsx7fB5jTuZaMzK/n8mYEwDEpMbAZXbB5FEizmrw56Tfq8e0H06j1cl37PmMLidTmQmqW1S0ObnMLsSkxtC2Z989/8RcTwcG4jYw5tS1owvzHp8XlJP9Re9mHPGUCyMp84J08p0jZt05izGnxhsbsfiVxYw5TfvhtKCcrENpyPAMo2Vjy4RxPV4n37GX98c8TjmNP0fMf3a+P6eKnR/i1O6XURp7AWbqzva27QfTeOWk26PD9Kum0+rkO/YCz8OB570Z182gzckx4PAvOBpuaEDi2EWhzaT3j6ekwiRanXzHXsEDBbQ5Ad56l0w56ffqobpVBe0uLfoqOpANI1wZCyaMp8WvLGbMKfOCTMac2oraaLUef46YXTybV06NNzdi4UsLJ+SE5li4iALOnuPec85PpjPOuWw59e/ux4wbZzDmFKh1+Z77sdDVjLYVf4bpWSesrVZM/8l0KJpS4FrwGgaQjdmf/BwlD70JW7cNimn/g4PEQvHfVUE5hVpHhJpzfccR05w748YZ/pysVd4LjM4X62n7hes6YvkHy0POuQceeA7LDhahlZqPlIWvwFkNwesIw2cGzLhuBuOcy5bT+HUElznXqEgH+vqx8B8LWdcRSYVJYa8jmObchheeRAxFENN0Ka3WXOdc33ha/Mpi3uuIpW8v5TTnejqTAQCGg8cErSP0e/VQ3aLilZNvHZHY14AeKhfWQ7aw59y2ojYsfmUx6zoiZXlKWOsIPnOuz7jMuePH0/gxEHjeS1mewrqOYDS624eBH5VqU4dKtek1mu8vU6k2ecY+bSrVJsPY/+8LB3cyP2yPjzb8vIH1NiubX0gsIYTUXVMnGTabn41XKPbRPy8lhx/8vmDuikd+RLq2LBKtXVy4CSGk5sHzSe+WQv+jo5Gu9aG3HiNEo/RunhImr1BuJl/rttXkyAPnCuKeLK3HPwI5Gdx0sR63mzT8dT3p08wlR35ew4u7d0uh9xHS1x/gHBuOn8538OXNhGiUZNQyMilaO+w20rptNRnQ5E+oRSXXeZQLd0dDBSEaJSn/YJco3OH2d0PpJ4RolKTmf2+GjS3meZQrNp1fr8nz11icDK27W2vJaPE0Uv3wxRNeHRjPPdDbQbq2LCIjxTNIY/kXpO8vBRM2DBGzbVz6u72u1Fu388MXaf1c2xWK+/M7niL24kzSvH0tGR7Sc8KWe84MtKbtp5EjD5wj6ZxJCCF1P68lvVsKSe0DZwf7ZJiv+XLreo9555w3H5Fd635NAePrRnKthUP55cQGw+Oj4V4UWlWqTffTfP+WSrXJrVJtumPs3/Eq1aZXx767MhzsyfpE6u6jbM8zS7kzUajnqPliG3Ra7/uEL20WzF36xI1B70dJ2d/j442GAWIvziQlz26cFG4xtK769BVCNErSWvN12LxCuZl26/PtAiaEe7K0Ln/sJ6Rny/xJ46aLrf1mr3dyfeNB3tyVOy717sR54F3OseH46XyH3n7c/4eIydDadxFa9cnLE/xynUe5cA/0dQbtLDsZ53DfH4u0HU1hY4s1tvhg0/k7ti4llY/8MCS3GFq7XS5Sd/8ZZFgzkxzvaZ/gD+Q+3tNOurcsIKPF08bee3xSkrZx6W//cfbGQ2FhC9H6yL7/EFvxNNK6bTUxGgY4Y8s9ZwZa1cOXkKNbV0g6ZxJCyOGv/hP0ByKfTfZ8LYTbNmrxvof50j2yaj2k6/e2Y/efOcdLrfWJuvuoG0AyzfffBzAC4DkA0Gp3OgBsBGAGcH2Y2LJbx186ePuFxIYyodiTzd1U/gW0/7gaAJCx9PuCuT0J6UgjFn+dP77t4sPdcuBNxFNuZK69clK4xdDaV3PPOnR80rjpfF1NlYin3IjPWyWIe7K0diVmQRnwLuZka03t34EBZGLlj+7kzW3PWQkPoTBz/mrOseH46XxxY+8vmce9vyQWdqC/s6kKa9p3oSr1bKy++Och+bhgC/GHG5uW4d2pz2M10Pql4vbommAlCZiRWxg2tlhjiw82nd8ao0RCwPvlYmEH+sv/8ziWOmrRtPwe5MwqYP19zqwCxN7yAQyKTDhILBacfY1kbQs31rejsmfcjspSaG02DWHe/36NbjIL02//GOmZwXV8I33ODDRnQhbSPNIfZ6Nf/hNGpGLZ97kvk6Vcr3LlTkhMhpkkgbIaZNX6yOOfAwCS81ZMOrec1w+SaE13pRj4Uak21apUmz4O+G7p2KOie2l+v0el2tQVDvZkfdjuFI52jbJeUbP5hcSeKNgtVftJzYPnE6JRkkFNLvn2n9tYd+sMF9u3o+LIsEGSdrP5qx6+hPRrCibsFhrpenS11Hj/+vjeM6Jjc/Edenunf5dFufskHP/BF//orUFmGw3yScEd6Ksv+dh7d/3f2wRhm01DpOaj92h94bQrlJ/O13joM0I0SnL4q7cl1drcPkIat68jQxo1GejrEhV7MrW2FmdP2Ol2MsbH4QfPIy3b1kiCzcfPJ7bq4UtI+9YVkmCPt46yOmIuziFHHjiHdg5jitf3d5P6L74UxC1mXkaNipQ+ebMk2D6rP/iR9zHV/74qOnY4fimwDz5/J3EUZxJLh1l0bJ8ZdFrvU0hP/1J07FB+KbB7t8wn5Y/9VNZ2f/viXwnRKIPu7IuBPdX0CDcWAu8U/g/ABWp10bxx3/0QAAGwj+b3nQCm87tMnXzTv6Pn7RcSCwC613WSYbP52XjDxT565CCqd2zA/Pcuw2xbI0rm/hYJv69FvvkWUArmQytcbkWKd2MG87i6XlL2ty/ebBrCEks5juWcP2G30EjXWjlNDQBwjQyEzSuUm85HtNUwIRmz5i4RxD0ZWgPf1VoyDR4P8knBHRjr/uphDCIdK3+8SRB3SloGMmtX8YoNx0/nS8ny1gd0mHSSal326oNY5GpE6+r7kD0zL8gv13mUK/cIlQqFfZjRz4U73P6eYeuAMWDn0VDYYo0tPth0fmdCJlLG7uZLpTXxeGD44A5QIJh27XO0cxgT97QZuVDWL+DNHcrPtb9NVDpix+2oLIXWpmMVAIDEdua8I33ODDQqdTriKDc6X2sRHdtnzZ/tQjzlxoxzb6P1T/Z8LZTbHJOOeIdRVq1JZy2GkYLpKvqdt6VeC/P1y4nNZOFeFD4JwAXgQ7W66Cq1uugKAL8b833IEONg+D7iTLleydsvJBYAdG8yH5BCsdn8bLyhYjsaK9A/8ivM++8GzLPWoGT2rxFTVIvTb9qOlLQMQTmP545NyQQAWIYHg3x82h0ud9PXbyOBckK55qecsOXWWpkxDU4SAxJQiF2I1qH8dL6M4UZ0xxeCUigEcU+G1gAQl+a9KBwZ6g/yScE9Prap4ksst1ehdd7NSEpJE8w92Vorxwo+u0YGJNO6t70R61z/wOHEU7H2sl/T/kau8yhXbosiDXFOE6OfC3c4/T0ybMAMDMKVFbxwFzL2JntcuxMzkU5GQDweybSu2PssVlNVOLJoE9QFi2h/M1Xma0tsBhIdQ2Fh89U65vgRDCATeWct5I0t95wZaLFp3lcwtPtbRccGvH94UB19C02KhShYcirtbyb7HC6UezQ2HUkuo6Rau5eYWWOnKdrREz+X8WaEXHNmKL+c2EwW1kWhVruzDcCdAOYDeAPe8hMzAXyl1e5spgmZDyD0SyYRYrZjNt5+IbGhTCi22NxdLTWoeOwnyH/jAixyVKI09/9AfleL02/x1goTg3d8fPzYFu6jpoEgH5d2c+VWNOyFHhlYuPYCTthya00pFBim0hAzOsj6WzG5A30upwOzne0YyVwqmHsytAaABKV3IWAx6oJ8UnCPj7V/+RCGkIbll9816dxc/XQ+ZXoW7R8ixMD2We/720EAzPgZ/d0bIbxi+LnEjsakId45zOgXm7vv6BEAQJI6eDwKGXvhcPPFpvNTSVlIoJywWoLrsArFBryL9dk1j6IRC3Ha1fewYojNHa6fS+xoXAaSXeEdZ3x5p480oTdp4Qm1PkpI9z754EweCvKJwd1eX+atr5p8OWu8FNzhxPLhdiRkIsU9LJnW1Z+9itnfrMORh76P5or/Bfk9bjfy3J0wK5nvWEfacRYJ2EwW9gyr1e58Ad6NZd4FcBjAqwB+Fvg7tbooHsC5ANj/1DJJFk6dQt1bOtY6hbq3dIz1hYwHjKx1Cnuf7QXAXF9o2mXTGGsm+WKZavH4+Jlq8VDxFGNOiXMSWWsmOYec/pxK7/wE5Tuvxqx/n4slpm+wz3oldEv+h+z625HkUgblZGmyMObUcEMD5hTPYa0vNNo+Cv1ePZz13npLhv2dYeXUvrkdLrOLtb5Q8oJk1vpC3W+0Y5H5EBrJmRhttE2o8TJcMsxap9CnF1MtnqyLsxjrC2l3aRlzsnXbMPTFEGt9IQKClo0tGFGkgzqum9CuBHUCa80kx3EHa30hRZKCMSfzEfOEnI6WlCORciJGtRKNNzdiTvEc1vpCtm5byJzodGoraoPH4WGvL7SMvb5Q/8v9cJldMH/mPVcYvjnmH08jFSOsdQp7n+tlzMlUZkLmeZmM9YX6X/b+vWzfr97AytEyVI78BClpGf5zxODHg6z1hTwOD2NO8dPjWWsmje93ulpxiiQFY04jFSNBOXVu78IwlQZ3R9+EcU1XK85x3MErpwx9I9oUS2B7MZaxpp+v7irXnHR7dDAfMbPWKQw8Dweey9PPTGes6Xf838cnxFotyUhyjfjHk/49PWudQpfZxVjTLzY9lrVO4Wj7KLSfHQIAuL6ZHnTeUyQpGHMaLhlmrVNIp/X4c4RzyMkrp8abG0HFU0E5OQZSAQDar45CuU7JOOey5dS/ux+WJgttTjV/KEEODOg8fg4UMTGM85NynVKSdUTgnBt43vPVWWNaRwTNubGZSHUbafuF6zpi9n2zg3KquaUSee5uDI0Uss65bDn52sVWpzD9zHTGOZctp/HrCC5zrm+ztuT1DtZ1hP49Pes64vi/j9Pm1Lf/GwBAKlnPOOf6tA53zg087ymSFLzXEXl353Gec+0OJdI9JkHrCP1ePTLPy6TNyfrV27CQROSOtmDhB1eg8r7z0Vr1tT+nIzu/QQplg91WGHLOpcsp93e5rOuIwY8Hw1pH8JlzQ60j6Obc8eNp/BgIPO+FWkcwGt2LhkI+KtWmm8c2oPmd2NhCPmwbzQztH2J9IZPNLySWEEIGPx2UDJvNz8bri9V2NJFDf7uOOIszyGjxNFLyzK+Ivr87JLaQnMfHazuaCNEovRuXhIktlPvgrue82/p/Hbxhx1TQuu7+s0jDX9eHzSuUO9BX9s5ThGiUpKOxUjC31Fr74um2cJ8Mrase3kCGNSpiMg7S+vlwT6bWPmvfuoJUPXyJJFq7XS5iKZ5Ovnn4F6zYcp1HuXKX7byG9GnmisIdTn+XPLuR2IqnEafDzglbrLHFB5vOP77cjhRaH9k3ViLgX2+xxkfyfD3efBum+DbLEVvrxrLPvaVhPn3lhFof+eaCA3/bLjo2IYSUPHMbsRZnE/1XweU7fCbHOVwId8lL3jJB/V9oebeLidvtcpEBTT4p++uPiNk0RA6+fC8xalTeY+/hDeRoban/3NBUzrzRk1xzZii/nNgQstGMWl00K5zfjVkjgC3w3lGcEmZtsvL2C4kFgPZ72yXDZvOz8eq1najb9xtMe/F0rDJ8isqcn8D8q0qs3/gcps3IDYktJOfx8akZ3r2KPKNDQT6puBXaDzEEJRatu4QzdiRobYvPRKrLGDavUO5An1tbAytJQG7hCsHcUmvti/9uC/fBIJ8U3NYmK9rrDmG19VvU5/9swqPXQrknU2ufWeIykOQ0SqJ1f3cbkik7nG720gBynUe5crsT0qEkI6Jwh9PfScNt6I2Zhdi4eE7YYo0tPth0/sR071wwOjwgidaW7sMAAKWJfbOYSJyv6WKplGzEUW6YjIMhsfloPdzu3WRGtXj9CbU+8s0Fo809omMDQKqxCT1xs2FvYd5yQ45zuBBuKmUaAEDfwHL3iSf3sfpDyIYR1pgzkJKWgdN/fj8URXUoyf8VCi1VmPv2RZhRshUAkLtwDS9uKfs7lF9ObCYL9/HRLrW6KKwH7bXanYe02p3btNqdnbxaJINlnJfB2y8kNpQJxebDrT3WBNeu7+M098eozr4MQ7eWYd1v/ols9cRdnYT0SSjzxaemZcBFFCDWoSCfFNy2UQuWeQ6hNfNs2kXUVNDalZgFJQldZ0ks7kBf+lA9uuLnISY2VjC3lP09Pj4uPgEmJEMx7l1MqbU2fnI/zCQJSy7/I2dsuY4zJp89PhMpAX+IEAt7oN27WM9ctjIkPh9eMfxcYklihvci12GfFO7ptg4MpQTvPBoKW6yxxQebzp+c4V2s200DQT6h2AAQM9AIHbKQe1EhrT8ci6TjLHasfqhpUBsSmw8v1XcYQ1Bixqy5J9T6KC4+AUakIj6e/Z1Cvtwq+zEMpc6PuHO4EO54pXejNmoh+0UIH25dtXcfy3nn/dj/XVp6Fk7/xQ54fleLklm3IN1jQjs1GylpzPiRdpxFAjaThXtRSAFIGP+FWl10vVpd9Cgv1ggz3Wvsuw+x+YXEAkD2ZcEFX8XCZvPT8fZ1NoPa/UMkwoYqvIh1d+7GjNx5NNHC+oQt5/HxlEIRtIW7lP3d9M17SKFsSFz5E17YkaC1Jzkb6bD4F52heIVyj/d53G7MdrRhOGOJKNxS9ndgvIlSIsY2ROsTm7vu1X1YNXIAtbnXIj0ruHKP2ONaLGwmnyvB+4cIKbQe1dYDAGIqVazYcp1HuXIrkr07KpuGBgRzh+rv7n8fw0yPDg6anUdDYYs5trhi0/nTMr3jxG0elETrLHMr+hPnRcQ5nCsvXWy8b/Msw/GQ2HxynjbShO7E+d4dpk+w9dGwIgPJyhFanxBsfX83pmEYnpwlEXcOF8KdkO79g43+mw7e7WLiVvbux9GYArg+SQjypWdNx+m3/g2uTfWwkH/y5payv0P55cRmNLpnSgM/Y+8IFgd8p1GpNrkZfn+7SrXpzXCwJ+vD9k6hc8TJ+uwtm19ILCGE2HXB73qIhc3mD+Tt724jPVvmk2GNirTWfC1pn7DlHBjftWURqXj0x2FjC+Eue/xKYtSoiMNu44UdCVqXvvEQIRql992IMHiFco/3dTZXe98B/e8TonBL2d+B8U3bTyNHHjh3UrjLH7mCWIqnE4OO/h0MMce1mNhMvoP/vJsQjZKYe0yiYx/aeS0Z0ORHxNgSg7t8r/ed5c7masHcobRuPnSAEI2SVH70ImdsMccWV2w6v9NhJ0SjJAf/+QfRtXY67MRenEVKnt0YsccZ1/5urfna/85fKGyuOdtGLcRRnEkOPn+HYOxI7O/6+88g9du/Jzr2kf3v+vcriLRzuBDujsZKQjRKUvbf53i3i457ZNjgPc6euyOixpaYfjmxIbB4PVebDuCnIX8VIdZ+N/szxWx+IbEAcGTDEcmw2fzjeQe0HXD+81IoPSb0/+h1FK48U9I+Ycs5MH40JhXxjmFan5jcDrsNC4e/Qb1jHeLig/8qJSW3GNg+v6/O0shgX1i8QrnH+3QtZQCA7PmnicItZX8Hxo/GZSDJJf1x1t16GKtH/ocjqiuROZ3+7pdY41psbCafItX7l9aKn30jOna6uR3H42dHxNgSgzturMyOdVgvmDuU1l1vfgsAmDZnBWdsMccWV2w6f2xcPExIgcI6KLrWvW21iKdciFUvj9jjjGt/p2aNlVYw6UJic825q6kScZQbCXlrBGNHYn/b4jORZGN/fJQPtqW7BgCgXnBKxJ3DhXCnjb2HOVzC/sYYV+7W0g8RR7mhXHZJRI0tMf1yYjOZVBeFEWPhlKSIzYhl3Uo6NiOWcSvp1NWprCUpXCMuAMxbSbtMLsbyDb5Ypm2XU1ensm4lXfhkIWNO9l47bN021D9WjtFdP0CmZwjNec8hx7wS2l1a5N6Vy7qV9IJnFzDmlH1FNmtJCractLu0yLkux5/TKJRIsJnCyql9czvy781n3UraoXXQ5vTNpj1Qwgqb+zzG7bGV65WsOfn0Ytp22WV0MW4l7R51M+Zk67YheXEy61bScx+Zi5aNLUgc21LbMnTc3y57t521JIX6djWrTgueXcBcZmPDd2U2rJVlsJM4uF5NZx1P47ddnnnLzJA50R03bUVtmLNtDutW0o5+B+uxB8Cfk51KR6rru63ZM87NYC1J4Ta7GXMylZngNDhpc+r96HE4EYvCyzczbiWdWJDIupX0nG1zGHOyddpYt8eeectM1pIUC55dwJhTxrkZtDnFJHo3GnAkD7GWpFDfruaUU/0NdVA7OzE4kou5j8xlzMlUZoJD5+CVk26PDlkbslhLUgSehwPPe85BJ2P5BkWiYkKsYY/3sW7t661wmV2InxnPWpIi/958xvINtg4ba0mKxJk9cJIYpHsKaM/lC55dwJiTcr2StSQF3bE3/hyRe1cur5wab25E4ZOFtDkNIw3UsAHOASfjnMuWU//ufmRfkR2U0/E676YpmQWrQ865zgGnJOuIUHOur+ZYuHNuCrxj0mPWB/UL13UEIWRCTkNt5QCAnMLTQs65QtYRhs8McA46GY89tpwC1xFc5lyLMQ0ZimHWdUT8zHjWkhSKREVQTlRHLfQkA46PY1jnXJ/W4cy5dOfyBc8u4L2OIG72MlB0c66rMhYeQiExd5T3OkK/Vw+nwTkhJ/OBD2EhifD8I5c1p1DrCJfZO6aZciJOwrqOSCxIDHsdwXXODbWOYJpzfcfe+DHAdR3BaHS3DwM/PB4fZfTJ9WF7fLT+Z/Wst1nZ/EJiCSHkyI+PSIbN5j/y4yNE399Njm1dSizF00lD6SeiYQvJOTC+/NErSPeWBaK0i4279M1HCNEoybe/+ITWLyW3GNg+/7GGckI0SlLxwT/C4hXKPd5Xd/9ZpHn72gl+IdxS9ndgfMmzG4m1OFtSbqvZREzFM8i3f/qxIOxQ41oqbCZf3bcfEqJRks9+8byo2Md72sdKhTwYEWNLDO6ulhpv6YP3nhXMHUrrss0Xko6tS3lhizm2uGIz+Zu2n0oOP3ie6Fof3PVb4ijOJLZRS8QeZ3z621ycQ0qevjUkNtecS5+4kQxrZvrLXZxo66ODL/yeEI2StoyLEOzWbavJ4QfPCxkvxzlcKPeQRk32/fE63u0K5Pa43aR3SyGpeviSsGIjdc4M5ZcTGwyPj0YvCk9SM+i0pH3rCmItziZ133wgd3MYrfTJm711aSS2kqf+j1iKpxO3yyU5l5Sm7+/2LqRff2BSeT1uNxnWzCSlT9w4qbxi2cHdfyZEoyRWM/t7cUKs7J0nCdEoSf3BjyTjkMPo/hAhhh3Z/1/vOzjf7BUVV04bPN5DiEZJSl67X3Kuzq2LSeWOH0rOM1lW89AFpGXbGtFxqx+6iLRvXS46rtzWu6WQlD/2U9Fxm7evJXX3nyU6bqQY3Xv5Qs3psBPbWK3nE9EC934QjDf2x7PSPQ+Lhhm1icZ0Ucjl8dFL1eqiO9TqonVqdRH9S1dT1MbfBubqFxILALWX10qGzeQfHtJD//dLoHL3ou38F7D0jEtF5RaSc2C8JzEDacQKj9stuF1s3CmmNvTG5aPp5828sSNB6/Qs7+OjHos+LF6h3D6ftqMZSlhBqVZN8AvhlrK/A+NjxmotDQ/2S8adWv8auik1yNN5grDZ/JOhdaCljb2/NPhJm6jYlh7vzqMz562KiLElBndahvf9SzJqFMzNxmu3WaF298GeyVxiYbLO4VyxmfyO+EykuodF13rG6FEMphSGFS/Xccanv80xGYh3DIXE5pKz02HHbOcxjGR+t8P0ibY+ihvbudWk76P188HuPVqHBMqJGNXykPFynMOFcltiMxBrGKT18eHurfgAAJC79rKwYiN1zgzllxOb0eiuFAM/Y3cKPSrVJvfYx65SbdKP/f8tKtWmNSrVpvhxv4/eKQzTyk8pn3TOklc0hGiU5PD/3pp0bkK45Vzy6hZCNEpiNAxIyn1cM4eUPX6VKBxcucW2IY2alD5x06TyVn70IiEaJWmpPjDhezmObz7cVZ++QohG6d21TwLujqZq7x2if/1FFPxweSfDXE4ncRenk89uv1VU3NInbiRDGrX/MTU2myrHGSGEjBTP8D/WJxVve32Z9zHV99kf6ZWCWyoreeY2Yi7OEZV7eEjv3dX05XvD+r1cxxkf3pqHLiCt21aLyn20ttR7XO1l32lSLJOjv9vrvDlWP3wxsdtGRcGs+PCFsfnlm5C/nUrnMp9VPXwJadu2UjTumocuIF1bFgnC48N7MhkE3imcBeDHAP4K4BMAQwCy4K1f+AKAcgBmtbqoTq0uehXAefwuUeUx3wukfPxCYkOZUGwmf6K2FJ1uFVacd6Uk3EJyDoz31fWyGPWC28VkJuMgcmCAa9rCE0JrkyIdcXb23dPE4vb57N3VcJIY5C86JWzecLH5+rlwJ41t0DNq1EnC3fe/5+AkMSi88FbJxnU4JsW4jomNxTCViri4YVo/X+w001H0xc0GpVBEzNgSg9tMpSLGbpSU29Dh3VEvczb9zqOhsMUcW1yxmfwkKRMplA2uGDutnw92b7N3k5nkvOVhxXPBFtvPNdYRn4VU93BIbC68+lbvDtM5C9eFFR+p5zI2f8HSdfi4+9dYZS1B3RNXTqj5yxfb0VsLF1Egd8GqsOK5YHPxSzWunfEZSHXxP/+PN9uoBQusNdBmfy/s2Kl4nMmNzWh0V4rhfFSqTXkq1aYrVKpN96tUmz4du3PoGX9XkS+2FB+2O4Wh6pSw+YXEyoHtcjrJsEZFSh9jfyk4Utrtu4PjuwMlRZ80ln/h/cvgZ/8+IbRu+OvppO7+MyXBZvIdfvD7pG3bKkmwJ8Pvq7Ho+wu4mNh22ygxaHJJ5Y5LRccWM1YIdsfWJaTi4ctEw/a43WRIoyaH/v6zkLFcsSfTT+dr27aSVD98sSTYPjv4wu+JqzidjFpGRMcW6ucbW7pnByEaJempaxYNu3TPw4RolKSvq1VQ24TGSoFd8uxGMlo8jXjcbtGwS5/6BTEX50x4934q9QkX7JJ/b/e+K/3Ij4I2neGKXfXwJeTYuE2fTrQ+O/jc7cQ2dqwJxfa9S17z5R7B7QrHH2nzw2RhQ+w6hVrtzm6tduc7Wu3O+7TanRdrtTuzAcwDcA2ARwHs44s92db/Uj9vv5BYANA+o5UMm87f0VAGJSywGFeJjh1uLFvOgfEJad53vWwmg+B2MXGPdNUBALILVpwQWtviM5HiMobFK5S7/6V+EI8HebYWDKYtCvIL4ZayvwPjlWPvxbnMetG56/73OjJhQuzam0XBZvNLrTWTWWIyEG9mf6eEC7ZhQIsMmOHJXhgyFpj886gQbluMEglOk2BuNt74oVZoMQOJyam8sMUcW1yxmfxxad73MXvebBEP+3g9TEjBjFlzw4qX6zjj1d/J05BIOWG1mETTOt3YgK74eVDExIQVP1XmTDre9df/GaWFm3CKeR9qnrwObpcrrFg6/4zRNv97q+Fw8213KL9U45pKnoaEsWNNKLel/lM4SCzmn3Zx2LGROmeG8suJzWRhXRSq1UULw/mdVrvzmFa78y2tducftdqd5/NqkcgWTp1Ca4uVtb6QtcXKWF/IbXGz1ik0HjACYK4v1P/vfsY6UL5Yplo8boubtb5Q5oWZQTm1734PAOB6p5C1ZlLykmTW+kJZG7IYc4pJj2GtbaV/X89aXyhuepw/J3I8CQAw2q9jzGl8jZfU1ams9YW0z2uDcnIfb4SdxEE1Z5G3vh1DzSSPw8Nar8unF1Mtnv5X+hnrCw1/O8yYk63b5q/hw1SLJ/2cdH9OZn0K0jzD/nb1PtPLWqcwcV4ia32hrA1ZjDkpEhVo3l2BTJhgMS8IqsWjf1/PWl8oPjc+rJzoavGkrUtjrS+k/YeWtb7QSMWIP6eYgWS4CQWXth/6vXqAAmudQuPXRsacTGUm9O3u8+ekOLQb/chGyv5VcJldGKkYYczJMeCAvc/OWl8obV0aY049T/Ww1kyKz41nrVOYtSGLMSdQYKyZZLRMxyx0ovaGI7Q5de3oQuK8xLBz6m87DABImbUULRtbkH5OOmudQu0/tbxy0u3RQZGoYK1TGHgeDjzv9b3cx1jTz1xrDtLJHqdEgt0El9kFW4eNtU5h6upUxpp+PU/2MOY0w9yE44oCxpx845opJ4/Dw1qncPw5nO5cnrwkmVdOjTc3IvPCTNqcXN3euUBf2skrp/7d/YhJj5mQU5q+GT2xc3D09+20Wgee9/pe7JNkHRFqzu15oocxJ6Y511Lv3Q/w8KaSCf3CdR0xsHcAjTc3wu1yId9xFMMZS8Kec4WsIwyfGdD3ch/jnMuWU+A6guucq/uPDj1P9GDJgt9jf+KtWGv6AhWPXIee57phbbXC1mFjrVM4fh1x/FA31EQHCwrDmnN9WrPNuWx1CrM2ZPFeRwy8O8BrznXpvJunlT2+k9c6Qr9Xj77dfTCVmTCj/xs0xy9H//2DYeUUah0Ras4d+O8A6zrC3mcPex3Bdc4NtY5gm3NbNrZMGANc1xGMRnf7MPAz9jioUaXa9KVKtelhlWrTlSrVptnhxEbKh+3x0b6X+1hvs7L5hcQSwv6Sq1BsOn/ljku9W1WHeLlWyj7hwt3f3ebdmvjNRwS3i4m75sHz/S9JnwhaH9z1O+IszvA/4iO11r5HfBvLvwjyC+GWsr/p4g2aXP8GPWJx9x5rIu7idG/tqzBjhfjlGte+jYbYyttwwfY91tff3RYylpDJP48K4T70t+uITjNbMDfjpllj9R2/eoB985TJOodzxWby+zY5+ej6h0TB9rjdZKR4Bil98uaw4+U6zvj0d/UXrxOiUZLmyq9E0bqjsZIQjdJbWifM+KkyZ4biPfjPuwnRKMmhv11P3C4XJ+zGQ595X0/54nVe3FzaHcov1bj2uN2kRnM2MRfnEG1HE2/uvq5W74Zsr2g4xUbqnBnKLyc2BD4+Wg8gBd4NZO4GsAdAu1pdpFOriz5Sq4u2qNVFZ4aJFXEWr4rn7RcSCwBzH5grGXagn3g8KLAcRq9yDSuvUG4hOQfGKzO9G4B4rEOC28XEPd3eiaHkuSHjp4rWVEo2YikPTEMDIXmFcser4mHrroabUJi9+LQgvxBuKfubLt6kUCLObhCVu/PLXQCAggt/FXasEL9c43rRWT/BKImHufptcbB1TTCTJOSoC0LGApN7HhXK7UnIQBoxC+Zm4u2q/hwAkJF3Fm9ssccWF2wmf1rWDABA6jq3KNh9Xa1IpUaBGUvDjpfrOOPT399tntUvitYDLd5NZrLnnxp2/FSZM0Pxnv6LHSjJ/QVOG/oA5c/8AnEzYsPGHu6oAQDMKFzNi5sNm6tfqnFNKRRIXPIoKBAMvH47iMfDi7urzFuKYsaaiWXSIm1sieWXE5vJwroo1Gp3LgeQDuBsAL/H2EUhgGwAlwAoBrBfrS5qUauLfsyrJSepeWzBg0cq62yuQiZMIHPOmFTeQOPCnZiUAgeJBWxGSbit5mGoiQ7OrPmi4HPhlspi06YD+K7mntS8Sfo6dMfkISklLcg3VY4zwPteXILDKBq32+XC3O53UJe0FjPzpT++fLxyWHJqOupjT8Vc/Vf+mqJCLGXkKLRxeaAU4f3dciodZyQpA4mUEzarWRJe97FvMEKSMCt7jSB8PtxSmnKsBisZ+8ONUNO1VQEA0uesCjtGruOMD29qlgoAYB8eEIXb1VsDG4lD3oLVISLEs0jq7/W/eAwlqpuwbvBdNB7eHD6YrgEjJAkz88KbA6bSuWy8Tc+YhyOLNmGFrQLl7z3Nizv22JfQIQtzOO5iLsTk7O9ItLA3mtFqd1q12p3faLU7d2q1O6/XanfOBzANwMUANPCWpSgE8F+1uuj34eJSFC6hKDRTFNooCn+i8f+aolBLUaihKHxDUVgy9n08ReGlMd9hisK54XIGmqPPwdsvJBYAOrZ1SIYd6D9e+z8AwKyVF7LyCuUWknNgPKVQwESlImbsolDs/u5t874DlaBaHDJ+qmidMPYXYstQf0heodyOPgfUoy0YoNlkRii3lP1NF2+Pz0Dy2LbaYnDXHfgvZmAQrlU3cIoV4pdzXA+0fA/TMYSWii8FY8+0d2A4dV7Y3JN5HhXKrUjyltkZMeol0XrmUCXak5fDfZx9sTNZ53Cu2Ez+hMRkDCIdiv5WUbBHe7zvreYuPIXWT2dyHWd8+js927t5lsfCfpyFm3PaUD064+YiNm7iHYgTYc4Mh5dSKLD+1r+jLPNSnG59C/r+7rCwlcPN6IkvmPAHrkg9hwsd16ddfQ8a45Zg0eEHodd2cuI++tc2FJor0JF5etAfAyNtbInllxObyXjvPgoAWu3OIa125+da7c7tWu3O9QDOAtAD4CG1umhVqHiKQgyApwFsALAEwHW+i75x9hohWE4IVgHYAeDxse9vBQBCsBzAhQAeoyh++aSuYd6hLZRfSGwoE4od6I/tPggdsqCeE3rfICn7hCu3RZGGWIdJcLvozNhZCwCYNmdFyPiponVyhvcxK/vwccm5nXOHkQMD3DOY66HxxZayv+ninQmZSPMMi8btqXgZg0jHsvOu5RQbqcdZqFhl+xlwkFgYK+kfIQ0Xe9gwgGwY4Z62IGxuNou0/o5NGau9OqwXnXvweA9me3owqlofUedwLths/s6UlZibXC8KdvxgE7TUDKQqM8OODxdbCj/X2JTUdDhILIiF/TgL61zmdiPf3gpj+mJO8ZF6LuPLTSkUmH7B7xBDERz9ek9IbOLxYJbzGEzKBbR+PhbJ41oRE4OUq55FAnGg+98THyMNxT2kroMSVsTMv5ATrxh+KbGlHB9SaC3oojDQtNqd3wK4FkAMgNvDCDkNQBshaCcEDgBvAJjw+CkhGL/HbQoAMvb/SwD8b+w3OgBGAGv5tNvwMfvjKGx+IbEAkHN1jmTY4/3E48HskWp0KdeAUihYeYVyC8mZLn40Jg3xLpPgdtFxu443wklioJ67NGT8VNE6dezdG4dpICSvUO5jX+0HACgL6IeeEG4p+5su3p2YhXQyAuLxCOZO+Kkbyy0laJ35Q8QnJHKKFeKXc1zn/XgeGlJOxRzdl7TvlISL3ddWAwBImrWE1k9nkzW2xOCOHyuzM2oaFF3rzqqx9wmXnBtR53Au2Gx+R+7pUMcMoL+L/m4hF+xsSxuOJ81j9NOZXMcZn/6mFAoYKSViRg2CtdYea0QaNQpKvYpT/FSZM7nwzll8KrqJCsltH4bEPt5zFEpYgRnLROEOFRvKPxnjOn/BKlTN24jVlm9Q9cnusLkTV3n3Jihc/0NOvEL9UvZ3KL+c2IxGt/uM0I9KtalEpdrUHOp3ALkSIC+M+/eNAHmK5ne/AchRgHQDZP7Yd7cB5C2AxAKkACBGgPyUiStSi9ePdo1Khj3e39VSM7aL56MheYVyC8mZLr7moQtJ67bVgttFxz2ZRWUnS2vbqIUQjZIcfOmekLxCub993rsrm8k4SOsXwi1lf9PFl7y6hRCNkgwP6QVzf/P8ZkI0StLZXM05Vohf7nFd9s5T/h0P+WIfevtxQjRK0tveyIk7HGwp/Fy5W6r2e3cj/Ozfomtd+uQtxFI8nTjstog6h3PBZvO3HSnx7oD57tOCsEetZuIsziAl/yjiFC/Xcca3v9u2rSLVD10kWOuKD18gRKMkrTVfc4qfKnMmF15CCPn2qTuIsziDDA3Q7/Dow/btANtY+qlo3FNhXDsddtKybQ3Ra/KIQacNi7t56ymk8a/rOfMK9UvZ36H8kVi8nn0LJf7WAuBKscAIwdMAnqYoXA/gzwB+DuBFAIsBVADoBHAQQNAOB/uofbcBuM0Sb8GBlAOIy45DwbYCJC9ORsvtY0VwPcCasjWoWu998TwmJQar969G482NsNRZYO+2Y/U3q2H42ID+f3nf08q/Ox/xqnjUX1WPhLwEZF2Yhfz78lFzbg0AIC47Dis/WYnqM6sRk+Yt9LrsnWXQva6D7k0dAGBO8Ry0/aENsUqvDNmXZUN9uxpHNnjfc3MNubD+6HrUXl4Le48dALDi4xXQPqOF/n097N12LH5lMTw2j/+56Jyrc5BzXQ7qrqiDvduOjPMysPSNpah/7D/ImwmY/5kH1wYXKk+rRMIsbx2jwscL4ehzoOsRb+2SmTfNxEjVCCx1FgBAyrIULH55MarPqYbb4u3i9DPSQSkof52VBc8sgLXRip4ne2DvtqNgewGUpyvRdEsTAEB5qhILnl2AynWVsNRZkLIsBaccOgUtG1tgKvfeBVz00iKYSkw49pdjSMhLQO6duUhenIwRXSKmJw+jragNxEP89YUCdfK1NW1NWpBObXd5a8zYe+1Y17rOr1POhmPQK+fDcm09RttGYe+245SKU4J0UiQq0HhjIxLyEoJ0SshNwPJ3l6NqfRViM2ODdAK8O1y13N6C2IzYIJ0AwD3ixrrmdTh8yWE49U4AwKp9q9B1fxcMnxtg77Zj6VtLg3TK2pCFhusaYO+2I2tDFha/vBgNFzdj/rlJGK3uhucGDyrXViIhLyFIJwBQ36bG8NfDsDRagnQibgIqhkL699LhsXmCdNLu0iLmnGp0paqR0RyLitsrAAAZZ2Vg7iNzUbW+CtZGK9LWpgXptOT1JTB8bEDn/Z1IyEsI0inrwiy4TC6YDpkmjKf6MZ0AIHVVKpIXJAfp1H6vt+aYQ+vA2sNrg3TyjSd7tx2n1p3q18mwlAAFQM8HzRj6PUFCXkKQTkmFSVj6xlJUnlqJuOy4IJ0AYN6jc5Hf9R/UuhfDfr0Lipt6/DoBgMfqwWkNp00YT2tK16D97nYYvzbC3m3H8veXB+nkG0/2bjum/2R6kE6nHDoFFWsqkDg7MUgnAMi9MxeGTwywtlqDdPKNp7Q1aXAOOYN06v9XP+zddsx/Yn6QTr7znrXRivjzl8O5Oga9r++G6Tbv4yu+8173Y91IyEsI0in7smzYtXaMVHprSVnOrcZoSjwGimzQ9niPqbRT0pCgTpgwnsaf9xz9DqwpWROk0+FLDsNcbUZCXkKQTr7zXutvW5GQlzBhPAHfnfcqVlcgfmZ8kE6+8VR/TT3ipsUF6QQAxEFw6pFTJ+g083XvncK+V1tguaEcK79YGaSTb36yd9sx4/oZQTqt3r8aFasrkDgncYJO03QVaHYtxKx3hjHw9gBGj40G6eQbTymLU2Dvs/vH0/j5yd5tx8J/LAzSyXfeszZakXlh5oTxBHx33ut9phcJeQlBOuVcnQNrixXmGnOQTr7znnKdErHK2CCduh7pggcUsn+QAnfLN6hYe9oEnarPqYa1yYqEvARanayNVhz941Ek5CWA/Kwfp1IeGL6ajornKvznvfLl5UjITfCPp8D5qfaHtYibHhekk3dg819HtN3VBnu3HapfqGjXERUrK5A4NzFIJ8B73ut/uR+2LluQTjEXpCLRPoiuB7owemyU1zrC2miF6/aDcCTHYHhzPCp03jHpG099L/YhIS9B9HUEADgHnFj+wXLadYStw4bEOYlhryN8OvnOe+XLypGQl0C7jvDYPFDfpmZcR1jUpyF2zb9w8O6nkVP746D5ybeOGPr6EABAf1cc9Pfo/ePJ3m3HrNtn0a4jypeXI6kwKUgnwHve632qF3atPUgn33hKUCXA0mjhtY6wNlox/arpQTr5xtPx144jIS8hrHVE7G+egfKtDSh7+DYkfr2FdR0xmjyEM85rw0HlL1CxNngdYe+2I3lRMq91RP59+X6t6dYRHocHM2+YybiOUMQpsGzvsrDXET6dfOPJ3m1H3u/zOK8jCh8vRNdDXXDoHEE6+cZTXGYcRqpGeK0jGI3uSjHwo1JtosL53bjfv65SbbKF+h1ATgfIp+P+vRkgm1l+rwDIMIPvIECWMMWy3SmU00LVSBGN57GfkgFNPvG43ZPKS9sWjtwlT/0fMRXPEJ171GomruL0oL8WS2WT2ec9W+aT8sd+IjmvVjOPVDx6OaN/Kh1nNV/uIUSjJE3lXwrirfvmA9a7GVJaJPT34QfPJz1b5vvPNVzt8IPf9z8ZwJVbDuPKbdT3e2tx/XubqLxGfb+3JubYEwJSmpz9XfbH80nXlkXCMN55kvFOPpvJlTdf3vLHfkJ6tswXzH3kgXM5j0kxLFL72+N2k94thaTmoQtYf1fx6OWkd0uhqNxSmtjcB1+4ixCNktR8+Trr78rff57xCROpTc7+ltMgsE6hSa0u2qdWFz2qVhddo1YXMRb2UKuLEuDdcKaT6TfjrBzAfIpCAUUhHt73EfeO/wFFYfw+vpcCaB37PpmikDL2/xcCcBGChjDzmWANN7CHsfmFxALw/0VICmyfn3g8yDVVoyt1pX9XJzZeodxCcqaLJ4kZSKNG4XI6RO1v7dE6xFAEsarvXp6f6lr7zByTgYSxrdul0tqo74cKA3DmLGeMFcItZX/TxSdleN8tsJl0grhHD70IE0nGsgtu5Bwr1B8J49o2/4eYRY7jaG0JL+wcWweMKQW8uENhS+Hnyp2aPg0eQoGMGkXVur3ycygogvRF54aMDeUXe2xxwQ7lNzpWIo9og3Y35ILt7quDjcRh1txltH4mk+s449vfrsQspHtMgvRILExArr0VBmXwJjOh4qfSnBkuLwA03tSErhkXYrG1EsNDekbsaZZW6ALeWxXKPZXG9Sk33I8ORT5UBzaj6ueljLGk9QsMkTTMW0Ff7jwSx5YYfjmxmYzyXjCym1pd5AD8j5r6AobgfXSzHEAVvHULcwD8AcAFAP6u1e68K2QDKPwAwN/g3ZzmRUJwP0VhG4AKQrCXovD3MTznGOcdhKCeojAHwKcAPAB6AfwfIcwXomvXriUVFRUhcz0RTdvRDPXLp+HQ4s1Yd01Q1Y+It9LX/or1LY9g6DdNyJyuEg234sN/YG35H9B+5WeYu2ydaLiRYDUPX4w0hw7z/lItGUfVp69gTckdqDv/X1h21tQvT9rb3ohZ/1qP8lX349TL7+CFYdT3I+nJZaiZfhnW3fGSyC2cGjY00Ie0p5agPPcmnH7r3znFmk1DSH18Dkrn/Abrb35AohbKb6YtajRmX4J1d7xI6zebhtD44kaofvhn5BYuo/1NoJU++yus6v8PqM1dSEhMFrO5EWWt1Qcw/73LUHHqo1h76a28MGofPA+JLhPm/6VS5NZFlpW8/Cec3vEs7H/qCzomHHYbqt97EgptJRbc9CTSs6bTYvR3tWLmi2txaMl9WHf1Hyej2VPCmiq+xKIPfoKKNQ9h7Y82BvntNitiHpyF8tyf4/Rb/zb5DYwQa6nah3nvXY7umDyMxE2j/c1cWyMalWdg7e//O8mtO3mNoqhKQkjQDoHh3ilMBbAOwG8AvAygFoASwEUA7gPwNrwXhp/AWx6iB0BYMzoh+IgQLCAE8wjB/WPfFRPivWNICH5HCJYSglWE4DxCUD/2fQchWEgIFhOCC9guCEOZnFf6hy85LBm2z99b492RLmfZeWHxCuUWkjNdfExKFgDAPKwXtb+d/U1wEwrqed8tuqa61j5zJGQhzW0MycuXu+3wNygsuQe9nhzMXX0uY6wQbin7my5eOc27a6vbzP84a3zvESRQTli/+QHnWDH8kTCuM6er0JS4ArnazybsQhoOttZXN1S9hNYfiptPu+UY12YqFTGOYUZ/7TuP4lTjx2j/x/aweafpK3A0YbF/8R9J53Au2KH8I5uTYSZJcLd/wxtbZW/HUNoCzvFyHWd8+1uR4r3Qq779oP87p8OOsv/8DYMPLsO6hr/iVOPHOP7MD2AyDtJi1GzzPryVMZd+h+kTZc4Ml9cXu2D1uTiOaYhp2kvr72mpQSzlQbw6+I86kX4OFxN7wZpzUbH0PlgcSYjz2Gk/nfGFGKq5lHe7I3XODOWPxDuFQnYYTVSpNq1XqTbdoVJtelGl2nRIpdpUpVJtelal2qTmiyvV52R+p/DQzmvJkEZN3C7XpPIyGVdu3w5eYjxvPp678pEfku4tCwVj8uGW2g4+dzuxF2cRj9stOm/bkRIypFGTPs080nusifW3U+k487jdxF6cSQ4+fwcvvuEhPRnWzCRVO34Qse/CTBZ36Z4dhGiUpL2+jBPGVHvXiy93y7Y1jO8jWUaMxKDJJUSjJCPFM4jZNBSSd3hIT1zF6d53eCbB5O7vww+eP2HXaC6m7+/2vtP56lZe3HIYX97KT/5FiEZJ2g5/S1xOJyl75ynSvWWBdz7dvpYc/uptUv3Zv4mjOJM0bl9HRoYNQRif3f5L4izOIKOWEaFpcLZI7++Sp39JbMXTaPvNtxNzR2OlJNxS2MnIHX2nkN87hUGm1e60abU7S7XanU9ptTt/odXuXKfV7lyj1e7cqNXu1PLFlcNaNrbw9guJDWVCsX3+WcOVOJa8EoqYmEnhFpIzXXxiqvdOoW1YL2p/Z1mPYSBp4rtLU11rn1Ep2YinXDCPGEXlPtZQjqz/XAU7EuC+6X2YH6ZC4vPllrK/6eIphQLDY3W9+HDXv7MDSliRdtFmzrFi+qXC5sI77+xr4CEU+kvf5ITtOt4IB4mBumAJrZ+PRWJ/22LTkOg00fqPvLsTmTChdN7vkEqNov6Lf4XkOVb1JWIogrSF54Td7sk8h3PBDofbolqHOZ5uGHS9nLH7WryPjKbmr+TFzYYtpZ9PrO896Z7Xn0PvAytwas29sCuSUXPmc5h/7yGsOPenSP7vWtR9bycKnc3ofvJSWALmjKzUNnTH5CExmb4Y9okyZ/LhVq75KRIoJ5q//k+Q391fDzuJw6x5wXcKp8I5XGzsqa61FH45sRmN7krxRPyw3Sl0jjhZr6jZ/EJiJwO7v7uN9q+ikd7u8dbRWEmIRknK339eNGyH3UYcNHeFprLW4813x6XnaJ1o2B2NlUSvySM6zWzS1XpEknaLFcvX3751Bal6eAPn2JFhAxnSqEn1QxeF5J6qfcY1tv7+M0j71uWcsKsfumhCDF9usWKl4q585DLSuXVxkH/UMkIGNPmk9oGzicftJh1bljDW7hof630yIJNYzSZJ2x0p2I1lnxOiUZKqT17mHFvy722EaJTeO4Yiti0S+6yzuZoQjdJ7137rclL1ycsTnhgaH1vx4QvEVZxO6u4/c8JxNKDJJ2WPXzWp7Z4q2C6nkwxo8knlIz8M8h9+8DzGHVsjPa8o9tTnDhULse8UsplaXfQHtbroGSmwudo+at9l+6h9u2xdNhgPGKHfq4dujw6mMhN6nuiBrduG2g218Lg8aLy5EQDQeOPYf29uhMflQe2GWti6beh5ogemMhN0e3TQ79XDeMCIxusaYW21omNrB1xml//q3Pc8b/X3vBt9tBW1wTHgQNeOLphrzejf3Q/DZwY0/7IZ/bv7Ya41o2tHFxwDDrQVtU2I9WG1bGyBy+xCx9YOWFutaLyukTGn9s3t6N7RjSOPvgUAcBxYNCGnIxceYcxJu0uLY/ceY8yp4YYG9DzWw5hT68ZWxpwabmhA1/1djDlpd2nRdmfbhJzMH3prsAyXdKN7RzejTu2b29FR3MGYk7XVitoNtXCZXSj/w+eIo9wYqZ85IbcjFx6hzcnwmQFNNzYx5jReL7qcjAeMaLq5iVYnj8uD6jOrGXOyddvQcFUDY04dWzvQ9UDXBJ3ildkAgBFDPw6fd5gxp/7d/Wj/YztjTj6tS377EVLeuAKEAEPnvIGY/02D8YARLbe2MObUeHMjuu7vYsyp54keHC06GnZO4//bVtSGzm2djDmZa82ou6yOMScAOHze4SCdzEhHwqgBzbc0M+YEANVnVE/Iqezh7ciAGTb8GqYyExpvaKTNyWV24fB5hxlzcgw4UH9FPWNOXTu60LmtkzGnmrNraI8933g6WnSUMafGGxvR81gPrU6mMhOab2lmzKllY8uEcd1wQwNMBRtQ4OlEe3kFunZ0of2P7SFzyrZ14PhobtB5r+uBLsacTGUm1F9Rzysn3R4dWm5tYcxp/LgO1Ml37DVc20Cbk2PAgSMXHqHVyTKYhBSP9xgdn1PZc48gG0Y4Cjah9ykt2mw/wCJnA9pry4LOe9WnV/tzytSVow3z4awn/vHUdmcbY06+cc2UU9ONTYw5tRW1BWkdeN47du8xWp16nuhBR3EHrU6+8dS9o5tWJ9+x13BVA2Ymr4GVJMDeciDsnPp396N1YyvQUQs9SUeCLoPznNtwVYMk64hQc66v9iHXOXfW3GX4ynoFvvJoYGn5N1Zf/HO03XGUdh0x2/VjHEzXYLG9Fi07LoWxVY+aP32LbBhh7CmgzSnUnCtkHWH4zICGaxsY59zxWoezjuAy57bf0866jqi7rA4uswtH72zH0ezvY9HIIYxaRiasI9S2oxhwFXCec31aM60jQs25PY/18F5HdGg6eM+54awjmHLS79Wj8YZGXjmFWkeEmnOP3XeMdR1Rf0U953VEuHNu4DqC65w7fgxwXUcwGt2VotCPSrXpHyrVJrcU2Hw/bHcKLS0W1itqNr+QWELYn2cWim1psZDSv99AhjUzics58a8GoZ6jlrJPuHLbRi2EaJTk4Ev3iNbfVZ+8TIhGSVqqD4QdH+laj7fmyq8I0ShJ9eevCda6bX8lOa6ZQwY1ueRYw0QssbXmgi0Fd8UjPyKdWxdz4raMGMmgJpccfvC8sLjF1nq8RdK49j+l8NLmsLBHLSPeOnv//INgbi6xcozrg8/fQezFmcTc9N17WqNWMzmumUPq7z/D/11vWRuxF2eSkmduY+S1jBjHnnq4k1O7J/McLla7xnPXPnA2adu2kjN2y7Y15MgD5wri5hM7VcZ12TtPEndxOjn84Pmk8uOXCNEoSUPpJ6Jgc/VPhf6u/fq9oLvWvWW+J7S2SMrN1T8VxrUU2HL1dyi/nNiYzDuFU82MXxl5+4XEhjKh2MavjJhprMKxpOWIiY1l/a2Y3EJypotPSEyGlSSAGjWK1t82rfevKrMKV4QdH+laj7fUTO8dUMfIgCBubUczkr+8CvFwwHjl25izeOIOdGJrzQVbCm5XYhbSPCZO3Efe+zuyYEL8eeGVexFbay42meN6Ru48NMUuRnb3J2Fh97YdgYIiiJ+5SDA3l1g5+ptKzEA85Ub/l9+9E3f4/aeRAwPcZ9793Q+rk1CXdiYWHv8QdpuVFuto1f8QR7mRsuCcCd9H0jmcC3a43CMz16PA1YFhw3fnuFCxg1/qkefqhCVjoSBuPrFTZVyfevkdqFixBSts5Zhf8id4CIW8xaeJgs3VLyR2srgXrbsEQ1DCXfee/7uO/d4arSl5KwJDReXm6p8K41oK7EjllhObyaIXhQCSF7HXdGLzC4kFgMLHCyXDdswyYranB6Oq4Bp8bLxCuYXkzBRvplKgsA+L1t9xhlb0YTqSU9PDjo9krQP96dnei0LPyABvrR12Gzy7f4QUjEJ/xZu0tRyl0DpcbCm4PUlZSCdmJCyIDyvWZjWjsPWfqI9fiUXrLgqLW2ytx1ukjWvjnEtQ6D6K3vbGkNhDnbUAgKw5wQupSBpbYnArkjMBACR3FIB3rM1u2IWm2MVYduZlE+LjTr0ZmRhB3Zev0fKaWw7ARRSYu+b7nNo92edwMdo1njt90blQUATHqr4IO9Y8vRuJlBMxKvraj5F6nE32uD7tp5twaMmfkUaNoodSI1WZKRo2F/9U6O/YuHi0Zp6NRaaDsI1aAADu+FYAgGoBfRmPSNJaLO6TQWux/XJiM1lYF4VqdVEuL/QpYi6ji7dfSCwAOPockmH3HvsKAJC55LwgHxuvUG4hOTPFWxRpiHMMi9bfmdZ26BLncIqPZK0D/ckpSthIHGAd5K11a8WXyCX9OJJ9LwpXnsGrXUKOMyn7mymeSp4GBUUwfFwXVuzhvU8iG0bgnHvC5hZb63B5hWLz6e/8M68DAHR/+3pIbGd/I1xEAfXcpaJwhxsrx7iOHdtR2TLovctV88GzmIkB2M/4PSiFYkL80jN/jD5MR/yRV2l504+XoT2uMGjhHmnncDHaNZ577qqzYSdxsLXuDzt2sG+slmbBakHcfGKn2rhed/XdqFj7CLoyfyc6drj+qdLfiSt/glRqFE3femsWxo40wQAlsmfmSc7NxT8VxrUU2HL1dyi/nNhMFu6dwi61ukirVhd9oFYXbVWriy5Tq4tUvBgj0DyjHt5+IbEA0PUI8wufQrExfAhWkoC5K4IX9Gy8QrmF5MwUPxqThgSXSZT+drtcyHX1YDQ9+C9EU1XrQD+lUMBIpSNmdJC31qb6T+AkMZiVcTHvdgk5zqTsb6b42DTvBj1mE/tFoWfUA7vNioKmXWiMW4olp28Im1tsrcPlFYrNp7/VcxaiNXY+sjo/DomdYGyDNkblL74ulDvcWDnGdbyvzI5lEC6nA7PqnkVr7HysOOenQfGKmBh05P8Ey+3V6G1vnMBrs5pR6GiCIftUzu2e7HO4GO0az52YlIKjCYswbbAifOwR7x8echesEsTNJ3Yqjuu1P7wNia+cIgl2OP6p0t+Lv/dDDCMFztp3AQDTnG3QJsydFG4u/qkwrqXAlqu/Q/nlxGaycC8KOwDMBPADAH8B8C6AHrW6qFetLtqrVhcVq9VFl6rVRTN4tUJmSyxI5O0XEhvKhGKrXDU4mrgUcfEJk8otJGemeHucEkmuEVH6u6+jCQmUEzEzFnOKj2St6fzmmHTEOwy8uXOOf4PWhCWYtpB5WEuhdbjYUnAnKKcDANzK4ZCxNe8/gxwY4DrzDxPu7vDhFdMvFTZfXn3exVjgasFIOnP52sSCREwbPYbBgLqhQrnDiZWjvxPTpgEASKoZ1R/+A7PIcVjW3RV0HPniCy68DW5CoevL5yf4j1bvRzzlQuL8szm3e7LP4eFic+E25ZyGuc42mE1DYcWmelrRGzMLiUkpgrm5xp5o41oM7BOlv+PiE9CSfhYWDn8N26gFeaQL5ozgd6Ol4Obij2odWdxyYjNZWCsZrXbnXADTAFwM4F4A7wDoBqAC8EMAWwDsBaBVq4u64b14nDJmKjXx9guJBYCZN82UBNuo78dc0gkzzfuEoXiFcgvJmSneFZ+OZM+IKP2t7/C+u5SWF/yY2lTUmslvjctEktPIq7/1/V2Y527H8KyzJ13rcLGl4E5K9xZ7HjnaG+Qbb4aDeuTXP4fm2IVYdtblnLil0DocXqHYfPs794xrAQDaQ39jjB08OIBZbi1sGfTvd0Ta2BLKnZLhvSM92tOPmYefwtGYAqw8/1rG+Jl5hahLPhXzet+Dy+nw85qa98NDKBSsuYBzu0+EcZ2y4BzEUATtlV+GFTvD0Q598jxRuLnGnmjjWgzsE6m/41ZcASUsqHn370iiHFDMDF5fSMUdrj+qNTdeqbnlxGaysP+8rdXuHNJqd36u1e58SKvdeaVWu3MOgOkANsB793AvgF4As+C9WIwIC6dOobXZylpfyNpsZawv5La6WesUGvcZATDXF4pJi2Gs6eeLZarF47a6GXOqfvxtAMBo+WLanCz1FtaaSSmrUljrC2Vfkc2YU+y0WNY6hVkbsljrC8Wr44NyclJpSCNmTLtsGmudwrR1aaz1hUbbRmHtqgMAqApXBdV4MR8xM9ZMIi7CWqfQpxdTLR4qnmKsL2T82siYk63bBueQk7W+UOZFmUE6mYdTkeo2YqR6hLVOYfLi5OB6XQ/+CwDg+HYVsq/IZsxJkaJgrZmUtSGLtb5QYkEip5zG1+JJPzudtb6QvcfOWl/IVG4Kygmd3jsIHjLEWqewue0VqDCAbt3PQTwIqi9EPISxvpCp3MSYk2PAAUe/g7W+UPrZ6cw5lQbnNP4ckViQyFqnMPuKbMaaSYgBa82k8eN6fE4z0hfiG/IDnEXewf4Hfkebk22BFrGUB3EzFtPWv8u8KJO1TqFjwMErJ90eHRQpCtY6hYHn4cDznnvUzVjTz3zETKuT7uERAIDa/l/kES1642/F8DemoPNe2ro0f0663kuRAwP23fWcv11p/YfQTuYgwZwadN6LV8ez1inMviKbMSfiIqx1Cum0Hn+OSFmVwlincHxOdOfyaZdNY61T6Bp2+XOavfBMOEkMLK37Q+bU/kILZlHHYaHm855zXcMuSdYRoebcoS+HaHMKd84d3y9c1hHtm9uReWEm7zlXyDrC8JkB7lE3r5yY1hHhzrkpq1JY1xG2DltQTovP+BHMJAnzmncBANKyl3OacwO15jvnZl+RzXsdkX52Ou85ly0nn15sdQqJh/DKKdQ6ItScm7YujXUd4eh3cF5HhDvn+s43fOfc8WOA6zqC0ejqVAj5qFSbclSqTT9QqTb9RWxsIR+2OoVH/3SUtZ4Hm19ILCHsNVKEYJc8cxsZLZ7mrfHHkVcot5CcmeIPvvhHQjRK0nRPI+92+bjLHr+KHNfM4RwfqVoz+UueuY1Yiqfz6u/yR68gA5p84na5Jl3rcLGl4LaaTYRolOTLe4sY45wOO+n+y3zSum018bjdnLml0DocXqHYQvrb7XKRb/98hbfe6Mubg/xfbn6cEI2StNZ8Izq3lP3Nl9vjdhNncQYhGiU5tnUZcbtcIeMddhsZ0OSTqocvIYQQUnLqt8RanE1Knv4lr3afKOO6cfs60rh9XcjYxvIv/LVbxeLmEnsijmuh2Cdaf5c/9hNCNEriKk4no5YR2t9IxR2OP6o1N16pueXEBkOdQtkv1ibrw3ZR6HbSL+zC8QuJJSTEQkoAduu21aTu/jN58QrlFpIzU3zpGw8SolESXU8n73b5uFu2n0KOPHAO5/hI1ZrJf/Dlzd5F+LqvOcW6nE5i0MwiZY9fGZJbCq3DxZaK21I8nRx85leMcWXvPu0tVvzpK7y4pdA6HF6h2EL72zFq9y+aSl4pnuA7+MIfiLs43XtRLjK3lP0thNugmUWIRknKP9gVdvzB5+4gzuIMous9Rj7+8T+CimZzafeJMq4PPncHcRRnEqvZxBpb+uajhGiUpPdYk2jcXGJP1HF9Is2Z4fCyxVZ+8i9CNErSuWXxpHOH449qzY1Xam45sZkuCkWvU6hWF2Wq1UXnq9VFd4f+dWRY8y+befuFxAJAyjL6F96FYJuMgyhwtaPfSF+LKRSvEO5wYvlwx4zV9WreVsO7XQCQtCwRs5xdsCjp312aalqz+WNSvO8tUctHOcUePfItMjECqvCCkNxSaB0utlTcJkoJT9fxoO9toxaUvvZXFFY/gKPu2Vh5/nW8uKXQOhxeodhC+7vt10ex6s7XUZl2Hta3/Q2lr233+6ijDehT5CApJU10bin7Wwi3SZGOTo8aqy++Jez43PNvQyzlwdHPdiG20Ptu9Jw1F9GFRtw5XIx20XEnLzgbcZQbR6v+xxjrdNiR3PY+RkgSVPnz9gdcNwAAgrBJREFURePmEnuijusTac4Mh5ctdslZV8BKEtBnnzPp3OH4o1pz45WaW05sJqO8F4z8bKwsxWoAa8Y+qwHk+/xa7c4Y3uAi29q1a0lFRUXoH54Advh/b2DlgV+h7sJXseyMy0IHTAE78tXbWLH//9D0g7ex6LQLeeP0d7Vi5otrcWjpX7Duqj+I2MLIs+rPXsXqg79B2xUfonDlmWHHlbx0D9Z1PA/jb+qRlTNLwhZGprVtXwNLfDZW3vMZAG9h8er3nkBBw7PIgQH18SuQcsVOzFlMX5g4auzmdNhR+/crscZyAIcWb8a6a/6EY9tWYjhhJlbd86nczZs0ayr/AklpWZi9aA2nuPoHzkKG8zgGE3KhdAxgTnGtRC2cGjYybEDy43NRnvcLrP/l40F+l9OBI3+/EmvM+3Fo8b1Yd809NChRi5o41lD6CdJz8jCLpt5q1KIWKUZRVCUhJGgRE/adQrW6aI5aXfQTtbpou1pd9KFaXaQF0APgfQDbAPwYwGwAZgClAJ5nRoss873AyscvJBYAqs+pFh3b2vo1HCQGrl1qXrxCuMOJ5cOdkOat69Xzrxbe7QKAmns/AQCk5dKfsKea1mx+306aR3dw6+9M7QEcjSv0XxBOttbhYkvFbY3LRPyIAU6HHWX/2QnDg8uwruF+DMbNRN2Fr2LpvV9j9AH2vy5Ottbh8ArFFqu/4+ITsOy3b6E6+XtY1/ggDu15GLPcPbClM+8KGWljSwzuRadeAOv9SZy5R5f9DLPIcaywVeJ4JvMFZSRoLXa76LjT0rPQHleINF1ZUKzb5ULNE9dijXk/SuffBeUHPxKVm0vsiT6uxW5XKO5I7e8l6y+BScO+tI5qzY07UrUW6pcTm9HonikN/KhUmwZVqk3usY9n7ONSqTY1q1Sb3lKpNhWPfbc7HDw5PmzvFMppoZ5n5mNN208lDX9dP+m84Rof7q6WGu/7N+89I4j741s3EaJREoNOKwiHj012n/v67KMbt4YdYxzUEVdxOjn4j02itGGqHWeEeDcLGNaoSPeWBYRolKR5+1py+Ku3GTeVEZNbqE2l/raNWkjNQxcQolESolGSQ/99YtK4xbTJ5h61jJBhjYoQjZJUfPCPSeX2WaT1d8kzvyK24mlk1Gr2f+dyOr3vRWuU5ODuP0vGPRkWaf19onOfjDmfrNxy5iynQeA7hZkACICXAPwfgNMApGm1OxdqtTuv0mp3bhv7nZ3fpam85tselo9fSGwo44NtGTFinrMVxpzTJp073Fg+3Cnp3vfjhvZ38m4XACSndsIAJTKn01dNmUpah/Irp3nvFFOJxrBjjx76ADEUQebyDWFxS6F1uNhScTtTZkIJC2yKFNSc9Tzm33sIK8796YTC4pGm9WRwi93fCYnJWPjbd3Ek8RQAQGbBKkm4p2p/M/kTk1PROH0DPITC7NXMj9JHktZcsPlwJxaejQTKiYN/eBMA4HG7UfnUjTh1+DOUzP41Tr9pu2Tc4cZGKvdU01os7JORO6p1ZHHLic1k4V4Uvj/220sBQKvdWaHV7hzlxTjJFk6dwvgZ8az1heJnxDPWF1KuU7LWKSQO7zubTLV45myfw1hfyBfLVItHuU4ZlFP1U+8ilvLAcWwZFjy3gDGnzPMzWWsm5d+Tz1pfaOELCxlzmn7NdNaaSWtK17DWF5rx8xlBOsVbvBtQpC52sdZMmq2ZzVpfaHpqD/riZjPW4olNj2WsmZR+djprnUKfXky1eGbfN5uxvhA8YMzJ1m1D6spU1vpChX8rDMqp57fH4SQxSJjGXHuxf3c/cn+X689p9JuPYEIynE/OmKA1U07ZP85mrZm0pnQNa30h1W0qTjmNr8VT8GABa32h6VdOZ60vpEhQ0OY0I+M3KJ22C9Mv/B+y68+EvdcRVF+IOAljTqYyE3LvymWsL6RIUDDm5BhwIHlRMmt9oYIHCxhzSjsljbVmkuo2FWudwoUvLGTMKfOiTNaaSePHNV2tuNzf5Qbl5GojSMazqFrzPLB7Bm1OLRtbUPi3QtY6hTN+NoNXTro9OmT/OJu1TmHgeTjwvKe+Xc1Y0y82PZYxJ5fZhcTZiYw59TzRg9ma2bQ5Lb3xUewf+BuycvIYa/rN+PkM1jqFC19YyJhT+tnprHUK6bQef47Ivyefc06+8bTguQWsdQpV/6cKyqnglAvhIRSSTmlE/Q11KH/mFpxm/AgHZ/wfZg38xj+epl8znTEnOq0Dz3uq/1NJso4INecmL0im1SncOXe81uGsI8afI1Z/s5r3nCtkHWH4zAD17WpeOTGtI8Kdc5e8tYR1HZE4O5ExJ4B9HRE45waey31a851zF76wkPc6YuUXK3nPuWw5hVpH6PfqkXtXLq+cQq0jQs25y95fxrqOSF6UzGsdEc6cG2odEWrOHT8GuK4jGI3u9iHdZ6z2YNvYI6QVKtWmMwL8HpVq065w8Sb7E6l1Cls3tYqKfXDX74izOIOMDBtY49l4+XKHG8uXe6R4Btl/34282+Vxu4mxWEVKn7iJV3ykaR2Of0CTT76+96qwYj1uN+nXFJDKHT8Mm1sqrcPBnqrcUtYmmorjWmi7QnFLXacwqrW42Hy5j25dQWr+fCYpffJm7yOjz98Z9Mh3VGvusZGotRjYcvW3nNxRrbnxSs095esUqlSbElSqTRqVapN17OLwDZVqUx6Z4heFo12jrJ3H5hcSSwj788x8sBv+ejpp3r42ZHyo56il7BO+3H2aeeTQjp/ybtdAb4e3Rtpr9/OKjzStw/G3b11BDt19flixxxrKve92vf142NxSaR0O9lTllkrrULxCsaP9HVncUa0nWulTv/C/n1ryzK9o3wGOas09NhK1FgNbrv6WkzuqNTdeqbnlxGa6KORUp1Cr3WnXanduBbAUwEcArgbQpFYXbWePjGzTv6Pn7RcSG8q4YtusZsxzNMOQfeqkc3OJ5cttiUlDrNnIKxYA+o/WAABSGXYeDRUfSVqH67fEZSAl1hRWbH/lBwCA/NMuo/XzaVcok6u/5eSWSmupuaP9ffJwT0WtExZ5368szbka6371zIR3gKXmFgM7qrV4vFFu8XmFYp+M/R3KLyc2k4V1UahWFz2sVhcl+/6t1e48ptXuvAzeMhTHAdwH70Y02bxaIbMp1yt5+4XEAsCCZxaIhn20ej/iKRcS558dMp6Nlw83l1i+3LaYNCTFWHi3y9xTDwCYOW8Vr/hI0jpcvz0+E8p49otCX2xK9350KPIwM68wbG6ptA4He6pyS6V1KF6h2NH+jizuqNYTbeV5V6Nh9ftY9+vnaS8IpeQWAzuqNTfuqdrfcnJHtebGKzW3nNhMFu6dwrsBNKrVRVeO/1Kr3fk+gMUAtsO78+iP1eqib9TqouW8WiOT2Y7ZePuFxAKAtdEqGrapeR88hELBmgtCxrPx8uHmEsuX2x6nRJJnhHe7KH0zTCQF02bm8YqPJK3D9bsSspBO2C8KbcdssJqHsdB2BP3Tg4vcy6F1ONhTlVsqrUPxCsWO9ndkcUe1nmiUQoFs2yLGC0IpucXAjmrNjXuq9rec3FGtufFKzS0nNpOFe1H4PIBZAPao1UWfqtVF832OsUdKNQCWAfgYwPcAVKrVRY/zapEMpkhi7wY2v5BYAOh5skc0bGX/IbTHzkV6ZnbIeDZePtxcYvlyu+LTkUrYLwrZuFNHjqLbmcu6aJgqWofr9yRPg5KywOlgrhajSFKgtexTxFMupCy9mBO3VFqHgz1VuaXSOhSvUOxof0cWd1Rr8bEjlTuqtfjYcvW3nNxRrbnxSs0tJzZjXDg/0mp3bgRwKoASABcCqFWrix5Qq4uSxv2mXavd+UMAVwDoBfA7Xi0S2cIpSaF7U8e6lbTuTR3jVtLGA0bWkhS9T/UCYN5K2mVyMW4l7Ytl2nbZx6/fq0fPv7swz96APs8K/9a3MakxjDnZe+2s22O7zW7WraRjM2IZc7K2Wlm3x2bLSbtLC1unjVYn60Ai0mBh3R7b4/DQ5lT7ZAkK7C3QmWazbiV9/NXjjFtJm0pNrDn59GLadtlldDFuJd37bC9jTrZuG4a+GGLdSpqKp+h1ypoNACh/7RHGbZedg04M7/8QoyQese8U0mrNlJOl3sK6PTZbTj1P9MDeZ+eck288AWDdStrR72A99vr+2ce4lfRI1QhrTj6tmbaSdhqcjFtJ9/2zjzEnx4ADgx8NhtxKmiknW6eNdXtse5+dVafYjFjGnEaqRli3x2bLqWtHF5yDTl45tWxsARVPsZakcOgcvHLS7dHBUm9hzSnwPBx43nMOOhnLNxx/9ThjTi6zC/p39awlKTwOD+N4snXYGHPS79XD1mlj1Sk2I5YxJ1OpibUkBVtOHVs74Da7eeXUeHMjYlJjWEtSOAecvHLq390Pa6uVNadQc65zwCnJOiLUnOv7qz/fOXd8v7CtI+jO5YQQXnOu0HWE4TMDnINOXjmxrSPCmXPZcrK2WqF/V897HeGbc5nGk09rvnNubEYs73UEcRPecy5bTqHWEfq9ejgNTl45hVpHhJpziZOwriMGPxrkvY4INeeGWkeEmnPHjwE+6whao9t9hu2jUm26QaXa1Du222iHSrXpJzS/SVSpNm3jii3lh2330d7ne1l36WHzC4kVE7ux9FNCNEpS+cm/RMfm6pcK++DLmwnRKMmoZYRTrMftJocfPJ9YiqeTw3//mnfbpmKfOew2UqG5iBCNkpS++ShjbNeWRaTmQfpdSk+0PpGbO4p94mDLyR3FnlxsObmj2CcOtpzcUezJxZaTO1QsxNh9FAC02p2vAlgA4BEAMwG8pVYXfaxWFxWO+41Nq91ZzBVbLss4L4O3X0gsAChPZ34ZlAu2sXEfAPjfJwwVz8bLlZtrLF9uRVImAMA0NMAptmLvM1hhK8eRRZsw97xVvLhD+cLxi6U1F39cfAIKLnsVh5NOw7r6bSj/79+DfmNbrEMe0WJ09nmcuaXSOhzsqcotldaheIViR/s7srijWouPHancUa3Fx5arv+XkjmrNjVdqbjmxmYzXQ6da7U6LVrvzHgArAHwK4GJ4Hyn9q1pdlMirJTKa7jUdb7+QWABouqVJFOzk/lJ0KPKROV0VVjwbL1durrF8uWNTswAAlmHmrXYDY/X9XVhQ8wAa45bgtKvvEZR3pGjN1d/+q04s/O27OJJ4Ck45rEH5e89M8Hd8/i4AQH3KpZyxpdI6HOypyi2l1lNxXAttVyhuKftbTu6o1uJjRyp3VGvxseXqbzm5o1pz45WaW05sRqO7fcj1o1Jtukyl2tQ29kjpMZVq0+XhxgLkEoA0A6QNIH+i8f8aILUAqQHINwBZMvZ9HEB2j/kaAbKZjYft8VHniJP1NiubX0gsIeyFM8PFdjrsxFI8nZQ+eXPY8aEKdkrZJ3y5myv3EaJRkrLHryJulyus2ModlxJb8TTS2VwtiDuULxy/GFrz8ft4Ry0jpPaBs4mrOJ2Uv/+831/10CWkd0shbbHnUNhSaR0O9lTlngytpcCO9ndkcUe1Fh87UrmjWouPLVd/y8kd1Zobr9TccmJDrMdH6WysNMUSAH8BMB3Af9Tqog9DxVEUYgA8DWDDWPx1FIUlAT97jRAsJwSrAOwA4NvV9CoACYRgOYBTAPyKojCHT/vb727n7RcSCwDKU5lvXYeL3V57EMmUHbFzz6T1c+Xlws0nli/3gjXn4H+jN+LU4U9R8dRN8LjdrLFVH7+ENZavUTVvI/IXrBLEHcoXjl8Mrfn4fbyJyamY99sP0JSwDKvK70HlRy/BYbdhgbUK3VnfY9yVVQ6tw8GeqtyTobUU2NH+jizuqNbiY0cqd1Rr8bHl6m85uaNac+OVmltObCajvBeMwkytLlIAmA9gOYAfALgZANFqd8awklM4HcAWQnDx2L83AwAheJDh99cBuIkQbBj7/+vh3e00Hd6dUdcTAgNd7Nq1a0lFRQWP7CLfSl8pxvqjf4f+17XInpkvd3Mkt5IXinB6z4s4NO1ynPabl2gvZoz6frifOg2GmOko+FMJYuPiZWhpZJplxIiuJzag0NGMyplXY/3x11H9vaex+qIb5G5a1KIWtahFLWpRi1rUJDSKoioJIWsDv+d8p1CtLlKr1UUXq9VFf1Cri15Wq4sqAZgBNADYA+8FIQCwV4T02iwA3eP+3TP2XUDj8RuKwlF47xT+duzrtwFYAPQB6ALwKNMFYSjzbeXKxy8kFgAq11UKxk7SlqKbUgddELLFs/Fy4eYTK4S74YYGrP/FYyhR3YR1g++i7NlbQTyeoNi2f90BJTFDcfnTEy4IhXLzjQ3FLRSbi9YpaRnIveNDtMfNx/rjr8NJYlC47ge8sKXWmm9sJHNPptZiYkf7O7K4o1qLjx2p3FGtxceWq7/l5I5qzY1Xam45sZksNpwfqdVFT8FbnH45gIyxr6lxP+kGcDjg08arRTRGCJ4G8DRF4XoAfwbwcwCnAXADUAPIBPA1ReELQjDhnuk+at9tAG6zxFtwIOUA4rLjULCtAMmLk9Fyu7ceS8ZZGfC4PKhaXwUAiEmJwer9q9F4cyMsdRYA8NY3+tiA/n/1AwDy785HvCoe1iYrKtZWIOvCLOTfl4+ac2sAAHHZcVj5yUoQF0HFWu8dymXvLIPudR10b3pfAJ1TPAfOIaffn31ZNtS3q3FkwxEAQEJuAgCg9vJa2Hu8xcdXfLwC2me00L/v3Wxl4BMdCqy1qDadgeNrK5BzdQ5yrstB3RV1AID6a+ux9I2lOHzJYTj1TgDAqn2rYOuy+XkLHy+Eo8+Brke8tUtm3jQTczRz/P6UZSlY/PJiVJ9TDbfF+8jmmtI1aCtqg/FrIwBgwTMLYG20+guBandpoTxd6X+JV3mqEgueXYDKdZWw1FlQua4Spxw6BS0bW2AqNwEAFr20CKYSk79Pc+/MDdJp0cuLUHlaFWJxI/53phHfx9vYt9mNlC9vBwUFlry+BF8/+ALOsn+Oz/XXYGXDXBiHjGi7y3s42nvtcJldQTrVX1uP0bZRAPDWnArQSZGo8LeLTqfl7y6H2+z291mgTnMfmAvn4HdaB+qUVOgt+RmoU9f9XTB87v1bh/GAMUinrA1ZaLjOO/gbb26k1cnW8Z3W43WyJTwE14X3wpY4DeT8dgDtE3QibgIqhmLUSbtLCwDo390fpNPcR+aian0VrI1WVJ9THTSelry+BIaPDf4+9Y0nn05ZF2ZhwXML/O2m02nZO8vQtaMrSKf2e72nAIfWW6cwUKfx48kx4AjSyWPz+NtFp9PSN5bCqf9Oy0CdCh8vhGPA4fcH6pSyLAUAgnRqv7vdP55MZaYJ40l9m3rCeGrZ2EKr02j7qJ83UKfcO3OR+9tcv3+8TgD9ec+nk++8p9ujC9LJd96zNlpx+JLDtDrpXtf5+zRQp+zLslG4s9DfLjqdVny8Ah1bO4J06tjW4dWx31unMFAn33iqWFtBr1Ofw98uOp0Wv7wY9m67v22BOi14ZgEcuu+0DtTJ92gS23gy15qDdBo/ntqK2mh1Gj36ndaBOuXfnQ/1bWq/n2l+otPJN570e/VBOvnOe9ZGK2ovr6XVSfuM1t+ngTrlXJ2DuQ/O9bcrUCfAO57aN7cH6eQ77zl0DlhbrUE6+cZTxdoKWp2sjVZ/u+h0WvDsAoy2evuU6bzn6P9Oa7r5ie86wjee2je30+rkaxedTnOK52DmTTNZ1xFMOvnGk+EzQ5BOvvOetdHKuI7our/L36dSrCOcA946hXTrCOImvNcRcx+Z6/fT6eSxeesUMumUdWEW73WEbzx1bO2g1cnabGVdR+RcncO6jmDSKZx1hLXRyriOaL+73d9ngTqpb1Mj/558f7v4rCOcg946hXTrCAC81xH59+X7/XQ6eRwe1nVE9mXZvNcRvvHUtaOL1zoi+7Js1nXEkleXCFpH0Brdi4aBn7ENZDwq1SarSrWpXKXa9IJKtelOlWrT2SrVpoxwMOg+ADkdIJ+O+/dmtg1jAKIAyPDY/z8NkBvH+V4EyNVMsWwbzdT/rJ71hUw2v5BYQgipOK1CEHZrzTeEaJSk/L1nOMWz8YbLzTdWCPd4n8ftJqVP/YIQjZIcfP4O4nG7SeVNJaRfU0CObV1G7LZRybj5+IVqzdcfKufanx3hjT1ZWp9I3HJqPRXGtdjcUva3nNxRrcXHjlTuqNbiY8vV33JyR7Xmxis1t5zYYNhoJqx3CtXqomvhvfvXrNXu9IT6fbhGUYgF0ALgfAC9AMoBXE8I6sf9Zj4haB37/8sAaAjBWorCPQAWEYJbKAopY7HXEoIjdFwn6juFpa9tx/qWR9H/f5WYmVcYOuAEM+LxoOzpW7Bu8F2U5v4fKOsA1g6+j6M/fhcL1pwrd/OiFrWoRS1qUYta1KIWtYgxQe8UarU739BqdzaKeUEIAITABeAOeGsdNgJ4kxDUUxS2URR+NPazOygK9RSFGgB3wfvoKODdtTSVolAP7wXhS0wXhKGsrYj9SVc2v5BYwPsYmBDshN5SaKkZtBeEbPFsvOFy840Vwh3ooxQKnHr7iyjL/CHW9/wT6wx7UT7zWsYLQjG5ufqFas3Xf6JofaJwR7UWt12huKXsbzm5o1qLjx2p3FGtxceWq7/l5I5qzY1Xam45sRmN7vbhifhhe3zUrrOz3mZl8wuJJYS9Rkqo2NF+KzFoZpGynddwjg9Vm0XKPhHCzeRzu1yk9ImbSNO2tcRqNk0qd7h+IVoL8Z9oWk917qjW4rYrFLeU/S0nd1Rr8bEjlTuqtfjYcvW3nNxRrbnxSs0tJzakrFM41a3/pX7efiGxoSxUbO0/vkImRkBmf2/SuYX0iRBuJp8iJgbr7tyNpIT/ICklbVK5w/ULiY1Ubjm0nurcUa25Y5+M/S0nd1Trk4c7qvXJwx3VOrK45cRmNLorxRPxw3ancOTICOsVNZtfSKxQ7G+e206IRkl6jjaIjj1V+2SqckexJxdbTu4o9omDLSd3FHtyseXkjmKfONhyckexJxdbTu5QsYjeKWQ2c5WZt19ILACYSky8Y+P6D+E4pkE9ZyHneDbecLiF9IkQbin7W05uodhRracOd1RrcdsVilvK/paTO6q1+NiRyh3VWnxsufpbTu6o1tx4peaWE5vJTviLwn3Uvsv2Uft22bpsMB4wQr9XD90eHUxlJvQ80QNbtw2DHw/C4/Kg8eZGAEDjjWP/vbkRHpcHgx8PwtZtQ88TPTCVmaDbo4N+rx7GA0aMVI/A2mpFx9YOuMwu/0urvsKRvjo3bUVtcAw40LWjC+ZaM/p398PwmQFdj3ahf3c/zLVmdO3ogmPA4X9B1Bfrw2rZ2AKX2YWOrR2wNJsx13MYnfErMPiBISin9s3tiMuJY8zpmOYYY07aXVoQD2HMqeGGBm8tGIac7Fo7Y04NNzRAu0tLm5O11QrtLi0cegetTqFyat/cDsSAMSdrqxUdW5hzAgD9e3ranAyfGWCuMzPmNF4vupyMB4zeWjg0OXlcHhz/93HGnGzdNpgOmZhz2tqBmIwYxpyO/eUYY079u/vhtrkZc/JpzZTTaMcoY06NNzdCu0vLmFPPEz1wDbt45dRW1AZFkoIxJ3OtGR3bOhhzAgDdWzranPR79bA2WxlzAoD+V/oZczKVmdD5UCdtTi6zC7q3dIw5OQYcMH5tZMypa0cXFEkKxpza72tnzEm3RwfXsIsxp8YbGxGvimfMydpsZcypZWPLhHEdmFPXji64bW5eObVsbEFMRgxjTqYyEzr+2sErJ90eHUY7RhlzGj+u6XIy15rR+WAnbU6OAQf07+kZc3KZXRj6cogxp54neoAY0OYEAO33tjPmpN+rh0PvYMzJN66ZcjLXmRlzaitqo9V6/DmCeAivnBpvbkRcThxjTsYDRnTe38krp/7d/bBr7Yw50WkdeN7rvL9TknVEqDm3fXM7Y07hzLnj+yUwp5HqEcac2je3o/f5Xt5zrpB1hOEzAzof7OSVU6h1RKg5t+epHtZ1xNCXQ7zXEaHmXJ/WfOfceFU873VE77O9vOdcIesI/V49Oh/q5JVTqHVEqDm39+le1nWE8Wsj73VEqDk31Doi1Jw7fgxwXUcwGt3twxPxw/b46OCng6y3Wdn8QmIJYX/JlS22s7maEI2SlL75KK/4UC/XStknQril7G85uYViR7WeOtxRrcVtVyhuKftbTu6o1uJjRyp3VGvxseXqbzm5o1pz45WaW05sRB8fZTZHn4O3X0gsAOTemcsrtv/I/wAAqpXn84pn4w0VG8ovJGeh2FOVWyh2VOupwx3VWtx2heKWsr/l5I5qLT52pHJHtRYfW67+lpM7qjU3Xqm55cRmsuhFIYDUNam8/UJiASB5cTKvWEXXt9AjHXmFK3jFs/GGig3lF5KzUOypyi0UO6r11OGOai1uu0JxS9nfcnJHtRYfO1K5o1qLjy1Xf8vJHdWaG6/U3HJiM1n0ohCA4WMDb7+QWABouZ25cCZb7MyRehwjS0ApmCVki2fjDRUbyi8kZ6HYU5VbKHZU66nDHdVa3HaF4payv+XkjmotPnakcke1Fh9brv6WkzuqNTdeqbnlxGY0umdKT8TPiVS8fnhIT4hGSb59/o+8uaMFUiOLe6oWZ52q/S0nd1RrcdsVijta0Jy7P6p1ZHFHtRYfO1ILmke1Fh87UrWOFq+PUOt6gGUnnhB+IbEAkHFWBufY7vpSAIC9dR5vbjbeULGh/EJyFoo9VbmFYke1njrcUa3FbVcobin7W07uqNbiY0cqd1Rr8bHl6m85uaNac+OVmltObCajvBeMJ76tXbuWVFRUyN2MIPO4PFDEcrs2L/33VqxvfRz6X9che2bepPGKZVHuk4M3yh3VOsp94vFGuaNan+jcJ2POJyu3nDnLaRRFVRJC1gZ+f8L3RDh1CsuWlrHWFypbWsZYX6hyXSVrncKSghIAzPWFypaUMdYX8sUG1uJBcxV0yELnj3Ws9YXqr69nzKkkr4S1ZtKRHx1hrS/UcEMDY07V51ez1kyqWl/FWl+o5sIaXjm1b25H7RW1rPWFSvNLWesLHVp8iLG+UOX3KlnrFPr0YqrFU7aojLG+UOm8UsacbN02VKytYK0vVHdNHWNOJbNKWGsmHbn0CGt9oYYbGhhzqj63mrVmUtX6Ktb6QocvOcwrp7aiNtRdWcdaX6h0TilrfaFDCw4x1heqOrOKtU6hT2um+kKHFhxirC90aMEhxpwcAw6Ury5nrS9Ud2UdY04HVQdZayYdvuQwa53ChhsaGHOqOrOKtWbS+HFNVyvuyKVHeOXUsrEFddfUsdYpLJ1byisn3R4dqs+tZq1TGHgeDjzvHZp/iLGm36HF9Fr7xlP5inLWOoW1V9Qy1vQ7OPMga53CmgtrWOsU+j50OVV+r5K1TiGd1uPPEUd+dIRXTo03N6L++nrWOoWH5h3ilVP/7n5Un1/NWqeQac71HXuH5h2SZB0Ras49OOMgY07hzLnj+yUwp8p1lazriMp1lbznXCHrCMNnBhyaf4hXTqHWEaHm3Io1FazriPIV5bzXEaHmXJ/WfOfchhsaeK8jKk+t5D3nCllH6PfqcWjBIV45hVpHhJpzK0+pZF1HlK8u572OCDXnhlpHhJpzx48BrusIRqN7pvRE/LC9UyinhXqemc46ti4l1Q9fPOm8YlmU++TgjXKfPLxR7pOHN8p98vCerNwnY84nK7ecOctpiL5TyGy+K2w+fiGxABCTEsMp1moeRp67B6PTlgniZuMNFRvKLyRnodhTlVsodlTrqcMd1VrcdoXilrK/5eSOai0+dqRyR7UWH1uu/paTO6o1N16pueXEZrLoO4VTzJoOfYZFH1+FmjOexaoLr5e7OVGLWtSiFrWoRS1qUYta1KaInbTvFIZjvmeo+fiFxALwP1scbqyx3Xthq15yuiBuNt5QsaH8QnIWij1VuYViR7WeOtxRrcVtVyhuKftbTu6o1uJjRyp3VGvxseXqbzm5o1pz45WaW05sRqN7pvRE/LC9U+gccbI+e8vmFxJLCPvzzHSxZTuvIXpNHvG43YK4Qz1HLWWfCOGWsr/l5BaKHdV66nBHtRa3XaG4pexvObmjWouPHancUa3Fx5arv+XkjmrNjVdqbjmxEX2nkNl6Huvh7RcSG8roYqeZGtGbuACUQjHp3OH6hfAKxZ6q3EKxo1pPHe6o1tyxT8b+lpM7qvXJwx3V+uThjmodWdxyYjMa3ZXiifhhu1NoabGwXlGz+YXEcsUetZqJsziDHNz1W9GxxfRPVWw5uaPYk4stJ3cU+8TBlpM7ij252HJyR7FPHGw5uaPYk4stJ3eoWJysdwrDqVPY+ttW1vpCrb9tZawv5KtNwlSLp/EGLxZTLZ6enT2M9YV8sT6sss0fIJbywNY9z8/JVl9o6Ishxpxabm9hrZmke0PHWl/I+JWRMafep3tZayYZPjaw1hfSPq/llVP75nYM/GeAtb5Q22/bWOsLtWxsYawv1PlAJ2udQp9eTLV4und0M9YXavp5E2NOtm4bjv3lGGt9IcPHBsacmm9tZq2ZdPzV46z1hYxfGRlz6vlbD2vNJMPHBtb6Qn0v9vHKqa2oDfq9etb6Qkf/cJS1vlDzL5sZ6wt17+hmrVPIlpOpzITO+zsZ6ws1/7KZMSfHgAPtf2pnrS+k36tnzKnplibWmkl9L/ax5mT8ysiYU/eObtaaSePHdWBOXTu6cPzV47xy8mGz1Sls/1M7r5x0e3To+VsPa53CwPNw4HmvY2sHY00/tvOMy+zC0buOstYpHPjPAPOxdwNzTvq9emif17LWKTR+ZWTMqfOBTtY6hXRajz9H6N7Q8cvp5kYMfTHEWqfw2F+O8cqpf3c/ep/uZa1TGKh14Hnv2F+OSbKOCDnnXsecUzhz7vh+Ccwp1Dpi8INB3nOukHWE4TPvmOeTU6h1RKg59/hrx1nXEUfvOsp7HRFyzr2OOadw5lzjV0be6wj9u3rec66QdYR+rx6d93fyyinUOiLUnDvw1gDrOqL9T+281xHhzLlMOYUz544fA1zXEYxGd6V4In7Y7hQO7R9ivaJm8wuJJYT9eebA2NI9OwjRKEnvsSbB3KGeo5ayT4RwS9nfcnILxY5qPXW4o1pzx47U/paTO6q1+NiRyh3VWnxsufpbTu6o1tx4peaWExsn653CcMxldPH2C4kNZYGxVP9hDCMFqvz5k87NxS+EVyj2VOUWih3VeupwR7Xmjn0y9rec3FGtTx7uqNYnD3dU68jilhObyaIXhQA8ox7efiGxAJB/d37YsZnDjehOmA9KoRDMzcYrFFtIzkKxpyq3UOyo1lOHO6q1uO0KxS1lf8vJHdVafOxI5Y5qLT62XP0tJ3dUa268UnPLic1k0YtCAIkFibz9QmIBIF4VH1asw27DbFcHzJlLReFm4xWKLSRnodhTlVsodlTrqcMd1VrcdoXilrK/5eSOai0+dqRyR7UWH1uu/paTO6o1N16pueXEZrLoRSEAU6mJt19ILAC03dUWVmx3cxXiKRdi81aLws3GKxRbSM5Csacqt1DsqNZThzuqtbjtCsUtZX/LyR3VWnzsSOWOai0+tlz9LSd3VGtuvFJzy4nNZNGLQgDZV2Tz9guJDWXjYw1t5QCAGQvXTzo3V78QXqHYU5VbKHZU66nDHdWaO/bJ2N9ycke1Pnm4o1qfPNxRrSOLW05sJoteFALQPqPl7RcSCwBZF2aFFevR1sBMkjCrYIko3Gy8QrGF5CwUe6pyC8WOaj11uKNai9uuUNxS9rec3FGtxceOVO6o1uJjy9XfcnJHtebGKzW3nNhMdsJfFIZTp5B4CGvtEOIhjPWFEvITWOsU2jptAJjrC6WtT2OsL+SLbbihARnGBnR4CuAZJf56KAn5Caz1hWZrZjPm5HF4WGsmzbhpBmt9oTnb5zDmpDxbyVozKf++fNb6QhkXZfDKqX1zO2beNpO1vhAUYK0v5La6GesLJc5PZK1T6NOLqRZPyqoUxvpC9l47Y062bhvicuJY6wvl3ZPHmJNrxMVaM2n61dNZ6wvN2T6HMae009JYaybl35fPWosn67IsXjm1FbVB/Vs1a30hRbKCtb6Q0+BkrC+UvDyZtU6hrcvGmJOpzISkRUmM9YWcBidjTo4BB2LSY1jrC6l/q2bMyTHgYK2ZlHVZFmvNpDnb5zDmlLw8mbVm0vhxTVcrbvrV03nl1LKxBXn35LHWKYzNiuWVk26PDmmnpbHWKRx/Hg7MyVxrRsKcBMaafm6rmzEnl9kFKp5irVM487aZjLWtbF021jqFGRdlsNYpnLN9DmNOifMTWesU0mk9/hwx46YZvHJqvLkRszWzWesUxqvieeXUv7sfyrOVrHUKA7UOPO/Fq+IlWUeEmnNHj40y5hTOnDu+XwJzCrWOyLsnj/ecK2QdYfjMgIQ5CbxyCrWOCDXnzvj5DNZ1BBVP8V5HhJpzfVrznXPnbJ/Dex0x63ezeM+5QtYR+r16JC1K4pVTqHVEqDlXdZuKdR0Rkx7Dex0Ras4NtY4INeeOHwNc1xGMRlen4kT8sNUpbPh5A2s9Dza/kFhC2Guk+GKdDjuxFmeTkqdvFY07VG0WKftECLeU/S0nt1DsqNZThzuqtbjtCsUtZX/LyR3VWnzsSOWOai0+tlz9LSd3VGtuvFJzy4kNhjqFsl+sTdaH7aJQTgt1QBJCyLGGckI0SlL27tOTyiuVRblPDt4o98nDG+U+eXij3CcP78nKfTLmfLJyy5mznMZ0UXjCPz4ajvluS/PxC4kFgLjsuJCxAy1lAICcBetE42bjFYotJGeh2FOVWyh2VOupwx3VWtx2heKWsr/l5I5qLT52pHJHtRYfW67+lpM7qjU3Xqm55cRmMsp7wSifURQuAfB3ADEAXiAEDwX4fw3gNwDcAMwAbiMEDRSFnwG4e9xPVwBYQwhq6HjWrl1LKioqJMhAeit95jasPP4O4v/Sh5jYWLmbE7WoRS1qUYta1KIWtahFbQoaRVGVhJC1gd/LeqeQohAD4GkAGwAsAXAdRWFJwM9eIwTLCcEqADsAPA4AhODfhGDV2Pc3AjjGdEEYynwvffLxC4kFgPpr60PGKofq0Rk3L+iCUAg3G69QbCE5C8WeqtxCsaNaTx3uqNbitisUt5T9LSd3VGvxsSOVO6q1+Nhy9bec3FGtufFKzS0nNqPRPVM6WR+AnA6QT8f9ezNANrP8/jqAfEzz/QMAuZ+Ni+2dQrfTzfrsLZtfSCwh7M8zu51u4na5iLk4h5Q+ebOo3KGeo5ayT4RwS9nfcnILxY5qPXW4o1qL265Q3FL2t5zcUa3Fx45U7qjW4mPL1d9ycke15sYrNbec2IjQdwpnAege9++ese8mGEXhNxSFo/DeKfwtDc41AF7n24iOv3Tw9guJDWUdf+lAb3sdUigbFOpVk87N1y+EVyj2VOUWih3VeupwR7Xmjn0y9rec3FGtTx7uqNYnD3dU68jilhObyabEC2qE4GkAT1MUrgfwZwA/9/koCusAWAlBHV3sPmrfbQBus8RbcCDlAOKy41CwrQDJi5PRcru3HkvqqlR4XB5Ura8CAMSkxGD1/tVovLkRljoLiINg5i9mwvCxAf3/6gcA5N+dj3hVPAY/HIThcwOyLsxC/n35qDm3BoD35dWVn6yEpd6CirXedxmXvbMMutd10L2pAwDMKZ6D3KJcvz/7smyob1fjyIYjXoxpcXCeWYY8AEMvTkPFAxVY8fEKaJ/RQv++HsRBkHFeBjw2Dzq2dQAAcq7OQc51Oai7og7EQTB6bBRL31iKw5cchlPvrdWyat8qKNcp/byFjxfC0edA1yPe2iUzb5qJzA2Zfn/KshQsfnkxqs+phtvirbu19K2laCtqg/FrIwBgwTMLYG20oufJHhAHQWJBIpSnK9F0SxMAQHmqEgueXYDKdZUgDoLKdZU45dApaNnYAlO5CQCw6KVFMJWY/H2ae2fuBJ0yzsrArN/O8rcrUCcAmPf4PPQ80ROkU9td3hozynVKuMyuIJ3qr63HaNsoiINAfbs6SCdFosLfrkCdEnITsPzd5RipGvG3bbxOADD3gblQb1T7/eN1AoAEdQIABOnUdX8XDJ8bQBwEWRuygnTK2pCFhusaQBwE9j57kE5rStcgdVWqn3e8TgCgvk2NjO9n+P0TdHITUDEUlr69lFYn7S4tiIMgeVFykE5zH5mLqvVVIA6C6nOqg3Ra8voSGD42+Ps0UKesC7Mw4+cz/O0K1AkA5j89H107uoJ0ar+33duOszPgGHAE6VR7eS3sPXYQB0HuXblBOnlsHn+7AnVKKkzC0jeWYrhk2N+28Tr5xtPMW2b6/eN1AoDEgkQACNKp/e52GL82gjgIsq/IDtLJN56Ig8BldAXpdMqhU5C8ONnPO14nAMi9MxfK73037sfr5BtPi19dTKtT/7/6QRwEqatSg3TynfeIg+DwJYeDdPKd93x9GqhT9mXZmH7NdH+7AnUCgIUvLETH1o4gnXznvayLsmDrtgXpdPiSw3BoHTB8bqDVydHn8LcrUCffeW/oqyF/28br5BtPM67/7jgdrxMAJC9OBoAgnXzjiTgIcq7PCdLJN56Ig8Bj8wTptHr/aiQWJPp5x+sEeM97qad8N+7p5qeF/1hIq5PuTR2Ig0B5ujJIJ995jzgIai+vDdLJd97z9WmgTjlX52Da5dP87Rqvk++8t/jVxWjf3B6kk++8N+2yabC2WoN0qj6nGq4hFwyfG2h1sjZa/e0K1Ml33hv8xOsP1Mk3nqZf+d1xGjg/CVlHtN3VBuLw7u9At45IUCewriOSlyUzriMSchMw/8n5tDqFu46ov7aedh3RdX+Xv0/FXkcAwPQrp8Nca6ZdR3hGPRj63xCvdcTcR+b6/ZO9jmi/tx3EQaBIVNCuI+KmxbGuIxILExnXEUmFSZj3yDxB64jGmxtp1xHtd7dj6MshGD43iL6OAIAZ18+AqczEuI4YPjjMax2Rf1++3z/Z64iObR0gDoLYzFhe64j43HjGdUTKshQUbC/gvY5gNLrbh5P14fH4qAIgwwHf7QTIvaG42B4f7f57N+ttVja/kFhCCOl8uJM1tuTZjcRenEUcdpuo3Gy8QrGF5CwUe6pyC8WOaj11uKNai9uuUNxS9rec3FGtxceOVO6o1uJjy9XfcnJHtebGKzW3nNiI0MdHywHMpygUUBTiAVwLYO/4H1AU5o/756UAWsf5FACuBvCGkEYo1yt5+4XEAvD/dYIpNnWoHp1xBYiLTxCVm41XKLaQnIViT1VuodhRracOd1RrcdsVilvK/paTO6q1+NiRyh3VWnxsufpbTu6o1tx4peaWE5vJZL0oJAQuAHcA+BRAI4A3CUE9RWEbReFHYz+7g6JQT1GoAXAXxj06CuBsAN2EoF1IO2zHbLz9QmJD2ehRK/LsrTAoF086t5R9IoRbypzl5BaKHdV66nBHteaOfTL2t5zcUa1PHu6o1icPd1TryOKWE5vJZH+nkBB8BOCjgO+Kx/3/71hi9wFYL7QNiiT2a2M2v5BYwPv8MpMZ3B1YDAugWik6NxuvUGwhOQvFnqrcQrGjWk8d7qjW4rYrFLeU/S0nd1Rr8bEjlTuqtfjYcvW3nNxRrbnxSs0tJzZjHK+oE8xiM9ivjdn8QmIBQJHILIFx9DAAIHPeqaJzs/EKxRaSs1DsqcotFDuq9dThjmotbrtCcUvZ33JyR7UWHztSuaNai48tV3/LyR3Vmhuv1NxyYjNZ9KIQgLXJytsvJBaAf4cjOnP018BJYpC/eK3o3Gy8QrGF5CwUe6pyC8WOaj11uKNai9uuUNxS9rec3FGtxceOVO6o1uJjy9XfcnJHtebGKzW3nNhMdsJfFO6j9l22j9q3y9Zlg/GAEfq9euj26GAqM6HniR7Yum0YqRyBx+VB482NAIDGG8f+e3MjPC4PRipHYOu2oeeJHpjKTNDt0UG/Vw/jASMcxx2wtlrRsbUDLrMLLRu92+k23ODdNnbww0EAQFtRGxwDDnTt6IK51oz+3f0wfGaAy+RC/+5+mGvN6NrRBceAA21F3u1000Yb0RWbj/ZbOwEALRtb4DK70LG1A9ZWKxzHHYw5tW9uR/pZ6Yw52XvtjDlpd2mRWJDImFPDDQ3IOC+DMScqjmLMaTwGXU7aXVooUhW8cmrf3I6kRUmMOVlbrXBoHYw5AcBwyTBtTobPDHAOOVlz8mlNl5PxgBEuo4s2J4/LA8MnBsacbN022LptjDl1bO1A2mlpjDnZu+2MOfXv7ke8Op5Vp4zzMhhzIh7CmBObTr5jLzYrlldObUVtSFmZwpiTudYMR7+D9dgz7jPS5qTfq4fb6mbNyac1XU6mMhOcBidtTi6zC8Z9RsacHAMOjLaNMubUtaMLKStTGHOyddoYc9Lt0SE2K5ZVp4zzMhhzclvdjDmx6eQ79uLV8bxyatnYgrTT0hhzMpWZ4NA5eOWk26MD8RDWnHxa0+VkrjXDOeikzckx4MBwyTBjTi6zC5Y6C2NOPU/0IGlREuN4snXYGHPS79VDkapg1SnjvAzmnIaYcwrnXJ5YkMgrp8abG5F+VjpjTsYDRjgHnLxy6t/dDyqOYs0pUOvA855zwCnJOiLUnOt7P4jvnDu+X7iuIwghvOdcIesIw2cGOAeZ51y2nEKtI0LNuR6Hh3UdYamz8F5HhJpzfVrznXMzzsvgvY4gbsJ7zhWyjtDv1cNpcPLKKdQ6ItScS5yEdR0x2jbKex0Ras4NtY4INeeOHwNc1xGMRrcl6Yn4YStJcWzLMdatW9n8QmLZ/B63mwwW55JDO6+VhFuqdk9lbDm5o9iTiy0ndxT7xMGWkzuKPbnYcnJHsU8cbDm5o9iTiy0nd6hYMJSkoLy+E9/Wrl1LKioqaH0uswuxqczP37L5hcQCgGPAgfjp8UHfH+85ihkvrMGhRX/Cums3i87NxCsGNt+cxcCeqtxCsaNaTx3uqNbitisUt5T9LSd3VGvxsSOVO6q1+Nhy9bec3FGtufFKzS0nNkVRlYSQoHfTTvjHR8Ox9rvZnylm8wuJBYAjG47Qft/XeAgAkD6X/n1CodxMvGJg881ZDOypyi0UO6r11OGOai1uu0JxS9nfcnJHtRYfO1K5o1qLjy1Xf8vJHdWaG6/U3HJiMxrd7cMT8cP2+KicVn5KOe33Jf8oIq7idGIZMU4q72RYlPvk4I1ynzy8Ue6ThzfKffLwnqzcJ2POJyu3nDnLaWB4fDR6pxDfvaDJxy8kFgASchNov08ZqMYxz2wkp6ZLws3EKwY235zFwJ6q3EKxo1pPHe6o1uK2KxS3lP0tJ3dUa/GxI5U7qrX42HL1t5zcUa258UrNLSc2k0XfKYxA87jdMG/PReO0C7Huzn/J3ZyoRS1qUYta1KIWtahFLWongEXfKWQxOa/0ay+vDfquu/UwlLBiuG2+IGw2Px2vWNh8chYLe6pyS/kXo6jWkcUd1VrcdoXilvruUVRrcbEjtb/l5I5qLT62XP0tJ3dUa268UnNH4p3CE/6iMJw6hYl5iaz1hRLzEhnrC2WcncFar8tnTLV4LA2W4PpCL30EAEi2rpiAFVjjJePsDNb6QoteXsSYk+mQibVm0hzNHNb6QkteXcKY08ybZrLWTLL32Blz0u7SQn2bmldO7ZvbMffBuaz1hUbKRljrC8VPj2esL5R1YRZrvS6fMdXisdRZGOsLUbEUY062bhuUpylZ6wsteG4BY06mEhNrzaT8e/JZ6wsteXUJY045V+ew1kyy99hZ6wvl/jaXV05tRW0o3FnIWl9opGKEtb5QbFosY32h7MuyWesUUhTFmJOp7P/bO/M4OYry/78r2dwnAUIy5CAhkAMItxxyiIiCCAoqKqIgKl9B9MuCoIjIJYogDnKIIKcXlwIigsCXHwHkyEEuckASNmETmpAsySZsDjZLnt8fVZN0Zqd7dmd6pmemn8/rNa85uut511NPT3VVd3XVWlpmtwSuL1TXry7Qp9aVrfSd2Dd0faEx6TGBPq15aU3omknDfjgsdJ3CCX+ZEOjTDifsELpmkv9/ne1T47WNjPjxiIJ8WnD2Anb/w+6h6xR+MP2Dgnxa8cAKBp8yOHSdwoxy+dTyegsts1oC1/TLzGqXy6e2ljZ679Y7dJ3C0b8aHbim35r/rgldpzB1Vip0ncIJf5kQ6NOgYwaFrlOYK9b+OmKXy3YpyKf5Z8xn3D3jQtcpbJnRUpBPy+9dzpBvDgldpzA71tn1XsuMlpK0I/Kdc9e8uCbQp46cc/3l0tl2xMalGws+5xbTjlj19CpaZrUU5FO+dkS+c+6Ghg2h7Yjeu/UuuB2R75ybiXWh59wJf5lQcDti49sbCz7nFtOOaHqsiZbZLQX5lK8dke+cu3HxxtB2RN+JfQtuR+Q75+ZrR+Q75/r/A51tRwQq14OGtfgKm2hm4XkLQx/IDNteTFqR3A+5vvq702TNZUPkzfPeKMp22PZ8D9eWskyKYZeivCuBXaxtjXX1sDXW0eYrH7uU5R0nW2Mdve1KZWuso7cdV3nHydZYd45banactgmYaCb2zlq5XmGdwg9XfBhaeGHbi0kbtH3RlXvLrF8dVRLbUaStVdtxstV2eW3HyVbbtWM7TrbaLq/tONlqu3Zsx8lW2+W1HSc7X9qgTmHNDx/tiJbfvbzg7cWkBfB+723zfd0HzezStoR1O+5btO2w7dncKG131ucobVcrW2Mdbb4qma2xjjZf+dilLO842Rrr6G1XKltjHb3tuMo7TrbGunPcUrPjtB0k7RQCg44bVPD2YtICNP2raZvvS2a/RFcj9B51UNG2w7Znc6O03Vmfo7RdrWyNdbT5qmS2xjrafOVjl7K842RrrKO3XalsjXX0tuMq7zjZGuvOcUvNjtN2kLRTCLRMbyl4ezFpc2ntopcBGDnxiKJtd5Ydle1iuMXarla2xrrztquVrbHuvO0klnecbI11ctga6+SwNdaVxY7TdpC0Uwh0H9q94O3FpAUY/cvR23zv+d4MlpoUA3cYUrTtsO3Z3Chtd9bnKG1XK1tjHW2+KpmtsY42X/nYpSzvONka6+htVypbYx297bjKO062xrpz3FKz47QdJO0UxqzNGzdv+SybNzN8/VyW99+zrNxyS9nJ4Co7OVxlJ4er7ORwk8pOos9JZcfpcyWq5juFHVmncNmNy0LXF/Lvl73Gy4r7VoSuU7jk8iVA8PpCi360aMvaIbOueYkdaGbte2O3SRu0Fs+K+1aEri/04bIPA31683/eDF0zqWVGS+j6Qq3vtgb6tPqZ1aFrJi25ckno+kLNkwrzqeHiBtbNWRe6vtCC7y0IXV9o6fVLA9cXWvnQytB1CjPxClqLZ9H5iwLXF1py5ZJAnzYu3cjyPy0PX19ocfD6Qm9+983QNZPWTlkbur5Q67utgT69/8T7oWsmLblySej6QmteWlOQT4vqF7H+zfWh6wstOGdB6PpCjdc0Bq4v1PRoU+g6hZlYB60vtPC8hYHrCzVe0xjoU+vKVt69893Q9YXWv7k+0Kc3vv1G6JpJa15aE7pOYeu7rYE+NT3aFLpmkv9/nWutuLVT1hbk04KzF7Bx8cbQdQoXnLugIJ9WPLCC9594P3Sdwux6OLveW/i/CwPX9Ft6/dJAn9pa2vBu9ULXKVw3Z13gmn5vnPlG6DqFzZOCfcr8r4N8WvnQytB1CnPF2l9HtMxoKcinzNqmYesULvzBwoJ8Wn7vclY/szp0ncJ859yFP1hYknZEvnPuG996I9Cnjpxz/eXS2XbE4isWF3zOLaYdserpVSz834UF+ZSvHZHvnNvws4bQdoR3q1dwOyLfOTcT60LPua3vthbcjlh82eKCz7nFtCOaHmti4XkLC/IpXzsi3zl38aWLQ9sR7975bsHtiHzn3HztiHznXP9/oLPtiEDlmpK0Fl9hS1J8MPuD0Klbw7YXk1Zk2zVSpj3+R5HL+suCGS9EYjtse761WUpZJsWwoyzvSmJrrKPNVyWzNdbR5isfu5TlHSdbYx297Upla6yjtx1XecfJ1lh3jltqdpy20SUpgrXqyVUFby8mLcDgUwZv+bypcSobpRu7TPhYJLbDtvu5UdvujM9R265WtsY62nxVMltjHW2+8rFLWd5xsjXW0duuVLbGOnrbcZV3nGyNdee4pWbHaTtQuXqKtfiq1MXrNzRu2PJ5/lUHybxfHBKZ7bDtfm7Utjvjc9S2q5WtsY42X5XM1lhHm6987FIvTqyxjtZ2pZZ3nGyNdfS24yrvONka685xS83WxesrVI2/DBlfm2d7MWkB5pw0B4APN65n1KZFrBk0MTLbYdsz3FLY7qjPpbBdrWyNdbT5qmS2xjrafOVjl7K842RrrKO3XalsjXX0tuMq7zjZGuvOcUvNjtN2oHL1FGvxFXanME5lxjO/+dpzIpf1l9eeuKus3Dik7GRwlZ0crrKTw1V2crhJZSfR56Sy4/Q5TqF3CoOVmbWnkO3FpAXoNaYXAKvefBmAnfc6IjLbYdsz3FLY7qjPpbBdrWyNdbT5qmS2xjrafOVjl7K842RrrKO3XalsjXX0tuMq7zjZGuvOcUvNjtN2kIztMNa+DjjgAJk2bVrc2QjUtN9+kRFrpzP48sVxZ0WlUqlUKpVKpVLVoIwxr4nIAdm/1/ydwo6sUzhljymh6wtN2WNK4PpCrx30Wug6ha+MegUIXl9o2gHTWH7vcoaufZ23Zdw266Fk0gatxfPaQa+Fri8099S5gT69OvrV0DWTZp84O3R9oXmnzQv0acbRM0LXTJp17KzQ9YVmHjOzIJ8aLm7g9ZNeD11f6NUxr4auLzR5/OTA9YVeO/S10HUKM/EKWotn6n5TA9cXenXXVwN92rh0I9MOmBa6vtCcr8wJ9OmVXV4JXTNp9vGzQ9cXmnfavECfZnxiRuiaSbOOnRW6vtCsY2cV5NOi+kXM+dKc0PWFJo+dHLq+0OTdJweuLzT9sOmh6xRmYh20vtDUvacGri80effJgT61rmxl6r5TQ9cXmvOlOYE+vTzi5dA1k2YdOyt0ncJ5p80L9Gn6YdND10zy/6+zfWq8tpHZx88uyKcFZy9gzlfmhK5TOHn85IJ8WvHACmZ8YkboOoXZ9XB2vTdlrymBa/pNHp871pn/09SJU0PXKXz9pNcD1/R7eeeXQ9cpnHnMzNB1CjOvXD69duhroesU5oq1v46YfeLsgnyaf8Z85p46N3Sdwil7TCnIp+X3LmfG0TNC1ynMd86dsseUkrQj8p1zX069HOhTR865/nLpbDti5qdnFnzOLaYdserpVUzZa0pBPuVrR+Q75844ckZoO2LqxKkFtyPynXMzsS70nDvvtHkFtyNmfmpmwefcYtoRTY81MXXvqQX5lK8dke+cO+OTM0LbEVP3nVpwOyLfOTdfOyLfOdf/H+hsOyJQucaU1uKrkp8pfP+9ZSKX9ZdX/nRpWblxSdnJ4Co7OVxlJ4er7ORwk8pOos9JZeszhfpMYTtlrowUsr2YtBk1vv4iAP13OzRS2x1hl8J2MdxibVcrW2PdedvVytZYd952Ess7TrbGOjlsjXVy2BrrymLHaTtI+kwh0NbSRl3fusC0YduLSZvZPvW+H3HgsnvZdFEjvfr0i9R2KfNda7bjZKvt8tqOk622a8d2nGy1XV7bcbLVdu3YjpOttstrO052vrQV+0yhMRxrDG8awyJj+EmO7d8zhteNYaYx/NcYJvi2TTSGV4xhrtunZyF5WHb9soK3F5MWoPHqRvo1zWBJ3ahtOoRR2A7b3nh1+BompSyTYthRlHclsjXW0earktka62jzlY9dyvKOk62xjt52pbI11tHbjqu842RrrDvHLTU7TtuByjWmtFwvkK4gb4GMBukOMgtkQtY+/X2fTwT5j/tcBzIbZG/3fXuQrkGssGcK1y1YFzr2Nmx7MWlFRF7d/xX54Oc7yas3fjNy22Hb842jLmWZFMMutkwqla2xjjZflczWWEebr3zsUpZ3nGyNdfS2K5WtsY7edlzlHSdbY905bqnZcdqmQp8p/BiwSIQGEVqB+4HP+3cQYa3vax8gM97108BsEWa5/d4X4aNCMtH8XHPB24tJC/DBjkvoazbQZfiBkdvOt71UtovhFmu7Wtka687brla2xrrztpNY3nGyNdbJYWusk8PWWFcWO07bQYq7U7gzsNT3fZn7bRsZw/eN4S3gWuCH7ufdATGGp4xhujFcVGgmeo/rXfD2YtIC9DzB3rreacJhkdsO2z7mt2NKZjtf2mLYxZZJpbI11tHmq5LZGuto85WPXcryjpOtsY7edqWyNdbR246rvONka6w7xy01O07bQQp+CrGCJMItwC3GcCrwM+B0bN4PAw4E1gPPGsNrIjzrTzvJTDoLOGtd93W80OcFuu3QjVFXjqL3+N4sOMfOztMj1YP+h/Zn+sHTAejapyv7Pr8v88+Yz7o562hrbmPikxNZ9eQqlv9pOQAjLhxB96HdeePMN6gbWMegYwYx4pIRzPzETAC67dCNvf+zN4svWcxHG+wNzD0f2ZMV961gxYMrANjl57uAN401Xfuw/NQP2XzCElLnpJh93GwATJ1h/1f35/UvvM6Hyz4EYOKTE/F+79H0rybamtvY/fe7s3njZpZcuQSAwacMZvDXBjPnpDm0NbfR74B+7HH/Hsw6dhabmjYBsM+kfVj626UsOt+uuzLmt2NofbeVxutsB3XIN4dgupst2/vs2Yfx94xnxpEz+Gid9WXkJSNZVL+I5hebAdj997uzfv56lt20jLbmNkZcNIL+h/TnjW+9AUD/A/uz+62789pBr7GpaRPddujG/pP3Z8HZC1g71d4MHnf3ONa+spbGaxupG1jHsB8M2yZOAw8fyIAjBjDtgGk54wQw9DtDaZnZ0i5OGV96DOtB3/36tovT3K/OZcOiDbQ1t7HPc/u0i1OXnl1YcM4C6gbWscMJO2wTpx7DerDXo3vx1kVvIW3SLk4Ao385mtXPrN6SD3+cALr26sq+L+7bLk6NVzey6plVtDW3Me6uce3iNOi4Qcz72jzamtsYcNiAdnHa79X9aLymcQvXHyeA1FkpNm/cvGW7P07ykWC6GkZeMjJnnLzbPdqa29jl0l3axWn0daOZfvB02la30WNYj3ZxmnDfBFY9uYplNy6jbmBduzgNOmYQffbpsyXW2XECSJ2dYu2ra9vFqeGnDQD0GtWL3uN7t4tT5v/U1tzGfq/s1y5OmfKoG1jXLk69xvRij/v3YNF5i7ZcUvPHKfN/WvmPlVt88ccJoG5AHfs8u0+7ODVc2EDzi820Nbcx4W8T2sUp839qa25j0DGD2sVp/8n78/ZVb7No9aJ2cQIY9oNhdr0lly9/nDL/p+EXDM8Zp+V/Wk5bcxujrx7dLk6Zeq9tdRu9duvVLk6Zes+73aNuYF27OO1wwg70HN1zS6yz4wQw7H+HseSKJe3ilKn3eo/tTc9RPdvFadaxs9iwaAN1A+tyxqn13VYaLmmgbmBduzhl6r0F5yygS88u7eKU+T+tuH/FlvLwxwmg+47dGXjEwHZxyvyf2prb2PORPdvFKfN/amtuY4cTdmgXp32f35fFly7ecvz44wS23tvQsCFnnDL/p53P2TlnnFY8uIK25jbG/HZMuzhl6r221W302atPuzhl6r3lf15O3cC6dnEafMpg6rav25Ivf5wy9d7wC4fTcHFDuzhl6r0+e/ah+9Du7eI048gZfPjOh9QNrMsZp/Xz17PkqiXUDaxrF6dMvffmd96ka/+u7eKU+T+99+f3tuQ9+/xUTDti0fmLaGtuY/CXB+dsRzT8pIHNrZvbxQlsvffBjA+25CvX+WnomUMLb0esbqPfgbnbEY1XN7LioRXUDayLvB0B0G/fftRtV5e7HfH+JrptX1g7YvR1o7e03XK1I3Y6bafQdkTPXXoW3I5o+GkDbc1tDPnGkNztiAveco9L5W5HND/fHNiO6DWmFzudulPh7YjVbQw4PHc7ouHCBpr+1UTdwLrI2xEA/T/Wf0s7yx+n6QdPp625jR47F9aOGHHJiC2xztWOGPqtoaHtiN679y64HbHkyiW0NbeROitVUDvi/X+9H9iO6LNnH3Y8eceC2xGByjWmtFwvkENAnvJ9vxjk4pD9u4CscZ+/CnKvb9ulIBcGpQ17pvC9+98LHXsbtr2YtCIiiy7ZU2b96uiS2A7bnm8cdSnLpBh2sWVSqWyNdbT5qmS2xjrafOVjl7K842RrrKO3XalsjXX0tuMq7zjZGuvOcUvNjtM2FfpM4VRgN2MYZQzdga8Cj/l3MIbdfF+PBxa6z08BexlDb2OoA44E5hWSiZ6jwictDdteTNqWtasZ1XUp63bcJ3LbHdleKtvFcIu1Xa1sjXXnbVcrW2PdedtJLO842Rrr5LA11slha6wrix2n7SDF2ikUoQ04F9vBmw88KMJcY7jSGE50u53rlpyYCZyPHTqKCKuB32I7ljOB6SL8u5B8rH11bcHbi0m7ZPaLdDFCn9EHR2473/Yh3xxSMtv50hbDLrZMKpWtsY42X5XM1lhHm6987FKWd5xsjXX0tiuVrbGO3nZc5R0nW2PdOW6p2XHaDlSu24e1+AobPrqhcUPobdaw7cWkfeXui0Uu6y/NTcsjt51ve77paktZJsWwiy2TSmVrrKPNVyWzNdbR5isfu5TlHSdbYx297Upla6yjtx1XecfJ1lh3jltqdpy2qdDhoxUh7/dewduLSdtjxQze/mgoA7bfKXLb+bZnHlYthe18aYthF1smlcrWWEebr0pma6yjzVc+dinLO062xjp625XK1lhHbzuu8o6TrbHuHLfU7DhtBypXT7EWX2F3Cj/a9FFojzpse6FpN3/0kTRdNlxePP/EkuQr3/Z8D9eWskyKYRdbJpXK1lhHm69KZmuso81XPnYpyztOtsY6etuVytZYR287rvKOk62x7hy31Ow4baN3CoP15nfeLHh7oWnffXsB27OGDbJXSfKVb3ufPfuUzHa+tMWwiy2TSmVrrKPNVyWzNdbR5isfu5TlHSdbYx297Upla6yjtx1XecfJ1lh3jltqdpy2g1TzncJJZtIJk8yk2zc2bqT5hWaaHmtixQMrWDtlLctuXMbGpRvpMbQHm9s2M/+M+QDM/4Z7P2M+m9s202NoDzYu3ciyG5exdspaVjywgqbHmmh+oZkBhw5g/cL1LLliCW0tbSw4266xMu80e0s6s27dovpFtK5spfHaRlpeb2Hjs4aXBv+CEYd8geX3Lqfl9RYar22064jVL9ombcbWgrMX0NbSxpIrlrB+4XoGHDog0KeGixsYe8fYgnzybvcYecnIQJ/mnTaP8feMb+fT8nuXs+rpVex06k6BPmXSBvnk3e4x9MyhBfnUcHEDo64aFejT+oXr6TWqV6BPAN2265bTp1VPr2K7o7YL9Mkfr1w+Nb/QzI4n75jTp81tm7ccr7l82rh0I/326xfo05IrlrDbzbsV5NPye5cz/ILhgT5l4lWIT/PPmM/4e8YH+rTsxmXsfM7OBfm0qH4Ru163a6BPLa+30GdCn0CfwK4Nmcunpsea2P647QN9AuAjAn1aO2Ut2x+3fU6f2lra6Nqra6BPrStb6TOhT6BPjdc2sut1uxbk04oHVrDzOTsH+jT/Gy5eBfi04OwF2/yvs31qvLaR4RcML8inBWcvYLebdwv0ae2UtfTbr19BPq14YAU7nrxjoE/+/3Uun1peb2G7o7bL6VPryla6bdct0Ke2ljZ6jeoV6NOyG5cx6qpROX3KKMinpseaGHrm0ECfsuvhzvi0qH5Rzlj764iRl4wsyKf5Z8xn7B1jA33Kd84N82n5vcvZ6dSdAn3KFevOnHOLaUfkO+cGtSOiOOcW047Id84tph2R75xbTDsi3zm3lO2IfOfcYtoR+c65xbQj8p1zi2lH5DvnFtOOyHfOLWU7It85t5h2RL5zbr52RKBy3T6sxVfY8NF5p80Lvc0atr2YtCIi04+YXjLbYdvDuMXaLsbnYm1XK1tjHW2+KpmtsY42X/nYpSzvONka6+htVypbYx297bjKO062xrpz3FKz47RNwPDR2Dtr5XqFdQrjVL7xzLXGVXZyuMpODlfZyeEqOzncpLKT6HNS2XH6HKeCOoU1P3y0I9pyK7eA7cWkzadibcfFLoZbrO1qZWusO2+7Wtka687bTmJ5x8nWWCeHrbFODltjXVnsOG0HydgOY+3rgAMOkGnTpuXctrltM13qgvvHYduLSau2K8t2nGy1XV7bcbLVdu3YjpOttstrO0622q4d23Gy1XZ5bcfJzpfWGPOaiByQ/bveKQSWXLqk4O3FpAVouLChZLbDtodxi7VdjM/F2q5Wtsa687arla2x7rztSi3vONka6+htVypbYx297bjKO062xrpz3FKz47QdqFxjSmvxFfZM4YbGDaFjb8O2F5NWJHw8c7G2w7bnG0ddyjIphl3K8o6TrbGONl+VzNZYR5uvfOxSlnecbI119LYrla2xjt52XOUdJ1tj3Tluqdlx2iapzxR2ZEmKN858I3Qq6TfOfCNwKum3LnwrdEmKOSfNAYKnkm5b2xY4lXQmbdC0y29d+FboVNIr/74y0KcP3/kwdHps7zYvdCrppkeaAn1q/FVj6PTYYT55t3ssvW5pQT41XNzAu3e9GzqVdKvXGjqV9PxvzA+cSrrh4oZQnzLxCpp2ua25LXAq6blfmhvo08alG1l03qLQqaRX3L8i0KcPl34YOj32O7e8ExqnpkeaAn16+8q3Q6fHDvNp2Y3LWHbDsoJ8WlS/iPf+/F7oVNKty1tDj715X5sXOJX04ksXh/o09+S5gT6tnbKWTas2BU4lPe9r8wJ9al3ZysJzF4ZOJf3en98L9Gnj2xtDp8dedsOy0Dg1PdIU6NPiSxeHTo8d5lPjtY28c8s7Bfm04OwFrLh/ReiSFK0rWgvyacUDK3j7yrdDfcquh7PrvU3vbwpcviFz7AfVewvOWhC6JMW7d70b+H/auGRjoE9NjzWx9LqloXFqeqQp0KeGixtCl6QI82nJFUvwbvMK8mn+GfNZ+feVoUtSbFq5qSCflt+7nMZfNYb6lO+cu2nlppK0I/Kdczcu3hjoU0fOuf5y6Ww7QkQKPucW045Y9fQqNr2/qSCf8rUj8p1zN7duDm1HLDhrQcHtiHzn3EysCz3nNj3SVHA7Qj6Sgs+5xbQjmh5rYtOqTQX5lK8dke+cK5sktB2x8NyFBbcj8p1z87Uj8p1z/f+BzrYjApWrp1iLr7A7hWsmrwntUYdtLyat2q4s23Gy1XZ5bcfJVtu1YztOttour+042Wq7dmzHyVbb5bUdJztfWpJ6p7AjylyZKWR7MWkB1s9fXzLbYdvDuMXaLsbnYm1XK1tjHW2+KpmtsY42X/nYpSzvONka6+htVypbYx297bjKO062xrpz3FKz47QdJO0UAl16hRdD2PZi0gIs/vniktkO2x7GLdZ2MT4Xa7ta2RrraPNVyWyNdbT5yscuZXnHydZYR2+7Utka6+htx1XecbI11p3jlpodp+3AdAWlqjHVDawreHsxaQE2NW0qme2w7WHcYm0X43OxtquVrbGONl+VzNZYR5uvfOxSlnecbI119LYrla2xjt52XOUdJ1tj3Tluqdlx2g5SYtYpNMasBN7Ota0f/QZ8wAdrgtKGbS8mLcAu7DJ+CUtyrjJZrO2w7WHcYm0X43OxtquVrbGONl+VzNZYR5uvfOxSlnecbI119LYrla2xjt52XOUdJ1tj3Tluqdlx2gZGisiO7X7N9aBh0l7P8dzthW4vJq3bnvNhz4hsh+U7kFuGMimYXcry1lhrrDXWtRfrUpa3xlpjrbGuvViXsrw11hrrSrAd9NLho1b/KmJ7MWkBbi+h7bDtYdxibRfjc7G2q5Wtse687Wpla6w7b7tSyztOtsY6etuVytZYR287rvKOk62x7hy31Ow4bedUYoaPqlQqlUqlUqlUKpWqvfROoUqlUqlUKpVKpVIlWNopVKlUKpVKpVKpVKoESzuFqqqXMZiksY3R/65KVWtKYl3m2FqflVFxxlqVHOlxVn3SirjKFfOJfJgxHBITu3fmswhSzkaFMfQ1hm4+dlliYAwTjOFex91cDmYlyBiGGMOYGLg9s76X8xjrVS5WDvZYY7gmJnZcF1m0LitjXebYsdRnes6M5TiLpT5LYl3m2Ik8zipBxjDIGHrEnY9Clahg1ZqMYTfg68awQwzsccA/gf4xsZ8whl8Ywx/BNirKUQkbw1jgCeB3vgZNyWdrMobdgXuBw43hy6XmZbF3NYYjjaFrObmOPQ54EkiVmTsWeNAYfmwMP4Mtx1jJ60zHfsQYfm0Ml5eal4P9IPAZYziwzOxY6jOty8pblzl2LPWZnjNjO87KXp8lsS5z7EQeZ748jDSGYzKd4jKzxwELgdPLxYxa2imsUhnDSGA6UA8cawztF6EsHXtX4N/A9SI8ld1QLmXD2RiGAH8B7gGuBnYyhhnGMLDUV6VcRf8Q8DfgeqCPMQz1bS8J2zWg7gNuA64BPlYKTgB7NDAXO23zEeXsGLq7gw8DvxXhhTLekR0G/AN7jD8JTDSGJ6D0HUMX6weAx4H/ArsaQ99S8bLYY4E/YWP9LHB8ObiOHUt9pnVZeesyZzuW+kzPmbEcZ7HUZ0msyxw7kceZLw/jsP5/BnjJGE4u44WuccCt2DbDqS4WVae6uDOgKli7AhcB7wJfBLoawxMirCwl1P2xj3Tc2e7ntDGsAwaL8J3MlaES/Rm7AbNEuMfl51zsyeZe4POlGoZkDHXAscB1IvzZndjGA2e4Su/HpfDb2GEIPwBucNwxwOXG8IoID0fFCWB3wTbYzgNagYuBLsYwSYSPSsyuA74ONGIreYArjaEF6C7CVSXEDwCeF+E2l5fbgL8Zw+MifK6Ex1h34FvA70S4211lvgL4hjuufl8KrmP3wp7IbxXhHmPYE3jUxXpSqbg+lb0+07qsvHWZY8dWn6HnzHtcfsp1nMVSnyWxLoPkHmcZGcNO2DvDN4hwhzHMAYYYw84ivFNi9jDsxY+0O+ZuB0YBbxlD11K3l6KUrlNYpXIn8q4irDGGk4GTsFfE/iPC8hKzhwKfBQ4BPoG9MvIotvPQU4RjSsgejb0a9QtgEvAVYASwD9AgwgUlZA9w5d0FSAM7uveLgUEifKJE3O1FeD9TuRjDmcAE4DJgfSmvhLkTuTj+2djj7FrgBRFaS8V17AnAKUAP7JXe/wIzgE8Da0X4dom4+2AXfv2SCJON4dtAb+Aw4EURbi4F17F3FGGluyN7D9AdeAQ4EWgT4ZslZA8RYbnvOLsEWCfCDcbQpZQn9bjqM63LyluXOXYs9ZmeM2M5zmKpz5JYlzl2Io8zx98JGCfC8+54W4jtHI8E7hDhlhKyDwL6ifB/7vvlwHHAweW6UxmZRERfVfoCMb7PJ4P8BeRzIF8GubnE7KEg9SBX+37rBvIAyIASs78M8jzIvSCvguwIsgfIr8pU7l1ADvB9rwN5EGS7csQb5DCQF0DGZh8HZTjOvgfyDMheIEeB/LzE7PEgN/pj6367E6SuhNxvgywGuQNkpjvevwhySZmOsTqQo33fx4LcBdKtDMdX5v1kkDdAhsZwnJWtPtO6rPx1WdZxVrb6TM+ZsR1nZa3PklqX6XEmXdz7USBXuM/7giwCOb4MfH/cHwA5qxx+R/nS4aNVLBE7W5yL5cPGsBz7jMYE4JwSs981hruATb6fD8deGSrJzEs+Xx8yhleAjcBmEVYZw2HAaDd05EMp4ZVAZ3ua76fDgeHYq6Alk4i94iTCf43hZeBmYzhRhA2l5maurorwB2NYCdyPvbvw3RKz5xvDL4EW388jgV2AfsbQnCmXKOQ7xu40hpewx9g6sVe764Bhxs7W2BYlN0ce2rBXlzPaBet336h9zsh3fGXeHzaGTwCXGEO9yDb/9cgVV32mdVn56zLHLnt9pufM8h9ncdRnSa3LHDuRxxlsqc8AXhLhOffbDGP4E5T+WdZM3N3X/wITS82MXHH3SvVV/Mt3NWwCyEeZKyKlvhqWlYcjQeaW42pMDvZRIO+CfLbM3DqQT4K8Xi6270rYOJC7QVJl9NfPbovpODsKZHYpyzuXP477Dsix5fI1kxeQT4HMKufx7Yv1Ce6qb8nvHGWXf1z1mdZlsRxnZa3P4j7GHCuJx1nZ67Mk12WOlbjjLCsPh4LMAzm0zNxBrs3w7bh8L+SlzxTWiNyzIYcBvcTOOlWqh4lzsbcHfgY8LcKT5WD62H2Ak4HlIjxTZr+HYB+cfzQGv+uAESI0lJnbDfgSsEaEJ8pc3oOBm4F7ROxsoGXiDgC+B8wU4alycR17GHaWxt+LbJlwp5z8HsAYEeaWmRtLfaZ1WfnrMscve32m58xYjrPY6rOk1WWOncjjzPF7YyfKuw04v5zHm+9Z1q9hn6ecXC52sdJOYYWrM38k/wPUJoIZjzrJ7i/C2pjY3cVNehIDe4AIa6Jgd5SbvV8UD8530udu4obfxMAeJMKqKNid5PYWYX0U3ALY24vwfrnZ2awY/I6sPtO6rEP7RlaXdYYddX2m58wO7RvncRZZfZbEuqwAds0cZyGcnPE0hkOBVhGmlapTGnYsGUOd2KHTVSPtFFaBjOE44AvAK8DLIizIsU/mysSWBruyq4/dSW6kFY4xfAo4BngGmCM5ZkrzsSOt4ONiJ9HnCmBXw3+rJuqTKmNHVp9Vkc+1xE5iHZ7UWMfJNiKIsXedN4nwnp+XtW9JOuMFsMt6l7QY6eL1FS5jF8T8KfAesAdwkbHr7mDcA62+P99A7Lo0kSyWquzysgvg3lAs12d3f+BXQD/s+oDfMXYRXv++fvbPjR1WWXXsJPocN9tnt1r+W1Vfn1Qhu+j6rEBuUsu7aHYS63Cf3UTFukLYmU7Z54AXgBuN4U6ATKfft2+d2HUZ+xnDmAg7hB1hd/Wxd62WDiGgE81U2gtke5C+7vPB7gHhE9z3PUF+AvJHkL3db13d+wCQSSBHKLs62DH7vJ3v8z4gk0EOd9+PAbkG5Gcgo3KwX6hGdhJ9rgB2Ev9bytZY1zo7iXV4UmMdG9uXh66+z7tjlzM5HGQwyL9B7vPv68vDQJDnQParRnYcr9gzoC9fMOxaMjf5KrSe2LXSnvLtMx7kcuwsWv3db5mD7zBlVwc7Zp+7gvwDZKT7Pg47S9a9vn2OBvkdyBUgvX3s/1eN7CT6XAHsJP63lK2xrnV2EuvwpMY6NrbP/o4gp4H0AtkB5BWQf4Js77b3Bnkc5J9Z6QaAPIu7aFBt7LhesWdAXy4QyCiQi9znESCXu889Qd4EudO37x4gu7rP3UFup7grQcouIztmn3cFOcd93g3kf93n8a7C8y94ewxbF5TuAfJQNbKT6HMFsJP431K2xrrW2Umsw5Ma69jYPrsDQE5y8d/B2T4G5EWQk0F6uv36gDwDsq/73h97F7mYDnFs7DhfsWdAXy4Qdk2T5SDfAUmBvATyU7etF3Zdn/sC0g5RdvWwY/Z5F5C3QD6HvdL6Hsj33bZx2Kt71wekHVGN7CT6XAHsJP63lK2xrnV2EuvwpMY6Nraz0QfkDpDPue9/APkp9u7lZ7FDU09ia+cssx6lATkW5KBqZMf9ij0DSX9hr2RlxiB/FuQ+kKHYqxNPg/zMbeuFvTozka0Loha1AKqyy8uO2ecuvs//A3KD+7wvyCKQc933PbBXXsf62F2qkZ1EnyuAncT/lrI11rXOTmIdntRYx8bOkZeLHGMAyBEgN4KcD1IHchzIVJAvZnNB6qqZHecr9gwk+QWyM8h8kDOxV8J6gtzM1mESe4I8CXKV+x7ZH07Z5WXH7PMIV4F9AjvWfzjIIyBfdNs/5vJ2vvvep9rZSfS5AthJ/G8pW2Nd6+wk1uFJjXVsbF8eUiCf8H2/GeRObGfsaOwzjudh79p9DuTgWmBXyiv2DCT5hb3VPAn78PQjIAeCHALyBlvHJ0/EPiQ9RtnVy46Dy9ard8OxM4bdg30Ify+QT4KsABnv9jkIZAnuIf5qZSfR57jZcR7jcXKVrbGuZXZcdUqS67Iks53tbo6/GORK99t47F26z7jvx2OfWbywVtiV9Io9A0l8gfTyfT4Ze0v6CJDnsVcj7nEH5xC3T19lVyc7Zp8z4927gdSDXA2yP8gUdzJNg9yNm+YbN3NYNbOT6HMFsJP431K2xrrW2Umsw5Ma69jYOfJyIMjfscOBf429U3wLyG99+3weZEItsSvlpYvXl1nGMByYaQxnGcNo4AngIGAw8BXgQGA1cBhwkjF0AdYru/rYMfs8xLG/AmwG7gYOAYYC3wC+BQwEjgcOdovOtlQzO4k+VwA7if8tZWusa52dxDo8qbGOje3Lw47GcLgxbC/CVOAR4HqgAdgT6Ap83xguBRDhnyLMq3Z2Jaou7gwkScbQCzgSe9AdBOwP/B+2ovsH8B5wHTDabVsowmZlVx87Zp/7Ax8HbgXOBvYB/gmcClzjXj8BDsBW+BtEEECqlZ1EnyuAncT/lrI11rXOTmIdntRYx8b25WE48EdgEfCRMfwGmAb8EPg5sNHl8QvAacZwhwjvVju7UuXGbqvKIWPoA1wIzAYeB04ALgZeBiYD44BrRfjAl8a4ik/ZVcSO2eftgZuBu4BZwGnAMcBCx+8jwp1u3zoR2qqdnUSfK4CdxP+Wsssb677Aj8rNTTg7iXV44v5bcbN99g4Bvgc8BHwMOB04CfgyMBb4mggfGsPhQKsIk2uBXbGKe/xqUl5sfXh6V5DXQD7lvvfBPtD7LMhqkKOVHXkeYmHH7PMeIC+w9eHw0dhFV98A8XCL+9YSO2afJySUncT/VlnZbDstfyLYIH1BesTkcyLZWfnYM6Y6PBauY42JsbwTyXa8nUEuwT2nB/IDkP+AfA/kZZCzcUtk1BK7Ul/6TGGJZQzdfZ+7ivAWcBHwSWMYIsI64IvYIRMzsOOXlV0ceztjGJn5Xi52XFzH3tEY9siUuwhzgVuAg9xV1QbgCOAPwCpgWLWzY/Z5pDF8NvNd7DMGSWCPNYZz3eey/a/j4lYAe0/gBmMY6GJb82xjGIe9cj8Yyl6PJpU9zBiOd3VLNxHmUJ46PBauYw8yhl0y30VYhB2eenQZyjuRbF8ejI//DrAB+LX7fhNwNdAb2AV7J69vLbCrQnH3Smv5hZ3O9iGQw9z3zF2zEdiZjcZl7d/Lv5+yC2a/5q72PMPW2bJGl5IdF9fZGAcyE+Qxd3Wrr/v9UOysYTtl7b9dtbNj5BrsrHgzQV4F+aZv28cde0itsZ2NsSCTQc7M+n0XkGtL+N+KhVsB7N1crM/L+n04yHW1yHbl/SrIKpC7fL+PLEM9mlT2HiCzXV36LMgw9/vBIPeWsD6LhetjzwR50LFOwN6l3aEM/61EsrOO9d9j10L8nO/334Fc4PvezeU3yhEIsbGr5aV3CkskY0gBTwLdgB8Zw+Eidhy2CI3AEuAmY+iXSSPCBvcuyi6IvQvwAPBbEY4FPOAmZ7uhVOy4uI49GvtA+PUinAiswT6sjwgvAzOB29yzAxn26mpmx+mzqzc3AY8CzwF7GcP/uG0vAVOBPxhD71piG8Mo7HMmvxDhLmOoM4aBzvYS4A3g5hL8t2Lhxs12Ohp4RIQbHPtAYxgGvANMB26pJbYxjMFOeHEtsBPQ2xgOdHbfprT1aFLZ/bB3SX7h6lIPOMQYuovwKjCF0tThsXB97F869inY5+kuA76OnWVzPqUt78SxfXkYiZ3d1ANSwHeM4Vq3+Q6gjzFb7syJCHNFeNZ/d68a2dUk7RSWTuuA3wHnAv8Bfmzsw6oAiHArtmF3pjGRxyGp7OHAXSL81X3/IVBnDN187EklYMfFBeiPreT/7L6PAk43hvuM4SgRbsB2Fj4ZMTdOdpw+Z/QOsAN2eM0exvAjYziPrUNujq4x9nqgJzDEff8bcLsxPGUMR4twF/YY/1bEx3hc3LjZAL3Yeo5+GNuA+x12eNND2Pq1Jtiu4bUncIUID2OHbH2Eu9gD29SjkfqcVLbTh9jhc5lG/97YCV4eNobvYpeEeIXo69K4uBn2ZuzFRES4HlgJTAQ+6f7X/6V05Z1EdkY7AJNE+AXwG+AC7FDhq0R4HTtk83SXv7ZMoog6pXGyq0baKSyBjKGLCGuA34mwDHgQO7PTj43hSLfPAOxV6BRsvTKj7KI0E7jfcbpi71aOw159xRh6As8DO0fMjouLCDNFuM9xfgBMEeFU4DXgAmNndFuEnVY5UsXFjtNnn54DlonwN+zseJcBE8VO1z23ltjGPkf3HjAeuMIY1mDL93RsY/U8Y6eTfwYYQUTHeFzcuNk+/RM43BjuAqaK8DkgDQzCPmc1qVbYruH1bxEecN9XY58juzBz18zpOexFuMh8Tirb2FkkW7GxvsAYngdeF+Hz2Asgn8AuRfCW+1zV3Awb2AS8iH2G7iRjOB1oxV5s+57b9WlKUN5JZGflQYDxxjBYhI1in5v9FrCfMRwFXAKcaEy0FzbjZFebtFNYArnGmf99FfB37K3rs43hIuA27OKYs11HStnF8z8QYbn7aoBmYLUIy4ydevg32Eb0rIj9joWbQ38U4ZsuT7/BPiA+HHgMWFZCbpzsuLjvAqPcxY6zsA2cVmM4FXsXpWbYInzkOkmN2LWs/ijCT0XYIMKvsBdBdsXepZwZ1TEeFzduNmy5wNaArS8nYCc+QIT/Yht3I7EXQSKvU+Jiix0ancmDEeFF7LAuf+doRtTcpLJ9d0AeAD4PvAT8y237G/YYH4WtX5ZWOzfDdvynsZ2hrwPHAd8T4ZdAN2OHrM6iBOWdRLbZOhGciDAdG+8njGGQ2+Udl6/9xa4B+BDQo9rZVau4H2pM2gvkJpCNIF/N+j2yB3mVvY39O0C+DTIN5EvlYsfF9TH2xz7Ev0dS2OXignTFTvpyDcjbID90v38dZL8aZtfl+O0AkNdBxpeKHRc3brazORjkXJD/YqdHP8Ad40fUMttn+1TsxBi9cmwrdX2SKDbIp7CTJx2OnchrLsghZTjOYuE6u11BerrPR4BMpf0EN8ounLPNpIO+33+DncArM3nQaSAPuM/DIvIxNnY1v2LPQJJe2NmMloOc6L6bcjTSk8jGra/lTjBrcWvwlMHXWLg+fneQo0DexDe7Vi2zY+R+AuRkfz6SwHY84xoTC8tc5rFwY/a5F8g+IH8FuQ3kpCSwfXm4P7uToOyS8MaCXImdCfTVcsU6Rq7xff40dh3EE5QdGScFsgTkUfc6PGv79dgZ268DaQQ5Piif1cSu9pdbKkBVDhnDdsAEEV7KzGgkUp6HWJPGdsNwxBjOB+aK8FQpeXFzffz+wLHAKhH+L5OfWmbH6bMvD13EDZsut+Jgu+dkPwVsLHOsY+HGzXb8LoARO7S15tmZ49oN4/2o1DxlgxtC2B3oL8LbZYx1LFwff2dgdxGeU3ZktgcAZ2KHZH7OvX4tdnh0Zp8jgTrgQ7HD06ueXe3STmGR8nUC+gCbRdiQq5GW/ZuxC7O2tTOo7MjYbt8+YhdjLZodF7cAdnexD/JXNbuKfN7yexQdtCpibynnMsc6Mm6Vsbc02qLoMMTFTqLP1cLO7hiU638dNbeT7ES2j0rBzrbtex8EnILtnF0nwvOu47YuKmYlsGtBOtFMkXJ/vs9jp+1+2BgOzRyQWbsa2HK1mSgORmWHs93JdF1U7Li4BbBba4FdRT5v9rGLvmNXRey2mGIdGbfK2OJjF30HKS52En2uFjauTVju/3XU3E6yE9k+KgXbl4d8kw5eiF12ZLuomJXArgnFPX612l/Yh1knYZ9pOgtkNcjH3bauWe8DsQ/0jlZ29bGT6LOWt7I11srWWCu7mrjKjofdwfxlJh38crmYlcCullfsGai2F8hQkM+6z7uB3Adym2/7mSArQY503+vc+wCQZ8h64FXZlctOos9a3srWWCtbY63sauIqOx52AXltN+lgEtjV9Io9A9X2AvksyASQvthpfa8GeRzkMN+f7Xsg69g65e122Cs3hym7ethJ9FnLW9kaa2VrrJVdTVxlx8MuIK/bsfWuZdlmwI+bXU2v2DNQLS+QkSAHuM8DQO4F+TJIF/cnvAHkULbelh/m3g3I+SBHKbs62En0Wctb2RprZWuslV1NXGXHw87Kh5uwUvrg1tXELc2VtV+XrO/t1oCtJnatvmLPQKW/3B+oP/a285u4deewC5PfAXIiSB3IVdh1nA5z27v40he0hpiyy8tOos9a3srWWCtbY61sjbWyC32BfB7kKZAnQQ7183z7ZDqnPWuFXYuvOlShEkGAtcZwN3AAcIEx9BDhTmNoBU4CBLgC+AWwyqXb7EvfquzKZyfR5zjZSfQ5qewk+pxUdhJ9Tio7iT4nmZ1LxjAeqHe83YB/G8PnxK5L3VXs2qaZ94HAM8bwFREaqpldq9JOYYiMoZsIm9zXScBA4EngO8aACH82hs3A14GuIvxE2dXJTqLPcbKT6HNS2Un0OansJPqcVHYSfU4y25eHocC+IjxhDLsBPwfeFOE54DljaAMeNYYviV0TsE7sUj4DsIvJn19opyxOdlKk6xQGyBjGAXcaw1Hup6eBAcDBwL3AOcZwrAh/xf4pIzvQlF1edhJ9jpOdRJ+Tyk6iz0llJ9HnpLKT6HOS2VnaF1hiDH0dowHY2RgOc52wu4BLgSeMYTvXKdsO+CdwhQgvVik7EdI7hcEaDJwG7GEMfwA2AxcDXwMmAz2BnxhDFxHuVXZVs5Poc5zsJPqcVHYSfU4qO4k+J5WdRJ+TzMYYRgI7urt0A4BbgMexnbCrgC8Bm41hsgh/MIbHRVhtDAb4FrZT9t9qYydN2ikMkAgvGMMRwFOABxwK3A/sDDwL/B17p3WlsqubnUSf42Qn0eekspPoc1LZSfQ5qewk+pxUtutY9cN2PNcYw/dF+D9jeAH4DPAhcJl7ne7y8F+Xx4xuFun8c4xxshOruGe6qfQXyGdA5oN0A/kYyM9BjnbbSjqtrbLLy06iz1reytZYK7sWuMrWWCu7pLH+FXax+ydBjne/fQPkbpATsDOeXgMyoZbYSXvFnoFqeIEc7/6EA9z3bsquTXYSfdbyVnYtc5WtsVZ27XGVXXq2367rjN6KXfLiEV/n7Osg94N8oVbYSX7p8NEOSIR/G8NHwAJjGCfCamXXJjuJPsfJTqLPSWUn0eekspPoc1LZSfQ5CWxjJ7b5qTHcLXZ2z6exQzT9E9t8JMJfjaGO6CfViYWddBnb21Z1RMZwPLBOhEnKrm12En2Ok51En5PKTqLPSWUn0eekspPocy2z3fOLk4AZsGVim//DTmxzL3Ak8D3gWhGeqBV20qWdwgJkDEaEWApO2cngJpWdRJ+Tyk6iz0llJ9HnpLKT6HOtso3hMOzENqdgJ7b5JHZimy9iO2ynAAtFmFpL7CRLO4UqlUqlUqlUKpVqGxnDZ4AbgInYdQKPBV4S4VnjFoevRXZSpZ1ClUqlUqlUKpVK1U5umOpvgINFWGMM3UTYVOvsJEonmlGpVCqVSqVSqVTtVOuT6qi2Su8UqlQqlUqlUqlUqkDV6qQ6qq3STqFKpVKpVCqVSqXKq1qcVEdlpZ1ClUqlUqlUKpVKpUqwusSdAZVKpVKpVCqVSqVSxSftFKpUKpVKpVKpVCpVgqWdQpVKpVKpVCqVSqVKsHRJCpVKpaoSpVL1TwPHuK//63npGwP2uxM403292/PSZ/q2/RS4GrjK89I/9/1+O/Bd9/VLnpf+Rw67mbQXeF76t0X4kQK+g12MeDQwCGgB5gL/Au7wvPSqHOl2culOAHYF+gNNwMvATZ6XfiFHmtHAW8ACz0uPDciPAZqBvsAAz0u3uN8PBV5yuz3ueekTAnx5B3jd89ITU6n6c4GbOlYSALzneekhHd05lao/BXgAOMvz0n8M2GcasD+wi+el387a1gNYAgwBfuF56UtzpL8JOBf4geelb3a/ZY69oz0v/f98+y4AdgM+AlKel16Rw14m7f6el56eI21Ggj0OlgJTgD95Xvq5POXRH3usnwTsDmwPrAamAbd5XvqxrP37Ah8ASzwvPSqHvX2AGe7rpzwv/WyOfWYDewETPS/9elj+VCqVqlqknUKVSqWqHu0PtGHr7om5dkil6g8GvoVtpHfFNo79OsC9B/0OcArQrlMYkrbDSqXqzwF+je2AzQIeA94HUtiOw6+BHwDDs9J9F7uIcX9sh+Hv2I7cWODzwBdTqfrLPS99ZUCep4Rka6yzOzfTIcxKC/DpVKp+gOel1wTYz5TJYuCKrH0+BXwceBp4JWvbkpB8baNUqr6rs/0WcHfIrpm1vAbk2HY6tkOYc3sqVb89tpPVBNzl27Q/ttP2mm/fAcAY97Ur8EXg1hzM/YEPgS0dKF/aZuB37ucuwEBsh+sbwBmpVP3fgDM8L91uwepUqv7zwB+cP3OAx4GVwEjsMfFZd4Hku56XFgDPS7ekUvVtuXx3usj3OVf5HOvy94R2CFUqVS1JO4UqlUpVBXJ3vAZh74rtSo5OYSpV3wW4BdswXgwcRAc6he7u0Z7ACqABOD6Vqu/teen1OdJuBqZTgFKp+suBy7B31k7wvPSkrO11wLeBkwPSvYW9ezM1a/tewIvAFalU/QzPS/8rK88A26TJ0oEB+2TSPgacCHwBuDdgn2kAnpf+N/DvrPzt7T7e5Hnpx0PykU+nAeOAb3peui1kv2b3vk2nxh0fFwDLsXfUBuZIew7QG7g2E3/fsbcwq1O8P2CAp4BPYC8mbNMp9KWdmtWxy6Sd7Hnpy7MzkUrVjwEeAU7FHi8XZW3/NnA7tvN6nOel/5O1fQTwAvZ4msW2d2+bge1yMEcCXwYagRHkLp8L3fu1ObapVCpV1UqfKVSpVKrqUKbz8Rp2eNserpHv1/8A+2Eb0LsCm7ANYgBSqfrB2Dtw73pe2vOl2xvo5mw/BPQBjvcb9qV9M+tuWoeUStWfjO3YvQcclN0hBPC8dJvnpW/Ddr4y6b7m0r3t0rXr3Lk7Npk7hD/K2tyZTmFQB/oS7J3XU3Kk7cjd0/3d+2sh+4QqlarvBvwcmA/8Nc/uQXcKv4AdYnkDsCZ7eypV3xM7bHQ9cLNvU767y/8Pexf0CDfEN9c+QWlzlonnpRcBX3Vfz02l6nv58nk4cBt2GOjHszuELn0jcL77ekHW5tVAVzeU1K8LsBfLM3d6s8tnP+CT2I7s87nyrVKpVNUq7RSqVCpVdcjfiJ6OvZuza2ZjKlW/A/Z5v5eB54EdgDmel/4wh42wu4d/d5+zO0AFDx11Dfpb3NfTPS/9Ttj+npfe4NL1x3ZOBPia56XfD0n2lHs/2Mc12E5yGzAzJG27O4WuwzAWWOp56TnYMj0mlaofmJV2f7I6335ldcTfDclDPp2Jff7y556X3pxn32b3nt0pvAjbGbzVvQ/M2n4GMBi4M6usO3LcPIRtU3wpZJ+O/L5Fnpeei7073AvYB7YMof0jdrjq91znMUiZY2Kke/Yzo2b3vqV83LDZbwOTsB1caF8+mbuVepdQpVLVnLRTqFKpVNUhfyM6MxGGfwjpNdhG7rlsvTPV2ecJp7k7LJOxz2P16UDajugM7HNfkzwv/VSeff36H+zQw8c8L539LF62lrr37r67SmOwZTIn09HMlhuyug/tO3b7Yc+RGX8fwt5NPcmXdgS2E5Xd+fYrKBYdlhve+zNs3HM965mtdncKU6n6I7HDiW/1vPRasu4UurvO52M70Ndn2Qs7bjLPGT6GfW6woxcTQu8U+pTpnA507ydjO+szPS99f1hCz0uvI/dd01y/nYu90HINtmy22Z5K1Y/CdngXAI/mybNKpVJVnbRTqFKpVBUu3x2v9cAbbH2mb6LbfhD2TtJtnpeeQeGdwkwD/UFsA/lzHUjbEX3Bvf+pk+kyzxb+pQP7bu/e1/s6gB0ZOroX0BM7e2iuu6qZMvkH7YeQlmXoKHA2MAx7l1A6sH+uTs9FwEbs0FGAtWx7J+wL2JlAH/TPWOo79rZ5ljSVqh8EjALe8rz0Gves4dPAYalU/dCstBuwM8tmp21yFyHClPEh0znMHBP5htBmhtz2d1+bfJu2KR93EeFcYIa7aNHi/B3oS1OPvTv5mw7cqVWpVKqqk3YKVSqVqvK1G7YBO9Pz0h9hJ5FZA0z0TS7zPvbZN+jEkD3XIJ6AHd6YGdaZawjpAdhO0Qw6r0zHKN/dvi3ydSggvFOXUWbYqH8SnCgmmclMILMSO4T0aNepabdPgIrqFKZS9b2BnwCvdmKSmmb3nun07AkcB9zjeen33LbsZwqDJlDJHHvZz5Lm8j17CGn2cZudNrRM3B3SkdgOWmaYaKY8O3JMHIDtyC118cuo2b1n/D8TO9z61wCu4/0BW8tvkNtnOZ2/sKFSqVRVIe0UqlQqVeUru4Mi2M7ZXsBZ2IbyxZ6XztwBybUEQAoYim0g+9eS2xfbcN7SQPcNIT0ularv60s7L2gYZpDcM2CZTtTyTiTtB3R3n1eH7eh0mnt/2PdbplMZNltqpjOZvWRFUKfHP4S0HHcKfwjshB0+2lFlymuge78I27G6zrfPGqB/KlVv3MQtBwP/8bx09rORnXkm8DGgla0XEwp+ntDpSOxd3Fd961bu4N47c0w8kvX7lvJxx+cF2GcX/+7bx//M5fexky/dEDJMWKVSqapa2ilUqVSqyleuOyszsBOP/BJ71+RO2PLs0yBgdtYSAJ19tush7AQfJ4SkzSt3h2ij+9qvE0lbsM+3AewctmMqVf9x4LPYu6X+Bd33cO8NAem6Ap9xX5/1/Z5ZQ+/trAlXHmbbIaTtOt9Z9ouaZMbl40Lss5jtFlEPUbN7H5BK1Q/HzuL5kOel/eWwBtsG6Ef4MgsdPm58Q0g/7i4kFPs8YWZil1t8vzW793zHxGjsxDGbaP+MZMbGAOwSFKOA67LuZq7Bll9mRta12DURVSqVqialnUKVSqWqfOVqRE/HrvM2APi+71mzgieZyfrdP4S02EXrM3fhPp1vRzdsFPfc1uR86VKp+h2B+9zXs7OGOHZz74MDkn8FSGE7XW/7fs+sobeNv+4O6wvAJ1Op+gPJ3fn2q9hJZs53jM7cJYRtn5mrx5bDNVn7rHXvB2OfHZ3qeennctjqyCQzfj2ELbsv50lLjrRblErVXwIcjT12/BPKvOzew46JXtjjtwdwSY7nFv3lcyH2DvY9WfusddtPxx4/t2Wt0ahSqVQ1Je0UqlQqVQXLPTO4L7AOu0ZdRk9ghzEenbV2X2c7fzl/d52kKcCxwFEBaTuqtHv/ZSpVf3CuHVKp+o+nUvVPYIeyZqe7LJWqH58jzSHYjuNw4FLPSz+UtUumU3mxuyvoT3s89g7UR8CPs9KFdYIfwq5l98uQfTIqeOioe47tPOyQzpc6mbzZvY8EvkvuYaGZDs5V2E5cu7uEvmNvm2dJfXdAF3he+oOsZP/EDiH9qkvbgp0cKTvtylyTzKRS9cNSqfp7gF9gF5E/OWtil99hh8J+3w17zU4/Httx3Be4A/hN9j5sLZ8vYocY5xoWuga7wP0Fzp8bcthRqVSqmlFd3BlQqVQqVajGAX2Bl/yNY/eM1aM59g+6O9Xud99afJ7npXM97/cQ8DHgMELW4ssnz0v/M5WqvwK7+PrLqVT9c8A8bEc3BRyKXXPxZc9Lt/nS/SOVqv8VcDEwM5Wq/xd2SYCB2I7bgdjG+3c8L31nDvRF2Dthp2NnxXwVezF0ArA3dljr6Z6XDnqeMFdn7mHs2omfct9L9Tzhj7EzZ3b2LiFsvRM2wb1n3yWErZ3CjwEL2fZZzIwyx97rWc+SBpaP56XXpFL1zwDHu59ezOrUZdJuTKXqL3ef67B3RPfBxrQO+Bdwpuel/bOG4nnpqalU/bnATcCkVKr+P8Bsl8+9gCOwnbifeV766hw+wdby+Rhb123MVqZTuB1wl+elvQBbKpVKVRPSO4UqlUpV2ero81cZ5VoCILOe3mLfhB2Zff1r8WXLf+ctbC2+vPK89OXA4dihnmOwE+Scg22Yz3Ofs9e4w/PSP8UOI3wc23n8kdtPgEuBsQEdQjwvPRvr453YIZSnAJ93n28CJnpeOtfSBmGdnvewQ0gzirxTmErVD8E+x/aI56U73aF0w1nXua+vel76+Ry7+YdCXh+wzEKhE8X4j5ugtMOBy9zrPOyzqxuxk+FM9Lz0idkdwow8L30rds3Fv2E7gudjJ5UZiJ1BdEJIhxC2naQms25jtjLlI2w7QY9KpVLVpIxIR5Y8UqlUKpVKVQ6lUvU3Yme83Nvz0nPizo9KpVKpal/aKVSpVCqVqkLkZgtdCDzseelT486PSqVSqZIh7RSqVCqVSqVSqVQqVYKlE82oVCqVqlNKpep7s3UNuXx60PPS80qZH5VKpVKpVMVJO4UqlUql6qx6YycI6YjmYCeSUalUKpVKVaHS4aMqlUqlUqlUKpVKlWDpkhQqlUqlUqlUKpVKlWBpp1ClUqlUKpVKpVKpEiztFKpUKpVKpVKpVCpVgqWdQpVKpVKpVCqVSqVKsLRTqFKpVCqVSqVSqVQJlnYKVSqVSqVSqVQqlSrB+v+++Nzyy+wyKwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 1080x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(15, 5))\n",
    "fig.suptitle('DCG@1 / Hits@1 with CBOW', fontsize=20, fontfamily = 'Verdana', fontstyle = 'normal', color = 'midnightblue')\n",
    "ax.plot(df2.index, df2[['DCG@1','Hits@1']])\n",
    "\n",
    "plt.xticks(np.arange(0, 100, 4),rotation=45)\n",
    "\n",
    "plt.legend(df2[['DCG@1','Hits@1']],loc = 'upper left', shadow = False,\n",
    "          facecolor = 'yellow', edgecolor = 'r', framealpha = .1, fontsize=16)\n",
    "plt.tick_params(axis = 'both', which = 'major', direction = 'inout', color = 'm',\n",
    "               labelcolor = 'b', bottom = True, left = True, labelbottom = True, labelleft = True)\n",
    "plt.tick_params(axis = 'both', which = 'minor', direction = 'out', color = 'm',\n",
    "               labelcolor = 'r', bottom = True, left = True)\n",
    "plt.grid(which='major', color = 'm', linestyle = '--')\n",
    "\n",
    "plt.minorticks_on()\n",
    "plt.grid(which='minor', color = 'm', linestyle = ':')\n",
    "\n",
    "ax.set_xlabel('', fontsize = 20, color = 'midnightblue', fontstyle = 'oblique')\n",
    "ax.set_ylabel('', fontsize = 20, color = 'midnightblue', fontstyle = 'oblique')\n",
    "plt.setp(ax, xlabel = 'MIN_COUNT / WINDOW', ylabel = 'VALUES')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 275,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>DCG@1</th>\n",
       "      <th>DCG@5</th>\n",
       "      <th>DCG@10</th>\n",
       "      <th>DCG@100</th>\n",
       "      <th>DCG@500</th>\n",
       "      <th>DCG@1000</th>\n",
       "      <th>Hits@1</th>\n",
       "      <th>Hits@5</th>\n",
       "      <th>Hits@10</th>\n",
       "      <th>Hits@100</th>\n",
       "      <th>Hits@500</th>\n",
       "      <th>Hits@1000</th>\n",
       "      <th>SUM</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Min_7/Win_9</th>\n",
       "      <td>0.421</td>\n",
       "      <td>0.509</td>\n",
       "      <td>0.536</td>\n",
       "      <td>0.58</td>\n",
       "      <td>0.592</td>\n",
       "      <td>0.595</td>\n",
       "      <td>0.421</td>\n",
       "      <td>0.587</td>\n",
       "      <td>0.669</td>\n",
       "      <td>0.884</td>\n",
       "      <td>0.976</td>\n",
       "      <td>1.0</td>\n",
       "      <td>7.77</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             DCG@1  DCG@5  DCG@10  DCG@100  DCG@500  DCG@1000  Hits@1  Hits@5  \\\n",
       "Min_7/Win_9  0.421  0.509   0.536     0.58    0.592     0.595   0.421   0.587   \n",
       "\n",
       "             Hits@10  Hits@100  Hits@500  Hits@1000   SUM  \n",
       "Min_7/Win_9    0.669     0.884     0.976        1.0  7.77  "
      ]
     },
     "execution_count": 275,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Строка с максимальным значением Hits@1\n",
    "df2[df2['Hits@1']==df2['Hits@1'].max()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 276,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>DCG@1</th>\n",
       "      <th>DCG@5</th>\n",
       "      <th>DCG@10</th>\n",
       "      <th>DCG@100</th>\n",
       "      <th>DCG@500</th>\n",
       "      <th>DCG@1000</th>\n",
       "      <th>Hits@1</th>\n",
       "      <th>Hits@5</th>\n",
       "      <th>Hits@10</th>\n",
       "      <th>Hits@100</th>\n",
       "      <th>Hits@500</th>\n",
       "      <th>Hits@1000</th>\n",
       "      <th>SUM</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Min_10/Win_10</th>\n",
       "      <td>0.416</td>\n",
       "      <td>0.513</td>\n",
       "      <td>0.535</td>\n",
       "      <td>0.581</td>\n",
       "      <td>0.593</td>\n",
       "      <td>0.595</td>\n",
       "      <td>0.416</td>\n",
       "      <td>0.597</td>\n",
       "      <td>0.666</td>\n",
       "      <td>0.886</td>\n",
       "      <td>0.979</td>\n",
       "      <td>1.0</td>\n",
       "      <td>7.777</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               DCG@1  DCG@5  DCG@10  DCG@100  DCG@500  DCG@1000  Hits@1  \\\n",
       "Min_10/Win_10  0.416  0.513   0.535    0.581    0.593     0.595   0.416   \n",
       "\n",
       "               Hits@5  Hits@10  Hits@100  Hits@500  Hits@1000    SUM  \n",
       "Min_10/Win_10   0.597    0.666     0.886     0.979        1.0  7.777  "
      ]
     },
     "execution_count": 276,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Строка с максимальным значением суммы всех метрик (DCG@1 + DCG@5 + .... + Hits@500 + Hits@1000)\n",
    "df2[df2['SUM']==df2['SUM'].max()]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Применим подобранные лучшие значения min_count, window и sg для рассчета значений метрик на разных токенайзерах."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "embeddings_trained = Word2Vec(words, # data for model to train on\n",
    "            vector_size=200,         # embedding vector size\n",
    "            min_count=9,             # Ignores all words with total frequency lower than this. Consider words that occured at least 5 times\n",
    "            window=10,               # Maximum distance between the current and predicted word within a sentence\n",
    "            workers=4,               # Use these many worker threads to train the model\n",
    "            sg=1).wv                 # sg=0 - CBOW, sg=1 - Skip-Gram"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MyTokenizer\n",
      "DCG@   1: 0.532 | Hits@   1: 0.532\n",
      "DCG@   5: 0.614 | Hits@   5: 0.683\n",
      "DCG@  10: 0.635 | Hits@  10: 0.749\n",
      "DCG@ 100: 0.670 | Hits@ 100: 0.913\n",
      "DCG@ 500: 0.680 | Hits@ 500: 0.986\n",
      "DCG@1000: 0.681 | Hits@1000: 1.000\n",
      "\n",
      "SpacyTokenizer\n",
      "DCG@   1: 0.508 | Hits@   1: 0.508\n",
      "DCG@   5: 0.592 | Hits@   5: 0.667\n",
      "DCG@  10: 0.610 | Hits@  10: 0.724\n",
      "DCG@ 100: 0.646 | Hits@ 100: 0.894\n",
      "DCG@ 500: 0.656 | Hits@ 500: 0.974\n",
      "DCG@1000: 0.659 | Hits@1000: 1.000\n",
      "\n",
      "WordPunctTokenizer\n",
      "DCG@   1: 0.505 | Hits@   1: 0.505\n",
      "DCG@   5: 0.598 | Hits@   5: 0.678\n",
      "DCG@  10: 0.614 | Hits@  10: 0.730\n",
      "DCG@ 100: 0.651 | Hits@ 100: 0.906\n",
      "DCG@ 500: 0.661 | Hits@ 500: 0.979\n",
      "DCG@1000: 0.663 | Hits@1000: 1.000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "tokenizers = [MyTokenizer(), SpacyTokenizer(), WordPunctTokenizer()]\n",
    "count=0\n",
    "\n",
    "for tokenizer in tokenizers:\n",
    "    wv_ranking = []\n",
    "    max_validation_examples = 1000\n",
    "    for i, line in enumerate(validation_data):\n",
    "        if i == max_validation_examples:\n",
    "            break\n",
    "        q, *ex = line\n",
    "        ranks = rank_candidates(q, ex, embeddings_trained, tokenizer)\n",
    "        wv_ranking.append([r[0] for r in ranks].index(0) + 1)\n",
    "    if count < 2:\n",
    "        print(str(tokenizer).split(\".\")[1].split(\" \")[0])\n",
    "    else:\n",
    "        print(str(tokenizer).split(\"(\")[0])\n",
    "    for k in [1, 5, 10, 100, 500, 1000]:\n",
    "        print(\"DCG@%4d: %.3f | Hits@%4d: %.3f\" % (k, dcg_score(wv_ranking, k), k, hits_count(wv_ranking, k)))\n",
    "    print()\n",
    "    count+=1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Эмбеддинги с нормализацией (лемматизацией)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [],
   "source": [
    "def preproc_nltk_lemma(text):\n",
    "    text = ' '.join([word for word in word_tokenize(text.lower()) if word not in stopWords])\n",
    "    return SpacyTokenizer().tokenize(text)\n",
    "words_lemma = [preproc_nltk_lemma(question) for question in couple_list]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['convert', 'string', 'list', 'convert', 'google', 'result', 'object', '(', 'pure', 'js', ')', 'python', 'object']\n",
      "\n",
      "['html', '5', 'canvas', 'javascript', 'use', 'make', 'interactive', 'drawing', 'tool', '?', 'event', 'handling', 'geometry', 'three.js', '?']\n",
      "\n",
      "['send', 'array', 'via', 'ajax', 'fail', 'get', 'list', 'item', 'unordered', 'list', 'php']\n",
      "\n",
      "['insert', 'cookiecollection', 'cookiecontainer', '?', 'c', '#', 'create', 'cookie', 'string', 'send']\n",
      "\n",
      "['update', 'one', 'element', 'bind', 'observable', 'collection', 'wpf-', 'update', 'change', 'list', 'item', 'list']\n",
      "\n"
     ]
    }
   ],
   "source": [
    "for i in range(5):\n",
    "    print(words_lemma[i])\n",
    "    print()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Обучим эмбеддинги с нормализацией на полученных ранее оптимальных параметрах:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [],
   "source": [
    "embeddings_trained_lemma = Word2Vec(words_lemma, # data for model to train on\n",
    "            vector_size=200,         # embedding vector size\n",
    "            min_count=9,             # Ignores all words with total frequency lower than this. Consider words that occured at least 5 times\n",
    "            window=10,               # Maximum distance between the current and predicted word within a sentence\n",
    "            workers=4,               # Use these many worker threads to train the model\n",
    "            sg=1).wv                 # sg=0 - CBOW, sg=1 - Skip-Gram"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Проверим качество ранжирования для каждого из трёх токенайзеров:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MyTokenizer\n",
      "DCG@   1: 0.366 | Hits@   1: 0.366\n",
      "DCG@   5: 0.441 | Hits@   5: 0.507\n",
      "DCG@  10: 0.459 | Hits@  10: 0.563\n",
      "DCG@ 100: 0.500 | Hits@ 100: 0.764\n",
      "DCG@ 500: 0.522 | Hits@ 500: 0.940\n",
      "DCG@1000: 0.529 | Hits@1000: 1.000\n",
      "\n",
      "SpacyTokenizer\n",
      "DCG@   1: 0.375 | Hits@   1: 0.375\n",
      "DCG@   5: 0.452 | Hits@   5: 0.520\n",
      "DCG@  10: 0.472 | Hits@  10: 0.583\n",
      "DCG@ 100: 0.512 | Hits@ 100: 0.779\n",
      "DCG@ 500: 0.533 | Hits@ 500: 0.940\n",
      "DCG@1000: 0.539 | Hits@1000: 1.000\n",
      "\n",
      "WordPunctTokenizer\n",
      "DCG@   1: 0.352 | Hits@   1: 0.352\n",
      "DCG@   5: 0.428 | Hits@   5: 0.492\n",
      "DCG@  10: 0.447 | Hits@  10: 0.552\n",
      "DCG@ 100: 0.487 | Hits@ 100: 0.753\n",
      "DCG@ 500: 0.511 | Hits@ 500: 0.937\n",
      "DCG@1000: 0.517 | Hits@1000: 1.000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "tokenizers = [MyTokenizer(), SpacyTokenizer(), WordPunctTokenizer()]\n",
    "count=0\n",
    "\n",
    "for tokenizer in tokenizers:\n",
    "    wv_ranking = []\n",
    "    max_validation_examples = 1000\n",
    "    for i, line in enumerate(validation_data):\n",
    "        if i == max_validation_examples:\n",
    "            break\n",
    "        q, *ex = line\n",
    "        ranks = rank_candidates(q, ex, embeddings_trained_lemma, tokenizer)\n",
    "        wv_ranking.append([r[0] for r in ranks].index(0) + 1)\n",
    "    if count < 2:\n",
    "        print(str(tokenizer).split(\".\")[1].split(\" \")[0])\n",
    "    else:\n",
    "        print(str(tokenizer).split(\"(\")[0])\n",
    "    for k in [1, 5, 10, 100, 500, 1000]:\n",
    "        print(\"DCG@%4d: %.3f | Hits@%4d: %.3f\" % (k, dcg_score(wv_ranking, k), k, hits_count(wv_ranking, k)))\n",
    "    print()\n",
    "    count+=1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "tY8PxB0j-ThG"
   },
   "source": [
    "### Замечание:\n",
    "Решить эту задачу с помощью обучения полноценной нейронной сети будет вам предложено, как часть задания в одной из домашних работ по теме \"Диалоговые системы\"."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "vymVj8IxO2PO"
   },
   "source": [
    "Напишите свой вывод о полученных результатах.\n",
    "* Какой принцип токенизации даёт качество лучше и почему?\n",
    "* Помогает ли нормализация слов?\n",
    "* Какие эмбеддинги лучше справляются с задачей и почему?\n",
    "* Почему получилось плохое качество решения задачи?\n",
    "* Предложите свой подход к решению задачи.\n",
    "\n",
    "## Вывод:\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "emODHztAQUQz"
   },
   "source": [
    "- Лучшее качество дает токенизация, основанная на регулярных выражениях и возвращающая слова без знаков пунктуации и без применения нормализации, потому что для используемого текста лемматизация дает более худшее ранжирование.\n",
    "- Нормализация (лемматизация) слов в данном конкретном случае не помогла и ухудшила полученные метрики.\n",
    "- С задачей лучше всего справляются эмбеддинги с применением технологии Skip-Gram, в которой распределенное представление входного слова используется для прогнозирования контекста. В модели Skip Gram контекстные слова предсказываются с использованием базового слова. Skip-Gram хорошо работает с небольшим объемом обучающих данных, хорошо представляет даже редкие слова или фразы. Модель с CBOW (Continuous Bag of Words), в которой распределенные представления контекста (или окружающих слов) объединяются для прогнозирования слова в середине, показала более худшие результаты. Но при этом CBOW обучается в несколько раз быстрее, чем Skip-Gram, и для частых слов может показать немного лучшую точность. Также более лучшее качество показали эмбеддинги с большим размером гипер-параметра window, характеризующим максимальное расстояние между текущим и прогнозируемым словом в предложении. Гипер-параметр min_count позволяет игнорировать слова, встречающиеся в тексте более чем n раз. Его увеличение приводит к незначительному росту метрик DCG и Hits при прочих равных параметрах. Эмбеддинги с лемматизацией показали более худшие результаты применительно к данному конкретному тексту, вероятно, потому что данный текст является \"техническим\". Предобученные эмбеддинги показали схожие значения метрик по сравнению с обучаемыми только в диапазоне с min_count=1, window=1, sg=1 (Skip Gram) и оказались хуже всех остальных метрик при увеличении min_count и window. Однако, они превосходят или равны почти всем полученным значениям метрик в случае использования CBOW.\n",
    "- Плохое качество решения задачи получилось, потому что word2vec не является современным решением. Оно разработано в 2013 году и имеет определенные недостатки: нет информации о предложении или контексте, в котором используется слово, игнорируется совместная встречаемость слов, модель не учитывает то, что слово может иметь различное значение в зависимости от контекста использования, не очень хорошо обрабатывает неизвестные и редкие слова. (по материалам статьи Ilias Miraoui - Master of Engineering Student at Berkeley)\n",
    "- В качестве улучшения решения можно предложить переход на использование модели GloVe (Global Vectors), которая улучшает Word2Vec, добавляет частоту встречаемости слов и опережает Word2Vec на большинстве бенчмарков. Еще один подход для улучшения результатов - использование библиотеки fastText, которая к основной модели Word2Vec добавлет модель символьных n-грамм. При этом в отличие от Word2Vec и Glove, модель способна генерировать эмбеддинги и для неизвестных слов."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [
    "BIWqBuEa6j0b",
    "uS9FwWNd5a3S",
    "MQk_rolFwT_h",
    "ai48-5vv6j1d",
    "Y60z4t6W6j16",
    "0sUSxk866j1_",
    "J5xWOORI6j2F",
    "tHZqgDTo6j0i",
    "ySQQp0oQt1Ep",
    "LL6_Rjg3InL8"
   ],
   "name": "[homework]simple_embeddings.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}