[Part_2]Hate-Speech-Detection_Neural-Learning-Models.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "[Part 2]Hate Speech Detection_Neural Learning Models.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true,
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/AbeerAbuZayed/QUIUG_Hate-Speech-Detection_OSACT4-Workshop/blob/master/%5BPart_2%5DHate-Speech-Detection_Neural-Learning-Models.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wrCuCYE3b7Zs",
        "colab_type": "text"
      },
      "source": [
        "#[Part 2] Hate speech Detection using Neural Learning Models\n",
        "In this notebook, we conduct a preliminary experiment on the detection of hate speech in Arabic tweets as part of our participation in the Hate Speech Detection subtask in [OSACT4 workshop](http://edinburghnlp.inf.ed.ac.uk/workshops/OSACT4/).\n",
        "\n",
        "We experimented with three types of neural models:\n",
        "### 1.   Recurrent Neural Networks (RNN):\n",
        ">*   Long Short-Term Memory (LSTM)\n",
        ">*   Bidirectional LSTM (BLSTM)\n",
        ">*   Gated Recurrent Unit (GRU)\n",
        "\n",
        "\n",
        "### 2.   Convolutional Neural Networks (CNN)\n",
        "\n",
        "### 3.   Combined CNN-RNN:\n",
        ">* CNN-LSTM\n",
        ">* CNN-BLSTM\n",
        ">* CNN-GRU\n",
        ">* CNN-LSTM with overSampling\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3psLV0AyPwSV",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "e8175ba8-eba8-44d0-e561-fbf70c889144"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jnSw0od87C8-",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 81
        },
        "outputId": "511ee536-8050-4448-90d4-bc728b8f3c67"
      },
      "source": [
        "from keras.layers import Embedding, Dense, Dropout, Input, LSTM, Bidirectional,GRU\n",
        "from keras.layers import MaxPooling1D, Conv1D, Flatten\n",
        "from keras.preprocessing import sequence, text\n",
        "from keras.models import Model\n",
        "from keras.utils import np_utils\n",
        "from keras.callbacks import Callback\n",
        "\n",
        "from gensim.models.keyedvectors import KeyedVectors\n",
        "from sklearn import preprocessing\n",
        "from time import time\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "import csv\n",
        "\n",
        "from keras import optimizers\n",
        "\n",
        "from sklearn.metrics import f1_score\n",
        "from sklearn.metrics import accuracy_score\n",
        "\n",
        "from sklearn.utils import class_weight\n",
        "\n",
        "from sklearn import preprocessing\n",
        "from sklearn.metrics import (\n",
        "    classification_report as creport\n",
        ")\n"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<p style=\"color: red;\">\n",
              "The default version of TensorFlow in Colab will soon switch to TensorFlow 2.x.<br>\n",
              "We recommend you <a href=\"https://www.tensorflow.org/guide/migrate\" target=\"_blank\">upgrade</a> now \n",
              "or ensure your notebook will continue to use TensorFlow 1.x via the <code>%tensorflow_version 1.x</code> magic:\n",
              "<a href=\"https://colab.research.google.com/notebooks/tensorflow_version.ipynb\" target=\"_blank\">more info</a>.</p>\n"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "n6_dt2cDtobO",
        "colab_type": "text"
      },
      "source": [
        "# Data and AraVec2.0 (pre-trained word embeddings model) Loading"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2Mn0KDJg-o5u",
        "colab_type": "code",
        "outputId": "d540318e-f1a6-459e-e5c4-43693491540a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 87
        }
      },
      "source": [
        "#pre-trained word embedding: https://github.com/bakrianoo/aravec/tree/master/AraVec%202.0\n",
        "\"\"\"\n",
        "Citation:\n",
        "Abu Bakr Soliman, Kareem Eisa, and Samhaa R. El-Beltagy, “AraVec:\n",
        "A set of Arabic Word Embedding Models for use in Arabic NLP”,\n",
        "in proceedings of the 3rd International Conference on \n",
        "Arabic Computational Linguistics (ACLing 2017), Dubai, UAE, 2017.\n",
        "\"\"\"\n",
        "! unzip '/content/drive/My Drive/New- test/tweets_sg_300.zip'  "
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Archive:  /content/drive/My Drive/New- test/tweets_sg_300.zip\n",
            "  inflating: tweets_sg_300           \n",
            "  inflating: tweets_sg_300.trainables.syn1neg.npy  \n",
            "  inflating: tweets_sg_300.wv.vectors.npy  \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "V3kKX6Ra-p1d",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Word_embedding_path\n",
        "embedding_path = '/content/tweets_sg_300'           #Twitter-Skipgram model-300d(trained on 77,600,000 Arabic tweets)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JpRSY6aVACVF",
        "colab_type": "code",
        "outputId": "011b9c19-bb39-4b5e-906c-9eca6bf65c96",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        }
      },
      "source": [
        "train_data = pd.read_csv('/content/drive/My Drive/OSACT4/train_data.csv')\n",
        "train_data"
      ],
      "execution_count": 3,
      "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>Tweet</th>\n",
              "      <th>Offensive</th>\n",
              "      <th>Hate</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>الحمدلله يارب فوز مهم يا زمالك.. كل الدعم ليكم...</td>\n",
              "      <td>NOT_OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>فدوه يا بخت فدوه يا زمن واحد منكم يجيبه</td>\n",
              "      <td>NOT_OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>RT @USER: يا رب يا واحد يا أحد بحق يوم الاحد ا...</td>\n",
              "      <td>OFF</td>\n",
              "      <td>HS</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>RT @USER: #هوا_الحرية يا وجع قلبي عليكي يا امي...</td>\n",
              "      <td>NOT_OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>يا بكون بحياتك الأهم يا إما ما بدي أكون 🎼</td>\n",
              "      <td>NOT_OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6995</th>\n",
              "      <td>@USER يا حمار ، يا جاهل ، نسبة الباطل ما بتتحس...</td>\n",
              "      <td>OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6996</th>\n",
              "      <td>RT @USER: @USER كل زق يا طاقية يا واطي يا حقير...</td>\n",
              "      <td>OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6997</th>\n",
              "      <td>@USER&lt;LF&gt;يا كبير يا ممتع يا نجم لابد أن تعي جي...</td>\n",
              "      <td>NOT_OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6998</th>\n",
              "      <td>يا رب الاتحاد يفوز يا رب. 😭😭 #الاتحاد_النصر</td>\n",
              "      <td>NOT_OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6999</th>\n",
              "      <td>يعز عليا ادخل الشارع وملاقيكش واقف مستنيني في ...</td>\n",
              "      <td>NOT_OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>7000 rows × 3 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                  Tweet Offensive    Hate\n",
              "0     الحمدلله يارب فوز مهم يا زمالك.. كل الدعم ليكم...   NOT_OFF  NOT_HS\n",
              "1               فدوه يا بخت فدوه يا زمن واحد منكم يجيبه   NOT_OFF  NOT_HS\n",
              "2     RT @USER: يا رب يا واحد يا أحد بحق يوم الاحد ا...       OFF      HS\n",
              "3     RT @USER: #هوا_الحرية يا وجع قلبي عليكي يا امي...   NOT_OFF  NOT_HS\n",
              "4             يا بكون بحياتك الأهم يا إما ما بدي أكون 🎼   NOT_OFF  NOT_HS\n",
              "...                                                 ...       ...     ...\n",
              "6995  @USER يا حمار ، يا جاهل ، نسبة الباطل ما بتتحس...       OFF  NOT_HS\n",
              "6996  RT @USER: @USER كل زق يا طاقية يا واطي يا حقير...       OFF  NOT_HS\n",
              "6997  @USER<LF>يا كبير يا ممتع يا نجم لابد أن تعي جي...   NOT_OFF  NOT_HS\n",
              "6998        يا رب الاتحاد يفوز يا رب. 😭😭 #الاتحاد_النصر   NOT_OFF  NOT_HS\n",
              "6999  يعز عليا ادخل الشارع وملاقيكش واقف مستنيني في ...   NOT_OFF  NOT_HS\n",
              "\n",
              "[7000 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "d5frVvfiL72D",
        "colab_type": "code",
        "outputId": "e6db2356-deb2-426a-c2ef-db09e13b560f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        }
      },
      "source": [
        "dev_data = pd.read_csv('/content/drive/My Drive/OSACT4/dev_data.csv')\n",
        "dev_data"
      ],
      "execution_count": 4,
      "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>Tweet</th>\n",
              "      <th>Offensive</th>\n",
              "      <th>Hate</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>فى حاجات مينفعش نلفت نظركوا ليها زى الاصول كده...</td>\n",
              "      <td>NOT_OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>RT @USER: وعيون تنادينا تحايل فينا و نقول يا ع...</td>\n",
              "      <td>NOT_OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>يا بلادي يا أم البلاد يا بلادي بحبك يا مصر بحب...</td>\n",
              "      <td>NOT_OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>RT @USER: يا رب يا قوي يا معين مدّني بالقوة و ...</td>\n",
              "      <td>NOT_OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>RT @USER: رحمك الله يا صدام يا بطل ومقدام. URL</td>\n",
              "      <td>NOT_OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>995</th>\n",
              "      <td>RT @USER: انتو بتوزعوا زيت وسكر فعلا يا عباس؟&lt;...</td>\n",
              "      <td>NOT_OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>996</th>\n",
              "      <td>RT @USER: كدا يا عمر متزعلهاش يا حبيبي 😂 URL</td>\n",
              "      <td>NOT_OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>997</th>\n",
              "      <td>هدا سكن اطفال امارتين من شارقة طالبين فزعتكم ي...</td>\n",
              "      <td>NOT_OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>998</th>\n",
              "      <td>RT @USER: ومدني بمدد من قوتك أواجه به ضعفي.. و...</td>\n",
              "      <td>NOT_OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>999</th>\n",
              "      <td>يا سلااااام يا يو خالد انت والطرب الاصيل URL</td>\n",
              "      <td>NOT_OFF</td>\n",
              "      <td>NOT_HS</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>1000 rows × 3 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                 Tweet Offensive    Hate\n",
              "0    فى حاجات مينفعش نلفت نظركوا ليها زى الاصول كده...   NOT_OFF  NOT_HS\n",
              "1    RT @USER: وعيون تنادينا تحايل فينا و نقول يا ع...   NOT_OFF  NOT_HS\n",
              "2    يا بلادي يا أم البلاد يا بلادي بحبك يا مصر بحب...   NOT_OFF  NOT_HS\n",
              "3    RT @USER: يا رب يا قوي يا معين مدّني بالقوة و ...   NOT_OFF  NOT_HS\n",
              "4       RT @USER: رحمك الله يا صدام يا بطل ومقدام. URL   NOT_OFF  NOT_HS\n",
              "..                                                 ...       ...     ...\n",
              "995  RT @USER: انتو بتوزعوا زيت وسكر فعلا يا عباس؟<...   NOT_OFF  NOT_HS\n",
              "996       RT @USER: كدا يا عمر متزعلهاش يا حبيبي 😂 URL   NOT_OFF  NOT_HS\n",
              "997  هدا سكن اطفال امارتين من شارقة طالبين فزعتكم ي...   NOT_OFF  NOT_HS\n",
              "998  RT @USER: ومدني بمدد من قوتك أواجه به ضعفي.. و...   NOT_OFF  NOT_HS\n",
              "999       يا سلااااام يا يو خالد انت والطرب الاصيل URL   NOT_OFF  NOT_HS\n",
              "\n",
              "[1000 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "s9yxPtQjKsPa",
        "colab_type": "code",
        "outputId": "5b2fc967-0ae6-4ac3-ce82-f84b8912b0e6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        }
      },
      "source": [
        "print(\"Train data shape: {} \\nDev data shape: {}\".format(train_data.shape,dev_data.shape))\n"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train data shape: (7000, 3) \n",
            "Dev data shape: (1000, 3)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ae5n6EhUcEDD",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def get_embedding_matrix(word_index, embedding_index, vocab_dim):\n",
        "    print('Building embedding matrix...')\n",
        "    embedding_matrix = np.zeros((len(word_index) + 1, vocab_dim))\n",
        "    for word, i in word_index.items():\n",
        "        try:\n",
        "            embedding_matrix[i] = embedding_index.get_vector(word)\n",
        "        except:\n",
        "            pass\n",
        "    print('Embedding matrix built.') \n",
        "    #print(\"Word index\", word_index.items())\n",
        "    #print(embedding_matrix) \n",
        "    return embedding_matrix\n",
        "\n",
        "\n",
        "def get_init_parameters(path, ext=None):\n",
        "    if ext == 'vec':\n",
        "        word_model = KeyedVectors.load_word2vec_format(path).wv\n",
        "    else:\n",
        "        word_model = KeyedVectors.load(path).wv\n",
        "    n_words = len(word_model.vocab)\n",
        "    vocab_dim = word_model[word_model.index2word[0]].shape[0]\n",
        "    index_dict = dict()\n",
        "    for i in range(n_words):\n",
        "        index_dict[word_model.index2word[i]] = i+1\n",
        "    print('Number of words in the word embedding',n_words)\n",
        "    #print('word_model', word_model)\n",
        "    #print(\"index_dict\",index_dict)\n",
        "    return word_model, index_dict, n_words, vocab_dim\n",
        "\n",
        "def get_max_length(text_data, return_line=False):\n",
        "    max_length = 0\n",
        "    long_line = \"\"\n",
        "    for line in text_data:\n",
        "        new = len(line.split())\n",
        "        if new > max_length:\n",
        "            max_length = new\n",
        "            long_line = line\n",
        "    if return_line:\n",
        "        return long_line, max_length\n",
        "    else:\n",
        "        return max_length\n",
        "    print(\"max\",long_line,max_length)\n",
        "\n",
        "def load_datasets(data_paths, header=True):\n",
        "    x = []\n",
        "    y = []\n",
        "    for data_path in data_paths:\n",
        "        with open(data_path, 'r') as f:\n",
        "            for line in f:\n",
        "                if header:\n",
        "                    header = False\n",
        "                else:\n",
        "                    temp = line.split(',')\n",
        "                    x.append(temp[0])\n",
        "                    y.append(temp[2].replace('\\n', ''))\n",
        "    max_length = get_max_length(x)\n",
        "    print('Max length:', max_length)\n",
        "    return x,y, max_length\n",
        "\n",
        "def get_train_test(train_raw_text, test_raw_text, n_words, max_length):\n",
        "    tokenizer = text.Tokenizer(num_words=n_words)\n",
        "    tokenizer.fit_on_texts(list(train_raw_text))\n",
        "    word_index = tokenizer.word_index\n",
        "   \n",
        "    train_tokenized = tokenizer.texts_to_sequences(train_raw_text)\n",
        "    test_tokenized = tokenizer.texts_to_sequences(test_raw_text)\n",
        "\n",
        "    return sequence.pad_sequences(train_tokenized, maxlen=max_length, padding='post', truncating='post'),\\\n",
        "           sequence.pad_sequences(test_tokenized, maxlen=max_length, padding='post', truncating='post'),\\\n",
        "           word_index\n",
        "\n",
        "def class_str_2_ind(x_train, x_test, y_train, y_test, classes, n_words, max_length):\n",
        "    print('Converting data to trainable form...')\n",
        "    y_encoder = preprocessing.LabelEncoder()\n",
        "    y_encoder.fit(classes)\n",
        "    y_train = y_encoder.transform(y_train)\n",
        "    y_test = y_encoder.transform(y_test)\n",
        "    #print(y_train)\n",
        "    #print(y_test)\n",
        "    train_y_cat = np_utils.to_categorical(y_train, len(classes))\n",
        "    x_vec_train, x_vec_test, word_index = get_train_test(x_train, x_test, n_words, max_length)\n",
        "    print('Number of training examples: ' + str(len(x_vec_train)))\n",
        "    print('Number of dev examples: ' + str(len(x_vec_test)))\n",
        "    return x_vec_train, x_vec_test, y_train, y_test, train_y_cat, word_index\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lT0r5MyUbtGC",
        "colab_type": "code",
        "outputId": "629270b0-e874-46e0-acef-b4d419458efa",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 90
        }
      },
      "source": [
        "WORD_MODEL, _, MAX_FEATURES, EMBED_SIZE = get_init_parameters(embedding_path) "
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/smart_open/smart_open_lib.py:402: UserWarning: This function is deprecated, use smart_open.open instead. See the migration notes for details: https://github.com/RaRe-Technologies/smart_open/blob/master/README.rst#migrating-to-the-new-open-function\n",
            "  'See the migration notes for details: %s' % _MIGRATION_NOTES_URL\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Number of words in the word embedding 331679\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cjwJ3aN3U1Mv",
        "colab_type": "code",
        "outputId": "5be8385e-658d-4eab-833f-42c30f5f7734",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        }
      },
      "source": [
        "# load train data\n",
        "train_data_path=[\"/content/drive/My Drive/OSACT4/train_data_cleaned.csv\"]\n",
        "x_train, y_train, MAX_TEXT_LENGTH = load_datasets(train_data_path)\n",
        "CLASSES_LIST = np.unique(y_train)\n",
        "print('Label categories: ' + str(CLASSES_LIST))\n",
        "#0= HS, 1= NOT_HS"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Max length: 84\n",
            "Label categories: ['HS' 'NOT_HS']\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VGShWPV6UwCE",
        "colab_type": "code",
        "outputId": "0a6050c4-014a-4ac4-85b2-bafcee220799",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        }
      },
      "source": [
        "# load dev data\n",
        "dev_data_path=[\"/content/drive/My Drive/OSACT4/dev_data_cleaned.csv\"]\n",
        "x_dev, y_dev, MAX_TEXT_LENGTH = load_datasets(dev_data_path)\n",
        "CLASSES_LIST = np.unique(y_dev)\n",
        "print('Label categories: ' + str(CLASSES_LIST))\n",
        "#0= HS, 1= NOT_HS"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Max length: 72\n",
            "Label categories: ['HS' 'NOT_HS']\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XcpaFQRJV8YL",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "MAX_TEXT_LENGTH=84"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "plOnSpgUb18i",
        "colab_type": "code",
        "outputId": "9344be4d-07fc-428f-ada7-80ca0286db0c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 70
        }
      },
      "source": [
        "x_train, x_dev, y_train, y_dev, train_y_cat, word_index = class_str_2_ind(x_train, x_dev, \n",
        "                                                                            y_train, y_dev,\n",
        "                                                                            CLASSES_LIST, MAX_FEATURES,\n",
        "                                                                            MAX_TEXT_LENGTH)\n",
        "dev_cat_y = np_utils.to_categorical(y_dev, len(CLASSES_LIST))"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Converting data to trainable form...\n",
            "Number of training examples: 7000\n",
            "Number of dev examples: 1000\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VQiq-BKShbLE",
        "colab_type": "code",
        "outputId": "85c0490c-8806-43df-8cdf-f269bd3dc108",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "print(\"Tokens number: \"+str(len(word_index)))"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Tokens number: 30125\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WjJd-CWUiNlQ",
        "colab_type": "code",
        "outputId": "f38a901d-d511-4dd0-d006-b8ee56860c1c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "# Sequence length\n",
        "print(\"Original sequence length: \"+str(MAX_TEXT_LENGTH))\n"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Original sequence length: 84\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TRIAZtDUidul",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def get_model(embedding_weights, word_index, vocab_dim, max_length,layer, dropout, optimizer, print_summary=True):\n",
        "    \"\"\"\n",
        "    Create Neural Network With an Embedding layer\n",
        "    \"\"\"\n",
        "    inp = Input(shape=(max_length,))\n",
        "    model = Embedding(input_dim=len(word_index)+1,\n",
        "                      output_dim=vocab_dim,\n",
        "                      trainable=False,\n",
        "                      weights=[embedding_weights])(inp)\n",
        "    model = layer(model)\n",
        "    model = Dropout(dropout)(model)       \n",
        "    model = Flatten()(model)\n",
        "    model = Dense(2, activation='sigmoid')(model)\n",
        "    model = Model(inputs=inp, outputs=model)\n",
        "    \n",
        "    model.compile(loss='categorical_crossentropy', optimizer=optimizer, metrics=['accuracy'])\n",
        "    if print_summary:\n",
        "        model.summary()\n",
        "    return model\n",
        "\n",
        "\n",
        "def get_main_model(word_index, WORD_MODEL, EMBED_SIZE, MAX_TEXT_LENGTH, \n",
        "                   layer, dropout,optimizer):\n",
        "   \n",
        "    tmp = get_embedding_matrix(word_index, WORD_MODEL, EMBED_SIZE)\n",
        "    model = get_model(tmp, word_index, EMBED_SIZE, MAX_TEXT_LENGTH, \n",
        "                      layer, dropout, optimizer= optimizer ,print_summary=True)\n",
        "    return model\n",
        "\n",
        "\n",
        "class TestCallback(Callback):\n",
        "    def __init__(self, test_data):\n",
        "        self.test_data = test_data\n",
        "\n",
        "    def on_epoch_end(self, epoch, logs={}):\n",
        "        x, y = self.test_data\n",
        "        loss, acc = self.model.evaluate(x, y, verbose=0)\n",
        "        print('\\nTesting loss: {}, acc: {}\\n'.format(loss, acc))\n",
        "\n",
        "def train_fit_predict(model, x_train, x_test, y_train, y_test,class_weight, batch_size, epochs, TestCallback=TestCallback):\n",
        "   \n",
        "    history = model.fit(x_train, y_train,\n",
        "                        batch_size=batch_size,\n",
        "                        epochs=epochs, verbose=1,\n",
        "                        validation_data=(x_test, y_test),\n",
        "                        class_weight=class_weight,\n",
        "                        callbacks=[TestCallback((x_test, y_test))])\n",
        "    return history, model"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "n51vTvUGQOP4",
        "colab_type": "text"
      },
      "source": [
        "#### Note: the prior work is the same for all of the neural learning models."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IXB7MDh9lLcM",
        "colab_type": "text"
      },
      "source": [
        "#  RNN (LSTM) Model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YKpEhF4ljGqR",
        "colab_type": "code",
        "outputId": "a1fc2a66-e0f0-4a4b-db1c-ef3b7cac55ba",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 407
        }
      },
      "source": [
        "model = get_main_model(word_index, WORD_MODEL, EMBED_SIZE, MAX_TEXT_LENGTH, \n",
        "                       layer= LSTM(units=16, return_sequences=True, return_state=False), dropout=0.5, \n",
        "                       optimizer= optimizers.Adam(0.001))"
      ],
      "execution_count": 42,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Building embedding matrix...\n",
            "Embedding matrix built.\n",
            "Model: \"model_4\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_4 (InputLayer)         (None, 84)                0         \n",
            "_________________________________________________________________\n",
            "embedding_4 (Embedding)      (None, 84, 300)           9037800   \n",
            "_________________________________________________________________\n",
            "lstm_4 (LSTM)                (None, 84, 16)            20288     \n",
            "_________________________________________________________________\n",
            "dropout_4 (Dropout)          (None, 84, 16)            0         \n",
            "_________________________________________________________________\n",
            "flatten_4 (Flatten)          (None, 1344)              0         \n",
            "_________________________________________________________________\n",
            "dense_4 (Dense)              (None, 2)                 2690      \n",
            "=================================================================\n",
            "Total params: 9,060,778\n",
            "Trainable params: 22,978\n",
            "Non-trainable params: 9,037,800\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nIdYkbZzjJpe",
        "colab_type": "code",
        "outputId": "09e1f791-457f-4afd-fc4e-5561fafb128b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "time_start = time()\n",
        "history, model = train_fit_predict(model,\n",
        "                               x_train[:, :MAX_TEXT_LENGTH],\n",
        "                               x_dev[:, :MAX_TEXT_LENGTH],\n",
        "                               train_y_cat, dev_cat_y, class_weight=None,\n",
        "                               batch_size=500, epochs=15)\n",
        "time_start = time() - time_start\n",
        "\n",
        "print(\"Took : \"+str(np.round(time_start, 2))+\" (s)\") "
      ],
      "execution_count": 43,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 7000 samples, validate on 1000 samples\n",
            "Epoch 1/15\n",
            "7000/7000 [==============================] - 6s 835us/step - loss: 0.5457 - acc: 0.8749 - val_loss: 0.3351 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.3351476159095764, acc: 0.956\n",
            "\n",
            "Epoch 2/15\n",
            "7000/7000 [==============================] - 5s 667us/step - loss: 0.2695 - acc: 0.9484 - val_loss: 0.2055 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.20551465719938278, acc: 0.956\n",
            "\n",
            "Epoch 3/15\n",
            "7000/7000 [==============================] - 5s 673us/step - loss: 0.2125 - acc: 0.9484 - val_loss: 0.1784 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.1783983495235443, acc: 0.956\n",
            "\n",
            "Epoch 4/15\n",
            "7000/7000 [==============================] - 5s 668us/step - loss: 0.1873 - acc: 0.9484 - val_loss: 0.1642 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.16415698558092118, acc: 0.956\n",
            "\n",
            "Epoch 5/15\n",
            "7000/7000 [==============================] - 5s 674us/step - loss: 0.1735 - acc: 0.9484 - val_loss: 0.1513 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.15130487981438637, acc: 0.956\n",
            "\n",
            "Epoch 6/15\n",
            "7000/7000 [==============================] - 5s 671us/step - loss: 0.1556 - acc: 0.9481 - val_loss: 0.1351 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.13512051886320114, acc: 0.956\n",
            "\n",
            "Epoch 7/15\n",
            "7000/7000 [==============================] - 5s 669us/step - loss: 0.1368 - acc: 0.9509 - val_loss: 0.1278 - val_acc: 0.9580\n",
            "\n",
            "Testing loss: 0.12779772283881902, acc: 0.958\n",
            "\n",
            "Epoch 8/15\n",
            "7000/7000 [==============================] - 5s 672us/step - loss: 0.1286 - acc: 0.9517 - val_loss: 0.1251 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.125084951069206, acc: 0.956\n",
            "\n",
            "Epoch 9/15\n",
            "7000/7000 [==============================] - 5s 682us/step - loss: 0.1235 - acc: 0.9520 - val_loss: 0.1227 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.12268184327334165, acc: 0.956\n",
            "\n",
            "Epoch 10/15\n",
            "7000/7000 [==============================] - 5s 672us/step - loss: 0.1179 - acc: 0.9521 - val_loss: 0.1205 - val_acc: 0.9580\n",
            "\n",
            "Testing loss: 0.12052161502093077, acc: 0.958\n",
            "\n",
            "Epoch 11/15\n",
            "7000/7000 [==============================] - 5s 678us/step - loss: 0.1131 - acc: 0.9540 - val_loss: 0.1241 - val_acc: 0.9580\n",
            "\n",
            "Testing loss: 0.12408628825843335, acc: 0.958\n",
            "\n",
            "Epoch 12/15\n",
            "7000/7000 [==============================] - 5s 662us/step - loss: 0.1086 - acc: 0.9561 - val_loss: 0.1194 - val_acc: 0.9600\n",
            "\n",
            "Testing loss: 0.11939282957091928, acc: 0.96\n",
            "\n",
            "Epoch 13/15\n",
            "7000/7000 [==============================] - 5s 673us/step - loss: 0.1060 - acc: 0.9559 - val_loss: 0.1190 - val_acc: 0.9610\n",
            "\n",
            "Testing loss: 0.11898698727786541, acc: 0.961\n",
            "\n",
            "Epoch 14/15\n",
            "7000/7000 [==============================] - 5s 674us/step - loss: 0.1027 - acc: 0.9584 - val_loss: 0.1273 - val_acc: 0.9620\n",
            "\n",
            "Testing loss: 0.1273266174942255, acc: 0.962\n",
            "\n",
            "Epoch 15/15\n",
            "7000/7000 [==============================] - 5s 677us/step - loss: 0.0986 - acc: 0.9597 - val_loss: 0.1184 - val_acc: 0.9590\n",
            "\n",
            "Testing loss: 0.11841939167305827, acc: 0.959\n",
            "\n",
            "Took : 79.31 (s)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Yt-l_Q9Aj44w",
        "colab_type": "code",
        "outputId": "24ee7896-fb7d-4146-97d5-0b5670725f03",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "history.history.keys()"
      ],
      "execution_count": 44,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "dict_keys(['val_loss', 'val_acc', 'loss', 'acc'])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 44
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-xsxi8A1j_Fq",
        "colab_type": "code",
        "outputId": "51b509d4-caee-416b-f204-92c599add468",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        }
      },
      "source": [
        "model.evaluate(x_dev[:, :MAX_TEXT_LENGTH], dev_cat_y, batch_size=1000)\n"
      ],
      "execution_count": 45,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\r1000/1000 [==============================] - 0s 306us/step\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.11841938644647598, 0.9589999914169312]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 45
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "X1e78k51kZ1S",
        "colab_type": "code",
        "outputId": "6fb462dd-bdb8-472d-8058-bbbe795b2fe5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 176
        }
      },
      "source": [
        "y_pred = np.argmax(model.predict(x_dev[:, :MAX_TEXT_LENGTH]), axis=1)\n",
        "\n",
        "print(creport(np.argmax(dev_cat_y, axis=1), y_pred,target_names=['HS', 'NOT_HS'],digits=4))"
      ],
      "execution_count": 46,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "          HS     0.5882    0.2273    0.3279        44\n",
            "      NOT_HS     0.9654    0.9927    0.9789       956\n",
            "\n",
            "    accuracy                         0.9590      1000\n",
            "   macro avg     0.7768    0.6100    0.6534      1000\n",
            "weighted avg     0.9488    0.9590    0.9502      1000\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "po5GA7KMkGUM",
        "colab_type": "code",
        "outputId": "e3fbc3da-890f-4d49-e885-91386e63a8e6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 105
        }
      },
      "source": [
        "n = np.argmin(history.history['val_loss'])\n",
        "\n",
        "print(\"Optimal epoch : {}\".format(n))\n",
        "print(\"Accuracy on train : {} %\".format(np.round(history.history['acc'][n]*100, 2)))\n",
        "print(\"Accuracy on val : {} %\".format(np.round(history.history['val_acc'][n]*100, 2)))\n",
        "print(\"Loss on train : {}\".format(np.round(history.history['loss'][n]*100, 2)))\n",
        "print(\"Loss on Val : {}\".format(np.round(history.history['val_loss'][n]*100, 2)))"
      ],
      "execution_count": 49,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Optimal epoch : 14\n",
            "Accuracy on train : 95.97 %\n",
            "Accuracy on val : 95.9 %\n",
            "Loss on train : 9.86\n",
            "Loss on Val : 11.84\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PlVrg_zJkkLx",
        "colab_type": "code",
        "outputId": "82a86c46-c07e-42d8-b581-fe1c80c0a603",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 404
        }
      },
      "source": [
        "plt.figure(\"Loss Plot\", figsize=(12, 6))\n",
        "plt.plot(range(1, len(history.history['loss'])+1), history.history['loss'], label=\"train loss\")\n",
        "plt.plot(range(1, len(history.history['val_loss'])+1), history.history['val_loss'], label=\"val loss\")\n",
        "plt.plot(n+1,history.history[\"val_loss\"][n],\"r*\", label=\"Lowest loss\")\n",
        "plt.legend()\n",
        "plt.title(\"Learning Curve\")\n",
        "plt.ylabel(\"loss (cross_entropy)\")\n",
        "plt.xlabel(\"epochs\")\n",
        "plt.show();"
      ],
      "execution_count": 50,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAtEAAAGDCAYAAADtZ0xmAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRK5CYII=\n",
            "text/plain": [
              "<Figure size 864x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qb4o3UXJkrLC",
        "colab_type": "code",
        "outputId": "8cb4a4e2-b59e-405c-f7ef-1088bb71ec76",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 404
        }
      },
      "source": [
        "plt.figure(\"Accuracy Plot\", figsize=(12, 6))\n",
        "plt.plot(range(1, len(history.history['acc'])+1), history.history['acc'], label=\"train accuracy\")\n",
        "plt.plot(range(1, len(history.history['val_acc'])+1), history.history['val_acc'], label=\"Val accuracy\")\n",
        "plt.plot(n+1,history.history[\"val_acc\"][n],\"r*\", label=\"Opt. Acc. (csp. Lowest loss)\")\n",
        "plt.legend()\n",
        "plt.title(\"Accuracy Curve\")\n",
        "plt.ylabel(\"accuracy\")\n",
        "plt.xlabel(\"epochs\")\n",
        "plt.show()"
      ],
      "execution_count": 51,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAtcAAAGDCAYAAADgeTwhAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 864x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kZp0EW06kwhy",
        "colab_type": "code",
        "outputId": "fcab617d-901b-4ee3-e92b-ce3237a0b271",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 564
        }
      },
      "source": [
        "from keras.utils import plot_model\n",
        "plot_model(model, to_file='RNN_LSTM_model.png', show_shapes=False, show_layer_names=False)"
      ],
      "execution_count": 52,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAHsAAAIjCAYAAAAjhWTMAAAABmJLR0QA/wD/AP+gvaeTAAAgAElE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\ne0pljjNcNkNw2QwhA1AgdhAcj1Dy/0kkBnNMsW8YAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<IPython.core.display.Image object>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 52
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "F-XRhkE5wNkw",
        "colab_type": "text"
      },
      "source": [
        "# RNN (BLSTM) Model \n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wOhvzj0WXZ-g",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 407
        },
        "outputId": "e8fb599c-3321-4433-b5cb-16bacc34c629"
      },
      "source": [
        "model = get_main_model(word_index, WORD_MODEL, EMBED_SIZE, MAX_TEXT_LENGTH, \n",
        "                       layer= Bidirectional(LSTM(units=32, return_sequences=True, return_state=False)), \n",
        "                       dropout=0.2, optimizer=optimizers.Adam())"
      ],
      "execution_count": 167,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Building embedding matrix...\n",
            "Embedding matrix built.\n",
            "Model: \"model_16\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_16 (InputLayer)        (None, 84)                0         \n",
            "_________________________________________________________________\n",
            "embedding_16 (Embedding)     (None, 84, 300)           9037800   \n",
            "_________________________________________________________________\n",
            "bidirectional_9 (Bidirection (None, 84, 64)            85248     \n",
            "_________________________________________________________________\n",
            "dropout_16 (Dropout)         (None, 84, 64)            0         \n",
            "_________________________________________________________________\n",
            "flatten_16 (Flatten)         (None, 5376)              0         \n",
            "_________________________________________________________________\n",
            "dense_16 (Dense)             (None, 2)                 10754     \n",
            "=================================================================\n",
            "Total params: 9,133,802\n",
            "Trainable params: 96,002\n",
            "Non-trainable params: 9,037,800\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "i62FHvSzXk7Q",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 941
        },
        "outputId": "91daf336-ec2a-44a7-b07b-2cc3f1795338"
      },
      "source": [
        "time_start = time()\n",
        "history, model = train_fit_predict(model,\n",
        "                               x_train[:, :MAX_TEXT_LENGTH],\n",
        "                               x_dev[:, :MAX_TEXT_LENGTH],\n",
        "                               train_y_cat, dev_cat_y, class_weight=None,\n",
        "                               batch_size=500, epochs=10)\n",
        "time_start = time() - time_start\n",
        "\n",
        "print(\"Took : \"+str(np.round(time_start, 2))+\" (s)\") "
      ],
      "execution_count": 168,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 7000 samples, validate on 1000 samples\n",
            "Epoch 1/10\n",
            "7000/7000 [==============================] - 17s 2ms/step - loss: 0.4376 - acc: 0.9081 - val_loss: 0.2274 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.2274200958609581, acc: 0.956\n",
            "\n",
            "Epoch 2/10\n",
            "7000/7000 [==============================] - 13s 2ms/step - loss: 0.2185 - acc: 0.9484 - val_loss: 0.1814 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.18138790315389633, acc: 0.956\n",
            "\n",
            "Epoch 3/10\n",
            "7000/7000 [==============================] - 13s 2ms/step - loss: 0.1825 - acc: 0.9484 - val_loss: 0.1589 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.15891621419787408, acc: 0.956\n",
            "\n",
            "Epoch 4/10\n",
            "7000/7000 [==============================] - 13s 2ms/step - loss: 0.1575 - acc: 0.9489 - val_loss: 0.1405 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.1404514151662588, acc: 0.956\n",
            "\n",
            "Epoch 5/10\n",
            "7000/7000 [==============================] - 13s 2ms/step - loss: 0.1369 - acc: 0.9513 - val_loss: 0.1263 - val_acc: 0.9590\n",
            "\n",
            "Testing loss: 0.12626913669705392, acc: 0.959\n",
            "\n",
            "Epoch 6/10\n",
            "7000/7000 [==============================] - 13s 2ms/step - loss: 0.1257 - acc: 0.9529 - val_loss: 0.1293 - val_acc: 0.9540\n",
            "\n",
            "Testing loss: 0.12929638730175794, acc: 0.954\n",
            "\n",
            "Epoch 7/10\n",
            "7000/7000 [==============================] - 13s 2ms/step - loss: 0.1188 - acc: 0.9539 - val_loss: 0.1243 - val_acc: 0.9540\n",
            "\n",
            "Testing loss: 0.12430897727608681, acc: 0.954\n",
            "\n",
            "Epoch 8/10\n",
            "7000/7000 [==============================] - 13s 2ms/step - loss: 0.1135 - acc: 0.9540 - val_loss: 0.1229 - val_acc: 0.9610\n",
            "\n",
            "Testing loss: 0.12289862802997231, acc: 0.961\n",
            "\n",
            "Epoch 9/10\n",
            "7000/7000 [==============================] - 13s 2ms/step - loss: 0.1076 - acc: 0.9556 - val_loss: 0.1223 - val_acc: 0.9580\n",
            "\n",
            "Testing loss: 0.12226111777871847, acc: 0.958\n",
            "\n",
            "Epoch 10/10\n",
            "7000/7000 [==============================] - 13s 2ms/step - loss: 0.1029 - acc: 0.9567 - val_loss: 0.1215 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.12154351579025388, acc: 0.956\n",
            "\n",
            "Took : 144.05 (s)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IWTJMbAqX5Gj",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "db1d6e5b-d39e-4a2f-fac7-af01c56abaad"
      },
      "source": [
        "model.evaluate(x_dev[:, :MAX_TEXT_LENGTH], dev_cat_y, batch_size=1000)"
      ],
      "execution_count": 169,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\r1000/1000 [==============================] - 1s 546us/step\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.1215435266494751, 0.9559999704360962]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 169
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Yf0CVyJrX-hu",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 176
        },
        "outputId": "f426df10-815a-42a7-e876-c8d5b92812aa"
      },
      "source": [
        "y_pred = np.argmax(model.predict(x_dev[:, :MAX_TEXT_LENGTH]), axis=1)\n",
        "\n",
        "print(creport(np.argmax(dev_cat_y, axis=1), y_pred,target_names=['HS', 'NOT_HS'],digits=4))"
      ],
      "execution_count": 170,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "          HS     0.5000    0.2727    0.3529        44\n",
            "      NOT_HS     0.9672    0.9874    0.9772       956\n",
            "\n",
            "    accuracy                         0.9560      1000\n",
            "   macro avg     0.7336    0.6301    0.6651      1000\n",
            "weighted avg     0.9467    0.9560    0.9498      1000\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Hjkr6-woi004",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 564
        },
        "outputId": "dbc2f094-67bc-40b0-8213-a1384758d95a"
      },
      "source": [
        "plot_model(model, to_file='RNN_BLSTM_model.png', show_shapes=False, show_layer_names=False)"
      ],
      "execution_count": 173,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAMMAAAIjCAYAAAC+pK8BAAAABmJLR0QA/wD/AP+gvaeTAAAgAElE\nQVR4nO3deVhU590+8PvAMMwMwgCKYGRREKNGjTWaKtFGY7NoXAEFl1DtL4lLUzXVhLrU+NpmMRox\nUYmvjbWpeS8E0VezNnmzmBWNJrEYoxixagwKiGwyyPr9/ZHHqRMWB8U5AvfnuuYPnvOc53xnuZlz\nnpk5RxMRARFtd9O7AqKbBcNApDAMRArDQKQY9C6gKdasWYOMjAy9yyAn/eEPf8DgwYP1LsNpLeqd\nISMjA3v37tW7DHJCeno6fvjhB73LaJIW9c4AAIMGDcL27dv1LoOuQtM0vUtoshb1zkB0IzEMRArD\nQKQwDEQKw0CkMAxECsNApDAMRArDQKQwDEQKw0CkMAxECsNApDAMREqrDsPevXvRs2dPuLm5QdM0\nBAYG4i9/+YveZTnYsWMHwsPDoWkaNE1DUFAQpk2bpndZbVKL+z1DUwwaNAhHjhzBAw88gHfffRdZ\nWVnw9fXVuywHMTExiImJQbdu3XD+/HmcO3dO75LarFb9znAzKi8vR1RUlN5lUD0YBhfbvHkz8vLy\n9C6D6tEmw5CcnAwvLy9YLBbs3r0bI0eOhI+PD4KDg5GSkmLv99JLL8FkMqFjx46YNWsWOnXqBJPJ\nhKioKOzbt8/eb+7cuTAajQgKCrK3/e53v4OXlxc0TcP58+cBAPPnz8eCBQuQnZ0NTdPQrVu3a6r/\n008/Ra9evWC1WmEymdCnTx+8++67AICHH37YfvwRERGBb775BgAwY8YMWCwWWK1WvP766wCAmpoa\nLFu2DKGhoTCbzejbty9SU1MBAM8//zwsFgu8vb2Rl5eHBQsWoHPnzsjKyrqmmlsEaUFiY2MlNja2\nyevdf//9AkAKCwvtbUuWLBEA8sEHH0hxcbHk5eXJ0KFDxcvLSyorK+39Zs6cKV5eXvLdd9/JpUuX\n5PDhwzJw4EDx9vaW06dP2/tNnTpVAgMDHba7atUqASD5+fn2tpiYGImIiKhTY0REhFitVqfuz/bt\n22X58uVy4cIFKSgokEGDBkn79u0dtuHu7i4//vijw3pTpkyR119/3f73woULxdPTU9LT06WwsFAW\nL14sbm5usn//fofHaN68ebJu3TqJjo6WI0eOOFUjAElNTXWq700irU2+M1wpKioKPj4+CAgIQHx8\nPMrKynD69GmHPgaDAT179oSnpyd69eqF5ORklJaWYsuWLbrUHBsbi6eeegp+fn7w9/fH2LFjUVBQ\ngPz8fADA7NmzUVNT41BfSUkJ9u/fj1GjRgEALl26hOTkZEyYMAExMTHw9fXF0qVL4eHhUed+Pffc\nc3jsscewY8cO9OjRw3V31MXafBiuZDQaAQBVVVWN9hswYAAsFguOHj3qirKuysPDA8BPuz0AcM89\n96B79+7429/+BlHnld62bRvi4+Ph7u4OAMjKyoLNZkPv3r3t45jNZgQFBd0098vVGIZr5Onpaf9P\n7GpvvfUWhg0bhoCAAHh6euLJJ590WK5pGmbNmoUTJ07ggw8+AAD84x//wP/7f//P3qesrAwAsHTp\nUvsxhqZpOHXqFGw2m+vuzE2EYbgGVVVVKCoqQnBwsEu298knnyApKQkAcPr0aUyYMAFBQUHYt28f\niouLsXLlyjrrTJ8+HSaTCa+88gqysrLg4+ODsLAw+/KAgAAAQFJSEkTE4dZWz1rYqj90u1H27NkD\nEcGgQYPsbQaD4aq7V9fqq6++gpeXFwDg0KFDqKqqwpw5cxAeHg6g/hN2+fn5IS4uDtu2bYO3tzce\neeQRh+UhISEwmUw4ePDgDam5JeI7gxNqa2tRWFiI6upqZGZmYv78+QgNDcX06dPtfbp164YLFy5g\n165dqKqqQn5+Pk6dOlVnLH9/f+Tk5ODkyZMoLS1tNEBVVVXIzc3Fnj177GEIDQ0FALz//vu4dOkS\nvv/+e4dp3ivNnj0bFRUVePPNNzFmzBiHZSaTCTNmzEBKSgqSk5NRUlKCmpoanDlzBmfPnm3qQ9Q6\n6DmX1VRNnVrdu3ev3HbbbeLm5iYAJCgoSJ5++mnZsGGDWCwWASCRkZGSnZ0tmzZtEh8fHwEgYWFh\ncuzYMRH5aWrVw8NDOnfuLAaDQXx8fGT8+PGSnZ3tsK2CggIZPny4mEwm6dq1q/z+97+XJ554QgBI\nt27d7NOwX3/9tYSFhYnZbJYhQ4bIyy+/LBEREQKg0dvOnTvt20pMTBR/f3/x9fWViRMnyvr16wWA\nREREOEz3ioj84he/kEWLFtX7+FRUVEhiYqKEhoaKwWCQgIAAiYmJkcOHD8vKlSvFbDYLAAkJCZGt\nW7c6/biLtMyp1VYdhuYwc+ZM8ff3d+k2m9OoUaPkxIkTLt9uSwwDd5OccHnKsiW4crcrMzMTJpMJ\nXbt21bGiloMH0K1MYmIiZs+eDRHBjBkzsHXrVr1LajH4ztCIxYsXY8uWLSguLkbXrl2Rnp6ud0lX\nZbFY0KNHD/z617/G8uXL0atXL71LajE0kZZz6duJEycCAK/P0AJomobU1FRMmjRJ71KcxUvfEl3G\nMBApDAORwjAQKQwDkcIwECkMA5HCMBApDAORwjAQKQwDkcIwECkMA5HS4n7PsHfvXvu3V4maU4sK\nw+DBg/Uuodnl5OTgwIEDGDt2rN6lNKvY2FiEhIToXUaTtKjfM7RGaWlpiIuLA58G3fH3DESXMQxE\nCsNApDAMRArDQKQwDEQKw0CkMAxECsNApDAMRArDQKQwDEQKw0CkMAxECsNApDAMRArDQKQwDEQK\nw0CkMAxECsNApDAMRArDQKQwDEQKw0CkMAxECsNApDAMRArDQKQwDEQKw0CkMAxECsNApLSoy1i1\ndD/++CPGjBmDqqoqe1tZWRnatWuHPn36OPTt168ftm7d6uoS2zSGwYU6d+6MS5cu4ciRI3WWffvt\ntw5/x8XFuaosUrib5GIJCQkwGK7+P4hhcD2GwcWmTJmCmpqaBpdrmob+/fsjMjLShVURwDC4XGho\nKAYOHAg3t/ofend3dyQkJLi4KgIYBl0kJCRA07R6l9XU1PCi7zphGHQwadKketvd3d1x991345Zb\nbnFxRQQwDLoICAjAsGHD4O7uXmfZQw89pENFBDAMunnooYcgIg5tbm5uiI6O1qkiYhh0Eh0d7TDF\najAYMHLkSPj6+upYVdvGMOjE29sbo0ePhoeHB4CfDpynTZumc1VtG8Ogo6lTp6K6uhoAYDKZMHr0\naJ0ratsYBh2NGjUKFosFABATEwOz2axzRW3bTf/dpDNnzuCLL77Qu4wbZuDAgdizZw9CQkKQlpam\ndzk3TEPTyTcTTX4+pXGTSUtL4/d0WoGb/GUGANtbzG6SiLTKW3V1NVasWKF7HTfqlpqaqvdLx2kt\nJgytlbu7OxYtWqR3GQSG4abgzFe66cZjGIgUhoFIYRiIFIaBSGEYiBSGgUhhGIgUhoFIYRiIFIaB\nSGEYiBSGgUhpdWHYsWMHwsPDoWlag7cuXbo0y7YGDhwId3d39OvXr1nGu9LDDz8Mb29vaJqGgwcP\nNrnf22+/DavVijfeeKPZa2utWl0YYmJicOLECURERMBqtTr8bsBmsyE3N9f+U8vrtX//fgwfPrxZ\nxvq5V155BX/961+vuV8L+DHNTafNfHfY3d0dZrMZZrMZ3bt3b9axGzpVpJ4efPBBFBcX611Gi9Lq\n3hmcsWvXrmYd7/LpXpqbsyFzRRhFBNu3b8emTZtu+Lb00ibDcKW1a9fCy8sLbm5uuOOOOxAYGAgP\nDw94eXmhf//+GDp0KEJCQmAymeDr64snn3yyzhjHjx9Hjx494OXlBbPZjKFDh+Kzzz5z6FNTU4Nl\ny5YhNDQUZrMZffv2dfhJpIhg1apVuPXWW+Hp6Qmr1Yonnniizrac6ffZZ58hNDQUmqZh/fr1AIDk\n5GR4eXnBYrFg9+7dGDlyJHx8fBAcHIyUlJQ6tT7zzDO49dZbYTab0aFDB3Tt2hXPPPNMi/hh/zWT\nm1xqaqpcS5kRERFitVod2ubNmyeHDh2q0/epp54SALJv3z4pKyuT8+fPywMPPCAA5K233pL8/Hwp\nKyuTuXPnCgA5ePCgfd0RI0ZIeHi4/Pvf/5aqqir59ttv5Ze//KWYTCY5duyYvd/ChQvF09NT0tPT\npbCwUBYvXixubm6yf/9+ERFZsmSJaJomL7zwghQWForNZpMNGzYIAPnmm2/s4zjb74cffhAAsm7d\nOod1AcgHH3wgxcXFkpeXJ0OHDhUvLy+prKy093v66afF3d1ddu/eLTabTb766isJDAyUYcOGNfl5\nuNbnTwdpN32V1xMGAHVujYWhtLTU3vbqq6/W6f/ll18KANm2bZu9bcSIEXL77bc7jJeZmSkAZOHC\nhSIiUl5eLhaLReLj4+19bDabeHp6ypw5c8Rms4nFYpF7773XYZyUlBSHF7mz/UQaD0N5ebm97XKQ\njh8/bm8bOHCg3HnnnQ7bePTRR8XNzU0qKirqPH6NaUlhaNW7SVfOJokI5s2b5/S6RqMRAOxnvAP+\nc2xw5QUK69OnTx9YrVZkZmYCALKysmCz2dC7d297H7PZjKCgIBw9ehTHjx+HzWbDiBEjGh3X2X5N\ncfl+XnmfLl26VGc2qqamBh4eHvWeOby1aNVh+Lm1a9c6vCBvJA8PD/sLrKysDACwdOlSh887Tp06\nBZvNhjNnzgD46VT1jXG23/UaNWoUvvrqK+zevRvl5eU4cOAAdu3ahdGjRzMM1DTV1dW4cOECQkND\nAfznxZuUlFTnvEIZGRkwmUwAgIqKikbHdbbf9Vq+fDnuueceTJ8+HT4+PoiOjsakSZOc+tyjJWuT\nYTh79ixmzJhxw8b/6KOPUFtbi/79+wOAfTaqoU+Se/fuDTc3N3z88ceNjutsv+t1+PBhZGdnIz8/\nH1VVVTh9+jSSk5Ph5+d3Q7ertzYVBhFBeXk5duzYAR8fn2Ybt7KyEsXFxaiursbXX3+NuXPnIiws\nDNOnTwfw03/0GTNmICUlBcnJySgpKUFNTQ3OnDmDs2fPIiAgADExMUhPT8fmzZtRUlKCzMzMOnP6\nzva7Xo899hhCQ0Nx8eLFZh33pqfbsbuTmjobsXPnzgZnkq68LV26VERE1q5dKxaLRQBIly5d5NNP\nP5XnnntOrFarAJDAwED5n//5H9m2bZsEBgYKAPHz85OUlBQREdmyZYsMHz5cOnbsKAaDQdq3by+T\nJ0+WU6dOOdRVUVEhiYmJEhoaKgaDQQICAiQmJkYOHz4sIiKlpaXy8MMPS/v27aVdu3YyZMgQWbZs\nmQCQ4OBg+de//uV0v3Xr1klQUJAAEIvFImPHjpUNGzbY72dkZKRkZ2fLpk2bxMfHRwBIWFiYfSr4\nww8/lPbt2zs8Xh4eHtKzZ0/ZsWPHDX3+dNR6p1bp2m3YsEHmz5/v0FZRUSGPP/64eHp6is1mc3qs\nFvT8pbWZ7yaRc86dO4e5c+fWOb4xGo0IDQ1FVVUVqqqqWuW1JNrUMQNdndlshoeHBzZv3ozc3FxU\nVVUhJycHr7zyCpYtW4b4+PhmPd66mTAM5MBqteK9997Dt99+i+7du8NsNqNXr17YsmULnnvuObz6\n6qt6l3jDcDeJ6hg6dCj+7//+T+8yXI7vDEQKw0CkMAxECsNApDAMRArDQKQwDEQKw0CkMAxECsNA\npDAMRArDQKQwDERKi/nWalpamt4l0DXIyMjQuwSntZgwxMXF6V0CtXKaCE/kr6e0tDTExcXxegr6\n285jBiKFYSBSGAYihWEgUhgGIoVhIFIYBiKFYSBSGAYihWEgUhgGIoVhIFIYBiKFYSBSGAYihWEg\nUhgGIoVhIFIYBiKFYSBSGAYihWEgUhgGIoVhIFIYBiKFYSBSGAYihWEgUhgGIoVhIFIYBiKFYSBS\nGAYipcVcxqo1yM3Nxd///neHtszMTADAypUrHdr9/Pzw6KOPuqo0Ai9j5VLV1dUIDAxEcXExDIb/\n/B8SEWiaZv+7oqICjzzyCDZt2qRHmW0VL2PlSgaDAfHx8XBzc0NFRYX9VllZ6fA3AEyZMkXnatse\nhsHFJk+ejKqqqkb7BAQEYOjQoS6qiC5jGFzsrrvuwi233NLgcqPRiISEBLi7u7uwKgIYBpfTNA3T\npk2Dh4dHvcsrKysxefJkF1dFAMOgi8Z2lcLCwnDHHXe4uCICGAZd9OvXD5GRkXXajUYjpk+f7vqC\nCADDoJuEhIQ6u0qVlZWIi4vTqSJiGHQyefJkVFdX2//WNA19+/ZFz549dayqbWMYdBIREYF+/frB\nze2np8BgMCAhIUHnqto2hkFHCQkJ9jBUV1dzF0lnDIOO4uLiUFtbCwAYPHgwgoODda6obWMYdNSp\nUyf7J82/+c1vdK6GID+TmpoqAHjjrVXf6pHW4Fe4U1NTG1pEzaisrAybNm3C448/rncpbUJGRgbW\nrl1b77IGwzBp0qQbVhA5uvfee3m84EINhYHHDDcBBuHmwDAQKQwDkcIwECkMA5HCMBApDAORwjAQ\nKQwDkcIwECkMA5HCMBApDAORwjAQKdcdhh07diA8PByapjncDAYDOnTogF//+tfYuXNnnfXefvtt\nWK1WvPHGGw2O/fDDD8Pb2xuapuHgwYNNWvdG0nv7q1evRseOHaFpGjZu3Fhvn/fffx+LFi2q8/wE\nBQVh2rRpV93Gv/71L8THx6Nr167w9PREhw4dcPvtt+Mvf/mLvU98fHyd572h25tvvlmnlj/96U+N\n1rBmzRpomgY3Nzf06NEDn3zyCV5//XWsXLkSNTU1TXvQnHDdYYiJicGJEycQEREBq9UKEYGIID8/\nH6mpqfjxxx8RExNT58dCzpwJ/5VXXsFf//rXOu16n0Vf7+0vXLgQX3zxRYPLn3rqKbz00ktYvHhx\nnefn3LlzeO211xod/9ChQ4iKikJQUBA++ugjFBcX44svvsADDzyAPXv2OPR97733UFRUhKqqKpw9\nexYAMHbsWFRWVqKsrAx5eXl45JFHADi+VoCfnt+GzixYU1ODl156CQBwzz334OjRo/jVr36FsWPH\nwmQyYcSIESgqKnLq8XLWDdtN8vPzw4gRI/Diiy8CANLS0hyWP/jggyguLsaYMWOaPPb1rNtU5eXl\niIqK0m37TfXcc89h27ZtSEtLg7e39zWNsXr1avj6+mLt2rXo0qULTCYTunfvjj//+c8wm832fpqm\n4a677oLVanW43oSmafDw8IDFYkFAQEC9p8u84447cO7cOezataveGnbs2IHOnTvXu2zevHm4/fbb\nMWrUKIdzT12vG37M0KVLFwC45hRfeREPPWzevBl5eXm61uCs48eP409/+hP+67/+CyaT6ZrHKSgo\nQHFxMS5cuODQbjQaHXYNU1JSYLFYrjrezJkzMXr0aIe2OXPmAABefvnletdZs2YNFixY0OCYy5cv\nx8GDBxv81dq1uOFhuHyZprvvvtve9tlnnyE0NBSapmH9+vX2dhHBqlWrcOutt8LT0xNWqxVPPPGE\nw3j1rfv888/DYrHA29sbeXl5WLBgATp37oysrCzU1NRg2bJlCA0NhdlsRt++fevssm3duhUDBgyA\nyWSCl5cXunTpgj//+c+YP38+FixYgOzsbGiahm7dujVa+5o1a9CzZ094enrCz88P48ePx9GjR+19\nkpOT4eXlBYvFgt27d2PkyJHw8fFBcHAwUlJSHGr69NNP0atXL1itVphMJvTp0wfvvvtuo4/1Sy+9\nBBHB2LFjnXlqGjRw4ECUlZXhnnvuweeff35dYzXknnvuQc+ePfHRRx8hKyvLYdnnn38Om82G++67\nr8H1/fz8cPfdd2Pt2rXNt9va0NkxmioiIkKsVqv9b5vNJu+8846EhYXJfffdJxcvXnTo/8MPPwgA\nWbdunb1tyZIlommavPDCC1JYWCg2m002bNggAOSbb7656roAZN68ebJu3TqJjo6WI0eOyMKFC8XT\n01PS09OlsLBQFi9eLG5ubrJ//34REUlKShIA8uyzz0pBQYFcuHBB/vu//1umTp0qIiIxMTESERFx\n1dqXLVsmRqNRtm7dKkVFRZKZmSn9+/eXDh06yLlz5+rU+cEHH0hxcbHk5eXJ0KFDxcvLSyorK+39\ntm/fLsuXL5cLFy5IQUGBDBo0SNq3b29f/v333wsAefnll+1t4eHh0qtXL6een8bYbDYZMGCA/UwS\nvXr1kpUrV0pBQUGj6509e1YAyLhx4xrtFxERIf/+97/lxRdfFAAyf/58h+UTJkyQLVu2SGlpqQCQ\nESNG1DvOokWL6rw2rqaR13das4YB9ZySo0+fPvLqq69KRUWFQ/+fv6BsNptYLBa59957HfqlpKQ0\nKQzl5eX2tvLycrFYLBIfH29vs9ls4unpKXPmzJHKykrx9fWV4cOHO2yzurpa1q5dKyLOhcFms0m7\ndu0ctiMi8uWXXwoAWbFiRaN1Xg788ePHG3p45ZlnnhEAkpeXJyJ1w3Dx4kXRNE3GjBlT7/pNCYOI\nSGVlpbz44ovSo0cP+3PZsWNH2bNnT4PrNDUMRUVF4uXlJX5+fmKz2UREJDs7W4KDg6WiouKqYfjb\n3/4mAOQf//iH0/ersTA0627SlbNJVVVVOHPmDB5//HHMnTsXffv2xfnz5xtc9/jx47DZbBgxYkSz\n1ZOVlQWbzYbevXvb28xmM4KCgnD06FFkZmaiqKgI999/v8N67u7umDdvntPbOXz4MC5evIgBAwY4\ntA8cOBBGoxH79u1rdH2j0QgAjV7e6vIZuxuaUszLy4OIOLUP7wwPDw/MnTsXR44cwd69ezF+/Hjk\n5eVh4sSJKCwsbJZtWK1WTJkyBYWFhdi2bRsAICkpCXPmzLE/Jo25fF9zc3ObpZ4bdsxgMBjQuXNn\nzJgxA6tXr0ZWVhaeffbZBvufOXMGwE/XM2suZWVlAIClS5c6zHmfOnUKNpsNJSUlAABfX9/r2s7l\nyYF27drVWebr64vS0tImj/nWW29h2LBhCAgIgKenJ5588slG+1+6dAkA4Onp2eRtXc0vf/lL/O//\n/i9mz56N/Px8fPTRR8029uUD6Y0bN6KoqAjbt2/HrFmznFr38szW5ft+vVzyCXSfPn0AAN99912D\nfS7Pfly+2mVzuByspKQk+zvW5VtGRob92mqNvWM543KY6nvRFxUVNflUMKdPn8aECRMQFBSEffv2\nobi4uM51on/u8gvjWj6M+uSTT5CUlGT/OyYmpt4py4ceeggAYLPZmryNhvTr1w+DBg3Cl19+iZkz\nZ2LixInw8/Nzat3KykoAcJjuvR4uCcNXX30FALj11lsb7NO7d2+4ubnh448/brbthoSEwGQyOXx6\nfaUuXbrA398f77333nVtp3fv3mjXrh0OHDjg0L5v3z5UVlY2+bJUhw4dQlVVFebMmYPw8HCYTKar\nTjFf/kS6uLi4yfV/9dVX8PLysv9dUVFR7z+uy7M+ffv2bfI2GnP53SE9Pb1JZxa8fF8DAwObpY5m\nD0N5eTlqa2shIsjJycGWLVuwdOlSdOjQodE7GhAQgJiYGKSnp2Pz5s0oKSlBZmbmdV0Y3GQyYcaM\nGUhJSUFycjJKSkpQU1ODM2fO4OzZs/D09MTixYvxySefYO7cufjxxx9RW1uL0tJS+4vB398fOTk5\nOHnyJEpLS+vdrzeZTFiwYAF27tyJ1157DSUlJTh06BBmz56NTp06YebMmU2qOzQ0FMBPX6m4dOkS\nvv/++6sed1gsFoSHh9t3N51RVVWF3Nxc7NmzxyEMADBhwgSkpaWhqKgIxcXF2L17N/74xz9i3Lhx\nzR6GSZMmoUOHDpgwYQLCw8OdXu/yfb2853HdmnC0Xa+dO3c2OJPk6ekpkZGRMmfOHDl9+rR9nXXr\n1klQUJAAEIvFImPHjhURkdLSUnn44Yelffv20q5dOxkyZIgsW7ZMAEhwcLD861//qnfdlStXitls\nFgASEhIiW7dutW+roqJCEhMTJTQ0VAwGgwQEBEhMTIwcPnzY3mf9+vXSp08fMZlMYjKZ5Be/+IVs\n2LBBRES+/vprCQsLE7PZLEOGDJGlS5fWW3ttba2sWrVKIiMjxcPDQ/z8/GTChAmSlZVl386GDRvE\nYrEIAImMjJTs7GzZtGmT+Pj4CAAJCwuTY8eOiYhIYmKi+Pv7i6+vr0ycOFHWr18vACQiIkLmz58v\ngYGBAkC8vLwkOjpaRETmzp0rHh4e9pmZqz0/V9527txpX+e9996TuLg4iYiIEE9PTzEajXLrrbfK\n8uXL5dKlS3VeAyUlJfKrX/1K/P39BYC4ublJt27d5Omnn27wtdKhQwd57LHH7MuefPJJ+eKLL+x/\nX/k4u7m5Sa9eveTTTz91GO/BBx+Uzp07S21tbf0vznq4ZGqV9Pf999+LwWBw+GfQWp0/f15MJpOs\nXr26Seu5bGqV9NWtWzesWLECK1aswMWLF/Uu54Zavnw5+vXrh7lz5zbbmAxDK7No0SJMnDgR8fHx\n13Qw3RKsWbMGBw8exNtvv93gxeWvBcPQCj399NOYO3duo5/rtFS7d+9GRUUF9uzZ4/QUrLMavD4D\ntWz33Xdfo190a6nGjRuHcePG3ZCx+c5ApDAMRArDQKQwDEQKw0CkMAxECsNApDAMRArDQKQwDEQK\nw0CkMAxECsNApDT4rVW9z3FK5Gp1whAVFVXnXKR042RkZGDt2rV8zG8CmojOFxto49LS0hAXF6f7\nNR8I23nMQKQwDEQKw0CkMAxECsNApDAMRArDQKQwDEQKw0CkMAxECsNApDAMRArDQKQwDEQKw0Ck\nMAxECsNApDAMRArDQKQwDEQKw0CkMAxECsNApDAMRArDQKQwDEQKw0CkMAxECsNApDAMRArDQKQw\nDERKg5exouZXXl6Os2fPOrTl5uYCAE6cOOHQ7u7ujrCwMJfVRrxyj0sVFBQgKCgI1dXVV+37wAMP\n4J133nFBVaTwyj2u1L59e9x7771wc2v8Ydc0DfHx8S6qii5jGFxs2rRpV12phv4AABMuSURBVL1+\nm8FgwPjx411UEV3GMLjYuHHj4Onp2eByg8GAsWPHwmq1urAqAhgGl/Py8sK4cePg4eFR7/KamhpM\nnTrVxVURwDDoYurUqaiqqqp3mdlsxsiRI11cEQEMgy4eeOAB+Pj41Gn38PBAXFwcTCaTDlURw6AD\nDw8PTJo0qc6uUlVVFaZMmaJTVcQw6GTKlCl1dpXat2+P4cOH61QRMQw6ufvuu9GxY0f730ajEdOm\nTYO7u7uOVbVtDINO3NzcMG3aNBiNRgBAZWUlJk+erHNVbRvDoKPJkyejsrISABAcHIw777xT54ra\nNoZBRwMGDEDXrl0BANOnT4emaTpX1La1ym+trlmzBhkZGXqX4RSz2QwA+PLLLzFx4kSdq3HOH/7w\nBwwePFjvMppdq3xnyMjIwN69e/UuwykhISGwWq31fu5wM0pPT8cPP/ygdxk3RKt8ZwCAQYMGYfv2\n7XqX4ZR3330X999/v95lOKU178q1yneGlqalBKG1YxiIFIaBSGEYiBSGgUhhGIgUhoFIYRiIFIaB\nSGEYiBSGgUhhGIgUhoFIYRiIFIYBwI4dOxAeHg5N0xxuRqMRHTt2xLBhw7Bq1SoUFhbqXSrdQAwD\ngJiYGJw4cQIRERGwWq0QEdTW1iIvLw9paWno2rUrEhMTcdttt+HAgQN6l0s3CMPQAE3T4Ovri2HD\nhmHLli1IS0tDbm4uHnzwQRQXF+td3nUrLy9HVFSU3mXcVBgGJ8XGxmL69OnIy8vDxo0b9S7num3e\nvBl5eXl6l3FTYRiaYPr06QBgv6LO888/D4vFAm9vb+Tl5WHBggXo3LkzsrKyICJYs2YNevbsCU9P\nT/j5+WH8+PE4evSofbyXXnoJJpMJHTt2xKxZs9CpUyeYTCZERUVh3759Dtt2Zry5c+fCaDQiKCjI\n3va73/0OXl5e0DQN58+fBwDMnz8fCxYsQHZ2NjRNQ7du3W7UQ9aySCsUGxsrsbGxTV4vIiJCrFZr\ng8tLSkoEgISEhNjblixZIgBk3rx5sm7dOomOjpYjR47IsmXLxGg0ytatW6WoqEgyMzOlf//+0qFD\nBzl37px9/ZkzZ4qXl5d89913cunSJTl8+LAMHDhQvL295fTp0/Z+zo43depUCQwMdKh71apVAkDy\n8/PtbTExMRIREdHkxwiApKamNnm9FiCN7wxN4O3tDU3TUFpaWmfZc889h8ceeww7duxAWFgY1qxZ\ng+joaEybNg1WqxV9+vTBxo0bcf78eWzatMlhXYPBYP+P36tXLyQnJ6O0tBRbtmwB8NP+fVPGo2vD\nMDRBWVkZROSqp3U5fPgwLl68iAEDBji0Dxw4EEajsc4u0M8NGDAAFovFvgt0veORcxiGJjh27BgA\noEePHo32KyoqAgC0a9euzjJfX99631l+ztPTE/n5+c02Hl0dw9AE//znPwHgqlfW8fX1BYB6X6RF\nRUUIDg5udP2qqiqHftc7HjmHYXDSuXPnkJSUhODgYPz2t79ttG/v3r3Rrl27Oh/Q7du3D5WVlbjj\njjsaXX/Pnj0QEQwaNKjJ4xkMhgYvkUWNYxh+RkRw8eJF1NbWQkSQn5+P1NRU3HXXXXB3d8euXbuu\nesxgMpmwYMEC7Ny5E6+99hpKSkpw6NAhzJ49G506dcLMmTMd+tfW1qKwsBDV1dXIzMzE/PnzERoa\nap/Kbcp43bp1w4ULF7Br1y5UVVUhPz8fp06dqlOjv78/cnJycPLkSZSWljJAAKdWRURef/116du3\nr1gsFjEajeLm5iYARNM08fX1lTvvvFNWrFghBQUFDuutXLlSzGazfbp169at9mW1tbWyatUqiYyM\nFA8PD/Hz85MJEyZIVlaWwxgzZ84UDw8P6dy5sxgMBvHx8ZHx48dLdna2Qz9nxysoKJDhw4eLyWSS\nrl27yu9//3t54oknBIB069bNPl379ddfS1hYmJjNZhkyZIjD9Gxj0IqnVjWRq1yhuwW6fDbrlnCu\n1VmzZmH79u0oKCjQuxSnaJqG1NRUTJo0Se9Smtt27ibdBGpqavQugcBjBiI7hkFHixcvxpYtW1Bc\nXIyuXbsiPT1d75LatFZ7fYaW4JlnnsEzzzyjdxmk8J2BSGEYiBSGgUhhGIgUhoFIYRiIFIaBSGEY\niBSGgUhhGIgUhoFIYRiIFIaBSGm131rdu3ev/RdvRM5olWEYPHiw3iU4LScnBwcOHMDYsWP1LsUp\nsbGxCAkJ0buMG6JV/ga6JUlLS0NcXBz4NOiOv4EmuoxhIFIYBiKFYSBSGAYihWEgUhgGIoVhIFIY\nBiKFYSBSGAYihWEgUhgGIoVhIFIYBiKFYSBSGAYihWEgUhgGIoVhIFIYBiKFYSBSGAYihWEgUhgG\nIoVhIFIYBiKFYSBSGAYihWEgUhgGIoVhIFIYBiKlVV7G6mb1448/YsyYMaiqqrK3lZWVoV27dujT\np49D3379+mHr1q2uLrFNYxhcqHPnzrh06RKOHDlSZ9m3337r8HdcXJyryiKFu0kulpCQAIPh6v+D\nGAbXYxhcbMqUKaipqWlwuaZp6N+/PyIjI11YFQEMg8uFhoZi4MCBcHOr/6F3d3dHQkKCi6sigGHQ\nRUJCAjRNq3dZTU0NL+auE4ZBB5MmTaq33d3dHXfffTduueUWF1dEAMOgi4CAAAwbNgzu7u51lj30\n0EM6VEQAw6Cbhx56CCLi0Obm5obo6GidKiKGQSfR0dEOU6wGgwEjR46Er6+vjlW1bQyDTry9vTF6\n9Gh4eHgA+OnAedq0aTpX1bYxDDqaOnUqqqurAQAmkwmjR4/WuaK2jWHQ0ahRo2CxWAAAMTExMJvN\nOlfUtrWp7yZlZGTghx9+0LsMBwMHDsSePXsQEhKCtLQ0vctxEBUVheDgYL3LcBlNfj6l0YpNnDgR\n6enpepfRYqSmpjb4mUgrtL3N7SbFxsZCRG6aW3V1NVasWKF7HT+/tUVtLgw3G3d3dyxatEjvMggM\nw03Bma90043HMBApDAORwjAQKQwDkcIwECkMA5HCMBApDAORwjAQKQwDkcIwECkMA5HCMDRix44d\nCA8Ph6ZpDd66dOkCAFi9ejU6duwITdOwceNGfQuna8IwNCImJgYnTpxAREQErFarw28QbDYbcnNz\n7T/bXLhwIb744gudK6brwTBcA3d3d5jNZnTs2BHdu3e/rrHKy8sRFRV11Ta68RiG67Rr167rWn/z\n5s3Iy8u7ahvdeAzDDfbpp5+iV69esFqtMJlM6NOnD959910AwPz587FgwQJkZ2dD0zR069at3jbg\np/MqLVu2DKGhoTCbzejbty9SU1MBAMnJyfDy8oLFYsHu3bsxcuRI+Pj4IDg4GCkpKbrd95aGYbhG\nH374IVavXn3Vfrm5uYiLi8PJkyeRk5ODdu3aYerUqQCAtWvXYsyYMYiIiICI4Pjx4/W2AcAf//hH\nPP/880hKSsLZs2cxZswYTJkyBQcOHMCcOXPw+OOPo7y8HN7e3khNTUV2djbCw8PxyCOPOFw2ixrG\nMDipuLjYYRZpxIgRTq0XGxuLp556Cn5+fvD398fYsWNRUFCA/Px8p7d96dIlJCcnY8KECYiJiYGv\nry+WLl0KDw8PbNmyxaFvVFQUfHx8EBAQgPj4eJSVleH06dNNuq9tFcPgpCtnk0QEH3300TWNc+Xp\nJJ2VlZUFm82G3r1729vMZjOCgoJw9OjRBtczGo0AwHcGJzEM12jYsGFYuHDhVfu99dZbGDZsGAIC\nAuDp6Yknn3yyydsqKysDACxdutTh3enUqVOw2WxNHo/qxzDcQKdPn8aECRMQFBSEffv2obi4GCtX\nrmzyOAEBAQCApKSkOuc3ysjIaO6y2yyeo+QGOnToEKqqqjBnzhyEh4cDQIOXr2pMSEgITCYTDh48\n2Nwl0hX4znADhYaGAgDef/99XLp0Cd9//z327dvn0Mff3x85OTk4efIkSktLUVVVVafN3d0dM2bM\nQEpKCpKTk1FSUoKamhqcOXMGZ8+e1eOutU7ShsTGxkpsbKzT/T///HPp3r27ABAAEhQUJCNGjKi3\n7wsvvCCBgYECQLy8vCQ6OlpERBITE8Xf3198fX1l4sSJsn79egEgERERcvr0afn6668lLCxMzGaz\nDBkyRM6dO1dvW0VFhSQmJkpoaKgYDAYJCAiQmJgYOXz4sGzYsEEsFosAkMjISMnOzpZNmzaJj4+P\nAJCwsDA5duxYkx4rAJKamtqkdVq4tDZ34mEA2L59u86V3Pw0TeOJh4naKoaBSGEYiBSGgUhhGIgU\nhoFIYRiIFIaBSGEYiBSGgUhhGIgUhoFIYRiIFIaBSGEYiBSGgUhhGIiUNndCgDNnziAtLU3vMugm\n1ObCsHfvXsTFxeldBt2E2tRvoG9GaWlpiIuLA58G3fE30ESXMQxECsNApDAMRArDQKQwDEQKw0Ck\nMAxECsNApDAMRArDQKQwDEQKw0CkMAxECsNApDAMRArDQKQwDEQKw0CkMAxECsNApDAMRArDQKQw\nDEQKw0CkMAxECsNApDAMRArDQKQwDEQKw0CkMAxECsNApLS5y1jpKTc3F3//+98d2jIzMwEAK1eu\ndGj38/PDo48+6qrSCLyMlUtVV1cjMDAQxcXFMBj+839IRKBpmv3viooKPPLII9i0aZMeZbZVvIyV\nKxkMBsTHx8PNzQ0VFRX2W2VlpcPfADBlyhSdq217GAYXmzx5MqqqqhrtExAQgKFDh7qoIrqMYXCx\nu+66C7fcckuDy41GIxISEuDu7u7CqghgGFxO0zRMmzYNHh4e9S6vrKzE5MmTXVwVAQyDLhrbVQoL\nC8Mdd9zh4ooIYBh00a9fP0RGRtZpNxqNmD59uusLIgAMg24SEhLq7CpVVlYiLi5Op4qIYdDJ5MmT\nUV1dbf9b0zT07dsXPXv21LGqto1h0ElERAT69esHN7efngKDwYCEhASdq2rbGAYdJSQk2MNQXV3N\nXSSdMQw6iouLQ21tLQBg8ODBCA4O1rmito1h0FGnTp3snzT/5je/0bkaanNf1EtLS+PuiBPa2MsC\nALa32a9wp6am6l0CAKCsrAybNm3C448/rncpAICMjAysXbtW7zJ00WbDMGnSJL1LsLv33ntvquOF\nthoGHjPcBG6mILRlDAORwjAQKQwDkcIwECkMA5HCMBApDAORwjAQKQwDkcIwECkMA5HCMBApDAOR\nwjA4YceOHQgPD4emaQ43o9GIjh07YtiwYVi1ahUKCwv1LpWuA8PghJiYGJw4cQIRERGwWq0QEdTW\n1iIvLw9paWno2rUrEhMTcdttt+HAgQN6l0vXiGG4RpqmwdfXF8OGDcOWLVuQlpaG3NxcPPjggygu\nLta7PLoGDEMziY2NxfTp05GXl4eNGzfqXQ5dA4ahGV0+T+o777xjb6upqcGyZcsQGhoKs9mMvn37\n2n9/nZycDC8vL1gsFuzevRsjR46Ej48PgoODkZKS4jD2xx9/jDvvvBMWiwU+Pj7o06cPSkpKrroN\nagJpY1JTU+Va73ZERIRYrdYGl5eUlAgACQkJsbctXLhQPD09JT09XQoLC2Xx4sXi5uYm+/fvFxGR\nJUuWCAD54IMPpLi4WPLy8mTo0KHi5eUllZWVIiJy8eJF8fHxkZUrV0p5ebmcO3dOoqOjJT8/36lt\nNMX1PD4tXFqbu9c3MgwiIpqmia+vr4iIlJeXi8Vikfj4ePtym80mnp6eMmfOHBH5TxjKy8vtfTZs\n2CAA5Pjx4yIi8u233woAefPNN+tsz5ltNEVbDgN3k5pRWVkZRAQ+Pj4AgKysLNhsNvTu3dvex2w2\nIygoCEePHm1wHKPRCAD2aziEh4ejY8eOmDZtGpYvX46TJ0/a+17rNqguhqEZHTt2DADQo0cPAD+F\nAwCWLl3q8PnEqVOnYLPZnB7XbDbjww8/xJAhQ/D0008jPDwc8fHxKC8vb7ZtEMPQrP75z38CAEaO\nHAngpwsVAkBSUhJExOGWkZHRpLFvu+02vPHGG8jJyUFiYiJSU1OxevXqZt1GW8cwNJNz584hKSkJ\nwcHB+O1vfwsACAkJgclkwsGDB69r7JycHHz33XcAfgrYs88+i/79++O7775rtm0Qw9BkIoKLFy+i\ntrYWIoL8/Hykpqbirrvugru7O3bt2mU/ZjCZTJgxYwZSUlKQnJyMkpIS1NTU4MyZMzh79qzT28zJ\nycGsWbNw9OhRVFZW4ptvvsGpU6cwaNCgZtsGoe1NG1zLbMnrr78uffv2FYvFIkajUdzc3ASAfebo\nzjvvlBUrVkhBQUGddSsqKiQxMVFCQ0PFYDBIQECAxMTEyOHDh2XDhg1isVgEgERGRkp2drZs2rRJ\nfHx8BICEhYXJsWPH5OTJkxIVFSV+fn7i7u4ut9xyiyxZskSqq6uvug1XPD6tRFqbPQt3G7vbTmvD\nj8927iYRKQwDkcIwECkMA5HCMBApDAORwjAQKQwDkcIwECkMA5HCMBApDAORwjAQKQwDkcIwECkM\nA5HCMBApBr0L0IumaXqXQDeZNheGqKgonoeU6tXmfgNN1AD+BproMoaBSGEYiBQDgO16F0F0E9j7\n/wEXWN8z8MiCIgAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<IPython.core.display.Image object>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 173
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "noAxq9pOwUrb",
        "colab_type": "text"
      },
      "source": [
        "# RNN (GRU) Model\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cq0ES8O8djoG",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 407
        },
        "outputId": "0e580e5f-9377-467a-ed6d-7d5fdb08f791"
      },
      "source": [
        "model = get_main_model(word_index, WORD_MODEL, EMBED_SIZE, MAX_TEXT_LENGTH, \n",
        "                       layer= GRU(units=16, return_sequences=True, return_state=False), \n",
        "                       dropout=0.5, optimizer=optimizers.Adam())"
      ],
      "execution_count": 181,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Building embedding matrix...\n",
            "Embedding matrix built.\n",
            "Model: \"model_18\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_18 (InputLayer)        (None, 84)                0         \n",
            "_________________________________________________________________\n",
            "embedding_18 (Embedding)     (None, 84, 300)           9037800   \n",
            "_________________________________________________________________\n",
            "gru_4 (GRU)                  (None, 84, 16)            15216     \n",
            "_________________________________________________________________\n",
            "dropout_18 (Dropout)         (None, 84, 16)            0         \n",
            "_________________________________________________________________\n",
            "flatten_18 (Flatten)         (None, 1344)              0         \n",
            "_________________________________________________________________\n",
            "dense_18 (Dense)             (None, 2)                 2690      \n",
            "=================================================================\n",
            "Total params: 9,055,706\n",
            "Trainable params: 17,906\n",
            "Non-trainable params: 9,037,800\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SGyK0YxHeHxU",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "83e3708b-7c0c-4a2b-c2fd-486150f72cd5"
      },
      "source": [
        "time_start = time()\n",
        "history, model = train_fit_predict(model,\n",
        "                               x_train[:, :MAX_TEXT_LENGTH],\n",
        "                               x_dev[:, :MAX_TEXT_LENGTH],\n",
        "                               train_y_cat, dev_cat_y, class_weight=None,\n",
        "                               batch_size=500, epochs=15)\n",
        "time_start = time() - time_start\n",
        "\n",
        "print(\"Took : \"+str(np.round(time_start, 2))+\" (s)\") "
      ],
      "execution_count": 182,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 7000 samples, validate on 1000 samples\n",
            "Epoch 1/15\n",
            "7000/7000 [==============================] - 8s 1ms/step - loss: 0.5246 - acc: 0.8624 - val_loss: 0.3077 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.3077390531301498, acc: 0.956\n",
            "\n",
            "Epoch 2/15\n",
            "7000/7000 [==============================] - 4s 540us/step - loss: 0.2563 - acc: 0.9484 - val_loss: 0.2055 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.20550610518455506, acc: 0.956\n",
            "\n",
            "Epoch 3/15\n",
            "7000/7000 [==============================] - 4s 536us/step - loss: 0.2139 - acc: 0.9484 - val_loss: 0.1780 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.17800627326965332, acc: 0.956\n",
            "\n",
            "Epoch 4/15\n",
            "7000/7000 [==============================] - 4s 538us/step - loss: 0.1917 - acc: 0.9484 - val_loss: 0.1686 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.16861123284697532, acc: 0.956\n",
            "\n",
            "Epoch 5/15\n",
            "7000/7000 [==============================] - 4s 536us/step - loss: 0.1811 - acc: 0.9484 - val_loss: 0.1600 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.16001157158613205, acc: 0.956\n",
            "\n",
            "Epoch 6/15\n",
            "7000/7000 [==============================] - 4s 538us/step - loss: 0.1708 - acc: 0.9484 - val_loss: 0.1496 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.14955879917740822, acc: 0.956\n",
            "\n",
            "Epoch 7/15\n",
            "7000/7000 [==============================] - 4s 542us/step - loss: 0.1549 - acc: 0.9486 - val_loss: 0.1383 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.13826877397298812, acc: 0.956\n",
            "\n",
            "Epoch 8/15\n",
            "7000/7000 [==============================] - 4s 544us/step - loss: 0.1430 - acc: 0.9489 - val_loss: 0.1290 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.12900964047014712, acc: 0.956\n",
            "\n",
            "Epoch 9/15\n",
            "7000/7000 [==============================] - 4s 544us/step - loss: 0.1343 - acc: 0.9511 - val_loss: 0.1220 - val_acc: 0.9560\n",
            "\n",
            "Testing loss: 0.12198059518635274, acc: 0.956\n",
            "\n",
            "Epoch 10/15\n",
            "7000/7000 [==============================] - 4s 538us/step - loss: 0.1270 - acc: 0.9524 - val_loss: 0.1200 - val_acc: 0.9570\n",
            "\n",
            "Testing loss: 0.12004500159621238, acc: 0.957\n",
            "\n",
            "Epoch 11/15\n",
            "7000/7000 [==============================] - 4s 539us/step - loss: 0.1245 - acc: 0.9520 - val_loss: 0.1159 - val_acc: 0.9610\n",
            "\n",
            "Testing loss: 0.11591215047240257, acc: 0.961\n",
            "\n",
            "Epoch 12/15\n",
            "7000/7000 [==============================] - 4s 544us/step - loss: 0.1194 - acc: 0.9551 - val_loss: 0.1161 - val_acc: 0.9590\n",
            "\n",
            "Testing loss: 0.11613988638296724, acc: 0.959\n",
            "\n",
            "Epoch 13/15\n",
            "7000/7000 [==============================] - 4s 548us/step - loss: 0.1142 - acc: 0.9570 - val_loss: 0.1180 - val_acc: 0.9600\n",
            "\n",
            "Testing loss: 0.11795939061045646, acc: 0.96\n",
            "\n",
            "Epoch 14/15\n",
            "7000/7000 [==============================] - 4s 534us/step - loss: 0.1107 - acc: 0.9567 - val_loss: 0.1143 - val_acc: 0.9570\n",
            "\n",
            "Testing loss: 0.11427535088360309, acc: 0.957\n",
            "\n",
            "Epoch 15/15\n",
            "7000/7000 [==============================] - 4s 539us/step - loss: 0.1083 - acc: 0.9557 - val_loss: 0.1155 - val_acc: 0.9570\n",
            "\n",
            "Testing loss: 0.11553058420866728, acc: 0.957\n",
            "\n",
            "Took : 66.89 (s)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8ENyBQVzejz9",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "4ad2ad2b-bf24-40ea-91a0-f18bdcff36f7"
      },
      "source": [
        "model.evaluate(x_dev[:, :MAX_TEXT_LENGTH], dev_cat_y, batch_size=1000)"
      ],
      "execution_count": 183,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\r1000/1000 [==============================] - 0s 229us/step\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.1155305728316307, 0.9570000171661377]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 183
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZoiPJ2MXejop",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 176
        },
        "outputId": "5f68fda3-d71e-4ea1-d2c8-b9118ad8078c"
      },
      "source": [
        "y_pred = np.argmax(model.predict(x_dev[:, :MAX_TEXT_LENGTH]), axis=1)\n",
        "\n",
        "print(creport(np.argmax(dev_cat_y, axis=1), y_pred,target_names=['HS', 'NOT_HS'],digits=4))"
      ],
      "execution_count": 184,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "          HS     0.5263    0.2273    0.3175        44\n",
            "      NOT_HS     0.9653    0.9906    0.9778       956\n",
            "\n",
            "    accuracy                         0.9570      1000\n",
            "   macro avg     0.7458    0.6089    0.6476      1000\n",
            "weighted avg     0.9460    0.9570    0.9487      1000\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4UMgjml0i4nQ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 564
        },
        "outputId": "a47588c4-74f7-4c98-9ba7-f3977b41ce7f"
      },
      "source": [
        "plot_model(model, to_file='RNN_GRU_model.png', show_shapes=False, show_layer_names=False)"
      ],
      "execution_count": 180,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAHsAAAIjCAYAAAAjhWTMAAAABmJLR0QA/wD/AP+gvaeTAAAgAElE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\n",
            "text/plain": [
              "<IPython.core.display.Image object>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 180
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EjBmZsD8wbF2",
        "colab_type": "text"
      },
      "source": [
        "# CNN Model\n",
        "[Colab Notebook](https://colab.research.google.com/drive/1_USllONVZIbf88aoPuQrzg9sg5HtY7K7)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ShqpL2T8wdaa",
        "colab_type": "text"
      },
      "source": [
        "# Combined CNN-RNN Models\n",
        "[Colab Notebook](https://colab.research.google.com/drive/1LbZdJNvTzTqu6qLjDbOaeccJl36anJEw)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PMeeL-Gqfy6p",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}