notebooks/roberta-finetuned-50split-dataset.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/mc-cat-tty/PlaceRank/blob/BERTembeddings/finetuneBERT.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "48jg1LDR-E-r"
      },
      "source": [
        "### finetuning BERT for sentiment analysis"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 254,
      "metadata": {
        "id": "4KivTNCL-E-t"
      },
      "outputs": [],
      "source": [
        "import os\n",
        "import re\n",
        "import string\n",
        "from tqdm import tqdm\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "import nltk\n",
        "from nltk.corpus import stopwords\n",
        "from nltk.tokenize import word_tokenize\n",
        "\n",
        "%matplotlib inline"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 255,
      "metadata": {
        "id": "rOgyVa3_Ir02",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "3a69fd79-c1f4-4471-d38d-ba2f77f3b2ad"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive', force_remount=True)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 257,
      "metadata": {
        "id": "VT-P7fU7-E-u"
      },
      "outputs": [],
      "source": [
        "file1_path = '/content/drive/MyDrive/inputs/nyc_listings.csv'\n",
        "file2_path = '/content/drive/MyDrive/inputs/nyc_reviews.csv'\n",
        "selected1_cols = ['id', 'review_scores_rating']\n",
        "selected2_cols = ['listing_id', 'comments']\n",
        "df1 = pd.read_csv(file1_path, usecols=selected1_cols)\n",
        "df2 = pd.read_csv(file2_path, usecols=selected2_cols)\n",
        "df1.rename(columns={'id': 'listing_id'}, inplace=True)\n",
        "df = pd.merge(df1, df2)\n",
        "df.rename(columns={'listing_id': 'id'}, inplace=True)\n",
        "df.rename(columns={'review_scores_rating': 'rating'}, inplace=True)\n",
        "df.rename(columns={'comments': 'review'}, inplace=True)\n",
        "new_cols = ['id', 'review', 'rating']\n",
        "dfs = df[new_cols]\n",
        "dfs['rating'] = dfs['rating'].round().astype(int)\n",
        "\n",
        "\n",
        "positive = [4, 5]\n",
        "negative = [1, 2, 3]\n",
        "\n",
        "def sentiment(rating):\n",
        "  if rating in positive:\n",
        "    return 1\n",
        "  else:\n",
        "    return 0\n",
        "\n",
        "dfs['target'] = dfs['rating'].apply(sentiment)\n",
        "dfs.rename(columns={'target': 'label'}, inplace=True)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 258,
      "metadata": {
        "id": "pQb6VeE8-E-v"
      },
      "outputs": [],
      "source": [
        "dfs.drop(['rating'], inplace=True, axis=1)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 259,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "id": "9eL5AKOz-E-v",
        "outputId": "9ee2aa1b-c8ac-48a6-ab0c-2e7e7d384efc"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "         id                                             review  label\n",
              "0  21935608  Mike was very responsive and accommodating. Th...      1\n",
              "1  21935608  I couldn’t be more satisfied with my stay at M...      1\n",
              "2  21935608  Mike's space was absolutely lovely and in a wo...      1\n",
              "3  21935608  The apartment was just exactly what I was look...      1\n",
              "4  21935608  Mike and Zoe were great hosts. Mike wrote a ve...      1"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-772b1935-acbf-479f-905d-f400cfad061a\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>id</th>\n",
              "      <th>review</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>21935608</td>\n",
              "      <td>Mike was very responsive and accommodating. Th...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>21935608</td>\n",
              "      <td>I couldn’t be more satisfied with my stay at M...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>21935608</td>\n",
              "      <td>Mike's space was absolutely lovely and in a wo...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>21935608</td>\n",
              "      <td>The apartment was just exactly what I was look...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>21935608</td>\n",
              "      <td>Mike and Zoe were great hosts. Mike wrote a ve...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-772b1935-acbf-479f-905d-f400cfad061a')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-772b1935-acbf-479f-905d-f400cfad061a button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-772b1935-acbf-479f-905d-f400cfad061a');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-80c98357-6128-4366-a9a6-7a0363083046\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-80c98357-6128-4366-a9a6-7a0363083046')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-80c98357-6128-4366-a9a6-7a0363083046 button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "    </div>\n",
              "  </div>\n"
            ]
          },
          "metadata": {},
          "execution_count": 259
        }
      ],
      "source": [
        "dfs.head()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 260,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "yxzxPw9V-E-v",
        "outputId": "9f8e90a6-e4ef-4350-a6a3-2492ea849d77"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "1    999846\n",
              "0      1449\n",
              "Name: label, dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 260
        }
      ],
      "source": [
        "balance_counts = dfs.groupby('label')['label'].agg('count').values\n",
        "dfs['label'].value_counts()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 261,
      "metadata": {
        "id": "HSfyiq2H-E-w"
      },
      "outputs": [],
      "source": [
        "dfs.review = dfs.review.astype(str)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mMx9Yidq-E-w"
      },
      "source": [
        "### building a test dataset"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 264,
      "metadata": {
        "id": "dFxSKtU3-E-w"
      },
      "outputs": [],
      "source": [
        "# negative sentiment\n",
        "neg_dfs = dfs[(dfs[\"label\"] == 0)]\n",
        "\n",
        "# drop the label\n",
        "# neg_dfs.drop(['label'], inplace=True, axis=1)\n",
        "\n",
        "new_dfs = dfs[(dfs[\"label\"] == 1)].sample(1500)\n",
        "positive_dfs_test = dfs[(dfs[\"label\"] == 1)].sample(105)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 263,
      "metadata": {
        "id": "LnASp28X-E-w"
      },
      "outputs": [],
      "source": [
        "# new_dfs.drop(['label'], inplace=True, axis=1)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 265,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "id": "WeWZQNrE-E-x",
        "outputId": "1d31fd5c-84e1-41f5-e177-b92ededf9846"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "           id                                             review  label\n",
              "321  19531174  Most horrible Airbnb experience I ever had, we...      0\n",
              "363  22306777  This location is ideal for fans of Brooklyn be...      0\n",
              "364  22306777  Bartolomeo is a nice and friendly host, the ap...      0\n",
              "454  28224871  I needed a place near NY Presbyterian Hospital...      0\n",
              "455  28224871  Unfortunately the host did not provide keys to...      0"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-694efe0f-c84d-4374-9844-f13525b0c99a\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>id</th>\n",
              "      <th>review</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>321</th>\n",
              "      <td>19531174</td>\n",
              "      <td>Most horrible Airbnb experience I ever had, we...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>363</th>\n",
              "      <td>22306777</td>\n",
              "      <td>This location is ideal for fans of Brooklyn be...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>364</th>\n",
              "      <td>22306777</td>\n",
              "      <td>Bartolomeo is a nice and friendly host, the ap...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>454</th>\n",
              "      <td>28224871</td>\n",
              "      <td>I needed a place near NY Presbyterian Hospital...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>455</th>\n",
              "      <td>28224871</td>\n",
              "      <td>Unfortunately the host did not provide keys to...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-694efe0f-c84d-4374-9844-f13525b0c99a')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-694efe0f-c84d-4374-9844-f13525b0c99a button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-694efe0f-c84d-4374-9844-f13525b0c99a');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-606ac686-204c-4999-99f4-b688c0bc5327\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-606ac686-204c-4999-99f4-b688c0bc5327')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-606ac686-204c-4999-99f4-b688c0bc5327 button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "    </div>\n",
              "  </div>\n"
            ]
          },
          "metadata": {},
          "execution_count": 265
        }
      ],
      "source": [
        "neg_dfs.head()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 266,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "id": "4KmeUyV--E-x",
        "outputId": "68385b4f-12cf-47b8-edd0-57edce2edf56"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "              id                                             review  label\n",
              "683338  37020421  Honestly, I would not stay here again. Our roo...      1\n",
              "661702  30713613           Lovely place, excellent stay experience!      1\n",
              "437280   6180121  Grover's house is my first ever Airbnb experie...      1\n",
              "659999  31291730  Not a clean hotel. Its only use is their showe...      1\n",
              "880006  22076961  Great place!  Very spacious and close to fanta...      1"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-2a73653d-0e92-4234-b3a9-a13cf2fa2908\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>id</th>\n",
              "      <th>review</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>683338</th>\n",
              "      <td>37020421</td>\n",
              "      <td>Honestly, I would not stay here again. Our roo...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>661702</th>\n",
              "      <td>30713613</td>\n",
              "      <td>Lovely place, excellent stay experience!</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>437280</th>\n",
              "      <td>6180121</td>\n",
              "      <td>Grover's house is my first ever Airbnb experie...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>659999</th>\n",
              "      <td>31291730</td>\n",
              "      <td>Not a clean hotel. Its only use is their showe...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>880006</th>\n",
              "      <td>22076961</td>\n",
              "      <td>Great place!  Very spacious and close to fanta...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-2a73653d-0e92-4234-b3a9-a13cf2fa2908')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-2a73653d-0e92-4234-b3a9-a13cf2fa2908 button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-2a73653d-0e92-4234-b3a9-a13cf2fa2908');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-aa58090e-5961-46ce-a137-723642da7c10\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-aa58090e-5961-46ce-a137-723642da7c10')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-aa58090e-5961-46ce-a137-723642da7c10 button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "    </div>\n",
              "  </div>\n"
            ]
          },
          "metadata": {},
          "execution_count": 266
        }
      ],
      "source": [
        "new_dfs.head()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 267,
      "metadata": {
        "id": "DmnOPb0y-E-x"
      },
      "outputs": [],
      "source": [
        "# new_dfs = new_dfs[['id', 'review']]\n",
        "# neg_dfs = neg_dfs[['id', 'review']]\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 268,
      "metadata": {
        "id": "rAMJkLzm-E-x"
      },
      "outputs": [],
      "source": [
        "test_dfs = pd.merge(new_dfs, neg_dfs, how='outer')\n",
        "new_test_dfs = pd.merge(neg_dfs, positive_dfs_test, how='outer')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 269,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "id": "KBpuJ60P-E-x",
        "outputId": "ce390628-1f9e-4ca6-f15b-c044f56e88a3"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "                      id                                             review  \\\n",
              "116   902324676342743242  The neighborhood is lovely & close to shops. T...   \n",
              "581             14037459  Our stay was great! We enjoyed Kathryn and Tob...   \n",
              "774   649185211357475650  Alexandra (our host) was very pleasant and com...   \n",
              "1510            24699301  This was our first trip since leaving NYC a ye...   \n",
              "733             45048137  Host was non existent, did not reply at all. C...   \n",
              "\n",
              "      label  \n",
              "116       0  \n",
              "581       0  \n",
              "774       0  \n",
              "1510      1  \n",
              "733       0  "
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-4521254f-07e2-46c2-af43-1b1d47f924f1\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>id</th>\n",
              "      <th>review</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>116</th>\n",
              "      <td>902324676342743242</td>\n",
              "      <td>The neighborhood is lovely &amp; close to shops. T...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>581</th>\n",
              "      <td>14037459</td>\n",
              "      <td>Our stay was great! We enjoyed Kathryn and Tob...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>774</th>\n",
              "      <td>649185211357475650</td>\n",
              "      <td>Alexandra (our host) was very pleasant and com...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1510</th>\n",
              "      <td>24699301</td>\n",
              "      <td>This was our first trip since leaving NYC a ye...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>733</th>\n",
              "      <td>45048137</td>\n",
              "      <td>Host was non existent, did not reply at all. C...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-4521254f-07e2-46c2-af43-1b1d47f924f1')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-4521254f-07e2-46c2-af43-1b1d47f924f1 button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-4521254f-07e2-46c2-af43-1b1d47f924f1');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-1f0738fe-9829-4449-86b5-f0244b4762de\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-1f0738fe-9829-4449-86b5-f0244b4762de')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-1f0738fe-9829-4449-86b5-f0244b4762de button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "    </div>\n",
              "  </div>\n"
            ]
          },
          "metadata": {},
          "execution_count": 269
        }
      ],
      "source": [
        "new_test_dfs.sample(5)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 270,
      "metadata": {
        "id": "ilqb_gpk-E-x"
      },
      "outputs": [],
      "source": [
        "test_dfs.review = test_dfs.review.astype(str)\n",
        "new_test_dfs.review = test_dfs.review.astype(str)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 271,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kPSgBHa4-E-y",
        "outputId": "97872a53-bc3e-439e-db4e-d54416b6c871"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "There are 1 GPU(s) available.\n",
            "Device name: Tesla T4\n"
          ]
        }
      ],
      "source": [
        "import torch\n",
        "\n",
        "if torch.cuda.is_available():\n",
        "    device = torch.device(\"cuda\")\n",
        "    print(f'There are {torch.cuda.device_count()} GPU(s) available.')\n",
        "    print('Device name:', torch.cuda.get_device_name(0))\n",
        "\n",
        "else:\n",
        "    print('No GPU available, using the CPU instead.')\n",
        "    device = torch.device(\"cpu\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 272,
      "metadata": {
        "id": "_3BvdRDY-E-y"
      },
      "outputs": [],
      "source": [
        "data = test_dfs\n",
        "test_data = new_test_dfs\n",
        "\n",
        "# drop label on the test data\n",
        "test_data.drop(['label'], inplace=True, axis=1)\n"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "data.label.value_counts()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "v04czaXDdu_K",
        "outputId": "875b4acf-fd02-4607-d12f-bb04d51c6e69"
      },
      "execution_count": 276,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "1    1500\n",
              "0    1449\n",
              "Name: label, dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 276
        }
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 277,
      "metadata": {
        "id": "lOw1QOoZ-E-y"
      },
      "outputs": [],
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "X = data.review.values\n",
        "y = data.label.values\n",
        "\n",
        "X_train, X_val, y_train, y_val = train_test_split(X, y, test_size=0.1, random_state=2020)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kpTxVyYWBYcX"
      },
      "source": [
        "### TF-IDF + Naive Bayes Classifier\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 278,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7QJobkJEBg3X",
        "outputId": "452f03e1-9226-4c6b-85e7-22a923f6734c"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "[nltk_data] Downloading package stopwords to /root/nltk_data...\n",
            "[nltk_data]   Package stopwords is already up-to-date!\n"
          ]
        }
      ],
      "source": [
        "import nltk\n",
        "# Uncomment to download \"stopwords\"\n",
        "nltk.download(\"stopwords\")\n",
        "from nltk.corpus import stopwords\n",
        "\n",
        "def text_preprocessing(s):\n",
        "    \"\"\"\n",
        "    - Lowercase the sentence\n",
        "    - Change \"'t\" to \"not\"\n",
        "    - Remove \"@name\"\n",
        "    - Isolate and remove punctuations except \"?\"\n",
        "    - Remove other special characters\n",
        "    - Remove stop words except \"not\" and \"can\"\n",
        "    - Remove trailing whitespace\n",
        "    \"\"\"\n",
        "    s = s.lower()\n",
        "    # Change 't to 'not'\n",
        "    s = re.sub(r\"\\'t\", \" not\", s)\n",
        "    # Remove @name\n",
        "    s = re.sub(r'(@.*?)[\\s]', ' ', s)\n",
        "    # Isolate and remove punctuations except '?'\n",
        "    s = re.sub(r'([\\'\\\"\\.\\(\\)\\!\\?\\\\\\/\\,])', r'  ', s)\n",
        "    s = re.sub(r'[^\\w\\s\\?]', ' ', s)\n",
        "    # Remove some special characters\n",
        "    s = re.sub(r'([\\;\\:\\|•«\\n])', ' ', s)\n",
        "    # Remove stopwords except 'not' and 'can'\n",
        "    s = \" \".join([word for word in s.split()\n",
        "                  if word not in stopwords.words('english')\n",
        "                  or word in ['not', 'can']])\n",
        "    # Remove trailing whitespace\n",
        "    s = re.sub(r'\\s+', ' ', s).strip()\n",
        "\n",
        "    return s"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 279,
      "metadata": {
        "id": "_QPlBgJKBwX-",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "a44d9c30-cbe6-4d7d-ff81-cafdf8af39ec"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "CPU times: user 16.8 s, sys: 2.18 s, total: 19 s\n",
            "Wall time: 19.8 s\n"
          ]
        }
      ],
      "source": [
        "%%time\n",
        "from sklearn.feature_extraction.text import TfidfVectorizer\n",
        "\n",
        "# Preprocess text\n",
        "X_train_preprocessed = np.array([text_preprocessing(text) for text in X_train])\n",
        "X_val_preprocessed = np.array([text_preprocessing(text) for text in X_val])\n",
        "\n",
        "# Calculate TF-IDF\n",
        "tf_idf = TfidfVectorizer(smooth_idf=False)\n",
        "X_train_tfidf = tf_idf.fit_transform(X_train_preprocessed)\n",
        "X_val_tfidf = tf_idf.transform(X_val_preprocessed)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 280,
      "metadata": {
        "id": "CHjVaXPoCLx5"
      },
      "outputs": [],
      "source": [
        "from sklearn.model_selection import StratifiedKFold, cross_val_score\n",
        "\n",
        "def get_auc_CV(model):\n",
        "    \"\"\"\n",
        "    Return the average AUC score from cross-validation.\n",
        "    \"\"\"\n",
        "    # Set KFold to shuffle data before the split\n",
        "    kf = StratifiedKFold(5, shuffle=True, random_state=1)\n",
        "\n",
        "    # Get AUC scores\n",
        "    auc = cross_val_score(\n",
        "        model, X_train_tfidf, y_train, scoring=\"roc_auc\", cv=kf)\n",
        "\n",
        "    return auc.mean()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 281,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 490
        },
        "id": "8Zs4Fug6CvJg",
        "outputId": "a70bce41-eed6-4d1a-843e-16a8d6a6ba6e"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Best alpha:  1.0\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "from sklearn.naive_bayes import MultinomialNB\n",
        "\n",
        "alpha_values = np.arange(1, 10, 0.1)\n",
        "res = pd.Series([get_auc_CV(MultinomialNB(alpha=i)) for i in alpha_values],\n",
        "                index=alpha_values)\n",
        "best_alpha = np.round(res.idxmax(), 2)\n",
        "print('Best alpha: ', best_alpha)\n",
        "\n",
        "plt.plot(res)\n",
        "plt.title('AUC vs. Alpha')\n",
        "plt.xlabel('Alpha')\n",
        "plt.ylabel('AUC')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 284,
      "metadata": {
        "id": "EJ-4zNmlNPSe"
      },
      "outputs": [],
      "source": [
        "from sklearn.metrics import accuracy_score, roc_curve, auc\n",
        "\n",
        "def evaluate_roc(probs, y_true):\n",
        "    \"\"\"\n",
        "    - Print AUC and accuracy on the test set\n",
        "    - Plot ROC\n",
        "    @params    probs (np.array): an array of predicted probabilities with shape (len(y_true), 2)\n",
        "    @params    y_true (np.array): an array of the true values with shape (len(y_true),)\n",
        "    \"\"\"\n",
        "    preds = probs[:, 1]\n",
        "    fpr, tpr, threshold = roc_curve(y_true, preds)\n",
        "    roc_auc = auc(fpr, tpr)\n",
        "    print(f'AUC: {roc_auc:.4f}')\n",
        "\n",
        "    # Get accuracy over the test set\n",
        "    y_pred = np.where(preds >= 0.5, 1, 0)\n",
        "    accuracy = accuracy_score(y_true, y_pred)\n",
        "    print(f'Accuracy: {accuracy*100:.2f}%')\n",
        "\n",
        "    # Plot ROC AUC\n",
        "    plt.title('Receiver Operating Characteristic')\n",
        "    plt.plot(fpr, tpr, 'b', label = 'AUC = %0.2f' % roc_auc)\n",
        "    plt.legend(loc = 'lower right')\n",
        "    plt.plot([0, 1], [0, 1],'r--')\n",
        "    plt.xlim([0, 1])\n",
        "    plt.ylim([0, 1])\n",
        "    plt.ylabel('True Positive Rate')\n",
        "    plt.xlabel('False Positive Rate')\n",
        "    plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 285,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 509
        },
        "id": "jCWCkvQrNUr2",
        "outputId": "bd6c7315-b715-43cd-b16e-f36d9c1cec9f"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "AUC: 0.8483\n",
            "Accuracy: 76.95%\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "# Compute predicted probabilities\n",
        "nb_model = MultinomialNB(alpha=1.0)\n",
        "nb_model.fit(X_train_tfidf, y_train)\n",
        "probs = nb_model.predict_proba(X_val_tfidf)\n",
        "\n",
        "# Evaluate the classifier\n",
        "evaluate_roc(probs, y_val)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zOwhNucAShSI"
      },
      "source": [
        "### fine-tuning BERT"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 380,
      "metadata": {
        "id": "QQWeCu_--E-z"
      },
      "outputs": [],
      "source": [
        "def text_preprocessing(text):\n",
        "    \"\"\"\n",
        "    - Remove entity mentions (eg. '@')\n",
        "    - Correct errors\n",
        "    @param    text (str): a string to be processed.\n",
        "    @return   text (Str): the processed string.\n",
        "    \"\"\"\n",
        "    # Remove '@'\n",
        "    text = re.sub(r'(@.*?)[\\s]',' ', text)\n",
        "\n",
        "    # Replace '<\\br>' with ' '\n",
        "    text = re.sub(r'<\\br>;', ' ', text)\n",
        "\n",
        "    # Remove trailing whitespace\n",
        "    text = re.sub(r'\\s+', ' ', text).strip()\n",
        "\n",
        "    return text"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 381,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7xuUgszU-E-z",
        "outputId": "02a36c57-2895-464a-afaf-079cf4041172"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Original:  Honestly, I would not stay here again. Our room wasn't clean when we arrived (dirty sheets, toilet super disgusting) and wasn't cleaned until a couple hours after we checked in.  Maintenance definitely needs to be done on the place. Our toilet didn't really work (I had to fix it one morning, because the handle was completely disconnected from the toilet, the water wouldn't stop running, & it wouldn't flush), the faucet handle on the sink kept falling off, & there's a TON of trash in the front yard.<br/>The room is also super noisy, the floor upstairs is extremely creaky so you can hear absolutely any movement from the neighbors on the upper floor, which can keep you up late into the night.<br/>Also important to note: a dog lives here, which doesn't seem to be mentioned anywhere but should be noted in case someone has an allergy.<br/>It's a good location, close to public transit, and the host is responsive, but it's waaaaay too frustrating to deal with everything else.\n",
            "Processed:  Honestly, I would not stay here again. Our room wasn't clean when we arrived (dirty sheets, toilet super disgusting) and wasn't cleaned until a couple hours after we checked in. Maintenance definitely needs to be done on the place. Our toilet didn't really work (I had to fix it one morning, because the handle was completely disconnected from the toilet, the water wouldn't stop running, & it wouldn't flush), the faucet handle on the sink kept falling off, & there's a TON of trash in the front yard.<br/>The room is also super noisy, the floor upstairs is extremely creaky so you can hear absolutely any movement from the neighbors on the upper floor, which can keep you up late into the night.<br/>Also important to note: a dog lives here, which doesn't seem to be mentioned anywhere but should be noted in case someone has an allergy.<br/>It's a good location, close to public transit, and the host is responsive, but it's waaaaay too frustrating to deal with everything else.\n"
          ]
        }
      ],
      "source": [
        "# Print sentence 0\n",
        "print('Original: ', X[0])\n",
        "print('Processed: ', text_preprocessing(X[0]))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 382,
      "metadata": {
        "id": "rXN4nbFZ-E-z"
      },
      "outputs": [],
      "source": [
        "from transformers import RobertaTokenizer\n",
        "\n",
        "# Load the BERT tokenizer\n",
        "tokenizer = RobertaTokenizer.from_pretrained('roberta-base', do_lower_case=True)\n",
        "\n",
        "# Create a function to tokenize a set of texts\n",
        "def preprocessing_for_bert(data):\n",
        "    \"\"\"Perform required preprocessing steps for pretrained BERT.\n",
        "    @param    data (np.array): Array of texts to be processed.\n",
        "    @return   input_ids (torch.Tensor): Tensor of token ids to be fed to a model.\n",
        "    @return   attention_masks (torch.Tensor): Tensor of indices specifying which\n",
        "                  tokens should be attended to by the model.\n",
        "    \"\"\"\n",
        "    # Create empty lists to store outputs\n",
        "    input_ids = []\n",
        "    attention_masks = []\n",
        "\n",
        "    # For every sentence...\n",
        "    for sent in data:\n",
        "        # `encode_plus` will:\n",
        "        #    (1) Tokenize the sentence\n",
        "        #    (2) Add the `[CLS]` and `[SEP]` token to the start and end\n",
        "        #    (3) Truncate/Pad sentence to max length\n",
        "        #    (4) Map tokens to their IDs\n",
        "        #    (5) Create attention mask\n",
        "        #    (6) Return a dictionary of outputs\n",
        "        encoded_sent = tokenizer.encode_plus(\n",
        "            text=text_preprocessing(sent),  # Preprocess sentence\n",
        "            add_special_tokens=True,        # Add `[CLS]` and `[SEP]`\n",
        "            max_length=MAX_LEN,             # Max length to truncate/pad\n",
        "            pad_to_max_length=True,         # Pad sentence to max length\n",
        "            #return_tensors='pt',           # Return PyTorch tensor\n",
        "            return_attention_mask=True      # Return attention mask\n",
        "            )\n",
        "\n",
        "        # Add the outputs to the lists\n",
        "        input_ids.append(encoded_sent.get('input_ids'))\n",
        "        attention_masks.append(encoded_sent.get('attention_mask'))\n",
        "\n",
        "    # Convert lists to tensors\n",
        "    input_ids = torch.tensor(input_ids)\n",
        "    attention_masks = torch.tensor(attention_masks)\n",
        "\n",
        "    return input_ids, attention_masks"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 392,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "9j3FNXn--E-0",
        "outputId": "272ce3f7-aed8-470b-e005-55801547336f"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Max length:  1674\n"
          ]
        }
      ],
      "source": [
        "# Concatenate train data and test data (test data is already a subset of train data)\n",
        "all_data = data.review.values\n",
        "\n",
        "# Encode our concatenated data\n",
        "encoded_data = [tokenizer.encode(sent, add_special_tokens=True) for sent in all_data]\n",
        "\n",
        "# Find the maximum length\n",
        "max_len = max([len(sent) for sent in encoded_data])\n",
        "print('Max length: ', max_len)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 393,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4Bb_-_Lo-E-0",
        "outputId": "1955bc76-1546-44fb-eff7-e6069a54926f"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/transformers/tokenization_utils_base.py:2614: FutureWarning: The `pad_to_max_length` argument is deprecated and will be removed in a future version, use `padding=True` or `padding='longest'` to pad to the longest sequence in the batch, or use `padding='max_length'` to pad to a max length. In this case, you can give a specific length with `max_length` (e.g. `max_length=45`) or leave max_length to None to pad to the maximal input size of the model (e.g. 512 for Bert).\n",
            "  warnings.warn(\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Original:  Honestly, I would not stay here again. Our room wasn't clean when we arrived (dirty sheets, toilet super disgusting) and wasn't cleaned until a couple hours after we checked in.  Maintenance definitely needs to be done on the place. Our toilet didn't really work (I had to fix it one morning, because the handle was completely disconnected from the toilet, the water wouldn't stop running, & it wouldn't flush), the faucet handle on the sink kept falling off, & there's a TON of trash in the front yard.<br/>The room is also super noisy, the floor upstairs is extremely creaky so you can hear absolutely any movement from the neighbors on the upper floor, which can keep you up late into the night.<br/>Also important to note: a dog lives here, which doesn't seem to be mentioned anywhere but should be noted in case someone has an allergy.<br/>It's a good location, close to public transit, and the host is responsive, but it's waaaaay too frustrating to deal with everything else.\n",
            "Token IDs:  [0, 30048, 6, 38, 74, 45, 1095, 259, 456, 4, 1541, 929, 938, 75, 2382, 77, 52, 2035, 36, 40087, 12208, 6, 11471, 2422, 21096, 43, 8, 938, 75, 17317, 454, 10, 891, 722, 71, 52, 7869, 11, 4, 29738, 2299, 782, 7, 28, 626, 15, 5, 317, 4, 1541, 11471, 399, 75, 269, 173, 36, 100, 56, 7, 4190, 24, 65, 662, 2]\n",
            "Tokenizing data...\n"
          ]
        }
      ],
      "source": [
        "# Specify `MAX_LEN`\n",
        "MAX_LEN = 64\n",
        "\n",
        "# Print sentence 0 and its encoded token ids\n",
        "token_ids = list(preprocessing_for_bert([X[0]])[0].squeeze().numpy())\n",
        "print('Original: ', X[0])\n",
        "print('Token IDs: ', token_ids)\n",
        "\n",
        "# Run function `preprocessing_for_bert` on the train set and the validation set\n",
        "print('Tokenizing data...')\n",
        "train_inputs, train_masks = preprocessing_for_bert(X_train)\n",
        "val_inputs, val_masks = preprocessing_for_bert(X_val)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 394,
      "metadata": {
        "id": "svydY_ta-E-0"
      },
      "outputs": [],
      "source": [
        "from torch.utils.data import TensorDataset, DataLoader, RandomSampler, SequentialSampler\n",
        "\n",
        "# Convert other data types to torch.Tensor\n",
        "train_labels = torch.tensor(y_train)\n",
        "val_labels = torch.tensor(y_val)\n",
        "\n",
        "# For fine-tuning BERT, the authors recommend a batch size of 16 or 32.\n",
        "batch_size = 32\n",
        "\n",
        "# Create the DataLoader for our training set\n",
        "train_data = TensorDataset(train_inputs, train_masks, train_labels)\n",
        "train_sampler = RandomSampler(train_data)\n",
        "train_dataloader = DataLoader(train_data, sampler=train_sampler, batch_size=batch_size)\n",
        "\n",
        "# Create the DataLoader for our validation set\n",
        "val_data = TensorDataset(val_inputs, val_masks, val_labels)\n",
        "val_sampler = SequentialSampler(val_data)\n",
        "val_dataloader = DataLoader(val_data, sampler=val_sampler, batch_size=batch_size)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 395,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fcloJVvo-E-0",
        "outputId": "9523acaf-5648-47fd-d1c0-01b8d6e13a84"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "CPU times: user 52 µs, sys: 0 ns, total: 52 µs\n",
            "Wall time: 55.8 µs\n"
          ]
        }
      ],
      "source": [
        "%%time\n",
        "import torch\n",
        "import torch.nn as nn\n",
        "from transformers import RobertaModel\n",
        "\n",
        "# Create the BertClassfier class\n",
        "class RobertaClassifier(nn.Module):\n",
        "    \"\"\"Bert Model for Classification Tasks.\n",
        "    \"\"\"\n",
        "    def __init__(self, freeze_bert=False):\n",
        "        \"\"\"\n",
        "        @param    bert: a BertModel object\n",
        "        @param    classifier: a torch.nn.Module classifier\n",
        "        @param    freeze_bert (bool): Set `False` to fine-tune the BERT model\n",
        "        \"\"\"\n",
        "        super(RobertaClassifier, self).__init__()\n",
        "        # Specify hidden size of BERT, hidden size of our classifier, and number of labels\n",
        "        D_in, H, D_out = 768, 50, 2\n",
        "\n",
        "        # Instantiate BERT model\n",
        "        self.bert = RobertaModel.from_pretrained('roberta-base')\n",
        "\n",
        "        # Instantiate an one-layer feed-forward classifier\n",
        "        self.classifier = nn.Sequential(\n",
        "            nn.Linear(D_in, H),\n",
        "            nn.ReLU(),\n",
        "            #nn.Dropout(0.5),\n",
        "            nn.Linear(H, D_out)\n",
        "        )\n",
        "\n",
        "        # Freeze the BERT model\n",
        "        if freeze_bert:\n",
        "            for param in self.bert.parameters():\n",
        "                param.requires_grad = False\n",
        "\n",
        "    def forward(self, input_ids, attention_mask):\n",
        "        \"\"\"\n",
        "        Feed input to BERT and the classifier to compute logits.\n",
        "        @param    input_ids (torch.Tensor): an input tensor with shape (batch_size,\n",
        "                      max_length)\n",
        "        @param    attention_mask (torch.Tensor): a tensor that hold attention mask\n",
        "                      information with shape (batch_size, max_length)\n",
        "        @return   logits (torch.Tensor): an output tensor with shape (batch_size,\n",
        "                      num_labels)\n",
        "        \"\"\"\n",
        "        # Feed input to BERT\n",
        "        outputs = self.bert(input_ids=input_ids,\n",
        "                            attention_mask=attention_mask)\n",
        "\n",
        "        # Extract the last hidden state of the token `[CLS]` for classification task\n",
        "        last_hidden_state_cls = outputs[0][:, 0, :]\n",
        "\n",
        "        # Feed input to classifier to compute logits\n",
        "        logits = self.classifier(last_hidden_state_cls)\n",
        "\n",
        "        return logits"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 396,
      "metadata": {
        "id": "S7mP3nBD-E-1"
      },
      "outputs": [],
      "source": [
        "from transformers import AdamW, get_linear_schedule_with_warmup\n",
        "\n",
        "def initialize_model(epochs=4):\n",
        "    \"\"\"Initialize the Bert Classifier, the optimizer and the learning rate scheduler.\n",
        "    \"\"\"\n",
        "    # Instantiate Bert Classifier\n",
        "    roberta_classifier = RobertaClassifier(freeze_bert=False)\n",
        "\n",
        "    # Tell PyTorch to run the model on GPU\n",
        "    roberta_classifier.to(device)\n",
        "\n",
        "    # Create the optimizer\n",
        "    optimizer = torch.optim.AdamW(roberta_classifier.parameters(),\n",
        "                      lr=5e-5,    # Default learning rate\n",
        "                      eps=1e-8    # Default epsilon value\n",
        "                      )\n",
        "\n",
        "    # Total number of training steps\n",
        "    total_steps = len(train_dataloader) * epochs\n",
        "\n",
        "    # Set up the learning rate scheduler\n",
        "    scheduler = get_linear_schedule_with_warmup(optimizer,\n",
        "                                                num_warmup_steps=0, # Default value\n",
        "                                                num_training_steps=total_steps)\n",
        "    return roberta_classifier, optimizer, scheduler"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 397,
      "metadata": {
        "id": "Asab23gm-E-1"
      },
      "outputs": [],
      "source": [
        "import random\n",
        "import time\n",
        "\n",
        "# Specify loss function\n",
        "loss_fn = nn.CrossEntropyLoss()\n",
        "\n",
        "def set_seed(seed_value=42):\n",
        "    \"\"\"Set seed for reproducibility.\n",
        "    \"\"\"\n",
        "    random.seed(seed_value)\n",
        "    np.random.seed(seed_value)\n",
        "    torch.manual_seed(seed_value)\n",
        "    torch.cuda.manual_seed_all(seed_value)\n",
        "\n",
        "def train(model, train_dataloader, val_dataloader=None, epochs=4, evaluation=False):\n",
        "    \"\"\"Train the BertClassifier model.\n",
        "    \"\"\"\n",
        "    # Start training loop\n",
        "    print(\"Start training...\\n\")\n",
        "    for epoch_i in range(epochs):\n",
        "        # =======================================\n",
        "        #               Training\n",
        "        # =======================================\n",
        "        # Print the header of the result table\n",
        "        print(f\"{'Epoch':^7} | {'Batch':^7} | {'Train Loss':^12} | {'Val Loss':^10} | {'Val Acc':^9} | {'Elapsed':^9}\")\n",
        "        print(\"-\"*70)\n",
        "\n",
        "        # Measure the elapsed time of each epoch\n",
        "        t0_epoch, t0_batch = time.time(), time.time()\n",
        "\n",
        "        # Reset tracking variables at the beginning of each epoch\n",
        "        total_loss, batch_loss, batch_counts = 0, 0, 0\n",
        "\n",
        "        # Put the model into the training mode\n",
        "        model.train()\n",
        "\n",
        "        # For each batch of training data...\n",
        "        for step, batch in enumerate(train_dataloader):\n",
        "            batch_counts +=1\n",
        "            # Load batch to GPU\n",
        "            b_input_ids, b_attn_mask, b_labels = tuple(t.to(device) for t in batch)\n",
        "\n",
        "            # Zero out any previously calculated gradients\n",
        "            model.zero_grad()\n",
        "\n",
        "            # Perform a forward pass. This will return logits.\n",
        "            logits = model(b_input_ids, b_attn_mask)\n",
        "\n",
        "            # Compute loss and accumulate the loss values\n",
        "            loss = loss_fn(logits, b_labels)\n",
        "            batch_loss += loss.item()\n",
        "            total_loss += loss.item()\n",
        "\n",
        "            # Perform a backward pass to calculate gradients\n",
        "            loss.backward()\n",
        "\n",
        "            # Clip the norm of the gradients to 1.0 to prevent \"exploding gradients\"\n",
        "            torch.nn.utils.clip_grad_norm_(model.parameters(), 1.0)\n",
        "\n",
        "            # Update parameters and the learning rate\n",
        "            optimizer.step()\n",
        "            scheduler.step()\n",
        "\n",
        "            # Print the loss values and time elapsed for every 20 batches\n",
        "            if (step % 20 == 0 and step != 0) or (step == len(train_dataloader) - 1):\n",
        "                # Calculate time elapsed for 20 batches\n",
        "                time_elapsed = time.time() - t0_batch\n",
        "\n",
        "                # Print training results\n",
        "                print(f\"{epoch_i + 1:^7} | {step:^7} | {batch_loss / batch_counts:^12.6f} | {'-':^10} | {'-':^9} | {time_elapsed:^9.2f}\")\n",
        "\n",
        "                # Reset batch tracking variables\n",
        "                batch_loss, batch_counts = 0, 0\n",
        "                t0_batch = time.time()\n",
        "\n",
        "        # Calculate the average loss over the entire training data\n",
        "        avg_train_loss = total_loss / len(train_dataloader)\n",
        "\n",
        "        print(\"-\"*70)\n",
        "        # =======================================\n",
        "        #               Evaluation\n",
        "        # =======================================\n",
        "        if evaluation == True:\n",
        "            # After the completion of each training epoch, measure the model's performance\n",
        "            # on our validation set.\n",
        "            val_loss, val_accuracy = evaluate(model, val_dataloader)\n",
        "\n",
        "            # Print performance over the entire training data\n",
        "            time_elapsed = time.time() - t0_epoch\n",
        "\n",
        "            print(f\"{epoch_i + 1:^7} | {'-':^7} | {avg_train_loss:^12.6f} | {val_loss:^10.6f} | {val_accuracy:^9.2f} | {time_elapsed:^9.2f}\")\n",
        "            print(\"-\"*70)\n",
        "        print(\"\\n\")\n",
        "\n",
        "    print(\"Training complete!\")\n",
        "\n",
        "\n",
        "def evaluate(model, val_dataloader):\n",
        "    \"\"\"After the completion of each training epoch, measure the model's performance\n",
        "    on our validation set.\n",
        "    \"\"\"\n",
        "    # Put the model into the evaluation mode. The dropout layers are disabled during\n",
        "    # the test time.\n",
        "    model.eval()\n",
        "\n",
        "    # Tracking variables\n",
        "    val_accuracy = []\n",
        "    val_loss = []\n",
        "\n",
        "    # For each batch in our validation set...\n",
        "    for batch in val_dataloader:\n",
        "        # Load batch to GPU\n",
        "        b_input_ids, b_attn_mask, b_labels = tuple(t.to(device) for t in batch)\n",
        "\n",
        "        # Compute logits\n",
        "        with torch.no_grad():\n",
        "            logits = model(b_input_ids, b_attn_mask)\n",
        "\n",
        "        # Compute loss\n",
        "        loss = loss_fn(logits, b_labels)\n",
        "        val_loss.append(loss.item())\n",
        "\n",
        "        # Get the predictions\n",
        "        preds = torch.argmax(logits, dim=1).flatten()\n",
        "\n",
        "        # Calculate the accuracy rate\n",
        "        accuracy = (preds == b_labels).cpu().numpy().mean() * 100\n",
        "        val_accuracy.append(accuracy)\n",
        "\n",
        "    # Compute the average accuracy and loss over the validation set.\n",
        "    val_loss = np.mean(val_loss)\n",
        "    val_accuracy = np.mean(val_accuracy)\n",
        "\n",
        "    return val_loss, val_accuracy"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 399,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "QafG5CJa-E-1",
        "outputId": "89f2829d-1897-48ff-f399-91a61d293b35"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "Some weights of RobertaModel were not initialized from the model checkpoint at roberta-base and are newly initialized: ['roberta.pooler.dense.bias', 'roberta.pooler.dense.weight']\n",
            "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Start training...\n",
            "\n",
            " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
            "----------------------------------------------------------------------\n",
            "   1    |   20    |   0.649911   |     -      |     -     |   7.18   \n",
            "   1    |   40    |   0.545615   |     -      |     -     |   7.03   \n",
            "   1    |   60    |   0.511445   |     -      |     -     |   7.17   \n",
            "   1    |   80    |   0.462830   |     -      |     -     |   7.27   \n",
            "   1    |   82    |   0.428538   |     -      |     -     |   0.72   \n",
            "----------------------------------------------------------------------\n",
            "   1    |    -    |   0.541000   |  0.476869  |   81.21   |   30.70  \n",
            "----------------------------------------------------------------------\n",
            "\n",
            "\n",
            " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
            "----------------------------------------------------------------------\n",
            "   2    |   20    |   0.429678   |     -      |     -     |   7.54   \n",
            "   2    |   40    |   0.489783   |     -      |     -     |   7.39   \n",
            "   2    |   60    |   0.468566   |     -      |     -     |   7.30   \n",
            "   2    |   80    |   0.433177   |     -      |     -     |   7.25   \n",
            "   2    |   82    |   0.474086   |     -      |     -     |   0.71   \n",
            "----------------------------------------------------------------------\n",
            "   2    |    -    |   0.455445   |  0.507368  |   77.90   |   31.45  \n",
            "----------------------------------------------------------------------\n",
            "\n",
            "\n",
            " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
            "----------------------------------------------------------------------\n",
            "   3    |   20    |   0.399999   |     -      |     -     |   7.28   \n",
            "   3    |   40    |   0.370226   |     -      |     -     |   7.10   \n",
            "   3    |   60    |   0.385052   |     -      |     -     |   7.09   \n",
            "   3    |   80    |   0.355291   |     -      |     -     |   7.09   \n",
            "   3    |   82    |   0.265021   |     -      |     -     |   0.69   \n",
            "----------------------------------------------------------------------\n",
            "   3    |    -    |   0.375198   |  0.498296  |   77.28   |   30.51  \n",
            "----------------------------------------------------------------------\n",
            "\n",
            "\n",
            " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
            "----------------------------------------------------------------------\n",
            "   4    |   20    |   0.306091   |     -      |     -     |   7.25   \n",
            "   4    |   40    |   0.329518   |     -      |     -     |   7.11   \n",
            "   4    |   60    |   0.305948   |     -      |     -     |   7.18   \n",
            "   4    |   80    |   0.286791   |     -      |     -     |   7.21   \n",
            "   4    |   82    |   0.395901   |     -      |     -     |   0.71   \n",
            "----------------------------------------------------------------------\n",
            "   4    |    -    |   0.309215   |  0.506252  |   78.71   |   30.75  \n",
            "----------------------------------------------------------------------\n",
            "\n",
            "\n",
            "Training complete!\n"
          ]
        }
      ],
      "source": [
        "set_seed(42)    # Set seed for reproducibility\n",
        "roberta_classifier, optimizer, scheduler = initialize_model(epochs=4)\n",
        "train(roberta_classifier, train_dataloader, val_dataloader, epochs=4, evaluation=True)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 400,
      "metadata": {
        "id": "sXSmYpn9VMgZ"
      },
      "outputs": [],
      "source": [
        "import torch.nn.functional as F\n",
        "\n",
        "def bert_predict(model, test_dataloader):\n",
        "    \"\"\"Perform a forward pass on the trained BERT model to predict probabilities\n",
        "    on the test set.\n",
        "    \"\"\"\n",
        "    # Put the model into the evaluation mode. The dropout layers are disabled during\n",
        "    # the test time.\n",
        "    model.eval()\n",
        "\n",
        "    all_logits = []\n",
        "\n",
        "    # For each batch in our test set...\n",
        "    for batch in test_dataloader:\n",
        "        # Load batch to GPU\n",
        "        b_input_ids, b_attn_mask = tuple(t.to(device) for t in batch)[:2]\n",
        "\n",
        "        # Compute logits\n",
        "        with torch.no_grad():\n",
        "            logits = model(b_input_ids, b_attn_mask)\n",
        "        all_logits.append(logits)\n",
        "\n",
        "    # Concatenate logits from each batch\n",
        "    all_logits = torch.cat(all_logits, dim=0)\n",
        "\n",
        "    # Apply softmax to calculate probabilities\n",
        "    probs = F.softmax(all_logits, dim=1).cpu().numpy()\n",
        "\n",
        "    return probs"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 401,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 509
        },
        "id": "i_3c0axz-E-1",
        "outputId": "9ec72a5d-da33-40b4-b2ec-4263df6091ca"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "AUC: 0.8723\n",
            "Accuracy: 79.32%\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "# Compute predicted probabilities on the test set\n",
        "probs = bert_predict(roberta_classifier, val_dataloader)\n",
        "\n",
        "# Evaluate the Bert classifier\n",
        "evaluate_roc(probs, y_val)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 402,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2PKo6SLH-E-2",
        "outputId": "aab6f294-be20-4778-b44e-c66df311a493"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "Some weights of RobertaModel were not initialized from the model checkpoint at roberta-base and are newly initialized: ['roberta.pooler.dense.bias', 'roberta.pooler.dense.weight']\n",
            "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Start training...\n",
            "\n",
            " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
            "----------------------------------------------------------------------\n",
            "   1    |   20    |   0.625199   |     -      |     -     |   7.28   \n",
            "   1    |   40    |   0.549731   |     -      |     -     |   7.24   \n",
            "   1    |   60    |   0.505859   |     -      |     -     |   7.30   \n",
            "   1    |   80    |   0.515395   |     -      |     -     |   7.31   \n",
            "   1    |   92    |   0.500592   |     -      |     -     |   4.28   \n",
            "----------------------------------------------------------------------\n",
            "\n",
            "\n",
            " Epoch  |  Batch  |  Train Loss  |  Val Loss  |  Val Acc  |  Elapsed \n",
            "----------------------------------------------------------------------\n",
            "   2    |   20    |   0.463795   |     -      |     -     |   7.43   \n",
            "   2    |   40    |   0.479584   |     -      |     -     |   7.18   \n",
            "   2    |   60    |   0.420411   |     -      |     -     |   7.17   \n",
            "   2    |   80    |   0.424868   |     -      |     -     |   7.15   \n",
            "   2    |   92    |   0.429631   |     -      |     -     |   4.17   \n",
            "----------------------------------------------------------------------\n",
            "\n",
            "\n",
            "Training complete!\n"
          ]
        }
      ],
      "source": [
        "# Concatenate the train set and the validation set\n",
        "full_train_data = torch.utils.data.ConcatDataset([train_data, val_data])\n",
        "full_train_sampler = RandomSampler(full_train_data)\n",
        "full_train_dataloader = DataLoader(full_train_data, sampler=full_train_sampler, batch_size=32)\n",
        "\n",
        "# Train the Bert Classifier on the entire training data\n",
        "set_seed(42)\n",
        "bert_classifier, optimizer, scheduler = initialize_model(epochs=2)\n",
        "train(bert_classifier, full_train_dataloader, epochs=2)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 341,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "id": "LZUnDt4x-E-2",
        "outputId": "815a1b60-ecac-48f6-9234-4d6fd07b05d1"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "                      id                                             review\n",
              "820             19244237                             Neto was a great host.\n",
              "313             30554355  Great stay, everything looks exactly like the ...\n",
              "1355            51799070  Eduardo is the nicest and friendliest of hosts...\n",
              "1423  821188169022006967  Really nice apartment, in a quiet residence an...\n",
              "1070            42876407  The apartment is lovely and very close to a su..."
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-326433f1-e84c-49f0-ad8c-3a428d93d71f\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>id</th>\n",
              "      <th>review</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>820</th>\n",
              "      <td>19244237</td>\n",
              "      <td>Neto was a great host.</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>313</th>\n",
              "      <td>30554355</td>\n",
              "      <td>Great stay, everything looks exactly like the ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1355</th>\n",
              "      <td>51799070</td>\n",
              "      <td>Eduardo is the nicest and friendliest of hosts...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1423</th>\n",
              "      <td>821188169022006967</td>\n",
              "      <td>Really nice apartment, in a quiet residence an...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1070</th>\n",
              "      <td>42876407</td>\n",
              "      <td>The apartment is lovely and very close to a su...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-326433f1-e84c-49f0-ad8c-3a428d93d71f')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-326433f1-e84c-49f0-ad8c-3a428d93d71f button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-326433f1-e84c-49f0-ad8c-3a428d93d71f');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-4d4d3fec-40f9-4330-b106-43f747b43726\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-4d4d3fec-40f9-4330-b106-43f747b43726')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-4d4d3fec-40f9-4330-b106-43f747b43726 button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "    </div>\n",
              "  </div>\n"
            ]
          },
          "metadata": {},
          "execution_count": 341
        }
      ],
      "source": [
        "test_data.sample(5)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 403,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "L7rKCAtQ-E-2",
        "outputId": "aa0ca01e-85a7-480b-f723-7e435eb985d8"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Tokenizing data...\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/transformers/tokenization_utils_base.py:2614: FutureWarning: The `pad_to_max_length` argument is deprecated and will be removed in a future version, use `padding=True` or `padding='longest'` to pad to the longest sequence in the batch, or use `padding='max_length'` to pad to a max length. In this case, you can give a specific length with `max_length` (e.g. `max_length=45`) or leave max_length to None to pad to the maximal input size of the model (e.g. 512 for Bert).\n",
            "  warnings.warn(\n"
          ]
        }
      ],
      "source": [
        "# Run `preprocessing_for_bert` on the test set\n",
        "print('Tokenizing data...')\n",
        "test_inputs, test_masks = preprocessing_for_bert(test_data.review)\n",
        "\n",
        "# Create the DataLoader for our test set\n",
        "test_dataset = TensorDataset(test_inputs, test_masks)\n",
        "test_sampler = SequentialSampler(test_dataset)\n",
        "test_dataloader = DataLoader(test_dataset, sampler=test_sampler, batch_size=32)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 417,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0OEGtAik-E-2",
        "outputId": "1636bb89-20c5-4686-a049-bac2a2c93ad3"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Number of reviews predicted non-negative:  108\n"
          ]
        }
      ],
      "source": [
        "# Compute predicted probabilities on the test set\n",
        "probs = bert_predict(bert_classifier, test_dataloader)\n",
        "\n",
        "# Get predictions from the probabilities\n",
        "threshold = 0.9177  # manually adjusted to get predicted non-negative reviews = 105\n",
        "preds = np.where(probs[:, 1] > threshold, 1, 0)\n",
        "\n",
        "# Number of reviews predicted non-negative\n",
        "print(\"Number of reviews predicted non-negative: \", preds.sum())"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 418,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "yIowZV74-E-3",
        "outputId": "94f3b901-bc3a-4308-a1cd-2d469725ac72"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['HIGHLY RECOMMEND! We flew to NY For my 14yr old birthday celebration. Location is perfect! Very clean, and very modern! Host is AMAZING!!',\n",
              " 'Amazing stay with a very modern vibe. Definitely a great stay to take a special someone.',\n",
              " 'Great place in a quiet neighbourhood with a sense of a home. A pieceful oasis in the big city. Amazing hospitality by Lidia which was there to help at any time.',\n",
              " 'Excellent place, just as described. Great location in a nice and quiet neighborhood, easy communication with host. Everything you might need is on the studio. Clean and cozy, will definitely stay here again.',\n",
              " 'I absolutely loved my stay. The host was extremely communicative and helpful. The place is beautiful, spacious, clean, and super close to a train station. The rooftop offers and incredible view of the city and I would gladly book again.',\n",
              " \"Josie is the most gracious host.  I arrived with my two teenage boys, and thoughtful touches like milk, juice and little cereal boxes, were waiting for us !! The Loft is absolutely gorgeous.  Beautiful lights, candles, the most comfortable beds, spotlessly clean, and pristine white towels.  Our stay exceeded our expectations.  <br/>Josie's Restaurant recommendations, and amazing coffee, were an added bonus.  Great location, close to Subway.  Can't WAIT to return\",\n",
              " \"Francis was an amazing host! She was very responsive and helpful before and during my stay. I had the chance to check-in early and immediately felt at home. The place exceeded my expectations in terms of size - it's much larger in person. Additionally, it's extremely clean and located in a vibrant part of Williamsburg. There are many restaurants and bars nearby, and the Bedford Ave Station is just around the corner, providing easy access to Manhattan on the L train. Thank you for hosting me, Francis!\",\n",
              " 'The stay with Ebbie was amazing! She was very friendly and kind hearted, and recommended places to go around the city and good plans. She was very reactive whenever we had a question and we felt really at home at her place. The room is cosy and the bed very comfortable, with a lots of natural light and peace. <br/>Also, the appartment is very well located as it is only 4-5mins walk away from the Ralph Av subway station (line C), which brings us directly to Downtown Brooklyn and Manhattan! <br/>I highly recommend staying at her place and will definitely book there again next time we come. <br/>Thanks again for the wonderful stay ☺️',\n",
              " 'A beautiful, clean home and excellent communication is what we were promised and that is what we received. Gretchen is an excellent host. Looking forward to visiting again.',\n",
              " 'We had a fantastic stay here.  Place is very clean and looks just like the photos.  Marc is a great host and very responsive.  The area is a perfect location for access to everywhere.',\n",
              " 'Laura’s place is adorable. Nice and clean and a great location. She’s a spectacular host',\n",
              " \"A good trip that was made great by our perfect place to stay.  From booking to check-out, Daphne and Jonathan were top notch hosts who were responsive and attentive to details.  La Couronne did not disappoint either.  It is as advertised and more!  My group of five was in town for the Brooklyn Half Marathon.  Our hosts left a great welcome card and a bowl of fresh fruits which was ideal for a group of runners. The weather on race day was rainy and cold so after running 13.1 miles in those conditions, returning to a hot waterfall shower, heated bathroom floor, and the most comfortable beds was exactly what we needed.  I'd not only recommend La Couronne, we'd love to come back next year.  Thanks Daphne and Jonathan.\",\n",
              " \"Stacy's house was the perfect accommodation for a great week in New York. Very close to the subway, always clean and quiet. I can only recommend it!! Stacy was very very kind and helpful - always! -, which made me feel really comfortable!\",\n",
              " 'Gilda was the perfect host and responded quickly to any questions I had. Her home was comfortable and in a wonderful neighborhood. Easily walked to restaurants, stores and subway. Enjoyed our visit to the NYC area. ',\n",
              " \"We truly enjoyed our stay at Michele's place! The apartment was immaculate, communication was a breeze, and we were a 10 minute walk from two different subway stations. Everything we wanted from a week in New York.\",\n",
              " 'Loved everything about this airbnb. Location could not be better, and the apartment itself was tastefully designed, airy, clean, and well appointed. Can not wait to return!',\n",
              " 'The most amazing place to stay in NYC - comfortable, quiet & friendly. Jane is such an amazing host!',\n",
              " 'This home is absolutely beautiful and exactly as described. It is open, airy and bright with beautiful minimalistic decor. It is very clean and had everything I was looking for, all in a great location. Public transportation and good food is within walking distance. The shower has great water pressure and the floor has heated tiles. The kitchen is updated and has a large center island if you wanted to cook. The bed was cozy and comfortable. The plants in the shower and bedroom make you feel right at home. The host was easy to communicate with and was able to accommodate me even when booking the same day! Had a wonderful stay!',\n",
              " 'Absolutely loved this cute place! There were so many small details that made me feel so welcomed! Snacks, water, shampoo, etc - Ingrid was extremely accommodating and very welcoming! The location was perfect for getting around town and being close to lots of fun areas. So grateful for her hospitality and for being able to stay at this beautiful Airbnb!',\n",
              " '2nd time staying at Joyanns place. Wonderful host and amazing and comfy home close to everything! Highly recommend this home!']"
            ]
          },
          "metadata": {},
          "execution_count": 418
        }
      ],
      "source": [
        "output = test_data[preds==1]\n",
        "list(output.sample(20).review)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 420,
      "metadata": {
        "id": "8KIKpgNVqYOs"
      },
      "outputs": [],
      "source": [
        "torch.save(bert_classifier, '/content/drive/MyDrive/inputs/roberta-50split-set')"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### testing on standard BERT (not fine-tuned)"
      ],
      "metadata": {
        "id": "rt9pzh_098JH"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "roberta_classifier, optimizer, scheduler = initialize_model(epochs=2)\n",
        "\n",
        "\n",
        "# Compute predicted probabilities on the validation set\n",
        "probs = bert_predict(roberta_classifier, val_dataloader)\n",
        "\n",
        "# Evaluate the Bert classifier on the validation set\n",
        "evaluate_roc(probs, y_val)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 565
        },
        "id": "Ku6K4Zvg-EZ6",
        "outputId": "5c9dbe5d-9301-458f-a8e5-2b64456181fc"
      },
      "execution_count": 343,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "Some weights of RobertaModel were not initialized from the model checkpoint at roberta-base and are newly initialized: ['roberta.pooler.dense.bias', 'roberta.pooler.dense.weight']\n",
            "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "AUC: 0.4958\n",
            "Accuracy: 50.51%\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAkIAAAHHCAYAAABTMjf2AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAABv3UlEQVR4nO3dd3jTVRvG8W9aumih7GmlgExBNsiUJUNFUYQyZImgshQEZSiIA0QEQUBRZIgvshQUFVFBUZbKFmQpQ2S0LGnpps15/4gNLR00JW067s915bK/mScJNU/Pec45FmOMQURERCQPcnN1ACIiIiKuokRIRERE8iwlQiIiIpJnKRESERGRPEuJkIiIiORZSoREREQkz1IiJCIiInmWEiERERHJs5QIiYiISJ6lREjESQIDA+nXr5+rw8hzWrZsScuWLV0dxk29/PLLWCwWLl686OpQsh2LxcLLL7/slHudPHkSi8XC4sWLnXI/yf2UCEmOsHjxYiwWi/2RL18+ypYtS79+/Thz5oyrw8vWIiIiePXVV7nrrrvInz8//v7+NG/enCVLlpBTVtg5ePAgL7/8MidPnnR1KMnEx8ezaNEiWrZsSZEiRfDy8iIwMJD+/fuzc+dOV4fnFJ988gkzZ850dRhJZMeYJGfK5+oARBzxyiuvUL58eaKjo/nll19YvHgxW7Zs4cCBA3h7e7s0tiNHjuDmlr3+tggJCaFNmzYcOnSI7t27M3ToUKKjo/nss8/o27cv69atY+nSpbi7u7s61DQdPHiQSZMm0bJlSwIDA5Mc++6771wTFBAVFcUjjzzC+vXradGiBePGjaNIkSKcPHmSlStX8tFHH3Hq1Cluu+02l8XoDJ988gkHDhzg2WefzZT7R0VFkS+fY19HqcVUrlw5oqKi8PDwcGKEkpspEZIcpWPHjtSvXx+AJ554gmLFijF16lTWrl1Lt27dXBqbl5dXlj9ndHQ0np6eqSZgffv25dChQ6xZs4YHH3zQvn/48OGMHj2at956izp16vDCCy9kVciArZXK19fXKffy9PR0yn0yYvTo0axfv56333472RfyxIkTefvtt7M0HmMM0dHR+Pj4ZOnzZoTVaiU2NhZvb2+n/hFjsVhc/keR5DBGJAdYtGiRAcyOHTuS7P/qq68MYCZPnpxk/6FDh0yXLl1M4cKFjZeXl6lXr5754osvkt3333//Nc8++6wpV66c8fT0NGXLljW9e/c2Fy5csJ8THR1tJkyYYCpWrGg8PT3NbbfdZkaPHm2io6OT3KtcuXKmb9++xhhjduzYYQCzePHiZM+5fv16A5gvv/zSvu/06dOmf//+pkSJEsbT09NUr17dLFiwIMl1P/74owHMsmXLzPjx402ZMmWMxWIx//77b4rv2fbt2w1gHn/88RSPX7t2zVSqVMkULlzYREZGGmOMOXHihAHMtGnTzIwZM8ztt99uvL29TYsWLcz+/fuT3SM973PCZ7dp0ybz9NNPm+LFi5tChQoZY4w5efKkefrpp03lypWNt7e3KVKkiHn00UfNiRMnkl1/4+PHH380xhhzzz33mHvuuSfZ+7RixQrz2muvmbJlyxovLy/TunVr8+effyZ7DXPmzDHly5c33t7epkGDBubnn39Ods+U/PPPPyZfvnzm3nvvTfO8BBMnTjSA+fPPP03fvn2Nv7+/KViwoOnXr5+JiIhIcu7ChQtNq1atTPHixY2np6epVq2aeffdd5Pds1y5cub+++8369evN/Xq1TNeXl7m7bffdugexhizbt0606JFC+Pn52cKFChg6tevb5YuXWqMsb2/N7735cqVs1+b3t8PwAwZMsT873//M9WrVzf58uUza9assR+bOHGi/dywsDDzzDPP2H8vixcvbtq2bWt27dp105gS/g0vWrQoyfMfOnTIdO3a1RQrVsx4e3ubypUrm3HjxqX1kUkeoRYhydESakYKFy5s3/fHH3/QtGlTypYty5gxY/D19WXlypV07tyZzz77jIcffhiA8PBwmjdvzqFDh3j88cepW7cuFy9eZO3atZw+fZpixYphtVp58MEH2bJlC4MGDaJatWrs37+ft99+m6NHj/L555+nGFf9+vWpUKECK1eupG/fvkmOrVixgsKFC9O+fXvA1n119913Y7FYGDp0KMWLF+ebb75hwIABhIWFJWtpePXVV/H09GTUqFHExMSk2iLy5ZdfAtCnT58Uj+fLl4+ePXsyadIktm7dStu2be3HlixZwtWrVxkyZAjR0dHMmjWL1q1bs3//fkqWLOnQ+5xg8ODBFC9enAkTJhAREQHAjh072LZtG927d+e2227j5MmTvPfee7Rs2ZKDBw+SP39+WrRowfDhw3nnnXcYN24c1apVA7D/NzVvvPEGbm5ujBo1itDQUN5880169erFr7/+aj/nvffeY+jQoTRv3pwRI0Zw8uRJOnfuTOHChW/anfXNN98QFxdH79690zzvRt26daN8+fJMmTKF3bt38+GHH1KiRAmmTp2aJK4777yTBx98kHz58vHll18yePBgrFYrQ4YMSXK/I0eO0KNHD5588kkGDhxIlSpVHLrH4sWLefzxx7nzzjsZO3YshQoVYs+ePaxfv56ePXsyfvx4QkNDOX36tL2Fy8/PD8Dh348ffviBlStXMnToUIoVK5asmzPBU089xaeffsrQoUOpXr06ly5dYsuWLRw6dIi6deumGVNKfv/9d5o3b46HhweDBg0iMDCQY8eO8eWXX/L666+n74OT3MvVmZhIeiS0CmzYsMFcuHDB/PPPP+bTTz81xYsXN15eXuaff/6xn9umTRtTs2bNJH+RWq1W06RJE1OpUiX7vgkTJhjArF69OtnzWa1WY4wxH3/8sXFzczObN29OcnzevHkGMFu3brXvS9wiZIwxY8eONR4eHuby5cv2fTExMaZQoUJJWmkGDBhgSpcubS5evJjkObp37278/f3trTUJLR0VKlSw70tL586dDZBqi5ExxqxevdoA5p133jHGXP9r2sfHx5w+fdp+3q+//moAM2LECPu+9L7PCZ9ds2bNTFxcXJLnT+l1JLRkLVmyxL5v1apVSVqBEkutRahatWomJibGvn/WrFkGsLdsxcTEmKJFi5oGDRqYa9eu2c9bvHixAW7aIjRixAgDmD179qR5XoKEFqEbW+gefvhhU7Ro0ST7Unpf2rdvbypUqJBkX7ly5Qxg1q9fn+z89NzjypUrpkCBAqZRo0YmKioqybkJvwPGGHP//fcnaQVK4MjvB2Dc3NzMH3/8kew+3NAi5O/vb4YMGZLsvMRSiymlFqEWLVqYAgUKmL///jvV1yh5V/aq7BS5ibZt21K8eHECAgJ49NFH8fX1Ze3atfa/3i9fvswPP/xAt27duHr1KhcvXuTixYtcunSJ9u3b8+eff9pHmX322WfUqlUrWcsF2OoMAFatWkW1atWoWrWq/V4XL16kdevWAPz444+pxhoUFMS1a9dYvXq1fd93333HlStXCAoKAmw1HZ999hmdOnXCGJPkOdq3b09oaCi7d+9Oct++ffumqwbk6tWrABQoUCDVcxKOhYWFJdnfuXNnypYta99u2LAhjRo1Yt26dYBj73OCgQMHJivKTvw6rl27xqVLl7jjjjsoVKhQstftqP79+ydpLWvevDkAx48fB2Dnzp1cunSJgQMHJinU7dWrV5IWxtQkvGdpvb8peeqpp5JsN2/enEuXLiX5DBK/L6GhoVy8eJF77rmH48ePExoamuT68uXL21sXE0vPPb7//nuuXr3KmDFjktXVJPwOpMXR34977rmH6tWr3/S+hQoV4tdff+Xs2bM3PfdmLly4wM8//8zjjz/O7bffnuRYel6j5H7qGpMcZe7cuVSuXJnQ0FAWLlzIzz//nKRI+a+//sIYw0svvcRLL72U4j3Onz9P2bJlOXbsGF26dEnz+f78808OHTpE8eLFU71XamrVqkXVqlVZsWIFAwYMAGzdYsWKFbN/UVy4cIErV67wwQcf8MEHH6TrOcqXL59mzAkSvqCvXr1KoUKFUjwntWSpUqVKyc6tXLkyK1euBBx7n9OKOyoqiilTprBo0SLOnDmTZDj/jV/4jrrxSy8hufn3338B+PvvvwG44447kpyXL1++VLtsEitYsCBw/T10RlwJ99y6dSsTJ05k+/btREZGJjk/NDQUf39/+3Zq/x7Sc49jx44BUKNGDYdeQwJHfz/S+2/3zTffpG/fvgQEBFCvXj3uu+8++vTpQ4UKFRyOMSHxzehrlNxPiZDkKA0bNrSPGuvcuTPNmjWjZ8+eHDlyBD8/P6xWKwCjRo1K8a9kSP7Flxar1UrNmjWZMWNGiscDAgLSvD4oKIjXX3+dixcvUqBAAdauXUuPHj3sLRAJ8T722GPJaokS3HXXXUm20zsiqFq1anz++ef8/vvvtGjRIsVzfv/9d4B0/ZWeWEbe55TiHjZsGIsWLeLZZ5+lcePG+Pv7Y7FY6N69u/05Miq1KQGMk+ZOqlq1KgD79++ndu3a6b7uZnEdO3aMNm3aULVqVWbMmEFAQACenp6sW7eOt99+O9n7ktL76ug9MsrR34/0/tvt1q0bzZs3Z82aNXz33XdMmzaNqVOnsnr1ajp27HjLcYskpkRIcix3d3emTJlCq1atmDNnDmPGjLH/xejh4ZGk+DclFStW5MCBAzc9Z9++fbRp0yZDzehBQUFMmjSJzz77jJIlSxIWFkb37t3tx4sXL06BAgWIj4+/abyOeuCBB5gyZQpLlixJMRGKj4/nk08+oXDhwjRt2jTJsT///DPZ+UePHrW3lDjyPqfl008/pW/fvkyfPt2+Lzo6mitXriQ5LzO6MMqVKwfYWrdatWpl3x8XF8fJkyeTJaA36tixI+7u7vzvf/9zuGA6LV9++SUxMTGsXbs2SetRWt2wGb1HxYoVAThw4ECafyCk9v7f6u9HWkqXLs3gwYMZPHgw58+fp27durz++uv2RCi9z5fwb/Vmv+uSd6lGSHK0li1b0rBhQ2bOnEl0dDQlSpSgZcuWvP/++5w7dy7Z+RcuXLD/3KVLF/bt28eaNWuSnZfw13m3bt04c+YM8+fPT3ZOVFSUffRTaqpVq0bNmjVZsWIFK1asoHTp0kmSEnd3d7p06cJnn32W4v+oE8frqCZNmtC2bVsWLVrEV199lez4+PHjOXr0KM8//3yyv9Q///zzJDU+v/32G7/++qv9S8iR9zkt7u7uyVpoZs+eTXx8fJJ9CXMO3Zgg3Yr69etTtGhR5s+fT1xcnH3/0qVL7d1naQkICGDgwIF89913zJ49O9lxq9XK9OnTOX36tENxJbQY3dhNuGjRIqffo127dhQoUIApU6YQHR2d5Fjia319fVPsqrzV34+UxMfHJ3uuEiVKUKZMGWJiYm4a042KFy9OixYtWLhwIadOnUpyzFmtg5KzqUVIcrzRo0fTtWtXFi9ezFNPPcXcuXNp1qwZNWvWZODAgVSoUIGQkBC2b9/O6dOn2bdvn/26Tz/9lK5du/L4449Tr149Ll++zNq1a5k3bx61atWid+/erFy5kqeeeooff/yRpk2bEh8fz+HDh1m5ciXffvutvasuNUFBQUyYMAFvb28GDBiQbPLDN954gx9//JFGjRoxcOBAqlevzuXLl9m9ezcbNmzg8uXLGX5vlixZQps2bXjooYfo2bMnzZs3JyYmhtWrV7Np0yaCgoIYPXp0suvuuOMOmjVrxtNPP01MTAwzZ86kaNGiPP/88/Zz0vs+p+WBBx7g448/xt/fn+rVq7N9+3Y2bNhA0aJFk5xXu3Zt3N3dmTp1KqGhoXh5edG6dWtKlCiR4ffG09OTl19+mWHDhtG6dWu6devGyZMnWbx4MRUrVkxXi8P06dM5duwYw4cPZ/Xq1TzwwAMULlyYU6dOsWrVKg4fPpykBTA92rVrh6enJ506deLJJ58kPDyc+fPnU6JEiRSTzlu5R8GCBXn77bd54oknaNCgAT179qRw4cLs27ePyMhIPvroIwDq1avHihUrGDlyJA0aNMDPz49OnTo55ffjRlevXuW2227j0UcfpVatWvj5+bFhwwZ27NiRpOUwtZhS8s4779CsWTPq1q3LoEGDKF++PCdPnuTrr79m7969DsUnuZBLxqqJOCi1CRWNMSY+Pt5UrFjRVKxY0T48+9ixY6ZPnz6mVKlSxsPDw5QtW9Y88MAD5tNPP01y7aVLl8zQoUNN2bJl7ZPB9e3bN8lQ9tjYWDN16lRz5513Gi8vL1O4cGFTr149M2nSJBMaGmo/78bh8wn+/PNP+6RvW7ZsSfH1hYSEmCFDhpiAgADj4eFhSpUqZdq0aWM++OAD+zkJw8JXrVrl0Ht39epV8/LLL5s777zT+Pj4mAIFCpimTZuaxYsXJxs+nHhCxenTp5uAgADj5eVlmjdvbvbt25fs3ul5n9P67P7991/Tv39/U6xYMePn52fat29vDh8+nOJ7OX/+fFOhQgXj7u6ergkVb3yfUpto75133jHlypUzXl5epmHDhmbr1q2mXr16pkOHDul4d42Ji4szH374oWnevLnx9/c3Hh4eply5cqZ///5JhtYnDJ9PPFln4vcn8SSSa9euNXfddZfx9vY2gYGBZurUqWbhwoXJzkuYUDEl6b1HwrlNmjQxPj4+pmDBgqZhw4Zm2bJl9uPh4eGmZ8+eplChQskmVEzv7wf/TaiYEhINn4+JiTGjR482tWrVMgUKFDC+vr6mVq1aySaDTC2m1D7nAwcOmIcfftgUKlTIeHt7mypVqpiXXnopxXgkb7EYo7ZBEbE5efIk5cuXZ9q0aYwaNcrV4biE1WqlePHiPPLIIyl2+YhI7qIaIRHJs6Kjo5PViSxZsoTLly/TsmVL1wQlIllKNUIikmf98ssvjBgxgq5du1K0aFF2797NggULqFGjBl27dnV1eCKSBZQIiUieFRgYSEBAAO+88w6XL1+mSJEi9OnThzfeeMOlq9qLSNZxaY3Qzz//zLRp09i1axfnzp1jzZo1dO7cOc1rNm3axMiRI/njjz8ICAjgxRdfpF+/flkSr4iIiOQuLq0RioiIoFatWsydOzdd5584cYL777+fVq1asXfvXp599lmeeOIJvv3220yOVERERHKjbDNqzGKx3LRF6IUXXuDrr79OMvFc9+7duXLlCuvXr8+CKEVERCQ3yVE1Qtu3b082nX/79u159tlnU70mJiYmyWykVquVy5cvU7RoUa08LCIikkMYY7h69SplypRJNjHtrchRiVBwcDAlS5ZMsi9h/aaoqKgUF/SbMmUKkyZNyqoQRUREJBP9888/3HbbbU67X45KhDJi7NixjBw50r4dGhrK7bffzj///EPBggVdGJmIiIgkZgxERl7fdv/qC+Lb3Eukyc8dd4QBARQoUMCpz5mjEqFSpUoREhKSZF9ISAgFCxZMsTUIwMvLCy8vr2T7CxYsqERIREQkmzAGmjWDbdsgPxHMZQj9+Ij5PMEgrs/y7uyylhyVCDVu3Jh169Yl2ff999/TuHFjF0UkIiIityKhFSgiwpYE3ckBVtKN6hwiHjdOcTu25Rozh0uHz4eHh7N371776r8nTpxg7969nDp1CrB1a/Xp08d+/lNPPcXx48d5/vnnOXz4MO+++y4rV65kxIgRrghfREREbkFCK5CfH5QsaXicBfxGQ6pzCGup0sSu28iY8JcID7dw9mzmxODSFqGdO3fSqlUr+3ZCLU/fvn1ZvHgx586dsydFAOXLl+frr79mxIgRzJo1i9tuu40PP/yQ9u3bZ3nsIiIicmsiI22tQL6EM4+neIylAJh27XD7+GN8SpSwnxsfnzkxZJt5hLJKWFgY/v7+hIaGqkZIRETEhSIibK1BZTnNqSK1sYRegVdfw/LC83DDEPnM+v7OUTVCIiIikvuc4TZiFi/Dp7CPra8sC7m0RkhERETyoLAw6N4d9y8/t++ytr43y5MgUIuQiIiIZKVduyAoCI4dw+uHH/GhHVHkd1k4ahESERGRzGcMzJ4NTZrAsWNQrhzRK75waRIEahESERGRzHblCgwYAKtXA2A6dyZy9kIiPAu7Ni6UCImIiEhmunIF6tSBkyfBwwMz7S2arRjGtoDssfC5usZEREQk8xQqhOnQEWv5CkRt3MaFHsPZtj1pEtS0KeR3UQ+ZWoRERETEuS5dgrg4KFkSY6D13hnsPhFDWAv/JKeFhICvry0JcvISYummFiERERFxnm3boE4dTI8eRITFc+ECbPrFmzCSJkFNm0Lx4rZEyFVJEKhFSERERJzBaoVp02D8eIiP5/QFLxr7n+MMt9lPSWgBAte2AiWmREhERERuzYUL0LcvfPMNAHFde1B91fuEU8B+SkILUHZIfhJTIiQiIiIZt3kzdO8OZ89ivL2JnfYOoV2fIHyVLePJDnVAaVGNkIiIiGRMfDwMHgxnz3LSpyp3Rf+G97CBlCx1PePx9XV9HVBalAiJiIhIhhg3dyIXLCOq1xPUiNrBAWomOe7KYfHppa4xERERSb8ffoA//8QMepJmzWDbthrAfPvh7FgQnRYlQiIiInJz8fHwyivw6qvg7k509Xps21Y/ySnZtSA6LUqEREREJG1nz0KvXrBpk227Xz+sVavbD2f3gui0KBESERGR1H37LfTubRsi7+cH778PPXtCxPVTEgqicyIVS4uIiEjKXn4ZOnSwJUG1asGuXbYkKBdRIiQiIiIpK1TI9t+nnoJffoHKlV0aTmZQ15iIiIhcFxFxvZ/rmWegTh245x7XxpSJ1CIkIiIicO0ajB4NdevC1au2fRZLsiTIGFuuFBGRwj1yICVCIiIied3ff0OLFvDWW3D0KHz+eYqnGQPNmtlqpkuWzNoQM4sSIRERkbzsiy+gdm1bDZC/P3z2mW2UWAoiI2HbtqT7csLs0WlRjZCIiEheFBsLzz8Ps2bZths2hOXLoXz5dF2ek+cOSkwtQiIiInnRCy9cT4Kee862inw6kyDI/oupppdahERERPKiMWPg++9hyhTo1AljIPImBdC5pUA6MSVCIiIieUF0NKxZAz162LZLloTffwc3N3sR9I31P3mBEiEREZHc7s8/oVs32LvXtp2QDLnZKmRSKoJOS04vkE5MiZCIiEhutmwZDBoE4eFQrBgUKZLm6QlF0GnJ6QXSiSkREhERyY2iomwzQ8+fb9tu0QI++QTKlk3zspy8gGpGKBESERHJbQ4ftnWF7d9va7oZPx4mToR817/2jbF1iUHuLIJOLyVCIiIiuc2xY7YkqEQJWLoU2rZNcjgvF0ffSImQiIhIbnP//bYusfvvh9Klkx1OrTg6NxVBp5cmVBQREcnp/vgDmje3rRmW4IknUkyCbhQSYqujDg+3zamYW4qg00uJkIiISE5lDCxcCA0awJYt8OyzDt8ioTg6N8wSnRHqGhMREcmJwsPhqadsNUAA7drB+++7NqYcSC1CIiIiOc2+fVCvni0JcneHyZPhm29sxdHiELUIiYiI5CSbN8O990JMjG1OoOXLbUPA0iFhyHxeHi5/IyVCIiIiOUmDBlC1qi0J+ugj22zR6aAh8ylTIiQiIpLdHToElSvbusG8vWHDBttSGW7pr3BJach8XhwufyPVCImIiGRXxsCcOVC7Nrz++vX9xYqlKwkyxtYNlvBIkDBkPi8Ol7+RWoRERESyoytXYMAAWL3atr1vH1it6W4FSqsrLK+tJ5YWtQiJiIhkN7/9BnXq2JIgDw+YORM+/fSWu8JA3WE3UouQiIhIdmGMLel54QW4dg3Kl4cVK2wF0rcgJOR6C1D+/OoOS0wtQiIiItnFiRMwbpwtCerSBXbvvuUkCDR7dFrUIiQiIpJdVKgAc+dCVBQMHqysJQsoERIREXEVqxWmT7ctmHr33bZ9jz/u2pjyGCVCIiIirnDhAvTta1sao1w5OHAA/Pxu+baaPdoxqhESERHJaj//bJsb6JtvbBMkjh/vlPHsCUPm/fygZMlbDzMvUCIkIiKSVaxW28SIrVrB2bNQpQr8+isMHOiUeiDNHu04dY2JiIhkhfBweOQR+P5723bv3vDuu5nWHZYwZF7D5dOmFiEREZGs4OsLPj62x6JFsGSJ05KglLrDNFw+fdQiJCIiklni4yE21pb8WCy2BCg4GKpXv6XbJrQAga0VSN1hGadESEREJDOcOwc9e0LZsvDxx7ZEqEgR2+MWpLWGmLrDHKdESERExNm++w4ee8w2RN7XF44fh4oVnXLrtNYQK15cCZCjVCMkIiLiLHFxtqHwHTrYkqC77oKdO52WBN0oJMRWgx0eDps3KwnKCLUIiYiIOMPp07ausM2bbdtPPglvv22rD8okCQXRknFKhERERG6V1QodO9pmhy5QAObPh6Ag++HExc23SjNGO5cSIRERkVvl5gYzZ8KYMbBsGdxxh/1QWsXN4nqqERIREcmIU6dsRdEJ2rSxzRKdKAmC1Iubb5WGyDuHWoREREQctXYt9OtnK47evft68uNma1+4cZ6fBAnD251BQ+SdQ4mQiIhIesXGwgsv2LrBABo0gHxJv0rT6gpTcXP24/Kusblz5xIYGIi3tzeNGjXit99+S/P8mTNnUqVKFXx8fAgICGDEiBFER0dnUbQiIpJnnThhy3ASkqARI2DLFggMBGwJUESEbdR8avP8qCsr+3Fpi9CKFSsYOXIk8+bNo1GjRsycOZP27dtz5MgRSpQokez8Tz75hDFjxrBw4UKaNGnC0aNH6devHxaLhRkzZrjgFYiISJ7w2WcwYACEhkLhwrB4MTz4oP1waq1AibvC1JWVPbm0RWjGjBkMHDiQ/v37U716debNm0f+/PlZuHBhiudv27aNpk2b0rNnTwIDA2nXrh09evS4aSuSiIjILdm2zZYENW4Me/cmSYIg5YLohJmetfhp9uayRCg2NpZdu3bRtm3b68G4udG2bVu2b9+e4jVNmjRh165d9sTn+PHjrFu3jvvuuy/V54mJiSEsLCzJQ0RE5KaMuf7zlCkwaxb89BPcfnualyXM9qyZnnMGlyVCFy9eJD4+npIlSybZX7JkSYKDg1O8pmfPnrzyyis0a9YMDw8PKlasSMuWLRk3blyqzzNlyhT8/f3tj4CAAKe+DhERyYWWL4f77oNr12zbnp4wfDh4eNz0UrUA5SwuL5Z2xKZNm5g8eTLvvvsuu3fvZvXq1Xz99de8+uqrqV4zduxYQkND7Y9//vknCyMWEZEcJSrKtjRGjx6wfr1thuhEEgqiU3pIzuSyYulixYrh7u5OSEhIkv0hISGUKlUqxWteeuklevfuzRNPPAFAzZo1iYiIYNCgQYwfPx43t+R5nZeXF15eXs5/ASIikrscOQLdusHvv9uac8aNg0GD7Ic1Q3Tu5LIWIU9PT+rVq8fGjRvt+6xWKxs3bqRx48YpXhMZGZks2XF3dwfAJO7LFRERcYD5+H+YevXg998xxUsQ9cW3RIx9jYiYfPYWn9SGxSemIfI5j0uHz48cOZK+fftSv359GjZsyMyZM4mIiKB///4A9OnTh7JlyzJlyhQAOnXqxIwZM6hTpw6NGjXir7/+4qWXXqJTp072hEhERMQR5rXXsbz0IgA/0IpeF5YS/GDpNK9JbYZoDZHPeVyaCAUFBXHhwgUmTJhAcHAwtWvXZv369fYC6lOnTiVpAXrxxRexWCy8+OKLnDlzhuLFi9OpUydef/11V70EERHJ4aLuf5RrL73JDEbyGi9iJe0/rBOGxSvhyR0sJo/1KYWFheHv709oaCgFCxZ0dTgiIpKJEq/5lXin24HfsdasBdi6vaqVvMRliqZrLTC1+rhGZn1/a60xERHJlVIqbvYlnPd4mp58Qmt+4Gfu+e9IUdtxrQWW5ygREhGRHC3FVh9sLT2Jk6Ca/M5KulGVI8TjRg0OJEqEVOicV+WoeYREREQSS2j18fNL/rg+X68hdNoH7PNqSFWOYC1TlthvN/Fm+BDCw7E/NBN03qQWIRERybFSWuMrsQKEsbrYkxQcvdy2o2NH3JYswadYsawJULI9JUIiIpLtpdX9lSClQud8y77Aa+BycHe3rRf23HOQwuS7kncpERIRkWwtvTM6p1joPOAxOLgHuna1rRwvcgOlxSIiki0lrOvl0IzOV67A0KHw77+2AxYLzJihJEhSpRYhERHJdlJrBUpzRuedOyAoCE6cgIsXbSvIi9yEEiEREcl2UiqCTnVGZ2Ng1ix4/nm4dg3Kl7fVAomkgxIhERHJFhIXRKdUBJ3ijM6XL0P//rB2rW27Sxf48EMoVCgrQpZcQDVCIiLicjfOB3R9DqDrRdDJkqD9+6FOHVsS5OkJc+bAqlVKgsQhahESERGXSWgFunEW6ARpzvZcpoztBhUrwsqVULdupsYquZMSIRERcYn0FEQn6w67etXWZGSxQNGi8M03EBAAWkRbMkhdYyIi4hJpFUSn2B22eTNUqwaLF1/fd+edSoLkligREhERlwsJSWO9L6sVJk+GVq3gzBmYPRvi410Sp+Q+SoRERMTlUi2IPn8eOnSA8eNtyc9jj8HPP9uWzBBxAtUIiYhI9vTjj9CzJwQHg4+PbVRY//5aIl6cSomQiIhkP3//De3aQVwcVK9uGxV2552ujkpyISVCIiKS/ZQrB2PHwunTtpqglNbVEHECJUIiIpI9bNgAgYFwxx227UmT1A0mmU7F0iIi4lLuxOEx6UVbV1hQEMTE2A4oCZIsoBYhERHJVInXEEssIgLKcIZl9MBz2mbbzgYNbBeIZBElQiIikmlSmz0aoAPfsJc+FOcipkABLB98AN27Z32Qkqepa0xERDJNSrNH5+Mab/AC33AfxbnIEd86sHOXkiBxCSVCIiLidMbYur4iIq7vS5g9+splw6h6PwJwbdAQKl/YhqVyJRdFKnmdusZERMSpUusO881v8PW1gK8nrFoBu3fj0aWLa4IU+Y9ahERExKlu7A7zIJZlZUaS//Xx13eWLw9KgiQbUCIkIiKZ5sJvJ4iq35zuZ9/GMvUNOHzY1SGJJKFESEREMsXDrKbovXVw3/kbFCoEa9ZA1aquDkskCSVCIiLiXDExvMMwVtMFS2go3H037N0LDz3k6shEklEiJCIizmMM3g+2YxhzAIgd8Tz8/LNt7TCRbEiJkIiIOI/FQly/J7hIUe7ja669OhU8PFwdlUiqlAiJiMitiYqCQ4fsm3E9elOZo3zDfS4MSiR9lAiJiEjGHTliqwFq2xYuXLDv/pciLgxKJP00oaKIiCSR2iKpN3Jf/j+8nnkKS0QEplhxog+ewFq/eJLZpEWyOyVCIiJil9YiqQl8iGQ2wxjAQgB+pCW9Li7lXMsyWRSliPMoERIRyeMStwBFRKSdBFXjICvpRg3+wIqFV5jAq7yEFfdk5zZtCvnzZ1LQIk6iREhEJA9LqwUoJAR8fZPu8xw0FY9P/sBashQxC5YyumVrRqdy7/z5wWJxesgiTqVESEQkD7txXbAETZtC8eIpJDLvvgPe+XCbPBmfkiWzJEaRzKRRYyIiAthagMLDbY/Nm/9Lgvbvh9GjbU1HAP7+sGABKAmSXEItQiIiAti6wexdYcbA/A9h+HCIjoYqVeCJJ1wan0hmUCIkIiJJhYXBk0/C8uW27Y4dtU6Y5FrqGhMRkev27IF69WxJkLs7TJ0KX31lKxgSyYVuqUUoOjoab29vZ8UiIiIulG/ZxzDkCYiNhYAAWzLUpImrwxLJVA63CFmtVl599VXKli2Ln58fx48fB+Cll15iwYIFTg9QREScwxjbPEE3PhJYy5WH+Hjo1An27lUSJHmCw4nQa6+9xuLFi3nzzTfx9PS0769RowYffvihU4MTERHnSJgvyM8v6aNSyVD7OdYmzWD7dvjiCyiitcIkb3A4EVqyZAkffPABvXr1wt39+kyitWrV4vDhw04NTkREnCP5fEGG4cziJIFU4+D1WaAbNNAsiJKnOJwInTlzhjvuuCPZfqvVyrVr15wSlIiIZJ7zhy9z7YGHmcWzFOYKe59dfH3eIJE8xuFEqHr16mzevDnZ/k8//ZQ6deo4JSgREckcjfiFovfWId9XX4CnJ8yejeeMqUqCJM9yeNTYhAkT6Nu3L2fOnMFqtbJ69WqOHDnCkiVL+OqrrzIjRhERuVVWK88xgymMxe2fOKhYEVassA2VF8nDHG4Reuihh/jyyy/ZsGEDvr6+TJgwgUOHDvHll19y7733ZkaMIiJyi/It/x9vMRoP4oh7pBvs2qUkSASwGJOwgEzeEBYWhr+/P6GhoRQsWNDV4YiIZBpjbEXSABGhcewpez9reJjpV5/E1099YZKzZNb3t8MtQhUqVODSpUvJ9l+5coUKFSo4JSgREbk1Jt7K1EofUsQvBj8/KFk2Hx1Yz/s8papokUQcToROnjxJfHx8sv0xMTGcOXPGKUGJiMgtOH+e+PYdGXNsIFN5IdEBy/Vh8iICOFAsvXbtWvvP3377Lf7+/vbt+Ph4Nm7cSGBgoFODExERB23aBD17ku/cOSLx4XfuIiTk+qry+fOrQUgksXTXCLm52RqPLBYLN17i4eFBYGAg06dP54EHHnB+lE6kGiERyZXi4+H112HSJLBasVapRs0jqzjInYSHX0+ERHKqzPr+TneLkNVqBaB8+fLs2LGDYsWKOS0IERG5BcHB0KsX/PCDbbt/f6LemM3Bksp+RG7G4XmETpw4kRlxiIhIRkVGws6dtn6vefOgd2+IuPllIpKBRAggIiKCn376iVOnThEbG5vk2PDhw50SmIiIpMGY68U+FSrAypVQrhxUrerauERyGIcToT179nDfffcRGRlJREQERYoU4eLFi+TPn58SJUooERIRyWxnzsBjj8HYsdCunW1f+/bA9bmDItQiJJIuDg+fHzFiBJ06deLff//Fx8eHX375hb///pt69erx1ltvZUaMIiKSYP16qF3bNjps8GCIi7MfMgaaNcM2b1BJl0UokqM4nAjt3buX5557Djc3N9zd3YmJiSEgIIA333yTcePGZUaMIiJy7RqMGQMdO8LFi7ZkaN06yHe9YT8yErZtS3qZ5g0SSZvDiZCHh4d9KH2JEiU4deoUAP7+/vzzzz/OjU5EROCff6BlS5g61bY9eDBs3w6VK6d6SUgIhIfD5s2aN0gkLQ4nQnXq1GHHjh0A3HPPPUyYMIGlS5fy7LPPUqNGDYcDmDt3LoGBgXh7e9OoUSN+++23NM+/cuUKQ4YMoXTp0nh5eVG5cmXWrVvn8POKiOQIZ87YWn+2bYOCBWHVKpg7F7y907zM19f2UBIkkjaHE6HJkydTunRpAF5//XUKFy7M008/zYULF3j//fcduteKFSsYOXIkEydOZPfu3dSqVYv27dtz/vz5FM+PjY3l3nvv5eTJk3z66accOXKE+fPnU7ZsWUdfhohIzlC2LHTqBPXrw5498Oijro5IJFdx6erzjRo1okGDBsyZMwewTdoYEBDAsGHDGDNmTLLz582bx7Rp0zh8+DAeHh4Zek7NLC0i2d7Jk7aK54SJayMjwd0dvLzSvCwiwnYZoNmkJdfJNqvPp2b37t0OLa8RGxvLrl27aNu27fVg3Nxo27Yt27dvT/GatWvX0rhxY4YMGULJkiWpUaMGkydPTnER2AQxMTGEhYUleYiIZFtr1ti6wvr2hf9m9Cd//lSTIGNsCVDCQ0Qc41Ai9O233zJq1CjGjRvH8ePHATh8+DCdO3emQYMG9mU40uPixYvEx8dT8oYxniVLliQ4ODjFa44fP86nn35KfHw869at46WXXmL69Om89tprqT7PlClT8Pf3tz8CAgLSHaOISJaJiYHhw+GRRyA0FC5dsv03DYmHy2vIvEjGpDsRWrBgAR07dmTx4sVMnTqVu+++m//97380btyYUqVKceDAgUwvWrZarZQoUYIPPviAevXqERQUxPjx45k3b16q14wdO5bQ0FD7QyPbRCTbOXbMNs599mzb9qhRtuFehQvbT7mx5SciAi5cSD5cHjRkXsQR6Z5ZetasWUydOpXRo0fz2Wef0bVrV959913279/Pbbfd5vATFytWDHd3d0JCQpLsDwkJoVSpUileU7p0aTw8PHB3d7fvq1atGsHBwcTGxuLp6ZnsGi8vL7xu0q8uIuIyK1fCE0/A1atQtCh89BHcf3+SUxJaflJKehKEhFyvCcqfX6PFRNIr3S1Cx44do2vXrgA88sgj5MuXj2nTpmUoCQLw9PSkXr16bNy40b7ParWyceNGGjdunOI1TZs25a+//krSBXf06FFKly6dYhIkIpKtRUfblsm4etXWjLN3b7IkCFKeKDGxpk2heHENmRfJiHQnQlFRUeT/r63VYrHg5eVlH0afUSNHjmT+/Pl89NFHHDp0iKeffpqIiAj69+8PQJ8+fRg7dqz9/KeffprLly/zzDPPcPToUb7++msmT57MkCFDbikOERGX8PaGFStg3Djbkhnp+MMyYaLExA9NmiiScQ4tuvrhhx/i99/YzLi4OBYvXkyxhOGd/3Fk0dWgoCAuXLjAhAkTCA4Opnbt2qxfv95eQH3q1Cn7LNYAAQEBfPvtt4wYMYK77rqLsmXL8swzz/DCCy848jJERFznk09sTTxPPGHbrl/f9kinhFYfEXGOdM8jFBgYiOUmf3JYLBb7aLLsSvMIiYhLREbCM8/Ahx9iPD2J2rYXU7Vaui6NiLg+IkzzA0lelVnf3+luETp58qTTnlREJE85dAi6dYMDB7Bi4ZXYsbxavzLpn3BERDKLQ11jIiLioI8+si2SGhmJtURJ2p7/hB9pnaFbaVi8iPMpERIRcTJjIDLC4Dl0IB4fLQAgvlVbLr/zP36saevjSjzcPb00LF7E+ZQIiYg40fU5fyyMpQKv4sZEJjHlx7FYa16fA01FzyLZgxIhERFnMYbIs6Fs21YIgDcYw3o6sIe6SU5TF5dI9qFESETEGa5ehSefxGfffnz4lSjyExzihq9v3WSnqotLJPvI0Orzx44d48UXX6RHjx6cP38egG+++YY//vjDqcGJiOQIe/dCvXqwbBmWI4dowc/A9e6vGx9KgkSyD4cToZ9++omaNWvy66+/snr1asLDwwHYt28fEydOdHqAIiLZljHw3ntw993w558QEED0tz/zLR1cHZmIpJPDidCYMWN47bXX+P7775Os79W6dWt++eUXpwYnIpJthYZCUJBtaHxMDHTqBHv2YL27iasjExEHOJwI7d+/n4cffjjZ/hIlSnDx4kWnBCUiku0NHQqrVkG+fDB9OubzL4jwLkpEhKsDExFHOJwIFSpUiHPnziXbv2fPHsqWLeuUoEREsr0pU2x1QVu2YEaMpFlzC35+15fCEJGcweFEqHv37rzwwgsEBwdjsViwWq1s3bqVUaNG0adPn8yIUUTE9f791zZLdILbboMdO6BRIyIjYdu2pKdriLxIzuDw8PnJkyczZMgQAgICiI+Pp3r16sTHx9OzZ09efPHFzIhRRMS1fv3VVg/0999Eexci/oGH/jtgG/6VuDssYcZoDZEXyRnSvfr8jU6dOsWBAwcIDw+nTp06VKpUydmxZQqtPi8i6WYMzJgBY8ZAXBynvSvyUPQKdlMv1Uu0OrxI5nD56vMJtmzZQrNmzbj99tu5/fbbnRaIiEi2cukS9OsHX30FQPSD3ai+dj5XSf1/wOoOE8l5HK4Rat26NeXLl2fcuHEcPHgwM2ISEXGtrVuhdm346itiLF48xXv4rF1uT4JCQmwtPzc+Nm9Wd5hITuNwInT27Fmee+45fvrpJ2rUqEHt2rWZNm0ap0+fzoz4RESy3tmzcPo01jsq0cj8wvs8RUI9UNOmULy4ZowWyS0yXCMEcOLECT755BOWLVvG4cOHadGiBT/88IMz43M61QiJSIqMSZrJfPQREe0fwa90AUBF0CKullnf3xlaayxB+fLlGTNmDG+88QY1a9bkp59+clZcIiJZ56efbHMCJZ4jrW9fKFDAvqlWH5HcKcOJ0NatWxk8eDClS5emZ8+e1KhRg6+//tqZsYmIZK74eHj1VWjdGvbsgQkTXB2RiGQxh0eNjR07luXLl3P27FnuvfdeZs2axUMPPUR+DZUQkZwkOBgeeww2brRt9+sHM2e6MiIRcQGHE6Gff/6Z0aNH061bN4oVK5YZMYmIZK6NG6FXL1vhT/78thXkNTO+SJ7kcCK0devWzIhDRCRrrFkDXbrYiqNr1ICVK6FaNfthYyAy0vazFlAVyf3SlQitXbuWjh074uHhwdq1a9M898EHH3RKYCIimeLee6FKFWjeHGbNAh8f+yFjoFmz5OuGiUjula7h825ubgQHB1OiRAnc3FKvr7ZYLMTHxzs1QGfT8HmRPGjHDtuosIT/f4WGgr9/stMiIsDPL/nlTZtqskQRV3Pp8Hmr1UqJEiXsP6f2yO5JkIjkMXFxMHYsNGxoWzMsQQpJ0I0Szx6tJEgk93J4+PySJUuIiYlJtj82NpYlS5Y4JSgRkVv2zz/QsiW88YZt28HZ7zVjtEje4HAi1L9/f0JDQ5Ptv3r1Kv3793dKUCIit+Trr21rhW3dCgULwqpVGhovIilyOBEyxmBJ4c+j06dP45+O5mYRkUwTGwujRsEDD8Dly1C/vm2ixEcfdXVkIpJNpXv4fJ06dbBYLFgsFtq0aUO+fNcvjY+P58SJE3To0CFTghQRSZdDh+Cdd2w/P/MMTJ0KXl6ujUlEsrV0J0KdO3cGYO/evbRv3x6/REMrPD09CQwMpEuXLk4PUEQk3WrVgjlzoEQJ+O//WSIiaUl3IjRx4kQAAgMDCQoKwtvbO9OCEhFJl5gYGDcOeve21QQBDBrk0pBEJGdxeGbpvn37ZkYcIiKOOXYMgoJg1y746is4cAA8PFwdlYjkMOlKhIoUKcLRo0cpVqwYhQsXTrFYOsHly5edFpyISIpWrYInnoCwMChSxDZHkJIgEcmAdCVCb7/9NgUKFLD/nFYiJCKSaaKjYeRI2yKpYJvyedkyCAhwbVwikmOla4mN3ERLbIjkUBcuQLt2sHevbXvsWHjlFcjncA9/qhIvsREebptMUUSyB5cusZHY7t272b9/v337iy++oHPnzowbN47Y2FinBSYikkSRIlCsGBQvDuvXw+TJTk2CRCRvcjgRevLJJzl69CgAx48fJygoiPz587Nq1Sqef/55pwcoInlYZCRERdl+dneHpUttLULt27s0LBHJPRxOhI4ePUrt/4aprlq1invuuYdPPvmExYsX89lnnzk7PhHJqw4dgkaN4Nlnr+8rUQLKlHFZSCKS+2RoiQ2r1QrAhg0buO+++wAICAjg4sWLzo1ORPKmjz6yLY9x4AB88YWtPkhEJBM4nAjVr1+f1157jY8//piffvqJ+++/H4ATJ05QsmRJpwcoInlIRAT062d7REZCmza2rrDixV0cmIjkVg4nQjNnzmT37t0MHTqU8ePHc8cddwDw6aef0qRJE6cHKCJ5xIED0KCBrTXIzQ1efRW+/RZKlXJ1ZCKSizlt+Hx0dDTu7u54ZPNJzTR8XiQbio2FihXh9GlbDdAnn8A992TZ0xtja4CKiICEhm0NnxfJXjLr+zvDY0937drFoUOHAKhevTp169Z1WlAiksd4esK8eTB3rq1FKAu7woyBZs1g27Yse0oRyUYcToTOnz9PUFAQP/30E4UKFQLgypUrtGrViuXLl1Ncffkikh779sH583Dvvbbt+++H++6DTJq5PqHV50YREcmToKZNIX/+TAlDRLIZh2uEhg0bRnh4OH/88QeXL1/m8uXLHDhwgLCwMIYPH54ZMYpIbmKMrfWnUSPboqmnTl0/lolJULNmtlmjb3wkHuMREmLrEtu8OdNCEZFsxuEWofXr17NhwwaqVatm31e9enXmzp1Lu3btnBqciOQyoaEwaBCsXGnbvvfeLCnEiYy8eddX06a2HjklQCJ5i8OJkNVqTbEg2sPDwz6/kIhIMrt2QbducPy4bWmMqVNhxAh75pFa15UzRERc/zkkJOXcK39+JUEieZHDiVDr1q155plnWLZsGWX+m+H1zJkzjBgxgjZt2jg9QBHJBWbPhlGjbKPDypWDFStsXWP/ycqCZV9fjQYTkescrhGaM2cOYWFhBAYGUrFiRSpWrEj58uUJCwtj9uzZmRGjiOR0f/xhS4I6d4Y9e5IkQZC+ritnUBG0iNzI4RahgIAAdu/ezcaNG+3D56tVq0bbtm2dHpyI5GDGXO9revttaNIEeve+af9Tal1XzqDuLxG5kUOJ0IoVK1i7di2xsbG0adOGYcOGZVZcIpJTGWNLfL7/Hr76yrZqvI8P9OmT4qkJExkmUNeViGSldCdC7733HkOGDKFSpUr4+PiwevVqjh07xrRp0zIzPhHJSS5dsq0T9tVXtu3Vq6Fr1xRP1USGIpIdpLtGaM6cOUycOJEjR46wd+9ePvroI959993MjE1EcpJt26BOHVsS5OUF770Hjz6a6ukp1QWphkdEslq61xrz8fHh0KFDBAYGArZh9D4+Ppw8eZLSpUtnZoxOpbXGRJzMaoVp02D8eIiPh0qVMCtWElm5dpqXJV7XK6EuSDU8IpIal681FhMTg2+ijns3Nzc8PT2JiopyWjAikgMNH25bIwygZ0/Me/No1rGAQ11eqgsSEVdxqFj6pZdeIn+iduvY2Fhef/11/P397ftmzJjhvOhEJPsbNAiWLYM334THHycy0uJQEqTuMBFxpXQnQi1atODIkSNJ9jVp0oTjx4/bty1q0xbJ/eLjYefO63MB3XUXnDwJBQokOzU9Q+HVHSYirpTuRGjTpk2ZGIaI5AghIfDYY7BpE2zZcj0ZSiEJAnV5iUj25/CEiiKSdyRe/8tt0w94Pd4Tt/MhmPz5ifnrNPE1GiW7JvGcQCIi2Z3DS2yISN6QMM9PQb943vKbiNcDbXE7H8J+alA9cic+j3XBz49kj4SRYCIiOYFahEQkicSzPZ/YdpYN9KIVmwCYzxM8wyyiuHl1s4qgRSQnUCIkInY3zvY8hNW0YhNWXz9iZ79Pz2496ZnOe6kIWkRygmzRNTZ37lwCAwPx9vamUaNG/Pbbb+m6bvny5VgsFjp37py5AYrkETfO9jyXIXxSZhSWXbvw7t/TXvycnoeSIBHJCTKUCG3evJnHHnuMxo0bc+bMGQA+/vhjtmzZ4vC9VqxYwciRI5k4cSK7d++mVq1atG/fnvPnz6d53cmTJxk1ahTNmzfPyEsQydOMsXV93fiI+vM0i+iHH1cJCYHwcAs9Tk/DUqWyq0MWEckUDidCn332Ge3bt8fHx4c9e/YQExMDQGhoKJMnT3Y4gBkzZjBw4ED69+9P9erVmTdvHvnz52fhwoWpXhMfH0+vXr2YNGkSFSpUcPg5RfKyhO6vG4ucu/l9DXVq04+PmM5zatkRkTzB4UTotddeY968ecyfPx8PDw/7/qZNm7J7926H7hUbG8uuXbto27bt9YDc3Gjbti3bt29P9bpXXnmFEiVKMGDAgJs+R0xMDGFhYUkeInlRQivQhQtJu7/ycY03Gc3XPEAxLrGTemyo+4IKnUUkT3A4ETpy5AgtWrRItt/f358rV644dK+LFy8SHx9PyRvG25YsWZLg4OAUr9myZQsLFixg/vz56XqOKVOm4O/vb38EBAQ4FKNIbpC4FSjxr9uFnX8T3bAFo3kLgGtPD6fapa2s2FlRLUEikic4nAiVKlWKv/76K9n+LVu2ZHo31dWrV+nduzfz58+nWLFi6bpm7NixhIaG2h///PNPpsYokh3dWAQN8HSNzRRtWxv3336BQoVgzRo83p2FbxEvJUEikmc4PHx+4MCBPPPMMyxcuBCLxcLZs2fZvn07o0aN4qWXXnLoXsWKFcPd3Z2QkJAk+0NCQihVqlSy848dO8bJkyfp1KmTfZ/VarW9kHz5OHLkCBUrVkxyjZeXF15eXg7FJZKbJaz/lT+sEpY6XrZlMpYvh8BAV4cmIpLlHE6ExowZg9VqpU2bNkRGRtKiRQu8vLwYNWoUw4YNc+henp6e1KtXj40bN9qHwFutVjZu3MjQoUOTnV+1alX279+fZN+LL77I1atXmTVrlrq9RG6iCJfw9S1qW//Lt5RtzbAKFcDT09WhiYi4hMOJkMViYfz48YwePZq//vqL8PBwqlevjp+fX4YCGDlyJH379qV+/fo0bNiQmTNnEhERQf/+/QHo06cPZcuWZcqUKXh7e1OjRo0k1xcqVAgg2X4RSaoLn7KAAbh/+gH0DbLtrFrVtUGJiLhYhmeW9vT0pHr16rccQFBQEBcuXGDChAkEBwdTu3Zt1q9fby+gPnXqFG5u2WLeR5GcKToazxHP8SnvAhD3yUfQp5vGxYuIABZjjHHkglatWmFJ43+gP/zwwy0HlZnCwsLw9/cnNDSUggULujockcz155/QrRvs3QvAFMYw/N9X8C3kkfZ1IiLZTGZ9fzvcIlS7du0k29euXWPv3r0cOHCAvn37OisuEblF5pNl8OQgLOHhWIsW475LH/MtHRiuHEhExM7hROjtt99Ocf/LL79MeHj4LQckIrfO7PsdSy/b8qg/0YKelz7hLGVdHJWISPbjcNdYav766y8aNmzI5cuXnXG7TKOuMcmJjLHNBZReERHwUcnRROHDK0wg/r+/eZo2hc2bVR4kIjlPtukaS8327dvx9vZ21u1E5D8Js0LfOCHijXqylM005x9u/2/Pm4DFPm8QQP78SoJERBJzOBF65JFHkmwbYzh37hw7d+50eEJFEbm5lGaFTiw/EcxmGI+ziK00oSWbiMMDsNC0KRQvruRHRCQ1DidC/v7+Sbbd3NyoUqUKr7zyCu3atXNaYCJ5WeKusIiI6/sTt+4AWA7+gXefbrgdPohxc6PB2PZced4N3G3H1QIkIpI2hxKh+Ph4+vfvT82aNSlcuHBmxSSSp6XVFebr+18iZAwsWgRDh0JUFJQujeWTT/Bs2RLNES0ikn4OzVTo7u5Ou3btHF5lXkRuzhhb68+FCyknQU2b2lp4iIiAPn1gwABbEtS+vW2eoJYtszhiEZGcz+Epm2vUqMHx48czIxaRPCuhFcjPD/6bVB2wdYWFh9se9tFebm7w++/g7g5TpsC6dVCihMtiFxHJyRyuEXrttdcYNWoUr776KvXq1cM3ccECaEi6SAakVBCdpNDZGNvD4gY+PrBypa3pqFkzl8QrIpJbpHseoVdeeYXnnnuOAgUKXL84URWmMQaLxUJ8fLzzo3QizSMk2cWNBdEJLUEJBdH2QufQUBg0CGrWhBdfdFm8IiKulFnf3+lOhNzd3Tl37hyHDh1K87x77rnHKYFlFiVCkh2kVRAdHp5oZNiuXRAUBMeOgbc3HD8OpUtnaawiItmByydUTMiXsnuiI5KdJbQCRUTcpCDaGJgzB0aNgthYKFcOli9XEiQi4mQO1Qilteq8iKQttVagZDM/h16xjQhbvdq2s3NnWLgQNGWFiIjTOZQIVa5c+abJUHZfa0zEVW5aEA0QFwdNmsChQ+DhAW+9BcOGaVZEEZFM4lAiNGnSpGQzS4tI+hZFTWmG6GQzP+fLB888A2++CStWQP36mRKviIjYpLtY2s3NjeDgYErk8PlKVCwtzpbeRVETS1IQffkynDsHd955/YaRkUnX0hARyeNcXiyt+iCR624c+u5IEmQviAbbhd272yZH3LMHChWyNREpCRIRyRIOjxoTyevSagG6cVHUlOTPDxZjhTenwfjxEB8PlSrB+fO2REhERLJMuhMhq9WamXGI5BgpFT1DCoXPqblwAfr2hW++sW336AHvvw+JJisVEZGs4fASGyJ5VeI5gBIkG/p+syTo559tic/Zs7YJEmfPtg2VV9eziIhLKBESSYfUusN8fR0s55kxw5YEVa1qWy+sZk2nxikiIo5RIiSSDqnNAWQvek6vBQugQgV45RXbUvMiIuJSbq4OQCSnCQmxDX/fvDkdPVo//ADPPWdrUgIoWtTWKqQkSEQkW1CLkIiD0tUdFh9va/V59VVbEtSoEXTrliXxiYhI+ikREklDSgXSN3X2LPTqBZs22bYHDIAHHsiM8ERE5Bapa0wkFQkF0n5+ULJkOi/67juoXduWBPn6wv/+Bx9+mIFiIhERyQpKhERS4XCB9LRp0KGDbZ6gWrVg925by5CIiGRb6hoTSYdUF0lNrE4d23+fftpWEO3tnWXxiYhIxigREkmHVAukz5+HhIWI27aF/fuvL54qIiLZnrrGRDLi2jUYPRoqV4Zjx67vVxIkIpKjKBEScdTff0Pz5vDWWxAaCl9+6eqIREQkg9Q1JuKIzz+H/v3hyhXw94eFC+GRR1wdlYiIZJBahETSIzYWnn0WHn7YlgQ1bAh79igJEhHJ4ZQIiaRDvvfnwKxZto2RI23ra5Qv79qgRETklqlrTCQd4p4citdP38PgwdCpk6vDERERJ1EiJJKS6GjyzX6XfAwjDg/w9IRvvnF1VCIi4mTqGhNJxBiI3Pcn8Xc3wWvsc7zCBFeHJCIimUiJkMh/jIGJVZcTX7su7vv2cIFi/EwLV4clIiKZSF1jIgBRUcQNeZZXjn4AwM80pwfLOEvZtNcXExGRHE2JkOQ5xtgWVE1g+fMo3o89iscf+7Fi4XXG8+SZiRz1t/16pLm+mIiI5GhKhCRPMQaaNUu6qnwVrOzkOBGUoBdL2UhbRvqnsraYiIjkKkqEJE+JjLQlQRasmP9K5I5QlUdYzX5qEkxpdYWJiOQhSoQkz6nOHyyjBwGfz8GzbUIxdDv7cXWFiYjkHUqEJEe5sb7H0Yvj5y9kB8PITxTxrzyH+4O/KesREcnDlAhJjpFSfU96+XGV93iax1gKwLe0o9mnH+OrJEhEJE9TIiQ5RkJ9j6PuYh8r6UYVjhKHOy/xKluavMDPgZpGS0Qkr1MiJDlSSEj6RnVZDh/Cp2kjLDExWMuU5dri5bzYpJnqgEREBFAiJDmUr286h7fXrQoPPggREbh99BE+xYplemwiIpJzKBGSbC+hQDoiIp0X7NkD5ctDoUK2Zp+PPgIvL3BTV5iIiCSlbwbJ1hIKpP38oGTJdJw8Zw7cfTc88YRtG8DHR0mQiIikSC1Ckq2lVCCd4oSHV67AgAGwerVtOy4OoqNtSZCIiEgqlAhJjpFQIJ2s0Pm33yAoCE6eBA8PmDYNhg9XNbSIiNyUEiHJMZIVSBsDM2fCCy/AtWu2uqAVK6BBA1eFKCIiOYwKJyTnCg2FGTNsSVCXLrB7t5IgERFxiFqEJOcqVAiWLYN9+2DwYHWFiYiIw5QISc5htcKbb0GpUtCnj21fs2a2h4iISAYoEZJsIbXFVBPmDirGBbwe7QvffWOrlm7VCgICsjZIERHJdZQIicvdbDHVZmxmOd3J991Z8Pa2FUjfdluWxigiIrmTiqXF5VJbTNWClXG8ziZaUpazmCpV4NdfYeBA1QOJiIhTqEVIshX7Yqrx8Xh1uZ98G74FwPTujeXdd21TTIuIiDiJWoQkW0mYK8i3oDv5GtW31QMtWoRlyRIlQSIi4nRKhMRljLEVQycURLsRDxcuXD/h5Zdh717o188F0YmISF6QLRKhuXPnEhgYiLe3N40aNeK3335L9dz58+fTvHlzChcuTOHChWnbtm2a50v2dONiqqU4x/fci/cjHSEmxnZSvnxQqZJrAxURkVzN5YnQihUrGDlyJBMnTmT37t3UqlWL9u3bc/78+RTP37RpEz169ODHH39k+/btBAQE0K5dO86cOZPFkYujErcAXbhwvUD6Xr5jH7VozY+4HT1smyBRREQkC1iMMcaVATRq1IgGDRowZ84cAKxWKwEBAQwbNowxY8bc9Pr4+HgKFy7MnDlz6JMwyV4awsLC8Pf3JzQ0lIIFC95y/JI+KQ2RdyeOSUxknGUKFmMwd92FZeVKqFLFdYGKiEi2lFnf3y5tEYqNjWXXrl20bdvWvs/NzY22bduyffv2dN0jMjKSa9euUaRIkRSPx8TEEBYWluQhWe/GIfJlOc0PtGY8k7EYA08+ieWXX5QEiYhIlnJpInTx4kXi4+MpWbJkkv0lS5YkODg4Xfd44YUXKFOmTJJkKrEpU6bg7+9vfwRoNmKXCwmBk/cOpAWbMQUKwPLlMG8e+Pi4OjQREcljXF4jdCveeOMNli9fzpo1a/D29k7xnLFjxxIaGmp//PPPP1kcpdzI1xfyzZsLrVph2b0bgoJcHZKIiORRLp1QsVixYri7uxMSEpJkf0hICKVKlUrz2rfeeos33niDDRs2cNddd6V6npeXF15eXk6JVzLO8s8pBvAdC3jCtqNCBfjhB9cGJSIieZ5LW4Q8PT2pV68eGzdutO+zWq1s3LiRxo0bp3rdm2++yauvvsr69eupX79+VoQqt2LtWnya1OYDBnEv37k6GhERETuXL7ExcuRI+vbtS/369WnYsCEzZ84kIiKC/v37A9CnTx/Kli3LlClTAJg6dSoTJkzgk08+ITAw0F5L5Ofnh59mHs5eYmPhhRdg5kwswA4a8CeaF0hERLIPlydCQUFBXLhwgQkTJhAcHEzt2rVZv369vYD61KlTuLldb7h67733iI2N5dFHH01yn4kTJ/Lyyy9nZejyH2Nso8ISs5w8gVffINx37QAg8skRNHv/Da7h6YIIRUREUubyeYSymuYRcq6U5gd6iM9ZTD8KEcplCtOPxXzJg/bj4eH/LawqIiKSTpn1/e3yFiHJvlJq6blRRETSJAigIGEUIpRtNKY7y/mH2+3Hmja1raMqIiKSHSgRkhSl1NKTFjfiORfi/l9LTx+iP/Om1oMPc8jDI8l5+fODxeL0cEVERDIkR88jJJnnxpmg0xLEcv7yqUlxy0V8fW3dXt59uuFbyMO+nfBQEiQiItmJEiG5qZAQW11PsseFKGL7P8lyelA+6hCWt2e4OlQRERGHqGtMbiqhNSeJw4ehWzfYv9/WzDNuHGjUnoiI5DBKhMRxH38MTz9tq5QuUQL+9z+4915XRyUiIuIwJULimPffh6eesv3cqhUsXQqlS7s2JhERkQxSjZA4pnt3uOMOWzfY998rCRIRkRxNLUJyEwa3H3+A+1vbaoH8/eH338HHx9WBiYiI3DK1CEmqfAnnI/ri06ktzJt3/YCSIBERySXUIiQpshz4nZ10oypHMG5uWCIiXB2SiIiI06lFSJIyBt5/H597GlKVI5ymLNHfbIJRo1wdmYiIiNMpEZLrwsKgRw946iksMTGsoyO12Yu1aXNXRyYiIpIp1DWWg6VnUVRHuO04gPeqVeDuTsT4KTzwynMY5coiIpKLKRHKoRxdFDV9mvAUc9hLbX55pbEzbywiIpIt6c/9HMqRRVFT488VltCbqhyy75vH0/zC9SSoaVPbivEiIiK5kVqEcoGQkBTWArsJt1078OobhNvJE/SsfZDozTtTXBo+f36tGC8iIrmXEqFcIMVFUVNjDMyaBc8/D9euQWAg7h/Mw9dP2Y6IiOQ9SoTyksuXoX9/WLvWtv3II7BgARQq5NKwREREXEWJUF5x4gS0bAmnToGnJ8yYAYMHq99LRETyNCVC2VB6hsU7PNFzQADcfjt4eMDKlVC3bobjExERyS2UCGUzTh0Wf+kSFChgawHKlw9WrbJVPxcs6ISbi4iI5HwaPp/NODosPtXh7Zs3Q61a8MIL1/eVKqUkSEREJBG1CGUhR7u80jMsPtnwdqsV3ngDJkyA+HhYvx5ee83x8fUiIiJ5gBKhLJKRLi+HhsUDnD8Pjz0G339v237sMXjvPSVBIiIiqVAilIkStwBFRDipyys1P/4IPXtCcDD4+MDcudCvn0aFiYiIpEGJUCZJqwUoQ11eaQkLgy5d4N9/oXp126iwO+90OGYREZG8RolQJkmt6LlpUyhe3MkNNQULwvvvwzffwOzZ6goTERFJJyVCWSBxC5DT1u7asAHc3KB1a9t21662h4iIiKSbhs9ngYSiZ19fJyRBcXHw4ovQrh306AHnzjklRhERkbxILUI5yZkztuRn82bbdufOWidMRETkFigRyim++Qb69IGLF8HPD+bPh+7dXR2ViIhIjqauMSczxjZU3uG1wFJjtdpmh77vPlsSVKcO7N6tJEhERMQJlAg5UcKQeT8/KFnSSTd1c7PNDQQwZIhtKFqlSk66uYiISN6mrjEnSmnIvMMTIyaIi7MtlAq2yRG7doUHHrjlGEVEsjNjDHFxccTHx7s6FHEBDw8P3N3ds/Q5lQhlkoQh8w4Pl4+NhTFj4K+/4IsvbBf7+SkJEpFcLzY2lnPnzhF5s0UZJdeyWCzcdttt+Pn5ZdlzKhHKJA6vEwZw4gQEBcGOHbbtTZugVStnhyYiku1YrVZOnDiBu7s7ZcqUwdPTE4uWCMpTjDFcuHCB06dPU6lSpSxrGVIilF2sXg2PPw6hobYh8YsXKwkSkTwjNjYWq9VKQEAA+TNUTyC5QfHixTl58iTXrl3LskRIxdKuFhMDw4bZ1goLDYW774a9e+Ghh1wdmYhIlnNz09dSXuaKVkD9i3O1Xr1gzhzbz6NHw88/Q7lyro1JREQkj1Ai5GovvAClS8NXX8Gbb4KHh6sjEhERyTOUCGW1qCj46afr2w0awPHjcP/9rotJRERuyfbt23F3d+f+FP5fvmnTJiwWC1euXEl2LDAwkJkzZybZ9+OPP3LfffdRtGhR8ufPT/Xq1Xnuuec4c+ZMJkUP0dHRDBkyhKJFi+Ln50eXLl0ICQlJ85p+/fphsViSPDp06JDknMuXL9OrVy8KFixIoUKFGDBgAOHh4Zn2OjJCiVBWOnLEVgPUvr2tDiiBt7fLQhIRkVu3YMEChg0bxs8//8zZs2czfJ/333+ftm3bUqpUKT777DMOHjzIvHnzCA0NZfr06U6MOKkRI0bw5ZdfsmrVKn766SfOnj3LI488ctPrOnTowLlz5+yPZcuWJTneq1cv/vjjD77//nu++uorfv75ZwYNGpRZLyNDNGosqyxdCk8+aVt7o3hxSOEvAxERyXnCw8NZsWIFO3fuJDg4mMWLFzNu3DiH73P69GmGDx/O8OHDefvtt+37AwMDadGiRYotSs4QGhrKggUL+OSTT2jdujUAixYtolq1avzyyy/cfffdqV7r5eVFqVKlUjx26NAh1q9fz44dO6hfvz4As2fP5r777uOtt96iTJkyzn8xGaAWocwWGQlPPAGPPWZLglq2tLUGtWzp4sBERLK3xGs3ZuXDGMfiXLlyJVWrVqVKlSo89thjLFy4EOPoTYBVq1YRGxvL888/n+LxQoUKpXptx44d8fPzS/Vx5513pnrtrl27uHbtGm3btrXvq1q1Krfffjvbt29PM+ZNmzZRokQJqlSpwtNPP82lS5fsx7Zv306hQoXsSRBA27ZtcXNz49dff03zvllJLUKZ6eBB6NYN/vjDNkP0hAnw0kuQxdOHi4jkRJGRton1s1p4uGMT4i5YsIDHHnsMsHUVhYaG8tNPP9HSwT94//zzTwoWLEjp0qUdug7gww8/JCoqKtXjHmkMxAkODsbT0zNZolWyZEmCE9a6TEGHDh145JFHKF++PMeOHWPcuHF07NjRXi8VHBxMiRIlklyTL18+ihQpkuZ9s5oSocz0xRe2JKhUKVvX2H9NjiIikjscOXKE3377jTVr1gC2L/qgoCAWLFjgcCJkjMnwPDply5bN0HW3onv37vafa9asyV133UXFihXZtGkTbdq0yfJ4MkqJUGZ6/nlbO+uwYU5cjl5EJG/In9/WOuOK502vBQsWEBcXl6TexRiDl5cXc+bMwd/fn4IFCwK2WpwbW12uXLmCv78/AJUrVyY0NJRz58453CrUsWNHNm/enOrxcuXK8ccff6R4rFSpUsTGxnLlypUk8YWEhKRa/5OSChUqUKxYMf766y/atGlDqVKlOH/+fJJz4uLiuHz5skP3zWxKhJzIcmA/K3mFPiwBfGxdYK+95uqwRERyJIslA2s2ZqG4uDiWLFnC9OnTadeuXZJjnTt3ZtmyZTz11FNUqlQJNzc3du3aRblEE+YeP36c0NBQKleuDMCjjz7KmDFjePPNN5MUSye4MVFJ7Fa6xurVq4eHhwcbN26kS5cugK2l69SpUzRu3DjV6250+vRpLl26ZE/iGjduzJUrV9i1axf16tUD4IcffsBqtdKoUaN03zfTmTwmNDTUACY0NNR5N7VajfngA2P19jYGzBs8b8LDnXd7EZHcLioqyhw8eNBERUW5OpR0W7NmjfH09DRXrlxJduz555839evXt28PGjTIBAYGmi+++MIcP37c/PTTT+buu+82d999t7Farfbz5s6daywWi3n88cfNpk2bzMmTJ82WLVvMoEGDzMiRIzPttTz11FPm9ttvNz/88IPZuXOnady4sWncuHGSc6pUqWJWr15tjDHm6tWrZtSoUWb79u3mxIkTZsOGDaZu3bqmUqVKJjo62n5Nhw4dTJ06dcyvv/5qtmzZYipVqmR69OiRahxp/TvIlO9vY4wSoVu/obF2726MbaCBWUcHU4zzSoRERByQExOhBx54wNx3330pHvv1118NYPbt22eMsb2+iRMnmqpVqxofHx9Tvnx5M2jQIHPhwoVk137//femffv2pnDhwsbb29tUrVrVjBo1ypw9ezbTXktUVJQZPHiwKVy4sMmfP795+OGHzblz55KcA5hFixYZY4yJjIw07dq1M8WLFzceHh6mXLlyZuDAgSY4ODjJNZcuXTI9evQwfn5+pmDBgqZ///7m6tWracaR1YmQxZgMjPHLwcLCwvD39yc0NNTeb5the/ZgunXD8tdfxOHOOCbzFqMwuDk86kBEJC+Ljo7mxIkTlC9fHm9NMptnpfXvwKnf34moRiij1qyB7t2xxMZyigC6s5ztNAGgaVPHiu1ERETENZQIZVT9+uDnR9zdTamzbhGXKUpIiK0VKH9+W5GfiIiIZG9KhBxx5gwkzNUQEAC//UZMyQpcLmDLenx91R0mIiKSk2iJjfQwBmbNggoVYO3a6/srVlTTj4iISA6mROhmLl+Ghx+GZ5+F2NikiZCIiIjkaOoaS8svv0BQEJw6hfH0JHbydOKeHAIR10+JiEj9chERcUweG8gsN3DF568WoZRYrfDWW9C8uS0JqliRAVW34T1qKH4FLPj5YX9o5QwRkVuXMPNxZGSkiyMRV4qNjQXAPQsXJ1eLUEp+/hlGj7b93K0bkTPns6hM2nMWaMi8iEjGubu7U6hQIfvaVPnz58/wAqSSM1mtVi5cuED+/PnJly/r0hMlQokYA5GRQIOWeA5+BmvlqsQNeJKIyOu/jAlD5G+kIfMiIrcmYSHOGxfqlLzDzc2N22+/PUuTYM0sDWC1YmbO4sFlPfhqZ9or4mrGaBGRzBUfH8+1a9dcHYa4gKenJ25uKVft5OqZpefOncu0adMIDg6mVq1azJ49m4YNG6Z6/qpVq3jppZc4efIklSpVYurUqdx3330Ze/Lz56F3byzffcezfMXXfI9JpXRK3V8iIpnP3d09S2tEJG9zebH0ihUrGDlyJBMnTmT37t3UqlWL9u3bp9o0um3bNnr06MGAAQPYs2cPnTt3pnPnzhw4cMDxJ9+0CWrXhu++w/j4sJReGCyEhNhafm58bN6s7i8REZHcxOVdY40aNaJBgwbMmTMHsBVLBQQEMGzYMMaMGZPs/KCgICIiIvjqq6/s++6++25q167NvHnzbvp89qa1sWMpOHWqbYRYtWpELl6Jb6MagLq/REREspvM6hpzaYtQbGwsu3btom3btvZ9bm5utG3blu3bt6d4zfbt25OcD9C+fftUz0/VlClgtXKtd38iNu0gPLCGw/GLiIhIzubSGqGLFy8SHx9PyRsm4ylZsiSHDx9O8Zrg4OAUzw8ODk7x/JiYGGJiYuzboaGhtvvgzUBmsfLj7vBxPBBmPycsDOLjM/KKREREJDOEhdm+p53dkZUtiqUz05QpU5g0aVKy/VWIBp7875FUmTKZH5eIiIg47tKlS/j7+zvtfi5NhIoVK4a7uzshISFJ9oeEhNjnk7hRqVKlHDp/7NixjBw50r595coVypUrx6lTp5z6RorjwsLCCAgI4J9//nFqf69kjD6P7EOfRfahzyL7CA0N5fbbb6dIkSJOva9LEyFPT0/q1avHxo0b6dy5M2Arlt64cSNDhw5N8ZrGjRuzceNGnn32Wfu+77//nsaNG6d4vpeXF15eXsn2+/v76x91NlGwYEF9FtmIPo/sQ59F9qHPIvtIbZ6hjHJ519jIkSPp27cv9evXp2HDhsycOZOIiAj69+8PQJ8+fShbtixTpkwB4JlnnuGee+5h+vTp3H///SxfvpydO3fywQcfuPJliIiISA7k8kQoKCiICxcuMGHCBIKDg6lduzbr16+3F0SfOnUqSfbXpEkTPvnkE1588UXGjRtHpUqV+Pzzz6lRQ6O+RERExDEuT4QAhg4dmmpX2KZNm5Lt69q1K127ds3Qc3l5eTFx4sQUu8ska+mzyF70eWQf+iyyD30W2UdmfRYun1BRRERExFVcvsSGiIiIiKsoERIREZE8S4mQiIiI5FlKhERERCTPypWJ0Ny5cwkMDMTb25tGjRrx22+/pXn+qlWrqFq1Kt7e3tSsWZN169ZlUaS5nyOfxfz582nevDmFCxemcOHCtG3b9qafnTjG0d+NBMuXL8disdgnPpVb5+hnceXKFYYMGULp0qXx8vKicuXK+n+Vkzj6WcycOZMqVarg4+NDQEAAI0aMIDo6Oouizb1+/vlnOnXqRJkyZbBYLHz++ec3vWbTpk3UrVsXLy8v7rjjDhYvXuz4E5tcZvny5cbT09MsXLjQ/PHHH2bgwIGmUKFCJiQkJMXzt27datzd3c2bb75pDh48aF588UXj4eFh9u/fn8WR5z6OfhY9e/Y0c+fONXv27DGHDh0y/fr1M/7+/ub06dNZHHnu5OjnkeDEiROmbNmypnnz5uahhx7KmmBzOUc/i5iYGFO/fn1z3333mS1btpgTJ06YTZs2mb1792Zx5LmPo5/F0qVLjZeXl1m6dKk5ceKE+fbbb03p0qXNiBEjsjjy3GfdunVm/PjxZvXq1QYwa9asSfP848ePm/z585uRI0eagwcPmtmzZxt3d3ezfv16h5431yVCDRs2NEOGDLFvx8fHmzJlypgpU6akeH63bt3M/fffn2Rfo0aNzJNPPpmpceYFjn4WN4qLizMFChQwH330UWaFmKdk5POIi4szTZo0MR9++KHp27evEiEncfSzeO+990yFChVMbGxsVoWYZzj6WQwZMsS0bt06yb6RI0eapk2bZmqceU16EqHnn3/e3HnnnUn2BQUFmfbt2zv0XLmqayw2NpZdu3bRtm1b+z43Nzfatm3L9u3bU7xm+/btSc4HaN++farnS/pk5LO4UWRkJNeuXXP6Ant5UUY/j1deeYUSJUowYMCArAgzT8jIZ7F27VoaN27MkCFDKFmyJDVq1GDy5MnEx8dnVdi5UkY+iyZNmrBr1y5799nx48dZt24d9913X5bELNc56/s7W8ws7SwXL14kPj7evjxHgpIlS3L48OEUrwkODk7x/ODg4EyLMy/IyGdxoxdeeIEyZcok+4cujsvI57FlyxYWLFjA3r17syDCvCMjn8Xx48f54Ycf6NWrF+vWreOvv/5i8ODBXLt2jYkTJ2ZF2LlSRj6Lnj17cvHiRZo1a4Yxhri4OJ566inGjRuXFSFLIql9f4eFhREVFYWPj0+67pOrWoQk93jjjTdYvnw5a9aswdvb29Xh5DlXr16ld+/ezJ8/n2LFirk6nDzParVSokQJPvjgA+rVq0dQUBDjx49n3rx5rg4tz9m0aROTJ0/m3XffZffu3axevZqvv/6aV1991dWhSQblqhahYsWK4e7uTkhISJL9ISEhlCpVKsVrSpUq5dD5kj4Z+SwSvPXWW7zxxhts2LCBu+66KzPDzDMc/TyOHTvGyZMn6dSpk32f1WoFIF++fBw5coSKFStmbtC5VEZ+N0qXLo2Hhwfu7u72fdWqVSM4OJjY2Fg8PT0zNebcKiOfxUsvvUTv3r154oknAKhZsyYREREMGjSI8ePHJ1kkXDJXat/fBQsWTHdrEOSyFiFPT0/q1avHxo0b7fusVisbN26kcePGKV7TuHHjJOcDfP/996meL+mTkc8C4M033+TVV19l/fr11K9fPytCzRMc/TyqVq3K/v372bt3r/3x4IMP0qpVK/bu3UtAQEBWhp+rZOR3o2nTpvz111/2ZBTg6NGjlC5dWknQLcjIZxEZGZks2UlIUI2W7sxSTvv+dqyOO/tbvny58fLyMosXLzYHDx40gwYNMoUKFTLBwcHGGGN69+5txowZYz9/69atJl++fOatt94yhw4dMhMnTtTweSdx9LN44403jKenp/n000/NuXPn7I+rV6+66iXkKo5+HjfSqDHncfSzOHXqlClQoIAZOnSoOXLkiPnqq69MiRIlzGuvveaql5BrOPpZTJw40RQoUMAsW7bMHD9+3Hz33XemYsWKplu3bq56CbnG1atXzZ49e8yePXsMYGbMmGH27Nlj/v77b2OMMWPGjDG9e/e2n58wfH706NHm0KFDZu7cuRo+n2D27Nnm9ttvN56enqZhw4bml19+sR+75557TN++fZOcv3LlSlO5cmXj6elp7rzzTvP1119nccS5lyOfRbly5QyQ7DFx4sSsDzyXcvR3IzElQs7l6Gexbds206hRI+Pl5WUqVKhgXn/9dRMXF5fFUedOjnwW165dMy+//LKpWLGi8fb2NgEBAWbw4MHm33//zfrAc5kff/wxxe+AhPe/b9++5p577kl2Te3atY2np6epUKGCWbRokcPPazFGbXkiIiKSN+WqGiERERERRygREhERkTxLiZCIiIjkWUqEREREJM9SIiQiIiJ5lhIhERERybOUCImIiEiepURIRJJYvHgxhQoVcnUYGWaxWPj888/TPKdfv3507tw5S+IRkexNiZBILtSvXz8sFkuyx19//eXq0Fi8eLE9Hjc3N2677Tb69+/P+fPnnXL/c+fO0bFjRwBOnjyJxWJh7969Sc6ZNWsWixcvdsrzpebll1+2v053d3cCAgIYNGgQly9fdug+StpEMleuWn1eRK7r0KEDixYtSrKvePHiLoomqYIFC3LkyBGsViv79u2jf//+nD17lm+//faW753aquGJ+fv73/LzpMedd97Jhg0biI+P59ChQzz++OOEhoayYsWKLHl+Ebk5tQiJ5FJeXl6UKlUqycPd3Z0ZM2ZQs2ZNfH19CQgIYPDgwYSHh6d6n3379tGqVSsKFChAwYIFqVevHjt37rQf37JlC82bN8fHx4eAgACGDx9OREREmrFZLBZKlSpFmTJl6NixI8OHD2fDhg1ERUVhtVp55ZVXuO222/Dy8qJ27dqsX7/efm1sbCxDhw6ldOnSeHt7U65cOaZMmZLk3gldY+XLlwegTp06WCwWWrZsCSRtZfnggw8oU6ZMkpXdAR566CEef/xx+/YXX3xB3bp18fb2pkKFCkyaNIm4uLg0X2e+fPkoVaoUZcuWpW3btnTt2pXvv//efjw+Pp4BAwZQvnx5fHx8qFKlCrNmzbIff/nll/noo4/44osv7K1LmzZtAuCff/6hW7duFCpUiCJFivDQQw9x8uTJNOMRkeSUCInkMW5ubrzzzjv88ccffPTRR/zwww88//zzqZ7fq1cvbrvtNnbs2MGuXbsYM2YMHh4eABw7dowOHTrQpUsXfv/9d1asWMGWLVsYOnSoQzH5+PhgtVqJi4tj1qxZTJ8+nbfeeovff/+d9u3b8+CDD/Lnn38C8M4777B27VpWrlzJkSNHWLp0KYGBgSne97fffgNgw4YNnDt3jtWrVyc7p2vXrly6dIkff/zRvu/y5cusX7+eXr16AbB582b69OnDM888w8GDB3n//fdZvHgxr7/+erpf48mTJ/n222/x9PS077Nardx2222sWrWKgwcPMmHCBMaNG8fKlSsBGDVqFN26daNDhw6cO3eOc+fO0aRJE65du0b79u0pUKAAmzdvZuvWrfj5+dGhQwdiY2PTHZOIQK5cfV4kr+vbt69xd3c3vr6+9sejjz6a4rmrVq0yRYsWtW8vWrTI+Pv727cLFChgFi9enOK1AwYMMIMGDUqyb/PmzcbNzc1ERUWleM2N9z969KipXLmyqV+/vjHGmDJlypjXX389yTUNGjQwgwcPNsYYM2zYMNO6dWtjtVpTvD9g1qxZY4wx5sSJEwYwe/bsSXJO3759zUMPPWTffuihh8zjjz9u337//fdNmTJlTHx8vDHGmDZt2pjJkycnucfHH39sSpcunWIMxhgzceJE4+bmZnx9fY23t7d9Je0ZM2akeo0xxgwZMsR06dIl1VgTnrtKlSpJ3oOYmBjj4+Njvv322zTvLyJJqUZIJJdq1aoV7733nn3b19cXsLWOTJkyhcOHDxMWFkZcXBzR0dFERkaSP3/+ZPcZOXIkTzzxBB9//LG9e6dixYqArdvs999/Z+nSpfbzjTFYrVZOnDhBtWrVUowtNDQUPz8/rFYr0dHRNGvWjA8//JCwsDDOnj1L06ZNk5zftGlT9u3bB9i6te69916qVKlChw4deOCBB2jXrt0tvVe9evVi4MCBvPvuu3h5ebF06VK6d++Om5ub/XVu3bo1SQtQfHx8mu8bQJUqVVi7di3R0dH873//Y+/evQwbNizJOXPnzmXhwoWcOnWKqKgoYmNjqV27dprx7tu3j7/++osCBQok2R8dHc2xY8cy8A6I5F1KhERyKV9fX+64444k+06ePMkDDzzA008/zeuvv06RIkXYsmULAwYMIDY2NsUv9JdffpmePXvy9ddf88033zBx4kSWL1/Oww8/THh4OE8++STDhw9Pdt3tt9+eamwFChRg9+7duLm5Ubp0aXx8fAAICwu76euqW7cuJ06c4JtvvmHDhg1069aNtm3b8umnn9702tR06tQJYwxff/01DRo0YPPmzbz99tv24+Hh4UyaNIlHHnkk2bXe3t6p3tfT09P+Gbzxxhvcf//9TJo0iVdffRWA5cuXM2rUKKZPn07jxo0pUKAA06ZN49dff00z3vDwcOrVq5ckAU2QXQriRXIKJUIieciuXbuwWq1Mnz7d3tqRUI+SlsqVK1O5cmVGjBhBjx49WLRoEQ8//DB169bl4MGDyRKum3Fzc0vxmoIFC1KmTBm2bt3KPffcY9+/detWGjZsmOS8oKAggoKCePTRR+nQoQOXL1+mSJEiSe6XUI8THx+fZjze3t488sgjLF26lL/++osqVapQt25d+/G6dety5MgRh1/njV588UVat27N008/bX+dTZo0YfDgwfZzbmzR8fT0TBZ/3bp1WbFiBSVKlKBgwYK3FJNIXqdiaZE85I477uDatWvMnj2b48eP8/HHHzNv3rxUz4+KimLo0KFs2rSJv//+m61bt7Jjxw57l9cLL7zAtm3bGDp0KHv37uXPP//kiy++cLhYOrHRo0czdepUVqxYwZEjRxgzZgx79+7lmWeeAWDGjBksW7aMw4cPc/ToUVatWkWpUqVSnASyRIkS+Pj4sH79ekJCQggNDU31eXv16sXXX3/NwoUL7UXSCSZMmMCSJUuYNGkSf/zxB4cOHWL58uW8+OKLDr22xo0bc9dddzF58mQAKlWqxM6dO/n22285evQoL730Ejt27EhyTWBgIL///jtHjhzh4sWLXLt2jV69elGsWDEeeughNm/ezIkTJ9i0aRPDhw/n9OnTDsUkkue5ukhJRJwvpQLbBDNmzDClS5c2Pj4+pn379mbJkiUGMP/++68xJmkxc0xMjOnevbsJCAgwnp6epkyZMmbo0KFJCqF/++03c++99xo/Pz/j6+tr7rrrrmTFzondWCx9o/j4ePPyyy+bsmXLGg8PD1OrVi3zzTff2I9/8MEHpnbt2sbX19cULFjQtGnTxuzevdt+nETF0sYYM3/+fBMQEGDc3NzMPffck+r7Ex8fb0qXLm0Ac+zYsWRxrV+/3jRp0sT4+PiYggULmoYNG5oPPvgg1dcxceJEU6tWrWT7ly1bZry8vMypU6dMdHS06devn/H39zeFChUyTz/9tBkzZkyS686fP29/fwHz448/GmOMOXfunOnTp48pVqyY8fLyMhUqVDADBw40oaGhqcYkIslZjDHGtamYiIiIiGuoa0xERETyLCVCIiIikmcpERIREZE8S4mQiIiI5FlKhERERCTPUiIkIiIieZYSIREREcmzlAiJiIhInqVESERERPIsJUIiIiKSZykREhERkTxLiZCIiIjkWf8H7kmd02poZggAAAAASUVORK5CYII=\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Concatenate the train set and the validation set\n",
        "full_train_data = torch.utils.data.ConcatDataset([train_data, val_data])\n",
        "full_train_sampler = RandomSampler(full_train_data)\n",
        "full_train_dataloader = DataLoader(full_train_data, sampler=full_train_sampler, batch_size=32)\n",
        "\n",
        "# Train the Bert Classifier on the entire training data\n",
        "set_seed(42)\n",
        "roberta_classifier, optimizer, scheduler = initialize_model(epochs=2)"
      ],
      "metadata": {
        "id": "twyEkT-7_Zvk",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "1376bd14-eda6-45aa-929b-106b2f6ce7e6"
      },
      "execution_count": 351,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "Some weights of RobertaModel were not initialized from the model checkpoint at roberta-base and are newly initialized: ['roberta.pooler.dense.bias', 'roberta.pooler.dense.weight']\n",
            "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Run `preprocessing_for_bert` on the test set\n",
        "print('Tokenizing data...')\n",
        "test_inputs, test_masks = preprocessing_for_bert(test_data.review)\n",
        "\n",
        "# Create the DataLoader for our test set\n",
        "test_dataset = TensorDataset(test_inputs, test_masks)\n",
        "test_sampler = SequentialSampler(test_dataset)\n",
        "test_dataloader = DataLoader(test_dataset, sampler=test_sampler, batch_size=32)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kOIA3Mhz-8MG",
        "outputId": "a3721756-07a2-4e16-c145-636918bb1053"
      },
      "execution_count": 352,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Tokenizing data...\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/transformers/tokenization_utils_base.py:2614: FutureWarning: The `pad_to_max_length` argument is deprecated and will be removed in a future version, use `padding=True` or `padding='longest'` to pad to the longest sequence in the batch, or use `padding='max_length'` to pad to a max length. In this case, you can give a specific length with `max_length` (e.g. `max_length=45`) or leave max_length to None to pad to the maximal input size of the model (e.g. 512 for Bert).\n",
            "  warnings.warn(\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Compute predicted probabilities on the test set\n",
        "probs = bert_predict(roberta_classifier, test_dataloader)\n",
        "\n",
        "# Get predictions from the probabilities\n",
        "threshold = 0.5111  # manually adjusted to get predicted non-negative reviews = 105\n",
        "preds = np.where(probs[:, 1] > threshold, 1, 0)\n",
        "\n",
        "# Number of reviews predicted non-negative\n",
        "print(\"Number of reviews predicted non-negative: \", preds.sum())"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rferO9hD_RMO",
        "outputId": "362a65e6-7b41-4650-93b6-9c2ea838f893"
      },
      "execution_count": 363,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Number of reviews predicted non-negative:  109\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "output = test_data[preds==1]\n",
        "list(output.sample(20).review)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "C2r_3boRCySh",
        "outputId": "46bfa90c-ff80-4975-aa8f-0e292d3ec1d2"
      },
      "execution_count": 370,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['NEVER GOT TO STAY HERE <br/>NO COMMUNICATION FROM THE BNB HOST.',\n",
              " 'Great host !!',\n",
              " 'Point positif: Près du métro, propre et lits confortables. Points négatifs: pas de salon ou cuisine assez grande pour se rassembler, manque de commodité dans la cuisine, Rez de chaussée non accessible sauf pour se rendre à l’étage.',\n",
              " '💯',\n",
              " 'Karen was an amazing host!',\n",
              " 'Luz es una anfitriona estupenda,esta súper atenta a cualquier duda que puedas tener y siempre dispuesta a ayudar igual que Andrew.<br/>Todo fue genial!',\n",
              " 'Great place..perfect location.<br/>Hotel looks very good and the rooms are nice and clean',\n",
              " '👍🏽',\n",
              " 'WORST COMMUNICATION EVER',\n",
              " \"Tres bon Airbnb, plus grand que sur les photos, calme et central. Nous avons vrainent apprecié notre séjour dans l'appartement de Chloé. Les articles pour cuisiner etaient un peu limités, mais Chloé nous a rapidement indiqué où nous pouvions nous en acheter dans son quartier. Arrivée très facile, air climatisé très puissante qui nous a permis de très bien dormir. Quartier proche de tout, et beaucoup de commerces à pied. Nous recommandons vivement le logement de Chloé!\",\n",
              " 'Great',\n",
              " \"The room was very clean and private -- LOVED that the room had its own bathroom! <br/>Check-in was flexible (thank god), as I arrived a few hours after I said I would, got stuck on the train.<br/>I was within walking distance of a train station and all the areas I needed to visit were an 8 - 15 minute Lyft/Uber ride away, so location was very convenient!<br/>Overall, I had a great experience and would recommend staying here if you're ever in Brooklyn. Would definitely stay here again.\",\n",
              " 'No ac in the room',\n",
              " \"I had a really good stay at Lana's place!\",\n",
              " 'Great place!',\n",
              " 'This is what I felt during stay<br/>1.Geographical drawbacks(trains after midnight)<br/>2. They did not informed that We were gonna live with one of their family.<br/>3. One of the woman smoked weed in the living room<br/>4. A few lights. Hard to see in the night<br/>5. Bath room near bed room has too much smokes coming from vent.<br/>6. No air circulation in the house.',\n",
              " 'Great place!',\n",
              " 'Normalde 2 gun kalacaktik fakat aniden bir işimiz ciktigi icin 1 gunu iptal etmek zorunda kaldik ve para iademizi hemen yapti , tesekkurler.Hersey soylendigi gibiydi.',\n",
              " \"Great big room that was very unique and inspired a creative headspace. Keyboard was available to use which was epic!\\n<br/>It's definitely not a place for someone looking for a classy hotel room, but for someone wanting to live with other people doing their thing in NY in a real home full of heaps of character and warmth. Loved it!\\n<br/>\",\n",
              " 'Hospitality']"
            ]
          },
          "metadata": {},
          "execution_count": 370
        }
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "cell_execution_strategy": "setup",
      "provenance": [],
      "gpuType": "T4"
    },
    "kernelspec": {
      "display_name": "Python 3",
      "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": "3.10.12"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}