LSTM_with_attention.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "LSTM_with_attention.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "machine_shape": "hm"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "7jCBe0GpsmoS",
        "colab_type": "code",
        "outputId": "6306ed71-b0e2-4e80-de90-f0d335e886e3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 121
        }
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Go to this URL in a browser: https://accounts.google.com/o/oauth2/auth?client_id=947318989803-6bn6qk8qdgf4n4g3pfee6491hc0brc4i.apps.googleusercontent.com&redirect_uri=urn%3aietf%3awg%3aoauth%3a2.0%3aoob&response_type=code&scope=email%20https%3a%2f%2fwww.googleapis.com%2fauth%2fdocs.test%20https%3a%2f%2fwww.googleapis.com%2fauth%2fdrive%20https%3a%2f%2fwww.googleapis.com%2fauth%2fdrive.photos.readonly%20https%3a%2f%2fwww.googleapis.com%2fauth%2fpeopleapi.readonly\n",
            "\n",
            "Enter your authorization code:\n",
            "··········\n",
            "Mounted at /content/drive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hEEkd5Qos20x",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "import os\n",
        "import re\n",
        "import warnings\n",
        "warnings.filterwarnings(\"ignore\")\n",
        "from sklearn import datasets, linear_model\n",
        "from sklearn.model_selection import train_test_split\n",
        "from matplotlib import pyplot as plt\n",
        "from bs4 import BeautifulSoup\n",
        "from nltk.corpus import stopwords"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Yb60OegVKFvA",
        "colab_type": "code",
        "outputId": "b2a4f648-0e44-44c0-bcc4-08bb9054f5e8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 67
        }
      },
      "source": [
        "import nltk\n",
        "nltk.download('stopwords')"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[nltk_data] Downloading package stopwords to /root/nltk_data...\n",
            "[nltk_data]   Unzipping corpora/stopwords.zip.\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "True"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 5
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "c61IdvLUtvKh",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# read data from the csv file (from the location it is stored)\n",
        "Data = pd.read_csv(r'/content/drive/My Drive/IDS_576_Project/wikihowAll.csv')\n",
        "Data = Data.astype(str)\n",
        "rows, columns = Data.shape"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JS1b4wJtw_xG",
        "colab_type": "code",
        "outputId": "b4f9c875-efe4-4b30-8891-08e10b4f17be",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 402
        }
      },
      "source": [
        "Data"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>headline</th>\n",
              "      <th>title</th>\n",
              "      <th>text</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>\\nKeep related supplies in the same area.,\\nMa...</td>\n",
              "      <td>How to Be an Organized Artist1</td>\n",
              "      <td>If you're a photographer, keep all the necess...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>\\nCreate a sketch in the NeoPopRealist manner ...</td>\n",
              "      <td>How to Create a Neopoprealist Art Work</td>\n",
              "      <td>See the image for how this drawing develops s...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>\\nGet a bachelor’s degree.,\\nEnroll in a studi...</td>\n",
              "      <td>How to Be a Visual Effects Artist1</td>\n",
              "      <td>It is possible to become a VFX artist without...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>\\nStart with some experience or interest in ar...</td>\n",
              "      <td>How to Become an Art Investor</td>\n",
              "      <td>The best art investors do their research on t...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>\\nKeep your reference materials, sketches, art...</td>\n",
              "      <td>How to Be an Organized Artist2</td>\n",
              "      <td>As you start planning for a project or work, ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>215360</th>\n",
              "      <td>\\nConsider changing the spelling of your name....</td>\n",
              "      <td>How to Pick a Stage Name3</td>\n",
              "      <td>If you have a name that you like, you might f...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>215361</th>\n",
              "      <td>\\nTry out your name.,\\nDon’t legally change yo...</td>\n",
              "      <td>How to Pick a Stage Name4</td>\n",
              "      <td>Your name might sound great to you when you s...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>215362</th>\n",
              "      <td>\\nUnderstand the process of relief printing.,\\...</td>\n",
              "      <td>How to Identify Prints1</td>\n",
              "      <td>Relief printing is the oldest and most tradit...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>215363</th>\n",
              "      <td>\\nUnderstand the process of intaglio printing....</td>\n",
              "      <td>How to Identify Prints2</td>\n",
              "      <td>Intaglio is Italian for \"incis­ing,\" and corr...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>215364</th>\n",
              "      <td>\\nUnderstand the different varieties of lithog...</td>\n",
              "      <td>How to Identify Prints3</td>\n",
              "      <td>Lithography is a big term often used to refer...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>215365 rows × 3 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                 headline  ...                                               text\n",
              "0       \\nKeep related supplies in the same area.,\\nMa...  ...   If you're a photographer, keep all the necess...\n",
              "1       \\nCreate a sketch in the NeoPopRealist manner ...  ...   See the image for how this drawing develops s...\n",
              "2       \\nGet a bachelor’s degree.,\\nEnroll in a studi...  ...   It is possible to become a VFX artist without...\n",
              "3       \\nStart with some experience or interest in ar...  ...   The best art investors do their research on t...\n",
              "4       \\nKeep your reference materials, sketches, art...  ...   As you start planning for a project or work, ...\n",
              "...                                                   ...  ...                                                ...\n",
              "215360  \\nConsider changing the spelling of your name....  ...   If you have a name that you like, you might f...\n",
              "215361  \\nTry out your name.,\\nDon’t legally change yo...  ...   Your name might sound great to you when you s...\n",
              "215362  \\nUnderstand the process of relief printing.,\\...  ...   Relief printing is the oldest and most tradit...\n",
              "215363  \\nUnderstand the process of intaglio printing....  ...   Intaglio is Italian for \"incis­ing,\" and corr...\n",
              "215364  \\nUnderstand the different varieties of lithog...  ...   Lithography is a big term often used to refer...\n",
              "\n",
              "[215365 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CpJ1lVDoE7TD",
        "colab_type": "text"
      },
      "source": [
        "**Cleaning** **the** **dataset**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2OXM62dBxipQ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "df=Data[Data['text'].isnull()==False]\n",
        "df=Data[Data['headline'].isnull()==False]\n",
        "df=df[df['text']!='nan']\n",
        "\n",
        "df.drop_duplicates(subset=['text'],inplace=True)  #dropping duplicates"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9-DOhC0UyBM4",
        "colab_type": "code",
        "outputId": "c51202a2-8fb2-4641-cb00-a4c86f2e47ce",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 402
        }
      },
      "source": [
        "df"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>headline</th>\n",
              "      <th>title</th>\n",
              "      <th>text</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>\\nKeep related supplies in the same area.,\\nMa...</td>\n",
              "      <td>How to Be an Organized Artist1</td>\n",
              "      <td>If you're a photographer, keep all the necess...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>\\nCreate a sketch in the NeoPopRealist manner ...</td>\n",
              "      <td>How to Create a Neopoprealist Art Work</td>\n",
              "      <td>See the image for how this drawing develops s...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>\\nGet a bachelor’s degree.,\\nEnroll in a studi...</td>\n",
              "      <td>How to Be a Visual Effects Artist1</td>\n",
              "      <td>It is possible to become a VFX artist without...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>\\nStart with some experience or interest in ar...</td>\n",
              "      <td>How to Become an Art Investor</td>\n",
              "      <td>The best art investors do their research on t...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>\\nKeep your reference materials, sketches, art...</td>\n",
              "      <td>How to Be an Organized Artist2</td>\n",
              "      <td>As you start planning for a project or work, ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>215360</th>\n",
              "      <td>\\nConsider changing the spelling of your name....</td>\n",
              "      <td>How to Pick a Stage Name3</td>\n",
              "      <td>If you have a name that you like, you might f...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>215361</th>\n",
              "      <td>\\nTry out your name.,\\nDon’t legally change yo...</td>\n",
              "      <td>How to Pick a Stage Name4</td>\n",
              "      <td>Your name might sound great to you when you s...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>215362</th>\n",
              "      <td>\\nUnderstand the process of relief printing.,\\...</td>\n",
              "      <td>How to Identify Prints1</td>\n",
              "      <td>Relief printing is the oldest and most tradit...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>215363</th>\n",
              "      <td>\\nUnderstand the process of intaglio printing....</td>\n",
              "      <td>How to Identify Prints2</td>\n",
              "      <td>Intaglio is Italian for \"incis­ing,\" and corr...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>215364</th>\n",
              "      <td>\\nUnderstand the different varieties of lithog...</td>\n",
              "      <td>How to Identify Prints3</td>\n",
              "      <td>Lithography is a big term often used to refer...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>209178 rows × 3 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                 headline  ...                                               text\n",
              "0       \\nKeep related supplies in the same area.,\\nMa...  ...   If you're a photographer, keep all the necess...\n",
              "1       \\nCreate a sketch in the NeoPopRealist manner ...  ...   See the image for how this drawing develops s...\n",
              "2       \\nGet a bachelor’s degree.,\\nEnroll in a studi...  ...   It is possible to become a VFX artist without...\n",
              "3       \\nStart with some experience or interest in ar...  ...   The best art investors do their research on t...\n",
              "4       \\nKeep your reference materials, sketches, art...  ...   As you start planning for a project or work, ...\n",
              "...                                                   ...  ...                                                ...\n",
              "215360  \\nConsider changing the spelling of your name....  ...   If you have a name that you like, you might f...\n",
              "215361  \\nTry out your name.,\\nDon’t legally change yo...  ...   Your name might sound great to you when you s...\n",
              "215362  \\nUnderstand the process of relief printing.,\\...  ...   Relief printing is the oldest and most tradit...\n",
              "215363  \\nUnderstand the process of intaglio printing....  ...   Intaglio is Italian for \"incis­ing,\" and corr...\n",
              "215364  \\nUnderstand the different varieties of lithog...  ...   Lithography is a big term often used to refer...\n",
              "\n",
              "[209178 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yW3xqCcEyhDb",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "stop_words = set(stopwords.words('english')) \n",
        "\n",
        "def text_cleaner(text,num):\n",
        "  str = text.lower()\n",
        "  str = BeautifulSoup(str, \"lxml\").text\n",
        "  str = re.sub(r'\\([^)]*\\)', '', str)\n",
        "  str = re.sub('\"','', str)\n",
        "  str = ' '.join([contraction_mapping[t] if t in contraction_mapping else t for t in str.split(\" \")])    \n",
        "  str = re.sub(r\"'s\\b\",\"\",str)\n",
        "  str = re.sub(\"[^a-zA-Z]\", \" \", str) \n",
        "  str = re.sub('[m]{2,}', 'mm', str)\n",
        "  if(num==0):\n",
        "    str = re.sub(r'\\.',' . ',str)\n",
        "  if(num==0):\n",
        "      tokens = [w for w in str.split() if not w in stop_words]\n",
        "      \n",
        "  else:\n",
        "      tokens=str.split()\n",
        "  long_words=[]\n",
        "  for i in tokens:\n",
        "      if len(i)>1:                                                 #removing short words\n",
        "          long_words.append(i)\n",
        "  return (\" \".join(long_words)).strip()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RUozoMqQ1Pes",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "contraction_mapping = {\"ain't\": \"is not\", \"aren't\": \"are not\",\"can't\": \"cannot\", \"'cause\": \"because\", \"could've\": \"could have\", \"couldn't\": \"could not\",\"didn't\": \"did not\", \"doesn't\": \"does not\", \"don't\": \"do not\", \"hadn't\": \"had not\", \"hasn't\": \"has not\", \"haven't\": \"have not\",\"he'd\": \"he would\",\"he'll\": \"he will\", \"he's\": \"he is\", \"how'd\": \"how did\", \"how'd'y\": \"how do you\", \"how'll\": \"how will\", \"how's\": \"how is\",\"I'd\": \"I would\", \"I'd've\": \"I would have\", \"I'll\": \"I will\", \"I'll've\": \"I will have\",\"I'm\": \"I am\", \"I've\": \"I have\", \"i'd\": \"i would\",\"i'd've\": \"i would have\", \"i'll\": \"i will\",  \"i'll've\": \"i will have\",\"i'm\": \"i am\", \"i've\": \"i have\", \"isn't\": \"is not\", \"it'd\": \"it would\",\"it'd've\": \"it would have\", \"it'll\": \"it will\", \"it'll've\": \"it will have\",\"it's\": \"it is\", \"let's\": \"let us\", \"ma'am\": \"madam\",\"mayn't\": \"may not\", \"might've\": \"might have\",\"mightn't\": \"might not\",\"mightn't've\": \"might not have\", \"must've\": \"must have\",\"mustn't\": \"must not\", \"mustn't've\": \"must not have\", \"needn't\": \"need not\", \"needn't've\": \"need not have\",\"o'clock\": \"of the clock\",\"oughtn't\": \"ought not\", \"oughtn't've\": \"ought not have\", \"shan't\": \"shall not\", \"sha'n't\": \"shall not\", \"shan't've\": \"shall not have\",\"she'd\": \"she would\", \"she'd've\": \"she would have\", \"she'll\": \"she will\", \"she'll've\": \"she will have\", \"she's\": \"she is\",\"should've\": \"should have\", \"shouldn't\": \"should not\", \"shouldn't've\": \"should not have\", \"so've\": \"so have\",\"so's\": \"so as\",\"this's\": \"this is\",\"that'd\": \"that would\", \"that'd've\": \"that would have\", \"that's\": \"that is\", \"there'd\": \"there would\",\"there'd've\": \"there would have\", \"there's\": \"there is\", \"here's\": \"here is\",\"they'd\": \"they would\", \"they'd've\": \"they would have\",\"they'll\": \"they will\", \"they'll've\": \"they will have\", \"they're\": \"they are\", \"they've\": \"they have\", \"to've\": \"to have\",\"wasn't\": \"was not\", \"we'd\": \"we would\", \"we'd've\": \"we would have\", \"we'll\": \"we will\", \"we'll've\": \"we will have\", \"we're\": \"we are\",\"we've\": \"we have\", \"weren't\": \"were not\", \"what'll\": \"what will\", \"what'll've\": \"what will have\", \"what're\": \"what are\",\"what's\": \"what is\", \"what've\": \"what have\", \"when's\": \"when is\", \"when've\": \"when have\", \"where'd\": \"where did\", \"where's\": \"where is\",\"where've\": \"where have\", \"who'll\": \"who will\", \"who'll've\": \"who will have\", \"who's\": \"who is\", \"who've\": \"who have\",\"why's\": \"why is\", \"why've\": \"why have\", \"will've\": \"will have\", \"won't\": \"will not\", \"won't've\": \"will not have\",\"would've\": \"would have\", \"wouldn't\": \"would not\", \"wouldn't've\": \"would not have\", \"y'all\": \"you all\",\"y'all'd\": \"you all would\",\"y'all'd've\": \"you all would have\",\"y'all're\": \"you all are\",\"y'all've\": \"you all have\",\"you'd\": \"you would\", \"you'd've\": \"you would have\", \"you'll\": \"you will\", \"you'll've\": \"you will have\",\"you're\": \"you are\", \"you've\": \"you have\"}"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1fmqZ6fb60np",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#call the function\n",
        "clean_text = []\n",
        "for t in df['text']:\n",
        "    clean_text.append(text_cleaner(t,0))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EVI9HLowAf62",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#call the function\n",
        "clean_summary = []\n",
        "for t in df['headline']:\n",
        "    clean_summary.append(text_cleaner(t,0))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "a50C4-IU9Gb7",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "df['text']=clean_text\n",
        "df['headline']=clean_summary\n",
        "\n",
        "df.replace('', np.nan, inplace=True)\n",
        "df.dropna(axis=0,inplace=True)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aC7uHaPoNcH5",
        "colab_type": "text"
      },
      "source": [
        "Updating the df"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qSCFwFWlHvbY",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7abXKKirMMvr",
        "colab_type": "code",
        "outputId": "f39e10b0-c72c-4d00-86a1-b2122900d057",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 402
        }
      },
      "source": [
        "df"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>headline</th>\n",
              "      <th>title</th>\n",
              "      <th>text</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>keep related supplies area make effort clean d...</td>\n",
              "      <td>How to Be an Organized Artist1</td>\n",
              "      <td>photographer keep necessary lens cords batteri...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>create sketch neopoprealist manner future mura...</td>\n",
              "      <td>How to Create a Neopoprealist Art Work</td>\n",
              "      <td>see image drawing develops step step however i...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>get bachelor degree enroll studio based progra...</td>\n",
              "      <td>How to Be a Visual Effects Artist1</td>\n",
              "      <td>possible become vfx artist without college deg...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>start experience interest art understand diffe...</td>\n",
              "      <td>How to Become an Art Investor</td>\n",
              "      <td>best art investors research pieces art buy som...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>keep reference materials sketches articles pho...</td>\n",
              "      <td>How to Be an Organized Artist2</td>\n",
              "      <td>start planning project work likely gathering s...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>215360</th>\n",
              "      <td>consider changing spelling name avoid symbols ...</td>\n",
              "      <td>How to Pick a Stage Name3</td>\n",
              "      <td>name like might fiddle spelling see alternate ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>215361</th>\n",
              "      <td>try name legally change name register stage na...</td>\n",
              "      <td>How to Pick a Stage Name4</td>\n",
              "      <td>name might sound great say loud bedroom find s...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>215362</th>\n",
              "      <td>understand process relief printing examine rim...</td>\n",
              "      <td>How to Identify Prints1</td>\n",
              "      <td>relief printing oldest traditional printing te...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>215363</th>\n",
              "      <td>understand process intaglio printing look plat...</td>\n",
              "      <td>How to Identify Prints2</td>\n",
              "      <td>intaglio italian incis ing correspondingly rev...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>215364</th>\n",
              "      <td>understand different varieties lithography mag...</td>\n",
              "      <td>How to Identify Prints3</td>\n",
              "      <td>lithography big term often used refer many dif...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>208675 rows × 3 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                 headline  ...                                               text\n",
              "0       keep related supplies area make effort clean d...  ...  photographer keep necessary lens cords batteri...\n",
              "1       create sketch neopoprealist manner future mura...  ...  see image drawing develops step step however i...\n",
              "2       get bachelor degree enroll studio based progra...  ...  possible become vfx artist without college deg...\n",
              "3       start experience interest art understand diffe...  ...  best art investors research pieces art buy som...\n",
              "4       keep reference materials sketches articles pho...  ...  start planning project work likely gathering s...\n",
              "...                                                   ...  ...                                                ...\n",
              "215360  consider changing spelling name avoid symbols ...  ...  name like might fiddle spelling see alternate ...\n",
              "215361  try name legally change name register stage na...  ...  name might sound great say loud bedroom find s...\n",
              "215362  understand process relief printing examine rim...  ...  relief printing oldest traditional printing te...\n",
              "215363  understand process intaglio printing look plat...  ...  intaglio italian incis ing correspondingly rev...\n",
              "215364  understand different varieties lithography mag...  ...  lithography big term often used refer many dif...\n",
              "\n",
              "[208675 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0fGQUB8kOyaO",
        "colab_type": "text"
      },
      "source": [
        "**Analyzing the sequence distribution**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "h306jOy2N6ui",
        "colab_type": "code",
        "outputId": "951268a3-70bd-4cd5-b1ba-466e7398bc5f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 281
        }
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "text_word_count = []\n",
        "headline_word_count = []\n",
        "\n",
        "# populate the lists with sentence lengths\n",
        "for i in df['text']:\n",
        "      temp=i.split()\n",
        "      text_word_count.append(len(temp))\n",
        "\n",
        "for j in df['headline']:\n",
        "  #print(j)\n",
        "  temp1=j.split()\n",
        "  headline_word_count.append(len(temp1))\n",
        "\n",
        "length_df = pd.DataFrame({'text':text_word_count, 'headline':headline_word_count})\n",
        "length_df.hist(bins = 30,range=[0,1000])\n",
        "plt.show()"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAY8AAAEICAYAAACnL3iHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAcgUlEQVR4nO3df7AV5Z3n8fdnIBrGJIIyuWWACaTC7izRiTGssmt26k5IEDU7mKqYwXUD/hgZKzqaLXYjZmaLVIwbnA1x1DXOYGBElxEdowUTcZAx3rLcGYhojPzS4YaQAIti5IchbjQ43/2jn6uH6znc+5zT955z7/m8qk6d7qef7n666cP3Pv08/bQiAjMzsxy/0ewCmJnZ0OPgYWZm2Rw8zMwsm4OHmZllc/AwM7NsDh5mZpbNwaPFSNop6VMDvI+JkkLSyDTfJemP0vTFkh4dyP2b2dDn4GFHiYgVETGj2eUw60tZf2gNxh9sw5GDh5mZZXPwaE2nS3pO0iFJ90l6N4Ckz0h6VtJBSf8o6Xd7VpC0QNKPJf1C0lZJn61YNkLSNyX9XNIO4PxaO5Z0iaQnK+ZD0pWStqf93i5JFcsvk7RN0gFJayV9sPSzYdaLpHuA3wb+TtJhSV+WNC39Lg5K+pGkzpT336drf0Ka/2i6Xn+n2naadlBDTUT400IfYCfwA+ADwEnANuBK4GPAPuAsYAQwN+U9Pq13YVrnN4A/BH4JnJKWXQk8D0xI23wcCGBkWt4F/FGavgR4sqI8AXwPGE3xI3sZmJmWzQK6gX8DjAT+DPjHZp9Df9rjk67/T6XpccArwHnpN/DpNP9bafmNwPeBUcAm4Opq2/Gn/x/XPFrTrRHxfyNiP/B3wOnAPOCvImJDRLwZEcuB14FpABHxt2mdf4mI+4DtwJlpe58H/iIidqVtfiOzPIsi4mBE/Iwi8Jye0q8EvhER2yLiCPA/KGpNrn3YYPvPwJqIWJN+A+uAjRTBBOCrwIkUf5jtAW5vSimHEQeP1vRixfRrwHuADwLzU5X8oKSDFDWJDwBImlNxS+sgcCowNm3jA8Cuim3+tITykMp0S8U+9wOi+CvQbDB9ELiw1+/jE8ApABHxa+Auit/F4khVDqvfyGYXwPptF3BjRNzYe0H6S/9OYDrwTxHxpqRnKf4jB9hLEWh6/HbJZVpR0vbMclQGgF3APRFxRbWMksYBC4G/BhZL+rcR8XqV7Vg/ueYxdNwJXCnpLBVOkHS+pPcCJ1D8AF4GkHQpxV9YPe4HrpE0XtIYYEFJZfpL4HpJH0n7PVHShSVt26wvLwEfStP/G/iPks5JHUTeLakzXfOiqHUsBS6n+GPqhhrbsX5y8BgiImIjcAXwv4ADFA3Vl6RlW4HFwD9R/BBOA/5Pxep3AmuBHwHPAA+WVKaHgJuAlZJeBTYD55axbbN++AbwZ+kW1R9SdOD4CsUfUbuA/0bxf9w1wPuB/55uV10KXCrpP/TejqT/OsjHMGTJt/7MzCyXax5mZpbNwcPMzLI5eJiZWTYHDzMzyzbsnvMYO3ZsTJw4seqyX/7yl5xwwgmDW6AW5PNQONZ5ePrpp38eEb81yEWqi6/5vvk8FMq85odd8Jg4cSIbN26suqyrq4vOzs7BLVAL8nkoHOs8SMp9Cr9pfM33zeehUOY179tWZmaWzcHDzMyyOXiYmVk2Bw8zM8vm4GFmZtkcPMzMLJuDh5mZZXPwMDOzbA4eZv0kaYKkxyVtlbRF0rUp/auS9qTXAD8r6byKda6X1C3pBUnnVKTPTGndkhZUpE+StCGl3yfpuME9SrP+GXZPmB/Lpj2HuGTBw2/N71x0fhNLY0PQEWB+RDyT3uD4tKR1adnNEfHNysySpgCzgY9QvEf+HyT9q7T4duDTwG7gKUmr00u9bkrbWinpLynefHdHvQX2NW8DxTUPs36KiL0R8Uya/gWwDRh3jFVmASsj4vWI+AnF2x/PTJ/uiNgREW8AK4FZ6XWpnwQeSOsvBy4YmKMxa0xb1TzMyiJpIvAxYANwNnC1pDnARorayQGKwLK+YrXdvB1sdvVKPws4GTgYEUeq5O+9/3nAPICOjg66urqqlrNjFMw/7chb87XyDXeHDx9u22OvVOZ5cPAwyyTpPcB3gS9FxKuS7gBuACJ9LwYuG8gyRMQSYAnA1KlTo9Zgd7etWMXiTW//zHdeXD3fcOeBEQtlngcHD7MMkt5FEThWRMSDABHxUsXyO4Hvpdk9wISK1cenNGqkvwKMljQy1T4q85u1FLd5mPVTapNYCmyLiG9VpJ9Ske2zwOY0vRqYLel4SZOAycAPgKeAyaln1XEUjeqrIyKAx4HPpfXnAqsG8pjM6uWah1n/nQ18Adgk6dmU9hXgIkmnU9y22gn8MUBEbJF0P7CVoqfWVRHxJoCkq4G1wAhgWURsSdu7Dlgp6evADymClVnLcfAw66eIeBJQlUVrjrHOjcCNVdLXVFsvInZQ9MYya2m+bWVmZtkcPMzMLJuDh5mZZXPwMDOzbA4eZmaWzcHDzMyy9Rk8JC2TtE/S5oq0/ynpeUnPSXpI0uiKZaUMQZ0erLovpW9IYwmZmVkL6E/N4y5gZq+0dcCpEfG7wD8D18M7hqCeCXxb0ghJIyiGoD4XmELxUNWUtK2eIag/DBygGIKa9H0gpd+c8pmZWQvoM3hExBPA/l5pj1aM/LmeYgweKHcI6llpnrR8espvZmZNVsYT5pcB96XpMoegHtezTkQckXQo5f957wJ4eOo8Hp664PNgVr+GgoekP6UYs2dFOcWpj4enzuPhqQs+D2b1qzt4SLoE+AwwPY0GCuUOQd2zrd2SRgInpvxmZtZkdXXVlTQT+DLwBxHxWsWiMoegXp3mScu/XxGkzMysifqseUi6F+gExkraDSyk6F11PLAutWGvj4grSx6Ceilwj6Ruigb72SUcr5mZlaDP4BERF1VJrvmOgbKGoI6IXwEX9lU+MzMbfH7C3MzMsjl4mJlZNgcPMzPL5tfQmtlRJi54+Kj5nYvOb1JJrJW55mFmZtkcPMzMLJtvW5m1kd63pMC3paw+rnmYmVk2Bw8zM8vm4GFmZtkcPMzMLJsbzM3aXLVGdLO+uOZhZmbZHDzMzCybg4eZmWVz8DAzs2xuMDezY/JT6VaNax5mGSRNkPS4pK2Stki6NqWfJGmdpO3pe0xKl6RbJXVLek7SGRXbmpvyb5c0tyL945I2pXVuVXrXs1krcfAwy3MEmB8RU4BpwFWSpgALgMciYjLwWJoHOBeYnD7zgDugCDbAQuAsitcwL+wJOCnPFRXrzRyE4zLL4uBhliEi9kbEM2n6F8A2YBwwC1iesi0HLkjTs4C7o7AeGC3pFOAcYF1E7I+IA8A6YGZa9r6IWB8RAdxdsS2zluE2D7M6SZoIfAzYAHRExN606EWgI02PA3ZVrLY7pR0rfXeV9N77nkdRk6Gjo4Ourq6qZewYBfNPO9L/g+qnWvtrVYcPHx5yZR4IZZ4HBw+zOkh6D/Bd4EsR8Wpls0REhKQYyP1HxBJgCcDUqVOjs7Ozar7bVqxi8abyf+Y7L66+v1bV1dVFrXPUTso8D75tZZZJ0rsoAseKiHgwJb+UbjmRvvel9D3AhIrVx6e0Y6WPr5Ju1lL6DB6SlknaJ2lzRdqA9yyptQ+zZkrX51JgW0R8q2LRaqDnup4LrKpIn5N+G9OAQ+n21lpghqQx6dqeAaxNy16VNC3ta07FtsxaRn9qHnfxzt4eg9GzpNY+zJrpbOALwCclPZs+5wGLgE9L2g58Ks0DrAF2AN3AncAXASJiP3AD8FT6fC2lkfJ8J63zY+CRwTgwsxx93gyNiCdSw2ClWUBnml4OdAHXUdGzBFgvqadnSSepZwmApJ6eJV2kniUpvadnySPH2IdZ00TEk0Ct5y6mV8kfwFU1trUMWFYlfSNwagPFNBtw9bakDUbPklr7eId6e560a+8L9zwp+DyY1a/hbhiD1LPkmPuot+fJUOsxUhb3PCn4PJjVr97eVoPRs6TWPszMrMnqDR6D0bOk1j7MzKzJ+rxtJeleiobrsZJ2U/SaWgTcL+ly4KfA51P2NcB5FL1EXgMuhaJniaSeniXwzp4ldwGjKBrKe3qW1NqHmZk1WX96W11UY9GA9iyJiFeq7cPMzJrPT5ibmVk2Bw8zM8vm4GFmZtkcPMzMLJuDh5mZZXPwMDOzbH4ZlJllm7jg4aPmdy46v0klsWZxzcPMzLI5eJiZWTYHDzMzy+bgYWZm2Rw8zMwsm4OHmZllc/AwM7NsDh5mZpbNwcPMzLI5eJiZWTYHDzMzy+bgYWZm2Rw8zMwsm4OHmZllc/AwM7NsDQUPSf9F0hZJmyXdK+ndkiZJ2iCpW9J9ko5LeY9P891p+cSK7Vyf0l+QdE5F+syU1i1pQSNlNTOz8tQdPCSNA64BpkbEqcAIYDZwE3BzRHwYOABcnla5HDiQ0m9O+ZA0Ja33EWAm8G1JIySNAG4HzgWmABelvGZm1mSN3rYaCYySNBL4TWAv8EnggbR8OXBBmp6V5knLp0tSSl8ZEa9HxE+AbuDM9OmOiB0R8QawMuU1M7Mmq/s1tBGxR9I3gZ8B/w94FHgaOBgRR1K23cC4ND0O2JXWPSLpEHBySl9fsenKdXb1Sj+rWlkkzQPmAXR0dNDV1VW1zB2jYP5pR96ar5VvuDt8+HDbHnslnwez+tUdPCSNoagJTAIOAn9Lcdtp0EXEEmAJwNSpU6Ozs7NqvttWrGLxprcPeefF1fMNd11dXdQ6R+2knvMgaRnwGWBful2LpK8CVwAvp2xfiYg1adn1FLds3wSuiYi1KX0mcAvF7d7vRMSilD6JopZ9MsUfY19INW+zltLIbatPAT+JiJcj4tfAg8DZwOh0GwtgPLAnTe8BJgCk5ScCr1Sm91qnVrpZM91F9T+Sbo6I09OnJ3DU055Xq83QrKU0Ejx+BkyT9Jup7WI6sBV4HPhcyjMXWJWmV6d50vLvR0Sk9NmpN9YkYDLwA+ApYHLqvXUcxY9wdQPlNWtYRDwB7O9n9qz2vPQ7qtVmaNZSGmnz2CDpAeAZ4AjwQ4pbRw8DKyV9PaUtTassBe6R1E3x45udtrNF0v0UgecIcFVEvAkg6WpgLUXVfllEbKm3vGYD7GpJc4CNwPyIOEB+e97J1G4zPEq97XwDpdXbjty+VSjzPNQdPAAiYiGwsFfyDoq/rHrn/RVwYY3t3AjcWCV9DbCmkTKaDYI7gBuASN+LgcsGcof1tvMNlFZvP3Q7X6HM8zDwV5XZMBcRL/VMS7oT+F6aPVa7XbX0V0hthqn24XY+a1kOHmYNknRKROxNs58FNqfp1cDfSPoW8AHebs8TqT2PIjjMBv5TRISknjbDlRzdZtjSJi54+B1pOxed34SS2GBx8DDLIOleoBMYK2k3xW3bTkmnU9y22gn8MdTdnncd1dsMzVqKg4dZhoi4qEpyzf/gc9vzIqJqm6FZq/GoumZmls3Bw8zMsjl4mJlZNgcPMzPL5uBhZmbZHDzMzCybg4eZmWVz8DAzs2wOHmZmls3Bw8zMsjl4mJlZNgcPMzPL5uBhZmbZHDzMzCybg4eZmWVz8DAzs2wOHmZmls3Bw8zMsjl4mJlZtoaCh6TRkh6Q9LykbZL+naSTJK2TtD19j0l5JelWSd2SnpN0RsV25qb82yXNrUj/uKRNaZ1bJamR8pqZWTkarXncAvx9RPwO8FFgG7AAeCwiJgOPpXmAc4HJ6TMPuANA0knAQuAs4ExgYU/ASXmuqFhvZoPlNTOzEtQdPCSdCPwesBQgIt6IiIPALGB5yrYcuCBNzwLujsJ6YLSkU4BzgHURsT8iDgDrgJlp2fsiYn1EBHB3xbbMzKyJRjaw7iTgZeCvJX0UeBq4FuiIiL0pz4tAR5oeB+yqWH93SjtW+u4q6e8gaR5FbYaOjg66urqqFrhjFMw/7chb87XyDXeHDx9u22Ov5PNgVr9GgsdI4AzgTyJig6RbePsWFQAREZKikQL2R0QsAZYATJ06NTo7O6vmu23FKhZvevuQd15cPd9w19XVRa1z1E58Hszq10ibx25gd0RsSPMPUASTl9ItJ9L3vrR8DzChYv3xKe1Y6eOrpJuZWZPVHTwi4kVgl6R/nZKmA1uB1UBPj6m5wKo0vRqYk3pdTQMOpdtba4EZksakhvIZwNq07FVJ01IvqzkV2zIzsyZq5LYVwJ8AKyQdB+wALqUISPdLuhz4KfD5lHcNcB7QDbyW8hIR+yXdADyV8n0tIvan6S8CdwGjgEfSx8zMmqyh4BERzwJTqyyaXiVvAFfV2M4yYFmV9I3AqY2U0czMytdozWNIm7jg4aPmdy46v0klMRt+/Psa3jw8iZmZZXPwMDOzbA4eZmaWzcHDLIOkZZL2SdpckebBQK3tOHiY5bmLdw7Q6cFAre04eJhliIgngP29kj0YqLWdtu6qa1aSITMYaDM1cxBKD4JZKPM8OHiYlajVBwNtpmYOROpBMAtlngfftjJrnAcDtbbj4GHWOA8Gam2nNeqzZkOEpHuBTmCspN0UvaYW4cFArc04eJhliIiLaizyYKDWVnzbyszMsjl4mJlZNgcPMzPL5uBhZmbZHDzMzCybg4eZmWVz8DAzs2x+zsPMBkXvd5qD32s+lLnmYWZm2RoOHpJGSPqhpO+l+UmSNqQ3od0n6biUfnya707LJ1Zs4/qU/oKkcyrSZ6a0bkkLeu/bzMyao4yax7XAtor5m4CbI+LDwAHg8pR+OXAgpd+c8iFpCjAb+AjFW9O+nQLSCOB2irexTQEuSnnNzKzJGgoeksYD5wPfSfMCPgk8kLL0fqtaz9vWHgCmp/yzgJUR8XpE/IRiELkz06c7InZExBvAypTXzMyarNGax18AXwb+Jc2fDByMiJ5Xl1W+Ce2tt6el5YdS/ty3rZmZWZPV3dtK0meAfRHxtKTO8opUV1lKeSVnu7ym0q/kLPg8mNWvka66ZwN/IOk84N3A+4BbgNGSRqbaReWb0HrenrZb0kjgROAVar9VjWOkH6WsV3I28zWZg8mv5Cz4PJjVr+7bVhFxfUSMj4iJFA3e34+Ii4HHgc+lbL3fqtbztrXPpfyR0men3liTgMnADyhelDM59d46Lu1jdb3lNTOz8gzEQ4LXASslfR34IbA0pS8F7pHUDeynCAZExBZJ9wNbgSPAVRHxJoCkqyle2TkCWBYRWwagvGZmlqmU4BERXUBXmt5B0VOqd55fARfWWP9G4MYq6WsoXuVpZmYtxE+Ym5lZNgcPMzPL5uBhZmbZHDzMzCybh2Q3s6bpPUy7h2gfOlzzMDOzbA4eZmaWzcHDzMyyOXiYmVk2Bw8zM8vm4GFmZtkcPMzMLJuDh5mZZXPwMDOzbA4eZiWRtFPSJknPStqY0k6StE7S9vQ9JqVL0q2SuiU9J+mMiu3MTfm3S5pba39mzeTgYVau34+I0yNiappfADwWEZOBx9I8wLkUb82cDMwD7oAi2AALgbMo3ouzsCfgmLUSBw+zgTULWJ6mlwMXVKTfHYX1wGhJpwDnAOsiYn9EHADWATMHu9BmffHAiGblCeBRSQH8VUQsAToiYm9a/iLQkabHAbsq1t2d0mqlH0XSPIoaCx0dHXR1dVUtUMcomH/akXqPZ9DdtmLVUfOnjTuxlO0ePny45jlqJ2WeBwcPs/J8IiL2SHo/sE7S85ULIyJSYGlYCkxLAKZOnRqdnZ1V8922YhWLNw3dn/nOiztL2U5XVxe1zlE7KfM8+LaVWUkiYk/63gc8RNFm8VK6HUX63pey7wEmVKw+PqXVSjdrKQ4eZiWQdIKk9/ZMAzOAzcBqoKfH1Fyg577MamBO6nU1DTiUbm+tBWZIGpMaymekNLOWMnTrs2atpQN4SBIUv6u/iYi/l/QUcL+ky4GfAp9P+dcA5wHdwGvApQARsV/SDcBTKd/XImL/4B2GWf84eJiVICJ2AB+tkv4KML1KegBX1djWMmBZ2WU0K1Pdt60kTZD0uKStkrZIujall/ZQlKSPp4euutO6auRgzcysHI20eRwB5kfEFGAacJWkKZT7UNQdwBUV67m/u5lZC6j7tlVq3Nubpn8haRtFf/RZQGfKthzoAq6j4qEoYL2knoeiOkkPRQFIWgfMlNQFvC89QIWkuykesHqk3jKb2dAyccHD70jbuej8JpTEeiulzUPSROBjwAbKeyhqXJrunV5t/6U8MNUuDxH5gamCz4NZ/RoOHpLeA3wX+FJEvFrZLFHmQ1HHUtYDU2U9kNTq/MBUwefBrH4NPech6V0UgWNFRDyYkst6KGpPmu6dbmZmTdZIbysBS4FtEfGtikWlPBSVlr0qaVra15yKbZmZWRM1ctvqbOALwCZJz6a0rwCLKO+hqC8CdwGjKBrK3VhuZtYCGult9SRQ67mLUh6KioiNwKn1ltHMhp/ePbDc+6o5PLaVmZllc/AwM7NsDh5mZpbNAyNW8NOsZkOPf7fN4ZqHmZllc83DzIad3rWRu2ae0KSSDF+ueZiZWTYHDzMzy+bbVmY27G3ac4hLKm5luUG9cQ4eZtZ23EOrcb5tZWZm2VzzMDPDY2blcvAwM6vCt7aOzcHDzKyfXDt5m9s8zMwsm2seZmZ1qnZrq7fhWjtx8DAzG0DD9VaXg4eZ2SAaLg3xDh5mZk3W1+2vVgwuDh59GK5VTjMbOlqxtuLgYWY2BDX7D1sHDzOzYaA/Pb/KfK+Jn/MwM7NsLR88JM2U9IKkbkkLml0es8Hg695aXUvftpI0Argd+DSwG3hK0uqI2NqsMrViw5UNL6143Zv11tLBAzgT6I6IHQCSVgKzgJb6EQ3FbnbW0obEdW/trdWDxzhgV8X8buCs3pkkzQPmpdnDkl6osb2xwM9LLWE/6KbB3mOfmnIeWtCxzsMHB7MgvfR53bf6Nd9qrvF5AOD3byrvmm/14NEvEbEEWNJXPkkbI2LqIBSppfk8FIbyefA1n8fnoVDmeWj1BvM9wISK+fEpzWw483VvLa/Vg8dTwGRJkyQdB8wGVje5TGYDzde9tbyWvm0VEUckXQ2sBUYAyyJiSwOb7LOa3yZ8HgoteR5Kvu5b8hibwOehUNp5UESUtS0zM2sTrX7byszMWpCDh5mZZWuL4NFuQz1I2ilpk6RnJW1MaSdJWidpe/oek9Il6dZ0bp6TdEZzS18/Scsk7ZO0uSIt+7glzU35t0ua24xjaZSv+fa45qGJ131EDOsPRYPjj4EPAccBPwKmNLtcA3zMO4GxvdL+HFiQphcAN6Xp84BHAAHTgA3NLn8Dx/17wBnA5nqPGzgJ2JG+x6TpMc0+tszz4Gu+jn/7ofpp1nXfDjWPt4Z6iIg3gJ6hHtrNLGB5ml4OXFCRfncU1gOjJZ3SjAI2KiKeAPb3Ss497nOAdRGxPyIOAOuAmQNf+lL5mi8M+2semnfdt0PwqDbUw7gmlWWwBPCopKfTMBYAHRGxN02/CHSk6eF+fnKPezicj+FwDLl8zR9twK/7ln7Ow+r2iYjYI+n9wDpJz1cujIiQ1HZ9tNv1uNuEr/kaBurY26Hm0XZDPUTEnvS9D3iI4jbGSz1V8/S9L2Uf7ucn97iHw/kYDseQxdf8Owz4dd8OwaOthnqQdIKk9/ZMAzOAzRTH3NODYi6wKk2vBuakXhjTgEMV1d3hIPe41wIzJI1JPVRmpLShxNd8e1/zMBjXfbN7CgxSb4TzgH+m6IHyp80uzwAf64coetf8CNjSc7zAycBjwHbgH4CTUrooXjz0Y2ATMLXZx9DAsd8L7AV+TXHP9vJ6jhu4DOhOn0ubfVx1ngtf821wzafjacp17+FJzMwsWzvctjIzs5I5eJiZWTYHDzMzy+bgYWZm2Rw8zMwsm4OHmZllc/AwM7Ns/x+bgJKS2ArGmgAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9ApVnUxJPXYT",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# From the graph\n",
        "# We can fix maximum length of text = 150 since most of the reviews have a length of 150 and maximum headline length of 50, since maximum headlines are of size 40\n",
        "\n",
        "max_len_text= 150\n",
        "max_len_headline=50"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YBPwN7lmGj6v",
        "colab_type": "code",
        "outputId": "efbfd575-9869-4ea1-ff68-da628b3a3404",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "cnt=0\n",
        "for i in df['text']:\n",
        "    if(len(i.split())<=150):\n",
        "        cnt=cnt+1\n",
        "print(cnt/len(df['text']))"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.4835318078351504\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dbjBRO7BIobl",
        "colab_type": "text"
      },
      "source": [
        "Selecting text and headlines below the maximum lengths "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DLEMfHWTIoHy",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "text1 =np.array(df['text'])\n",
        "headline1=np.array(df['headline'])\n",
        "\n",
        "short_text=[]\n",
        "short_summary=[]\n",
        "\n",
        "for i in range(len(text1)):\n",
        "    if(len(headline1[i].split())<=50 and len(text1[i].split())<=150):\n",
        "        short_text.append(text1[i])\n",
        "        short_summary.append(headline1[i])\n",
        "        \n",
        "df=pd.DataFrame({'text':short_text,'summary':short_summary})"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-2cdR4iXImwQ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Validating the lengths\n",
        "\n",
        "text1 =np.array(df['text'])\n",
        "headline1=np.array(df['summary'])\n",
        "\n",
        "\n",
        "for i in range(len(text1)):\n",
        "    if(len(headline1[i].split())>=150):\n",
        "      print(i)\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ytlW3ABFo2-M",
        "colab_type": "code",
        "outputId": "a83046a5-efe1-491b-f40d-c108f7a077c9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 70
        }
      },
      "source": [
        "print(df['text'][50],df['summary'][50],sep='\\n')"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "guy like every facebook status update instagram picture post better yet frequently comment posts sign wants interact may indicate likes look see comments people posts well comments frequently may reveal avid social media user however rarely likes comments posts sign might feelings publicly comments one photos posts respond eager start conversation way may reveal likes least enjoys messaging example may say great photo could respond saying vancouver last week beautiful city guy met starts liking commenting old photos sig interested means spent time looking back old photos probably wants know better enjoys looking pictures guy likes want connect follow variety different social media platforms example may add facebook snapchat start following twitter instagram adding number social media sites likely sign wants look posts photos selfies attempt get know better\n",
            "see interacts posts reply comments notice comments old pictures posts check see added multiple social media platforms\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dGoWWbOiXbxK",
        "colab_type": "text"
      },
      "source": [
        "**Splitting data into train, test  --  70 - 30**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "R8BhLDriRM_m",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "x_train,x_test,y_train,y_test=train_test_split(df['text'],df['summary'],test_size=0.3,random_state=0,shuffle=True) "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xiYMy1YSS1jT",
        "colab_type": "code",
        "outputId": "9aa75c74-31d3-4a13-ee8e-e003955ed415",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 50
        }
      },
      "source": [
        "print(len(x_train))\n",
        "print(len(x_test))"
      ],
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "59677\n",
            "25576\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KJZvHT5MchLm",
        "colab_type": "text"
      },
      "source": [
        "Language Translation"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QnQWC6Q-ZHtB",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#from torchtext.data import Field, BucketIterator"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ziCWLWO7Nf5s",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#pip install spacy\n",
        "#!python -m spacy download en"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rVcK0Er0NjA3",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from __future__ import unicode_literals, print_function, division\n",
        "from io import open\n",
        "import unicodedata\n",
        "import string\n",
        "import re\n",
        "import random\n",
        "\n",
        "import torch\n",
        "import torch.nn as nn\n",
        "from torch import optim\n",
        "import torch.nn.functional as F\n",
        "\n",
        "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rT0hCmdcZIby",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "SOS_token = 0\n",
        "EOS_token = 1\n",
        "\n",
        "class Lang:\n",
        "    def __init__(self, name):\n",
        "        self.name = name\n",
        "        self.word2index = {}\n",
        "        self.word2count = {}\n",
        "        self.index2word = {0: \"SOS\", 1: \"EOS\"}\n",
        "        self.n_words = 2  # Count SOS and EOS\n",
        "\n",
        "    def addSentence(self, sentence):\n",
        "        for word in sentence.split(' '):\n",
        "            self.addWord(word)\n",
        "\n",
        "    def addWord(self, word):\n",
        "        if word not in self.word2index:\n",
        "            self.word2index[word] = self.n_words\n",
        "            self.word2count[word] = 1\n",
        "            self.index2word[self.n_words] = word\n",
        "            self.n_words += 1\n",
        "        else:\n",
        "            self.word2count[word] += 1"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_2uvBCtacriQ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def readLangs(text, summary, reverse=False):\n",
        "    print(\"Reading lines...\")\n",
        "    \n",
        "    # Split every line into pairs and normalize\n",
        "    text=np.array(text)\n",
        "    summary=np.array(summary)\n",
        "    pairs = [[text[i],summary[i]] for i in range(len(text))]\n",
        "\n",
        "    # Reverse pairs, make Lang instances\n",
        "    if reverse:\n",
        "        pairs = [list(reversed(p)) for p in pairs]\n",
        "        input_lang = Lang(summary)\n",
        "        output_lang = Lang(text)\n",
        "    else:\n",
        "        input_lang = Lang(text)\n",
        "        output_lang = Lang(summary)\n",
        "\n",
        "    return input_lang, output_lang, pairs"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bi7I1f8VQxqy",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def prepareData(lang1, lang2, reverse=False):\n",
        "    input_lang, output_lang, pairs = readLangs(lang1, lang2, reverse)\n",
        "    print(\"Read %s sentence pairs\" % len(pairs))\n",
        "    print(\"Counting words...\")\n",
        "    for pair in pairs:\n",
        "        input_lang.addSentence(pair[0])\n",
        "        output_lang.addSentence(pair[1])\n",
        "    print(\"Counted words:\")\n",
        "    print(input_lang.name,\"--------------------\", input_lang.n_words)\n",
        "    #print(output_lang.name, output_lang.n_words)\n",
        "    return input_lang, output_lang, pairs"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lQEMoa3jQyDz",
        "colab_type": "code",
        "outputId": "ac0464f4-eb56-4086-b1bb-04578d5d2aa8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 222
        }
      },
      "source": [
        "input_lang, output_lang, pairs = prepareData(x_train, y_train , False)\n",
        "#print(random.choice(pairs))\n"
      ],
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Reading lines...\n",
            "Read 59677 sentence pairs\n",
            "Counting words...\n",
            "Counted words:\n",
            "['save hardware later use remove styrofoam lining necessary proper fit grind sand ensure perfect fit'\n",
            " 'use mug microwave safe suitable drinking handle good shape mug must large enough make brownie make sure mug also hard sturdy glass mug mix wet ingredients together metal spoon break egg yolk pushing breaking yolk gently slowly add dry ingredient mixture stir well combined beating egg yolk adding wet ingredients mixture make brownie light fluffy place mug microwave operate minute seconds bakes puff rise fully cooked remove microwave oven mitts place safe place gets burned mistakenly cool minutes avoid spills brownie cooking place small dish bowl beneath mug helps catch brownie mix leaking microwave floor test brownie ready poke toothpick middle comes dry ready scoop vanilla ice cream sprinkle chocolate sprinkles add chopped walnuts spread marshmallow fluff creative many topping decisions choose go choosing serve enjoy eating brownie either eat brownie mug remove enjoy sweet delicious brownie'\n",
            " 'consolidator make sure travel agent use member professional organization like universal federation travel agents association national association career travel agents also look reviews travel agents consumer affairs website better business bureau whether booking business flight group travel others best meet travel agent first time person gives chance meet agent discuss specific travel needs longer conversation phone email allow make sure ask travel agents plenty questions clarify details trip know fees top price ticket help hotline call travel plans disrupted also ask travel agent best way communicate prefer phone email particularly important need contact travel met travel agent create schedule due dates phases trip includes knowing need traveler information agent need copies passports ids payments final payments due'\n",
            " ...\n",
            " 'app gray gear usually located home screen top half page near top page phone ringer sound adjusting able hear volume level setting want change alarm volume future slide change buttons setting position setting underneath volume slider enable adjust alarm volume using iphone volume buttons phone unlocked'\n",
            " 'make sure latest version downloaded google play store options bottom screen based availability vehicles area uberx cars seat passengers least expensive uber service get picked car like prius honda accord similar uberxl cars seat passengers cost uberx rides get picked either suv minivan uberselect entry level luxury car service uberselect cars seat get picked bmw mercedes audi something similar leather interior uberpool allows share ride split cost trip another rider headed direction uberblack uber luxury service expensive expect picked professional driver typically driving black suv luxury sedan box magnifying glass left side pickup location automatically set based gps need change tap pickup location box green dot left side type pickup address tap pickup address update pickup location request sent ride approved information driver appear receive message driver minute away'\n",
            " 'matter kind long yummy repeat necessary confectioner sugar'] -------------------- 59885\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KzJDRhw1Q0uT",
        "colab_type": "code",
        "outputId": "407c901e-ccbf-48f1-b010-80d9f003f92f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 70
        }
      },
      "source": [
        "pairs[1]"
      ],
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['use mug microwave safe suitable drinking handle good shape mug must large enough make brownie make sure mug also hard sturdy glass mug mix wet ingredients together metal spoon break egg yolk pushing breaking yolk gently slowly add dry ingredient mixture stir well combined beating egg yolk adding wet ingredients mixture make brownie light fluffy place mug microwave operate minute seconds bakes puff rise fully cooked remove microwave oven mitts place safe place gets burned mistakenly cool minutes avoid spills brownie cooking place small dish bowl beneath mug helps catch brownie mix leaking microwave floor test brownie ready poke toothpick middle comes dry ready scoop vanilla ice cream sprinkle chocolate sprinkles add chopped walnuts spread marshmallow fluff creative many topping decisions choose go choosing serve enjoy eating brownie either eat brownie mug remove enjoy sweet delicious brownie',\n",
              " 'find good mug cooking mix ingredients heat microwave minute seconds add desired toppings serve']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 31
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "78iDN5buRyU2",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#---- Deep Model"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PiIp2TKggfCT",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "SOS_token = 0\n",
        "EOS_token = 1"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QzhoiaCFTXrP",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def indexesFromSentence(lang, sentence):\n",
        "    return [lang.word2index[word] for word in sentence.split(' ')]\n",
        "\n",
        "def tensorFromSentence(lang, sentence):\n",
        "    indexes = indexesFromSentence(lang, sentence)\n",
        "    indexes.append(EOS_token)\n",
        "    return torch.tensor(indexes, dtype=torch.long, device=device).view(-1, 1)\n",
        "\n",
        "def tensorsFromPair(pair):\n",
        "    input_tensor = tensorFromSentence(input_lang, pair[0])\n",
        "    target_tensor = tensorFromSentence(output_lang, pair[1])\n",
        "    return (input_tensor, target_tensor)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6wMWZ-Vmfvpj",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Z-i6UICKO20T",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from __future__ import unicode_literals, print_function, division\n",
        "from io import open\n",
        "import unicodedata\n",
        "import string\n",
        "import re\n",
        "import random\n",
        "\n",
        "import torch\n",
        "import torch.nn as nn\n",
        "from torch import optim\n",
        "import torch.nn.functional as F\n",
        "\n",
        "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DfJ_bgxzVmiL",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class EncoderRNN(nn.Module):\n",
        "    def __init__(self, input_size, hidden_size):\n",
        "        super(EncoderRNN, self).__init__()\n",
        "        self.hidden_size = hidden_size\n",
        "\n",
        "        self.embedding = nn.Embedding(input_size, hidden_size)\n",
        "        self.LSTM = nn.LSTM(hidden_size, hidden_size)\n",
        "\n",
        "    def forward(self, input, hidden):\n",
        "        embedded = self.embedding(input).view(1, 1, -1)\n",
        "        output = embedded\n",
        "        output, hidden = self.LSTM(output, hidden)\n",
        "        return output, hidden\n",
        "\n",
        "    def initHidden(self):\n",
        "        return (torch.zeros(1, 1, self.hidden_size, device=device),torch.zeros(1, 1, self.hidden_size, device=device)) "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5I2dCJd8TbMJ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class DecoderRNN(nn.Module):\n",
        "    def __init__(self, hidden_size, output_size):\n",
        "        super(DecoderRNN, self).__init__()\n",
        "        self.hidden_size = hidden_size\n",
        "\n",
        "        self.embedding = nn.Embedding(output_size, hidden_size)\n",
        "        self.LSTM = nn.LSTM(hidden_size, hidden_size)\n",
        "        self.out = nn.Linear(hidden_size, output_size)\n",
        "        self.softmax = nn.LogSoftmax(dim=2)\n",
        "\n",
        "    def forward(self, input, hidden):\n",
        "        output = self.embedding(input).view(1, 1, -1)\n",
        "        output = F.relu(output)\n",
        "        output, hidden = self.LSTM(output, hidden)\n",
        "        output = self.softmax(self.out(output[0]))\n",
        "        return output, hidden\n",
        "\n",
        "    def initHidden(self):\n",
        "        return torch.zeros(1, 1, self.hidden_size, device=device)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "payXbX3mnHGh",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "MAX_LENGTH=150"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XqVupGeYm7Rb",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class AttnDecoderRNN(nn.Module):\n",
        "    def __init__(self, hidden_size, output_size, dropout_p=0.1, max_length=MAX_LENGTH):\n",
        "        super(AttnDecoderRNN, self).__init__()\n",
        "        self.hidden_size = hidden_size\n",
        "        self.output_size = output_size\n",
        "        self.dropout_p = dropout_p\n",
        "        self.max_length = max_length\n",
        "\n",
        "        self.embedding = nn.Embedding(self.output_size, self.hidden_size)\n",
        "        self.attn = nn.Linear(self.hidden_size*2 , self.max_length)\n",
        "        self.attn_combine = nn.Linear(self.hidden_size*2 , self.hidden_size)\n",
        "        self.dropout = nn.Dropout(self.dropout_p)\n",
        "        self.LSTM = nn.LSTM(self.hidden_size, self.hidden_size)\n",
        "        self.out = nn.Linear(self.hidden_size, self.output_size)\n",
        "\n",
        "        #print('Decoder --- atndecoder')\n",
        "\n",
        "    def forward(self, input, hidden, encoder_outputs):\n",
        "        #print(\"inside forward decoder\")\n",
        "        embedded = self.embedding(input).view(1, 1, -1)\n",
        "        embedded = self.dropout(embedded)\n",
        "\n",
        "        #print(\"embedded size\", embedded.size())\n",
        "        #print(embedded[0].size()) #1,300\n",
        "        #print(hidden[0].size()) # 1, 1, 300\n",
        "        #print(\"diff or not\")\n",
        "\n",
        "        #temp = torch.cat((embedded[0], hidden[0]), 1)\n",
        "        \n",
        "        #print(temp)\n",
        "        #print(temp.size())\n",
        "\n",
        "        attn_weights = F.softmax(self.attn(torch.cat((embedded[0], hidden[0][0]), 1)), dim=1)\n",
        "        #print(\"after getting attn weights softmax\")\n",
        "        #print(attn_weights.size())\n",
        "        attn_applied = torch.bmm(attn_weights.unsqueeze(0),\n",
        "                                 encoder_outputs.unsqueeze(0))\n",
        "\n",
        "        output = torch.cat((embedded[0], attn_applied[0]), 1)\n",
        "        output = self.attn_combine(output).unsqueeze(0)\n",
        "\n",
        "        output = F.relu(output)\n",
        "        output, hidden = self.LSTM(output, hidden)\n",
        "\n",
        "        output = F.log_softmax(self.out(output[0]), dim=1)\n",
        "        return output, hidden, attn_weights\n",
        "\n",
        "    def initHidden(self):\n",
        "        return torch.zeros(1, 1, self.hidden_size, device=device)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5F8X6eQoRXpX",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0GCmpRFEO65P",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#teacher_forcing_ratio = 0.5\n",
        "def train(input_tensor, target_tensor, encoder, decoder, encoder_optimizer, decoder_optimizer, criterion, max_length=MAX_LENGTH):\n",
        "    encoder_hidden = encoder.initHidden()\n",
        "\n",
        "    encoder_optimizer.zero_grad()\n",
        "    decoder_optimizer.zero_grad()\n",
        "\n",
        "    input_length = input_tensor.size(0)\n",
        "    target_length = target_tensor.size(0)\n",
        "\n",
        "    encoder_outputs = torch.zeros(max_length, encoder.hidden_size, device=device)\n",
        "\n",
        "    loss = 0\n",
        "    #print('bbbbbbb-->>> input length', input_length)\n",
        "\n",
        "    for ei in range(input_length):\n",
        "        encoder_output, encoder_hidden = encoder(input_tensor[ei], encoder_hidden)\n",
        "        #print(\"priting before error\")\n",
        "        #print(encoder_output.size())\n",
        "        #print(encoder_outputs.size())\n",
        "        temp = encoder_output[0, 0]\n",
        "        #print(temp)\n",
        "        encoder_outputs[ei] = temp\n",
        "\n",
        "    decoder_input = torch.tensor([[SOS_token]], device=device)\n",
        "\n",
        "    decoder_hidden = encoder_hidden\n",
        "\n",
        "    #use_teacher_forcing = True if random.random() < teacher_forcing_ratio else False\n",
        "    #Without teacher forcing: use its own predictions as the next input\n",
        "\n",
        "    #print('aaaaa-->>>')\n",
        "\n",
        "    for di in range(target_length):\n",
        "      decoder_output, decoder_hidden, decoder_attention = decoder(\n",
        "      decoder_input, decoder_hidden, encoder_outputs)\n",
        "\n",
        "\n",
        "      topv, topi = decoder_output.topk(1)\n",
        "      \n",
        "      decoder_input = topi.squeeze().detach()  # detach from history as input\n",
        "      \n",
        "      loss += criterion(decoder_output, target_tensor[di])\n",
        "      \n",
        "      if decoder_input.item() == EOS_token:\n",
        "             break\n",
        "\n",
        "    loss.backward()\n",
        "\n",
        "    encoder_optimizer.step()\n",
        "    decoder_optimizer.step()\n",
        "\n",
        "    return loss.item() / target_length"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VjMd-P0CRVqb",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import time\n",
        "import math\n",
        "\n",
        "\n",
        "def asMinutes(s):\n",
        "    m = math.floor(s / 60)\n",
        "    s -= m * 60\n",
        "    return '%dm %ds' % (m, s)\n",
        "\n",
        "\n",
        "def timeSince(since, percent):\n",
        "    now = time.time()\n",
        "    s = now - since\n",
        "    es = s / (percent)\n",
        "    rs = es - s\n",
        "    return '%s (- %s)' % (asMinutes(s), asMinutes(rs))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gs3wu28ydLtt",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "plt.switch_backend('agg')\n",
        "import matplotlib.ticker as ticker\n",
        "import numpy as np\n",
        "\n",
        "\n",
        "def showPlot(points):\n",
        "    plt.figure()\n",
        "    fig, ax = plt.subplots()\n",
        "    # this locator puts ticks at regular intervals\n",
        "    loc = ticker.MultipleLocator(base=0.2)\n",
        "    ax.yaxis.set_major_locator(loc)\n",
        "    plt.plot(points)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LFkw0w3NdNpR",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def trainIters(encoder, decoder, n_iters, print_every=1000, plot_every=100, learning_rate=0.01):\n",
        "    print(\"Training....\")\n",
        "    start = time.time()\n",
        "    plot_losses = []\n",
        "    print_loss_total = 0  # Reset every print_every\n",
        "    plot_loss_total = 0  # Reset every plot_every\n",
        "\n",
        "    encoder_optimizer = optim.SGD(encoder.parameters(), lr=learning_rate)\n",
        "    decoder_optimizer = optim.SGD(decoder.parameters(), lr=learning_rate)\n",
        "    training_pairs = [tensorsFromPair(random.choice(pairs))\n",
        "                      for i in range(n_iters)]\n",
        "    criterion = nn.NLLLoss()\n",
        "\n",
        "    for iter in range(1, n_iters + 1):\n",
        "        if iter% 1000 == 0:\n",
        "            print(iter,\"/\",n_iters + 1)\n",
        "        training_pair = training_pairs[iter - 1]\n",
        "        input_tensor = training_pair[0]\n",
        "        target_tensor = training_pair[1]\n",
        "\n",
        "        input_length = input_tensor.size(0)\n",
        "        if(input_length > 150):\n",
        "          #print(input_length)\n",
        "          continue\n",
        "      \n",
        "        loss = train(input_tensor, target_tensor, encoder,decoder, encoder_optimizer, decoder_optimizer, criterion)\n",
        "        print_loss_total += loss\n",
        "        plot_loss_total += loss\n",
        "\n",
        "        if iter % print_every == 0:\n",
        "            print_loss_avg = print_loss_total / print_every\n",
        "            print_loss_total = 0\n",
        "            print('%s (%d %d%%) %.4f' % (timeSince(start, iter / n_iters),\n",
        "                                         iter, iter / n_iters * 100, print_loss_avg))\n",
        "\n",
        "        if iter % plot_every == 0:\n",
        "            plot_loss_avg = plot_loss_total / plot_every\n",
        "            plot_losses.append(plot_loss_avg)\n",
        "            plot_loss_total = 0\n",
        "\n",
        "    showPlot(plot_losses)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "R7i2tbx2dTNv",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def evaluate(encoder, decoder, sentence, max_length=MAX_LENGTH):\n",
        "    with torch.no_grad():\n",
        "        input_tensor = tensorFromSentence(input_lang, sentence)\n",
        "        input_length = input_tensor.size()[0]\n",
        "        encoder_hidden = encoder.initHidden()\n",
        "\n",
        "        encoder_outputs = torch.zeros(max_length, encoder.hidden_size, device=device)\n",
        "\n",
        "        for ei in range(input_length):\n",
        "            encoder_output, encoder_hidden = encoder(input_tensor[ei],encoder_hidden)\n",
        "            encoder_outputs[ei] += encoder_output[0, 0]\n",
        "\n",
        "        decoder_input = torch.tensor([[SOS_token]], device=device)  # SOS\n",
        "\n",
        "        decoder_hidden = encoder_hidden\n",
        "\n",
        "        decoded_words = []\n",
        "        decoder_attentions = torch.zeros(max_length, max_length)\n",
        "\n",
        "        #print('cccccc->>>>')\n",
        "\n",
        "        for di in range(max_length):\n",
        "            decoder_output, decoder_hidden, decoder_attention = decoder(\n",
        "                decoder_input, decoder_hidden, encoder_outputs)\n",
        "            decoder_attentions[di] = decoder_attention.data\n",
        "            topv, topi = decoder_output.data.topk(1)\n",
        "            if topi.item() == EOS_token:\n",
        "                decoded_words.append('<EOS>')\n",
        "                break\n",
        "            else:\n",
        "                decoded_words.append(output_lang.index2word[topi.item()])\n",
        "\n",
        "            decoder_input = topi.squeeze().detach()\n",
        "\n",
        "        return decoded_words, decoder_attentions[:di + 1]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uO94wpqTe7PZ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def evaluateRandomly(encoder, decoder, n=5):\n",
        "    text=list()\n",
        "    headline=list()\n",
        "    pred_headline=list()    \n",
        "    for i in range(n):\n",
        "        pair = random.choice(pairs)\n",
        "        \n",
        "        if(len(pair[0].split())>=150):\n",
        "          continue\n",
        "        else:\n",
        "          if(i%1000==0):\n",
        "            print(i*100/n,\"% complete\")\n",
        "          \n",
        "          #print('>', pair[0])\n",
        "          text.append(pair[0])\n",
        "          #print('=', pair[1])\n",
        "          headline.append(pair[1])\n",
        "          output_words, attentions = evaluate(encoder, decoder, pair[0])\n",
        "          output_sentence = ' '.join(output_words)\n",
        "          pred_headline.append(output_sentence)\n",
        "          #print('<', output_sentence)\n",
        "          #print('')\n",
        "    return(text,headline,pred_headline)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KuKv9Q9xe7-Y",
        "colab_type": "code",
        "outputId": "82b51a52-03a6-4eba-a7e9-c9994e01dfa8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "hidden_size = 300\n",
        "encoder1 = EncoderRNN(input_lang.n_words, hidden_size).to(device)\n",
        "attn_decoder1 = AttnDecoderRNN(hidden_size, output_lang.n_words, dropout_p=0.1).to(device)\n",
        "\n",
        "trainIters(encoder1, attn_decoder1, 150000, print_every=1000)"
      ],
      "execution_count": 46,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Training....\n",
            "1000 / 150001\n",
            "1m 35s (- 236m 10s) (1000 0%) 2.2282\n",
            "2000 / 150001\n",
            "2m 59s (- 221m 26s) (2000 1%) 2.3739\n",
            "3000 / 150001\n",
            "4m 26s (- 217m 23s) (3000 2%) 2.7115\n",
            "4000 / 150001\n",
            "5m 54s (- 215m 51s) (4000 2%) 3.0980\n",
            "5000 / 150001\n",
            "7m 23s (- 214m 10s) (5000 3%) 3.0176\n",
            "6000 / 150001\n",
            "8m 53s (- 213m 23s) (6000 4%) 3.3115\n",
            "7000 / 150001\n",
            "10m 24s (- 212m 40s) (7000 4%) 3.3324\n",
            "8000 / 150001\n",
            "11m 48s (- 209m 43s) (8000 5%) 3.0211\n",
            "9000 / 150001\n",
            "13m 16s (- 207m 55s) (9000 6%) 3.0845\n",
            "10000 / 150001\n",
            "14m 43s (- 206m 5s) (10000 6%) 2.9608\n",
            "11000 / 150001\n",
            "16m 10s (- 204m 25s) (11000 7%) 3.0854\n",
            "12000 / 150001\n",
            "13000 / 150001\n",
            "19m 5s (- 201m 11s) (13000 8%) 5.7726\n",
            "14000 / 150001\n",
            "20m 37s (- 200m 25s) (14000 9%) 3.1662\n",
            "15000 / 150001\n",
            "22m 8s (- 199m 14s) (15000 10%) 3.2424\n",
            "16000 / 150001\n",
            "23m 35s (- 197m 34s) (16000 10%) 3.1002\n",
            "17000 / 150001\n",
            "25m 4s (- 196m 10s) (17000 11%) 2.9817\n",
            "18000 / 150001\n",
            "26m 34s (- 194m 51s) (18000 12%) 3.1510\n",
            "19000 / 150001\n",
            "28m 1s (- 193m 16s) (19000 12%) 2.9268\n",
            "20000 / 150001\n",
            "29m 31s (- 191m 56s) (20000 13%) 2.9329\n",
            "21000 / 150001\n",
            "31m 2s (- 190m 43s) (21000 14%) 3.4017\n",
            "22000 / 150001\n",
            "32m 31s (- 189m 16s) (22000 14%) 2.9702\n",
            "23000 / 150001\n",
            "33m 58s (- 187m 36s) (23000 15%) 3.0523\n",
            "24000 / 150001\n",
            "35m 30s (- 186m 26s) (24000 16%) 3.3077\n",
            "25000 / 150001\n",
            "36m 58s (- 184m 50s) (25000 16%) 3.2750\n",
            "26000 / 150001\n",
            "38m 27s (- 183m 23s) (26000 17%) 3.3451\n",
            "27000 / 150001\n",
            "39m 58s (- 182m 8s) (27000 18%) 3.3137\n",
            "28000 / 150001\n",
            "41m 30s (- 180m 53s) (28000 18%) 3.4586\n",
            "29000 / 150001\n",
            "42m 58s (- 179m 20s) (29000 19%) 3.1795\n",
            "30000 / 150001\n",
            "44m 26s (- 177m 47s) (30000 20%) 3.2433\n",
            "31000 / 150001\n",
            "45m 57s (- 176m 24s) (31000 20%) 3.2852\n",
            "32000 / 150001\n",
            "47m 26s (- 174m 57s) (32000 21%) 3.2248\n",
            "33000 / 150001\n",
            "48m 58s (- 173m 39s) (33000 22%) 3.1108\n",
            "34000 / 150001\n",
            "50m 28s (- 172m 11s) (34000 22%) 3.2950\n",
            "35000 / 150001\n",
            "51m 59s (- 170m 49s) (35000 23%) 3.2612\n",
            "36000 / 150001\n",
            "53m 31s (- 169m 29s) (36000 24%) 3.2373\n",
            "37000 / 150001\n",
            "54m 57s (- 167m 50s) (37000 24%) 3.0123\n",
            "38000 / 150001\n",
            "56m 29s (- 166m 29s) (38000 25%) 3.1568\n",
            "39000 / 150001\n",
            "57m 59s (- 165m 3s) (39000 26%) 3.1628\n",
            "40000 / 150001\n",
            "59m 27s (- 163m 31s) (40000 26%) 3.1985\n",
            "41000 / 150001\n",
            "60m 59s (- 162m 10s) (41000 27%) 3.2270\n",
            "42000 / 150001\n",
            "62m 32s (- 160m 48s) (42000 28%) 3.4607\n",
            "43000 / 150001\n",
            "64m 2s (- 159m 22s) (43000 28%) 3.4258\n",
            "44000 / 150001\n",
            "65m 33s (- 157m 57s) (44000 29%) 3.2201\n",
            "45000 / 150001\n",
            "67m 6s (- 156m 34s) (45000 30%) 3.3627\n",
            "46000 / 150001\n",
            "68m 36s (- 155m 7s) (46000 30%) 3.1791\n",
            "47000 / 150001\n",
            "70m 8s (- 153m 43s) (47000 31%) 3.3432\n",
            "48000 / 150001\n",
            "71m 41s (- 152m 20s) (48000 32%) 3.2489\n",
            "49000 / 150001\n",
            "73m 17s (- 151m 4s) (49000 32%) 3.4051\n",
            "50000 / 150001\n",
            "74m 47s (- 149m 34s) (50000 33%) 3.1962\n",
            "51000 / 150001\n",
            "76m 20s (- 148m 12s) (51000 34%) 3.4068\n",
            "52000 / 150001\n",
            "77m 51s (- 146m 43s) (52000 34%) 3.2382\n",
            "53000 / 150001\n",
            "79m 26s (- 145m 24s) (53000 35%) 3.4143\n",
            "54000 / 150001\n",
            "80m 59s (- 143m 59s) (54000 36%) 3.3419\n",
            "55000 / 150001\n",
            "82m 33s (- 142m 36s) (55000 36%) 3.3606\n",
            "56000 / 150001\n",
            "84m 4s (- 141m 7s) (56000 37%) 3.4175\n",
            "57000 / 150001\n",
            "85m 39s (- 139m 45s) (57000 38%) 3.2866\n",
            "58000 / 150001\n",
            "87m 13s (- 138m 22s) (58000 38%) 3.2496\n",
            "59000 / 150001\n",
            "88m 46s (- 136m 55s) (59000 39%) 3.5253\n",
            "60000 / 150001\n",
            "90m 17s (- 135m 26s) (60000 40%) 3.2081\n",
            "61000 / 150001\n",
            "91m 50s (- 134m 0s) (61000 40%) 3.3567\n",
            "62000 / 150001\n",
            "93m 23s (- 132m 33s) (62000 41%) 3.3920\n",
            "63000 / 150001\n",
            "94m 56s (- 131m 6s) (63000 42%) 3.4786\n",
            "64000 / 150001\n",
            "96m 31s (- 129m 41s) (64000 42%) 3.4872\n",
            "65000 / 150001\n",
            "98m 0s (- 128m 10s) (65000 43%) 3.3215\n",
            "66000 / 150001\n",
            "99m 33s (- 126m 43s) (66000 44%) 3.3852\n",
            "67000 / 150001\n",
            "101m 9s (- 125m 18s) (67000 44%) 3.4052\n",
            "68000 / 150001\n",
            "102m 41s (- 123m 49s) (68000 45%) 3.4371\n",
            "69000 / 150001\n",
            "104m 11s (- 122m 19s) (69000 46%) 3.2270\n",
            "70000 / 150001\n",
            "105m 43s (- 120m 49s) (70000 46%) 3.2318\n",
            "71000 / 150001\n",
            "107m 20s (- 119m 25s) (71000 47%) 3.3133\n",
            "72000 / 150001\n",
            "108m 55s (- 118m 0s) (72000 48%) 3.4244\n",
            "73000 / 150001\n",
            "110m 28s (- 116m 31s) (73000 48%) 3.3908\n",
            "74000 / 150001\n",
            "112m 3s (- 115m 4s) (74000 49%) 3.5163\n",
            "75000 / 150001\n",
            "113m 35s (- 113m 35s) (75000 50%) 3.4336\n",
            "76000 / 150001\n",
            "115m 5s (- 112m 3s) (76000 50%) 3.2750\n",
            "77000 / 150001\n",
            "116m 36s (- 110m 33s) (77000 51%) 3.2347\n",
            "78000 / 150001\n",
            "118m 10s (- 109m 5s) (78000 52%) 3.2876\n",
            "79000 / 150001\n",
            "119m 41s (- 107m 34s) (79000 52%) 3.1272\n",
            "80000 / 150001\n",
            "121m 14s (- 106m 5s) (80000 53%) 3.2811\n",
            "81000 / 150001\n",
            "122m 49s (- 104m 38s) (81000 54%) 3.5721\n",
            "82000 / 150001\n",
            "83000 / 150001\n",
            "125m 56s (- 101m 39s) (83000 55%) 6.8086\n",
            "84000 / 150001\n",
            "127m 30s (- 100m 11s) (84000 56%) 3.3857\n",
            "85000 / 150001\n",
            "129m 1s (- 98m 40s) (85000 56%) 3.3643\n",
            "86000 / 150001\n",
            "130m 33s (- 97m 9s) (86000 57%) 3.4090\n",
            "87000 / 150001\n",
            "132m 7s (- 95m 40s) (87000 57%) 3.4813\n",
            "88000 / 150001\n",
            "133m 43s (- 94m 12s) (88000 58%) 3.3977\n",
            "89000 / 150001\n",
            "135m 16s (- 92m 43s) (89000 59%) 3.5371\n",
            "90000 / 150001\n",
            "136m 53s (- 91m 15s) (90000 60%) 3.6539\n",
            "91000 / 150001\n",
            "138m 28s (- 89m 46s) (91000 60%) 3.4408\n",
            "92000 / 150001\n",
            "140m 5s (- 88m 18s) (92000 61%) 3.4202\n",
            "93000 / 150001\n",
            "141m 39s (- 86m 49s) (93000 62%) 3.3559\n",
            "94000 / 150001\n",
            "143m 14s (- 85m 20s) (94000 62%) 3.5368\n",
            "95000 / 150001\n",
            "144m 50s (- 83m 51s) (95000 63%) 3.4900\n",
            "96000 / 150001\n",
            "146m 26s (- 82m 22s) (96000 64%) 3.4950\n",
            "97000 / 150001\n",
            "147m 57s (- 80m 50s) (97000 64%) 3.3198\n",
            "98000 / 150001\n",
            "149m 28s (- 79m 18s) (98000 65%) 3.3886\n",
            "99000 / 150001\n",
            "151m 2s (- 77m 48s) (99000 66%) 3.4311\n",
            "100000 / 150001\n",
            "152m 36s (- 76m 18s) (100000 66%) 3.4107\n",
            "101000 / 150001\n",
            "154m 9s (- 74m 47s) (101000 67%) 3.2804\n",
            "102000 / 150001\n",
            "155m 43s (- 73m 16s) (102000 68%) 3.2886\n",
            "103000 / 150001\n",
            "157m 17s (- 71m 46s) (103000 68%) 3.4991\n",
            "104000 / 150001\n",
            "158m 51s (- 70m 15s) (104000 69%) 3.5060\n",
            "105000 / 150001\n",
            "160m 26s (- 68m 45s) (105000 70%) 3.3939\n",
            "106000 / 150001\n",
            "161m 58s (- 67m 14s) (106000 70%) 3.3733\n",
            "107000 / 150001\n",
            "163m 32s (- 65m 43s) (107000 71%) 3.5016\n",
            "108000 / 150001\n",
            "165m 6s (- 64m 12s) (108000 72%) 3.3665\n",
            "109000 / 150001\n",
            "166m 40s (- 62m 41s) (109000 72%) 3.3724\n",
            "110000 / 150001\n",
            "168m 12s (- 61m 9s) (110000 73%) 3.3769\n",
            "111000 / 150001\n",
            "169m 42s (- 59m 37s) (111000 74%) 3.2020\n",
            "112000 / 150001\n",
            "171m 14s (- 58m 6s) (112000 74%) 3.3834\n",
            "113000 / 150001\n",
            "172m 46s (- 56m 34s) (113000 75%) 3.3952\n",
            "114000 / 150001\n",
            "174m 16s (- 55m 2s) (114000 76%) 3.4067\n",
            "115000 / 150001\n",
            "175m 51s (- 53m 31s) (115000 76%) 3.4807\n",
            "116000 / 150001\n",
            "177m 24s (- 52m 0s) (116000 77%) 3.2409\n",
            "117000 / 150001\n",
            "178m 56s (- 50m 28s) (117000 78%) 3.2525\n",
            "118000 / 150001\n",
            "180m 27s (- 48m 56s) (118000 78%) 3.3466\n",
            "119000 / 150001\n",
            "181m 59s (- 47m 24s) (119000 79%) 3.3394\n",
            "120000 / 150001\n",
            "183m 33s (- 45m 53s) (120000 80%) 3.5735\n",
            "121000 / 150001\n",
            "185m 7s (- 44m 22s) (121000 80%) 3.4494\n",
            "122000 / 150001\n",
            "186m 42s (- 42m 51s) (122000 81%) 3.4648\n",
            "123000 / 150001\n",
            "188m 19s (- 41m 20s) (123000 82%) 3.4556\n",
            "124000 / 150001\n",
            "189m 52s (- 39m 48s) (124000 82%) 3.3927\n",
            "125000 / 150001\n",
            "191m 27s (- 38m 17s) (125000 83%) 3.3544\n",
            "126000 / 150001\n",
            "193m 2s (- 36m 46s) (126000 84%) 3.2639\n",
            "127000 / 150001\n",
            "194m 35s (- 35m 14s) (127000 84%) 3.2714\n",
            "128000 / 150001\n",
            "196m 8s (- 33m 42s) (128000 85%) 3.4043\n",
            "129000 / 150001\n",
            "197m 46s (- 32m 11s) (129000 86%) 3.5456\n",
            "130000 / 150001\n",
            "199m 21s (- 30m 40s) (130000 86%) 3.3111\n",
            "131000 / 150001\n",
            "200m 55s (- 29m 8s) (131000 87%) 3.2618\n",
            "132000 / 150001\n",
            "202m 31s (- 27m 37s) (132000 88%) 3.4920\n",
            "133000 / 150001\n",
            "204m 5s (- 26m 5s) (133000 88%) 3.2542\n",
            "134000 / 150001\n",
            "205m 39s (- 24m 33s) (134000 89%) 3.4589\n",
            "135000 / 150001\n",
            "207m 11s (- 23m 1s) (135000 90%) 3.4521\n",
            "136000 / 150001\n",
            "208m 45s (- 21m 29s) (136000 90%) 3.4473\n",
            "137000 / 150001\n",
            "210m 20s (- 19m 57s) (137000 91%) 3.3711\n",
            "138000 / 150001\n",
            "211m 54s (- 18m 25s) (138000 92%) 3.4012\n",
            "139000 / 150001\n",
            "213m 29s (- 16m 53s) (139000 92%) 3.5054\n",
            "140000 / 150001\n",
            "215m 5s (- 15m 21s) (140000 93%) 3.3953\n",
            "141000 / 150001\n",
            "216m 38s (- 13m 49s) (141000 94%) 3.2942\n",
            "142000 / 150001\n",
            "218m 13s (- 12m 17s) (142000 94%) 3.4037\n",
            "143000 / 150001\n",
            "219m 46s (- 10m 45s) (143000 95%) 3.2801\n",
            "144000 / 150001\n",
            "221m 22s (- 9m 13s) (144000 96%) 3.3766\n",
            "145000 / 150001\n",
            "222m 59s (- 7m 41s) (145000 96%) 3.4959\n",
            "146000 / 150001\n",
            "224m 33s (- 6m 9s) (146000 97%) 3.4519\n",
            "147000 / 150001\n",
            "226m 7s (- 4m 36s) (147000 98%) 3.4316\n",
            "148000 / 150001\n",
            "227m 41s (- 3m 4s) (148000 98%) 3.4501\n",
            "149000 / 150001\n",
            "229m 18s (- 1m 32s) (149000 99%) 3.4763\n",
            "150000 / 150001\n",
            "230m 54s (- 0m 0s) (150000 100%) 3.5219\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 0 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_uH0qi_PS2fv",
        "colab_type": "code",
        "outputId": "d4d27ec9-50aa-44ae-bbf3-2f4bdfd1a72d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 269
        }
      },
      "source": [
        "text,headline,pred_headline=evaluateRandomly(encoder1, attn_decoder1,15000)\n",
        "\n",
        "pred_df_LSTM=pd.DataFrame()\n",
        "\n",
        "pred_df_LSTM['text']=text\n",
        "pred_df_LSTM['headline']=headline\n",
        "pred_df_LSTM['pred_headline']=pred_headline"
      ],
      "execution_count": 47,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.0 % complete\n",
            "6.666666666666667 % complete\n",
            "13.333333333333334 % complete\n",
            "20.0 % complete\n",
            "26.666666666666668 % complete\n",
            "33.333333333333336 % complete\n",
            "40.0 % complete\n",
            "46.666666666666664 % complete\n",
            "53.333333333333336 % complete\n",
            "60.0 % complete\n",
            "66.66666666666667 % complete\n",
            "73.33333333333333 % complete\n",
            "80.0 % complete\n",
            "86.66666666666667 % complete\n",
            "93.33333333333333 % complete\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sFw6prOqtsE2",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 289
        },
        "outputId": "c61aacf0-38d2-475f-f8ba-e028e5815731"
      },
      "source": [
        "for i in range(105,110):\n",
        "  print(\"original Summary>>>\",pred_df_LSTM.iloc[i]['headline'])\n",
        "  print(\"Predicted Summary>>>\",pred_df_LSTM.iloc[i]['pred_headline'])\n",
        "  print('-----------------------------------------------------------------------')"
      ],
      "execution_count": 48,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "original Summary>>> click profile picture upper right corner select settings select privacy permissions scroll defaults new uploads section edit able see comment add notes add people change default setting anyone allow friends family see images click save settings\n",
            "Predicted Summary>>> open click website click select click click select click click select click click <EOS>\n",
            "-----------------------------------------------------------------------\n",
            "original Summary>>> conditioning key know basics learn fundamentals move elite wrestler moves learn counters moves hopefully get position less dangerous situation stand key escape move perform review moves know head get practice little early work practice partner never starve ask help practice makes perfect heart urge win always one step front opponents\n",
            "Predicted Summary>>> talk parents time take look get <EOS>\n",
            "-----------------------------------------------------------------------\n",
            "original Summary>>> apply pressure defecating soften stools water foods medicine soak warm bathtub apply witch hazel ask doctor medications\n",
            "Predicted Summary>>> take skin water water water water water water <EOS>\n",
            "-----------------------------------------------------------------------\n",
            "original Summary>>> pour thai coffee powder saucepan medium heat add sugar leave coffee mixture cool strain coffee cool chill ready serve\n",
            "Predicted Summary>>> pour ingredients add add add add add add add add add add <EOS>\n",
            "-----------------------------------------------------------------------\n",
            "original Summary>>> click apple menu select system preferences adjust mouse settings selecting mouse option adjust magic mouse settings adjust trackpad settings selecting trackpad option system preferences\n",
            "Predicted Summary>>> open settings open open settings open <EOS>\n",
            "-----------------------------------------------------------------------\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WibVGx9zwqaR",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import os\n",
        "os.chdir ('/content/drive/My Drive/IDS_576_Project')\n",
        "pred_df_LSTM.to_csv('Attention_LSTM_Pred.csv',sep='\\t',index=False,header=True)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1p6Jki1bwqX2",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OV7y3sPRyfo8",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def calculate_rogue(src_trg, pred_trg):\n",
        "      \n",
        "  #cut off <eos> token\n",
        "  pred_trg = pred_trg[:-6]\n",
        "\n",
        "  \n",
        "  if (len(pred_trg) == 0):\n",
        "    rogue_score = 0.0\n",
        "  else:\n",
        "    s = rouge.get_scores(pred_trg, src_trg, avg= True)\n",
        "    rogue_score = s['rouge-1']['f']\n",
        "    \n",
        "  return rogue_score"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "geOZvrpTyf9T",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "attn_plot_threshold = 0.45\n",
        "\n",
        "def evaluateRandomlyprint_1(encoder, decoder, n=5):\n",
        "    text=list()\n",
        "    headline=list()\n",
        "    pred_headline=list()    \n",
        "\n",
        "    for i in range(n):\n",
        "        pair = random.choice(pairs)\n",
        "        \n",
        "        if(len(pair[0].split())>=150):\n",
        "          continue\n",
        "        else:\n",
        "          #if(i%1000==0):\n",
        "           # print(i*100/n,\"% complete\")\n",
        "          \n",
        "          tokenized_input = nltk.word_tokenize(pair[0])\n",
        "          #print(len(tokenized_input))\n",
        "          output_words, attentions = evaluate(encoder, decoder, pair[0])\n",
        "\n",
        "          output_sentence = ' '.join(output_words)\n",
        "\n",
        "          #get rogue f-score -- need to pass tokenized version here\n",
        "          #print(pair[0])\n",
        "          #print(output_sentence)\n",
        "          score = calculate_rogue(pair[0], output_sentence)\n",
        "          #if(score>0.25):\n",
        "           # print(score)\n",
        "\n",
        "          if score > attn_plot_threshold:\n",
        "            plot_attention(tokenized_input, output_words, attentions, i)\n",
        "          \n",
        "          #print('<', output_sentence)\n",
        "          #print('')\n",
        "    #return(text,headline,pred_headline)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "G9_ILKf8yh9m",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#sentence, translation, attention\n",
        "def plot_attention(sentence, predicted_sentence, attention, i):\n",
        "  plt.rcParams.update({'font.size': 18})\n",
        "\n",
        "  fig = plt.figure(figsize=(30,30))\n",
        "\n",
        "  ax = fig.add_subplot(1,1,1)\n",
        "  #print(len(attention))  \n",
        "  ax.matshow(attention, cmap='bone')\n",
        "  #fig.colorbar(cax)\n",
        "  #fontdict = {'fontsize': 25}\n",
        "\n",
        "  #ax.set_xticklabels([''] + sentence.split(' ') +['<EOS>'], rotation=90)\n",
        "  #ax.set_yticklabels([''] + predicted_sentence)\n",
        "\n",
        "  ax.set_xticklabels(['']+['<sos>']+[t.lower() for t in sentence]+['<eos>'], rotation=90)\n",
        "  ax.set_yticklabels(['']+predicted_sentence)\n",
        "\n",
        "  ax.xaxis.set_major_locator(ticker.MultipleLocator(3))\n",
        "  ax.yaxis.set_major_locator(ticker.MultipleLocator(3))\n",
        "  #xlim(right=3)\n",
        "\n",
        "  plt.show()\n",
        "  #plt.close()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WVb9wutXyjqy",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 573
        },
        "outputId": "f5e7c84b-ad09-49de-a10e-8cc857b76e2b"
      },
      "source": [
        "!pip install rouge\n",
        "import nltk\n",
        "nltk.download('punkt')\n",
        "from rouge import Rouge\n",
        "\n",
        "rouge = Rouge()\n",
        "\n",
        "\n",
        "evaluateRandomlyprint_1(encoder1, attn_decoder1,15000)\n",
        "#plt.close()"
      ],
      "execution_count": 54,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: rouge in /usr/local/lib/python3.6/dist-packages (1.0.0)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from rouge) (1.12.0)\n",
            "[nltk_data] Downloading package punkt to /root/nltk_data...\n",
            "[nltk_data]   Package punkt is already up-to-date!\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 2160x2160 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 2160x2160 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 2160x2160 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ATc2C77jqDzr",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 300
        },
        "outputId": "971bd429-c056-43d9-f9fa-84a90f97e399"
      },
      "source": [
        "!pip install rouge\n",
        "import nltk\n",
        "nltk.download('punkt')\n",
        "from rouge import Rouge\n",
        "\n",
        "rouge = Rouge()\n",
        "\n",
        "\n",
        "evaluateRandomlyprint_1(encoder1, attn_decoder1,15000)\n",
        "#plt.close()"
      ],
      "execution_count": 55,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: rouge in /usr/local/lib/python3.6/dist-packages (1.0.0)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from rouge) (1.12.0)\n",
            "[nltk_data] Downloading package punkt to /root/nltk_data...\n",
            "[nltk_data]   Package punkt is already up-to-date!\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 2160x2160 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}