Solution_764b.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "611de0d7",
   "metadata": {},
   "outputs": [],
   "source": [
    "import sys, os\n",
    "from tqdm import tqdm\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.model_selection import train_test_split\n",
    "from torch.utils.data import Dataset, DataLoader\n",
    "import torch\n",
    "import torch.nn.functional as F\n",
    "import torch.nn as nn\n",
    "import math\n",
    "\n",
    "os.environ[\"TOKENIZERS_PARALLELISM\"] = \"false\"\n",
    "\n",
    "def seed_everything(seed):\n",
    "    os.environ['PYTHONHASHSEED'] = str(seed)\n",
    "    np.random.seed(seed)\n",
    "    torch.manual_seed(seed)\n",
    "    torch.cuda.manual_seed(seed)\n",
    "    torch.backends.cudnn.deterministic = True\n",
    "    torch.backends.cudnn.benchmark = False\n",
    "    \n",
    "seed_everything(0)\n",
    "MAX_LEN = 64"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "9b5061ee",
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import AutoConfig, AutoTokenizer, AutoModelForMaskedLM, AutoModel\n",
    "\n",
    "\n",
    "class VecModel(nn.Module):\n",
    "    def __init__(self, model_name, size, has_top=True):\n",
    "        super(VecModel, self).__init__()\n",
    "        conf = AutoConfig.from_pretrained(model_name)\n",
    "        self.backbone = AutoModel.from_config(conf)\n",
    "        \n",
    "        self.has_top = has_top\n",
    "        \n",
    "        if self.has_top:\n",
    "            self.bn = nn.BatchNorm1d(size)\n",
    "            self.top = nn.Linear(size, size)\n",
    "\n",
    "    def forward(self, ids, mask):\n",
    "        out = self.backbone(ids, mask)[0]\n",
    "        out = (out[:, 1:MAX_LEN//2, :]*mask[:, 1:MAX_LEN//2, None]).mean(axis=1)\n",
    "        \n",
    "        if self.has_top:\n",
    "            out = self.top(self.bn(out))\n",
    "        \n",
    "        return F.normalize(out)\n",
    "\n",
    "    def save(self, path):\n",
    "        torch.save(self.state_dict(), path)\n",
    "\n",
    "    def load(self, path):\n",
    "        self.load_state_dict(torch.load(path, map_location='cpu'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "9d65f903",
   "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>topic_id</th>\n",
       "      <th>content_ids</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>t_00004da3a1b2</td>\n",
       "      <td>c_1108dd0c7a5d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>t_00004da3a1b2</td>\n",
       "      <td>c_376c5a8eb028</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>t_00004da3a1b2</td>\n",
       "      <td>c_5bc0e1e2cba0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>t_00004da3a1b2</td>\n",
       "      <td>c_76231f9d0b5e</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>t_00068291e9a4</td>\n",
       "      <td>c_639ea2ef9c95</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>279914</th>\n",
       "      <td>t_fff9e5407d13</td>\n",
       "      <td>c_d64037a72376</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>279915</th>\n",
       "      <td>t_fffbe1d5d43c</td>\n",
       "      <td>c_46f852a49c08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>279916</th>\n",
       "      <td>t_fffbe1d5d43c</td>\n",
       "      <td>c_6659207b25d5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>279917</th>\n",
       "      <td>t_fffe14f1be1e</td>\n",
       "      <td>c_cece166bad6a</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>279918</th>\n",
       "      <td>t_fffe811a6da9</td>\n",
       "      <td>c_92b8fad372ee</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>279919 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "              topic_id     content_ids\n",
       "0       t_00004da3a1b2  c_1108dd0c7a5d\n",
       "1       t_00004da3a1b2  c_376c5a8eb028\n",
       "2       t_00004da3a1b2  c_5bc0e1e2cba0\n",
       "3       t_00004da3a1b2  c_76231f9d0b5e\n",
       "4       t_00068291e9a4  c_639ea2ef9c95\n",
       "...                ...             ...\n",
       "279914  t_fff9e5407d13  c_d64037a72376\n",
       "279915  t_fffbe1d5d43c  c_46f852a49c08\n",
       "279916  t_fffbe1d5d43c  c_6659207b25d5\n",
       "279917  t_fffe14f1be1e  c_cece166bad6a\n",
       "279918  t_fffe811a6da9  c_92b8fad372ee\n",
       "\n",
       "[279919 rows x 2 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "corr_df = pd.read_csv(\"data/correlations.csv\")\n",
    "\n",
    "corr_df[\"content_ids\"] = corr_df[\"content_ids\"].apply(lambda x: x.split())\n",
    "\n",
    "corr_df = corr_df.explode('content_ids').reset_index(drop=True)\n",
    "corr_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "70e52483",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(61517, 9)\n",
      "(61517, 10)\n",
      "(61517, 12)\n",
      "(61517, 14)\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>id</th>\n",
       "      <th>title</th>\n",
       "      <th>description</th>\n",
       "      <th>channel</th>\n",
       "      <th>category</th>\n",
       "      <th>level</th>\n",
       "      <th>language</th>\n",
       "      <th>parent</th>\n",
       "      <th>has_content</th>\n",
       "      <th>parent_title</th>\n",
       "      <th>grandpa</th>\n",
       "      <th>grandpa_title</th>\n",
       "      <th>ggrandpa</th>\n",
       "      <th>ggrandpa_title</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>t_00004da3a1b2</td>\n",
       "      <td>Откриването на резисторите</td>\n",
       "      <td>Изследване на материали, които предизвикват на...</td>\n",
       "      <td>000cf7</td>\n",
       "      <td>source</td>\n",
       "      <td>4</td>\n",
       "      <td>bg</td>\n",
       "      <td>t_16e29365b50d</td>\n",
       "      <td>True</td>\n",
       "      <td>Открития и проекти</td>\n",
       "      <td>t_c85886762db7</td>\n",
       "      <td>Физика</td>\n",
       "      <td>t_27a0a3c7ee9a</td>\n",
       "      <td>Наука</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>t_00068291e9a4</td>\n",
       "      <td>Entradas e saídas de uma função</td>\n",
       "      <td>Entenda um pouco mais sobre funções.</td>\n",
       "      <td>8e286a</td>\n",
       "      <td>source</td>\n",
       "      <td>4</td>\n",
       "      <td>pt</td>\n",
       "      <td>t_d14b6c2a2b70</td>\n",
       "      <td>True</td>\n",
       "      <td>Álgebra: funções</td>\n",
       "      <td>t_461c9a8dd863</td>\n",
       "      <td>9º Ano</td>\n",
       "      <td>t_a6420bcff569</td>\n",
       "      <td>Matemática por ano (Alinhada à BNCC)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>t_00069b63a70a</td>\n",
       "      <td>Transcripts</td>\n",
       "      <td>NaN</td>\n",
       "      <td>6e3ba4</td>\n",
       "      <td>source</td>\n",
       "      <td>3</td>\n",
       "      <td>en</td>\n",
       "      <td>t_4054df11a74e</td>\n",
       "      <td>True</td>\n",
       "      <td>Flow Charts: Logical Thinking?</td>\n",
       "      <td>t_acbbd893e6af</td>\n",
       "      <td>Engineering</td>\n",
       "      <td>t_9c5f09334d83</td>\n",
       "      <td>MIT Blossoms</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>t_0006d41a73a8</td>\n",
       "      <td>Графики на експоненциални функции (Алгебра 2 н...</td>\n",
       "      <td>Научи повече за графиките на сложните показате...</td>\n",
       "      <td>000cf7</td>\n",
       "      <td>source</td>\n",
       "      <td>4</td>\n",
       "      <td>bg</td>\n",
       "      <td>t_e2452e21d252</td>\n",
       "      <td>True</td>\n",
       "      <td>Показателни и логаритмични функции</td>\n",
       "      <td>t_5f4cc8e02423</td>\n",
       "      <td>Алгебра (цялото съдържание)</td>\n",
       "      <td>t_afd8db5e4f44</td>\n",
       "      <td>Математика</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>t_0008768bdee6</td>\n",
       "      <td>100 સુધીનો સરવાળો</td>\n",
       "      <td>37 અને 49 જેવી બે-અંકની સંખ્યાઓ ઉમેરતા શીખો.</td>\n",
       "      <td>5223e0</td>\n",
       "      <td>supplemental</td>\n",
       "      <td>4</td>\n",
       "      <td>gu</td>\n",
       "      <td>t_0da7a331d666</td>\n",
       "      <td>True</td>\n",
       "      <td>100 સુધીના સરવાળા અને બાદબાકી</td>\n",
       "      <td>t_d20fc01fdf38</td>\n",
       "      <td>પ્રારંભિક ગણિત</td>\n",
       "      <td>t_38bca5075c51</td>\n",
       "      <td>ધોરણ ૩</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               id                                              title  \\\n",
       "0  t_00004da3a1b2                         Откриването на резисторите   \n",
       "1  t_00068291e9a4                    Entradas e saídas de uma função   \n",
       "2  t_00069b63a70a                                        Transcripts   \n",
       "3  t_0006d41a73a8  Графики на експоненциални функции (Алгебра 2 н...   \n",
       "4  t_0008768bdee6                                  100 સુધીનો સરવાળો   \n",
       "\n",
       "                                         description channel      category  \\\n",
       "0  Изследване на материали, които предизвикват на...  000cf7        source   \n",
       "1               Entenda um pouco mais sobre funções.  8e286a        source   \n",
       "2                                                NaN  6e3ba4        source   \n",
       "3  Научи повече за графиките на сложните показате...  000cf7        source   \n",
       "4       37 અને 49 જેવી બે-અંકની સંખ્યાઓ ઉમેરતા શીખો.  5223e0  supplemental   \n",
       "\n",
       "   level language          parent  has_content  \\\n",
       "0      4       bg  t_16e29365b50d         True   \n",
       "1      4       pt  t_d14b6c2a2b70         True   \n",
       "2      3       en  t_4054df11a74e         True   \n",
       "3      4       bg  t_e2452e21d252         True   \n",
       "4      4       gu  t_0da7a331d666         True   \n",
       "\n",
       "                         parent_title         grandpa  \\\n",
       "0                  Открития и проекти  t_c85886762db7   \n",
       "1                    Álgebra: funções  t_461c9a8dd863   \n",
       "2      Flow Charts: Logical Thinking?  t_acbbd893e6af   \n",
       "3  Показателни и логаритмични функции  t_5f4cc8e02423   \n",
       "4       100 સુધીના સરવાળા અને બાદબાકી  t_d20fc01fdf38   \n",
       "\n",
       "                 grandpa_title        ggrandpa  \\\n",
       "0                       Физика  t_27a0a3c7ee9a   \n",
       "1                       9º Ano  t_a6420bcff569   \n",
       "2                  Engineering  t_9c5f09334d83   \n",
       "3  Алгебра (цялото съдържание)  t_afd8db5e4f44   \n",
       "4               પ્રારંભિક ગણિત  t_38bca5075c51   \n",
       "\n",
       "                         ggrandpa_title  \n",
       "0                                 Наука  \n",
       "1  Matemática por ano (Alinhada à BNCC)  \n",
       "2                          MIT Blossoms  \n",
       "3                            Математика  \n",
       "4                                ધોરણ ૩  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "topic_df = pd.read_csv(\"data/topics.csv\")\n",
    "topic_df[\"title\"] = topic_df[\"title\"].fillna(\"\")\n",
    "\n",
    "title_map = topic_df.set_index(\"id\")[\"title\"].to_dict()\n",
    "parent_map = topic_df.set_index(\"id\")[\"parent\"].to_dict()\n",
    "\n",
    "topic_df = topic_df[topic_df[\"has_content\"]].reset_index(drop=True)\n",
    "print(topic_df.shape)\n",
    "\n",
    "topic_df[\"parent_title\"] = topic_df[\"parent\"].apply(lambda x: title_map.get(x, \"\"))\n",
    "print(topic_df.shape)\n",
    "\n",
    "topic_df[\"grandpa\"] = topic_df[\"parent\"].apply(lambda x: parent_map.get(x))\n",
    "topic_df[\"grandpa_title\"] = topic_df[\"grandpa\"].apply(lambda x: title_map.get(x, \"\"))\n",
    "print(topic_df.shape)\n",
    "\n",
    "topic_df[\"ggrandpa\"] = topic_df[\"grandpa\"].apply(lambda x: parent_map.get(x))\n",
    "topic_df[\"ggrandpa_title\"] = topic_df[\"ggrandpa\"].apply(lambda x: title_map.get(x, \"\"))\n",
    "print(topic_df.shape)\n",
    "\n",
    "topic_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "7d560d0a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    20506\n",
       "1    20506\n",
       "2    20505\n",
       "Name: fold, dtype: int64"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.model_selection import KFold\n",
    "\n",
    "\n",
    "N_FOLDS = 3\n",
    "\n",
    "splits = KFold(N_FOLDS, shuffle=True, random_state=0).split(topic_df, topic_df[\"has_content\"])\n",
    "\n",
    "topic_df[\"fold\"] = 0\n",
    "\n",
    "for f, (train_ind, val_ind) in enumerate(splits):\n",
    "    topic_df.loc[val_ind, \"fold\"] = f\n",
    "\n",
    "topic_df[\"fold\"].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "5f2a048f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1    53269\n",
       "0     8248\n",
       "Name: fold, dtype: int64"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "topic_df.loc[(topic_df[\"category\"] == \"source\") | (topic_df[\"fold\"] > 0), \"fold\"] = 1\n",
    "topic_df[\"fold\"].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "175a86b8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "41024"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "le_topic = LabelEncoder()\n",
    "topic_df[\"t\"] = topic_df[\"title\"].fillna(\"\") + \" | \" + topic_df[\"description\"].fillna(\"\")\n",
    "topic_df[\"label\"] = le_topic.fit_transform(topic_df[\"t\"])\n",
    "\n",
    "len(le_topic.classes_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "1fd28ef2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(154047, 9)\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>id</th>\n",
       "      <th>title</th>\n",
       "      <th>description</th>\n",
       "      <th>kind</th>\n",
       "      <th>text</th>\n",
       "      <th>language</th>\n",
       "      <th>copyright_holder</th>\n",
       "      <th>license</th>\n",
       "      <th>t</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>c_00002381196d</td>\n",
       "      <td>Sumar números de varios dígitos: 48,029+233,930</td>\n",
       "      <td>Suma 48,029+233,930 mediante el algoritmo está...</td>\n",
       "      <td>video</td>\n",
       "      <td>NaN</td>\n",
       "      <td>es</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Sumar números de varios dígitos: 48,029+233,93...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>c_000087304a9e</td>\n",
       "      <td>Trovare i fattori di un numero</td>\n",
       "      <td>Sal trova i fattori di 120.\\n\\n</td>\n",
       "      <td>video</td>\n",
       "      <td>NaN</td>\n",
       "      <td>it</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Trovare i fattori di un numero | video | Sal t...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>c_0000ad142ddb</td>\n",
       "      <td>Sumar curvas de demanda</td>\n",
       "      <td>Cómo añadir curvas de demanda\\n\\n</td>\n",
       "      <td>video</td>\n",
       "      <td>NaN</td>\n",
       "      <td>es</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Sumar curvas de demanda | video | Cómo añadir ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>c_0000c03adc8d</td>\n",
       "      <td>Nado de aproximação</td>\n",
       "      <td>Neste vídeo você vai aprender o nado de aproxi...</td>\n",
       "      <td>document</td>\n",
       "      <td>\\nNado de aproximação\\nSaber nadar nas ondas ...</td>\n",
       "      <td>pt</td>\n",
       "      <td>Sikana Education</td>\n",
       "      <td>CC BY-NC-ND</td>\n",
       "      <td>Nado de aproximação | document | Neste vídeo v...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>c_00016694ea2a</td>\n",
       "      <td>geometry-m3-topic-a-overview.pdf</td>\n",
       "      <td>Estándares Comunes del Estado de Nueva York\\n\\...</td>\n",
       "      <td>document</td>\n",
       "      <td>Estándares Comunes del Estado de Nueva York\\n\\...</td>\n",
       "      <td>es</td>\n",
       "      <td>Engage NY</td>\n",
       "      <td>CC BY-NC-SA</td>\n",
       "      <td>geometry-m3-topic-a-overview.pdf | document | ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               id                                             title  \\\n",
       "0  c_00002381196d  Sumar números de varios dígitos: 48,029+233,930    \n",
       "1  c_000087304a9e                    Trovare i fattori di un numero   \n",
       "2  c_0000ad142ddb                           Sumar curvas de demanda   \n",
       "3  c_0000c03adc8d                               Nado de aproximação   \n",
       "4  c_00016694ea2a                  geometry-m3-topic-a-overview.pdf   \n",
       "\n",
       "                                         description      kind  \\\n",
       "0  Suma 48,029+233,930 mediante el algoritmo está...     video   \n",
       "1                    Sal trova i fattori di 120.\\n\\n     video   \n",
       "2                  Cómo añadir curvas de demanda\\n\\n     video   \n",
       "3  Neste vídeo você vai aprender o nado de aproxi...  document   \n",
       "4  Estándares Comunes del Estado de Nueva York\\n\\...  document   \n",
       "\n",
       "                                                text language  \\\n",
       "0                                                NaN       es   \n",
       "1                                                NaN       it   \n",
       "2                                                NaN       es   \n",
       "3  \\nNado de aproximação\\nSaber nadar nas ondas ...       pt   \n",
       "4  Estándares Comunes del Estado de Nueva York\\n\\...       es   \n",
       "\n",
       "   copyright_holder      license  \\\n",
       "0               NaN          NaN   \n",
       "1               NaN          NaN   \n",
       "2               NaN          NaN   \n",
       "3  Sikana Education  CC BY-NC-ND   \n",
       "4         Engage NY  CC BY-NC-SA   \n",
       "\n",
       "                                                   t  \n",
       "0  Sumar números de varios dígitos: 48,029+233,93...  \n",
       "1  Trovare i fattori di un numero | video | Sal t...  \n",
       "2  Sumar curvas de demanda | video | Cómo añadir ...  \n",
       "3  Nado de aproximação | document | Neste vídeo v...  \n",
       "4  geometry-m3-topic-a-overview.pdf | document | ...  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "content_df = pd.read_csv(\"data/content.csv\")\n",
    "content_df.loc[content_df[\"title\"] == content_df[\"description\"], \"description\"] = None\n",
    "content_df[\"description\"].fillna(content_df[\"text\"].fillna(\"\").apply(lambda x: x[:256]), inplace=True)\n",
    "content_df[\"t\"] = content_df[\"title\"].fillna(\"\") + \" | \" + content_df[\"kind\"].fillna(\"\") + \" | \" + content_df[\"description\"].fillna(\"\")\n",
    "\n",
    "print(content_df.shape)\n",
    "content_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "6dddd282",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.9733569582391859, 0.9165319675164073)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def extract_number(x, is_subtopic=False):\n",
    "    chapter, rest = x.split(\":\", 1)\n",
    "    \n",
    "    if is_subtopic:\n",
    "        chapter = chapter.split(\".\", 1)[0]\n",
    "        x = f\"{chapter}:{rest}\"\n",
    "    chapter = int(chapter)\n",
    "\n",
    "    return chapter, x\n",
    "\n",
    "\n",
    "def extract_chapters(df, is_subtopic):\n",
    "    titles = df[\"title\"].fillna(\"\").values\n",
    "    \n",
    "    chapters = np.zeros(len(titles))\n",
    "    new_titles = np.array(titles)\n",
    "    \n",
    "    for i, title in enumerate(titles):\n",
    "        try:\n",
    "            chapters[i], new_titles[i] = extract_number(title, is_subtopic)\n",
    "        except:\n",
    "            pass\n",
    "        \n",
    "    df[\"chapter\"], df[\"title\"] = chapters, new_titles\n",
    "    df[\"chapter\"] = df[\"chapter\"].astype(int)\n",
    "    \n",
    "    df.loc[df[\"chapter\"] == 0, \"chapter\"] = None\n",
    "    return df\n",
    "    \n",
    "topic_df = extract_chapters(topic_df, False)\n",
    "content_df = extract_chapters(content_df, True)\n",
    "\n",
    "topic_df[\"chapter\"].isnull().mean(), content_df[\"chapter\"].isnull().mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "08a09a37",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((53269, 18), (8248, 18))"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "TEST_FOLD = 0\n",
    "\n",
    "topic_df[\"t\"] = topic_df[\"title\"] + \" @ \" + topic_df[\"parent_title\"] + \" @ \" + topic_df[\"grandpa_title\"] + \" @ \" + topic_df[\"ggrandpa_title\"] + \" | \" + topic_df[\"description\"].fillna(\"\")\n",
    "\n",
    "topic_df_test = topic_df[topic_df[\"fold\"] == TEST_FOLD].reset_index(drop=True)\n",
    "topic_df_train = topic_df[topic_df[\"fold\"] != TEST_FOLD].reset_index(drop=True)\n",
    "\n",
    "topic_df_train.shape, topic_df_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "3fe9791a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(141107, 2)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_contents = corr_df.merge(topic_df_train[[\"id\"]], left_on=\"topic_id\", right_on=\"id\")[[\"content_ids\"]].drop_duplicates()\n",
    "train_contents.rename(columns={\"content_ids\": \"id\"}, inplace=True)\n",
    "train_contents[\"is_train\"] = True\n",
    "train_contents.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "e14aa955",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9159996624406836"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "content_df = content_df.merge(train_contents, on=\"id\", how=\"left\")\n",
    "content_df[\"is_train\"].fillna(False, inplace=True)\n",
    "content_df[\"is_train\"].mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "87570f15",
   "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>id</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>c_00002381196d</td>\n",
       "      <td>[27817, 27750]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>c_000087304a9e</td>\n",
       "      <td>[13283, 13284]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>c_0000ad142ddb</td>\n",
       "      <td>[11785]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>c_0000c03adc8d</td>\n",
       "      <td>[23664]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>c_00016694ea2a</td>\n",
       "      <td>[21301, 32510]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154042</th>\n",
       "      <td>c_fffcbdd4de8b</td>\n",
       "      <td>[2312]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154043</th>\n",
       "      <td>c_fffe15a2d069</td>\n",
       "      <td>[27466, 27467, 16294]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154044</th>\n",
       "      <td>c_fffed7b0d13a</td>\n",
       "      <td>[22032, 22033, 22034]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154045</th>\n",
       "      <td>c_ffff04ba7ac7</td>\n",
       "      <td>[3455]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154046</th>\n",
       "      <td>c_ffffe5254266</td>\n",
       "      <td>[10100]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>154047 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                    id                  label\n",
       "0       c_00002381196d         [27817, 27750]\n",
       "1       c_000087304a9e         [13283, 13284]\n",
       "2       c_0000ad142ddb                [11785]\n",
       "3       c_0000c03adc8d                [23664]\n",
       "4       c_00016694ea2a         [21301, 32510]\n",
       "...                ...                    ...\n",
       "154042  c_fffcbdd4de8b                 [2312]\n",
       "154043  c_fffe15a2d069  [27466, 27467, 16294]\n",
       "154044  c_fffed7b0d13a  [22032, 22033, 22034]\n",
       "154045  c_ffff04ba7ac7                 [3455]\n",
       "154046  c_ffffe5254266                [10100]\n",
       "\n",
       "[154047 rows x 2 columns]"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#corr_df[\"label\"] = le_topic.transform(corr_df[\"topic_id\"])\n",
    "corr_df = corr_df.merge(topic_df[[\"id\", \"label\"]].rename(columns={\"id\": \"topic_id\"}), on=\"topic_id\")\n",
    "\n",
    "\n",
    "content_labels = corr_df.groupby(\"content_ids\").agg({\"label\": list}).reset_index()\n",
    "content_labels = content_labels.rename(columns={\"content_ids\": \"id\"})\n",
    "content_labels[\"label\"] = content_labels[\"label\"].apply(lambda x: list(set(x)))\n",
    "\n",
    "topic_df_train[\"label\"] = topic_df_train[\"label\"].apply(lambda x: [x])\n",
    "topic_df_test[\"label\"] = topic_df_test[\"label\"].apply(lambda x: [x])\n",
    "\n",
    "content_labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "52c928a4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(154047, 11)\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>id</th>\n",
       "      <th>title</th>\n",
       "      <th>description</th>\n",
       "      <th>kind</th>\n",
       "      <th>text</th>\n",
       "      <th>language</th>\n",
       "      <th>copyright_holder</th>\n",
       "      <th>license</th>\n",
       "      <th>t</th>\n",
       "      <th>chapter</th>\n",
       "      <th>is_train</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>c_00002381196d</td>\n",
       "      <td>Sumar números de varios dígitos: 48,029+233,930</td>\n",
       "      <td>Suma 48,029+233,930 mediante el algoritmo está...</td>\n",
       "      <td>video</td>\n",
       "      <td>NaN</td>\n",
       "      <td>es</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Sumar números de varios dígitos: 48,029+233,93...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>True</td>\n",
       "      <td>[27817, 27750]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>c_000087304a9e</td>\n",
       "      <td>Trovare i fattori di un numero</td>\n",
       "      <td>Sal trova i fattori di 120.\\n\\n</td>\n",
       "      <td>video</td>\n",
       "      <td>NaN</td>\n",
       "      <td>it</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Trovare i fattori di un numero | video | Sal t...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>True</td>\n",
       "      <td>[13283, 13284]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>c_0000ad142ddb</td>\n",
       "      <td>Sumar curvas de demanda</td>\n",
       "      <td>Cómo añadir curvas de demanda\\n\\n</td>\n",
       "      <td>video</td>\n",
       "      <td>NaN</td>\n",
       "      <td>es</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Sumar curvas de demanda | video | Cómo añadir ...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>True</td>\n",
       "      <td>[11785]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>c_0000c03adc8d</td>\n",
       "      <td>Nado de aproximação</td>\n",
       "      <td>Neste vídeo você vai aprender o nado de aproxi...</td>\n",
       "      <td>document</td>\n",
       "      <td>\\nNado de aproximação\\nSaber nadar nas ondas ...</td>\n",
       "      <td>pt</td>\n",
       "      <td>Sikana Education</td>\n",
       "      <td>CC BY-NC-ND</td>\n",
       "      <td>Nado de aproximação | document | Neste vídeo v...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>True</td>\n",
       "      <td>[23664]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>c_00016694ea2a</td>\n",
       "      <td>geometry-m3-topic-a-overview.pdf</td>\n",
       "      <td>Estándares Comunes del Estado de Nueva York\\n\\...</td>\n",
       "      <td>document</td>\n",
       "      <td>Estándares Comunes del Estado de Nueva York\\n\\...</td>\n",
       "      <td>es</td>\n",
       "      <td>Engage NY</td>\n",
       "      <td>CC BY-NC-SA</td>\n",
       "      <td>geometry-m3-topic-a-overview.pdf | document | ...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>True</td>\n",
       "      <td>[21301, 32510]</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154042</th>\n",
       "      <td>c_fffcbdd4de8b</td>\n",
       "      <td>2: Diffusion</td>\n",
       "      <td>What will eventually happen to these dyes?\\n\\n...</td>\n",
       "      <td>html5</td>\n",
       "      <td>What will eventually happen to these dyes?\\n\\n...</td>\n",
       "      <td>en</td>\n",
       "      <td>CSU and Merlot</td>\n",
       "      <td>CC BY-NC-SA</td>\n",
       "      <td>2. 12: Diffusion | html5 | What will eventuall...</td>\n",
       "      <td>2.0</td>\n",
       "      <td>True</td>\n",
       "      <td>[2312]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154043</th>\n",
       "      <td>c_fffe15a2d069</td>\n",
       "      <td>Sommare facendo gruppi da 10</td>\n",
       "      <td>Sal somma 5+68 spezzando il 5 in un 2 e un 3.\\n\\n</td>\n",
       "      <td>video</td>\n",
       "      <td>NaN</td>\n",
       "      <td>it</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Sommare facendo gruppi da 10 | video | Sal som...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>True</td>\n",
       "      <td>[27466, 27467, 16294]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154044</th>\n",
       "      <td>c_fffed7b0d13a</td>\n",
       "      <td>Introdução à subtração</td>\n",
       "      <td>Sal fala sobre o que significa subtrair. Os ex...</td>\n",
       "      <td>video</td>\n",
       "      <td>NaN</td>\n",
       "      <td>pt</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Introdução à subtração | video | Sal fala sobr...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>True</td>\n",
       "      <td>[22032, 22033, 22034]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154045</th>\n",
       "      <td>c_ffff04ba7ac7</td>\n",
       "      <td>SA of a Cone</td>\n",
       "      <td></td>\n",
       "      <td>video</td>\n",
       "      <td>NaN</td>\n",
       "      <td>en</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>SA of a Cone | video |</td>\n",
       "      <td>NaN</td>\n",
       "      <td>True</td>\n",
       "      <td>[3455]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154046</th>\n",
       "      <td>c_ffffe5254266</td>\n",
       "      <td>The Jats</td>\n",
       "      <td></td>\n",
       "      <td>video</td>\n",
       "      <td>NaN</td>\n",
       "      <td>en</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>The Jats | video |</td>\n",
       "      <td>NaN</td>\n",
       "      <td>True</td>\n",
       "      <td>[10100]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>154047 rows × 12 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                    id                                             title  \\\n",
       "0       c_00002381196d  Sumar números de varios dígitos: 48,029+233,930    \n",
       "1       c_000087304a9e                    Trovare i fattori di un numero   \n",
       "2       c_0000ad142ddb                           Sumar curvas de demanda   \n",
       "3       c_0000c03adc8d                               Nado de aproximação   \n",
       "4       c_00016694ea2a                  geometry-m3-topic-a-overview.pdf   \n",
       "...                ...                                               ...   \n",
       "154042  c_fffcbdd4de8b                                      2: Diffusion   \n",
       "154043  c_fffe15a2d069                      Sommare facendo gruppi da 10   \n",
       "154044  c_fffed7b0d13a                            Introdução à subtração   \n",
       "154045  c_ffff04ba7ac7                                      SA of a Cone   \n",
       "154046  c_ffffe5254266                                          The Jats   \n",
       "\n",
       "                                              description      kind  \\\n",
       "0       Suma 48,029+233,930 mediante el algoritmo está...     video   \n",
       "1                         Sal trova i fattori di 120.\\n\\n     video   \n",
       "2                       Cómo añadir curvas de demanda\\n\\n     video   \n",
       "3       Neste vídeo você vai aprender o nado de aproxi...  document   \n",
       "4       Estándares Comunes del Estado de Nueva York\\n\\...  document   \n",
       "...                                                   ...       ...   \n",
       "154042  What will eventually happen to these dyes?\\n\\n...     html5   \n",
       "154043  Sal somma 5+68 spezzando il 5 in un 2 e un 3.\\n\\n     video   \n",
       "154044  Sal fala sobre o que significa subtrair. Os ex...     video   \n",
       "154045                                                        video   \n",
       "154046                                                        video   \n",
       "\n",
       "                                                     text language  \\\n",
       "0                                                     NaN       es   \n",
       "1                                                     NaN       it   \n",
       "2                                                     NaN       es   \n",
       "3       \\nNado de aproximação\\nSaber nadar nas ondas ...       pt   \n",
       "4       Estándares Comunes del Estado de Nueva York\\n\\...       es   \n",
       "...                                                   ...      ...   \n",
       "154042  What will eventually happen to these dyes?\\n\\n...       en   \n",
       "154043                                                NaN       it   \n",
       "154044                                                NaN       pt   \n",
       "154045                                                NaN       en   \n",
       "154046                                                NaN       en   \n",
       "\n",
       "        copyright_holder      license  \\\n",
       "0                    NaN          NaN   \n",
       "1                    NaN          NaN   \n",
       "2                    NaN          NaN   \n",
       "3       Sikana Education  CC BY-NC-ND   \n",
       "4              Engage NY  CC BY-NC-SA   \n",
       "...                  ...          ...   \n",
       "154042    CSU and Merlot  CC BY-NC-SA   \n",
       "154043               NaN          NaN   \n",
       "154044               NaN          NaN   \n",
       "154045               NaN          NaN   \n",
       "154046               NaN          NaN   \n",
       "\n",
       "                                                        t  chapter  is_train  \\\n",
       "0       Sumar números de varios dígitos: 48,029+233,93...      NaN      True   \n",
       "1       Trovare i fattori di un numero | video | Sal t...      NaN      True   \n",
       "2       Sumar curvas de demanda | video | Cómo añadir ...      NaN      True   \n",
       "3       Nado de aproximação | document | Neste vídeo v...      NaN      True   \n",
       "4       geometry-m3-topic-a-overview.pdf | document | ...      NaN      True   \n",
       "...                                                   ...      ...       ...   \n",
       "154042  2. 12: Diffusion | html5 | What will eventuall...      2.0      True   \n",
       "154043  Sommare facendo gruppi da 10 | video | Sal som...      NaN      True   \n",
       "154044  Introdução à subtração | video | Sal fala sobr...      NaN      True   \n",
       "154045                            SA of a Cone | video |       NaN      True   \n",
       "154046                                The Jats | video |       NaN      True   \n",
       "\n",
       "                        label  \n",
       "0              [27817, 27750]  \n",
       "1              [13283, 13284]  \n",
       "2                     [11785]  \n",
       "3                     [23664]  \n",
       "4              [21301, 32510]  \n",
       "...                       ...  \n",
       "154042                 [2312]  \n",
       "154043  [27466, 27467, 16294]  \n",
       "154044  [22032, 22033, 22034]  \n",
       "154045                 [3455]  \n",
       "154046                [10100]  \n",
       "\n",
       "[154047 rows x 12 columns]"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(content_df.shape)\n",
    "\n",
    "content_df = content_df.merge(content_labels, on=\"id\") #, how=\"left\")\n",
    "content_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "4908271c",
   "metadata": {},
   "outputs": [],
   "source": [
    "MODEL_NAME = \"xlm-roberta-large\"\n",
    "\n",
    "roberta_tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "cd222b11",
   "metadata": {},
   "outputs": [],
   "source": [
    "MODEL_NAME = \"bert-base-multilingual-uncased\"\n",
    "\n",
    "bert_tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "4f84911e",
   "metadata": {},
   "outputs": [],
   "source": [
    "NW = 8\n",
    "BS = 64\n",
    "\n",
    "class LECRDataset(Dataset):\n",
    "    def __init__(self, df, max_len=512):\n",
    "        self.df = df.reset_index(drop=True)\n",
    "        self.max_len = max_len\n",
    "\n",
    "    def __len__(self):\n",
    "        return self.df.shape[0]\n",
    "\n",
    "    def __getitem__(self, index):\n",
    "        row = self.df.iloc[index]\n",
    "        encoding = roberta_tokenizer(str(row.t), padding='max_length', \n",
    "                                  truncation=True, max_length=self.max_len)\n",
    "\n",
    "        encoding = {key: torch.as_tensor(val) for key, val in encoding.items()}\n",
    "\n",
    "        encoding2 = bert_tokenizer(str(row.t), padding='max_length', \n",
    "                              truncation=True, max_length=self.max_len)\n",
    "        encoding2 = {f\"bert_{key}\": torch.as_tensor(val) for key, val in encoding2.items()}\n",
    "        \n",
    "        encoding.update(encoding2)\n",
    "\n",
    "        return encoding\n",
    "    \n",
    "model_w = [0.3, 0.3, 0.4]\n",
    "\n",
    "models = [VecModel(\"bert-base-multilingual-uncased\", 768, has_top=False),\n",
    "          VecModel(\"xlm-roberta-base\", 768, has_top=False), \n",
    "          VecModel(\"xlm-roberta-large\", 1024)]\n",
    "for model in models:\n",
    "    model.cuda()\n",
    "    model.eval()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "e59400c0",
   "metadata": {},
   "outputs": [],
   "source": [
    "models[0].load(\"models/vec_model_v36bert.pth\")\n",
    "models[1].load(\"models/vec_model_v36.pth\")\n",
    "models[2].load(\"models/vec_model_v35large.pth\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "0e0c701b",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.neighbors import NearestNeighbors\n",
    "\n",
    "\n",
    "def to_vec(models, ds):\n",
    "    val_loader = DataLoader(ds, batch_size=BS, shuffle=False, num_workers=NW,\n",
    "                             pin_memory=False, drop_last=False)\n",
    "\n",
    "    tbar = tqdm(val_loader, file=sys.stdout)\n",
    "\n",
    "    vectors = []\n",
    "\n",
    "    with torch.no_grad():\n",
    "        for idx, data in enumerate(tbar):\n",
    "            ids, mask = data[\"input_ids\"].cuda(), data[\"attention_mask\"].cuda()\n",
    "            bert_ids, bert_mask = data[\"bert_input_ids\"].cuda(), data[\"bert_attention_mask\"].cuda()\n",
    "            \n",
    "            vec_list = [model_w[0]*models[0](bert_ids, bert_mask)]\n",
    "            for i in range(1, 3):\n",
    "                vec_list.append(model_w[i]*models[i](ids, mask))\n",
    "                \n",
    "\n",
    "            vec = torch.cat(vec_list, axis=1)\n",
    "            vectors.append(vec.detach().cpu().numpy())\n",
    "\n",
    "    V = np.concatenate(vectors)\n",
    "    \n",
    "    return V\n",
    "\n",
    "\n",
    "\n",
    "def get_matches(V_topic, V_content, topic_ids, content_ids, n_neighbors=5):\n",
    "    \n",
    "    neighbors_model = NearestNeighbors(n_neighbors=n_neighbors, metric='cosine', n_jobs=-1)\n",
    "    neighbors_model.fit(V_content)\n",
    "    dists, indices = neighbors_model.kneighbors(V_topic)\n",
    "    \n",
    "    res_df = pd.DataFrame({\"topic_id\": np.repeat(topic_ids, n_neighbors),\n",
    "                           'content_id': content_ids[indices.ravel()],\n",
    "                           'vec_dist': dists.ravel()\n",
    "                          })\n",
    "    return res_df\n",
    "\n",
    "\n",
    "def f2(recall, precision):\n",
    "    return 5*precision*recall/(4*precision + recall)\n",
    "\n",
    "def metric(pred_df):\n",
    "    scores = np.zeros(pred_df.shape[0])\n",
    "\n",
    "    for i, (gt, pred) in enumerate(zip(pred_df[\"content_ids\"].values, pred_df[\"pred\"].values)):\n",
    "        hits = 0\n",
    "\n",
    "        gt, pred = gt.split(), pred.split()\n",
    "\n",
    "        for l in gt:\n",
    "            if l in pred:\n",
    "                hits += 1\n",
    "\n",
    "        if hits > 0:           \n",
    "            recall = hits / len(gt)\n",
    "            precision = hits / len(pred)\n",
    "\n",
    "            scores[i] = f2(recall, precision)\n",
    "\n",
    "    return scores.mean()\n",
    "    \n",
    "\n",
    "def evaluate(topic_df, content_df):    \n",
    "    topic_ds = LECRDataset(topic_df, tokenizer, MAX_LEN, aug=False)\n",
    "    content_ds = LECRDataset(content_df, tokenizer, MAX_LEN, aug=False)\n",
    "\n",
    "\n",
    "    V_topic = to_vec(model, topic_ds)\n",
    "    V_content = to_vec(model, content_ds)\n",
    "\n",
    "    res_df = get_matches(V_topic, V_content, topic_df[\"id\"].values, content_df[\"id\"].values)\n",
    "\n",
    "    res_df[\"rank\"] = res_df.groupby(\"topic_id\")[\"vec_dist\"].rank(method=\"first\", ascending=True)\n",
    "\n",
    "    pred_df = res_df.copy() #res_df[(res_df[\"rank\"] == 1) | (res_df[\"vec_dist\"] < 0.1)]\n",
    "    pred_df = pred_df.groupby(\"topic_id\")[\"content_id\"].apply(lambda x: \" \".join(list(x)))\n",
    "    pred_df = pred_df.reset_index().rename(columns={\"content_id\": 'pred'})\n",
    "    pred_df = pred_df.merge(pd.read_csv(\"data/correlations.csv\"), on=\"topic_id\")\n",
    "\n",
    "    return metric(pred_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "e09dc327",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████████████| 129/129 [00:25<00:00,  4.96it/s]\n",
      "100%|███████████████████████████████████████| 2407/2407 [07:44<00:00,  5.18it/s]\n"
     ]
    }
   ],
   "source": [
    "topic_ds = LECRDataset(topic_df_test, MAX_LEN)\n",
    "content_ds = LECRDataset(content_df, MAX_LEN)\n",
    "\n",
    "\n",
    "V_topic = to_vec(models, topic_ds)\n",
    "V_content = to_vec(models, content_ds)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "4e421798",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(164960, 3)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "res_dfs = []\n",
    "\n",
    "for lang in topic_df_test[\"language\"].unique():\n",
    "    topic_ix = np.where(topic_df_test[\"language\"] == lang)[0]\n",
    "    content_ix = np.where(content_df[\"language\"] == lang)[0]\n",
    "    \n",
    "    res_dfs.append(get_matches(V_topic[topic_ix], V_content[content_ix], \n",
    "                               topic_df_test[\"id\"].values[topic_ix], content_df[\"id\"].values[content_ix],\n",
    "                    n_neighbors=20))\n",
    "res_df = pd.concat(res_dfs)\n",
    "res_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "0c6cd8a8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████████████| 833/833 [02:39<00:00,  5.22it/s]\n"
     ]
    }
   ],
   "source": [
    "topic_ds = LECRDataset(topic_df_train, MAX_LEN)\n",
    "\n",
    "V_topic_train = to_vec(models, topic_ds)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "181bf8a0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def get_max_train(df, V):\n",
    "    max_train_scores = np.zeros(df.shape[0])\n",
    "    matches = np.zeros(df.shape[0])\n",
    "\n",
    "    for lang in df[\"language\"].unique():\n",
    "        topic_ix = np.where(topic_df_train[\"language\"] == lang)[0]\n",
    "        ix = np.where(df[\"language\"] == lang)[0]\n",
    "\n",
    "        neighbors_model = NearestNeighbors(n_neighbors=1, metric='cosine', n_jobs=4)\n",
    "        neighbors_model.fit(V_topic_train[topic_ix])\n",
    "        dists, idx = neighbors_model.kneighbors(V[ix])\n",
    "\n",
    "        max_train_scores[ix] = dists.ravel()\n",
    "        matches[ix] = topic_ix[idx.ravel()]  \n",
    "        \n",
    "    return max_train_scores, matches\n",
    "\n",
    "\n",
    "content_max_train_scores, _ = get_max_train(content_df, V_content)\n",
    "    \n",
    "content_df[\"max_train_score\"] = content_max_train_scores\n",
    "content_df[\"max_train_score\"].hist(bins=50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "a6f7b184",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "topic_max_train_scores, topic_matches = get_max_train(topic_df_test, V_topic)\n",
    "    \n",
    "topic_df_test[\"max_train_score\"] = topic_max_train_scores\n",
    "topic_df_test[\"matched_topic\"] = topic_df_train[\"id\"].values[topic_matches.astype(int)]\n",
    "\n",
    "topic_df_test[\"max_train_score\"].hist(bins=50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "16e991d1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(35979, 3)"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "second_degree_match_df = topic_df_test.merge(corr_df, left_on=\"matched_topic\", right_on=\"topic_id\")[[\"id\", \"content_ids\"]]\n",
    "second_degree_match_df[\"second_degree\"] = True\n",
    "second_degree_match_df.rename(columns={\"id\": \"topic_id\", \"content_ids\": \"content_id\"}, inplace=True)\n",
    "second_degree_match_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "b869bb41",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "en\n",
      "136622 28053 65939\n",
      "----\n",
      "es\n",
      "81953 11769 30844\n",
      "----\n",
      "gu\n",
      "30561 1809 3677\n",
      "----\n",
      "hi\n",
      "37631 1373 4042\n",
      "----\n",
      "bn\n",
      "23102 1731 2513\n",
      "----\n",
      "fr\n",
      "33965 3034 10682\n",
      "----\n",
      "pt\n",
      "37485 3425 10435\n",
      "----\n",
      "fil\n",
      "4291 224 516\n",
      "----\n",
      "as\n",
      "9677 126 641\n",
      "----\n",
      "sw\n",
      "21455 2082 1447\n",
      "----\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(164960, 3)"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "from sparse_dot_topn import awesome_cossim_topn\n",
    "\n",
    "\n",
    "def get_matches(topic_df, content_df):\n",
    "    tfidf = TfidfVectorizer(analyzer=\"char\", ngram_range=(4, 4), min_df=2)\n",
    "\n",
    "    tfidf.fit(pd.concat([content_df[\"t\"], topic_df[\"t\"]]))\n",
    "    print(len(tfidf.vocabulary_), topic_df.shape[0], content_df.shape[0])\n",
    "    \n",
    "    topic_df = topic_df[topic_df[\"fold\"] == TEST_FOLD].reset_index(drop=True)\n",
    "    \n",
    "    V_topic = tfidf.transform(topic_df[\"title\"] + \" \" + topic_df[\"t\"])\n",
    "    V_content = tfidf.transform(content_df[\"title\"] + \" \" + content_df[\"t\"])\n",
    "\n",
    "    d = awesome_cossim_topn(V_topic, V_content.T, 20, 0.01, use_threads=True, n_jobs=4)\n",
    "\n",
    "    non_zeros = d.nonzero()\n",
    "\n",
    "    name_indices = non_zeros[0]\n",
    "    gt_indices = non_zeros[1]\n",
    "\n",
    "    content_ids = content_df[\"id\"].values\n",
    "    topic_ids = topic_df[\"id\"].values\n",
    "\n",
    "    left_side = np.empty(gt_indices.size, dtype=object)\n",
    "    right_side = np.empty(gt_indices.size, dtype=object)\n",
    "    match_score = np.zeros(gt_indices.size)\n",
    "\n",
    "    for index in range(gt_indices.size):\n",
    "        left_side[index] = topic_ids[name_indices[index]]\n",
    "        right_side[index] = content_ids[gt_indices[index]]\n",
    "        match_score[index] = d.data[index]\n",
    "\n",
    "    res_df = pd.DataFrame({\"topic_id\": left_side,\n",
    "                           'content_id': right_side,\n",
    "                           'match_score': match_score\n",
    "                          })\n",
    "    return res_df\n",
    "\n",
    "\n",
    "res_dfs = []\n",
    "\n",
    "for lang in topic_df[topic_df[\"fold\"] == TEST_FOLD][\"language\"].unique():\n",
    "    print(lang)\n",
    "    content_df_lang = content_df[(content_df[\"language\"] == lang)].reset_index(drop=True)\n",
    "    topic_df_lang = topic_df[(topic_df[\"language\"] == lang)].reset_index(drop=True)\n",
    "    \n",
    "    if content_df_lang.shape[0] > 0:\n",
    "        res_dfs.append(get_matches(topic_df_lang, content_df_lang))\n",
    "        \n",
    "    print(\"----\")\n",
    "    \n",
    "res_df2 = pd.concat(res_dfs)\n",
    "res_df2.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "bcdacf37",
   "metadata": {},
   "outputs": [],
   "source": [
    "corr_df[\"target\"] = 1\n",
    "corr_df.rename(columns={\"content_ids\": \"content_id\"}, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "5b92a024",
   "metadata": {},
   "outputs": [],
   "source": [
    "topic_df.columns = [f\"topic_{col}\" for col in topic_df.columns]\n",
    "content_df.columns = [f\"content_{col}\" for col in content_df.columns]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "a8e4a481",
   "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>topic_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>dup_count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>t_0016d30772f3</td>\n",
       "      <td>c_061d9f90bb06</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>t_0016d30772f3</td>\n",
       "      <td>c_242ddc729eec</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>t_0016d30772f3</td>\n",
       "      <td>c_61b851222e17</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>t_0016d30772f3</td>\n",
       "      <td>c_b4fa764f5218</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>t_0016d30772f3</td>\n",
       "      <td>c_c25053d6fafd</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>385495</th>\n",
       "      <td>t_ffa56a59eb42</td>\n",
       "      <td>c_6d9c9523ad59</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>385496</th>\n",
       "      <td>t_ffba5459a977</td>\n",
       "      <td>c_04a421dba8aa</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>385497</th>\n",
       "      <td>t_ffba5459a977</td>\n",
       "      <td>c_787a7a2e7217</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>385498</th>\n",
       "      <td>t_ffba5459a977</td>\n",
       "      <td>c_a46e0ec1377b</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>385499</th>\n",
       "      <td>t_ffba5459a977</td>\n",
       "      <td>c_a6c82a21c03a</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>385500 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "              topic_id      content_id  dup_count\n",
       "0       t_0016d30772f3  c_061d9f90bb06          2\n",
       "1       t_0016d30772f3  c_242ddc729eec          2\n",
       "2       t_0016d30772f3  c_61b851222e17          3\n",
       "3       t_0016d30772f3  c_b4fa764f5218          1\n",
       "4       t_0016d30772f3  c_c25053d6fafd          3\n",
       "...                ...             ...        ...\n",
       "385495  t_ffa56a59eb42  c_6d9c9523ad59          1\n",
       "385496  t_ffba5459a977  c_04a421dba8aa          1\n",
       "385497  t_ffba5459a977  c_787a7a2e7217          1\n",
       "385498  t_ffba5459a977  c_a46e0ec1377b          1\n",
       "385499  t_ffba5459a977  c_a6c82a21c03a          1\n",
       "\n",
       "[385500 rows x 3 columns]"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dup_df = []\n",
    "\n",
    "topic_df[\"key\"] = topic_df[\"topic_title\"].fillna(\"\") + \" | \" + topic_df[\"topic_description\"].fillna(\"\")\n",
    "\n",
    "\n",
    "train_df, val_df = topic_df[topic_df[\"topic_fold\"] != TEST_FOLD].reset_index(drop=True), topic_df[topic_df[\"topic_fold\"] == TEST_FOLD].reset_index(drop=True)\n",
    "\n",
    "\n",
    "lookup = train_df.merge(corr_df, on=\"topic_id\").groupby([\"key\", \"content_id\"])[\"topic_channel\"].count().reset_index()\n",
    "lookup.rename(columns={\"topic_channel\": \"dup_count\"}, inplace=True)\n",
    "\n",
    "\n",
    "dup_df.append(val_df[[\"topic_id\", \"key\"]].merge(lookup, on=[\"key\"])[[\"topic_id\", 'content_id', \"dup_count\"]])\n",
    "    \n",
    "dup_df = pd.concat(dup_df).reset_index(drop=True)\n",
    "dup_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "e7a203de",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(658256, 6)"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "res_df = res_df.merge(res_df2, how=\"outer\", on=[\"topic_id\", \"content_id\"])\n",
    "res_df = res_df.merge(dup_df, how=\"outer\", on=[\"topic_id\", \"content_id\"])\n",
    "res_df = res_df.merge(second_degree_match_df, how=\"outer\", on=[\"topic_id\", \"content_id\"])\n",
    "\n",
    "res_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "094ff50b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(730924, 7)"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dup_df = []\n",
    "\n",
    "topic_df[\"key\"] = topic_df[\"topic_title\"].fillna(\"\")\n",
    "\n",
    "\n",
    "train_df, val_df = topic_df[topic_df[\"topic_fold\"] != TEST_FOLD].reset_index(drop=True), topic_df[topic_df[\"topic_fold\"] == TEST_FOLD].reset_index(drop=True)\n",
    "\n",
    "\n",
    "lookup = train_df.merge(corr_df, on=\"topic_id\").groupby([\"key\", \"content_id\"])[\"topic_channel\"].count().reset_index()\n",
    "lookup.rename(columns={\"topic_channel\": \"tdup_count\"}, inplace=True)\n",
    "\n",
    "\n",
    "dup_df.append(val_df[[\"topic_id\", \"key\"]].merge(lookup, on=[\"key\"])[[\"topic_id\", 'content_id', \"tdup_count\"]])\n",
    "    \n",
    "dup_df = pd.concat(dup_df).reset_index(drop=True)\n",
    "\n",
    "\n",
    "res_df = res_df.merge(dup_df, how=\"outer\", on=[\"topic_id\", \"content_id\"])\n",
    "res_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "ba8e166c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(838146, 8)"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dup_df = []\n",
    "\n",
    "topic_df[\"key\"] = topic_df[\"topic_parent\"].fillna(topic_df[\"topic_id\"])\n",
    "\n",
    "train_df, val_df = topic_df[topic_df[\"topic_fold\"] != TEST_FOLD].reset_index(drop=True), topic_df[topic_df[\"topic_fold\"] == TEST_FOLD].reset_index(drop=True)\n",
    "\n",
    "\n",
    "lookup = train_df.merge(corr_df, on=\"topic_id\").groupby([\"key\", \"content_id\"])[\"topic_channel\"].count().reset_index()\n",
    "lookup.rename(columns={\"topic_channel\": \"pdup_count\"}, inplace=True)\n",
    "\n",
    "\n",
    "dup_df.append(val_df[[\"topic_id\", \"key\"]].merge(lookup, on=[\"key\"])[[\"topic_id\", 'content_id', \"pdup_count\"]])\n",
    "\n",
    "dup_df = pd.concat(dup_df).reset_index(drop=True)\n",
    "\n",
    "\n",
    "res_df = res_df.merge(dup_df, how=\"outer\", on=[\"topic_id\", \"content_id\"])\n",
    "res_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "e27a5aca",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8926849851546789"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def best_case_metric(pred_df):\n",
    "    scores = np.zeros(pred_df.shape[0])\n",
    "\n",
    "    for i, (gt, pred) in enumerate(zip(pred_df[\"content_ids\"].values, pred_df[\"pred\"].values)):\n",
    "        hits = 0\n",
    "\n",
    "        gt, pred = gt.split(), pred.split()\n",
    "\n",
    "        for l in gt:\n",
    "            if l in pred:\n",
    "                hits += 1\n",
    "\n",
    "        if hits > 0:           \n",
    "            recall = hits / len(gt)\n",
    "            precision = 1 #hits / len(pred)\n",
    "\n",
    "            scores[i] = f2(recall, precision)\n",
    "\n",
    "    return scores.mean()\n",
    "\n",
    "\n",
    "pred_df = res_df.groupby(\"topic_id\")[\"content_id\"].apply(lambda x: \" \".join(list(x)))\n",
    "pred_df = pred_df.reset_index().rename(columns={\"content_id\": 'pred'})\n",
    "pred_df = pred_df.merge(pd.read_csv(\"data/correlations.csv\"), on=\"topic_id\")\n",
    "\n",
    "best_case_metric(pred_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "10fbf59a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(838146, 38)"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "res_df = res_df.merge(topic_df, on=\"topic_id\", how=\"left\")\n",
    "res_df = res_df.merge(content_df, on=\"content_id\", how=\"left\")\n",
    "\n",
    "res_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "0720ca6a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.03136923638602344, 26292.0)"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "res_df = res_df.merge(corr_df[[\"topic_id\", \"content_id\", \"target\"]], on=[\"topic_id\", \"content_id\"], how=\"left\")\n",
    "res_df[\"target\"].fillna(0, inplace=True)\n",
    "\n",
    "res_df[\"target\"].mean(), res_df[\"target\"].sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "1dce824f",
   "metadata": {},
   "outputs": [],
   "source": [
    "res_df[\"topic_language\"] = res_df[\"topic_language\"].astype(\"category\")\n",
    "res_df[\"topic_category\"] = res_df[\"topic_category\"].astype(\"category\")\n",
    "res_df[\"content_kind\"] = res_df[\"content_kind\"].astype(\"category\")\n",
    "res_df[\"topic_channel\"] = res_df[\"topic_channel\"].astype(\"category\")\n",
    "\n",
    "res_df[\"len_topic_title\"] = res_df[\"topic_title\"].fillna(\"\").apply(len)\n",
    "res_df[\"len_topic_description\"] = res_df[\"topic_description\"].fillna(\"\").apply(len)\n",
    "res_df[\"len_content_title\"] = res_df[\"content_title\"].fillna(\"\").apply(len)\n",
    "res_df[\"len_content_description\"] = res_df[\"content_description\"].fillna(\"\").apply(len)\n",
    "res_df[\"len_content_text\"] = res_df[\"content_text\"].fillna(\"\").apply(len)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "e7306e51",
   "metadata": {},
   "outputs": [],
   "source": [
    "res_df[\"vec_dist_max\"] = res_df.groupby(\"topic_id\")[\"vec_dist\"].transform(\"max\")\n",
    "res_df[\"vec_dist_min\"] = res_df.groupby(\"topic_id\")[\"vec_dist\"].transform(\"min\")\n",
    "\n",
    "res_df[\"dup_count\"] = res_df[\"dup_count\"].fillna(0)\n",
    "res_df[\"total_count\"] = res_df.groupby(\"topic_id\")[\"content_id\"].transform(\"count\")\n",
    "res_df[\"dup_count_mean\"] = res_df.groupby(\"topic_id\")[\"dup_count\"].transform(\"mean\")\n",
    "\n",
    "res_df[\"tdup_count\"] = res_df[\"tdup_count\"].fillna(0)\n",
    "res_df[\"tdup_count_mean\"] = res_df.groupby(\"topic_id\")[\"tdup_count\"].transform(\"mean\")\n",
    "\n",
    "res_df[\"pdup_count\"] = res_df[\"pdup_count\"].fillna(0)\n",
    "res_df[\"pdup_count_mean\"] = res_df.groupby(\"topic_id\")[\"pdup_count\"].transform(\"mean\")\n",
    "\n",
    "res_df[\"same_chapter\"] = res_df[\"topic_chapter\"] == res_df[\"content_chapter\"]\n",
    "res_df[\"starts_same\"] = res_df[\"topic_title\"].apply(lambda x: x.split(\" \", 1)[0]) == res_df[\"content_title\"].apply(lambda x: x.split(\" \", 1)[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "99c7da5d",
   "metadata": {},
   "outputs": [],
   "source": [
    "res_df.loc[~res_df[\"content_is_train\"], \"content_max_train_score\"] = None\n",
    "\n",
    "res_df[\"second_degree\"].fillna(False, inplace=True)\n",
    "res_df[\"topic_max_train_score\"] = res_df[\"topic_id\"].map(topic_df_test.set_index(\"id\")[\"max_train_score\"].to_dict())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "b215c127",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fold 0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/aerdem/venv/lib/python3.8/site-packages/lightgbm/engine.py:239: UserWarning: 'verbose_eval' argument is deprecated and will be removed in a future release of LightGBM. Pass 'log_evaluation()' callback via 'callbacks' argument instead.\n",
      "  _log_warning(\"'verbose_eval' argument is deprecated and will be removed in a future release of LightGBM. \"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Info] Number of positive: 19846, number of negative: 608763\n",
      "[LightGBM] [Warning] Auto-choosing row-wise multi-threading, the overhead of testing was 0.009678 seconds.\n",
      "You can set `force_row_wise=true` to remove the overhead.\n",
      "And if memory is not enough, you can set `force_col_wise=true`.\n",
      "[LightGBM] [Info] Total Bins 3718\n",
      "[LightGBM] [Info] Number of data points in the train set: 628609, number of used features: 27\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/aerdem/venv/lib/python3.8/site-packages/lightgbm/basic.py:1780: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/home/aerdem/venv/lib/python3.8/site-packages/lightgbm/basic.py:1513: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning(f'{cat_alias} in param dict is overridden.')\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Info] [binary:BoostFromScore]: pavg=0.031571 -> initscore=-3.423427\n",
      "[LightGBM] [Info] Start training from score -3.423427\n",
      "[200]\tvalid_0's auc: 0.991417\n",
      "[400]\tvalid_0's auc: 0.992422\n",
      "[600]\tvalid_0's auc: 0.992663\n",
      "[800]\tvalid_0's auc: 0.992792\n",
      "[1000]\tvalid_0's auc: 0.992825\n",
      "[1200]\tvalid_0's auc: 0.992867\n",
      "[1400]\tvalid_0's auc: 0.992869\n",
      "[1600]\tvalid_0's auc: 0.992867\n",
      "\n",
      "Fold 1\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/aerdem/venv/lib/python3.8/site-packages/lightgbm/engine.py:239: UserWarning: 'verbose_eval' argument is deprecated and will be removed in a future release of LightGBM. Pass 'log_evaluation()' callback via 'callbacks' argument instead.\n",
      "  _log_warning(\"'verbose_eval' argument is deprecated and will be removed in a future release of LightGBM. \"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Info] Number of positive: 19688, number of negative: 608921\n",
      "[LightGBM] [Warning] Auto-choosing row-wise multi-threading, the overhead of testing was 0.008941 seconds.\n",
      "You can set `force_row_wise=true` to remove the overhead.\n",
      "And if memory is not enough, you can set `force_col_wise=true`.\n",
      "[LightGBM] [Info] Total Bins 3721\n",
      "[LightGBM] [Info] Number of data points in the train set: 628609, number of used features: 27\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/aerdem/venv/lib/python3.8/site-packages/lightgbm/basic.py:1780: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/home/aerdem/venv/lib/python3.8/site-packages/lightgbm/basic.py:1513: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning(f'{cat_alias} in param dict is overridden.')\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Info] [binary:BoostFromScore]: pavg=0.031320 -> initscore=-3.431679\n",
      "[LightGBM] [Info] Start training from score -3.431679\n",
      "[200]\tvalid_0's auc: 0.990122\n",
      "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
      "[400]\tvalid_0's auc: 0.99137\n",
      "[600]\tvalid_0's auc: 0.99166\n",
      "[800]\tvalid_0's auc: 0.991803\n",
      "[1000]\tvalid_0's auc: 0.991959\n",
      "[1200]\tvalid_0's auc: 0.992032\n",
      "[1400]\tvalid_0's auc: 0.992074\n",
      "[1600]\tvalid_0's auc: 0.992095\n",
      "\n",
      "Fold 2\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/aerdem/venv/lib/python3.8/site-packages/lightgbm/engine.py:239: UserWarning: 'verbose_eval' argument is deprecated and will be removed in a future release of LightGBM. Pass 'log_evaluation()' callback via 'callbacks' argument instead.\n",
      "  _log_warning(\"'verbose_eval' argument is deprecated and will be removed in a future release of LightGBM. \"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Info] Number of positive: 19646, number of negative: 608965\n",
      "[LightGBM] [Warning] Auto-choosing row-wise multi-threading, the overhead of testing was 0.011888 seconds.\n",
      "You can set `force_row_wise=true` to remove the overhead.\n",
      "And if memory is not enough, you can set `force_col_wise=true`.\n",
      "[LightGBM] [Info] Total Bins 3722\n",
      "[LightGBM] [Info] Number of data points in the train set: 628611, number of used features: 27\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/aerdem/venv/lib/python3.8/site-packages/lightgbm/basic.py:1780: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/home/aerdem/venv/lib/python3.8/site-packages/lightgbm/basic.py:1513: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning(f'{cat_alias} in param dict is overridden.')\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Info] [binary:BoostFromScore]: pavg=0.031253 -> initscore=-3.433887\n",
      "[LightGBM] [Info] Start training from score -3.433887\n",
      "[200]\tvalid_0's auc: 0.991305\n",
      "[400]\tvalid_0's auc: 0.992475\n",
      "[600]\tvalid_0's auc: 0.99281\n",
      "[800]\tvalid_0's auc: 0.992919\n",
      "[1000]\tvalid_0's auc: 0.992946\n",
      "[1200]\tvalid_0's auc: 0.992976\n",
      "[1400]\tvalid_0's auc: 0.992981\n",
      "[1600]\tvalid_0's auc: 0.99298\n",
      "\n",
      "Fold 3\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/aerdem/venv/lib/python3.8/site-packages/lightgbm/engine.py:239: UserWarning: 'verbose_eval' argument is deprecated and will be removed in a future release of LightGBM. Pass 'log_evaluation()' callback via 'callbacks' argument instead.\n",
      "  _log_warning(\"'verbose_eval' argument is deprecated and will be removed in a future release of LightGBM. \"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Info] Number of positive: 19696, number of negative: 608913\n",
      "[LightGBM] [Warning] Auto-choosing row-wise multi-threading, the overhead of testing was 0.010951 seconds.\n",
      "You can set `force_row_wise=true` to remove the overhead.\n",
      "And if memory is not enough, you can set `force_col_wise=true`.\n",
      "[LightGBM] [Info] Total Bins 3721\n",
      "[LightGBM] [Info] Number of data points in the train set: 628609, number of used features: 27\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/aerdem/venv/lib/python3.8/site-packages/lightgbm/basic.py:1780: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/home/aerdem/venv/lib/python3.8/site-packages/lightgbm/basic.py:1513: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning(f'{cat_alias} in param dict is overridden.')\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Info] [binary:BoostFromScore]: pavg=0.031333 -> initscore=-3.431260\n",
      "[LightGBM] [Info] Start training from score -3.431260\n",
      "[200]\tvalid_0's auc: 0.989075\n",
      "[400]\tvalid_0's auc: 0.990405\n",
      "[600]\tvalid_0's auc: 0.990883\n",
      "[800]\tvalid_0's auc: 0.991088\n",
      "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
      "[1000]\tvalid_0's auc: 0.991099\n",
      "[1200]\tvalid_0's auc: 0.991201\n",
      "[1400]\tvalid_0's auc: 0.991188\n",
      "[1600]\tvalid_0's auc: 0.991151\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import GroupKFold\n",
    "from sklearn.metrics import roc_auc_score\n",
    "import lightgbm as lgb\n",
    "\n",
    "N_FOLDS = 4\n",
    "gkf = GroupKFold(N_FOLDS)\n",
    "\n",
    "\n",
    "features = [\"match_score\", #\"match_score_max\", \"match_score_min\",\n",
    "            \"vec_dist\", \"vec_dist_max\", \"vec_dist_min\", \"len_content_text\",\n",
    "            \"len_topic_title\", \"len_topic_description\", \"len_content_title\", \"len_content_description\",\n",
    "             \"dup_count\", \"total_count\", \"dup_count_mean\",\n",
    "            \"tdup_count\", \"tdup_count_mean\", \"pdup_count\", \"pdup_count_mean\",\n",
    "            \"topic_language\", \"topic_category\", \n",
    "            \"content_kind\", \"topic_level\",\n",
    "            \"same_chapter\", \"starts_same\", \"content_is_train\", \"topic_channel\",\n",
    "            \"content_max_train_score\", \"topic_max_train_score\", \"second_degree\"\n",
    "           ]\n",
    "\n",
    "mc = -(np.array(features) == \"vec_dist\").astype(int)\n",
    "fc = (np.array(features) == \"vec_dist\")*0.5 + 0.5\n",
    "\n",
    "lgb_param = {\"objective\": \"binary\",\n",
    "             \"metric\": 'auc',\n",
    "             \"boosting\": \"gbdt\",\n",
    "             \"learning_rate\": 0.02,\n",
    "             \"is_unbalance\": True,\n",
    "            \"min_data_in_leaf\": 1024,\n",
    "            \"num_leaves\": 31,\n",
    "            \"feature_fraction\": 0.8,\n",
    "             \"feature_fraction_bynode\": 0.8,\n",
    "             \"subsample\": 0.5,\n",
    "             \"subsample_freq\": 1,\n",
    "             \"monotone_constraints\": mc,\n",
    "             \"monotone_constraints_method\": \"basic\",\n",
    "             \"feature_contri\": fc\n",
    "                    }\n",
    "\n",
    "y_oof = np.zeros(res_df.shape[0])\n",
    "\n",
    "\n",
    "\n",
    "scores = []\n",
    "\n",
    "\n",
    "for f, (train_ind, val_ind) in enumerate(gkf.split(res_df, res_df[\"target\"], groups=res_df[\"topic_id\"])):\n",
    "    print(\"Fold\", f)\n",
    "    train_df, val_df = res_df.iloc[train_ind].reset_index(drop=True), res_df.iloc[val_ind].reset_index(drop=True)\n",
    "        \n",
    "    lgb_train = lgb.Dataset(train_df[features], train_df[\"target\"])\n",
    "    lgb_val = lgb.Dataset(val_df[features], val_df[\"target\"])\n",
    "\n",
    "    model = lgb.train(lgb_param, lgb_train, num_boost_round=1600, #early_stopping_rounds=20, \n",
    "                      valid_sets=[lgb_val],\n",
    "                      verbose_eval=200)\n",
    "    model.save_model(f'models/lgb_{f}.txt')\n",
    "    \n",
    "    y_oof[val_ind] += model.predict(val_df[features])\n",
    "    scores.append(roc_auc_score(val_df[\"target\"], y_oof[val_ind]))\n",
    "    print()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "bf74fb81",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9922733268568036"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "ba11a622",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:title={'center':'Feature importance'}, xlabel='Feature importance', ylabel='Features'>"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "lgb.plot_importance(importance_type=\"gain\", booster=model, figsize=(8, 8))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "fff559cf",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:title={'center':'Feature importance'}, xlabel='Feature importance', ylabel='Features'>"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAl4AAAHwCAYAAAB332GFAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAAsTAAALEwEAmpwYAACsj0lEQVR4nOzdd3xX1f3H8debraACBSyjGgVkJIEICkUoBhFc1D2KqAytxYkDlRYHuADFCiLFWUFQVFDE0R+CQBRxMCRMBa2kMlQEFAkgJOHz++PexG/CN4NMCJ/n4/F95N5zz7ontPl47vneIzPDOeecc86VvApl3QHnnHPOuUOFB17OOeecc6XEAy/nnHPOuVLigZdzzjnnXCnxwMs555xzrpR44OWcc845V0o88HLOuYOQpH9Ieq6s++Gc2z/y93g55w41klKAo4GMiOQTzGxjEeu8xszeL1rvDj6ShgBNzOyKsu6Lcwc6n/Fyzh2q/mxmNSI+hQ66ioOkSmXZfmEdrP12rqx44OWccyFJR0l6XtJ3kjZIelBSxfBaY0lzJG2RtFnSS5JqhtcmAscAb0tKlXSnpERJ63PUnyLp9PB4iKSpkiZJ+gXok1f7Ufo6RNKk8DhGkknqK2mdpJ8k9Zd0sqRlkn6W9GRE2T6S5kt6UtI2SV9K6hpxvYGktyRtlfS1pL/maDey3/2BfwCXhfe+NMzXV9IXkrZL+kbS3yLqSJS0XtLtkjaF99s34vphkh6T9L+wfx9JOiy89kdJH4f3tFRSYiF+1c6VGQ+8nHPuN+OBdKAJcCLQHbgmvCZgGNAAaAH8ARgCYGZXAt/y2yzaIwVs7zxgKlATeCmf9guiPdAUuAwYBQwGTgdigUslnZoj73+BOsB9wBuSaofXXgHWh/d6MfCwpNNy6ffzwMPAq+G9tw7zbAJ6AEcCfYHHJbWJqOP3wFFAQ+BqYKykWuG1kUBb4BSgNnAnsFdSQ+Bd4MEwfSDwuqS6+zFGzpUpD7ycc4eqN8NZk58lvSnpaOBs4BYz22Fmm4DHgb8AmNnXZjbLzHab2Y/AP4FTc6++QD4xszfNbC9BgJJr+wX0gJn9amYzgR3AZDPbZGYbgHkEwVymTcAoM0szs1eB1cA5kv4AdATuCutKBp4DrorWbzPbFa0jZvaumf3XAh8AM4E/RWRJA+4P2/8PkAo0k1QB6AcMMLMNZpZhZh+b2W7gCuA/ZvafsO1ZwKJw3Jw7KPizeefcoer8yIXwktoBlYHvJGUmVwDWhdePBkYTBA9HhNd+KmIf1kUcH5tX+wX0Q8TxrijnNSLON1j2b1f9j2CGqwGw1cy257h2Ui79jkrSWQQzaScQ3MfhwPKILFvMLD3ifGfYvzpANYLZuJyOBS6R9OeItMrA3Pz649yBwgMv55wLrAN2A3VyBASZHgYMiDezrZLOB56MuJ7zK+I7CIINAMK1WjkfiUWWya/94tZQkiKCr2OAt4CNQG1JR0QEX8cAGyLK5rzXbOeSqgKvE8ySTTezNElvEjyuzc9m4FegMbA0x7V1wEQz++s+pZw7SPijRuecA8zsO4LHYY9JOlJShXBBfebjxCMIHodtC9ca3ZGjih+A4yPO1wDVJJ0jqTJwN1C1CO0Xt3rAzZIqS7qEYN3af8xsHfAxMExSNUmtCNZgTcqjrh+AmPAxIUAVgnv9EUgPZ7+6F6RT4WPXfwP/DBf5V5TUIQzmJgF/lnRGmF4tXKjfaP9v37my4YGXc8795iqCoGEVwWPEqUD98NpQoA2wjWCB9xs5yg4D7g7XjA00s23A9QTrozYQzICtJ295tV/cPiNYiL8ZeAi42My2hNd6AjEEs1/TgPvyeT/ZlPDnFkmfhzNlNwOvEdzH5QSzaQU1kOCx5EJgKzACqBAGhecRfIvyR4IZsDvwv2XuIOIvUHXOuUOMpD4EL3vtVNZ9ce5Q4/+V4JxzzjlXSjzwcs4555wrJf6o0TnnnHOulPiMl3POOedcKfHAyznnnHOulPgLVF2Jq1mzpjVp0qSsu3FQ2rFjB9WrVy/rbhy0fPwKz8euaHz8Cq88jN3ixYs3m1nUPUQ98HIl7uijj2bRokVl3Y2DUlJSEomJiWXdjYOWj1/h+dgVjY9f4ZWHsZP0v9yu+aNG55xzzrlS4oGXc84551wp8cDLOeecc66UeODlnHPOOVdKPPByzjnnnCslHng555xzzpUSD7ycc84550qJB17OOeecc6XEAy/nnHPOuVLigZdzzjnnDjr9+vWjXr16xMXFZaVt3bqVbt260bRpU7p168ZPP/0EwEsvvUSrVq2Ij4/nlFNOYenSpQCsW7eOLl260LJlS2JjYxk9enRWXVOmTCE2NpYKFSoU6+4rHng555xz7qDTp08fZsyYkS1t+PDhdO3ala+++oquXbsyfPhwAI477jg++OADli9fzj333MO1114LQKVKlXjsscdYtWoVn376KWPHjmXVqlUAxMXF8cYbb9C5c+di7bcHXoUgqaak64tQ/j+SahZDP2IkrShqPUVof7yki8uqfeecc4euzp07U7t27Wxp06dPp3fv3gD07t2bN998E4BTTjmFWrVqAfDHP/6R9evXA1C/fn3atGkDwBFHHEGLFi3YsGEDAC1atKBZs2bF3m/fJLtwagLXA/8qTGEzO7tYe3OA25WWQcygd8u6Gwel2+PT6eNjV2g+foXnY1c0Pn6FV5CxSxl+TtT0H374gfr16wPw+9//nh9++GGfPM8//zxnnXXWvnWmpLBkyRLat29fiF4XnAdehTMcaCwpGZgVpp0FGPCgmb0qKRG4H9gONAHmAteb2V5JKcBJZrZZ0lXAwLDsMjO7MlqDko4GngKOD5OuAzYCFSU9C5wCbADOM7Ndkv4KXAtUAb4GrjSznZLGA78AJwG/B+40s6lhf4cAm4E4YDFwhZmZpLbAP4Ea4fU+ZvZdXgMk6dqwferUqcu98el5ZXe5OPqw4P+EXOH4+BWej13R+PgVXkHGLikpCYDvv/+eHTt2ZJ2np6dnHQNkZGRkO1+yZAljxozhiSeeyJa+a9cuBgwYwDXXXMPnn3+era2ff/6ZxYsXk5qaWpTbyuKBV+EMAuLMLEHSRUB/oDVQB1go6cMwXzugJfA/YAZwITA1sxJJscDdwClhEJZ9zjS7J4APzOwCSRUJgqBaQFOgp5n9VdJrwEXAJOANM3s2bOdB4GpgTFhXfaAT0Bx4K6JPJwKxBAHdfKCjpM/CcueZ2Y+SLgMeAvrlNUBm9gzwDMAxxzexx5b7P7XCuD0+HR+7wvPxKzwfu6Lx8Su8goxdSq/E4GdKCtWrVycxMThv2LAhzZo1o379+nz33Xc0aNAg69qyZct48sknmTVrFieccEJWXWlpafTo0YP+/ftz22237dNWzZo1adu2LSeddFKx3J//qyi6TsBkM8sAfpD0AXAywazSAjP7BkDS5DDv1IiypwFTzGwzgJltzaOd04CrwnwZwDZJtYC1ZpYc5lkMxITHcWHAVZMgSHsvoq43zWwvsCqcScu0wMzWh/1NDuv6mWAGbJYkgIpAnrNdOR1WuSKrc5kWdnlLSkrK+j8Yt/98/ArPx65ofPwKryhjd+655zJhwgQGDRrEhAkTOO+88wD49ttvufDCC5k4cWK2oMvMuPrqq2nRokXUoKsk+OL6kmX5nBeH3RHHGfwWTI8HbjSzeGAoUC2XMsqnLgErzSwh/MSbWffi6rxzzjlXGD179qRDhw6sXr2aRo0a8fzzzzNo0CBmzZpF06ZNef/99xk0aBAA999/P1u2bOH6668nISEha/Zq/vz5TJw4kTlz5pCQkEBCQgL/+c9/AJg2bRqNGjXik08+4ZxzzuGMM84oln77jFfhbAeOCI/nAX+TNAGoDXQG7iB4jNdO0nEEjxovI3z0FmEOME3SP81si6Taecx6zSZY1zUq4lFjXo4AvpNUGehFsP6rMFYDdSV1MLNPwvpOMLOVhazPOeecK7LJkydHTZ89e/Y+ac899xzPPffcPumdOnXCLPqcyAUXXMAFF1xQtE5G4TNehWBmW4D54ascOgDLgKUEgdSdZvZ9mHUh8CTwBbAWmJajnpUE66U+kLSUYAF7bgYAXSQtJ3ik2DKfbt4DfEawVuvLgt9ddma2B7gYGBH2MZlgIb9zzjnn9pPPeBWSmV2eI+mOKNl+MbMeUcrGRBxPACYUoL0fgPOiXIqLyDMy4ngcMC5KPX1ynNcIfyYBSRHpN0YcJxPM5OVZl3POOefy5jNezjnnnHOlxGe8SkjOGaSCkjQYuCRH8hQze6gYuuWcc865MuSB1wEmDLA8yHLOOefKIX/U6JxzzjlXSjzwcs45V+J+/vlnLr74Ypo3b06LFi345JNPsq499thjSGLz5s3ZyixcuJBKlSoxdepv752eMGECTZs2pWnTpkyYkO/3kpw74PijRueccyVuwIABnHnmmUydOpU9e/awc+dOANatW8fMmTM55phjsuXPyMjgrrvuonv3397XvHXrVoYOHcqiRYuQRNu2bTn33HOpVatWqd6Lc0XhM14ui6TU8GcDSVPzyFdT0vWl1zPn3MFs27ZtfPjhh1x99dUAVKlShZo1awJw66238sgjjxBuSZZlzJgxXHTRRdSrVy8r7b333qNbt27Url2bWrVq0a1bN2bMmFFq9+FccfAZL7cPM9tI8NLU3NQErgf+VZD6dqVlEDPo3WLo2aHn9vh0+vjYFZqPX+EV19ilDD+HtWvXUrduXfr27cvSpUtp27Yto0eP5v3336dhw4a0bt06W5kNGzYwbdo05s6dy8KFC7Ol/+EPf8g6b9SoERs2FHZTDufKhs94HQQkDZd0Q8T5EEkDJd0haaGkZZKGRly/KkxbKmliHvUeJ+kTScvDDbUz02PCt/IjKVbSAknJYZ1NgeFA4zDt0ZK5a+dceZGens7nn3/Oddddx5IlS6hevTpDhgzh4Ycf5v77798n/y233MKIESOoUMH/RLnyx2e8Dg6vAqOAseH5pcAIoCPQjmAj67ckdQa2AHcDp5jZZkm186h3NDDOzF6MDOxy6A+MNrOXJFUBKgKDgDgzS8itYknXAtcC1KlTl3vj0wt0oy67ow8LZh5c4fj4FV5xjV1SUhJbt26lTp067Nq1i6SkJBo3bsz48eNZu3YtzZo1A+DHH38kNjaWcePG8dFHHzFv3jwgeEw5ffp0vvzyS3bv3k1ycjJJSUkALFiwgISEhKzzA0lqauoB2a+DQXkfOw+8DgJmtkRSPUkNgLrAT0A80B1YEmarATQFWhO8cHVzWDa3TbchCNwuCo8nEgRzOX0CDJbUCHjDzL7KuRYjlz4/Q7gp+DHHN7HHlvs/tcK4PT4dH7vC8/ErvOIau5ReiQA8/vjj1K9fn2bNmpGUlETXrl159NHfJsxjYmJYtGgRderU4cILL8xK79OnDz169ODiiy9m69attG3bNuvR5IoVK5gwYQK1a+f135dlIykpicTExLLuxkGpvI+d/z/SwWMKwbqr3xPMgB0LDDOzpyMzSbppP+uNvi175kWzlyV9BpwD/EfS34Bv9qeBwypXZPXwc/azWw6C/wPK/MPl9p+PX+EV99iNGTOGXr16sWfPHo4//nheeOGF/a6jdu3a3HPPPZx88skA3HvvvQdk0OVcXjzwOni8CjwL1AFOJZjxekDSS2aWKqkhkAbMAaZJ+qeZbZFUO49Zr/nAX4BJQK9oGSQdD3xjZk9IOgZoBSwFjijOm3POlW8JCQksWrQo1+spKSlR08ePH5/tvF+/fvTr168Ye+Zc6fKViwcJM1tJEOxsMLPvzGwm8DLwiaTlwFTgiDDfQ8AHkpYC/8yj2gHADWH5hrnkuRRYISkZiANeNLMtwHxJK3xxvXPOOVdwPuN1EDGz+BznowkWyOfMNwHI95XOZrYW6BCRdHeYnkIQZGFmwwm+xZiz7OX70XXnnHPO4TNezjnnnHOlxme8DgGSBgOX5EieYmYPlUV/nHPOuUOVB16HgDDA8iDLOeecK2P+qNE555xzrpR44OWcc845V0o88HLOuQgZGRmceOKJ9OjRA4BevXrRrFkz4uLi6NevH2lpaQA8+uijJCQkkJCQQFxcHBUrVmTr1q38+uuvtGvXjtatWxMbG8t9991XlrfjnDvAeODlnHMRRo8eTYsWLbLOe/XqxZdffsny5cvZtWsXzz33HAB33HEHycnJJCcnM2zYME499VRq165N1apVmTNnDkuXLiU5OZkZM2bw6aefltXtOOcOMAd04CUpRlKR3hcl6RZJhxdXn4pCUoKkswtRroGkqSXRJ+fcb9avX8+7777LNddck5V29tlnIwlJtGvXjvXr1+9TbvLkyfTs2RMASdSoUQOAtLQ00tLSKMj+ps65Q8OB/q3GGOBygje0F9YtBFvi7CyG/hRVAnAS8J+cFyRVMrP0aIXMbCPBPo1lKq8+5mVXWgYxg94tiS6Ve7fHp9PHx67Q9mf8Uoafwy233MIjjzzC9u3b97melpbGxIkTGT06+zuLd+7cyYwZM3jyySez0jIyMmjbti1ff/01N9xwA+3bty/ajTjnyo0SnfGSdJWkZZKWSpoYzmDNCdNmh3v/IWm8pCckfSzpG0mZQcZw4E+SkiXdKqmipEclLQzr+FtYPlFSkqSpkr6U9JICNwMNgLmS5ubRz9Sw3pWS3pfULqzvG0nnhnliJM2T9Hn4OSVMvyC8F0mqL2mNpN9HaaMKcD9wWXg/l0kaEo7LfGBiHm3ESFoRHveR9IakGZK+kvRIHvdVMRzbFZKWS7o1TG8S3ufSsJ3GYf8fjch7WcTYzpP0FrAqt9+Bcwe7d955h3r16tG2bduo16+//no6d+7Mn/70p2zpb7/9Nh07dsy2WXPFihVJTk5m/fr1LFiwgBUrVpRo351zBw+ZWclULMUC04BTzGyzpNoE29hMNbMJkvoB55rZ+ZLGA9WBy4DmwFtm1kRSIjDQzHqEdV4L1DOzByVVJdjk+RLgWGA6EAtsDNPvMLOPJKUAJ5nZ5jz6asDZZvZ/kqaFfTkHaAlMMLOE8HHlXjP7VVJTYLKZnRSWnwR8CpwJvGRmk3Npp0/YlxvD8yHAn4FOZrYrtzYkxQDvmFlcWMe9wInAbmB1WH5dlPbaAsPNrFt4XtPMfpb0WZg+TVI1ggD8LKB/eA91gIVAe6AZ8C4QZ2Zrc/sdhNsPRbZ9LXAtQJ06ddveO+rZ3Ibf5eHow+CHXWXdi4PX/ozfp/95jZkzZ1KxYkX27NnDzp07+dOf/sTgwYOZMGECX331Fffffz8VKmT/79V77rmHU089ldNPPz1qvRMmTKBatWpcdtllRb2dUpWampr1yNTtPx+/wisPY9elS5fFmTHCPsysRD7ATcBDOdI2A5XD48rA5vB4PNArIt/28GciQcCRmT4VWAMkh5+1QPcw36yIfOOAK8LjFKBOPn3dzW9B6P3A4PC4AvBzeHwUMBFYHra9M6J8LWAD8Ho+7fQBnow4HwLcF3EetQ2CR64rIup4NqLM/xEEXtHaqwX8FxhDEFBVINhoe32UvI8D/SLOJwLnhmM7N7/fQV73fcIJJ5grnLlz55Z1Fw5qhR2/uXPn2jnnnGNmZs8++6x16NDBdu7cuU++n3/+2WrVqmWpqalZaZs2bbKffvrJzMx27txpnTp1srfffrtQ/ShL/m+vaHz8Cq88jB2wyHL5m3ggrfHaHXGc20pUATeZ2XvZEoOZscjyGezf+rW0cKAA9mbWZWZ7JWXWcyvwA9CaIID5NaJ8o7Dc0ZIqmNne/Wh7R8RxXm1EKtC9mtlPkloDZxDMZl0KDNiPvkXrY9TfgXPlVf/+/Tn22GPp0CHYT/7CCy/k3nvvBWDatGl0796d6tWrZ+X/7rvv6N27NxkZGezdu5dLL70069UUzjlXkoHXHGCapH+a2ZbwUePHwF8IZlN6AfPyqWM7wQxNpveA6yTNMbM0SScQzDQVpI5cHzUW0FEEM0V7JfUGKkKw4Bz4N9AT6A3cBozMpy/71UZhSaoD7DGz1yWtBiaZ2XZJ6yWdb2Zvho8LKxL8Lv4maQJQG+gM3EHw6DdS1N+Bme3AuXIiMTGRxMREANLTc/8+SZ8+fejTp0+2tFatWrFkyZIS7J1z7mBWYoGXma2U9BDwgaQMYAnB48cXJN0B/Aj0zaeaZUCGpKUEjyNHEzx2+1ySwjrOz6eOZ4AZkjaaWZdC3g7Av4DXJV0FzOC3WaB/APMsWE+2FFgo6V0z+yJKHXOBQZKSgWH70UZhNSQY78xFKX8Pf14JPC3pfiCNYJ3cNKADsBQw4E4z+15SzsDrOfb/d+Ccc845SnBxvXOZmjVrZqtXry7rbhyUkpKSsmZe3P7z8Ss8H7ui8fErvPIwdpJyXVx/QL9A1TnnnHOuPDmQFteXuPA1ClVzJF9pZsuLuZ0zgBE5ktea2QXF2U6ONkvl3pxzzjlXeIdU4GVmpfL66PAbf6X6rb/SujfnnHPOFZ4/anTOOeecKyUeeDnnSsyvv/5Ku3btaN26NbGxsdx3330APPnkkzRp0gRJbN7825tetm3bxp///Oes/C+88AIAycnJdOjQgdjYWFq1asWrr75aJvfjnHNFdUg9anTOla6qVasyZ84catSoQVpaGp06deKss86iY8eO9OjRY59vLo0dO5aWLVvy9ttv8+OPP9KsWTN69erF4YcfzosvvkjTpk3ZuHEjbdu25YwzzqBmzZplcl/OOVdYPuN1kJGUGv5sIGlqHvlqSrq+mNrsH75bzLn9Iilrz7W0tDTS0tKQxIknnkhMTEzU/Nu3b8fMSE1NpXbt2lSqVIkTTjiBpk2bAtCgQQPq1avHjz/+WJq34pxzxcJnvA5SZrYRuDiPLDWB6wleylrUtp4qSvldaRnEDHq3qN04JN0en06fg3TsUoafA0BGRgZt27bl66+/5oYbbqB9+9y/B3LjjTdy7rnn0qBBA7Zv386rr766z6bUCxYsYM+ePTRu3LhE+++ccyXBZ7yKiaThkm6IOB8iaaCkOyQtlLRM0tCI61eFaUslTcyj3uMkfSJpuaQHI9JjJK0Ij2MlLZCUHNbZFBgONA7THs2l7kRJH0iaLumb8B56hXUtl9Q48l7C4yRJI8I8ayT9qahj58q3ihUrkpyczPr161mwYAErVqzINe97771HQkICGzduJDk5mRtvvJFffvkl6/p3333HlVdeyQsvvLBPQOaccwcDn/EqPq8Co4Cx4fmlBO/y6gi0I9hc+i1JnYEtwN3AKWa2OdzHMjejgXFm9mJkYJdDf2C0mb0kqQrB3ouDgDgzS8in362BFsBW4BvgOTNrJ2kAwRZPt0QpUynMczZwH3B6zgySrgWuBahTpy73xue+353L3dGHBbNeB6OkpKR90mJiYhg7diyXXXYZECy+nz9/PkcddRQAI0eO5PLLL+eDDz4AoFatWrz00ku0aNGCHTt2cOutt9KrVy9+/fXXqPXnlJqaWqB8bl8+dkXj41d45X3sPPAqJma2RFI9SQ2AusBPQDzQnWCfSoAaQFOCYGeKmW0Oy27No+qOwEXh8UT2fTErwCfAYEmNgDfM7KtgG8UCWWhm3wFI+i8wM0xfDuS2t+Ub4c/FBPs27sPMniHYJ5Njjm9ijy33f2qFcXt8Ogfr2KX0SuTHH3+kcuXK1KxZk127dnHPPfdw1113ZS2qr1atGh07dqROnToAnHjiiWzdupXExER++OEHfvjhBy655BKOPPJIzjrrLK6//npuueWWAvehPGw9UlZ87IrGx6/wyvvYHZz/j37gmkKw7ur3BDNgxwLDzOzpyEySbtrPevPcUNPMXg7fXH8O8B9JfyOYvSqI3RHHeyPO95L7v4/MPBl55MlyWOWKrA7X+7j9k5SUREqvxLLuRqF999139O7dm4yMDPbu3cull15Kjx49eOKJJ3jkkUf4/vvvadWqFWeffTbPPfcc99xzD3369CE+Ph4zY8SIEdSpU4dJkybx4YcfsmXLFsaPHw/A+PHjSUhIKNP7c865/eWBV/F6FXgWqAOcSjDj9YCkl8wsVVJDIA2YA0yT9E8z2yKpdh6zXvOBvwCTgF7RMkg6HvjGzJ6QdAzQClgKHFGcN+fc/mrVqhVLlizZJ/3mm2/m5ptv3ie9QYMGzJw5c5/0K664giuuuKJE+uicc6XJV6cWIzNbSRDsbDCz78xsJvAy8Imk5cBU4Igw30PAB5KWAv/Mo9oBwA1h+Ya55LkUWCEpGYgDXjSzLcB8SStyW1zvnHPOudLlM17FzMzic5yPJlggnzPfBGBCAepbC3SISLo7TE8hCLIws+EE32LMWfbyfOpOApIizhOjXTOzIbnk2Uwua7ycc845ty+f8XLOOeecKyU+43WAkDQYuCRH8hQze6gY6o4n+EZkpN1mlvubLJ1zzjlX7DzwOkCEAVaRg6xc6l4OJJRE3c4555wrOH/U6JxzzjlXSjzwcs4555wrJR54OecKZN26dXTp0oWWLVsSGxvL6NHBl3WTk5P54x//SEJCAieddBILFizIVm7hwoVUqlSJqVOnZqXdeeedxMbG0qJFC26++WbM8nxHsHPOlRseeDnnCqRSpUo89thjrFq1ik8//ZSxY8eyatUq7rzzTu677z6Sk5O5//77ufPOO7PKZGRkcNddd9G9e/estI8//pj58+ezbNkyVqxYwcKFC7P2ZnTOufLOA6+DgKTU8GcDSVPzyFdT0vWl1zN3KKlfvz5t2rQB4IgjjqBFixZs2LABSfzyyy8AbNu2jQYNGmSVGTNmDBdddBH16tXLSpPEr7/+yp49e9i9ezdpaWkcffTRpXszzjlXRvxbjQcRM9tIsBdkbmoC1wP/KpUOFdCutAxiBr1b1t04KN0en06fA2DsUnLstZmSksKSJUto3749o0aN4owzzmDgwIHs3buXjz/+GIANGzYwbdo05s6dy8KFC7PKdujQgS5dulC/fn3MjBtvvJEWLVqU6v0451xZ8RmvIpA0XNINEedDJA2UdIekhZKWSRoacf2qMG2ppJzv1Yqs9zhJn0haLunBiPQYSSvC41hJCyQlh3U2JXh7feMwLeo2QZISJX0gabqkb8J76BXWtVxS4zDfnyV9JmmJpPclHR2mj5Z0b3h8hqQPJfm/o0NIamoqF110EaNGjeLII49k3LhxPP7446xbt47HH3+cq6++GoBbbrmFESNGUKFC9n8eX3/9NV988QXr169nw4YNzJkzh3nz5pXFrTjnXKmTL2otPEknAqPM7NTwfBUwAugI/A0Q8BbwCLAFmAacYmab89oYW9JbwFQzezEM7EaYWQ1JMcA7ZhYnaQzwqZm9JKkKUBE4OvN6Hn1OBN4EWgBbgW+A58zsPkkDgOPM7BZJtYCfzcwkXQO0MLPbJR0OLARuBJ4Czjaz/0Zp51rgWoA6deq2vXfUswUbVJfN0YfBD7vKuhcQ3/AoANLT0/n73//OySefzKWXXgpAjx49ePvtt5GEmdGjRw/effddevbsmbVoftu2bVSrVo3bb7+d9evXs2fPHq666ioAJkyYQJUqVejZs2ex9zs1NZUaNWoUe72HAh+7ovHxK7zyMHZdunRZbGYnRb1oZv4pwgf4AmgAtAbmAyOBFCA5/HwNXA3cBDxUwDq3AJXD4yOB1PA4BlgRHl8OrATuAprmvJ5H3YnArIjzD4GO4fFpwJvhcTwwE1gOrAZmRJQ5BUgHbirI/ZxwwgnmCmfu3Lll3YUse/futSuvvNIGDBiQLb158+ZZ/Xz//fetTZs2+5Tt3bu3TZkyxczMXnnlFevataulpaXZnj177LTTTrO33nqrRPp8II3fwcbHrmh8/AqvPIwdsMhy+Zvoa7yKbgrBuqvfA68CxwLDzOzpyEySbtrPevOcijSzlyV9BpwD/EfS3whmrwpid8Tx3ojzvfy27m8M8E8zeyucJRsSUSaeIDhsgDtkzJ8/n4kTJxIfH09CQgIADz/8MM8++ywDBgwgPT2datWq8cwzz+RZz8UXX8ycOXOIj49HEmeeeSZ//vOfS+EOnHOu7HngVXSvAs8CdYBTCYKSByS9ZGapkhoCacAcYJqkf5rZlrweNRLMnP0FmAT0ipZB0vHAN2b2hKRjgFbAUuCIYrqvo4AN4XHviHaPBW4HTiQI+N40s8+KqU13AOvUqVOu79tavHhxnmXHjx+fdVyxYkWefvrp3DM751w55ouii8jMVhIEOxvM7Dszmwm8DHwiaTkwFTgizPcQ8IGkpcA/86h2AHBDWL5hLnkuBVZISgbigBfNbAswX9KK3BbX74chwBRJi4HNAJIEPA8MtOAbllcDz0mqVsS2nHPOuUOCz3gVAzOLz3E+GhgdJd8EYEIB6lsLdIhIujtMTyEIsjCz4QTfYsxZ9vJ86k4CkiLOE6NdM7PpwPQoVZwekX8xwQyfc8455wrAZ7ycc84550qJz3iVIUmDgUtyJE8xs4eKoe54IOe7wnabWfui1u2cc865wvHAqwyFAVaRg6xc6l4OJJRE3c4555wrHH/U6JxzzjlXSjzwcs5FtW7dOrp06ULLli2JjY1l9Ojg+yKXXXYZCQkJJCQkEBMTk/VOr0zffvstNWrUYOTIkVlpo0ePJi4ujtjYWEaNGlWKd+GccwcWf9TonIuqUqVKPPbYY7Rp04bt27fTtm1bunXrxquvvpqV5/bbb+eoo47KVu62227jrLPOyjpfsWIFzz77LAsWLKBKlSqceeaZ9OjRgyZNmpTavTjn3IHigJrxklRT0vVFKP8fSTWLsUvFStI/ClnuOUkti7s/zuWlfv36tGnTBoAjjjiCFi1asGHDhqzrZsZrr72WbY/FN998k+OOO47Y2NistC+++IL27dtz+OGHU6lSJU499VTeeOON0rsR55w7gBxQgRdQEyh04GVmZ5vZz8XWm+IXNfBSINffhZldY2arSq5bBSOpYln3wZWNlJQUlixZQvv2v30pdt68eRx99NE0bdoUCDa2HTFiBPfdd1+2snFxccybN48tW7awc+dO/vOf/7Bu3bpS7b9zzh0oDrRHjcOBxuHb2GeFaWcR7Fv4oJm9Gu4beD+wHWgCzAWuN7O9klKAk8xss6SrgIFh2WVmdmW0BiWNB3YRbIFTD+gHXEXwAtPPzKxPmG8ccDJwGDDVzO6TdBSwADjXzFZLmgzMMbNno7QzHDgsvLeVwGDgPeAzoC1wtqRBOdsIyyYRvC1+kaRUgpez9gj7fZ6Z/ZDLvV0C3AdkANvMrHMYPI0AziTYm/FZMxsjqSvBBt+VgIXAdWa2OxzTV4FuwCOStgJDgarAf4G+ZpYarf1Mu9IyiBn0bl5ZXC5uj0+nTxmMXcrwc7KOU1NTueiiixg1ahRHHnlkVvrkyZOzzXYNGTKEW2+9lRo1amSrq0WLFtx11110796d6tWrk5CQQMWKHsM75w5Nym3vtbIgKQZ4x8ziJF0E9CcIEOoQBAPtgWbADKAl8L/w+Gkzm5oZeAFHA9OAU8IgLNd9EcPAqxrQEziX4N1XHQmCo4XA1WaWnFlHGLjMBm42s2WSuhEEgqOBPmZ2Zh73l2pmNSLu9Zuwj5+Gabm1kcRvgZcRBHpvS3oE+MXMHsylveXAmWa2QVJNM/tZ0nVAV+AvZpYuqTawE/gK6GpmayS9CHxuZqPCMf2XmT0iqQ7wBnCWme2QdBdQ1czuj9L2tcC1AHXq1G1776h9YlFXAEcfBj/sKv124xsG67bS09P5+9//zsknn8yll16adT0jI4NLLrmEp59+mrp16wJw8803s2nTJiAI1ipUqEDfvn254IILstX97LPPUrduXc4///wSv4/U1NR9AkFXMD52RePjV3jlYey6dOmy2MxOinbtQJvxitQJmGxmGcAPkj4gmA36BVhgZt8AhLNMnQj2RMx0GsGLSDcD5LEZdaa3zczCQOWH8B1YSFoJxADJwKVhMFEJqE8Q+C0zs1nhzNJYoPV+3uP/MoOuUNQ2cpTZA7wTHi8mmInKzXxgvKTXCAImCLb8ecrM0iEYG0mtgbVmtibMMwG4ARgVnmeupv5j2Kf5wbaNVAE+idawmT0DPANwzPFN7LHlB/I/tQPX7fHplMXYpfRKxMzo3bs3HTt23OebiDNmzCA+Pp5LLvnt/b/Llv32T3XIkCHUqFGDgQMHArBp0ybq1avHt99+y+LFi/n000+pWbNmid9HUlISiYmJJd5OeeRjVzQ+foVX3sfuYP1rmHOarqjTdrvDn3sjjjPPK0k6juCx5clm9lPELBnh2qwWBLNGtYD1+9HujsyDvNrIIc1+m6bMII/foZn1l9QeOAdYLKntfvQtWj8FzDKznnllzumwyhVZHfHoyhVcUlISKb0Sy6Tt+fPnM3HiROLj47NeGfHwww9z9tln88orr2R7zJifiy66iC1btlC5cmXGjh1bKkGXc84diA60wGs7cER4PA/4m6QJQG2gM3AH0BxoFwYq/wMuI5xZiTAHmCbpn2a2Ja9HjQV0JEHwsU3S0QTrzpLCa7cCXxAsnH9BUgczS8ulnjRJlXO5nlcbhSKpsZl9Bnwm6SzgDwRr5/4maW7Eo8bVQIykJmb2NXAl8EGUKj8Fxmbmk1QdaBgxU+bKkU6dOpHbUoTx48fnWXbIkCHZzufNm1dMvXLOuYPbAfWtRjPbQvAYawXB4vZlwFKCQOpOM/s+zLoQeJIg4FlLsJ4rsp6VBFvxfCBpKfDPIvZrKbAE+BJ4meARHpKaAdcAt5vZPOBD4O48qnoGWCbppYK2UUSPSloejufHBGP5HPBt2I+lwOVm9ivQF5gSPm7dCzwVpY8/An2AyZKWETxmbF4M/XTOOecOCQfajBdmdnmOpDuiZPvFzHpEKRsTcTyBYK1Sfu31iThOAeJyudaH6FpE5Lktn7buAu6KSIrLcT1qG2aWGHFcI+J4KtnXtuUsd2GU5HTgtvATmXc2wTc7c9YRk+N8DsFaO+ecc87tpwNqxss555xzrjw74Ga88mNmSRRi7ZOkwcAlOZKnmNlDxdCtnG19RvCeq0hXZn5bsgTaK7V7c84551zhHXSBV2GFQUipBCJm1j7/XMXaXqndm3POOecKzx81Ouecc86VEg+8nHPOOedKiQdezh3k1q1bR5cuXWjZsiWxsbGMHj0agClTphAbG0uFChVYtGhRVv60tDR69+5NfHw8LVq0YNiwYdnqy8jI4MQTT6RHj32+OOycc66IDpk1Xs6VV5UqVeKxxx6jTZs2bN++nbZt29KtWzfi4uJ44403+Nvf/pYt/5QpU9i9ezfLly9n586dtGzZkp49exITEwPA6NGjadGiBb/88ksZ3I1zzpVvh/yMl6TUsu5DNJJukXR4EconSjolnzznS2oZcX6/pNOjtS8pJdwk2x1g6tevT5s2bQA44ogjaNGiBRs2bKBFixY0a9Zsn/yS2LFjB+np6ezatYsqVapw5JFHArB+/XreffddrrnmmlK9B+ecO1T4jNeB6xZgEsEekIWRCKQSvLE+N+cTbLi9CsDM7i3G9rPsSssgZtC7Ra3mkHR7fDp98hi7lBx7YKakpLBkyRLat8/9i7UXX3wx06dPp379+uzcuZPHH3+c2rVrA3DLLbfwyCOPsH379uK5Aeecc9kc8jNekSTdIWmhpGWShoZpMZK+kPSspJWSZko6LI86mkh6X9JSSZ9LaqzAo5JWhFv4XBbmTZSUJGmqpC8lvRTmvRloAMyVNDfM213SJ2GdUyTVCNNTJA0N05dLai4pBugP3CopWdKfovTzFOBcgm2FksN+jpd0cbT2c5S9QtKCsNzTkioWcehdMUhNTeWiiy5i1KhRWTNY0SxYsICKFSuyceNG1q5dy2OPPcY333zDO++8Q7169WjbtrB7qTvnnMuPz3iFJHUHmgLtAAFvSepMsK9hU6Cnmf1V0mvARQSzQdG8BAw3s2mSqhEEtxcCCUBroA6wUNKHYf4TgVhgI8H+jB3N7AlJtwFdzGxz+IjvbuB0M9sh6S6CLX/uD+vYbGZtJF0PDDSzayQ9BaSa2chonTSzjyW9BbwTbj2EpMxr2drPMU4tCDYm72hmaZL+BfQCXsyR71rgWoA6depyb3x6LsPl8nL0YcGsV26SkpIASE9P5+9//zvt27endu3aWekAP//8M4sXLyY1NXiqPmrUKFq2bMn8+cF2oMcffzwTJkzg66+/ZubMmbzxxhvs2bOHnTt30q1bNwYPHlxi91fSUlNTs42FKzgfu6Lx8Su88j52Hnj9pnv4WRKe1yAIuL4F1ppZcpi+GIiJVoGkI4CGZjYNINx8GkmdgMlmlgH8IOkDgv0OfwEWmNn6MF9yWPdHOar+I9CSYANxgCoEG1RneiOib9H2ZyxOXYG2BMEjwGHAppyZzOwZgk3Badasmd3U67wS7lb5lJSUxKWJiXnmMTN69+5Nx44dGTVq1D7Xa9asSdu2bTnppJMA+Oyzz/jyyy9JTExkx44d/O9//2PEiBG0atUqW7sjR47knXfeKc7bKXVJSUkk5jN+Ljofu6Lx8Su88j52Hnj9RsAwM3s6W2Lw2G53RFIGQbBRXHLWHe13ImCWmfXMp47cyhcnARPM7O8l3I4roPnz5zNx4kTi4+NJSEgA4OGHH2b37t3cdNNN/Pjjj5xzzjkkJCTw3nvvccMNN9C3b19iY2MxM/r27Zst6HLOOVdyPPD6zXvAA5JeMrNUSQ2BtP2pwMy2S1ov6Xwze1NSVaAiMA/4m6QJQG2gM3AH0DyP6rYDRwCbgU+BsZKamNnXkqoTzKytyad87gt9sreRX/uRZgPTJT1uZpsk1QaOMLP/5dOWKyGdOnXCzKJeu+CCC/ZJq1GjBlOmTMmzzsTExHL9X5zOOVdWfHF9yMxmAi8Dn0haDkwl96AkL1cCN0taRvCNwt8D04BlwFJgDnCnmX2fTz3PADMkzTWzH4E+wOSw3k/IO2gDeBu4ILfF9aFXgDskLZHUOLf2IxPNbBXBerOZYV9mAfXz6Ytzzjnn8BkvzKxGxPFoYHSUbHEReaIuVo+4/hVwWpRLd4SfyLxJQFLE+Y0Rx2OAMRHncwjWheVsLybieBHBayQIZ8PyfH5kZvMJ1o5l6pNH+5HtvAq8mlfdzjnnnNuXz3g555xzzpWSQ37Gq7AkjQU65kgebWYvlEV/8iJpMHBJjuQpZvZQWfTHOeecO1R54FVIZnZDWfehoMIAy4Ms55xzroz5o0bnnHPOuVLigZdzB7ARI0ZQr1494uKyvt/B0qVL6dChA/Hx8fz5z3/ml19+ybq2bNkyOnToQGxsLPHx8fz6668ATJ48mfj4eFq1asWZZ57J5s053xLinHOuNHjg5dwB7Mwzz2TGjBnZ0q655hqGDx/O8uXLueCCC3j00UeBYNugK664gqeeeoqVK1eSlJRE5cqVSU9PZ8CAAcydO5dly5bRqlUrnnzyybK4HeecO+R54FVKJNUM91LMK0+MpMsLUFeMpBXF17v9U5B7ccWjdevW1K5dO1vamjVr6Ny5MwDdunXj9ddfB2DmzJm0atWK1q1bA/C73/2OihUrYmaYGTt27MDM+OWXX2jQoEHp3ohzzjnAA6/SVBPIL1iJAfINvA4ANcn/XlwJiY2NZfr06QBMmTKFdevWAUFAJokzzjiDNm3a8MgjjwBQuXJlxo0bR3x8PA0aNGDVqlVcffXVZdZ/55w7lPm3GkvPcKBxuBH2rDDtLMCAB8OXkg4HWoR5JhC88X4iUD3Mf6OZfZxfQ5IqAiOAM4G9wLNmNkZSV2Akwe99IXCdme2WlAKcZGabJZ0EjDSzRElDgGOA48Ofo8zsiZz3YmZ3kIddaRnEDHo3v267HFKGnxM1/d///jc333wzDzzwAOeeey5VqlQBgkeNH330EQsXLuTwww+na9eutG3bls6dOzNu3DiWLFnC8ccfz0033cSwYcO4++67S/N2nHPO4YFXaRoExJlZgqSLgP5Aa6AOsFDSh2GegWbWA0DS4UA3M/tVUlNgMnBSAdq6lmD2LMHM0iXVllQNGA90NbM1kl4ErgNG5VNXc6ALwfZJqyWNi7yX3ApJujbsB3Xq1OXe+PQCdNtFSkpKIjU1lU8//ZQdO3aQlJSUde0f//gHAOvWraNevXokJSXxyy+/cMIJJ7BiRfAUukWLFkyZMoWvvvqKn376iXXr1rFu3TqaNm3K5MmT6dSpU1ncVqlKTU3NNm6u4HzsisbHr/DK+9h54FU2OgGTzSwD+EHSBwTbAf2SI19l4ElJCUAGcEIB6z8deMrM0gHMbKuk1sDaiI21JwA3kH/g9a6Z7QZ2S9oEHF2QDpjZMwT7PXLM8U3sseX+T21/pfRKJCkpibi4OKpXr561afWmTZuoV68ee/fupU+fPtxxxx0kJibSunVrunbtSrt27ahSpQoPPvggt956KyeeeCJDhw4lNjaWunXrMnv2bDp27HhIbIKdlJR0SNxnSfCxKxofv8Ir72Pnfw0PbLcCPxDMjFUAfi2hdtL5bb1ftRzXdkccZ1CIfzOHVa7I6lwem7m8PfDAA6xatYrNmzfTqFEjhg4dSmpqKmPHjgXgwgsvpG/fvgDUqlWL2267jZNPPhlJnH322ZxzTjDu9913H507d6Zy5coce+yxjB8/vqxuyTnnDmkeeJWe7QSP6wDmAX+TNAGoDXQm2EC7YUQegKOA9Wa2V1JvoGIB25oV1j8381EjsBqIkdTEzL4GrgQ+CPOnAG2B/wMu2s97cSXonnvuifpffgMGDIia/4orruCKK67YJ71///7079+/uLvnnHNuP/m3GkuJmW0B5oevgegALAOWAnOAO83s+zAtQ9JSSbcC/wJ6S1pKsNZqRwGbew74FlgWlr3czH4F+gJTJC0nWHT/VJh/KDBa0iKCWa0C34ukRwvYJ+ecc+6Q5zNepcjMcr4q4o4c19OA03LkaRVxfFeYLwWIIxfh2q7bwk9k+mzgxCj55xFl/ZiZDclxHhdxfDC89sI555w7oPiMl3POOedcKfEZr4OYpDMI3tcVaa2ZXVAW/XHOOedc3jzwOoiZ2XvAe2XdD+ecc84VjD9qdM4555wrJR54Oeecc86VEg+8nDuAjRgxgnr16hEX99uXWJcuXUqHDh2Ij4/nz3/+M7/8kn3Dg2+//ZYaNWowcuTIrLTHH3+c2NhY4uLi6NmzJ7/+WlLv4nXOOZcXD7ycO4CdeeaZzJgxI1vaNddcw/Dhw1m+fDkXXHABjz6a/VVqt912G2eddVbW+YYNG3jiiSdYtGgRK1asICMjg1deeaVU+u+ccy47D7wOIJISJJ1dgHyppdEfV/Zat25N7dq1s6WtWbOGzp07A9CtWzdef/31rGtvvvkmxx13HLGxsdnKpKens2vXLtLT09m5cycNGjQo+c4755zbh3+r8cCSAJwE/KeM+5ErSRXDzb0LbFdaBjGD3i2pLpVLKXnsbRkbG8v06dM5//zzmTJlCuvWrQMgNTWVESNGMGvWrGyPGRs2bMjAgQM55phjOOyww+jevTvdu3cv8Xtwzjm3L5/xKmaSYiR9KWm8pDWSXpJ0uqT5kr6S1C78fCJpiaSPJTWTVAW4H7hMUrKkyyTVkPSCpOWSlkm6KKKdh8KthT6VdHQe/bkk3NpnqaQPw7SKkkaG6csk3RSmdw37tFzSvyVVDdNTJI2Q9DlwiaTuYf8/lzRFUo0SHVSXzb///W/+9a9/0bZtW7Zv306VKlUAGDJkCLfeeis1amT/dfz0009Mnz6dtWvXsnHjRnbs2MGkSZPKouvOOXfI8xmvktEEuAToBywELgc6AecC/wCuAv4UbmB9OvCwmV0k6V7gJDO7EUDSCGCbmcWH57XC+qsDn5rZYEmPAH8FHsylL/cCZ5jZBkk1w7RrgRggIXMTbUnVgPFAVzNbI+lF4DpgVFhmi5m1kVQHeAM43cx2SLqLYGui+yMblXRt2A516tTl3vj0/RzCQ1tSUhIQzGJ9+umn7NixIysN4B//+AcA69ato169eiQlJTFz5kwmTZrEzTffTGpqKhUqVGDdunXUqlWLatWqsXLlSgBatGjBlClTaNSoUWnfVqlLTU3NNm6u4HzsisbHr/DK+9h54FUy1prZcgBJK4HZZmbh5tQxwFHABElNAQMq51LP6cBfMk/M7KfwcA/wTni8GOiWR1/mA+MlvUYQMGXW+1S4pyNmtlVS67Dfa8I8E4Ab+C3wejX8+UegJcEm2QBVgE9yNmpmzwDPADRr1sxu6nVeHl10uUlKSiIuLo7q1auTmJgIwKZNm6hXrx579+6lT58+3HHHHSQmJrJs2bKsckOGDKFGjRoMHDiQzz77jClTptCuXTsOO+wwXnjhBU4//fSs+sqzpKSkQ+I+S4KPXdH4+BVeeR87D7xKxu6I470R53sJxvwBYK6ZXSApBkjaz/rTzMzC4wzy+D2aWX9J7YFzgMWS2u5nW5l2hD8FzDKznoWsx+2HBx54gFWrVrF582YaNWrE0KFDSU1NZezYsQBceOGF9O3bN8862rdvz8UXX0ybNm2oVKkSJ554Itdee21pdN8551wOHniVjaOADeFxn4j07cAREeezCGadboHgUWPErFeBSGpsZp8Bn0k6C/hDWO/fJM3NfNQIrAZiJDUxs6+BK4EPolT5KTA2M5+k6kDDiJkyV4zuueeeqP/lN2DAgDzLDRkyJNv50KFDGTp0aDH2zDnnXGH44vqy8QgwTNISsge/c4GWmYvrCdZt1cpcHA90KURbj4aL5VcAHwNLgeeAb4FlYb2Xm9mvQF9gSvhIdC/wVM7KzOxHgmBxsqRlBI8ZmxeiX84559whx2e8ipmZpQBxEed9crl2QkSxu8PrW4GTc1TZO0obNSKOpwJT8+jPhVGS0wkWxN+WI+9s4MQodcTkOJ8TpZ/OOeecy4fPeDnnnHPOlRKf8SonJA0meIVFpClm9lBZ9Mc555xz+/LAq5wIAywPspxzzrkDmD9qdM4555wrJR54Oeecc86VEg+8nCtj/fr1o169esTFZX0Zlssuu4yEhASuueYaYmJiSEhIACAtLY3evXsTHx9PixYtGDZsGACrV68mISEh63PkkUcyatSoMrgb55xzefE1XoUkKTXytQ7FUF8fYKaZbSxk+f7ATjN7cT/KxACnmNnL4flJwFVmdrOkRGCPmX0cXhsCpJrZyML0z+WuT58+3HjjjVx11VVZaa++GuzQlJSUxNtvv81RRx0FwJQpU9i9ezfLly9n586dtGzZkp49e9KsWTOSk5MByMjIoGHDhlxwwQWlfi/OOefy5oHXgaMPsAIoVOBlZvu87LQAYgg28H45rGMRsCi8lgikErx01ZWgzp07k5KSEvWamfHaa68xZ84cACSxY8cO0tPT2bVrF1WqVOHII4/MVmb27Nk0btyYY489tqS77pxzbj954FUMJN0BXApUBaaZ2X3hbNL/AR8BpxBsEXSeme2KUv5i4CTgJUm7gA5hmZEEv6OFwHVmtltSCvAacBawi+Ct819HzkhJakLw1vm6BHs5XmJm/43S9eFAC0nJBJtiLwEGAjcC/YEMSVcAN+Xob2NgbFj/TuCvZvZlbuOzKy2DmEHv5j6Ah7CU4efkeX3ZsmUcffTRNG3aFICLL76Y6dOnU79+fXbu3Mnjjz9O7dq1s5V55ZVX6NnTt9J0zrkDkQdeRSSpO9AUaEewgfRbkjoTbMnTFOhpZn+V9BpwETApZx1mNlXSjcBAM1skqRowHuhqZmskvQhcB4wKi2wzs3hJV4VpPXJU+RIw3MymhXXltpZvUNhmj/BeEsP+pEh6iohHi5K6RpR7BuhvZl+FG3D/Czgtx7hcC1wLUKdOXe6NT8+lC4e2pKQkAL7//nt27NiRdZ7pvffeo127dlnpy5cvZ/PmzUyePJnt27czYMAAatSoQYMGDYBgDdjrr79Ojx499qnrUJSamurjUEg+dkXj41d45X3sPPAquu7hZ0l4XoMg4PoWWGtmyWH6YoJHewXRLCybufH0BILNskeF55Mjfj4eWVDSEQSbVk8DCPdgLDaSahDMxk2RlJlcNWc+M3uGIEDjmOOb2GPL/Z9aNCm9EoOfKSlUr14924bY6enpXHjhhTzzzDM0atQICNZ49e7dm9NPPx2At99+m0qVKmWVmz59Ou3bt+fCC6PtFHXoSUpKirrJuMufj13R+PgVXnkfO/9rWHQChpnZ09kSg0eNuyOSMoDDiqlNy+W4NFQAfjazhIIWOKxyRVbn80jN7ev999/nD3/4Q1bQBXDMMccwZ84crrzySnbs2MGnn37KLbfcknV98uTJ/pjROecOYP46iaJ7D+gXzgQhqaGkeoWoZztwRHi8GogJ12oBXAl8EJH3soifn0RWYmbbgfWSzg/7U1XS4QVos0DXzOwXYK2kS8L6Jal1Hvfl8tGzZ086dOjA6tWradSoEc8//zwQrNXq2rVrtrw33HADqampxMbGcvLJJ9O3b19atWoFwI4dO5g1a5bPdjnn3AHMZ7yKyMxmSmoBfBI+eksFriCY4dof44GnIhbX9yV4nJe5uD7yW4u1JC0jmFGLNr1xJfC0pPuBNII9HL+Jkm8ZwQL6pWH7SyKuvQ1MlXQeORbXA72AcZLuBioDrwBLC36rLtLkyZOjpo8fP36fdQ41atRgypQpUfNXr16dLVu2FHf3nHPOFSMPvAop8h1eZjYaGB0lW1xEnjzff2VmrwOvRyTNBk7MJfujZnZXjvJDIo6/Isdi91zaTIuSLym8tgZoFZE+L6LcWuDM/Op3zjnnXHb+qNE555xzrpT4jFcpkzQW6JgjebSZvVCQ8mYWU4g244GJOZJ3m1n7/a3LOeecc4XngVcpM7MbyqDN5UBCabfrnHPOuez8UaNzzjnnXCnxwMs555xzrpR44OVcGejXrx/16tUjLi4uW/qYMWNo3rw5sbGx3HnnnQDs2bOHvn37Eh8fT+vWrbO9YmLy5MnEx8fTqlUrzjzzTDZv3lyat+Gcc24/eeDlXBno06cPM2bMyJY2d+5cpk+fztKlS1m5ciUDBw4E4NlnnwWCfRpnzZrF7bffzt69e0lPT2fAgAHMnTuXZcuW0apVK5588slSvxfnnHMFd8gHXpISJb1T1v3YH5LOl9SyrPvhCq9z587Url07W9q4ceMYNGgQVasGW1/WqxdsgLBq1SpOO+20rLSaNWuyaNEizAwzY8eOHZgZv/zyS9Zm2c455w5M/q3Gg9P5wDvAqjLuR4HsSssgZtC7Zd2NA0JKHntWrlmzhnnz5jF48GCqVavGyJHBO3dbt27NW2+9Rc+ePVm3bh2LFy9m3bp1tGvXjnHjxhEfH0/16tVp2rQpY8eOLa1bcc45VwjlesZLUoykLyW9JOkLSVMlHS7pzDD9c+DCiPxDJA2MOF8R1hG1njzaPVnSx5KWSlog6QhJ1SS9IGm5pCWSuoR5+0h6MqLsO5ISw+NUSQ+F9Xwq6WhJpwDnAo9KSpbUOJc+JEl6XNKisM8nS3pD0leSHozId0XYx2RJT0uqGKaPC8uulDQ0In+KpKGSPg/vpfl+/2JcVOnp6WzdupVPP/2URx99lEsvvRQzo1+/fjRq1IiTTjqJW265hVNOOYWKFSuSlpbGuHHjWLJkCRs3bqRVq1YMGzasrG/DOedcHg6FGa9mwNVmNl/Sv4HbgL8RbJXzNfBqIeu5HthnGyBJVcI6LzOzhZKOBHYBAwAzs/gwWJkp6YR82qwOfGpmgyU9AvzVzB6U9BbwjplNzaf8HjM7SdIAYDrQFtgK/FfS40A9go22O5pZmqR/EezD+CIw2My2hoHYbEmtzGxZWO9mM2sj6XpgIHBNlHG4FrgWoE6dutwbn55PVw8NkQvjv//+e3bs2JGVdvjhh3P88cfzwQfBfuh79uxh48aNfPTRR5x33nmcd955ANx44438/PPPPP/88/z000+sW7eOdevW0bRpUyZPnkynTp1K+7YOWKmpqfvsd+kKxseuaHz8Cq+8j92hEHitM7P54fEk4GZgbbifIZImEQYIhagn2v6LzYDvzGwhgJn9ErbTCRgTpn0p6X9AfoHXHoJHigCLgW4F6Gekt8Kfy4GVZvZd2JdvgD8AnQiCsYXhBt+HAZvCMpeGwVMloD7QkmBTbYA3IvqUNWMYycyeAZ4BaNasmd3U67z97Hr5l5KSQvXq1UlMTASCbzpu3LiRxMRE1qxZQ4UKFWjQoAHt2rXDzKhevTqzZs2idu3a9OnTh40bNzJ06FBiY2OpW7cus2fPpmPHjln1uSDQ9fEoHB+7ovHxK7zyPnaHQuBlOc6PyiNvOtkfv1bLo56c54WVV5tpZpbZTgb7//vaHf7cG3GceV4JEDDBzP4eWUjScQQzWSeb2U+SxufoV2ZdhemTA3r27ElSUhKbN2+mUaNGDB06lH79+tGvXz/i4uKoUqUKEyZMQBKbNm3ijDPOoEKFCjRs2JCJE4Pdnxo0aMB9991H586dqVy5Msceeyzjx48v2xtzzjmXp0Phj+YxkjqY2SfA5cD7wN8kNTaz/wI9I/KmAD0AJLUBjsujno9yaW81UF/SyeGjxiMIHjXOI3iMNyd8xHhMmPdI4HpJFYCGQLsC3NN24IgC5MvPbGC6pMfNbJOk2mG9RwI7gG2SjgbOApKKoT0Xmjx5ctT0SZMmZTtPSkoiJiaG1atXR83fv39/+vfvX+z9c845VzLK9eL60GrgBklfALWAxwkeLb4bLq7fFJH3daC2pJXAjcCaPOoZF60xM9tDsG5qjKSlwCyC2aJ/ARUkLSdYA9bHzHYD84G1BN9QfAL4vAD39ApwR7hIP+ri+oIws1XA3QTrzZaFfa1vZkuBJcCXwMthH51zzjlXRIfCjFe6mV2RI20GsM+38cxsF9A9Z7qkmFzqiSpc3/XHKJf6RslrBDNh0eqpEXE8FZgaHs8nWHOVVx8SI46TiJixynHtVaJ8wcDM+uRSb0zE8SIgMVo+55xzzu3rUJjxcs4555w7IJTrGS8zSwHi8stX2HokTSP7OjCAu8zsvaK2WVCSxgIdcySPNrMXSqsPzjnnnCuYch14lTQzu+AA6MMNZd0H55xzzhWMP2p0zjnnnCslHng555xzzpUSD7ycKwH9+vWjXr16xMX9tjRwyJAhNGzYkISEBBISEvjPf/4DwJYtW+jSpQs1atTgxhtvzFbPnXfeSevWrYmNjaV///5kZGSU6n0455wrXqUeeElKLe02C0LSLXltfF2A8onhBtb7U6bExkJSA0l57uUoqWa432KBy7iC6dOnDzNmzNgn/dZbbyU5OZnk5GTOPvtsAKpVq8YDDzzAyJH77kB13333sXTpUlasWMGPP/7IlClTSrzvzjnnSo7PeP3mFqDQgRfB+6z2K/AqKZIqmdlGM7s4n6w1CTb7BqCAZVwBdO7cmdq1axcob/Xq1enUqRPVqlWLeg0gPT2dPXv2EO6p6Zxz7iBVpt9qlHQHcClQFZhmZveFLyv9P4IteU4BNgDnhS83jVZHE+ApoC7B3oGXAN8AjxBsdWPAg2b2qqREYAiwmeD1EIuBK4CbgAbAXEmbzayLpO7A0LBv/wX6mlmqpBRgAvBnoHLY3q9AfyBD0hXATWY2L0pfjyN4E3wNYHoBxqI68BrQCKgIPBDex8nAaKA6wb6JXYGLCDasrgFUlNQbeMfM4iT1AS4g2KeyITDJzIYCw4HGkpIJ3lo/NqJMNYK3859EsJ/kbWY2N6zrXIIgtXHY1zuj/W4y7UrLIGbQu3llKVdShp+T67Unn3ySF198kZNOOonHHnuMWrVq5VvfGWecwYIFCzjrrLO4+GKPi51z7mBWZoFXGNg0JdibUMBbkjoD34bpPc3sr5JeIwgqJuVS1UvAcDObFgYLFQgCkASgNVAHWCjpwzD/iUAssJFgK5yOZvaEpNuALma2WVIdgq10TjezHZLuAm4D7g/r2GxmbcLHdAPN7BpJTwGpZrbv86LfjAbGmdmLkrJeA5HHWNQFNprZOWG+oyRVIXjT/GXhXpBHEuwFCdAGaGVmW8MANlI7gmBzZzge7wKDgDgzSwjrjyxzA8GL9eMlNSfYVuiE8FpCOI67gdWSxpjZusjGJF1LsDUTderU5d749DyGpXxJSkoC4Pvvv2fHjh1Z561ateL5559HEv/+97+5/PLLueuuu7LKffnll2zYsCErP0Bqaip///vf2bNnDw8++CCPP/44J510UinezcEtNTU123i6gvOxKxofv8Ir72NXljNe3cPPkvC8BkHw8S2w1sySw/TFQEy0CsINqBua2TQAM/s1TO8ETDazDOAHSR8AJwO/AAvMbH2YLzmsO+eG138k2JJnfvhopwrwScT1NyL6duF+3HNHgiASYCIwIjzObSzmAY9JGkEwEzVPUjzwXbgtEWb2S3gvALPMbGsubc8ysy1h3jeATsCbefS1EzAmbONLSf8DMgOv2Wa2LaxrFXAskC3wMrNngGcAjjm+iT22/NB5ZVxKr8TgZ0oK1atXJzExcZ88xx9/PD169Mh2LSUlhdTU1GxpSUlJWefff/89CxYsYODAgSXX+XImcvzc/vGxKxofv8Ir72NXln8NBQwzs6ezJQazLrsjkjKAw4qx3Zx1RxsDEQQqPfOpI7fyebFc2ttnLAAktQHOBh6UNBuYlkfdO/aj3Wj9KKiCjGGWwypXZHUej98OFd999x3169cHYNq0adm+8RhNamoqW7ZsAYI1Xu+++y5/+tOfSryfzjnnSk5ZLq5/D+gnqQaApIaS6u1PBWa2HVgv6fywjqrhNxPnAZdJqiipLtAZWJBPdduBI8LjT4GO4foxJFWPeMxWkPK5mQ/8JTyO3Bg76lhIagDsNLNJwKMEjxJXA/XDdV5IOkJSQYK/bpJqSzoMOD/sS159npfZx/DejwnbdgXQs2dPOnTowOrVq2nUqBHPP/88d955J/Hx8bRq1Yq5c+fy+OOPZ+WPiYnhtttuY/z48TRq1IhVq1axY8cOBg8eTKtWrUhISKBevXr079+/DO/KOedcUZXZjJeZzZTUAvgkfEyWSrDQfX9fVHQl8LSk+4E0gsXu04AOwFKCmZ07zez7cK1Sbp4BZkjaGC6u7wNMllQ1vH43sCaP8m8DUyWdRy6L64EBwMvhmrGsxfV5jEUT4FFJe8N7u87M9ki6DBgTBlG7gNPz6FemBcDrBAv1J5nZIgBJ8yWtIPhCw9iI/P8CxklaTrC4vo+Z7fZv1RXM5MmT90m7+uqrc82fkpISNf2pp54q11Puzjl3qJFZUZ44uYNBGESeZGY35pe3JDRr1sxWr/bJssIo72sdSpqPX+H52BWNj1/hlYexk7TYzKJ+E8rf4+Wcc845V0oOmq+aSRpL8K3ASKPN7IWy6E9eJA0meOQZaYqZPVQW/TGz8cD4smjbOeecc785aAIvM7sh/1wHhjDAKpMgyznnnHMHLn/U6JxzzjlXSjzwcs4555wrJR54OVeM+vXrR7169aK+HPWxxx5DEps3b86WvnDhQipVqsTUqVMB+N///kebNm1ISEigT58+PPXUU6XSd+eccyWvQIGXpMaZ77OSlCjpZkk1S7Rnzh2E+vTpw4wZM/ZJX7duHTNnzuSYY47Jlp6RkcFdd91F9+7ds9Lq16/PJ598QnJyMuPGjWP48OFs3LixxPvunHOu5BV0xut1ICN8k/szwB+Al0usV6VIUmpZ9yEaSbeEb+EvbPlESafkk+d8SS2L0EaR+lgede7cmdq1a++Tfuutt/LII4+Q8wW0Y8aM4aKLLqJevd82bahSpQpVqwbv7d2zZw979+4t2U4755wrNQX9VuNeM0uXdAEwxszGSFqSbylXFLcAk4CdhSyfSPAG/I/zyHM+8A6wqpBt3EIB+rgrLYOYQe8WsomDQ0oee1FOnz6dhg0b0rp162zpGzZsYNq0acydO5eFCxdmu7Zu3TrOOecc1qxZw2OPPUaDBg1KpN/OOedKV0FnvNIk9QR6E/yhBqhcMl0qO5LukLRQ0jJJQ8O0GElfSHpW0kpJM8OtenKro4mk9yUtlfR5+JhWkh6VtELS8nDLn8xZqSRJUyV9KemlMO/NQANgrqS5Yd7ukj4J65wSsa9jiqShYfpySc3Djcb7A7dKSpa0z87K4WzYuQRbEiWH/WwsaYakxZLmhXVVCsckMSw3TNJD0fro9rVz504efvhh7r///n2u3XLLLYwYMYIKFfb9n+Ef/vAHli1bxqRJk5gwYQI//PBDaXTXOedcCSvojFdfgj/kD5nZWknHARNLrlulT1J3oCnQDhDwlqTOwLdhek8z+6uk14CLCGZ6onkJGG5m0yRVIwhuLwQSgNZAHWChpA/D/CcCscBGgo2rO5rZE5JuA7qY2WZJdQj2ijzdzHaEez3eBmT+Nd9sZm0kXQ8MNLNrJD0FpJrZyGidNLOPJb0FvGNmU8MxmA30N7OvJLUH/mVmp4VbDk2VdBNwJtA+3DMyq49RxvNa4FqAOnXqcm98eu6DXw4kJSVlHX///ffs2LGDpKQkvvnmG9asWUOzZs0A+PHHH4mNjWXcuHF89NFHzJsXbOm5bds2pk+fzpdffkmnTp2y6qpWrRq/+93veOqppzj11FNL9Z7Kg9TU1Gy/G1dwPnZF4+NXeOV+7MysQB/gMKBZQfMfLB+C4ARgJJACJIefr4GrgRjgq4j8dwF351LXEcD6KOmPA/0izicSzDYlArMi0scBV4THKUCd8LgHsDmib6uA5yPyNQyP2wPvh8dDCIKwvO59PHBxeFyDYMPt5IjPFxF5/wH8CpwYkZbVx7w+J5xwgh1K1q5da7GxsVGvHXvssfbjjz/uk967d2+bMmWKmZmtW7fOdu7caWZmb731ljVt2tSWLVtWch0ux+bOnVvWXTho+dgVjY9f4ZWHsQMWWS5/Ews04yXpz2FgUgU4TlICcL+ZnVuQ8gcJAcPM7OlsicFju90RSRkEQWhxyVl3tN+JCAK0nvnUkVv5gqgA/GxmCblcjwd+Burlct0BPXv2JCkpic2bN9OoUSOGDh3K1VdfvV91fPHFF9x+++1IIjU1lbvuuov4+PgS6rFzzrnSVNA/0kMIHsElAZhZsqTjS6hPZeU94AFJL5lZqqSGQNr+VGBm2yWtl3S+mb0ZvoKjIjAP+JukCUBtoDNwB9A8j+q2E8ygbQY+BcZKamJmX0uqTjDLtSaf8kfm0+XMNjCzXyStlXSJmU1R8PW7Vma2VNKFEf1+R1I7M/s5Rx8dMHny5Dyvp6SkRE0fP3581nG3bt1YtmwZEDzCTExMLKbeOeecK2sFXlxvZttypJWr77ib2UyCV2R8Imk5MJUwKNlPVwI3S1pG8I3C3wPTgGXAUmAOcKeZfZ9PPc8AMyTNNbMfgT7A5LDeT8g7aAN4G7ggt8X1oVeAOyQtkdQY6AVcLWkpsBI4L1xfNhy4Jgz0ngRG5+xjPn1xzjnnHAWf8Vop6XKgoqSmwM3k/ZqCg4aZ1Yg4Hs1vQUWkuIg8URerR1z/CjgtyqU7wk9k3iTCWcTw/MaI4zHAmIjzOcDJUdqLiTheRLBujDBIapVPX+cDOd/jdWaUrCdElHkitz4655xzLm8FnfG6ieCbd7sJZoW2EbzDyTnnnHPOFVC+M16SKgLvmlkXYHDJd+ngIGks0DFH8mgze6Es+pMXSYOBS3IkTzGzh8qiP84559yhKt/Ay8wyJO2VdFSUdV6HLDO7oaz7UFBhgOVBlnPOOVfGCrrGKxVYLmkWsCMz0cxuLpFeOeecc86VQwUNvN4IP84555xzrpAKtLjezCZE+5R055wrqtWrV5OQkJD1OfLIIxk1ahQAY8aMoXnz5sTGxnLnnXcCsGfPHvr27Ut8fDytW7cu39tWOOecK3UFfXP9WsBypptZeXuJqitnmjVrRnJyMgAZGRk0bNiQCy64gLlz5zJ9+nSWLl1K1apV2bRpEwDPPvssAMuXL2fTpk2cddZZLFy4MOpG1s4559z+Kuhfk5MI3iF1MvAn4Aly3yT6gCOpZriBdLRr4yVdXNp9KgpJ/yjrPhyMZs+eTePGjTn22GMZN24cgwYNomrVqgDUqxfshLRq1SpOO+20rLSaNWuyaNGiMuuzc8658qVAM15mtiVH0ihJi4F7i79LJaImcD3wrzLuR3H5B/BwWXeioHalZRAz6N0yaTtl+DlZx6+88go9ewbbXa5Zs4Z58+YxePBgqlWrxsiRIzn55JNp3bo1b731Fj179mTdunUsXryYdevW0a5duzLpv3POufKloI8a20ScViCYASvsZsxlYTjQWFIyMItgk+tuwDpgT2YmSSnASWa2WdJJwEgzS5Q0BGgMNAHqAI+Y2bO5NSbpLuAKgm2V/s/MBoUbiz8FHA78F+hnZj9JSgIGmtmicHueRWYWI6kPcG6YvzEwzczulDQcOCy8l5Vm1itK+zHADII9Hk8BFgIvAEMJNrnuZWYLwj0fxxC8mb8yMMTMpoflJwLVwypvNLOPJSUS7Nu5OSyzGLgi3Ik9Zx+uBa4FqFOnLvfGp+c2XCUqc41WWloar7/+Oj169CApKYlt27axfPlyhg8fzpdffsm5557Lyy+/TOPGjZk1axbNmzfn6KOPpnnz5nzxxRdlttYrNTXV15kVgY9f4fnYFY2PX+GV97EraPD0WMRxOrAWuLT4u1NiBgFxZpYQbvh8HcFWOUcDq4B/F6COVsAfCYKRJZLeNbONOTNJOgs4D2hvZjsl1Q4vvQjcZGYfSLofuI/83/6fAJxIsGPAakljwiDuRjNLyKdsE4KXpvYjCLwuBzoRBHP/AM4neCHuHDPrJ6kmsEDS+8AmoJuZ/RpuETWZINgm7E8ssBGYT/AS2Y9yNm5mzxDs5cgxxzexx5aXTZye0isRgOnTp9O+fXsuvPBCIFj7ddNNN9GlSxe6dOnCyJEjiYuLo27dunTt2jWr/CmnnMKFF15Iy5Y5d1YqHb5JdtH4+BWej13R+PgVXnkfu4L+NbzazL6JTJB0XAn0pzR0BiabWQawUdKcApabbma7gF3hptDtgDej5DsdeMHMdgKY2VZJRwE1zeyDMM8EYEoB2pyd+dJaSauAYwlm6QpirZktD8uuDOuycAPwmDBPd+BcSQPD82rAMQRB1ZPhLF0GEXs1AgvMbH1Yb3JY1z6BV6TDKldkdcQjv7IwefLkrMeMAOeffz5z586lS5curFmzhj179lCnTh127tyJmVG9enVmzZpFpUqVyizocs45V/4UNPCaCrSJkta2eLtT5tL57QsH1XJcy/k4bZ/HayXQ5u6I4wz27/FuZNm9Eed7I+oRcJGZrY4sGD5a/QFoHfbt12LqU5nYsWMHs2bN4umnn85K69evH/369SMuLo4qVaowYcIEJLFp0ybOOOMMKlSoQMOGDZk4cWIZ9tw551x5k+cfTUnNCR4rHRU+ost0JPsGCQey7cAR4fGHwN8kTSBY79SFYONvgBSCYPL/gIty1HGepGEEjxoTCR5fRjMLuFfSS5mPGsNZr58k/cnM5gFXApmzX5ltLgAK+u3KNEmVzSytgPlz8x5wk6SbwtmwE81sCXAUsN7M9krqDVQsYjtlqnr16mzZkv37IVWqVGHSpH2/mBsTE8Pq1av3SXfOOeeKQ36vk2gG9CD4VuCfIz5tgL+WaM+KUfitzPmSVgAdgK8I1na9CHwSkXUoMFrSIoLZnEjLgLkEC9YfiLa+K2xrBvAWsCh8FJf5GK838KikZQRrt+4P00cC10laQrBwvyCeAZZJeqmA+XPzAMGi+mXh48gHwvR/Ab0lLQWaE7FNlHPOOecKL88ZLzObDkyX1MHMPskr74HOzC4vQJ55ZF/PFGmZmV1VwLaGE3yTMjItmWBxfs68XxIs3M90d5g+Hhgfka9HxPFdwF15tJ9C8K3DzPM+0a6Fa9b+FqX8Vzn6dFeYngQkReS7Mbc+OOecc25fBV2fs0TSDQSPHbMeMZpZvxLplXPOOedcOVTQwGsi8CVwBsEjsl7AFyXVqQONmQ3JmSYpnmBcIu02s/al0qmgD78DZke51DXKS2+dc845V8YKGng1MbNLJJ1nZhMkvQzMK8mOHejCVzUklHEftpR1H5xzzjlXcAXdqzHz23M/S4oj+NZbvZLpknPOOedc+VTQwOsZSbWAewi+sbcKeKTEeuVcMVm9ejUJCQlZnyOPPJJRo0YBMGbMGJo3b05sbCx33nlnVplly5bRoUMHYmNjiY+P59dff82lduecc27/FHST7OfCww+A40uuO84Vr2bNmpGcnAxARkYGDRs25IILLmDu3LlMnz6dpUuXUrVqVTZt2gRAeno6V1xxBRMnTqR169Zs2bKFypUrl+EdOOecK08KNOMl6WhJz0v6v/C8paSrS7ZrTlKipHf2I39KuNG2i2L27Nk0btyYY489lnHjxjFo0CCqVq0KQL16wZPzmTNn0qpVK1q3bg3A7373OypWPKjfH+ucc+4AUtDF9eOBFwg2VQZYA7wKPF8CfXJlTFIlM0svrvp2pWUQM+jd4qquwFJy7A/5yiuvZO3XuGbNGubNm8fgwYOpVq0aI0eO5OSTT2bNmjVI4owzzuDHH3/kL3/5S7bHkM4551xRFHSNVx0ze41gnz/CP8o53+xe7kiqLuldSUslrZB0maS2kj6QtFjSe5Lqh3mbSHo/zPu5pMYKPBqWXS7psjBvoqQkSVMlfSnpJUkKr50Zpn0OXJhH95D0O0kzJa2U9BzB3ouZ166QtEBSsqSnJVUM06+WtCa89qykJ8P08ZKekvQZ8EjY/xnhfc4Lt49CUl1Jr0taGH46lsDQF7s9e/bw1ltvcckllwDBI8WtW7fy6aef8uijj3LppZdiZqSnp/PRRx/x0ksv8dFHHzFt2jRmz472xg7nnHNu/xV0xmtH+M4oA5D0R2BbifXqwHEmsNHMzgGQdBTBPo7nmdmPYSD1ENAPeAkYbmbTJFUjCGovJHjdQ2uC7YAWSvowrPtEghfSbgTmAx3DrYqeBU4DviaYVczLfcBHZna/pHOAq8N+tgAuAzqaWZqkfwG9JL1P8AWJNgT7V84BlkbU1wg4xcwyJM0G+pvZV5LaE2wjdBowGnjczD6SdAzBfo8tcnZM0rXAtQB16tTl3vhim0ArsKSkpKzjjz76iOOOO44vvviCL774gsMPP5zjjz+eDz4Itszcs2cP06dP55dffuGEE05gxYoVALRo0YIpU6aU2ePG1NTUbPfh9o+PX+H52BWNj1/hlfuxM7N8PwR/qOcTBFvzCR41tipI2YP5Q7B9UAowAvgTwVY7vwDJ4Wc5MJNgA+71Uco/DvSLOJ8InEuwyfasiPRxwBUEQdqHEennAu/k0b9k4PiI860EAd6NBAFdZj9XA0OA84EJEflvBp4Mj8cDvcPjGsCuiPLJwBfhtU050jcANfIaxxNOOMHK2mWXXWb//ve/s87HjRtn99xzj5mZrV692ho1amR79+61rVu32oknnmg7duywtLQ069q1q73zzjtl1W2bO3dumbVdHvj4FZ6PXdH4+BVeeRg7YJHl8jcxzxkvSceY2bdm9rmkUwk2zRaw2szS8ipbHpjZGkltgLOBBwlmiFaaWYfIfJKOKET1uyOOMyj47GNBiCDA+nu2ROn8fMplboZdAfjZzBKi5KkA/NHMDpp3LOzYsYNZs2bx9NNPZ6X169ePfv36ERcXR5UqVZgwYQKSqFWrFrfddhsnn3wykjj77LM555xz8qjdOeecK7j81ni9GXH8qpmtNLMVh0LQBSCpAbDTzCYBjwLtgbqSOoTXK0uKNbPtwPrMwEZSVUmHE7zd/zJJFSXVBToDC/Jo8ksgRlLj8LxnPl38ELg8bPMsoFaYPhu4WFK98FptSccCC4FTJdWSVAm4KFqlZvYLsFbSJWF5SWodXp4J3BQxRgn59LHMVa9enS1btnDUUUdlpVWpUoVJkyaxYsUKPv/8c0477bSsa1dccQUrV65kxYoVPPKIv67OOedc8ckv8FLE8aH4/q54YIGkZIL1VPcCFwMjJC0leNR2Spj3SuBmScuAj4HfA9OAZQTrqOYAd5rZ97k1Fs4iXQu8Gy6u35RP/4YCnSWtJFhP9m1YzyrgbmBm2J9ZQH0z2wA8TBD8zSd4jJrbWr1ewNXhfa4EzgvTbwZOkrRM0iqgfz59dM4551wov8dblsvxIcHM3iNYPJ5T5yh5vyJYfJ7THeEnMm8SkBRxfmPE8QygeQH7twXonsu1V4m+OP9lM3smnPGaRjiraWZ9cpRfS/Dlgpz1biZYuO+cc865/ZRf4NVa0i8EM1+HhceE52ZmR5Zo71xJGCLpdKAawWPDN8u2O84559yhI8/Ay8z8ld0HAEl9gQE5kueb2Q37W5eZDSyeXjnnnHNufxXnN+lcCTGzFwh2DnDOOefcQaygb653zjnnnHNF5IGXc84551wp8cDLlWurV68mISEh63PkkUcyatQoAMaMGUPz5s2JjY3NthH2sGHDaNKkCc2aNeO996J9qdU555wrHF/j5cq1Zs2akZycDEBGRgYNGzbkggsuYO7cuUyfPp2lS5dStWpVNm0KXpm2atUqXnnlFVauXMnGjRs5/fTTWbNmTZnt1eicc6588RmvA4ikGEmXF7GOW8K35ueVJzVKWgNJU/ezrT6SntzfPpaV2bNn07hxY4499ljGjRvHoEGDqFq1KgD16tUDYPr06fzlL3+hatWqHHfccTRp0oQFC/LabMA555wrOJ/xOrDEEGwB9HIR6rgFmATs3J9CZraR4K38xW5XWgYxg94tiarzlDI8+x6Lr7zyCj17BrswrVmzhnnz5jF48GCqVavGyJEjOfnkk9mwYQN//OMfs8o0atSIDRs2lGq/nXPOlV8eeBUjSVcBAwne8r8MuAf4N1AH+BHoa2bfShoP/AKcRLC10J1mNhUYDrQItyiaADwRpiUCVYGxZva0pERgCLAZiAMWA1cQ7KHYAJgrabOZdcmnv3WAtwk2AF8JvGNmcZL6AOcChwONgWlmdmdYpi/wd+Bngq2Qdu9bM0i6lmD7I+rUqcu98en5jl9xS0pKyjpOS0vj9ddfp0ePHiQlJbFt2zaWL1/O8OHD+fLLLzn33HN5+eWX2bBhA1988UVW2e+++46VK1dSp06dUu8/QGpqarb7cPvHx6/wfOyKxsev8Mr72HngVUwkxRLsj3iKmW2WVJsgeJpgZhMk9SMIpM4Pi9QHOhFsD/QWMBUYBAw0sx5hndcC28zsZElVgfmSZoblTwRigY0E+y52NLMnJN0GdAm39smrv0eH7d5tZrMkxeTIkhC2sRtYLWkMkE6wP2Rbgj0e5wJLotVvZs8AzwAcc3wTe2x56f9TS+mVmHU8ffp02rdvz4UXXggEa79uuukmunTpQpcuXRg5ciRxcXG0a9cOgMTEoOywYcPo3r07HTp0KO3uA0HwmNkXt/98/ArPx65ofPwKr7yPnQdexec0YEpmwGNmWyV1INi8GmAi8EhE/jfNbC+wKgyCoukOtJKU+QjwKKApsAdYYGbrAcIZshjgowL2tTIwG7jBzD7IJc9sM9sW1r8KOJZg5i7JzH4M018FTsivscMqV2R1jsd+pW3y5MlZjxkBzj//fObOnUuXLl1Ys2YNe/bsoU6dOpx77rlcfvnl3HbbbWzcuJGvvvoqKxhzzjnnisoDr7IT+YhOueQRcFO4WfdvicGjxsjyGezf7zKd4PHkGUBugVdR6j+g7Nixg1mzZvH0009npfXr149+/foRFxdHlSpVmDBhApKIjY3l0ksvpWXLllSqVImxY8f6Nxqdc84Vm4P2j+kBaA4wTdI/zWxL+KjxY+AvBLNdvYB5+dSxHTgi4vw94DpJc8wsTdIJQH4rvTPryOtRowH9gCmS7jKzEfnUmekzYLSk3xGsUbuEYJ3XAa169eps2bIlW1qVKlWYNGlS1PyDBw9m8ODBpdE155xzhxgPvIqJma2U9BDwgaQMgrVPNwEvSLqDcHF9PtUsAzIkLQXGA6MJHiF+LklhHefnU8czwAxJG/NaXG9mGZJ6Am9J2g78J596MbPvJA0BPiFYXJ+cXxnnnHPO/cYDr2JkZhMIFtRHOi1Kvj45zmuEP9Oi5P9H+ImUFH4yy98YcTwGGJNPPzPb203wuDFTXJg+niDwy8zfI+LYN+x2zjnnCslfoOqcc845V0p8xqsck/QZwfu/Il1pZsvLoj/OOefcoc4Dr3LMzNqXdR+cc8459xt/1Oicc845V0o88HLlzs8//8zFF19M8+bNadGiBZ988gmXXXYZCQkJJCQkEBMTQ0JCQlb+YcOG0aRJE5o1a8Z7772Xe8XOOedcEfmjRlfuDBgwgDPPPJOpU6eyZ88edu7cyauvvpp1/fbbb+eoo44CYNWqVbzyyiusXLmSjRs3cvrpp7NmzRp/aapzzrkScdDMeElKLeb6+khqUITy/cNNsYvSh/ER2wEVO0nPSWqZT55s41CQMgeybdu28eGHH3L11VcDwYtSa9asmXXdzHjttdeytg+aPn06f/nLX6hatSrHHXccTZo0YcGCBWXRdeecc4eAQ3nGqw+wgmCT6f1mZk8Va2+KmaSKZnZNAbL2IWIcClhmv+xKyyBm0LvFXW1Ub/6lIXXr1qVv374sXbqUtm3bMnr0aKpXrw7AvHnzOProo2natCkAGzZs4I9//GNW+UaNGrFhQ36bAzjnnHOFc9DMeEWSdIekhZKWSRoapsVI+kLSs5JWSpop6bBcyl8MnAS8JClZ0mGSukpaImm5pH9LqhrmTZH0SJi+QFKTMH2IpIHhcRNJ70taKulzSY1zaVeSnpS0WtL7QL2Ia20lfSBpsaT3JNUP02+WtCq811fCtBqSXgj7tEzSRWF6qqTHwjffd5CUJOmkiGuPh2MzW1LdXMYhskzPsI0VkkZE9DVV0kPh/X6axybfpS49PZ3PP/+c6667jiVLllC9enWGDx+edT3nZtnOOedcaTroZrwkdQeaAu0INpF+S1Jn4NswvaeZ/VXSa8BFwD4b8pnZVEk3AgPNbJGkagRvau9qZmskvQhcB4wKi2wzs/jw0eIooEeOKl8ChpvZtLCu3ALaC4BmQEvgaGAV8G9JlQneNn+emf0o6TLgIYL9FAcBx5nZbkk1w3ruyexTOCa1wvTqwGdmdnuYHtl2dWCRmd0q6V7gPjO7MXIcIsuEjx9HAG2Bn4CZks43szfDuj41s8GSHgH+CjwY2Zika4FrAerUqcu98em5DEnx+vbbb6lTpw67du0iKSmJxo0b8/LLL9O1a1cyMjJ49dVXefrpp0lKSgJg9+7dfPDBBzRq1AiAZcuW0aZNm6zrZS01NfWA6cvByMev8HzsisbHr/DK+9gddIEX0D38LAnPaxAEXN8Ca80sOUxfTLDPYUE0C8uuCc8nADfwW+A1OeLn45EFJR0BNDSzaQBm9mse7XQGJptZBrBR0pyI9uOAWWHgUxH4Lry2jGBG6k3gzTDtdILNtwnb/Ck8zABez6XtvUDmCvNJwBt59BPgZCDJzH4M7/OlsP9vAnuAd8J8i4FuOQub2TME+0ZyzPFN7LHlpfNPLWX4eTz++OPUr1+fZs2akZSUxJ/+9CcSExOZMWMG8fHxXHLJJVn569aty+WXX86TTz7Jxo0b2bJlC/379z9gFtcnJSWRmJhY1t04aPn4FZ6PXdH4+BVeeR+7gzHwEjDMzJ7OlijFALsjkjKAqI8aC8FyOS4uAlaaWYco184hCHj+DAyWFJ9HPb+GQV1BFOU+0swss3wG+fw7OqxyRVYPP6cIze2fMWPG0KtXL/bs2cPxxx/PCy8EW0u+8sor+zxmjI2N5dJLL6Vly5ZUqlSJsWPHHjBBl3POufLnYFzj9R7QT1INAEkNJdXLp0w024EjwuPVQEzm+i3gSuCDiLyXRfz8JLISM9sOrJd0ftifqpIOz6XND4HLJFUM13B1iWi/rqQOYR2VJcVKqgD8wczmAncBRxHM8M0imJEjzF+L/FUAMr9BeTnwUXgcOQ6RFgCnSqojqSLQk+xjcsBKSEhg0aJFLFu2jDfffJNatYLhGT9+PP37998n/+DBg/nvf//L6tWrOeuss0q7u8455w4hB13gZWYzgZeBTyQtB6YSPXDIz3jgKUnJBDNOfYEpYZ17gchvLdaStAwYANwapa4rgZvDPB8Dv8+lzWnAVwRru14kDOLMbA9BUDQiXBifDJxC8MhxUtinJcATZvYzwXqqWuGi96X8FsDlZQfQTtIK4DTg/pzjEPllBDP7jmB92VxgKbDYzKYXoB3nnHPO5eKgedRoZjUijkcDo6Nki4vIMzKf+l4n+3qo2cCJuWR/1MzuylF+SMTxVwTBTJ7Cx3M35nItmeCRYk6douRNBXpHSa+R4zwxx/ltUcrkHIfEiGuT+W19W9R2zGwqQfDrnHPOuXwcdDNezjnnnHMHq4NmxquwJI0FOuZIHm1mLxSkvJnFFKLNeGBijuTdZtZ+f+sqLjlnw5xzzjlX+sp94GVmN+Sfq9jbXA4klHa7zjnnnDuw+aNG55xzzrlS4oGXc84551wp8cDLlRs///wzF198Mc2bN6dFixZ88sknbN26lW7dutG0aVO6devGTz8FL/l/9NFHSUhIICEhgbi4OCpWrMjWrVvL+A6cc86Vdx54uXJjwIABnHnmmXz55ZcsXbqUFi1aMHz4cLp27cpXX31F165dszbMvuOOO0hOTiY5OZlhw4Zx6qmnUrt27TK+A+ecc+WdB16ApCGSBpZ1PwpK0i15vB3/kLRt2zY+/PBDrr76agCqVKlCzZo1mT59Or17B6886927N2+++eY+ZSdPnrzPVkLOOedcSSj332osp24h2Oh6Zxn3o0B2pWUQM+jdEqs/Zfg5rF27lrp169K3b1+WLl1K27ZtGT16ND/88AP169cH4Pe//z0//PBDtrI7d+5kxowZPPnkkyXWP+eccy7TIRt4SRpM8Pb3TcA6YLGkJGCgmS2SVAdYZGYxkvoAFxDsldgQmGRmQ/Oo+ypgIMFG1MvM7MpwE+9/A3WAH4G+ZvatpPHAO+Eb4JGUamY1JCUCQ4DNBG/kXwxcAdwENADmStpsZlG3C5KUCowDzga+A/4BPAIcA9xiZm+FezAOJ3hbfVVgrJk9He6DOR2oBVQG7jaz6eE9/B/BPo+nABuA88xsV5T2rwWuBahTpy73xqfnNlxFlpSUxOrVq1m8eDF9+vShT58+jBkzhuuuu4709HSSkpKy8mZkZGQ7nzNnDs2bN2fZsmUl1r+iSE1NzdZft398/ArPx65ofPwKr7yP3SEZeElqC/yF4F1blYDPCQKbvLQjCIB2AgslvWtmi6LUHQvcDZxiZpslZS4cGgNMMLMJkvoBTwDn59PmiUAssBGYD3Q0syck3QZ0MbPNeZStDswxszskTSPY37Eb0BKYALwFXA1sM7OTJVUF5kuaSRCIXmBmv4QB6KeS3grrbQr0NLO/SnoNuIhg9i0bM3sGeAbgmOOb2GPLS+6fWkqvRJo3b86wYcO4/vrrAahYsSLDhw+nYcOGNGvWjPr16/Pdd9/RoEEDEhMTs8qOHj2aG2+8MVvagSQpKemA7dvBwMev8HzsisbHr/DK+9gdkoEX8CdgmpntBIgIKvIyy8y2hPnfINhDcZ/Ai2DPximZQZGZZX5VrgNwYXg8kWD2KT8LzGx92GYyEEMw21QQe4AZ4fFygjfnp4UbbseE6d2BVpIuDs+PIgis1gMPS+pMsGF4Q+DoMM/acF9JCILVzLpydVjliqwefk4Bu104v//97/nDH/7A6tWradasGbNnz6Zly5a0bNmSCRMmMGjQICZMmMB5552XVWbbtm188MEHTJq0T9zonHPOlYhDNfDKTTq/feGgWo5rls95kduUVAGoEnFtd8RxBvv3+0oLN+WGIHjaDWBmeyVl1iPgJjN7L7Jg+Gi1LtA2DNZS+G08cvbpsP3oU4kaM2YMvXr1Ys+ePRx//PG88MIL7N27l0svvZTnn3+eY489ltdeey0r/7Rp0+jevTvVq1cvw14755w7lByqgdeHwHhJwwjG4M/A00AK0BZYAFyco0y38LHhLoJHhP1yqXsOME3SP81si6Ta4azXxwSPNycCvYB5Yf7MNl8DziVYU5Wf7cARBOu/iuI94DpJc8IA6wSCdVtHAZvCtC7AsUVsp1QkJCSwaNG+k5CzZ8+Omj9zPZhzzjlXWg7JwMvMPpf0KrCUYHH9wvDSSOC1cGF4zq/hLQBeBxoRLK6P9pgRM1sp6SHgA0kZwBKgD8Gi+Bck3UG4uD4s8iwwXdJSgkeDOwpwC88AMyRtzG1xfQE9R/Co8HNJCvt1PvAS8Hb4WHIR8GUR2nDOOedc6JAMvADM7CHgoSiXWkUc3x1xvN7Mzi9g3RMIFrBHpv2PYP1Xzrw/AH+MSLorTE8CkiLy3RhxPIZgsX5efagRcTwk2jUz20vwbcd/RKmiQy5Vx0XUMzKvPjjnnHMuO3+BqnPOOedcKTlkZ7z2h5mNB8ZHpkn6HRBt8VDXzG8/lgZJnxG8gyvSlWa2vLT64JxzzrmC8cCrkMLgKuEA6Ef7su6Dc8455wrGHzU655xzzpUSD7zcQSsjI4MTTzyRHj16AMH2P23atCEuLo7evXuTnp59m6KFCxdSqVIlpk6dWhbddc455zzwcgev0aNH06JFCwD27t1L7969eeWVV1ixYgXHHnssEyb89sXSjIwM7rrrLrp3715W3XXOOec88DpQSEqU9E5Z92N/SDpfUsuyaHv9+vW8++67XHPNNQBs2bKFKlWqcMIJJwDQrVs3Xn/99az8Y8aM4aKLLqJevXpl0V3nnHMO8MX1rmjOB94BVuWVaVdaBjGDcr6PtvBShp/DLbfcwiOPPML27dsBqFOnDunp/9/encdXVZ37H/98GUSZjBT0StUGhyIEMBIKDlSDLQ7VW0RQVKpQ9IdKtQ5XBYvFqVYUbaGoRbQWil6gTA6oWAocRBwAZRZRFHoRsUwFiSKG8Pz+2CvhEE6Sk+kkxOf9ep1X9ll7TXu5Wx/XXmevPSxatIgOHTowefJk1q9fD8CGDRuYNm0ac+bMYeHChcVV7ZxzzlUqn/FKAUnpkj6U9LykVZImS6ov6fyQ/j77NtBG0r2Sbo/7viLUkbCeYtr9kaS3JC2VtEBSI0mHSvqrpOWSFoctgZDUV9LjcWWnS8oOxzmSHgz1vCPpKElnEG1xNEzSEkknVPjAFWH69OkceeSRZGVlxV8rEyZM4NZbb6Vjx440atSI2rVrA3DLLbfw8MMPU6uW3+7OOeeqls94pU5L4Bozmy/pWeA24Dqit9mvASaWsZ4BRFsd7UfSIaHOXma2UFJjon0mbwbMzNpKOhn4R9ijsTgNgHfMbLCkR4D/Z2a/k/QSMN3MDlitHrZd6g/QtGkzhrTdUzhLmY0f/3f+8Y9/MHXqVL799lu+/vprunbtyuDBg3nggQeAaCF9WloasViMN998k3nzoq0xd+zYwYsvvsiHH35I586dK6xPlSUnJ4dYLFbV3Tho+fiVnY9d+fj4lV1NHzsPvFJnvZnND8fPAb8G1prZxwCSniMEKmWoJ9HWPS2BjWa2EMDMvgztdCZsN2RmH0r6F1BS4PUt0SNFgPeAriV10sxGE+0pyXHHn2iPLa+4W23d888XHMdiMR599FGmT5/Opk2bOPLII9m9ezcPPPAAQ4YMITs7m40bNxbk79u3LxdddBE9exbeA716isViZGdnV3U3Dlo+fmXnY1c+Pn5lV9PHzgOv1LFC3w8vJu8e9n8MfGgx9RT+XlbFtZlrZvnt5FHK++awurVZPfTCcnavZMOGDWP69Ons3buXG264gXPOOWBrTOecc65KeeCVOsdJOt3M3gauBP4JXCfpBDP7BLgiLu864CIASe2BFsXU82YR7a0Gjpb0o/CosRHRo8Z5QG9gdnjEeFzI2xgYIKkW8H2gYxLXtBNolES+SpOdnV3wX0bDhg1j2LBhxeYfM2ZM5XfKOeecK4KvNk6d1cCvJK0CjgD+SPRo8ZWwuH5TXN4pQBNJK4EbgY+KqefPiRozs2+BXsBISUuBmUSzWE8CtSQtJ1oD1tfMdgPzgbVEv1D8E/B+Etc0AbgjLNJP2eJ655xz7mDlM16ps8fMflEobQZwcuGMZrYLOOBNn5LSi6gnobC+67QEp36ZIK8RzYQlqqdh3PFkYHI4ng9UyXu8nHPOuYORz3g555xzzqWIz3ilgJmtA9pUVj2SprH/OjCAgWb2ennbdM4551zF8cCrBjCz7lXdB+ecc86VzB81Ouecc86liAdezjnnnHMp4oGXq/a++eYbOnbsyCmnnEJGRgb33HMPAD/+8Y/JzMwkMzOT5s2bc/HFFwPw/PPP065dO9q2bcsZZ5zB0qVLq7D3zjnn3D6+xstVe/Xq1WP27Nk0bNiQ3NxcOnfuzAUXXFCw/yJAjx496NatGwAtWrRg7ty5HHHEEbz22mv079+fd999t6q675xzzhXwGa8qIClN0oBylH9VUlopy9wr6faytpmqOotoh4YNo1eJ5ebmkpubi6SC819++SWzZ88umPE644wzOOKIIwA47bTT+Oyzzyq7i84551xSfMaraqQBA4jeIl9qZvazCu1NJduVm0f6oFfKVHZd2OMxLy+PrKws1qxZw69+9Ss6depUkOeFF17gJz/5CY0bNz6g/F/+8hcuuOCCsnXcOeecq2A+41U1hgInSFoiaVj4rJC0XFIvAEnZkt6Q9Iqk1ZJGhX0UkbROUtNwfLWkZZKWShqXTOOSTpA0Q9J7kuZJOlnS4ZL+FddGA0nrJdVNlL+SxqVItWvXZsmSJXz22WcsWLCAFStWFJwbP348V1xxxQFl5syZw1/+8hcefvjhVHbVOeecK5LPeFWNQUAbM8uU1AO4HjgFaAoslPRGyNeRaEuefxFtL3QJYbseAEkZwN3AGWa2RVKTJNsfDVxvZh9L6gQ8aWbnSFoCnA3MIdqk+3Uzy5V0QH7gnOIakNSfaC9KmjZtxpC2e5Ls2v5isdgBaenp6TzxxBP06tWLHTt28NZbb3Hrrbful/eTTz5hyJAhDB06lOXLl5ep7eogJycn4Ri45Pj4lZ2PXfn4+JVdTR87D7yqXmdgvJnlAf+WNBf4EfAlsMDMPgWQND7knRxX9hxgkpltATCzbSU1JqkhcAYwKW6dVL3wdyLRxtpzgMuBJ0vIXyQzG00U4NGyZUu7qXe3kooUafPmzdStW5e0tDR27drFb3/7WwYOHEh2djajRo3i4osv5txz921t+X//939ce+21TJo0iTPOOKPM7VYHsViM7Ozsqu7GQcvHr+x87MrHx6/savrYeeBVvVkJ38uiFrDdzDITnHsJ+H2YOcsCZgMNismfEhs3bqRPnz7k5eWxd+9eLrvsMi666CIAJkyYwKBBg/bLf//997N161YGDIh+v1CnTh0WLVqU8n4755xzhXngVTV2Ao3C8TzgOkljgSbAWcAdwMlAR0ktiB419iLMIMWZDUyT9Acz2yqpSUmzXmb2paS1ki41s0mKprHamdlSM8uRtBAYAUwPs3BF5q+YoShZu3btWLx4ccJziaajn3nmGZ555plK7pVzzjlXer64vgqY2VZgvqQVwOnAMmApUSB1p5l9EbIuBB4HVgFrgWmF6lkJPAjMlbQU+EOSXegNXBPKrATinwNOBH4R/iaT3znnnHNJ8hmvKmJmVxZKuiNBti/N7KIEZdPjjscCY5No796447XA+UXkmwyoUFrC/PF1Ouecc65kPuPlnHPOOZciPuNVTZlZDIiVtpykwcClhZInmdmDFdAt55xzzpWDB141TAiwPMhyzjnnqiF/1Oicc845lyIeeLlq7ZtvvqFjx46ccsopZGRkcM899wDQt29fWrRoQWZmJpmZmSxZsgSAYcOGFaS1adOG2rVrs21bie+Vdc4551LCHzW6aq1evXrMnj2bhg0bkpubS+fOnQs2vR42bBg9e/bcL/8dd9zBHXdEPxB9+eWX+eMf/0iTJsnupOScc85VLp/xShFJaZIGFHFujKSeic5VV5J+k6J2aNiwIQC5ubnk5uYSt3VRsYraPNs555yrKh54pU4akDDwOkilJPACyMvLIzMzkyOPPJKuXbvSqVMnAAYPHky7du249dZb2b17935lvv76a2bMmEGPHj1S1U3nnHOuRP6oMXWGAidIWgLMBA4DugLrgW/zM0laB3Qwsy2SOgCPmlm2pHuBE4ATgabAI2b2dFGNSRpI9Ab6vcBrZjZIUiYwCqgPfAL0M7P/SIoBt5vZIklNgUVmli6pL/DzkP8EYJqZ3SlpKHBYuJaVZta7uAvflZtH+qBXkh+pYN3QCwGoXbs2S5YsYfv27XTv3p0VK1bw0EMP8V//9V98++239O/fn4cffpghQ4YUlH355Zc588wz/TGjc865asUDr9QZBLQxs0xJlwA3AK2Bo4APgGeTqKMdcBrRxtWLJb1iZp8XziTpAqJtfTqZ2ddh02uAvwE3mdlcSfcD9wC3lNBmJnAqsBtYLWlkCOJuLG7jbEn9gf4ATZs2Y0jbPUlc3v4S7cOYnp7OE088Qa9evVi9ejUAp556KhMnTuSss84qyPf4449z9tlnJ6zjYJKTk3PQX0NV8vErOx+78vHxK7uaPnYeeFWNs4DxYRPqzyXNTrLci2a2C9glaQ7QEXghQb6fAn81s68BzGybpMOBNDObG/KMBSYl0eYsM9sBIOkD4AdEs3TFMrPRhE29jzv+RHtseelvtXW9s9m8eTN169YlLS2NXbt28dvf/paBAwfSsmVLjj76aMyMF154gbPPPpvs7GwAduzYwcqVK5kxYwYNGjQodbvVSSwWK7guV3o+fmXnY1c+Pn5lV9PHzgOv6mcP+9beHVronJXwvTLajF88lUcZ7pnD6tZmdXhsWFobN26kT58+5OXlsXfvXi677DIuuugizjnnHDZv3oyZkZmZyahRowrKTJs2jXPPPfegD7qcc87VPB54pc5OoFE4fgO4TtJY4EigC/C/4dw6IAt4DSi8MrybpIeIHjVmEz2+TGQmMETS8/mPGsOs138k/djM5gFXAfmzX/ltLgCS/XVlrqS6ZpabZP4yadeuHYsXLz4gffbsoicJ+/btS9++fSuxV84551zZ+K8aU8TMtgLzJa0ATgc+Jlrb9Tfg7bis9wEjJC0immGKtwyYA7wDPJBofVdoawbwErAoLIC/PZzqAwyTtIxo7db9If1R4AZJi4kW7idjNLBM0vNJ5nfOOee+83zGK4XM7Mok8swDfljE6WVmdnWSbQ0l+iVlfNoSosX5hfN+SLRwP9/dIX0MMCYu30VxxwOBgcn0xTnnnHMRn/FyzjnnnEsRn/E6SJjZvYXTJLUFxhVK3m1mnVLSKeecc86VigdeBzEzW060Vss555xzBwF/1Oicc845lyIeeDnnnHPOpYgHXq7aWb9+PV26dKF169ZkZGQwYsQIAHr16kVmZiaZmZmkp6eTmZkJwMyZM8nKyqJt27ZkZWUV+44v55xzrir5Gi9X7dSpU4fHHnuM9u3bs3PnTrKysujatSsTJ04syPM///M/HH744QA0bdqUl19+mebNm7NixQrOO+88NmzYUFXdd84554rkM17VhKR7Jd1ecs7qQdItkupXRt1HH3007du3B6BRo0a0atVqv0DKzPj73//OFVdcAUSbZDdv3hyAjIwMdu3axe7duw+s2DnnnKtiPuPlyuoW4Dng65Iy7srNI33QK0lVuq7Qno7r1q1j8eLFdOq07w0Z8+bN46ijjuKkk046oPyUKVNo37499erVS6o955xzLpV8xqsKSRos6SNJbwItQ1pMUodw3FTSunDcV9KL4fzHku4poe6rJS2TtFTSuJCWLml2SJ8l6biQPkZSz7iyOeFvdmhvsqQPJT2vyK+B5sAcSXMqfmQiOTk59OjRg+HDh9O4ceOC9PHjxxfMdsVbuXIlAwcO5KmnnqqsLjnnnHPl4jNeVURSFnA50Xu46gDvA++VUKwj0IZolmmhpFfMbFGCujOItv05w8y2SGoSTo0ExprZWEn9gD8BF5fQ5qlABvA5MB8408z+JOk2oIuZbSni+voD/QGaNm3GkLZ7SmgmEovFANizZw933XUXnTp1okmTJgXpeXl5TJw4kaeeeqogDWDz5s3cdttt3Hnnnaxfv57169cn1V51l5OTs991utLx8Ss7H7vy8fEru5o+dh54VZ0fA9PM7GsASS8lUWZm2GwbSVOBzsABgRdwDjApPygys20h/XTgknA8DngkiTYXmNlnoc0lQDrwZkmFzGw00UbatGzZ0m7q3S2JpgrK0qdPH84880yGDx++37kZM2bQtm1bLr300oK07du3c/bZZzNixAguueQSapJYLEZ2dnZVd+Og5eNXdj525ePjV3Y1fez8UWP1s4d9/1wOLXTOSvhe7jYl1QIOiTsXv0o9jxQE6/Pnz2fcuHHMnj274PURr776KgATJkw44DHj448/zpo1a7j//vsL8m/atKmyu+mcc86Vms94VZ03gDGSHiL65/DfwFPAOiALWAD0LFSma3hsuIvoEWG/IuqeDUyT9Acz2yqpSZj1eovo8eY4oDcwL+TPb/PvwM+Bukn0fyfQCEj4qLE8OnfujFnimHLMmDEHpN19993cfffdFd0N55xzrsL5jFcVMbP3gYnAUuA1YGE49Shwg6TFQNNCxRYAU4BlwJRE67tC3SuBB4G5kpYCfwinbgJ+KWkZcBVwc0h/Gjg75D0d+CqJSxgNzKjMxfXOOedcTeMzXlXIzB4kCpAKaxd3HD+V85mZXZxk3WOBsYXS/kW0/qtw3n8Dp8UlDQzpMSAWl+/GuOORRIv1nXPOOZckn/FyzjnnnEsRn/E6SJjZGGBMfJqk7wGzEmT/Sf6vH51zzjlXfXjgdRALwVVmVffDOeecc8nxR43OOeeccynigZerFtavX0+XLl1o3bo1GRkZjBgxAoBJkyaRkZFBrVq1WLRo3484FyxYUPDOrlNOOYVp06ZVVdedc865pPmjRlct1KlTh8cee4z27duzc+dOsrKy6Nq1K23atGHq1Klcd911++Vv06YNixYtok6dOmzcuJFTTjmF//7v/6ZOHb+lnXPOVV8+41UESWmSBpSj/KuS0kpZ5l5Jt5e1zYPZ0UcfTfv27QFo1KgRrVq1YsOGDbRq1YqWLVsekL9+/foFQdY333yDpJT21znnnCsLD7yKlgaUOfAys5+Z2fYK6813yLp161i8eDGdOnUqNt+7775LRkYGbdu2ZdSoUT7b5Zxzrtrzf1MVbShwQtgYemZIu4Bof8TfmdlESdnA/UTb55wIzAEGmNleSeuADma2RdLVwO2h7DIzu6qkxiX9P6A/0b6Ja4CrzOxrSWOAL4EOwH8Bd5rZ5LDH4uNEL0hdD+QCz4Zz8X3pADxqZtmSOgIjiPaE3AX80sxWS6pP9OqKNsBqoDnwKzNbJOlc4D6gHvBJKJNT3LXsys0jfdArxV7vuqEXAtGu9D169GD48OE0bty42DKdOnVi5cqVrFq1ij59+nDBBRdw6KGFt7d0zjnnqg8PvIo2CGhjZpmSegDXA6cQbeOzUNIbIV9HoDXwL2AGcAkwOb8SSRlEb58/IwQ+TZJsf6qZPR3q+B1wDfveFH800Bk4GXgptHcJkB76ciSwCni2hDY+BH5sZnsk/RT4PdCDaKbvP2bWWlIbYEnoR9NwLT81s68kDQRuIwo+9yOpP1HgSNOmzRjSdk+xHYnFYuzZs4e77rqLTp060aRJE2KxWMH57du3895775GTkzjG27NnD2PHjk34WPJglpOTs984uNLx8Ss7H7vy8fEru5o+dh54JaczMN7M8oB/S5oL/Iho5mmBmX0KIGl8yDs5ruw5wCQz2wIQNqtORpsQcKUBDYHX4869YGZ7gQ8kHRXXx0kh/Ysk91A8HBgr6SSi2bj8zbE7E82EYWYrwt6OEG0r1BqYH9ZUHQK8nahiMxtNtJ8jxx1/oj22vPhbbe2VZ9OnTx/OPPNMhg8ffsD5tLQ0srKy6NChQ5R/7VqOPfZY6tSpw7/+9S+++OILevToQdOmhbe3PLjFYjGys7OruhsHLR+/svOxKx8fv7Kr6WPngVf5WQnfy2oMcLGZLZXUF8iOO7c77jiZVeV72LeeL/5Z3APAHDPrLimduH0ZiyBgppldkUSbBQ6rW5vV4VFiUd58803GjRtH27ZtyczMBOD3v/89u3fv5qabbmLz5s1ceOGFZGZm8vrrr/Pmm28ydOhQ6tatS61atXjyySdrXNDlnHOu5vHAq2g7gUbheB5wnaSxQBPgLOAOokd9HSW1IHrU2IswyxNnNjBN0h/MbKukJknOejUCNkqqC/QGNpSQfz7QJ/SxGVGg9r/h3DogC3iN6FFivsPj6u1bqK7LgDmSWgNtQ/o7wBOSTjSzNZIaAN83s4+SuJ5ide7cGbPEMWv37t0PSLvqqqu46qoSl8o555xz1Yr/qrEIYTue+ZJWAKcDy4ClRIHUnWb2Rci6kGhR+ypgLTCtUD0rgQeBuZKWAn9Isgu/Bd4lCoI+TCL/FOAz4APgOeB9YEc4dx8wQtIiIC+uzCPAQ5IWs38Q/iTQTNIHwO+AlcAOM9tMFKCND48f3yYKPp1zzjmXBJ/xKoaZXVko6Y4E2b40s4sSlE2POx4LjE2ivXvjjv8M/DlBnr6FvjcMf/dKut3McsLm2QuA5eHcPOCHCep6u1D63eHvN8AvzOwbSScA/ySa0cPMZhOtb3POOedcKXngVbNMDy9tPQR4IG5WrrTqEz1mrEu0rmuAmX1bQX10zjnnvrM88CoHM4tR8oL0A0gaDFxaKHmSmT1Yzv5kl6d8XD07id4T5pxzzrkK5IFXFQgBVrmCLOecc84dfHxxvXPOOedcinjg5ZxzzjmXIh54uWph/fr1dOnShdatW5ORkcGIESMAmDRpEhkZGdSqVYtFixYV5J85cyZZWVm0bduWrKwsZs+eXVVdd84555Lma7xctVCnTh0ee+wx2rdvz86dO8nKyqJr1660adOGqVOnct111+2Xv2nTprz88ss0b96cFStWcN5557FhQ0nvmHXOOeeqls94VTBJ6ZIKv/+rtHXcIql+CXleDa+OqNB6iyh3f9hEu9IcffTRtG/fHoBGjRrRqlUrNmzYQKtWrRJufH3qqafSvHlzADIyMti1axe7d+8+IJ9zzjlXnfiMV8VLB65k33Y9ZXEL0dvnvy4qg5n9rCLrlVQ7bAKeqK0hZWirwK7cPNIHvVLk+XWF9nFct24dixcvplOnTknVP2XKFNq3b0+9evXK003nnHOu0vmMVyGSrpa0TNJSSePCDNbskDZL0nEh3xhJf5L0lqRPJfUMVQwFfixpiaRbJdWWNEzSwlDHdaF8tqSYpMmSPpT0vCK/BpoTvcB0TjH9XCepqaQGkl4J/V0hqVcR+Q+oV1KOpMfCVkanSxoS+rlC0mhJirvWnnHt3ifpfUnLJVXolkE5OTn06NGD4cOH07hx4xLzr1y5koEDB/LUU09VZDecc865SuEzXnEkZRBtm3OGmW2R1IRoq5+xZjZWUj/gT8DFocjRQGei/QpfAiYDg4Db87cRktSfaJ/DH0mqR7T/4z9C+VOBDOBzoj0ZzzSzP0m6DehiZluS6Pb5wOdmdmFo7/BEmYqotwHwrpn9Tyj7gZndH47HARcBLyeobouZtZc0ALgduLZwhnDd/QGaNm3GkLZ7iryAWCwGwJ49e7jrrrvo1KkTTZo0KUgH2L59O++99x45OTkFaZs3b+a2227jzjvvZP369axfv77INg5WOTk5+42DKx0fv7LzsSsfH7+yq+lj54HX/s4heoP8FgAz2ybpdOCScH4c0cbS+V4ws73AB5KOKqLOc4F2cTNihwMnAd8CC8zsMwBJS4geU75Zyj4vBx6T9DAwPezLmKw8os2183WRdCfRlkFNiDbHThR4TQ1/32Pf2OzHzEYDowFatmxpN/XuVmxHzIw+ffpw5plnMnz48APOp6WlkZWVRYcO0Qv1t2/fztlnn82IESO45JKEXagRYrEY2dnZVd2Ng5aPX9n52JWPj1/Z1fSx80eN5RO/mltF5BFwk5llhk8LM8uf8Yovn0cZAmEz+whoTxSA/U5SadZjfZO/rkvSocCTQE8zaws8DRxaRLn8fpepz4nMnz+fcePGMXv2bDIzM8nMzOTVV19l2rRpHHPMMbz99ttceOGFnHfeeQA8/vjjrFmzhvvvv78g/6ZNmyqiK84551yl8Rmv/c0Gpkn6g5ltDY8a3wIuJ5rt6g2UNKO0E2gU9/114AZJs80sV9IPgZLee5BfR4mPGiU1B7aZ2XOStpPgsV+S9eYHWVskNQR6Ej06TYnOnTtjZgnPde/e/YC0u+++m7vvvruyu+Wcc85VKA+84pjZSkkPAnMl5QGLgZuAv0q6A9gM/LKEapYBeWHB+hhgBNEjxPfDYvXN7FsjVpTRwAxJn5tZlxLytgWGSdoL5AI3lKVeM9su6WlgBfAFsLCEdp1zzjlXSh54FWJmY4kW1Mc7J0G+voW+Nwx/cxPk/034xIuFT375G+OORwIjS+hnejh8PXxKVLje/D7Hfb+b6McFhcv1TdAuZrYIyE6mbeecc875Gi/nnHPOuZTxGa9qTtK7QOE3g15lZsuLKTMNaFEoeaCZJTUz5pxzzrnK4YFXNWdmyb2+ff8yB65Gd84551yV80eNzjnnnHMp4oGXqzL9+vXjyCOPpE2bNgVpkyZNIiMjg1q1arFo0aL98j/00EOceOKJtGzZktdf96emzjnnDj4eeKWQpFsk1S9Dub7hfV01St++fZkxY8Z+aW3atGHq1KmcddZZ+6V/8MEHTJgwgZUrVzJjxgwGDBhAXl7CPb2dc865assDr9S6hWg7nqRJqg30JdrgukY566yzaNKkyX5prVq1omXLlgfkffHFF7n88supV68eLVq04MQTT2TBggWp6qpzzjlXITzwqiSSGkh6RdJSSSsk3UMUPM2RNCfk+bOkRZJWSrovruw6SQ9Leh+4AugAPC9piaTDJA2V9IGkZZIeLaYPl4a2l0p6I6SlS5on6f3wOSOkZ0uaK+lFSZ+GNnpLWiBpuaQTQr5mkqZIWhg+Z1baIMbZsGEDxx57bMH3Y445hg0bStoAwDnnnKte/FeNled84HMzuxBA0uFEb73vkr8JNzA4bMRdG5glqZ2ZLQvntppZ+1D2WuB2M1sk6XtAd+BkMzNJacX0YQhwnpltiMu3CehqZt9IOgkYTxTYAZwCtAK2AZ8Cz5hZR0k3E73B/xaiN/H/0czelHQc0ctbWxU3ELty80gf9ErB93VDLywuu3POOVdjeeBVeZYDj0l6GJhuZvOiHYP2c5mk/kT/HI4GWhNtOQQwsYh6dwDfAH+RNB2YXkwf5gNjJP0dmBrS6gKPS8ok2uT6h3H5F5rZRgBJnwD5m3kvB/K3GPop0DruWhpLamhmOfENh+vqD9C0aTOGtN1TcC4WixUcf/HFF3z11Vf7pQFs376d9957j5ycqNrdu3czd+5cjjnmGACWLVtG+/btDyhX0+Tk5NT4a6xMPn5l52NXPj5+ZVfTx84Dr0piZh9Jag/8DPidpFnx5yW1AG4HfmRm/5E0hn0bVQN8VUS9eyR1BH5CtJH1jSTY0ijkvV5SJ+BC4D1JWUQzV/8mmt2qRRTE5dsdd7w37vte9t0rtYDTzCy+XKK2RxPtDclxx59ojy3fd6ut652973jdOho0aEB2dvZ+5dPS0sjKyqJDh2gyrlmzZlx55ZU8/vjjfP7552zdupXrr7+e2rVrF9eNg14sFjtgbFzyfPzKzseufHz8yq6mj50HXpUk/Apxm5k9J2k7cC2wE2gEbAEaEwVXOyQdBVxA3N6NheSXQ1JDoL6ZvSppPtEjwaL6cIKZvQu8K+kC4FjgcOAzM9srqQ9Q2sjlH0TB27DQRqaZLSmuwGF1a7M6wePFK664glgsxpYtWzjmmGO47777aNKkCTfddBObN2/mwgsvJDMzk9dff52MjAwuu+wyWrduTZ06dXjiiSdqfNDlnHOu5vHAq/K0BYZJ2gvkAjcApwMzJH1uZl0kLQY+BNYTPRYsyhhglKRdRAHai5IOBQTcVky5YWEdl4BZwFLgSWCKpKuBGRQxs1aMXwNPSFpGdP+8AVxfyjoAGD9+fML07t0Tv3h/8ODBDB48uCxNOeecc9WCB16VJOyLWPgtn4uAkXF5+hZRNr3Q9ynAlLikjkn24ZIEyR8D7eK+Dwx5Y8TNuJlZdtxxwbnww4BeybTvnHPOuf356yScc84551LEZ7xqAEmDgUsLJU8ysweroj/OOeecS8wDrxogBFgeZDnnnHPVnD9qdM4555xLEQ+8nHPOOedSxAMv55xzzrkU8cDLpdzq1avJzMws+DRu3Jjhw4cDMHLkSE4++WQyMjK48847q7ajzjnnXAXzxfUu5Vq2bMmSJUsAyMvL4/vf/z7du3dnzpw5vPjiiyxdupR69eqxadOmqu2oc845V8F8xqsGkHSvpNsrqK6+YbujlJg1axYnnHACP/jBD/jzn//MoEGDqFevHgBHHnlkqrrhnHPOpYQHXq6wvkCpAi9Jxc6c7srNI33QK6QPeuWAcxMmTOCKK64A4KOPPmLevHl06tSJs88+m4ULF5amG84551y154FXKUlqIOkVSUslrZDUS9IQSQvD99GSFPLGJP1R0iJJqyT9SNJUSR9L+l1cnb+QtEDSEklPSSpy92dJ50t6P7Q/K+5U69Dep5J+HZf/BUnvSVopqX9cek7o20pJsyQ1k9QT6AA8H/pymKQsSXNDHa9LOjru2oZLWgTcXJax/Pbbb3nppZe49NLo3a979uxh27ZtvPPOOwwbNozLLrsMMytL1c4551y15Gu8Su984HMzuxBA0uHATDO7P3wfB1wEvBzyf2tmHSTdDLwIZAHbgE8k/RE4kmjvwzPNLFfSk0Bv4G+FG5bUDHgaOMvM1kpqEnf6ZKAL0AhYLenPZpYL9DOzbZIOAxZKmmJmW4EGwCIzu1XSEOAeM7tR0o3A7Wa2SFJdor0lu5nZZkm9iF7U2i+0eYiZdUg0SCHI6w/QtGkzhrTdA0AsFivI8+abb9KiRQtWrVrFqlWrqF+/Pscffzxz586NBu7bb3nxxRdJS0sr9h9ITZaTk7PfmLnS8fErOx+78vHxK7uaPnYeeJXecuAxSQ8D081snqQeku4E6gNNgJXsC7xeiiu30sw2Akj6FDgW6EwUjC0ME2WHAUWtKj8NeMPM1gKY2ba4c6+Y2W5gt6RNwFHAZ8CvJXUPeY4FTgK2AnuBiSH9OWBqgvZaAm2AmaFvtYGNcecnJihD6NtoYDRAy5Yt7abe3Q7IM2rUKAYMGEB2djYA/fr14/PPPyc7O5uPPvqIWrVq0a1bN0Lb30mxWKxgfFzp+fiVnY9d+fj4lV1NHzsPvErJzD6S1B74GfC78LjvV0AHM1sv6V7g0Lgiu8PfvXHH+d/rAALGmtld5exafN15QB1J2cBPgdPN7GtJsUJ9i5fomZ6IgsXTiyjzVdm6Cl999RUzZ87kqaeeKkjr168f/fr1o02bNhxyyCGMHTv2Ox10Oeecq3k88Cql8Iu/bWb2nKTtwLXh1BZJDYGewORSVDkLeFHSH81sU3h82MjM/pUg7zvAk5Ja5D9qLDTrVdjhwH9C0HUy0YxZvlqhrxOAK4E3Q/pOoseVAKuBZpJON7O3w6PHH5rZylJcX0INGjRg69at+6UdcsghPPfcc+Wt2jnnnKu2PPAqvbbAMEl7gVzgBuBiYAXwBVCqn+KZ2QeS7gb+IalWqPNXwAGBV1hn1R+YGvJuAroWU/0M4HpJq4iCqHfizn0FdAxtbyJaZwYwBhglaRdwOlFw9qewlq0OMJzoUapzzjnnSskDr1Iys9eB1wslLwLuTpA3O+44BsSKODeRYtZLFarzNeC1Qmn3FvreJu7rBcXUdVuCtCnAlLikJcBZCfJlJ9Nf55xzzu3jr5NwzjnnnEsRn/GqpiS9C9QrlHyVmS2viPrNrGFF1OOcc8655HngVU2ZWaeq7oNzzjnnKpY/anTOOeecSxEPvJxzzjnnUsQDL5dy27dvp2fPnpx88sm0atWKt99+mzvuuIOTTz6Zdu3a0b17d7Zv317V3XTOOecqnAdeSZCUJmlAOcq/KimtAruUqI3fVGb9Fenmm2/m/PPP58MPP2Tp0qW0atWKrl27smLFCpYtW8YPf/hDHnrooarupnPOOVfhPPBKThpQ5sDLzH5mZtsrrDeJVXrgJancP8bYsWMHb7zxBtdccw0Qva0+LS2Nc889lzp1oupPO+00Pvvss/I25ZxzzlU7HnglZyhwgqQlkoaFzwpJyyX1ApCULekNSa9IWi1pVHi7PJLWSWoajq+WtEzSUknjimpQ0lGSpoV8SyWdEdJfkPSepJXhLfZIGgocFvr3fEj7haQFIe0pSbVD+jWSPgrnnpb0eEhPlzQ79G2WpONC+phwLe8Cj0j6WFKzcK6WpDX534uyKzeP9EGvkD7oFdauXUuzZs345S9/yamnnsq1117LV1/tv+Xjs88+ywUXFPneV+ecc+6gJbNEeyO7eJLSgelm1kZSD+B64HygKdEWQZ2AlkRb9LQm2u5nBvCUmU2WtA7oABwFTAPOMLMtxe21KGki8LaZDQ9BU0Mz25FfRtJhoe2zzWyrpJz8d3NJagU8AlxiZrmSniTaLuifwFtAe6I9GWcDS83sRkkvA5PNbKykfsDPzexiSWPCdXYzszxJ9wA7Qr/OBa4zsx4J+t8f6A/QtGmzrCHDnwbgkJwvGDBgACNHjqR169aMHDmSBg0a0K9fPwCee+45Vq9ezf333+8bZAM5OTk0bOivXCsrH7+y87ErHx+/sqsJY9elS5f3zKxDonP+Hq/S6wyMN7M84N+S5gI/Ar4EFpjZpwCSxoe88RtmnwNMMrMtACVscH0OcHXIlwfsCOm/ltQ9HB8LnARsLVT2J0AWsDAEL4cR7cfYEZib366kScAPQ5nTgUvC8TiiwC3fpNAHgGeBF4n2bOwH/DVR581sNDAa4LjjT7THlke32ju3dOOhhx5iwIDoyW3t2rUZOnQo2dnZjBkzhpUrVzJr1izq169fzNB8d8RiMbKzs6u6GwctH7+y87ErHx+/sqvpY+eBV8UqPH1YodOJkrKBnwKnm9nXkmLAoYmyAmPN7K5C5S8uY9MFzwLNbL2kf0s6hyiQ611S4cPq1mb10AsLvh977LGsXr2ali1bMmvWLFq3bs2MGTN45JFHmDt3rgddzjnnaixf45WcnUCjcDwP6CWpdljbdBawIJzrKKlFWNvVC3izUD2zgUslfQ9AUpNi2pwF3BDy1ZZ0OHA48J8QdJ0MnBaXP1dS3biyPSUdmd+OpB8QHk1KOiIslI9/RPgWcHk47h2usyjPAM+x/0xY0kaOHEnv3r1p164dS5Ys4Te/+Q033ngjO3fupGvXrmRmZnL99deXtlrnnHOu2vMZrySENVTzJa0AXgOWAUuJZrTuNLMvQiC0EHgcOBGYQ7SeK76elZIeBOZKygMWA32LaPZmYLSka4A8oiBsBnC9pFXAaqJ1W/lGA8skvW9mvSXdDfwjBIG5wK/M7B1JvycKFLcBH7LvEeZNwF8l3QFsBn5ZzJC8RPSIMeFjxpJkZmayaNGi/dLWrFlTlqqcc865g4oHXkkysysLJd2RINuXZnZRgrLpccdjgbFJtPdvoFuCUwl/7mdmA4GBcd8nAhMTZP1fMxsdZrymAS+E/P8iWldWuN6+Ceo4hWhR/ofFX4Vzzjnn4vmjxu+eeyUtAVYAawmBV7IkDQKmAHeVlNc555xz+/MZrwpiZjEgVtpykgYDlxZKnmRmD1ZAtw5gZreXs/xQoveaOeecc66UPPCqYiHAqpQgyznnnHPViz9qdM4555xLEQ+8nHPOOedSxAMv55xzzrkU8cDLOeeccy5FPPByzjnnnEsRD7ycc84551LEAy/nnHPOuRSRmVV1H1wNJ2kn0d6SrvSaAluquhMHMR+/svOxKx8fv7KrCWP3AzNrluiEv0DVpcJqM+tQ1Z04GEla5GNXdj5+ZedjVz4+fmVX08fOHzU655xzzqWIB17OOeeccynigZdLhdFV3YGDmI9d+fj4lZ2PXfn4+JVdjR47X1zvnHPOOZciPuPlnHPOOZciHni5SiXpfEmrJa2RNKiq+1MdSDpW0hxJH0haKenmkN5E0kxJH4e/R4R0SfpTGMNlktrH1dUn5P9YUp+quqZUk1Rb0mJJ08P3FpLeDWM0UdIhIb1e+L4mnE+Pq+OukL5a0nlVdCkpJSlN0mRJH0paJel0v++SJ+nW8L/ZFZLGSzrU772iSXpW0iZJK+LSKux+k5QlaXko8ydJSu0VlpGZ+cc/lfIBagOfAMcDhwBLgdZV3a+q/gBHA+3DcSPgI6A18AgwKKQPAh4Oxz8DXgMEnAa8G9KbAJ+Gv0eE4yOq+vpSNIa3Af8LTA/f/w5cHo5HATeE4wHAqHB8OTAxHLcO92M9oEW4T2tX9XWlYNzGAteG40OANL/vkh677wNrgcPi7rm+fu8VO2ZnAe2BFXFpFXa/AQtCXoWyF1T1NSfz8RkvV5k6AmvM7FMz+xaYAHSr4j5VOTPbaGbvh+OdwCqi/1PvRvQvRsLfi8NxN+BvFnkHSJN0NHAeMNPMtpnZf4CZwPmpu5KqIekY4ELgmfBdwDnA5JCl8Njlj+lk4CchfzdggpntNrO1wBqi+7XGknQ40b8I/wJgZt+a2Xb8viuNOsBhkuoA9YGN+L1XJDN7A9hWKLlC7rdwrrGZvWNRFPa3uLqqNQ+8XGX6PrA+7vtnIc0F4fHDqcC7wFFmtjGc+gI4KhwXNY7f1fEdDtwJ7A3fvwdsN7M94Xv8OBSMUTi/I+T/Lo5dC2Az8NfwmPYZSQ3w+y4pZrYBeBT4P6KAawfwHn7vlVZF3W/fD8eF06s9D7ycqyKSGgJTgFvM7Mv4c+G/4Pwnx4VIugjYZGbvVXVfDkJ1iB77/NnMTgW+InrUU8Dvu6KFtUjdiALY5kADvjszfZXiu3q/eeDlKtMG4Ni478eEtO88SXWJgq7nzWxqSP53mD4n/N0U0osax+/i+J4J/FzSOqJH1+cAI4geS+RvgRY/DgVjFM4fDmzluzl2nwGfmdm74ftkokDM77vk/BRYa2abzSwXmEp0P/q9VzoVdb9tCMeF06s9D7xcZVoInBR+9XMI0QLTl6q4T1UurPP4C7DKzP4Qd+olIP8XO32AF+PSrw6/+jkN2BGm6l8HzpV0RPiv8XNDWo1lZneZ2TFmlk50P802s97AHKBnyFZ47PLHtGfIbyH98vDLsxbASUQLdWssM/sCWC+pZUj6CfABft8l6/+A0yTVD/8bzh8/v/dKp0Lut3DuS0mnhX8eV8fVVb1V9ep+/9TsD9EvVT4i+uXO4KruT3X4AJ2JpteXAUvC52dE6z9mAR8D/wSahPwCnghjuBzoEFdXP6LFuWuAX1b1taV4HLPZ96vG44n+5bUGmATUC+mHhu9rwvnj48oPDmO6moPk11AVMGaZwKJw771A9Csxv++SH7/7gA+BFcA4ol8m+r1X9HiNJ1oPl0s043pNRd5vQIfwz+IT4HHCS+Gr+8ffXO+cc845lyL+qNE555xzLkU88HLOOeecSxEPvJxzzjnnUsQDL+ecc865FPHAyznnnHMuRTzwcs65IkjKk7Qk7pNehjoultS6ErqHpOaSJpecs0LbzJT0s1S26VxNUqfkLM459521y8wyy1nHxcB0opdtJkVSHdu3/1+RzOxz9r28s9KFN7BnEr0/6dVUtetcTeIzXs45VwqSsiTNlfSepNfjtj/5f5IWSloqaUp4w/kZwM+BYWHG7ARJMUkdQpmmYfsjJPWV9JKk2cAsSQ0kPStpQdjUuluCvqRLWhFX/gVJMyWtk3SjpNtC2XckNQn5YpJGhP6skNQxpDcJ5ZeF/O1C+r2SxkmaT/TS0PuBXqF8L0kdJb0d2nkr/834oT9TJc2Q9LGkR+L6fb6k98NYzQppJV6vczWBz3g551zRDpO0JByvBS4DRgLdzGyzpF7Ag0Rv1p5qZk8DSPodcI2ZjZT0EtEb9ieHc8W11x5oZ2bbJP2eaJuZfpLSgAWS/mlmXxVTvg1wKtFb09cAA83sVEl/JNpSZXjIV9/MMiWdBTwbyt0HLDaziyWdA/yNaHYLoDXQ2cx2SepL9FbxG8P1NAZ+bGZ7JP0U+D3QI5TLDP3ZDayWNBL4BngaOMvM1uYHhERvcy/t9Tp30PHAyznnirbfo0ZJbYiClJkhgKpNtCUKQJsQcKUBDSnb/oUzzWxbOD6XaEPw28P3Q4HjgFXFlJ9jZjuBnZJ2AC+H9OVAu7h84wHM7A1JjUOg05kQMJnZbEnfC0EVwEtmtquINg8Hxko6iWgrrLpx52aZ2Q4ASR8APyDapugNM1sb2irP9Tp30PHAyznnkidgpZmdnuDcGOBiM1saZoWyi6hjD/uWeRxa6Fz87I6AHma2uhT92x13vDfu+172///7wnvFlbR3XHGzTg8QBXzdw48PYkX0J4/i/51Tlut17qDja7yccy55q4Fmkk4HkFRXUkY41wjYKKku0DuuzM5wLt86ICscF7cw/nXgJoWpNUmnlr/7BXqFOjsDO8Ks1DxCvyVlA1vM7MsEZQtfz+HAhnDcN4m23wHOktQitJX/qLEyr9e5asMDL+ecS5KZfUsULD0saSmwBDgjnP4t8C4wH/gwrtgE4I6wYPwE4FHgBkmLgabFNPcA0WO7ZZJWhu8V5ZvQ/ijgmpB2L5AlaRkwFOhTRNk5QOv8xfXAI8BDob4Sn6KY2WagPzA1jOHEcKoyr9e5akNmJc0wO+ecqykkxYDbzWxRVffFue8in/FyzjnnnEsRn/FyzjnnnEsRn/FyzjnnnEsRD7ycc84551LEAy/nnHPOuRTxwMs555xzLkU88HLOOeecSxEPvJxzzjnnUuT/AxADNhqDYv38AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "lgb.plot_importance(booster=model, figsize=(8, 8))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "9a54c183",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9922440428757475"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "res_df[\"pred\"] = y_oof\n",
    "\n",
    "roc_auc_score(res_df[\"target\"], res_df[\"pred\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "9502b0bf",
   "metadata": {},
   "outputs": [],
   "source": [
    "#res_df.loc[~res_df[\"content_is_train\"], \"pred\"] *= 0.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 468,
   "id": "92947283",
   "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>topic_id</th>\n",
       "      <th>pred</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>t_000d1fb3f2f5</td>\n",
       "      <td>c_b8d730238789 c_1ec97b588bce c_7868f655c31e c...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>t_0016d30772f3</td>\n",
       "      <td>c_c25053d6fafd c_61b851222e17 c_ea312de91d4f c...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>t_001bcbb22694</td>\n",
       "      <td>c_1d9dfc709413</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>t_0020bde404c1</td>\n",
       "      <td>c_aa8f7f0cbeb3 c_085d324f601b c_978da5beda25</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>t_00260f878951</td>\n",
       "      <td>c_86f126d7f1f8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8243</th>\n",
       "      <td>t_ffb0a70e5e48</td>\n",
       "      <td>c_f3b75faf5480 c_81f88e566297 c_f7c218334f49 c...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8244</th>\n",
       "      <td>t_ffb1d959761e</td>\n",
       "      <td>c_9b993553de45</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8245</th>\n",
       "      <td>t_ffba5459a977</td>\n",
       "      <td>c_a46e0ec1377b c_04a421dba8aa c_787a7a2e7217 c...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8246</th>\n",
       "      <td>t_ffdc013937fc</td>\n",
       "      <td>c_2e959f78384e</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8247</th>\n",
       "      <td>t_ffe8df837f62</td>\n",
       "      <td>c_ea0a2006e2af c_37e4bfb234da c_a0a99e889a41 c...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>8248 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            topic_id                                               pred\n",
       "0     t_000d1fb3f2f5  c_b8d730238789 c_1ec97b588bce c_7868f655c31e c...\n",
       "1     t_0016d30772f3  c_c25053d6fafd c_61b851222e17 c_ea312de91d4f c...\n",
       "2     t_001bcbb22694                                     c_1d9dfc709413\n",
       "3     t_0020bde404c1       c_aa8f7f0cbeb3 c_085d324f601b c_978da5beda25\n",
       "4     t_00260f878951                                     c_86f126d7f1f8\n",
       "...              ...                                                ...\n",
       "8243  t_ffb0a70e5e48  c_f3b75faf5480 c_81f88e566297 c_f7c218334f49 c...\n",
       "8244  t_ffb1d959761e                                     c_9b993553de45\n",
       "8245  t_ffba5459a977  c_a46e0ec1377b c_04a421dba8aa c_787a7a2e7217 c...\n",
       "8246  t_ffdc013937fc                                     c_2e959f78384e\n",
       "8247  t_ffe8df837f62  c_ea0a2006e2af c_37e4bfb234da c_a0a99e889a41 c...\n",
       "\n",
       "[8248 rows x 2 columns]"
      ]
     },
     "execution_count": 468,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "res_df[\"rank\"] = res_df.groupby(\"topic_id\")[\"pred\"].rank(method=\"first\", ascending=False)\n",
    "\n",
    "res_df[\"gap\"] = res_df.groupby(\"topic_id\")[\"pred\"].transform(\"max\") - res_df[\"pred\"]\n",
    "\n",
    "res_df[\"content_dist\"] = res_df[\"gap\"] + res_df.groupby(\"content_id\")[\"pred\"].transform(\"max\") - res_df[\"pred\"]\n",
    "\n",
    "res_df[\"content_rank\"] = res_df.groupby(\"content_id\")[\"content_dist\"].rank(method=\"first\")\n",
    "\n",
    "pred_df = res_df[(res_df[\"rank\"] == 1) | \n",
    "                 ((res_df[\"gap\"] < 0.25) & res_df[\"content_is_train\"] & (res_df[\"pred\"] > 0.05)) | \n",
    "                 (~res_df[\"content_is_train\"] & (res_df[\"content_rank\"] == 1) & (res_df[\"content_dist\"] < 0.65)) | \n",
    "                 (~res_df[\"content_is_train\"] & (res_df[\"gap\"] < 0.05))]\n",
    "\n",
    "\n",
    "pred_df = pred_df.groupby(\"topic_id\")[\"content_id\"].apply(lambda x: \" \".join(list(x)))\n",
    "pred_df = pred_df.reset_index().rename(columns={\"content_id\": 'pred'})\n",
    "\n",
    "pred_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 469,
   "id": "057ab1ff",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(8248, 3)"
      ]
     },
     "execution_count": 469,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pred_df = pred_df.merge(pd.read_csv(\"data/correlations.csv\"), on=\"topic_id\")\n",
    "pred_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 470,
   "id": "df6faec5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7642550450912223"
      ]
     },
     "execution_count": 470,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metric(pred_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e48730f0",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c0089d22",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "5ff63b03",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(354, (8248,))"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x = res_df.groupby(\"topic_id\")[\"target\"].max()\n",
    "ix = x[x == 0].index\n",
    "len(ix), x.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "eee417b8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1260, 34)"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "difficult_df = topic_df.set_index(\"topic_id\").loc[ix].merge(corr_df, on=\"topic_id\").merge(content_df, on=\"content_id\")\n",
    "difficult_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "45697286",
   "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>5</th>\n",
       "      <th>494</th>\n",
       "      <th>52</th>\n",
       "      <th>988</th>\n",
       "      <th>186</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>topic_id</th>\n",
       "      <td>t_0139f08b50fc</td>\n",
       "      <td>t_62242dcf093f</td>\n",
       "      <td>t_0d5145117aa3</td>\n",
       "      <td>t_b3adbe9834d1</td>\n",
       "      <td>t_21c8907f5d3c</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_title</th>\n",
       "      <td>Lectura 5</td>\n",
       "      <td>6 Las lluvias intensas y los problemas que oca...</td>\n",
       "      <td>B4.2.2.2 Use reading readiness and word identi...</td>\n",
       "      <td>Trigonometric Identity Proofs</td>\n",
       "      <td>Multiplication and Division</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_description</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>B4.2.2.2.1 Recognise and read words using a va...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_channel</th>\n",
       "      <td>6e90a7</td>\n",
       "      <td>f65044</td>\n",
       "      <td>9d5527</td>\n",
       "      <td>01dbc1</td>\n",
       "      <td>7db7f2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_category</th>\n",
       "      <td>aligned</td>\n",
       "      <td>supplemental</td>\n",
       "      <td>aligned</td>\n",
       "      <td>supplemental</td>\n",
       "      <td>supplemental</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_level</th>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_language</th>\n",
       "      <td>es</td>\n",
       "      <td>es</td>\n",
       "      <td>en</td>\n",
       "      <td>en</td>\n",
       "      <td>en</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_parent</th>\n",
       "      <td>t_74ea1148a50c</td>\n",
       "      <td>t_007239952dfd</td>\n",
       "      <td>t_5de8e7f60cf4</td>\n",
       "      <td>t_e41d316189a6</td>\n",
       "      <td>t_02359b8e34ca</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_has_content</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_parent_title</th>\n",
       "      <td>01 Unidad 1</td>\n",
       "      <td>Exploramos y aprendemos</td>\n",
       "      <td>Sub-Strand 2: Phonics</td>\n",
       "      <td>Understanding Trigonometric Functions</td>\n",
       "      <td>Mathematics</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_grandpa</th>\n",
       "      <td>t_c0125c7d0a1f</td>\n",
       "      <td>t_8ea654fa6e92</td>\n",
       "      <td>t_44cc3efd077e</td>\n",
       "      <td>t_7004866e0ffe</td>\n",
       "      <td>t_97faaaa83d71</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_grandpa_title</th>\n",
       "      <td>Español Lectura 1ro</td>\n",
       "      <td>1° y 2° grado</td>\n",
       "      <td>Strand 2</td>\n",
       "      <td>Trigonometric Functions</td>\n",
       "      <td>Level 1 (Grade 4)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_ggrandpa</th>\n",
       "      <td>t_7eb0dc459971</td>\n",
       "      <td>t_6e977f750f41</td>\n",
       "      <td>t_b17d1e99f128</td>\n",
       "      <td>t_0396c73309a7</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_ggrandpa_title</th>\n",
       "      <td>01 Primero</td>\n",
       "      <td>Aprendo en Casa - Primaria</td>\n",
       "      <td>Basic 4 - English</td>\n",
       "      <td>Algebra 2</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_fold</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_t</th>\n",
       "      <td>Lectura 5 @ 01 Unidad 1 @ Español Lectura 1ro ...</td>\n",
       "      <td>6 Las lluvias intensas y los problemas que oca...</td>\n",
       "      <td>B4.2.2.2 Use reading readiness and word identi...</td>\n",
       "      <td>Trigonometric Identity Proofs @ Understanding ...</td>\n",
       "      <td>Multiplication and Division @ Mathematics @ Le...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_label</th>\n",
       "      <td>18549</td>\n",
       "      <td>3680</td>\n",
       "      <td>6576</td>\n",
       "      <td>30308</td>\n",
       "      <td>20861</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_chapter</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>key</th>\n",
       "      <td>t_74ea1148a50c</td>\n",
       "      <td>t_007239952dfd</td>\n",
       "      <td>t_5de8e7f60cf4</td>\n",
       "      <td>t_e41d316189a6</td>\n",
       "      <td>t_02359b8e34ca</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>content_id</th>\n",
       "      <td>c_a4236dd7ded9</td>\n",
       "      <td>c_4f65d6c00631</td>\n",
       "      <td>c_93b8b3ea4c10</td>\n",
       "      <td>c_deb3dfaeb7d1</td>\n",
       "      <td>c_657cacfea9c5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>label</th>\n",
       "      <td>18549</td>\n",
       "      <td>3680</td>\n",
       "      <td>6576</td>\n",
       "      <td>30308</td>\n",
       "      <td>20861</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>target</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>content_title</th>\n",
       "      <td>Ramón el ratón</td>\n",
       "      <td>02-A-Orientaciones para mi planficación</td>\n",
       "      <td>Word Stems</td>\n",
       "      <td>A proof for all real numbers</td>\n",
       "      <td>Dividing 3 and 4 Digit Numbers by 2 Digit Numbers</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>content_description</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>content_kind</th>\n",
       "      <td>document</td>\n",
       "      <td>video</td>\n",
       "      <td>html5</td>\n",
       "      <td>document</td>\n",
       "      <td>video</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>content_text</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>content_language</th>\n",
       "      <td>es</td>\n",
       "      <td>es</td>\n",
       "      <td>en</td>\n",
       "      <td>en</td>\n",
       "      <td>en</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>content_copyright_holder</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>content_license</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>content_t</th>\n",
       "      <td>Ramón el ratón | document |</td>\n",
       "      <td>02-A-Orientaciones para mi planficación  | vid...</td>\n",
       "      <td>Word Stems | html5 |</td>\n",
       "      <td>A proof for all real numbers | document |</td>\n",
       "      <td>Dividing 3 and 4 Digit Numbers by 2 Digit Numb...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>content_chapter</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>content_is_train</th>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>content_label</th>\n",
       "      <td>[18549]</td>\n",
       "      <td>[3680, 1755]</td>\n",
       "      <td>[6576, 23638, 32087]</td>\n",
       "      <td>[30308]</td>\n",
       "      <td>[10921, 20861]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>content_max_train_score</th>\n",
       "      <td>0.616942</td>\n",
       "      <td>0.393206</td>\n",
       "      <td>0.147531</td>\n",
       "      <td>0.650264</td>\n",
       "      <td>0.162085</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                        5    \\\n",
       "topic_id                                                     t_0139f08b50fc   \n",
       "topic_title                                                       Lectura 5   \n",
       "topic_description                                                       NaN   \n",
       "topic_channel                                                        6e90a7   \n",
       "topic_category                                                      aligned   \n",
       "topic_level                                                               6   \n",
       "topic_language                                                           es   \n",
       "topic_parent                                                 t_74ea1148a50c   \n",
       "topic_has_content                                                      True   \n",
       "topic_parent_title                                              01 Unidad 1   \n",
       "topic_grandpa                                                t_c0125c7d0a1f   \n",
       "topic_grandpa_title                                     Español Lectura 1ro   \n",
       "topic_ggrandpa                                               t_7eb0dc459971   \n",
       "topic_ggrandpa_title                                             01 Primero   \n",
       "topic_fold                                                                0   \n",
       "topic_t                   Lectura 5 @ 01 Unidad 1 @ Español Lectura 1ro ...   \n",
       "topic_label                                                           18549   \n",
       "topic_chapter                                                           NaN   \n",
       "key                                                          t_74ea1148a50c   \n",
       "content_id                                                   c_a4236dd7ded9   \n",
       "label                                                                 18549   \n",
       "target                                                                    1   \n",
       "content_title                                                Ramón el ratón   \n",
       "content_description                                                           \n",
       "content_kind                                                       document   \n",
       "content_text                                                            NaN   \n",
       "content_language                                                         es   \n",
       "content_copyright_holder                                                NaN   \n",
       "content_license                                                         NaN   \n",
       "content_t                                      Ramón el ratón | document |    \n",
       "content_chapter                                                         NaN   \n",
       "content_is_train                                                      False   \n",
       "content_label                                                       [18549]   \n",
       "content_max_train_score                                            0.616942   \n",
       "\n",
       "                                                                        494  \\\n",
       "topic_id                                                     t_62242dcf093f   \n",
       "topic_title               6 Las lluvias intensas y los problemas que oca...   \n",
       "topic_description                                                       NaN   \n",
       "topic_channel                                                        f65044   \n",
       "topic_category                                                 supplemental   \n",
       "topic_level                                                               3   \n",
       "topic_language                                                           es   \n",
       "topic_parent                                                 t_007239952dfd   \n",
       "topic_has_content                                                      True   \n",
       "topic_parent_title                                  Exploramos y aprendemos   \n",
       "topic_grandpa                                                t_8ea654fa6e92   \n",
       "topic_grandpa_title                                           1° y 2° grado   \n",
       "topic_ggrandpa                                               t_6e977f750f41   \n",
       "topic_ggrandpa_title                             Aprendo en Casa - Primaria   \n",
       "topic_fold                                                                0   \n",
       "topic_t                   6 Las lluvias intensas y los problemas que oca...   \n",
       "topic_label                                                            3680   \n",
       "topic_chapter                                                           NaN   \n",
       "key                                                          t_007239952dfd   \n",
       "content_id                                                   c_4f65d6c00631   \n",
       "label                                                                  3680   \n",
       "target                                                                    1   \n",
       "content_title                      02-A-Orientaciones para mi planficación    \n",
       "content_description                                                           \n",
       "content_kind                                                          video   \n",
       "content_text                                                            NaN   \n",
       "content_language                                                         es   \n",
       "content_copyright_holder                                                NaN   \n",
       "content_license                                                         NaN   \n",
       "content_t                 02-A-Orientaciones para mi planficación  | vid...   \n",
       "content_chapter                                                         NaN   \n",
       "content_is_train                                                      False   \n",
       "content_label                                                  [3680, 1755]   \n",
       "content_max_train_score                                            0.393206   \n",
       "\n",
       "                                                                        52   \\\n",
       "topic_id                                                     t_0d5145117aa3   \n",
       "topic_title               B4.2.2.2 Use reading readiness and word identi...   \n",
       "topic_description         B4.2.2.2.1 Recognise and read words using a va...   \n",
       "topic_channel                                                        9d5527   \n",
       "topic_category                                                      aligned   \n",
       "topic_level                                                               3   \n",
       "topic_language                                                           en   \n",
       "topic_parent                                                 t_5de8e7f60cf4   \n",
       "topic_has_content                                                      True   \n",
       "topic_parent_title                                    Sub-Strand 2: Phonics   \n",
       "topic_grandpa                                                t_44cc3efd077e   \n",
       "topic_grandpa_title                                                Strand 2   \n",
       "topic_ggrandpa                                               t_b17d1e99f128   \n",
       "topic_ggrandpa_title                                      Basic 4 - English   \n",
       "topic_fold                                                                0   \n",
       "topic_t                   B4.2.2.2 Use reading readiness and word identi...   \n",
       "topic_label                                                            6576   \n",
       "topic_chapter                                                           NaN   \n",
       "key                                                          t_5de8e7f60cf4   \n",
       "content_id                                                   c_93b8b3ea4c10   \n",
       "label                                                                  6576   \n",
       "target                                                                    1   \n",
       "content_title                                                    Word Stems   \n",
       "content_description                                                           \n",
       "content_kind                                                          html5   \n",
       "content_text                                                            NaN   \n",
       "content_language                                                         en   \n",
       "content_copyright_holder                                                NaN   \n",
       "content_license                                                         NaN   \n",
       "content_t                                             Word Stems | html5 |    \n",
       "content_chapter                                                         NaN   \n",
       "content_is_train                                                       True   \n",
       "content_label                                          [6576, 23638, 32087]   \n",
       "content_max_train_score                                            0.147531   \n",
       "\n",
       "                                                                        988  \\\n",
       "topic_id                                                     t_b3adbe9834d1   \n",
       "topic_title                                   Trigonometric Identity Proofs   \n",
       "topic_description                                                       NaN   \n",
       "topic_channel                                                        01dbc1   \n",
       "topic_category                                                 supplemental   \n",
       "topic_level                                                               3   \n",
       "topic_language                                                           en   \n",
       "topic_parent                                                 t_e41d316189a6   \n",
       "topic_has_content                                                      True   \n",
       "topic_parent_title                    Understanding Trigonometric Functions   \n",
       "topic_grandpa                                                t_7004866e0ffe   \n",
       "topic_grandpa_title                                 Trigonometric Functions   \n",
       "topic_ggrandpa                                               t_0396c73309a7   \n",
       "topic_ggrandpa_title                                              Algebra 2   \n",
       "topic_fold                                                                0   \n",
       "topic_t                   Trigonometric Identity Proofs @ Understanding ...   \n",
       "topic_label                                                           30308   \n",
       "topic_chapter                                                           NaN   \n",
       "key                                                          t_e41d316189a6   \n",
       "content_id                                                   c_deb3dfaeb7d1   \n",
       "label                                                                 30308   \n",
       "target                                                                    1   \n",
       "content_title                                  A proof for all real numbers   \n",
       "content_description                                                           \n",
       "content_kind                                                       document   \n",
       "content_text                                                            NaN   \n",
       "content_language                                                         en   \n",
       "content_copyright_holder                                                NaN   \n",
       "content_license                                                         NaN   \n",
       "content_t                        A proof for all real numbers | document |    \n",
       "content_chapter                                                         NaN   \n",
       "content_is_train                                                      False   \n",
       "content_label                                                       [30308]   \n",
       "content_max_train_score                                            0.650264   \n",
       "\n",
       "                                                                        186  \n",
       "topic_id                                                     t_21c8907f5d3c  \n",
       "topic_title                                     Multiplication and Division  \n",
       "topic_description                                                       NaN  \n",
       "topic_channel                                                        7db7f2  \n",
       "topic_category                                                 supplemental  \n",
       "topic_level                                                               2  \n",
       "topic_language                                                           en  \n",
       "topic_parent                                                 t_02359b8e34ca  \n",
       "topic_has_content                                                      True  \n",
       "topic_parent_title                                              Mathematics  \n",
       "topic_grandpa                                                t_97faaaa83d71  \n",
       "topic_grandpa_title                                       Level 1 (Grade 4)  \n",
       "topic_ggrandpa                                                          NaN  \n",
       "topic_ggrandpa_title                                                         \n",
       "topic_fold                                                                0  \n",
       "topic_t                   Multiplication and Division @ Mathematics @ Le...  \n",
       "topic_label                                                           20861  \n",
       "topic_chapter                                                           NaN  \n",
       "key                                                          t_02359b8e34ca  \n",
       "content_id                                                   c_657cacfea9c5  \n",
       "label                                                                 20861  \n",
       "target                                                                    1  \n",
       "content_title             Dividing 3 and 4 Digit Numbers by 2 Digit Numbers  \n",
       "content_description                                                          \n",
       "content_kind                                                          video  \n",
       "content_text                                                            NaN  \n",
       "content_language                                                         en  \n",
       "content_copyright_holder                                                NaN  \n",
       "content_license                                                         NaN  \n",
       "content_t                 Dividing 3 and 4 Digit Numbers by 2 Digit Numb...  \n",
       "content_chapter                                                         NaN  \n",
       "content_is_train                                                       True  \n",
       "content_label                                                [10921, 20861]  \n",
       "content_max_train_score                                            0.162085  "
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "difficult_df.sample(5).T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "cef782a0",
   "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>content_t</th>\n",
       "      <th>vec_dist</th>\n",
       "      <th>pred</th>\n",
       "      <th>content_id</th>\n",
       "      <th>target</th>\n",
       "      <th>content_kind</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>72240</th>\n",
       "      <td>Organic Chemistry | html5 |</td>\n",
       "      <td>0.243220</td>\n",
       "      <td>0.995024</td>\n",
       "      <td>c_0c5af4baa365</td>\n",
       "      <td>0.0</td>\n",
       "      <td>html5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72241</th>\n",
       "      <td>Organic Chemistry Practice | exercise |</td>\n",
       "      <td>0.310441</td>\n",
       "      <td>0.991823</td>\n",
       "      <td>c_5b913fce12b7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>exercise</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72242</th>\n",
       "      <td>What Are Alcohols  Organic Chemistry  Chemistr...</td>\n",
       "      <td>0.582684</td>\n",
       "      <td>0.397149</td>\n",
       "      <td>c_c43bb1c60af9</td>\n",
       "      <td>0.0</td>\n",
       "      <td>video</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72243</th>\n",
       "      <td>End of Unit 1 Introduction to Organic Chemistr...</td>\n",
       "      <td>0.604150</td>\n",
       "      <td>0.665597</td>\n",
       "      <td>c_458601acecea</td>\n",
       "      <td>0.0</td>\n",
       "      <td>exercise</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72244</th>\n",
       "      <td>Alkanes  Alkenes  Organic Chemistry  Chemistry...</td>\n",
       "      <td>0.617299</td>\n",
       "      <td>0.188268</td>\n",
       "      <td>c_fcb0f55828c0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>video</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72245</th>\n",
       "      <td>Organic Farming | video |</td>\n",
       "      <td>0.617362</td>\n",
       "      <td>0.505613</td>\n",
       "      <td>c_b65cdc45b048</td>\n",
       "      <td>0.0</td>\n",
       "      <td>video</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72246</th>\n",
       "      <td>Pure and Applied Chemistry | html5 |</td>\n",
       "      <td>0.623326</td>\n",
       "      <td>0.326792</td>\n",
       "      <td>c_9e2105174dd6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>html5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72247</th>\n",
       "      <td>Straight-Chain Alkanes | html5 |</td>\n",
       "      <td>0.633136</td>\n",
       "      <td>0.204842</td>\n",
       "      <td>c_1e640d849f1b</td>\n",
       "      <td>0.0</td>\n",
       "      <td>html5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72248</th>\n",
       "      <td>Branched Alkanes | html5 |</td>\n",
       "      <td>0.645685</td>\n",
       "      <td>0.184800</td>\n",
       "      <td>c_36dfb7e7ddd6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>html5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72249</th>\n",
       "      <td>Organic Compounds | video |</td>\n",
       "      <td>0.652957</td>\n",
       "      <td>0.122906</td>\n",
       "      <td>c_f8cfe5c58d75</td>\n",
       "      <td>0.0</td>\n",
       "      <td>video</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72250</th>\n",
       "      <td>Isomers of 1,5-Dibromohexane | html5 |</td>\n",
       "      <td>0.655746</td>\n",
       "      <td>0.084381</td>\n",
       "      <td>c_1fc5acd9ee12</td>\n",
       "      <td>0.0</td>\n",
       "      <td>html5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72251</th>\n",
       "      <td>First Organic Molecules | html5 |</td>\n",
       "      <td>0.657255</td>\n",
       "      <td>0.125482</td>\n",
       "      <td>c_41e8232ab486</td>\n",
       "      <td>0.0</td>\n",
       "      <td>html5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72252</th>\n",
       "      <td>Organic Compounds Practice | exercise |</td>\n",
       "      <td>0.661732</td>\n",
       "      <td>0.098266</td>\n",
       "      <td>c_e6050f74de1c</td>\n",
       "      <td>0.0</td>\n",
       "      <td>exercise</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72253</th>\n",
       "      <td>Isomers | html5 |</td>\n",
       "      <td>0.664043</td>\n",
       "      <td>0.109077</td>\n",
       "      <td>c_c8777c8323c1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>html5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72254</th>\n",
       "      <td>Polymerization – Addition Polymers | html5 |</td>\n",
       "      <td>0.666670</td>\n",
       "      <td>0.095566</td>\n",
       "      <td>c_6acc431cc8df</td>\n",
       "      <td>0.0</td>\n",
       "      <td>html5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72255</th>\n",
       "      <td>Carbohydrates  | video | Learn about carbohydr...</td>\n",
       "      <td>0.667489</td>\n",
       "      <td>0.255139</td>\n",
       "      <td>c_440f8b11b1ea</td>\n",
       "      <td>0.0</td>\n",
       "      <td>video</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72256</th>\n",
       "      <td>Alkyl Halides | html5 |</td>\n",
       "      <td>0.668653</td>\n",
       "      <td>0.113681</td>\n",
       "      <td>c_a7d5c225e6c6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>html5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72257</th>\n",
       "      <td>Alcohols | html5 |</td>\n",
       "      <td>0.680825</td>\n",
       "      <td>0.088748</td>\n",
       "      <td>c_8b4521c342b5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>html5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72258</th>\n",
       "      <td>Condensation Polymerisation  | video | Learn t...</td>\n",
       "      <td>0.688858</td>\n",
       "      <td>0.136319</td>\n",
       "      <td>c_f5e52ce26e32</td>\n",
       "      <td>0.0</td>\n",
       "      <td>video</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72259</th>\n",
       "      <td>Formulae of Organic Molecules  | video | Learn...</td>\n",
       "      <td>0.697331</td>\n",
       "      <td>0.222246</td>\n",
       "      <td>c_347290e64831</td>\n",
       "      <td>0.0</td>\n",
       "      <td>video</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                               content_t  vec_dist      pred  \\\n",
       "72240                       Organic Chemistry | html5 |   0.243220  0.995024   \n",
       "72241           Organic Chemistry Practice | exercise |   0.310441  0.991823   \n",
       "72242  What Are Alcohols  Organic Chemistry  Chemistr...  0.582684  0.397149   \n",
       "72243  End of Unit 1 Introduction to Organic Chemistr...  0.604150  0.665597   \n",
       "72244  Alkanes  Alkenes  Organic Chemistry  Chemistry...  0.617299  0.188268   \n",
       "72245                         Organic Farming | video |   0.617362  0.505613   \n",
       "72246              Pure and Applied Chemistry | html5 |   0.623326  0.326792   \n",
       "72247                  Straight-Chain Alkanes | html5 |   0.633136  0.204842   \n",
       "72248                        Branched Alkanes | html5 |   0.645685  0.184800   \n",
       "72249                       Organic Compounds | video |   0.652957  0.122906   \n",
       "72250            Isomers of 1,5-Dibromohexane | html5 |   0.655746  0.084381   \n",
       "72251                 First Organic Molecules | html5 |   0.657255  0.125482   \n",
       "72252           Organic Compounds Practice | exercise |   0.661732  0.098266   \n",
       "72253                                 Isomers | html5 |   0.664043  0.109077   \n",
       "72254      Polymerization – Addition Polymers | html5 |   0.666670  0.095566   \n",
       "72255  Carbohydrates  | video | Learn about carbohydr...  0.667489  0.255139   \n",
       "72256                           Alkyl Halides | html5 |   0.668653  0.113681   \n",
       "72257                                Alcohols | html5 |   0.680825  0.088748   \n",
       "72258  Condensation Polymerisation  | video | Learn t...  0.688858  0.136319   \n",
       "72259  Formulae of Organic Molecules  | video | Learn...  0.697331  0.222246   \n",
       "\n",
       "           content_id  target content_kind  \n",
       "72240  c_0c5af4baa365     0.0        html5  \n",
       "72241  c_5b913fce12b7     0.0     exercise  \n",
       "72242  c_c43bb1c60af9     0.0        video  \n",
       "72243  c_458601acecea     0.0     exercise  \n",
       "72244  c_fcb0f55828c0     0.0        video  \n",
       "72245  c_b65cdc45b048     0.0        video  \n",
       "72246  c_9e2105174dd6     0.0        html5  \n",
       "72247  c_1e640d849f1b     0.0        html5  \n",
       "72248  c_36dfb7e7ddd6     0.0        html5  \n",
       "72249  c_f8cfe5c58d75     0.0        video  \n",
       "72250  c_1fc5acd9ee12     0.0        html5  \n",
       "72251  c_41e8232ab486     0.0        html5  \n",
       "72252  c_e6050f74de1c     0.0     exercise  \n",
       "72253  c_c8777c8323c1     0.0        html5  \n",
       "72254  c_6acc431cc8df     0.0        html5  \n",
       "72255  c_440f8b11b1ea     0.0        video  \n",
       "72256  c_a7d5c225e6c6     0.0        html5  \n",
       "72257  c_8b4521c342b5     0.0        html5  \n",
       "72258  c_f5e52ce26e32     0.0        video  \n",
       "72259  c_347290e64831     0.0        video  "
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "res_df[res_df[\"topic_id\"] == \"t_adf5741d6a79\"].sort_values(\"vec_dist\")[[\"content_t\", \"vec_dist\", \"pred\", \"content_id\", \"target\", \"content_kind\"]].head(20)#.T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "ffa16c1a",
   "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>content_id</th>\n",
       "      <th>content_title</th>\n",
       "      <th>content_description</th>\n",
       "      <th>content_kind</th>\n",
       "      <th>content_text</th>\n",
       "      <th>content_language</th>\n",
       "      <th>content_copyright_holder</th>\n",
       "      <th>content_license</th>\n",
       "      <th>content_t</th>\n",
       "      <th>content_chapter</th>\n",
       "      <th>content_is_train</th>\n",
       "      <th>content_label</th>\n",
       "      <th>content_max_train_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>128089</th>\n",
       "      <td>c_d4cd86bb4256</td>\n",
       "      <td>Fungi</td>\n",
       "      <td></td>\n",
       "      <td>html5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>en</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Fungi | html5 |</td>\n",
       "      <td>NaN</td>\n",
       "      <td>True</td>\n",
       "      <td>[14088, 22550]</td>\n",
       "      <td>0.165047</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>149895</th>\n",
       "      <td>c_f9132f87dc46</td>\n",
       "      <td>Fungi</td>\n",
       "      <td></td>\n",
       "      <td>video</td>\n",
       "      <td>NaN</td>\n",
       "      <td>en</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Fungi | video |</td>\n",
       "      <td>NaN</td>\n",
       "      <td>True</td>\n",
       "      <td>[20204, 31637, 20430]</td>\n",
       "      <td>0.318020</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            content_id content_title content_description content_kind  \\\n",
       "128089  c_d4cd86bb4256         Fungi                            html5   \n",
       "149895  c_f9132f87dc46         Fungi                            video   \n",
       "\n",
       "       content_text content_language content_copyright_holder content_license  \\\n",
       "128089          NaN               en                      NaN             NaN   \n",
       "149895          NaN               en                      NaN             NaN   \n",
       "\n",
       "               content_t  content_chapter  content_is_train  \\\n",
       "128089  Fungi | html5 |               NaN              True   \n",
       "149895  Fungi | video |               NaN              True   \n",
       "\n",
       "                content_label  content_max_train_score  \n",
       "128089         [14088, 22550]                 0.165047  \n",
       "149895  [20204, 31637, 20430]                 0.318020  "
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "content_df[content_df[\"content_id\"].isin({\"c_d4cd86bb4256\", \"c_f9132f87dc46\"})]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "c39f7172",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['Hydrocarbons | ', 'Organic Chemistry | '], dtype=object)"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "le_topic.inverse_transform([15081, 22469])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "582ee949",
   "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>t</th>\n",
       "      <th>content_t</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1993</th>\n",
       "      <td>Alkenes: Properties @ Hydrocarbons @ Chemistry...</td>\n",
       "      <td>Addition of water (acid-catalyzed) mechanism |...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1995</th>\n",
       "      <td>Alkenes: Properties @ Hydrocarbons @ Chemistry...</td>\n",
       "      <td>Markovnikov's rule and carbocations | video | ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1997</th>\n",
       "      <td>Alkenes: Properties @ Hydrocarbons @ Chemistry...</td>\n",
       "      <td>Ozonolysis | video | Alkene cleavage using ozo...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1998</th>\n",
       "      <td>Alkene reactions @ Hydrocarbons (5) @ Form 4 C...</td>\n",
       "      <td>Ozonolysis | video | Alkene cleavage using ozo...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2000</th>\n",
       "      <td>Alkenes: Properties @ Hydrocarbons @ Chemistry...</td>\n",
       "      <td>Introduction to reaction mechanisms | video | ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>195211</th>\n",
       "      <td>Cyclic Hydrocarbons @ Carbon and its Compounds...</td>\n",
       "      <td>Cyclic Hydrocarbons | video |</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>208168</th>\n",
       "      <td>Unsaturated Hydrocarbons @ Organic Chemistry @...</td>\n",
       "      <td>Alkenes - Example 1 | video |</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>218706</th>\n",
       "      <td>Unsaturated Hydrocarbons @ Organic Chemistry @...</td>\n",
       "      <td>Unsaturated Hydrocarbons Practice | exercise |</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>234760</th>\n",
       "      <td>Saturated Hydrocarbons @ Organic Chemistry @ P...</td>\n",
       "      <td>Saturated Hydrocarbons Practice | exercise |</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>234761</th>\n",
       "      <td>Saturated Hydrocarbons @ Organic Chemistry @ P...</td>\n",
       "      <td>Saturated Hydrocarbons | html5 |</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>80 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                        t  \\\n",
       "1993    Alkenes: Properties @ Hydrocarbons @ Chemistry...   \n",
       "1995    Alkenes: Properties @ Hydrocarbons @ Chemistry...   \n",
       "1997    Alkenes: Properties @ Hydrocarbons @ Chemistry...   \n",
       "1998    Alkene reactions @ Hydrocarbons (5) @ Form 4 C...   \n",
       "2000    Alkenes: Properties @ Hydrocarbons @ Chemistry...   \n",
       "...                                                   ...   \n",
       "195211  Cyclic Hydrocarbons @ Carbon and its Compounds...   \n",
       "208168  Unsaturated Hydrocarbons @ Organic Chemistry @...   \n",
       "218706  Unsaturated Hydrocarbons @ Organic Chemistry @...   \n",
       "234760  Saturated Hydrocarbons @ Organic Chemistry @ P...   \n",
       "234761  Saturated Hydrocarbons @ Organic Chemistry @ P...   \n",
       "\n",
       "                                                content_t  \n",
       "1993    Addition of water (acid-catalyzed) mechanism |...  \n",
       "1995    Markovnikov's rule and carbocations | video | ...  \n",
       "1997    Ozonolysis | video | Alkene cleavage using ozo...  \n",
       "1998    Ozonolysis | video | Alkene cleavage using ozo...  \n",
       "2000    Introduction to reaction mechanisms | video | ...  \n",
       "...                                                   ...  \n",
       "195211                     Cyclic Hydrocarbons | video |   \n",
       "208168                     Alkenes - Example 1 | video |   \n",
       "218706    Unsaturated Hydrocarbons Practice | exercise |   \n",
       "234760      Saturated Hydrocarbons Practice | exercise |   \n",
       "234761                  Saturated Hydrocarbons | html5 |   \n",
       "\n",
       "[80 rows x 2 columns]"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x = corr_df.merge(topic_df_train, left_on=\"topic_id\", right_on=\"id\")[[\"t\", \"content_id\"]].merge(content_df[content_df[\"content_is_train\"]], on=\"content_id\")\n",
    "\n",
    "x = x[[\"t\", \"content_t\"]]\n",
    "x[x[\"t\"].apply(lambda x: \"Hydrocarbons\" in x)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "708cf8af",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([['Properties of benzene @ Hydrocarbons @ Chemistry class 11 @ Science (India) | ',\n",
       "        'Friedel crafts acylation addendum | video | Friedel-Crafts acylation addendum.\\n\\n'],\n",
       "       ['Alkene reactions @ Hydrocarbons (5) @ Form 4 Chemistry @ Chemistry | In this tutorial, Jay explains the addition reactions of alkenes.',\n",
       "        'Hydroboration-oxidation: Mechanism | video | Electron-pushing mechanism of a hydroboration-oxidation reaction.\\n\\n'],\n",
       "       ['Alkynes: Properties @ Hydrocarbons @ Chemistry class 11 @ Science (India) | ',\n",
       "        'Synthesis using alkynes | video | Using a flowsheet showing different reactions of alkanes, alkenes, and alkynes\\nto solve an organic synthesis problem.\\n\\n'],\n",
       "       ['Mechanism of electrophilic substitution reactions @ Hydrocarbons @ Chemistry class 11 @ Science (India) | ',\n",
       "        'Aromatic halogenation | video | The halogenation of benzene.\\n\\n'],\n",
       "       ['Alkanes: Nomenclature and isomerism @ Hydrocarbons @ Chemistry class 11 @ Science (India) | ',\n",
       "        'Representing structures of organic molecules | video | Representing structures of organic molecules using line (or line-angle)\\ndiagrams.\\n\\n']],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x[x[\"t\"].apply(lambda x: \"Hydrocarbons\" in x)].sample(5).values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "7b92e345",
   "metadata": {},
   "outputs": [],
   "source": [
    "np.save(\"cache/V_content.npy\", V_content)\n",
    "np.save(\"cache/V_topic.npy\", V_topic)\n",
    "np.save(\"cache/V_topic_train.npy\", V_topic_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6ee848a9",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "eb711019",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.6983007937982826"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "res_df[\"rank\"] = res_df.groupby(\"topic_id\")[\"vec_dist\"].rank(method=\"first\", ascending=True)\n",
    "\n",
    "res_df[\"gap\"] = res_df[\"vec_dist\"] - res_df.groupby(\"topic_id\")[\"vec_dist\"].transform(\"min\")\n",
    "\n",
    "pred_df = res_df[res_df[\"gap\"] < 0.15]\n",
    "\n",
    "pred_df = pred_df.groupby(\"topic_id\")[\"content_id\"].apply(lambda x: \" \".join(list(x)))\n",
    "pred_df = pred_df.reset_index().rename(columns={\"content_id\": 'pred'})\n",
    "\n",
    "pred_df = pred_df.merge(pd.read_csv(\"data/correlations.csv\"), on=\"topic_id\")\n",
    "\n",
    "metric(pred_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cb8ff12f",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "98e10585",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}