Text_Classification_with_Keras.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Text Classification with Keras.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "P03ep-160G_n"
      },
      "source": [
        "### This notebook involves the Multi label classification on Toxic comments data."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "CCfwISAlDxSQ",
        "outputId": "fc648dbf-0c85-44e5-c78c-56a280e5d0f5"
      },
      "source": [
        "!pip install scikit-plot"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting scikit-plot\n",
            "  Downloading https://files.pythonhosted.org/packages/7c/47/32520e259340c140a4ad27c1b97050dd3254fdc517b1d59974d47037510e/scikit_plot-0.3.7-py3-none-any.whl\n",
            "Requirement already satisfied: scipy>=0.9 in /usr/local/lib/python3.7/dist-packages (from scikit-plot) (1.4.1)\n",
            "Requirement already satisfied: matplotlib>=1.4.0 in /usr/local/lib/python3.7/dist-packages (from scikit-plot) (3.2.2)\n",
            "Requirement already satisfied: joblib>=0.10 in /usr/local/lib/python3.7/dist-packages (from scikit-plot) (1.0.1)\n",
            "Requirement already satisfied: scikit-learn>=0.18 in /usr/local/lib/python3.7/dist-packages (from scikit-plot) (0.22.2.post1)\n",
            "Requirement already satisfied: numpy>=1.13.3 in /usr/local/lib/python3.7/dist-packages (from scipy>=0.9->scikit-plot) (1.19.5)\n",
            "Requirement already satisfied: kiwisolver>=1.0.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib>=1.4.0->scikit-plot) (1.3.1)\n",
            "Requirement already satisfied: python-dateutil>=2.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib>=1.4.0->scikit-plot) (2.8.1)\n",
            "Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.7/dist-packages (from matplotlib>=1.4.0->scikit-plot) (0.10.0)\n",
            "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib>=1.4.0->scikit-plot) (2.4.7)\n",
            "Requirement already satisfied: six>=1.5 in /usr/local/lib/python3.7/dist-packages (from python-dateutil>=2.1->matplotlib>=1.4.0->scikit-plot) (1.15.0)\n",
            "Installing collected packages: scikit-plot\n",
            "Successfully installed scikit-plot-0.3.7\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Lmk0CHGiw1Rf"
      },
      "source": [
        "# Import required libraries\n",
        "from numpy import array\n",
        "from keras.preprocessing.text import one_hot\n",
        "from keras.preprocessing.sequence import pad_sequences\n",
        "from keras.models import Sequential\n",
        "from keras.layers.core import Activation, Dropout, Dense\n",
        "from keras.layers import Flatten, LSTM\n",
        "from keras.layers import GlobalMaxPooling1D\n",
        "from keras.models import Model\n",
        "from keras.layers.embeddings import Embedding\n",
        "from sklearn.model_selection import train_test_split\n",
        "from keras.preprocessing.text import Tokenizer\n",
        "from keras.layers import Input\n",
        "from keras.layers.merge import Concatenate\n",
        "\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import re\n",
        "\n",
        "import matplotlib.pyplot as plt"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NzzI3uxG3oh_"
      },
      "source": [
        "Read the dataset:"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "siOwiCfh34_G"
      },
      "source": [
        "We are considering the only 1000 comments for this task."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IxTmb_dI3j80"
      },
      "source": [
        "toxic_comments = pd.read_csv('toxic_comments.csv')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 191
        },
        "id": "KIe_pffa3zRX",
        "outputId": "e25bfccd-e08c-40bd-e93a-27dead3bcbf0"
      },
      "source": [
        "toxic_comments.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>id</th>\n",
              "      <th>comment_text</th>\n",
              "      <th>toxic</th>\n",
              "      <th>severe_toxic</th>\n",
              "      <th>obscene</th>\n",
              "      <th>threat</th>\n",
              "      <th>insult</th>\n",
              "      <th>identity_hate</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0000997932d777bf</td>\n",
              "      <td>Explanation\\nWhy the edits made under my usern...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>000103f0d9cfb60f</td>\n",
              "      <td>D'aww! He matches this background colour I'm s...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>000113f07ec002fd</td>\n",
              "      <td>Hey man, I'm really not trying to edit war. It...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0001b41b1c6bb37e</td>\n",
              "      <td>\"\\nMore\\nI can't make any real suggestions on ...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0001d958c54c6e35</td>\n",
              "      <td>You, sir, are my hero. Any chance you remember...</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                 id  ... identity_hate\n",
              "0  0000997932d777bf  ...             0\n",
              "1  000103f0d9cfb60f  ...             0\n",
              "2  000113f07ec002fd  ...             0\n",
              "3  0001b41b1c6bb37e  ...             0\n",
              "4  0001d958c54c6e35  ...             0\n",
              "\n",
              "[5 rows x 8 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "H-Fhx-4C32us",
        "outputId": "d2529f9c-1fa3-4cbd-8a8a-bf20be54f1d1"
      },
      "source": [
        "print(f'Toxic comments data contains {toxic_comments.shape[0]} comments & {toxic_comments.shape[1]} features')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Toxic comments data contains 999 comments & 8 features\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RxIM7aya5Ffj"
      },
      "source": [
        "Let's remove all the records where any row contain a null value or empty string."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VFZKLx_94ez2"
      },
      "source": [
        "filter = toxic_comments[\"comment_text\"] != \"\"\n",
        "toxic_comments = toxic_comments[filter]\n",
        "toxic_comments = toxic_comments.dropna()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7_XFpQ3l5QRY"
      },
      "source": [
        "The comment_text column contains text comments. Let's print a random comment and then see the labels for the comments."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "wXi_uVFW5KSG",
        "outputId": "b70fde73-7579-4eca-996c-a293b79764a1"
      },
      "source": [
        "print(toxic_comments[\"comment_text\"][168])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "You should be fired, you're a moronic wimp who is too lazy to do research. It makes me sick that people like you exist in this world.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nxSD6ept5bkw"
      },
      "source": [
        "This is clearly a toxic comment. Let's see the associated labels with this comment:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "b5q989yK5XCr",
        "outputId": "d2b8ae5f-7885-4816-e5fa-70a434cc17e3"
      },
      "source": [
        "print(\"Toxic:\" + str(toxic_comments[\"toxic\"][168]))\n",
        "print(\"Severe_toxic:\" + str(toxic_comments[\"severe_toxic\"][168]))\n",
        "print(\"Obscene:\" + str(toxic_comments[\"obscene\"][168]))\n",
        "print(\"Threat:\" + str(toxic_comments[\"threat\"][168]))\n",
        "print(\"Insult:\" + str(toxic_comments[\"insult\"][168]))\n",
        "print(\"Identity_hate:\" + str(toxic_comments[\"identity_hate\"][168]))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Toxic:1\n",
            "Severe_toxic:0\n",
            "Obscene:0\n",
            "Threat:0\n",
            "Insult:1\n",
            "Identity_hate:0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SqUejRyQ5nT7"
      },
      "source": [
        "Let's now plot the comment count for each label. To do so, we will first filter all the label or output columns."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 191
        },
        "id": "mL4-1qAd5g9f",
        "outputId": "67da9ad5-e0de-42bf-95ff-1ee5d64570e0"
      },
      "source": [
        "toxic_comments_labels = toxic_comments[[\"toxic\", \"severe_toxic\", \"obscene\", \"threat\", \"insult\", \"identity_hate\"]]\n",
        "toxic_comments_labels.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>toxic</th>\n",
              "      <th>severe_toxic</th>\n",
              "      <th>obscene</th>\n",
              "      <th>threat</th>\n",
              "      <th>insult</th>\n",
              "      <th>identity_hate</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   toxic  severe_toxic  obscene  threat  insult  identity_hate\n",
              "0      0             0        0       0       0              0\n",
              "1      0             0        0       0       0              0\n",
              "2      0             0        0       0       0              0\n",
              "3      0             0        0       0       0              0\n",
              "4      0             0        0       0       0              0"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mVTg6pcG50-M"
      },
      "source": [
        "Using the toxic_comments_labels dataframe we will plot bar plots that show the total comment counts for different labels."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 555
        },
        "id": "Wd7d8xjv5uax",
        "outputId": "1e551409-ffba-4e58-a22c-f0e79f71fbed"
      },
      "source": [
        "fig_size = plt.rcParams[\"figure.figsize\"]\n",
        "fig_size[0] = 10\n",
        "fig_size[1] = 8\n",
        "plt.rcParams[\"figure.figsize\"] = fig_size\n",
        "\n",
        "toxic_comments_labels.sum(axis=0).plot.bar()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7ff44f765d10>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "x8qRuIlp6FJD"
      },
      "source": [
        "You can see that the \"toxic\" comment has the highest frequency of occurrence followed by \"insult\" and \"obscene\", respectively."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KZxaMm1I6e7f"
      },
      "source": [
        "We will create a **multi-label text classification** model where each output label will have a dedicated output dense layer. Let's first define our preprocessing function:"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "07lZgZZ36nAz"
      },
      "source": [
        "Let's first define our preprocessing function:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "P6320sui579w"
      },
      "source": [
        "def preprocess_text(sen):\n",
        "    # Remove punctuations and numbers\n",
        "    sentence = re.sub('[^a-zA-Z]', ' ', sen)\n",
        "\n",
        "    # Single character removal\n",
        "    sentence = re.sub(r\"\\s+[a-zA-Z]\\s+\", ' ', sentence)\n",
        "\n",
        "    # Removing multiple spaces\n",
        "    sentence = re.sub(r'\\s+', ' ', sentence)\n",
        "\n",
        "    return sentence"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "B3JRhoco6xNw"
      },
      "source": [
        "The second step is to create inputs and output for the model. The input to the model will be the text comments, whereas the output will be six labels. The following script creates the input layer and the combined output layer:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BWvbgBOb6rsb"
      },
      "source": [
        "X = []\n",
        "sentences = list(toxic_comments[\"comment_text\"])\n",
        "for sen in sentences:\n",
        "    X.append(preprocess_text(sen))\n",
        "\n",
        "y = toxic_comments[[\"toxic\", \"severe_toxic\", \"obscene\", \"threat\", \"insult\", \"identity_hate\"]]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "S6G0rZMZ7CWw"
      },
      "source": [
        "Let's divide the data into training and testing sets:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nJxg6dvd6-Gq"
      },
      "source": [
        "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.20, random_state=42)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1OyaKr4Q7Rz4"
      },
      "source": [
        "The y variable contains the combined output from 6 labels. However, we want to create individual output layer for each label. We will create 6 variables that store individual labels from the training data and 6 variables that store individual label values for the test data."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PtU9Orl07XWb"
      },
      "source": [
        "Look at the following script:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nwnW6swp7HnS"
      },
      "source": [
        "# First output\n",
        "y1_train = y_train[[\"toxic\"]].values\n",
        "y1_test =  y_test[[\"toxic\"]].values\n",
        "\n",
        "# Second output\n",
        "y2_train = y_train[[\"severe_toxic\"]].values\n",
        "y2_test =  y_test[[\"severe_toxic\"]].values\n",
        "\n",
        "# Third output\n",
        "y3_train = y_train[[\"obscene\"]].values\n",
        "y3_test =  y_test[[\"obscene\"]].values\n",
        "\n",
        "# Fourth output\n",
        "y4_train = y_train[[\"threat\"]].values\n",
        "y4_test =  y_test[[\"threat\"]].values\n",
        "\n",
        "# Fifth output\n",
        "y5_train = y_train[[\"insult\"]].values\n",
        "y5_test =  y_test[[\"insult\"]].values\n",
        "\n",
        "# Sixth output\n",
        "y6_train = y_train[[\"identity_hate\"]].values\n",
        "y6_test =  y_test[[\"identity_hate\"]].values"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wjsvJ9rD7gfX"
      },
      "source": [
        "The next step is to convert textual inputs to embedded vectors. The following script does that:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kYXykGxM7cWx"
      },
      "source": [
        "tokenizer = Tokenizer(num_words=5000)\n",
        "tokenizer.fit_on_texts(X_train)\n",
        "\n",
        "X_train = tokenizer.texts_to_sequences(X_train)\n",
        "X_test = tokenizer.texts_to_sequences(X_test)\n",
        "\n",
        "vocab_size = len(tokenizer.word_index) + 1\n",
        "\n",
        "maxlen = 200\n",
        "\n",
        "X_train = pad_sequences(X_train, padding='post', maxlen=maxlen)\n",
        "X_test = pad_sequences(X_test, padding='post', maxlen=maxlen)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5xjP85SI73od"
      },
      "source": [
        "We will be using GloVe word embeddings to convert text inputs to their numeric counterparts."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pyxyBowx94PS"
      },
      "source": [
        "We are downloading Glove embedding **glove6B.100d.txt** from kaggle."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "resources": {
            "http://localhost:8080/nbextensions/google.colab/files.js": {
              "data": "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",
              "ok": true,
              "headers": [
                [
                  "content-type",
                  "application/javascript"
                ]
              ],
              "status": 200,
              "status_text": ""
            }
          },
          "base_uri": "https://localhost:8080/",
          "height": 70
        },
        "id": "yae1kA5Y7k33",
        "outputId": "5c9e9c66-591a-4073-e8cc-a180f236c764"
      },
      "source": [
        "from google.colab import files\n",
        "uploaded_json=files.upload()                  #Upload kaggle API json file"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "     <input type=\"file\" id=\"files-5b9347aa-6b02-47f3-944b-bc75ac3aa1e0\" name=\"files[]\" multiple disabled\n",
              "        style=\"border:none\" />\n",
              "     <output id=\"result-5b9347aa-6b02-47f3-944b-bc75ac3aa1e0\">\n",
              "      Upload widget is only available when the cell has been executed in the\n",
              "      current browser session. Please rerun this cell to enable.\n",
              "      </output>\n",
              "      <script src=\"/nbextensions/google.colab/files.js\"></script> "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Saving kaggle.json to kaggle.json\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MpFIxIl9-bJa"
      },
      "source": [
        "Saving kaggle.json to kaggle.json"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ezG-3Ei3-V98"
      },
      "source": [
        "!pip install -q kaggle        #install kaggle API"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pfjSu1Dy-f2B"
      },
      "source": [
        "!mkdir -p ~/.kaggle\n",
        "!cp kaggle.json ~/.kaggle/         #move kaggle json file to /.kaggle"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ZBNt9Bjw-j_s",
        "outputId": "14c07aac-b86a-4fb6-d625-2d30a87e145d"
      },
      "source": [
        "!ls                    #list directory"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "kaggle.json  sample_data  toxic_comments.csv\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fg5LuQlH-oUh"
      },
      "source": [
        "!chmod 600 /root/.kaggle/kaggle.json"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1gjQqpPO-sQ5",
        "outputId": "8aaf3d15-ed27-47f2-c5c0-887a737de045"
      },
      "source": [
        "!kaggle datasets download -d danielwillgeorge/glove6b100dtxt"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading glove6b100dtxt.zip to /content\n",
            " 95% 125M/131M [00:03<00:00, 48.1MB/s]\n",
            "100% 131M/131M [00:04<00:00, 32.1MB/s]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KiOgH1QT-x89",
        "outputId": "a72e4322-8092-4af5-8042-7e74afd1ef3f"
      },
      "source": [
        "!unzip \\*.zip"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Archive:  glove6b100dtxt.zip\n",
            "  inflating: glove.6B.100d.txt       \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xXuiV6xV-5Up",
        "outputId": "95ad4544-4559-4efb-bbb0-f5524722beb4"
      },
      "source": [
        "!ls"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "glove.6B.100d.txt   kaggle.json  toxic_comments.csv\n",
            "glove6b100dtxt.zip  sample_data\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LP05e0Zr_Nir"
      },
      "source": [
        "from numpy import array\n",
        "from numpy import asarray\n",
        "from numpy import zeros\n",
        "\n",
        "embeddings_dictionary = dict()\n",
        "\n",
        "glove_file = open('glove.6B.100d.txt', encoding=\"utf8\")\n",
        "\n",
        "for line in glove_file:\n",
        "    records = line.split()\n",
        "    word = records[0]\n",
        "    vector_dimensions = asarray(records[1:], dtype='float32')\n",
        "    embeddings_dictionary[word] = vector_dimensions\n",
        "glove_file.close()\n",
        "\n",
        "embedding_matrix = zeros((vocab_size, 100))\n",
        "for word, index in tokenizer.word_index.items():\n",
        "    embedding_vector = embeddings_dictionary.get(word)\n",
        "    if embedding_vector is not None:\n",
        "        embedding_matrix[index] = embedding_vector"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WsNKptwG_kMk"
      },
      "source": [
        "We will creates a model with one input layer, one embedding layer, one LSTM layer with 128 neurons and one output layer with 6 neurons since we have 6 labels in the output."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xh1ffAfF_YGb"
      },
      "source": [
        "deep_inputs = Input(shape=(maxlen,))\n",
        "embedding_layer = Embedding(vocab_size, 100, weights=[embedding_matrix], trainable=False)(deep_inputs)\n",
        "LSTM_Layer_1 = LSTM(128)(embedding_layer)\n",
        "dense_layer_1 = Dense(6, activation='sigmoid')(LSTM_Layer_1)\n",
        "model = Model(inputs=deep_inputs, outputs=dense_layer_1)\n",
        "\n",
        "model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['acc'])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bx284mqfAChu"
      },
      "source": [
        "Let's print the model summary:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Ba45uwk0_5cA",
        "outputId": "c4fe9391-3fd6-46a5-a2b4-b6316c4fd839"
      },
      "source": [
        "print(model.summary())"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"model\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_1 (InputLayer)         [(None, 200)]             0         \n",
            "_________________________________________________________________\n",
            "embedding (Embedding)        (None, 200, 100)          774800    \n",
            "_________________________________________________________________\n",
            "lstm (LSTM)                  (None, 128)               117248    \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 6)                 774       \n",
            "=================================================================\n",
            "Total params: 892,822\n",
            "Trainable params: 118,022\n",
            "Non-trainable params: 774,800\n",
            "_________________________________________________________________\n",
            "None\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ScV0bYvlApnp"
      },
      "source": [
        "Let's now train our model:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "sC5_QjZfAUIk",
        "outputId": "69677884-f006-484a-ac9a-d3c4d2142be5"
      },
      "source": [
        "history = model.fit(X_train, y_train, batch_size=128, epochs=5, verbose=1, validation_split=0.2)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/5\n",
            "5/5 [==============================] - 21s 842ms/step - loss: 0.6833 - acc: 0.0426 - val_loss: 0.6492 - val_acc: 0.0000e+00\n",
            "Epoch 2/5\n",
            "5/5 [==============================] - 3s 647ms/step - loss: 0.6096 - acc: 0.0032 - val_loss: 0.2153 - val_acc: 0.0000e+00\n",
            "Epoch 3/5\n",
            "5/5 [==============================] - 3s 635ms/step - loss: 0.2160 - acc: 0.0654 - val_loss: 0.1283 - val_acc: 0.9937\n",
            "Epoch 4/5\n",
            "5/5 [==============================] - 3s 656ms/step - loss: 0.1899 - acc: 0.9871 - val_loss: 0.1148 - val_acc: 1.0000\n",
            "Epoch 5/5\n",
            "5/5 [==============================] - 3s 641ms/step - loss: 0.1689 - acc: 0.9962 - val_loss: 0.1125 - val_acc: 1.0000\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IvZAx4ltA8kV"
      },
      "source": [
        "Let's now evaluate our model on the test set:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vCYtDlGcAvNG",
        "outputId": "fcc406fe-04e5-4644-8c16-881ce0621bab"
      },
      "source": [
        "score = model.evaluate(X_test, y_test, verbose=1)\n",
        "\n",
        "print(\"Test Accuracy:\", round(score[1], 3))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "7/7 [==============================] - 0s 58ms/step - loss: 0.1420 - acc: 0.9950\n",
            "Test Accuracy: 0.995\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lcXSyeMGBPmP"
      },
      "source": [
        "Our model achieves an accuracy of around 99.5% which is pretty impressive."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XdTckBiIBVjg"
      },
      "source": [
        "Finally, we will plot the loss and accuracy values for training and test sets to see if our model is overfitting."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "2U6EEDMABBt-",
        "outputId": "8f13fdfa-dade-4dc9-d9b7-565cf22d6f4f"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "plt.plot(history.history['acc'])\n",
        "plt.plot(history.history['val_acc'])\n",
        "\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train','test'], loc='upper left')\n",
        "plt.show()\n",
        "\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train','test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmEAAAHwCAYAAADuJ7gwAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzdeXSU153u++/WLIEQIDGLecbYBgzCjucZG2ziOLGNkr4n3efESeckndzuTsfpk850zr0rOd0nKz0k3Z3umx4tbMdOTAmI53mSmDxglZhRaWAQg4TQrKp9/yhhCyygJOqtXfXW81mLVapRj2yrePzu37vLWGsRERERkcTKcB1AREREJB2phImIiIg4oBImIiIi4oBKmIiIiIgDKmEiIiIiDqiEiYiIiDigEiYiKcEY86/GmP8V42MPGmNu8zqTiMilUAkTERERcUAlTEQkgYwxWa4ziEhyUAkTkbjpXwb8ljHmfWNMuzHm/zPGTDDG/M4Y02aMecEYM2bA4+81xnxojGkxxrxijFk44L6lxpjt/c97HMg753utMca82//ct4wxV8SYcbUxZocx5pQxpt4Y84Nz7r+u//Va+u//Yv/t+caY/2OMqTPGtBpj3ui/7SZjTMMg/xxu6//6B8aYJ40x/2mMOQV80RhTZox5u/97HDLG/J0xJmfA8y8zxjxvjDlhjDlijPlzY8xEY0yHMaZ4wOOWGWOajTHZsfzsIpJcVMJEJN7uB24H5gH3AL8D/hwYR/Q9548AjDHzgPXAN/vv2wxUGmNy+gvJ08B/AGOBX/e/Lv3PXQr8CvgyUAz8IxAwxuTGkK8d+L+A0cBq4A+NMZ/uf93p/Xn/tj/TEuDd/uf9FXAV8Kn+TH8GRGL8Z7IWeLL/ez4KhIH/GygBrgFuBb7an6EQeAF4BpgMzAFetNYeBl4BHhjwur8HPGat7Y0xh4gkEZUwEYm3v7XWHrHWNgKvA1XW2h3W2i7gt8DS/sc9CGyy1j7fXyL+CsgnWnKuBrKBn1lre621TwJbBnyPh4F/tNZWWWvD1tp/A7r7n3dB1tpXrLUfWGsj1tr3iRbBG/vvLgdesNau7/++x6217xpjMoA/AL5hrW3s/55vWWu7Y/xn8ra19un+79lprd1mrX3HWttnrT1ItESeybAGOGyt/T/W2i5rbZu1tqr/vn8DvgBgjMkE1hEtqiKSglTCRCTejgz4unOQ6yP7v54M1J25w1obAeqBKf33NVpr7YDn1g34ejrwJ/3LeS3GmBZgav/zLsgYs9IY83L/Ml4r8BWiR6Tof419gzythOhy6GD3xaL+nAzzjDEbjTGH+5co/98YMgBsABYZY2YSPdrYaq2tHmYmEXFMJUxEXGkiWqYAMMYYogWkETgETOm/7YxpA76uB/4fa+3oAX8KrLXrY/i+FUAAmGqtLQL+ATjzfeqB2YM85xjQdZ772oGCAT9HJtGlzIHsOdf/HqgF5lprRxFdrh2YYdZgwfuPJj5B9GjY76GjYCIpTSVMRFx5AlhtjLm1f7D8T4guKb4FvA30AX9kjMk2xnwGKBvw3H8CvtJ/VMsYY0b0D9wXxvB9C4ET1touY0wZ0SXIMx4FbjPGPGCMyTLGFBtjlvQfpfsV8FNjzGRjTKYx5pr+GbTdQF7/988GvgtcbDatEDgFnDbGLAD+cMB9G4FJxphvGmNyjTGFxpiVA+7/d+CLwL2ohImkNJUwEXHCWruL6BGdvyV6pOke4B5rbY+1tgf4DNGycYLo/NhvBjx3K/Al4O+Ak8De/sfG4qvAj4wxbcD3iJbBM68bAu4mWghPEB3Kv7L/7j8FPiA6m3YC+AmQYa1t7X/NfyZ6FK8dOOtsyUH8KdHy10a0UD4+IEMb0aXGe4DDwB7g5gH3v0n0hIDt1tqBS7QikmLM2SMXIiKS7IwxLwEV1tp/dp1FRIZPJUxEJIUYY1YAzxOdaWtznUdEhk/LkSIiKcIY829E9xD7pgqYSOrTkTARERERB3QkTERERMQBlTARERERB7JcBxiqkpISO2PGDNcxRERERC5q27Ztx6y1527gDKRgCZsxYwZbt251HUNERETkoowx593PT8uRIiIiIg6ohImIiIg4oBImIiIi4kDKzYQNpre3l4aGBrq6ulxH8VReXh6lpaVkZ2e7jiIiIiKXyBclrKGhgcLCQmbMmIExxnUcT1hrOX78OA0NDcycOdN1HBEREblEvliO7Orqori42LcFDMAYQ3Fxse+P9omIiKQLX5QwwNcF7Ix0+BlFRETShW9KmEstLS384he/GPLz7r77blpaWjxIJCIiIslOJSwOzlfC+vr6Lvi8zZs3M3r0aK9iiYiISBLzxWC+a4888gj79u1jyZIlZGdnk5eXx5gxY6itrWX37t18+tOfpr6+nq6uLr7xjW/w8MMPAx/v/n/69GnuuusurrvuOt566y2mTJnChg0byM/Pd/yTiYiIiFd8V8J+WPkhNU2n4vqaiyaP4vv3XHbe+3/84x+zc+dO3n33XV555RVWr17Nzp07PzqL8Ve/+hVjx46ls7OTFStWcP/991NcXHzWa+zZs4f169fzT//0TzzwwAM89dRTfOELX4jrzyEiIiLJw3clLBmUlZWdtY3E3/zN3/Db3/4WgPr6evbs2fOJEjZz5kyWLFkCwFVXXcXBgwcTlldEREQSz3cl7EJHrBJlxIgRH339yiuv8MILL/D2229TUFDATTfdNOg2E7m5uR99nZmZSWdnZ0KyioiIiBsazI+DwsJC2traBr2vtbWVMWPGUFBQQG1tLe+8806C04mIiEgy8uxImDHmV8Aa4Ki1dvEg9xvgr4G7gQ7gi9ba7V7l8VJxcTHXXnstixcvJj8/nwkTJnx036pVq/iHf/gHFi5cyPz587n66qsdJhUREZFkYay13rywMTcAp4F/P08Juxv4OtESthL4a2vtyou97vLly+3WrVvPui0YDLJw4cK45E526fSzioiIpDpjzDZr7fLB7vPsSJi19jVjzIwLPGQt0YJmgXeMMaONMZOstYe8yiQiIoK1EOmDcC9EeqPXJT1l5UF2nrtv7+w7wxSgfsD1hv7bVMJERJJNJBItLGeKS7hvwPW+s0vNWfcN8tiPbj/3eoIeF7nwRtqSRm77IVz3TWffPiXOjjTGPAw8DDBt2jTHaUREYhQJx1gQhlssLvS8OD/ORhL0D81AZjZkZENmVv/lYNezPr49KwcyRnzy9k9cP8/rGZ2jNhiLJWIhYm20g1v70R975vb+SxuBsLXYAbd9fGmxkU/ebu2A55z1+mCtJfzRZf/3i5x9/1mv/9H9n/ze57vfWsvszvnc7vCfscsS1ghMHXC9tP+2T7DW/hL4JURnwryPJiJpZf8rsPM3wzz6coESQ4LerkzmxYvKuQUkp2BoRSXWgvPRbcN9zczE/DMbRCRi6YtYwhFLXyRCJAJ9kUj/9ejt53595rHh89x/1utZS1/449vPfz0SfZ61hMMDXu+s6xHCluhjwwPuj9izrp95bF+4//UjHz/m7OsRIpazfpZIEv1tm5lhyMwwZGUYMo0hM7P/67OuZ3x8PcOQlWk+cT3DnHleBlkZhpmTpzj9uVyWsADwNWPMY0QH81s1DyYiTvzuETh5EAqKL1wQsnKHX06Ge5TmYs/LyIIMHckZDmst/+VftvDW3mP0JVPjgI8KxkdFY0BxyBz0PjPgORlkZEBORuZHt2ecdf/A6xlkZvBRgbnYYwe7/+zvH329c7Oe77EX/n79P4uB6IYK/uPlFhXrgZuAEmNMA/B9IBvAWvsPwGaiZ0buJbpFxe97lUVE5LyO7YHmIKz6CVz9FddpJIHeb2jltd3N3H35ROaMGxktBYMcPfnk9YyPr2dEj8JkmgHF4aMjLhmDHIHpfw3zccnIPOt6tIRIevDy7Mh1F7nfAv/dq++fSC0tLVRUVPDVr351yM/92c9+xsMPP0xBQYEHyUTkooKB6OXCNW5zSMJVVIXIz87kJ/dfQWFetus4koZ0DDsOWlpa+MUvfjGs5/7sZz+jo6MjzolEJGY1AZhyFRSVuk4iCXSqq5fAe03ce+VkFTBxJiXOjkx2jzzyCPv27WPJkiXcfvvtjB8/nieeeILu7m7uu+8+fvjDH9Le3s4DDzxAQ0MD4XCYv/iLv+DIkSM0NTVx8803U1JSwssvv+z6RxFJLy0hOPRu9DR1SSsbdjTS2RumfKXOuBd3/FfCfvcIHP4gvq858XK468fnvfvHP/4xO3fu5N133+W5557jySefpLq6Gmst9957L6+99hrNzc1MnjyZTZs2AdHPlCwqKuKnP/0pL7/8MiUlJfHNLCIXF6yMXi68x20OSShrLY9Whbhs8iiuKC1yHUfSmJYj4+y5557jueeeY+nSpSxbtoza2lr27NnD5ZdfzvPPP8+3v/1tXn/9dYqK9Isv4lxNACYshuLZrpNIAr1b30Lt4TbKV07z7Vl3khr8dyTsAkesEsFay3e+8x2+/OUvf+K+7du3s3nzZr773e9y66238r3vfc9BQhEBoO0w1FfBTd9xnUQSrKIqxIicTNYucbtHlIiOhMVBYWEhbW1tANx555386le/4vTp0wA0NjZy9OhRmpqaKCgo4Atf+ALf+ta32L59+yeeKyIJVLsRsFqKTDOtnb1Uvt/EvUumMDLXf8chJLXov8A4KC4u5tprr2Xx4sXcddddlJeXc8011wAwcuRI/vM//5O9e/fyrW99i4yMDLKzs/n7v/97AB5++GFWrVrF5MmTNZgvkkg1ASieA+MXuk4iCfT0jka6eiN8XgP5kgSMTbFPj1++fLndunXrWbcFg0EWLkyPN9J0+llFPNNxAv5yDlz7Dbjt+67TSIJYa1n1s9fJzc4g8LXrXMeRNGGM2WatXT7YfVqOFJH0s2sz2LCWItPM9tBJdh1po7xMR8EkOaiEiUj6qQlA0TSYvNR1EkmgR6tCjMzN4p4rJ7uOIgKohIlIuuk6Bftfjh4F0/YEaaO1o5dN7x9i7ZLJjNBAviQJ35SwVJttG450+BlFPLfnOQj3aCkyzfxmRwPdfRHtkC9JxRclLC8vj+PHj/u6pFhrOX78OHl5ea6jiKS2YABGToCpK10nkQSx1lJRFeLKqaO5bLI2ypbk4YtjsqWlpTQ0NNDc3Ow6iqfy8vIoLdWHDIsMW08H7HkernwIMnzx/6ASg611J9lz9DT/+/4rXEcROYsvSlh2djYzZ850HUNEkt2+F6G3Axbe6zqJJFBFVYjC3CzWXDnJdRSRs+h/BUUkfQQrIX8MzNAeUeniZHsPmz44xH3LplCQ44vjDuIjKmEikh76emDXMzD/bsjMdp1GEuSp7Q30aCBfkpRKmIikhwOvQnerliLTiLWWiuoQy6aNZsHEUa7jiHyCSpiIpIdgAHIKYdZNrpNIglQdOMH+5nbWaYd8SVIqYSLif+E+qN0E8+6AbG3zki4qqkIU5mWx5grtkC/JSSVMRPwv9BZ0HNdSZBo50d7DMzsPc/+yUvJzMl3HERmUSpiI+F+wErLyYO7trpNIgjy1rYGesAbyJbmphImIv0Ui0RI25zbIGeE6jSSAtZb11SGWTx/DvAmFruOInJdKmIj4W+M2aDukpcg08vb+4+w/1q6jYJL0VMJExN+CGyAjG+bd6TqJJEhFVYii/Gzuvlw75EtyUwkTEf+yFmoCMOtGyB/tOo0kwLHT3Tz7YXQgPy9bA/mS3FTCRMS/Dn8ALXVaikwjT25roDdsKV851XUUkYtSCRMR/woGwGTAgtWuk0gCRCLRgfyyGWOZM14D+ZL8VMJExL9qAjD9WhhR4jqJJMBb+45Td7xDA/mSMlTCRMSfmnfBsV1aikwj66tDjCnIZtXiia6jiMREJUxE/CkYiF4uXOM2hyREc5sG8iX1qISJiD/VBKB0BYzS5wamg19vq6cvYlmnpUhJISphIuI/Jw/C4fe1FJkmIhHLY9X1XD1rLLPHjXQdRyRmKmEi4j/Byujlwnvc5pCEeGPvMUInOihfOd11FJEhUQkTEf8JVsLEy2HsTNdJJAEqqkKMHZHDnZdNcB1FZEhUwkTEX04dgvoqLUWmiaOnung+eITPXVVKbpYG8iW1qISJiL/UboxeqoSlhSe21hOOWB4q00C+pB6VMBHxl2AASubB+AWuk4jHwhHL+up6PjW7mJklI1zHERkylTAR8Y/243DwTQ3kp4nX9jTT2NKpHfIlZamEiYh/7NoENqylyDSxvipEycgc7likHfIlNamEiYh/BCth9DSYdKXrJOKxI6e6eLH2KJ+9aio5WfqrTFKT/ssVEX/oaoV9L0ePghnjOo147PEt0YH8dWVTXUcRGTaVMBHxh93PQqRXS5FpIByxPFYd4vq5JUwv1kC+pC6VMBHxh2AARk6Mfl6k+Nqru4/S1NpFubalkBSnEiYiqa+nHfa8AAvXQIbe1vyuoirEuMJcblukHfIltendSkRS394Xoa9TS5FpoKmlk5dqj/LA8lKyM/VXmKQ2/RcsIqkvGID8sTD9WtdJxGOPb6nHAg+t0FKkpD6VMBFJbX3d0aH8BXdDZpbrNOKhvnCEx7fUc/3ccUwdW+A6jsglUwkTkdS2/1XoPgUL17pOIh57ZVczh09pIF/8QyVMRFJbcAPkjoJZN7pOIh6rqA4xvjCXWxeOdx1FJC5UwkQkdYX7oHYzzLsTsnJdpxEPNbZ08squozy4YqoG8sU39F+yiKSuujeh84TOikwDj1eHsMCDK7RDvviHSpiIpK5gALLyYc6trpOIh/rCER7fWs9N88ZROkYD+eIfKmEikpoiEQhuhLm3QY4+usbPXqw9ypFT3ZSvnO46ikhcqYSJSGpq2AKnD+usyDRQURVi4qg8bp4/znUUkbhSCROR1BQMQEY2zLvDdRLxUP2JDl7b08wDK6aSpYF88Rn9Fy0iqcfaaAmbfTPkFblOIx56bEsIAzykgXzxIZUwEUk9h96DlpDOivS53nCEJ7Y2cPP88Uwene86jkjcqYSJSOoJBsBkwvy7XScRD70YPEJzWzflK7VDvviTSpiIpJ5gJcy4FkYUu04iHnq0KsTkojxumq8d8sWfVMJEJLUcrYVju7UU6XOh4x28vucYD66YRmaGcR1HxBMqYSKSWoKB6OWCNW5ziKfWbwmRmWG0Q774mkqYiKSWYACmroRRk1wnEY/09EX49dZ6blkwnolFea7jiHhGJUxEUseJA3D4A1h4j+sk4qHna45w7HSPBvLF91TCRCR1BCujlyphvlZRXceU0fncMFc75Iu/qYSJSOoIBmDSlTBmhusk4pGDx9p5c+9xHloxVQP54nsqYSKSGlobo58XqaNgvra+OjqQ/4AG8iUNqISJSGqo3RS91Ad2+1Z3X5hfb2vgtoXjmTBKA/nifyphIpIaggEYtwDGzXOdRDzy3IdHONHeQ/nK6a6jiCSESpiIJL/2Y1D3ppYifa6iKsTUsflcP6fEdRSRhPC0hBljVhljdhlj9hpjHhnk/mnGmJeNMTuMMe8bY/RBcCLySbWbwEa0S76P7W8+zdv7j/PQimlkaCBf0oRnJcwYkwn8HLgLWASsM8YsOudh3wWesNYuBR4CfuFVHhFJYcEAjJ4OEy93nUQ8sr46RFaG4XPLS11HEUkYL4+ElQF7rbX7rbU9wGPAuRO1FhjV/3UR0ORhHhFJRZ0tsP9VWHQvGB0h8aOu3jBPbmvgjssmML5QA/mSPrI8fO0pQP2A6w3AynMe8wPgOWPM14ERwG0e5hGRVLT7WYj06qxIH3v2w8Oc7OhlXZl2yJf04nowfx3wr9baUuBu4D+MMZ/IZIx52Biz1Riztbm5OeEhRcShYAAKJ8GUq1wnEY88WhVi2tgCrp2tgXxJL16WsEZg4G57pf23DfRfgScArLVvA3nAJ34LrbW/tNYut9YuHzdOH2MhkjZ62mHvC9GzIjNc/z+jeGHv0TaqD5xgXZkG8iX9ePmutgWYa4yZaYzJITp4HzjnMSHgVgBjzEKiJUyHukQkas/z0NelsyJ9bH11PdmZGsiX9ORZCbPW9gFfA54FgkTPgvzQGPMjY8yZd9Q/Ab5kjHkPWA980VprvcokIikmGICCYph2jesk4oGu3jBPbW/gjssmUjIy13UckYTzcjAfa+1mYPM5t31vwNc1wLVeZhCRFNXbFR3KX/wZyPT0rUoc+d3OQ7R09PJ5DeRLmtKQhYgkp/2vQM9pnRXpYxVVIWaWjOCa2cWuo4g4oRImIskpGIDcIph5g+sk4oHdR9rYcvAk68qmYrT/m6QplTARST7hXti1Geavgqwc12nEAxVVIXIyM/jsVVMv/mARn1IJE5Hkc/AN6DypD+z2qa7eML/Z3sCqxRMZO0IlW9KXSpiIJJ9gALILYPatrpOIBza+f4hTXX3aIV/SnkqYiCSXSBiCG2Hu7ZBT4DqNeKCiqo5Z40Zw9ayxrqOIOKUSJiLJpb4a2o9qg1afqj18iu2hFsrLpmkgX9KeSpiIJJdgJWTmwNw7XCcRD6yvCpGTlcH9y7RDvohKmIgkD2ujJWz2LZA3ynUaibPOnjC/2dHI3YsnMkYD+SIqYSKSRJp2QGtIZ0X6VOX7TbR19VG+crrrKCJJQSVMRJJHsBJMJsy/23US8UBFVYg540eyYsYY11FEkoJKmIgkB2ujW1PMvB4KdNac39Q0neLdeg3kiwykEiYiyeFoEI7v1VKkT1VU15GblcFnlk1xHUUkaaiEiUhyCFYCBhascZ1E4qy9u4+ndzSx+vJJjC7QQL7IGSphIpIcggGYuhIKJ7pOInFW+V4Tp7v7KF+pHfJFBlIJExH3ju+DIzthkTZo9aP11SHmTRjJVdM1kC8ykEqYiLgXrIxeah7Md3Y2tvJeQ6sG8kUGoRImIu4FAzBpCYzWcpXfVFSHyMvO4D7tkC/yCSphIuJWawM0btNSpA+d7u5jw45G1lwxmaL8bNdxRJKOSpiIuBXcGL1cuNZtDom7wLtNtPeENZAvch4qYSLiVjAA4xZCyRzXSSTOKqrrWDCxkKVTR7uOIpKUVMJExJ3TR6HuLS1F+tD7DS3sbDzF51dqIF/kfFTCRMSd2k2AhYUqYX5TURUiPzuTtUu1Q77I+aiEiYg7wQCMmQkTLnOdROKorauXwHtN3HPlJEblaSBf5HxUwkTEjc6TcOC16FKklqt85el3m+joCVO+crrrKCJJTSVMRNzY9QxE+rQU6TPWWiqqQiyaNIorS4tcxxFJaiphIuJGMACjpsDkZa6TSBy919BK8NApyjWQL3JRKmEiknjdp2Hvi9GPKcrQ25CfVFTVUZCTydolk11HEUl6evcTkcTb8xyEu/VZkT5zqquXyvcOsXbJZAo1kC9yUSphIpJ4wUoYMQ6mXeM6icTR0zsa6ewNU16mgXyRWKiEiUhi9XZFj4QtWA0Zma7TSJycGci/fEoRl2sgXyQmKmEiklj7XoKe01qK9JntoRZqD7exrkyfEykSK5UwEUmsYCXkFcGMG1wnkTiqqAoxIieTezWQLxIzlTARSZxwL+zaDPPvhqwc12kkTlo7etn4fhNrl05hZG6W6zgiKUMlTEQS58Br0NWipUif+e2OBrr7IpRrKVJkSFTCRCRxgpWQPQJm3+I6icSJtZaK6hBXlhaxeIoG8kWGQiVMRBIjEobajTDvDsjOd51G4mRb3Ul2HzlN+UodBRMZKpUwEUmM0DvQ3qylSJ+pqApRmJvFPVdqIF9kqFTCRCQxgpWQmQtz73CdROKkpaOHjR8c4tNLp1CQo4F8kaFSCRMR71kbLWGzb4HcQtdpJE6e2t5IT19Ee4OJDJNKmIh4r2k7nGqARfe6TiJxEt0hv44lU0ezaPIo13FEUpJKmIh4ryYAGVkwb5XrJBIn1QdOsK+5XQP5IpdAJUxEvGUtBAMw43ooGOs6jcRJRXWIwrws7rlCA/kiw6USJiLeOloDJ/ZrKdJHTrb38LsPDvOZpVPIz9GHsIsMl0qYiHirJgAYWLDGdRKJk6e2N9ATjlC+crrrKCIpTSVMRLwVDMC0a2DkeNdJJA7O7JB/1fQxzJ+oM11FLoVKmIh459je6HKkliJ94539J9jf3K7PiRSJA5UwEfFOMBC91C75vlFRHaIoP5vVV0xyHUUk5amEiYh3ggGYvAyKSl0nkTg4frqbZ3Ye4jPLppCXrYF8kUulEiYi3miph6YdWor0kSe3NdAbtlqKFIkTlTAR8UawMnq5UCXMDyIRy/rqECtmjGHuBA3ki8SDSpiIeCNYCeMvg+LZrpNIHLy9/zgHj3doh3yROFIJE5H4azsCobe1FOkjFVUhRhdkc9diDeSLxItKmIjEX+1GwOqsSJ9obuvm2Q8Pc/+yUg3ki8SRSpiIxF+wEsbOhvGLXCeROHhyWwN9Ecs6DeSLxJVKmIjEV8cJOPh6dCnSGNdp5BKdGchfOXMsc8aPdB1HxFdUwkQkvnb9DiJ9Wor0iTf3HSN0QgP5Il5QCROR+ApWQtHU6CatkvIqqkKMHZHDqsUTXUcR8R2VMBGJn+422PdS9CiYliJT3tG2Lp6vOcJnryolN0sD+SLxphImIvGz5zkId2sp0id+vTU6kP/Qiqmuo4j4kkqYiMRPTQBGjIepK10nkUt0ZiD/mlnFzBqngXwRL6iEiUh89HbCnudhwWrI0NJVqnttTzMNJzs1kC/iIZUwEYmPfS9Bb7t2yfeJ9dUhikfkcOdlGsgX8YpKmIjER00A8kbDjOtdJ5FLdORUFy8Ej/LZ5aXkZOmvCRGv6LdLRC5dX090f7D5d0Nmtus0come2FJPOGJZt0JLkSJeUgkTkUt38DXobtVSpA+EI5bHttRz3ZwSZpSMcB1HxNdUwkTk0tUEIGckzLrZdRK5RK/tbqaxRQP5IomgEiYilyYShtpNMPcOyM5znUYu0aNVIUpG5nL7ogmuo4j4nkqYiFya0NvQcUxLkT5wqLWTl2qP8LnlpWRn6q8HEa/pt0xELk1NALLyYM7trpPIJXp8Sz0RiwbyRRLE0xJmjFlljNlljNlrjHnkPI95wBhTY4z50BhT4WUeEYmzSCT6gd2zb4Vc7aqeyvrCER7fUqjrpO0AACAASURBVM/1c0uYVlzgOo5IWvCshBljMoGfA3cBi4B1xphF5zxmLvAd4Fpr7WXAN73KIyIeaNoObU1aivSBV3c3c6i1i89rIF8kYbw8ElYG7LXW7rfW9gCPAWvPecyXgJ9ba08CWGuPephHROKtZgNkZMG8O10nkUtUURViXGEuty7UQL5IonhZwqYA9QOuN/TfNtA8YJ4x5k1jzDvGmFUe5hGReLI2uhQ580bIH+M6jVyCppZOXt51lAeXT9VAvkgCuf5tywLmAjcB64B/MsaMPvdBxpiHjTFbjTFbm5ubExxRRAZ1ZCecPKClSB94bEs9FniobKrrKCJpxcsS1ggM/I0u7b9toAYgYK3ttdYeAHYTLWVnsdb+0lq73Fq7fNy4cZ4FFpEhqAmAyYD5q10nkUsQHcgPceO8cZSO0UC+SCJ5WcK2AHONMTONMTnAQ0DgnMc8TfQoGMaYEqLLk/s9zCQi8RKshGmfgpH6H6NU9lLtUY6c6qa8TAP5IonmWQmz1vYBXwOeBYLAE9baD40xPzLGnFm/eBY4boypAV4GvmWtPe5VJhGJk2N7oDmopUgfqKgOMWFULrcsGO86ikjayfLyxa21m4HN59z2vQFfW+CP+/+ISKqo2RC9XLDGbQ65JPUnOnh1dzNfv3kOWRrIF0k4/daJyNAFK2HKcig694RnSSWPb6nHAA9qKVLECZUwERmak3Vw6F0tRaa43nCEJ7bWc9P88UwZne86jkhaUgkTkaGp3Ri9XHiP2xxySV4MHuVomwbyRVxSCRORoakJwITLYews10nkElRUh5hUlMdN83V2q4grKmEiEru2w1BfpaNgKa7+RAev72nmwRVTNZAv4pB++0QkdrUbAat5sBS3vjoUHchfoR3yRVxSCROR2NUEoHgujFvgOokMU3Qgv4FbFkxgUpEG8kVcUgkTkdh0nICDb0SXIo1xnUaG6fmaIxw73U35Sh0FE3FNJUxEYrNrM9iwliJTXEVViCmj87lxnnbIF3FNJUxEYlMTgKJpMGmJ6yQyTAePtfPG3mM8uGIqmRk6minimkqYiFxc1ynY/7KWIlPc+i0hMjOMBvJFkoRKmIhc3J7nINyjpcgU1tMX4cmtDdy6YDwTRuW5jiMiqISJSCxqNsDICVBa5jqJDNNzNYc53t5D+UrtkC+SLFTCROTCejpg7wuwYA1k6C0jVVVUhSgdk88Nc7VDvkiyiOkd1RjzG2PMamOM3oFF0s2+F6G3Q0uRKezAsXbe2necdWXTyNBAvkjSiLVU/QIoB/YYY35sjJnvYSYRSSY1AcgfA9OvdZ1Ehml9dYisDMPnlpe6jiIiA8RUwqy1L1hrPw8sAw4CLxhj3jLG/L4xJtvLgCLiUF8P7H4G5q+GTP2qp6LuvjBPbmvg9kUTGF+ogXyRZBLz8qIxphj4IvDfgB3AXxMtZc97kkxE3DvwKnSf0lJkCntm52FOtPewrkwD+SLJJiuWBxljfgvMB/4DuMdae6j/rseNMVu9CicijtVsgJxCmHWT6yQyTBVVIaaNLeC6OSWuo4jIOWIqYcDfWGtfHuwOa+3yOOYRkWQR7ot+VNG8OyEr13UaGYa9R09TdeAEf7ZqvgbyRZJQrMuRi4wxo89cMcaMMcZ81aNMIpIMQm9Bx3EtRaawx84M5F+lHfJFklGsJexL1tqWM1estSeBL3kTSUSSQk0AsvJhzm2uk8gwdPWGeXJ7A3deNpFxhTqSKZKMYi1hmcZ8/IFxxphMIMebSCLiXCQCtRthzq2QM8J1GhmGZ3YepqWjVzvkiySxWGfCniE6hP+P/de/3H+biPhR41ZoOwSL1rpOIsNUURViRnEB18wqdh1FRM4j1hL2baLF6w/7rz8P/LMniUTEvWAAMrJh7h2uk8gw7DnSRvXBE3znrgUayBdJYjGVMGttBPj7/j8i4mfWRufBZt0E+aMv9mhJQhXVIbIzDZ+9SjvkiySzWD87cq4x5kljTI0xZv+ZP16HExEHDr8PLXU6KzJFdfWGeWpbdCC/eKQG8kWSWayD+f9C9ChYH3Az8O/Af3oVSkQcClaCyYD5d7tOIsOw6f1DnOrq00C+SAqItYTlW2tfBIy1ts5a+wNgtXexRMSZmkD0w7pHaIf1VFRRHWJWyQgN5IukgFhLWLcxJgPYY4z5mjHmPmCkh7lExIXmXXBsFyzUUmQq2nW4jW11J1lXNo0BuwqJSJKKtYR9AygA/gi4CvgC8F+8CiUijgQD0cuFa9zmkGFZXx0iJzOD+zWQL5ISLnp2ZP/GrA9aa/8UOA38vuepRMSNmgCUlsGoya6TyBB19oR5ansDd10+kbEjtJe2SCq46JEwa20YuC4BWUTEpRMHomdGLrzHdRIZho3vN9HW1Ud5mQbyRVJFrJu17jDGBIBfA+1nbrTW/saTVCKSeLUbo5famiIlVVSHmDN+JGUzx7qOIiIxirWE5QHHgVsG3GYBlTARv6gJwMQrYMwM10lkiIKHTrEj1MJfrFmkgXyRFBLrjvmaAxPxs1OHoKEabv6u6yQyDBVVIXKyMrh/2RTXUURkCGIqYcaYfyF65Oss1to/iHsiEUk8LUWmrI6ePp7e0cjqyycxukAD+SKpJNblyI0Dvs4D7gOa4h9HRJyo2QAl82HcfNdJZIgq32uirVs75IukoliXI58aeN0Ysx54w5NEIpJY7ceh7k247o9dJ5FhqKiuZ+74kSyfPsZ1FBEZolg3az3XXGB8PIOIiCO7NoGNaCkyBX3Y1Mp79S2Ur9QO+SKpKNaZsDbOngk7DHzbk0Qiklg1ARg9LXpmpKSUiqoQuVkZfGapdsgXSUWxLkcWeh1ERBzoaoX9r8DKL4OOpKSU9u4+NrzbxJorJlNUkO06jogMQ0zLkcaY+4wxRQOujzbGfNq7WCKSELufhUgvLFrrOokMUeC9Jk5rIF8kpcU6E/Z9a23rmSvW2hbg+95EEpGEqdkAhZNgynLXSWSIKqpCLJhYyLJpo11HEZFhirWEDfa4WLe3EJFk1NMOe1+EBWsgY7jn6IgLHzS08kFjK+vKNJAvkspifefdaoz5qTFmdv+fnwLbvAwmIh7b+wL0deqsyBRUUV1HXnYGn16qHfJFUlmsJezrQA/wOPAY0AX8d69CiUgCBCshfyxM+5TrJDIEbV29bHi3iXuumExRvgbyRVJZrGdHtgOPeJxFRBKlrzs6lL9oLWRqsiCVbHi3iY6esAbyRXwg1rMjnzfGjB5wfYwx5lnvYomIp/a/At2ndFZkirHWUlEVYuGkUSyZqoF8kVQX63JkSf8ZkQBYa0+iHfNFUlcwALmjYOYNrpPIELzf0ErNoVPaIV/EJ2ItYRFjzEfHvo0xMzh7B30RSRXhPqjdDPNWQVau6zQyBBVVIQpyMvn0ksmuo4hIHMQ6DPI/gDeMMa8CBrgeeNizVCLinbo3oPMELLzHdRIZglNdvQTea2LtkskU5mkgX8QPYh3Mf8YYs5xo8doBPA10ehlMRDwSrITsAphzm+skMgQbdjTS2auBfBE/ifUDvP8b8A2gFHgXuBp4G7jFu2giEneRCAQ3RgtYToHrNBIjay2PVoW4bPIoLp9SdPEniEhKiHUm7BvACqDOWnszsBRoufBTRCTpNGyB04dhoTZoTSU76luoPdymgXwRn4m1hHVZa7sAjDG51tpaYL53sUTEE8EAZObAvDtdJ5EhqKgKMSInk7VLtEO+iJ/EOpjf0L9P2NPA88aYk0Cdd7FEJO6shZoAzLoZ8ka5TiMxau3sZeP7Tdy3tJSRudpYV8RPYh3Mv6//yx8YY14GioBnPEslIvF36D1oDcGNf+Y6iQzB0zsa6eqN8HkN5Iv4zpD/t8pa+6oXQUTEY8EAmExYsNp1EonRmR3yrygtYrEG8kV8J9aZMBFJZWeWImdcBwVjXaeRGG0PnWTXkTbKy3QUTMSPVMJE0kHzLji+Rxu0pphHq0KMzM3iniu1Q76IH6mEiaSDYAAwKmEppLWjl03vH+LTSyczQgP5Ir6kEiaSDmoCMLUMCie6TiIxemp7A919EcrLpruOIiIeUQkT8bsT++HIB9qgNYVYa6moDnHl1NEsmqztRET8SiVMxO+CldFLLUWmjC0HT7L36Gk+r4F8EV9TCRPxu2AlTLoSxmhZK1VUVNVRmJvFmisnuY4iIh5SCRPxs9bG6OdFaikyZZxs72HzzsPct2wKBTkayBfxM5UwET+r3Ri9XLTWbQ6J2VPbG+jpi1CuHfJFfE8lTMTPgpUwbgGUzHWdRGJwZiB/2bTRLJiogXwRv1MJE/Gr9mNQ96aWIlNI1YET7G9up3yl5vdE0oGnJcwYs8oYs8sYs9cY88gFHne/McYaY5Z7mUckrdRuBBuBRSphqaKiKsSovCzWXKGBfJF04FkJM8ZkAj8H7gIWAeuMMYsGeVwh8A2gyqssImkpWAljZsCExa6TSAxOtPfwzM7DfGZZKXnZma7jiEgCeHkkrAzYa63db63tAR4DBpsO/p/AT4AuD7OIpJfOFtj/anQp0hjXaSQGT26rpyesgXyRdOJlCZsC1A+43tB/20eMMcuAqdbaTR7mEEk/u5+BSK/mwVKEtZb11fUsnz6GeRMKXccRkQRxNphvjMkAfgr8SQyPfdgYs9UYs7W5udn7cCKpLlgJhZNhylWuk0gM3t53nAPH2nUUTCTNeFnCGoGpA66X9t92RiGwGHjFGHMQuBoIDDacb639pbV2ubV2+bhx4zyMLOID3adh7wvRjynK0AnQqeDR6hBF+dncfbkG8kXSiZfv0FuAucaYmcaYHOAhIHDmTmttq7W2xFo7w1o7A3gHuNdau9XDTCL+t/cF6OvSZ0WmiGOnu3nuw8Pcr4F8kbTjWQmz1vYBXwOeBYLAE9baD40xPzLGaFBFxCvBABSUwPRPuU4iMXhyWwO9YUv5yqkXf7CI+IqnH0xmrd0MbD7ntu+d57E3eZlFJC30dsHuZ2Hx/ZChoyrJLhKxrK8OUTZzLHPGayBfJN1oYETET/a/Aj2ndVZkinhr33HqjnfweQ3ki6QllTARPwkGILcIZt7gOonEoKK6jjEF2dx52UTXUUTEAZUwEb8I90LtJph/F2TluE4jF3G0rYvnPjyigXyRNKYSJuIXB9+ArhadFZkifr21gb6IZZ2WIkXSlkqYiF8EA5BdAHNudZ1ELiISsTy2JcTVs8Yye9xI13FExBGVMBE/iIQhuBHm3g7Z+a7TyEW8vvcY9Sc6KV853XUUEXFIJUzED+qrof2ozopMEeurQowdkcOdl01wHUVEHFIJE/GDYAAyc2Dena6TyEUcPdXF88EjfO6qUnKzNJAvks5UwkRSnbXRD+yefQvkasPPZPfE1nrCEcu6Mg3ki6Q7lTCRVNe0A1rrtRSZAsIRy/rqeq6dU8yMkhGu44iIYyphIqkuGACTGd0fTJLaa3uaaWzppLxMA/kiohImktqshZoAzLweCsa6TiMXUVEVomRkDrcv0kC+iKiEiaS2o0E4sU9LkSngcGsXL9Ue5bNXTSUnS2+9IqISJpLaggHAwII1rpPIRTy+5cxA/lTXUUQkSaiEiaSyYCVMuxoKtbyVzMIRy+NbQlw/t4TpxRrIF5EolTCRVHV8HxzZqaXIFPDq7qM0tXZRrm0pRGQAlTCRVBUMRC/1gd1Jr6IqxLjCXG7TQL6IDKASJpKqgpUweSmM1oxRMmtq6eSl2qM8sLyU7Ey95YrIx/SOIJKKWhugcZuWIlPA41vqscBDK7QUKSJnUwkTSUXBjdFLlbCk1heO8PiWem6YO46pYwtcxxGRJKMSJpKKggEYvwhK5rhOIhfw8q5mDp/q0udEisigVMJEUs3po1D3lo6CpYCKqjrGF+Zy68LxrqOISBJSCRNJNbWbAKuzIpNcw8kOXtndzIMrpmogX0QGpXcGkVQTDMDYWTDhMtdJ5AIe31IPwIMrdPaqiAxOJUwklXSehAOvRZcijXGdRs7jzED+TfPGUTpGA/kiMjiVMJFUsusZiPRpHizJvVh7lKNt3ZSvnO46iogkMZUwkVQSDMCoUpiyzHUSuYCKqhATR+Vx8/xxrqOISBJTCRNJFd1tsPfF6EC+liKTVv2JDl7bEx3Iz9JAvohcgN4hRFLFnuch3K2zIpPcY1tCGOChMg3ki8iFqYSJpIpgAEaMg2lXu04i59EbjvDE1gZuWTCeSUX5ruOISJJTCRNJBb1dsPs5WLAaMjJdp5HzeKHmCM1t3dohX0RiohImkgr2vQS97TorMslVVIeYXJTHTfO1Q76IXJxKmEgqCAYgrwhm3uA6iZxH3fF2Xt9zjAdXTCMzQydOiMjFqYSJJLtwL+zaDPPvhsxs12nkPNZX15OZYbRDvojETCVMJNkdeA26WrUUmcR6+iI8ua2eWxaMZ2JRnus4IpIiVMJEkl0wANkjYPYtrpPIeTxfc4Rjp3soX6mBfBGJnUqYSDKLhKF2E8y7A7J1hCVZVVTXMWV0PjfM1Q75IhI7lTCRZBZ6B9qbtRSZxA4ea+fNvcdZVzZVA/kiMiQqYSLJLFgJmbkw9w7XSeQ81leHyMwwPLBcA/kiMjQqYSLJytpoCZtzK+SOdJ1GBtHdF+bX2xq4beF4xo/ScrGIDI1KmEiyatwOpxq0FJnEnv3wCCfaeyhfOd11FBFJQSphIskqGICMLJi/ynUSOY+Kqjqmjs3n+jklrqOISApSCRNJRtZGS9jMGyB/jOs0Moh9zad5Z/8JHloxjQwN5IvIMKiEiSSjIx/Cif1aikxij1WHyMowfG55qesoIpKiVMJEklGwEjCwYLXrJDKIrt4wT25r4I7LJjC+UAP5IjI8KmEiySgYgOmfgpHjXSeRQTz74WFOdvRSXqaBfBEZPpUwkWRzbC8crdFSZBJ7tCrE9OICPjW72HUUEUlhKmEiySYYiF4uXOM2hwxq79E2qg+cYF2ZBvJF5NKohIkkm2AAplwFRRr4TkYVVfVkZxo+e5X+/YjIpVEJE0kmLfXQtENLkUmqqzfMU9sbuOOyiZSMzHUdR0RSnEqYSDIJVkYvF97jNocMavMHh2jt7OXzZdNcRxERH1AJE0kmwQBMWAzFs10nkUFUVIWYWTKCazSQLyJxoBImkizajkDoHR0FS1K7j7Sxte4k68qmYowG8kXk0qmEiSSL2o2A1TxYkqqoCpGTmcFnr5rqOoqI+IRKmEiyCAageA6MX+g6iZyjqzfMb7Y3sGrxRMaOyHEdR0R8QiVMJBl0nIADr0eXIrXUlXQ2vn+IU119lK/UQL6IxI9KmEgy2PU7sGEtRSapiqo6Zo0bwcqZY11HEREfUQkTSQbBABRNhclLXSeRc9QePsX2UAvlZdM0kC8icaUSJuJadxvse0lLkUmqoipETlYG9y/TDvkiEl8qYSKu7X4Wwj1aikxCHT19/HZ7I3cvnsgYDeSLSJyphIm4FqyEkRNg6krXSeQcG987RFt3H+Urp7uOIiI+pBIm4lJvJ+x5Hhashgz9OiabR6tDzBk/khUzxriOIiI+pHd9EZf2vgi97VqKTEIfNrXyXr0G8kXEOyphIi4FKyFvNMy4znUSOcf66hC5GsgXEQ+phIm40tcT3R9swWrIzHadRgZo7+7j6R1NrL5iEkUF+ncjIt5QCRNx5cBr0N2qpcgkVPleE6e7+/i8dsgXEQ+phIm4EgxAzkiYdZPrJHKOiuoQ8yaMZNk0DeSLiHdUwkRciIShdhPMuxOy81ynkQF2NrbyfkOrBvJFxHMqYSIu1L0FHce0FJmEHq0KkZedwX0ayBcRj6mEibgQrISsPJhzm+skMsDp7j4C7zay5orJFOVrIF9EvOVpCTPGrDLG7DLG7DXGPDLI/X9sjKkxxrxvjHnRGKNtqcX/IpFoCZtzG+SOdJ1GBtjwbiPtPWHKNZAvIgngWQkzxmQCPwfuAhYB64wxi8552A5gubX2CuBJ4H97lUckaTRth7YmLUUmofXVIRZMLGTp1NGuo4hIGvDySFgZsNdau99a2wM8Bqwd+ABr7cvW2o7+q+8AGsIQ/6vZABnZ0aF8SRrvN7Sws/EUn1+pgXwRSQwvS9gUoH7A9Yb+287nvwK/8zCPiHvWRremmHUj5OtoSzKpqAqRn53J2qUXepsSEYmfpBjMN8Z8AVgO/OV57n/YGLPVGLO1ubk5seFE4unITjh5EBbe4zqJDNDW1UvgvSbuvXIyo/I0kC8iieFlCWsEpg64Xtp/21mMMbcB/wO411rbPdgLWWt/aa1dbq1dPm7cOE/CiiRETQBMBixY4zqJDPDrrQ10aCBfRBIsy8PX3gLMNcbMJFq+HgLKBz7AGLMU+EdglbX2qIdZRJJDMADTr4URJa6TCHDsdDc/+V0tv97WwFXTx3BFaZHrSCKSRjwrYdbaPmPM14BngUzgV9baD40xPwK2WmsDRJcfRwK/7h+EDVlrdcqY+FPzbmiuheV/4DpJ2gtHLI9W1fFXz+6ioyfMV26czddvmaOBfBFJKC+PhGGt3QxsPue27w34WjtVSvoIBqKXmgdzalvdCf7i6Q+pOXSK6+aU8IN7L2POeO3XJiKJ52kJE5EBggEoXQGjJrtOkpYGLj1OKsrj5+XLuPvyiTr6JSLOqISJJMLJOjj0Htz+I9dJ0s75lh5H5OrtT0Tc0ruQSCIEK6OX2iU/obT0KCLJTCVMJBGClTDxchg703WStDBw6XHiKC09ikhyUgkT8VrbYaivgpv/3HUS39PSo4ikEr0ziXgtWAlYLUV6bFvdSb63YScfNmnpUURSg0qYiNeClVAyD8YvcJ3El7T0KCKpSiVMxEsdJ+DgG3DdN10n8R0tPYpIqtO7lYiXajeBDWspMs4GLj1eO6eYH957GXPGF7qOJSIyJCphIl4KVsLoaTDpStdJfEFLjyLiJyphIl7pOgX7X4ayh0El4ZJo6VFE/EjvYCJe2f0shHu0FHmJtPQoIn6lEibilWAARk6Mfl6kDJmWHkXE71TCRLzQ0wF7X4Al5ZCR4TpNSjl36fHLN87ij26Zq6VHEfEdvauJeGHfi9DboaXIIdLSo4ikE5UwES/UBCB/LEy/1nWSlKClRxFJRyphIvHW1w27n4FF90KmfsUuREuPIpLO9E4nEm8HXoPuU7BwreskSU1LjyKS7lTCROKtZgPkjoJZN7pOkpTOXXr8u/KlrL58kpYeRSTtqISJxFO4L/pRRfPuhKxc12mSipYeRUTOpnc/kXgKvQWdJ2DhPa6TJBUtPYqIfJJKmEg81QQgKx/m3OY6SVLQ0qOIyPmphInESyQS/cDuubdBzgjXaZzS0qOIyMXpHVEkXhq3wunDab9B68Clx0/NLuZHa7X0KCIyGJUwkXip2QAZ2dGh/DSkpUcRkaFRCROJB2ujS5Gzb4a8ItdpEiocsVRU1fGXWnoUERkSvUuKxMPh96GlDm74U9dJEkpLjyIiw6cSJhIPNQEwmTB/teskCaGlRxGRS6cSJhIPwUqYcS2MKHadxFNaehQRiR+9c4pcquZdcGwXlH3JdRJPaelRRCS+VMJELlVNIHq5YI3bHB7R0qOIiDdUwkQuVTAApWUwapLrJHGlpUcREW/p3VTkUpw4ED0z8o7/5TpJXGnpUUTEeyphIpeidmP00icf2H38dDc/eaaWJ7Y2MGFUrpYeRUQ8pBImcilqAjDxChgzw3WSS/KJpccbZvH1W+cyUkuPIiKe0TusyHCdaoKGarjlu66TXBItPYqIuKESJjJctZuilwvXus0xTFp6FBFxSyVMZLhqNkDJfBg3z3WSIdHSo4hIctC7rshwtB+Dujfh+j9xnWRItPQoIpI8VMJEhmPXZrARWHiv6yQxOXfp8W/XLWXNFVp6FBFxSSVMZDhqAjB6Oky83HWSC9LSo4hI8tI7schQdbbA/lfg6q9AEh9J0tKjiEhyUwkTGao9z0GkN2mXIrX0KCKSGlTCRIaqZgMUToIpy10nOYuWHkVEUovenUWGoqcd9r4Iy34PMjJcp/nIuUuPP7z3MuZO0NKjiEgyUwkTGYq9L0BfZ9J8VqSWHkVEUpdKmMhQ1ASgoBimfcppDC09ioikPr1ji8Sqrxt2PwuL74NMd78620PRpcedjVp6FBFJZSphIrHa/wr0tDk7K1JLjyIi/qISJhKrmgDkFsHMGxP6bbX0KCLiT3oXF4lFuA92bYL5qyArJ2HfVkuPIiL+pRImEou6N6DzZMLOitTSo4iI/6mEicSiJgDZBTD7Vk+/jZYeRUTSh97ZRS4mEoHajTDnNsgp8OzbDFx6vGZW9LMetfQoIuJfKmEiF9NQDaePwKK1nry8lh5FRNKTSpjIxQQrITMH5t4R15fV0qOISHrTu73IhVgbnQebdTPkjYrby2rpUUREVMLOUX3gBD+s/JDSMflMGV3AlDH5TBmdT+mY6J+i/GwtE6WTQ+9Cawhu+nZcXk5LjyIicoZK2DkyDIwrzGV/czuv7zlGR0/4rPtH5GQOKGYfl7QpY/IpHZ1PychcMjL0F6pvBCvBZML8uy/pZbT0KCIi59LfAOdYPmMs//r7ZQBYaznZ0UvjyU4aWzpoONlJw8lOGls6aTzZyfZQC62dvWc9PycrI1rKRn98BG1gUZs4Ko+szAwXP5oM1ZmlyBnXQcHYYb+Mlh5FRGQwKmEXYIxh7Igcxo7I4fLSokEf09bV+1EpO3PZcLKThpZOXqw9yrHT3Wc9PjPDMHFU3kdHzs49qjZ5dB65WZmJ+PHkYppr4fgeuPorw3q6lh5FRORCVMIuUWFeNgsmZrNg4uBD2129YZpazj6C1tjS+f+3d/cxdlRlHMe/vy5Lt6WlLbRC05aWAlHAIC9NQVFCJESCSI2UUBGkRCVRCRITXzCKyh9GYuILQgIEiLwKWsEstQhYsISYFgoW+0IhpeFlK9gXykJpt3R3H/+4s2X3cpe93e7cAGQXJgAACqdJREFUs535fZIb5s6c3XuenuXMc+ecM0Pb1u0sXb+FN97uoDv6/syksSOzOWmj+iRrUyeMZsr4URzgIazGeP5BQPCxc/boxzz0aGZm9fBZIWctzU3MnDSGmZPG1Dy+q6ubN9o7ssRsx+6hzw1v7WDlhnYeXv0Gu7r6ZmnjRze/P9TpxQP5WdMK006GsYfW/SMeejQzs3o5CUusuWkE0w4azbSDat+Jvbs72LRt5+4raW1bt+++mrZ+07s88eJmduzy4oEh9+Z6+N9K+Nwv6iruoUczM9tTTsKGuREjxCEHtnDIgS2cNH3CB44PxeKBKRNG9Rn+9OIBsqFIBhyKrB56vOy0mVzhoUczM6uDzxT7OC8eyMmaVph8PEyY3m8RDz2amdnecBJWAvUsHqhO0upZPND7FhyFWjzQvgE2LIczrq552EOPZmY2FPbhM6UNlZbmJo6YNIYjBlg80HeFZ32LB6rnpfUMe44fPYwXD6xdWPnv0ef22d3VHdzz1Kv86u9rPfRoZmZ7zWcPG9CeLB5o27q9z9W09ZsrTx74sMUDva+gDYvFA2taYdLRMPGo3bs89GhmZkPNSZjttb1dPPDMK1t5u6Ozz8/UWjzQe/gzt8UD2zbBq/+C074HfHDo8bovn8AXPPRoZmZDINckTNJZwO+AJuCWiPhl1fGRwB3AScAW4IKIeDnPOlnjDWbxQM890/Z08UDPVbXJ41poaR7E4oEX/gbRTddHz+Gepa946NHMzHKT2xlFUhNwA3Am0AY8Lak1Itb0KvY1YGtEHClpHnAtcEFedbLhazCLB3qGPgdaPNBzC473k7XKHLWaCdWaVjrGTmfugrdY9d/XPPRoZma5yfNr/WxgXUSsB5B0LzAH6J2EzQF+lm0vAK6XpIioOp1a2Q128UDb1h2s2tDOIwMsHpgyYRQzx3Ry4Uv/5A+dZ7Np9HseejQzs1zlmYRNAV7r9b4NOLm/MhHRKakdOBjYnGO9Ptyry2DJtck+3ganGZiWvfpoAQ6FOBTe6+ymo7OLjl3d7NzVRUdnFzu3d9PR3kXHS92MjXdoGtFFy3FzWDzndA89mplZrvaJs4yky4DLAA477LB8P6x7F3S05/sZ1nACRmavcc1Usrbdmgia6OqewM7xX2L+3PNgRMmfGGBmZrnLMwnbQN8LE1OzfbXKtEnaDxhHZYJ+HxFxM3AzwKxZs/IdqpzxafjG4lw/woYfUfmfYZ/4VmJmZoWQ59f9p4GjJB0uaX9gHtBaVaYVuCTbngs85vlgZmZmVga5ffHP5nhdDjxM5RYVt0XEaknXAMsjohW4FbhT0jrgTSqJmpmZmVnh5Tr6EhGLgEVV+67utd0BnJ9nHczMzMyGI88+NjMzM0vASZiZmZlZAk7CzMzMzBJwEmZmZmaWgJMwMzMzswSchJmZmZkl4CTMzMzMLAEnYWZmZmYJOAkzMzMzS8BJmJmZmVkCTsLMzMzMEnASZmZmZpaAkzAzMzOzBJyEmZmZmSXgJMzMzMwsAUVE6jrsEUmbgFdy/piJwOacP2M4K3P8ZY4dyh2/Yy+vMsdf5tihMfFPj4hJtQ7sc0lYI0haHhGzUtcjlTLHX+bYodzxO/Zyxg7ljr/MsUP6+D0caWZmZpaAkzAzMzOzBJyE1XZz6gokVub4yxw7lDt+x15eZY6/zLFD4vg9J8zMzMwsAV8JMzMzM0ug1EmYpLMkvSBpnaQf1jg+UtJ92fFlkmY0vpb5qSP++ZI2SVqRvb6eop5DTdJtkjZKWtXPcUm6Lvt3+Y+kExtdxzzVEf/pktp7tfvVja5jXiRNk/S4pDWSVkv6To0yhWz/OmMvctu3SHpK0nNZ/D+vUaaQfX6dsReyv+8hqUnSvyUtrHEsXbtHRClfQBPwEjAT2B94Djimqsy3gBuz7XnAfanr3eD45wPXp65rDrGfBpwIrOrn+NnAQ4CAU4Blqevc4PhPBxamrmdOsU8GTsy2xwIv1vi7L2T71xl7kdtewJhsuxlYBpxSVaaQfX6dsReyv+8V33eBe2r9fads9zJfCZsNrIuI9RHxHnAvMKeqzBzg9mx7AXCGJDWwjnmqJ/5CiogngDc/pMgc4I6oWAqMlzS5MbXLXx3xF1ZEvB4Rz2bb7wDPA1OqihWy/euMvbCy9tyWvW3OXtWTogvZ59cZe2FJmgp8HrilnyLJ2r3MSdgU4LVe79v4YIe0u0xEdALtwMENqV3+6okf4LxsSGaBpGmNqVpy9f7bFNkns6GLhyQdm7oyeciGHE6gclWgt8K3/4fEDgVu+2xIagWwEXg0Ivpt+6L1+XXEDsXt738LfB/o7ud4snYvcxJmA3sQmBERxwGP8v43BSu2Z6k8ZuMTwO+Bvyauz5CTNAb4C3BlRLyduj6NNEDshW77iOiKiOOBqcBsSR9PXadGqSP2Qvb3ks4BNkbEM6nrUkuZk7ANQO9Mf2q2r2YZSfsB44AtDald/gaMPyK2RMTO7O0twEkNqltq9fxtFFZEvN0zdBERi4BmSRMTV2vISGqmkoTcHRH31yhS2PYfKPait32PiHgLeBw4q+pQkft8oP/YC9zfnwqcK+llKtNuPivprqoyydq9zEnY08BRkg6XtD+VyXitVWVagUuy7bnAY5HN3CuAAeOvmgdzLpU5JGXQCnw1WyV3CtAeEa+nrlSjSDq0Zz6EpNlU+olCnIiyuG4Fno+IX/dTrJDtX0/sBW/7SZLGZ9ujgDOBtVXFCtnn1xN7Ufv7iLgqIqZGxAwq57nHIuKiqmLJ2n2/RnzIcBQRnZIuBx6mslLwtohYLekaYHlEtFLpsO6UtI7KROZ56Wo8tOqM/wpJ5wKdVOKfn6zCQ0jSH6msApsoqQ34KZWJqkTEjcAiKivk1gHbgUvT1DQfdcQ/F/impE5gBzCvCCeizKnAxcDKbH4MwI+Aw6Dw7V9P7EVu+8nA7ZKaqCSXf4qIhSXp8+uJvZD9fX+GS7v7jvlmZmZmCZR5ONLMzMwsGSdhZmZmZgk4CTMzMzNLwEmYmZmZWQJOwszMzMwScBJmZlYHSadLWpi6HmZWHE7CzMzMzBJwEmZmhSLpIklPSVoh6abswcXbJP1G0mpJiyVNysoeL2lp9tDiByRNyPYfKekf2YOsn5V0RPbrx2QPN14r6e6eu8ubmQ2GkzAzKwxJRwMXAKdmDyvuAr4CHEDl7tjHAkuoPCUA4A7gB9lDi1f22n83cEP2IOtPAT2PLToBuBI4BphJ5S70ZmaDUtrHFplZIZ1B5cHDT2cXqUYBG4Fu4L6szF3A/ZLGAeMjYkm2/3bgz5LGAlMi4gGAiOgAyH7fUxHRlr1fAcwAnsw/LDMrIidhZlYkAm6PiKv67JR+UlVusM9r29lruwv3oWa2FzwcaWZFshiYK+kjAJIOkjSdSl83NytzIfBkRLQDWyV9Jtt/MbAkIt4B2iR9MfsdIyWNbmgUZlYK/hZnZoUREWsk/Rh4RNIIYBfwbeBdYHZ2bCOVeWMAlwA3ZknWeuDSbP/FwE2Srsl+x/kNDMPMSkIRg70qb2a2b5C0LSLGpK6HmVlvHo40MzMzS8BXwszMzMwS8JUwMzMzswSchJmZmZkl4CTMzMzMLAEnYWZmZmYJOAkzMzMzS8BJmJmZmVkC/weShmlgTqZKAgAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 720x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "S9DQaKIKDOBB"
      },
      "source": [
        "Since we are using small dataset, it is looks like overfitting."
      ]
    }
  ]
}