Pretrained embedding imdb.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import os\n",
    "base_dir ='/Users/Rajiv/Desktop/ML/code/aclImdb'\n",
    "train_dir =os.path.join(base_dir ,'train')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "labels =[]                       # creating empty list of output \n",
    "texts =[]                        # creating empty list for storing text"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "for label_type in ['neg' ,'pos']:\n",
    "    dir_name = os.path.join(train_dir ,label_type)       \n",
    "    for fname in os.listdir(dir_name):               # create a list of names in the directory of dir_name\n",
    "        if fname[-4:] == '.txt':                      # checking last four word is text file or not\n",
    "            f = open(os.path.join(dir_name ,fname) ,encoding=\"utf8\")\n",
    "            texts.append(f.read())                  # read and append the text in text list\n",
    "            f.close()\n",
    "            if label_type == 'neg':                 # creating real ouput for imdb datases\n",
    "                labels.append(0)\n",
    "            else:\n",
    "                labels.append(1)\n",
    "        \n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 88582 unique tokens\n",
      "shape of data tensor (25000, 100)\n",
      "shape of labels (25000,)\n"
     ]
    }
   ],
   "source": [
    "from keras.preprocessing.text import Tokenizer\n",
    "from keras.preprocessing.sequence import pad_sequences\n",
    "import numpy as np\n",
    "maxlen = 100                                                      # cut off sequencess after 100th word\n",
    "max_words = 10000                                                  # considering only top 10000 word\n",
    "training_sample = 200                                              # since we are using pretrained embedding so it will work on smaller dataset\n",
    "validation_sample = 10000\n",
    "tokenizer = Tokenizer(num_words =max_words) \n",
    "tokenizer.fit_on_texts(texts)\n",
    "sequences = tokenizer.texts_to_sequences(texts)\n",
    "word_index = tokenizer.word_index\n",
    "print('Found %s unique tokens'%len(word_index))\n",
    "data = pad_sequences(sequences , maxlen =maxlen)\n",
    "labels = np.asarray(labels)\n",
    "print('shape of data tensor' ,data.shape)\n",
    "print('shape of labels' ,labels.shape)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "indices = np.arange(data.shape[0])\n",
    "np.random.shuffle(indices)\n",
    "data =data[indices]\n",
    "labels = labels[indices]\n",
    "x_train =data[:training_sample]\n",
    "y_train = labels[:training_sample]\n",
    "x_val = data[training_sample:training_sample + validation_sample]\n",
    "y_val = labels[training_sample:training_sample +validation_sample]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "glove_dir = r'\\Users\\Rajiv\\Desktop\\ML\\code\\glove.6B'\n",
    "f= open(os.path.join(glove_dir ,'glove.6B.100d.txt') ,encoding=\"utf8\")                   # will read the file and stored in it\n",
    "embeddings_index = {}                                                   # making a empty dictionary to store word and their embedding coefficient\n",
    "for line in f:\n",
    "    values = line.split()                                          # make a column for every word and their embedding coefficient\n",
    "    word = values[0]                                             # extracting word value\n",
    "    coefs = np.asarray(values[1:] ,dtype ='float')\n",
    "    embeddings_index[word] = coefs                               # storing word and coeffiecient in dictionary\n",
    "f.close()\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "found 400000 word vectors\n"
     ]
    }
   ],
   "source": [
    "print('found %s word vectors' %len(embeddings_index))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "shape  of embedding matrix (10000, 100)\n"
     ]
    }
   ],
   "source": [
    "# buildind a embedding matrx , shape->(max_word ,emd_dims)\n",
    "emd_dims =100                                                      # since we are using 100.d\n",
    "embedding_matrix = np.zeros((max_words ,emd_dims))              # only considering 10000 most common word\n",
    "for word , i in word_index.items():                            # return two things word from most common word and integer\n",
    "    if i < max_words:\n",
    "        embedding_vector = embeddings_index.get(word)        # return coef corresponding tro the word\n",
    "        if embedding_vector is not None:\n",
    "            embedding_matrix[i] = embedding_vector\n",
    "print('shape  of embedding matrix' ,embedding_matrix.shape)\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding_3 (Embedding)      (None, 100, 100)          1000000   \n",
      "_________________________________________________________________\n",
      "flatten_3 (Flatten)          (None, 10000)             0         \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 32)                320032    \n",
      "_________________________________________________________________\n",
      "dense_6 (Dense)              (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 1,320,065\n",
      "Trainable params: 1,320,065\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# defining model\n",
    "from keras.models import  Sequential\n",
    "from keras.layers import Flatten ,Dense ,Embedding\n",
    "model = Sequential()\n",
    "model.add(Embedding(max_words ,emd_dims ,input_length =maxlen))                 # will give 3D output\n",
    "model.add(Flatten())                                                            \n",
    "model.add(Dense(32 ,activation ='relu'))\n",
    "model.add(Dense(1 ,activation ='sigmoid'))\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#LOADING THE GLOVE EMBEDDINGS IN THE MODEL\n",
    "model.layers[0].set_weights([embedding_matrix])\n",
    "model.layers[0].trainable = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 200 samples, validate on 10000 samples\n",
      "Epoch 1/10\n",
      "200/200 [==============================] - 1s 4ms/step - loss: 2.2140 - acc: 0.5250 - val_loss: 0.7159 - val_acc: 0.4960\n",
      "Epoch 2/10\n",
      "200/200 [==============================] - 1s 3ms/step - loss: 0.6366 - acc: 0.6650 - val_loss: 0.7410 - val_acc: 0.4986\n",
      "Epoch 3/10\n",
      "200/200 [==============================] - 1s 4ms/step - loss: 0.4107 - acc: 0.7900 - val_loss: 1.0432 - val_acc: 0.4942\n",
      "Epoch 4/10\n",
      "200/200 [==============================] - 1s 3ms/step - loss: 0.4976 - acc: 0.7450 - val_loss: 0.7576 - val_acc: 0.5065\n",
      "Epoch 5/10\n",
      "200/200 [==============================] - 1s 4ms/step - loss: 0.1645 - acc: 0.9900 - val_loss: 0.7588 - val_acc: 0.5055\n",
      "Epoch 6/10\n",
      "200/200 [==============================] - 1s 4ms/step - loss: 0.1253 - acc: 0.9950 - val_loss: 0.9686 - val_acc: 0.5102\n",
      "Epoch 7/10\n",
      "200/200 [==============================] - 1s 4ms/step - loss: 0.4975 - acc: 0.7350 - val_loss: 0.8074 - val_acc: 0.5095\n",
      "Epoch 8/10\n",
      "200/200 [==============================] - 1s 4ms/step - loss: 0.0501 - acc: 1.0000 - val_loss: 0.8209 - val_acc: 0.5064\n",
      "Epoch 9/10\n",
      "200/200 [==============================] - 1s 4ms/step - loss: 0.0380 - acc: 1.0000 - val_loss: 0.8330 - val_acc: 0.5114\n",
      "Epoch 10/10\n",
      "200/200 [==============================] - 1s 3ms/step - loss: 0.0279 - acc: 1.0000 - val_loss: 0.9382 - val_acc: 0.5030\n"
     ]
    }
   ],
   "source": [
    "model.compile(optimizer ='rmsprop' ,loss= 'binary_crossentropy' ,metrics =['acc'])\n",
    "history = model.fit(x_train ,y_train ,epochs =10 ,batch_size =32 ,validation_data =(x_val ,y_val))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model.save_weights('pre_trained_glove_model.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEKCAYAAAD9xUlFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nhYKIiIQUCiIiEvr/X6WhUFCcO64AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x239e2d6d9b0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEKCAYAAAD9xUlFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAIABJREFUeJzt3Xt4VOW59/HvDYnGAEJAfIsiBFu1\nEAgBsi0tVkStBbXiARUFK75a1Nptbav1QIvnVt2oiLpV3OpGTVWKJ/oWD5uKIq2nAMpWsfUAaMRD\nQBPAgAq53z+eYRFiDgNkZk0yv891zUVmzZo1d2bI/NbzrGc9y9wdERERgHZxFyAiIplDoSAiIhGF\ngoiIRBQKIiISUSiIiEhEoSAiIhGFgoiIRBQKIiISUSiIiEgkJ+4CttVuu+3mhYWFcZchItKqLFy4\ncJW7d29uvVYXCoWFhZSXl8ddhohIq2JmK5JZT91HIiISUSiIiEhEoSAiIpFWd0xBRNLv66+/pqKi\ngg0bNsRdijQjLy+Pnj17kpubu13PVyiISLMqKiro1KkThYWFmFnc5Ugj3J3Vq1dTUVFBnz59tmsb\nWdF9VFYGhYXQrl34t6ws7opEWpcNGzbQrVs3BUKGMzO6deu2Qy26Nt9SKCuDiROhpibcX7Ei3AcY\nNy6+ukRaGwVC67Cjn1ObbylMmrQlEDarqQnLRURka20+FN5/f9uWi0jmqaqq4j//8z+367mHH344\nVVVVTa4zefJk5s6du13br6+wsJBVq1a1yLbi0OZDoVevbVsuIjuupY/jNRUKmzZtavK5c+bMoUuX\nLk2uc8UVV3DooYdud31tSZsPhauvhvz8rZfl54flItLyNh/HW7EC3Lccx9uRYLjooot49913KSkp\n4YILLuDZZ59lxIgRnHzyyQwYMACAo48+miFDhlBUVMT06dOj527ec1++fDl9+/blZz/7GUVFRRx2\n2GGsX78egAkTJjBr1qxo/UsvvZTBgwczYMAA3nrrLQAqKyv50Y9+xODBgznzzDPp3bt3sy2CG264\ngf79+9O/f3+mTp0KwBdffMERRxzBwIED6d+/Pw899FD0O/br14/i4mLOP//87X+zdpS7t6rbkCFD\nfFvdf797797uZuHf++/f5k2IZLU333wz6XV793YPcbD1rXfv7X/9ZcuWeVFRUXR/3rx5np+f7++9\n9160bPXq1e7uXlNT40VFRb5q1apEPb29srLSly1b5u3bt/fFixe7u/vxxx/v9913n7u7n3rqqf7n\nP/85Wn/atGnu7n7rrbf66aef7u7u55xzjv/hD39wd/cnnnjCAa+srGzg9w+vV15e7v379/d169b5\n2rVrvV+/fr5o0SKfNWuWn3HGGdH6VVVVvnr1at933329trbW3d0///zz7X+zvOHPCyj3JL5j23xL\nAcIoo+XLobY2/KtRRyKpk67jePvvv/9WY/GnTZvGwIEDGTp0KB988AFvv/32N57Tp08fSkpKABgy\nZAjLly9vcNvHHnvsN9ZZsGABY8eOBWDkyJEUFBQ0Wd+CBQs45phj6NChAx07duTYY4/l+eefZ8CA\nAcydO5cLL7yQ559/ns6dO7PrrruSl5fHGWecwSOPPEJ+/e6NNMqKUBCR9EnXcbwOHTpEPz/77LPM\nnTuXF154gddee41BgwY1OFZ/5513jn5u3749GzdubHDbm9eru07Y2U5eY+vvu+++LFy4kAEDBnDx\nxRdzxRVXkJOTw8svv8xxxx3HY489xsiRI7fptVqSQkFEWlQqjuN16tSJtWvXNvp4dXU1BQUF5Ofn\n89Zbb/Hiiy9u/4s14oADDmDmzJkAPP3003z++edNrn/ggQfy2GOPUVNTwxdffMGjjz7KD3/4Q1au\nXEl+fj7jx4/n/PPPZ9GiRaxbt47q6moOP/xwpk6dyquvvtri9SerzZ+8JiLptbl7dtKk0GXUq1cI\nhB3ptu3WrRvDhg2jf//+jBo1iiOOOGKrx0eOHMntt99OcXEx++23H0OHDt2B36Bhl156KSeddBIP\nPfQQw4cPp0ePHnTq1KnR9QcPHsyECRPYf//9ATjjjDMYNGgQTz31FBdccAHt2rUjNzeX2267jbVr\n1zJ69Gg2bNiAu3PjjTe2eP3Jsm1tEsWttLTUdZEdkfRaunQpffv2jbuMWH355Ze0b9+enJwcXnjh\nBc4+++xY9+ib0tDnZWYL3b20ueeqpSAikoT333+fE044gdraWnbaaSfuvPPOuEtKCYWCiEgS9tln\nHxYvXhx3GSmnA80iIhJRKIiISEShICIiEYWCiIhEFAoi0iZ17NgRgJUrVzJmzJgG1znooINoboj7\n1KlTqalzUZZkpuJOxmWXXcaUKVN2eDstLWWhYGZ7mdk8M1tqZm+Y2S8bWMfMbJqZvWNmS8xscKrq\nEZHstMcee0QzoG6P+qGQzFTcrVkqWwobgd+4e19gKHCOmfWrt84oYJ/EbSJwWwrrEZFW6sILL9zq\negqXXXYZ119/PevWreOQQw6Jprl+/PHHv/Hc5cuX079/fwDWr1/P2LFjKS4u5sQTT4ymzgY4++yz\nKS0tpaioiEsvvRQIk+ytXLmSESNGMGLECGDri+g0NDV2U1N0N+bVV19l6NChFBcXc8wxx0RTaEyb\nNi2aTnvzZHzPPfccJSUllJSUMGjQoCan/9geKTtPwd0/Aj5K/LzWzJYCewJv1lltNHBvYlrXF82s\ni5n1SDxXRDLQeedBS5/IW1ICie/UBo0dO5bzzjuPn//85wDMnDmTJ598kry8PB599FF23XVXVq1a\nxdChQznqqKMavU7xbbfdRn5+PkuWLGHJkiUMHrylc+Lqq6+ma9eubNq0iUMOOYQlS5Zw7rnncsMN\nNzBv3jx22223rba1cOFC7rnnHl566SXcne9973sMHz6cgoIC3n77bR544AHuvPNOTjjhBB5++GHG\njx/f6O/305/+lJtvvpnhw4czefJkLr/8cqZOnco111zDsmXL2HnnnaMuqylTpnDrrbcybNgw1q1b\nR15eXrJvc1LSckzBzAqBQcBL9R7aE/igzv2KxDIRkcigQYP49NNPWblyJa+99hoFBQX06tULd+eS\nSy6huLiYQw89lA8//JBPPvmk0e3Mnz8/+nIuLi6muLg4emzmzJkMHjyYQYMG8cYbb/Dmm282thmg\n8amxIfkpuiFM5ldVVcXw4cMBOPXUU5k/f35U47hx47j//vvJyQn78MOGDePXv/4106ZNo6qqKlre\nUlJ+RrOZdQQeBs5z9zX1H27gKd+YjMnMJhK6l+il62iKxKqpPfpUGjNmDLNmzeLjjz+OulLKysqo\nrKxk4cKF5ObmUlhY2OCU2XU11IpYtmwZU6ZM4ZVXXqGgoIAJEyY0u52m5o2rP0V3c91HjfnrX//K\n/PnzmT17NldeeSVvvPEGF110EUcccQRz5sxh6NChzJ07l+9+97vbtf2GpLSlYGa5hEAoc/dHGlil\nAtirzv2ewMr6K7n7dHcvdffS7t27p6ZYEcloY8eO5cEHH2TWrFnRaKLq6mp23313cnNzmTdvHitW\nrGhyGwceeCBlieuCvv766yxZsgSANWvW0KFDBzp37swnn3zCE088ET2nsWm7G5sae1t17tyZgoKC\nqJVx3333MXz4cGpra/nggw8YMWIE1113HVVVVaxbt453332XAQMGcOGFF1JaWhpdLrSlpKylYCGO\n7wKWuvsNjaw2G/iFmT0IfA+o1vEEEWlIUVERa9euZc8996RHjx4AjBs3jp/85CeUlpZSUlLS7B7z\n2WefzWmnnUZxcTElJSXRtNYDBw5k0KBBFBUVsffeezNs2LDoORMnTmTUqFH06NGDefPmRcsbmxq7\nqa6ixsyYMYOzzjqLmpoa9t57b+655x42bdrE+PHjqa6uxt351a9+RZcuXfj973/PvHnzaN++Pf36\n9WPUqFHb/HpNSdnU2WZ2APA88L9AbWLxJUAvAHe/PREctwAjgRrgNHdvctCwps4WST9Nnd26ZOTU\n2e6+gIaPGdRdx4FzUlWDiIhsG53RLCIiEYWCiCSltV2lMVvt6OekUBCRZuXl5bF69WoFQ4Zzd1av\nXr1DJ7Tpymsi0qyePXtSUVFBZWVl3KVIM/Ly8ujZs+d2P1+hICLNys3NpU+fPnGXIWmg7iMREYko\nFEREJKJQEBGRiEJBREQiCgUREYkoFEREJKJQEBGRiEJBREQiCgUREYkoFEREJKJQEBGRiEJBREQi\nCgUREYkoFEREJKJQEBGRiEJBREQiCgUREYkoFEREJKJQEBGRiEJBREQiCgUREYkoFEREJKJQEBGR\niEJBREQiCgUREYkoFEREJKJQEBGRiEJBREQiCgUREYkoFEREJKJQEBGRSMpCwczuNrNPzez1Rh4/\nyMyqzezVxG1yqmoREZHk5KRw2/8N3ALc28Q6z7v7kSmsQUREtkHKWgruPh/4LFXbFxGRlhf3MYXv\nm9lrZvaEmRXFXIuISNZLZfdRcxYBvd19nZkdDjwG7NPQimY2EZgI0KtXr/RVKCKSZWJrKbj7Gndf\nl/h5DpBrZrs1su50dy9199Lu3buntU4RkWwSWyiY2bfMzBI/75+oZXVc9YiISAq7j8zsAeAgYDcz\nqwAuBXIB3P12YAxwtpltBNYDY93dU1WPiIg0L2Wh4O4nNfP4LYQhqyIikiHiHn0kIiIZRKEgIiIR\nhYKIiEQUCiIiElEoiIhIRKEgIiIRhYKIiEQUCiIiElEoiIhIRKEgIiIRhYKIiEQUCiIiElEoiIhI\nRKEgIiIRhYKIiEQUCiIiElEoiIhIRKEgIiIRhYKIiEQUCiIiElEoiIhIRKEgIiKRpELBzH5pZrta\ncJeZLTKzw1JdnIiIpFeyLYX/6+5rgMOA7sBpwDUpq0pERGKRbChY4t/DgXvc/bU6y0REpI1INhQW\nmtnThFB4ysw6AbWpK0tEROKQk+R6pwMlwHvuXmNmXQldSCIi0oYk21L4PvBPd68ys/HA74Dq1JUl\nIiJxSDYUbgNqzGwg8FtgBXBvyqoSEZFYJBsKG93dgdHATe5+E9ApdWWJiEgckj2msNbMLgZOAX5o\nZu2B3NSVJSIicUi2pXAi8CXhfIWPgT2B/0hZVSIiEoukQiERBGVAZzM7Etjg7jqmICLSxiQ7zcUJ\nwMvA8cAJwEtmNiaVhYmISPole0xhEvBv7v4pgJl1B+YCs1JVmIiIpF+yxxTabQ6EhNXb8FzJMO4w\nYwbceCNs2hR3NSKSSZL9Yn/SzJ4yswlmNgH4KzCnqSeY2d1m9qmZvd7I42Zm08zsHTNbYmaDt610\n2R5ffglnngkTJsCvfw2HHw6VlXFXJSKZItkDzRcA04FiYCAw3d0vbOZp/w2MbOLxUcA+idtEwgly\nkkIffQQHHwx33gmTJsEdd8Bzz8GgQfCPf8RdnYhkgmSPKeDuDwMPb8P6882ssIlVRgP3Jk6Ke9HM\nuphZD3f/KNnXkOS9/DIccwxUVcHMmXD88WH5v/0bjBkDw4fDddfBeeeBaf5bkazVZEvBzNaa2ZoG\nbmvNbM0OvvaewAd17lcklkkLmzEDDjwQdtoptAg2BwKEVsLChXDkkaE7acwYqNasViJZq8lQcPdO\n7r5rA7dO7r7rDr52Q/uj3uCKZhPNrNzMyivVAZ60jRvDnv+ECTBsGJSXw8CB31yvSxd45BGYMgUe\nfxxKS+HVV9NerohkgDhHEFUAe9W53xNY2dCK7j7d3UvdvbR79+5pKa61W7UKfvxjuOmmEAxPPQXd\nujW+vhn85jfw7LNQUwNDh8Jdd4WRSiKSPeIMhdnATxOjkIYC1Tqe0DJeey0cK/j737cMPc1J8ujR\nAQfA4sXwwx/CGWfAaaeFkJDU+vJLuOGG0J3329/CihVxVyTZKmWhYGYPAC8A+5lZhZmdbmZnmdlZ\niVXmAO8B7wB3Aj9PVS3ZZOZM+MEP4Ouv4fnn4ac/3fZt7L47PPkkTJ4M994bWg3/+lfL1yqhJfbw\nw1BUFFpqtbUhHPbeG447LowOU2tN0srdW9VtyJAhLt+0caP7xRe7g/sPfuD+0Ucts92nnnLv1s29\nY0f3hx5qmW1K8PLL7gccED6zoiL3J58My1escL/oIveuXcNjAwe633WXe01NvPVKvFauDLftBZR7\nEt+xOiu5DaiqgqOOgj/+EX72M3jmGfjWt1pm24cdFrqTBgyAE0+Ec8+Fr75qmW1nq/ffh/HjYf/9\nQwvsjjvCgf0f/zg83qtX+CwrKsI5JbW1cPrpsNde4fySiop465f0+OqrcIzvoougpAT22AOmTUvD\nCyeTHJl0U0tha0uXuu+7r3tOjvttt6Xudb780v2888Ke6/e+F/ZmZdusWeN+ySXueXnhdsklYVlz\namvd581zP/po93bt3Nu3dz/hBPe//z08Jm3HsmXh73j06NA6B/fcXPcRI9yvvTb8vW8vkmwpxP4l\nv603hcIWf/mL+667uu++u/v8+el5zVmzwmt27eo+Z056XrO1+/pr9zvuCJ8TuI8bt/2h+t577uef\n796lS9jWkCHu997rvmFDy9Ys6VFT4/7EE+6//KX7fvuFzxTcCwvdzz7b/fHHk9txSIZCoQ2rrXW/\n8kp3M/fBg93ffz+9r/+vf7kXF4f/PZMmheMZ0rAnn3Tv3z+8Vwcc4P7SSy2z3XXrwh5l375h27vv\n7j558o71OUvq1daGvf0bb3T/8Y9DixHCv6NGud90k/s//5maFqBCoY1au9b9uOO27HHGdfCxpsb9\n9NNDHQcf7P7xx/HUkalef9195Mjw/uy9d2hhpeIPvbbW/emn3Y88Muwk5OaG/xctFT6y46qr3R99\n1P3MM917997SGujb1/1XvwqDOdLxd6xQaIPefdd9wIDQrzxlSmb0J99zj/suu7j36OH+3HNxVxO/\njz8Of/zt2rl37ux+/fXp69p5++3QDdGpU/jLHjrU/U9/CseDJH1qa90XL3b/4x/dhw8Px/sgfC7H\nHBO6EpcvT39dCoU25n/+J/TjFxSEPcNMsmSJ+z77hAOg117rvmlT3BWlX02N+x/+EP7wc3Lczz3X\nfdWqeGpZs8b95pvDZwIhsK+80v2TT+KpJxusWuX+wAPup57q/q1vbWkNDBoUhoo/95z7V1/FW6NC\noY2orXW/4Yaw59m/v/s778RdUcOqq92PPz78j/rJT9w/+yzuitJj0yb3sjL3Xr3C7z56dOgTzgSb\nNoXBAJu7sXbaKXxpLVoUd2Wt38aN7i+8EI7j7L9/6LqDsON20knuM2a03LlCLUWh0AbU1Lifckr4\nlI49NhxPyGS1teFAWW5uGD1RXh53Ram1YEH4QgD3khL3Z56Ju6LGLV3qfs457h06eHTQe+bMMDJK\nkvPhh+533x2GAxcUhPexXTv373/f/fLLw3GcTB50kWwoWFi39SgtLfXy8vK4y0i5Dz6AY48NM5te\ncUU4aaldKznV8MUX4YQT4JNPYOpUOOustnWNhnffDScUzZoVTii6+mo45RRo3z7uyppXXQ133w23\n3ALvvRdOiPv5z8NJj01NmNja1NaG+aTWr4cNG8Kt7s/17ze33qJFsGRJ2HaPHjByZLgdeih07Rrv\n75osM1vo7qXNrqdQyDwLFoR5b9avh/vvD2crtzarV4ezdp98Ek4+OZy127Fj3FXtmKoquOqqcFZp\nbm6YuO7886FDh7gr23abNsGcOWEW3b/9DfLywud17rnh7PWWfq3168PEiuvXb32rv6yhdbbnS3xH\nz7pv3x522SW8L3l58J3vbAmC4uLWuZOjUGil7rgD/v3fobAwXNugb9+4K9p+tbVhuobJk2G//cKe\ndb9+cVe17b7+Gm6/HS6/HD77LFyf4qqrQiuhLXjjDbj55jD54fr1MGJEaPnk5Gz/F3ndZV9/vX11\ntWsXvpB32WXrL+jNyxr6uaUeS3ZW4dZEodDKfPVV2Eu74w4YNQr+9Kdw8Zu24Jln4KSTYN26MJfP\nySfHXVFy3OEvf4ELLghzFB18MFx/fZiHpi367LNwDY1bbgnzMzUkP3/Ll/TmW6qW5ea2zj3yTKVQ\naEU++SRcBnPBgtBXfdVVraN/elusXAljx4bpvM86K1zjIS8v7qoat3hxmMp63rzQyvmP/wiXLM2G\nL6mNG+Htt8PlW+t+We+8c3b8/m1VsqHQBhtJrUt5ORx9dNhLe/DBMBNpW7THHqHFMGkSXHcdvPxy\n6E7q0yfuyrb24Yfwu9+FixN17Rq6Vc48M+y1ZoucnNbdbSk7ppWMZ2mb7rsvXOksJwf+8Y+2Gwib\n5eTAtdeGYyXvvguDB8Ps2XFXFXzxBVx2Gey7b+i6+81v4J134Be/yK5AEFFLIQYbN4aRKzfeCAcd\nFK6Wlk2Xnj7qqDDE7/jjYfTocNUx2HwO6JYrjdW9v73Lk33O5hEsxx8P11wTrnwmko0UCim2aVMY\nG/7551tu114bhgGeey5MmZKde6J77x2uIX3VVfDmm2GZWcO3dDyWkxOO6wwblt73QSTTKBSSUFsL\na9aEfv+6X+51b409Vl39zWvs7rRTOIHotNPi+X0yRV5eCAURyRxZEwru4Yu9uS/xhh6rqmr64uk7\n7QQFBVtuPXqE8fh1l22+de0K3/52WEdEJNNkTSg88ACMG9f447m5W3957757GIrYtWvDX+6bv+AL\nCsJwPQ3VE5G2IGtCobQ09N839iXfoYO+2EVEsiYU9t03DDMUEZHG6TwFERGJKBRERCSiUEijsrIw\n+2m7duHfsrK4KxIR2VrWHFOIW1kZTJwYphQGWLEi3IemR0WJiKSTWgppMmnSlkDYrKYmLBcRyRQK\nhTRpbH76xpaLiMRBoZAmvXpt23IRkTgoFNLk6qvDFaXqys8Py0VEMoVCIU3GjYPp06F373DmdO/e\n4b4OMotIJtHoozQaN04hICKZTS0FERGJKBRERCSiUBARkYhCQUREIikNBTMbaWb/NLN3zOyiBh6f\nYGaVZvZq4nZGKuuRzKK5oDKPPhNJ2egjM2sP3Ar8CKgAXjGz2e7+Zr1VH3L3X6SqDslMmgsq8+gz\nEUhtS2F/4B13f8/dvwIeBEan8PWkFdFcUJlHn4lAakNhT+CDOvcrEsvqO87MlpjZLDPbq6ENmdlE\nMys3s/LKyspU1CppprmgMo8+E4HUhkJDVzz2evf/AhS6ezEwF5jR0Ibcfbq7l7p7affu3Vu4TImD\n5oLKPPpMBFIbChVA3T3/nsDKuiu4+2p3/zJx905gSArrkQyiuaAyjz4TgdSGwivAPmbWx8x2AsYC\ns+uuYGY96tw9Cliawnokg2guqMyjz0QAzL1+j04LbtzscGAq0B64292vNrMrgHJ3n21mfySEwUbg\nM+Bsd3+rqW2WlpZ6eXl5ymoWEWmLzGyhu5c2t15Kz1Nw9znuvq+7f9vdr04sm+zusxM/X+zuRe4+\n0N1HNBcI0jI0Fl1EGqNZUrOMxqKLSFM0zUWW0Vh0EWmKQiHLaCy6iDRFoZBlNBZdRJqiUMgyGosu\nIk1RKGQZjUUXkaZo9FEW0rWiRaQxaimIiEhEoSAiIhGFgoiIRBQKIiISUSiIiEhEoSAiIhGFgoiI\nRBQKIiISUSiIiEhEoSAiIhGFgoiIRBQKIiISUSiIiEhEoSAiIhGFgoiIRBQKkvXKyqCwENq1C/+W\nlcVdkUh8dJEdyWplZTBxItTUhPsrVoT7oAsRSXZSS0Gy2qRJWwJhs5qasFwkGykUJKu9//62LU8l\ndWNJJlAoSFbr1WvblqfK5m6sFSvAfUs3loJB0k2hIFnt6qshP3/rZfn5YXk6qRtLMoVCQbLauHEw\nfTr07g1m4d/p09N/kDmTurEku2n0kWS9cePiH2nUq1foMmpouUg6qaUgkgEypRtLRKEgkgEypRtL\nRKEgkiHGjYPly6G2NvybzYGg4bnxUSiISEbJpOG52RhOCgURySiZMjw3W8PJ3D11W0+B0tJSLy8v\nj7sMEUmRdu3Cl3B9ZqFrLV0KCxseEda7d+jeS5f683NBGISwrceczGyhu5c2t15KWwpmNtLM/mlm\n75jZRQ08vrOZPZR4/CUzK0xlPSKS+TLlLPNMOXck3S2nlIWCmbUHbgVGAf2Ak8ysX73VTgc+d/fv\nADcC16aqHhFpHTJleG62hlMqWwr7A++4+3vu/hXwIDC63jqjgRmJn2cBh5iZpbAmEclwmTI8N1vD\nKZWhsCfwQZ37FYllDa7j7huBaqBb/Q2Z2UQzKzez8srKyhSVKyKZIhOG52ZrOKUyFBra469/+CiZ\ndXD36e5e6u6l3bt3b5HiRESak43hlMq5jyqAverc7wmsbGSdCjPLAToDn6WwJhGRVied83OlsqXw\nCrCPmfUxs52AscDseuvMBk5N/DwGeMZb2xhZEZE2JGUtBXffaGa/AJ4C2gN3u/sbZnYFUO7us4G7\ngPvM7B1CC2FsquoREZHmpXTqbHefA8ypt2xynZ83AMensgYREUmeprkQEZGIQkFERCKtbu4jM6sE\nGpiRpFXZDVgVdxEZRO/H1vR+bKH3Yms78n70dvdmx/S3ulBoC8ysPJmJqbKF3o+t6f3YQu/F1tLx\nfqj7SEREIgoFERGJKBTiMT3uAjKM3o+t6f3YQu/F1lL+fuiYgoiIRNRSEBGRiEIhjcxsLzObZ2ZL\nzewNM/tl3DXFzczam9liM/t/cdcSNzPrYmazzOytxP+R78ddU5zM7FeJv5PXzewBM8uLu6Z0MrO7\nzexTM3u9zrKuZvY/ZvZ24t+Cln5dhUJ6bQR+4+59gaHAOQ1cjS7b/BJYGncRGeIm4El3/y4wkCx+\nX8xsT+BcoNTd+xPmT8u2udH+GxhZb9lFwN/cfR/gb4n7LUqhkEbu/pG7L0r8vJbwR1//wkNZw8x6\nAkcA/xV3LXEzs12BAwmTROLuX7l7VbxVxS4H2CUxrX4+35x6v01z9/l881ICda9WOQM4uqVfV6EQ\nEzMrBAYBL8VbSaymAr8FauMuJAPsDVQC9yS60/7LzDrEXVRc3P1DYArwPvARUO3uT8dbVUb4P+7+\nEYSdTGD3ln4BhUIMzKwj8DBwnruvibueOJjZkcCn7r4w7loyRA4wGLjN3QcBX5CCroHWItFXPhro\nA+wBdDCz8fFWlR0UCmlmZrmmfIhoAAACqUlEQVSEQChz90firidGw4CjzGw58CBwsJndH29JsaoA\nKtx9c8txFiEkstWhwDJ3r3T3r4FHgB/EXFMm+MTMegAk/v20pV9AoZBGZmaEPuOl7n5D3PXEyd0v\ndvee7l5IOID4jLtn7Z6gu38MfGBm+yUWHQK8GWNJcXsfGGpm+Ym/m0PI4gPvddS9WuWpwOMt/QIp\nvciOfMMw4BTgf83s1cSySxIXIxL5d6Ascfna94DTYq4nNu7+kpnNAhYRRu0tJsvObjazB4CDgN3M\nrAK4FLgGmGlmpxOCs8UvUqYzmkVEJKLuIxERiSgUREQkolAQEZGIQkFERCIKBRERiSgURNLIzA7S\njLCSyRQKIiISUSiINMDMxpvZy2b2qpndkbjuwzozu97MFpnZ38yse2LdEjN70cyWmNmjm+e4N7Pv\nmNlcM3st8ZxvJzbfsc51E8oSZ+yKZASFgkg9ZtYXOBEY5u4lwCZgHNABWOTug4HnCGeYAtwLXOju\nxcD/1lleBtzq7gMJ8/Z8lFg+CDgP6EeYHXVYyn8pkSRpmguRbzoEGAK8ktiJ34Uw8Vgt8FBinfuB\nR8ysM9DF3Z9LLJ8B/NnMOgF7uvujAO6+ASCxvZfdvSJx/1WgEFiQ+l9LpHkKBZFvMmCGu1+81UKz\n39dbr6k5YprqEvqyzs+b0N+hZBB1H4l809+AMWa2O0TXxe1N+HsZk1jnZGCBu1cDn5vZDxPLTwGe\nS1wno8LMjk5sY2czy0/rbyGyHbSHIlKPu79pZr8DnjazdsDXwDmEC98UmdlCoJpw3AHCFMa3J770\n685uegpwh5ldkdhGi89oKdLSNEuqSJLMbJ27d4y7DpFUUveRiIhE1FIQEZGIWgoiIhJRKIiISESh\nICIiEYWCiIhEFAoiIhJRKIiISOT/A9+SWn0g+iddAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x239e2d78ac8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "acc = history.history['acc']\n",
    "val_acc = history.history['val_acc']\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "epoch = range(1 ,len(acc) + 1)\n",
    "plt.plot(epoch ,acc ,'bo' ,label = 'training accuracy')\n",
    "plt.plot(epoch , val_acc ,'b-' ,label= 'validation accuracy')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('accuracy')\n",
    "plt.legend()\n",
    "plt.show()\n",
    "plt.plot(epoch ,loss ,'bo' ,label ='training loss')\n",
    "plt.plot(epoch ,val_loss ,'b-' ,label ='validation loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.ylabel('loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding_4 (Embedding)      (None, 100, 100)          1000000   \n",
      "_________________________________________________________________\n",
      "flatten_4 (Flatten)          (None, 10000)             0         \n",
      "_________________________________________________________________\n",
      "dense_7 (Dense)              (None, 32)                320032    \n",
      "_________________________________________________________________\n",
      "dense_8 (Dense)              (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 1,320,065\n",
      "Trainable params: 1,320,065\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Train on 200 samples, validate on 10000 samples\n",
      "Epoch 1/10\n",
      "200/200 [==============================] - 1s 6ms/step - loss: 0.6948 - acc: 0.4950 - val_loss: 0.6944 - val_acc: 0.4999\n",
      "Epoch 2/10\n",
      "200/200 [==============================] - 1s 4ms/step - loss: 0.5028 - acc: 0.9750 - val_loss: 0.7096 - val_acc: 0.4959\n",
      "Epoch 3/10\n",
      "200/200 [==============================] - 1s 5ms/step - loss: 0.2909 - acc: 0.9800 - val_loss: 0.7013 - val_acc: 0.5096\n",
      "Epoch 4/10\n",
      "200/200 [==============================] - 1s 4ms/step - loss: 0.1352 - acc: 1.0000 - val_loss: 0.7194 - val_acc: 0.4992\n",
      "Epoch 5/10\n",
      "200/200 [==============================] - 1s 4ms/step - loss: 0.0648 - acc: 1.0000 - val_loss: 0.7187 - val_acc: 0.4973\n",
      "Epoch 6/10\n",
      "200/200 [==============================] - 1s 4ms/step - loss: 0.0343 - acc: 1.0000 - val_loss: 0.7271 - val_acc: 0.4997\n",
      "Epoch 7/10\n",
      "200/200 [==============================] - 1s 4ms/step - loss: 0.0193 - acc: 1.0000 - val_loss: 0.7266 - val_acc: 0.5099\n",
      "Epoch 8/10\n",
      "200/200 [==============================] - 1s 4ms/step - loss: 0.0112 - acc: 1.0000 - val_loss: 0.7423 - val_acc: 0.5018\n",
      "Epoch 9/10\n",
      "200/200 [==============================] - 1s 5ms/step - loss: 0.0066 - acc: 1.0000 - val_loss: 0.7434 - val_acc: 0.5099\n",
      "Epoch 10/10\n",
      "200/200 [==============================] - 1s 4ms/step - loss: 0.0040 - acc: 1.0000 - val_loss: 0.7579 - val_acc: 0.5025\n"
     ]
    }
   ],
   "source": [
    "#Training the same model without pretrained word embeddings\n",
    "from keras.models import  Sequential\n",
    "from keras.layers import Flatten ,Dense ,Embedding\n",
    "model1 = Sequential()\n",
    "model1.add(Embedding(max_words ,emd_dims ,input_length =maxlen))                 # will give 3D output\n",
    "model1.add(Flatten())                                                            \n",
    "model1.add(Dense(32 ,activation ='relu'))\n",
    "model1.add(Dense(1 ,activation ='sigmoid'))\n",
    "model1.summary()\n",
    "model1.compile(optimizer ='rmsprop' ,loss= 'binary_crossentropy' ,metrics =['acc'])\n",
    "history1 = model1.fit(x_train ,y_train ,epochs =10 ,batch_size =32 ,validation_data =(x_val ,y_val))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEKCAYAAAD9xUlFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x239e2d6d978>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEKCAYAAAD9xUlFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAIABJREFUeJzt3X10VPW97/H3lwBywpOIsYsCIajY\nQiAQmMOlpYpcsQvEilofsKFLvNK0Lj31tEevWGpx0UVPr9XWshZ9SJ+uR3OKlKrlnuLBS4sP3KWV\ngGAL6CrPRHyIHBLBCPLwvX/syc4kmSSTkJ09IZ/XWrNm9p7f3vPNKL/P7KffNndHREQEoEfcBYiI\nSPZQKIiISEihICIiIYWCiIiEFAoiIhJSKIiISEihICIiIYWCiIiEFAoiIhLqGXcBbXX++ed7QUFB\n3GWIiHQpmzZtet/d81pr1+VCoaCggIqKirjLEBHpUsxsXybttPtIRERCCgUREQkpFEREJKRQEBGR\nkEJBRERCCgUREQkpFEREJNTlrlMQEekOjh6FAwfqH5WVcPXVMGlStJ+rUBAR6WQffVTf0ad2/KkB\nUF3ddLlPfEKhICLSpRw/Dm+9lb6jr3t96FDT5c4/H4YPhwsvhGnTgtfDhgXPw4fD0KHQu3f09SsU\nREQydOIEHDzY8q/8995rutygQfWd+5QpTTv8YcOgT5/O/3vSUSiISNY4eTLodPftg717g10op0+n\nf5w61fx7Z/p+6nunTsH77wcd/ttvg3vDmgcMqO/cJ05s2NnXdfh9+8bydbaLQkFEOs2xY0Hnundv\n0PE3flRWBp1wW5lBjx7pHzk57Xuv7n2zYNfOzJnpO/wBAzr8a4qVQkFEOsyRI/W/8tN1+u+807B9\njx5BxzpiBFx6afCc+jjvvKBjbq3zNovlzz0rKRREJCPuwQHSljr9w4cbLnPOOZCfH3Tws2c37fSH\nDoVevWL5c6QZCgWRTnb6NNTUBPvLDx+uf1RXB7+03YNfvnWPul/C7X20Z/na2vSdfm1tw7+lf//6\nDv6zn23Y4RcUwAUXBJ8vXYdCQaQdTpwIOvHGHXtd597SvJqapgcrs9X55wcd/OjRwT71xp3+uedq\n183ZRqEgne7UqeCA4s6dsGtXcPCx7tds6j7iTJ7b0ralZdzhgw8y79iPHm35b+zTJ+gwBw0KHkOG\nwJgxDefVPVLn9e8f1OTe9HH6dPr5mTzaumyfPsFun6501ox0DIWCROLEiWB3w65dQeef+ti9Gz7+\nOO4KW9evX8MOe+TI4JTD1jr2QYOy55xzkbZSKEi7HT8Oe/Y07fR37gwORKaeWti3L1x8MRQWwpw5\nweuLLw6u3uzbt+Gv2brzwxvPa+35TJZxD04tTO3kdQBUuqNIQ8HMZgI/BnKAX7r79xu9/yNgenIy\nF7jA3c+NsqbO4B5c7LJ/f/1j37761x9+WP/LMtPn/v3jOWBXWxv8sk/X8e/f33Df+IABMGoUJBIw\nd259x3/xxcGYLdr3LJL9IgsFM8sBlgNXApXARjNb7e7b69q4+zdS2v8TUBxVPR3p2LFgn3i6Dr/u\ncexYw2Vyc4N9tPn5wUUvNTXB1ZE7dtTvr27p4GOPHjBwYPrQyCRYWtqdceRI+t08O3cGY7ikGjw4\n6OQ/97mGnf7FFwfvqeMX6dqi3FKYDOx0990AZrYCmANsb6b9LcDiKAopL4dFi4LOOj8fli6FkpL0\nbVv7lb9/P7z7btPlhgwJ1j1+PFxzTX0A1D3OO6/lDvP06aBzrjuYme658bw33qifbnyqYGOpBz7r\nnmtqgo6/8d/ziU8EnfyMGQ07/YsuCpYTkbNXlKEwFDiQMl0J/Ld0Dc1sBDAS+HMz75cCpQD5+flt\nKqK8HEpL6zvNfftgwQLYtg0uuaTtv/LHjw9Ox0vt8IcODS7SORN1WwIDBwbrb6vjx4NOvrlAaRws\n774b7Mu/+uqmHX///mf2t4hI1xVlKKT7XdzcDpK5wCp3TzvqibuXAWUAiUSiTWd4L1rU9Ff0sWPw\nr/9aP32mv/KzwTnnBBcKXXBB3JWISFcWZShUAsNTpocBB5tpOxe4M4oi9u9v/r1duzrmV76IyNki\nyvNZNgKjzGykmfUm6PhXN25kZp8CBgEvR1FEc3ubRowITodUIIiI1IssFNz9JHAXsBbYAax0921m\ntsTMrklpeguwwj2aC/+XLg2OCaTKzQ3mi4hIQ5Fep+Dua4A1jeZ9p9H0g1HWUHeWUaZnH4mIdGfd\n4ormkhKFgIhIJjSorYiIhBQKIiISUiiIiEhIoSAiIiGFgoiIhBQKIiISUiiIiEhIoSAiIiGFgoiI\nhBQKIiISUiiIiEhIoSAiIiGFgoiIhBQKIiISUiiIiEhIoSAiIqFIQ8HMZprZm2a208wWNtPmJjPb\nbmbbzOzfo6xHRERaFtmd18wsB1gOXAlUAhvNbLW7b09pMwq4H5jq7ofN7IKo6hERkdZFuaUwGdjp\n7rvd/WNgBTCnUZuvAMvd/TCAu78XYT0iItKKKENhKHAgZboyOS/VJcAlZvb/zOwVM5sZYT0iItKK\nyHYfAZZmnqf5/FHA5cAw4CUzG+vu1Q1WZFYKlALk5+d3fKUiIgJEu6VQCQxPmR4GHEzT5g/ufsLd\n9wBvEoREA+5e5u4Jd0/k5eVFVrCISHcXZShsBEaZ2Ugz6w3MBVY3avMMMB3AzM4n2J20O8KaRESk\nBZGFgrufBO4C1gI7gJXuvs3MlpjZNclma4FDZrYdWA/c6+6HoqpJRERaZu6Nd/Nnt0Qi4RUVFXGX\nISLSpZjZJndPtNZOVzSLiEhIoSAiIiGFgoiIhBQKIiISUiiIiEhIoSAiIiGFgoiIhBQKIiISUiiI\niEhIoSAiIiGFgoiIhBQKIiISUiiIiEhIoSAiIiGFgoiIhBQKIiISUiiIiEgo0lAws5lm9qaZ7TSz\nhWnen29mVWa2JflYEGU9cSsvh4IC6NEjeC4vj7siEZGGeka1YjPLAZYDVwKVwEYzW+3u2xs1fdLd\n74qqjmxRXg6lpVBbG0zv2xdMA5SUxFeXiEiqKLcUJgM73X23u38MrADmRPh5WW3RovpAqFNbG8wX\nEckWUYbCUOBAynRlcl5jXzSz181slZkNj7CeWO3f37b5IiJxiDIULM08bzT9f4ACdy8C1gGPpV2R\nWamZVZhZRVVVVQeX2Tny89s2X0QkDlGGQiWQ+st/GHAwtYG7H3L348nJXwCT0q3I3cvcPeHuiby8\nvEiKjdrSpZCb23Bebm4wX0QkW0QZChuBUWY20sx6A3OB1akNzGxIyuQ1wI4I64lVSQmUlcGIEWAW\nPJeV6SCziGSXyM4+cveTZnYXsBbIAX7t7tvMbAlQ4e6rga+b2TXASeC/gPlR1ZMNSkoUAiKS3cy9\n8W7+7JZIJLyioiLuMkREuhQz2+Tuidba6YpmEREJKRRERCSkUBARkZBCQUREQgoFEREJKRRERCSk\nUBARkZBCQUREQgoFEREJKRRERCSkUBARkZBCQUREQgoFEREJKRRERCSkUBARkZBCQUREQgoFEREJ\nRRoKZjbTzN40s51mtrCFdjeYmZtZq3cFEhGR6EQWCmaWAywHZgFjgFvMbEyadv2BrwN/iaoWERHJ\nTEahYGZ3m9kAC/zKzDab2edbWWwysNPdd7v7x8AKYE6adt8FHgKOtalyERHpcJluKfwPd/8A+DyQ\nB9wGfL+VZYYCB1KmK5PzQmZWDAx39//IsA4REYlQpqFgyeergN+4+9aUea0tk8rDN816AD8C/qXV\nDzcrNbMKM6uoqqrKsGQREWmrTENhk5k9RxAKa5PHAU63skwlMDxlehhwMGW6PzAWeN7M9gJTgNXp\nDja7e5m7J9w9kZeXl2HJIiLSVj0zbHc7MAHY7e61ZnYewS6klmwERpnZSOAtYC7wpbo33b0GOL9u\n2syeB+5x94rMyxcRkY6U6ZbCZ4A33b3azOYB3wZqWlrA3U8CdwFrgR3ASnffZmZLzOyaMylaRESi\nYe7eeiOz14HxQBHwOPAr4Hp3nxZteU0lEgmvqNDGhIhIW5jZJndv9VqwTLcUTnqQHnOAH7v7jwmO\nCYiIyFkk02MKR8zsfuDLwKXJC9N6RVeWiIjEIdMthZuB4wTXK7xDcL3BDyKrSkREYpFRKCSDoBwY\naGZXA8fc/d8irUxERDpdpsNc3AS8CtwI3AT8xcxuiLIwERHpfJkeU1gE/KO7vwdgZnnAOmBVVIWJ\niEjny/SYQo+6QEg61IZlRUSki8h0S+E/zWwt8Nvk9M3AmmhKEhGRuGQUCu5+r5l9EZhKMNBdmbs/\nHWllIiLS6TLdUsDdfw/8PsJaREQkZi2GgpkdIWW469S3AHf3AZFUJSIisWgxFNxdQ1mIiHQjOoNI\nRERCCgUREQkpFEREJKRQEBGRkEJBRERCCgUREQlFGgpmNtPM3jSznWa2MM37XzOzv5rZFjPbYGZj\noqxHRERaFlkoJO/OthyYBYwBbknT6f+7u49z9wnAQ8APo6pHRERaF+WWwmRgp7vvdvePgRUE93gO\nufsHKZN9SX/1tIiIdJIoQ2EocCBlujI5rwEzu9PMdhFsKXw93YrMrNTMKsysoqqqKpJiu5Pycigo\ngB49gufy8rgrEpFsEWUoWJp5TbYE3H25u18E3Ad8O92K3L3M3RPunsjLy+vgMruX8nIoLYV9+8A9\neC4tVTCISCDKUKgEhqdMDwMOttB+BXBthPUIsGgR1NY2nFdbG8wXEYkyFDYCo8xspJn1BuYCq1Mb\nmNmolMnZwN8jrEeA/fvbNl9EupeM76fQVu5+0szuAtYCOcCv3X2bmS0BKtx9NXCXmc0ATgCHgVuj\nqkcC+fnBLqN080VEIgsFAHdfQ6Pbdrr7d1Je3x3l50tTS5cGxxBSdyHl5gbzRUR0RXM3U1ICZWUw\nYgSYBc9lZcF8EZFItxQkO5WUKAREJD1tKYiISEihICIiIYWCiIiEFAoiIhJSKIiISEihICIiIYWC\niIiEFAoiIhJSKIiISEihICIiIYWCiIiEFAoiIhJSKIiISEihICIiIYWCiIiEIg0FM5tpZm+a2U4z\nW5jm/W+a2XYze93M/mRmI6KsR0REWhZZKJhZDrAcmAWMAW4xszGNmr0GJNy9CFgFPBRVPSIi0roo\ntxQmAzvdfbe7fwysAOakNnD39e5ed7fgV4BhEdYjIiKtiDIUhgIHUqYrk/OaczvwbIT1iIhIK6K8\nR7OlmedpG5rNAxLAtGbeLwVKAfLz8zuqPhERaSTKLYVKYHjK9DDgYONGZjYDWARc4+7H063I3cvc\nPeHuiby8vEiKFRGRaENhIzDKzEaaWW9gLrA6tYGZFQM/JwiE9yKsRUREMhBZKLj7SeAuYC2wA1jp\n7tvMbImZXZNs9gOgH/A7M9tiZqubWZ2IiHSCKI8p4O5rgDWN5n0n5fWMKD9fRETaRlc0i4hISKEg\nIiIhhYKIiIQUCiIiElIoiIhISKEgIiIhhYKIiIQUCiIiElIoiIhISKEgsSkvh4IC6NEjeC4vj7si\nEYl0mAuR5pSXQ2kp1CZvsbRvXzANUFISX10i3Z22FCQWixbVB0Kd2tpgvojER6Egsdi/v23zRaRz\nKBQkFs3dQE831hOJl0JBYrF0KeTmNpyXmxvMF5H4KBQkFiUlUFYGI0aAWfBcVqaDzCJx09lHEpuS\nEoWASLaJdEvBzGaa2ZtmttPMFqZ5/zIz22xmJ83shihrERGR1kW2pWBmOcBy4EqgEthoZqvdfXtK\ns/3AfOCeM/msEydOUFlZybFjx85kNdIJ+vTpw7Bhw+jVq1fcpYhIGlHuPpoM7HT33QBmtgKYA4Sh\n4O57k++dPpMPqqyspH///hQUFGBmZ7IqiZC7c+jQISorKxk5cmTc5YhIGlHuPhoKHEiZrkzO63DH\njh1j8ODBCoQsZ2YMHjxYW3QiWSzKUEjXQ3u7VmRWamYVZlZRVVXVXJv2rFo6mf47iWS3KEOhEhie\nMj0MONieFbl7mbsn3D2Rl5fXIcV1pOrqan7yk5+0a9mrrrqK6urqFtt85zvfYd26de1af2MFBQW8\n//77HbIuETn7RBkKG4FRZjbSzHoDc4HVEX5exjp6dM6WQuHUqVMtLrtmzRrOPffcFtssWbKEGTNm\ntLs+EZFMRRYK7n4SuAtYC+wAVrr7NjNbYmbXAJjZP5pZJXAj8HMz2xZVPXXqRufctw/c60fnPJNg\nWLhwIbt27WLChAnce++9PP/880yfPp0vfelLjBs3DoBrr72WSZMmUVhYSFlZWbhs3S/3vXv3Mnr0\naL7yla9QWFjI5z//eT766CMA5s+fz6pVq8L2ixcvZuLEiYwbN4433ngDgKqqKq688komTpzIV7/6\nVUaMGNHqFsEPf/hDxo4dy9ixY3n00UcB+PDDD5k9ezbjx49n7NixPPnkk+HfOGbMGIqKirjnnjM6\nWUxEspm7d6nHpEmTvLHt27c3mdecESPcgzho+BgxIuNVNLFnzx4vLCwMp9evX++5ubm+e/fucN6h\nQ4fc3b22ttYLCwv9/fffT9YzwquqqnzPnj2ek5Pjr732mru733jjjf7444+7u/utt97qv/vd78L2\ny5Ytc3f35cuX++233+7u7nfeead/73vfc3f3Z5991gGvqqpK8/cHn1dRUeFjx471o0eP+pEjR3zM\nmDG+efNmX7VqlS9YsCBsX11d7YcOHfJLLrnET58+7e7uhw8fbv+X5W377yUiHQOo8Az62G43zEVn\njc45efLkBqddLlu2jPHjxzNlyhQOHDjA3//+9ybLjBw5kgkTJgAwadIk9u7dm3bd119/fZM2GzZs\nYO7cuQDMnDmTQYMGtVjfhg0buO666+jbty/9+vXj+uuv56WXXmLcuHGsW7eO++67j5deeomBAwcy\nYMAA+vTpw4IFC3jqqafIbTxokYicNbpdKHTW6Jx9+/YNXz///POsW7eOl19+ma1bt1JcXJz2tMxz\nzjknfJ2Tk8PJkyfTrruuXWqb4IdA5pprf8kll7Bp0ybGjRvH/fffz5IlS+jZsyevvvoqX/ziF3nm\nmWeYOXNmmz4r2+kOcCL1ul0oRDE6Z//+/Tly5Eiz79fU1DBo0CByc3N54403eOWVV9r/Yc343Oc+\nx8qVKwF47rnnOHz4cIvtL7vsMp555hlqa2v58MMPefrpp7n00ks5ePAgubm5zJs3j3vuuYfNmzdz\n9OhRampquOqqq3j00UfZsmVLh9cflyiOMYl0Zd1uQLy6AdgWLQp2GeXnB4FwJgOzDR48mKlTpzJ2\n7FhmzZrF7NmzG7w/c+ZMfvazn1FUVMSnPvUppkyZcgZ/QXqLFy/mlltu4cknn2TatGkMGTKE/v37\nN9t+4sSJzJ8/n8mTJwOwYMECiouLWbt2Lffeey89evSgV69e/PSnP+XIkSPMmTOHY8eO4e786Ec/\n6vD649LSHeA0WJ90R9bW3Q5xSyQSXlFR0WDejh07GD16dEwVZYfjx4+Tk5NDz549efnll7njjjuy\n9hd9Nv336tEj2EJozAxOn9HgKyLZxcw2uXuitXbdbkvhbLV//35uuukmTp8+Te/evfnFL34Rd0ld\nQn5+sMso3XyR7kihcJYYNWoUr732WtxldDlLlwbHEFJ3IekOcNKddbsDzSKpdAc4kYa0pSDdnu4A\nJ1JPWwoiWUDXSki20JaCSMzqrpWoO65Rd60EaAtGOp+2FGLSr18/AA4ePMgNN6S/PfXll19O49Nv\nG3v00UepTTlKmslQ3Jl48MEHefjhh894PdK6lq6VEOlsCoWYffKTnwxHQG2PxqGQyVDckl06azwu\nkUwoFDrAfffd1+B+Cg8++CCPPPIIR48e5YorrgiHuf7DH/7QZNm9e/cyduxYAD766CPmzp1LUVER\nN998czh0NsAdd9xBIpGgsLCQxYsXA8EgewcPHmT69OlMnz4daHgTnXRDY7c0RHdztmzZwpQpUygq\nKuK6664Lh9BYtmxZOJx23WB8L7zwAhMmTGDChAkUFxe3OPyHBDprPK5M6NiGxD4UdlsfrQ2dfffd\n7tOmdezj7rtbHpJ28+bNftlll4XTo0eP9n379vmJEye8pqbG3d2rqqr8oosuCoef7tu3r7s3HHb7\nkUce8dtuu83d3bdu3eo5OTm+ceNGd68fevvkyZM+bdo037p1q7vXD4Vdp7WhsVsaojvV4sWL/Qc/\n+IG7u48bN86ff/55d3d/4IEH/O7kFzJkyBA/duyYu9cPp3311Vf7hg0b3N39yJEjfuLEiSbr1tDZ\nDT3xhHtubsOh3HNzg/ndsQ6JBho6u/MUFxfz3nvvcfDgQbZu3cqgQYPIz8/H3fnWt75FUVERM2bM\n4K233uLdd99tdj0vvvgi8+bNA6CoqIiioqLwvZUrVzJx4kSKi4vZtm0b27dvb7Gm5obGhsyH6IZg\nML/q6mqmTZsGwK233sqLL74Y1lhSUsITTzxBz57BOQtTp07lm9/8JsuWLaO6ujqcL83LlmslsunY\nhrZY4nPW/YtN7iXpdDfccAOrVq3inXfeCXellJeXU1VVxaZNm+jVqxcFBQVph8xOle7G9nv27OHh\nhx9m48aNDBo0iPnz57e6Hm9hTKvGQ3S3tvuoOX/84x958cUXWb16Nd/97nfZtm0bCxcuZPbs2axZ\ns4YpU6awbt06Pv3pT7dr/d1JNlwrkS3HNrLpbKzy8o4dPLMriHRLwcxmmtmbZrbTzBamef8cM3sy\n+f5fzKwgynqiNHfuXFasWMGqVavCs4lqamq44IIL6NWrF+vXr2dfukF2Ulx22WWUJ38S/e1vf+P1\n118H4IMPPqBv374MHDiQd999l2effTZcprlhu5sbGrutBg4cyKBBg8KtjMcff5xp06Zx+vRpDhw4\nwPTp03nooYeorq7m6NGj7Nq1i3HjxnHfffeRSCTC24VK9suWYxvZssWSTcOqd+aWU2RbCmaWAywH\nrgQqgY1mttrdU/d73A4cdveLzWwu8L+Am6OqKUqFhYUcOXKEoUOHMmTIEABKSkr4whe+QCKRYMKE\nCa3+Yr7jjju47bbbKCoqYsKECeGw1uPHj6e4uJjCwkIuvPBCpk6dGi5TWlrKrFmzGDJkCOvXrw/n\nNzc0dku7iprz2GOP8bWvfY3a2louvPBCfvOb33Dq1CnmzZtHTU0N7s43vvENzj33XB544AHWr19P\nTk4OY8aMYdasWW3+PIlHtowDlS1bLNkyrHqnbzllcuChPQ/gM8DalOn7gfsbtVkLfCb5uifwPsnh\nvJt7nOk9miV++u+VvZ54IrhfuVnwHMdB5ijuo94eZunrMOvcOjrq+yALDjQPBQ6kTFcm56Vt4+4n\ngRpgcIQ1iUgLSkpg797gXhJ798az/zyKuyO2R7bsTuvsLacoQ6HpEVNofPQzkzaYWamZVZhZRVVV\nVYcUJyLZKVvOxuqu4RRlKFQCw1OmhwEHm2tjZj2BgcB/NV6Ru5e5e8LdE3l5eRGVKyLZIhu2WLpr\nOEUZChuBUWY20sx6A3OB1Y3arAZuTb6+Afhzct9Xm7VzMelk+u8kXUl3DKfIzj5y95NmdhfBweQc\n4Nfuvs3MlhAc8FgN/Ap43Mx2EmwhzG3PZ/Xp04dDhw4xePDgtOf5S3Zwdw4dOkSfPn3iLkWkS+nM\n61isq/1ySyQS3njk0BMnTlBZWdnqBV0Svz59+jBs2DB69eoVdyki3YqZbXL3RGvtzoormnv16sXI\nkSPjLkNEpMvT2EciIhJSKIiISEihICIioS53oNnMqoCWR5bLfucTDOkhAX0f9fRdNKTvo6Ez+T5G\nuHurF3p1uVA4G5hZRSZnAXQX+j7q6btoSN9HQ53xfWj3kYiIhBQKIiISUijEoyzuArKMvo96+i4a\n0vfRUOTfh44piIhISFsKIiISUih0IjMbbmbrzWyHmW0zs7vjriluZpZjZq+Z2X/EXUvczOxcM1tl\nZm8k/x/5TNw1xcnMvpH8d/I3M/utmXWbkRTN7Ndm9p6Z/S1l3nlm9n/N7O/J50FRfLZCoXOdBP7F\n3UcDU4A7zWxMzDXF7W5gR9xFZIkfA//p7p8GxtONvxczGwp8HUi4+1iCkZbbNYpyF/W/gZmN5i0E\n/uTuo4A/Jac7nEKhE7n72+6+Ofn6CME/+sa3KO02zGwYMBv4Zdy1xM3MBgCXEQwnj7t/7O7V8VYV\nu57APyRvwJVL05t0nbXc/UWa3nBsDvBY8vVjwLVRfLZCISZmVgAUA3+Jt5JYPQr8T+B03IVkgQuB\nKuA3yd1pvzSzvnEXFRd3fwt4GNgPvA3UuPtz8VYVu0+4+9sQ/MAELojiQxQKMTCzfsDvgX929w/i\nricOZnY18J67b4q7lizRE5gI/NTdi4EPiWj3QFeQ3F8+BxgJfBLoa2bz4q2qe1AodDIz60UQCOXu\n/lTc9cRoKnCNme0FVgD/3cyeiLekWFUCle5et+W4iiAkuqsZwB53r3L3E8BTwGdjrilu75rZEIDk\n83tRfIhCoRNZcK/QXwE73P2HcdcTJ3e/392HuXsBwQHEP7t7t/0l6O7vAAfM7FPJWVcA22MsKW77\ngSlmlpv8d3MF3fjAe1LqPe1vBf4QxYecFXde60KmAl8G/mpmW5LzvuXua2KsSbLHPwHlZtYb2A3c\nFnM9sXH3v5jZKmAzwVl7r9GNrm42s98ClwPnm1klsBj4PrDSzG4nCM0bI/lsXdEsIiJ1tPtIRERC\nCgUREQkpFEREJKRQEBGRkEJBRERCCgWRTmRml2tEWMlmCgUREQkpFETSMLN5ZvaqmW0xs58n7/tw\n1MweMbPNZvYnM8tLtp1gZq+Y2etm9nTdOPdmdrGZrTOzrcllLkquvl/KfRPKk1fsimQFhYJII2Y2\nGrgZmOruE4BTQAnQF9js7hOBFwiuMgX4N+A+dy8C/poyvxxY7u7jCcbteTs5vxj4Z2AMweioUyP/\no0QypGEuRJq6ApgEbEz+iP8HgsHHTgNPJts8ATxlZgOBc939heT8x4DfmVl/YKi7Pw3g7scAkut7\n1d0rk9NbgAJgQ/R/lkjrFAoiTRnwmLvf32Cm2QON2rU0RkxLu4SOp7w+hf4dShbR7iORpv4E3GBm\nF0B4b9wRBP9ebki2+RKwwd1Moun+AAAAl0lEQVRrgMNmdmly/peBF5L3yag0s2uT6zjHzHI79a8Q\naQf9QhFpxN23m9m3gefMrAdwAriT4MY3hWa2CaghOO4AwTDGP0t2+qmjm34Z+LmZLUmuI5JRLUU6\nkkZJFcmQmR11935x1yESJe0+EhGRkLYUREQkpC0FEREJKRRERCSkUBARkZBCQUREQgoFEREJKRRE\nRCT0/wHI+XS2G1l1UAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x239e2d69b00>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "acc = history1.history['acc']\n",
    "val_acc = history1.history['val_acc']\n",
    "loss = history1.history['loss']\n",
    "val_loss = history1.history['val_loss']\n",
    "epoch = range(1 ,len(acc) + 1)\n",
    "plt.plot(epoch ,acc ,'bo' ,label = 'training accuracy')\n",
    "plt.plot(epoch , val_acc ,'b-' ,label= 'validation accuracy')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('accuracy')\n",
    "plt.legend()\n",
    "plt.show()\n",
    "plt.plot(epoch ,loss ,'bo' ,label ='training loss')\n",
    "plt.plot(epoch ,val_loss ,'b-' ,label ='validation loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.ylabel('loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}