TF 1 Tutorials/text_classification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### [Text classification with movie reviews](https://www.tensorflow.org/tutorials/keras/basic_text_classification)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from __future__ import absolute_import, division, print_function, unicode_literals\n",
    "\n",
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "\n",
    "import numpy as np\n",
    "\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensorflow: 1.12.0\n",
      "Numpy: 1.16.2\n"
     ]
    }
   ],
   "source": [
    "print('Tensorflow: {}\\nNumpy: {}'.format(tf.__version__, np.version.version))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Load the IMDB dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "imdb = keras.datasets.imdb\n",
    "(train_data, train_labels), (test_data, test_labels) = imdb.load_data(num_words=10000)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Note:** If it throws `ValueError: Object arrays cannot be loaded when allow_pickle=False`, try degrading **Numpy** to 1.16.2 using `!pip install numpy==1.16.1`."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### About the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training data: 25000, labels: 25000\n"
     ]
    }
   ],
   "source": [
    "print('Training data: {}, labels: {}'.format(len(train_data), len(train_labels)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1, 14, 22, 16, 43, 530, 973, 1622, 1385, 65, 458, 4468, 66, 3941, 4, 173, 36, 256, 5, 25, 100, 43, 838, 112, 50, 670, 2, 9, 35, 480, 284, 5, 150, 4, 172, 112, 167, 2, 336, 385, 39, 4, 172, 4536, 1111, 17, 546, 38, 13, 447, 4, 192, 50, 16, 6, 147, 2025, 19, 14, 22, 4, 1920, 4613, 469, 4, 22, 71, 87, 12, 16, 43, 530, 38, 76, 15, 13, 1247, 4, 22, 17, 515, 17, 12, 16, 626, 18, 2, 5, 62, 386, 12, 8, 316, 8, 106, 5, 4, 2223, 5244, 16, 480, 66, 3785, 33, 4, 130, 12, 16, 38, 619, 5, 25, 124, 51, 36, 135, 48, 25, 1415, 33, 6, 22, 12, 215, 28, 77, 52, 5, 14, 407, 16, 82, 2, 8, 4, 107, 117, 5952, 15, 256, 4, 2, 7, 3766, 5, 723, 36, 71, 43, 530, 476, 26, 400, 317, 46, 7, 4, 2, 1029, 13, 104, 88, 4, 381, 15, 297, 98, 32, 2071, 56, 26, 141, 6, 194, 7486, 18, 4, 226, 22, 21, 134, 476, 26, 480, 5, 144, 30, 5535, 18, 51, 36, 28, 224, 92, 25, 104, 4, 226, 65, 16, 38, 1334, 88, 12, 16, 283, 5, 16, 4472, 113, 103, 32, 15, 16, 5345, 19, 178, 32]\n"
     ]
    }
   ],
   "source": [
    "print(train_data[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(218, 189)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(train_data[0]), len(train_data[1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Converting integers back to words"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "word_index = imdb.get_word_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "word_index = {k:(v+3) for k,v in word_index.items()}\n",
    "word_index['<PAD>'] = 0\n",
    "word_index['<START>'] = 1\n",
    "word_index['<UNK>'] = 2\n",
    "word_index['<UNUSED>'] = 3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "reverse_word_index = dict([(value, key) for  (key, value) in word_index.items()])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def decode_review(text):\n",
    "    return ' '.join([reverse_word_index.get(i, '?') for i in text])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\"<START> this film was just brilliant casting location scenery story direction everyone's really suited the part they played and you could just imagine being there robert <UNK> is an amazing actor and now the same being director <UNK> father came from the same scottish island as myself so i loved the fact there was a real connection with this film the witty remarks throughout the film were great it was just brilliant so much that i bought the film as soon as it was released for <UNK> and would recommend it to everyone to watch and the fly fishing was amazing really cried at the end it was so sad and you know what they say if you cry at a film it must have been good and this definitely was also <UNK> to the two little boy's that played the <UNK> of norman and paul they were just brilliant children are often left out of the <UNK> list i think because the stars that play them all grown up are such a big profile for the whole film but these children are amazing and should be praised for what they have done don't you think the whole story was so lovely because it was true and was someone's life after all that was shared with us all\""
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "decode_review(train_data[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Prepare the data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Since the movie reviews must be the same length, we will use the `pad_sequences` function to standardize the lengths."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_data = keras.preprocessing.sequence.pad_sequences(train_data, value=word_index['<PAD>'], padding='post', maxlen=256)\n",
    "test_data = keras.preprocessing.sequence.pad_sequences(test_data, value=word_index['<PAD>'], padding='post', maxlen=256)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(256, 256)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(train_data[0]), len(train_data[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[   1   14   22   16   43  530  973 1622 1385   65  458 4468   66 3941\n",
      "    4  173   36  256    5   25  100   43  838  112   50  670    2    9\n",
      "   35  480  284    5  150    4  172  112  167    2  336  385   39    4\n",
      "  172 4536 1111   17  546   38   13  447    4  192   50   16    6  147\n",
      " 2025   19   14   22    4 1920 4613  469    4   22   71   87   12   16\n",
      "   43  530   38   76   15   13 1247    4   22   17  515   17   12   16\n",
      "  626   18    2    5   62  386   12    8  316    8  106    5    4 2223\n",
      " 5244   16  480   66 3785   33    4  130   12   16   38  619    5   25\n",
      "  124   51   36  135   48   25 1415   33    6   22   12  215   28   77\n",
      "   52    5   14  407   16   82    2    8    4  107  117 5952   15  256\n",
      "    4    2    7 3766    5  723   36   71   43  530  476   26  400  317\n",
      "   46    7    4    2 1029   13  104   88    4  381   15  297   98   32\n",
      " 2071   56   26  141    6  194 7486   18    4  226   22   21  134  476\n",
      "   26  480    5  144   30 5535   18   51   36   28  224   92   25  104\n",
      "    4  226   65   16   38 1334   88   12   16  283    5   16 4472  113\n",
      "  103   32   15   16 5345   19  178   32    0    0    0    0    0    0\n",
      "    0    0    0    0    0    0    0    0    0    0    0    0    0    0\n",
      "    0    0    0    0    0    0    0    0    0    0    0    0    0    0\n",
      "    0    0    0    0]\n"
     ]
    }
   ],
   "source": [
    "print(train_data[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Build the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "VOCAB_SIZE = 10000"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = keras.Sequential()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.add(keras.layers.Embedding(VOCAB_SIZE, 16))\n",
    "model.add(keras.layers.GlobalAveragePooling1D())\n",
    "model.add(keras.layers.Dense(16, activation=tf.nn.relu))\n",
    "model.add(keras.layers.Dense(1, activation=tf.nn.sigmoid))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding (Embedding)        (None, None, 16)          160000    \n",
      "_________________________________________________________________\n",
      "global_average_pooling1d (Gl (None, 16)                0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 16)                272       \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 1)                 17        \n",
      "=================================================================\n",
      "Total params: 160,289\n",
      "Trainable params: 160,289\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Compile the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(optimizer='adam',\n",
    "             loss='binary_crossentropy',\n",
    "             metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Create validation set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_val = train_data[:10000]\n",
    "partial_x_train = train_data[10000:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_val = train_labels[:10000]\n",
    "partial_y_train = train_labels[10000:]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Training the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\harshil\\AppData\\Roaming\\Python\\Python36\\site-packages\\tensorflow\\python\\ops\\gradients_impl.py:112: UserWarning: Converting sparse IndexedSlices to a dense Tensor of unknown shape. This may consume a large amount of memory.\n",
      "  \"Converting sparse IndexedSlices to a dense Tensor of unknown shape. \"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 15000 samples, validate on 10000 samples\n",
      "Epoch 1/40\n",
      "15000/15000 [==============================] - 3s 181us/step - loss: 0.6920 - acc: 0.5236 - val_loss: 0.6903 - val_acc: 0.5420\n",
      "Epoch 2/40\n",
      "15000/15000 [==============================] - 2s 133us/step - loss: 0.6864 - acc: 0.6844 - val_loss: 0.6819 - val_acc: 0.7115\n",
      "Epoch 3/40\n",
      "15000/15000 [==============================] - 2s 117us/step - loss: 0.6736 - acc: 0.7516 - val_loss: 0.6659 - val_acc: 0.7517\n",
      "Epoch 4/40\n",
      "15000/15000 [==============================] - 2s 117us/step - loss: 0.6507 - acc: 0.7603 - val_loss: 0.6400 - val_acc: 0.7646\n",
      "Epoch 5/40\n",
      "15000/15000 [==============================] - 2s 124us/step - loss: 0.6165 - acc: 0.7903 - val_loss: 0.6032 - val_acc: 0.7825\n",
      "Epoch 6/40\n",
      "15000/15000 [==============================] - 2s 125us/step - loss: 0.5731 - acc: 0.8078 - val_loss: 0.5612 - val_acc: 0.7964\n",
      "Epoch 7/40\n",
      "15000/15000 [==============================] - 2s 133us/step - loss: 0.5249 - acc: 0.8282 - val_loss: 0.5170 - val_acc: 0.8153\n",
      "Epoch 8/40\n",
      "15000/15000 [==============================] - 2s 120us/step - loss: 0.4769 - acc: 0.8452 - val_loss: 0.4752 - val_acc: 0.8320\n",
      "Epoch 9/40\n",
      "15000/15000 [==============================] - 2s 121us/step - loss: 0.4332 - acc: 0.8578 - val_loss: 0.4386 - val_acc: 0.8428\n",
      "Epoch 10/40\n",
      "15000/15000 [==============================] - 2s 122us/step - loss: 0.3940 - acc: 0.8734 - val_loss: 0.4073 - val_acc: 0.8509\n",
      "Epoch 11/40\n",
      "15000/15000 [==============================] - 2s 134us/step - loss: 0.3607 - acc: 0.8810 - val_loss: 0.3840 - val_acc: 0.8557\n",
      "Epoch 12/40\n",
      "15000/15000 [==============================] - 2s 124us/step - loss: 0.3333 - acc: 0.8889 - val_loss: 0.3617 - val_acc: 0.8643\n",
      "Epoch 13/40\n",
      "15000/15000 [==============================] - 2s 123us/step - loss: 0.3087 - acc: 0.8958 - val_loss: 0.3464 - val_acc: 0.8686\n",
      "Epoch 14/40\n",
      "15000/15000 [==============================] - 2s 130us/step - loss: 0.2886 - acc: 0.9017 - val_loss: 0.3329 - val_acc: 0.8725\n",
      "Epoch 15/40\n",
      "15000/15000 [==============================] - 2s 146us/step - loss: 0.2712 - acc: 0.9060 - val_loss: 0.3227 - val_acc: 0.8745\n",
      "Epoch 16/40\n",
      "15000/15000 [==============================] - 2s 152us/step - loss: 0.2564 - acc: 0.9101 - val_loss: 0.3143 - val_acc: 0.8770\n",
      "Epoch 17/40\n",
      "15000/15000 [==============================] - 2s 140us/step - loss: 0.2422 - acc: 0.9156 - val_loss: 0.3075 - val_acc: 0.8774\n",
      "Epoch 18/40\n",
      "15000/15000 [==============================] - 2s 152us/step - loss: 0.2298 - acc: 0.9208 - val_loss: 0.3018 - val_acc: 0.8802\n",
      "Epoch 19/40\n",
      "15000/15000 [==============================] - 2s 148us/step - loss: 0.2183 - acc: 0.9253 - val_loss: 0.2975 - val_acc: 0.8814\n",
      "Epoch 20/40\n",
      "15000/15000 [==============================] - 2s 150us/step - loss: 0.2081 - acc: 0.9280 - val_loss: 0.2937 - val_acc: 0.8829\n",
      "Epoch 21/40\n",
      "15000/15000 [==============================] - 2s 152us/step - loss: 0.1983 - acc: 0.9324 - val_loss: 0.2908 - val_acc: 0.8835\n",
      "Epoch 22/40\n",
      "15000/15000 [==============================] - 2s 151us/step - loss: 0.1893 - acc: 0.9363 - val_loss: 0.2890 - val_acc: 0.8841\n",
      "Epoch 23/40\n",
      "15000/15000 [==============================] - 2s 141us/step - loss: 0.1811 - acc: 0.9395 - val_loss: 0.2877 - val_acc: 0.8845\n",
      "Epoch 24/40\n",
      "15000/15000 [==============================] - 2s 142us/step - loss: 0.1728 - acc: 0.9445 - val_loss: 0.2860 - val_acc: 0.8842\n",
      "Epoch 25/40\n",
      "15000/15000 [==============================] - 2s 140us/step - loss: 0.1657 - acc: 0.9471 - val_loss: 0.2854 - val_acc: 0.8846\n",
      "Epoch 26/40\n",
      "15000/15000 [==============================] - 2s 143us/step - loss: 0.1585 - acc: 0.9500 - val_loss: 0.2854 - val_acc: 0.8851\n",
      "Epoch 27/40\n",
      "15000/15000 [==============================] - 2s 144us/step - loss: 0.1525 - acc: 0.9527 - val_loss: 0.2862 - val_acc: 0.8847\n",
      "Epoch 28/40\n",
      "15000/15000 [==============================] - 2s 137us/step - loss: 0.1461 - acc: 0.9559 - val_loss: 0.2859 - val_acc: 0.8866\n",
      "Epoch 29/40\n",
      "15000/15000 [==============================] - 2s 140us/step - loss: 0.1403 - acc: 0.9565 - val_loss: 0.2866 - val_acc: 0.8858\n",
      "Epoch 30/40\n",
      "15000/15000 [==============================] - 2s 144us/step - loss: 0.1353 - acc: 0.9595 - val_loss: 0.2881 - val_acc: 0.8860\n",
      "Epoch 31/40\n",
      "15000/15000 [==============================] - 2s 146us/step - loss: 0.1292 - acc: 0.9619 - val_loss: 0.2896 - val_acc: 0.8855\n",
      "Epoch 32/40\n",
      "15000/15000 [==============================] - 2s 142us/step - loss: 0.1245 - acc: 0.9639 - val_loss: 0.2915 - val_acc: 0.8856\n",
      "Epoch 33/40\n",
      "15000/15000 [==============================] - 2s 139us/step - loss: 0.1191 - acc: 0.9664 - val_loss: 0.2935 - val_acc: 0.8851\n",
      "Epoch 34/40\n",
      "15000/15000 [==============================] - 2s 145us/step - loss: 0.1147 - acc: 0.9679 - val_loss: 0.2962 - val_acc: 0.8843\n",
      "Epoch 35/40\n",
      "15000/15000 [==============================] - 2s 144us/step - loss: 0.1107 - acc: 0.9693 - val_loss: 0.2979 - val_acc: 0.8853\n",
      "Epoch 36/40\n",
      "15000/15000 [==============================] - 2s 137us/step - loss: 0.1058 - acc: 0.9709 - val_loss: 0.3009 - val_acc: 0.8839\n",
      "Epoch 37/40\n",
      "15000/15000 [==============================] - 2s 140us/step - loss: 0.1019 - acc: 0.9725 - val_loss: 0.3039 - val_acc: 0.8845 loss: 0.1006 - acc\n",
      "Epoch 38/40\n",
      "15000/15000 [==============================] - 2s 149us/step - loss: 0.0985 - acc: 0.9733 - val_loss: 0.3072 - val_acc: 0.8828\n",
      "Epoch 39/40\n",
      "15000/15000 [==============================] - 2s 142us/step - loss: 0.0942 - acc: 0.9753 - val_loss: 0.3095 - val_acc: 0.8833\n",
      "Epoch 40/40\n",
      "15000/15000 [==============================] - 2s 145us/step - loss: 0.0905 - acc: 0.9770 - val_loss: 0.3131 - val_acc: 0.8812\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(partial_x_train, partial_y_train, epochs=40, batch_size=512, validation_data=(x_val, y_val), verbose=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "25000/25000 [==============================] - 2s 63us/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.3332369383621216, 0.87292]"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate(test_data, test_labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Accuracy and Loss over time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['val_loss', 'val_acc', 'loss', 'acc'])"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "history_dict = history.history\n",
    "history_dict.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "acc = history_dict['acc']\n",
    "val_acc = history_dict['val_acc']\n",
    "loss = history_dict['loss']\n",
    "val_loss = history_dict['val_loss']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "epochs = range(1, len(acc) + 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.xlabel('Epochs')\n",
    "\n",
    "plt.plot(epochs, acc, 'g--', label='Training accuracy')\n",
    "plt.plot(epochs, val_acc, 'g', label='Validation accuracy')\n",
    "\n",
    "plt.plot(epochs, loss, 'r--', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'r', label='Validation loss')\n",
    "\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}