Abstractive_Text_Summary_Hindi.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "id": "nHOtn5-Yzutf",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "d56c86c1-4555-4b8c-9517-a432ba3e435e"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "id": "arwL40pnz0T1"
      },
      "outputs": [],
      "source": [
        "default_path = \"<drive path to files>\""
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cGei2JacaO6C"
      },
      "source": [
        "### Loading Data"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "id": "rVXbBfAy0bxJ"
      },
      "outputs": [],
      "source": [
        "import pandas as pd\n",
        "import torch\n",
        "import random\n",
        "import numpy as np"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xor2OxjJL8SJ"
      },
      "source": [
        "Training Data"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "id": "La5ar_12zt8J",
        "outputId": "cf0024ce-ee83-4f06-d3d6-937333536b18"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "                                            headline  \\\n",
              "0  EXCLUSIVE: दिल्ली में डीजल टैक्सियों पर बैन से...   \n",
              "1  जॉर्डन: राष्ट्रपति मुखर्जी ने 86 करोड़ डॉलर के...   \n",
              "2  UN में पाकिस्तान की राजदूत मलीहा लोधी ने कराई ...   \n",
              "3  38 देशों में पीएम नरेंद्र मोदी बायोपिक को रिली...   \n",
              "4           13 अगस्त 2011: दिनभर की बड़ी खबरें पढ़ें   \n",
              "\n",
              "                                             article  \n",
              "0  दिल्ली में सुप्रीम कोर्ट के डीज़ल टैक्सियों को...  \n",
              "1  जॉर्डन के ऐतिहासिक दौरे पर पहुंचे राष्ट्रपति प...  \n",
              "2  पाकिस्तानी नेताओं को विवादित और हास्यास्पद बया...  \n",
              "3  पीएम नरेंद्र मोदी बायोपिक में विवेक ओबेरॉय ने ...  \n",
              "4  देश, दुनिया, महानगर, खेल, आर्थिक और बॉलीवुड मे...  "
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-17570157-1891-4056-8086-dea82d1c4763\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>headline</th>\n",
              "      <th>article</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>EXCLUSIVE: दिल्ली में डीजल टैक्सियों पर बैन से...</td>\n",
              "      <td>दिल्ली में सुप्रीम कोर्ट के डीज़ल टैक्सियों को...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>जॉर्डन: राष्ट्रपति मुखर्जी ने 86 करोड़ डॉलर के...</td>\n",
              "      <td>जॉर्डन के ऐतिहासिक दौरे पर पहुंचे राष्ट्रपति प...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>UN में पाकिस्तान की राजदूत मलीहा लोधी ने कराई ...</td>\n",
              "      <td>पाकिस्तानी नेताओं को विवादित और हास्यास्पद बया...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>38 देशों में पीएम नरेंद्र मोदी बायोपिक को रिली...</td>\n",
              "      <td>पीएम नरेंद्र मोदी बायोपिक में विवेक ओबेरॉय ने ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>13 अगस्त 2011: दिनभर की बड़ी खबरें पढ़ें</td>\n",
              "      <td>देश, दुनिया, महानगर, खेल, आर्थिक और बॉलीवुड मे...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-17570157-1891-4056-8086-dea82d1c4763')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-17570157-1891-4056-8086-dea82d1c4763 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-17570157-1891-4056-8086-dea82d1c4763');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 4
        }
      ],
      "source": [
        "train = pd.read_csv(default_path+\"train.csv\")\n",
        "train.head()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "pbKAsewzzt8P",
        "outputId": "d7066e97-de8a-4e50-8c7b-d186de71418c"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0    EXCLUSIVE: दिल्ली में डीजल टैक्सियों पर बैन से...\n",
            "Name: headline, dtype: object\n"
          ]
        }
      ],
      "source": [
        "print(train.head(1)[\"headline\"])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "WexyzTOH3guJ",
        "outputId": "97d4dec0-25a6-4857-83e3-22f31b0952b3"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(266607, 2)"
            ]
          },
          "metadata": {},
          "execution_count": 6
        }
      ],
      "source": [
        "train.shape"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BP8lJ9HqMATI"
      },
      "source": [
        "Testing Data"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "U6XSQtvQMB8x",
        "outputId": "a24a3967-28c7-4ae9-907a-1cba46d9c4a3"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(66653, 2)"
            ]
          },
          "metadata": {},
          "execution_count": 7
        }
      ],
      "source": [
        "test = pd.read_csv(default_path+\"news_data.csv\")\n",
        "test.shape"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "id": "eeJBqzfZkKpK",
        "outputId": "9e4faadd-b930-469b-c064-5a582fb0dcc5"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "                                            headline  \\\n",
              "0  पठानकोट पहुंचे PM मोदी, एयरबेस का जायजा ले बॉर...   \n",
              "1  सचिन ने देशवासियों को समर्पित किया अपना दोहरा शतक   \n",
              "2  एनआईए करेगी छत्तीसगढ़ में सुरक्षा खामियों की ज...   \n",
              "3  सीधी बात:  शाह बोले- हमारा बस चलता तो अब तक मं...   \n",
              "4  ऋषभ पंत के पास यूनिक टैलेंट, उसके साथ छेड़छाड़ न...   \n",
              "\n",
              "                                             article  \n",
              "0  प्रधानमंत्री नरेंद्र मोदी पठानकोट एयरबेस पहुंच...  \n",
              "1  सचिन तेंदुलकर ने एकदिवसीय अंतरराष्ट्रीय क्रिके...  \n",
              "2  केंद्रीय गृह राज्य मंत्री आर. पी. एन. सिंह ने ...  \n",
              "3  भारतीय जनता पार्टी (बीजेपी) के राष्ट्रीय अध्यक...  \n",
              "4  ऋषभ पंत की कभी कभार इस बात के लिए आलोचना की जा...  "
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-96c29489-5c61-4cf2-b041-ad48ca5478bb\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>headline</th>\n",
              "      <th>article</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>पठानकोट पहुंचे PM मोदी, एयरबेस का जायजा ले बॉर...</td>\n",
              "      <td>प्रधानमंत्री नरेंद्र मोदी पठानकोट एयरबेस पहुंच...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>सचिन ने देशवासियों को समर्पित किया अपना दोहरा शतक</td>\n",
              "      <td>सचिन तेंदुलकर ने एकदिवसीय अंतरराष्ट्रीय क्रिके...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>एनआईए करेगी छत्तीसगढ़ में सुरक्षा खामियों की ज...</td>\n",
              "      <td>केंद्रीय गृह राज्य मंत्री आर. पी. एन. सिंह ने ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>सीधी बात:  शाह बोले- हमारा बस चलता तो अब तक मं...</td>\n",
              "      <td>भारतीय जनता पार्टी (बीजेपी) के राष्ट्रीय अध्यक...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>ऋषभ पंत के पास यूनिक टैलेंट, उसके साथ छेड़छाड़ न...</td>\n",
              "      <td>ऋषभ पंत की कभी कभार इस बात के लिए आलोचना की जा...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-96c29489-5c61-4cf2-b041-ad48ca5478bb')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-96c29489-5c61-4cf2-b041-ad48ca5478bb button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-96c29489-5c61-4cf2-b041-ad48ca5478bb');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 8
        }
      ],
      "source": [
        "test.head(5)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "id": "ZcbOoPvRnGPp"
      },
      "outputs": [],
      "source": [
        "news = pd.concat([train, test], ignore_index=True, sort=False)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "9ZgquVOmC8hz",
        "outputId": "5db9d80e-85f3-4158-c5c7-869f0b7a41a7"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(333260, 2)"
            ]
          },
          "metadata": {},
          "execution_count": 10
        }
      ],
      "source": [
        "news.shape"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oEnvEYxgxvOS"
      },
      "source": [
        "### Cleaning"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "UN3pXFKE18Qf"
      },
      "outputs": [],
      "source": [
        "import re\n",
        "def remove_emojis(data):\n",
        "    emoj = re.compile(\"[\"\n",
        "        u\"\\U0001F600-\\U0001F64F\"  # emoticons\n",
        "        u\"\\U0001F300-\\U0001F5FF\"  # symbols & pictographs\n",
        "        u\"\\U0001F680-\\U0001F6FF\"  # transport & map symbols\n",
        "        u\"\\U0001F1E0-\\U0001F1FF\"  # flags (iOS)\n",
        "        u\"\\U00002500-\\U00002BEF\"  # chinese char\n",
        "        u\"\\U00002702-\\U000027B0\"\n",
        "        u\"\\U00002702-\\U000027B0\"\n",
        "        u\"\\U000024C2-\\U0001F251\"\n",
        "        u\"\\U0001f926-\\U0001f937\"\n",
        "        u\"\\U00010000-\\U0010ffff\"\n",
        "        u\"\\u2640-\\u2642\" \n",
        "        u\"\\u2600-\\u2B55\"\n",
        "        u\"\\u200d\"\n",
        "        u\"\\u23cf\"\n",
        "        u\"\\u23e9\"\n",
        "        u\"\\u231a\"\n",
        "        u\"\\ufe0f\"  # dingbats\n",
        "        u\"\\u3030\"\n",
        "                      \"]+\", re.UNICODE)\n",
        "    return re.sub(emoj, ' ', data)\n",
        "\n",
        "def preprocess_tokenize(text):\n",
        "      # for removing punctuation from sentencesc\n",
        "    text = str(text)\n",
        "    text = re.sub(r'(\\d+)', r'', text)\n",
        "    \n",
        "    text = text.replace('\\n', ' ')\n",
        "    text = text.replace('\\r', ' ')\n",
        "    text = text.replace('\\t', ' ')\n",
        "    text = text.replace('\\u200d', '')\n",
        "    text=re.sub(\"(__+)\", ' ', str(text)).lower()   #remove _ if it occors more than one time consecutively\n",
        "    text=re.sub(\"(--+)\", ' ', str(text)).lower()   #remove - if it occors more than one time consecutively\n",
        "    text=re.sub(\"(~~+)\", ' ', str(text)).lower()   #remove ~ if it occors more than one time consecutively\n",
        "    text=re.sub(\"(\\+\\++)\", ' ', str(text)).lower()   #remove + if it occors more than one time consecutively\n",
        "    text=re.sub(\"(\\.\\.+)\", ' ', str(text)).lower()   #remove . if it occors more than one time consecutively\n",
        "    text=re.sub(r\"[<>()|&©@#ø\\[\\]\\'\\\",;:?.~*!]\", ' ', str(text)).lower() #remove <>()|&©ø\"',;?~*!\n",
        "    text = re.sub(r\"[‘’।:]\", \" \", str(text)) #removing other special characters\n",
        "    text = re.sub(\"([a-zA-Z])\",' ',str(text)).lower()\n",
        "    text = re.sub(\"(\\s+)\",' ',str(text)).lower()\n",
        "    text = remove_emojis(text)\n",
        "    return text\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "vktOAFhA0rTY"
      },
      "outputs": [],
      "source": [
        "train_data_trg = news['headline']\n",
        "train_data_src = news['article']"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "8bIq8jkh09z_"
      },
      "outputs": [],
      "source": [
        "tokenized_corpus_trg = [preprocess_tokenize(x) for x in train_data_trg]  #these are headlines\n",
        "tokenized_corpus_src = [preprocess_tokenize(x) for x in train_data_src]  #these are articles"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "UcAIeAle0_OP",
        "outputId": "5c204f72-e7a9-49dd-8cc6-e2cb94be6edd"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'जॉर्डन के ऐतिहासिक दौरे पर पहुंचे राष्ट्रपति प्रणब मुखर्जी ने करोड़ डॉलर की लागत से निर्मित भारत-जॉर्डन उवर्रक संयंत्र का जॉर्डन के शाह अब्दुल्ला द्वितीय इब्न अल हुसैन के साथ उद्घाटन किया यह संयंत्र एक साल से कम समय में बनकर तैयार हुआ है राष्ट्रपति मुखर्जी के यहां एयर इंडिया की उड़ान से दोपहर पहुंचने के कुछ देर बाद ही शाह के महल से इस संयंत्र का रिमोट के जरिए उद्घाटन किया गया अधिकारियों ने बताया कि भारतीय उर्वरक कंपनी इफको और जॉर्डन के फास्फेट्स माइन कंपनी ने इस संयंत्र के लिए में एक संयुक्त उद्यम कंपनी जॉर्डन इंडिया फर्टिलाइजर कंपनी बनाया संयुक्त उद्यम में इफको की हिस्सेदारी प्रतिशत है इस संयंत्र से प्रति वर्ष करोड़ टन सल्फ्यूरिक एसिड और करोड़ टन फास्फेरिक एसिड के उत्पादन का अनुमान है राष्ट्रपति का इससे पहले यहां पारंपरिक स्वागत किया गया और राष्ट्रपति भवन के सामने उन्हें तोपों की सलामी दी गई इसके बाद वह शाह अब्दुल्ला द्वितीय इब्न अल हुसैन के साथ वार्ता में व्यस्त हो गए वार्ता के बाद दोनों नेताओं ने इंडो-जर्मन उवर्रक संयंत्र का संयुक्त रूप से उद्घाटन किया इस संयंत्र से कच्चे माल का उत्पादन किया जाएगा इसमें फास्फोरिक एसिड और सल्फ्यूरिक एसिड प्रमुख हैं यहां पहुंचने से पहले राष्ट्रपति ने कहा दोनों देशों के क्षेत्रीय एवं अंतर्राष्ट्रीय मुद्दे मिलते-जुलते हैं और दोनों सीरिया के साथ ही मध्य पूर्व में शांति प्रक्रिया का समर्थन करते हैं उन्होंने कहा कि दोनों देश उग्रवाद और आतंकवाद के सभी रूपों की निदा करते हैं और धार्मिक सौहार्द्र में भरोसा करते हैं राष्ट्रपति के इस दौरे के दौरान व्यापार एवं निवेश पर भी जोर है उन्होंने कहा कि दोनों देश द्विपक्षीय व्यापार को पांच अरब डॉलर करना चाहते हैं अभी दोनों देशों के बीच व्यापार दो अरब डॉलर है प्रणब मुखर्जी ने जिस संयंत्र का उद्घाटन किया उससे भारत करोड़ टन फास्फोरिक एसिड का आयात करेगी भारत बड़ी मात्रा में पोटाश एवं फास्फेट जॉर्डन से हासिल करता है भारत और जॉर्डन ने में सामंजस्य के लिए द्विपक्षीय समझौते पर हस्ताक्षर किया था हालांकि इसे औपचारिक रूप में दिया गया जब पूर्ण कूटनीकि संबंध दोनों देश के बीच बने शाह अब्दुल्ला और बेगम रानिया ने अक्टूबर में भारत का दौरा किया था राष्ट्रपति के इस दौरे से पूर्व करीब साल पहले तत्कालीन प्रधानमंत्री राजीव गांधी ने इस देश का दौरा किया था इनपुट- '"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            }
          },
          "metadata": {},
          "execution_count": 15
        }
      ],
      "source": [
        "tokenized_corpus_src[1] #cleaned sentences"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7g7k7soMyBpQ"
      },
      "source": [
        "### Deciding max len and dictionary size"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 350
        },
        "id": "wRks5E7I1UJI",
        "outputId": "94cc27c3-d6b8-4a1b-f6f6-0efe90d09fb9"
      },
      "outputs": [
        {
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAnAAAAFNCAYAAACAH1JNAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3de7wWZb338c83EA8Ygsp2K2iLkq2RlimiPe22paioJfRsNe0guklqq2llzw7TnWfD9pOWPepWA8VDItlBDIrIQ1k7kCWaCEgsEQVCREE84An8PX/MteBmca+17nux7sMsvu/Xa15r5pprrvnNLBx/65q5ZhQRmJmZmVl+vKfWAZiZmZlZeZzAmZmZmeWMEzgzMzOznHECZ2ZmZpYzTuDMzMzMcsYJnJmZmVnOOIHrAiT9t6T/7KS29pL0mqRuafkhSV/ujLZTe7+RNLKz2itjv5dLelHS8xXcx2uS3t/Zdbcwpk9KWlrp/ZiZWXU5gatzkhZLekPSq5JelvQ/kr4qacPvLiK+GhGXldjW0LbqRMRzEbFjRKzvhNgvlnRHi/aPiYgJW9p2mXHsBZwHDIqIfyyyvlOSnHTeFnV23WqRdJqkP9U6DjMza58TuHz4TES8F3gfMBb4NjCus3ciqXtnt1kn9gJeiogXOtpAFz43ZmaWQ07gciQi1kTEZOBzwEhJ+wFIulXS5Wl+V0m/Tr11qyQ9LOk9km4nS2TuS7fv/kNSg6SQNErSc8ADBWWFCcsHJD0i6RVJ90raOe1rs56r5l4+ScOA7wCfS/v7a1q/4ZZsiutCSc9KekHSbZJ2Suua4xgp6bl0+/OC1s6NpJ3S9itTexem9ocC04E9Uhy3ttiuJ/CbgvWvSdoj9R7eI+kOSa8Ap0kaIukv6dwul/T/JPUoaCsk7V3wO7lO0pTUezpT0gc6WPcoSQskrZF0vaQ/tHZbW9L2qb3VkuYBB7dYP0bS02k/8yR9NpV/EPhv4GPpHLycyo+T9Fj63S+RdHFrvwMzM6seJ3A5FBGPAEuBTxRZfV5a1xfYjSyJioj4EvAcWW/ejhHx/YJtDgM+CBzdyi5PBf4N2B1YB1xbQoy/Ba4E7k77+0iRaqel6VPA+4Edgf/Xos4/A/sARwDfTYlGMT8GdkrtHJZiPj0ifg8cA/w9xXFaizhfb7F+x4j4e1o9HLgH6A3cCawHvgHsCnwsxXRmG6fhZOASoA/QBFxRbl1Ju6YYzgd2ARYA/6uNdi4CPpCmo4GWzxs+TfbvZqe0vzsk7R4R84GvAn9J56B3qv862bnsDRwH/LukEW3s38zMqsAJXH79Hdi5SPk7ZInW+yLinYh4ONr/4O3FEfF6RLzRyvrbI+LJlOz8J3CS0iCHLfQF4OqIWBQRr5ElKSe36P27JCLeiIi/An8FNksEUywnA+dHxKsRsRj4AfClLYzvLxHxq4h4N8XwaETMiIh1aR83kiWLrfllRDwSEevIEsADOlD3WGBuRPwirbsWaGsgxknAFRGxKiKW0CLZjoifRcTf0zHdDSwEhrTWWEQ8FBFzUv0ngLvaOWYzM6sCJ3D51Q9YVaT8v8h6cH4naZGkMSW0taSM9c8C25D1Qm2pPVJ7hW13J+s5bFaYrKwl66VradcUU8u2+m1hfJucF0n/lG5PP59uq15J2+ehlNjbq7tHYRwpGW9rwMUm9dn0nCDpVEmPp9vALwP70cYxSDpE0oPp1vQasl66zvjdm5nZFnACl0OSDiZLTjYbMZh6oM6LiPcDxwPflHRE8+pWmmyvh27Pgvm9yHr5XiS7vbZDQVzdyG7dltru38kGZhS2vQ5Y0c52Lb2YYmrZ1rISty/1vNwAPAUMjIheZLenVUacHbEc6N+8IEmFy63Ub/n7at72fcDNwNnALuk26ZNsPIZi5+GnwGRgz4jYiew5uUofs5mZtcMJXI5I6iXp08BE4I6ImFOkzqcl7Z3+R7+G7Lmtd9PqFWTPiJXri5IGSdoBuBS4J71m5G/AdulB922AC4FtC7ZbATSo4JUnLdwFfEPSAEk7svGZuXXlBJdimQRcIem9KVH5JnBH21tuEucuzQMo2vBe4BXgNUn7Av9eTpwdNAXYX9KIdGv5LGCzV6EUmAScL6mPpP7A1wrW9SRL0lYCSDqdrAeu2Qqgf+HADLJjXhURb0oaAnx+i4/IzMy2mBO4fLhP0qtkt8YuAK4GTm+l7kDg98BrwF+A6yPiwbTue8CF6fbZt8rY/+3ArWS3+bYDzoFsVCzZQ/w/Ievtep1Nb+/9LP18SdLsIu2OT23/EXgGeJNNE45yfC3tfxFZz+RPU/vtioinyJLJRenc7NFK1W+RJTCvkvVk3d3BWEsWES8CJwLfB14CBgGNwFutbHIJ2W3TZ4DfkZ3f5rbmkT0b+BeyZG1/4M8F2z4AzAWel/RiKjsTuDT9+/suWYJoZmY1pvafbzezepF6M5cCXyhIzM3MbCvjHjizOifpaEm9JW3LxufuZtQ4LDMzqyEncGb172Nk7297EfgMMKKNV75YGSSNV/YS6ScLyv5L0lOSnpD0S0m9C9adL6kpvVj56ILyYamsqXDkd3q+c2Yqv7v5+UJJ26blprS+oTpHbGZdhRM4szoXERdHxC4R8d6IOCQiZtY6pi7kVmBYi7LpwH4R8WGygTrnA0gaRPa+wQ+lba6X1C2Nvr6O7IXQg4BTUl2Aq4BrImJvYDUwKpWPAlan8mtSPTOzkjmBM7OtVkT8kRbvU4yI3xWMhJ7Bxte2DAcmRsRbEfEM2fsWh6SpKb2Q+m2yUeLD00jww8m+pAEwARhR0NaENH8PcESqb2ZWEidwZmat+zeyb+VC9u7FwpckL01lrZXvArxckAw2l2/SVlq/JtU3MytJ9/ardC277rprNDQ01DoMM6uSRx999MWI6Nt+zU1JuoDsxdJ3dn5UZcUxGhgN0LNnz4P23XffWoZjZlXW2jVsq0vgGhoaaGxsrHUYZlYlkp5tv9Zm25wGfBo4ouBbwsvY9CsX/dn4tY9i5S8BvSV1T71shfWb21qaXtC8U6q/mYi4CbgJYPDgweHrl9nWpbVrmG+hmpkVkDQM+A/g+IhYW7BqMnByGkE6gOyl2Y8As4CBacRpD7KBDpNT4vcgcELafiRwb0FbI9P8CcAD4ZdymlkZtroeODOzZpLuAj4J7CppKXAR2ajTbYHpaVzBjIj4akTMlTQJmEd2a/Ws9Bk3JJ0NTAO6AeMjYm7axbeBiZIuBx4DxqXyccDtkprIBlGcXPGDNbMuxQmcmW21IuKUIsXjipQ1178CuKJI+VRgapHyRWSjVFuWv0n2iTQzsw7xLVQzMzOznHECZ2ZmZpYzTuDMzMzMcsYJnJmZmVnOOIEzMzMzyxkncGZmZmY54wTOzMzMLGecwJmZmZnljF/kuxVrGDOl3TqLxx5XhUjMrJ75WmFWf9wDZ2ZmZpYzTuDMzMzMcsYJnJmZmVnOOIEzMzMzyxkncGZmZmY54wTOzMzMLGecwJmZmZnljBM4MzMzs5xxAmdmZmaWM07gzMzMzHLGCZyZmZlZzjiBMzMzM8sZJ3BmZmZmOeMEzszMzCxnnMCZmZmZ5UzFEjhJ20l6RNJfJc2VdEkqHyBppqQmSXdL6pHKt03LTWl9Q0Fb56fyBZKOLigflsqaJI2p1LGYmZmZ1ZNK9sC9BRweER8BDgCGSToUuAq4JiL2BlYDo1L9UcDqVH5NqoekQcDJwIeAYcD1krpJ6gZcBxwDDAJOSXXNzMzMurSKJXCReS0tbpOmAA4H7knlE4ARaX54WiatP0KSUvnEiHgrIp4BmoAhaWqKiEUR8TYwMdU1MzMz69K6V7Lx1Ev2KLA3WW/Z08DLEbEuVVkK9Evz/YAlABGxTtIaYJdUPqOg2cJtlrQoP6QCh5E7DWOm1DoEMzMzq6CKDmKIiPURcQDQn6zHbN9K7q81kkZLapTUuHLlylqEYGZmZtZpqjIKNSJeBh4EPgb0ltTc89cfWJbmlwF7AqT1OwEvFZa32Ka18mL7vykiBkfE4L59+3bKMZmZmZnVSiVHofaV1DvNbw8cCcwnS+ROSNVGAvem+clpmbT+gYiIVH5yGqU6ABgIPALMAgamUa09yAY6TK7U8ZiZmZnVi0o+A7c7MCE9B/ceYFJE/FrSPGCipMuBx4Bxqf444HZJTcAqsoSMiJgraRIwD1gHnBUR6wEknQ1MA7oB4yNibgWPx8zMzKwuVCyBi4gngI8WKV9E9jxcy/I3gRNbaesK4Ioi5VOBqVscrJmZmVmOVHQUquVfKSNaF489rgqRmJmZWTN/SsvMzMwsZ5zAmZmZmeWMEzgzMzOznHECZ2ZbNUnjJb0g6cmCsp0lTZe0MP3sk8ol6VpJTZKekHRgwTYjU/2FkkYWlB8kaU7a5tr0icBW92FmVgoncGa2tbsVGNaibAxwf0QMBO5PywDHkL2LciAwGrgBsmQMuIjsc35DgIsKErIbgDMKthvWzj7MzNrlBM7MtmoR8Ueyd08WGg5MSPMTgBEF5bdFZgbZl2V2B44GpkfEqohYDUwHhqV1vSJiRnox+W0t2iq2DzOzdjmBMzPb3G4RsTzNPw/slub7AUsK6i1NZW2VLy1S3tY+NuFvOZtZMU7gzMzakHrOolb78LeczawYJ3BmZptbkW5/kn6+kMqXAXsW1Oufytoq71+kvK19mJm1y19iMDPb3GRgJDA2/by3oPxsSRPJBiysiYjlkqYBVxYMXDgKOD8iVkl6RdKhwEzgVODH7ewjl/zVFrPqcgJnZls1SXcBnwR2lbSUbDTpWGCSpFHAs8BJqfpU4FigCVgLnA6QErXLgFmp3qUR0Tww4kyyka7bA79JE23sw8ysXU7gzGyrFhGntLLqiCJ1AzirlXbGA+OLlDcC+xUpf6nYPszMSuFn4MzMzMxyxgmcmZmZWc44gTMzMzPLGT8DlzOljPQyMzOzrs09cGZmZmY54wTOzMzMLGecwJmZmZnljBM4MzMzs5xxAmdmZmaWM07gzMzMzHLGCZyZmZlZzjiBMzMzM8sZJ3BmZmZmOeMEzszMzCxnnMCZmZmZ5YwTODMzM7OccQJnZmZmljNO4MzMzMxypmIJnKQ9JT0oaZ6kuZLOTeUXS1om6fE0HVuwzfmSmiQtkHR0QfmwVNYkaUxB+QBJM1P53ZJ6VOp4zMzMzOpFJXvg1gHnRcQg4FDgLEmD0rprIuKANE0FSOtOBj4EDAOul9RNUjfgOuAYYBBwSkE7V6W29gZWA6MqeDxmZmZmdaFiCVxELI+I2Wn+VWA+0K+NTYYDEyPirYh4BmgChqSpKSIWRcTbwERguCQBhwP3pO0nACMqczRmZmZm9aMqz8BJagA+CsxMRWdLekLSeEl9Ulk/YEnBZktTWWvluwAvR8S6FuVmZmZmXVrFEzhJOwI/B74eEa8ANwAfAA4AlgM/qEIMoyU1SmpcuXJlpXdnZmZmVlEVTeAkbUOWvN0ZEb8AiIgVEbE+It4Fbia7RQqwDNizYPP+qay18peA3pK6tyjfTETcFBGDI2Jw3759O+fgzMzMzGqkkqNQBYwD5kfE1QXluxdU+yzwZJqfDJwsaVtJA4CBwCPALGBgGnHag2ygw+SICOBB4IS0/Ujg3kodj5mZmVm96N5+lQ77OPAlYI6kx1PZd8hGkR4ABLAY+ApARMyVNAmYRzaC9ayIWA8g6WxgGtANGB8Rc1N73wYmSroceIwsYTQzMzPr0iqWwEXEnwAVWTW1jW2uAK4oUj612HYRsYiNt2DNzMzMtgr+EoOZmZlZzjiBMzMzM8sZJ3BmZmZmOeMEzszMzCxnnMCZmZmZ5YwTODMzM7OccQJnZmZmljNO4MzMzMxyxgmcmZmZWc44gTMzK0LSNyTNlfSkpLskbZe+yTxTUpOku9P3mUnfcL47lc+U1FDQzvmpfIGkowvKh6WyJkljqn+EZpZnTuDMzFqQ1A84BxgcEfuRfYf5ZOAq4JqI2BtYDYxKm4wCVqfya1I9JA1K230IGAZcL6mbpG7AdcAxwCCyb0QPqtbxmVn+OYEzMyuuO7C9pO7ADsBy4HDgnrR+AjAizQ9Py6T1R0hSKp8YEW9FxDNAE9n3m4cATRGxKCLeBiamumZmJanYx+xt69EwZkq7dRaPPa4KkZh1johYJun/As8BbwC/Ax4FXo6IdanaUqBfmu8HLEnbrpO0Btgllc8oaLpwmyUtyg+pwKGYWRflHjgzsxYk9SHrERsA7AH0JLsFWotYRktqlNS4cuXKWoRgZnXICZyZ2eaGAs9ExMqIeAf4BfBxoHe6pQrQH1iW5pcBewKk9TsBLxWWt9imtfLNRMRNETE4Igb37du3M47NzLoAJ3BmZpt7DjhU0g7pWbYjgHnAg8AJqc5I4N40Pzktk9Y/EBGRyk9Oo1QHAAOBR4BZwMA0qrUH2UCHyVU4LjPrIvwMnJlZCxExU9I9wGxgHfAYcBMwBZgo6fJUNi5tMg64XVITsIosISMi5kqaRJb8rQPOioj1AJLOBqaRjXAdHxFzq3V8ZpZ/TuDMzIqIiIuAi1oULyIbQdqy7pvAia20cwVwRZHyqcDULY/UzLZGvoVqZmZmljNO4MzMzMxyxgmcmZmZWc74Gbg6UsoLcc3MzMzcA2dmZmaWM07gzMzMzHLGCZyZmZlZzjiBMzMzM8sZJ3BmZmZmOeMEzszMzCxnnMCZmZmZ5YwTODMzM7OccQJnZmZmljMVS+Ak7SnpQUnzJM2VdG4q31nSdEkL088+qVySrpXUJOkJSQcWtDUy1V8oaWRB+UGS5qRtrpWkSh2PmZmZWb2oZA/cOuC8iBgEHAqcJWkQMAa4PyIGAvenZYBjgIFpGg3cAFnCB1wEHAIMAS5qTvpSnTMKthtWweMxMzMzqwsVS+AiYnlEzE7zrwLzgX7AcGBCqjYBGJHmhwO3RWYG0FvS7sDRwPSIWBURq4HpwLC0rldEzIiIAG4raMvMzMysyyopgZO0/5bsRFID8FFgJrBbRCxPq54Hdkvz/YAlBZstTWVtlS8tUm5mW6E5c+bUOgQzs6optQfuekmPSDpT0k7l7EDSjsDPga9HxCuF61LPWZTTXkdIGi2pUVLjypUrK707M6uBM888kyFDhnD99dezZs2aWodjZlZRJSVwEfEJ4AvAnsCjkn4q6cj2tpO0DVnydmdE/CIVr0i3P0k/X0jly1L7zfqnsrbK+xcpLxb/TRExOCIG9+3bt72wzSyHHn74Ye68806WLFnCQQcdxOc//3mmT59e67DMzCqi5GfgImIhcCHwbeAw4FpJT0n638XqpxGh44D5EXF1warJQPNI0pHAvQXlp6bRqIcCa9Kt1mnAUZL6pMELRwHT0rpXJB2a9nVqQVtmthUaOHAgl19+OVdddRV/+MMfOOeccwA+1Np1yswsr0p9Bu7Dkq4hG4hwOPCZiPhgmr+mlc0+DnwJOFzS42k6FhgLHClpITA0LQNMBRYBTcDNwJkAEbEKuAyYlaZLUxmpzk/SNk8Dvyn1wM2sa3niiSf4xje+wQc/+EEeeOAB7rvvPubPnw/wN1q/TpmZ5VL3Euv9mCxR+k5EvNFcGBF/l3RhsQ0i4k9Aa+9lO6JI/QDOaqWt8cD4IuWNwH7tRm9mXd7XvvY1vvzlL3PllVey/fbbF656h+zugZlZl1FqAncc8EZErAeQ9B5gu4hYGxG3Vyw6M7MSTZkyhe23355u3boB8O677/Lmm28C4OuUmXU1pT4D93ug8E/aHVKZmVldGDp0KG+8seEGAWvXrmXo0KE1jMjMrHJKTeC2i4jXmhfS/A6VCcnMrHxvvvkmO+6444blHXfckbVr19YwIjOzyik1gXu9xbdJDwLeaKO+mVlV9ezZk9mzZ29YfvTRR1s+C2dm1mWU+gzc14GfSfo72cCEfwQ+V7GozMzK9MMf/pATTzyRPfbYg4jg+eef5+6772bw4MG1Ds3MrNOVlMBFxCxJ+wL7pKIFEfFO5cIyMyvPwQcfzFNPPcWCBQsA2Geffdhmm21qHJWZWWWU2gMHcDDQkLY5UBIRcVtFojIz64BZs2axePFi1q1bt8ntVDOzrqakBE7S7cAHgMeB9ak4ACdwZlYXvvSlL/H0009zwAEHbHiVSPaRFjOzrqfUHrjBwKD0sl0zs7rT2NjIvHnzNkvafvzjH9coIjOzyil1FOqTZAMXzMzq0n777cfzzz9f6zDMzKqi1B64XYF5kh4B3moujIjjKxKVmVmZXnzxRQYNGsSQIUPYdtttax2OmVlFlZrAXVzJIMzMttTFF19ctPy+++6rbiBmZlVQ6mtE/iDpfcDAiPi9pB2AbpUNzcysdIcddhjPPvssCxcuZOjQoaxdu5b169e3v6GZWQ6V9AycpDOAe4AbU1E/4FeVCsrMrFw333wzJ5xwAl/5ylcAWLZsGSNGjKhxVGZmlVHqIIazgI8DrwBExELgHyoVlJlZua677jr+/Oc/06tXLwAGDhzICy+80OH2JPWWdI+kpyTNl/QxSTtLmi5pYfrZJ9WVpGslNUl6osWnB0em+gsljSwoP0jSnLTNtfI7T8ysDKUmcG9FxNvNC5K6k70HzsysLmy77bb06NFjw/K6deu29D1wPwJ+GxH7Ah8B5gNjgPsjYiBwf1oGOAYYmKbRwA0AknYGLgIOAYYAFzUnfanOGQXbDduSYM1s61JqAvcHSd8Btpd0JPAzwE8Gm1ndOOyww7jyyit54403mD59OieeeCKf+cxnOtSWpJ2AfwHGAUTE2xHxMjAcmJCqTQCa79EOB26LzAygt6TdgaOB6RGxKiJWA9OBYWldr4iYkd6veVtBW2Zm7Sp1FOoYYBQwB/gKMBX4SaWCsq6nYcyUdussHntcFSKxrmrs2LGMGzeO/fffnxtvvJFjjz2WL3/5y1x55ZUdaW4AsBK4RdJHgEeBc4HdImJ5qvM8sFua7wcsKdh+aSprq3xpkXIzs5KUOgr1XeDmNJmZ1Z33vOc9nHHGGZxxxhmd0Vx34EDgaxExU9KP2Hi7FICICEkVf5RE0miy27Lstddeld6dmeVEqd9CfYYiz7xFxPs7PSIzsw4YMGBAZ377dCmwNCJmpuV7yBK4FZJ2j4jl6TZo8yiJZcCeBdv3T2XLgE+2KH8olfcvUn8zEXETcBPA4MGD/eyxmQHlfQu12XbAicDOnR+OmVnHNDY2bph/8803+dnPfsaqVau47LLLym4rIp6XtETSPhGxADgCmJemkcDY9PPetMlk4GxJE8kGLKxJSd404MqCgQtHAedHxCpJr0g6FJgJnAr4o61mVrJSb6G+1KLoh5IeBb7b+SGZmZVvl1122WT561//OgcddNCWNPk14E5JPYBFwOlkA78mSRoFPAuclOpOBY4FmoC1qS4pUbsMmJXqXRoRq9L8mcCtwPbAb9JkZlaSUm+hHliw+B6yHrlSe+/MzCpu9uzZG+bfffddGhsbWbduXYfbi4jH2fTuQ7MjitQNsvdlFmtnPDC+SHkjsF+HAzSzrVqpSdgPCubXAYvZ+JenmVnNnXfeeRvmu3fvTkNDA5MmTWLfffetYVRmZpVR6i3UT1U6EDOzLfHggw/WOgQzs6op9RbqN9taHxFXd044ZmYdc/XVrV6GdpP0TV+nzKwrKWcU6sFkI60APgM8AiysRFBmZuVqbGxk1qxZHH/88QDcd999DBkyBLLndt9by9jMzDpbqQlcf+DAiHgVQNLFwJSI+GKlAjMzK8fSpUuZPXs2731vlqtdfPHFHHfccQDLI+KSmgZnZtbJSv0W6m7A2wXLb7PxEzJmZjW3YsWKTT5m36NHD1asWFHDiMzMKqfUHrjbgEck/TItj2DjB53NzGru1FNPZciQIXz2s58F4Fe/+hUjR47kggsuqHFkZmadr6QeuIi4guzFlKvTdHpEdOgL0WZmlXDBBRdwyy230KdPH/r06cMtt9zCd77znVqHZWZWEaXeQgXYAXglIn4ELJU0oK3KksZLekHSkwVlF0taJunxNB1bsO58SU2SFkg6uqB8WCprkjSmoHyApJmp/O70tnQz24qtXbuWXr16ce6559K/f3+eeeaZWodkZlYRJSVwki4Cvg2cn4q2Ae5oZ7NbgWFFyq+JiAPSNDW1Pwg4GfhQ2uZ6Sd0kdQOuA44BBgGnpLoAV6W29ibrFRxVyrGYWdd0ySWXcNVVV/G9730PgHfeeYcvftHjrMysayq1B+6zwPHA6wAR8XfaGZYfEX8EVrVVp8BwYGJEvBURz5B9T3BImpoiYlFEvA1MBIZLEnA4cE/afgLZc3lmtpX65S9/yeTJk+nZsycAe+yxB6+++mqNozIzq4xSE7i307f+AkBSzy3Y59mSnki3WPuksn7AkoI6S1NZa+W7AC9HxLoW5UVJGi2pUVLjypUrtyB0M6tXPXr0QBLZ33fw+uuv1zgiM7PKKTWBmyTpRqC3pDOA3wM3d2B/NwAfAA4AlrPpN1YrJiJuiojBETG4b9++1dilmVXZSSedxFe+8hVefvllbr75ZoYOHcoZZ5xR67DMzCqi3deIpNuVdwP7Aq8A+wDfjYjp5e4sIja8lEnSzcCv0+IyYM+Cqv1TGa2Uv0SWTHZPvXCF9c1sKxMRfO5zn+Opp56iV69eLFiwgEsvvZQjjzySc845p9bhmZl1unYTuIgISVMjYn+g7KStkKTdI2J5Wvws0DxCdTLwU0lXA3sAA8k+1SVgYBrxuoxsoMPnU0wPAieQPRc3Erh3S2Izs/ySxLHHHsucOXM48sgjax2OmVnFlXoLdbakg8tpWNJdwF+AfSQtlTQK+L6kOZKeAD4FfAMgIuYCk4B5wG+BsyJifepdOxuYBswHJqW6kI2K/aakJrJn4saVE5+ZdS0HHnggs2bNqnUYZmZVUeqXGA4BvihpMdlIVJF1zn24tQ0i4pQixa0mWellwVcUKZ8KTC1SvohslKqZGTNnzuSOO+6goaGBnj17EhEbBjSYmXU1bSZwkvaKiOeAo9uqZ2ZWK8899xx77bUX06ZNK7q+oaGhugGZmVVBez1wvwIOjIhnJf08Iv61GkGZmZVqxIgRzJ49m/e9733867/+Kz//+c9rHcFrdWMAABVXSURBVJKZWcW19wxc4f2H91cyEDOzjsheUZlZtGhRDSMxM6ue9hK4aGXezKwuFD7n5mfezGxr0d4t1I9IeoWsJ277NA8bBzH0qmh0Zmbt+Otf/0qvXr2ICN544w169couSx7EYGZdWZsJXER0q1YgZmYdsX79+jbXO4kzs66o1PfAmZmZmVmdcAJnZmZmljNO4MzMzMxyptQvMZhVXMOYKSXVWzz2uApHYmZmVt/cA2dmZmaWM07gzMzMzHLGCZyZmZlZzvgZuCop9fkuMzMzs/a4B87MzMwsZ5zAmZmZmeWMEzgzMzOznHECZ2ZmZpYzTuDMzMzMcsYJnJlZKyR1k/SYpF+n5QGSZkpqknS3pB6pfNu03JTWNxS0cX4qXyDp6ILyYamsSdKYah+bmeWbEzgzs9adC8wvWL4KuCYi9gZWA6NS+ShgdSq/JtVD0iDgZOBDwDDg+pQUdgOuA44BBgGnpLpmZiVxAmdmVoSk/sBxwE/SsoDDgXtSlQnAiDQ/PC2T1h+R6g8HJkbEWxHxDNAEDElTU0Qsioi3gYmprplZSZzAmZkV90PgP4B30/IuwMsRsS4tLwX6pfl+wBKAtH5Nqr+hvMU2rZWbmZXECZyZWQuSPg28EBGP1kEsoyU1SmpcuXJlrcMxszrhBM7MbHMfB46XtJjs9ubhwI+A3pKaP0HYH1iW5pcBewKk9TsBLxWWt9imtfLNRMRNETE4Igb37dt3y4/MzLoEJ3BmZi1ExPkR0T8iGsgGITwQEV8AHgROSNVGAvem+clpmbT+gYiIVH5yGqU6ABgIPALMAgamUa090j4mV+HQzKyL8MfszcxK921goqTLgceAcal8HHC7pCZgFVlCRkTMlTQJmAesA86KiPUAks4GpgHdgPERMbeqR2JmueYEzsysDRHxEPBQml9ENoK0ZZ03gRNb2f4K4Ioi5VOBqZ0YqpltRXwL1czMzCxnnMCZmZmZ5UzFbqFKGg80D8XfL5XtDNwNNACLgZMiYnV64eWPgGOBtcBpETE7bTMSuDA1e3lETEjlBwG3AtuT3YY4Nz00bGZmJWoYM6XWIZhZB1SyB+5Wsk/HFBoD3B8RA4H70zJkn5MZmKbRwA2wIeG7CDiE7LmTiyT1SdvcAJxRsF3LfZmZmZl1SRVL4CLij2SjsQoVfm6m5WdobovMDLJ3Le0OHA1Mj4hVEbEamA4MS+t6RcSM1Ot2W0FbZmZmZl1atZ+B2y0ilqf554Hd0ny5n5vpl+ZblpuZmZl1eTV7jUhEhKSqPLMmaTTZrVn22muvauzSKqiUZ3YWjz2uCpGYmZnVRrV74Fak25+kny+k8nI/N7MszbcsL8qfojEzM7OupNoJXOHnZlp+huZUZQ4F1qRbrdOAoyT1SYMXjgKmpXWvSDo0jWA9taAtMzMzsy6tkq8RuQv4JLCrpKVko0nHApMkjQKeBU5K1aeSvUKkiew1IqcDRMQqSZeRfTcQ4NKIaB4YcSYbXyPymzSZmZmZdXkVS+Ai4pRWVh1RpG4AZ7XSznhgfJHyRmC/LYnRzMzMLI/8JQYzMzOznHECZ2ZmZpYzTuDMzMzMcsYJnJmZmVnOOIEzMzMzyxkncGZmZmY54wTOzMzMLGecwJmZmZnljBM4MzMzs5xxAmdmZmaWM07gzMzMzHLGCZyZmZlZzjiBMzMzM8uZ7rUOwKwSGsZMabfO4rHHVSESMzOzzuceODMzM7OccQJnZmZmljNO4MzMzMxyxgmcmZmZWc44gTMzMzPLGSdwZmZmZjnjBM7MzMwsZ5zAmZmZmeWMEzgzMzOznHECZ2ZmZpYzTuDMzMzMcsYJnJmZmVnOOIEzM2tB0p6SHpQ0T9JcSeem8p0lTZe0MP3sk8ol6VpJTZKekHRgQVsjU/2FkkYWlB8kaU7a5lpJqv6RmlleOYEzM9vcOuC8iBgEHAqcJWkQMAa4PyIGAvenZYBjgIFpGg3cAFnCB1wEHAIMAS5qTvpSnTMKthtWheMysy6ie60DMKuVhjFT2q2zeOxxVYjE6k1ELAeWp/lXJc0H+gHDgU+mahOAh4Bvp/LbIiKAGZJ6S9o91Z0eEasAJE0Hhkl6COgVETNS+W3ACOA31Tg+M8s/98CZmbVBUgPwUWAmsFtK7gCeB3ZL8/2AJQWbLU1lbZUvLVJuZlYSJ3BmZq2QtCPwc+DrEfFK4brU2xZViGG0pEZJjStXrqz07swsJ2pyC1XSYuBVYD2wLiIGp2dF7gYagMXASRGxOj3Y+yPgWGAtcFpEzE7tjAQuTM1eHhETqnkcUNptODPLH0nbkCVvd0bEL1LxCkm7R8TydIv0hVS+DNizYPP+qWwZG2+5Npc/lMr7F6m/mYi4CbgJYPDgwRVPGM0sH2rZA/epiDggIgan5c58ONjMrMPSH47jgPkRcXXBqslA80jSkcC9BeWnptGohwJr0q3WacBRkvqk69NRwLS07hVJh6Z9nVrQlplZu+rpFupwsoeCST9HFJTfFpkZQPPDwUeTHg6OiNXAdDyKy8w6x8eBLwGHS3o8TccCY4EjJS0EhqZlgKnAIqAJuBk4EyANXrgMmJWmS5sHNKQ6P0nbPI0HMJhZGWo1CjWA30kK4MZ0i6CzHg42M9siEfEnoLX3sh1RpH4AZ7XS1nhgfJHyRmC/LQjTzLZitUrg/jkilkn6B2C6pKcKV0ZEpOSuU0gaTXb7lb322quzmjUzMzOriZrcQo2IZennC8AvyZ5hW5FujVLGw8HFyovt76aIGBwRg/v27duZh2JmZmZWdVVP4CT1lPTe5nmyh3qfpJMeDq7ioZiZmZnVRC1uoe4G/DJ99q878NOI+K2kWcAkSaOAZ4GTUv2pZK8QaSJ7jcjpkD0cLKn54WDY9OFgs05R6mti/MUGMzOrpqoncBGxCPhIkfKX6KSHg83MzMy6snp6jYiZmZmZlcAfszczs6rwIwlmncc9cGZmZmY54wTOzMzMLGecwJmZmZnljBM4MzMzs5zxIAazTlDKw9l+MNvMzDqLe+DMzMzMcsYJnJmZmVnOOIEzMzMzyxkncGZmZmY54wTOzMzMLGecwJmZmZnljF8jYlYlftWImZl1FvfAmZmZmeWMEzgzMzOznPEtVLM64tusZmZWCvfAmZmZmeWMEzgzMzOznPEtVLOc8W1WMzNzD5yZmZlZzjiBMzMzM8sZ30I164JKuc0KvtVqZpZX7oEzMzMzyxn3wJltxTwgwswsn9wDZ2ZmZpYz7oEzsza5l87MrP44gWtDqQ+Cm23tnOSZmVWXEzgzqwqPjDUz6zxO4Mysrrg3z8ysfR7EYGZmZpYzuU/gJA2TtEBSk6QxtY7HzKxUvn6ZWUfl+haqpG7AdcCRwFJglqTJETGvtpGZmbWtGtcvD8Qy67ry3gM3BGiKiEUR8TYwERhe45jMzErh65eZdViue+CAfsCSguWlwCE1isXMrBy+frXCA1nM2pf3BK4kkkYDo9Pia5IWlLjprsCLlYmqohx39eU19lzGravKivt9lYyl0rbg+gU5/f1SQty6qkqRlKfLnu86tbXEXfQalvcEbhmwZ8Fy/1S2iYi4Cbip3MYlNUbE4I6HVxuOu/ryGrvjrqmKXr8gv+fJcVeX466uzoo778/AzQIGShogqQdwMjC5xjGZmZXC1y8z67Bc98BFxDpJZwPTgG7A+IiYW+OwzMza5euXmW2JXCdwABExFZhaoeY7dNuiDjju6str7I67hip8/YL8nifHXV2Ou7o6JW5FRGe0Y2ZmZmZVkvdn4MzMzMy2Ok7gWlHPn7iRtKekByXNkzRX0rmpfGdJ0yUtTD/7pHJJujYdyxOSDqxx/N0kPSbp12l5gKSZKb670wPdSNo2LTel9Q01jLm3pHskPSVpvqSP5eF8S/pG+jfypKS7JG1Xj+db0nhJL0h6sqCs7PMraWSqv1DSyGrFX2/q+fpVqNxrWb0p9VpWT8q5ltWTcq5ltdZZ17P2OIErQhs/cXMMMAg4RdKg2ka1iXXAeRExCDgUOCvFNwa4PyIGAvenZciOY2CaRgM3VD/kTZwLzC9Yvgq4JiL2BlYDo1L5KGB1Kr8m1auVHwG/jYh9gY+QxV/X51tSP+AcYHBE7Ef2oPzJ1Of5vhUY1qKsrPMraWfgIrKX4Q4BLqrH/xFVWg6uX4XKvZbVm1KvZfWknGtZXejAtazWbmULr2cliQhPLSbgY8C0guXzgfNrHVcb8d5L9j3FBcDuqWx3YEGavxE4paD+hno1iLV/+sd7OPBrQGQvNOze8tyTjc77WJrvnuqpBjHvBDzTct/1fr7Z+Kb/ndP5+zVwdL2eb6ABeLKj5xc4BbixoHyTelvLlLfrV4vY27yW1dNUzrWsXqZyr2X1MpV7LauHaUuvZ6Xswz1wxRX7xE2/GsXSpnSb66PATGC3iFieVj0P7Jbm6+l4fgj8B/BuWt4FeDki1qXlwtg2xJ3Wr0n1q20AsBK4Jd0u+YmkntT5+Y6IZcD/BZ4DlpOdv0ep//PdrNzzWxfnvQ7k8jyUeC2rJ+Vcy+pFudeyutCBa1k96vT/XziByzFJOwI/B74eEa8Urossla+rIcaSPg28EBGP1jqWMnUHDgRuiIiPAq/T4hZDnZ7vPmQfRx8A7AH0ZPNu/Vyox/NrncfXsqrxtawOdNY5dgJXXEmfuKklSduQXfDujIhfpOIVknZP63cHXkjl9XI8HweOl7QYmEh26+FHQG9Jze8kLIxtQ9xp/U7AS9UMOFkKLI2ImWn5HrKLYL2f76HAMxGxMiLeAX5B9juo9/PdrNzzWy/nvdZydR7KvJbVi3KvZfWi3GtZvSj3WlaPOv3/F07giqvrT9xIEjAOmB8RVxesmgw0j7wbSfY8SXP5qWm0y6HAmoKu3KqJiPMjon9ENJCd0wci4gvAg8AJrcTdfDwnpPpV/8swIp4HlkjaJxUdAcyjzs832e2GQyXtkP7NNMdd1+e7QLnndxpwlKQ+6S/2o1LZ1qaur1+FOnAtqwsduJbVhQ5cy+pFudeyetT5/7+o9YN+9ToBxwJ/A54GLqh1PC1i+2ey7tcngMfTdCzZMxj3AwuB3wM7p/oiG5X2NDCHbCRPrY/hk8Cv0/z7gUeAJuBnwLapfLu03JTWv7+G8R4ANKZz/iugTx7ON3AJ8BTwJHA7sG09nm/gLrJnW94h6yUY1ZHzC/xbir8JOL3W/85r+Huv2+tXizjLupbV41TKtayepnKuZfU0lXMtq/XUWdez9iZ/icHMzMwsZ3wL1czMzCxnnMCZmZmZ5YwTODMzM7OccQJnZmZmljNO4MzMzMxyxgmclUxSSPpBwfK3JF1cgf3cJekJSd/o7LZb2d9iSbuWWPdSSUNLrHu0pMfT9JqkBWn+tjJia5D0+VLrm3V1vg6Vdx1K9XeQdKekOZKelPSn9PULy7Hu7Vcx2+At4H9L+l5EvFiJHUj6R+DgiNi7Qu13j43fzitbRHy3jLrTSC+SlfQQ8K2IaCxzlw3A54GflrmdWVfl61AZ16HkXGBFROyf9r8P2TvK6saWnpOtkXvgrBzrgJuAzf4iTT1FD6S/WO+XtFdbDUnaTtIt6S/CxyR9Kq36HdAv9VR9oqB+N0nPpLdV95a0XtK/pHV/lDRQ0s6SfpVimCHpw2n9xZJul/Rn4HZJu0j6naS5kn5C9iJFJPWUNEXSX9NfqZ8rEvetkk5I84slXSJpdjqOfUs5iZK+KOmRdIw3pmM7OMW9XYpjrqT9gLHAJ1LdqvQEmNU5X4fKvw7tTsHnmSJiQUS8lc7XkwXtbujNlPSQpGskNUqan65Rv5C0UNLlBef7qRTP35T18g2V9OdUb0iqN0TSX9I5/h+lL0FIOk3SZEkPAPdLuk3SiIJ47pQ0vK3f4dbMCZyV6zrgC5J2alH+Y2BCRHwYuBO4tp12ziL7pu/+wCnABEnbAccDT0fEARHxcHPliFgPLAAGkb29fTZZYrMtsGdELCR7U/djKYbvAIW3KgcBQyPiFOAi4E8R8SHgl0DzRX4Y8PeI+EhE7Af8toTz8WJEHAjcAHyrvcqSPgh8Dvh4RBwArAe+EBGzyD6pcjnwfeCOiHiS7EPTD6fzcU0J8ZhtDXwd2lR716HxwLdTEnW5pIEltAnwdkQMBv6b7NNPZwH7AadJ2iXV2Rv4AbBvmj5Pdm6+RXb8kH1B4RMR8VHgu8CVBfs4EDghIg4j+6zaaQDpd/u/gCklxrrVcQJnZYmIV8guSOe0WPUxNt7mu53sP+C2/DNwR2rzKeBZ4J/a2eZh4F/S9L3UxsFk335sbvP21OYDwC6SeqV1kyPijTT/LwX7ngKsTuVzgCMlXSXpExGxpp14IPuoMsCjZLc723MEcBAwS9Ljafn9ad2lwJHAYLIkzsyK8HVoM21ehyLicbLrzH8BO5Ndfz5YQrvN39CdA8yNiOUR8RawiI0fYH8mIuZExLvAXOD+yD7xNKcglp2An6XevmuADxXsY3pErEpx/oHsO759yRLqn/u2auucwFlH/JDs2249q7zfPwKfAIYAU4HeZN8hfLiNbZq93l6FiPgb2V+Dc4DLJZXynMlb6ed6SnumVGQ9BAekaZ+IuDit2wXYEXgv2XdJzax1vg5t1O51KCJei4hfRMSZZInjsWS3owvzgJbXneZ23y2Yb17u3qJOy3qFdS4DHkw9ip9psZ+W5+Q24IvA6WQ9h9YKJ3BWtvTX0iSyi2ez/wFOTvNfoP2L2cOpHpL+iez2wYJ2tnmErEv93Yh4k+zD118hu6C2bPOTZLcVXinSzh/JuvmRdAzZx5yRtAewNiLuIPtL9cB24umI+4ETJP1D2ufOkt6X1t0I/CfZrZ+rUtmrZAmdmRXwdah0kj4uqbn9HmS3cp8FVgD/kJ7H2xb49JbuqxU7sfEZvNPaqXsr8HWAiJhXoXi6BI9CtY76AXB2wfLXgFsk/R9gJdlfT0j6KkBE/HeL7a8HbpA0h+yvwNPSQ7Wt7jCtXwLMSEUPk3Wzz0nLFwPjJT0BrAVGttLUJcBdkuaSXfCfS+X7A/8l6V2yEVr/3mowHRQR8yRdCPxO0nvSfs6SdBjwTkT8VFI34H8kHZ6Ocb2kvwK3+jk4s034OlSaD5Adp8g6bqaQ3Z4MSZeSJaXLyJ5Vq4Tvkz1feCHtPNMWESskzQd+VaFYugxlt6rNzMzMakvSDmTJ8IElPv+31fItVDMzM6s5ZS8nng/82Mlb+9wDZ2ZmZpYz7oEzMzMzyxkncGZmZmY54wTOzMzMLGecwJmZmZnljBM4MzMzs5xxAmdmZmaWM/8fTj8v6QdUDMcAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 720x360 with 2 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "text_word_count = []\n",
        "summary_word_count = []\n",
        "\n",
        "# populate the lists with sentence lengths\n",
        "for i in tokenized_corpus_src[:train.shape[0]]:\n",
        "      text_word_count.append(len(i.split()))\n",
        "\n",
        "for i in tokenized_corpus_trg[:train.shape[0]]:\n",
        "      summary_word_count.append(len(i.split()))\n",
        "\n",
        "fig , ax = plt.subplots(1,2)\n",
        "fig.set_size_inches(10,5)\n",
        "ax[0].hist(text_word_count, bins=25, range=[0, 1000])\n",
        "ax[1].hist(summary_word_count, bins=25, range=[0, 100])\n",
        "ax[0].set_xlabel(\"No. of words in Text\")\n",
        "ax[0].set_ylabel(\"Frequency\")\n",
        "ax[1].set_xlabel(\"No. of words in Summary\")\n",
        "ax[1].set_ylabel(\"Frequency\")\n",
        "ax[0].set_title('Distribution of training data')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 350
        },
        "id": "H8N750Lg6ePt",
        "outputId": "aa834cbd-d168-4a34-c438-ffc873e6ccd2"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x360 with 2 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "for i in tokenized_corpus_src[:-test.shape[0]]:\n",
        "      text_word_count.append(len(i.split()))\n",
        "\n",
        "for i in tokenized_corpus_trg[:-test.shape[0]]:\n",
        "      summary_word_count.append(len(i.split()))\n",
        "\n",
        "fig , ax = plt.subplots(1,2)\n",
        "fig.set_size_inches(10,5)\n",
        "ax[0].hist(text_word_count, bins=25, range=[0, 1000])\n",
        "ax[1].hist(summary_word_count, bins=25, range=[0, 100])\n",
        "ax[0].set_xlabel(\"No. of words in Text\")\n",
        "ax[0].set_ylabel(\"Frequency\")\n",
        "ax[1].set_xlabel(\"No. of words in Summary\")\n",
        "ax[1].set_ylabel(\"Frequency\")\n",
        "ax[0].set_title('Distribution of Testing data')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fx80oeV91e-B",
        "outputId": "34ddd93e-2375-4ba2-9e9b-1734b071ffe2"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "::::: Text_Cleaned :::::\n",
            "0    दिल्ली में सुप्रीम कोर्ट के डीज़ल टैक्सियों को...\n",
            "1    जॉर्डन के ऐतिहासिक दौरे पर पहुंचे राष्ट्रपति प...\n",
            "2    पाकिस्तानी नेताओं को विवादित और हास्यास्पद बया...\n",
            "3    पीएम नरेंद्र मोदी बायोपिक में विवेक ओबेरॉय ने ...\n",
            "4    देश दुनिया महानगर खेल आर्थिक और बॉलीवुड में क्...\n",
            "Name: Text_Cleaned, dtype: object \n",
            "\n",
            "::::: Summary :::::\n",
            "0     दिल्ली में डीजल टैक्सियों पर बैन से मुश्किल म...\n",
            "1    जॉर्डन राष्ट्रपति मुखर्जी ने करोड़ डॉलर के संय...\n",
            "2     में पाकिस्तान की राजदूत मलीहा लोधी ने कराई फज...\n",
            "3     देशों में पीएम नरेंद्र मोदी बायोपिक को रिलीज ...\n",
            "4                      अगस्त दिनभर की बड़ी खबरें पढ़ें\n",
            "Name: Summary_Cleaned, dtype: object \n",
            "\n"
          ]
        }
      ],
      "source": [
        "news['Text_Cleaned'] = tokenized_corpus_src  \n",
        "print(\"::::: Text_Cleaned :::::\")\n",
        "print(news['Text_Cleaned'][0:5], \"\\n\")\n",
        "\n",
        "\n",
        "news['Summary_Cleaned'] =  tokenized_corpus_trg \n",
        "print(\"::::: Summary :::::\")\n",
        "print(news['Summary_Cleaned'][0:5], \"\\n\")\n",
        "\n",
        "corpus = list(news['Text_Cleaned'])\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "5yvUrDGQ25z1"
      },
      "outputs": [],
      "source": [
        "text_count = []\n",
        "summary_count = []\n",
        "\n",
        "for sent in news['Text_Cleaned']:\n",
        "    text_count.append(len(sent.split()))\n",
        "for sent in news['Summary_Cleaned']:\n",
        "    summary_count.append(len(sent.split()))\n",
        "\n",
        "graph_df = pd.DataFrame()\n",
        "graph_df['text'] = text_count\n",
        "graph_df['summary'] = summary_count"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "PoQEg3gb2_9E",
        "outputId": "0f3919db-249d-4a97-fa97-4c1efee662dc"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "count    333260.000000\n",
              "mean        348.910493\n",
              "std         314.028859\n",
              "min           0.000000\n",
              "25%         197.000000\n",
              "50%         282.000000\n",
              "75%         399.000000\n",
              "max       15089.000000\n",
              "Name: text, dtype: float64"
            ]
          },
          "execution_count": 33,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "graph_df['text'].describe()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ikYNF9wh3FtH",
        "outputId": "f5ae86e9-f9d4-4801-e7c2-1c148e289b22"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "count    333260.000000\n",
              "mean         10.842576\n",
              "std           3.001873\n",
              "min           0.000000\n",
              "25%           9.000000\n",
              "50%          11.000000\n",
              "75%          13.000000\n",
              "max         162.000000\n",
              "Name: summary, dtype: float64"
            ]
          },
          "execution_count": 34,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "graph_df['summary'].describe()\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "id": "jubcT55k3OYb",
        "outputId": "31bdf0da-82cf-4364-85a2-d7dd17f70c14"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "graph_df['text'].hist(bins = 25, range=(0, 300))\n",
        "plt.show()\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "id": "Rbb2zU523UvI",
        "outputId": "ad8ee58a-57de-4077-9b90-4a5dbdc50db8"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "graph_df['summary'].hist(bins = 15, range=(0, 20))\n",
        "plt.show()\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "k6jbgjJ03aP1",
        "outputId": "cbaf25b4-fd1a-4edb-c3ac-2a501ca3bc31"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "0.5509842165276361\n"
          ]
        }
      ],
      "source": [
        "count = 0\n",
        "for i in graph_df['text']:\n",
        "    if i > 10 and i <= 300:\n",
        "        count = count + 1\n",
        "print(count / len(graph_df['text']))\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "3n8lE86g3dVp",
        "outputId": "22a6c81a-899b-456f-99b4-d113ddabae57"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "0.9994118706115346\n"
          ]
        }
      ],
      "source": [
        "count = 0\n",
        "for i in graph_df['summary']:\n",
        "    if i > 1 and i <= 20:\n",
        "        count = count + 1\n",
        "print(count / len(graph_df['summary']))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "sKVyb0_I3hz6"
      },
      "outputs": [],
      "source": [
        "max_text_len = 300\n",
        "max_summary_len = 16\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "2KPtuQ263pWr"
      },
      "outputs": [],
      "source": [
        "cleaned_text = np.array(news['Text_Cleaned'])\n",
        "cleaned_summary = np.array(news['Summary_Cleaned'])\n",
        "\n",
        "short_text = []\n",
        "short_summary = []\n",
        "\n",
        "for i in range(len(cleaned_text)):\n",
        "    if(len(cleaned_summary[i].split()) <= max_summary_len \n",
        "       and len(cleaned_summary[i].split()) > 1 \n",
        "       and len(cleaned_text[i].split()) <= max_text_len):\n",
        "        short_text.append(cleaned_text[i])\n",
        "        short_summary.append(cleaned_summary[i])\n",
        "    elif((len(cleaned_summary[i].split()) > max_summary_len) or (len(cleaned_text[i].split()) > max_text_len)):\n",
        "        summ = cleaned_summary[i].split()\n",
        "        txt = cleaned_text[i].split()\n",
        "        if(len(summ)>max_summary_len):\n",
        "          short_summary.append(' '.join(summ[:max_summary_len]))\n",
        "        else:\n",
        "          short_summary.append(cleaned_summary[i])\n",
        "        if(len(txt)>max_text_len):\n",
        "          short_text.append(' '.join(txt[:max_text_len]))\n",
        "        else:\n",
        "          short_text.append(cleaned_text[i]) \n",
        "    \n",
        "post_pre = pd.DataFrame({'text':short_text,'summary':short_summary})"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "YO4gf-0J3uMl"
      },
      "outputs": [],
      "source": [
        "post_pre['summary'] = post_pre['summary'].apply(lambda x : 'sostok '+ x + ' eostok')\n",
        "post_pre['text'] = post_pre['text'].apply(lambda x: 'sostok ' + x + ' eostok')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "PkEfUcfK3z8f",
        "outputId": "28f1c910-0f44-4ed2-c27e-85a5b16e195b"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "(333143, 2)"
            ]
          },
          "execution_count": 42,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "post_pre.shape"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "QLjXenrh33bJ",
        "outputId": "242a0edf-15b8-42a3-d64a-4aca958efcca"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "cleaned text's size  (333143,)\n",
            "cleaned summary's size  (333143,)\n"
          ]
        }
      ],
      "source": [
        "cleaned_summary = np.array(post_pre['summary'])\n",
        "cleaned_text = np.array(post_pre['text'])\n",
        "print(\"cleaned text's size \",cleaned_text.shape)\n",
        "print(\"cleaned summary's size \",cleaned_summary.shape)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Td7_wzZZyH0K"
      },
      "source": [
        "### Tokenizer and Padding"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Xrr2qffo-Q--"
      },
      "outputs": [],
      "source": [
        "from keras.preprocessing.text import Tokenizer \n",
        "from keras.preprocessing.sequence import pad_sequences"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "QH8XCxHB-jMj"
      },
      "outputs": [],
      "source": [
        "#Tokenize text to get the vocab count\n",
        "#prepare a tokenizer for training data\n",
        "text_tokenizer = Tokenizer() \n",
        "text_tokenizer.fit_on_texts(list(cleaned_text))\n",
        "\n",
        "#prepare a tokenizer for reviews on training data\n",
        "summary_tokenizer = Tokenizer()   \n",
        "summary_tokenizer.fit_on_texts(list(cleaned_summary))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "J-2pwY5G_xgW",
        "outputId": "bea0c779-17af-4b44-9804-f280748900b2"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "{'“याचिकाकर्ता': 221385, 'केशवचन्द्र': 158522, 'रामधनु': 126226, 'एंग': 94482, 'बर्टोलूसी': 208227, 'मतांजिमा': 277157, 'बाहडेई': 353641, 'मुस्लीमानी': 293851, 'टीएटी': 321408, 'हबाईपुर': 207593}\n"
          ]
        }
      ],
      "source": [
        "vocab = text_tokenizer.word_index\n",
        "import random\n",
        "def sample_from_dict(d, sample=10):\n",
        "    keys = random.sample(list(d), sample)\n",
        "    values = [d[k] for k in keys]\n",
        "    return dict(zip(keys, values))\n",
        "\n",
        "print(sample_from_dict(vocab))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kNPUJiel379i",
        "outputId": "4ce30229-5e98-473e-f3fa-41595b36c8f1"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "353646"
            ]
          },
          "execution_count": 47,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "len(vocab)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mR6KfrRUAuMG",
        "outputId": "cb70c677-6723-4cd7-fa3c-1651cf26e1d2"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "% of rare words in vocabulary: 68.68167602630879\n",
            "Total Coverage of rare words: 0.41351097036187695\n"
          ]
        }
      ],
      "source": [
        "thresh=4\n",
        "\n",
        "cnt=0\n",
        "tot_cnt=0\n",
        "freq=0\n",
        "tot_freq=0\n",
        "\n",
        "for key,value in text_tokenizer.word_counts.items():\n",
        "    tot_cnt=tot_cnt+1\n",
        "    tot_freq=tot_freq+value\n",
        "    if(value<thresh):\n",
        "        cnt=cnt+1\n",
        "        freq=freq+value\n",
        "    \n",
        "print(\"% of rare words in vocabulary:\",(cnt/tot_cnt)*100)\n",
        "print(\"Total Coverage of rare words:\",(freq/tot_freq)*100)\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "CcKsljAwB0kM",
        "outputId": "04f5eecc-ac46-4b00-d29a-a5c5b0c0a98e"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Size of vocabulary in text = 110757\n"
          ]
        }
      ],
      "source": [
        "# #prepare a tokenizer for reviews on training data\n",
        "text_tokenizer = Tokenizer(num_words = tot_cnt - cnt) \n",
        "text_tokenizer.fit_on_texts(list(cleaned_text))\n",
        "\n",
        "#convert text sequences into integer sequences (i.e one-hot encodeing all the words)\n",
        "text__seq    =   text_tokenizer.texts_to_sequences(cleaned_text)\n",
        "#padding zero upto maximum length\n",
        "cleaned_text    =   pad_sequences(text__seq,  maxlen=max_text_len, padding='post')\n",
        "\n",
        "#size of vocabulary ( +1 for padding token)\n",
        "text_voc   =  text_tokenizer.num_words + 1\n",
        "\n",
        "print(\"Size of vocabulary in text = {}\".format(text_voc))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Jx6Xk8l2HKxc",
        "outputId": "dc41e8ae-707d-4eb5-ff1d-ef0003760624"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "% of rare words in vocabulary: 58.05363524028988\n",
            "Total Coverage of rare words: 1.1603857371100994\n"
          ]
        }
      ],
      "source": [
        "thresh=4\n",
        "\n",
        "cnt=0\n",
        "tot_cnt=0\n",
        "freq=0\n",
        "tot_freq=0\n",
        "\n",
        "for key,value in summary_tokenizer.word_counts.items():\n",
        "    tot_cnt=tot_cnt+1\n",
        "    tot_freq=tot_freq+value\n",
        "    if(value<thresh):\n",
        "        cnt=cnt+1\n",
        "        freq=freq+value\n",
        "    \n",
        "print(\"% of rare words in vocabulary:\",(cnt/tot_cnt)*100)\n",
        "print(\"Total Coverage of rare words:\",(freq/tot_freq)*100)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Ez9ZOpk7GuE7",
        "outputId": "9e0deca9-407c-4e83-be97-2507e44eda7b"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Size of vocabulary in summary = 25121\n"
          ]
        }
      ],
      "source": [
        "#prepare a tokenizer for reviews on training data\n",
        "summary_tokenizer = Tokenizer(num_words = tot_cnt - cnt) \n",
        "summary_tokenizer.fit_on_texts(list(cleaned_summary))\n",
        "\n",
        "#convert text sequences into integer sequences (i.e one-hot encodeing all the words)\n",
        "summary__seq    =   summary_tokenizer.texts_to_sequences(cleaned_summary)\n",
        "#padding zero upto maximum length\n",
        "cleaned_summary    =   pad_sequences(summary__seq,  maxlen=max_summary_len, padding='post')\n",
        "\n",
        "#size of vocabulary ( +1 for padding token)\n",
        "summary_voc   =  summary_tokenizer.num_words + 1\n",
        "\n",
        "print(\"Size of vocabulary in summary = {}\".format(summary_voc))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "URQqx2HBHTaS",
        "outputId": "c1dddbe6-bb83-40d3-bdf2-0ec1b8d25020"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "(333143, 333143)"
            ]
          },
          "execution_count": 52,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "summary_tokenizer.word_counts['sostok'],len(cleaned_summary)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "G6IQjt1W3JoJ"
      },
      "outputs": [],
      "source": [
        "import pickle\n",
        "\n",
        "with open(default_path+'text_tokenizer.pickle', 'wb') as handle:\n",
        "    pickle.dump(text_tokenizer, handle, protocol=pickle.HIGHEST_PROTOCOL)\n",
        "\n",
        "with open(default_path+'summary_tokenizer.pickle', 'wb') as handle:\n",
        "    pickle.dump(summary_tokenizer, handle, protocol=pickle.HIGHEST_PROTOCOL)\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "w0ri_-I_HhbW"
      },
      "outputs": [],
      "source": [
        "ind=[]\n",
        "for i in range(len(cleaned_summary)):\n",
        "    cnt=0\n",
        "    for j in cleaned_summary[i]:\n",
        "        if j!=0:\n",
        "            cnt=cnt+1\n",
        "    if(cnt==2):\n",
        "        ind.append(i)\n",
        "\n",
        "cleaned_text = np.delete(cleaned_text, ind, axis=0)\n",
        "cleaned_summary = np.delete(cleaned_summary, ind, axis=0)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_0fJWmTj3YFk",
        "outputId": "5fe346b5-b01c-4c99-f302-41bb6b360b7b"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "cleaned text's size  (333130, 300)\n",
            "cleaned summary's size  (333130, 16)\n"
          ]
        }
      ],
      "source": [
        "print(\"cleaned text's size \",cleaned_text.shape)\n",
        "print(\"cleaned summary's size \",cleaned_summary.shape)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "p2MuokRi48mI",
        "outputId": "d0ecb7fe-44ea-483a-92fc-663db2feb748"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "array([[   24,    56,     2, ...,     0,     0,     0],\n",
              "       [11094,     1,  1708, ...,   162,  3278,    25],\n",
              "       [   24,   580,   333, ...,     0,     0,     0],\n",
              "       ...,\n",
              "       [    6,   580,   665, ..., 18583,     8,    25],\n",
              "       [   24,  2442,  1083, ...,     0,     0,     0],\n",
              "       [   24,   310,     1, ...,     0,     0,     0]], dtype=int32)"
            ]
          },
          "execution_count": 56,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "cleaned_text"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ZuPh_5IL7ITE",
        "outputId": "9118651a-4a97-46a0-8125-8fcf4d83de16"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "array([[    1,    21,     3, ...,     2,     0,     0],\n",
              "       [    1,  9770,   206, ...,     0,     0,     0],\n",
              "       [    1,     3,    68, ...,     2,     0,     0],\n",
              "       ...,\n",
              "       [    1,    68,     9, ...,     0,     0,     0],\n",
              "       [   94,    23,    12, ..., 17783, 11079,     2],\n",
              "       [    1,     4,    26, ...,     2,     0,     0]], dtype=int32)"
            ]
          },
          "execution_count": 57,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "cleaned_summary"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9mBMOlMy6wR6"
      },
      "source": [
        "### Saving Final Input Text"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "pL1QJgb96svJ"
      },
      "outputs": [],
      "source": [
        "with open(default_path+'cleaned_text.pickle', 'wb') as handle:\n",
        "    pickle.dump(cleaned_text, handle, protocol=pickle.HIGHEST_PROTOCOL)\n",
        "\n",
        "with open(default_path+'cleaned_summary.pickle', 'wb') as handle:\n",
        "    pickle.dump(cleaned_summary, handle, protocol=pickle.HIGHEST_PROTOCOL)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xP44gjHSx8U_"
      },
      "source": [
        "### Embedding Space"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bRvaKAo0l93y",
        "outputId": "0bb731d5-3184-4369-bf24-6a143f7861cc"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Collecting fasttext\n",
            "  Downloading fasttext-0.9.2.tar.gz (68 kB)\n",
            "\u001b[K     |████████████████████████████████| 68 kB 3.0 MB/s \n",
            "\u001b[?25hCollecting pybind11>=2.2\n",
            "  Using cached pybind11-2.9.2-py2.py3-none-any.whl (213 kB)\n",
            "Requirement already satisfied: setuptools>=0.7.0 in /usr/local/lib/python3.7/dist-packages (from fasttext) (57.4.0)\n",
            "Requirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from fasttext) (1.21.6)\n",
            "Building wheels for collected packages: fasttext\n",
            "  Building wheel for fasttext (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for fasttext: filename=fasttext-0.9.2-cp37-cp37m-linux_x86_64.whl size=3145175 sha256=bfc82251de90f7ed268ed54b54ca7b1077f3f37e407b5f27071ce05fff6c959d\n",
            "  Stored in directory: /root/.cache/pip/wheels/4e/ca/bf/b020d2be95f7641801a6597a29c8f4f19e38f9c02a345bab9b\n",
            "Successfully built fasttext\n",
            "Installing collected packages: pybind11, fasttext\n",
            "Successfully installed fasttext-0.9.2 pybind11-2.9.2\n"
          ]
        }
      ],
      "source": [
        "!pip install fasttext"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "saN0iEShoNzP"
      },
      "outputs": [],
      "source": [
        "import fasttext.util"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "1vjJw6w5wY_e"
      },
      "outputs": [],
      "source": [
        "#fasttext.util.download_model('hi', if_exists='ignore')  # Hindi"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "zz9kFK6Hvzao",
        "outputId": "f775eee3-8a14-4f10-ea46-d36414dffd58"
      },
      "outputs": [
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "Warning : `load_model` does not return WordVectorModel or SupervisedModel any more, but a `FastText` object which is very similar.\n"
          ]
        }
      ],
      "source": [
        "ft = fasttext.load_model(default_path+'cc.hi.300.bin')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tPjpXAG5wMm8",
        "outputId": "329e7859-d348-45e4-f8a2-360ba113718a"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Embedding vector shape is 300\n"
          ]
        }
      ],
      "source": [
        "word = \"नृत्य\"\n",
        "print(\"Embedding vector shape is {}\".format(ft.get_word_vector(word).shape[0]))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "V6_ZUkajyZR2"
      },
      "source": [
        "### Embedding Matrix - Get Vectors"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Vrbxj9G9yY1e"
      },
      "outputs": [],
      "source": [
        "def create_weight_matrix(model, DICT_SIZE, tokenizer):\n",
        "  '''\n",
        "  Accepts word embedding model\n",
        "  and the second model, if provided\n",
        "  Returns weight matrix of size m*n, where\n",
        "  m - size of the dictionary\n",
        "  n - size of the word embedding vector\n",
        "\n",
        "  '''\n",
        "  vector_size = model.get_word_vector(\"नृत्य\").shape[0]\n",
        "  w_matrix = np.zeros((DICT_SIZE, vector_size))\n",
        "  unk_words = []\n",
        "  text_words = []\n",
        "  for word, index in tokenizer.word_index.items():\n",
        "    if index < DICT_SIZE:\n",
        "      if word in model.words: \n",
        "        w_matrix[index] = model.get_word_vector(word)\n",
        "        text_words.append(word)\n",
        "      else:\n",
        "        w_matrix[index] = np.random.rand(1, vector_size)\n",
        "        unk_words.append(word)\n",
        "        \n",
        "  \n",
        "  print('{} words were considered as they were found in embedding space:'.format(len(text_words)))\n",
        "  print('{} are some such words'.format(text_words[:10]))\n",
        "  print()  \n",
        "  print('{} words were randomly alloted as they were not found in embedding space:'.format(len(unk_words)))\n",
        "  print('{} are some such words'.format(unk_words[:10]))\n",
        "\n",
        "  return w_matrix"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "background_save": true
        },
        "id": "41UFCTW_yctB",
        "outputId": "9a40842d-289d-44e0-b19b-6002c7cd19c5"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "101084 words were considered as they were found in embedding space:\n",
            "['के', 'में', 'है', 'की', 'को', 'ने', 'से', 'और', 'का', 'पर'] are some such words\n",
            "\n",
            "9672 words were randomly alloted as they were not found in embedding space:\n",
            "['sostok', 'eostok', 'सिंबा', 'जेजेपी', 'एयरस्ट्राइक', 'रियलमी', 'इनपुटः', 'मीटू', 'एनालिटिका', 'आर्काइव्ड'] are some such words\n"
          ]
        }
      ],
      "source": [
        "text_matrix = create_weight_matrix(ft, text_voc, text_tokenizer)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "background_save": true
        },
        "id": "HqnYKxuBvG4L",
        "outputId": "3d14ca9f-191d-44b4-b186-4de587d84e83"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "24917 words were considered as they were found in embedding space:\n",
            "['में', 'के', 'की', 'का', 'ने', 'से', 'को', 'पर', 'है', 'नहीं'] are some such words\n",
            "\n",
            "203 words were randomly alloted as they were not found in embedding space:\n",
            "['sostok', 'eostok', 'एयरस्ट्राइक', 'दीपवीर', 'सिंबा', 'फरीदाबादः', 'रसूखदारः', 'ओडिशाः', 'मिजाजः', 'तेलंगानाः'] are some such words\n"
          ]
        }
      ],
      "source": [
        "summ_matrix = create_weight_matrix(ft, summary_voc, summary_tokenizer)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Knr120_e3LEZ"
      },
      "source": [
        "### Saving Embedding Matrices"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "background_save": true
        },
        "id": "7ejdQ-tM0jE2"
      },
      "outputs": [],
      "source": [
        "with open(default_path+'text_matrix.pickle', 'wb') as handle:\n",
        "    pickle.dump(text_matrix, handle, protocol=pickle.HIGHEST_PROTOCOL)\n",
        "\n",
        "with open(default_path+'summ_matrix.pickle', 'wb') as handle:\n",
        "    pickle.dump(summ_matrix, handle, protocol=pickle.HIGHEST_PROTOCOL)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vFf9ANuoqS7B"
      },
      "source": [
        "### Model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "metadata": {
        "id": "Zbxhyl_zFlWL"
      },
      "outputs": [],
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "import tensorflow as tf\n",
        "import keras\n",
        "import time\n",
        "import re\n",
        "import pickle\n",
        "import matplotlib.pyplot as plt "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yH5cg5pSIHaZ"
      },
      "source": [
        "### Loading Cleaned Dataset"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {
        "id": "slZ5f4P4DurS"
      },
      "outputs": [],
      "source": [
        "BUFFER_SIZE = 20000\n",
        "BATCH_SIZE = 64"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 13,
      "metadata": {
        "id": "aF1bqvLy7BQ-"
      },
      "outputs": [],
      "source": [
        "with open(default_path+'cleaned_text.pickle', 'rb') as handle:\n",
        "    cleaned_text = pickle.load(handle)\n",
        "with open(default_path+'cleaned_summary.pickle', 'rb') as handle:\n",
        "    cleaned_summary = pickle.load(handle)"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "cleaned_text.shape"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fT9kdiFOWjzr",
        "outputId": "cb1a9a35-be28-404c-9f68-6fecd8a60c17"
      },
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(86629, 300)"
            ]
          },
          "metadata": {},
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kDGbc5KEtBDw"
      },
      "source": [
        "#### Train Test Split"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "metadata": {
        "id": "hM-jngKwtAa4"
      },
      "outputs": [],
      "source": [
        "x_train, y_train = cleaned_text[:cleaned_text.shape[0]-test.shape[0], :], cleaned_summary[:cleaned_summary.shape[0]-test.shape[0], :]\n",
        "x_test, y_test = cleaned_text[:-test.shape[0], :], cleaned_summary[:-test.shape[0], :]"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {
        "id": "wI-fV7eABWN6"
      },
      "outputs": [],
      "source": [
        "dataset = tf.data.Dataset.from_tensor_slices((x_train, y_train)).shuffle(BUFFER_SIZE).batch(BATCH_SIZE)\n",
        "test_dataset = tf.data.Dataset.from_tensor_slices((x_test, y_test)).batch(BATCH_SIZE)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "isN1CpAXLfsl"
      },
      "source": [
        "### Positional Encoding for adding notion of position among words as unlike RNN this is non-directional"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 17,
      "metadata": {
        "id": "Purv7oyhETDZ"
      },
      "outputs": [],
      "source": [
        "def get_angles(position, i, d_model):\n",
        "    angle_rates = 1 / np.power(10000, (2 * (i // 2)) / np.float32(d_model))\n",
        "    return position * angle_rates"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 18,
      "metadata": {
        "id": "40J2pc2NEXp5"
      },
      "outputs": [],
      "source": [
        "def positional_encoding(position, d_model):\n",
        "    angle_rads = get_angles(\n",
        "        np.arange(position)[:, np.newaxis],\n",
        "        np.arange(d_model)[np.newaxis, :],\n",
        "        d_model\n",
        "    )\n",
        "\n",
        "    # apply sin to even indices in the array; 2i\n",
        "    angle_rads[:, 0::2] = np.sin(angle_rads[:, 0::2])\n",
        "\n",
        "    # apply cos to odd indices in the array; 2i+1\n",
        "    angle_rads[:, 1::2] = np.cos(angle_rads[:, 1::2])\n",
        "\n",
        "    pos_encoding = angle_rads[np.newaxis, ...]\n",
        "\n",
        "    return tf.cast(pos_encoding, dtype=tf.float32)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "24Pe01DMMWHc"
      },
      "source": [
        "### Masking\n",
        "\n",
        "- Padding mask for masking \"pad\" sequences\n",
        "- Lookahead mask for masking future words from contributing in prediction of current words in self attention"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 19,
      "metadata": {
        "id": "hN1wVQAdMVYy"
      },
      "outputs": [],
      "source": [
        "def create_padding_mask(seq):\n",
        "    seq = tf.cast(tf.math.equal(seq, 0), tf.float32)\n",
        "    return seq[:, tf.newaxis, tf.newaxis, :]"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 20,
      "metadata": {
        "id": "UmjAPLWuMREE"
      },
      "outputs": [],
      "source": [
        "def create_look_ahead_mask(size):\n",
        "    mask = 1 - tf.linalg.band_part(tf.ones((size, size)), -1, 0)\n",
        "    return mask"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "n8DqUBc4NFOy"
      },
      "source": [
        "### Building the Model"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WfknVF7hNKf7"
      },
      "source": [
        "#### Scaled Dot Product"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 21,
      "metadata": {
        "id": "w_B6M9OBNBKB"
      },
      "outputs": [],
      "source": [
        "def scaled_dot_product_attention(q, k, v, mask):\n",
        "    matmul_qk = tf.matmul(q, k, transpose_b=True)\n",
        "\n",
        "    dk = tf.cast(tf.shape(k)[-1], tf.float32)\n",
        "    scaled_attention_logits = matmul_qk / tf.math.sqrt(dk)\n",
        "\n",
        "    if mask is not None:\n",
        "        scaled_attention_logits += (mask * -1e9)  \n",
        "\n",
        "    attention_weights = tf.nn.softmax(scaled_attention_logits, axis=-1)\n",
        "\n",
        "    output = tf.matmul(attention_weights, v)\n",
        "    return output, attention_weights"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Rf7_a5uQOfJk"
      },
      "source": [
        "#### Multi-Headed Attention"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 22,
      "metadata": {
        "id": "iIuFrdXnNZEC"
      },
      "outputs": [],
      "source": [
        "class MultiHeadAttention(tf.keras.layers.Layer):\n",
        "    def __init__(self, d_model, num_heads):\n",
        "        super(MultiHeadAttention, self).__init__()\n",
        "        self.num_heads = num_heads\n",
        "        self.d_model = d_model\n",
        "\n",
        "        assert d_model % self.num_heads == 0\n",
        "\n",
        "        self.depth = d_model // self.num_heads\n",
        "\n",
        "        self.wq = tf.keras.layers.Dense(d_model)\n",
        "        self.wk = tf.keras.layers.Dense(d_model)\n",
        "        self.wv = tf.keras.layers.Dense(d_model)\n",
        "\n",
        "        self.dense = tf.keras.layers.Dense(d_model)\n",
        "        \n",
        "    def split_heads(self, x, batch_size):\n",
        "        x = tf.reshape(x, (batch_size, -1, self.num_heads, self.depth))\n",
        "        return tf.transpose(x, perm=[0, 2, 1, 3])\n",
        "    \n",
        "    def call(self, v, k, q, mask):\n",
        "        batch_size = tf.shape(q)[0]\n",
        "\n",
        "        q = self.wq(q)\n",
        "        k = self.wk(k)\n",
        "        v = self.wv(v)\n",
        "\n",
        "        q = self.split_heads(q, batch_size)\n",
        "        k = self.split_heads(k, batch_size)\n",
        "        v = self.split_heads(v, batch_size)\n",
        "\n",
        "        scaled_attention, attention_weights = scaled_dot_product_attention(\n",
        "            q, k, v, mask)\n",
        "\n",
        "        scaled_attention = tf.transpose(scaled_attention, perm=[0, 2, 1, 3])\n",
        "\n",
        "        concat_attention = tf.reshape(scaled_attention, (batch_size, -1, self.d_model))\n",
        "        output = self.dense(concat_attention)\n",
        "            \n",
        "        return output, attention_weights"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "A49tXMVvOkOZ"
      },
      "source": [
        "### Feed Forward Network"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 23,
      "metadata": {
        "id": "d9-qoKuTNwKq"
      },
      "outputs": [],
      "source": [
        "def point_wise_feed_forward_network(d_model, dff):\n",
        "    return tf.keras.Sequential([\n",
        "        tf.keras.layers.Dense(dff, activation='relu'),\n",
        "        tf.keras.layers.Dense(d_model)\n",
        "    ])"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "B2RRmn2bOpW9"
      },
      "source": [
        "#### Fundamental Unit of Transformer encoder"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 24,
      "metadata": {
        "id": "HNuoJoFWO335"
      },
      "outputs": [],
      "source": [
        "class EncoderLayer(tf.keras.layers.Layer):\n",
        "    def __init__(self, d_model, num_heads, dff, rate=0.1):\n",
        "        super(EncoderLayer, self).__init__()\n",
        "\n",
        "        self.mha = MultiHeadAttention(d_model, num_heads)\n",
        "        self.ffn = point_wise_feed_forward_network(d_model, dff)\n",
        "\n",
        "        self.layernorm1 = tf.keras.layers.LayerNormalization(epsilon=1e-6)\n",
        "        self.layernorm2 = tf.keras.layers.LayerNormalization(epsilon=1e-6)\n",
        "\n",
        "        self.dropout1 = tf.keras.layers.Dropout(rate)\n",
        "        self.dropout2 = tf.keras.layers.Dropout(rate)\n",
        "    \n",
        "    def call(self, x, training, mask):\n",
        "        attn_output, _ = self.mha(x, x, x, mask)\n",
        "        attn_output = self.dropout1(attn_output, training=training)\n",
        "        out1 = self.layernorm1(x + attn_output)\n",
        "\n",
        "        ffn_output = self.ffn(out1)\n",
        "        ffn_output = self.dropout2(ffn_output, training=training)\n",
        "        out2 = self.layernorm2(out1 + ffn_output)\n",
        "\n",
        "        return out2\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9i6Zh8gnPqdW"
      },
      "source": [
        "#### Fundamental Unit of Transformer decoder"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 25,
      "metadata": {
        "id": "7CVmvs6dPMRC"
      },
      "outputs": [],
      "source": [
        "class DecoderLayer(tf.keras.layers.Layer):\n",
        "    def __init__(self, d_model, num_heads, dff, rate=0.1):\n",
        "        super(DecoderLayer, self).__init__()\n",
        "\n",
        "        self.mha1 = MultiHeadAttention(d_model, num_heads)\n",
        "        self.mha2 = MultiHeadAttention(d_model, num_heads)\n",
        "\n",
        "        self.ffn = point_wise_feed_forward_network(d_model, dff)\n",
        "\n",
        "        self.layernorm1 = tf.keras.layers.LayerNormalization(epsilon=1e-6)\n",
        "        self.layernorm2 = tf.keras.layers.LayerNormalization(epsilon=1e-6)\n",
        "        self.layernorm3 = tf.keras.layers.LayerNormalization(epsilon=1e-6)\n",
        "\n",
        "        self.dropout1 = tf.keras.layers.Dropout(rate)\n",
        "        self.dropout2 = tf.keras.layers.Dropout(rate)\n",
        "        self.dropout3 = tf.keras.layers.Dropout(rate)\n",
        "    \n",
        "    \n",
        "    def call(self, x, enc_output, training, look_ahead_mask, padding_mask):\n",
        "        attn1, attn_weights_block1 = self.mha1(x, x, x, look_ahead_mask)\n",
        "        attn1 = self.dropout1(attn1, training=training)\n",
        "        out1 = self.layernorm1(attn1 + x)\n",
        "\n",
        "        attn2, attn_weights_block2 = self.mha2(enc_output, enc_output, out1, padding_mask)\n",
        "        attn2 = self.dropout2(attn2, training=training)\n",
        "        out2 = self.layernorm2(attn2 + out1)\n",
        "\n",
        "        ffn_output = self.ffn(out2)\n",
        "        ffn_output = self.dropout3(ffn_output, training=training)\n",
        "        out3 = self.layernorm3(ffn_output + out2)\n",
        "\n",
        "        return out3, attn_weights_block1, attn_weights_block2\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6zt5MUc_QNid"
      },
      "source": [
        "#### Encoder consisting of multiple EncoderLayer(s)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 26,
      "metadata": {
        "id": "BrbnTwijQJ-h"
      },
      "outputs": [],
      "source": [
        "class Encoder(tf.keras.layers.Layer):\n",
        "    def __init__(self, num_layers, d_model, num_heads, dff, input_vocab_size, maximum_position_encoding, emb_matrix, max_len, rate=0.1):\n",
        "        super(Encoder, self).__init__()\n",
        "\n",
        "        self.d_model = d_model\n",
        "        self.num_layers = num_layers\n",
        "\n",
        "        self.embedding = tf.keras.layers.Embedding(input_dim = input_vocab_size, \n",
        "                      output_dim = d_model,\n",
        "                      input_length = max_len, # max_len of text sequence - 300\n",
        "                      weights=[emb_matrix],\n",
        "                      trainable=False) # static weights to be assigned from pretrained embedding\n",
        "        #self.embedding = tf.keras.layers.Embedding(input_vocab_size, d_model)\n",
        "        self.pos_encoding = positional_encoding(maximum_position_encoding, self.d_model)\n",
        "\n",
        "        self.enc_layers = [EncoderLayer(d_model, num_heads, dff, rate) for _ in range(num_layers)]\n",
        "\n",
        "        self.dropout = tf.keras.layers.Dropout(rate)\n",
        "        \n",
        "    def call(self, x, training, mask):\n",
        "        seq_len = tf.shape(x)[1]\n",
        "\n",
        "        x = self.embedding(x)\n",
        "        x *= tf.math.sqrt(tf.cast(self.d_model, tf.float32))\n",
        "        x += self.pos_encoding[:, :seq_len, :]\n",
        "\n",
        "        x = self.dropout(x, training=training)\n",
        "    \n",
        "        for i in range(self.num_layers):\n",
        "            x = self.enc_layers[i](x, training, mask)\n",
        "    \n",
        "        return x\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4N5LrNrvRexg"
      },
      "source": [
        "#### Decoder consisting of multiple DecoderLayer(s)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 27,
      "metadata": {
        "id": "UmeqkZrIRbSB"
      },
      "outputs": [],
      "source": [
        "class Decoder(tf.keras.layers.Layer):\n",
        "    def __init__(self, num_layers, d_model, num_heads, dff, target_vocab_size, maximum_position_encoding, emb_matrix, max_len, rate=0.1):\n",
        "        super(Decoder, self).__init__()\n",
        "\n",
        "        self.d_model = d_model\n",
        "        self.num_layers = num_layers\n",
        "\n",
        "        #self.embedding = tf.keras.layers.Embedding(target_vocab_size, d_model)\n",
        "        self.embedding = tf.keras.layers.Embedding(input_dim = target_vocab_size, \n",
        "                      output_dim = d_model,\n",
        "                      input_length = max_len, # max_len of summ sequence - 16\n",
        "                      weights=[emb_matrix],\n",
        "                      trainable=False)\n",
        "        self.pos_encoding = positional_encoding(maximum_position_encoding, d_model)\n",
        "\n",
        "        self.dec_layers = [DecoderLayer(d_model, num_heads, dff, rate) for _ in range(num_layers)]\n",
        "        self.dropout = tf.keras.layers.Dropout(rate)\n",
        "    \n",
        "    def call(self, x, enc_output, training, look_ahead_mask, padding_mask):\n",
        "        seq_len = tf.shape(x)[1]\n",
        "        attention_weights = {}\n",
        "\n",
        "        x = self.embedding(x)\n",
        "        x *= tf.math.sqrt(tf.cast(self.d_model, tf.float32))\n",
        "        x += self.pos_encoding[:, :seq_len, :]\n",
        "\n",
        "        x = self.dropout(x, training=training)\n",
        "\n",
        "        for i in range(self.num_layers):\n",
        "            x, block1, block2 = self.dec_layers[i](x, enc_output, training, look_ahead_mask, padding_mask)\n",
        "\n",
        "            attention_weights['decoder_layer{}_block1'.format(i+1)] = block1\n",
        "            attention_weights['decoder_layer{}_block2'.format(i+1)] = block2\n",
        "    \n",
        "        return x, attention_weights\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lbMNK_bzSHnh"
      },
      "source": [
        "#### Finally, the Transformer"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 28,
      "metadata": {
        "id": "FXHRG-o4R9Mc"
      },
      "outputs": [],
      "source": [
        "class Transformer(tf.keras.Model):\n",
        "    def __init__(self, num_layers, d_model, num_heads, dff, input_vocab_size, target_vocab_size, pe_input, pe_target, text_matrix, max_len_text, summ_matrix, max_len_summ, rate=0.1):\n",
        "        super(Transformer, self).__init__()\n",
        "\n",
        "        self.encoder = Encoder(num_layers, d_model, num_heads, dff, input_vocab_size, pe_input, text_matrix, max_len_text, rate)\n",
        "\n",
        "        self.decoder = Decoder(num_layers, d_model, num_heads, dff, target_vocab_size, pe_target, summ_matrix, max_len_summ, rate)\n",
        "\n",
        "        self.final_layer = tf.keras.layers.Dense(target_vocab_size)\n",
        "    \n",
        "    def call(self, inp, tar, training, enc_padding_mask, look_ahead_mask, dec_padding_mask):\n",
        "        enc_output = self.encoder(inp, training, enc_padding_mask)\n",
        "\n",
        "        dec_output, attention_weights = self.decoder(tar, enc_output, training, look_ahead_mask, dec_padding_mask)\n",
        "\n",
        "        final_output = self.final_layer(dec_output)\n",
        "\n",
        "        return final_output, attention_weights\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UndsMPZXTdSr"
      },
      "source": [
        "### Training"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 29,
      "metadata": {
        "id": "6zUhf3Wb2N6J"
      },
      "outputs": [],
      "source": [
        "with open(default_path+'text_matrix.pickle', 'rb') as handle:\n",
        "    text_matrix = pickle.load(handle)\n",
        "with open(default_path+'summ_matrix.pickle', 'rb') as handle:\n",
        "    summ_matrix = pickle.load(handle)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 31,
      "metadata": {
        "id": "lMTZJdIoSbuy"
      },
      "outputs": [],
      "source": [
        "# hyper-params\n",
        "num_layers = 4\n",
        "d_model = text_matrix.shape[1]\n",
        "dff = 512\n",
        "num_heads = 10\n",
        "EPOCHS = 20\n",
        "max_len_text = 300\n",
        "max_len_summ = 16"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 30,
      "metadata": {
        "id": "lpjRGbUkl-Q0"
      },
      "outputs": [],
      "source": [
        "encoder_vocab_size = 110757\n",
        "decoder_vocab_size = 25121"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uOGvkYDNTjIj"
      },
      "source": [
        "#### Adam optimizer with custom learning rate scheduling"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 33,
      "metadata": {
        "id": "tfiynCLlTL8C"
      },
      "outputs": [],
      "source": [
        "class CustomSchedule(tf.keras.optimizers.schedules.LearningRateSchedule):\n",
        "    def __init__(self, d_model, warmup_steps=4000):\n",
        "        super(CustomSchedule, self).__init__()\n",
        "\n",
        "        self.d_model = d_model\n",
        "        self.d_model = tf.cast(self.d_model, tf.float32)\n",
        "\n",
        "        self.warmup_steps = warmup_steps\n",
        "    \n",
        "    def __call__(self, step):\n",
        "        arg1 = tf.math.rsqrt(step)\n",
        "        arg2 = step * (self.warmup_steps ** -1.5)\n",
        "\n",
        "        return tf.math.rsqrt(self.d_model) * tf.math.minimum(arg1, arg2)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "DsVdrENTUERY"
      },
      "source": [
        "#### Custom Loss and Accuracy Functions"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 34,
      "metadata": {
        "id": "Ip1-943kTXXK"
      },
      "outputs": [],
      "source": [
        "learning_rate = CustomSchedule(d_model)\n",
        "\n",
        "optimizer = tf.keras.optimizers.Adam(learning_rate, beta_1=0.9, beta_2=0.98, epsilon=1e-9)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 35,
      "metadata": {
        "id": "ktKwyvKtTvF6"
      },
      "outputs": [],
      "source": [
        "loss_object = tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True, reduction='none')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 36,
      "metadata": {
        "id": "uW4LA_45T4Aa"
      },
      "outputs": [],
      "source": [
        "def loss_function(real, pred):\n",
        "    mask = tf.math.logical_not(tf.math.equal(real, 0))\n",
        "    loss_ = loss_object(real, pred)\n",
        "\n",
        "    mask = tf.cast(mask, dtype=loss_.dtype)\n",
        "    loss_ *= mask\n",
        "\n",
        "    return tf.reduce_sum(loss_)/tf.reduce_sum(mask)\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 37,
      "metadata": {
        "id": "Ze0u6xxXT7dI"
      },
      "outputs": [],
      "source": [
        "train_loss = tf.keras.metrics.Mean(name='train_loss')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 38,
      "metadata": {
        "id": "IjuEY22pv6Iq"
      },
      "outputs": [],
      "source": [
        "val_loss = tf.keras.metrics.Mean(name='val_loss')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9XvKy3v6ULnO"
      },
      "source": [
        "#### Transformer"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 39,
      "metadata": {
        "id": "d5-RcxqFUCuk"
      },
      "outputs": [],
      "source": [
        "transformer = Transformer(\n",
        "    num_layers, \n",
        "    d_model, \n",
        "    num_heads, \n",
        "    dff,\n",
        "    encoder_vocab_size, \n",
        "    decoder_vocab_size, \n",
        "    pe_input = encoder_vocab_size, \n",
        "    pe_target = decoder_vocab_size,\n",
        "    text_matrix = text_matrix, \n",
        "    max_len_text = max_len_text, \n",
        "    summ_matrix = summ_matrix, \n",
        "    max_len_summ = max_len_summ\n",
        ")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "f56BGiVXU_Dk"
      },
      "source": [
        "#### Masks"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 40,
      "metadata": {
        "id": "FZxHuyZxU5Pa"
      },
      "outputs": [],
      "source": [
        "def create_masks(inp, tar):\n",
        "    enc_padding_mask = create_padding_mask(inp)\n",
        "    dec_padding_mask = create_padding_mask(inp)\n",
        "\n",
        "    look_ahead_mask = create_look_ahead_mask(tf.shape(tar)[1])\n",
        "    dec_target_padding_mask = create_padding_mask(tar)\n",
        "    combined_mask = tf.maximum(dec_target_padding_mask, look_ahead_mask)\n",
        "  \n",
        "    return enc_padding_mask, combined_mask, dec_padding_mask\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SYIotvaBVI0d"
      },
      "source": [
        "#### Checkpoints"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 41,
      "metadata": {
        "id": "tOc1_3c-VGaL",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "f6c932ca-23e8-4844-d0d2-d1bd3454c119"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Latest checkpoint restored!!\n"
          ]
        }
      ],
      "source": [
        "checkpoint_path = default_path+\"checkpoints\"\n",
        "\n",
        "ckpt = tf.train.Checkpoint(transformer=transformer, optimizer=optimizer)\n",
        "\n",
        "ckpt_manager = tf.train.CheckpointManager(ckpt, checkpoint_path, max_to_keep=5)\n",
        "\n",
        "if ckpt_manager.latest_checkpoint:\n",
        "    ckpt.restore(ckpt_manager.latest_checkpoint)\n",
        "    print ('Latest checkpoint restored!!')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WfpI0gS4c06c"
      },
      "source": [
        "#### Training steps"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "xmVOMzkrczgl"
      },
      "outputs": [],
      "source": [
        "@tf.function\n",
        "def train_step(inp, tar):\n",
        "    tar_inp = tar[:, :-1]\n",
        "    tar_real = tar[:, 1:]\n",
        "\n",
        "    enc_padding_mask, combined_mask, dec_padding_mask = create_masks(inp, tar_inp)\n",
        "\n",
        "    with tf.GradientTape() as tape:\n",
        "        predictions, _ = transformer(\n",
        "            inp, tar_inp, \n",
        "            True, \n",
        "            enc_padding_mask, \n",
        "            combined_mask, \n",
        "            dec_padding_mask\n",
        "        )\n",
        "        loss = loss_function(tar_real, predictions)\n",
        "        \n",
        "    gradients = tape.gradient(loss, transformer.trainable_variables)    \n",
        "    optimizer.apply_gradients(zip(gradients, transformer.trainable_variables))\n",
        "\n",
        "    train_loss(loss)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lmGP2ar_HrfJ"
      },
      "source": [
        "### Test Validation Steps"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "EOL61k00HqpF"
      },
      "outputs": [],
      "source": [
        "@tf.function\n",
        "def val_step(inp, tar):\n",
        "    tar_inp = tar[:, :-1]\n",
        "    tar_real = tar[:, 1:]\n",
        "\n",
        "    enc_padding_mask, combined_mask, dec_padding_mask = create_masks(inp, tar_inp)\n",
        "\n",
        "    with tf.GradientTape() as tape:\n",
        "        predictions, _ = transformer(\n",
        "            inp, tar_inp, \n",
        "            False, \n",
        "            enc_padding_mask, \n",
        "            combined_mask, \n",
        "            dec_padding_mask\n",
        "        )\n",
        "        loss = loss_function(tar_real, predictions)\n",
        "    val_loss(loss)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "xORKpv69dSW5"
      },
      "outputs": [],
      "source": [
        "train_loss_values = []\n",
        "test_loss_values = []\n",
        "for epoch in range(EPOCHS):\n",
        "    start = time.time()\n",
        "\n",
        "    train_loss.reset_states()\n",
        "    val_loss.reset_states()\n",
        "    #training set - forward prop, loss fn, (gradient descent and optimization) backprop\n",
        "    for (batch, (inp, tar)) in enumerate(dataset):\n",
        "        train_step(inp, tar)\n",
        "    \n",
        "        # 85k samples\n",
        "        # we display 3 batch results -- 0th, middle and last one (approx)\n",
        "        # 85k / 64 ~ 1328; 1328 / 2 ~ 664\n",
        "        if batch % 2601 == 0:\n",
        "            print ('Epoch {} ========> Batch {} Loss {:.4f}'.format(epoch + 1, batch, train_loss.result().numpy()))\n",
        "      \n",
        "    if (epoch + 1) % 2 == 0:\n",
        "        ckpt_save_path = ckpt_manager.save()\n",
        "        print ('Saving checkpoint for epoch {} at {}'.format(epoch+1, ckpt_save_path))\n",
        "\n",
        "    train_loss_values.append(train_loss.result().numpy())\n",
        "    \n",
        "    #test set - forward pass, loss fn (no backward pass)\n",
        "    for (batch, (inp, tar)) in enumerate(test_dataset):\n",
        "        val_step(inp, tar)\n",
        "\n",
        "    test_loss_values.append(val_loss.result().numpy())\n",
        "    \n",
        "    print ('Epoch {}/{} ========> Training Loss {:.4f} Test/Validation Loss {:.4f}'.format(epoch + 1, EPOCHS, train_loss.result().numpy(), val_loss.result().numpy()))\n",
        "    \n",
        "    print ('Time taken for epoch {}: {:.2f} secs\\n'.format(epoch+1, time.time() - start))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "8yK0U1QTHOGw",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 404
        },
        "outputId": "87d416ea-c7c1-43de-dd13-66cd04c7b39b"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 720x432 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "no_of_epochs = [i for i in range(EPOCHS)]\n",
        "fig , ax = plt.subplots(1,1)\n",
        "fig.set_size_inches(10,6)\n",
        "ax.plot(no_of_epochs, test_loss_values, label = 'Testing Loss')\n",
        "ax.plot(no_of_epochs, train_loss_values, label = 'Training Loss')\n",
        "ax.set_title('Categorical Crossentropy Loss vs No. of Epochs')\n",
        "ax.legend()\n",
        "ax.set_xlabel(\"No. of Epochs\")\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YMbqGTixu1cl"
      },
      "source": [
        "#### Predicting one word at a time at the decoder and appending it to the output; then taking the complete sequence as an input to the decoder and repeating until maxlen or stop keyword appears"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 42,
      "metadata": {
        "id": "gHPLy2CmNSLc"
      },
      "outputs": [],
      "source": [
        "with open(default_path+'text_tokenizer.pickle', 'rb') as handle:\n",
        "    text_tokenizer = pickle.load(handle)\n",
        "with open(default_path+'summary_tokenizer.pickle', 'rb') as handle:\n",
        "    summary_tokenizer = pickle.load(handle)\n",
        "max_text_len = 300\n",
        "max_summary_len = 16"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 43,
      "metadata": {
        "id": "F5D5cv2Jd8-6"
      },
      "outputs": [],
      "source": [
        "def predict(input_document):\n",
        "    # #clean\n",
        "    #input_document = preprocess_tokenize(input_document)\n",
        "    input_document = \"sostok \"+input_document+\" eostok\"\n",
        "    #tokenize\n",
        "    input_document = text_tokenizer.texts_to_sequences([input_document])\n",
        "    #padding\n",
        "    input_document = tf.keras.preprocessing.sequence.pad_sequences(input_document, maxlen=max_text_len, padding='post', truncating='post')\n",
        "\n",
        "    encoder_input = tf.expand_dims(input_document[0], 0)\n",
        "\n",
        "    decoder_input = [summary_tokenizer.word_index[\"sostok\"]]\n",
        "    output = tf.expand_dims(decoder_input, 0)\n",
        "    \n",
        "    for i in range(max_summary_len):\n",
        "        enc_padding_mask, combined_mask, dec_padding_mask = create_masks(encoder_input, output)\n",
        "\n",
        "        predictions, attention_weights = transformer(\n",
        "            encoder_input, \n",
        "            output,\n",
        "            False,\n",
        "            enc_padding_mask,\n",
        "            combined_mask,\n",
        "            dec_padding_mask,\n",
        "        )\n",
        "\n",
        "        predictions = predictions[: ,-1:, :]\n",
        "        predicted_id = tf.cast(tf.argmax(predictions, axis=-1), tf.int32)\n",
        "\n",
        "        if predicted_id == summary_tokenizer.word_index[\"eostok\"]:\n",
        "            return tf.squeeze(output, axis=0), attention_weights\n",
        "\n",
        "        output = tf.concat([output, predicted_id], axis=-1)\n",
        "\n",
        "    return tf.squeeze(output, axis=0), attention_weights"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 44,
      "metadata": {
        "id": "UkpdiW6wnmiS"
      },
      "outputs": [],
      "source": [
        "def summarize(input_document):\n",
        "    # not considering attention weights for now, can be used to plot attention heatmaps in the future\n",
        "    summarized = predict(input_document=input_document)[0].numpy()\n",
        "    summarized = np.expand_dims(summarized[1:], 0)  # not printing <go> token\n",
        "    return summary_tokenizer.sequences_to_texts(summarized)[0]  # since there is just one translated document"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "summarize(\"पूर्व प्रधानमंत्री और कांग्रेस के दिग्गज नेता राजीव गांधी की आज 75वीं जयंती है. इस मौके पर कांग्रेस अध्यक्ष राहुल गांधी, यूपीए चेयरपर्सन सोनिया गांधी, प्रियंका गांधी, पूर्व प्रधानमंत्री मनमोहन सिंह समेत कई बड़े कांग्रेस नेताओं ने उन्हें श्रद्धांजलि अर्पित की. प्रधानमंत्री नरेंद्र मोदी ने भी ट्वीट कर उन्हें श्रद्धांजलि दी. बता दें कि राजीव गांधी का जन्म आज ही के दिन 1944 में हुआ था. 21 मई, 1991 को उनकी हत्या कर दी गई थी. राजीव गांधी 1984 से 1989 तक प्रधानमंत्री रहे. तमिलनाडु के श्रीपेरंबुदुर में 21 मई, 1991 को आम चुनाव के प्रचार के दौरान एलटीटीई के एक आत्मघाती हमलावर ने राजीव गांधी की हत्या कर दी थी. 1984 में इंदिरा गांधी की हत्या के बाद वह भारी बहुमत के साथ प्रधानमंत्री बने. राजीव गांधी की राजनीति में कोई रूचि नहीं थी और वो एक एयरलाइन पायलट की नौकरी करते थे और उसी में खुश थे. लेकिन आपातकाल के उपरान्त जब इंदिरा गांधी को सत्ता छोड़नी पड़ी थी. वहीं साल 1980 में छोटे भाई संजय गांधी की हवाई जहाज दुर्घटना में मृत्यु हो जाने के बाद माता इंदिरा का सहयोग देने के लिए उन्होंने राजनीति में प्रवेश कर लिया.\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "id": "EszFpEG98joV",
        "outputId": "749eb767-83b8-4357-ded9-cc076f0d520a"
      },
      "execution_count": 45,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'राजीव गांधी की पुण्यतिथि पर प्रधानमंत्री नरेंद्र मोदी ने दी श्रद्धांजलि'"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            }
          },
          "metadata": {},
          "execution_count": 45
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "#### Output"
      ],
      "metadata": {
        "id": "ghFFn6Ix9und"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "documents = test[\"article\"]\n",
        "summary = test[\"headline\"]"
      ],
      "metadata": {
        "id": "os6gTaJkDdup"
      },
      "execution_count": 46,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "samples = [17, 35, 69, 21, 26]\n",
        "docs = []\n",
        "summ = []\n",
        "pred = []\n",
        "for i in samples:\n",
        "  pred.append(summarize(documents[i]))\n",
        "  docs.append(' '.join(documents[i].split(' ')[:100]))\n",
        "  summ.append(summary[i])\n",
        "\n",
        "\n",
        "result_df = pd.DataFrame({'Input': docs, 'Reference Summary': summ, 'Model Output Summary': pred, 'Padding':pred })\n",
        "result_df[['Input', 'Reference Summary', 'Model Output Summary']] #17, 69, 21, 26, 35\n",
        "result_df.style.set_properties(subset=['Input', 'Padding'], **{'width': '400px'})"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 883
        },
        "id": "HbWHHKge9t8B",
        "outputId": "40cec07b-07da-476b-d3ed-4578ab36f971"
      },
      "execution_count": 50,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<pandas.io.formats.style.Styler at 0x7fc040d01850>"
            ],
            "text/html": [
              "<style type=\"text/css\">\n",
              "#T_c48e8_row0_col0, #T_c48e8_row0_col3, #T_c48e8_row1_col0, #T_c48e8_row1_col3, #T_c48e8_row2_col0, #T_c48e8_row2_col3, #T_c48e8_row3_col0, #T_c48e8_row3_col3, #T_c48e8_row4_col0, #T_c48e8_row4_col3 {\n",
              "  width: 400px;\n",
              "}\n",
              "</style>\n",
              "<table id=\"T_c48e8_\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr>\n",
              "      <th class=\"blank level0\" >&nbsp;</th>\n",
              "      <th class=\"col_heading level0 col0\" >Input</th>\n",
              "      <th class=\"col_heading level0 col1\" >Reference Summary</th>\n",
              "      <th class=\"col_heading level0 col2\" >Model Output Summary</th>\n",
              "      <th class=\"col_heading level0 col3\" >Padding</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th id=\"T_c48e8_level0_row0\" class=\"row_heading level0 row0\" >0</th>\n",
              "      <td id=\"T_c48e8_row0_col0\" class=\"data row0 col0\" >इराक की राजधानी बगदाद में सोमवार को दो आत्मघाती हमले हुए. अधिकारियों ने जानकारी देते हुए कहा कि इस दोहरे आत्मघाती हमले में 26 लोगों की मौत हो गई. आपको बता दें कि इराकी राजधानी में पिछले तीन दिनों में हुआ यह दूसरा हमला है.\n",
              "पूर्वी बगदाद के स्वास्थ्य प्रमुख डॉ अब्देल गनी अल-सादी ने ‘‘26 लोगों के मारे जाने और 90 के घायल होने” की जानकारी दी. सेना और पुलिस की संयुक्त ऑपरेशन्स कमान के प्रवक्ता जनरल साद मान ने बताया, “मध्य बगदाद के अल-तय्यारन चौराहे पर दो\n",
              "आत्मघाती हमलावरों\n",
              "ने खुद को धमाका कर उड़ा लिया.”\n",
              "इराक के बगदाद का तय्यारन चौराहा</td>\n",
              "      <td id=\"T_c48e8_row0_col1\" class=\"data row0 col1\" >इराक के बगदाद में दोहरा आत्मघाती हमला, 26 लोगों की मौत</td>\n",
              "      <td id=\"T_c48e8_row0_col2\" class=\"data row0 col2\" >इराक में आत्मघाती हमले में लोगों की मौत</td>\n",
              "      <td id=\"T_c48e8_row0_col3\" class=\"data row0 col3\" >इराक में आत्मघाती हमले में लोगों की मौत</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th id=\"T_c48e8_level0_row1\" class=\"row_heading level0 row1\" >1</th>\n",
              "      <td id=\"T_c48e8_row1_col0\" class=\"data row1 col0\" >दिल्ली से सटे नोएडा के ग्राम बंबावड़ में एक महिला की मौत के मामले में ससुराल पक्ष पर दहेज हत्या का मामला दर्ज किया गया है. महिला के पिता ने उसके पति समेत छह लोगों के खिलाफ थाना बादलपुर में मामला दर्ज कराया है.\n",
              "पुलिस उपाधीक्षक दादरी निशांक शर्मा ने बताया कि गांव सौराजपुर के रहने वाले विनोद कुमार ने थाना बादलपुर में रिपोर्ट दर्ज कराई कि उनकी बेटी मोनिका की शादी गांव बंबावड़ के रहने वाले अक्षय के साथ हुई थी.\n",
              "शादी के समय से ही उसके ससुराल वाले कार और नगदी की मांग कर रहे थे. मांग पूरी ना होने</td>\n",
              "      <td id=\"T_c48e8_row1_col1\" class=\"data row1 col1\" >ससुराल में महिला की संदिग्ध मौत, दहेज हत्या का केस दर्ज</td>\n",
              "      <td id=\"T_c48e8_row1_col2\" class=\"data row1 col2\" >दहेज के लिए महिला की हत्या</td>\n",
              "      <td id=\"T_c48e8_row1_col3\" class=\"data row1 col3\" >दहेज के लिए महिला की हत्या</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th id=\"T_c48e8_level0_row2\" class=\"row_heading level0 row2\" >2</th>\n",
              "      <td id=\"T_c48e8_row2_col0\" class=\"data row2 col0\" >पूर्व प्रधानमंत्री और\n",
              "कांग्रेस\n",
              "के दिग्गज नेता\n",
              "राजीव गांधी\n",
              "की आज 75वीं जयंती है. इस मौके पर कांग्रेस अध्यक्ष राहुल गांधी, यूपीए चेयरपर्सन सोनिया गांधी, प्रियंका गांधी, पूर्व प्रधानमंत्री मनमोहन सिंह समेत कई बड़े कांग्रेस नेताओं ने उन्हें श्रद्धांजलि अर्पित की. प्रधानमंत्री नरेंद्र मोदी ने भी ट्वीट कर उन्हें श्रद्धांजलि दी.\n",
              "बता दें कि राजीव गांधी का जन्म आज ही के दिन 1944 में हुआ था. 21 मई, 1991 को उनकी हत्या कर दी गई थी.\n",
              "Tributes to our former Prime Minister Shri Rajiv Gandhi Ji on his birth anniversary. We remember his efforts towards the nation.\n",
              "— Narendra Modi (@narendramodi)\n",
              "August 20, 2018\n",
              "Delhi: Sonia Gandhi, Rahul Gandhi,</td>\n",
              "      <td id=\"T_c48e8_row2_col1\" class=\"data row2 col1\" >राजीव गांधी की जयंती पर पीएम मोदी का ट्वीट, देश के लिए योगदान यादगार</td>\n",
              "      <td id=\"T_c48e8_row2_col2\" class=\"data row2 col2\" >राजीव गांधी की पुण्यतिथि पर प्रधानमंत्री नरेंद्र मोदी ने दी श्रद्धांजलि</td>\n",
              "      <td id=\"T_c48e8_row2_col3\" class=\"data row2 col3\" >राजीव गांधी की पुण्यतिथि पर प्रधानमंत्री नरेंद्र मोदी ने दी श्रद्धांजलि</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th id=\"T_c48e8_level0_row3\" class=\"row_heading level0 row3\" >3</th>\n",
              "      <td id=\"T_c48e8_row3_col0\" class=\"data row3 col0\" >सैमसंग इंडिया ने शुक्रवार को मिनिस्ट्री ऑफ माइक्रो, स्मॉल एंड मिडियम इंटरप्राइजेस (MSME) के साथ एमओयू पर हस्ताक्षर किए, जिसके तहत दो और टेक्निकल स्कूल खोलने और देश भर में चल रहे 10 वर्तमान स्कूलों को चलाने के लिए पार्टनरशिप को रिन्यू किया गया. कंपनी ने एक बयान में कहा कि एमओयू के मुताबिक वह बेंगलुरू और जमशेदपुर में दो नए MSME-Samsung टेक्निकल स्कूल खोलेगी.\n",
              "सूक्ष्म, लघु एवं मध्यम उद्यम मंत्री कलराज मिश्र ने एक बयान में कहा, 'हम उद्योग के लिए तैयार कार्यबल के निर्माण के लिए प्रतिबद्ध हैं और स्किल इंडिया कार्यक्रम के तहत हम\n",
              "युवाओं को इंडस्ट्री\n",
              "को ध्यान में</td>\n",
              "      <td id=\"T_c48e8_row3_col1\" class=\"data row3 col1\" >Samsung जमशेदपुर और बंगलुरू में खोलेगा टेक्निकल स्कूल</td>\n",
              "      <td id=\"T_c48e8_row3_col2\" class=\"data row3 col2\" >सैमसंग ने किया स्कूलों में फ्री पढ़ाई का आयोजन</td>\n",
              "      <td id=\"T_c48e8_row3_col3\" class=\"data row3 col3\" >सैमसंग ने किया स्कूलों में फ्री पढ़ाई का आयोजन</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th id=\"T_c48e8_level0_row4\" class=\"row_heading level0 row4\" >4</th>\n",
              "      <td id=\"T_c48e8_row4_col0\" class=\"data row4 col0\" >एक अध्ययन की रिपोर्ट में सामने आया है कि जो पुरुष हरी पत्तेदार सब्जियां, गहरे नारंगी और लाल रंग वाली सब्जियां, बेरीज (स्ट्रॉबेरी, ब्लैकबेरी, ब्लूबेरी) खाते हैं और संतरे का जूस पीते हैं, उनके बुढ़ापे में\n",
              "याददाश्त\n",
              "खोने का खतरा कम हो जाता है.\n",
              "शोध के निष्कर्षों से पता चलता है कि जो पुरुष बुढ़ापे से 20 साल पहले ज्यादा मात्रा में फल और सब्जियां खाते हैं, उनमें सोच और\n",
              "याददाश्त\n",
              "से जुड़ी परेशानियां कम होती हैं, चाहे वे बाद में अधिक मात्रा में फल और सब्जियां खाएं या नहीं.\n",
              "शोधकर्ताओं ने पाया कि जो पुरुष रोजाना संतरे का जूस पीते हैं, उनमें याददाश्त कमजोर होने</td>\n",
              "      <td id=\"T_c48e8_row4_col1\" class=\"data row4 col1\" >डाइट में शामिल करें ये चीजें, बुढ़ापे में भी तेज रहेगी याददाश्त</td>\n",
              "      <td id=\"T_c48e8_row4_col2\" class=\"data row4 col2\" >वजन घटाने के लिए डाइट में शामिल करें ये चीजें</td>\n",
              "      <td id=\"T_c48e8_row4_col3\" class=\"data row4 col3\" >वजन घटाने के लिए डाइट में शामिल करें ये चीजें</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n"
            ]
          },
          "metadata": {},
          "execution_count": 50
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rAaTWjd0oWzG"
      },
      "source": [
        "### Summary Evaluation - ROUGE"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 51,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Tyci_Q8cohOa",
        "outputId": "6bffe3bc-ded1-4bdf-9c70-827535651419"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Collecting rouge\n",
            "  Downloading rouge-1.0.1-py3-none-any.whl (13 kB)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.7/dist-packages (from rouge) (1.15.0)\n",
            "Installing collected packages: rouge\n",
            "Successfully installed rouge-1.0.1\n"
          ]
        }
      ],
      "source": [
        "!pip install rouge"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 52,
      "metadata": {
        "id": "gvtLSnJboZI8"
      },
      "outputs": [],
      "source": [
        "from rouge import Rouge"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 53,
      "metadata": {
        "id": "O9_KEUvGodH3"
      },
      "outputs": [],
      "source": [
        "rouge = Rouge()"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "summary = result_df['Model Output Summary'][0]\n",
        "reference = result_df['Reference Summary'][0]"
      ],
      "metadata": {
        "id": "GJlBSl5s-Jyy"
      },
      "execution_count": 55,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "rouge.get_scores(summary,reference)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mknCgyrN9sMK",
        "outputId": "52857ca0-e368-4890-8a8a-48512255eec9"
      },
      "execution_count": 56,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[{'rouge-1': {'f': 0.6666666619135801,\n",
              "   'p': 0.8571428571428571,\n",
              "   'r': 0.5454545454545454},\n",
              "  'rouge-2': {'f': 0.23529411280276827, 'p': 0.2857142857142857, 'r': 0.2},\n",
              "  'rouge-l': {'f': 0.6666666619135801,\n",
              "   'p': 0.8571428571428571,\n",
              "   'r': 0.5454545454545454}}]"
            ]
          },
          "metadata": {},
          "execution_count": 56
        }
      ]
    }
  ],
  "metadata": {
    "colab": {
      "collapsed_sections": [
        "oEnvEYxgxvOS",
        "7g7k7soMyBpQ",
        "Td7_wzZZyH0K",
        "9mBMOlMy6wR6",
        "xP44gjHSx8U_",
        "V6_ZUkajyZR2",
        "isN1CpAXLfsl",
        "24Pe01DMMWHc"
      ],
      "name": "Abstractive_Text_Summary_Hindi.ipynb",
      "provenance": []
    },
    "gpuClass": "standard",
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.8.3"
    },
    "accelerator": "GPU"
  },
  "nbformat": 4,
  "nbformat_minor": 0
}