brief-review-of-gain-prediction.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "4778c95f",
   "metadata": {},
   "source": [
    "Gain Extraction Model analyzes Patient Discharge Summaries and predicts the progress of a patient’s health. Gain Extraction Model can determine whether a patient has made any improvements in their health and if they have met their goals. This model offers an efficient and effective way to track a patient’s progress and ensure that they are receiving the best possible care."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "085eec06",
   "metadata": {},
   "source": [
    "## Importing Necessary Libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "0b3d63a8",
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "import os, evaluate, torch, accelerate, nltk, re, datetime \n",
    "import pandas as pd\n",
    "from collections import Counter\n",
    "import numpy as np\n",
    "from datasets import load_dataset, Dataset, DatasetDict\n",
    "from transformers import BertTokenizer\n",
    "from transformers import AutoTokenizer, AutoModel\n",
    "from transformers import AutoModelForSequenceClassification\n",
    "from datasets import load_metric\n",
    "from transformers import TrainingArguments, Trainer\n",
    "from transformers import pipeline\n",
    "from nltk.corpus import stopwords\n",
    "from nltk.stem.porter import PorterStemmer\n",
    "from nltk.stem import WordNetLemmatizer\n",
    "from nltk.tokenize import word_tokenize\n",
    "from sklearn.metrics import confusion_matrix, accuracy_score\n",
    "from sklearn import metrics\n",
    "import seaborn as sns\n",
    "from pynvml import *\n",
    "from numba import cuda\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "60c9835c",
   "metadata": {},
   "source": [
    "## Get Current Working Directory"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "01756186",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CWD: /home/bilal/NLP/discharge_report_models/\n"
     ]
    }
   ],
   "source": [
    "my_dir = os.getcwd() + '/' \n",
    "print(f\"CWD: {my_dir}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a2833e0b",
   "metadata": {},
   "source": [
    "## GPU Configuration"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "14f3b562",
   "metadata": {},
   "source": [
    "<p> Reset Cuda Memory</p>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "45ce30da",
   "metadata": {},
   "outputs": [],
   "source": [
    "device = cuda.get_current_device()\n",
    "device.reset()\n",
    "with torch.no_grad():\n",
    "    torch.cuda.empty_cache()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "55ec71aa",
   "metadata": {},
   "source": [
    "<p>Check Memory</p>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "e36552e9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total    : 11176 MB\n",
      "Free     : 11155 MB\n",
      "Used     : 21 MB\n"
     ]
    }
   ],
   "source": [
    "nvmlInit()\n",
    "h = nvmlDeviceGetHandleByIndex(0)\n",
    "info = nvmlDeviceGetMemoryInfo(h)\n",
    "print(f'Total    : {info.total//1024**2} MB')\n",
    "print(f'Free     : {info.free//1024**2} MB')\n",
    "print(f'Used     : {info.used//1024**2} MB')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c8ab82b5",
   "metadata": {},
   "source": [
    "### Get Cuda Device"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "6196c9c4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "device(type='cuda', index=0)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "devices = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\") \n",
    "devices"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "141073b8",
   "metadata": {},
   "source": [
    "## Load Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "8eed33b5",
   "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>label</th>\n",
       "      <th>text</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>Patient has shown improvement with functional ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>DC from skilled PT services d/t patient reache...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>Reconnect after 3 months.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>This patient has been seen since 11/18/22 by s...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>During his POC, he had a transition of a prima...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   label                                               text\n",
       "0      1  Patient has shown improvement with functional ...\n",
       "1      0  DC from skilled PT services d/t patient reache...\n",
       "2      0                          Reconnect after 3 months.\n",
       "3      0  This patient has been seen since 11/18/22 by s...\n",
       "4      0  During his POC, he had a transition of a prima..."
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv('brog_train1_Updated.csv')\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "43eafab1",
   "metadata": {},
   "source": [
    "<p>Check Empty Rows</p>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "9db2bb3d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Empty Label: 0\n",
      "Empty Sentence: 0\n"
     ]
    }
   ],
   "source": [
    "print(f\"Empty Label: {len(df[pd.isnull(df['label'])])}\\nEmpty Sentence: {len(df[pd.isnull(df['text'])])}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a262ee46",
   "metadata": {},
   "source": [
    "## Visualizing and Summarizing Label Distribution in DataFrame 'df'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "1d121464",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "label\n",
      "0    1750\n",
      "1     716\n",
      "Name: count, dtype: int64\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.countplot(x='label', data=df, hue='label')\n",
    "print(df['label'].value_counts())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c467e2c2",
   "metadata": {},
   "source": [
    "It's evident that we're facing a class imbalance issue here. To tackle this, we need to apply majority downsampling, which involves reducing the number of instances in the majority class to align with the number of instances in the minority class. In this scenario, it appears that the majority class comprises 1750 instances, whereas the minority class only has 716 instances."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "96b01fc1",
   "metadata": {},
   "source": [
    "## Cleansing Processes"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "92f28c4a",
   "metadata": {},
   "source": [
    "All I did was do some cleansing processes, including removing some patterns, punctuations, English stop words  and rows with word count less then 3."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "90d2d41e",
   "metadata": {},
   "outputs": [],
   "source": [
    "wordnet=WordNetLemmatizer()\n",
    "corpus=[]\n",
    "def preprossing(row):\n",
    "    my_words1=[]\n",
    "    my_words1 = re.sub('[^a-zA-Z]',' ', row)\n",
    "    my_words1 = my_words1.lower().strip()\n",
    "    my_words1 = my_words1.split()\n",
    "    my_words1 = [wordnet.lemmatize(word) for word in my_words1 if not word in set(stopwords.words('english'))]\n",
    "    my_words1 =  \" \".join(my_words1)\n",
    "    return my_words1\n",
    "\n",
    "for i in range(len(df)):\n",
    "    corpus.append(preprossing(df['text'][i]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "4d7d914a",
   "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>sentence</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>patient shown improvement functional transfer ...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>dc skilled pt service patient reached maximum ...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>reconnect month</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>patient seen since skilled pt</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>poc transition primary physical therapist may ...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                            sentence  label\n",
       "0  patient shown improvement functional transfer ...      1\n",
       "1  dc skilled pt service patient reached maximum ...      0\n",
       "2                                    reconnect month      0\n",
       "3                      patient seen since skilled pt      0\n",
       "4  poc transition primary physical therapist may ...      0"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "processed_df = pd.DataFrame({'sentence': corpus, 'label': df['label']})\n",
    "processed_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "66d9ff1c",
   "metadata": {},
   "source": [
    "### Majority Downsampling"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7a267794",
   "metadata": {},
   "source": [
    "To perform majority downsampling, we will remove rows with words count less than 8, as they contribute less to model performance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "6fd0cab8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of rows with word length less then 3: 833\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>sentence</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>reconnect month</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>patient seen since skilled pt</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>made slow progress throughout poc continues fa...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>continues fall however fall preventative use u...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>complained cramping time rle since change wate...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>follow month</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>patient discharged today</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>pt report falling secondary impairment listed ...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>patient met applicable goal therapy discharged...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>pt also found mass lung biopsied result pending</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                             sentence  label\n",
       "2                                     reconnect month      0\n",
       "3                       patient seen since skilled pt      0\n",
       "5   made slow progress throughout poc continues fa...      0\n",
       "7   continues fall however fall preventative use u...      0\n",
       "12  complained cramping time rle since change wate...      0\n",
       "16                                       follow month      0\n",
       "25                           patient discharged today      0\n",
       "27  pt report falling secondary impairment listed ...      0\n",
       "34  patient met applicable goal therapy discharged...      0\n",
       "36    pt also found mass lung biopsied result pending      0"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_filtered = processed_df[processed_df['label'] == 0]\n",
    "df_filtered = df_filtered[df_filtered['sentence'].apply(lambda x: len(x.split()) <= 8)]\n",
    "print(f\"Number of rows with word length less then 3: {len(df_filtered)}\")\n",
    "df_filtered.head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3c6ce3af",
   "metadata": {},
   "source": [
    "### Remove rows with word count less then 8"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "b8883078",
   "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>sentence</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>patient shown improvement functional transfer ...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>dc skilled pt service patient reached maximum ...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>poc transition primary physical therapist may ...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>however demonstrated mild improvement confiden...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>majority fall occurred without using ustep due...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                            sentence  label\n",
       "0  patient shown improvement functional transfer ...      1\n",
       "1  dc skilled pt service patient reached maximum ...      0\n",
       "2  poc transition primary physical therapist may ...      0\n",
       "3  however demonstrated mild improvement confiden...      1\n",
       "4  majority fall occurred without using ustep due...      0"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "indices_to_remove = df_filtered.index\n",
    "processed_df = processed_df.drop(indices_to_remove)\n",
    "processed_df = processed_df.reset_index(drop=True)\n",
    "processed_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cc282803",
   "metadata": {},
   "source": [
    "### Visualizing and Summarizing Label Distribution in DataFrame 'processed_df' after Majority Downsampling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "939c572e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "label\n",
      "0    917\n",
      "1    716\n",
      "Name: count, dtype: int64\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.countplot(x='label', data=processed_df, hue='label')\n",
    "print(processed_df['label'].value_counts())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "2ff5ca4d",
   "metadata": {},
   "outputs": [],
   "source": [
    "# processed_df.to_csv('processed_df.csv', index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4e303730",
   "metadata": {},
   "source": [
    "## Most Common Phrases"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "01fff758",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Tokenize sentences and count occurrences for each class label\n",
    "token_counts = {}\n",
    "for label in [0, 1]:\n",
    "    label_data = processed_df[processed_df['label'] == label]['sentence']\n",
    "    all_tokens = [token for sent in label_data for token in word_tokenize(sent)]\n",
    "    token_counts[label] = Counter(all_tokens)\n",
    "\n",
    "# Extract 10 most common phrases for each class label\n",
    "plot_data = []\n",
    "for label, counts in token_counts.items():\n",
    "    top_phrases = counts.most_common(10)\n",
    "    for phrase, count in top_phrases:\n",
    "        plot_data.append({'Phrase': phrase, 'Count': count})\n",
    "\n",
    "plot_df_0 = pd.DataFrame(plot_data[:10])\n",
    "plot_df_1 = pd.DataFrame(plot_data[10:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "efb7639a",
   "metadata": {},
   "source": [
    "### 10 most common phrases for 0 class label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "2df54811",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABNYAAAGlCAYAAAA2+3t3AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8pXeV/AAAACXBIWXMAAA9hAAAPYQGoP6dpAABpT0lEQVR4nO3dd3gUVf/+8XtDCjVBEBICoSMQehMiiEgLXQRROkhTpFdBpUoRlCLSRHq3IQJKDUqR0EKvUiJBIAkPJaElIcn5/cEv8yWCzyMrZFPer+va62LnnNn9TIYtc++ZMzZjjBEAAAAAAACAJ+Lk6AIAAAAAAACAlIhgDQAAAAAAALADwRoAAAAAAABgB4I1AAAAAAAAwA4EawAAAAAAAIAdCNYAAAAAAAAAOxCsAQAAAAAAAHYgWAMAAAAAAADsQLAGAAAAAAAA2IFgDQAAAAAAALCDQ4O1uLg4DRs2TAUKFFCGDBlUqFAhffzxxzLGWH2MMRo+fLhy5cqlDBkyqHbt2jpz5kyix7l+/bratGkjd3d3Zc2aVZ07d9bt27eTenMAAAAAAACQhjg0WJswYYJmzZql6dOn6+TJk5owYYImTpyoL774wuozceJETZs2TbNnz9aePXuUKVMm+fv7KyoqyurTpk0bHT9+XJs3b9a6deu0fft2devWzRGbBAAAAAAAgDTCZh4eHpbEGjVqJE9PT82bN89a1rx5c2XIkEFLly6VMUbe3t4aMGCABg4cKEmKiIiQp6enFi5cqJYtW+rkyZPy9fXVvn37VLFiRUnShg0b1KBBA/3555/y9vZ2yLYBAAAAAAAgdXN25JO/9NJLmjNnjn7//Xe98MILOnz4sHbu3KnJkydLkoKDgxUaGqratWtb63h4eKhy5coKDAxUy5YtFRgYqKxZs1qhmiTVrl1bTk5O2rNnj15//fVHnjc6OlrR0dHW/fj4eF2/fl3Zs2eXzWZ7hlsMAAAAAACA5MwYo1u3bsnb21tOTv/9ZE+HBmtDhgxRZGSkihUrpnTp0ikuLk5jx45VmzZtJEmhoaGSJE9Pz0TreXp6Wm2hoaHKmTNnonZnZ2dly5bN6vNX48eP16hRo5725gAAAAAAACCVuHjxovLkyfNf+zg0WPvmm2+0bNkyLV++XCVKlNChQ4fUt29feXt7q0OHDs/seYcOHar+/ftb9yMiIpQ3b15dvHhR7u7uz+x5AQAAAAAAkLxFRkbKx8dHWbJk+Z99HRqsDRo0SEOGDFHLli0lSaVKldKFCxc0fvx4dejQQV5eXpKksLAw5cqVy1ovLCxMZcuWlSR5eXkpPDw80ePGxsbq+vXr1vp/5ebmJjc3t0eWu7u7E6wBAAAAAADgH00X5tCrgt69e/eRc1XTpUun+Ph4SVKBAgXk5eWlgIAAqz0yMlJ79uyRn5+fJMnPz083b95UUFCQ1Wfr1q2Kj49X5cqVk2ArAAAAAAAAkBY5dMRa48aNNXbsWOXNm1clSpTQwYMHNXnyZHXq1EnSg2Swb9++GjNmjIoUKaICBQpo2LBh8vb2VtOmTSVJxYsXV7169dS1a1fNnj1b9+/fV8+ePdWyZUuuCAoAAAAAAIBnxqHB2hdffKFhw4bpvffeU3h4uLy9vfXOO+9o+PDhVp/Bgwfrzp076tatm27evKlq1appw4YNSp8+vdVn2bJl6tmzp2rVqiUnJyc1b95c06ZNc8QmAQAAAAAAII2wGWOMo4twtMjISHl4eCgiIoI51gAAAAAAANKwJ8mJHDrHGgAAAAAAAJBSEawBAAAAAAAAdiBYAwAAAAAAAOxAsAYAAAAAAADYgWANAAAAAAAAsAPBGgAAAAAAAGAHgjUAAAAAAADADgRrAAAAAAAAgB2cHV1ASlRh0GJHl5DmBH3a3tElAAAAAAAAJMKINQAAAAAAAMAOBGsAAAAAAACAHQjWAAAAAAAAADsQrAEAAAAAAAB2IFgDAAAAAAAA7ECwBgAAAAAAANiBYA0AAAAAAACwA8EaAAAAAAAAYAeCNQAAAAAAAMAOBGsAAAAAAACAHQjWAAAAAAAAADsQrAEAAAAAAAB2IFgDAAAAAAAA7ECwBgAAAAAAANiBYA0AAAAAAACwA8EaAAAAAAAAYAeCNQAAAAAAAMAOBGsAAAAAAACAHRwarOXPn182m+2RW48ePSRJUVFR6tGjh7Jnz67MmTOrefPmCgsLS/QYISEhatiwoTJmzKicOXNq0KBBio2NdcTmAAAAAAAAIA1xaLC2b98+Xblyxbpt3rxZktSiRQtJUr9+/bR27Vp9++232rZtmy5fvqxmzZpZ68fFxalhw4aKiYnRrl27tGjRIi1cuFDDhw93yPYAAAAAAAAg7bAZY4yji0jQt29frVu3TmfOnFFkZKRy5Mih5cuX64033pAknTp1SsWLF1dgYKCqVKmi9evXq1GjRrp8+bI8PT0lSbNnz9b777+vq1evytXV9R89b2RkpDw8PBQRESF3d/f/2b/CoMX2byTsEvRpe0eXAAAAAAAA0oAnyYmSzRxrMTExWrp0qTp16iSbzaagoCDdv39ftWvXtvoUK1ZMefPmVWBgoCQpMDBQpUqVskI1SfL391dkZKSOHz/+t88VHR2tyMjIRDcAAAAAAADgSSSbYG316tW6efOmOnbsKEkKDQ2Vq6ursmbNmqifp6enQkNDrT4Ph2oJ7Qltf2f8+PHy8PCwbj4+Pk9vQwAAAAAAAJAmJJtgbd68eapfv768vb2f+XMNHTpUERER1u3ixYvP/DkBAAAAAACQujg7ugBJunDhgrZs2aJVq1ZZy7y8vBQTE6ObN28mGrUWFhYmLy8vq8/evXsTPVbCVUMT+jyOm5ub3NzcnuIWAAAAAAAAIK1JFiPWFixYoJw5c6phw4bWsgoVKsjFxUUBAQHWstOnTyskJER+fn6SJD8/Px09elTh4eFWn82bN8vd3V2+vr5JtwEAAAAAAABIcxw+Yi0+Pl4LFixQhw4d5Oz8f+V4eHioc+fO6t+/v7JlyyZ3d3f16tVLfn5+qlKliiSpbt268vX1Vbt27TRx4kSFhobqo48+Uo8ePRiRBgAAAAAAgGfK4cHali1bFBISok6dOj3SNmXKFDk5Oal58+aKjo6Wv7+/Zs6cabWnS5dO69atU/fu3eXn56dMmTKpQ4cOGj16dFJuAgAAAAAAANIgmzHGOLoIR4uMjJSHh4ciIiLk7u7+P/tXGLQ4CarCw4I+be/oEgAAAAAAQBrwJDlRsphjDQAAAAAAAEhpCNYAAAAAAAAAOxCsAQAAAAAAAHYgWAMAAAAAAADsQLAGAAAAAAAA2IFgDQAAAAAAALADwRoAAAAAAABgB4I1AAAAAAAAwA4EawAAAAAAAIAdCNYAAAAAAAAAOxCsAQAAAAAAAHYgWAMAAAAAAADsQLAGAAAAAAAA2IFgDQAAAAAAALADwRoAAAAAAABgB4I1AAAAAAAAwA4EawAAAAAAAIAdCNYAAAAAAAAAOxCsAQAAAAAAAHYgWAMAAAAAAADsQLAGAAAAAAAA2IFgDQAAAAAAALADwRoAAAAAAABgB4I1AAAAAAAAwA4EawAAAAAAAIAdCNYAAAAAAAAAOxCsAQAAAAAAAHYgWAMAAAAAAADs4PBg7dKlS2rbtq2yZ8+uDBkyqFSpUtq/f7/VbozR8OHDlStXLmXIkEG1a9fWmTNnEj3G9evX1aZNG7m7uytr1qzq3Lmzbt++ndSbAgAAAAAAgDTEocHajRs3VLVqVbm4uGj9+vU6ceKEJk2apOeee87qM3HiRE2bNk2zZ8/Wnj17lClTJvn7+ysqKsrq06ZNGx0/flybN2/WunXrtH37dnXr1s0RmwQAAAAAAIA0wmaMMY568iFDhui3337Tjh07HttujJG3t7cGDBiggQMHSpIiIiLk6emphQsXqmXLljp58qR8fX21b98+VaxYUZK0YcMGNWjQQH/++ae8vb3/Zx2RkZHy8PBQRESE3N3d/2f/CoMWP8FW4mkI+rS9o0sAAAAAAABpwJPkRA4dsbZmzRpVrFhRLVq0UM6cOVWuXDl99dVXVntwcLBCQ0NVu3Zta5mHh4cqV66swMBASVJgYKCyZs1qhWqSVLt2bTk5OWnPnj2Pfd7o6GhFRkYmugEAAAAAAABPwqHB2vnz5zVr1iwVKVJEGzduVPfu3dW7d28tWrRIkhQaGipJ8vT0TLSep6en1RYaGqqcOXMmand2dla2bNmsPn81fvx4eXh4WDcfH5+nvWkAAAAAAABI5RwarMXHx6t8+fIaN26cypUrp27duqlr166aPXv2M33eoUOHKiIiwrpdvHjxmT4fAAAAAAAAUh+HBmu5cuWSr69vomXFixdXSEiIJMnLy0uSFBYWlqhPWFiY1ebl5aXw8PBE7bGxsbp+/brV56/c3Nzk7u6e6AYAAAAAAAA8CYcGa1WrVtXp06cTLfv999+VL18+SVKBAgXk5eWlgIAAqz0yMlJ79uyRn5+fJMnPz083b95UUFCQ1Wfr1q2Kj49X5cqVk2ArAAAAAAAAkBY5O/LJ+/Xrp5deeknjxo3Tm2++qb1792rOnDmaM2eOJMlms6lv374aM2aMihQpogIFCmjYsGHy9vZW06ZNJT0Y4VavXj3rFNL79++rZ8+eatmy5T+6IigAAAAAAABgD4cGa5UqVdIPP/ygoUOHavTo0SpQoICmTp2qNm3aWH0GDx6sO3fuqFu3brp586aqVaumDRs2KH369FafZcuWqWfPnqpVq5acnJzUvHlzTZs2zRGbBAAAAAAAgDTCZowxji7C0SIjI+Xh4aGIiIh/NN9ahUGLk6AqPCzo0/aOLgEAAAAAAKQBT5ITOXSONQAAAAAAACClIlgDAAAAAAAA7ECwBgAAAAAAANiBYA0AAAAAAACwA8EaAAAAAAAAYAeCNQAAAAAAAMAOBGsAAAAAAACAHQjWAAAAAAAAADsQrAEAAAAAAAB2IFgDAAAAAAAA7ECwBgAAAAAAANiBYA0AAAAAAACwA8EaAAAAAAAAYAeCNQAAAAAAAMAOBGsAAAAAAACAHQjWAAAAAAAAADsQrAEAAAAAAAB2IFgDAAAAAAAA7ECwBgAAAAAAANiBYA0AAAAAAACwA8EaAAAAAAAAYAeCNQAAAAAAAMAOBGsAAAAAAACAHQjWAAAAAAAAADsQrAEAAAAAAAB2IFgDAAAAAAAA7ECwBgAAAAAAANjBocHayJEjZbPZEt2KFStmtUdFRalHjx7Knj27MmfOrObNmyssLCzRY4SEhKhhw4bKmDGjcubMqUGDBik2NjapNwUAAAAAAABpjLOjCyhRooS2bNli3Xd2/r+S+vXrp59++knffvutPDw81LNnTzVr1ky//fabJCkuLk4NGzaUl5eXdu3apStXrqh9+/ZycXHRuHHjknxbAAAAAAAAkHY4PFhzdnaWl5fXI8sjIiI0b948LV++XDVr1pQkLViwQMWLF9fu3btVpUoVbdq0SSdOnNCWLVvk6empsmXL6uOPP9b777+vkSNHytXVNak3BwAAAAAAAGmEw+dYO3PmjLy9vVWwYEG1adNGISEhkqSgoCDdv39ftWvXtvoWK1ZMefPmVWBgoCQpMDBQpUqVkqenp9XH399fkZGROn78+N8+Z3R0tCIjIxPdAAAAAAAAgCfh0GCtcuXKWrhwoTZs2KBZs2YpODhYL7/8sm7duqXQ0FC5uroqa9asidbx9PRUaGioJCk0NDRRqJbQntD2d8aPHy8PDw/r5uPj83Q3DAAAAAAAAKmeQ08FrV+/vvXv0qVLq3LlysqXL5+++eYbZciQ4Zk979ChQ9W/f3/rfmRkJOEaAAAAAAAAnojDTwV9WNasWfXCCy/o7Nmz8vLyUkxMjG7evJmoT1hYmDUnm5eX1yNXCU24/7h52xK4ubnJ3d090Q0AAAAAAAB4EskqWLt9+7bOnTunXLlyqUKFCnJxcVFAQIDVfvr0aYWEhMjPz0+S5Ofnp6NHjyo8PNzqs3nzZrm7u8vX1zfJ6wcAAAAAAEDa4dBTQQcOHKjGjRsrX758unz5skaMGKF06dKpVatW8vDwUOfOndW/f39ly5ZN7u7u6tWrl/z8/FSlShVJUt26deXr66t27dpp4sSJCg0N1UcffaQePXrIzc3NkZsGAAAAAACAVM6hwdqff/6pVq1a6dq1a8qRI4eqVaum3bt3K0eOHJKkKVOmyMnJSc2bN1d0dLT8/f01c+ZMa/106dJp3bp16t69u/z8/JQpUyZ16NBBo0ePdtQmAQAAAAAAII2wGWOMo4twtMjISHl4eCgiIuIfzbdWYdDiJKgKDwv6tL2jSwAAAAAAAGnAk+REyWqONQAAAAAAACClIFgDAAAAAAAA7ECwBgAAAAAAANiBYA0AAAAAAACwA8EaAAAAAAAAYAeCNQAAAAAAAMAOBGsAAAAAAACAHQjWAAAAAAAAADsQrAEAAAAAAAB2IFgDAAAAAAAA7ECwBgAAAAAAANiBYA0AAAAAAACwA8EaAAAAAAAAYAeCNQAAAAAAAMAOdgVrBQsW1LVr1x5ZfvPmTRUsWPBfFwUAAAAAAAAkd3YFa3/88Yfi4uIeWR4dHa1Lly7966IAAAAAAACA5M75STqvWbPG+vfGjRvl4eFh3Y+Li1NAQIDy58//1IoDAAAAAAAAkqsnCtaaNm0qSbLZbOrQoUOiNhcXF+XPn1+TJk16asUBAAAAAAAAydUTBWvx8fGSpAIFCmjfvn16/vnnn0lRAAAAAAAAQHL3RMFaguDg4KddBwAAAAAAAJCi2BWsSVJAQIACAgIUHh5ujWRLMH/+/H9dGAAAAAAAAJCc2RWsjRo1SqNHj1bFihWVK1cu2Wy2p10XAAAAAAAAkKzZFazNnj1bCxcuVLt27Z52PQAAAAAAAECK4GTPSjExMXrppZeedi0AAAAAAABAimFXsNalSxctX778adcCAAAAAAAApBh2nQoaFRWlOXPmaMuWLSpdurRcXFwStU+ePPmpFAcAAAAAAAAkV3YFa0eOHFHZsmUlSceOHUvUxoUMAAAAAAAAkBbYFaz98ssvT7sOAAAAAAAAIEWxa461Z+GTTz6RzWZT3759rWVRUVHq0aOHsmfPrsyZM6t58+YKCwtLtF5ISIgaNmyojBkzKmfOnBo0aJBiY2OTuHoAAAAAAACkNXaNWHv11Vf/6ymfW7dufaLH27dvn7788kuVLl060fJ+/frpp59+0rfffisPDw/17NlTzZo102+//SZJiouLU8OGDeXl5aVdu3bpypUrat++vVxcXDRu3Lgn3zAAAAAAAADgH7JrxFrZsmVVpkwZ6+br66uYmBgdOHBApUqVeqLHun37ttq0aaOvvvpKzz33nLU8IiJC8+bN0+TJk1WzZk1VqFBBCxYs0K5du7R7925J0qZNm3TixAktXbpUZcuWVf369fXxxx9rxowZiomJsWfTAAAAAAAAgH/ErhFrU6ZMeezykSNH6vbt20/0WD169FDDhg1Vu3ZtjRkzxloeFBSk+/fvq3bt2tayYsWKKW/evAoMDFSVKlUUGBioUqVKydPT0+rj7++v7t276/jx4ypXrtxjnzM6OlrR0dHW/cjIyCeqGQAAAAAAAHiqc6y1bdtW8+fP/8f9V65cqQMHDmj8+PGPtIWGhsrV1VVZs2ZNtNzT01OhoaFWn4dDtYT2hLa/M378eHl4eFg3Hx+ff1wzAAAAAAAAID3lYC0wMFDp06f/R30vXryoPn36aNmyZf94nadl6NChioiIsG4XL15M0ucHAAAAAABAymfXqaDNmjVLdN8YoytXrmj//v0aNmzYP3qMoKAghYeHq3z58tayuLg4bd++XdOnT9fGjRsVExOjmzdvJhq1FhYWJi8vL0mSl5eX9u7dm+hxE64amtDncdzc3OTm5vaP6gQAAAAAAAAex65gzcPDI9F9JycnFS1aVKNHj1bdunX/0WPUqlVLR48eTbTs7bffVrFixfT+++/Lx8dHLi4uCggIUPPmzSVJp0+fVkhIiPz8/CRJfn5+Gjt2rMLDw5UzZ05J0ubNm+Xu7i5fX197Ng0AAAAAAAD4R+wK1hYsWPCvnzhLliwqWbJkomWZMmVS9uzZreWdO3dW//79lS1bNrm7u6tXr17y8/NTlSpVJEl169aVr6+v2rVrp4kTJyo0NFQfffSRevTowYg0AAAAAAAAPFN2BWsJgoKCdPLkSUlSiRIl/vYqnPaaMmWKnJyc1Lx5c0VHR8vf318zZ8602tOlS6d169ape/fu8vPzU6ZMmdShQweNHj36qdYBAAAAAAAA/JXNGGOedKXw8HC1bNlSv/76qzX/2c2bN/Xqq69q5cqVypEjx9Ou85mKjIyUh4eHIiIi5O7u/j/7Vxi0OAmqwsOCPm3v6BIAAAAAAEAa8CQ5kV1XBe3Vq5du3bql48eP6/r167p+/bqOHTumyMhI9e7d266iAQAAAAAAgJTErlNBN2zYoC1btqh48eLWMl9fX82YMeMfX7wAAAAAAAAASMnsGrEWHx8vFxeXR5a7uLgoPj7+XxcFAAAAAAAAJHd2BWs1a9ZUnz59dPnyZWvZpUuX1K9fP9WqVeupFQcAAAAAAAAkV3YFa9OnT1dkZKTy58+vQoUKqVChQipQoIAiIyP1xRdfPO0aAQAAAAAAgGTHrjnWfHx8dODAAW3ZskWnTp2SJBUvXly1a9d+qsUBAAAAAAAAydUTjVjbunWrfH19FRkZKZvNpjp16qhXr17q1auXKlWqpBIlSmjHjh3PqlYAAAAAAAAg2XiiYG3q1Knq2rWr3N3dH2nz8PDQO++8o8mTJz+14gAAAAAAAIDk6omCtcOHD6tevXp/2163bl0FBQX966IAAAAAAACA5O6JgrWwsDC5uLj8bbuzs7OuXr36r4sCAAAAAAAAkrsnCtZy586tY8eO/W37kSNHlCtXrn9dFAAAAAAAAJDcPVGw1qBBAw0bNkxRUVGPtN27d08jRoxQo0aNnlpxAAAAAAAAQHLl/CSdP/roI61atUovvPCCevbsqaJFi0qSTp06pRkzZiguLk4ffvjhMykUAAAAAAAASE6eKFjz9PTUrl271L17dw0dOlTGGEmSzWaTv7+/ZsyYIU9Pz2dSKAAAAAAAAJCcPFGwJkn58uXTzz//rBs3bujs2bMyxqhIkSJ67rnnnkV9AAAAAAAAQLL0xMFagueee06VKlV6mrUAAAAAAAAAKcYTXbwAAAAAAAAAwAMEawAAAAAAAIAdCNYAAAAAAAAAOxCsAQAAAAAAAHYgWAMAAAAAAADsQLAGAAAAAAAA2MHZ0QUAyUHI6FKOLiHNyTv8qKNLAAAAAADgX2HEGgAAAAAAAGAHgjUAAAAAAADADgRrAAAAAAAAgB0I1gAAAAAAAAA7ODRYmzVrlkqXLi13d3e5u7vLz89P69evt9qjoqLUo0cPZc+eXZkzZ1bz5s0VFhaW6DFCQkLUsGFDZcyYUTlz5tSgQYMUGxub1JsCAAAAAACANMahwVqePHn0ySefKCgoSPv371fNmjX12muv6fjx45Kkfv36ae3atfr222+1bds2Xb58Wc2aNbPWj4uLU8OGDRUTE6Ndu3Zp0aJFWrhwoYYPH+6oTQIAAAAAAEAaYTPGGEcX8bBs2bLp008/1RtvvKEcOXJo+fLleuONNyRJp06dUvHixRUYGKgqVapo/fr1atSokS5fvixPT09J0uzZs/X+++/r6tWrcnV1/UfPGRkZKQ8PD0VERMjd3f1/9q8waLH9Gwi7BH3a/pk+fsjoUs/08fGovMOPOroEAAAAAAAe8SQ5UbKZYy0uLk4rV67UnTt35Ofnp6CgIN2/f1+1a9e2+hQrVkx58+ZVYGCgJCkwMFClSpWyQjVJ8vf3V2RkpDXq7XGio6MVGRmZ6AYAAAAAAAA8CYcHa0ePHlXmzJnl5uamd999Vz/88IN8fX0VGhoqV1dXZc2aNVF/T09PhYaGSpJCQ0MThWoJ7Qltf2f8+PHy8PCwbj4+Pk93owAAAAAAAJDqOTxYK1q0qA4dOqQ9e/aoe/fu6tChg06cOPFMn3Po0KGKiIiwbhcvXnymzwcAAAAAAIDUx9nRBbi6uqpw4cKSpAoVKmjfvn36/PPP9dZbbykmJkY3b95MNGotLCxMXl5ekiQvLy/t3bs30eMlXDU0oc/juLm5yc3N7SlvCQAAAAAAANISh49Y+6v4+HhFR0erQoUKcnFxUUBAgNV2+vRphYSEyM/PT5Lk5+eno0ePKjw83OqzefNmubu7y9fXN8lrBwAAAAAAQNrh0BFrQ4cOVf369ZU3b17dunVLy5cv16+//qqNGzfKw8NDnTt3Vv/+/ZUtWza5u7urV69e8vPzU5UqVSRJdevWla+vr9q1a6eJEycqNDRUH330kXr06MGINAAAAAAAADxTDg3WwsPD1b59e125ckUeHh4qXbq0Nm7cqDp16kiSpkyZIicnJzVv3lzR0dHy9/fXzJkzrfXTpUundevWqXv37vLz81OmTJnUoUMHjR492lGbBAAAAAAAgDTCZowxji7C0SIjI+Xh4aGIiAi5u7v/z/4VBi1OgqrwsKBP2z/Txw8ZXeqZPj4elXf4UUeXAAAAAADAI54kJ0p2c6wBAAAAAAAAKQHBGgAAAAAAAGAHgjUAAAAAAADADgRrAAAAAAAAgB0I1gAAAAAAAAA7EKwBAAAAAAAAdiBYAwAAAAAAAOxAsAYAAAAAAADYgWANAAAAAAAAsAPBGgAAAAAAAGAHgjUAAAAAAADADgRrAAAAAAAAgB0I1gAAAAAAAAA7EKwBAAAAAAAAdiBYAwAAAAAAAOxAsAYAAAAAAADYgWANAAAAAAAAsAPBGgAAAAAAAGAHgjUAAAAAAADADgRrAAAAAAAAgB0I1gAAAAAAAAA7EKwBAAAAAAAAdiBYAwAAAAAAAOxAsAYAAAAAAADYgWANAAAAAAAAsIOzowsAgKet6hdVHV1CmvNbr98cXQIAAAAAJDlGrAEAAAAAAAB2IFgDAAAAAAAA7ODQYG38+PGqVKmSsmTJopw5c6pp06Y6ffp0oj5RUVHq0aOHsmfPrsyZM6t58+YKCwtL1CckJEQNGzZUxowZlTNnTg0aNEixsbFJuSkAAAAAAABIYxwarG3btk09evTQ7t27tXnzZt2/f19169bVnTt3rD79+vXT2rVr9e2332rbtm26fPmymjVrZrXHxcWpYcOGiomJ0a5du7Ro0SItXLhQw4cPd8QmAQAAAAAAII1w6MULNmzYkOj+woULlTNnTgUFBal69eqKiIjQvHnztHz5ctWsWVOStGDBAhUvXly7d+9WlSpVtGnTJp04cUJbtmyRp6enypYtq48//ljvv/++Ro4cKVdXV0dsGgAAAAAAAFK5ZDXHWkREhCQpW7ZskqSgoCDdv39ftWvXtvoUK1ZMefPmVWBgoCQpMDBQpUqVkqenp9XH399fkZGROn78+GOfJzo6WpGRkYluAAAAAAAAwJNINsFafHy8+vbtq6pVq6pkyZKSpNDQULm6uipr1qyJ+np6eio0NNTq83ColtCe0PY448ePl4eHh3Xz8fF5ylsDAAAAAACA1C7ZBGs9evTQsWPHtHLlymf+XEOHDlVERIR1u3jx4jN/TgAAAAAAAKQuDp1jLUHPnj21bt06bd++XXny5LGWe3l5KSYmRjdv3kw0ai0sLExeXl5Wn7179yZ6vISrhib0+Ss3Nze5ubk95a0AAAAAAABAWuLQEWvGGPXs2VM//PCDtm7dqgIFCiRqr1ChglxcXBQQEGAtO336tEJCQuTn5ydJ8vPz09GjRxUeHm712bx5s9zd3eXr65s0GwIAAAAAAIA0x6Ej1nr06KHly5frxx9/VJYsWaw50Tw8PJQhQwZ5eHioc+fO6t+/v7JlyyZ3d3f16tVLfn5+qlKliiSpbt268vX1Vbt27TRx4kSFhobqo48+Uo8ePRiVBgAAAAAAgGfGocHarFmzJEk1atRItHzBggXq2LGjJGnKlClycnJS8+bNFR0dLX9/f82cOdPqmy5dOq1bt07du3eXn5+fMmXKpA4dOmj06NFJtRkAAAAAAABIgxwarBlj/mef9OnTa8aMGZoxY8bf9smXL59+/vnnp1kaAAAAAAAA8F8li4sXAADwd7ZVf8XRJaQ5r2zf5ugSAAAAgBTBoRcvAAAAAAAAAFIqRqwBAIAkM33AWkeXkOb0nNTY0SUAAACkWoxYAwAAAAAAAOxAsAYAAAAAAADYgWANAAAAAAAAsAPBGgAAAAAAAGAHgjUAAAAAAADADgRrAAAAAAAAgB0I1gAAAAAAAAA7EKwBAAAAAAAAdiBYAwAAAAAAAOxAsAYAAAAAAADYwdnRBQAAACDlGtv2DUeXkOZ8uPQ7R5cAAAD+P4I1AAAAAJKkk2O3OrqENKf4hzUdXQIA4F/gVFAAAAAAAADADgRrAAAAAAAAgB0I1gAAAAAAAAA7EKwBAAAAAAAAduDiBQAAAACQCo0cOdLRJaQ5/M2BtIcRawAAAAAAAIAdCNYAAAAAAAAAOxCsAQAAAAAAAHYgWAMAAAAAAADsQLAGAAAAAAAA2IFgDQAAAAAAALADwRoAAAAAAABgB4cGa9u3b1fjxo3l7e0tm82m1atXJ2o3xmj48OHKlSuXMmTIoNq1a+vMmTOJ+ly/fl1t2rSRu7u7smbNqs6dO+v27dtJuBUAAAAAAABIixwarN25c0dlypTRjBkzHts+ceJETZs2TbNnz9aePXuUKVMm+fv7KyoqyurTpk0bHT9+XJs3b9a6deu0fft2devWLak2AQAAAAAAAGmUsyOfvH79+qpfv/5j24wxmjp1qj766CO99tprkqTFixfL09NTq1evVsuWLXXy5Elt2LBB+/btU8WKFSVJX3zxhRo0aKDPPvtM3t7eSbYtAAAAAAAASFscGqz9N8HBwQoNDVXt2rWtZR4eHqpcubICAwPVsmVLBQYGKmvWrFaoJkm1a9eWk5OT9uzZo9dff/2xjx0dHa3o6GjrfmRk5LPbEAAAAAAA/qVvvn3R0SWkOW+22OvoEpACJNtgLTQ0VJLk6emZaLmnp6fVFhoaqpw5cyZqd3Z2VrZs2aw+jzN+/HiNGjXqKVcMAAAAAADwz5T5bqOjS0hzDr/h/9QfM01eFXTo0KGKiIiwbhcvXnR0SQAAAAAAAEhhkm2w5uXlJUkKCwtLtDwsLMxq8/LyUnh4eKL22NhYXb9+3erzOG5ubnJ3d090AwAAAAAAAJ5Esg3WChQoIC8vLwUEBFjLIiMjtWfPHvn5+UmS/Pz8dPPmTQUFBVl9tm7dqvj4eFWuXDnJawYAAAAAAEDa4dA51m7fvq2zZ89a94ODg3Xo0CFly5ZNefPmVd++fTVmzBgVKVJEBQoU0LBhw+Tt7a2mTZtKkooXL6569eqpa9eumj17tu7fv6+ePXuqZcuWXBEUAAAAAAAAz5RDg7X9+/fr1Vdfte73799fktShQwctXLhQgwcP1p07d9StWzfdvHlT1apV04YNG5Q+fXprnWXLlqlnz56qVauWnJyc1Lx5c02bNi3JtwUAAAAAAABpi0ODtRo1asgY87ftNptNo0eP1ujRo/+2T7Zs2bR8+fJnUR4AAAAAAADwt5LtHGsAAAAAAABAckawBgAAAAAAANiBYA0AAAAAAACwA8EaAAAAAAAAYAeCNQAAAAAAAMAOBGsAAAAAAACAHQjWAAAAAAAAADsQrAEAAAAAAAB2IFgDAAAAAAAA7ECwBgAAAAAAANiBYA0AAAAAAACwA8EaAAAAAAAAYAeCNQAAAAAAAMAOBGsAAAAAAACAHQjWAAAAAAAAADsQrAEAAAAAAAB2IFgDAAAAAAAA7ECwBgAAAAAAANiBYA0AAAAAAACwA8EaAAAAAAAAYAeCNQAAAAAAAMAOBGsAAAAAAACAHQjWAAAAAAAAADsQrAEAAAAAAAB2IFgDAAAAAAAA7ECwBgAAAAAAANiBYA0AAAAAAACwA8EaAAAAAAAAYIdUE6zNmDFD+fPnV/r06VW5cmXt3bvX0SUBAAAAAAAgFUsVwdrXX3+t/v37a8SIETpw4IDKlCkjf39/hYeHO7o0AAAAAAAApFKpIlibPHmyunbtqrffflu+vr6aPXu2MmbMqPnz5zu6NAAAAAAAAKRSzo4u4N+KiYlRUFCQhg4dai1zcnJS7dq1FRgY+Nh1oqOjFR0dbd2PiIiQJEVGRv6j54yLvvcvKoY9/um+sdetqLhn+vh41LPcp7H3Yp/ZY+PxnuX+vBPL/kxqz3J/3ou++8weG4/3rD9Do+7ff6aPj0c9y316O+rOM3tsPN6z3J8PH/MgaTzL/Xn3LscsSe1Zf4bG3eU9N6n9032a0M8Y8z/72sw/6ZWMXb58Wblz59auXbvk5+dnLR88eLC2bdumPXv2PLLOyJEjNWrUqKQsEwAAAAAAACnIxYsXlSdPnv/aJ8WPWLPH0KFD1b9/f+t+fHy8rl+/ruzZs8tmszmwsmcrMjJSPj4+unjxotzd3R1dDv4l9mfqwv5MfdinqQv7M3Vhf6Yu7M/Uh32aurA/U5e0sj+NMbp165a8vb3/Z98UH6w9//zzSpcuncLCwhItDwsLk5eX12PXcXNzk5ubW6JlWbNmfVYlJjvu7u6p+gWQ1rA/Uxf2Z+rDPk1d2J+pC/szdWF/pj7s09SF/Zm6pIX96eHh8Y/6pfiLF7i6uqpChQoKCAiwlsXHxysgICDRqaEAAAAAAADA05TiR6xJUv/+/dWhQwdVrFhRL774oqZOnao7d+7o7bffdnRpAAAAAAAASKVSRbD21ltv6erVqxo+fLhCQ0NVtmxZbdiwQZ6eno4uLVlxc3PTiBEjHjkNFikT+zN1YX+mPuzT1IX9mbqwP1MX9mfqwz5NXdifqQv781Ep/qqgAAAAAAAAgCOk+DnWAAAAAAAAAEcgWAMAAAAAAADsQLAGAAAAAAAA2IFgDQAAAECawPTSAICnjWANAAAA+BcIa1KG+/fvy2azKS4uztGlAABSEYK1NIQvfQAAAE/PpEmT9N1338lms/E9K5lbsGCBihYtqtu3bytdunSEawCQhFL7ZyTBWhphjJHNZnN0GUgmYmNjHV0CAAApWkREhA4dOqQOHTpo3bp1hGvJXJ48eZQ1a1bVqlWLcC2V43UIJD8JWcSSJUu0detWB1fz9BGspQGBgYE6ePCgJOmdd97R1KlTHVsQklxgYKD++OMPSdKwYcP0008/ObYg/GN8OUw9EvblrVu39J///OexbUj+EvbVqVOntH//fm3atMnBFcFRPDw8NGrUKL399ttq27at1q5dS7iWjL366quaNm2aXFxcVL16dd25c4dwLZVIeM3duXNH0dHRio6OfqQNqUd8fLyjS4CdLl26pPHjx1vZRGp6/7UZ3m1SLWOMrl69qrJly6pGjRpydnbWqlWr9Ntvv6lMmTKOLg9J5Pz582rdurUKFCigzJkza968eTpy5IhKlizp6NLwGPHx8XJyevQ3j79bjpQhYdTwmjVrNHnyZP3xxx8qVaqUatSooffee08ZMmRwdIn4BxL246pVqzRgwABlz55dwcHBqlixogYPHqxatWo5ukQ4QHBwsCZOnKgVK1ZoyZIlaty4MWcKJDNxcXFKly6dDh06pJ07d6p379565ZVXtG7dOmXKlMlqR8qT8Fr76aefNH/+fJ05c0aVKlVSnTp11LJlS0eXh6foxx9/VLly5ZQ3b15Hl4J/YezYsZo+fboOHz6snDlzOrqcp4ajtFTMZrMpZ86c2rRpkzZv3qyVK1dq3rx5VqhGppo2FCxYUL1799a2bdu0ZMkSrVu3TiVLlkxVvxCkFsYYKzxbuHChRo4cqdGjR+vkyZOEaimczWbT+vXr1apVK9WvX1/r1q2Th4eHxo8fr+3btzu6PPxDNptNv/32mzp37qzhw4dbI9Y2b96skJAQR5eHJJbwOVqgQAENHDhQrVq1Urt27Ri5lgylS5dO3377rV577TUdO3ZML730kg4fPqwaNWpwWmgKZ7PZtHbtWr3xxhuqVKmSBg8erPj4eLVu3VqHDh1ydHl4Sj744AP169dPK1euVHR0NO+vKcBf91HCVERt2rRRwYIFtXr1akmpZwQiR2qpVMJ/0NjYWDk5Oem5556Tp6en1q5dq3379kl68EGUWv4j4/ES9m/CvCK+vr5asWKFzp49q3Tp0rH/k5GHRzcMHDhQ/fv3V0BAgFavXq0yZcpo7ty5ioqKcnCVsEd8fLzu3bunefPmadCgQXr//ffl4+Ojbdu2qXXr1vL397f6Ifnbt2+f/P399fbbb+v3339Xy5Yt1blzZ7399tuSpJiYGAdXiGct4WDh4R88ChUqpP79+xOuJVN//vmnBg0apL59+2r27NnasWOH5s2bp7i4ONWsWZPTQlOwW7du6csvv9TYsWM1ZMgQ1atXT5s3b1aPHj1UtmxZR5eHp2DUqFGaM2eOVq5cqffee09ubm6MCE4BEvbR/PnzdeLECev7Uf78+VWoUCEtWLBAklLN4IHUsRVI5OFTxi5evKjcuXNrz5492r59uwIDA/Xpp59q//79klLPf2QklnCAnrB/y5Ytqz179qhPnz66cOGCPvroI507dy7R/r9//75DasUDCR8+Z86c0YULFxQQEKBff/1V+/fv1wcffKD33ntPa9eulUQAk9I4OTkpQ4YMunXrll555RVdunRJvr6+atCggaZNmyZJWrdunfW+jOTtxIkTypYtm4wxqlWrlmrWrKk5c+ZIkhYtWqQlS5Y4uEI8Swk/gmzbtk09e/ZU9+7d9dlnn0mSihQpogEDBljh2k8//US4lkxERETo7t27qlq1qqQHn7kNGjTQBx98oFOnTqlp06bWyDWkLMYYnT17Vi+++KIuX76ssmXLqkGDBvriiy8kSd99952OHz/u4Cphr9DQUP3yyy+aO3euXnzxRUVGRmrnzp3q0qWLFi9erGvXrjm6RPwX58+f19y5c1W5cmW99957WrRokSRpzJgxViieWpCqpDIPn0o2ZMgQ1a1bV4ULF1aTJk10+PBhbd68WUFBQZo8ebJ2794t6cFkrgkHBUj5Hg5W169fr59++kkHDx5UlixZ1KFDB7Vp00aXL1/WiBEjdO7cOUlSu3btFBAQ4MiyIWn58uVq3Lixzp8/L29vb9lsNjk5OWnkyJHq3bu3evbsqbCwMALxFCYuLk73799XTEyMlixZoho1aqhRo0aaMWOGJOnatWtasmSJjhw5wgF4MhUWFqZ79+5Jkl577TVt2LBBWbNmVdOmTfXll19awXhgYKB27txp9UXqY7PZ9MMPP6hp06aKiIiQk5OTJk6cqHfeeUeSVLhwYQ0YMEBt27ZV48aNtWHDBkZWJAN58+aVp6dnoivRubm5qXHjxipatKgCAgLUsGFDGWN4H07mEvZPwg/Crq6u8vX11YEDB1StWjU1aNBAs2fPliRduXJF69ev14kTJ9ivKZSbm5vOnTunffv2ac+ePerfv7/69u2r06dPq2PHjvruu+8cXSIe8tcf/wsWLKhdu3bpq6++0nPPPafu3burWbNmmjt3rsqVK6dTp05JSiVTVBmkGnFxcda/V6xYYby8vMzq1avNwoULzcCBA42Tk5NZtGiROX/+vClevLipWLGiKVWqlHnhhRdMdHS0AyvH0xIfH2/9e8CAASZbtmwmb968Jk+ePOajjz6y2r788ktTo0YN4+vra6pWrWpy5cpl7t+/74iS8f/Fx8ebr776yvj5+ZmsWbOasLAwY4wxUVFRxhhjDh06ZHLnzm1+++03R5aJfyDhdXj9+nUTFxdnvbYCAgJMtmzZTNmyZRP1//DDD03hwoXN+fPnk7xW/G8HDx40efLkMcuWLTP37t0zly9fNm3btjUFChQwa9asMcYYc+3aNfPBBx8YT09Pc/LkSQdXjGfpwIEDpkCBAmbWrFnGGGPOnTtncuTIYWw2m3njjTesfqdOnTL9+/c3p06dclSpaVbCe3BcXJz1GRoXF2feffddU7VqVbN27Vqrb3R0tGnfvr1ZvHixCQkJcUi9+OcS9u3GjRvNhx9+aH1ufvTRR8Zms5kGDRqY2NhYq//QoUNNsWLFzIULFxxSL+y3YcMG8/vvvxtjjPniiy9Mrly5TMaMGc3AgQPNhg0bjDHGtGvXznTq1MmRZeIhD2cRW7ZsMevWrTMBAQGJ+pw4ccL06tXL1KlTx9hsNmOz2cyuXbuSutRngmAtFfrll19Mly5dzOTJk61lkZGR5vPPPzfp06c3O3fuNKdPnzZTpkwxY8eOtQ76CFZSjwsXLpjy5cubQ4cOmcOHD5vp06cbFxcXM2DAAKvP2rVrzccff2z69u1r7fuHv4zg2Xr4wydBbGys+e6770zhwoVN9erVzX/+8x+r7cyZMyZ37tzm119/TcoyYafVq1cbPz8/U7FiRTNhwgRz5swZY4wxkyZNMk5OTqZFixame/fupl27diZr1qzmwIEDDq4Y/039+vVNoUKFzHfffWeMMWb37t2mefPmJmvWrKZ06dKmSpUqxsfHh/2YBnz77bemf//+xhhjQkJCTIECBUzXrl3N999/b5ydnU23bt2svjExMY4qM81KCF5+/vln0759e1OqVCkzbNgws2vXLhMZGWlq1qxpXn75ZfPRRx+ZzZs3mz59+pjChQubP//808GV45/6/vvvTZYsWczAgQPN8ePHreXvvvuuyZw5s/nwww/N8OHDTZcuXYy7u7s5ePCg44qFXYYMGWJ8fX3N1KlTTUxMjLlz5465ePGiOXHihNUnLi7OvPLKK2bkyJEOrBQJHh7c0bdvX5MjRw6TM2dOkydPHlOtWjUTHh5utUdHR5uoqCgzffp0U6VKFdOjRw8TGxub6DFSIpsxqWHcHRKEhoaqWrVqCg8P1/vvv68PP/zQartx44Y6duwoHx8fTZ8+PdEpg1xmPPWYNGmS9u3bp2zZsmn69OlycnLS3bt3tWzZMvXo0UN9+vTRp59++sh6/B9IOg+/9jZu3KgbN27IGKNGjRopS5Ys+v777/XJJ58oPj5eEyZMUHR0tGbNmqXLly9r37597Kdk7vDhw6pVq5b69eun33//XWfOnFGePHk0btw4FS5cWJs3b9bUqVPl4uKi/Pnz691331WxYsUcXTb+v9jYWDk7Oz+y/I033tDevXs1ZcoUNW/eXGFhYTp06JACAwNVvHhxValSRfny5XNAxUhK9+/f1+HDh1WuXDk1adJEzz//vBYtWqRr166pWrVqOn36tFq3bq2lS5c6utQ0a82aNWrVqpX69u2rvHnzasmSJbp69ap+/fVX2Ww2jRs3Ttu3b9eNGzeUOXNmLVmyROXLl3d02fgHTp06pbp162rYsGHq2rXrI+0jR47U3r17de3aNZUuXVp9+/ZViRIlHFAp7DVy5EhNnz5da9asUZkyZZQpU6ZE7Xfv3tXhw4c1duxYXbx4UUFBQY/9zEbSMQ9dgC0wMFDdu3e3Tv0MCwvTe++9JycnJwUGBip9+vSJvmd9/vnnmjx5so4ePSp3d3dHbsa/59hcD8/C4cOHTaFChUz58uUf+fW8c+fOpl69eg6qDM/a7du3zdChQ427u7upVatWorY7d+6Yr776yqRPn9688847DqoQDxs8eLDJkyePqVWrlsmdO7d55ZVXzMaNG018fLxZuXKleeGFF4yrq6tp0aKFGTFihLl7964xhpGFydHDv7Lt2LHD9O3b17q/dOlSU6NGDdOsWTPr1/WEUaKPG7mIpLd48WIzY8YM6/4vv/xiduzY8cj+adasmfH09DTfffeduXPnTlKXiSSW8Lq+efPmI7+mX7582ZQrV85s2rTJGGPMrVu3TKdOncwPP/xgjVBF0kl4rYaHh5tq1aqZzz//3BhjzL1790z27NlN//79E72eo6Ojzfnz582NGzccUS7+gVWrVj0yVc22bdtMqVKlzJ9//ml9F/rr+3R0dLSJiYnhTJwU6MKFC6Zy5crmp59+MsY8eJ/dtWuX6d27t1mxYoWJiYkx69atM6+//rqpU6eONSqY78XJw9dff22aNWtm2rdvn2h5SEiIKVKkiGnZsqW1LOF1GxERYQoWLGh27NiRpLU+C8yAnQqVLl1aq1atUlxcnKZOnapDhw5JenA56pMnT8rHx8exBeKp+esEkZkyZVKPHj00aNAgbd26VVOmTLHaMmbMqNatW+uTTz7RyZMnU8ckkSnYvHnztGTJEq1evVpbtmzRsGHDtHPnTsXFxclms6lFixYaPXq0qlatqlu3bqlXr17KkCGDoqKiGLGWzJj//0vdzp07NX36dP3888+JJitv06aNunTpouvXr2vkyJE6dOiQ9Usdk5o73p07d7R48WItXbpU8+fPlySNGDFCTZs21e7duxO9z37//fcqVKiQRo0apVWrVikqKspRZSMJ2Gw2rVmzRs2bN1flypU1bdo0nTlzRpLk4uKiP/74Q+vWrdN//vMfjRkzRnv37lXVqlVVuHBhB1eeNixevFiff/55olHgbm5uunv3rho1aqTz58+rcOHCatasmSZNmiQnJydt3rxZwcHBcnV1VYECBZQ1a1bHbgQe69ixY+rSpYuuXr2aaPmlS5d06tQpZcqUSenSpVNcXJy174OCgqx96+LiwiimFMjd3V3Xrl3Tvn37tHfvXg0cOFDvvfee9u/fr9atW+vbb7/Vq6++qiFDhmjDhg1ycXFRbGws34uTgRs3buj777/Xzp07deHCBWt5bGysfHx81KtXL506dUrXr1+XJOt1+8UXX+g///mPChYs6JC6nypHJ3t4dg4cOGB8fX2Nl5eXadSokWnWrJkpV66c9etPSj+POa17+Be648ePm+3bt5vQ0FATExNjoqOjzfDhw03mzJnN1KlTE60XFRVl7Xv+DzjOwIEDTa9evYwxxqxcudJ4eHiYmTNnGmMejDyMiIgwcXFxZunSpeall14yTZo0STQ/AZKXH374waRPn94UL17ceHh4mBw5ciSa+8UYY5YvX27KlStn2rVrxwVjkpnLly+bFi1amOrVq1vzqNWqVcvkz5/f7NixI9Gv4b169TKurq6mXLlyJjIy0lElIwns2bPHZMqUyXz00UemdevWpkyZMqZdu3bWnE3z58836dOnN/nz5zdeXl7MsZdE4uLizN27d029evVM5cqVzZw5c6zvRBcuXDDFixc3y5YtM4ULFzZdunSx2s6cOWNat25tNm/e7Mjy8Q9FREQYYx5Mdp5wEYrg4GBTqlQp069fP3Pt2jVjzP+NVnr77bfNmDFjGAmewo0cOdIULFjQuLq6mv79+5v169cbY4xp0aKFee+99xL1ZV87zuOOIc+dO2feeecd89xzz5lPP/00UdvKlStN4cKFzZUrVxItX7hwYar57CRYS+WOHj1qChQoYF5++WXrClbGMKFuSvfwm1nCFY9y585typcvbzp37mwuXbpkrl+/bkaNGmU8PDzMtGnT/utj4Nn66986Li7OvPnmm2by5MkmKCjIZM6c2Xp9xsXFmc8//9zMnz/fur9ixQrj6+tr3nzzTb5EJBMP79MbN26YESNGmHnz5pm4uDjz448/mtq1axs/Pz9z7NixROt988035o8//kjqcvE34uPjrc/D48ePm/r165sXX3zRumpgjRo1rHDt3r17xpgHkyrv3LmTyc5TufPnz5tRo0aZCRMmWMuWLFliqlWrZlq3bm1d/fXs2bNm06ZN5tKlS44qNc1J+JEpPDzctGrVylStWtXMnj3b+nwcPHiwsdls5vXXX0+03gcffGBKly5tLl68mOQ14597+KquoaGhJl26dKZbt27m/v37Jj4+3gwdOtSa8PzSpUvmxIkT5oMPPjA5cuRINLk9UoZvvvnGjBw50owZM8YEBgYaYx68rx4+fNjqEx8fb6pVq2bGjh3rqDLxkIePRf46Tc25c+dM586dTYUKFcyoUaPMrVu3zLlz50ydOnVM7dq1U/XxJ8FaGnDw4EFTuXJl07VrV+b9SGUmT55sPD09zdatW40xxnTs2NFky5bNOk89LCzMjB492thsNvPNN984stQ06+GRLufOnTNhYWHGmAfzEKRPn97YbDazfPlyq8+tW7dMnTp1zIcffmitGxsba7799lsTHBycpLXjUT///HOi+0FBQcbT09NUrlzZ7N6921q+fv16U79+fVO5cuVHRq4h+Uj4gvf111+bN9980/j5+ZmMGTOafPnymR9//NEY82DkWuHChU3r1q1N+/btTaZMmXgtpnLnz583lSpVMt7e3uaTTz5J1LZkyRLz0ksvmXbt2pn9+/c7qMK0a+nSpaZWrVrWQXd4eLh58803TdWqVc2sWbNMXFycuX79umndurVJnz69mTVrlvn8889Njx49TJYsWcyhQ4ccvAX4p65fv26MeTDSJVOmTKZnz57GmAcH9WPHjjWVKlUyNpvN+Pr6msKFC6eaUS9pyaBBg0zevHmteblsNptZtGiR1X7nzh2zf/9+U79+fVOmTBnmzUsGHg7VpkyZYpo1a2Zq1aplxo0bZ/3ocfr0adO5c2fj4uJifHx8zJtvvmmaNGli/UCZWgcJEKylEQcOHDAvvviiadmypfUrK1KWhw/O4+PjTVRUlGnSpIk1Gu2nn34yWbJkMV9++aUx5sEpnzExMebq1atm/vz5fBglsZkzZya6xPuQIUNMiRIlTPbs2c2gQYPMmjVrzKBBg4y3t7cJCAgwd+/eNWfPnjX16tUzFSpUsPYXE7ImH4GBgSZr1qzmypUrViCzd+9e07hxY+Pq6vrIxKsbNmwwjRs3NkWLFuV9NxnbvXu3yZgxo5k3b545deqUOXPmjKlRo4apWLGiFa598MEHpkmTJqZevXrmyJEjDq4YSWHChAnGx8fH+Pv7mwsXLiRqW7ZsmSlRooTp2rVroukV8Owk/I2//PJL8/LLL5sWLVpYr8WEcO2ll14yc+bMMcYYc/XqVTN48GBTtGhRU6lSJfPGG2/w2k1Bjh07ZrJnz25OnTpljHkw3YKrq6vp0aOH1SciIsJs3LjRHDhw4JHTy5D8rVq1yuTOndv6UXLlypXGZrNZZ2wkLHv99ddNrVq1uFBBMjNkyBDz/PPPm88++8wMGjTIVKhQwbz++uvWa/HcuXOmS5cupnz58mbEiBHWegnhWmpEsJaG7N2717zyyivm8uXLji4FT6hz585myJAhiZbFxMSYOnXqmKCgILNp0yaTOXNmM3v2bGPMgysiffnll9ZItgSEa0nj/PnzJk+ePNYo0R9//NHkzp3b/PDDD2bUqFHGz8/PvPXWW2bChAmmd+/extnZ2eTNm9eUKVPGVKtWjS8PyVR0dLQ1p8vp06et5UFBQaZu3brG09PTOghIsGbNGtOiRQtGOCVjX375pfH19bVOZzDGmD///NNUq1bN5MuXz6xevdpanpq/EKZlfxeMTZo0yZQqVcr079//kVO4v/76a17XSSggIMD69/Lly82rr75qXn/9dWvk2tWrV81bb71lqlSpkmjOtdDQUGtONqQc586dM35+ftb3WmOMWb16tXF1dTU9e/ZkSptUYMqUKdZVIr///nuTOXNmKxiPiIgwwcHB5tatW2bbtm3W65njmORh+fLlpmjRombfvn3GGGPWrl1r3NzcTJEiRYy/v791ds6pU6dM586dH3ktp1YEa2kMBwUp09atW60vEQ/P49KoUSNTpEgR4+7unugXnsuXL5tXX33VGr2GpHfw4EFToUIF07dvX9O/f38zd+5cqy1hDq4WLVqYTZs2mWPHjplvvvnG/PLLL3x5SGYeN1w9ODjYODs7m/79+1vLgoKCTIMGDYyPj88j4drt27efeZ2w3+LFi03RokWtUxgS3muPHDliMmfObEqWLGkWLlxojGFuytQoYZ/u2LHDfPTRR2bEiBGJ3q8nTpxoypYta/r27fvIyDUkjTVr1phy5colGpW0aNGivw3XEg7iHv5xitduytOtWzdTsmTJRMtWr15tMmXKZDp37sxFgFKor7/+2pw/f958+eWXpm3btmblypWJ5ho25sG8a3369El0gaDUevpgSvDw3z42NtZs2LDB9O3b1xjz4JgmW7ZsZvr06Wbu3Lkma9aspmnTptbx6rlz50y3bt1MsWLFzFdffeWQ+pMKwRqQzD38ZXDu3LmmYcOGZufOncaYB5N7li1b1vrice/ePXP9+nVTv359U61aNUY8OVhQUJCpWLGiee6558yUKVMSta1Zs8bUrFnTNG3a1JqsNQH7LXkJCQmx5ihcsWKFadOmjZk2bZrJkCGD+eCDD6x++/fvNw0aNDAFCxZkXrUU5MyZMyZ9+vRm2LBhiZbv37/fvPLKK6ZVq1YEKqnc999/bzJmzGgaNmxo/Pz8TObMmU3Tpk2tg4nx48ebSpUqma5du5qQkBAHV5v2XLhwwTpIe3iu4L8L11q3bm18fX3NggULHFEunlDC99x79+4l+s4bERFhChQo8MiE9V9//bXx9PQ0oaGhSVon/r2xY8caHx8fc/z4cbN27VpTsGBBkyFDBvP5559bfW7dumXq169vevXq5cBK8Tgff/yx+eKLL4wxD0YD37x501SpUsWMHz/eGPPgYl5FixY1zz//vOnVq5f1ev79999Nr169Uv0ob4I1IIW4d++eWb9+vSlXrpxp27at2bNnjzHmwRwFnp6epnDhwqZy5crGz8/PlCtXjtMJk4kjR46YggULmjp16jwyv8tPP/1kSpYsaZ3myy/qyU9MTIxp2bKleemll0y/fv2MzWazDtbmzp1rnJ2dE4VrBw4cMNWqVTMlS5Y0MTEx7NMUYsmSJcbFxcV88MEHJjg42Ny4ccMMGzbMdOjQwURERDi6PDxDISEhpkCBAtaB3d27d82OHTuMt7d3oqtKjhw50lSvXp2DeQc6deqUKVSokBk9erS17HHhWlhYmOnUqVOqP4hL6eLj463PyC1btpjKlSubzz77zJqyJi4uzgwcONC89tpr1sjvhLD71q1bjikadjt37pzp2rWrddVtYx7MX+rs7GymTZtmdu/ebfbu3Wvq1q1rypYta525wfcox3l4pNrKlSuNl5dXoov2HDlyxHh7e5tdu3YZYx7s47feest8//331roJ+y8tnIlDsAYkU6tXr7bmcRo8eLA1YeuPP/5oKlWqZFq2bGldASk8PNyMHj3aTJgwwSxcuNAK09LCm1hKcOjQIVOuXDnTtWtXc+zYsURtv/32G+FnMnfjxg1TuXJlY7PZTPfu3a3l9+7de2y4dujQIUa1pDDx8fFm+fLlJnPmzKZAgQKmUKFCJlu2bCYoKMjRpeEpGj58eKKLyhhjzNGjR03+/PnNiRMnEi3/9ddfjbu7u1mxYoW1LGGORSSdhIOymJgYExYWZnr37m1KlixpJkyYYPVJCNdatGhhvWY5bSzl2LRpk5k2bZpp3ry5qV27tnn++efNtGnTzJkzZ8zFixdN+vTpE109HSnPihUrjM1mM/ny5XvkQk+9evUy5cqVM+nSpTN+fn6mTp06DA5IZn744QczZswYM2nSJGPM/+2X4OBgU6FCBdOpUyezc+dO4+/vb5o0aWK9/6a192GCNSAZun37tqlTp47JnDmz6dixo8mYMWOig4HVq1db4dpfTyNMwIdR8nLgwAFTvnx507Vr18eeJsj+Sr5iYmJMzZo1TdmyZU2dOnXM0qVLrba7d++auXPnmgwZMpg+ffo4rkg8FcHBwebHH380K1euZLRLKhMTE2Nef/11c/To0UTLL1++bLJkyWLNo5fgxo0bpkSJEmbq1KlJWSb+v4dHqaxbt846Vfv8+fPm/fffN0WLFk0Uri1ZssSUK1fOtGvXzkRHRzPKJYXYt2+fsdlsZu3atSYuLs5cvnzZjBs3zpQrV874+vqasWPHWvPmcfG1lK1ly5bGZrOZadOmmaioqERtFy5cMHv37jXBwcHMNZzMREZGGmdnZ2Oz2Uzv3r0Ttd2/f99MmjTJlClTxuTJk8e8/PLLViia1kI1Y4yxGWOMACQ7d+/eVaFChXTjxg19/fXXeu211xQTEyNXV1dJ0po1azR27FgVKVJEXbt21SuvvOLgivG/HDx4UO+8847y5cuniRMnqkCBAo4uCf9QdHS0bty4oS5duuju3bvq1KmT2rZta7VPmTJFEyZM0NGjR5UjRw4HVgrg7xhjZLPZtHXrVnl4eKhChQqKiYnRO++8o8uXL2vw4MGqVauW1f+VV15R06ZN1a9fPwdWnbZ89913KliwoMqXL6/4+Hg5OTmpTZs2euGFFzRixAhJ0oULFzRr1iytXr1anTp10uDBgyVJK1eulJ+fn/Lly+fITcA/dOTIEf3+++86dOiQxowZk6jtxIkTCgoK0siRIxUcHKw8efLo0KFDypYtm4OqhT02bdqk559/XuXLl5ckvf7669q2bZuWLVumOnXqyNnZ+bHrJbz2kfQSPicfdunSJVWvXl1ubm5avny5ypYta7XFxcXpP//5j65evSpfX185OTkpNjb2b/dtasb/WCCZun37tvLmzaty5cqpW7duOnHihFxdXXX//n1JUpMmTfThhx9q586d2rp1q4OrxT9Rrlw5TZ8+XVmyZOGLfwrj5uYmLy8vTZs2TRkzZtTChQu1ZMkSSdKIESN0+PBhnThxglANSEYe/u04NjZWNptN8fHx+vTTT/Xqq6/qwIEDcnV1VefOnRUfH6+xY8dq1qxZCgwM1IABA3T06FE1adLEgVuQtpw8eVLjx4/XiBEjdPToUTk5OckYo7CwMLm4uFj98uXLp+7du6tp06ZasmSJRo0aJUlq2bIln60pxJ07d1SvXj29+eabunjxorU8Li5OkuTr66t27drp0KFDmj9/vn755RdCtRTm4MGDev311zV//nydOHFCkvTDDz/opZdeUocOHbRlyxbFxsY+dl1CNceIj4+3QrUrV67o+vXrCg0NVe7cufXLL7/o+vXr6t+/v06dOmWtky5dOnl6eqpkyZJycnJSfHx8mgzVJIkRa0Ay8bhfZ6KionTnzh21a9dO+/fv17Zt21S8eHGr3Rijw4cPq1SpUkqXLl1Slww7JfwaxC9yKVNwcLAGDBigM2fOKH369Dpz5ow2btyoypUrO7o0AP9fwvvstWvXlD17dknS9u3blStXLuXOnVtvvfWWgoKCtHbtWlWoUEG//fablixZopUrVypXrlxycXHR4sWLE/0yj2dvxYoVWrBggTJmzKhRo0apTJkyatKkiZo0aaIuXbooKipK6dOnlySdO3dOU6ZM0Z49e7RhwwZly5btkZEWSL6OHTum1q1byxijtWvXKn/+/Ina4+Li+G6bQn388ce6f/++Zs+erYiICLVv3159+/ZViRIlJEmNGzdWUFCQZs6cqcaNG7Ofk4GHR6qNGTNGmzZt0tWrV+Xp6al+/frptdde08WLF1WpUiWVLFlSM2bMUNGiRR1cdfJCsAYkAw8HLPv27ZPNZpOTk5M1dPrPP/9Ut27ddPDgQW3YsEHFixdXu3btVLBgQY0fP14SX0BSmscNtUbKcenSJW3cuFF//vmn3nrrLb5cAMlQeHi42rZtK39/f+XLl09vvvmmNm7cqDp16ujevXtq1qyZDh8+bIVrxhhdv35dd+7cUZYsWfTcc885ehPSjIdPHfruu+80Y8YMubu7a9y4cZo0aZIaNWqkZs2aPbJeRESEoqOjlTNnzqQuGU8g4TvPvXv35OrqqtjYWLm5uenYsWOqW7euSpcurWXLllkhOFKuiRMnaty4cfrhhx+UMWNGHTp0SIMHD1arVq3Uu3dv+fr6SpJeeuklZcuWTevWrXNwxXjY8OHDNXPmTC1YsEDZs2fX8OHDtX37dp05c0b58uXTn3/+qRdffFE5cuTQ2rVrlTdvXkeXnGwQrAEO9nDAMmzYMK1YsUJOTk4KCwvTiBEj1KdPH6VLl06XLl3Se++9p3Xr1qls2bKKjIzUiRMnEp0eAQAAHggLC9O4ceO0YcMGXbhwQXPmzFH79u2tECchXDty5IjWrl1r/ZiFpJfwXejEiRPKlSuXtm3bpunTp8vV1VXbtm1Tnjx5rNAlXbp0io6OVu7cubVixQprBBuSp4R9+/PPP2vp0qU6c+aMXnzxRTVo0EANGzbUsWPH5O/vr9KlS2vp0qWEaylYXFycGjVqpOLFi2vy5MnW8hUrVqhjx47q0KGDevfurZIlS1r9GRSQfISFhalFixb68MMP5e/vr3Xr1qldu3YaN26cunfvbo0Y/uOPP9SrVy/9+OOPnHnzEP4SgIM9POz2q6++0vz583Xo0CF17NhRAwcO1MiRIxUXF6fcuXPrxx9/1Pz589WtWzedPHlSLi4ufzs/AQAAaVV8fLw8PT3VqFEjhYSEyMvLS9euXZMkOTs7KzY2VhkyZNCqVatUvnx5Va1aVYcPH3Zw1WlTQvCyevVq1axZU9OmTVOTJk3UuXNnRUdHy8fHR35+fhowYIDat2+vJk2aqGXLlho3bhyhWgpgs9m0Zs0aNW/eXCVKlFDnzp117do1NWnSRCdPnlTJkiW1adMmnThxQo0bN9b169cdXTLsEB8fr/j4eEVHR1tz5cXExCg+Pl6tWrVSjx49tGLFCs2bN0/nz5+X9CAkj4+Pd2TZeMidO3d07NgxlShRQhs2bFCrVq00fvx4de/eXffu3dOMGTN09uxZ5c+fX2vXrpWTk5O1ryGlzZnlgGTm1KlT2r17t+bNm6fq1avrxx9/1OLFi9W+fXuNHz9eNptNQ4cOVYYMGdShQwdrvbi4uDQ7QSQAAI9jjJGTk5MiIiL03HPPadWqVdq6datWrlypqKgoDR06NFG4tnLlSnXq1EkZM2Z0dOlpks1m008//aTWrVtr2rRpqlevnpycnNSqVStlyJBBX331lW7evKmyZcuqUKFCji4XT+jmzZv64osvNH78ePXt21dXr17Vxx9/rO7du1vzBpcoUUJr1qxRq1atdPv2bS5UkAI5OTnJyclJNWrU0IQJE9SzZ08VKVLECl6yZ8+uF198UYsWLVLu3Lk1cOBA670aSe9xU9LkzJlTr776qiZPnqy5c+dq0qRJ6tatm6QHV2Pevn27SpUqpcKFC1vrM+Lw//A/GXCAv/46kz17djVu3Fg1atTQjh071KNHD40ZM0YLFy5U+/btNWbMGH3wwQfWFUET8GYGAMD/Sfiyv3btWrVq1Up37txR/fr1NWDAAFWuXFmrV6/WhAkTJD0YubZ8+XJduXJFK1euVJEiRRxcfdoUFRWlRYsWqV+/furSpYuyZcumM2fO6NNPP1XGjBn18ssvKy4uTl26dNHx48cdXS6e0P379/XHH3+oevXqunz5ssqVK6eGDRtq+vTpkqTvv/9eZ86cUZkyZXTo0CHmbEphDh06pMDAQOvKn8OHD9fLL7+s6tWr68iRI4qOjlZMTIz27dunoUOHatCgQRo9erSuXbvGXMMOcv/+fetvHxERoYiICElS5syZ5e3tralTp6p9+/ZWqHbr1i0NGDBA9+7dU+3atSWJffcYDHUBktjDFyo4e/asMmTIoOeff17vvPOOJOmbb77Rq6++qs6dO0v6v18PDhw4wOg0AAD+i4TTzlq1aqUhQ4YoR44ckiQvLy99+OGHGjdunFavXq1z584pR44cGj9+vM6cOcNBggMZYxQcHCwvLy9dv35dI0aM0NGjR/X7778rXbp06tevn9544w1999138vDwcHS5+B8Swu1Dhw4pe/bs8vT0VPHixXXgwAGNHTtWDRo00KxZsyQ9uDjXTz/9JGdnZxUuXFiurq4Orh5PYsiQIfrhhx8UGhoqHx8fFSxYUGvWrNGiRYv07rvv6sUXX1Tx4sV1+/ZtazRbZGSkcufOLTc3N0eXn+asWLFCrVq1subnHjVqlNauXStjjBo1aqRRo0bpiy++0KVLl/Tjjz/q2rVrev7553XkyBHduHFDQUFBcnJySnQsi//DXwRIQg8PeR4yZIgaN26ssmXLyt/fXzNmzJAkHT9+XMYYpU+fXvfv39epU6c0YMAAbdu2TTabTVxvBACAxwsPD9fIkSM1YsQIDRs2zLoCXVxcnDw9PfXhhx+qQYMGOn36tDZv3qwDBw5weqGDZciQQb169dLcuXNVoEABXbp0SZ06ddLly5fVrFkzbdy4UR06dNCKFSuUJ08eR5eL/+Lh+fIaNmyoOXPmyNnZWXnz5lW3bt1Urlw5zZ492zrjYsaMGdqzZ4/Kly9PuJ3CfP755/rqq680Z84cbdu2TcOGDdPvv/+uqlWrytPTUz/88IMWLVqkDh06qE+fPjp+/LicnZ31yy+/yNPTk+OZJBYQEKA2bdpo+PDhkqTp06dr5syZatmypWrWrKmJEyda0w2tWrVKvXr1kouLi8LDw60BHglzexOqPR5XBQWSyMPp/sqVK9WvXz/Nnj1bN2/e1PHjxzVlyhTNnDlTefPmVf369dWoUSNduHBBxhhrtNrjzocHAAAPXLx4UTVq1NDixYtVtWpV6+Dt4c/O2NhYxcfH6969e4yASkZOnDihS5cuqU6dOtZ3pp49e+rmzZuaP38+o5lSiJ9++kktWrSw5stLCEM7duyon3/+Wf369ZOTk5POnz+vFStWaMeOHSpTpoyDq8aTatOmjQoUKKAxY8ZIenCcExQUpDZt2qhevXqaNm1aov4XLlzQJ598oq+//lrbtm1TqVKlHFF2mnXv3j2tWLFC3bt314cffihvb289//zzatq0qSRpw4YNeuutt9SkSRMtWbLksY/BVVz/O84rA5JIQqj266+/KiAgQIMHD9Zrr70m6cG56z4+Purbt6/mz5+vb775RqtXr1bt2rU1YcIEOTs782YGAMD/YLPZFB4eruDgYFWtWlU2m03x8fGy2WwKCgrSjRs3VLNmTTk7OxPUJDO+vr7WCMPff/9dS5Ys0dKlS7Vz5072VQrx1/ny7t69q99//11r1qxRs2bNdPPmTW3fvl1hYWEqWbKkdu3apZIlSzq6bDyBgIAAVa9eXf/5z39069Yta7mTk5MqVaqkpk2bKigoSNHR0dbpntevX9f27dsVHBysX375hVAtiRljlCFDBnXq1ElRUVEaNGiQ4uLitHjxYqtPvXr19M033+jNN99Up06dNG/evEcGc3Ac+t8RrAFJKDQ0VF26dFF4eLjef/99a3mWLFnUpk0bbd26Vb/99pt1qfmEL5KxsbHMrwYAwEMSRnE/PJo7e/bsatSokRYtWiQfHx+98sor1g9bCxYs0JUrV1S1alVlyJDBkaXjvwgKCtKkSZN06NAhbdu2jeAlBXncfHlHjhzR2bNn5eLiot69e6tbt25ycnIi3E6Bhg8frlWrVlmn+i5fvlwbN26Uv7+/1adgwYL69ddfEwVr2bJl02uvvaamTZsqS5Ysjio/Tfrll1906dIltW3bVu+9956io6M1c+ZM9erVSzt37tSbb75p9fX399e3336revXqqUiRIho6dKgDK095OBUUSGJHjhxRs2bN5OHhoblz56pcuXJWW5cuXXTp0iWtX7/egRUCAJC8JYRpmzdv1qZNm3Tr1i31799fL7zwgjXfT7p06fTWW2/J29tbmzZt0rJly7R9+3ZGSyRz9+7d0/79+5U/f375+Pg4uhw8ocWLF+vdd9+Vi4uLatWqpaZNm6p9+/bq06ePjh49qk2bNvFjcQp09OhRffDBBxo0aJCqV6+u4OBgtWnTRjlz5lT79u3VtGlT3bhxQ61atVL27Nm1fPlypq9xIGOMbt++rebNmysmJkZZsmTRjh07tGvXLhUvXlwLFizQO++8oyFDhujjjz9OtO7evXtVvnx5XqdPiGANcIAjR46offv2KlOmjPr166eyZcvq1q1bqlevnkqUKKE5c+Y4ukQAAJK1DRs2qEmTJmrQoIGOHj2qiIgITZs2Ta1bt9bu3bu1cOFCffPNN8qVK5eyZs2qmTNnMpcTkAT+br68W7duac6cOVwRMoWZOXOmvv76a8XFxWnVqlXKmTOnpAcXXOvTp48uXbqkmzdvKleuXIqLi9P+/fvl4uLC3NDJwPXr1/XSSy/p999/17hx4zRkyBBJD07bXrZsmd59910NHTpUo0ePfmRdzph6MgRrgIMcPHhQbdu21fXr11WxYkW5uroqODhYu3fvlqurKx9GAAD8RcJn482bNzVq1Cj5+vqqa9eukqS3335b69ev16effqp27dpJkm7cuKH4+Hi5urpyChLgAKdOndKSJUs0Y8YM7dy5k1N7U6CtW7fq7bffVnh4uL7//ns1aNDAagsNDVVISIh+++03eXt764033lC6dOkIZZKJmzdvqk2bNrp9+7bc3NzUvn17tW3bVtL/XdCgR48e6tatmz7//HMHV5uyEawBDnTs2DE1adJEefLkUevWrfXuu+9Kku7fvy8XFxcHVwcAQPKzd+9evf766/L29taYMWMSze/TqVMn/fzzz5o0aZIaNWrEVT8BB3p4vrwVK1YwYjSFOXv2rNzc3OTj46Pz58+rTp068vX11YgRI1SxYsW/XY8LriU/oaGh6ty5s+7du6fOnTurTZs2kh4cc06dOlU///yztm7dyqCOf8HJ0QUAaVnJkiW1atUqxcTE6MCBAzp79qwkEaoBAPA3XnzxRZUsWVJBQUE6f/684uLirLb58+ercePG6tSpkzZs2CB+PwYcx9fXV927d9fGjRsJ1VKYIUOGqHHjxipXrpyqV6+uI0eOaMuWLTpx4oQmTpyooKAgq+9f32cJ1ZIfLy8vTZ8+XRkzZtSiRYu0YMECxcXFqX79+goLC7NCNT4z7ceINSAZOHjwoN59910VLFhQI0aMULFixRxdEgAAycrx48d19+5dVapUSZJUv359HTx4UEuXLtWrr76a6GCuZ8+e6tOnj4oUKeKocgEgRVq5cqX69eun2bNn6+bNmzp27JgmT56sBQsWqFq1aqpbt65efPFF9e7dW1WqVHF0uXgCwcHBGjhwoE6ePKmoqChlypRJQUFBTEP0FBCsAcnEvn37NGjQIK1YsUK5cuVydDkAACQLxhhdvXpVL7/8sipUqKABAwaoQoUKkqTatWvr5MmTWrRo0SPhGgDgyfz6669atmyZfH191a9fP0nSrVu3tGDBAr3//vsKCAhQhgwZVK1aNQ0cOFCjRo1ycMV4UleuXFFQUJDCwsLUoUMHOTs7MyfeU0CwBiQjUVFRSp8+vaPLAADAYW7duvXYCw2sXLlS48aNU6VKldS9e3drjp/atWvr7NmzmjVrlurWrUu4BgB2CA0NVbVq1RQeHq73339fH374odV248YNdezYUT4+Ppo+fboOHTqkUqVK8X6bCjAn3tPBHGtAMkKoBgBIy7p166Y+ffooJiZG0oOrliVo2bKlRowYocDAQM2ePdua42fLli3KkSOHBgwYoOjoaIfUDQApnZeXl1atWqWcOXNq1apVOnjwoNX23HPPKUeOHNZ80GXLllW6dOkSzXGJlIlQ7ekgWAMAAIDDrVy5UqtXr1bv3r3l6uqqXbt2ady4cTp//rzVp3nz5ho9erQ2btyoqVOnWgd++/bt04YNG5QxY0ZHlQ8AKV7p0qW1atUqxcXFaerUqTp06JCkByOJT548qbx58ybqTygDPMCJtAAAAHC4ixcvKnv27Cpbtqw2btyoTp06KTo6WvHx8erWrZvy5csnSXrjjTcUGhqqDz74QLGxsRo4cKAqVKjwyAEfAODJlS5dWgsWLFDbtm1Vv359VaxYUa6urrp3756mT58uSUx0D/wFI9YAAADgcDVq1JAxRjVr1lSDBg2sOdUWLVqk2bNn648//rD65syZUwUKFNCff/4pb29vxxUNAKlQuXLl9PXXXytDhgyKiIhQnTp1dODAAbm6uur+/fuEasBfEKwBAADA4SpVqqRatWrp119/VaVKlfTyyy+rW7duGjRokJYsWaIvv/xSR48elSQdOXJE7777rtatW8eVtAHgGShZsqRWrVqlmJgYHThwwJpfzcXFxcGVAckPVwUFAACAw927d0+NGjVSwYIFtWvXLpUuXVorVqyQJE2fPl2zZ89WdHS0cubMqSNHjigwMFAlS5Z0cNUAkLodPHhQ7777rgoWLKgRI0aoWLFiji4JSHYI1gAAAJAs3L17VxkzZtT8+fM1ceJElStXzgrXNm/erBMnTujq1atq166dihYt6uBqASBt2LdvnwYNGqQVK1YwShh4DII1AAAAJCu3b9/Wt99+a4Vry5cvd3RJAJCmRUVFKX369I4uA0iWCNYAAACQ7Ny5c0fffPONJk+erAIFCmjNmjWOLgkAAOARXLwAAAAAyU6mTJn05ptv6r333lNYWJguX77s6JIAAAAewYg1AAAAJFt3797V/fv35eHh4ehSAAAAHkGwBgAAAAAAANiBU0EBAAAAAAAAOxCsAQAAAAAAAHYgWAMAAAAAAADsQLAGAAAAAAAA2IFgDQAAAAAAALADwRoAAAAAAABgB4I1AAAAAAAAwA4EawAAAKnMwoULlTVrVkeXAQAAkOoRrAEAAKRAHTt2lM1mk81mk6urqwoXLqzRo0crNjbW0aUBAACkGc6OLgAAAAD2qVevnhYsWKDo6Gj9/PPP6tGjh1xcXJQrV64nfqyYmBi5uro+gyoBAABSL0asAQAApFBubm7y8vJSvnz51L17d9WuXVtr1qyx2jdu3KjixYsrc+bMqlevnq5cuWK1dezYUU2bNtXYsWPl7e2tokWLSpKWLFmiihUrKkuWLPLy8lLr1q0VHh5urXfjxg21adNGOXLkUIYMGVSkSBEtWLDAar948aLefPNNZc2aVdmyZdNrr72mP/7449n/MQAAAByAYA0AACCVyJAhg2JiYiRJd+/e1WeffaYlS5Zo+/btCgkJ0cCBAxP1DwgI0OnTp7V582atW7dOknT//n19/PHHOnz4sFavXq0//vhDHTt2tNYZNmyYTpw4ofXr1+vkyZOaNWuWnn/+eWtdf39/ZcmSRTt27NBvv/1mhXoJdQEAAKQmnAoKAACQwhljFBAQoI0bN6pXr16SHoRcs2fPVqFChSRJPXv21OjRoxOtlylTJs2dOzfRKaCdOnWy/l2wYEFNmzZNlSpV0u3bt5U5c2aFhISoXLlyqlixoiQpf/78Vv+vv/5a8fHxmjt3rmw2myRpwYIFypo1q3799VfVrVv3mWw/AACAozBiDQAAIIVat26dMmfOrPTp06t+/fp66623NHLkSElSxowZrVBNknLlypXolE5JKlWq1CPzqgUFBalx48bKmzevsmTJoldeeUWSFBISIknq3r27Vq5cqbJly2rw4MHatWuXte7hw4d19uxZZcmSRZkzZ1bmzJmVLVs2RUVF6dy5c8/iTwAAAOBQjFgDAABIoV599VXNmjVLrq6u8vb2lrPz/321c3FxSdTXZrPJGJNoWaZMmRLdv3Pnjvz9/eXv769ly5YpR44cCgkJkb+/v3UqZ/369XXhwgX9/PPP2rx5s2rVqqUePXros88+0+3bt1WhQgUtW7bskVpz5MjxtDYbAAAg2SBYAwAASKEyZcqkwoULP7XHO3XqlK5du6ZPPvlEPj4+kqT9+/c/0i9Hjhzq0KGDOnTooJdfflmDBg3SZ599pvLly+vrr79Wzpw55e7u/tTqAgAASK44FRQAAACSpLx588rV1VVffPGFzp8/rzVr1ujjjz9O1Gf48OH68ccfdfbsWR0/flzr1q1T8eLFJUlt2rTR888/r9dee007duxQcHCwfv31V/Xu3Vt//vmnIzYJAADgmSJYAwAAgKQHI9EWLlyob7/9Vr6+vvrkk0/02WefJerj6uqqoUOHqnTp0qpevbrSpUunlStXSnowr9v27duVN29eNWvWTMWLF1fnzp0VFRXFCDYAAJAq2cxfJ9sAAAAAAAAA8D8xYg0AAAAAAACwA8EaAAAAAAAAYAeCNQAAAAAAAMAOBGsAAAAAAACAHQjWAAAAAAAAADsQrAEAAAAAAAB2IFgDAAAAAAAA7ECwBgAAAAAAANiBYA0AAAAAAACwA8EaAAAAAAAAYAeCNQAAAAAAAMAO/w+5WuAdE8xM6AAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 1500x400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15, 4))\n",
    "ax = sns.barplot(data=plot_df_0, x=\"Phrase\", y=\"Count\")\n",
    "ax.set(ylabel='Count')\n",
    "plt.xticks(rotation=45)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "14a03712",
   "metadata": {},
   "source": [
    "### 10 most common phrases for 1 class label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "6bc40acf",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1500x400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15, 4))\n",
    "ax = sns.barplot(data=plot_df_1, x=\"Phrase\", y=\"Count\")\n",
    "ax.set(ylabel='Count')\n",
    "plt.xticks(rotation=45)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "323de154",
   "metadata": {},
   "source": [
    "## Load Pre-Processed Dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "024b1f1c",
   "metadata": {},
   "source": [
    "Split dataset to train, test and validation sets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "05368490",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "DatasetDict({\n",
       "    train: Dataset({\n",
       "        features: ['sentence', 'label'],\n",
       "        num_rows: 1306\n",
       "    })\n",
       "    validate: Dataset({\n",
       "        features: ['sentence', 'label'],\n",
       "        num_rows: 163\n",
       "    })\n",
       "    test: Dataset({\n",
       "        features: ['sentence', 'label'],\n",
       "        num_rows: 164\n",
       "    })\n",
       "})"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# dataset = load_dataset('csv', data_files={'train': './processed_df.csv'}, split=\"train\")\n",
    "# dataset = load_dataset('csv', data_files={'train': './brog_train1.csv', 'test': './brog_test1.csv'})\n",
    "data_dict = processed_df.to_dict(orient=\"list\")\n",
    "dataset = Dataset.from_dict(data_dict)\n",
    "train_remaining_split = dataset.train_test_split(test_size=0.20)\n",
    "test_validation_split = train_remaining_split['test'].train_test_split(test_size=0.50)\n",
    "\n",
    "# Now you have train, validation, and test sets\n",
    "train_set = train_remaining_split['train']\n",
    "validation_set = test_validation_split['train']\n",
    "test_set = test_validation_split['test']\n",
    "\n",
    "dataset = DatasetDict({\n",
    "    'train': train_set,\n",
    "    'validate': validation_set,\n",
    "    'test': test_set\n",
    "    })\n",
    "\n",
    "\n",
    "dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "314edeb1",
   "metadata": {},
   "source": [
    "<h2>Bio_ClinicalBERT</h2>\n",
    "\n",
    "<p>The Bio_ClinicalBERT model was trained on all notes from MIMIC III, a database containing electronic health records from ICU patients at the Beth Israel Hospital in Boston, MA. For more details on MIMIC, see here. All notes from the NOTEEVENTS table were included (~880M words).</p>\n",
    "<p>Each note in MIMIC was first split into sections using a rules-based section splitter (e.g. discharge summary notes were split into \"History of Present Illness\", \"Family History\", \"Brief Hospital Course\", etc. sections). Then each section was split into sentences using SciSpacy (en core sci md tokenizer).</p>\n",
    "\n",
    "<a href=\"https://huggingface.co/emilyalsentzer/Bio_ClinicalBERT\"><p>For More Information Click Here</p> <a/>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d207d44a",
   "metadata": {},
   "source": [
    "## Load & Save Pre-Trained Tokenizer"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "743dfaf1",
   "metadata": {},
   "source": [
    "If the tokenizer is not saved in the local directory, please uncomment the following cell code and execute it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "f0b97521",
   "metadata": {},
   "outputs": [],
   "source": [
    "# tokenizer = AutoTokenizer.from_pretrained('emilyalsentzer/Bio_ClinicalBERT')\n",
    "# tokenizer.save_pretrained(my_dir+\"save_tokenizer_bcb\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "219aad0d",
   "metadata": {},
   "source": [
    "### Load Pre-Trained Tokenizer locally"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "cd6ae6ab",
   "metadata": {},
   "outputs": [],
   "source": [
    "tokenizer = AutoTokenizer.from_pretrained('./save_tokenizer_bcb/')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "05033eb7",
   "metadata": {},
   "source": [
    "## Load & Save Pre-Trained Model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8d6d9ce4",
   "metadata": {},
   "source": [
    "If the model is not saved in the local directory, please uncomment the following cell code and execute it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "f8b01a16",
   "metadata": {},
   "outputs": [],
   "source": [
    "# pre_trained_model = AutoModelForSequenceClassification.from_pretrained(\"emilyalsentzer/Bio_ClinicalBERT\", num_labels=2)\n",
    "# pre_trained_model.save_pretrained(my_dir+\"save_model_bcb\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "765c7c4e",
   "metadata": {},
   "source": [
    "### Load Pre-Trained Model locally"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "fa3dd105",
   "metadata": {},
   "outputs": [],
   "source": [
    "pre_trained_model = AutoModelForSequenceClassification.from_pretrained(\"./save_model_bcb/\", num_labels=2)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3b3abcc5",
   "metadata": {},
   "source": [
    "## Before Fine-Tuning Result (Bio_ClinicalBERT):"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bfe8f82c",
   "metadata": {},
   "source": [
    "#### Check Max Length of sequences in dataset and Update max_length paraameter in tokenizer."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "af57f521",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Maximum sequence lengths for each split:\n",
      "{'train': 112, 'validate': 60, 'test': 52}\n"
     ]
    }
   ],
   "source": [
    "max_lengths = {}\n",
    "\n",
    "# Iterate over each split\n",
    "for split in dataset.keys():\n",
    "    max_length = 0\n",
    "    \n",
    "    # Iterate over examples in the split\n",
    "    for example in dataset[split]:\n",
    "        # Tokenize the text\n",
    "        tokenized_text = tokenizer(example['sentence'])\n",
    "        \n",
    "        # Get the length of the tokenized sequence\n",
    "        sequence_length = len(tokenized_text['input_ids'])\n",
    "        \n",
    "        # Update max_length if necessary\n",
    "        if sequence_length > max_length:\n",
    "            max_length = sequence_length\n",
    "    \n",
    "    # Store the max length for the split\n",
    "    max_lengths[split] = max_length\n",
    "\n",
    "print(\"Maximum sequence lengths for each split:\")\n",
    "print(max_lengths)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d5098a12",
   "metadata": {},
   "source": [
    "### Use confusion matrix to evaluate the accuracy of pre-trained model result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "7831b4c8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Set Length: 164\n",
      "Actual/True +ve: 70 \n",
      "Actual/True -ve: 94\n"
     ]
    }
   ],
   "source": [
    "print(f\"Test Set Length: {len(dataset['test']['label'])}\")\n",
    "true_pos = len([i for i in dataset['test']['label'] if i == 1])\n",
    "true_neg = len([i for i in dataset['test']['label'] if i == 0])\n",
    "print(f\"Actual/True +ve: {true_pos} \\nActual/True -ve: {true_neg}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0855676a",
   "metadata": {},
   "source": [
    "Confusion Matrix"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e2fef857",
   "metadata": {},
   "source": [
    "<img src=\"https://miro.medium.com/v2/resize:fit:1218/1*jMs1RmSwnYgR9CsBw-z1dw.png\" alt=\"Girl in a jacket\" width=\"500\" height=\"600\">"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c63cf75b",
   "metadata": {},
   "source": [
    "Based on the maximum sequence we you have obtained for each split:\n",
    "\n",
    "For the train split, the maximum sequence length is 112.\n",
    "For the validate split, the maximum sequence length is 55.\n",
    "For the test split, the maximum sequence length is 63.\n",
    "You should choose a max_length that is larger than the maximum sequence length across all splits to ensure that no sequences get truncated during tokenization. A common approach is to choose a max_length that is slightly larger than the longest sequence in your dataset.\n",
    "\n",
    "In this case, we could set the max_length to a value slightly higher than the maximum sequence length across all splits, such as 120 or 130, to provide some buffer room for longer sequences.\n",
    "\n",
    "So, a suitable choice for max_length in your tokenizer could be 130. This ensures that all sequences in your dataset will be accommodated without truncation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "08ab5b4f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+ve Pred: 120 \n",
      "-ve Pred: 44 \n",
      "\n",
      "\n",
      "Accuracy: 0.4634146341463415\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def tokenize_data(data):\n",
    "    tokenized_data = tokenizer(data[\"sentence\"], padding='max_length', max_length=130, truncation=True, return_tensors=\"pt\")\n",
    "    return tokenized_data\n",
    "\n",
    "test_data = tokenize_data(dataset['test'])\n",
    "# Define a function for evaluating the model\n",
    "def evaluate_model(model, test_data):\n",
    "    # Set the model to evaluation mode\n",
    "    model.eval()\n",
    "    \n",
    "    # Disable gradients\n",
    "    with torch.no_grad():\n",
    "        # Move input tensors to the device\n",
    "        input_ids = test_data['input_ids'].to(model.device)\n",
    "        attention_mask = test_data['attention_mask'].to(model.device)\n",
    "        \n",
    "        # Forward pass\n",
    "        outputs = model(input_ids, attention_mask=attention_mask)\n",
    "        logits = outputs.logits\n",
    "        \n",
    "        # Get predicted labels\n",
    "        predicted_labels = np.argmax(logits.cpu().numpy(), axis=1)\n",
    "    \n",
    "    return predicted_labels\n",
    "\n",
    "# Evaluate the model on test data\n",
    "predicted_labels = evaluate_model(pre_trained_model, test_data)\n",
    "\n",
    "# Total Positive and Negative Prediction made:\n",
    "pos_pred = len([i for i in predicted_labels if i == 1])\n",
    "neg_pred = len([i for i in predicted_labels if i == 0])\n",
    "print(f\"+ve Pred: {pos_pred} \\n-ve Pred: {neg_pred} \\n\\n\")\n",
    "\n",
    "# Get true labels\n",
    "true_labels = dataset['test']['label']\n",
    "\n",
    "# Compute confusion matrix\n",
    "conf_matrix = confusion_matrix(true_labels, list(predicted_labels))\n",
    "\n",
    "\n",
    "# Compute accuracy\n",
    "accuracy = accuracy_score(true_labels, list(predicted_labels))\n",
    "print(\"Accuracy:\", accuracy)\n",
    "\n",
    "cm_display = metrics.ConfusionMatrixDisplay(confusion_matrix = conf_matrix, display_labels = [False, True])\n",
    "cm_display.plot()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6998e685",
   "metadata": {},
   "source": [
    "Here you can see that before fine-tuning, we achieved a 50% accuracy on the test dataset."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1420863a",
   "metadata": {},
   "source": [
    "## Tokenization"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ebbfc04c",
   "metadata": {},
   "source": [
    "Now create a function that will tokenize the text. You should also truncate longer sequences in the text to be no longer than the model’s maximum input length."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "1e3f87a3",
   "metadata": {},
   "outputs": [],
   "source": [
    "def tokenize_function(examples):\n",
    "    return tokenizer(examples['sentence'], padding='max_length', max_length=130, truncation=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "064b48bb",
   "metadata": {},
   "source": [
    "Use Huggingface Datasets map function to apply the preprocessing function to the entire dataset. You can also set batched=True to apply the preprocessing function to multiple elements of the dataset at once for faster preprocessing:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "56c8cd37",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Map:   0%|          | 0/1306 [00:00<?, ? examples/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Map:   0%|          | 0/163 [00:00<?, ? examples/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Map:   0%|          | 0/164 [00:00<?, ? examples/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "DatasetDict({\n",
       "    train: Dataset({\n",
       "        features: ['sentence', 'label', 'input_ids', 'token_type_ids', 'attention_mask'],\n",
       "        num_rows: 1306\n",
       "    })\n",
       "    validate: Dataset({\n",
       "        features: ['sentence', 'label', 'input_ids', 'token_type_ids', 'attention_mask'],\n",
       "        num_rows: 163\n",
       "    })\n",
       "    test: Dataset({\n",
       "        features: ['sentence', 'label', 'input_ids', 'token_type_ids', 'attention_mask'],\n",
       "        num_rows: 164\n",
       "    })\n",
       "})"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tokenized_datasets = dataset.map(tokenize_function, batched=True)\n",
    "tokenized_datasets"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "aa17e70b",
   "metadata": {},
   "source": [
    "Now loads the accuracy evaluation metric, defines a function to compute metrics based on model predictions and true labels, and uses the loaded accuracy metric to compute the accuracy of the model's predictions."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "9e50fa1b",
   "metadata": {},
   "outputs": [],
   "source": [
    "metric = evaluate.load(\"accuracy\")\n",
    "def compute_metrics(eval_pred):\n",
    "    logits, labels = eval_pred\n",
    "    predictions = np.argmax(logits, axis=-1)\n",
    "    return metric.compute(predictions=predictions, references=labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6471dfa9",
   "metadata": {},
   "source": [
    "1) Define training hyperparameters in TrainingArguments.\n",
    "2) Pass the training arguments to Trainer along with the model, dataset and tokenizer.\n",
    "3) Call train() to fine-tune your model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "c11d9dce",
   "metadata": {},
   "outputs": [],
   "source": [
    "training_args = TrainingArguments(\n",
    "    output_dir=\"./fine_tuned_model_bcb/\",\n",
    "    per_device_train_batch_size=4,\n",
    "    per_device_eval_batch_size=4,\n",
    "    gradient_accumulation_steps=1,\n",
    "    learning_rate=1e-5,\n",
    "    warmup_steps=500,\n",
    "    max_steps=1000,\n",
    "    evaluation_strategy=\"steps\",\n",
    "    save_steps=500,\n",
    "    eval_steps=5,\n",
    "    logging_steps=5,\n",
    "    report_to=[\"tensorboard\"],\n",
    "    load_best_model_at_end=True,\n",
    "    greater_is_better=False,\n",
    "    push_to_hub=False,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "973dff98",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Detected kernel version 4.15.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.\n"
     ]
    }
   ],
   "source": [
    "trainer = Trainer(\n",
    "    model=pre_trained_model,\n",
    "    args=training_args,\n",
    "    train_dataset=tokenized_datasets['train'],\n",
    "    eval_dataset=tokenized_datasets['validate'],\n",
    "    compute_metrics=compute_metrics,\n",
    "    tokenizer=tokenizer,\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "098c030c",
   "metadata": {},
   "source": [
    "Before training, navigate to the directory where checkpoints will be saved, which in my case is <b>/fine_tuned_model_bcb</b>, and execute the following command: <b>tensorboard --logdir=runs</b>. \n",
    "This command will serve TensorBoard on localhost at http://localhost:6006/."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a9d9f35c",
   "metadata": {},
   "source": [
    "## Start Training"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bd0862ae",
   "metadata": {},
   "source": [
    "To fine-tune the model on our dataset, we just have to call the train() method of our Trainer:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "id": "bd60cf35",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "    <div>\n",
       "      \n",
       "      <progress value='1000' max='1000' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      [1000/1000 04:56, Epoch 3/4]\n",
       "    </div>\n",
       "    <table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       " <tr style=\"text-align: left;\">\n",
       "      <th>Step</th>\n",
       "      <th>Training Loss</th>\n",
       "      <th>Validation Loss</th>\n",
       "      <th>Accuracy</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.744200</td>\n",
       "      <td>0.693007</td>\n",
       "      <td>0.490798</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.697900</td>\n",
       "      <td>0.692967</td>\n",
       "      <td>0.490798</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.711000</td>\n",
       "      <td>0.692786</td>\n",
       "      <td>0.490798</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.755800</td>\n",
       "      <td>0.692275</td>\n",
       "      <td>0.496933</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.718900</td>\n",
       "      <td>0.691386</td>\n",
       "      <td>0.503067</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.697300</td>\n",
       "      <td>0.690529</td>\n",
       "      <td>0.515337</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.640400</td>\n",
       "      <td>0.690097</td>\n",
       "      <td>0.503067</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.677500</td>\n",
       "      <td>0.689422</td>\n",
       "      <td>0.503067</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.725700</td>\n",
       "      <td>0.688619</td>\n",
       "      <td>0.496933</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>50</td>\n",
       "      <td>0.735800</td>\n",
       "      <td>0.687395</td>\n",
       "      <td>0.509202</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>55</td>\n",
       "      <td>0.639900</td>\n",
       "      <td>0.685551</td>\n",
       "      <td>0.533742</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>60</td>\n",
       "      <td>0.700400</td>\n",
       "      <td>0.683457</td>\n",
       "      <td>0.539877</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>65</td>\n",
       "      <td>0.728300</td>\n",
       "      <td>0.681561</td>\n",
       "      <td>0.564417</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>70</td>\n",
       "      <td>0.680600</td>\n",
       "      <td>0.680150</td>\n",
       "      <td>0.570552</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>75</td>\n",
       "      <td>0.743100</td>\n",
       "      <td>0.679161</td>\n",
       "      <td>0.570552</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>80</td>\n",
       "      <td>0.719700</td>\n",
       "      <td>0.678400</td>\n",
       "      <td>0.570552</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>85</td>\n",
       "      <td>0.706300</td>\n",
       "      <td>0.677039</td>\n",
       "      <td>0.576687</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>90</td>\n",
       "      <td>0.694300</td>\n",
       "      <td>0.674931</td>\n",
       "      <td>0.588957</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>95</td>\n",
       "      <td>0.729000</td>\n",
       "      <td>0.672461</td>\n",
       "      <td>0.607362</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>100</td>\n",
       "      <td>0.658900</td>\n",
       "      <td>0.670780</td>\n",
       "      <td>0.613497</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>105</td>\n",
       "      <td>0.690400</td>\n",
       "      <td>0.669362</td>\n",
       "      <td>0.644172</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>110</td>\n",
       "      <td>0.671200</td>\n",
       "      <td>0.667144</td>\n",
       "      <td>0.674847</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>115</td>\n",
       "      <td>0.655100</td>\n",
       "      <td>0.664970</td>\n",
       "      <td>0.631902</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>120</td>\n",
       "      <td>0.705500</td>\n",
       "      <td>0.662833</td>\n",
       "      <td>0.644172</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>125</td>\n",
       "      <td>0.646000</td>\n",
       "      <td>0.660700</td>\n",
       "      <td>0.650307</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>130</td>\n",
       "      <td>0.679300</td>\n",
       "      <td>0.658474</td>\n",
       "      <td>0.638037</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>135</td>\n",
       "      <td>0.705000</td>\n",
       "      <td>0.656345</td>\n",
       "      <td>0.619632</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>140</td>\n",
       "      <td>0.722500</td>\n",
       "      <td>0.654594</td>\n",
       "      <td>0.631902</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>145</td>\n",
       "      <td>0.623300</td>\n",
       "      <td>0.652951</td>\n",
       "      <td>0.631902</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>150</td>\n",
       "      <td>0.660500</td>\n",
       "      <td>0.650990</td>\n",
       "      <td>0.625767</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>155</td>\n",
       "      <td>0.660300</td>\n",
       "      <td>0.648605</td>\n",
       "      <td>0.638037</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>160</td>\n",
       "      <td>0.648600</td>\n",
       "      <td>0.646236</td>\n",
       "      <td>0.644172</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>165</td>\n",
       "      <td>0.671600</td>\n",
       "      <td>0.644284</td>\n",
       "      <td>0.644172</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>170</td>\n",
       "      <td>0.655000</td>\n",
       "      <td>0.644009</td>\n",
       "      <td>0.662577</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>175</td>\n",
       "      <td>0.664800</td>\n",
       "      <td>0.643787</td>\n",
       "      <td>0.687117</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>180</td>\n",
       "      <td>0.654600</td>\n",
       "      <td>0.641812</td>\n",
       "      <td>0.742331</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>185</td>\n",
       "      <td>0.660500</td>\n",
       "      <td>0.638974</td>\n",
       "      <td>0.736196</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>190</td>\n",
       "      <td>0.606500</td>\n",
       "      <td>0.636357</td>\n",
       "      <td>0.760736</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>195</td>\n",
       "      <td>0.619200</td>\n",
       "      <td>0.630388</td>\n",
       "      <td>0.711656</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>200</td>\n",
       "      <td>0.614600</td>\n",
       "      <td>0.627909</td>\n",
       "      <td>0.730061</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>205</td>\n",
       "      <td>0.674300</td>\n",
       "      <td>0.628246</td>\n",
       "      <td>0.736196</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>210</td>\n",
       "      <td>0.647200</td>\n",
       "      <td>0.628909</td>\n",
       "      <td>0.730061</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>215</td>\n",
       "      <td>0.623900</td>\n",
       "      <td>0.618264</td>\n",
       "      <td>0.748466</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>220</td>\n",
       "      <td>0.595300</td>\n",
       "      <td>0.606238</td>\n",
       "      <td>0.736196</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>225</td>\n",
       "      <td>0.657300</td>\n",
       "      <td>0.603433</td>\n",
       "      <td>0.748466</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>230</td>\n",
       "      <td>0.641400</td>\n",
       "      <td>0.602268</td>\n",
       "      <td>0.736196</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>235</td>\n",
       "      <td>0.638700</td>\n",
       "      <td>0.600379</td>\n",
       "      <td>0.742331</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>240</td>\n",
       "      <td>0.658600</td>\n",
       "      <td>0.594677</td>\n",
       "      <td>0.736196</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>245</td>\n",
       "      <td>0.551800</td>\n",
       "      <td>0.588871</td>\n",
       "      <td>0.742331</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>250</td>\n",
       "      <td>0.627800</td>\n",
       "      <td>0.582078</td>\n",
       "      <td>0.748466</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>255</td>\n",
       "      <td>0.590700</td>\n",
       "      <td>0.574285</td>\n",
       "      <td>0.754601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>260</td>\n",
       "      <td>0.595500</td>\n",
       "      <td>0.566926</td>\n",
       "      <td>0.773006</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>265</td>\n",
       "      <td>0.520400</td>\n",
       "      <td>0.558995</td>\n",
       "      <td>0.773006</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>270</td>\n",
       "      <td>0.595400</td>\n",
       "      <td>0.551564</td>\n",
       "      <td>0.779141</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>275</td>\n",
       "      <td>0.516400</td>\n",
       "      <td>0.541577</td>\n",
       "      <td>0.779141</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>280</td>\n",
       "      <td>0.447700</td>\n",
       "      <td>0.529196</td>\n",
       "      <td>0.791411</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>285</td>\n",
       "      <td>0.561100</td>\n",
       "      <td>0.521000</td>\n",
       "      <td>0.785276</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>290</td>\n",
       "      <td>0.529200</td>\n",
       "      <td>0.507430</td>\n",
       "      <td>0.779141</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>295</td>\n",
       "      <td>0.448800</td>\n",
       "      <td>0.495192</td>\n",
       "      <td>0.791411</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>300</td>\n",
       "      <td>0.493600</td>\n",
       "      <td>0.492257</td>\n",
       "      <td>0.791411</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>305</td>\n",
       "      <td>0.382700</td>\n",
       "      <td>0.475768</td>\n",
       "      <td>0.797546</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>310</td>\n",
       "      <td>0.549600</td>\n",
       "      <td>0.467008</td>\n",
       "      <td>0.803681</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>315</td>\n",
       "      <td>0.491300</td>\n",
       "      <td>0.453660</td>\n",
       "      <td>0.815951</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>320</td>\n",
       "      <td>0.476400</td>\n",
       "      <td>0.432001</td>\n",
       "      <td>0.828221</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>325</td>\n",
       "      <td>0.349200</td>\n",
       "      <td>0.423475</td>\n",
       "      <td>0.809816</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>330</td>\n",
       "      <td>0.293900</td>\n",
       "      <td>0.452242</td>\n",
       "      <td>0.809816</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>335</td>\n",
       "      <td>0.408500</td>\n",
       "      <td>0.463348</td>\n",
       "      <td>0.797546</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>340</td>\n",
       "      <td>0.603600</td>\n",
       "      <td>0.386835</td>\n",
       "      <td>0.840491</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>345</td>\n",
       "      <td>0.328700</td>\n",
       "      <td>0.390751</td>\n",
       "      <td>0.834356</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>350</td>\n",
       "      <td>0.495600</td>\n",
       "      <td>0.430570</td>\n",
       "      <td>0.815951</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>355</td>\n",
       "      <td>0.558800</td>\n",
       "      <td>0.383178</td>\n",
       "      <td>0.846626</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>360</td>\n",
       "      <td>0.195000</td>\n",
       "      <td>0.371515</td>\n",
       "      <td>0.852761</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>365</td>\n",
       "      <td>0.252400</td>\n",
       "      <td>0.353394</td>\n",
       "      <td>0.877301</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>370</td>\n",
       "      <td>0.277500</td>\n",
       "      <td>0.355028</td>\n",
       "      <td>0.858896</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>375</td>\n",
       "      <td>0.662000</td>\n",
       "      <td>0.359918</td>\n",
       "      <td>0.852761</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>380</td>\n",
       "      <td>0.341600</td>\n",
       "      <td>0.366626</td>\n",
       "      <td>0.840491</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>385</td>\n",
       "      <td>0.397000</td>\n",
       "      <td>0.352394</td>\n",
       "      <td>0.852761</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>390</td>\n",
       "      <td>0.458000</td>\n",
       "      <td>0.347865</td>\n",
       "      <td>0.871166</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>395</td>\n",
       "      <td>0.332600</td>\n",
       "      <td>0.365231</td>\n",
       "      <td>0.852761</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>400</td>\n",
       "      <td>0.261200</td>\n",
       "      <td>0.392296</td>\n",
       "      <td>0.840491</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>405</td>\n",
       "      <td>0.160700</td>\n",
       "      <td>0.371293</td>\n",
       "      <td>0.852761</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>410</td>\n",
       "      <td>0.282400</td>\n",
       "      <td>0.362860</td>\n",
       "      <td>0.889571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>415</td>\n",
       "      <td>0.098100</td>\n",
       "      <td>0.443543</td>\n",
       "      <td>0.822086</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>420</td>\n",
       "      <td>0.643200</td>\n",
       "      <td>0.496372</td>\n",
       "      <td>0.803681</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>425</td>\n",
       "      <td>0.683900</td>\n",
       "      <td>0.381552</td>\n",
       "      <td>0.883436</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>430</td>\n",
       "      <td>0.078100</td>\n",
       "      <td>0.359784</td>\n",
       "      <td>0.871166</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>435</td>\n",
       "      <td>0.410100</td>\n",
       "      <td>0.367683</td>\n",
       "      <td>0.871166</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>440</td>\n",
       "      <td>0.393400</td>\n",
       "      <td>0.404336</td>\n",
       "      <td>0.858896</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>445</td>\n",
       "      <td>0.055900</td>\n",
       "      <td>0.420995</td>\n",
       "      <td>0.852761</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>450</td>\n",
       "      <td>0.220800</td>\n",
       "      <td>0.380446</td>\n",
       "      <td>0.865031</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>455</td>\n",
       "      <td>0.424200</td>\n",
       "      <td>0.391700</td>\n",
       "      <td>0.871166</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>460</td>\n",
       "      <td>0.469200</td>\n",
       "      <td>0.405871</td>\n",
       "      <td>0.871166</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>465</td>\n",
       "      <td>0.659700</td>\n",
       "      <td>0.398162</td>\n",
       "      <td>0.877301</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>470</td>\n",
       "      <td>0.426300</td>\n",
       "      <td>0.380889</td>\n",
       "      <td>0.877301</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>475</td>\n",
       "      <td>0.312000</td>\n",
       "      <td>0.384659</td>\n",
       "      <td>0.877301</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>480</td>\n",
       "      <td>0.230500</td>\n",
       "      <td>0.390052</td>\n",
       "      <td>0.871166</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>485</td>\n",
       "      <td>0.468100</td>\n",
       "      <td>0.402907</td>\n",
       "      <td>0.877301</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>490</td>\n",
       "      <td>0.218800</td>\n",
       "      <td>0.414927</td>\n",
       "      <td>0.877301</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>495</td>\n",
       "      <td>0.122700</td>\n",
       "      <td>0.419077</td>\n",
       "      <td>0.858896</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>500</td>\n",
       "      <td>0.859600</td>\n",
       "      <td>0.379329</td>\n",
       "      <td>0.889571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>505</td>\n",
       "      <td>0.327000</td>\n",
       "      <td>0.370161</td>\n",
       "      <td>0.889571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>510</td>\n",
       "      <td>0.101900</td>\n",
       "      <td>0.385530</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>515</td>\n",
       "      <td>0.417200</td>\n",
       "      <td>0.423369</td>\n",
       "      <td>0.877301</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>520</td>\n",
       "      <td>0.440400</td>\n",
       "      <td>0.584981</td>\n",
       "      <td>0.803681</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>525</td>\n",
       "      <td>1.039100</td>\n",
       "      <td>0.628674</td>\n",
       "      <td>0.791411</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>530</td>\n",
       "      <td>0.467000</td>\n",
       "      <td>0.349918</td>\n",
       "      <td>0.889571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>535</td>\n",
       "      <td>0.793400</td>\n",
       "      <td>0.277590</td>\n",
       "      <td>0.901840</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>540</td>\n",
       "      <td>0.792800</td>\n",
       "      <td>0.275456</td>\n",
       "      <td>0.920245</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>545</td>\n",
       "      <td>0.779400</td>\n",
       "      <td>0.268480</td>\n",
       "      <td>0.901840</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>550</td>\n",
       "      <td>0.360800</td>\n",
       "      <td>0.297239</td>\n",
       "      <td>0.907975</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>555</td>\n",
       "      <td>0.047900</td>\n",
       "      <td>0.275004</td>\n",
       "      <td>0.901840</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>560</td>\n",
       "      <td>0.524400</td>\n",
       "      <td>0.292898</td>\n",
       "      <td>0.907975</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>565</td>\n",
       "      <td>0.734900</td>\n",
       "      <td>0.300608</td>\n",
       "      <td>0.920245</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>570</td>\n",
       "      <td>0.321600</td>\n",
       "      <td>0.288637</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>575</td>\n",
       "      <td>0.274200</td>\n",
       "      <td>0.315449</td>\n",
       "      <td>0.889571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>580</td>\n",
       "      <td>0.472100</td>\n",
       "      <td>0.336927</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>585</td>\n",
       "      <td>0.185200</td>\n",
       "      <td>0.338075</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>590</td>\n",
       "      <td>0.169800</td>\n",
       "      <td>0.342598</td>\n",
       "      <td>0.883436</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>595</td>\n",
       "      <td>0.240400</td>\n",
       "      <td>0.356347</td>\n",
       "      <td>0.871166</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>600</td>\n",
       "      <td>0.182500</td>\n",
       "      <td>0.392413</td>\n",
       "      <td>0.883436</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>605</td>\n",
       "      <td>0.300500</td>\n",
       "      <td>0.422733</td>\n",
       "      <td>0.883436</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>610</td>\n",
       "      <td>0.306400</td>\n",
       "      <td>0.444664</td>\n",
       "      <td>0.883436</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>615</td>\n",
       "      <td>0.205300</td>\n",
       "      <td>0.386557</td>\n",
       "      <td>0.901840</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>620</td>\n",
       "      <td>0.018200</td>\n",
       "      <td>0.382426</td>\n",
       "      <td>0.901840</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>625</td>\n",
       "      <td>0.030200</td>\n",
       "      <td>0.302081</td>\n",
       "      <td>0.926380</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>630</td>\n",
       "      <td>0.685200</td>\n",
       "      <td>0.262986</td>\n",
       "      <td>0.920245</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>635</td>\n",
       "      <td>0.038800</td>\n",
       "      <td>0.288522</td>\n",
       "      <td>0.926380</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>640</td>\n",
       "      <td>0.032500</td>\n",
       "      <td>0.308538</td>\n",
       "      <td>0.907975</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>645</td>\n",
       "      <td>1.096300</td>\n",
       "      <td>0.296322</td>\n",
       "      <td>0.914110</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>650</td>\n",
       "      <td>0.457300</td>\n",
       "      <td>0.251022</td>\n",
       "      <td>0.944785</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>655</td>\n",
       "      <td>0.172400</td>\n",
       "      <td>0.268562</td>\n",
       "      <td>0.926380</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>660</td>\n",
       "      <td>0.264400</td>\n",
       "      <td>0.284824</td>\n",
       "      <td>0.920245</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>665</td>\n",
       "      <td>0.097000</td>\n",
       "      <td>0.297774</td>\n",
       "      <td>0.920245</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>670</td>\n",
       "      <td>0.272700</td>\n",
       "      <td>0.345816</td>\n",
       "      <td>0.920245</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>675</td>\n",
       "      <td>0.014100</td>\n",
       "      <td>0.371806</td>\n",
       "      <td>0.920245</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>680</td>\n",
       "      <td>0.004700</td>\n",
       "      <td>0.381311</td>\n",
       "      <td>0.914110</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>685</td>\n",
       "      <td>0.110300</td>\n",
       "      <td>0.351944</td>\n",
       "      <td>0.920245</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>690</td>\n",
       "      <td>0.026600</td>\n",
       "      <td>0.282707</td>\n",
       "      <td>0.914110</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>695</td>\n",
       "      <td>0.244800</td>\n",
       "      <td>0.278290</td>\n",
       "      <td>0.914110</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>700</td>\n",
       "      <td>0.121400</td>\n",
       "      <td>0.288564</td>\n",
       "      <td>0.914110</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>705</td>\n",
       "      <td>0.459000</td>\n",
       "      <td>0.319986</td>\n",
       "      <td>0.914110</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>710</td>\n",
       "      <td>0.011900</td>\n",
       "      <td>0.330732</td>\n",
       "      <td>0.926380</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>715</td>\n",
       "      <td>0.224400</td>\n",
       "      <td>0.332564</td>\n",
       "      <td>0.926380</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>720</td>\n",
       "      <td>0.080000</td>\n",
       "      <td>0.307664</td>\n",
       "      <td>0.920245</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>725</td>\n",
       "      <td>0.004400</td>\n",
       "      <td>0.299673</td>\n",
       "      <td>0.907975</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>730</td>\n",
       "      <td>0.300000</td>\n",
       "      <td>0.307511</td>\n",
       "      <td>0.914110</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>735</td>\n",
       "      <td>0.006700</td>\n",
       "      <td>0.326433</td>\n",
       "      <td>0.907975</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>740</td>\n",
       "      <td>0.010000</td>\n",
       "      <td>0.398898</td>\n",
       "      <td>0.920245</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>745</td>\n",
       "      <td>0.083300</td>\n",
       "      <td>0.447397</td>\n",
       "      <td>0.914110</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>750</td>\n",
       "      <td>0.242700</td>\n",
       "      <td>0.461221</td>\n",
       "      <td>0.914110</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>755</td>\n",
       "      <td>0.002600</td>\n",
       "      <td>0.459562</td>\n",
       "      <td>0.914110</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>760</td>\n",
       "      <td>0.003300</td>\n",
       "      <td>0.462796</td>\n",
       "      <td>0.914110</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>765</td>\n",
       "      <td>0.547200</td>\n",
       "      <td>0.433146</td>\n",
       "      <td>0.914110</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>770</td>\n",
       "      <td>0.288800</td>\n",
       "      <td>0.393483</td>\n",
       "      <td>0.914110</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>775</td>\n",
       "      <td>0.761500</td>\n",
       "      <td>0.360419</td>\n",
       "      <td>0.901840</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>780</td>\n",
       "      <td>0.051400</td>\n",
       "      <td>0.382489</td>\n",
       "      <td>0.907975</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>785</td>\n",
       "      <td>0.012900</td>\n",
       "      <td>0.418005</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>790</td>\n",
       "      <td>0.123700</td>\n",
       "      <td>0.403719</td>\n",
       "      <td>0.901840</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>795</td>\n",
       "      <td>0.565600</td>\n",
       "      <td>0.385832</td>\n",
       "      <td>0.907975</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>800</td>\n",
       "      <td>0.286600</td>\n",
       "      <td>0.418646</td>\n",
       "      <td>0.914110</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>805</td>\n",
       "      <td>0.271700</td>\n",
       "      <td>0.482516</td>\n",
       "      <td>0.907975</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>810</td>\n",
       "      <td>0.954300</td>\n",
       "      <td>0.478548</td>\n",
       "      <td>0.907975</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>815</td>\n",
       "      <td>0.082400</td>\n",
       "      <td>0.453762</td>\n",
       "      <td>0.901840</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>820</td>\n",
       "      <td>0.006000</td>\n",
       "      <td>0.459372</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>825</td>\n",
       "      <td>0.528200</td>\n",
       "      <td>0.473601</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>830</td>\n",
       "      <td>0.826400</td>\n",
       "      <td>0.455546</td>\n",
       "      <td>0.889571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>835</td>\n",
       "      <td>0.152100</td>\n",
       "      <td>0.444457</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>840</td>\n",
       "      <td>0.186200</td>\n",
       "      <td>0.446537</td>\n",
       "      <td>0.889571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>845</td>\n",
       "      <td>0.422500</td>\n",
       "      <td>0.453951</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>850</td>\n",
       "      <td>0.692400</td>\n",
       "      <td>0.466219</td>\n",
       "      <td>0.889571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>855</td>\n",
       "      <td>0.375600</td>\n",
       "      <td>0.463933</td>\n",
       "      <td>0.883436</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>860</td>\n",
       "      <td>0.007500</td>\n",
       "      <td>0.455944</td>\n",
       "      <td>0.877301</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>865</td>\n",
       "      <td>0.647600</td>\n",
       "      <td>0.459176</td>\n",
       "      <td>0.889571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>870</td>\n",
       "      <td>0.479700</td>\n",
       "      <td>0.458758</td>\n",
       "      <td>0.883436</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>875</td>\n",
       "      <td>0.331000</td>\n",
       "      <td>0.446983</td>\n",
       "      <td>0.889571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>880</td>\n",
       "      <td>0.194700</td>\n",
       "      <td>0.441735</td>\n",
       "      <td>0.883436</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>885</td>\n",
       "      <td>0.025300</td>\n",
       "      <td>0.436627</td>\n",
       "      <td>0.889571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>890</td>\n",
       "      <td>0.238900</td>\n",
       "      <td>0.434400</td>\n",
       "      <td>0.889571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>895</td>\n",
       "      <td>0.408200</td>\n",
       "      <td>0.437151</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>900</td>\n",
       "      <td>0.005200</td>\n",
       "      <td>0.438662</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>905</td>\n",
       "      <td>0.005600</td>\n",
       "      <td>0.440164</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>910</td>\n",
       "      <td>0.551000</td>\n",
       "      <td>0.442990</td>\n",
       "      <td>0.889571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>915</td>\n",
       "      <td>0.374800</td>\n",
       "      <td>0.438991</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>920</td>\n",
       "      <td>0.354300</td>\n",
       "      <td>0.431233</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>925</td>\n",
       "      <td>0.003500</td>\n",
       "      <td>0.425595</td>\n",
       "      <td>0.889571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>930</td>\n",
       "      <td>0.008900</td>\n",
       "      <td>0.419712</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>935</td>\n",
       "      <td>0.030600</td>\n",
       "      <td>0.415346</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>940</td>\n",
       "      <td>0.397400</td>\n",
       "      <td>0.412214</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>945</td>\n",
       "      <td>0.003100</td>\n",
       "      <td>0.408848</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>950</td>\n",
       "      <td>0.026800</td>\n",
       "      <td>0.406165</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>955</td>\n",
       "      <td>0.883200</td>\n",
       "      <td>0.402410</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>960</td>\n",
       "      <td>0.505800</td>\n",
       "      <td>0.397569</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>965</td>\n",
       "      <td>0.092200</td>\n",
       "      <td>0.394934</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>970</td>\n",
       "      <td>0.208500</td>\n",
       "      <td>0.395384</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>975</td>\n",
       "      <td>0.003500</td>\n",
       "      <td>0.396456</td>\n",
       "      <td>0.901840</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>980</td>\n",
       "      <td>0.003900</td>\n",
       "      <td>0.397296</td>\n",
       "      <td>0.901840</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>985</td>\n",
       "      <td>0.002300</td>\n",
       "      <td>0.398135</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>990</td>\n",
       "      <td>0.288400</td>\n",
       "      <td>0.398815</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>995</td>\n",
       "      <td>0.037600</td>\n",
       "      <td>0.399045</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1000</td>\n",
       "      <td>0.286100</td>\n",
       "      <td>0.398943</td>\n",
       "      <td>0.895706</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table><p>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "TrainOutput(global_step=1000, training_loss=0.4120384232858196, metrics={'train_runtime': 297.1089, 'train_samples_per_second': 13.463, 'train_steps_per_second': 3.366, 'total_flos': 266821332351600.0, 'train_loss': 0.4120384232858196, 'epoch': 3.06})"
      ]
     },
     "execution_count": 70,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "trainer.train()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c8b097ea",
   "metadata": {},
   "source": [
    "Save the fine-tuned model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "743786ab",
   "metadata": {},
   "outputs": [],
   "source": [
    "# trainer.save_model()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "id": "6328a3f4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "predictions = trainer.predict(tokenized_datasets['test'])\n",
    "predictions = list(np.argmax(predictions.predictions, axis=-1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "id": "fbe92eb5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Get true labels\n",
    "true_labels = dataset['test']['label']\n",
    "\n",
    "# Compute confusion matrix\n",
    "conf_matrix = confusion_matrix(true_labels, predictions)\n",
    "cm_display = metrics.ConfusionMatrixDisplay(confusion_matrix = conf_matrix, display_labels = [False, True])\n",
    "cm_display.plot()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "885fcdd5",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9085365853658537"
      ]
     },
     "execution_count": 73,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy = accuracy_score(true_labels, predictions)\n",
    "accuracy"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cf950627",
   "metadata": {},
   "source": [
    "Here you can see that after fine-tuning, we achieved a 90% accuracy on the test dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "950e6cda",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e03d76d9",
   "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": "3.7.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}