classifying_newswires.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "#Loading the Reuters dataset\n",
    "#The data is stored as an array of lists of integers\n",
    "#Every list is an example of a newswire, encoded in integers, i.e. a map integers to words\n",
    "from keras.datasets import reuters\n",
    "(train_data, train_labels), (test_data, test_labels) = reuters.load_data(num_words=10000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7511"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Decoding newswire back into text\n",
    "word_index = reuters.get_word_index()\n",
    "reverse_word_index = dict([(value, key) for (key, value) in word_index.items()])\n",
    "def decoded(x): #function for translating the newswires back into text\n",
    "    print(' '.join([reverse_word_index.get(i-3, '?') for i in train_data[x]]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Vectorize the data\n",
    "import numpy as np\n",
    "\n",
    "def vectorize_sequences(sequences, dimension=10000):\n",
    "    results = np.zeros((len(sequences), dimension))\n",
    "    for i, sequence in enumerate(sequences):\n",
    "        results[i,sequence] = 1\n",
    "    return results\n",
    "\n",
    "x_train = vectorize_sequences(train_data)\n",
    "x_test = vectorize_sequences(test_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "#vectorize the labels\n",
    "#there are two possibilities\n",
    "#we use one-hot encoding, i.e. we embed each label as an all-zero vector with 1 in the place of the label index\n",
    "def to_one_hot(labels, dimension = 46):\n",
    "    results = np.zeros((len(labels), dimension))\n",
    "    for i, label in enumerate(labels):\n",
    "        results[i, label] = 1\n",
    "    return results\n",
    "\n",
    "one_hot_train_labels = to_one_hot(train_labels)\n",
    "one_hot_test_labels = to_one_hot(test_labels)\n",
    "\n",
    "#we could do this also with a built-in way from Keras:\n",
    "#from keras.utils.np_utils import to_categorical\n",
    "\n",
    "#one_hot_train_labels1 = to_categorical(train_labels)\n",
    "#one_hot_test_labels2 = to_categorical(test_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Defining the model\n",
    "#We use 64 hidden layers, since 16 could be to limited to learn to seperate 46 different classes\n",
    "#i.e. information could get lost\n",
    "from tensorflow.keras import models\n",
    "from tensorflow.keras import layers\n",
    "\n",
    "model = models.Sequential()\n",
    "model.add(layers.Dense(64, activation='relu', input_shape=(10000,)))\n",
    "model.add(layers.Dense(64, activation='relu'))\n",
    "model.add(layers.Dense(46, activation='softmax'))\n",
    "#we end the network with a Dense layer of size 46, i.e. for each input sample we get a 46dim vector\n",
    "#each dimension will encode a different output class\n",
    "#softmax activation will lead to a output of a probability distribution\n",
    "#each dimension i of the output vector gives us the probability that the sample belongs to class i"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tensorflow.keras.utils import plot_model\n",
    "from IPython.display import SVG\n",
    "\n",
    "plot_model(model, to_file='model.png', show_shapes=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "#compiling the model\n",
    "model.compile(optimizer='rmsprop', loss='categorical_crossentropy', metrics=['acc'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "#setting a validation set\n",
    "x_val = x_train[:1000]\n",
    "partial_x_train = x_train[1000:]\n",
    "y_val = one_hot_train_labels[:1000]\n",
    "partial_y_train = one_hot_train_labels[1000:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 7982 samples, validate on 1000 samples\n",
      "Epoch 1/20\n",
      "7982/7982 [==============================] - 1s 181us/sample - loss: 2.6964 - acc: 0.5202 - val_loss: 1.7559 - val_acc: 0.6620\n",
      "Epoch 2/20\n",
      "7982/7982 [==============================] - 1s 132us/sample - loss: 1.4036 - acc: 0.7249 - val_loss: 1.2749 - val_acc: 0.7390\n",
      "Epoch 3/20\n",
      "7982/7982 [==============================] - 1s 133us/sample - loss: 1.0161 - acc: 0.7889 - val_loss: 1.1046 - val_acc: 0.7680\n",
      "Epoch 4/20\n",
      "7982/7982 [==============================] - 1s 133us/sample - loss: 0.7950 - acc: 0.8345 - val_loss: 1.0231 - val_acc: 0.7820\n",
      "Epoch 5/20\n",
      "7982/7982 [==============================] - 1s 135us/sample - loss: 0.6310 - acc: 0.8687 - val_loss: 0.9410 - val_acc: 0.8150\n",
      "Epoch 6/20\n",
      "7982/7982 [==============================] - 1s 133us/sample - loss: 0.5057 - acc: 0.8963 - val_loss: 0.9124 - val_acc: 0.8130\n",
      "Epoch 7/20\n",
      "7982/7982 [==============================] - 1s 134us/sample - loss: 0.4086 - acc: 0.9156 - val_loss: 0.9155 - val_acc: 0.8020\n",
      "Epoch 8/20\n",
      "7982/7982 [==============================] - 1s 134us/sample - loss: 0.3367 - acc: 0.9296 - val_loss: 0.9100 - val_acc: 0.8220\n",
      "Epoch 9/20\n",
      "7982/7982 [==============================] - 1s 134us/sample - loss: 0.2775 - acc: 0.9397 - val_loss: 0.9200 - val_acc: 0.8120\n",
      "Epoch 10/20\n",
      "7982/7982 [==============================] - 1s 135us/sample - loss: 0.2374 - acc: 0.9460 - val_loss: 0.9337 - val_acc: 0.8160\n",
      "Epoch 11/20\n",
      "7982/7982 [==============================] - 1s 135us/sample - loss: 0.2089 - acc: 0.9486 - val_loss: 0.8979 - val_acc: 0.8190\n",
      "Epoch 12/20\n",
      "7982/7982 [==============================] - 1s 135us/sample - loss: 0.1787 - acc: 0.9538 - val_loss: 0.9515 - val_acc: 0.8180\n",
      "Epoch 13/20\n",
      "7982/7982 [==============================] - 1s 135us/sample - loss: 0.1636 - acc: 0.9543 - val_loss: 0.9480 - val_acc: 0.8160\n",
      "Epoch 14/20\n",
      "7982/7982 [==============================] - 1s 135us/sample - loss: 0.1491 - acc: 0.9550 - val_loss: 0.9880 - val_acc: 0.8130\n",
      "Epoch 15/20\n",
      "7982/7982 [==============================] - 1s 151us/sample - loss: 0.1430 - acc: 0.9551 - val_loss: 1.0064 - val_acc: 0.8100\n",
      "Epoch 16/20\n",
      "7982/7982 [==============================] - 1s 153us/sample - loss: 0.1334 - acc: 0.9559 - val_loss: 1.0582 - val_acc: 0.8010\n",
      "Epoch 17/20\n",
      "7982/7982 [==============================] - 1s 147us/sample - loss: 0.1273 - acc: 0.9567 - val_loss: 1.0851 - val_acc: 0.8030\n",
      "Epoch 18/20\n",
      "7982/7982 [==============================] - 1s 149us/sample - loss: 0.1265 - acc: 0.9563 - val_loss: 1.0784 - val_acc: 0.8040\n",
      "Epoch 19/20\n",
      "7982/7982 [==============================] - 1s 151us/sample - loss: 0.1162 - acc: 0.9568 - val_loss: 1.0577 - val_acc: 0.8030\n",
      "Epoch 20/20\n",
      "7982/7982 [==============================] - 1s 156us/sample - loss: 0.1167 - acc: 0.9577 - val_loss: 1.0861 - val_acc: 0.8010\n"
     ]
    }
   ],
   "source": [
    "#training the model\n",
    "history = model.fit(partial_x_train, partial_y_train, epochs=20, batch_size=512, validation_data=(x_val, y_val))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAIABJREFUeJzt3XucXHV9//HXO5sQCQQCJIFwyS63HyIWuWzBiJZLxSJaoGoRDBQwIcJDbuqvP6y01FrS1l5AbtWGi4hEoIootYBQUYoWAiEG5CoBAsaEZEFya0JCks/vj+/ZyWQzszub2TNnduf9fDzOY87le8585uzs+cz3fM/5HkUEZmZmAMOKDsDMzJqHk4KZmZU4KZiZWYmTgpmZlTgpmJlZiZOCmZmVOCnYgJLUJmmlpIkDWbZIkvaRNODXbkv6oKT5ZdPPS/pALWW34L2ul/SlLV2/l+1eJummgd6uFWd40QFYsSStLJscBawB1mfTn4mImf3ZXkSsB7Yd6LKtICL2G4jtSJoKnBYRR5Vte+pAbNuGPieFFhcRpYNy9kt0akT8V7XykoZHxLpGxGZmjefTR9ar7PTA7ZJulbQCOE3SJEmPSFoqaZGkqySNyMoPlxSSOrLpW7Ll90haIelhSXv2t2y2/MOSfi1pmaSrJf1C0plV4q4lxs9ImifpTUlXla3bJukKSW9IehE4rpf985eSbusx71pJl2fjUyU9m32eF7Nf8dW2tUDSUdn4KEnfzmJ7Gji0wvu+lG33aUknZPN/D7gG+EB2au71sn375bL1z8k++xuSfiBpQi37pi+STsriWSrpAUn7lS37kqSFkpZLeq7ss75X0pxs/mJJ/1Tr+1kOIsKDByICYD7wwR7zLgPWAn9M+hGxNfD7wOGkmuZewK+B87Lyw4EAOrLpW4DXgU5gBHA7cMsWlB0PrABOzJZ9HngbOLPKZ6klxh8C2wMdwO+6PztwHvA0sDuwE/Df6V+l4vvsBawEtinb9hKgM5v+46yMgGOA1cCB2bIPAvPLtrUAOCob/2fgZ8AOQDvwTI+yJwMTsr/Jp7IYds6WTQV+1iPOW4AvZ+MfymI8CHgH8K/AA7Xsmwqf/zLgpmx8/yyOY7K/0Zey/T4COAB4BdglK7snsFc2/hhwajY+Gji86P+FVh5cU7Ba/Dwi/iMiNkTE6oh4LCJmRcS6iHgJmAEc2cv634uI2RHxNjCTdDDqb9mPAnMj4ofZsitICaSiGmP8+4hYFhHzSQfg7vc6GbgiIhZExBvAP/TyPi8BT5GSFcCxwNKImJ0t/4+IeCmSB4CfABUbk3s4GbgsIt6MiFdIv/7L3/ffI2JR9jf5Dimhd9awXYDJwPURMTci3gK+CBwpafeyMtX2TW9OAe6KiAeyv9E/ANuRkvM6UgI6IDsF+XK27yAl930l7RQRKyJiVo2fw3LgpGC1+E35hKR3SvpPSa9JWg58BRjby/qvlY2vovfG5Wpldy2PIyKC9Mu6ohpjrOm9SL9we/Md4NRs/FOkZNYdx0clzZL0O0lLSb/Se9tX3Sb0FoOkMyU9kZ2mWQq8s8btQvp8pe1FxHLgTWC3sjL9+ZtV2+4G0t9ot4h4HvgC6e+wJDsduUtW9CzgXcDzkh6VdHyNn8Ny4KRgteh5Oea/kX4d7xMR2wGXkk6P5GkR6XQOAJLEpgexnuqJcRGwR9l0X5fM3g58MPulfSIpSSBpa+B7wN+TTu2MAe6rMY7XqsUgaS/g68C5wE7Zdp8r225fl88uJJ2S6t7eaNJpqt/WEFd/tjuM9Df7LUBE3BIRR5BOHbWR9gsR8XxEnEI6RfgvwB2S3lFnLLaFnBRsS4wGlgH/K2l/4DMNeM8fAYdI+mNJw4ELgXE5xfjvwEWSdpO0E3Bxb4UjYjHwc+CbwPMR8UK2aCSwFdAFrJf0UeAP+xHDlySNUbqP47yyZduSDvxdpPw4lVRT6LYY2L27Yb2CW4Epkg6UNJJ0cH4oIqrWvPoR8wmSjsre+89J7UCzJO0v6ejs/VZnw3rSBzhd0tisZrEs+2wb6ozFtpCTgm2JLwBnkP7h/430SzlX2YH3k8DlwBvA3sAvSfdVDHSMXyed+/8VqRH0ezWs8x1Sw/F3ymJeCnwOuJPUWPsJUnKrxV+TaizzgXuAm8u2+yRwFfBoVuadQPl5+PuBF4DFkspPA3Wvfy/pNM6d2foTSe0MdYmIp0n7/OukhHUccELWvjAS+EdSO9BrpJrJX2arHg88q3R12z8Dn4yItfXGY1tG6dSs2eAiqY10uuITEfFQ0fGYDRWuKdigIek4SdtnpyD+inRFy6MFh2U2pDgp2GDyfuAl0imI44CTIqLa6SMz2wI+fWRmZiWuKZiZWcmg6xBv7Nix0dHRUXQYZmaDyuOPP/56RPR2GTcwCJNCR0cHs2fPLjoMM7NBRVJfd+YDPn1kZmZlnBTMzKzEScHMzEqcFMzMrMRJwczMSloiKcycCR0dMGxYep3Zr0fRm5m1jkF3SWp/zZwJ06bBqlVp+pVX0jTA5Lr7hTQzG1qGfE3hkks2JoRuq1al+WZmtqkhnxRefbV/883MWtmQTwoTqzxIsdp8M7NWNuSTwvTpMGrUpvNGjUrzzcxsU0M+KUyeDDNmQHs7SOl1xgw3MpuZVTLkrz6ClACcBMzM+jbkawpmZlY7JwUzMytxUjAzsxInBTMzK3FSMDOzEicFMzMrcVIwM7OS3JKCpD0k/VTSs5KelnRhhTJHSVomaW42XJpXPGZm1rc8b15bB3whIuZIGg08Lun+iHimR7mHIuKjOcZhZmY1yq2mEBGLImJONr4CeBbYLa/3MzOz+jWkTUFSB3AwMKvC4kmSnpB0j6QDqqw/TdJsSbO7urpyjNTMrLXlnhQkbQvcAVwUEct7LJ4DtEfEe4CrgR9U2kZEzIiIzojoHDduXL4Bm5m1sFyTgqQRpIQwMyK+33N5RCyPiJXZ+N3ACElj84zJzMyqy/PqIwE3AM9GxOVVyuySlUPSYVk8b+QVk5mZ9S7Pq4+OAE4HfiVpbjbvS8BEgIj4BvAJ4FxJ64DVwCkRETnGZGZmvcgtKUTEzwH1UeYa4Jq8YjAzs/7xHc1mZlbipGBmZiVOCmZmVuKkYGZmJU4KZmZW4qRgZmYlTgpmZlbipGBmZiVOCmZmVuKkYGZmJU4KZmZW4qRgZmYlTgpmZlbipGBmZiVOCmZmVuKkYGZmJU4KZmZW4qRgZmYlTgpmZlbipGBmZiVOCmZmVuKkYGZmJU4KZmZW4qRgZmYlTgpmZlbipGBmZiVOCmZmVuKkYGZmJU4KZmZWkltSkLSHpJ9KelbS05IurFBGkq6SNE/Sk5IOySseMzPr2/Act70O+EJEzJE0Gnhc0v0R8UxZmQ8D+2bD4cDXs1czMytAbjWFiFgUEXOy8RXAs8BuPYqdCNwcySPAGEkT8orJzMx615A2BUkdwMHArB6LdgN+Uza9gM0Th5mZNUjuSUHStsAdwEURsbzn4gqrRIVtTJM0W9Lsrq6uPMI0MzNyTgqSRpASwsyI+H6FIguAPcqmdwcW9iwUETMiojMiOseNG5dPsGZmluvVRwJuAJ6NiMurFLsL+LPsKqT3AssiYlFeMZmZWe/yvProCOB04FeS5mbzvgRMBIiIbwB3A8cD84BVwFk5xmNmZn3ILSlExM+p3GZQXiaAz+YVg5mZ9Y/vaDYzsxInBTMzK3FSMDOzEicFMzMrcVIwM7MSJwUzMytxUjAzsxInBTMzK3FSMDOzEicFMzMrcVIwM7MSJwUzMytpmaTwyitw6aWwbl3RkZiZNa+WSQpz5sDf/i388IdFR2Jm1rxaJimccAJ0dMCVVxYdiZlZ82qZpNDWBuedBw89lGoNZma2uZZJCgBTpsA227i2YGZWTUslhTFj4Mwz4bbbYPHioqMxM2s+LZUUAM4/H9auhW98o+hIzMyaT8slhf32g+OPh3/9V1izpuhozMyaS8slBYALL4QlS+D224uOxMysubRkUjj2WNh//9TgHFF0NGZmzaMlk4KUagtz5sAvflF0NGZmzaMlkwLA6afDDjvA175WdCRmZs2jZZPCqFFw9tlw552pXyQzM2vhpADw2c+mU0nXXlt0JGZmzaGlk8LEifCxj8F118H//m/R0ZiZFa+lkwKkBuelS+Hmm4uOxMyseC2fFN73Pjj0ULjqKtiwoehozMyK1fJJQYKLLoLnnoP77y86GjOzYuWWFCTdKGmJpKeqLD9K0jJJc7Ph0rxi6cvJJ8Muu/jyVDOzPGsKNwHH9VHmoYg4KBu+kmMsvdpqKzj3XLj33lRjMDNrVTUlBUl7SxqZjR8l6QJJY3pbJyL+G/jdAMTYEOeck5LD1VcXHYmZWXFqrSncAayXtA9wA7An8J0BeP9Jkp6QdI+kA6oVkjRN0mxJs7u6ugbgbTc3fjx86lNw003w5pu5vIWZWdOrNSlsiIh1wJ8AX4uIzwET6nzvOUB7RLwHuBr4QbWCETEjIjojonPcuHF1vm11F14Iq1bBDTfk9hZmZk2t1qTwtqRTgTOAH2XzRtTzxhGxPCJWZuN3AyMkja1nm/U66CA48ki45hpYt67ISMzMilFrUjgLmARMj4iXJe0J3FLPG0vaRZKy8cOyWN6oZ5sD4cILU19Id91VdCRmZo1XU1KIiGci4oKIuFXSDsDoiPiH3taRdCvwMLCfpAWSpkg6R9I5WZFPAE9JegK4CjglovinG5xwAnR0bHp56syZad6wYel15syCgjMzy9nwWgpJ+hlwQlZ+LtAl6cGI+Hy1dSLi1N62GRHXANfUHmpjtLWl5zh/4Qvwy1/CM8/AtGmprQFSLWLatDQ+eXJxcZqZ5aHW00fbR8Ry4GPANyPiUOCD+YVVrE9/GrbZJj2Z7ZJLNiaEbqtWpflmZkNNrUlhuKQJwMlsbGgessaMgTPPhFtvrf6shVdfbWhIZmYNUWtS+ArwY+DFiHhM0l7AC/mFVbzzz4e1a2H77SsvnzixsfGYmTVCrQ3N342IAyPi3Gz6pYj4eL6hFWu//eD449P41ltvumzUKJg+vfExmZnlrdZuLnaXdGfWwd1iSXdI2j3v4Ip24YWwbBmccQa0t6ceVdvbYcYMNzKb2dBU09VHwDdJ3Vr8aTZ9Wjbv2DyCahbHHgv77w+PPgovv5ySgpnZUFZrm8K4iPhmRKzLhpuA/PqbaBJSqi3MmQO/+EXR0ZiZ5a/WpPC6pNMktWXDaTTB3ceNcPrpsMMO6fJUM7Ohrtak8GnS5aivAYtIdyOflVdQzWTUKDj7bPj+96tfnmpmNlTUevXRqxFxQkSMi4jxEXES6Ua2lvDZz6ZTSddeW3QkZmb5qufJa1W7uBhqJk6Ej30MrrsOcnqcg5lZU6gnKbTUtTgXXwxvvQW///swd27R0ZiZ5aOepFB4j6aNdOih8NBD6TkLRxwB3/1u0RGZmQ28XpOCpBWSllcYVgC7NijGptHZCbNnw3veAyefDH/1V7BhQ9FRmZkNnF6TQkSMjojtKgyjI6LWG9+GlF12gZ/+FKZMgcsug5NOguXLi47KzGxg1HP6qGWNHJkana++Gu6+GyZNgnnzio7KzKx+TgpbSILzzoP77oPFi1MD9H33FR2VmVl9nBTqdMwx8NhjsMce8OEPwxVXQPEPFTUz2zJOCgNgzz3hf/4ntS98/vPpAT1vvVV0VGZm/eekMEC23TZdpvo3fwM33wxHHgm//W3RUZmZ9Y+TwgAaNgwuvRTuvBOeeSa1MzzySNFRmZnVzkkhByedBA8/DO94R6ox3HRT0RGZmdXGSSEn7353aoD+wAfgrLPgoovS3dBmZs3MSSFHO+0E996bHtRz5ZXwR3+UnuLmq5PMrFk5KeRs+HD42tfgxhth1iw4/PDUTcaVV8IbLfGYIrPmsGpVGtavLzqS/tuwAV59FV57Lf/3Ugyyn62dnZ0xe/bsosPYIsuXw223wfXXp1NLW22VuuSeOhWOPjo1VJvZllmzJh04X3658vD66xvLjhiR2vze8Q7YeuuN431Njx0LEybArrum1wkT0ryB+N9dvTrF+eKLG4eXXkqvL78Ma9fCX/wF/N3fbdn2JT0eEZ19lnNSKMYTT8ANN8C3vw1Ll6Z7HaZMSfc47LZb0dGZNZ/169Nl3tUO+gsXbnpqdsQIaG9P/1t77pnGhw9P9xB1D6tXbzpdaV739OrVsGLF5nENH576ROtOEj2TRvf0+PHw5pubH/C7h4ULN93u6NGw996bDpMmwe/93pbtPyeFQWL16nQJ6/XXp472hg1Ld0ZPnQof+Uj6Ypu1ojVrUhvcz34GDz6YbhBdvXrjcin9gOo+6Pccdt0V2toGNqa33oJFizYdFi7cfLq8VtKbXXdNB/u99to8Aey0U/qMA8VJoYnMnAmXXJKqthMnwvTpMHny5uVefDG1PXzzm+nLtfPOqeYwZQrsu2/DwzZrqLfeSvf1PPhgGh5+eGPPAAcemC7vfve7Nx7099gjdU7ZjNauTX2ilSeN116DMWM2HvT33DM9A75RCk8Kkm4EPgosiYh3V1gu4ErgeGAVcGZEzOlru4MtKcycCdOmpQaubqNGwYwZlRMDpEtX77kn1R7+8z9TtfnII1Pt4eMfT+c4zQa71avTgf/BB1NtYNasVDuQ4KCD0nf+qKPSZd077lh0tINfMySFPwBWAjdXSQrHA+eTksLhwJURcXhf2x1sSaGjA155ZfP57e0wf37f6y9cCN/6Vmp/ePFF2H57OO00OPvsdBWT2WCxcmWqCXSfDpo1C95+O50yPfjglACOPBLe/37YYYeiox16Ck8KWRAdwI+qJIV/A34WEbdm088DR0XEot62OdiSwrBhle9LkPr31LYNG9I/0vXXwx13pF9UnZ0pOZxyCmy33cDFbFaLiHRF3eLFsGRJei0f7/na/TCqtrb0eNvumsARR6QfO5avWpNCkU9P2w34Tdn0gmzeZklB0jRgGsDEiRMbEtxAmTixck2hvx9j2LB02erRR6eH+9xyS3rQz2c+A5/7XEoMU6fCe987sI1T1tpWrkxXyv3yl/Dkk+nqn+4D/ZIl6cdJT1JqJB0/PrWLHXJIeu0eP+KIdGWNNacik0KlQ1fFaktEzABmQKop5BnUQJs+vXKbwvTpW77NHXeECy6A889PV2dcfz3cemtqpD7ggJQcTj89/WOa1aqrKx38y4cXXthY091pp3Tac/z41ODbfdDfeeeN4+PHw7hx6TJNG5yK/NMtAPYom94dWFil7KDV3Zhcy9VH/SWlO6QPPxwuvxxuvz3VHj73Obj44nRj3Nlnpyq6b4yzbhGp9tozAZR39d7ens7zT56cXg8+OF3+6Vro0Fdkm8JHgPPY2NB8VUQc1tc2B1ubQhGefDLVHrpvjNt773RZ6xlnpOuibfB4883UlvTAA+k6/bVr06/wESPSa3/GI+C552Du3LRdSD8W9t9/44H/oIPS4Kt9hp7CG5ol3QocBYwFFgN/DYwAiIhvZJekXgMcR7ok9ayI6PNo76RQu+4b4667Ll3xAenuyp43yeR1s8xgtG4dLFuWDprlw9Kl6fx6R0c6dbL33vmcIlmxAh56KN3I+MAD6Rd8RLoMedKk1CC7bl26amfduv6Nb9gA++yzMQEcfHC6O9aXOLeGwpNCXpwUtswLL6Srln7964231fd8Mtx226WD3T77bJ4wdt+9f6egItJBqPugtG5dmh4xIvX5NGLEwCagiHSj08qV6cC6cmXl8aVLNz/Yl09X6sagkpEj4Z3vTAmifJg4sX/7afXqVAN44IGUCB59NN2XstVWKQkcc0waDjsszTPbUk4K1qdKHXDNm5de589Pvy67bbVVuoO0rW3TA/369ZtOl8/vy8iRabsjR9Y2PmLExv5nug/05Qf+Wi/x3WabdB38mDHptXvoa3rUqNRfzVNPbTr8puwaum23TY393Umie3yXXVISXLs2Hfi7k0D3KaG2tvSkvu4kMGlSY+92taHPScHqsn59OtiVJ4xXX03Lus9Tt7VtHK9letiwlGjWrEkHwjVrqo9Xmvf22+lAue22aRg9uvJ4b9Pbbz/wv7iXLYOnn948WXR1bSyz446pf5tnnklXonXftdudBN7/ft9rYvlyUjAr2JIlmyaLF15INYejj043bvmSYWukwXDzmtmQNn58Go4+uuhIzGrnq9fNzKzEScHMzEqcFMzMrMRJYRCYOTPdNDVsWHqdObPoiMxsqHJDc5Pr+ZCeV15J0zAw/SeZmZVzTaHJXXLJpj2sQpq+5JJi4jGzoc1Jocl13zBW63wzs3o4KTS5ag/jGWTPGjKzQcJJoclNn755Hzj1PqTHzKwaJ4UmN3kyzJiRHnoipdcZM9zIbGb58NVHg8DkyU4CZtYYrimYmVmJk4KZmZU4KZiZWYmTgpmZlTgptAD3nWRmtfLVR0Oc+04ys/5wTWGIc99JZtYfTgpDnPtOMrP+cFIY4tx3kpn1h5PCEOe+k8ysP5wUhjj3nWRm/eGrj1qA+04ys1q5pmBmZiVOCmZmVpJrUpB0nKTnJc2T9MUKy8+U1CVpbjZMzTMe23K+K9qsNeTWpiCpDbgWOBZYADwm6a6IeKZH0dsj4ry84rD6+a5os9aRZ03hMGBeRLwUEWuB24ATc3w/y4nvijZrHXkmhd2A35RNL8jm9fRxSU9K+p6kPXKMx7aQ74o2ax15JgVVmBc9pv8D6IiIA4H/Ar5VcUPSNEmzJc3u6uoa4DCtL74r2qx15JkUFgDlv/x3BxaWF4iINyJiTTZ5HXBopQ1FxIyI6IyIznHjxuUSrFXnu6LNWkeeSeExYF9Je0raCjgFuKu8gKQJZZMnAM/mGI9tId8VbdY6crv6KCLWSToP+DHQBtwYEU9L+gowOyLuAi6QdAKwDvgdcGZe8Vh9fFe0WWvI9T6FiLg7Iv5PROwdEdOzeZdmCYGI+IuIOCAi3hMRR0fEc3nGY8XxfQ5mg4P7PrLc+T4Hs8HD3VxY7nyfg9ng4aRgufN9DmaDh5OC5c73OZgNHk4KlruBuM/BDdVmjeGkYLmr9z6H7obqV16BiI0N1U4MZgNPET17nmhunZ2dMXv27KLDsAbq6EiJoKf2dpg/v9HRmA1Okh6PiM6+yrmmYE3PDdVmjeOkYE3PDdVmjeOkYE3PDdVmjeOkYE3PDdVmjeOGZhvy3FBt5oZms5KBaKj26SdrFU4KNuTV21Dt00/WSpwUbMirt6F6IDr0c03DBgsnBRvy6m2orvf0k2saNpg4KVhLmDw5NSpv2JBe+/Mch3pPP7mmYYOJk4JZH+o9/dQMNQ0nFauVk4JZH+o9/VR0TcNJxfolIgbVcOihh4bZYHLLLRGjRkWkQ3IaRo1K82shbbpu9yDVtn57e+X129sbE3/3NtrbU8zt7f1b1wYGMDtqOMa6pmCWs6JrGvWevhoKNZWi1x9UaskczTS4pmCtpt5f6vXWFAZ7TaXo9bu3UU9NaSBqWtRYUyj8IN/fwUnBWlE9B4VWTypFr98MSSnCScHMyrRyUil6/aKTUrdak4LbFMxaQD33adTbJlLvJb31tqkUvX69bTqNfsiUk4KZ9WkwJ5Wi1y86KfVbLdWJZhp8+sis9RTdUFvk6bdGtyn4eQpmZjmbOTNdwvvqq+kX/vTp/att1bs+1P48BScFM7MW4IfsmJlZv+WaFCQdJ+l5SfMkfbHC8pGSbs+Wz5LUkWc8ZmbWu9ySgqQ24Frgw8C7gFMlvatHsSnAmxGxD3AF8NW84jEzs77lWVM4DJgXES9FxFrgNuDEHmVOBL6VjX8P+ENJyjEmMzPrRZ5JYTfgN2XTC7J5FctExDpgGbBTzw1JmiZptqTZXV1dOYVrZmbDc9x2pV/8PS91qqUMETEDmAEgqUvSK/WHl4uxwOtFB9GLZo8Pmj9Gx1cfx1efeuJrr6VQnklhAbBH2fTuwMIqZRZIGg5sD/yut41GxLiBDHIgSZpdyyVfRWn2+KD5Y3R89XF89WlEfHmePnoM2FfSnpK2Ak4B7upR5i7gjGz8E8ADMdhunDAzG0JyqylExDpJ5wE/BtqAGyPiaUlfId1ufRdwA/BtSfNINYRT8orHzMz6lufpIyLibuDuHvMuLRt/C/jTPGNosBlFB9CHZo8Pmj9Gx1cfx1ef3OMbdN1cmJlZftzNhZmZlTgpmJlZiZNCP0naQ9JPJT0r6WlJF1Yoc5SkZZLmZsOllbaVY4zzJf0qe+/NupRVclXW59STkg5pYGz7le2XuZKWS7qoR5mG7z9JN0paIumpsnk7Srpf0gvZ6w5V1j0jK/OCpDMqlckpvn+S9Fz2N7xT0pgq6/b6fcgxvi9L+m3Z3/H4Kuv22kdajvHdXhbbfElzq6yb6/6rdkwp7PtXy0MXPGwcgAnAIdn4aODXwLt6lDkK+FGBMc4Hxvay/HjgHtLNg+8FZhUUZxvwGtBe9P4D/gA4BHiqbN4/Al/Mxr8IfLXCejsCL2WvO2TjOzQovg8Bw7Pxr1aKr5bvQ47xfRn4vzV8B14E9gK2Ap7o+f+UV3w9lv8LcGkR+6/aMaWo759rCv0UEYsiYk42vgJ4ls2772h2JwI3R/IIMEbShALi+EPgxYgo/A71iPhvNr9xsrxvrm8BJ1VY9Y+A+yPidxHxJnA/cFwj4ouI+yJ1DwPwCOkG0UJU2X+1qKWPtLr1Fl/W39rJwK0D/b616OWYUsj3z0mhDllX3wcDsyosniTpCUn3SDqgoYGlrkLuk/S4pGkVltfSL1UjnEL1f8Qi91+3nSNiEaR/XGB8hTLNsi8/Tar9VdLX9yFP52Wnt26scvqjGfbfB4DFEfFCleUN2389jimFfP+cFLaQpG2BO4CLImJ5j8VzSKdE3gNcDfygweEdERGHkLot/6ykP+ixvKY+p/KU3eV+AvDdCouL3n/90Qz78hJgHTCzSpG+vg95+TqwN3AQsIh0iqanwvcfcCq91xIasv/6OKZUXa3CvLr2n5PCFpA0gvTHmxkR3++5PCKWR8TKbPxuYISksY2KLyIWZq9LgDtJVfRytfRLlbcPA3MiYnHPBUXvvzKLu0+rZa9LKpQpdF9mDYsfBSZHdpK5pxob/+1rAAADL0lEQVS+D7mIiMURsT4iNgDXVXnfovffcOBjwO3VyjRi/1U5phTy/XNS6Kfs/OMNwLMRcXmVMrtk5ZB0GGk/v9Gg+LaRNLp7nNQY+VSPYncBf5ZdhfReYFl3NbWBqv46K3L/9VDeN9cZwA8rlPkx8CFJO2SnRz6UzcudpOOAi4ETImJVlTK1fB/yiq+8nepPqrxvLX2k5emDwHMRsaDSwkbsv16OKcV8//JqUR+qA/B+UvXsSWBuNhwPnAOck5U5D3iadCXFI8D7GhjfXtn7PpHFcEk2vzw+kZ6K9yLwK6CzwftwFOkgv33ZvEL3HylBLQLeJv36mkJ6tsdPgBey1x2zsp3A9WXrfhqYlw1nNTC+eaTzyd3fw29kZXcF7u7t+9Cg+L6dfb+eJB3gJvSML5s+nnTFzYuNjC+bf1P3966sbEP3Xy/HlEK+f+7mwszMSnz6yMzMSpwUzMysxEnBzMxKnBTMzKzEScHMzEqcFMwyktZr0x5cB6zHTkkd5T10mjWrXB/HaTbIrI6Ig4oOwqxIrimY9SHrT/+rkh7Nhn2y+e2SfpJ1+PYTSROz+TsrPd/giWx4X7apNknXZX3m3ydp66z8BZKeybZzW0Ef0wxwUjArt3WP00efLFu2PCIOA64BvpbNu4bUBfmBpM7orsrmXwU8GKlDv0NId8IC7AtcGxEHAEuBj2fzvwgcnG3nnLw+nFktfEezWUbSyojYtsL8+cAxEfFS1nHZaxGxk6TXSV03vJ3NXxQRYyV1AbtHxJqybXSQ+r3fN5u+GBgREZdJuhdYSeoN9geRdQZoVgTXFMxqE1XGq5WpZE3Z+Ho2tul9hNQX1aHA41nPnWaFcFIwq80ny14fzsb/h9SrJ8Bk4OfZ+E+AcwEktUnartpGJQ0D9oiInwL/DxgDbFZbMWsU/yIx22hrbfrw9nsjovuy1JGSZpF+SJ2azbsAuFHSnwNdwFnZ/AuBGZKmkGoE55J66KykDbhF0vak3muviIilA/aJzPrJbQpmfcjaFDoj4vWiYzHLm08fmZlZiWsKZmZW4pqCmZmVOCmYmVmJk4KZmZU4KZiZWYmTgpmZlfx/bKlCg5Ulph8AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plotting the training and validation loss\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "epochs = range(1,len(loss)+1)\n",
    "\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAIABJREFUeJzt3XmYFNXVx/HvYV8FBBUDCmiIGwHEETXivgSMglGjEpKoqKgJqElMQtRXiVvco0ZfAxqjkYnKG0UlUVwI0Rg3IAoIREAEHEEEZN8HzvvHrWmapnumh5nq7pn5fZ6nn+6qulV9uqanTt97q26ZuyMiIgJQL98BiIhI4VBSEBGRBCUFERFJUFIQEZEEJQUREUlQUhARkQQlBdmJmdU3s7Vmtm91ls0nM/u6mVX7+ddmdrKZzU+a/tjMjsmm7C6816Nmdu2uri+SjQb5DkCqzszWJk02AzYBW6Ppy9y9uDLbc/etQIvqLlsXuPsB1bEdM7sE+IG7H5+07UuqY9si5VFSqAXcPXFQjn6JXuLur2cqb2YN3L00F7GJVETfx8Ki5qM6wMxuMbNnzOwpM1sD/MDMjjKzd81spZktNrMHzKxhVL6BmbmZdY6mR0fLXzazNWb2jpl1qWzZaHk/M5ttZqvM7Pdm9m8zuzBD3NnEeJmZzTWzFWb2QNK69c3sd2a23Mw+AfqWs3+uN7OnU+Y9ZGb3Rq8vMbNZ0ef5JPoVn2lbJWZ2fPS6mZk9GcU2AzgszfvOi7Y7w8z6R/O/CTwIHBM1zS1L2rcjkta/PPrsy83seTPbO5t9U5n9XBaPmb1uZl+Z2Rdm9suk9/mfaJ+sNrPJZva1dE11ZvZW2d852p9vRu/zFXC9mXU1s4nRZ1kW7bdWSet3ij7j0mj5/WbWJIr5oKRye5vZejNrm+nzSgXcXY9a9ADmAyenzLsF2AycQfgh0BQ4HDiCUFvcD5gNDI3KNwAc6BxNjwaWAUVAQ+AZYPQulN0TWAMMiJb9DNgCXJjhs2QT4wtAK6Az8FXZZweGAjOAjkBb4M3wdU/7PvsBa4HmSdv+EiiKps+IyhhwIrAB6B4tOxmYn7StEuD46PXdwD+BNkAnYGZK2XOBvaO/yfejGPaKll0C/DMlztHAiOj1qVGMPYEmwP8C/8hm31RyP7cClgBXAY2B3YDe0bJfA1OBrtFn6AnsDnw9dV8Db5X9naPPVgpcAdQnfB+/AZwENIq+J/8G7k76PB9F+7N5VP7oaNko4Nak9/k5MDbf/4c1+ZH3APSo5j9o5qTwjwrWuwb4v+h1ugP9H5LK9gc+2oWyg4F/JS0zYDEZkkKWMR6ZtPw54Jro9ZuEZrSyZaelHqhStv0u8P3odT9gdjll/wb8JHpdXlJYmPy3AH6cXDbNdj8CvhO9rigpPAHclrRsN0I/UseK9k0l9/MPgckZyn1SFm/K/GySwrwKYjgHmBS9Pgb4AqifptzRwKeARdMfAmdV9/9VXXqo+aju+Cx5wswONLO/R80Bq4GbgHblrP9F0uv1lN+5nKns15Lj8PBfXJJpI1nGmNV7AQvKiRfgL8DA6PX3gUTnvJmdbmbvRc0nKwm/0svbV2X2Li8GM7vQzKZGTSArgQOz3C6Ez5fYnruvBlYAHZLKZPU3q2A/7wPMzRDDPoTEsCtSv4/tzWyMmX0exfB4SgzzPZzUsAN3/zeh1tHHzLoB+wJ/38WYBPUp1CWpp2OOJPwy/bq77wbcQPjlHqfFhF+yAJiZseNBLFVVYlxMOJiUqeiU2WeAk82sI6F56y9RjE2BvwK/JTTttAZezTKOLzLFYGb7AQ8TmlDaRtv9b9J2Kzp9dhGhSapsey0JzVSfZxFXqvL282fA/hnWy7RsXRRTs6R57VPKpH6+OwhnzX0ziuHClBg6mVn9DHH8GfgBoVYzxt03ZSgnWVBSqLtaAquAdVFH3WU5eM+/Ab3M7Awza0Bop94jphjHAFebWYeo0/FX5RV29yWEJo4/AR+7+5xoUWNCO/dSYKuZnU5o+842hmvNrLWF6ziGJi1rQTgwLiXkx0sINYUyS4COyR2+KZ4CLjaz7mbWmJC0/uXuGWte5ShvP78I7GtmQ82skZntZma9o2WPAreY2f4W9DSz3QnJ8AvCCQ31zWwISQmsnBjWAavMbB9CE1aZd4DlwG0WOu+bmtnRScufJDQ3fZ+QIKQKlBTqrp8DFxA6fkcSfinHKjrwngfcS/gn3x/4gPALsbpjfBiYAEwHJhF+7VfkL4Q+gr8kxbwS+CkwltBZew4huWXjRkKNZT7wMkkHLHefBjwAvB+VORB4L2nd14A5wBIzS24GKlt/PKGZZ2y0/r7AoCzjSpVxP7v7KuAU4GxCx/Zs4Lho8V3A84T9vJrQ6dskaha8FLiWcNLB11M+Wzo3Ar0JyelF4NmkGEqB04GDCLWGhYS/Q9ny+YS/82Z3f7uSn11SlHXOiORc1BywCDjH3f+V73ik5jKzPxM6r0fkO5aaThevSU6ZWV9Cc8BGwimNpYRfyyK7JOqfGQB8M9+x1AZqPpJc6wPMIzQr9AXOVMeg7Coz+y3hWonb3H1hvuOpDdR8JCIiCaopiIhIQo3rU2jXrp137tw532GIiNQoU6ZMWebu5Z0CDtTApNC5c2cmT56c7zBERGoUM6voqn5AzUciIpJESUFERBKUFEREJEFJQUREEpQUREQkQUlBRKTAFRdD585Qr154Li6uaI1dp6QgIhWq6kFJ6+/6+sXFMGQILFgA7uF5yJAYE0O+b/1W2cdhhx3mInXN6NHunTq5m4Xn0aNzt/7o0e7NmrmHQ1J4NGuW/Ta0ftXW79Rpx3XLHp06Zbd+GTLcVjX1kfeDfGUfSgpS19T0g5LWr9r6ZunXN8tu/TLZJoUaNyBeUVGR64pmqUmKi+G662DhQth3X7j1VhhUidvhdO4cmgxSdeoE8+fHv369euEwlMoMtm3T+nGvX9W/3/b3synuXlRROfUpiMSoOtqDF2YYEDrT/Opef98Md7fONF/rV+/6t94KzZrtOK9ZszA/DkoKIhWoSifhddfB+vU7zlu/PszPVk0/KGn9qq0/aBCMGhVqBmbhedSoytU2KyWbNqZCeqhPQXKpqu3x1dEenO8+hbJt5KujW+tXD9TRLBJU5R8y352M1fEZqmN9qfmyTQrqaJZaraxNP7kJp1mz7KvfVe0krOr7i1QXdTSLUPU2/aq2x+e8PVikipQUpFar6pk31XHmx6BB4dTBbdvCsxKCFDIlBanV9EtfpHKUFKTgVeWUUP3SF6kcJQUpaFW9+Eu/9EUqR2cfSUGrrkv8Reo6nX0ktUJVO4pFpHKUFKSgVbWjWEQqR0lBClquBwMTqeuUFCR2VTl7SB3FIrnVIN8BSO2WOsxD2dlDkP2BfdAgJQGRXFFNQWJVHUNHi0juKClIrHT2kEjNoqQgsdLZQyI1i5KCxEpnD4nULEoKEiudPSRSs8SaFMysr5l9bGZzzWx4muWdzGyCmU0zs3+aWcc445H80IByIjVHbEnBzOoDDwH9gIOBgWZ2cEqxu4E/u3t34Cbgt3HFIyIiFYuzptAbmOvu89x9M/A0MCClzMHAhOj1xDTLpQBU5eIzEalZ4kwKHYDPkqZLonnJpgJnR6+/C7Q0s7apGzKzIWY22cwmL126NJZgJb2qDl0tIjVLnEnB0sxLHaf7GuA4M/sAOA74HCjdaSX3Ue5e5O5Fe+yxR/VHKhnp4jORuiXOYS5KgH2SpjsCi5ILuPsi4CwAM2sBnO3uq2KMSSpJF5+J1C1x1hQmAV3NrIuZNQLOB15MLmBm7cysLIZfA4/FGI/sAl18JlK3xJYU3L0UGAq8AswCxrj7DDO7ycz6R8WOBz42s9nAXoAuaSowuvhMpG7R7TilQsXFoQ9h4cJQQ7j1Vl1rIFLTZHs7Tg2dLRXS0NUidYeGuRARkQQlBRERSVBSEBGRBCUFERFJUFKoAzR2kYhkS2cf1XJlYxeVDVVRNnYR6IwiEdmZagq1nMYuEpHKUFKo5TR2kYhUhpJCLaexi0SkMpQUajmNXSQilaGkUMsNGgSjRkGnTmAWnkeNUieziKSns4/qAI1dJCLZUk1BREQSlBRERCRBSUFERBKUFEREJEFJQUREEpQUREQkQUlBREQSlBRERCRBSaEG0P0QRCRXdEVzgdP9EEQkl1RTKHC6H4KI5JKSQoHT/RBEJJeUFAqc7ocgIrmkpFDgdD8EEcklJYUCp/shiEgu6eyjGkD3Q4Bt2+CTT2C//aB+/XxHI1J7qaYgBe/NN+Hww+Eb34A99oDzz4fHH4fFi/MdmUjto5qCFKy5c+FXv4LnnoOOHeGuu2DmTBg/Hp55JpTp0QP69oV+/eCoo6BRo9zHuXkzLFsGX34JS5eG57JH8vSKFdCgATRpAo0bb3+kTqeb16RJeLRsCa1awW67hUfy6wb6b5ZqoK9RHbJ5MzRsGPomCtnKlXDzzfD734eD/M03w89+tr3D3R2mTQvJYfx4uOceuOOOcMA86aSQJPr2Df0vu8odVq2Czz6DkpLtz+kO/CtXpt9GgwahZrPnnuHRqRNs3QqbNoXHxo0hUZRNl81Lni4tzT7mZs12TBSpyaNVK2jXbns8ZbG1basmOdnO3D3fMVRKUVGRT548Od9h1CjucMMNcNttYTrdgSPdr8/UeXvuCfvsE1+cW7bAyJEwYgR89RUMHhwSwt57l7/e6tXwj3+EBPHyy9uv4TjwwFCD6NsXjj02/NIus3ZtONAnP8oO/mWPtWt3fB+z9AfVTK9bt656Ak5NIqtXb3+sWlX+c+q8NWvCdyGVWUgMqfFn+nytW4chV6RmMbMp7l5UYTklhdpt2zYYOhQefhjOOQcOOKDiA8nGjZm3d+CBcOaZMGAA9O5dPQcHd3jpJbjmGvjvf+GEE+Dee6Fnz13b1scfb08Qb7wRDqhNm8Jhh4Vf9SUlO/+6N4O99gpJL/nRseP213vvXbObaLZuDck2tbaT6fWKFem3U79+xYkj+XXLloVfO60LlBSELVvgggvgqafgl7+E22/P7p9z8+b0vzg//RTGjQsH2tJSaN8e+vcPCeLEE3f8JZ6t6dND09Drr0PXrnD33XDGGdV3EFm/PsQ7fjxMmhQOVukO+h065Kc/opBt2bJjX8mSJeE5U7/JmjXpt9OoUUgO7dvDEUeEpH/ccaHWJbmjpFDHrV8P3/te+AV+++2hw7a6rFgRtvvCC+HX+Nq10KJFaKY580w47TRo06b8bSxZEpq0Hn00NE3deCNccYUOzDXZxo3l10A++wzefRfWrQvlu3cPCeKEE0LzXkXfGamagkgKZtYXuB+oDzzq7renLN8XeAJoHZUZ7u4vlbdNJYWKrVoVfm2/9Rb84Q/bR1WNw6ZNoT3/hRfC44svQhPLscdub2ZKHpJj40b43e9C/8bGjaFp63/+B3bfPb4YpXBs2RJqbBMnhse//x2+B2Zw6KHbk8Qxx4S+rHxzDzFv2rT9RI3GjcOPl5rWJJb3pGBm9YHZwClACTAJGOjuM5PKjAI+cPeHzexg4CV371zedpUUyvfll/Dtb8OMGfDkk3Deebl7723bwj/8Cy/A88/DrFlh/qGHhuTQsWPoOF6wIEzfeWe49kDqrk2b4L33tieJd94JB9/69UMf0PHHhyTRp0+ojabaujU0W5XXwZ7cOb9hQ/lne6Wbl+kQ2ahR5tOHM51inHxKcUUndzRsWL37uhCSwlHACHf/djT9awB3/21SmZHAPHe/Iyp/j7t/q7ztKilktnAhnHJKqKY/91xozsmn2bO31yDefjv8c/XsGU4hPfHE/MYmhWnDhpAYypLEe++F/qsGDcJ3p379HQ/2qWeIpVOvXjjItmwZTjjI5rqQdPMbNQqxZJtMUudt3BgS2KpV2Z1q3LTpzgnjqqtCK8CuyDYpxHkuRQfgs6TpEuCIlDIjgFfNbBjQHDg53YbMbAgwBGBfDQ+a1n//GxLCmjXw2mtw9NH5jijUAn7xi/BYsiQkiW99S+fES2ZNm4YfDGU/GtatC01MZQmiYcNwYkBlTqlu3rywmnrcdzy9ONtTilevDjWjuMWZFNL9GVKrJQOBx939nqim8KSZdXP3bTus5D4KGAWhphBLtDXYlCmhVlCvXjjTpkePfEe0s732Cg+RymjeHE49NTxqC7OQ/Jo2Lcz/iTgvQSkBki916ggsSilzMTAGwN3fAZoAte5EtTjvsfzPf4Y21+bNQ8dyISYEEak54kwKk4CuZtbFzBoB5wMvppRZCJwEYGYHEZLC0hhjyrmyeywvWBCqjWX3WK6OxDBuXKghdOwYEkLXrlXfpojUbbElBXcvBYYCrwCzgDHuPsPMbjKz/lGxnwOXmtlU4CngQq9pF05UIK57LI8eDd/9Lnzzm2EU0Y4dq7Y9ERHQxWuxq1cv83gz27btPD8bDz4Iw4aF0/VeeKEwzucWkcKW7dlHGtYqZtV5j2X3cJ7/sGFheImXX1ZCEJHqpaQQs+q6x/K2bWGMoBtugB/+EJ59dtfGGhIRKU+FScHMhpqZRiXZRVW9x/IXX8BvfxvO+b/vvlBLePzxmj1ap4gUrmwOLe2BSWb2H+Ax4JXa1hkct8reY3nrVnj1VXjkkXCGUWlpGFXy1lvh3HML60IcEaldKkwK7n69mf0PcCpwEfCgmY0B/ujun8QdYF3y2Wfw2GPhsXBhGOb56qvhkkvCfRBEROKWVSOEu7uZfQF8AZQCbYC/mtlr7v7LOAOs7bZsgb//PdQKxo8PfQennBLuKzBggIaSFpHcqjApmNmVwAXAMuBR4BfuvsXM6gFzACWFXTBvXriXwOOPw+LF4a5ev/41XHwxdOmS7+hEpK7KpqbQDjjL3Rckz3T3bWZ2ejxh1U6bNoXrCh55JNxprF69cEOaSy8Nz+o8FpF8y+Yw9BLwVdmEmbUEDnb399x9VmyR1TLFxaF/YNmycAbSTTfBRRfpSmQRKSzZJIWHgV5J0+vSzJMMtm4NzUJ33RWGsx49Gk4+WcNHi0hhyiYpWPIpqFGzkRo6srByJXz/++HK4x//OFxnUN13UxIRqU7ZXNE8z8yuNLOG0eMqYF7cgdV0s2fDkUeGG96MHAkPPaSEICKFL5ukcDnwLeBztt89LcZbwdd848dD796wfDlMmBCGyhYRqQmyuXjtS8K9EKQC7nDvvfDLX0K3buFMo86d8x2ViEj2srlOoQnhDmmHEG6CA4C7D44xrhpn48ZQI3jySTj7bHjiiXA3NBGRmiSb5qMnCeMffRt4g3BbzTVxBlXTLFoUxiZ68kn4zW9gzBglBBGpmbI5i+jr7v49Mxvg7k+Y2V8Id1MT4P334cwzYfVqeO65cDc0EZGaKpuawpboeaWZdQNaAZ1ji6gGGT0ajj0WGjeGd95RQhCRmi+bpDAqup/C9cCLwEzgjlijKnBbt4bO5B/+EI46CiZNCvdKFhGp6cptPooGvVvt7iuAN4H9chJVAdMFaSJSm5VbU3D3bcDQHMVS8HRBmojUdtl0NL9mZtcAzxDGPQLA3b/KvErtM3lyGLOoYcNwQdqxx+Y7IhGR6pdNUii7HuEnSfOcOtaUdO210KQJvPuuLkgTkdormyua6/wtXz74IDQZ3X67EoKI1G7ZXNH8o3Tz3f3P1R9OYbrrLmjZEi67LN+RiIjEK5tTUg9PehwDjAD6xxhTQZk/H555Bsxg991DTaG4ON9RiYjEI5vmo2HJ02bWijD0RZ1w2WWwbVu4YhlgwYLto54OGpS/uERE4pBNTSHVeqBrdQdSiJYvD30Jqdavh+uuy308IiJxy6ZPYRzhbCMISeRgYEycQRWK//3fMBx2OgsX5jYWEZFcyOaU1LuTXpcCC9y9JKZ4CsaGDfDAA9C0aXidat99cx+TiEjcsmk+Wgi85+5vuPu/geVm1jnWqArA44/DsmXw859Ds2Y7LmvWDG69NS9hiYjEKpuk8H/AtqTprdG8WmvrVrj7bjjiCLjpJhg1Cjp1CmcgdeoUptXJLCK1UTbNRw3cfXPZhLtvNrNGMcaUd889B/PmhesTzEICUBIQkbogm5rCUjNLXJdgZgOAZfGFlF/ucOed0LUrDBiQ72hERHIrm5rC5UCxmT0YTZcAaa9yrg3eeCMMfjdyJNSvn+9oRERyK5uL1z4BjjSzFoC5e62+P/Odd8Kee8KPam3aExHJrMLmIzO7zcxau/tad19jZm3M7JZcBJdr06aFm+dceWUYEVVEpK7Jpk+hn7uvLJuI7sJ2WjYbN7O+Zvaxmc01s+Fplv/OzD6MHrPNbGW67eTK3XdD8+ZwxRX5jEJEJH+y6VOob2aN3X0TgJk1BRpXtJKZ1QceAk4h9ENMMrMX3X1mWRl3/2lS+WHAoZWMv9osXAhPPQVDh4aB70RE6qJsksJoYIKZ/Smavgh4Iov1egNz3X0egJk9DQwAZmYoPxC4MYvtxuK++8KZRz/9acVlRURqq2w6mu80s2nAyYAB44FOWWy7A/BZ0nQJcES6gmbWCegC/CPD8iHAEIB9YxhfYsWKcEHawIEavkJE6rZsR0n9gnBV89nAScCsLNaxNPMyDC/H+cBf3X1ruoXuPsrdi9y9aI899sgm3kp5+GFYtw5+8Ytq37SISI2SsaZgZt8gHKwHAsuBZwinpJ6Q5bZLgH2SpjsCizKUPZ8d7wGdMxs3wv33Q9++0L17PiIQESkc5dUU/kuoFZzh7n3c/feEcY+yNQnoamZdomExzgdeTC1kZgcAbYB3KrHtavPnP8OXX6qWICIC5SeFswnNRhPN7BEzO4n0TUJpuXspMBR4hdDcNMbdZ5jZTcnDZhBqIk+7Z7pzQXy2boV77oHDDoMTsq3/iIjUYhmbj9x9LDDWzJoDZwI/BfYys4eBse7+akUbd/eXgJdS5t2QMj1iF+KuFi++CLNnb78Hs4hIXVdhR7O7r3P3Ync/ndAv8CGw04VoNY073HEH7LcfnHVWvqMRESkMlbpHs7t/5e4j3f3EuALKlbfegvfeCzfRaZDN1RoiInVApZJCbXLnndCuHVx4Yb4jEREpHHUyKcyYAX/7GwwbtvOtNkVE6rI6mRTuvjskg5/k5coIEZHCVeeSQkkJFBfD4MHQtm2+oxERKSx1Lincf3+4PuFnP8t3JCIihadOJYVVq8JtNs89F7p0yXc0IiKFp04lhZEjYc0aDWkhIpJJnUkKmzaFeyacfDL06pXvaEREClOduWyruBgWL4Ynsrk9kIhIHVVnagr77w+XXhpqCiIikl6dqSkcd1x4iIhIZnWmpiAiIhVTUhARkQQlBRERSVBSEBGRBCUFERFJUFIQEZEEJQUREUlQUhARkQQlBRERSVBSEBGRBCUFERFJUFIQEZEEJQUREUlQUhARkQQlBRERSVBSEBGRBCUFERFJUFIQEZEEJQUREUlQUhARkQQlBRERSVBSEBGRBCUFERFJiDUpmFlfM/vYzOaa2fAMZc41s5lmNsPM/hJnPCIiUr4GcW3YzOoDDwGnACXAJDN70d1nJpXpCvwaONrdV5jZnnHFIyIiFYuzptAbmOvu89x9M/A0MCClzKXAQ+6+AsDdv4wxHhERqUCcSaED8FnSdEk0L9k3gG+Y2b/N7F0z65tuQ2Y2xMwmm9nkpUuXxhSuiIjEmRQszTxPmW4AdAWOBwYCj5pZ651Wch/l7kXuXrTHHntUe6AiIhLEmRRKgH2SpjsCi9KUecHdt7j7p8DHhCQhIiJ5EGdSmAR0NbMuZtYIOB94MaXM88AJAGbWjtCcNC/GmEREpByxJQV3LwWGAq8As4Ax7j7DzG4ys/5RsVeA5WY2E5gI/MLdl8cVk4iIlM/cU5v5C1tRUZFPnjw532GIiNQoZjbF3YsqKqcrmkVEJEFJQUREEmK7ollEapctW7ZQUlLCxo0b8x2KlKNJkyZ07NiRhg0b7tL6SgoikpWSkhJatmxJ586dMUt3GZLkm7uzfPlySkpK6NKlyy5tQ81HIpKVjRs30rZtWyWEAmZmtG3btkq1OSUFEcmaEkLhq+rfSElBREQSlBREJBbFxdC5M9SrF56Li6u2veXLl9OzZ0969uxJ+/bt6dChQ2J68+bNWW3joosu4uOPPy63zEMPPURxVYOtwdTRLCLVrrgYhgyB9evD9IIFYRpg0KBd22bbtm358MMPARgxYgQtWrTgmmuu2aGMu+Pu1KuX/vfun/70pwrf5yc/+cmuBVhLqKYgItXuuuu2J4Qy69eH+dVt7ty5dOvWjcsvv5xevXqxePFihgwZQlFREYcccgg33XRTomyfPn348MMPKS0tpXXr1gwfPpwePXpw1FFH8eWX4XYu119/Pffdd1+i/PDhw+nduzcHHHAAb7/9NgDr1q3j7LPPpkePHgwcOJCioqJEwkp24403cvjhhyfiKxtBYvbs2Zx44on06NGDXr16MX/+fABuu+02vvnNb9KjRw+ui2NnZUFJQUSq3cKFlZtfVTNnzuTiiy/mgw8+oEOHDtx+++1MnjyZqVOn8tprrzFz5syd1lm1ahXHHXccU6dO5aijjuKxxx5Lu2135/333+euu+5KJJjf//73tG/fnqlTpzJ8+HA++OCDtOteddVVTJo0ienTp7Nq1SrGjx8PwMCBA/npT3/K1KlTefvtt9lzzz0ZN24cL7/8Mu+//z5Tp07l5z//eTXtncpRUhCRarfvvpWbX1X7778/hx9+eGL6qaeeolevXvTq1YtZs2alTQpNmzalX79+ABx22GGJX+upzjrrrJ3KvPXWW5x//vkA9OjRg0MOOSTtuhMmTKB379706NGDN954gxkzZrBixQqWLVvGGWecAYSLzZo1a8brr7/O4MGDadq0KQC777575XdENVBSEJFqd+ut0KzZjvOaNQvz49C8efPE6zlz5nD//ffzj3/8g2nTptG3b9+05+03atQo8bp+/fqUlpam3Xbjxo13KpPNQKLr169n6NChjB07lmnTpjF48OBEHOlOG3X3gjjlV0lBRKrdoEEwahR06gRm4XnUqF3vZK6M1atX07JlS3bbbTcWL17MK6+8Uu3v0adPH8aMGQPA9OnT09ZENmzYQL169WjXrh1r1qyx51LXAAANyklEQVTh2WefBaBNmza0a9eOcePGAeGiwPXr13Pqqafyxz/+kQ0bNgDw1VdfVXvc2dDZRyISi0GDcpMEUvXq1YuDDz6Ybt26sd9++3H00UdX+3sMGzaMH/3oR3Tv3p1evXrRrVs3WrVqtUOZtm3bcsEFF9CtWzc6derEEUcckVhWXFzMZZddxnXXXUejRo149tlnOf3005k6dSpFRUU0bNiQM844g5tvvrnaY6+I7qcgIlmZNWsWBx10UL7DKAilpaWUlpbSpEkT5syZw6mnnsqcOXNo0KAwfmen+1tlez+FwvgEIiI1yNq1aznppJMoLS3F3Rk5cmTBJISqqh2fQkQkh1q3bs2UKVPyHUYs1NEsIiIJSgoiIpKgpCAiIglKCiIikqCkICI1wvHHH7/ThWj33XcfP/7xj8tdr0WLFgAsWrSIc845J+O2KzrV/b777mN90ih/p512GitXrswm9BpFSUFEaoSBAwfy9NNP7zDv6aefZuDAgVmt/7WvfY2//vWvu/z+qUnhpZdeonXr1ru8vUKlU1JFpNKuvhrSjBRdJT17QjRidVrnnHMO119/PZs2baJx48bMnz+fRYsW0adPH9auXcuAAQNYsWIFW7Zs4ZZbbmHAgAE7rD9//nxOP/10PvroIzZs2MBFF13EzJkzOeiggxJDSwBcccUVTJo0iQ0bNnDOOefwm9/8hgceeIBFixZxwgkn0K5dOyZOnEjnzp2ZPHky7dq14957702MsnrJJZdw9dVXM3/+fPr160efPn14++236dChAy+88EJiwLsy48aN45ZbbmHz5s20bduW4uJi9tprL9auXcuwYcOYPHkyZsaNN97I2Wefzfjx47n22mvZunUr7dq1Y8KECdX3R0BJQURqiLZt29K7d2/Gjx/PgAEDePrppznvvPMwM5o0acLYsWPZbbfdWLZsGUceeST9+/fPOMDcww8/TLNmzZg2bRrTpk2jV69eiWW33noru+++O1u3buWkk05i2rRpXHnlldx7771MnDiRdu3a7bCtKVOm8Kc//Yn33nsPd+eII47guOOOo02bNsyZM4ennnqKRx55hHPPPZdnn32WH/zgBzus36dPH959913MjEcffZQ777yTe+65h5tvvplWrVoxffp0AFasWMHSpUu59NJLefPNN+nSpUss4yMpKYhIpZX3iz5OZU1IZUmh7Ne5u3Pttdfy5ptvUq9ePT7//HOWLFlC+/bt027nzTff5MorrwSge/fudO/ePbFszJgxjBo1itLSUhYvXszMmTN3WJ7qrbfe4rvf/W5ipNazzjqLf/3rX/Tv358uXbrQs2dPIPPw3CUlJZx33nksXryYzZs306VLFwBef/31HZrL2rRpw7hx4zj22GMTZeIYXrtO9ClU971iRSQ/zjzzTCZMmMB//vMfNmzYkPiFX1xczNKlS5kyZQoffvghe+21V9rhspOlq0V8+umn3H333UyYMIFp06bxne98p8LtlDd+XNmw25B5eO5hw4YxdOhQpk+fzsiRIxPvl24o7VwMr13rk0LZvWIXLAD37feKVWIQqXlatGjB8ccfz+DBg3foYF61ahV77rknDRs2ZOLEiSxYsKDc7Rx77LEURweBjz76iGnTpgFh2O3mzZvTqlUrlixZwssvv5xYp2XLlqxZsybttp5//nnWr1/PunXrGDt2LMccc0zWn2nVqlV06NABgCeeeCIx/9RTT+XBBx9MTK9YsYKjjjqKN954g08//RSIZ3jtWp8UcnmvWBGJ38CBA5k6dWrizmcAgwYNYvLkyRQVFVFcXMyBBx5Y7jauuOIK1q5dS/fu3bnzzjvp3bs3EO6iduihh3LIIYcwePDgHYbdHjJkCP369eOEE07YYVu9evXiwgsvpHfv3hxxxBFccsklHHrooVl/nhEjRvC9732PY445Zof+iuuvv54VK1bQrVs3evTowcSJE9ljjz0YNWoUZ511Fj169OC8887L+n2yVeuHzq5XL9QQUpnBtm3VGJhILaehs2uOqgydXetrCrm+V6yISE1W65NCru8VKyJSk9X6pJDPe8WK1DY1rbm5Lqrq36hOXKeQr3vFitQmTZo0Yfny5bRt2zb20yJl17g7y5cvp0mTJru8jTqRFESk6jp27EhJSQlLly7NdyhSjiZNmtCxY8ddXj/WpGBmfYH7gfrAo+5+e8ryC4G7gM+jWQ+6+6NxxiQiu6Zhw4aJK2ml9ootKZhZfeAh4BSgBJhkZi+6+8yUos+4+9C44hARkezF2dHcG5jr7vPcfTPwNDCggnVERCSP4kwKHYDPkqZLonmpzjazaWb2VzPbJ92GzGyImU02s8lqzxQRiU+cfQrpTk9IPVdqHPCUu28ys8uBJ4ATd1rJfRQwCsDMlppZ+QOb5E87YFm+gyiH4quaQo8PCj9GxVc1VYmvUzaF4kwKJUDyL/+OwKLkAu6+PGnyEeCOijbq7ntUS3QxMLPJ2VxGni+Kr2oKPT4o/BgVX9XkIr44m48mAV3NrIuZNQLOB15MLmBmeydN9gdmxRiPiIhUILaagruXmtlQ4BXCKamPufsMM7sJmOzuLwJXmll/oBT4CrgwrnhERKRisV6n4O4vAS+lzLsh6fWvgV/HGUOOjcp3ABVQfFVT6PFB4ceo+Kom9vhq3NDZIiISn1o/IJ6IiGRPSUFERBKUFCrJzPYxs4lmNsvMZpjZVWnKHG9mq8zsw+hxQ7ptxRjjfDObHr33Treps+ABM5sbXTjYK4exHZC0Xz40s9VmdnVKmZzvPzN7zMy+NLOPkubtbmavmdmc6LlNhnUviMrMMbMLchTbXWb23+jvN9bMWmdYt9zvQswxjjCzz5P+jqdlWLevmX0cfR+H5zC+Z5Jim29mH2ZYN9Z9mOmYkrfvn7vrUYkHsDfQK3rdEpgNHJxS5njgb3mMcT7QrpzlpwEvEy4wPBJ4L09x1ge+ADrle/8BxwK9gI+S5t0JDI9eDwfuSLPe7sC86LlN9LpNDmI7FWgQvb4jXWzZfBdijnEEcE0W34FPgP2ARsDU1P+nuOJLWX4PcEM+9mGmY0q+vn+qKVSSuy929/9Er9cQrq1IN3xHIRsA/NmDd4HWKdeM5MpJwCfunvcr1N39TcJp0ckGEK6yJ3o+M82q3wZec/ev3H0F8BrQN+7Y3P1Vdy+NJt8lXByaNxn2XzZyMkZaefFZuDnEucBT1f2+2SjnmJKX75+SQhWYWWfgUOC9NIuPMrOpZvaymR2S08DCcCKvmtkUMxuSZnm241LF7Xwy/yPmc/+V2cvdF0P4xwX2TFOmEPblYELNL52KvgtxGxo1cT2WofmjEPbfMcASd5+TYXnO9mHKMSUv3z8lhV1kZi2AZ4Gr3X11yuL/EJpEegC/B57PcXhHu3svoB/wEzM7NmV5NuNSxSq6yr0/8H9pFud7/1VGXvelmV1HuPizOEORir4LcXoY2B/oCSwmNNGkyvt3ERhI+bWEnOzDCo4pGVdLM69K+09JYReYWUPCH6/Y3Z9LXe7uq919bfT6JaChmbXLVXzuvih6/hIYS6iiJ6twXKoc6Af8x92XpC7I9/5LsqSsWS16/jJNmbzty6hT8XRgkEcNzKmy+C7Ext2XuPtWd99GGNss3Xvn9btoZg2As4BnMpXJxT7McEzJy/dPSaGSovbHPwKz3P3eDGXaR+Uws96E/bw8XdkY4mtuZi3LXhM6JD9KKfYi8KPoLKQjgVVl1dQcyvjrLJ/7L8WLQNnZHBcAL6Qp8wpwqpm1iZpHTo3mxcrCXQ1/BfR39/UZymTzXYgzxuR+qu9meO8Kx0iL2cnAf929JN3CXOzDco4p+fn+xdWjXlsfQB9C9Wwa8GH0OA24HLg8KjMUmEE4k+Jd4Fs5jG+/6H2nRjFcF81Pjs8Id8X7BJgOFOV4HzYjHORbJc3L6/4jJKjFwBbCr6+LgbbABGBO9Lx7VLaIcHvZsnUHA3Ojx0U5im0uoS257Dv4h6js14CXyvsu5HD/PRl9v6YRDnB7p8YYTZ9GOOPmk7hiTBdfNP/xsu9dUtmc7sNyjil5+f5pmAsREUlQ85GIiCQoKYiISIKSgoiIJCgpiIhIgpKCiIgkKCmIRMxsq+04gmu1jdhpZp2TR+gUKVSx3o5TpIbZ4O498x2ESD6ppiBSgWg8/TvM7P3o8fVoficzmxAN+DbBzPaN5u9l4R4HU6PHt6JN1TezR6Ix8181s6ZR+SvNbGa0nafz9DFFACUFkWRNU5qPzktattrdewMPAvdF8x4kDEHenTAg3QPR/AeANzwM6NeLcCUsQFfgIXc/BFgJnB3NHw4cGm3n8rg+nEg2dEWzSMTM1rp7izTz5wMnuvu8aOCyL9y9rZktIwzdsCWav9jd25nZUqCju29K2kZnwrj3XaPpXwEN3f0WMxsPrCWMBvu8R4MBiuSDagoi2fEMrzOVSWdT0uutbO/T+w5hLKrDgCnRyJ0ieaGkIJKd85Ke34lev00Y1RNgEPBW9HoCcAWAmdU3s90ybdTM6gH7uPtE4JdAa2Cn2opIrugXich2TW3Hm7ePd/ey01Ibm9l7hB9SA6N5VwKPmdkvgKXARdH8q4BRZnYxoUZwBWGEznTqA6PNrBVh9NrfufvKavtEIpWkPgWRCkR9CkXuvizfsYjETc1HIiKSoJqCiIgkqKYgIiIJSgoiIpKgpCAiIglKCiIikqCkICIiCf8PrHHz9r3pFewAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plotting the training and validation accuracy\n",
    "plt.clf()\n",
    "\n",
    "acc = history.history['acc']\n",
    "val_acc = history.history['val_acc']\n",
    "\n",
    "plt.plot(epochs, acc, 'bo', label='Training acc')\n",
    "plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 7982 samples, validate on 1000 samples\n",
      "Epoch 1/9\n",
      "7982/7982 [==============================] - 1s 160us/sample - loss: 2.8004 - acc: 0.4629 - val_loss: 1.8114 - val_acc: 0.6540\n",
      "Epoch 2/9\n",
      "7982/7982 [==============================] - 1s 123us/sample - loss: 1.4506 - acc: 0.7150 - val_loss: 1.2971 - val_acc: 0.7210\n",
      "Epoch 3/9\n",
      "7982/7982 [==============================] - 1s 130us/sample - loss: 1.0562 - acc: 0.7790 - val_loss: 1.1187 - val_acc: 0.7530\n",
      "Epoch 4/9\n",
      "7982/7982 [==============================] - 1s 131us/sample - loss: 0.8259 - acc: 0.8225 - val_loss: 1.0103 - val_acc: 0.7780\n",
      "Epoch 5/9\n",
      "7982/7982 [==============================] - 1s 126us/sample - loss: 0.6578 - acc: 0.8573 - val_loss: 0.9503 - val_acc: 0.7980\n",
      "Epoch 6/9\n",
      "7982/7982 [==============================] - 1s 123us/sample - loss: 0.5302 - acc: 0.8875 - val_loss: 0.9203 - val_acc: 0.8060\n",
      "Epoch 7/9\n",
      "7982/7982 [==============================] - 1s 120us/sample - loss: 0.4261 - acc: 0.9103 - val_loss: 0.8966 - val_acc: 0.8080\n",
      "Epoch 8/9\n",
      "7982/7982 [==============================] - 1s 121us/sample - loss: 0.3461 - acc: 0.9271 - val_loss: 0.8998 - val_acc: 0.8100\n",
      "Epoch 9/9\n",
      "7982/7982 [==============================] - 1s 120us/sample - loss: 0.2900 - acc: 0.9360 - val_loss: 0.8985 - val_acc: 0.8150\n"
     ]
    }
   ],
   "source": [
    "#Retraining a model from scratch with less epochs, avoiding overfit\n",
    "model = models.Sequential()\n",
    "model.add(layers.Dense(64, activation='relu', input_shape=(10000,)))\n",
    "model.add(layers.Dense(64, activation='relu'))\n",
    "model.add(layers.Dense(46, activation='softmax'))\n",
    "\n",
    "model.compile(optimizer='rmsprop', loss='categorical_crossentropy', metrics=['acc'])\n",
    "\n",
    "history1 = model.fit(partial_x_train, partial_y_train, epochs=9, batch_size=512, validation_data=(x_val,y_val))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAH/1JREFUeJzt3XmUXGWd//H3J52QELIBiSYkJA3DIqAITAtGHQdlEQFBfiBbwAHNCXgEQXFBMgwjkjOugIg/JLLJL01ACDCgrCrIIlsnBlkii5CEQIAA2SBsSb6/P57blUqnurt6qb7VXZ/XOfdU3aVufas76U/d57n3uYoIzMzMAPrlXYCZmVUPh4KZmRU4FMzMrMChYGZmBQ4FMzMrcCiYmVmBQ8G6laQ6SW9KGt+d2+ZJ0jaSuv3cbUl7S5pfNP+UpH8rZ9tOvNclks7o7Ovb2O85kq7o7v1afvrnXYDlS9KbRbODgXeBNdn8CRHR2JH9RcQaYEh3b1sLImL77tiPpMnAMRGxZ9G+J3fHvq3vcyjUuIgo/FHOvolOjog/tra9pP4RsbonajOznufmI2tT1jxwjaSZklYCx0iaKOlBScskLZZ0gaQB2fb9JYWk+mx+Rrb+VkkrJT0gaauObput/7ykpyUtl/RLSfdLOq6Vusup8QRJz0paKumCotfWSTpP0uuS/gns18bP5z8lXd1i2a8knZs9nyxpXvZ5/pl9i29tX4sk7Zk9Hyzp/2W1PQH8a4n3fS7b7xOSDsqWfwS4EPi3rGnutaKf7X8Xvf7E7LO/LulGSWPK+dm0R9IXs3qWSfqzpO2L1p0h6SVJKyT9o+izflzSnGz5K5J+Wu77WQVEhCdPRATAfGDvFsvOAd4DvkD6ErEx8DFgD9KR5tbA08BJ2fb9gQDqs/kZwGtAAzAAuAaY0YltPwCsBA7O1n0LeB84rpXPUk6N/wsMB+qBN5o/O3AS8AQwDtgcuCf9Vyn5PlsDbwKbFO37VaAhm/9Cto2AzwJvAztn6/YG5hftaxGwZ/b8Z8DdwKbABODJFtseDozJfidHZzV8MFs3Gbi7RZ0zgP/Onu+b1bgLMAj4v8Cfy/nZlPj85wBXZM93yOr4bPY7OiP7uQ8AdgIWAKOzbbcCts6ePwIclT0fCuyR9/+FWp58pGDluC8ibo6ItRHxdkQ8EhEPRcTqiHgOmA78exuvvy4imiLifaCR9Meoo9seCMyNiP/N1p1HCpCSyqzxfyJieUTMJ/0Bbn6vw4HzImJRRLwO/KiN93kOeJwUVgD7AMsioilbf3NEPBfJn4E/ASU7k1s4HDgnIpZGxALSt//i9/1dRCzOfidXkQK9oYz9AkwCLomIuRHxDnA68O+SxhVt09rPpi1HAjdFxJ+z39GPgGGkcF5NCqCdsibI57OfHaRw31bS5hGxMiIeKvNzWAU4FKwcLxTPSPqQpD9IelnSCuBsYGQbr3+56Pkq2u5cbm3bLYrriIggfbMuqcway3ov0jfctlwFHJU9P5oUZs11HCjpIUlvSFpG+pbe1s+q2Zi2apB0nKRHs2aaZcCHytwvpM9X2F9ErACWAmOLtunI76y1/a4l/Y7GRsRTwGmk38OrWXPk6GzT44EdgackPSxp/zI/h1WAQ8HK0fJ0zItJ3463iYhhwH+RmkcqaTGpOQcASWL9P2ItdaXGxcCWRfPtnTJ7DbB39k37YFJIIGlj4Drgf0hNOyOAO8qs4+XWapC0NXAR8DVg82y//yjab3unz75EapJq3t9QUjPVi2XU1ZH99iP9zl4EiIgZEfFJUtNRHennQkQ8FRFHkpoIfw7MkjSoi7VYJzkUrDOGAsuBtyTtAJzQA+/5e2A3SV+Q1B84BRhVoRp/B5wqaaykzYHvtbVxRLwC3AdcDjwVEc9kqwYCGwFLgDWSDgT26kANZ0gaoXQdx0lF64aQ/vAvIeXjZNKRQrNXgHHNHeslzAS+KmlnSQNJf5zvjYhWj7w6UPNBkvbM3vs7pH6ghyTtIOkz2fu9nU1rSB/gWEkjsyOL5dlnW9vFWqyTHArWGacB/0H6D38x6ZtyRWV/eI8AzgVeB/4F+BvpuorurvEiUtv/Y6RO0OvKeM1VpI7jq4pqXgZ8E7iB1Fl7GCncynEW6YhlPnArcGXRfv8OXAA8nG3zIaC4Hf5O4BngFUnFzUDNr7+N1IxzQ/b68aR+hi6JiCdIP/OLSIG1H3BQ1r8wEPgJqR/oZdKRyX9mL90fmKd0dtvPgCMi4r2u1mOdo9Q0a9a7SKojNVccFhH35l2PWV/hIwXrNSTtJ2l41gRxJumMlodzLsusT3EoWG/yKeA5UhPEfsAXI6K15iMz6wQ3H5mZWYGPFMzMrKDXDYg3cuTIqK+vz7sMM7NeZfbs2a9FRFuncQO9MBTq6+tpamrKuwwzs15FUntX5gNuPjIzsyIOBTMzK3AomJlZgUPBzMwKHApmZlZQE6HQ2Aj19dCvX3ps7NCt6M3MakevOyW1oxobYcoUWLUqzS9YkOYBJnV5XEgzs76lzx8pTJ26LhCarVqVlpuZ2fr6fCgsXNix5WZmtazPh8L4Vm6k2NpyM7Na1udDYdo0GDx4/WWDB6flZma2vj4fCpMmwfTpMGECSOlx+nR3MpuZldLnzz6CFAAOATOz9vX5IwUzMyufQ8HMzAocCmZmVuBQMDOzAoeCmZkVOBTMzKzAoWBmZgUOBTMzK3AomJlZgUPBzMwKHApmZlbgUDAzswKHgpmZFTgUzMyswKFgZmYFDgUzMyuoWChI2lLSXZLmSXpC0iklttlT0nJJc7PpvypVj5mZta+Sd15bDZwWEXMkDQVmS7ozIp5ssd29EXFgBeswM7MyVexIISIWR8Sc7PlKYB4wtlLvZ2ZmXdcjfQqS6oFdgYdKrJ4o6VFJt0raqZXXT5HUJKlpyZIlFazUzKy2VTwUJA0BZgGnRsSKFqvnABMi4qPAL4EbS+0jIqZHRENENIwaNaqyBZuZ1bCKhoKkAaRAaIyI61uuj4gVEfFm9vwWYICkkZWsyczMWlfJs48EXArMi4hzW9lmdLYdknbP6nm9UjWZmVnbKnn20SeBY4HHJM3Nlp0BjAeIiF8DhwFfk7QaeBs4MiKigjWZmVkbKhYKEXEfoHa2uRC4sFI1mJlZx/iKZjMzK3AomJlZgUPBzMwKHApmZlbgUDAzswKHgpmZFTgUzMyswKFgZmYFDgUzMytwKJiZWYFDwczMChwKZmZW4FAwM7MCh4KZmRU4FMzMrMChYGZmBQ4FMzMrcCiYmVmBQ8HMzAocCmZmVuBQMDOzAoeCmZkVOBTMzKzAoWBmZgUOBTMzK3AomJlZgUPBzMwKHApmZlbgUDAzswKHgpmZFVQsFCRtKekuSfMkPSHplBLbSNIFkp6V9HdJu1WqHjMza1//Cu57NXBaRMyRNBSYLenOiHiyaJvPA9tm0x7ARdmjmZnloGJHChGxOCLmZM9XAvOAsS02Oxi4MpIHgRGSxlSqJjMza1uP9ClIqgd2BR5qsWos8ELR/CI2DA4kTZHUJKlpyZIllSrTzKzmVTwUJA0BZgGnRsSKlqtLvCQ2WBAxPSIaIqJh1KhRlSjTzMyocChIGkAKhMaIuL7EJouALYvmxwEvVbImMzNrXSXPPhJwKTAvIs5tZbObgC9nZyF9HFgeEYsrVZOZmbWtkmcffRI4FnhM0txs2RnAeICI+DVwC7A/8CywCji+gvWYmVk7KhYKEXEfpfsMircJ4OuVqsHMzDrGVzSbmVmBQ8HMzAocCmZmVuBQMDOzAoeCmZkVOBTMzKzAoWBmZgUOBTMzK6iZUHj6aTj4YFi2LO9KzMyqV82EwoIFcOutcMAB8NZbeVdjZladaiYU9tkHZs6EBx+EQw6Bd9/NuyIzs+pTM6EAcOihcMklcOedcPTRsHp13hWZmVWXmgoFgOOPh/PPh+uvh8mTYe3avCsyM6selRw6u2qdcgosXw5nnQXDhsEvfgFqczxXM7PaUJOhAHDmmelMpPPOgxEj4Oyz867IzCx/NRsKEvz857BiBfzwhzB8OJx2Wt5VmZnlq2ZDAVIwXHxxCoZvfzsFw+TJeVdlZpafmg4FgLo6mDEDVq6EKVNg6FA44oi8qzIzy0fNnX1UykYbwaxZ8KlPwTHHwC235F2RmVk+HAqZwYPh5pth553T9Qz33JN3RWZmPc+hUGT4cLjtNqivhwMPhKamvCsyM+tZDoUWRo2CP/4RNt8c9tsPnnwy74rMzHpOWaEg6V8kDcye7ynpG5JGVLa0/Iwdm4bCGDAgjZn0/PN5V2Rm1jPKPVKYBayRtA1wKbAVcFXFqqoC22yTguHtt2HvveGll/KuyMys8soNhbURsRo4BDg/Ir4JjKlcWdXhwx9OfQyvvgr77guvv553RWZmlVVuKLwv6SjgP4DfZ8sGVKak6rL77nDTTfDss6mPYcWKvCsyM6ucckPheGAiMC0inpe0FTCjcmVVl898Bq69FubOhYMOSk1KZmZ9UVmhEBFPRsQ3ImKmpE2BoRHxowrXVlW+8AW48sp0/cKXvgTvv593RWZm3a/cs4/uljRM0mbAo8Dlks6tbGnV56ij4KKL4A9/gC9/GdasybsiM7PuVe7YR8MjYoWkycDlEXGWpL9XsrBqdcIJ6V4M3/teuhfDr3/tezGYWd9Rbp9Cf0ljgMNZ19HcJkmXSXpV0uOtrN9T0nJJc7Ppv8qsJXff/S58//swfXoKh4i8KzIz6x7lHimcDdwO3B8Rj0jaGnimnddcAVwIXNnGNvdGxIFl1lBVpk1LRww//Wm6Sc8ZZ+RdkZlZ15UVChFxLXBt0fxzwKHtvOYeSfVdKa6aSfDLX6ZTVKdOTU1JJ52Ud1VmZl1TbkfzOEk3ZM1Br0iaJWlcN7z/REmPSrpV0k7dsL8e1a8fXH45HHwwnHxyOjvJzKw3K7dP4XLgJmALYCxwc7asK+YAEyLio8AvgRtb21DSFElNkpqWLFnSxbftXv37w9VXw157wVe+AjfckHdFZmadV24ojIqIyyNidTZdAYzqyhtHxIqIeDN7fgswQNLIVradHhENEdEwalSX3rYiBg2CG2+EhgY48sg0ymo5GhvTMN39+qXHxsZKVmlm1r5yQ+E1ScdIqsumY4AujQQkabSUTuaUtHtWS68dXWjIkHTHtu23T81JDzzQ9vaNjen2nwsWpLOXFixI8w4GM8tTuaHwFdLpqC8Di4HDSENftErSTOABYHtJiyR9VdKJkk7MNjkMeFzSo8AFwJERvfvkzs02gzvugC22gP33h0cfbX3bqVNh1ar1l61alZabmeVFnf07LOnUiDi/m+tpV0NDQzRV+S3RFixI93t+7z24917YbrsNt+nXr/T1DRKsXVv5Gs2stkiaHREN7W3XlTuvfasLr+3TJkxI92KISPdiWLhww23Gjy/92taWm5n1hK6Eggd3aMOHPgS3356uY9hnH3jllfXXT5sGgwevv2zw4LTczCwvXQmFXt3+3xN23TUNnvfCC/C5z8GyZevWTZqUhsmYMCE1GU2YkOYnTcqvXjOzNvsUJK2k9B9/ARtHRLnDZHSb3tCn0NLtt6ehtz/2sdQRvckmeVdkZrWmW/oUImJoRAwrMQ3NIxB6q899DmbOhAcfhEMOgXffzbsiM7PSutJ8ZB1w6KFw6aWpA/qoo2D16rwrMjPbkEOhBx13HJx/fhoK46tf9amnZlZ93ATUw045JQ25fdZZMHw4/OIXvkmPmVUPh0IOzjwzBcO556Zg+OEP867IzCxxKORAgp/9LAXDOeekYPj2t/OuyszMfQq5keDii+FLX4LvfCdd03D++Rte5GZm1pMcCjmqq4MZM9Id3Orq4JvfhLFj0zUN114L77yTd4VmVmscCjnbaKN0G8+mJnj8cTjtNJgzBw4/HMaMgRNPTMNw9+7xY82st3AoVJGddoIf/zgNoHf77XDAAekWn5/4RLpPwznnpBFYzcwqxaFQherqYN99U9PSyy/DZZelezSceWa6Q9tnPgNXXAErV+ZdqZn1NQ6FKjdsGBx/PNx9Nzz/PJx9NixalJaNHg3HHpuukl6zJu9KzawvcCj0IvX16Wjh6afh/vvhmGPg5pvTUcWECXD66TBvXt5Vmllv5lDohaTUz3Dxxal56ZprYJdd0rUPO+4Iu+8OF14Ir/faO16bWV4cCr3coEHpTKXf/z41K517broN6Mknp7OXDjkEbrwxLTMza49DoQ8ZPTpd6zB3bppOPjmdznrIIamj+uST4ZFHfHqrmbXOodBHffSj8POfp6OHP/wB9toLfvOb1LS0007wox+ldWZmxRwKfVz//rD//qnf4eWXUz/EZpvB978P48en+0fPmAFvvZV3pWZWDRwKNWTECJgyBe67D555Jp3J9Oyz6bTW0aPXnfrq+zyY1S6HQo3aZhv4wQ/gn/+Ev/wldVbPmpUujNt663WnvppZbVH0sl7HhoaGaGpqyruMPmnVqnSm0pVXpgvi1q6FceNg4sR0CuzEiWk01402yrtSM+soSbMjoqHd7RwKVsqLL8L118Nf/5qmhQvT8oEDoaFh/aAYPTrfWs2sfQ4F61YvvphOb22eZs9ed+3DVlutHxI775w6uM2sepQbCv6vayU1NsLUqekIYfx4mDYNJk2Cww5L6999Nw3x/de/ppC46y646qq0bvDgdOprc1B8/OMwcmR+n8XMyudQsA00NqazlFatSvMLFqR5SMEAqRlp4sQ0QbogbuHCFBDNQfGTn6wbqG+77dY/mthxxzQarJlVFzcf2Qbq60vft2HCBJg/v/z9rFqVbh7UHBIPPABLlqR1Q4emI4jmoNhjj3TKrJlVhvsUrNP69Ss9FIbUtWsYItIpsMVHE489lvYpwQ47rDuS+MQn0tFFP580bdYtcg8FSZcBBwKvRsSHS6wX8Atgf2AVcFxEzGlvvw6FyuuuI4VyrFwJDz+8LiQefBCWLk3rNt10XRPVxImpn2Lo0O59f7NaUQ0dzVcAFwJXtrL+88C22bQHcFH2aDmbNm39PgVIncfTpnX/ew0dmsZl2muvNL92LTz11PpHE7fcktb16wcf+UhqdtpuuxRezdOmm6ajDTPrmoqFQkTcI6m+jU0OBq6MdKjyoKQRksZExOJK1WTlae5MLnX2UaX165eakXbYAb7ylbRs6VJ46KF1QXH11bB8+fqvGzo0HckUB0V9/bplm2/u0DArR55nH40FXiiaX5Qt2yAUJE0BpgCMHz++R4qrdZMm9UwIlGPTTWG//dIEqW9i6dLUlLVgQXosnu65B1asWH8fm2yyYVAUTyNHOjTMIN9QKPVfsGQHR0RMB6ZD6lOoZFFW/aQ00utmm8Fuu5XeZtmy9YOiODzuvz+tLzZ48IZhUTz/gQ84NKw25BkKi4Ati+bHAS/lVIv1MSNGpFuU7rJL6fXLl5c+ypg/PzVVvfHG+tsPGtT6UUZ9PXzwgw4N6xvyDIWbgJMkXU3qYF7u/gTrKcOHp+E4dt659PoVK9YPjeLnTU0b3v964MDU9zJyZAqk4cPTY1tT8zaDBlX2s5p1RMVCQdJMYE9gpKRFwFnAAICI+DVwC+l01GdJp6QeX6lazDpq2LB0ptNHPlJ6/ZtvbniksWBB6utYsiTdr2LZsjStXt32ew0c2H5wtDUNGuSjFOs+vnjNrIIi4O231wVEa9Py5aWXL10K77/f9nsMGND+EcnQoeumIUNKPzpc+rZquE7BrOZJqRN78GDYYouOvz4C3nmn7eAoFS4vvLBu2TvvlPdedXUpIJpDoq0AaflYatkmm3h8q97IoWBWxSTYeOM0jRnTuX28+266cnzlytTs1dHHBQvWza9cmY58yjV4cOsBMmRIui5l7doUfmvXrpuK58td1x37aJ4v/vk3T8Xzba3ryLYd3c/RR8MJJ5T/8+8Mh4JZHzdwYJq6a/jyNWtSSHQmYFauTH0uzz+f5iPSH7x+/dZNxfOdXVdX1/l9NE8R6yYo/by9+c6ua23bnuBQMLMOqatL/RTDh+ddiVWCx6A0M7MCh4L1Ko2N6WKxfv3SY2Nj3hWZ9S1uPrJeo5w7wplZ1/hIwXqNqVPXH84b0vzUqfnUY9YXORSs11i4sGPLzazjHArWa7Q2arpHUzfrPg4F6zWmTUsXQxWr1B3hzGqVQ8F6jUmTYPr0NHy1lB6nT3cns1l38tlH1qtU0x3hzPoiHymYmVmBQ8HMzAocCmZmVuBQMDOzAoeCmZkVOBTMuoEH6rO+wqekmnWRB+qzvsRHCmZd5IH6rC9xKJh1kQfqs77EoWDWRR6oz/oSh4JZF3mgPutLHApmXeSB+qwv8dlHZt3AA/VZX+EjBTMzK3AomJlZgUPBrA/zldbWUe5TMOujfKW1dUZFjxQk7SfpKUnPSjq9xPrjJC2RNDebJleyHrNa4iutrTMqdqQgqQ74FbAPsAh4RNJNEfFki02viYiTKlWHWa3yldbWGZU8UtgdeDYinouI94CrgYMr+H5mVsRXWltnVDIUxgIvFM0vypa1dKikv0u6TtKWpXYkaYqkJklNS5YsqUStZn2Or7S2zqhkKKjEsmgxfzNQHxE7A38EfltqRxExPSIaIqJh1KhR3VymWd/kK62tMyp59tEioPib/zjgpeINIuL1otnfAD+uYD1mNcdXWltHVfJI4RFgW0lbSdoIOBK4qXgDSWOKZg8C5lWwHjOrEr5+onpV7EghIlZLOgm4HagDLouIJySdDTRFxE3ANyQdBKwG3gCOq1Q9ZlYdfP1EdVNEy2b+6tbQ0BBNTU15l2FmnVRfn4KgpQkTYP78nq6mdkiaHREN7W3nYS7MrEf5+onq5lAwsx7l6yeqm0PBzHqUr5+obg4FM+tR1Xz9hM+K8iipZpaDarx+wmdFJT5SMDPDo8o2cyiYmeGzopo5FMzM8FlRzRwKZmZU91lRPdkB7lAwM6N6z4pq7gBfsAAi1nWAVyoYPMyFmVkV665hQTzMhZlZH9DTHeAOBTOzKtbTHeAOBTOzKtbTHeAOBTOzKtbTHeAe5sLMrMr15LAgPlIwM7MCh4KZmRU4FMzMrMChYGZmBQ4FMzMr6HXDXEhaApS46LssI4HXurGc7lKtdUH11ua6OsZ1dUxfrGtCRIxqb6NeFwpdIampnLE/elq11gXVW5vr6hjX1TG1XJebj8zMrMChYGZmBbUWCtPzLqAV1VoXVG9trqtjXFfH1GxdNdWnYGZmbau1IwUzM2uDQ8HMzApqIhQkXSbpVUmP511LMUlbSrpL0jxJT0g6Je+aACQNkvSwpEezun6Qd03FJNVJ+puk3+ddSzNJ8yU9JmmupKq5X6ykEZKuk/SP7N/ZxCqoafvs59Q8rZB0at51AUj6ZvZv/nFJMyUNyrsmAEmnZDU9UemfVU30KUj6NPAmcGVEfDjveppJGgOMiYg5koYCs4EvRsSTOdclYJOIeFPSAOA+4JSIeDDPuppJ+hbQAAyLiAPzrgdSKAANEVFVFzxJ+i1wb0RcImkjYHBELMu7rmaS6oAXgT0iorMXpXZXLWNJ/9Z3jIi3Jf0OuCUirsi5rg8DVwO7A+8BtwFfi4hnKvF+NXGkEBH3AG/kXUdLEbE4IuZkz1cC84Cx+VYFkbyZzQ7Ipqr49iBpHHAAcEnetVQ7ScOATwOXAkTEe9UUCJm9gH/mHQhF+gMbS+oPDAZeyrkegB2AByNiVUSsBv4CHFKpN6uJUOgNJNUDuwIP5VtJkjXRzAVeBe6MiKqoCzgf+C6wNu9CWgjgDkmzJU3Ju5jM1sAS4PKsue0SSZvkXVQLRwIz8y4CICJeBH4GLAQWA8sj4o58qwLgceDTkjaXNBjYH9iyUm/mUKgCkoYAs4BTI2JF3vUARMSaiNgFGAfsnh3C5krSgcCrETE771pK+GRE7AZ8Hvh61mSZt/7AbsBFEbEr8BZwer4lrZM1Zx0EXJt3LQCSNgUOBrYCtgA2kXRMvlVBRMwDfgzcSWo6ehRYXan3cyjkLGuznwU0RsT1edfTUtbccDewX86lAHwSOChrv78a+KykGfmWlETES9njq8ANpPbfvC0CFhUd5V1HColq8XlgTkS8knchmb2B5yNiSUS8D1wPfCLnmgCIiEsjYreI+DSpKbwi/QngUMhV1qF7KTAvIs7Nu55mkkZJGpE935j0n+Uf+VYFEfH9iBgXEfWkZoc/R0Tu3+QkbZKdKEDWPLMv6ZA/VxHxMvCCpO2zRXsBuZ7E0MJRVEnTUWYh8HFJg7P/m3uR+vlyJ+kD2eN44P9QwZ9b/0rtuJpImgnsCYyUtAg4KyIuzbcqIH3zPRZ4LGu/BzgjIm7JsSaAMcBvszND+gG/i4iqOf2zCn0QuCH9HaE/cFVE3JZvSQUnA41ZU81zwPE51wNA1ja+D3BC3rU0i4iHJF0HzCE1z/yN6hnuYpakzYH3ga9HxNJKvVFNnJJqZmblcfORmZkVOBTMzKzAoWBmZgUOBTMzK3AomJlZgUPBLCNpTYvRO7vt6l9J9dU2Sq9ZKTVxnYJZmd7OhvYwq1k+UjBrR3avhB9n95h4WNI22fIJkv4k6e/Z4/hs+Qcl3ZDdj+JRSc1DJdRJ+k02Jv4d2dXiSPqGpCez/Vyd08c0AxwKZsU2btF8dETRuhURsTtwIWmkVrLnV0bEzkAjcEG2/ALgLxHxUdJYQ09ky7cFfhUROwHLgEOz5acDu2b7ObFSH86sHL6i2Swj6c2IGFJi+XzgsxHxXDaA4csRsbmk10g3SXo/W744IkZKWgKMi4h3i/ZRTxqCfNts/nvAgIg4R9JtpJtA3QjcWHQvC7Me5yMFs/JEK89b26aUd4uer2Fdn94BwK+AfwVmZzd4McuFQ8GsPEcUPT6QPf8rabRWgEmkWzkC/An4GhRuVjSstZ1K6gdsGRF3kW4eNALY4GjFrKf4G4nZOhsXjVYLcFtENJ+WOlDSQ6QvUkdly74BXCbpO6Q7nDWPQHoKMF3SV0lHBF8j3cmrlDpghqThgIDzqvCWmVZD3Kdg1o6sT6EhIl7LuxazSnPzkZmZFfhIwczMCnykYGZmBQ4FMzMrcCiYmVmBQ8HMzAocCmZmVvD/AbsMj+iMgl9IAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plotting the training and validation loss of the retrained model\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "loss = history1.history['loss']\n",
    "val_loss = history1.history['val_loss']\n",
    "\n",
    "epochs = range(1,len(loss)+1)\n",
    "\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "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": [
    "#plotting the training and validation accuracy of the retrained model\n",
    "plt.clf()\n",
    "\n",
    "acc = history1.history['acc']\n",
    "val_acc = history1.history['val_acc']\n",
    "\n",
    "plt.plot(epochs, acc, 'bo', label='Training acc')\n",
    "plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2246/2246 [==============================] - 0s 137us/sample - loss: 1.0000 - acc: 0.7850\n"
     ]
    }
   ],
   "source": [
    "results = model.evaluate(x_test, one_hot_test_labels)"
   ]
  }
 ],
 "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": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}