lstm_and_gru/lstm_and_gru_keras.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "lstm_and_gru_keras.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "metadata": {
        "id": "ZIC0nEEjEjGG",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# LSTM\n",
        "\n",
        "In this notebook, we will go through basics of LSTM, GRU, CuDNNLSTM, CuDNNGRU and Bi-LSTM using pretrained embeddings like word2vec, glove and fasttext and  also custom embeddings on IMDB dataset. \n",
        "\n",
        "Here we will use [keras](https://keras.io/  \"Keras Homepage\").\n",
        "\n",
        "\n",
        "Everything is explained in-detail in [blog post](https://dudeperf3ct.github.io/lstm/gru/nlp/2019/01/28/Force-of-LSTM-and-GRU/). This is notebook which replicates the result of blog and runs in colab. Enjoy!\n",
        "\n",
        "Let's see at [nlpprogess](http://nlpprogress.com/english/sentiment_analysis.html) what is the current state-of-the-art in sentiment analysis.\n",
        "\n",
        "\n",
        "Model | Accuracy | Paper |\n",
        "----- | -------- | ------|\n",
        "ULMFit| 95.4     | [Universal Language Model Fine-tuning for Text Classification](https://arxiv.org/abs/1801.06146)|\n",
        "Block-sparse LSTM| 94.99 | [GPU Kernels for Block-Sparse Weights](https://s3-us-west-2.amazonaws.com/openai-assets/blocksparse/blocksparsepaper.pdf)|\n",
        "oh-LSTM | 94.1 | [Supervised and Semi-Supervised Text Categorization using LSTM for Region Embeddings](https://arxiv.org/abs/1602.02373) |\n",
        "Virtual adversarial training  | 94.1 | [Adversarial Training Methods for Semi-Supervised Text Classification](https://arxiv.org/abs/1605.07725) |\n",
        "BCN+Char+CoVe | 91.8 | [Learned in Translation: Contextualized Word Vectors](https://arxiv.org/abs/1708.00107) |\n",
        "\n",
        "\n",
        "#### Run in Colab\n",
        "\n",
        "You can run this notebook in google colab.\n",
        "\n",
        "[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/dudeperf3ct/DL_notebooks/blob/master/lstm_and_gru/lstm_and_gru_keras.ipynb)\n"
      ]
    },
    {
      "metadata": {
        "id": "oYAs-iIiT5wG",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "## Download data and pretrained models"
      ]
    },
    {
      "metadata": {
        "id": "W7RxSTNzFP17",
        "colab_type": "code",
        "outputId": "ea407a66-4c35-4ee6-92be-760df8efc6fa",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 88
        }
      },
      "cell_type": "code",
      "source": [
        "# download word2vec pretrained model on Google News\n",
        "! gdown \"https://drive.google.com/uc?id=0B7XkCwpI5KDYNlNUTTlSS21pQmM\"\n",
        "! gunzip \"/content/GoogleNews-vectors-negative300.bin.gz\""
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading...\n",
            "From: https://drive.google.com/uc?id=0B7XkCwpI5KDYNlNUTTlSS21pQmM\n",
            "To: /content/GoogleNews-vectors-negative300.bin.gz\n",
            "1.65GB [00:11, 138MB/s]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "JL6ipmq0JpoM",
        "colab_type": "code",
        "outputId": "fe521b5b-ba26-47df-bfaf-59d4a7b748de",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 285
        }
      },
      "cell_type": "code",
      "source": [
        "# download pretrained glove model\n",
        "! wget \"http://nlp.stanford.edu/data/glove.6B.zip\"\n",
        "! unzip -q -n \"glove.6B.zip\""
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2019-02-26 13:47:33--  http://nlp.stanford.edu/data/glove.6B.zip\n",
            "Resolving nlp.stanford.edu (nlp.stanford.edu)... 171.64.67.140\n",
            "Connecting to nlp.stanford.edu (nlp.stanford.edu)|171.64.67.140|:80... connected.\n",
            "HTTP request sent, awaiting response... 302 Found\n",
            "Location: https://nlp.stanford.edu/data/glove.6B.zip [following]\n",
            "--2019-02-26 13:47:33--  https://nlp.stanford.edu/data/glove.6B.zip\n",
            "Connecting to nlp.stanford.edu (nlp.stanford.edu)|171.64.67.140|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 862182613 (822M) [application/zip]\n",
            "Saving to: ‘glove.6B.zip’\n",
            "\n",
            "glove.6B.zip        100%[===================>] 822.24M  9.26MB/s    in 78s     \n",
            "\n",
            "2019-02-26 13:48:52 (10.5 MB/s) - ‘glove.6B.zip’ saved [862182613/862182613]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "ctnBICUp42WV",
        "colab_type": "code",
        "outputId": "cd1e589d-bf6d-45d6-af0f-f26bd99c8c60",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 214
        }
      },
      "cell_type": "code",
      "source": [
        "# download pretrained fasttext model\n",
        "! wget \"https://s3-us-west-1.amazonaws.com/fasttext-vectors/wiki.simple.zip\"\n",
        "! unzip -q -n \"wiki.simple.zip\""
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2019-02-26 13:49:21--  https://s3-us-west-1.amazonaws.com/fasttext-vectors/wiki.simple.zip\n",
            "Resolving s3-us-west-1.amazonaws.com (s3-us-west-1.amazonaws.com)... 52.219.116.80\n",
            "Connecting to s3-us-west-1.amazonaws.com (s3-us-west-1.amazonaws.com)|52.219.116.80|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 2618496202 (2.4G) [application/zip]\n",
            "Saving to: ‘wiki.simple.zip’\n",
            "\n",
            "wiki.simple.zip     100%[===================>]   2.44G  21.4MB/s    in 2m 0s   \n",
            "\n",
            "2019-02-26 13:51:22 (20.8 MB/s) - ‘wiki.simple.zip’ saved [2618496202/2618496202]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "bBecD2yP-M8q",
        "colab_type": "code",
        "outputId": "549d1519-232c-4bcb-82e8-3d25372bf371",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 214
        }
      },
      "cell_type": "code",
      "source": [
        "# ! wget \"https://dl.fbaipublicfiles.com/fasttext/vectors-english/wiki-news-300d-1M-subword.vec.zip\"\n",
        "# ! unzip -q -n \"wiki-news-300d-1M-subword.vec.zip\"\n",
        "! wget \"https://dl.fbaipublicfiles.com/fasttext/vectors-english/wiki-news-300d-1M.vec.zip\"\n",
        "! unzip -q -n \"wiki-news-300d-1M.vec.zip\""
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2019-02-26 13:52:01--  https://dl.fbaipublicfiles.com/fasttext/vectors-english/wiki-news-300d-1M.vec.zip\n",
            "Resolving dl.fbaipublicfiles.com (dl.fbaipublicfiles.com)... 104.20.22.166, 104.20.6.166, 2606:4700:10::6814:16a6, ...\n",
            "Connecting to dl.fbaipublicfiles.com (dl.fbaipublicfiles.com)|104.20.22.166|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 681808098 (650M) [application/zip]\n",
            "Saving to: ‘wiki-news-300d-1M.vec.zip’\n",
            "\n",
            "wiki-news-300d-1M.v 100%[===================>] 650.22M  8.54MB/s    in 56s     \n",
            "\n",
            "2019-02-26 13:52:59 (11.6 MB/s) - ‘wiki-news-300d-1M.vec.zip’ saved [681808098/681808098]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "rrOw6W0dO8Mh",
        "colab_type": "code",
        "outputId": "ff586012-17b6-405a-f87b-5f4f625e0b4b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 412
        }
      },
      "cell_type": "code",
      "source": [
        "# install spacy\n",
        "! python -m spacy download en\n",
        "! python -m spacy download en_core_web_md"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: en_core_web_sm==2.0.0 from https://github.com/explosion/spacy-models/releases/download/en_core_web_sm-2.0.0/en_core_web_sm-2.0.0.tar.gz#egg=en_core_web_sm==2.0.0 in /usr/local/lib/python3.6/dist-packages (2.0.0)\n",
            "\n",
            "\u001b[93m    Linking successful\u001b[0m\n",
            "    /usr/local/lib/python3.6/dist-packages/en_core_web_sm -->\n",
            "    /usr/local/lib/python3.6/dist-packages/spacy/data/en\n",
            "\n",
            "    You can now load the model via spacy.load('en')\n",
            "\n",
            "Collecting en_core_web_md==2.0.0 from https://github.com/explosion/spacy-models/releases/download/en_core_web_md-2.0.0/en_core_web_md-2.0.0.tar.gz#egg=en_core_web_md==2.0.0\n",
            "\u001b[?25l  Downloading https://github.com/explosion/spacy-models/releases/download/en_core_web_md-2.0.0/en_core_web_md-2.0.0.tar.gz (120.8MB)\n",
            "\u001b[K    100% |████████████████████████████████| 120.9MB 3.3MB/s \n",
            "\u001b[?25hInstalling collected packages: en-core-web-md\n",
            "  Running setup.py install for en-core-web-md ... \u001b[?25ldone\n",
            "\u001b[?25hSuccessfully installed en-core-web-md-2.0.0\n",
            "\n",
            "\u001b[93m    Linking successful\u001b[0m\n",
            "    /usr/local/lib/python3.6/dist-packages/en_core_web_md -->\n",
            "    /usr/local/lib/python3.6/dist-packages/spacy/data/en_core_web_md\n",
            "\n",
            "    You can now load the model via spacy.load('en_core_web_md')\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "ZjmAYyBxaT94",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "You need to upload your credentials from Kaggle on your instance. Login to kaggle and click on your profile picture on the top left corner, then 'My account'. Scroll down until you find a button named 'Create New API Token' and click on it. This will trigger the download of a file named 'kaggle.json'.\n",
        "\n",
        "Upload this file to the directory this notebook is running in, by clicking \"Upload\" on your main Jupyter page, then uncomment and execute the next two commands (or run them in a terminal)."
      ]
    },
    {
      "metadata": {
        "id": "VNEfTV3nSMlX",
        "colab_type": "code",
        "outputId": "e498229c-29ad-473c-f23c-9398c87827b3",
        "colab": {
          "resources": {
            "http://localhost:8080/nbextensions/google.colab/files.js": {
              "data": "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",
              "ok": true,
              "headers": [
                [
                  "content-type",
                  "application/javascript"
                ]
              ],
              "status": 200,
              "status_text": "OK"
            }
          },
          "base_uri": "https://localhost:8080/",
          "height": 83
        }
      },
      "cell_type": "code",
      "source": [
        "# #uncomment this once to upload kaggle.json\n",
        "\n",
        "from google.colab import files\n",
        "def getLocalFiles():\n",
        "    _files = files.upload()\n",
        "    if len(_files) >0:\n",
        "       for k,v in _files.items():\n",
        "          open(k,'wb').write(v)\n",
        "getLocalFiles()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "     <input type=\"file\" id=\"files-3d70605e-9c71-4a19-9e2f-ff9ff0cc308b\" name=\"files[]\" multiple disabled />\n",
              "     <output id=\"result-3d70605e-9c71-4a19-9e2f-ff9ff0cc308b\">\n",
              "      Upload widget is only available when the cell has been executed in the\n",
              "      current browser session. Please rerun this cell to enable.\n",
              "      </output>\n",
              "      <script src=\"/nbextensions/google.colab/files.js\"></script> "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Saving kaggle.json to kaggle.json\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "05RB8_zDXhL3",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# #uncomment and run this once when you upload kaggle.json\n",
        "\n",
        "! mkdir -p ~/.kaggle/\n",
        "! mv kaggle.json ~/.kaggle/"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "H8qeSt13EN5J",
        "colab_type": "code",
        "outputId": "2927f7e1-6af7-470d-ff2b-383b2b3cbe4e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "cell_type": "code",
      "source": [
        "from pathlib import Path\n",
        "\n",
        "path = Path(\"data\")\n",
        "path.mkdir(parents=True, exist_ok=True)\n",
        "path"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "PosixPath('data')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "metadata": {
        "id": "v9dF18MWaY2G",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "You're all set to download the data from [Bags of Model](https://www.kaggle.com/c/word2vec-nlp-tutorial) competition. You first need to go to its main page and accept its rules, and run the two cells below (uncomment the shell commands to download and unzip the data). If you get a 403 forbidden error it means you haven't accepted the competition rules yet (you have to go to the competition page, click on Rules tab, and then scroll to the bottom to find the accept button)."
      ]
    },
    {
      "metadata": {
        "id": "YbF1Z1KUSMsd",
        "colab_type": "code",
        "outputId": "9680aaa2-959c-404e-e758-3839ba167d4d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 249
        }
      },
      "cell_type": "code",
      "source": [
        "# download imdb dataset from kaggle competition\n",
        "!kaggle competitions download -c word2vec-nlp-tutorial -p {path}"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Warning: Your Kaggle API key is readable by other users on this system! To fix this, you can run 'chmod 600 /root/.kaggle/kaggle.json'\n",
            "Downloading sampleSubmission.csv to data\n",
            "  0% 0.00/276k [00:00<?, ?B/s]\n",
            "100% 276k/276k [00:00<00:00, 42.0MB/s]\n",
            "Downloading unlabeledTrainData.tsv.zip to data\n",
            " 92% 24.0M/26.0M [00:00<00:00, 12.3MB/s]\n",
            "100% 26.0M/26.0M [00:00<00:00, 38.0MB/s]\n",
            "Downloading testData.tsv.zip to data\n",
            " 40% 5.00M/12.6M [00:00<00:00, 9.11MB/s]\n",
            "100% 12.6M/12.6M [00:00<00:00, 21.5MB/s]\n",
            "Downloading labeledTrainData.tsv.zip to data\n",
            " 39% 5.00M/13.0M [00:00<00:00, 9.16MB/s]\n",
            "100% 13.0M/13.0M [00:00<00:00, 20.4MB/s]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "vAOqhVlhZZvL",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "! unzip -q -n {path}/labeledTrainData.tsv.zip -d {path}\n",
        "! unzip -q -n {path}/testData.tsv.zip -d {path}"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "CPX0rk2CUG-f",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "## IMDB data"
      ]
    },
    {
      "metadata": {
        "id": "pNkINkerEmzY",
        "colab_type": "code",
        "outputId": "e9472829-6f2d-44b3-e184-997fe8603ecd",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        }
      },
      "cell_type": "code",
      "source": [
        "from pathlib import Path\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "from sklearn.model_selection import train_test_split\n",
        "from keras.layers import LSTM, Embedding, Dense, Input, CuDNNLSTM, GRU, CuDNNGRU, Bidirectional\n",
        "from keras.models import Model\n",
        "from keras.preprocessing.text import Tokenizer\n",
        "from keras.preprocessing.sequence import pad_sequences\n",
        "from keras.initializers import Constant\n",
        "from keras.utils import to_categorical\n",
        "from keras.callbacks import EarlyStopping\n",
        "path = Path(\"data\")\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline\n",
        "\n",
        "data = pd.read_csv(path/\"labeledTrainData.tsv\", sep='\\t', header=0, quoting=3)\n",
        "data.head()\n",
        "\n",
        "# test = pd.read_csv(path/\"testData.tsv\", sep='\\t', header=0, quoting=3)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>id</th>\n",
              "      <th>sentiment</th>\n",
              "      <th>review</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>\"5814_8\"</td>\n",
              "      <td>1</td>\n",
              "      <td>\"With all this stuff going down at the moment ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>\"2381_9\"</td>\n",
              "      <td>1</td>\n",
              "      <td>\"\\\"The Classic War of the Worlds\\\" by Timothy ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>\"7759_3\"</td>\n",
              "      <td>0</td>\n",
              "      <td>\"The film starts with a manager (Nicholas Bell...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>\"3630_4\"</td>\n",
              "      <td>0</td>\n",
              "      <td>\"It must be assumed that those who praised thi...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>\"9495_8\"</td>\n",
              "      <td>1</td>\n",
              "      <td>\"Superbly trashy and wondrously unpretentious ...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "         id  sentiment                                             review\n",
              "0  \"5814_8\"          1  \"With all this stuff going down at the moment ...\n",
              "1  \"2381_9\"          1  \"\\\"The Classic War of the Worlds\\\" by Timothy ...\n",
              "2  \"7759_3\"          0  \"The film starts with a manager (Nicholas Bell...\n",
              "3  \"3630_4\"          0  \"It must be assumed that those who praised thi...\n",
              "4  \"9495_8\"          1  \"Superbly trashy and wondrously unpretentious ..."
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "metadata": {
        "id": "2Z2FYcgwYaW3",
        "colab_type": "code",
        "outputId": "e2d21a04-7851-46cf-dc53-f96bc60a4f7d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "cell_type": "code",
      "source": [
        "print (data['review'].shape, data['sentiment'].shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(25000,) (25000,)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "ymLgvGIzOfGQ",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "## Gensim\n",
        "\n",
        "[Gensim](https://radimrehurek.com/gensim/) is a Python library for topic modelling, document indexing and similarity retrieval with large corpora."
      ]
    },
    {
      "metadata": {
        "id": "oVTKPknKULmG",
        "colab_type": "code",
        "outputId": "3a78f894-ebb1-464f-8baa-0c2b19c9ff58",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "cell_type": "code",
      "source": [
        "NUM_WORDS = 20000\n",
        "MAX_SEQUENCE_LENGTH = 200\n",
        "\n",
        "# convert each word into token(integer) with maximum number of words based on frequency is 20000\n",
        "tokenizer = Tokenizer(num_words=NUM_WORDS, filters='!\"#$%&()*+,-./:;<=>?@[\\\\]^_`{|}~\\t\\n\\'', lower=True)\n",
        "tokenizer.fit_on_texts(data['review'])\n",
        "\n",
        "# stitch back the words into sequnces with filtering stop words and converting to lower #text_to_word_sequence\n",
        "sequences = tokenizer.texts_to_sequences(data['review'])\n",
        "# convert all sentences to equal length of 200\n",
        "sentences = pad_sequences(sequences, maxlen=MAX_SEQUENCE_LENGTH)\n",
        "\n",
        "word_index = tokenizer.word_index\n",
        "print('Found %s unique tokens.' % len(word_index))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 75899 unique tokens.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "1qPBZqC9d-n0",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "### Pretrained Word2Vec"
      ]
    },
    {
      "metadata": {
        "id": "s5O7BVzGEnHM",
        "colab_type": "code",
        "outputId": "c41e59e7-8d1a-4c08-c900-0f3182905ad2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        }
      },
      "cell_type": "code",
      "source": [
        "from gensim.models import KeyedVectors\n",
        "\n",
        "%time word_vectors = KeyedVectors.load_word2vec_format('/content/GoogleNews-vectors-negative300.bin', binary=True)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 55 s, sys: 5.73 s, total: 1min\n",
            "Wall time: 1min 2s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "hIZ5EX_9Ohz3",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "EMBEDDING_DIM = 300\n",
        "VOCAB_SIZE = min(len(word_index)+1, NUM_WORDS)\n",
        "embedding_matrix = np.zeros((VOCAB_SIZE, EMBEDDING_DIM))\n",
        "\n",
        "for word, i in word_index.items():\n",
        "  \n",
        "    if i>=NUM_WORDS:\n",
        "        continue\n",
        "    try:\n",
        "        embedding_vector = word_vectors[word]\n",
        "        embedding_matrix[i] = embedding_vector\n",
        "        \n",
        "    except KeyError:\n",
        "        embedding_matrix[i]=np.random.normal(0,np.sqrt(0.25), EMBEDDING_DIM)\n",
        "\n",
        "del(word_vectors)\n",
        "\n",
        "# load pre-trained word embeddings into an Embedding layer\n",
        "# note that we set trainable = False so as to keep the embeddings fixed\n",
        "embedding_layer = Embedding(VOCAB_SIZE,\n",
        "                            EMBEDDING_DIM,\n",
        "                            embeddings_initializer=Constant(embedding_matrix),\n",
        "                            input_length=MAX_SEQUENCE_LENGTH,\n",
        "                            trainable=False)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "ur-zL_mreCaf",
        "colab_type": "code",
        "outputId": "b9b889fa-49cd-473a-802e-bb4d15233b07",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        }
      },
      "cell_type": "code",
      "source": [
        "labels = data['sentiment'].values\n",
        "\n",
        "X, Y = sentences, labels\n",
        "\n",
        "train_x, val_x, train_y, val_y = train_test_split(X, Y, test_size=0.2, stratify=Y)\n",
        "\n",
        "train_y = to_categorical(train_y, num_classes=2)\n",
        "val_y = to_categorical(val_y, num_classes=2)\n",
        "\n",
        "print ('Training:', train_x.shape, train_y.shape)\n",
        "print ('Validation:', val_x.shape, val_y.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Training: (20000, 200) (20000, 2)\n",
            "Validation: (5000, 200) (5000, 2)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "5HndsxUQOshP",
        "colab_type": "code",
        "outputId": "e6b13245-bb4f-4792-d94a-e112c69ec0ed",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 412
        }
      },
      "cell_type": "code",
      "source": [
        "num_classes = 2\n",
        "\n",
        "# input layer with shape = (MAX_SEQUENCE_LENGTH,)\n",
        "sequence_input = Input(shape=(MAX_SEQUENCE_LENGTH,), dtype='int32')\n",
        "\n",
        "# convert the sequence input to embedding from (MAX_SEQUENCE_LENGTH, ) -> (None, MAX_SEQUENCE_LENGTH, EMBEDDING_DIM)\n",
        "embedded_sequences = embedding_layer(sequence_input)\n",
        "\n",
        "x = LSTM(128, dropout=0.2, recurrent_dropout=0.2)(embedded_sequences)\n",
        "preds = Dense(num_classes, activation='softmax')(x)\n",
        "\n",
        "model = Model(inputs=sequence_input, outputs=preds)\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer='adam',\n",
        "              metrics=['acc'])\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Colocations handled automatically by placer.\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:3445: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_1 (InputLayer)         (None, 200)               0         \n",
            "_________________________________________________________________\n",
            "embedding_1 (Embedding)      (None, 200, 300)          6000000   \n",
            "_________________________________________________________________\n",
            "lstm_1 (LSTM)                (None, 128)               219648    \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 2)                 258       \n",
            "=================================================================\n",
            "Total params: 6,219,906\n",
            "Trainable params: 219,906\n",
            "Non-trainable params: 6,000,000\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "5YZf_1x3OsoP",
        "colab_type": "code",
        "outputId": "d7f540d7-59b4-48f1-d9cd-cd13366e8159",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 466
        }
      },
      "cell_type": "code",
      "source": [
        "es = EarlyStopping(monitor='val_acc', patience=3)\n",
        "\n",
        "history = model.fit(train_x, train_y,\n",
        "                    batch_size=128,\n",
        "                    epochs=10,\n",
        "                    validation_data=(val_x, val_y),\n",
        "                    callbacks =[es])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use tf.cast instead.\n",
            "Train on 20000 samples, validate on 5000 samples\n",
            "Epoch 1/10\n",
            "20000/20000 [==============================] - 81s 4ms/step - loss: 0.5923 - acc: 0.6876 - val_loss: 0.4976 - val_acc: 0.7806\n",
            "Epoch 2/10\n",
            "20000/20000 [==============================] - 84s 4ms/step - loss: 0.5056 - acc: 0.7634 - val_loss: 0.4467 - val_acc: 0.8110\n",
            "Epoch 3/10\n",
            "20000/20000 [==============================] - 73s 4ms/step - loss: 0.4960 - acc: 0.7690 - val_loss: 0.4755 - val_acc: 0.7766\n",
            "Epoch 4/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.4647 - acc: 0.7855 - val_loss: 0.4011 - val_acc: 0.8294\n",
            "Epoch 5/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.4603 - acc: 0.7865 - val_loss: 0.3782 - val_acc: 0.8430\n",
            "Epoch 6/10\n",
            "20000/20000 [==============================] - 73s 4ms/step - loss: 0.4057 - acc: 0.8179 - val_loss: 0.3547 - val_acc: 0.8462\n",
            "Epoch 7/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.3800 - acc: 0.8339 - val_loss: 0.3889 - val_acc: 0.8266\n",
            "Epoch 8/10\n",
            "20000/20000 [==============================] - 73s 4ms/step - loss: 0.3596 - acc: 0.8455 - val_loss: 0.3594 - val_acc: 0.8388\n",
            "Epoch 9/10\n",
            "20000/20000 [==============================] - 73s 4ms/step - loss: 0.3425 - acc: 0.8516 - val_loss: 0.3199 - val_acc: 0.8630\n",
            "Epoch 10/10\n",
            "20000/20000 [==============================] - 73s 4ms/step - loss: 0.3333 - acc: 0.8574 - val_loss: 0.3400 - val_acc: 0.8554\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "bZAVPamE06pG",
        "colab_type": "code",
        "outputId": "027d46bb-7dc3-4acd-959f-c91ddf1c5e9a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 571
        }
      },
      "cell_type": "code",
      "source": [
        "# summarize history for accuracy\n",
        "plt.plot(history.history['acc'])\n",
        "plt.plot(history.history['val_acc'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "# summarize history for loss\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEVCAYAAAARjMm4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nolHKIiIiGqUsIiKi0V+QH+mrdcaF5wAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEVCAYAAAAPRfkLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "8115BEW3eC4Z",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "### Pretrained Glove"
      ]
    },
    {
      "metadata": {
        "id": "QaMDO5POJ9en",
        "colab_type": "code",
        "outputId": "c86ba1cc-9ef1-4103-8d9f-92f2a0351a21",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "cell_type": "code",
      "source": [
        "from gensim.scripts.glove2word2vec import glove2word2vec\n",
        "\n",
        "# to convert the GloVe file format to the word2vec file format\n",
        "glove_input_file = 'glove.6B.100d.txt'\n",
        "word2vec_output_file = 'glove.6B.100d.txt.word2vec'\n",
        "glove2word2vec(glove_input_file, word2vec_output_file)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(400000, 100)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 24
        }
      ]
    },
    {
      "metadata": {
        "id": "cSWpEqqVeDSq",
        "colab_type": "code",
        "outputId": "fc253c98-5069-489b-8ba3-f323ca16eb09",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        }
      },
      "cell_type": "code",
      "source": [
        "from gensim.models import KeyedVectors\n",
        "\n",
        "%time word_vectors = KeyedVectors.load_word2vec_format('/content/glove.6B.100d.txt.word2vec', binary=False)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 37.5 s, sys: 455 ms, total: 37.9 s\n",
            "Wall time: 37.9 s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "GLzp322yeDW5",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "EMBEDDING_DIM = 100\n",
        "VOCAB_SIZE = min(len(word_index)+1, NUM_WORDS)\n",
        "embedding_matrix = np.zeros((VOCAB_SIZE, EMBEDDING_DIM))\n",
        "\n",
        "for word, i in word_index.items():\n",
        "  \n",
        "    if i>=NUM_WORDS:\n",
        "        continue\n",
        "    try:\n",
        "        embedding_vector = word_vectors[word]\n",
        "        embedding_matrix[i] = embedding_vector\n",
        "        \n",
        "    except KeyError:\n",
        "        embedding_matrix[i]=np.random.normal(0,np.sqrt(0.25), EMBEDDING_DIM)\n",
        "\n",
        "del(word_vectors)\n",
        "\n",
        "# load pre-trained word embeddings into an Embedding layer\n",
        "# note that we set trainable = False so as to keep the embeddings fixed\n",
        "embedding_layer = Embedding(VOCAB_SIZE,\n",
        "                            EMBEDDING_DIM,\n",
        "                            embeddings_initializer=Constant(embedding_matrix),\n",
        "                            input_length=MAX_SEQUENCE_LENGTH,\n",
        "                            trainable=False)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "UsxI2qegLb-a",
        "colab_type": "code",
        "outputId": "0421de83-3e1b-4a40-f398-070c7f2d10ea",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        }
      },
      "cell_type": "code",
      "source": [
        "labels = data['sentiment'].values\n",
        "\n",
        "X, Y = sentences, labels\n",
        "\n",
        "train_x, val_x, train_y, val_y = train_test_split(X, Y, test_size=0.2, stratify=Y)\n",
        "\n",
        "train_y = to_categorical(train_y, num_classes=2)\n",
        "val_y = to_categorical(val_y, num_classes=2)\n",
        "\n",
        "print ('Training:', train_x.shape, train_y.shape)\n",
        "print ('Validation:', val_x.shape, val_y.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Training: (20000, 200) (20000, 2)\n",
            "Validation: (5000, 200) (5000, 2)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "-rN3w5sdKvCa",
        "colab_type": "code",
        "outputId": "89ca6147-9f71-44be-9b41-08242f442e64",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 285
        }
      },
      "cell_type": "code",
      "source": [
        "num_classes = 2\n",
        "\n",
        "# input layer with shape = (MAX_SEQUENCE_LENGTH,)\n",
        "sequence_input = Input(shape=(MAX_SEQUENCE_LENGTH,), dtype='int32')\n",
        "# convert the sequence input to embedding from (MAX_SEQUENCE_LENGTH, ) -> (VOCAB_SIZE, EMBEDDING_DIM)\n",
        "embedded_sequences = embedding_layer(sequence_input)\n",
        "\n",
        "x = LSTM(100, dropout=0.2, recurrent_dropout=0.2)(embedded_sequences)\n",
        "preds = Dense(num_classes, activation='softmax')(x)\n",
        "\n",
        "model = Model(inputs=sequence_input, outputs=preds)\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer='adam',\n",
        "              metrics=['acc'])\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_2 (InputLayer)         (None, 200)               0         \n",
            "_________________________________________________________________\n",
            "embedding_2 (Embedding)      (None, 200, 100)          2000000   \n",
            "_________________________________________________________________\n",
            "lstm_2 (LSTM)                (None, 100)               80400     \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 2)                 202       \n",
            "=================================================================\n",
            "Total params: 2,080,602\n",
            "Trainable params: 80,602\n",
            "Non-trainable params: 2,000,000\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "YzstQbbYLRU4",
        "colab_type": "code",
        "outputId": "ac77f0f9-351d-427d-a391-9051161f2a52",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 392
        }
      },
      "cell_type": "code",
      "source": [
        "es = EarlyStopping(monitor='val_acc', patience=3)\n",
        "\n",
        "history = model.fit(train_x, train_y,\n",
        "                    batch_size=128,\n",
        "                    epochs=10,\n",
        "                    validation_data=(val_x, val_y),\n",
        "                    callbacks =[es])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 20000 samples, validate on 5000 samples\n",
            "Epoch 1/10\n",
            "20000/20000 [==============================] - 73s 4ms/step - loss: 0.6395 - acc: 0.6289 - val_loss: 0.5903 - val_acc: 0.7126\n",
            "Epoch 2/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.5615 - acc: 0.7141 - val_loss: 0.4477 - val_acc: 0.8000\n",
            "Epoch 3/10\n",
            "20000/20000 [==============================] - 71s 4ms/step - loss: 0.4700 - acc: 0.7782 - val_loss: 0.3923 - val_acc: 0.8268\n",
            "Epoch 4/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.4180 - acc: 0.8078 - val_loss: 0.3586 - val_acc: 0.8434\n",
            "Epoch 5/10\n",
            "20000/20000 [==============================] - 71s 4ms/step - loss: 0.3763 - acc: 0.8320 - val_loss: 0.3474 - val_acc: 0.8502\n",
            "Epoch 6/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.3546 - acc: 0.8454 - val_loss: 0.3369 - val_acc: 0.8538\n",
            "Epoch 7/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.3444 - acc: 0.8500 - val_loss: 0.3319 - val_acc: 0.8582\n",
            "Epoch 8/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.3323 - acc: 0.8548 - val_loss: 0.3490 - val_acc: 0.8470\n",
            "Epoch 9/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.3219 - acc: 0.8613 - val_loss: 0.3193 - val_acc: 0.8638\n",
            "Epoch 10/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.3105 - acc: 0.8684 - val_loss: 0.3072 - val_acc: 0.8700\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "bIuLC13S01v-",
        "colab_type": "code",
        "outputId": "b7c14a3f-edd8-4479-bc85-a49592b31a02",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 571
        }
      },
      "cell_type": "code",
      "source": [
        "# summarize history for accuracy\n",
        "plt.plot(history.history['acc'])\n",
        "plt.plot(history.history['val_acc'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "# summarize history for loss\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEVCAYAAAAPRfkLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJzt3Xd8XMd16PHfLnrvHQRIguRhl0iK\nnWIRRarHliVbkq3YkiM7jp3Eiu3np5fYefFz/HGK/WSnvchxjyxZkiXL6iIpihR7b6KIARsAohCF\nWHRggd29749dgAArSGJxF4vz/Xzw4e7de3fPDoE5d2bunXFYloVSSikF4LQ7AKWUUqFDk4JSSql+\nmhSUUkr106SglFKqnyYFpZRS/TQpKKWU6qdJQSlARH4qIn93lX0eE5ENIxSSUrbQpKCUUqpfpN0B\nKHWtRGQ8sAN4GvgTwAF8Fvg2cDPwrjHm84F9Pwn8b/y/6zXAF4wxJ0UkA3gemAx8BHQCVYFjpgP/\nD8gD3MDjxpi9V4np28Cjgc85BjxqjGkWkTjgGeBWoBv4njHm2Sts/yVwwhjz94H37X8uIuXAz4HP\nAGuAOOBnQAYQBXzbGPN84Lg7gR8GtpcFyucZYJcx5geBfWYC7wN5xhjP0EpfhTttKajRKhM4a4wR\n4DDwAvA5YDbwaREpEZEi4L+AjxtjpgJv4q8YAf4n0GCMmQB8BbgDQEScwKvAr40xU4AvAX8Qkcue\nQInIPODPgfn4k0xM4DnA14HowOesAf5NRPKvsP1qCo0xYoypBH4AvGGMmQZ8HviZiESJSALwG+Ch\nwHc4AXwXfxL89ID3uh94WROCGkiTghqtIoGXAo+PAHuMMY3GmHNALZCPv7J93xhzIrDfT4FVgQp+\nOfAigDGmHNgc2GcqkI3/jBxjzDagAVhyuUCMMfuAccaYVmOMD9gOTAy8fDfw28B+Vfgr9ZorbL+a\nNwY8/hjwz4HHW4FY/K2bpcAZY8yHgde+CfwV8BZQIiIS2H4//mSqVD/tPlKjldcY09X3GGgf+BoQ\nAWQBrr6NxpgWEXHgb2WkAy0DjunbLxWIB46drztJxt9Fc0kiEg88LSIrA5vS8bdKCHxW84AY2q+y\n/WqaBjy+A/iWiGQBPvzdaM5LvHfPgFh/j78l9TP8CWQzSg2gSUGFszpgcd8TEUnDX3k24k8CKQP2\nzQJO4R93aA10Nw0iIo9d5nOexN9tNM8Y0y4i3wMKAq814q+k+96jEH/FfrntfQmtT9qlPlBEovC3\nlD5ljHlLRGKAviR54XvHA+mBFsnz+MdiWoDfBVo2SvXT7iMVztYDy0WkryvnS8C6QB/6DvzdJ4hI\nCbAssE8FUCUiDwZeyxSR5wP99JeTDZQGEkIx/q6hxMBrrwGfFRGHiOQCB/BX2JfbXgvcFPjsiQPi\nulBC4KdvAPyrQE/gc7cCuSIyP/Dat4G/DTzegL/V85do15G6BE0KKmwFzoyfwD9QXIp/HOFPAy9/\nHygWkdPAvwKvBI6xgIeBPw8c8wHwnjGm4wof9Z/AChEx+K/4+RqwWkSexH9WXo8/2WwCvhEYJL7c\n9v8CxovI8UCMv7vMd2sG/gk4ICIHgJP4B8jfwN+N9ADwrIiU4R98/+vAcV78LYwIYNvVS1GNNQ5d\nT0GpsUVEvglkGmO+aXcsKvTomIJSY0hgUPqLwFq7Y1GhSbuPlBojRORP8Y9B/KMx5pTd8ajQpN1H\nSiml+mlLQSmlVL9RP6bQ0NB23U2dtLR4XK7O4QxnVNPyOE/LYjAtj8HCoTyyspIcl9o+plsKkZER\nV99pDNHyOE/LYjAtj8HCuTzGdFJQSik1mCYFpZRS/TQpKKWU6qdJQSmlVD9NCkoppfppUlBKKdVP\nk4JSSql+o/7mNaWUGgu6PF3UdzZS19lAfWcDbm8P9028k+iIqGH9HE0KQbJp03usXLn6qvv9+Mc/\n5JOffJj8/IKr7quUCm8en4fGribqOxsClX8gCXQ10NYzeMVWJ06W5C0gLzFnWGPQpBAEtbU1bNjw\n7pCSwle/+vURiEgpFSosy6KlpzVQ8TdSHzjzr+ts4Fy3C581eIVUBw5iSSLRk09vRzwdzdF4u+Lx\ndSbhmZlwfo2/YaJJIQj+7//9R44dO8qtt85n7dq7qK2t4Uc/+g++//3/Q0NDPV1dXXz+819k6dJb\n+fM//yJf+9o3ef/99+joaKeysoLq6ir+8i+/zuLFS+3+Kkqp69Tl6Q5U+Oe7fOq7Gvu7fi4U64wj\nxZGDszcBd1scLU1R9HbEY7nj6bT8w7+x0REUZSVQODGRiXnJ5GfGD3vcYZ8UXtx4gj2l9Zd8LSLC\ngdd77fPpzZ+azadum3TZ1x955I955ZUXmTChhMrKcv7jP36Ky9XEggWLuOuue6muruLb336KpUtv\nHXRcfX0dP/jBv7Bz53b+8IeXNSkoFeK8Pi+N3QO7e84ngdaetov2j3RGkhqZTqYzGcudQHdrLK7G\nSDpbYujyRuPq2y/CQV5GAgUlCRRmJVKQmUBBVgIZybE4HJecx27YhH1SsNu0aTMASEpK5tixo7z2\n2is4HE5aW1su2nf27JsByM7Opr29/aLXlVL2qm6vZc/ZA5wrPUdVcy2NXU2X7O5Ji0llQkIJUb4k\nfF0JdLRE01QfSbPLSRuOAftCdlocU0v8FX9htv/fnPQ4Ipz2XBwa9knhU7dNuuxZfVZWEg0NF2fz\n4RQV5b8yYP36d2htbeXf//2ntLa28sQTf3zRvhER52de1MWPlAoNPd5eDtQfZkv1Tk63VvRvT4iK\npyipkOSIdKI8SfR2xtHuiqahIYKaJjfVF/wJpyXFMHOC/4y/MCuRgqwE8jISiIkKrRlXwz4p2MHp\ndOL1egdta25uJi8vH6fTyebNG+nt7bUpOqXUUNR1NrC1eie7avfR4fGvnVAYO4Ec31SsrgzOlHdz\n/FwnHu/AloKH+BiYXJBCQaDiL8xKJD8zgcS44b10NFg0KQRBcfEEjCklLy+f1NRUAFauvI2nnvoa\nH330Iffc80dkZ2fzi1/8l82RKqUGau9ys7XiILsbdlPXewYAhzcGb/1EeusKOd4Tz3EsoJGoSKe/\nyycrYVACSE2MDnq/fzCN+jWab2TltZHoPhpNtDzO07IYLJzKw7IsXG1uas91Unuug9qmTqpcDdRy\nDE9KJY5oNwDe1nS89eOI7swnNy2JvIx48jLiyU1PYJZkE+nz4XSO3sr/ciuvaUtBqTHMsiy6PF00\nu1tpdrcM+Gmlxd2Cy91Ci7uVlNgkJqVMZFr6FCanTiQ2Mtbu0K+q1+OlrqmL2iZ/5X/2XCe15zo5\n29SJu9cLWDhTGojMPoMzswGHAyJ8UeT4ZjArZQ5TSsaRlxFPSsLFZ/5ZWYlhkyQvpElBqTDls3y0\n9rT5K/nulgEVfyvN7mZaAs97fJcf34qOiCY1OpnGLhfVbdvZXLUdp8PJhORipqVPZmr6FIqTC3E6\n7LlSxrIs2rp6AxV+R3+lX3uug8bmbi7sRoiKdJKV6SQqu47W2ON047/KryixkOXjljAvezbREdEj\n/0VCiCYFpUahHm/PoLP6wf/6z+5b3K1YF1WL5yVFJZITn0VKTAqpsSmkRqeQGpPsfxzjfxwb4b8u\nPi09jt0nj1LaVMYx13FOtZRzsuU0b5xeR1xkHJI2ianpk5mWPoXMuPTh/769Xs61dlPn6jpf+QcS\nQUe356L9k+OjmDwu1d/lkx5PTno87ph6Djfv41DjUXyWj+iIaJbmLGRZwUKKkgqHPebRSpOCUiGk\nx9tDa087bT3ttPe2B87mB1f2LncLXZ6uy75HpCOClJhkJqQUkxaTQkpMcqCSP1/ZJ8ckE+Uc+p9/\nZEQkk9MmMjltIvdxJx29nRjXCX+SaDrOwYYjHGw4AkBmXAbT0qcwNX0yU1JLiI+Ku+J7+yyLts5e\nmlq7OdfSTVNrN42t3TS1ujnX6n/e1nlxa8bpcJCVFsfkQn/ln5sRT15GArnp8f1X+nT0drKrdi+v\n1uykvrMRgPyEXG4tWMz83DnEjYJusJEW1KQgIk8DiwAL+KoxZs+A174CPAp4gb3GmCdF5DHgu8DJ\nwG7rjTHfC2aMSgWTz/LR0dvZX8n3V/g9gce9gx/3XGL6g4HiImNJiUlhfPK4Cyr7848TouKD3p2T\nEBXP3OzZzM2ejWVZNHQ1Utp0nGNNxylznWRL9Q62VO/A6XBSlFjIuPgJZDrHEelOp7mtJ1DZu/v/\nHXxZ53lRkU7Sk2MZl51IelIsOelx5KYnkJcRT3ZaHJERF39Py7I43VLBluqd7K8/RK/PQ6QzkgW5\nc7m1YBETkotH9dVBwRa0pCAiK4DJxpjFIjIN+DmwOPBaMvA/gEnGGI+IrBORRYFDXzDGfCNYcSl1\no3q8vQMq+Tbaejr8FXtvW6DC7/Bv7/U/vlIXDkCEI4Kk6ERy4jJJjE4kOTqJxOgEkqL8j/sq/ZSY\nFGIjY0boWw6Nz7Jo6+ihvTWG+PZJjO8uJKF3ITUd1TT4ztAZVctp3xnK2yoBsDyR+NrS8bZk4mvJ\nIDEylXHZCaQnx5KRHBv4N6b/eVJ81JAr8G5PN3vqDrCleifV7bUAZMVlsKxgEYtybyExOiFo5RBO\ngtlSWA28CmCMOSYiaSKSbIxpBXoCP4ki0g7EA01BjGXEDXXq7D4HD+6nuHg8aWnD3x+rhsayLFzu\nZmo76ulubqemqfGSZ/XdXvdV3ysuMpakqESyUjJJjk4kMToxUMkPfpwUnUhcZFzInrl293hwtbmp\nauri1BmXv4un7yy/pZumtm48l5w/zEFU5HjSk6eSmuIgMsVFT2wdzdFVtKfVE5Hmn48sJTaNksCA\ntaQVkBB17RO8VbXVsKVmJ3vO7sft7cHpcDInaxbLChYxJa3EtkHw0SqYSSEX2DfgeUNgW6sxpltE\nvgOcArqA3xpjykRkCbBCRN4BooBvGGMOBDHGoLiWqbP7vPnmazzyyKOaFEaAz/LR1N3M2Y46ajvq\nONtRT21nHWc76i45eyWA0+EkKSqBjLh0kqISSYpOIik6gaTowOMo/+Pk6CQSoxKIGuaFT4Khy+2h\nqc2Nq60bV6sbV5ubprbuwDY3rlY3ne6LB3H7JCdEMy47adCZfXpyLBkp/udJcZc+y2/sOsexpuOU\nNh3HuE6wrWY322p248BBUXIh09L8SWJCShGRlxn3OD/1xA5Ot/pbIWkxqawpWsWS/PmkxCQPTyGN\nQSM50Nz/2xHoPvprYArQCmwUkZuAnUCDMeZNEVkM/BqYdaU3TUuLJzLy+ucOycpKuu5jL+dv/uaH\nHD58mBde+BVlZWW0tLTg9Xr51re+xdSpU/nJT37C+vXrcTqdrFq1ilmzZrF162bOnCnnX//1X8nP\nzx/2mIYqGOVhF5/PR31HI1WttVS1nqWqpZaq1lqqW89eVPlHOCPIT8qhMDmPwuRc8pNzSI1NISU2\niZSYJBKig99PP1wsy6Kz20NjSxeNzV00Nndzrv9xF40t/uedl7hqp09CbCSZaXFkpsaRmRJHZkos\nWWnxZKXFkZXm3xZ9nXP2ZJHEtKLxwBp8Ph8nXRUcOnuMI3XHKGs8RUXrGd6p2EhMZAwzsqcwO2cq\nN+VOJz8ph9q2Otaf3Mqm8h109HTiwMGcvJmsKbmVuXkzcY7gJHLh9LcyUDCTQg3+lkGffKA28Hga\ncMoY0wggIluAecaYnwOlAMaYHSKSJSIRxpjBEwkN4HJ1XjGIV068wYH6I5d8LcLpwOu79hui52TP\n4hOT7r3s6w888AgORwRdXb3cfPN87rvv45w+fYrvfvd7/OhH/8HPfvYzXn31HSIiInj11ZeZMmU2\nJSWT+drXvklUlH13jo7Wu1b7pi/2n/WfP/uv66yn1ze44ot0RpITn0VeQg658TnkJeaQF59NZlwG\nEc7zlVx/Wbih223RTcdIf61LsiyLTrcHV6ubpsCZfd9Zvitwlt/U5sbdc9k/GRJiI0lPiqEkP4X0\n5BjSkvw/6UmxpCfHkJoYQ1zM4Kph0O+GZdHSfOW/u2uRSiYrsm9lRfatdHu6Od58KtCSKGN/zRH2\n1/j/fpOiEmnrbe9/fEfxbSzNX0BG4BLYc+dG7v9otP6tDHS5pBbMpLAO+A7wjIjMBWqMMX2lWA5M\nE5E4Y0wXcAvwloh8EzhjjHleRGbibzVc/rc7xB05cpjmZhfvvvsWAG53NwArV67mySe/zJo1d7J2\n7Z12hjiqeH1eGroaqe2op7bjrL/bp6OO+s4GPNbgX5MoZxS5fRV/QrY/CSTkkBmXPirO+Ht6vRwt\nb+J0bWt/Aujr3unpvfSVOgCJcVFkp8YFKvkY0pJj/f8OqPhjokNrVs6BYiNjmZU5nVmZ0wFo6nZR\nGuhqOtF8mimpJSwrWMRNWTMu27WkbkxQ5z4SkX8AlgM+4CvAHKDFGPN7EflT4HHAA2w3xnxTRAqB\n/wac+BPWXxljdl/pM0Jx7qP9+/fyyisv4vF4ePTRx5g5c/ZF+1RUlLNx43q2bNnET37yK5588st8\n7WvfZOLEyy/eE2yhcvbT6/PQ0NlIbf9Zfx21nfXUdzZcNHd9dEQ0efE55PZX/NnkJeSSHpt6Q5W/\nHWXR3ePh8Mlz7DMNHD55LjAVw3mJcVGkJ/n76/sr+eQY0pL8FX9qUkzQpmEOld+NUBEO5WHL3EfG\nmKcu2HRowGvPAM9csH8VsCqYMY2Evqmzp0+fyQcfbGLmzNmcPn2KXbu2c++9H+ell57n8ce/wOOP\nf4GDBw/Q2dlxyem2xxJXdzPba3azv/4w9V2NF1X+sRGxFCcV+s/+AwkgLyGH1JiUUXHmfzmd3R4O\nnWhkr6nnw9NN9Hr83zs7NY55ksWMCelkpviTQNQNjJ0pNVTa/gqCgVNn19Wd5ctffgKfz8eTT36D\nxMREmptdfOELnyUuLp6ZM2eTnJzCzTfP5Vvf+p98//s/ZOLEEru/wojwWT4+OmfYWrOTDxtLsbCI\njohmfHIReQnZ5CbkkBfo90+JTg7ZyzavVVtnDweON7LPNPBReVP/uFZ+ZgLzpmQxT7IYl50YNt9X\njS46dfYobwIOp5Eqj2Z3Cztq9rCtZjcudzMARUmFLCtYyLzsm0PiBq3hLouWdjf7yxrYaxowlc34\nAn93RdmJzJMs5kk2+Zmhe3OV/q0MFg7loVNnK1v5LB/Hmo6zrXonR84dw2f5iImIZml++E5Idq6l\nm31lDewz9Zyoaum/r3lifrI/EUzJIjvt2m/WUiqYNCmooGpxt7Gjdg/ba3ZxrtsFwLjEfJYWLGJ+\nzs2jYl7+a1Hn6mSf8SeC07X+M0kHMLkwhXmSzTzJIj05vL6zCi+aFNSw81k+jOsEW6t3cbhvmmJn\nFEvy5rOsYBFFSYVh1V9e3djBPlPP3tIGqhr819E7HQ6mj09jnmQzd3ImKYn2d4kpNRSaFNSwaetp\nZ0etf6ygsescAAWJeSzLXxiYpvjKUyiPFpZlUVnXzr6yevaZBmrP+W/kinA6mF2SwTzJYs7krFGz\nULtSA2lSUDfEsizKXCfZWrOTQw1H8VpeopxRLMq9hWUFCxmfXBQWrQKfZXG6tpV9pQ3sK6unodl/\nI2J0pJO5gSuGbirJJD5W/6TU6Ka/weq6tPd0sPPsXrZV76K+y794SV5CDsvyF7Egd+5VF1YZDXw+\ni+NVzf4xgrIGXG3+2VFjoiNYMC2bWySbWRMzQvoOYaWulSYFNWSWZXGi+RRba3ZxsP4IHsvbv3jJ\nsvxFTEwJj8VLqhs7eHHzKbYfrqG1wz9xXnxMJEtm5jJPspg5IV1vJFNhS5OCuqqO3k52nd3H1upd\n1HX658HPic9mWcFCFubOu6458ENRZ3cvr249zcZ91fgsi6T4KJbflM8tksXU4rRLrvKlVLjRpKAu\nybIsTraUs61mF/vrD+PxeYh0RHBLzs0sy1/EpNQJYdEqAH830dYjtfxu00nau3rJTo3jiY/PYkJ2\nPBEjOBWzUqFAk4IapLO3k11n97OtZhe1HXUAZMdnsjR/YVguaXiiqoXfbCij4mwbMVERPLBiImvn\nF5GflzLq71hV6npoUlAAlLdW8tKpfWyr3Euvr5cIRwTzsm9iWcFCJqeWhE2roI+rzc3vNp1gx1F/\n4ls8I4cHV04iLUnvJ1BjmyaFMa6jt5OXj7/OrrP+lVMz4zJYlr+QRXm3kBSdaHN0w6/X42P93jO8\nvq0cd6+X4pwkPr1mMpMLU+0OTamQoElhDDtQf4QXyn5PW08745IK+NzcB8hx5o/qqaiv5NCJRp5/\n7zj1ri4S46J4ePUkbp2dj9MZXq0gpW6EJoUxqMXdyotlr3Kw4UOinJF8vORubht3K7k5qWHZj362\nqZPfvnecwyfP4XQ4uP2WQj62bAIJsXrHsVIX0qQwhliWxc7avbx84g26PF2UpEzgM9MeJCc+y+7Q\ngqLL7eH17eWs33MGr89iWnEan759MgVZ4dctptRw0aQwRpzrauK50pcpdR0nJiKah6bcz7KChWHZ\nVeSzLHZ8eJbfbTpJS0cPGcmxPLx6EnOnZIXdgLlSw02TQpjzWT42V23ntVPv0OPtYXqG8Ih8gvTY\nNLtDC4rTta08t76MkzWtREc6+fitE7hzQRHRQVq7WKlwo0khjJ3tqOM3pb/jVEsFCZHxPDL9E8zP\nmROWZ8utHT28vPkkWw/XYgHzp2bzqVWTyEjRtQuUuhaaFMKQ1+dlfeUm3j69AY/lZW72bD415eNh\neYmpx+tj4/5q/rD1FF1uL4VZCXz69ilMLQ7PlpBSwaZJIcxUtlXx7LGXqG6vJSU6iYfkfm7Kmml3\nWEFx9HQTz20oo/ZcJwmxkXxmzRRWzsnXqSmUugGaFMJEj7eXt06v570zH+CzfCzJm8/9k+4Niyms\nL1Tf3MUL7x3nwPFGHMDKOQXcf+sEkuKj7Q5NqVFPk0IYONF8mt+UvkR9ZyMZsel8euoDTE2fbHdY\nw87d4+XNneW8s+sMHq+PKYUpfHrNFIpykuwOTamwoUlhFOvydPPaybf5oHoHDhysGreM+ybeSUxE\neJ0xW5bF7mP1vPj+CVxtbtKSYvjUqkksmJYdloPmStlJk8IodfRcKc+XvoLL3UxuQg6PTn2QCSnF\ndoc17Crr2nhufRllVS1ERji4d0kx9ywar6udKRUkmhRGmfbeDl4+/jq7z+7H6XBy1/jV3DF+NVHO\n8PqvbO/q5fcfnGLTwWosC+ZMzuSh2yaRnRYeC/ooFarCqyYJY5Zlsb/+MC+WvUp7bwdFSQU8Ou1T\nFCTm2R3asPL6fGw6UMOrW07R0e0hLyOeR26fzMwJGXaHptSYoElhFGh2t/CieZVDjUcHTWAX4Qyv\nLpTys638/M1SqhraiYuJ4OHbJnHbvEJdBlOpEaRJIYRZlsWO2j28cuINujzdTEqdwGemPkh2mE1g\n57Ms3t1dySubT+H1WSybnccDK0pISQivAXOlRgNNCiGqsescz5W+jHGdIDYihoflEyzNXxB2E9i5\n2tz89I2POFbhIiUhmifunc6MCel2h6XUmKVJIcT4LB+bqrbx+sl36PH1MjNjKg/LJ0iLDb+VwQ4c\nb+AXb5XS3tXLTSUZPH7PNJL1BjSlbBXUpCAiTwOLAAv4qjFmz4DXvgI8CniBvcaYJ0UkCvglUBzY\n/rgx5lQwYwwltR11/ObYS5xurSQhKp5PT32QW3JuDrtr8Xt6vbzw/gne319NZISTz6yZwm1zC8Lu\neyo1GgUtKYjICmCyMWaxiEwDfg4sDryWDPwPYJIxxiMi60RkESBAszHmMyKyFvg+8FCwYgwVHp+H\n9RWbeKf8PTyWl3nZN/HJKR8LywnsztS388xrR6lp7KAgK4E//aMZFOqiN0qFjGC2FFYDrwIYY46J\nSJqIJBtjWoGewE+iiLQD8UBT4JhfB47fgD+RhLXK1iqeLe2bwC6Zh+V+ZmfNsDusYWdZFu/tq+LF\n90/i8fpYPbeQT64q0XUOlAoxwRy1zAUaBjxvCGzDGNMNfAc4BVQAu4wxZQOPMcb4AEtEwraTubWn\njacP/CfV7bUszV/Atxd9PSwTQmtHDz/+3WGe23Cc2OgI/vLB2Xxm7RRNCEqFoJEcaO7vMA50H/01\nMAVoBTaKyE1XOuZy0tLiiYy8/solK8u+ydTWH36PHm8Pn735Ae6V222LY6DhLo/9pfU8/dv9NLe5\nuXlKFn/1yFzSk0fHwjd2/m6EIi2PwcK1PIKZFGoItAwC8oHawONpwCljTCOAiGwB5g045lBg0Nlh\njOm50oe4XJ3XHWBWVhINDW3XffyN6PJ08U7ZZpKiEpmTMte2OAYazvLo9fh4efNJ1u05Q4TTwadW\nTWLtgnF43b00NPQOy2cEk52/G6FIy2OwcCiPyyW1YHYfrQMeBBCRuUCNMaavFMuBaSLSN9n/LcDx\nwDGfDGy7D3g/iPHZakvVTrq93awat4zoiCi7wxlWtec6+N6v97Juzxly0uP51mdv4c6FRTj16iKl\nQl7QWgrGmO0isk9EtgM+4Csi8hjQYoz5vYj8M/C+iHiA7caYLSISAawRka2AG3gsWPHZqcfby8Yz\nW4iNiGV54WK7wxk2lmXxwaEant9wnB6Pj+U35fHI6ik6o6lSo0hQxxSMMU9dsOnQgNeeAZ65YH8v\n8HgwYwoFO2v30tbbztriVcRFhsfKaO1dvfzy7VL2lzWQEBvJE/dO55ap2XaHpZS6RnpH8wjz+rxs\nqNxMlDOSVeOW2R3OsDhW4eKnb3yEq82NjEvlC/dNHzWDyUqpwTQpjLB99Yc4193E8oLFJEeP7qsX\nPF4ff9h6mrd2VOBwOLh/+UTuWVSM06ljB0qNVpoURpDP8rG+YhNOh5PVRSvsDueG1Ls6eea1jzhd\n20pWaixfvG8GJQUpdoellLpBmhRG0NFzpdR0nGV+zhwy40bnTKCWZbH9w7M8u74Md4+XxTNyeXTt\nFOJi9FdJqXCgf8kjxLIs3i33X2G7tniVzdFcn85uD/+9zrDrozpioyP4wn3TWTwj9+oHKqVGDU0K\nI+RE8ylOt1YwK3Ma+YmjryI9UdXCT14/SmNLNyX5yXzhj2aQnRoeV04ppc7TpDBC1lVsAmBt8W32\nBnKNvD4fb26v4LVt5VhY3LfyMOahAAAaIElEQVRkPPctHa9LZCoVpjQpjIAzbdV81GSYnDqRiSnF\ndoczZI0tXfzX6x9xvKqF9OQYvnDvdKQoze6wlFJBpElhBKyr8I8lrBlFYwm7j9Xxq3cMXW4Pt0gW\nn7trKgmx4TUdh1LqYpoUgqy+s4ED9UcYl5jP9PQpdodzVd09Hp5bf5ytR2qJjnLy+F1TWTY7T1dF\nU2qM0KQQZOsrNmNhsaZ4VchXrMfPuPjHX+2hztVFcU4SX/yj6eRlJNgdllJqBGlSCKJmdwu7zu4j\nOy6TOdmz7A7nsizL4p3dlbyy+RRen8WdC4v4xPKJOpis1BikSSGINlZuwWt5ub14BU5H6Faw2z88\ny0vvnyQ9OYbH757GjPGj88Y6pdSN06QQJB29nWyp2UlKdDILcufZHc5ltXb08Nv3jhMTFcE//cVy\nnF6v3SEppWwUuqevo9zmqm30eHtYXbScKGfo5t7nNpTR0e3hgRUTyUmPtzscpZTNNCkEgdvbw6aq\nbcRHxrE0f4Hd4VzWweON7D5WT0lBMrfNLbQ7HKVUCBhSUhCR0L5sJsRsq9lFR28nKwqXEhsZmusK\ndLn98xhFOB08dudUne5aKQUMvaVQISJ/LyITgxpNGPD4PLxX+QHRzihWjltqdziX9btNJ3G1ubl3\nyXgKshLtDkcpFSKG2tm9AHgQ+LmI9AK/AH5njOkJWmSj1O6zB2h2t7Bq3DISo0LzGv+yM828f6Ca\ngswE7lk8eqbdUEoF35BaCsaYs8aYfzPGrAT+LPBTG2g9hGb/iA18lo8NlZuIcESwetxyu8O5pF6P\nl1+8XYoDeOyuqXovglJqkCHXCCKyXER+DrwNbAOWAc3AS0GKbdQ51HCUus4GFuTOJS021e5wLun1\n7eXUNXWyel6hrpSmlLrIkLqPROQEUA78BPhTY0xv4KVjIvLxIMU2qliWxbqKjThwsCZEl9o8U9/O\n2zsryUiO5RMrdHhIKXWxoY4p3Ak4jDHHAURkjjHmQOC1W4MS2ShT6jpOZVs1N2fNIich2+5wLuL1\n+fjFW8fw+iw+d6cQGx26904opewz1O6jx4D/NeD5UyLyDwDGGGu4gxqN1gWW2rwjRKfHXr+nivKz\nbSyekcvMiRl2h6OUClFDTQqrjDGf73tijHkI/5iCAk63VFLWfJKpaZMpSg69m8Dqm7t4dcspkuKj\neOT2yXaHo5QKYUNNCtEiEt33REQSAV1xJaBvEZ07xodeK8GyLH79Tik9Hh+P3D6ZxDj9b1NKXd5Q\nO5b/E/+g8l4gApgP/F2wghpNajvqONx4lPHJRUxOLbE7nItsPVLLR+UuZpdksHBajt3hKKVC3JCS\ngjHmZyKyHn8ysIC/AlqDGdhosb5iEwBri1eG3CI6Le1uXnjvBLHREXz2Dgm5+JRSoeda7lxKBBqA\nRmAqsDMoEY0i57pc7Kk7QG5CDrMyp9sdzkV+s76MTreHB1eWkJ6s9xgqpa5uqPcp/BhYC+QCJ4AS\n4AdBjGtUeO/MZnyWj7VFK0NuEZ39ZQ3sNQ1MKkxh5ZwCu8NRSo0SQ63JFhhjpgEHjTHzgTXAmJ58\nv62nne01u0mPTeOWnJvtDmeQzm4Pz64zREY4ePyuqTi120gpNURDTQruwL8xIuIwxuwDQncK0BHw\n/pmt9Po8rC5aToQzwu5wBnlp0wma23u4b8l48jJCc1I+pVRoGurVR0ZEvgx8AKwXEQNcdXIfEXka\nWIR/cPqrxpg9ge0FwG8G7DoReAqIBr4LnAxsX2+M+d4QYxwxXZ5uPqjeTmJUAkvy5tsdziCm0sXm\ngzUUZiVw1yKdAVUpdW2GmhS+BKThnwDvYSAH+P6VDhCRFcBkY8xiEZkG/BxYDGCMqQZWBvaLBDYB\nr+GfnvsFY8w3rvWLjKSt1Tvp8nRz38Q7iI6IvvoBI6Sn18sv3y7F4YDH7pqmM6Aqpa7ZUJPC08aY\nJwOPnxviMauBVwGMMcdEJE1Eko0xF17K+hjwsjGmXUSG+Nb26fX2svHMFmIjYlhesMTucAZ5bVs5\nda4u1s4fx8T8ZLvDUUqNQkNNCl4RuQ3YDvQvrGOM8V3hmFxg34DnDYFtFyaFJ/Bf2dRnhYi8g/+O\n6W8MmHgvJOw8u5fWnjbWFK0kPirO7nD6VZxt451dlWSmxHL/rToDqlLq+gw1KTwBPAkMvIzFwn93\n81BddAmMiCwGSge0HnYCDcaYNwOv/RqYdaU3TUuLJzLy+gd6s7KShryv1+dl464PiHJG8smb7yQ1\nbujHBpPX6+N7z+7DZ1n85UNzKCy4/rUcrqU8wp2WxWBaHoOFa3kM9Y7m61mNpQZ/y6BPPlB7wT73\nAhsGfE4pUBp4vENEskQkwhjjvdyHuFyd1xGaX1ZWEg0NbUPef8/ZA9R3nGNZwSJ62500tA/92GB6\ne2cFJ6taWDozl8L0uGv6TgNda3mEMy2LwbQ8BguH8rhcUhvqzWv/51LbjTF/e4XD1gHfAZ4RkblA\njTHmwlKcD/x2wOd8EzhjjHleRGbibzVcNiGMJP8iOu+H3CI6da5OXt16muT4KB5arTOgKqVuzFAv\nT/EO+IkAVgFXbD0YY7YD+0RkO/AvwFdE5DERuX/AbnlA/YDnzwFfFJHNwDPAnwwxvqD78NwxajrO\nMi/nJjLjQmM9Asuy+NXbpfR6fHx6zRSdAVUpdcOG2n30nYHPRSQCeHkIxz11waZDF7w+64LnVfgT\nTshZ1z/xXeiEt+VwLaWVzdw8KZP5U0NvtTel1OhzvReyRwGThjOQUHai+TSnWsqZmTGVgsQ8u8MB\nwNXm5oWNJ4iLieCPdQZUpdQwGeqYwhn8Vxv1SQd+GYyAQtG7FRsBWFt8m82RnPfc+jK63B7++A4h\nLSnG7nCUUmFiqJekDlx60wJajTHNQYgn5Jxpq+Gjc4aSlAmUpI63OxwA9pl69pU1MGVcKituzrc7\nHKVUGBlq91EC8CVjTIUxphJ4WkRmBDGukLE+xJba7Oju5dl1ZURGOHlMZ0BVSg2zoSaFfwfeGvD8\nZ4FtYa2+s5H99YcpSMxjenpoTMHx4sYTtHT08LFl48lNH9OzlyulgmCoSSHSGLOl74kxZiuXuEM5\n3Gyo3ISFxdriVSExkHusvIkth2sZl53IHQuK7A5HKRWGhjqm0CIif4Z/NlMncCcwum/nu4pmdwu7\naveRGZfBnKwrzrQxIty9Xn71jgnMgDpVZ0BVSgXFUGuWx4F5wIvA8/gvR308WEGFgo1ntuCxvKwp\nWhESi+j8Yetp6pu7uGN+ERPydAZUpVRwDCkpGGMagH80xswyxswGfhLYFpY6ezvZWr2TlOgkFubd\nYnc4lJ9t5d3dlWSlxvKxWyfYHY5SKowNKSmIyPeA/zVg01Mi8g/BCcl+m6u24/b2cFvRcqKcQ+1h\nCw6P18cv3irFsuBzd04lJsr+VotSKnwNtftopTHm831PjDEPMfjehbDh9vbwftVW4iLjWJa/0O5w\neHd3JWfq21k2O4/p49PtDkcpFeaGmhSiRaR/3UkRScQ/1UXY2V6zm47eTlYULiE2MtbWWM42dfKH\nreWkJETz0G1jZlYRpZSNhto38p/AMRHZi3+W1PnAj4IWlU08Pg/vVX5AlDOKlYVLbY3FZ1n88u1S\nPF4fn1kzhYTYsMzBSqkQM9RZUn8mIseBTPzTXLyGf4zh6SDGNuL21B3E5W5mZeFSkqITbY3lg4M1\nlJ1pZs7kTOZJlq2xKKXGjqFOiPcj4A78K6mdAEqAHwQxrhHns3ysr9iE0+FkddFyW2Nxtbl5adMJ\n4mIieXStzoCqlBo5Qx1TWGiMmQYcNMbMB9YAYTXHwuGGo9R11rMgZy7psWm2xWFZFv/9rqHL7eVT\nq0p0BlSl1IgaalJwB/6NERGHMWYfYG+n+zCyLIt3+5baLLZ3qc29poGDJxqZWpTK8pt0BlSl1Mga\n6kCzEZEvAx8A60XEAKnBC2tkGdcJKtuquClrJrkJObbF0d7Vy2/WGaIinXzuzqnabaSUGnFDTQpf\nAtKAZuBhIAf4frCCGmnv9k2PbfNSmy9sPE5rZy8PriwhR2dAVUrZYKhXH1lAU+Dpc8ELZ+SVt1ZS\n5jqBpE2iOHmcbXEcLW9i25GzFOUkcscC++JQSo1tY36qzXUVmwBYa2Mrwd3j5Vdvl+J0OHj8rmlE\nOMf8f4tSyiZjuvapaq3lUMOHFCeNQ9Lsu2P491tO0djSzR0Lx1Gcm2RbHEopNaaTwh+OrQNg7Xj7\nFtE5XdvK+r1nyE6L42NLdQZUpZS9xmxSaOp2sbViNznx2czOnG5LDD6f1T8D6mN3TiVaZ0BVStls\nzCaF9yo/wGv5WFO8EqfDnmLYa+qpamhn6cxcphbbd8OcUkr1GbNJ4Vy3i/ykHObn3GzL5/ssize2\nl+NwwH1Lx9sSg1JKXcjeFWRs9MTMR8nISKDF5b76zkFw6EQjVQ0dLJ6RQ3aa3pOglAoNY7alEOmM\nJDoy+uo7BoEVaCUA3L14vC0xKKXUpYzZpGCno+VNnK5tY55kUZCZYHc4SinVT5OCDd7YVg7AvdpK\nUEqFGE0KI8xUuiiramF2SYbeqKaUCjmaFEbYGzsqALh3yXh7A1FKqUsI6tVHIvI0sAj/Ep5fNcbs\nCWwvAH4zYNeJwFPAS8AvgWLACzxujDkVzBhH0qmaVo6ebmJacRqTClLsDkcppS4StJaCiKwAJhtj\nFgN/AvxL32vGmGpjzEpjzErgdqAS/7rPnwaajTHLgO8RRtNzA/1XHN2nrQSlVIgKZvfRauBVAGPM\nMSBNRJIvsd9jwMvGmPbAMb8PbN9AGK3uVlnXxsETjUwqTEGKwmZ9IqVUmAlmUsgFGgY8bwhsu9AT\nwM8uPMYY4wMsEbHnZoJh9mZgLOG+JeN1RTWlVMgayTuaL6oJRWQxUGqMaR3qMRdKS4snMvL6J5LL\nygr+FUBn6trYa+qZVJjCqgXFIZ0URqI8Rgsti8G0PAYL1/IIZlKoYXDLIB+ovWCfe/F3E114zCER\niQIcxpieK32Iy9V53QFmZSXR0NB23ccP1bNvfYRlwR3zi2hsbA/6512vkSqP0UDLYjAtj8HCoTwu\nl9SC2X20DngQQETmAjXGmAtLcT5w6IJjPhl4fB/wfhDjGxH1zV3sPFpHQWYCc6Zk2h2OUkpdUdBa\nCsaY7SKyT0S2Az7gKyLyGNBijOkbTM4D6gcc9gKwRkS2Am78g9Cj2ts7K/BZFvcsKcYZwt1GSikF\nQR5TMMY8dcGmQxe8PuuC517g8WDGNJKaWrvZdqSWnLQ4FkzNsTscpZS6Kr2jOYje2V2Jx2tx9+Ji\nnE5tJSilQp8mhSBp6ejhg4M1ZCTHsHjGpa7EVUqp0KNJIUjW7amkx+Pj7kXFREZoMSulRgetrYKg\nvauXjfurSUmMZtnsPLvDUUqpIdOkEAQb9p7B3ePlzgVFRN3AjXVKKTXSNCkMsy63hw17q0iMi2Ll\nzQV2h6OUUtdEk8Iw27i/ik63h7XzxxETra0EpdTooklhGLl7vLy7+wxxMZHcNrfQ7nCUUuqaaVIY\nRpsP1dDe1cvt8wqJjx3JuQaVUmp4aFIYJr0eH+/sqiAmKoI188fZHY5SSl0XTQrDZNuRWprbe1g1\nt4DEuCi7w1FKqeuiSWEYeLw+3tpZQVSkkzu0laCUGsU0KQyDXR/V0djSzfLZ+aQkxtgdjlJKXTdN\nCjfI57N4Y0cFEU4Hdy4ssjscpZS6IZoUbtBeU09dUydLZ+WSkRJrdzhKKXVDNCncAJ9l8fr2chwO\nuHtRsd3hKKXUDdOkcAMOHW+kuqGDRdNzyE6LtzscpZS6YZoUrpPV10oA7l483u5wlFJqWGhSuE5H\nTzdRfraNeZJFQWaC3eEopdSw0KRwnd7YXg7AvUvG2xqHUkoNJ00K18FUuiiramF2SQZFOUl2h6OU\nUsNGk8J10FaCUipcaVK4RidrWjha7mJacRqTClLsDkcppYaVJoVr9Ob2CgDu01aCUioMaVK4BpV1\nbRw80cikwhSkKNXucJRSathpUrgGb+w430pwOBw2R6OUUsNPk8IQ1TR2sK+0nuLcJGZOSLc7HKWU\nCgpNCkP05o4KLLSVoJQKb5oUhqC+uYtdH9VRkJXAzZMz7Q5HKaWCRpPCELy9swKfZXHP4mKc2kpQ\nSoUxTQpX0dTazdbDteSkxbFgao7d4SilVFBpUriKd3ZV4vVZ3L24GKdTWwlKqfAWGcw3F5GngUWA\nBXzVGLNnwGvjgOeBaGC/MeZLIrISeAk4GtjtiDHmL4IZ45W0dPSw+VANGckxLJ6Ra1cYSik1YoKW\nFERkBTDZGLNYRKYBPwcWD9jlh8APjTG/F5F/F5G+BY43G2MeDFZc12Ld7kp6PT7uXlRMZIQ2qpRS\n4S+YNd1q4FUAY8wxIE1EkgFExAncCrwWeP0rxpjKIMZyzdq7etl4oJqUxGiWzc6zOxyllBoRwUwK\nuUDDgOcNgW0AWUAb8LSIbBWR7w/Yb7qIvBbYviaI8V3Rhr1ncPd4uWtBEVGREXaFoZRSIyqoYwoX\ncFzwuAD4MVAOvCki9wAHge8ALwITgfdFZJIxpudyb5qWFk/kDVTaWVkXr4fQ0dXLe/urSU6I5oHb\nhdiYkSwme12qPMYqLYvBtDwGC9fyCGZtV8P5lgFAPlAbeNwIVBhjTgKIyHvADGPMm8ALgX1OishZ\n/Mnj9OU+xOXqvO4As7KSaGhou2j7mzvK6ejq5RPLJ9LW2sXFe4Sny5XHWKRlMZiWx2DhUB6XS2rB\n7D5aBzwIICJzgRpjTBuAMcYDnBKRyYF95wFGRD4jIt8IHJML5ADVQYzxIu4eL+/uPkNcTCS3zS0c\nyY9WSinbBa2lYIzZLiL7RGQ74AO+IiKPAS3GmN8DTwK/DAw6HwFeBxKA50TkY/gvVf2zK3UdBcPm\nQzW0d/Vy35LxxMeOnW4jpZSCII8pGGOeumDToQGvnQCWXfB6G3BfMGO6kl6Pl3d2VRATFcGa+ePs\nCkMppWyjF98PsPXIWZrbe1g1t4DEuCi7w1FKqRGnSSHA4/Xx1o4KoiKd3KGtBKXUGKVJIWDn0TrO\ntXaz/KZ8UhJj7A5HKaVsoUkB8Pks3txRToTTwV0Li666v1JKhStNCsCe0nrqXF0snZVHenKs3eEo\npZRtxnxS8FkWb+wox+GAuxdpK0EpNbaN+aRw8Hgj1Q0dLJqeQ3ZavN3hKKWUrcZ0UrAsi9e3l+MA\n7l483u5wlFLKdmM6KRwwDVScbWOeZFGQmWB3OEopZbsxmxQsy+K36w0A9y4Zb28wSikVIsZsUig7\n08yx8iZuKsmgKCc8p8BVSqlrNWaTwro9ZwBtJSil1EBjdhrQKeNSKRmXRklBit2hKKVUyBizSeGO\nBUVhsVCGUkoNpzHbfaSUUupimhSUUkr106SglFKqnyYFpZRS/TQpKKWU6qdJQSmlVD9NCkoppfpp\nUlBKKdXPYVmW3TEopZQKEdpSUEop1U+TglJKqX6aFJRSSvXTpKCUUqqfJgWllFL9NCkopZTqp0lB\nKaVUvzG7yI6IPA0sAizgq8aYPTaHZBsR+SfgVvy/D983xrxic0i2E5E44EPgu8aYX9ocjq1E5DPA\nNwEP8LfGmDdtDskWIpII/BpIA2KA7xhj3rU3quE3JlsKIrICmGyMWQz8CfAvNodkGxFZBcwMlMWd\nwI9sDilUfAtosjsIu4lIBvC/gWXAvcDH7I3IVo8BxhizCngQ+LG94QTHmEwKwGrgVQBjzDEgTUSS\n7Q3JNh8Anww8bgYSRCTCxnhsJyJTgenAmDwjvsDtwAZjTJsxptYY80W7A7JRI5AReJwWeB52xmpS\nyAUaBjxvCGwbc4wxXmNMR+DpnwBvGWO8dsYUAn4IfM3uIELEeCBeRF4TkS0istrugOxijPktUCQi\nJ/CfTH3D5pCCYqwmhQs57A7AbiLyMfxJ4c/tjsVOIvJZYIcx5rTdsYQIB/6z40/g7z75hYiMyb8X\nEXkUqDTGTAJuA/7N5pCCYqwmhRoGtwzygVqbYrGdiNwB/A1wlzGmxe54bHYP8DER2Qk8AXxbRG63\nOSY71QHbjTEeY8xJoA3IsjkmuywF3gUwxhwC8sOxq3WsXn20DvgO8IyIzAVqjDFtNsdkCxFJAf4Z\nuN0YM+YHVo0xD/U9FpG/A8qNMRvsi8h264Bfisg/4u9HTyRM+9KH4ASwEHhZRIqB9nDsah2TScEY\ns11E9onIdsAHfMXumGz0EJAJvCgifds+a4yptC8kFSqMMdUi8jtgZ2DTXxhjfHbGZKNngJ+LyGb8\ndeeXbI4nKHQ9BaWUUv3G6piCUkqpS9CkoJRSqp8mBaWUUv00KSillOqnSUEppVQ/TQpK2UhEHhOR\nZ+2OQ6k+mhSUUkr10/sUlBoCEfkL4FP4b1oqBf4JeAN4G7gpsNvDgZu97gH+FugM/HwxsH0h/qnJ\ne/BPy/1Z4AH88wq14p+ZtQL4hDFG/zCVLbSloNRViMgC4H5geWDdiWb8U0pPBH5hjLkV2AR8XUTi\ngZ8CDwTm3X8b+PvAWz0LfMEYswLYjH+eJYAZwBeBecBMYO5IfC+lLmVMTnOh1DVaCUwC3g9MBZIA\nFADnjDH7AvtsA54EpgB1xpiqwPZNwJdEJBNINcZ8CGCM+RH4xxSAPcaYzsDzaiA1+F9JqUvTpKDU\n1bmB14wx/dOKi8h4YP+AfRz4l3a9sNtn4PbLtcw9lzhGKVto95FSV7cNuCuwRi8i8mUgD/+KfXMC\n+ywDDgNlQLaIFAW23w7sNMacAxpFZH7gPb4eeB+lQoomBaWuwhizF/h3YJOIbMXfndQCVAOPichG\n/HPtP22M6cK/WNELIrIJ/9Kv3wq81R8DPw7Msrkc/xiDUiFFrz5S6joEuo+2GmMK7Y5FqeGkLQWl\nlFL9tKWglFKqn7YUlFJK9dOkoJRSqp8mBaWUUv00KSillOqnSUEppVS//w9xvhDmSUhd0AAAAABJ\nRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEVCAYAAAAPRfkLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nBSGEEP3+f0jydKzqgafeAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "text",
        "id": "bd3qOrYN4Bvz"
      },
      "cell_type": "markdown",
      "source": [
        "### Pretrained fasttext"
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "b0039303-e8b6-4af1-c560-673452642c21",
        "id": "J5SeYpnn4Bwk",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        }
      },
      "cell_type": "code",
      "source": [
        "from gensim.models.wrappers import FastText\n",
        "\n",
        "%time word_vectors = FastText.load_fasttext_format('wiki.simple')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 23.9 s, sys: 1.58 s, total: 25.5 s\n",
            "Wall time: 39 s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "id": "XM9XpRkL4Bw3",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "EMBEDDING_DIM = 300\n",
        "VOCAB_SIZE = min(len(word_index)+1, NUM_WORDS)\n",
        "embedding_matrix = np.zeros((VOCAB_SIZE, EMBEDDING_DIM))\n",
        "\n",
        "for word, i in word_index.items():\n",
        "  \n",
        "    if i>=NUM_WORDS:\n",
        "        continue\n",
        "    try:\n",
        "        embedding_vector = word_vectors[word]\n",
        "        embedding_matrix[i] = embedding_vector\n",
        "        \n",
        "    except KeyError:\n",
        "        embedding_matrix[i]=np.random.normal(0,np.sqrt(0.25), EMBEDDING_DIM)\n",
        "\n",
        "del(word_vectors)\n",
        "\n",
        "# load pre-trained word embeddings into an Embedding layer\n",
        "# note that we set trainable = False so as to keep the embeddings fixed\n",
        "embedding_layer = Embedding(VOCAB_SIZE,\n",
        "                            EMBEDDING_DIM,\n",
        "                            embeddings_initializer=Constant(embedding_matrix),\n",
        "                            input_length=MAX_SEQUENCE_LENGTH,\n",
        "                            trainable=False)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "0b994aa7-16a9-4871-c465-07e28e006fbd",
        "id": "faPFIxnZ4BxC",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        }
      },
      "cell_type": "code",
      "source": [
        "labels = data['sentiment'].values\n",
        "\n",
        "X, Y = sentences, labels\n",
        "\n",
        "train_x, val_x, train_y, val_y = train_test_split(X, Y, test_size=0.2, stratify=Y)\n",
        "\n",
        "train_y = to_categorical(train_y, num_classes=2)\n",
        "val_y = to_categorical(val_y, num_classes=2)\n",
        "\n",
        "print ('Training:', train_x.shape, train_y.shape)\n",
        "print ('Validation:', val_x.shape, val_y.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Training: (20000, 200) (20000, 2)\n",
            "Validation: (5000, 200) (5000, 2)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "8ca92722-e68a-4c95-93f8-0f84873f114e",
        "id": "UjokcuoI4BxN",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 285
        }
      },
      "cell_type": "code",
      "source": [
        "num_classes = 2\n",
        "\n",
        "# input layer with shape = (MAX_SEQUENCE_LENGTH,)\n",
        "sequence_input = Input(shape=(MAX_SEQUENCE_LENGTH,), dtype='int32')\n",
        "# convert the sequence input to embedding from (MAX_SEQUENCE_LENGTH, ) -> (VOCAB_SIZE, EMBEDDING_DIM)\n",
        "embedded_sequences = embedding_layer(sequence_input)\n",
        "\n",
        "x = LSTM(256, dropout=0.2, recurrent_dropout=0.2)(embedded_sequences)\n",
        "preds = Dense(num_classes, activation='softmax')(x)\n",
        "\n",
        "model = Model(inputs=sequence_input, outputs=preds)\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer='adam',\n",
        "              metrics=['acc'])\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_3 (InputLayer)         (None, 200)               0         \n",
            "_________________________________________________________________\n",
            "embedding_3 (Embedding)      (None, 200, 300)          6000000   \n",
            "_________________________________________________________________\n",
            "lstm_3 (LSTM)                (None, 256)               570368    \n",
            "_________________________________________________________________\n",
            "dense_3 (Dense)              (None, 2)                 514       \n",
            "=================================================================\n",
            "Total params: 6,570,882\n",
            "Trainable params: 570,882\n",
            "Non-trainable params: 6,000,000\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "584e4775-4e19-465f-9ed8-b84ff60fb660",
        "id": "fuNYZwmL4Bxc",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 392
        }
      },
      "cell_type": "code",
      "source": [
        "es = EarlyStopping(monitor='val_acc', patience=3)\n",
        "\n",
        "history = model.fit(train_x, train_y,\n",
        "                    batch_size=128,\n",
        "                    epochs=10,\n",
        "                    validation_data=(val_x, val_y),\n",
        "                    callbacks =[es])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 20000 samples, validate on 5000 samples\n",
            "Epoch 1/10\n",
            "20000/20000 [==============================] - 75s 4ms/step - loss: 0.6177 - acc: 0.6426 - val_loss: 0.5850 - val_acc: 0.6876\n",
            "Epoch 2/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.5628 - acc: 0.7018 - val_loss: 0.4878 - val_acc: 0.7802\n",
            "Epoch 3/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.5206 - acc: 0.7367 - val_loss: 0.5119 - val_acc: 0.7462\n",
            "Epoch 4/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.4621 - acc: 0.7782 - val_loss: 0.3869 - val_acc: 0.8200\n",
            "Epoch 5/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.4011 - acc: 0.8175 - val_loss: 0.3578 - val_acc: 0.8456\n",
            "Epoch 6/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.3574 - acc: 0.8410 - val_loss: 0.3418 - val_acc: 0.8524\n",
            "Epoch 7/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.3355 - acc: 0.8499 - val_loss: 0.3282 - val_acc: 0.8584\n",
            "Epoch 8/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.3173 - acc: 0.8636 - val_loss: 0.3257 - val_acc: 0.8560\n",
            "Epoch 9/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.2990 - acc: 0.8694 - val_loss: 0.3172 - val_acc: 0.8624\n",
            "Epoch 10/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.2815 - acc: 0.8781 - val_loss: 0.3164 - val_acc: 0.8616\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "18538546-0153-432f-e8ae-6954359d1142",
        "id": "SyxjMmgD4Bxs",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 571
        }
      },
      "cell_type": "code",
      "source": [
        "# summarize history for accuracy\n",
        "plt.plot(history.history['acc'])\n",
        "plt.plot(history.history['val_acc'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "# summarize history for loss\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEVCAYAAAAPRfkLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEVCAYAAAAPRfkLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nJxV9OhvlAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "eBfTvO4Zd2u5",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "### Custom Word2Vec\n",
        "\n",
        "Creating a custom word2vec model i.e. from our own dataset instead of using pretrained model"
      ]
    },
    {
      "metadata": {
        "id": "i07q_-woEnEF",
        "colab_type": "code",
        "outputId": "54ffe839-10c4-40ea-b98e-795b836e72e3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        }
      },
      "cell_type": "code",
      "source": [
        "# BeautifulSoup is used to remove html tags from the text\n",
        "from bs4 import BeautifulSoup \n",
        "import re # For regular expressions\n",
        "\n",
        "# Initializing the train model\n",
        "from gensim.models import word2vec\n",
        "\n",
        "# Stopwords can be useful to undersand the semantics of the sentence.\n",
        "# Therefore stopwords are not removed while creating the word2vec model.\n",
        "# But they will be removed  while averaging feature vectors.\n",
        "from nltk.corpus import stopwords\n",
        "\n",
        "# word2vec expects a list of lists.\n",
        "# Using punkt tokenizer for better splitting of a paragraph into sentences.\n",
        "\n",
        "import nltk\n",
        "nltk.download('punkt')\n",
        "import nltk.data\n",
        "\n",
        "tokenizer = nltk.data.load('tokenizers/punkt/english.pickle')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[nltk_data] Downloading package punkt to /root/nltk_data...\n",
            "[nltk_data]   Unzipping tokenizers/punkt.zip.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "5tG7RqYZeJO6",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# This function converts a text to a sequence of words.\n",
        "def review_wordlist(review, remove_stopwords=False):\n",
        "    \n",
        "    # Removing html tags\n",
        "    review_text = BeautifulSoup(review).get_text()\n",
        "    \n",
        "    # Removing non-letter.\n",
        "    review_text = re.sub(\"[^a-zA-Z]\",\" \",review_text)\n",
        "    \n",
        "    # Converting to lower case and splitting\n",
        "    words = review_text.lower().split()\n",
        "    \n",
        "    # Optionally remove stopwords\n",
        "    if remove_stopwords:\n",
        "        stops = set(stopwords.words(\"english\"))     \n",
        "        words = [w for w in words if not w in stops]\n",
        "    \n",
        "    return(words)\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "s9soj9nreJha",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# This function splits a review into sentences\n",
        "def review_sentences(review, tokenizer, remove_stopwords=True):\n",
        "    \n",
        "    # Using nltk tokenizer\n",
        "    raw_sentences = tokenizer.tokenize(review.strip())\n",
        "    sentences = []\n",
        "    \n",
        "    # Loop for each sentence\n",
        "    for raw_sentence in raw_sentences:\n",
        "        if len(raw_sentence)>0:\n",
        "            sentences.append(review_wordlist(raw_sentence, remove_stopwords))\n",
        "\n",
        "    # This returns the list of lists\n",
        "    return sentences\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "cZ85IkX0eegd",
        "colab_type": "code",
        "outputId": "8bf506ba-41a8-4158-ac6e-b9905d5c55db",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 126
        }
      },
      "cell_type": "code",
      "source": [
        "sentences = []\n",
        "print(\"Parsing sentences from training set\")\n",
        "for review in data[\"review\"]:\n",
        "    sentences += review_sentences(review, tokenizer)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Parsing sentences from training set\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/bs4/__init__.py:273: UserWarning: \"b'.'\" looks like a filename, not markup. You should probably open this file and pass the filehandle into Beautiful Soup.\n",
            "  ' Beautiful Soup.' % markup)\n",
            "/usr/local/lib/python3.6/dist-packages/bs4/__init__.py:336: UserWarning: \"http://www.happierabroad.com\"\" looks like a URL. Beautiful Soup is not an HTTP client. You should probably use an HTTP client like requests to get the document behind the URL, and feed that document to Beautiful Soup.\n",
            "  ' that document to Beautiful Soup.' % decoded_markup\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "metadata": {
        "id": "usvGk0OtOh4Z",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Importing the built-in logging module\n",
        "import logging\n",
        "logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.INFO)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "MFBa_hd0Em2g",
        "colab_type": "code",
        "outputId": "33b47876-4ab7-4024-c58b-4a67a68f7561",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1752
        }
      },
      "cell_type": "code",
      "source": [
        "# Creating the model and setting values for the various parameters\n",
        "num_features = 150  # Word vector dimensionality\n",
        "min_word_count = 40 # Minimum word count\n",
        "num_workers = 4     # Number of parallel threads\n",
        "context = 10        # Context window size\n",
        "downsampling = 1e-3 # (0.001) Downsample setting for frequent words\n",
        "\n",
        "print(\"Training model....\")\n",
        "model = word2vec.Word2Vec(sentences,\n",
        "                          workers=num_workers,\n",
        "                          size=num_features,\n",
        "                          min_count=min_word_count,\n",
        "                          window=context,\n",
        "                          sample=downsampling)\n",
        "\n",
        "# To make the model memory efficient\n",
        "model.init_sims(replace=True)\n",
        "\n",
        "# Saving the model for later use. Can be loaded using Word2Vec.load()\n",
        "model_name = \"150features_40minwords_10context_imdb\"\n",
        "model.save(model_name)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2019-02-26 14:39:56,238 : WARNING : consider setting layer size to a multiple of 4 for greater performance\n",
            "2019-02-26 14:39:56,240 : INFO : collecting all words and their counts\n",
            "2019-02-26 14:39:56,241 : INFO : PROGRESS: at sentence #0, processed 0 words, keeping 0 word types\n",
            "2019-02-26 14:39:56,301 : INFO : PROGRESS: at sentence #10000, processed 114931 words, keeping 17627 word types\n",
            "2019-02-26 14:39:56,343 : INFO : PROGRESS: at sentence #20000, processed 228988 words, keeping 24797 word types\n",
            "2019-02-26 14:39:56,383 : INFO : PROGRESS: at sentence #30000, processed 339533 words, keeping 29883 word types\n",
            "2019-02-26 14:39:56,424 : INFO : PROGRESS: at sentence #40000, processed 453983 words, keeping 34196 word types\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Training model....\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "2019-02-26 14:39:56,467 : INFO : PROGRESS: at sentence #50000, processed 565006 words, keeping 37609 word types\n",
            "2019-02-26 14:39:56,508 : INFO : PROGRESS: at sentence #60000, processed 676637 words, keeping 40571 word types\n",
            "2019-02-26 14:39:56,549 : INFO : PROGRESS: at sentence #70000, processed 789005 words, keeping 43180 word types\n",
            "2019-02-26 14:39:56,591 : INFO : PROGRESS: at sentence #80000, processed 899771 words, keeping 45561 word types\n",
            "2019-02-26 14:39:56,631 : INFO : PROGRESS: at sentence #90000, processed 1013453 words, keeping 47982 word types\n",
            "2019-02-26 14:39:56,672 : INFO : PROGRESS: at sentence #100000, processed 1125135 words, keeping 50054 word types\n",
            "2019-02-26 14:39:56,713 : INFO : PROGRESS: at sentence #110000, processed 1236261 words, keeping 51928 word types\n",
            "2019-02-26 14:39:56,756 : INFO : PROGRESS: at sentence #120000, processed 1348541 words, keeping 53966 word types\n",
            "2019-02-26 14:39:56,797 : INFO : PROGRESS: at sentence #130000, processed 1461911 words, keeping 55694 word types\n",
            "2019-02-26 14:39:56,837 : INFO : PROGRESS: at sentence #140000, processed 1568503 words, keeping 57193 word types\n",
            "2019-02-26 14:39:56,881 : INFO : PROGRESS: at sentence #150000, processed 1682622 words, keeping 58902 word types\n",
            "2019-02-26 14:39:56,922 : INFO : PROGRESS: at sentence #160000, processed 1794988 words, keeping 60464 word types\n",
            "2019-02-26 14:39:56,966 : INFO : PROGRESS: at sentence #170000, processed 1907744 words, keeping 61924 word types\n",
            "2019-02-26 14:39:57,007 : INFO : PROGRESS: at sentence #180000, processed 2018412 words, keeping 63343 word types\n",
            "2019-02-26 14:39:57,050 : INFO : PROGRESS: at sentence #190000, processed 2131820 words, keeping 64641 word types\n",
            "2019-02-26 14:39:57,092 : INFO : PROGRESS: at sentence #200000, processed 2245187 words, keeping 65934 word types\n",
            "2019-02-26 14:39:57,136 : INFO : PROGRESS: at sentence #210000, processed 2357380 words, keeping 67237 word types\n",
            "2019-02-26 14:39:57,178 : INFO : PROGRESS: at sentence #220000, processed 2470883 words, keeping 68544 word types\n",
            "2019-02-26 14:39:57,220 : INFO : PROGRESS: at sentence #230000, processed 2582936 words, keeping 69805 word types\n",
            "2019-02-26 14:39:57,264 : INFO : PROGRESS: at sentence #240000, processed 2697681 words, keeping 71014 word types\n",
            "2019-02-26 14:39:57,305 : INFO : PROGRESS: at sentence #250000, processed 2805851 words, keeping 72198 word types\n",
            "2019-02-26 14:39:57,354 : INFO : PROGRESS: at sentence #260000, processed 2916657 words, keeping 73325 word types\n",
            "2019-02-26 14:39:57,384 : INFO : collected 74065 word types from a corpus of 2988095 raw words and 266551 sentences\n",
            "2019-02-26 14:39:57,385 : INFO : Loading a fresh vocabulary\n",
            "2019-02-26 14:39:57,445 : INFO : effective_min_count=40 retains 8160 unique words (11% of original 74065, drops 65905)\n",
            "2019-02-26 14:39:57,446 : INFO : effective_min_count=40 leaves 2627279 word corpus (87% of original 2988095, drops 360816)\n",
            "2019-02-26 14:39:57,475 : INFO : deleting the raw counts dictionary of 74065 items\n",
            "2019-02-26 14:39:57,478 : INFO : sample=0.001 downsamples 30 most-common words\n",
            "2019-02-26 14:39:57,480 : INFO : downsampling leaves estimated 2494390 word corpus (94.9% of prior 2627279)\n",
            "2019-02-26 14:39:57,506 : INFO : estimated required memory for 8160 words and 150 dimensions: 13872000 bytes\n",
            "2019-02-26 14:39:57,508 : INFO : resetting layer weights\n",
            "2019-02-26 14:39:57,609 : INFO : training model with 4 workers on 8160 vocabulary and 150 features, using sg=0 hs=0 sample=0.001 negative=5 window=10\n",
            "2019-02-26 14:39:58,636 : INFO : EPOCH 1 - PROGRESS: at 16.53% examples, 409219 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:39:59,651 : INFO : EPOCH 1 - PROGRESS: at 33.96% examples, 418120 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:00,661 : INFO : EPOCH 1 - PROGRESS: at 51.81% examples, 425483 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:01,664 : INFO : EPOCH 1 - PROGRESS: at 69.21% examples, 426784 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:02,687 : INFO : EPOCH 1 - PROGRESS: at 86.44% examples, 426283 words/s, in_qsize 6, out_qsize 1\n",
            "2019-02-26 14:40:03,387 : INFO : worker thread finished; awaiting finish of 3 more threads\n",
            "2019-02-26 14:40:03,402 : INFO : worker thread finished; awaiting finish of 2 more threads\n",
            "2019-02-26 14:40:03,403 : INFO : worker thread finished; awaiting finish of 1 more threads\n",
            "2019-02-26 14:40:03,419 : INFO : worker thread finished; awaiting finish of 0 more threads\n",
            "2019-02-26 14:40:03,420 : INFO : EPOCH - 1 : training on 2988095 raw words (2494328 effective words) took 5.8s, 430025 effective words/s\n",
            "2019-02-26 14:40:04,432 : INFO : EPOCH 2 - PROGRESS: at 16.53% examples, 415910 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:05,453 : INFO : EPOCH 2 - PROGRESS: at 33.96% examples, 421616 words/s, in_qsize 8, out_qsize 1\n",
            "2019-02-26 14:40:06,465 : INFO : EPOCH 2 - PROGRESS: at 51.82% examples, 426100 words/s, in_qsize 6, out_qsize 1\n",
            "2019-02-26 14:40:07,482 : INFO : EPOCH 2 - PROGRESS: at 69.21% examples, 426240 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:08,509 : INFO : EPOCH 2 - PROGRESS: at 86.44% examples, 425473 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:09,188 : INFO : worker thread finished; awaiting finish of 3 more threads\n",
            "2019-02-26 14:40:09,203 : INFO : worker thread finished; awaiting finish of 2 more threads\n",
            "2019-02-26 14:40:09,221 : INFO : worker thread finished; awaiting finish of 1 more threads\n",
            "2019-02-26 14:40:09,230 : INFO : worker thread finished; awaiting finish of 0 more threads\n",
            "2019-02-26 14:40:09,231 : INFO : EPOCH - 2 : training on 2988095 raw words (2494528 effective words) took 5.8s, 430141 effective words/s\n",
            "2019-02-26 14:40:10,264 : INFO : EPOCH 3 - PROGRESS: at 16.53% examples, 407553 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:11,315 : INFO : EPOCH 3 - PROGRESS: at 35.33% examples, 426271 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:12,329 : INFO : EPOCH 3 - PROGRESS: at 54.15% examples, 437642 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:13,343 : INFO : EPOCH 3 - PROGRESS: at 72.83% examples, 443623 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:14,387 : INFO : EPOCH 3 - PROGRESS: at 90.72% examples, 441182 words/s, in_qsize 6, out_qsize 1\n",
            "2019-02-26 14:40:14,820 : INFO : worker thread finished; awaiting finish of 3 more threads\n",
            "2019-02-26 14:40:14,827 : INFO : worker thread finished; awaiting finish of 2 more threads\n",
            "2019-02-26 14:40:14,834 : INFO : worker thread finished; awaiting finish of 1 more threads\n",
            "2019-02-26 14:40:14,848 : INFO : worker thread finished; awaiting finish of 0 more threads\n",
            "2019-02-26 14:40:14,854 : INFO : EPOCH - 3 : training on 2988095 raw words (2494720 effective words) took 5.6s, 444614 effective words/s\n",
            "2019-02-26 14:40:15,900 : INFO : EPOCH 4 - PROGRESS: at 17.49% examples, 427107 words/s, in_qsize 8, out_qsize 0\n",
            "2019-02-26 14:40:16,953 : INFO : EPOCH 4 - PROGRESS: at 35.99% examples, 431361 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:17,983 : INFO : EPOCH 4 - PROGRESS: at 54.15% examples, 433415 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:19,021 : INFO : EPOCH 4 - PROGRESS: at 71.81% examples, 431808 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:20,029 : INFO : EPOCH 4 - PROGRESS: at 89.71% examples, 434700 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:20,557 : INFO : worker thread finished; awaiting finish of 3 more threads\n",
            "2019-02-26 14:40:20,564 : INFO : worker thread finished; awaiting finish of 2 more threads\n",
            "2019-02-26 14:40:20,566 : INFO : worker thread finished; awaiting finish of 1 more threads\n",
            "2019-02-26 14:40:20,577 : INFO : worker thread finished; awaiting finish of 0 more threads\n",
            "2019-02-26 14:40:20,578 : INFO : EPOCH - 4 : training on 2988095 raw words (2494537 effective words) took 5.7s, 436840 effective words/s\n",
            "2019-02-26 14:40:21,604 : INFO : EPOCH 5 - PROGRESS: at 17.49% examples, 435168 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:22,605 : INFO : EPOCH 5 - PROGRESS: at 35.65% examples, 442329 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:23,611 : INFO : EPOCH 5 - PROGRESS: at 53.82% examples, 444211 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:24,626 : INFO : EPOCH 5 - PROGRESS: at 71.81% examples, 444303 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:25,633 : INFO : EPOCH 5 - PROGRESS: at 89.40% examples, 443218 words/s, in_qsize 7, out_qsize 0\n",
            "2019-02-26 14:40:26,142 : INFO : worker thread finished; awaiting finish of 3 more threads\n",
            "2019-02-26 14:40:26,171 : INFO : worker thread finished; awaiting finish of 2 more threads\n",
            "2019-02-26 14:40:26,183 : INFO : worker thread finished; awaiting finish of 1 more threads\n",
            "2019-02-26 14:40:26,187 : INFO : worker thread finished; awaiting finish of 0 more threads\n",
            "2019-02-26 14:40:26,188 : INFO : EPOCH - 5 : training on 2988095 raw words (2494414 effective words) took 5.6s, 445537 effective words/s\n",
            "2019-02-26 14:40:26,191 : INFO : training on a 14940475 raw words (12472527 effective words) took 28.6s, 436397 effective words/s\n",
            "2019-02-26 14:40:26,193 : INFO : precomputing L2-norms of word weight vectors\n",
            "2019-02-26 14:40:26,275 : INFO : saving Word2Vec object under 150features_40minwords_10context_imdb, separately None\n",
            "2019-02-26 14:40:26,277 : INFO : not storing attribute vectors_norm\n",
            "2019-02-26 14:40:26,280 : INFO : not storing attribute cum_table\n",
            "2019-02-26 14:40:26,399 : INFO : saved 150features_40minwords_10context_imdb\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "metadata": {
        "id": "93Rh5xfIEm9-",
        "colab_type": "code",
        "outputId": "6043d1fe-2af7-4cbb-b25c-3fa8347f3d81",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 251
        }
      },
      "cell_type": "code",
      "source": [
        "model.wv.most_similar(\"awful\")"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/gensim/matutils.py:737: FutureWarning: Conversion of the second argument of issubdtype from `int` to `np.signedinteger` is deprecated. In future, it will be treated as `np.int64 == np.dtype(int).type`.\n",
            "  if np.issubdtype(vec.dtype, np.int):\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[('terrible', 0.8913172483444214),\n",
              " ('horrible', 0.8555285930633545),\n",
              " ('dreadful', 0.8114152550697327),\n",
              " ('sucks', 0.7771377563476562),\n",
              " ('lousy', 0.7676631808280945),\n",
              " ('abysmal', 0.7665973901748657),\n",
              " ('horrid', 0.7587701678276062),\n",
              " ('horrendous', 0.7528089284896851),\n",
              " ('crappy', 0.7453817129135132),\n",
              " ('atrocious', 0.7430322766304016)]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 45
        }
      ]
    },
    {
      "metadata": {
        "id": "UPejLLyDs0uQ",
        "colab_type": "code",
        "outputId": "ed4a58c3-439a-4434-d33d-f44da3cd4c26",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 251
        }
      },
      "cell_type": "code",
      "source": [
        "model.wv.most_similar(\"amazing\")"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/gensim/matutils.py:737: FutureWarning: Conversion of the second argument of issubdtype from `int` to `np.signedinteger` is deprecated. In future, it will be treated as `np.int64 == np.dtype(int).type`.\n",
            "  if np.issubdtype(vec.dtype, np.int):\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[('fantastic', 0.8467389345169067),\n",
              " ('incredible', 0.8339853882789612),\n",
              " ('wonderful', 0.8116568922996521),\n",
              " ('outstanding', 0.7767613530158997),\n",
              " ('superb', 0.7459939122200012),\n",
              " ('excellent', 0.7354440093040466),\n",
              " ('stunning', 0.7323859333992004),\n",
              " ('magnificent', 0.7316502332687378),\n",
              " ('remarkable', 0.7282607555389404),\n",
              " ('brilliant', 0.7208505868911743)]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 46
        }
      ]
    },
    {
      "metadata": {
        "id": "6ZBWU7F7LPz8",
        "colab_type": "code",
        "outputId": "8a82a76d-05e1-4910-af10-5d9681485ee4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        }
      },
      "cell_type": "code",
      "source": [
        "# This will give the total number of words in the vocabolary created from this dataset\n",
        "print (model.wv.vectors.shape)\n",
        "\n",
        "# Unique words in custom word vectors\n",
        "word_vectors = model.wv\n",
        "print ('Number of unique words in word vector: {}'.format(len(word_vectors.vocab)))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(8160, 150)\n",
            "Number of unique words in word vector: 8160\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "bxtbvlTPs4Bp",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "EMBEDDING_DIM = 150\n",
        "VOCAB_SIZE = min(len(word_index)+1, NUM_WORDS)\n",
        "embedding_matrix = np.zeros((VOCAB_SIZE, EMBEDDING_DIM))\n",
        "\n",
        "for word, i in word_index.items():\n",
        "  \n",
        "    if i>=NUM_WORDS:\n",
        "        continue\n",
        "    try:\n",
        "        embedding_vector = word_vectors[word]\n",
        "        embedding_matrix[i] = embedding_vector\n",
        "        \n",
        "    except KeyError:\n",
        "        embedding_matrix[i] = np.random.normal(0,np.sqrt(0.25), EMBEDDING_DIM)\n",
        "\n",
        "del(word_vectors)\n",
        "\n",
        "# load pre-trained word embeddings into an Embedding layer\n",
        "# note that we set trainable = False so as to keep the embeddings fixed\n",
        "embedding_layer = Embedding(VOCAB_SIZE,\n",
        "                            EMBEDDING_DIM,\n",
        "                            embeddings_initializer=Constant(embedding_matrix),\n",
        "                            input_length=MAX_SEQUENCE_LENGTH,\n",
        "                            trainable=False)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "ywh_iEXbs4Pa",
        "colab_type": "code",
        "outputId": "dc73b778-ac24-49d0-fcfc-4262796358ca",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 71
        }
      },
      "cell_type": "code",
      "source": [
        "from keras.preprocessing.text import Tokenizer\n",
        "from keras.preprocessing.sequence import pad_sequences\n",
        "from keras.initializers import Constant\n",
        "from keras.utils import to_categorical\n",
        "\n",
        "NUM_WORDS = 20000\n",
        "MAX_SEQUENCE_LENGTH = 200\n",
        "\n",
        "# convert each word into token(integer) with maximum number of words based on frequency is 20000\n",
        "tokenizer = Tokenizer(num_words=NUM_WORDS, filters='!\"#$%&()*+,-./:;<=>?@[\\\\]^_`{|}~\\t\\n\\'', lower=True)\n",
        "tokenizer.fit_on_texts(data['review'])\n",
        "\n",
        "# stitch back the words into sequnces with filtering stop words and converting to lower text_to_word_sequence\n",
        "sequences = tokenizer.texts_to_sequences(data['review'])\n",
        "# convert all sentences to equal length of 200\n",
        "sentences = pad_sequences(sequences, maxlen=MAX_SEQUENCE_LENGTH)\n",
        "\n",
        "word_index = tokenizer.word_index\n",
        "print('Found %s unique tokens.' % len(word_index))\n",
        "\n",
        "labels = data['sentiment'].values\n",
        "\n",
        "X, Y = sentences, labels\n",
        "\n",
        "train_x, val_x, train_y, val_y = train_test_split(X, Y, test_size=0.2, stratify=Y)\n",
        "\n",
        "train_y = to_categorical(train_y, num_classes=2)\n",
        "val_y = to_categorical(val_y, num_classes=2)\n",
        "\n",
        "print ('Training:', train_x.shape, train_y.shape)\n",
        "print ('Validation:', val_x.shape, val_y.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 75899 unique tokens.\n",
            "Training: (20000, 200) (20000, 2)\n",
            "Validation: (5000, 200) (5000, 2)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "6NFbaVfBPp2i",
        "colab_type": "code",
        "outputId": "14afb3fb-9eb3-4925-ec02-9c84f6562f2a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 285
        }
      },
      "cell_type": "code",
      "source": [
        "num_classes = 2\n",
        "\n",
        "# input layer with shape = (MAX_SEQUENCE_LENGTH,)\n",
        "sequence_input = Input(shape=(MAX_SEQUENCE_LENGTH,), dtype='int32')\n",
        "# convert the sequence input to embedding from (MAX_SEQUENCE_LENGTH, ) -> (None, MAX_SEQUENCE_LENGTH, EMBEDDING_DIM)\n",
        "embedded_sequences = embedding_layer(sequence_input)\n",
        "\n",
        "x = LSTM(128, dropout=0.2, recurrent_dropout=0.2)(embedded_sequences)\n",
        "preds = Dense(num_classes, activation='softmax')(x)\n",
        "\n",
        "model = Model(inputs=sequence_input, outputs=preds)\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer='adam',\n",
        "              metrics=['acc'])\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_4 (InputLayer)         (None, 200)               0         \n",
            "_________________________________________________________________\n",
            "embedding_4 (Embedding)      (None, 200, 150)          3000000   \n",
            "_________________________________________________________________\n",
            "lstm_4 (LSTM)                (None, 128)               142848    \n",
            "_________________________________________________________________\n",
            "dense_4 (Dense)              (None, 2)                 258       \n",
            "=================================================================\n",
            "Total params: 3,143,106\n",
            "Trainable params: 143,106\n",
            "Non-trainable params: 3,000,000\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "rcIzO8AdPsWP",
        "colab_type": "code",
        "outputId": "c32b8793-05d6-4257-a69a-132db4cc1eb3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 392
        }
      },
      "cell_type": "code",
      "source": [
        "es = EarlyStopping(monitor='val_acc', patience=3)\n",
        "\n",
        "history = model.fit(train_x, train_y,\n",
        "                    batch_size=128,\n",
        "                    epochs=10,\n",
        "                    validation_data=(val_x, val_y),\n",
        "                    callbacks =[es])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 20000 samples, validate on 5000 samples\n",
            "Epoch 1/10\n",
            "20000/20000 [==============================] - 74s 4ms/step - loss: 0.6215 - acc: 0.6460 - val_loss: 0.5230 - val_acc: 0.7406\n",
            "Epoch 2/10\n",
            "20000/20000 [==============================] - 73s 4ms/step - loss: 0.5553 - acc: 0.7139 - val_loss: 0.4860 - val_acc: 0.7726\n",
            "Epoch 3/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.5448 - acc: 0.7227 - val_loss: 0.5194 - val_acc: 0.7532\n",
            "Epoch 4/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.5065 - acc: 0.7510 - val_loss: 0.4257 - val_acc: 0.8050\n",
            "Epoch 5/10\n",
            "20000/20000 [==============================] - 73s 4ms/step - loss: 0.4844 - acc: 0.7659 - val_loss: 0.4134 - val_acc: 0.8192\n",
            "Epoch 6/10\n",
            "20000/20000 [==============================] - 73s 4ms/step - loss: 0.4588 - acc: 0.7823 - val_loss: 0.4457 - val_acc: 0.7974\n",
            "Epoch 7/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.4403 - acc: 0.7963 - val_loss: 0.4072 - val_acc: 0.8214\n",
            "Epoch 8/10\n",
            "20000/20000 [==============================] - 73s 4ms/step - loss: 0.4240 - acc: 0.8045 - val_loss: 0.3800 - val_acc: 0.8312\n",
            "Epoch 9/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.4079 - acc: 0.8127 - val_loss: 0.3834 - val_acc: 0.8388\n",
            "Epoch 10/10\n",
            "20000/20000 [==============================] - 73s 4ms/step - loss: 0.3932 - acc: 0.8219 - val_loss: 0.3705 - val_acc: 0.8390\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "Jyir-rjO0wxQ",
        "colab_type": "code",
        "outputId": "a7f699b5-ff2d-4b46-8b19-1e64beac4e3d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 571
        }
      },
      "cell_type": "code",
      "source": [
        "# summarize history for accuracy\n",
        "plt.plot(history.history['acc'])\n",
        "plt.plot(history.history['val_acc'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "# summarize history for loss\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEVCAYAAAARjMm4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEVCAYAAAAPRfkLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngiRpMBQkSYOhIEkaDAVJ0nAGZMMo0udEgmQAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "ix6Y7FYsOiEI",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "## spaCy"
      ]
    },
    {
      "metadata": {
        "id": "NwxCr9HxOlCn",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "import spacy\n",
        "import string\n",
        "from time import time\n",
        "from keras.utils import to_categorical\n",
        "\n",
        "nlp = spacy.load('en_core_web_md')\n",
        "\n",
        "# Clean text before feeding it to spaCy\n",
        "punctuations = string.punctuation\n",
        "\n",
        "import nltk\n",
        "nltk.download('stopwords')\n",
        "from nltk.corpus import stopwords\n",
        "stopwords = stopwords.words('english')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "ynBtp7iGOlfm",
        "colab_type": "code",
        "outputId": "83feeb39-4df5-4c61-c457-a6023f8c5207",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        }
      },
      "cell_type": "code",
      "source": [
        "%%time\n",
        "\n",
        "# Define function to cleanup text by removing personal pronouns, stopwords, and puncuation\n",
        "def cleanup_text(sentences, logging=False):\n",
        "    texts = []\n",
        "    counter = 1\n",
        "    for word in sentences:\n",
        "        if counter % 1000 == 0 and logging:\n",
        "            print(\"Processed %d out of %d documents.\" % (counter, len(word)))\n",
        "        counter += 1\n",
        "        doc = nlp(word, disable=['parser', 'ner'])\n",
        "        tokens = [tok.lemma_.lower().strip() for tok in doc if tok.lemma_ != '-PRON-']\n",
        "        tokens = [tok for tok in tokens if tok not in stopwords and tok not in punctuations]\n",
        "        tokens = ' '.join(tokens)\n",
        "        texts.append(tokens)\n",
        "    return texts\n",
        "  \n",
        "sent = cleanup_text(data['review'])  "
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 14min 10s, sys: 9min 2s, total: 23min 12s\n",
            "Wall time: 17min 52s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "Q2sex3AYOli_",
        "colab_type": "code",
        "outputId": "7fc6e703-119e-4e54-e2bb-ae7ceaf121aa",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "cell_type": "code",
      "source": [
        "NUM_WORDS = 20000\n",
        "MAX_SEQUENCE_LENGTH = 200\n",
        "\n",
        "# convert each word into token(integer) with maximum number of words based on frequency is 20000\n",
        "tokenizer = Tokenizer(num_words=NUM_WORDS, filters='!\"#$%&()*+,-./:;<=>?@[\\\\]^_`{|}~\\t\\n\\'', lower=True)\n",
        "tokenizer.fit_on_texts(sent)\n",
        "\n",
        "# stitch back the words into sequnces with filtering stop words and converting to lower text_to_word_sequence\n",
        "sequences = tokenizer.texts_to_sequences(sent)\n",
        "# convert all sentences to equal length of 200\n",
        "sentences = pad_sequences(sequences, maxlen=MAX_SEQUENCE_LENGTH)\n",
        "\n",
        "word_index = tokenizer.word_index\n",
        "print('Found %s unique tokens.' % len(word_index))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 66179 unique tokens.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "_Xz3eln4d8YZ",
        "colab_type": "code",
        "outputId": "fe94cfb7-228a-47f8-c219-7d0e1c492957",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        }
      },
      "cell_type": "code",
      "source": [
        "labels = data['sentiment'].values\n",
        "\n",
        "X, Y = sentences, labels\n",
        "\n",
        "train_x, val_x, train_y, val_y = train_test_split(X, Y, test_size=0.2, stratify=Y)\n",
        "\n",
        "train_y = to_categorical(train_y, num_classes=2)\n",
        "val_y = to_categorical(val_y, num_classes=2)\n",
        "\n",
        "print ('Training:', train_x.shape, train_y.shape)\n",
        "print ('Validation:', val_x.shape, val_y.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Training: (20000, 200) (20000, 2)\n",
            "Validation: (5000, 200) (5000, 2)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "Fsjco59ed8kb",
        "colab_type": "code",
        "outputId": "1c47958c-84e4-45dc-cc64-ec1036fe3767",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 412
        }
      },
      "cell_type": "code",
      "source": [
        "num_classes = 2\n",
        "EMBED_DIM = 200\n",
        "VOCAB_SIZE = min(len(word_index)+1, NUM_WORDS)\n",
        "\n",
        "# input layer with shape = (MAX_SEQUENCE_LENGTH,)\n",
        "sequence_input = Input(shape=(MAX_SEQUENCE_LENGTH,), dtype='int32')\n",
        "# convert the sequence input to embedding from (MAX_SEQUENCE_LENGTH, ) -> (None, MAX_SEQUENCE_LENGTH, EMBEDDING_DIM)\n",
        "embedded_sequences = Embedding(VOCAB_SIZE, EMBED_DIM)(sequence_input)\n",
        "\n",
        "x = LSTM(128, dropout=0.2, recurrent_dropout=0.2)(embedded_sequences)\n",
        "preds = Dense(num_classes, activation='softmax')(x)\n",
        "\n",
        "model = Model(inputs=sequence_input, outputs=preds)\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer='adam',\n",
        "              metrics=['acc'])\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Colocations handled automatically by placer.\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:3445: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_2 (InputLayer)         (None, 200)               0         \n",
            "_________________________________________________________________\n",
            "embedding_1 (Embedding)      (None, 200, 200)          4000000   \n",
            "_________________________________________________________________\n",
            "lstm_1 (LSTM)                (None, 128)               168448    \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 2)                 258       \n",
            "=================================================================\n",
            "Total params: 4,168,706\n",
            "Trainable params: 4,168,706\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "cizpw2U9d8gQ",
        "colab_type": "code",
        "outputId": "1b522ae7-8564-4880-d090-98b0260372a5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 287
        }
      },
      "cell_type": "code",
      "source": [
        "es = EarlyStopping(monitor='val_acc', patience=3)\n",
        "\n",
        "history = model.fit(train_x, train_y,\n",
        "                    batch_size=128,\n",
        "                    epochs=10,\n",
        "                    validation_data=(val_x, val_y),\n",
        "                    callbacks =[es])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use tf.cast instead.\n",
            "Train on 20000 samples, validate on 5000 samples\n",
            "Epoch 1/10\n",
            "20000/20000 [==============================] - 162s 8ms/step - loss: 0.4376 - acc: 0.7972 - val_loss: 0.3452 - val_acc: 0.8582\n",
            "Epoch 2/10\n",
            "20000/20000 [==============================] - 140s 7ms/step - loss: 0.2356 - acc: 0.9116 - val_loss: 0.3932 - val_acc: 0.8700\n",
            "Epoch 3/10\n",
            "20000/20000 [==============================] - 141s 7ms/step - loss: 0.1681 - acc: 0.9406 - val_loss: 0.3703 - val_acc: 0.8480\n",
            "Epoch 4/10\n",
            "20000/20000 [==============================] - 139s 7ms/step - loss: 0.1188 - acc: 0.9587 - val_loss: 0.4036 - val_acc: 0.8540\n",
            "Epoch 5/10\n",
            "20000/20000 [==============================] - 141s 7ms/step - loss: 0.0905 - acc: 0.9679 - val_loss: 0.5185 - val_acc: 0.8586\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "YQm4WawWzkVH",
        "colab_type": "code",
        "outputId": "66f72c5c-566d-4d0a-acf2-2835e47c6346",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 571
        }
      },
      "cell_type": "code",
      "source": [
        "# summarize history for accuracy\n",
        "plt.plot(history.history['acc'])\n",
        "plt.plot(history.history['val_acc'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "# summarize history for loss\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEVCAYAAAARjMm4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nLIwxxvTp/we5LCUJ7cK59AAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAEVCAYAAADpbDJPAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "text",
        "id": "QIg8QlOX94AC"
      },
      "cell_type": "markdown",
      "source": [
        "## fasttext"
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "id": "dk5n37D694AR",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "import io\n",
        "from tqdm import tqdm\n",
        "\n",
        "# def load_vectors(fname): \n",
        "#     fin = io.open(fname, 'r', encoding='utf-8', newline='\\n', errors='ignore')\n",
        "#     n, d = map(int, fin.readline().split())\n",
        "#     data = {}\n",
        "#     for line in fin:\n",
        "#         tokens = line.rstrip().split(' ')\n",
        "#         data[tokens[0]] = map(float, tokens[1:])\n",
        "#     return data\n",
        "  \n",
        "def load_vectors(fname):  \n",
        "    embeddings_index = {}\n",
        "    f = io.open(fname, encoding='utf-8', newline='\\n', errors='ignore')\n",
        "    for line in tqdm(f):\n",
        "        values = line.rstrip().rsplit(' ')\n",
        "        word = values[0]\n",
        "        coefs = np.asarray(values[1:], dtype='float32')\n",
        "        embeddings_index[word] = coefs\n",
        "    f.close()\n",
        "    return embeddings_index"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "OvUnMByj9-Rz",
        "colab_type": "code",
        "outputId": "91a097f2-8490-404e-cd11-d35fe416a304",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 71
        }
      },
      "cell_type": "code",
      "source": [
        "%time word_vectors = load_vectors('wiki.simple.vec')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "111052it [00:09, 11540.02it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 9.43 s, sys: 356 ms, total: 9.79 s\n",
            "Wall time: 9.63 s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "metadata": {
        "id": "6CRCG3uAEYnL",
        "colab_type": "code",
        "outputId": "7c2e48e2-7864-4aaa-844f-5566ff90737f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "cell_type": "code",
      "source": [
        "from keras.preprocessing.text import Tokenizer\n",
        "from keras.preprocessing.sequence import pad_sequences\n",
        "from keras.initializers import Constant\n",
        "from keras.utils import to_categorical\n",
        "\n",
        "NUM_WORDS = 20000\n",
        "MAX_SEQUENCE_LENGTH = 200\n",
        "\n",
        "# convert each word into token(integer) with maximum number of words based on frequency is 20000\n",
        "tokenizer = Tokenizer(num_words=NUM_WORDS, filters='!\"#$%&()*+,-./:;<=>?@[\\\\]^_`{|}~\\t\\n\\'', lower=True)\n",
        "tokenizer.fit_on_texts(data['review'])\n",
        "\n",
        "# stitch back the words into sequnces with filtering stop words and converting to lower text_to_word_sequence\n",
        "sequences = tokenizer.texts_to_sequences(data['review'])\n",
        "# convert all sentences to equal length of 200\n",
        "sentences = pad_sequences(sequences, maxlen=MAX_SEQUENCE_LENGTH)\n",
        "\n",
        "word_index = tokenizer.word_index\n",
        "print('Found %s unique tokens.' % len(word_index))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 75899 unique tokens.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "id": "RfjVy8o594Ai",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "EMBEDDING_DIM = 300\n",
        "VOCAB_SIZE = min(len(word_index)+1, NUM_WORDS)\n",
        "embedding_matrix = np.zeros((VOCAB_SIZE, EMBEDDING_DIM))\n",
        "\n",
        "for word, i in word_index.items():\n",
        "  \n",
        "    if i>=NUM_WORDS:\n",
        "        continue\n",
        "    try:\n",
        "        embedding_vector = word_vectors[word]\n",
        "        embedding_matrix[i] = embedding_vector\n",
        "        \n",
        "    except KeyError:\n",
        "        embedding_matrix[i] = np.random.normal(0,np.sqrt(0.25), EMBEDDING_DIM)\n",
        "\n",
        "del(word_vectors)\n",
        "\n",
        "# load pre-trained word embeddings into an Embedding layer\n",
        "# note that we set trainable = False so as to keep the embeddings fixed\n",
        "embedding_layer = Embedding(VOCAB_SIZE,\n",
        "                            EMBEDDING_DIM,\n",
        "                            embeddings_initializer=Constant(embedding_matrix),\n",
        "                            input_length=MAX_SEQUENCE_LENGTH,\n",
        "                            trainable=False)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "colab_type": "code",
        "id": "3G_CRwba94At",
        "outputId": "5e4b2a75-fdaf-41e0-a6a8-a28787e52614",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        }
      },
      "cell_type": "code",
      "source": [
        "labels = data['sentiment'].values\n",
        "\n",
        "X, Y = sentences, labels\n",
        "\n",
        "train_x, val_x, train_y, val_y = train_test_split(X, Y, test_size=0.2, stratify=Y)\n",
        "\n",
        "train_y = to_categorical(train_y, num_classes=2)\n",
        "val_y = to_categorical(val_y, num_classes=2)\n",
        "\n",
        "print ('Training:', train_x.shape, train_y.shape)\n",
        "print ('Validation:', val_x.shape, val_y.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Training: (20000, 200) (20000, 2)\n",
            "Validation: (5000, 200) (5000, 2)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "id": "0N9X5u0m94A4",
        "outputId": "2397008d-a691-463d-ba05-8e71665ace10",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 285
        }
      },
      "cell_type": "code",
      "source": [
        "num_classes = 2\n",
        "\n",
        "# input layer with shape = (MAX_SEQUENCE_LENGTH,)\n",
        "sequence_input = Input(shape=(MAX_SEQUENCE_LENGTH,), dtype='int32')\n",
        "# convert the sequence input to embedding from (MAX_SEQUENCE_LENGTH, ) -> (None, MAX_SEQUENCE_LENGTH, EMBEDDING_DIM)\n",
        "embedded_sequences = embedding_layer(sequence_input)\n",
        "\n",
        "x = LSTM(256, dropout=0.2, recurrent_dropout=0.2)(embedded_sequences)\n",
        "preds = Dense(num_classes, activation='softmax')(x)\n",
        "\n",
        "model = Model(inputs=sequence_input, outputs=preds)\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer='adam',\n",
        "              metrics=['acc'])\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_7 (InputLayer)         (None, 200)               0         \n",
            "_________________________________________________________________\n",
            "embedding_6 (Embedding)      (None, 200, 300)          6000000   \n",
            "_________________________________________________________________\n",
            "lstm_7 (LSTM)                (None, 256)               570368    \n",
            "_________________________________________________________________\n",
            "dense_7 (Dense)              (None, 2)                 514       \n",
            "=================================================================\n",
            "Total params: 6,570,882\n",
            "Trainable params: 570,882\n",
            "Non-trainable params: 6,000,000\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "id": "CDgzVAA694BE",
        "outputId": "c510b9ba-857f-4138-d71d-02677efb2acc",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 392
        }
      },
      "cell_type": "code",
      "source": [
        "es = EarlyStopping(monitor='val_acc', patience=3)\n",
        "\n",
        "history = model.fit(train_x, train_y,\n",
        "                    batch_size=128,\n",
        "                    epochs=10,\n",
        "                    validation_data=(val_x, val_y),\n",
        "                    callbacks =[es])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 20000 samples, validate on 5000 samples\n",
            "Epoch 1/10\n",
            "20000/20000 [==============================] - 75s 4ms/step - loss: 0.6032 - acc: 0.6583 - val_loss: 0.5221 - val_acc: 0.7390\n",
            "Epoch 2/10\n",
            "20000/20000 [==============================] - 73s 4ms/step - loss: 0.5401 - acc: 0.7190 - val_loss: 0.5058 - val_acc: 0.7574\n",
            "Epoch 3/10\n",
            "20000/20000 [==============================] - 73s 4ms/step - loss: 0.4962 - acc: 0.7547 - val_loss: 0.4520 - val_acc: 0.7884\n",
            "Epoch 4/10\n",
            "20000/20000 [==============================] - 73s 4ms/step - loss: 0.4576 - acc: 0.7776 - val_loss: 0.3925 - val_acc: 0.8184\n",
            "Epoch 5/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.3966 - acc: 0.8175 - val_loss: 0.3760 - val_acc: 0.8264\n",
            "Epoch 6/10\n",
            "20000/20000 [==============================] - 73s 4ms/step - loss: 0.3589 - acc: 0.8391 - val_loss: 0.3451 - val_acc: 0.8416\n",
            "Epoch 7/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.3363 - acc: 0.8486 - val_loss: 0.3283 - val_acc: 0.8512\n",
            "Epoch 8/10\n",
            "20000/20000 [==============================] - 73s 4ms/step - loss: 0.3150 - acc: 0.8628 - val_loss: 0.3169 - val_acc: 0.8620\n",
            "Epoch 9/10\n",
            "20000/20000 [==============================] - 73s 4ms/step - loss: 0.2896 - acc: 0.8776 - val_loss: 0.3135 - val_acc: 0.8642\n",
            "Epoch 10/10\n",
            "20000/20000 [==============================] - 73s 4ms/step - loss: 0.2716 - acc: 0.8863 - val_loss: 0.3200 - val_acc: 0.8602\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "id": "JehEq5na94BO",
        "outputId": "48b5a64b-466a-4253-d020-2e8b9cdf8706",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 571
        }
      },
      "cell_type": "code",
      "source": [
        "# summarize history for accuracy\n",
        "plt.plot(history.history['acc'])\n",
        "plt.plot(history.history['val_acc'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "# summarize history for loss\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEVCAYAAAAPRfkLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSinVSZOCUkqpTv8PQ6vDV5JOzW0AAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEVCAYAAAAPRfkLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\npCgIIYToIUVBCCFEDykKQgghevw/F/w2iKXIYFcAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "wBph_OxEOnkH",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "## Custom Embedding Layer using LSTM"
      ]
    },
    {
      "metadata": {
        "id": "qUM5YxmHOveh",
        "colab_type": "code",
        "outputId": "252b52be-937a-4c53-ab13-d808981199ab",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "cell_type": "code",
      "source": [
        "from keras.preprocessing.text import Tokenizer\n",
        "from keras.preprocessing.sequence import pad_sequences\n",
        "from keras.initializers import Constant\n",
        "from keras.utils import to_categorical\n",
        "\n",
        "NUM_WORDS = 20000\n",
        "MAX_SEQUENCE_LENGTH = 200\n",
        "EMBED_DIM = 300\n",
        "\n",
        "# convert each word into token(integer) with maximum number of words based on frequency is 20000\n",
        "tokenizer = Tokenizer(num_words=NUM_WORDS, filters='!\"#$%&()*+,-./:;<=>?@[\\\\]^_`{|}~\\t\\n\\'', lower=True)\n",
        "tokenizer.fit_on_texts(data['review'])\n",
        "\n",
        "# stitch back the words into sequnces with filtering stop words and converting to lower text_to_word_sequence\n",
        "sequences = tokenizer.texts_to_sequences(data['review'])\n",
        "# convert all sentences to equal length of 200\n",
        "sentences = pad_sequences(sequences, maxlen=MAX_SEQUENCE_LENGTH)\n",
        "\n",
        "word_index = tokenizer.word_index\n",
        "print('Found %s unique tokens.' % len(word_index))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 75899 unique tokens.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "VyHGIfQYOwDH",
        "colab_type": "code",
        "outputId": "8d0459a8-d245-48e2-c484-85fc06e2b327",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        }
      },
      "cell_type": "code",
      "source": [
        "labels = data['sentiment'].values\n",
        "\n",
        "X, Y = sentences, labels\n",
        "\n",
        "train_x, val_x, train_y, val_y = train_test_split(X, Y, test_size=0.2, stratify=Y)\n",
        "\n",
        "train_y = to_categorical(train_y, num_classes=2)\n",
        "val_y = to_categorical(val_y, num_classes=2)\n",
        "\n",
        "print ('Training:', train_x.shape, train_y.shape)\n",
        "print ('Validation:', val_x.shape, val_y.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Training: (20000, 200) (20000, 2)\n",
            "Validation: (5000, 200) (5000, 2)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "gpECYps3ywdo",
        "colab_type": "code",
        "outputId": "96e660cb-1491-4a4f-e7a9-f48553c8fae1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 285
        }
      },
      "cell_type": "code",
      "source": [
        "num_classes = 2\n",
        "VOCAB_SIZE = min(len(word_index)+1, NUM_WORDS)\n",
        "\n",
        "# input layer with shape = (MAX_SEQUENCE_LENGTH,)\n",
        "sequence_input = Input(shape=(MAX_SEQUENCE_LENGTH,), dtype='int32')\n",
        "# convert the sequence input to embedding from (MAX_SEQUENCE_LENGTH, ) -> (None, MAX_SEQUENCE_LENGTH,  EMBEDDING_DIM)\n",
        "embedded_sequences = Embedding(VOCAB_SIZE, EMBED_DIM)(sequence_input)\n",
        "\n",
        "x = LSTM(128, dropout=0.2, recurrent_dropout=0.2)(embedded_sequences)\n",
        "preds = Dense(num_classes, activation='softmax')(x)\n",
        "\n",
        "model = Model(inputs=sequence_input, outputs=preds)\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer='adam',\n",
        "              metrics=['acc'])\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_8 (InputLayer)         (None, 200)               0         \n",
            "_________________________________________________________________\n",
            "embedding_7 (Embedding)      (None, 200, 300)          6000000   \n",
            "_________________________________________________________________\n",
            "lstm_8 (LSTM)                (None, 128)               219648    \n",
            "_________________________________________________________________\n",
            "dense_8 (Dense)              (None, 2)                 258       \n",
            "=================================================================\n",
            "Total params: 6,219,906\n",
            "Trainable params: 6,219,906\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "KJOXsiC-ywk6",
        "colab_type": "code",
        "outputId": "cacba0aa-9571-4c12-f901-036b5ba968cb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 178
        }
      },
      "cell_type": "code",
      "source": [
        "es = EarlyStopping(monitor='val_acc', patience=3)\n",
        "\n",
        "history = model.fit(train_x, train_y,\n",
        "                    batch_size=128,\n",
        "                    epochs=10,\n",
        "                    validation_data=(val_x, val_y),\n",
        "                    callbacks =[es])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 20000 samples, validate on 5000 samples\n",
            "Epoch 1/10\n",
            "20000/20000 [==============================] - 81s 4ms/step - loss: 0.4678 - acc: 0.7759 - val_loss: 0.3535 - val_acc: 0.8464\n",
            "Epoch 2/10\n",
            "20000/20000 [==============================] - 79s 4ms/step - loss: 0.2824 - acc: 0.8909 - val_loss: 0.4198 - val_acc: 0.7996\n",
            "Epoch 3/10\n",
            "20000/20000 [==============================] - 79s 4ms/step - loss: 0.2216 - acc: 0.9157 - val_loss: 0.4070 - val_acc: 0.8460\n",
            "Epoch 4/10\n",
            "20000/20000 [==============================] - 80s 4ms/step - loss: 0.2083 - acc: 0.9193 - val_loss: 0.4524 - val_acc: 0.8076\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "FdUnppQNzSFx",
        "colab_type": "code",
        "outputId": "ab2605ed-2624-45af-b2cd-0e45548e79c7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 571
        }
      },
      "cell_type": "code",
      "source": [
        "# summarize history for accuracy\n",
        "plt.plot(history.history['acc'])\n",
        "plt.plot(history.history['val_acc'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "# summarize history for loss\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEVCAYAAAAPRfkLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEVCAYAAAAPRfkLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nGi8+qy+l5gAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "B46HEf3g-7NF",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "colab_type": "text",
        "id": "9ZFom7kTbPvW"
      },
      "cell_type": "markdown",
      "source": [
        "## Custom Embedding Layer using Bidirectional LSTM"
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "211af85d-cd94-463a-8e81-4f41499e8589",
        "id": "19ZUcumKbPvb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "cell_type": "code",
      "source": [
        "from keras.preprocessing.text import Tokenizer\n",
        "from keras.preprocessing.sequence import pad_sequences\n",
        "from keras.initializers import Constant\n",
        "from keras.utils import to_categorical\n",
        "\n",
        "NUM_WORDS = 20000\n",
        "MAX_SEQUENCE_LENGTH = 200\n",
        "EMBED_DIM = 300\n",
        "\n",
        "# convert each word into token(integer) with maximum number of words based on frequency is 20000\n",
        "tokenizer = Tokenizer(num_words=NUM_WORDS, filters='!\"#$%&()*+,-./:;<=>?@[\\\\]^_`{|}~\\t\\n\\'', lower=True)\n",
        "tokenizer.fit_on_texts(data['review'])\n",
        "\n",
        "# stitch back the words into sequnces with filtering stop words and converting to lower text_to_word_sequence\n",
        "sequences = tokenizer.texts_to_sequences(data['review'])\n",
        "# convert all sentences to equal length of 200\n",
        "sentences = pad_sequences(sequences, maxlen=MAX_SEQUENCE_LENGTH)\n",
        "\n",
        "word_index = tokenizer.word_index\n",
        "print('Found %s unique tokens.' % len(word_index))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 75899 unique tokens.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "57897dda-945b-40f7-e7c2-dee095340958",
        "id": "jezPTb7tbPvi",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        }
      },
      "cell_type": "code",
      "source": [
        "labels = data['sentiment'].values\n",
        "\n",
        "X, Y = sentences, labels\n",
        "\n",
        "train_x, val_x, train_y, val_y = train_test_split(X, Y, test_size=0.2, stratify=Y)\n",
        "\n",
        "train_y = to_categorical(train_y, num_classes=2)\n",
        "val_y = to_categorical(val_y, num_classes=2)\n",
        "\n",
        "print ('Training:', train_x.shape, train_y.shape)\n",
        "print ('Validation:', val_x.shape, val_y.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Training: (20000, 200) (20000, 2)\n",
            "Validation: (5000, 200) (5000, 2)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "eb1fa613-e9c0-4ccb-d54f-1c1f5738053d",
        "id": "pJEh9cxpbPvp",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 359
        }
      },
      "cell_type": "code",
      "source": [
        "num_classes = 2\n",
        "VOCAB_SIZE = min(len(word_index)+1, NUM_WORDS)\n",
        "\n",
        "# input layer with shape = (MAX_SEQUENCE_LENGTH,)\n",
        "sequence_input = Input(shape=(MAX_SEQUENCE_LENGTH,), dtype='int32')\n",
        "# convert the sequence input to embedding from (MAX_SEQUENCE_LENGTH, ) -> (None, MAX_SEQUENCE_LENGTH, EMBEDDING_DIM)\n",
        "embedded_sequences = Embedding(VOCAB_SIZE, EMBED_DIM)(sequence_input)\n",
        "\n",
        "x = Bidirectional(LSTM(128, dropout=0.2, recurrent_dropout=0.2))(embedded_sequences)\n",
        "preds = Dense(num_classes, activation='softmax')(x)\n",
        "\n",
        "model = Model(inputs=sequence_input, outputs=preds)\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer='adam',\n",
        "              metrics=['acc'])\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:3445: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_2 (InputLayer)         (None, 200)               0         \n",
            "_________________________________________________________________\n",
            "embedding_2 (Embedding)      (None, 200, 300)          6000000   \n",
            "_________________________________________________________________\n",
            "bidirectional_1 (Bidirection (None, 256)               439296    \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 2)                 514       \n",
            "=================================================================\n",
            "Total params: 6,439,810\n",
            "Trainable params: 6,439,810\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "38b49ca1-a207-478d-a18c-77b626672aca",
        "id": "x0bAj-FzbPvu",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 287
        }
      },
      "cell_type": "code",
      "source": [
        "es = EarlyStopping(monitor='val_acc', patience=3)\n",
        "\n",
        "history = model.fit(train_x, train_y,\n",
        "                    batch_size=128,\n",
        "                    epochs=10,\n",
        "                    validation_data=(val_x, val_y),\n",
        "                    callbacks =[es])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use tf.cast instead.\n",
            "Train on 20000 samples, validate on 5000 samples\n",
            "Epoch 1/10\n",
            "20000/20000 [==============================] - 163s 8ms/step - loss: 0.4896 - acc: 0.7560 - val_loss: 0.3622 - val_acc: 0.8462\n",
            "Epoch 2/10\n",
            "20000/20000 [==============================] - 170s 8ms/step - loss: 0.2948 - acc: 0.8813 - val_loss: 0.3490 - val_acc: 0.8670\n",
            "Epoch 3/10\n",
            "20000/20000 [==============================] - 312s 16ms/step - loss: 0.2422 - acc: 0.9084 - val_loss: 0.3454 - val_acc: 0.8608\n",
            "Epoch 4/10\n",
            "20000/20000 [==============================] - 330s 17ms/step - loss: 0.1917 - acc: 0.9272 - val_loss: 0.3974 - val_acc: 0.8514\n",
            "Epoch 5/10\n",
            "20000/20000 [==============================] - 266s 13ms/step - loss: 0.1250 - acc: 0.9551 - val_loss: 0.4385 - val_acc: 0.8376\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "06c37177-7efc-4fbd-b21a-6a491ca8be4f",
        "id": "i04onOIQbPvx",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 571
        }
      },
      "cell_type": "code",
      "source": [
        "# summarize history for accuracy\n",
        "plt.plot(history.history['acc'])\n",
        "plt.plot(history.history['val_acc'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "# summarize history for loss\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEVCAYAAAARjMm4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nmFFZsjDGGDMqSxbGGGNG9f8BFKsIxZ8h0JAAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEVCAYAAAAPRfkLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "id": "kGQU1ec8bPv1",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "colab_type": "text",
        "id": "AqqAGLaJDmP2"
      },
      "cell_type": "markdown",
      "source": [
        "## Custom Embedding Layer using GRU"
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "eb3ac4a1-117b-4221-8e0a-df6920a32cd5",
        "id": "JKOcSi7KDmQ7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 285
        }
      },
      "cell_type": "code",
      "source": [
        "num_classes = 2\n",
        "VOCAB_SIZE = min(len(word_index)+1, NUM_WORDS)\n",
        "\n",
        "# input layer with shape = (MAX_SEQUENCE_LENGTH,)\n",
        "sequence_input = Input(shape=(MAX_SEQUENCE_LENGTH,), dtype='int32')\n",
        "# convert the sequence input to embedding from (MAX_SEQUENCE_LENGTH, ) -> (None, MAX_SEQUENCE_LENGTH, EMBEDDING_DIM)\n",
        "embedded_sequences = Embedding(VOCAB_SIZE, EMBED_DIM)(sequence_input)\n",
        "\n",
        "x = GRU(128, dropout=0.2, recurrent_dropout=0.2)(embedded_sequences)\n",
        "preds = Dense(num_classes, activation='softmax')(x)\n",
        "\n",
        "model = Model(inputs=sequence_input, outputs=preds)\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer='adam',\n",
        "              metrics=['acc'])\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_11 (InputLayer)        (None, 200)               0         \n",
            "_________________________________________________________________\n",
            "embedding_10 (Embedding)     (None, 200, 300)          6000000   \n",
            "_________________________________________________________________\n",
            "gru_1 (GRU)                  (None, 128)               164736    \n",
            "_________________________________________________________________\n",
            "dense_9 (Dense)              (None, 2)                 258       \n",
            "=================================================================\n",
            "Total params: 6,164,994\n",
            "Trainable params: 6,164,994\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "44f60515-beb4-4191-fa36-2b6979840159",
        "id": "r14VWVIcDmRN",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 285
        }
      },
      "cell_type": "code",
      "source": [
        "es = EarlyStopping(monitor='val_acc', patience=3)\n",
        "\n",
        "history = model.fit(train_x, train_y,\n",
        "                    batch_size=128,\n",
        "                    epochs=10,\n",
        "                    validation_data=(val_x, val_y),\n",
        "                    callbacks =[es])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 20000 samples, validate on 5000 samples\n",
            "Epoch 1/10\n",
            "20000/20000 [==============================] - 72s 4ms/step - loss: 0.5299 - acc: 0.7245 - val_loss: 0.3942 - val_acc: 0.8226\n",
            "Epoch 2/10\n",
            "20000/20000 [==============================] - 69s 3ms/step - loss: 0.3156 - acc: 0.8728 - val_loss: 0.3766 - val_acc: 0.8374\n",
            "Epoch 3/10\n",
            "20000/20000 [==============================] - 69s 3ms/step - loss: 0.2430 - acc: 0.9035 - val_loss: 0.4216 - val_acc: 0.8254\n",
            "Epoch 4/10\n",
            "20000/20000 [==============================] - 69s 3ms/step - loss: 0.1556 - acc: 0.9452 - val_loss: 0.4551 - val_acc: 0.8522\n",
            "Epoch 5/10\n",
            "20000/20000 [==============================] - 69s 3ms/step - loss: 0.1047 - acc: 0.9634 - val_loss: 0.5046 - val_acc: 0.8318\n",
            "Epoch 6/10\n",
            "20000/20000 [==============================] - 69s 3ms/step - loss: 0.0745 - acc: 0.9761 - val_loss: 0.5858 - val_acc: 0.8364\n",
            "Epoch 7/10\n",
            "20000/20000 [==============================] - 69s 3ms/step - loss: 0.0534 - acc: 0.9831 - val_loss: 0.6552 - val_acc: 0.8330\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "d5071b43-81ab-4374-fd35-453d2dd3a059",
        "id": "Lu35QyfVDmRc",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 571
        }
      },
      "cell_type": "code",
      "source": [
        "# summarize history for accuracy\n",
        "plt.plot(history.history['acc'])\n",
        "plt.plot(history.history['val_acc'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "# summarize history for loss\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEVCAYAAAAPRfkLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAEVCAYAAADpbDJPAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\noSQAhBDCQ0kACCGEh/r/kTYvtGsdIyoAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "id": "peMfRaebDmRv",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "colab_type": "text",
        "id": "C1sP7KCJDr4U"
      },
      "cell_type": "markdown",
      "source": [
        "## Custom Embedding Layer using cuDNNLSTM"
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "c574fa2f-3f49-412a-ca4e-3be070a5ad11",
        "id": "tctGKWCqDr5S",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 285
        }
      },
      "cell_type": "code",
      "source": [
        "num_classes = 2\n",
        "VOCAB_SIZE = min(len(word_index)+1, NUM_WORDS)\n",
        "\n",
        "# input layer with shape = (MAX_SEQUENCE_LENGTH,)\n",
        "sequence_input = Input(shape=(MAX_SEQUENCE_LENGTH,), dtype='int32')\n",
        "# convert the sequence input to embedding from (MAX_SEQUENCE_LENGTH, ) -> (None, MAX_SEQUENCE_LENGTH, EMBEDDING_DIM)\n",
        "embedded_sequences = Embedding(VOCAB_SIZE, EMBED_DIM)(sequence_input)\n",
        "\n",
        "x = CuDNNLSTM(128)(embedded_sequences)\n",
        "preds = Dense(num_classes, activation='softmax')(x)\n",
        "\n",
        "model = Model(inputs=sequence_input, outputs=preds)\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer='adam',\n",
        "              metrics=['acc'])\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_14 (InputLayer)        (None, 200)               0         \n",
            "_________________________________________________________________\n",
            "embedding_13 (Embedding)     (None, 200, 300)          6000000   \n",
            "_________________________________________________________________\n",
            "cu_dnnlstm_1 (CuDNNLSTM)     (None, 128)               220160    \n",
            "_________________________________________________________________\n",
            "dense_10 (Dense)             (None, 2)                 258       \n",
            "=================================================================\n",
            "Total params: 6,220,418\n",
            "Trainable params: 6,220,418\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "a75e9d8c-2799-4354-8932-698b05f5a73b",
        "id": "Fc8WQXetDr5h",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 249
        }
      },
      "cell_type": "code",
      "source": [
        "es = EarlyStopping(monitor='val_acc', patience=3)\n",
        "\n",
        "history = model.fit(train_x, train_y,\n",
        "                    batch_size=128,\n",
        "                    epochs=10,\n",
        "                    validation_data=(val_x, val_y),\n",
        "                    callbacks =[es])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 20000 samples, validate on 5000 samples\n",
            "Epoch 1/10\n",
            "20000/20000 [==============================] - 14s 707us/step - loss: 0.4499 - acc: 0.7802 - val_loss: 0.3344 - val_acc: 0.8628\n",
            "Epoch 2/10\n",
            "20000/20000 [==============================] - 10s 496us/step - loss: 0.2284 - acc: 0.9130 - val_loss: 0.3415 - val_acc: 0.8592\n",
            "Epoch 3/10\n",
            "20000/20000 [==============================] - 10s 494us/step - loss: 0.1359 - acc: 0.9531 - val_loss: 0.3786 - val_acc: 0.8630\n",
            "Epoch 4/10\n",
            "20000/20000 [==============================] - 10s 496us/step - loss: 0.0851 - acc: 0.9715 - val_loss: 0.4559 - val_acc: 0.8590\n",
            "Epoch 5/10\n",
            "20000/20000 [==============================] - 10s 496us/step - loss: 0.0586 - acc: 0.9818 - val_loss: 0.4890 - val_acc: 0.8542\n",
            "Epoch 6/10\n",
            "20000/20000 [==============================] - 10s 498us/step - loss: 0.0570 - acc: 0.9800 - val_loss: 0.5604 - val_acc: 0.8520\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "44d57027-4fad-4628-cde4-4db3da76c2b3",
        "id": "kMdUpXFvDr5s",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 571
        }
      },
      "cell_type": "code",
      "source": [
        "# summarize history for accuracy\n",
        "plt.plot(history.history['acc'])\n",
        "plt.plot(history.history['val_acc'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "# summarize history for loss\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEVCAYAAAARjMm4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ncVmyMMYYE9f/B2Vj2w0ZgLOAAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAEVCAYAAADpbDJPAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nlAJARCRI/X+vMPof2QhqUAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "id": "FP_fEqm9Dr58",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "colab_type": "text",
        "id": "9l-J6x40ENGj"
      },
      "cell_type": "markdown",
      "source": [
        "## Custom Embedding Layer using cuDNNGRU"
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "de45e9e1-7715-471c-bced-882783b78a08",
        "id": "wHGBO_g4ENHT",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 285
        }
      },
      "cell_type": "code",
      "source": [
        "num_classes = 2\n",
        "VOCAB_SIZE = min(len(word_index)+1, NUM_WORDS)\n",
        "\n",
        "# input layer with shape = (MAX_SEQUENCE_LENGTH,)\n",
        "sequence_input = Input(shape=(MAX_SEQUENCE_LENGTH,), dtype='int32')\n",
        "# convert the sequence input to embedding from (MAX_SEQUENCE_LENGTH, ) -> (None, MAX_SEQUENCE_LENGTH, EMBEDDING_DIM)\n",
        "embedded_sequences = Embedding(VOCAB_SIZE, EMBED_DIM)(sequence_input)\n",
        "\n",
        "x = CuDNNGRU(128)(embedded_sequences)\n",
        "preds = Dense(num_classes, activation='softmax')(x)\n",
        "\n",
        "model = Model(inputs=sequence_input, outputs=preds)\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer='adam',\n",
        "              metrics=['acc'])\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_15 (InputLayer)        (None, 200)               0         \n",
            "_________________________________________________________________\n",
            "embedding_14 (Embedding)     (None, 200, 300)          6000000   \n",
            "_________________________________________________________________\n",
            "cu_dnngru_1 (CuDNNGRU)       (None, 128)               165120    \n",
            "_________________________________________________________________\n",
            "dense_11 (Dense)             (None, 2)                 258       \n",
            "=================================================================\n",
            "Total params: 6,165,378\n",
            "Trainable params: 6,165,378\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "fcc350f8-20d0-42fc-f9d1-b5b04b027a77",
        "id": "b2V70IILENHi",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 214
        }
      },
      "cell_type": "code",
      "source": [
        "es = EarlyStopping(monitor='val_acc', patience=3)\n",
        "\n",
        "history = model.fit(train_x, train_y,\n",
        "                    batch_size=128,\n",
        "                    epochs=10,\n",
        "                    validation_data=(val_x, val_y),\n",
        "                    callbacks =[es])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 20000 samples, validate on 5000 samples\n",
            "Epoch 1/10\n",
            "20000/20000 [==============================] - 11s 573us/step - loss: 0.4450 - acc: 0.7833 - val_loss: 0.3272 - val_acc: 0.8580\n",
            "Epoch 2/10\n",
            "20000/20000 [==============================] - 9s 439us/step - loss: 0.2002 - acc: 0.9262 - val_loss: 0.3840 - val_acc: 0.8680\n",
            "Epoch 3/10\n",
            "20000/20000 [==============================] - 9s 430us/step - loss: 0.1106 - acc: 0.9621 - val_loss: 0.3761 - val_acc: 0.8494\n",
            "Epoch 4/10\n",
            "20000/20000 [==============================] - 9s 436us/step - loss: 0.0649 - acc: 0.9790 - val_loss: 0.5475 - val_acc: 0.8428\n",
            "Epoch 5/10\n",
            "20000/20000 [==============================] - 9s 441us/step - loss: 0.0663 - acc: 0.9770 - val_loss: 0.5244 - val_acc: 0.8470\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "outputId": "87fb0166-444e-480e-a81d-60f2d812a183",
        "id": "V-Cp7HqsENHu",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 571
        }
      },
      "cell_type": "code",
      "source": [
        "# summarize history for accuracy\n",
        "plt.plot(history.history['acc'])\n",
        "plt.plot(history.history['val_acc'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "# summarize history for loss\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEVCAYAAAARjMm4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nLFkYY4wJy5KFMcaYsP5/A9fUU/t/ElUAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAEVCAYAAADpbDJPAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIYSXkgIQQggvJQUghBBeSgpACCG81P8HJEYBBCjc4k4AAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "8VrXWl6rL0Iw",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "## Prediction"
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "id": "LNQ8FVLyENH-",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "test = pd.read_csv(path/\"testData.tsv\", sep='\\t', header=0, quoting=3)\n",
        "\n",
        "review = test[:10]\n",
        "\n",
        "# stitch back the words into sequnces with filtering stop words and converting to lower text_to_word_sequence\n",
        "sequence = tokenizer.texts_to_sequences(review['review'])\n",
        "\n",
        "# convert all sentences to equal length of 200\n",
        "sentence = pad_sequences(sequence, maxlen=MAX_SEQUENCE_LENGTH)\n",
        "\n",
        "preds = model.predict(sentence)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "oztOE82-D1cg",
        "colab_type": "code",
        "outputId": "b57ac6e1-c4f3-41f5-fa4c-5899839c6307",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 394
        }
      },
      "cell_type": "code",
      "source": [
        "for i in range(len(review)):\n",
        "  print ('Sentence:', review['review'].iloc[i])\n",
        "  print ('Prediction:', preds[i])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Sentence: \"Naturally in a film who's main themes are of mortality, nostalgia, and loss of innocence it is perhaps not surprising that it is rated more highly by older viewers than younger ones. However there is a craftsmanship and completeness to the film which anyone can enjoy. The pace is steady and constant, the characters full and engaging, the relationships and interactions natural showing that you do not need floods of tears to show emotion, screams to show fear, shouting to show dispute or violence to show anger. Naturally Joyce's short story lends the film a ready made structure as perfect as a polished diamond, but the small changes Huston makes such as the inclusion of the poem fit in neatly. It is truly a masterpiece of tact, subtlety and overwhelming beauty.\"\n",
            "Prediction: [7.4012915e-04 9.9925989e-01]\n",
            "Sentence: \"This movie is a disaster within a disaster film. It is full of great action scenes, which are only meaningful if you throw away all sense of reality. Let's see, word to the wise, lava burns you; steam burns you. You can't stand next to lava. Diverting a minor lava flow is difficult, let alone a significant one. Scares me to think that some might actually believe what they saw in this movie.<br /><br />Even worse is the significant amount of talent that went into making this film. I mean the acting is actually very good. The effects are above average. Hard to believe somebody read the scripts for this and allowed all this talent to be wasted. I guess my suggestion would be that if this movie is about to start on TV ... look away! It is like a train wreck: it is so awful that once you know what is coming, you just have to watch. Look away and spend your time on more meaningful content.\"\n",
            "Prediction: [0.99533975 0.00466027]\n",
            "Sentence: \"All in all, this is a movie for kids. We saw it tonight and my child loved it. At one point my kid's excitement was so great that sitting was impossible. However, I am a great fan of A.A. Milne's books which are very subtle and hide a wry intelligence behind the childlike quality of its leading characters. This film was not subtle. It seems a shame that Disney cannot see the benefit of making movies from more of the stories contained in those pages, although perhaps, it doesn't have the permission to use them. I found myself wishing the theater was replaying \\\"Winnie-the-Pooh and Tigger too\\\", instead. The characters voices were very good. I was only really bothered by Kanga. The music, however, was twice as loud in parts than the dialog, and incongruous to the film.<br /><br />As for the story, it was a bit preachy and militant in tone. Overall, I was disappointed, but I would go again just to see the same excitement on my child's face.<br /><br />I liked Lumpy's laugh....\"\n",
            "Prediction: [0.02207627 0.9779237 ]\n",
            "Sentence: \"Afraid of the Dark left me with the impression that several different screenplays were written, all too short for a feature length film, then spliced together clumsily into this Frankenstein's monster.<br /><br />At his best, the protagonist, Lucas, is creepy. As hard as it is to draw a bead on the secondary characters, they're far more sympathetic.<br /><br />Afraid of the Dark could have achieved mediocrity had it taken just one approach and seen it through -- and had it made Lucas simply psychotic and confused instead of ghoulish and off-putting. I wanted to see him packed off into an asylum so the rest of the characters could have a normal life.\"\n",
            "Prediction: [0.9347983  0.06520173]\n",
            "Sentence: \"A very accurate depiction of small time mob life filmed in New Jersey. The story, characters and script are believable but the acting drops the ball. Still, it's worth watching, especially for the strong images, some still with me even though I first viewed this 25 years ago.<br /><br />A young hood steps up and starts doing bigger things (tries to) but these things keep going wrong, leading the local boss to suspect that his end is being skimmed off, not a good place to be if you enjoy your health, or life.<br /><br />This is the film that introduced Joe Pesce to Martin Scorsese. Also present is that perennial screen wise guy, Frank Vincent. Strong on characterizations and visuals. Sound muddled and much of the acting is amateurish, but a great story.\"\n",
            "Prediction: [0.08774225 0.9122578 ]\n",
            "Sentence: \"...as valuable as King Tut's tomb! (OK, maybe not THAT valuable, but worth hunting down if you can). I notice no one has commented on this movie for some years, and I hope a fresh post will spark some new comments. This is a film that I remembered only snippets of from childhood, and only saw recently when I tired of waiting for Fox to honour its own past, and hunted down the Korean DVD (in English, but with unremovable Korean subtitles). I won't go through another long plot description - suffice to say that seeing it for the first time in its proper widescreen format left me agape at the vistas and the scope of the film. The matte paintings still hold up, and the palace sets are truly breathtaking. But it is the smaller scale details that lend this film its depth and richness, offering a glimpse into the lifestyles of Egypt's poor as well as its elite. The bazaars, hovels, docks, embalming houses, and taverns are as fascinating as Pharaoh's throne room. While errors abound on the large scale (most notably the dynastic succession), the details are more meticulously researched than the vast majority of Hollywood's films. Visually, it's not without its flaws - the interiors are often too overly lit and colourful to blend seamlessly with the exteriors. Nevertheless, this is a movie that should be credited for being as audacious in the small as it is in the large. Tedious? In parts, absolutely. Overacted? Underacted? Yes, both - though 'understated' might be a more apt description. Too long? Absolutely not. I wished they had spent more time with Sinuhe's experiences in the House of Death, and among the Hittites, and less with his 'romance' with Nefer, though. Historically inaccurate? Yes, that too, but so was Shakespeare. Nobody chastises him for it. I appreciate historical accuracy as much as the next guy, but ultimately it has to be remembered that cinema is theater, not a history lesson.\"\n",
            "Prediction: [0.8392459  0.16075405]\n",
            "Sentence: \"This has to be one of the biggest misfires ever...the script was nice and could have ended a lot better.the actors should have played better and maybe then i would have given this movie a slightly better grade. maybe Hollywood should remake this movie with some little better actors and better director.sorry guys for disappointment but the movie is bad.<br /><br />If i had to re-watch it it would be like torture. I don't want to spoil everyone's opinion with mine so..my advice is watch the movie first..see if u like it and after vote(do not vote before you watch it ! ) and by the way... Have fun watching it ! Don't just peek...watch it 'till the end :))))))))) !!\"\n",
            "Prediction: [0.23039062 0.76960933]\n",
            "Sentence: \"This is one of those movies I watched, and wondered, why did I watch it? What did I find so interesting about it? Being a truck driver myself, I didn't find it very realistic. No, I've never used a 'lot lizard', nor have I ever seen, nor heard about one traveling around the country in a brand new seventy thousand dollar RV, either.<br /><br />Same thing about a pimp whom has never sampled the lady in question (until the end of the movie, and well, he still really didn't...), and only getting 50 bucks 'a cut', when the prostitute gets $200.00 (well, $150.00 after his cut, yeah...).<br /><br />I still laugh at the lot lizard comment Ivey made (them's Lot Lizards, they'll screw anything with 20 bucks, and some are men dressed as woman... or something equally as weird), meaning, we're better then them, as we may still be prostitutes, but we get paid BETTER.<br /><br />Other then that, it's just a story of a young woman whom wanted something more from life then a dead end job while living at home (she's 18, remember?) and embarrassed by her mother basically doing the same thing (dead end job). At least she had a roof over her head and a job. She turned FIVE tricks on the road... I wonder if the $750.00 she made was worth it? I'd guess not.\"\n",
            "Prediction: [0.8161818  0.18381816]\n",
            "Sentence: \"The worst movie i've seen in years (and i've seen a lot of movies). Acting is terrible, there is no plot whatsoever, there is no point whatsoever, i felt robbed after i rented this movie. they recommended it to me mind you! a disgrace for terrible movies! stay away from this terrible piece of c**p. save your money !\"\n",
            "Prediction: [9.9955624e-01 4.4377038e-04]\n",
            "Sentence: \"Five medical students (Kevin Bacon, David Labraccio; William Baldwin, Dr. Joe Hurley; Oliver Platt, Randy Steckle; Julia Roberts, Dr. Rachel Mannus; Kiefer Sutherland, Nelson) experiment with clandestine near death & afterlife experiences, (re)searching for medical & personal enlightenment. One by one, each medical student's heart is stopped, then revived.<br /><br />Under temporary death spells each experiences bizarre visions, including forgotten childhood memories. Their flashbacks are like children's nightmares. The revived students are disturbed by remembering regretful acts they had committed or had done against them. As they experience afterlife, they bring real life experiences back into the present. As they continue to experiment, their remembrances dramatically intensify; so much so, some are physically overcome. Thus, they probe & transcend deeper into the death-afterlife experiences attempting to find a cure.<br /><br />Even though the DVD was released in 2007, this motion picture was released in 1990. Therefore, Kevin Bacon, William Baldwin, Julia Roberts & Kiefer Sutherland were in the early stages of their adult acting careers. Besides the plot being extremely intriguing, the suspense building to a dramatic climax & the script being tight & convincing, all of the young actors make \\\"Flatliners,\\\" what is now an all-star cult semi-sci-fi suspense. Who knew 17 years ago that the film careers of this young group of actors would skyrocket? I suspect that director Joel Schumacher did.\"\n",
            "Prediction: [0.7998832  0.20011681]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "fnz2jqQAJGaD",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}