model1-Conv_2-batchnormalisation-augmentation-1.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "import keras\n",
    "from keras.callbacks import *\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "from keras.models import Sequential\n",
    "from keras.layers import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "traindata = pd.read_csv('sign_mnist_train.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainlabel=traindata['label'].values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "traindata.drop('label',inplace=True,axis=1)\n",
    "trainimages = traindata.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainimages=trainimages.reshape(-1,28,28,1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "testdata = pd.read_csv('sign_mnist_test.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "testlabel=testdata['label'].values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "testdata.drop('label',inplace=True,axis=1)\n",
    "testimages = testdata.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "testimages=testimages.reshape(-1,28,28,1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.preprocessing.image import ImageDataGenerator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "traingen=ImageDataGenerator(rotation_range=40,zoom_range=0.2,width_shift_range=0.2,height_shift_range=0.2,\n",
    "                  shear_range=0.2,horizontal_flip=True,rescale=1/255.0,validation_split=0.2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "traindata_generator = traingen.flow(trainimages,trainlabel,subset='training')\n",
    "validationdata_generator = traingen.flow(trainimages,trainlabel,subset='validation')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "testgen=ImageDataGenerator(rescale=1/255.0)\n",
    "testdata_generator = testgen.flow(testimages,testlabel)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "model=Sequential([])\n",
    "\n",
    "model.add(Conv2D(64,(3,3),activation=\"relu\",input_shape=(28,28,1)))\n",
    "model.add(BatchNormalization())\n",
    "model.add(MaxPooling2D(2,2))\n",
    "\n",
    "model.add(Conv2D(128,(3,3),activation=\"relu\"))\n",
    "model.add(BatchNormalization())\n",
    "model.add(MaxPooling2D(2,2))\n",
    "\n",
    "model.add(Flatten())\n",
    "model.add(BatchNormalization())\n",
    "model.add(Dense(256,activation=\"relu\"))\n",
    "model.add(BatchNormalization())\n",
    "model.add(Dense(26,activation=\"softmax\"))\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss=\"sparse_categorical_crossentropy\",optimizer='adam',metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_1 (Conv2D)            (None, 26, 26, 64)        640       \n",
      "_________________________________________________________________\n",
      "batch_normalization_1 (Batch (None, 26, 26, 64)        256       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 13, 13, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 11, 11, 128)       73856     \n",
      "_________________________________________________________________\n",
      "batch_normalization_2 (Batch (None, 11, 11, 128)       512       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 5, 5, 128)         0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 3200)              0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_3 (Batch (None, 3200)              12800     \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 256)               819456    \n",
      "_________________________________________________________________\n",
      "batch_normalization_4 (Batch (None, 256)               1024      \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 26)                6682      \n",
      "=================================================================\n",
      "Total params: 915,226\n",
      "Trainable params: 907,930\n",
      "Non-trainable params: 7,296\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define a Callback class that stops training once accuracy reaches 97.0%\n",
    "class myCallback(Callback):\n",
    "  def on_epoch_end(self, epoch, logs={}):\n",
    "    if(logs.get('accuracy')>0.985):\n",
    "      print(\"\\nReached 98.5% accuracy so cancelling training!\")\n",
    "      self.model.stop_training = True\n",
    "callback=myCallback()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/40\n",
      "687/687 [==============================] - 10s 15ms/step - loss: 2.1503 - accuracy: 0.3446 - val_loss: 2.7383 - val_accuracy: 0.2564\n",
      "Epoch 2/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 1.2079 - accuracy: 0.6066 - val_loss: 0.9692 - val_accuracy: 0.6389\n",
      "Epoch 3/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.8788 - accuracy: 0.7109 - val_loss: 1.4328 - val_accuracy: 0.4238\n",
      "Epoch 4/40\n",
      "687/687 [==============================] - 8s 12ms/step - loss: 0.7686 - accuracy: 0.7465 - val_loss: 365.7654 - val_accuracy: 0.0517\n",
      "Epoch 5/40\n",
      "687/687 [==============================] - 8s 12ms/step - loss: 0.6430 - accuracy: 0.7855 - val_loss: 2.6605 - val_accuracy: 0.5626\n",
      "Epoch 6/40\n",
      "687/687 [==============================] - 8s 12ms/step - loss: 0.5696 - accuracy: 0.8093 - val_loss: 0.2284 - val_accuracy: 0.7687\n",
      "Epoch 7/40\n",
      "687/687 [==============================] - 8s 12ms/step - loss: 0.5237 - accuracy: 0.8242 - val_loss: 0.8860 - val_accuracy: 0.7414\n",
      "Epoch 8/40\n",
      "687/687 [==============================] - 8s 12ms/step - loss: 0.4750 - accuracy: 0.8385 - val_loss: 0.4206 - val_accuracy: 0.7356\n",
      "Epoch 9/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.4489 - accuracy: 0.8498 - val_loss: 1.5462 - val_accuracy: 0.7711\n",
      "Epoch 10/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.4095 - accuracy: 0.8632 - val_loss: 101.0115 - val_accuracy: 0.1078\n",
      "Epoch 11/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.3894 - accuracy: 0.8669 - val_loss: 0.9741 - val_accuracy: 0.7427\n",
      "Epoch 12/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.3643 - accuracy: 0.8773 - val_loss: 0.5976 - val_accuracy: 0.7130\n",
      "Epoch 13/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.3392 - accuracy: 0.8863 - val_loss: 1.2287 - val_accuracy: 0.8086\n",
      "Epoch 14/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.3443 - accuracy: 0.8849 - val_loss: 0.2035 - val_accuracy: 0.8734\n",
      "Epoch 15/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.3329 - accuracy: 0.8878 - val_loss: 69.3496 - val_accuracy: 0.1275\n",
      "Epoch 16/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.3095 - accuracy: 0.8971 - val_loss: 2.4212 - val_accuracy: 0.4884\n",
      "Epoch 17/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.3073 - accuracy: 0.8948 - val_loss: 0.4124 - val_accuracy: 0.7775\n",
      "Epoch 18/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.2865 - accuracy: 0.9047 - val_loss: 3.0841 - val_accuracy: 0.4750\n",
      "Epoch 19/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.2932 - accuracy: 0.9032 - val_loss: 16.9515 - val_accuracy: 0.3764\n",
      "Epoch 20/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.2567 - accuracy: 0.9149 - val_loss: 1.7882 - val_accuracy: 0.6225\n",
      "Epoch 21/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.2535 - accuracy: 0.9141 - val_loss: 1.7072 - val_accuracy: 0.7791\n",
      "Epoch 22/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.2508 - accuracy: 0.9140 - val_loss: 0.2841 - val_accuracy: 0.9073\n",
      "Epoch 23/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.2370 - accuracy: 0.9201 - val_loss: 0.2343 - val_accuracy: 0.9139\n",
      "Epoch 24/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.2330 - accuracy: 0.9230 - val_loss: 0.6872 - val_accuracy: 0.8334\n",
      "Epoch 25/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.2205 - accuracy: 0.9268 - val_loss: 0.1677 - val_accuracy: 0.8656\n",
      "Epoch 26/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.2204 - accuracy: 0.9265 - val_loss: 12.2205 - val_accuracy: 0.4832\n",
      "Epoch 27/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.2130 - accuracy: 0.9285 - val_loss: 0.9914 - val_accuracy: 0.8561\n",
      "Epoch 28/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.2077 - accuracy: 0.9293 - val_loss: 0.2614 - val_accuracy: 0.9037\n",
      "Epoch 29/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.1975 - accuracy: 0.9343 - val_loss: 0.5602 - val_accuracy: 0.8871\n",
      "Epoch 30/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.2001 - accuracy: 0.9331 - val_loss: 0.5115 - val_accuracy: 0.8456\n",
      "Epoch 31/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.1916 - accuracy: 0.9338 - val_loss: 95.8431 - val_accuracy: 0.1930\n",
      "Epoch 32/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.1859 - accuracy: 0.9383 - val_loss: 0.6307 - val_accuracy: 0.8304\n",
      "Epoch 33/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.1709 - accuracy: 0.9429 - val_loss: 0.3663 - val_accuracy: 0.8346\n",
      "Epoch 34/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.1754 - accuracy: 0.9420 - val_loss: 0.0512 - val_accuracy: 0.9190\n",
      "Epoch 35/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.1700 - accuracy: 0.9434 - val_loss: 1.1317 - val_accuracy: 0.7425\n",
      "Epoch 36/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.1711 - accuracy: 0.9419 - val_loss: 0.3102 - val_accuracy: 0.9177\n",
      "Epoch 37/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.1692 - accuracy: 0.9435 - val_loss: 0.2111 - val_accuracy: 0.8345\n",
      "Epoch 38/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.1707 - accuracy: 0.9439 - val_loss: 1.6201 - val_accuracy: 0.8614\n",
      "Epoch 39/40\n",
      "687/687 [==============================] - 8s 11ms/step - loss: 0.1611 - accuracy: 0.9460 - val_loss: 0.2324 - val_accuracy: 0.8638\n",
      "Epoch 40/40\n",
      "687/687 [==============================] - 8s 12ms/step - loss: 0.1610 - accuracy: 0.9458 - val_loss: 0.4312 - val_accuracy: 0.9310\n"
     ]
    }
   ],
   "source": [
    "history=model.fit(traindata_generator,epochs=40,validation_data=validationdata_generator,callbacks=[callback])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAEICAYAAABPgw/pAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nO2deXhU5fXHPwfCEkjCKgRZBESFiOzghoobFRfcRaq2atFWq9a2tlprq7W1i3v91VqXVqvigm3dqIqoWBdwAQQElD1IIAlhCVsCZHl/f5y5yWQyM7kzmZl7Z/J+nmeeO8ude8/cufOdc897znnFGIPFYrFY0p9WXhtgsVgslsRgBd1isVgyBCvoFovFkiFYQbdYLJYMwQq6xWKxZAhW0C0WiyVDsIKewYhIaxHZLSL9Ermul4jIIBFJeK6tiJwiIoVBj1eIyHFu1o1jX0+IyK3xvt9iiUSW1wZY6hGR3UEPOwD7gJrA4+8bY6bHsj1jTA2Qk+h1WwLGmMMSsR0RmQZcaoyZELTtaYnYtsUSihV0H2GMqRPUgAc4zRjzTqT1RSTLGFOdCtsslqaw56P32JBLGiEivxORF0XkeRHZBVwqIkeLyCciUi4ixSLykIi0CayfJSJGRPoHHj8beP1NEdklIvNEZECs6wZenyQiK0Vkh4j8n4h8LCKXR7DbjY3fF5HVIrJdRB4Kem9rEXlARLaKyBrgtCjH5zYReSHkuYdF5P7A/Wki8lXg86wJeM+RtlUkIhMC9zuIyDMB25YBo8Psd21gu8tEZHLg+SOAvwDHBcJZW4KO7R1B7/9B4LNvFZFXRKSXm2MTy3F27BGRd0Rkm4iUiMjPg/bzq8Ax2Ski80XkwHDhLRH5yPmeA8fzg8B+tgG3icghIjIn8Fm2BI5bp6D3HxT4jGWB1/8sIu0DNg8JWq+XiFSISLdIn9cSBmOMvfnwBhQCp4Q89ztgP3AW+mecDYwFjkSvtgYCK4HrAutnAQboH3j8LLAFGAO0AV4Eno1j3R7ALuDswGs/AaqAyyN8Fjc2vgp0AvoD25zPDlwHLAP6AN2AD/S0DbufgcBuoGPQtjcDYwKPzwqsI8BJQCUwLPDaKUBh0LaKgAmB+/cC7wNdgIOA5SHrXgT0Cnwn3w7Y0DPw2jTg/RA7nwXuCNyfGLBxBNAe+CvwnptjE+Nx7gSUAj8C2gF5wLjAa78AFgOHBD7DCKArMCj0WAMfOd9z4LNVA9cArdHz8VDgZKBt4Dz5GLg36PMsDRzPjoH1jw289hhwV9B+fgq87PXvMN1unhtgbxG+mMiC/l4T77sJeClwP5xI/y1o3cnA0jjWvRL4MOg1AYqJIOgubTwq6PX/ADcF7n+Ahp6c104PFZmQbX8CfDtwfxKwMsq6M4EfBu5HE/Rvgr8L4NrgdcNsdylwRuB+U4L+T+D3Qa/loeMmfZo6NjEe58uA+RHWW+PYG/K8G0Ff24QNFwCfB+4fB5QArcOsdyywDpDA40XAeYn+XWX6zYZc0o8NwQ9EZLCI/DdwCb0TuBPoHuX9JUH3K4g+EBpp3QOD7TD6CyyKtBGXNrraF7A+ir0AzwFTA/e/DdQNJIvImSLyaSDkUI56x9GOlUOvaDaIyOUisjgQNigHBrvcLujnq9ueMWYnsB3oHbSOq++siePcF1gdwYa+qKjHQ+j5mC8iM0RkY8CGp0JsKDQ6AN8AY8zHqLc/XkSGAv2A/8ZpU4vFCnr6EZqy9yjqEQ4yxuQBv0Y95mRSjHqQAIiI0FCAQmmOjcWoEDg0lVb5InCKiPRBQ0LPBWzMBv4F/AENh3QG3nZpR0kkG0RkIPAIGnboFtju10HbbSrFchMaxnG2l4uGdja6sCuUaMd5A3BwhPdFem1PwKYOQc/lh6wT+vn+hGZnHRGw4fIQGw4SkdYR7HgauBS9mphhjNkXYT1LBKygpz+5wA5gT2BQ6fsp2OdMYJSInCUiWWhc9oAk2TgDuFFEegcGyG6OtrIxphQNCzwJrDDGrAq81A6N65YBNSJyJhrrdWvDrSLSWTRP/7qg13JQUStD/9umoR66QynQJ3hwMoTnge+JyDARaYf+4XxojIl4xROFaMf5NaCfiFwnIm1FJE9ExgVeewL4nYgcLMoIEemK/pGVoIPvrUXkaoL+fKLYsAfYISJ90bCPwzxgK/B70YHmbBE5Nuj1Z9AQzbdRcbfEiBX09OenwHfRQcpHUQ81qQREcwpwP/oDPRj4AvXMEm3jI8C7wJfA56iX3RTPoTHx54JsLgd+DLyMDixegP4xueF29EqhEHiTILExxiwBHgI+C6wzGPg06L2zgVVAqYgEh06c97+FhkZeDry/H3CJS7tCiXicjTE7gFOB89FB2JXACYGX7wFeQY/zTnSAsn0glHYVcCs6QD4o5LOF43ZgHPrH8hrw7yAbqoEzgSGot/4N+j04rxei3/N+Y8zcGD+7hfoBCIslbgKX0JuAC4wxH3ptjyV9EZGn0YHWO7y2JR2xhUWWuBCR09BL6L1o2ls16qVaLHERGI84GzjCa1vSFRtyscTLeGAteil+GnCOHcSyxIuI/AHNhf+9MeYbr+1JV2zIxWKxWDIE66FbLBZLhuBZDL179+6mf//+Xu3eYrFY0pIFCxZsMcaETRP2TND79+/P/Pnzvdq9xWKxpCUiErFa2oZcLBaLJUOwgm6xWCwZghV0i8ViyRCsoFssFkuGYAXdYrFYMgQr6BaLxZIhWEG3WCyWDME257JYLBY3GAO1tXqrrITy8sa3HTt0WVUFrVrpTaTx/YkTYcSIhJtoBd1isXhHbS2UlMC6dbB2rS6d+7t2Qbt2emvbtvH9qiqoqGh427NHl5WVuk7Hjo1vOTn6WkUF7N7d8LZrly4rK6Gmpl7AaxrNmtc88vKsoFssljTDGNiyBQoL9bZuXcP769bBvpAmnb17w4AButy/X1/ftUu34zzetw/atIEOHVSkO3SA/Pz6x+3b6zp79tTftm6F9ev1/t699eKekwO5uXDggfWPs7OhdWv1pp1l8P127aBLF+jcufEtL09tM6beqw/27o3R15OAFXSLpaVTUaEe8erVDW9r1qgXnJ2tAhm8zM5WUaup0XWc2/799ff37IFvvtFlMF27Qv/+UFAAZ56p4j1woC4POkj3kQmI1IdYgpg7F448EiJNrNocrKBbLH6kshI2btRwRE1NeC+xVSv19nbt0tjtzp2Nl7t3q8g6nm3o/dJS2LSp4b67d4dBg+DYY1VcKyv1tnevLnfs0Pft3au2tGmjt7Ztddmxoy779YNvfUuFun9/vR10EHTq5MUR9QXLlsH48fDHP8LPf5747VtBt1hShTE6YFZW1vC2aRMUFamAO8tt25q3rzZtVDg7dmwYe3ZunTrpctgwOOQQFfBBg+DggzVsYEkKv/qVRne+973kbN8KusWSSPbtgxUrYOlS+PJLXa5fr8K9ZQtUVzd+jwj06AF9+qg3O3683u/dG3r1gqyshoNzoQN1eXl669SpftmunW43Q3nlFRgyBA47LPb3GqOh+4EDE29XND77DF5+Ge68E7p1S84+rKBbWi6VlTBrFrz9tsZ8nZBB6LJNm/p4aLhbebkK99KlsHIl1NRggHVZh7LkwNMYPKgfg4/cDwccEP7Ws6fuy+KKDRvg/PNhyhR47rnY3//GGxq6/+ADOO64xNsXiV/+Ur/uG29M3j6soFtaFpWV8Oab8K9/weuva4w5L08zG5wBPSfO7DJVrZJslvWeyKIDrmXxyDEs2nUwSzZ2Zefu1vANHNwGVryt4WZL83n0Ub1AWbAgvvd/9JEu//zn1An6e+/BO+/AAw9oyCVZeDan6JgxY4yd4MLSbPbtg48/1pBGcApb8LJNG3j/fXjpJZg5U7MuunWD886DCy6AE08Mn0ZWW1ufseGkoIXcHn6iLT+6tSM1NRreyMmB4cP1NmIEbN8ON98Mr70GZ52V2kPTFHv36sfIzvbaEvfs26djrVu26NdTXh77GOupp6q4tm6toZe+fZNjq4MxcPTROlSycmXzk3hEZIExZkyEnRlPbqNHjzYWS8zU1hqzYoUxDz1kzOmnG9OhQwSpDXPr3t2Yq682ZvZsY6qqEmLOmDHGDB5szL/+ZcyqVcbU1DR8varKmL59jTnxxITsLmFUVBgzcqQxnTsb88tfGlNa6rVF7pg+Xb/KH/9Yl3PmxPb+2lpjunQx5tRTjWnVyphbb02KmQ145RW19YknErM9YL6JoKvWQ7d4R3U1zJ8P776rA4g5OfXFGaFFG2VlGu+eNUvdKtCsjNNO09S4gQMbVgoGLysr1WU+4QQdYEwQW7boWOYdd8Cvfx15vXvu0RS1RYvUDD9wzTXwt7/Ve6vt2mnmxU03aXahXznmGD0VPv5Yhx7uvRd++lP371+7VhN5Hn1UY+kff6wx+WSlvtfU6JXa/v2aspiI08966BZ/UFtrzNKlxvz5z8ZMnmxMXl699zxwoDEHHhjd4+7YUd/38MPGrF7t9acxzz+vZs2bF329bdv0Y11xRWrsaooXX1S7f/5zffz118ZceaUxbdoY07q1MZdeasySJd7aGI6FC9XuBx7Qx337GjN1amzbcD77/PnGvPOO3n/qqcTb6vDMM7qPF19M3DaxHrqlWVRXq3u5davGpbOzGy47dFAXZ/fu+vQ8Z+ncLyqC//1PC2VA3aSTT9bbiSfq8L/D/v31TY62b9dldraW1/koG+TKKzUNbcuWpgc8r7sOHn9cCyd79kyNfeFYswZGjdIizQ8+aDh0UFSkg3aPPqoXNmeeqR5wPKmByWDaNHj+eU3T79xZh0CWLdMsUbfcfDM8+KDWXLVtC4cfrqfv558nPstz/35NrczL0wHcVgnqbWs9dEtsVFcbs2CBMffea8wZZzT0pOO5ZWcb07+/MRdfrIHEdeu8/oTNprZWLyguuMDd+itW6KG4447k2hWNvXuNGT1a4+aFhZHX27rVmDvv1Fhznz7GbNyYOhuj2ZSdbcxVV9U/97vf6TEtL3e/nZNO0mPg8PDD7q6y4uGvf9Vtv/FGYrdLFA/dpi1aVHaXL4fZs2HOHHXdysv1tUMPhalTYcIETQdwysCdjnbBy5wc9bS7d2+47NDB04+XDJYv16yFb33L3fqHHgpnnAF//SvccovGrFPNLbeop/jyy1qBH4muXbWi8ayztMZp8mS9uOrYMXW2hvLkk3qa/fCH9c+NHq3LL77Q07MpjNHPf/HF9c9ddpkel7/8BY46KnH2VlTAb3+rx++00xK33SaJpPTJvlkP3WMqK415801jrrtOvWfHmx40yJhp0zSdoKjI9eb++19jSkqSaK/PuO8+PVzr17t/z+zZyY/ZRuLVV3XfN9wQ2/tef12zQc45p3EGT6qoqTHm4IONGT++4fObN+tnuvded9tZtUrXf/zxhs/fcIOOHxQXJ8ZeY4z50590Xx98kLhtOhDFQ7eC3pIoKjLmscd0YNEZfMzO1sePPhqbOgUxd65u6je/SbC9PuZb39J0xViorTVm6FBjhg/X+6li/XoNn4wapWGXWHngAdNgEDXVvPGG7v/55xu/1q+f+4FRZxB74cKGzzvhsESdv+XlerwnTUrM9kKJJug25JIpGKOl54WFOmoU7uaEUQ46CK64QmMAEyY0q7LEGPjZz/S+s/lMp7JSQxBXXx3b+0S07HvaNH1/U2GCDRs0HfL22+NPJayq0hBDdTW8+GJ8oZ4f/UgLYu6+W/t4TZsWny3x8vDDOpB83nmNXxs92n3F6IIF+vkPP7zh84ceqqGzv/0NfvGL2FuVV1bq2P22bbqcPl2Xd90V23YSQiSlT/bNeugJYsMGY+66y5hDDqkPm4BeJx94oDFjx+r18g9/qHGCpUsT6h46RRPQcMAqk3n7bf28//1v7O+tqND6prPPjr7e8uWalhcuRBALN99sEpI2V1VlzMSJxmRlabpfqlizxhgRY371q/CvxzIwOmGC/hzCMXOmbueFF6JvY+9eY266yZiCAmPy841p167hz865XXpp0/bEC9ZDzzD27oVXX9WRotmztQb6hBN0dGfoUO3S17NnQotowlFdrbscPFhLsnfvTurufMOsWZrydsIJsb83Oxt+8AP13lav1tqoUD79FE4/vf7rKy2Nz85334U//Qm+/3246KL4tuGQlQUzZmiL9AsugHnz9HtPNo88oul+3/9++NfdDozW1sLChXDJJeFfnzRJa9P+7/+06Vc4Cgv1OH7+uX4/xxyj9W9du+rSuXXtmpTZ5dwRSemTfbMeeozU1hrz2WfGXHutBuhAA4i/+pVnRTaPPqpmvPKKlpGfdZYnZqScI47Q9Ld42bRJB+HCDVC+9ZYObxx8sH6tnTvruHU8/OAHxnTqpFcFiWLdOmN69NA6sLKyxG03HBUVxnTtGj011O3AqBMn//vfI6/jDHSHxtiNUQ++SxfN4P3Pf9zZnyywg6JpSnW1DpPfeKOKNxjTvr0x3/62pkx4lXZgjNm9Wy85jz1W/2uOO85//UqSwcaN+jX88Y/N285llxmTk9MwVDB9uoY0Royoz7g47DBjLrwwvn2cc44OwiaaefM01HDssfENsrrlH/8wrvq1uBkYdXrALFoUeZ3t2/XP9Mor65+rqjLmF7/Q944Y4YsCZSvoacW+fcbMmqVNpHr00K+obVtjzjxTz/Dt27220BijhSegGS7G6Ih+pPhkJvHUU/q5v/iieduZP1+3c//9+vjBB/XxhAkNRf6EE/TPMh6OOsqYU05pnp2RcEram4o5x0ttrRYAHX5400M+556rQ0jR+MlP1Bfavz/6et//vq63ZYv+qZ5wgqkbH0rklU5ziCboNobuB4zR6Uweewz+8x9NF+nYUbNQzjtPA3bJbKIcI5s3a8bDeedpW1DQmqLCQk/NSgmzZmlDrmHDmred0aO1F/dDD2lnhD/8Ac49VydsCG4UlZ+vsd94KClJXr9vp6CqrCw52//sM81K+etfmy7JHz1ai6V27IjcSnf+fG2M1lQGyw9/qK0PbrhBe5jv2AH//Cd85zvxfY5UYwXdS3bt0hynRx/VXikdO+pULOefr23wfNqo+s47NVXr97+vfy4nJ/MHRWtrdQz6tNMS05fjxhv1q/7DH+Cqq3QAMLQnTM+e8Q2KGgPFxfqHkAzy8nS5a1f826iuVhvXr9ceN998U39/8WL1YS69tOntNDUw6gyIfve7TW/riCN0G889pz1sZs/WPIN0wQq6F3zxhYr49OmqgsOH66/529+u/6X4lFWr1PSrr27YtCknp3k/7nTgiy+0EdfEiYnZ3tlna3n92LFw223hPdH8fG0kVVERWweFHTs08yhZgu7MOb1zZ3zvf+YZbW4WOsVqt246gcXYsSrAbi5MHUFfsCC8oK9cqT+zMeHbWTXigQd0LpRbbvHVhbErrKCnispKrex45BG9nmzfXis+vv997SKYJhP63nqr/phvv73h846HbkzafJSYefttXZ56amK217q1zmQUDUeQS0t1/mi3OE0tkyXooGIXr6B//rmGPx5+WOvc+vXTWzz9Yg44QN8bqcDIaerqCH9TjBjhYdphM7GCnmwKC1XE//53bT87ZIhOZnjZZZq0mkZ8+qlOxXnHHY1bwObkqLe1f783jadSwdtv68VUMkUyFOc4xyvovXol3iaHvLz4BX3nThXiWKttIzF6dL1wh7JggUYvhwxJzL78TII69FoaUFurv/6zz9Zqhfvu02vB997TBs433JB2Ym4CJf49e4afIca5NM3UOPru3Tq7jdvuionC+fNwBNotxcUN358MmiPoO3YkNro4erSGA3fsaPza/PnqcSe5zs4XuBJ0ETlNRFaIyGoRuSXM6/1EZI6IfCEiS0Tk9MSbmgZUVGjawpAh+sv/5BONUaxbp67tiSembTxi5kz48EMNteTkNH7deS5TBf3997UvSqLi524J9tBjIRUhl+Z66IkWdNBxjmBqavQ5t/HzdKdJQReR1sDDwCSgAJgqIgUhq90GzDDGjAQuBv6aaEN9z8cf6/X4j36ktb/PPqvD9b/7XfKnFU8Bt9+uTYwiNWbKdEGfNUsv28ePT+1+e/TQZaweekmJDlp27px4mxyaK+iRUgzjIXhgNJgVK3T2Jbfx83THjYc+DlhtjFlrjNkPvACcHbKOAZz/207ApsSZ6HMqK3Vm3eOOU3fgvfe00cUll2RMMHnxYvVybrghch5vpgv6229r1CzVX2mbNjpPSDyCnp+f3AvCvLz4M5sS7aE7A6OhcXTnsfXQ6+kNbAh6XBR4Lpg7gEtFpAh4A7g+Idb5nU8/1Qka77tPs1WWLNGwSobxz3+qsATP9BJKJgt6YaGmvqU6fu4QTy56cXFyB0TBXyEXCN9Kd8ECTfdMRSMxP+BG0MP9x4fOLD0VeMoY0wc4HXhGRBptW0SuFpH5IjK/LFklZqlg3z6NjR9zjF7Pvf22ZrKECy6nOdXVmi5/5pmaIxwJ56NnYi66k66Y6vi5Q35+/B56MmlO2mKyBD10YHT+fBg5sulJvDMFN4JeBAQHgfvQOKTyPWAGgDFmHtAe6B66IWPMY8aYMcaYMQcEz/KeTixcqNdvf/gDXH45fPll4hKTfcisWVrq31TpcyZ76LNm6TCIV16eXwU9L08jjlVVsb2vqkrzBxIt6E5YxRkYra7WAuyWEm4Bd4L+OXCIiAwQkbbooGdoOcQ3wMkAIjIEFfQ0dsEj8MorWgS0bRv897+aW57IkR0f8vTT6pmf3kTeUqYKenW19hWfONG7BCUn5GJCr4sjUFWlFa2pEHSI/arMWT/RP53QgdGvv9Y/jpYyIAouBN0YUw1cB8wCvkKzWZaJyJ0iMjmw2k+Bq0RkMfA8cHmgK1jm8MEHGkQePVqnemtK4TKA7dt1Ho2pUzVjIhrJzEOvqIh/kofm8tlnegnvVfwcVJgrKtwf282bVfxTEUOH2MMuzvqJ9tC7d284MNrSBkTBZaWoMeYNdLAz+LlfB91fDhybWNN8xJIlMHmylur9979pVxQULy+9pMMFbpoaOX1GkiHov/qVdtNbuzbx226K99/X5UknpX7fDsHFRW56i6QiBx38J+jQcGB0wQK9cjz00MTvx6/YStGmKCzU9no5ORpMjTIyWFWlIfVkUFqqjZzmzEnO9sPxz39CQYG7S9ZWrbQPRzIEfelSrc2qrEz8tpvCmWot2oBwsom1uCjVgh5ryCWZgj5mTP3AaEsbEAUr6NEpK9Nr7cpKeOstvZ6LwiOPaJ/sa65RzzZRbN6sHuLMmakT9NWrYe5cHQx1GztOVgvd9et1uWFD9PUSjTF6DI45JrX7DSXW8v+W7qGDNv9qaQOiYAU9Mrt36wQT33wDr7/uqiny/Pmar/23v2lFoSNEzaGsTMV83TrddrxpYrHy9NMq5G76UTskQ9CNqT+O33yT2G03xcqVOv7tTOLhFcEdF92QKkF3wj+xnpNOWmEyBf2ZZ3Qu9ZY0IApW0MNTVaVTmy9YoC1vXdZ7L1umdUUvv6xiMGqURmnixRHztWs1dN+jR2ryvGtr9QdxyinQO7SELArJEPTNm/WHCan30OfN06XXHnq3bhrScuuhFxfrME+yq1qb66EnI0HMGRidMUMfWw+9pVNbq533Z83SKeEmT276PWjV//Ll6sifc47+F/TpA5MmwW9+o5uNhS1b4OSTNfQxc6b+UTSnMi8WPvxQhw7cDIYGk4xJLoKvclLtoc+dq71QvK4ybN1a/8xjCbmkosWvH0MuoF753r16BXHIIcnZh1+xgh6M0yP22Wfhrrvge99z/da1a/UkciIzgwaph3fZZdo//IwztB26GxwxX7VKxdzJsGhOZV4sPP20ivM558T2vmR46F4L+tFHJ2a6ueYSS/l/qgTdqT2IR9BF4pvMwg2OVz5qlD++u1TSwj5uFIzR/Lj774frr4df/CKmty9dqsvgUHuHDvDUUzpl23vv1YdgwvVsdti6VUMdK1dq6P7kk+tfa04zpE2b9KM54YtIVFRouuKFF8b+g0umoA8enFpBLy/XKy6v4+cOsVSLlpQkPwcdVCzjcTKcsv9kFWo5cfOWFm4BK+j13HGHeuXTpsGDD8Z8tjmCXhDSWFhEZ2X5+GO9f9ppehnfp49WH/74x/D44/r62rUq5l9/rVOTnXJKw201x0N/6SWdmOKkk6J7ei+/rH8a8cxynpubeEEvLNQf/xFHpFbQP/1U/+O9jp87uBX0ZE8OHUo8TkYy+rgEc9RRGmo566zk7cOvtIA5PFzwm9/oVPZXXqnudBzXaUuX6uREkbzaMWO0Pul//1PPb9kyXT72mHrFDu3aqZiHaw/T3BliQFO5xo3TUM4RRzRe7+mndY7H44+PfR/J8tD799eBrtdfT92cpXPn6mkwblzy9+WG4PL/aJ9/9249n1Ip6PFkuSSzY0anTnqF2xKxgv6736l3fvnl6irHGXRbuhQOPzz6Onl56jUEew61tep5Llumnvlxx0UWkeb2n+7QQTsYTJ6snucLL2hs32HjRnjnHW0kGc9hSLag792r4wup6Os2b57+4fll1vf8fJ2vtbw8eqFyqlIWHZoTcrEknpYdcvn97zVuftll8MQTcYv5/v3qEbhIVW9Eq1YqWGecoSGRaB6h8+OJp0uOM0PM6NHan+TQQ1XYH3igfnvTp+sfTDzhFlBB379fb4nAyUF3ZoWH1IRdamp09kC/hFvAfXFRqgU9Hg/dCnryaLmC/sc/wi9/qTMLPflks+qDV67UrnzxCHos5OWp4MZTAh/8I+rdWz31c86Bn/xE5+aoqtJS/6OPjj/Vy8l62LMnvveHUl6udgcLeipy0Zct0yshPwm62/J/Z3LoVAyKghV0v9EyBf2eezSLZepUVbFmNnsIl+GSDOLN+4XGs6x37KgDpbfeqpGmceM0ph9r7nkwiZ7kwslwcUIukBoPfe5cXfolwwWsh25xR8sT9Icegp//HKZM0RHABHTuWbpUN3PYYQmwLwrxllo77wn9EbVqpYk9Tz+tYt6uHVx0Ufz2JbonuiPoBx2k1ZLt26dG0OfN00KegQOTvy+3uC3/LymBrCydpzwVxDsoagU9ObSsQQWkIOUAACAASURBVNGKCnVJJ03S4qGsxHz8pUs1Jp2qUut4POCdO+sv20O57DId0C0ra15n4EQLemGhLg86SDM7+vVLnYd+zDHeTWgRji5dtJePGw+9Z8/UFdQ4A/Vus4+qq/VnmOHzwnhGyxL0N97QAO9NNyVMzEEFfdSohG0uIs0JuTiDopFIhP2JnuRi/XrIzq7PakmFoG/erO0Wrr46ufuJFREV6qYEPRWTQwfjjOtUVLgrRHOcEeuhJ4eWFXKZMUOvpU84IWGbrKjQgqCmUhYTQXNCLqm4zE1GyMXxziE1gv7JJ7r0U/zcIT/fXcglVfFziP2cTHYfl5ZOyxH03bu1muaCCxLa8f6rr/RyM9kDohB/yMWY1AxEJUvQHfr1Uw80kb3mQ5k7V0Mbfmy76sZDT7Wgx3rVaAU9ubQcQZ85U/P9pkxJ6GZTleEC8Xvoe/aoqKeboBcWNhZ00AKoZDF3roafsrOTt494aar8v6ZGQ0ZW0FsuLUfQZ8zQ4KLL3uZuWbpUB0MPPjihmw1Lc6f8SvZAVCIFfc8ebVTWv3/9c8nORa+q0plu/BhuAfXQN2+O3Iq5rExfS3UMHdwLutOCwg6KJoeWIeg7d+qA6IUXJnz4f+lSGDIkoWOsEcnO1miRn2aICcYZFEuEoAenLDokOxd90SJtL+CngqJg8vPVC4/UhjnVOehgPXS/0TIE/bXXNPCa4HALqKCnItwCOjgYb+8MSP6PqHVr/dNJRGFROEHv00eXyRJ0PxYUBdNUcZGXgu72O7eCnlxahqDPmKFqcNRRCd1seTkUFaVO0CH+dqXOe5NNohp0OTnowSEXJ4UxmYLer1/9H4ffaKr833rolswX9PJyeOstLYFMcLhl+XJdpiJl0SHeUmtITdwyUT3R16/XbJPQeHAyUxfnzfOvdw5Ne+hOHxe/py0mc7ailk7mC/qrr+poV5LCLZBaDz2ekEuqYuiQOA99/Xro27fxf3CyBH3DBr35NX4OTZf/l5Tod9yhQ+psatcO2raNTdBzc1ve1HCpIvMP64sv6nX72LEJ3/TSpSpgzmBdKmhJIZfg+LmDI+jxtBCOxrx5uvSzoOfmaj+baDH0VHrnDrFcNSZ7couWTmYL+tatMHu2hluS0JjDmdQild5GcwZFUzFZQyI99OD4uUO/frr98vLm7yOYefM0Rj98eGK3m0hEoueip4Og206LySWzBf2VV7QbUBLCLZDaDBeHeD30Dh1Sk1qZCEHft0/jwZE8dEh8LvrcuXoR16ZNYrebaKKV/6dyLtFgrKD7h8wW9Bdf1IqfkSMTvunNm7WQwwtB99scjsEkQtAdsY4m6ImMo1dWwsKF/g63OEQr/y8pSW1RkUMsToYV9OSScYI+d65O3EBZGbz3nnrnSQq3QOoFPTdXfzyRqgXDkcofUU5O8/PQg9vmhtK3ry4TKegLFuiFXDoIeqSQy549etyth96yyThBv/tuuPZa4D//0bK65szYEIVly3SZypRFqP8xxDLNW6oFvbkeevBMRaH07KlhkUQKulNQlOAyhaSQn68TZVdXN3zeCcN4IeixjOs01cbZ0jwyTtBLSvSEL3vmLZ1CaNiwpOxn6VKdFSbVP6B4eqKnWtD37dNM0XhZv14HmsMV+LRqpV56ogX9kEPq+677mZ49NcOnrKzh817koDvEmuViPfTkkZGCDvDV3O1JC7dA/YBoqme1iafjYiq9Ise+5kwUvX49HHhg5AHKROaiG1M/Q1E6ECkX3TnvvYqhuzkfa2r0vLCCnjwyStCNqT+xl5vBSQu3GONNhgvE13ExlV5RIjouFhaGD7c4JFLQ161Tb9fPFaLBOOX/oXF0L8r+HfLytKnZ/v3R17OzFSWfjBL0nTvrJz9Y3vW4pAW4i4p0X14IerweejoJeujEFqH066c90UPjyPHw1Ve6POKI5m8rFUQq/y8p0XBU9+6pt8mtk2H7uCSfjBL04JN8eefkuVxeZbhA7B56qmYrcmiuoFdX6x9mU4JeW1sfN24Oq1bp8pBDmr+tVBCpQVdxsc6umMDJuFxjBd0/uBJ0ETlNRFaIyGoRuSXCOheJyHIRWSYizyXWTHc4gt6Xb1i+K3n1+F5luEDsg6LObEWpzEOH+AV940aNtTYl6JCYsMvq1XpsvPBs46FjRz3G4Tx0L+Ln4P6ctJNbJJ8mBV1EWgMPA5OAAmCqiBSErHMI8AvgWGPM4cCNSbC1SRyv5aS8BRSXZbF9e3L2s3Sp/ni6dk3O9qMRa8gllY25oF7Q481Fj5ay6JDIXPRVq2DQoNQPbjeHcLnoXpX9g/tz0nroyceNhz4OWG2MWWuM2Q+8AJwdss5VwMPGmO0AxpjNiTXTHc5JfuLB+kt34qOJxqsBUfD/hALN9dDDTWwRSqIFPV3CLQ7hyv+9FHS3HroV9OTjRtB7A8GdM4oCzwVzKHCoiHwsIp+IyGnhNiQiV4vIfBGZXxaaSJsASooNWVQxfoSqidOvPJHU1Oh2vRL0eNqVQvoJerQOlrm50KVL8wV9/37dX7oJemj5f22tCrwVdIsbQQ93MRravDQLOASYAEwFnhCRzo3eZMxjxpgxxpgxByShiqN0fSU9KWXAyM5kZydH0Net094fXgk6xF6ZB6nPQ49X0AsLVbCys6Ovl4jUxXXrVAzTTdBDQy5bt+pgst9j6FbQk48bQS8C+gY97gNsCrPOq8aYKmPMOmAFKvAppaRwLz0ppdWggQwZkhxB9zLDxSGWZkipjqE3d6LoplIWHRIh6E6Gy6BBzdtOqsnPh+3b61N0vcxBh9gEXaT+Ks6SeNwI+ufAISIyQETaAhcDr4Ws8wpwIoCIdEdDMGsTaagbSooN+ZTAwQdTUJBcQS8oiL5eMonHQ0+VoGdl6SQM6STo6eahO6mLmwMjVV4LeseOKtRNORk7dtjZipJNk4fWGFMNXAfMAr4CZhhjlonInSIyObDaLGCriCwH5gA/M8ZsTZbRkSjd2oZ8SuGggygo0DassbaabYply2DAAG+9jFjblTrvSRXxNuiqrVWRdivo5eXN6+y4ahV07gzdusW/DS8ILf/3WtBbtXLnZNhOi8nH1ZQHxpg3gDdCnvt10H0D/CRw84TaWijd3YGeeRXQrl2dB/311zBuXOL248xS5CV5eZEnOQgllbMVOcQr6KWlGkaIlrLoEDzRRbxXS6tXq3eeTimL0Lha1Cmw8iqGDlbQ/ULGXPxs3w7VJov8Hjpe6/zIExl22b9f/yC8jJ9DbCGXHTv0kjgVsxU5xCvoblIWHRKRuujkoKcbof1cSkrqC468wk2DLivoySdjBN05uXv2bQtoWKRdu8QK+qpVmk3gtaD7fYaYeCe5iEXQm1stum+fvjfd4ufQuPzfyxx0Byvo/iBzBH1dJQD5B2uaRVaWtkNPpKAXFelywIDEbTMe/D5DTCo89F69tG9JvIKerimLoI5K584NPfR0EXRb9p9cMkbQS5dqoVJ+QX09fqIzXbZt06UXJf/B5OZCRYW7boNeCHpubnyCXlioBUNu7M3Kgt694xf0dM1wcQjORfdqcuhg3Ai6ndwi+WSMoJes1LOp5/D6M7ugQEWiOZMtBOMXQXd+FG5EM5UTRDs0x0N34507NCd1MV1z0B2Cy/+9bMzl4CYMaEMuySdzBH1dJe3YS6fh/eueKyjQToMrViRmH06zry5dErO9eIml42K6hVxSKehduqRfyqKDU/6/d6+mb/rdQ6+p0XPCCnpyyRhBLy2uJV82I13r1TbRmS7btqlYRZoaLVU4KYhuBh7TRdCNaXqmolD69dNxjdra2PYF9SmL6YoTcvFycuhgnMwrE9oUJIBzPlhBTy4ZI+glW7Lomb2zQVLxoEEaa02koHsdboH08NArK9Urc8u2bRoai9VDr6pyn5MfTDp2WQwmP1//0NeurX/sJXl5KuaRwpu2j0tqyBxB39WR/E6VDZ5r0wYOPTRxgr59uz8E3W3/aWe2Ii9i6BDb2EUsGS4O8eai792r70nX+DnUpy4uWqRLP8TQIfI5meomcS2VzBD0mhpK93emZ4/G196JzHTZts37+Dm474m+e7eKuhceOsSWix6PoMebi752rR6XdPfQARYvbvjYK5oS9FQ3iWupZISg13yzkTIOIL9P43LIggJYs0a9suaSbiEXry5z4+mJXlioy1hj6BC7oK9erctMEPRFizTKmIRu1DHh1kO3gp5cMkLQtyz8hlpakz+wY6PXCgp00Gzlyubvxy+C7vcpv+IR9PXrtXw9luPbqZMei1gFPd1z0KE+5LJ8uc6H6vVAfVNXjVbQU0NGCHrJl1pU1HNw43hIojJdjNEYuh9CLm6zXLyKW8YzyYWTshhLoyyR+FIXV63SPw4/fJfxcsAB+vmrqryPn4P10P1CZgj61+UA5B/eOKn40EO1veeyZc3bR2Wl9v/wg4fepo3O6ON2lvV08dBjCbc4xCvo6eydg54D3bvrfa/j59D0VaMdFE0NGSHopesqAMLG0Nu102yG5nrofqkSdcjNde+hp4OgFxbGNiDqEI+gp3sOuoMTdvGDoLv10O1sRcklIwS9ZKMmPDsneCiJyHTxS5Wog9tmSM66qSRWQd+1S49vPILety9s2aJXUG7Yu1d7qGeCoDtC7gdBb8pDt7MVpYaMOLwlW7LomLU34r9/QYFeZu/fH/8+/OihZ4qgx5Oy6BA80YUb1qzR8ZB0zkF3cBwYP8TQ27XTWzQP3cbPk0/6C3p5OaX7OtEzL3JeYkGBVi062Q3x4DdBd9MMyasYeqwTRceTsugQa+piJmS4OPjJQ4foV41W0FND+gv62rWUkE/+AZHrzJ0p45oTdnFCLn4SdDceeseO2jc8lbRtqze3hUWOd+2IcyzEKuhODnomeOh+FPRoaYtW0JNP+gv6mjUq6H0iq9Zhh2mKV3ME3fHQ/RJD9/scjrE06Nq0SWOrkcZAotG7t363sXjo3br553tsDgMH6meP58omGTTlodsMl+ST/oK+di2l9KRn/w4RV8nO1pO/uYKeleWfUXq/95+OZZKL4mLo0SO+K4m2bTWGHIugZ0K4BeCcc7RS1C+CHs3JsJNbpIa0F/SqVYVspTv5/dpGXa+5mS5OYy6/zBDvdoYYr7yiWDz04uLmDewNGgQLF7pbN1NSFkGvaoYN89qKemwM3XvSXtA3f62xkKbiiAUFOtGFm2nbwuGXxlwOubla6BQtcyddQi7NFfTzz9cmVU0Vj1VWZk7Koh+xgu49aS/oJWu1qKip+GtBgZZJr1kT33780sfFwU3HxZYi6FOmaLhm+vTo6znffSYMiPqRSIJeW6vnqRX05JPegl5VVTdRrhsPHeIPu/hN0N006EoHQa+pgc2bmyfoPXvCxIkq6NFmL8qklEU/EknQnfPADoomn/QW9PXrKTXaN7QpD33wYF3GK+h+aczl4MZDT4cY+ubNKsLNLY655BIdGP3oo8jrZELbXD+Tlxc+DGgbc6WO9Bb0QA46NC3oOTlaiZgpHnpTvTO8vszNyXGXh15crMvmCvo552jOfbSwy6pV2qXQeorJIZKTYSe3SB3pLeiBHPROebVkZze9eryZLjU1elL6SdCbCrns2ePNbEUObj30RAl6x45w7rkwY4Z6ieFYtcrGz5NJpHPSeuipI70Ffe1aSlv1ome+u1zCggL4+uvYJi8GKNfuvGkVcvH6R5SbCxUVTR/rRAk6aNilvBzeeCP865mUg+5HIl01en0utiTSW9DXrKGkfX/yYxD0vXvre4e4xW99XMD/Ewo4BVgVFdHXcwQ9EeXrp5yiBUrPPtv4tYoK2LjRCnoysYLuPekt6GvXUiK9XItBvJkufhT0pmYtcuKWXg6KQtNhl+JiPa7t2jV/n1lZMHUqzJxZ33vHwUlZtIKePJoSdDt2kXzSV9CNgTVrKK3q6roHyJAhuoxV0P3WmAvqBdPvHrobQU9k+9dLL9Usi3//u+HzTsqijaEnD+uhe0/6CvqWLVTurmbH/g6uPfROneDAA+Grr2Lbld8ac4EW0uTkWEEPZfRonXYwNOxic9CTT1NZLn7pg5TJpK+gr1lDKeqax9Klr18/jaXGgh9DLhB9GrqWKugi6qX/738NG3atXq3xdeslJo9oHnpOTurbOLdE0lfQA10WIbYBtfx86qpL3eK36eccovXO8EsMPVouujH6XSR6xp1LLtHlc8/VP2czXJJPx476hxpO0O0faWpwJegicpqIrBCR1SJyS5T1LhARIyJjEmdiBAI56BCboPfqVZ9Z4ZZt29Qbzmo8B7WnRGtX6jzvDJ6mGjce+rZtGu8+8MDE7nvgQDjmGA27GKPP2Rz05CMS/py0vdBTR5OCLiKtgYeBSUABMFVECsKslwvcAHyaaCPDsnYtpZ0OA2ILueTnw9atsc0v6rcqUYemZojxYrYiB+ePJJqgJzIHPZRLLtHui0uWaJHVpk3WQ08F4a4arYeeOtx46OOA1caYtcaY/cALwNlh1vstcDcQeXLPRLJmDSUBQe/Rw/3bHG9+82b373F6ofsNP7crdeOhJ1PQL7pIr6iefdamLKYSK+je4kbQewPBc6oXBZ6rQ0RGAn2NMTOjbUhErhaR+SIyv6ysLGZjG7B2LSXZ/enWTWetcYsj6LHE0f3WC90h2qCol425wHtB794dJk3SOPqKFfqcFfTkE07Q7WxFqcONoIcrwzR1L4q0Ah4AftrUhowxjxljxhhjxhxwwAHurQylshI2bqS09YExz0PpiEcscXQ/h1z86qG3bQtt2kQX9E2bdJkMQQfNdtm0CR5/XB/bGHryCRcG9PpcbEm4EfQioG/Q4z7ApqDHucBQ4H0RKQSOAl5L6sBooHa/pKp7zCXj8XrofhZ0Yxq/5ocfUVMNuoqL9SqjY8fk7P+ss3T7s2frOItXA8QtCRty8RY3gv45cIiIDBCRtsDFwGvOi8aYHcaY7saY/saY/sAnwGRjzPykWAx1QdGSityYPXRnfbeCboz/eqE75ObqlHrhugv64UfkRtCT5Z2DTg5+/vl634ZbUkNolovTxtlmuaSGJgXdGFMNXAfMAr4CZhhjlonInSIyOdkGhmXtWgBKt7eN2UNv2xa6dXMfcqmo0IwYv3roED7s4nUMHbwXdNCwC1hBTxWhHrrXbZxbGq4yq40xbwBvhDz36wjrTmi+WU2wZg27O/Zkz55WcXXpi6W4yK9VotCw/3Ropo9fPPRohUXFxTB2bHJtmDABzjhDwy+W5OPE0GtroVUrO7lFqvFZqYxL1qyhpM8YWBFbDrpDLILux8ZcDpF6Z3g9W5FDNA/dmNR46K1ba/dFS2rIy9Pvds+ehuEXr8/FlkJ6lv6vXUtpz2FAfH20Y6kW9WNjLodIIRe/XObm5kYW9F27NJyVbEG3pJbQc9IKempJP0GvrdUc9EBRUXNCLuGyQ0JJh5CLX+dwjOahJzMH3eIdoVeNthd6akk/QS8uhn37KO0wAIg/5LJ3b+Qc7mDSIeTi1wkFrKC3PKyH7i3pJ+hOymLr3rRqpbO4x4rj1bsJu6RjyMUvPyIr6C2P0Imi/XIuthTST9ADKYslNd3p3j2+5lOOiLgZGN22TSsek1X80hwihVz88iPKydF4fm1t49esoGcmoU6GX8J/LYX0E/RAt/zSPblxTywcS7Wo05hL3M1DnVKi9Z8G739EOTk6TlFZ2fi14mKdR7Rz59TbZUkekUIutko3NaSfoN9wA+zcScnm+HLQITZB92tjLojcf9rryS0cok1y4aQs+vGP0hI/4QTdyzbOLY30E3QAEUpK4hsQBRXotm3dx9D9OCDqEKkZkvOal0TruJiKHHRL6gkXQ/fasWhJpKWgGwOlpfGlLIJ6hW6Li9JB0COFXLy+zI02yYUV9MykbVto375h2qLXjkVLIi0FfccObUgVr6CDe0H3a2Muh3A90f1ymWs99JZJsJNhBT21pKWgO0Icb8gF3FeLpqOH7ofGXBBZ0CsrobzcCnqmEjyuYye3SC1pKeilpbpMtodeXa0nZroJul+8okiC7hz3RE8ObfEH1kP3jrQU9ER46Pn5sGULVFVFXqe8XJd+FvRIIRc//IgiCbrNQc9sQgXdD1eLLYW0FvTmeujGRJ8s2s9Vog7p6KFbQc9srIfuHWkp6KWlOqN7czxnN9Wifm7M5eB46MGNxvwSt7SC3jIJnhrRCnpqSUtBLynRCR1aNcN6N8VFfm7M5ZCXp6X1FRX1z/nlMrddO820CQ0JFRfrH3L37t7YZUkuTm2EX9o4tyTSUtCbk4Pu4EbQ0yXkAg3DLn7xipxK1nAees+ezftDtvgXx0O3fVxST1r+pEpKEifo0VIX0yXkAvVesF9mK3II13HR5qBnNrm5WidSVqaP/XIutgTSVtCbk+ECGg7o0sVdyMXPDaRCPfTdu/11mWsFveXhnHsbN+rSD+G/lkLaCXptrWamNNdDh6Zz0bdt05Mzy8czr0bqbueXH5EV9JaHc04WFTV8bEk+aSfo27ZpwU9zPXRoulrU71Wi0Djk4pfGXA6hgl5VpZfiVtAzFyvo3pF2gp6IHHQHNx663wXd71N+hQp6aamGhKygZy5W0L0j7QQ9EWX/Dk1NFu33xlzg/xliQgXd5qBnPlbQvSPtBD0RZf8OvXpp/na4CRggPTz0SCEXP8XQg4+vFfTMxwq6d6StoCfKQw/eZijpIOjt2+ugbbqEXKygZz6Ok7Fhgz/aOLck0k7QTzoJ7r8/MR5oNEE3pn4+UT/jFO/4dVDUKSxywlrFxWpzIq6wLP7EOff27PHPedhS8HFCXnhGjtRbIohWXLRnj2Zk+D2GDo2bIYH3sxU5BE8U3aGDHuvu3aFNG68tsyQLZ/JyP9VDtBTSzkNPJNEadKVDlahDsKDv2KEi6pfL3NAGXTYHPfMRqRdyK+ippUULepcu6imGE/R0aMzlEBpy8dOPyAp6y8QKuje0aEFv1UpjudE89HQMufjpR2QFvWXinIN+ybZqKbRoQYfI1aLpGnLxs6DX1modgRX0zMd66N7Q4gU9UrVouoZc/DK5hYMj6Lt26ZR/1dVW0FsCzqC8n87FloAV9AiCns4hFz9d5gZ76DYHveVgPXRvaPGC3quXdm+srm74/LZt0Latptr5HSfXu7bWfyEXx1MLFvQDD/TOHktqsILuDa4EXUROE5EVIrJaRG4J8/pPRGS5iCwRkXdF5KDEm5ocnMminWb8Dk5RkYg3dsWC86PZvdt/gm499JaJHRT1hiYFXURaAw8Dk4ACYKqIFISs9gUwxhgzDPgXcHeiDU0WkapFt21Lj3AL1P94ysv9NVsRWEFvqVgP3RvceOjjgNXGmLXGmP3AC8DZwSsYY+YYY5xpij8B+iTWzOQRqVo0Hfq4ODhhjeJivdrwk1fUvr2mhzqC3rmzPmfJbKyge4MbQe8NbAh6XBR4LhLfA94M94KIXC0i80VkfllojMMjIlWLppOg+7m7nUh9gy6bg95ysILuDW4EPVwUOWwHcRG5FBgD3BPudWPMY8aYMcaYMQcccIB7K5OI0yQqVNDToTGXg58FHaygt0S6dWu4tKQGN4JeBPQNetwH2BS6koicAvwSmGyM2ZcY85JPdraGKNI5hu6EXKygW/zCWWfB66/DkCFeW9KycCPonwOHiMgAEWkLXAy8FryCiIwEHkXFfHPizUwuodWiVVU6uJiuHrqfYuhQP8mFFfSWQ5s2cOaZXlvR8mhS0I0x1cB1wCzgK2CGMWaZiNwpIpMDq90D5AAvicgiEXktwuZ8SWhxUXm5LtNF0B0PfeNGXfrNQ8/N1T+bvXutoFssycRVP3RjzBvAGyHP/Tro/ikJtiul5OfD/Pn1j9OpShTSI+TiHF8r6I2pqqqiqKiIvXv3em2KxUe0b9+ePn360CaGyQPSboKLZBAackmnxlwA7drpza8eek6OThgCVtDDUVRURG5uLv3790fSoZLNknSMMWzdupWioiIGDBjg+n0tvvQf1EPfs6e+xWs6NeZyyM2F/fvr7/sJp7gIrKCHY+/evXTr1s2KuaUOEaFbt24xX7VZQadxtWi6hVyg3iv302xFDlbQm8aKuSWUeM4JK+jUi4wTdkm3kAv4u5DDEfQOHfx39WCxZBJW0GnsoTshl86dvbEnHvzcf9oR9F690qPZWUtj69atjBgxghEjRpCfn0/v3r3rHu934nhNcMUVV7BixYqo6zz88MNMnz49ESZbImAHRQkfcunc2X+hi2ikg4duwy3+pFu3bixatAiAO+64g5ycHG666aYG6xhjMMbQqlV4H/DJJ59scj8//OEPm29siqmuriYrK31k0nroaHlyVlbDkEs6xc+h3kP3W1ER1NtmBd0FN94IEyYk9nbjjXGZsnr1aoYOHcoPfvADRo0aRXFxMVdffTVjxozh8MMP584776xbd/z48SxatIjq6mo6d+7MLbfcwvDhwzn66KPZvFlrDW+77TYefPDBuvVvueUWxo0bx2GHHcbcuXMB2LNnD+effz7Dhw9n6tSpjBkzpu7PJpjbb7+dsWPH1tlnjHYjWblyJSeddBLDhw9n1KhRFBYWAvD73/+eI444guHDh/PLX/6ygc0AJSUlDBo0CIAnnniCiy++mDPPPJNJkyaxc+dOTjrpJEaNGsWwYcOYOXNmnR1PPvkkw4YNY/jw4VxxxRWUl5czcOBAqgMTLJSXlzNgwABqamri+g5ixQo6jSeLTqfGXA7WQ7ckg+XLl/O9732PL774gt69e/PHP/6R+fPns3jxYmbPns3y5csbvWfHjh2ccMIJLF68mKOPPpp//OMfYbdtjOGzzz7jnnvuqftz+L//+z/y8/NZvHgxt9xyC1988UXY9/7oRz/i888/58svv2THjh289dZbAEydOpUf//jHLF68mLlz59KjRw9ef/113nzzTT777DMWL17MT3/60yY/9uULVQAADyBJREFU97x583jmmWeYPXs22dnZvPrqqyxcuJB33nmHH//4xwAsXryYP/3pT7z//vssXryY++67j86dO3PsscfW2fPcc89x0UUX0TpFl/vpcy2RZIKrRdOpMZeDFfQMIeDB+oWDDz6YsWPH1j1+/vnn+fvf/051dTWbNm1i+fLlFBQ0nB4hOzubSZMmATB69Gg+/PDDsNs+77zz6tZxPOmPPvqIm2++GYDhw4dz+OGHh33vu+++yz333MPevXvZsmULo0eP5qijjmLLli2cddZZgBbmALzzzjtceeWVZGdnA9DVxY974sSJdAlcphtjuPnmm/noo49o1aoVGzZsYMuWLbz33ntMmTKlbnvOctq0aTz00EOceeaZPPnkkzzzzDNN7i9RWEEPkJ8PmwItx7Ztg759o6/vN9JlUNSSXnTs2LHu/qpVq/jzn//MZ599RufOnbn00kvD5km3bdu27n7r1q3rwg+htGvXrtE6TugkGhUVFVx33XUsXLiQ3r17c9ttt9XZES7VzxgT9vmsrCxqa2sBGn2O4M/99NNPs2PHDhYuXEhWVhZ9+vRh7969Ebd7wgkncN111zFnzhzatGnD4MGDm/xMicKGXAIEV4umc8jFjzH0wYNh9Gg49livLbE0h507d5Kbm0teXh7FxcXMmjUr4fsYP348M2bMAODLL78MG9KprKykVatWdO/enV27dvHvf/8bgC5dutC9e3def/11QEW6oqKCiRMn8ve//53KykoAtgXykvv378+CBQsA+Ne//hXRph07dtCjRw+ysrKYPXs2GwMl2aeccgovvPBC3facJcCll17KJZdcwhVXXNGs4xErVtAD5OfrZNE1NTbkkmi6ddNeLoExJ0uaMmrUKAoKChg6dChXXXUVxybhH/r6669n48aNDBs2jPvuu4+hQ4fSKcRL6datG9/97ncZOnQo5557LkceeWTda9OnT+e+++5j2LBhjB8/nrKyMs4880xOO+00xowZw4gRI3jggQcA+NnPfsaf//xnjjnmGLY7ucphuOyyy5g7dy5jxozhpZde4pBDDgFg2LBh/PznP+f4449nxIgR/OxnP6t7zyWXXMKOHTuYMmVKIg9P0zjpSKm+jR492viJv/zFGDBm1Spd3nOP1xbFxowZavdjj3ltiSVWli9f7rUJvqGqqspUVlYaY4xZuXKl6d+/v6mqqvLYqth5/vnnzeWXX97s7YQ7N4D5JoKu2hh6ACe+61zhWQ/dYkk9u3fv5uSTT6a6uhpjDI8++mha5YEDXHPNNbzzzjt1mS6pJL2OVBJxiovSVdB7B2Z57ZM203NbLI3p3LlzXVw7XXnkkUc827cV9AChgp5uhUVDh8KKFRAI71kslhaIFfQA6e6hAxx6qNcWWCwWL7FZLgE6dND481df6eN0FHSLxdKysYIeRH4+VFTofSvoFosl3bCCHoQTdmnXDgJVwhZLxjNhwoRGRUIPPvgg1157bdT35QRKgDdt2sQFF1wQcdvzgyfsDcODDz5IheNJAaeffjrlzkztlpiwgh6Ek7povXNLS2Lq1Km88MILDZ574YUXmDp1qqv3H3jggVErLZsiVNDfeOMNOqfRZATGmLoWAl5jBT0Ix0O3gm7xCi+6515wwQXMnDmTffv2AVBYWMimTZsYP358XV74qFGjOOKII3j11Vcbvb+wsJChQ4cCWpZ/8cUXM2zYMKZMmVJXbg+an+203r399tsBeOihh9i0aRMnnngiJ554IqAl+Vu2bAHg/vvvZ+jQoQwdOrSu9W5hYSFDhgzhqquu4vDDD2fixIkN9uPw+uuvc+SRRzJy5EhOOeUUSktLAc11v+KKKzjiiCMYNmxYXeuAt956i1GjRjF8+HBOPvlkQPvD33vvvXXbHDp0KIWFhXU2XHvttYwaNYoNGzaE/XwAn3/+OccccwzDhw9n3Lhx7Nq1i+OOO65BW+Bjjz2WJUuWRP+iXGCzXIJwBD3dUhYtlubQrVs3xo0bx1tvvcXZZ5/NCy+8wJQpUxAR2rdvz8svv0xeXh5btmzhqKOOYvLkyRHnu3zkkUfo0KEDS5YsYcmSJYwaNarutbvuuouuXbtSU1PDySefzJIlS7jhhhu4//77mTNnDt27d2+wrQULFvDkk0/y6aefYozhyCOP5IQTTqBLly6sWrWK559/nscff5yLLrqIf//731x66aUN3j9+/Hg++eQTRIQnnniCu+++m/vuu4/f/va3dOrUiS+//BKA7du3U1ZWxlVXXcUHH3zAgAEDGvRlicSKFSt48skn+etf/xrx8w0ePJgpU6bw4osvMnbsWHbu3El2djbTpk3jqaee4sEHH2TlypXs27ePYcOGxfS9hcMKehA25GLxGq+65zphF0fQnR7mxhhuvfVWPvjgA1q1asXGjRspLS0l3/F+Qvjggw+44YYbAO11EixSM2bM4LHHHqO6upri4mKWL18eVcQ++ugjzj333LrOh+eddx4ffvghkydPZsCAAYwYMQJo2H43mKKiIqZMmUJxcTH79+9nwIABgLbTDQ4xdenShddff53jjz++bh03LXYPOuggjjrqqKifT0To1atXXQvivEAp94UXXshvf/tb7rnnHv7xj39w+eWXN7k/N9iQSxA25GJpqZxzzjm8++67LFy4kMrKyjrPevr06ZSVlbFgwQIWLVpEz549w7bMDSac975u3Truvfde3n33XZYsWcIZZ5zR5HZMlFa6TutdiNyi9/rrr+e6667jyy+/5NFHH63bnwnT9jbcc9CwxS40bLMb3GI30ueLtN0OHTpw6qmn8uqrrzJjxgy+/e1vR/yssWAFPQgbcrG0VHJycpgwYQJXXnllg8FQp3VsmzZtmDNnDuvXr4+6neOPP75uIuilS5fWxYV37txJx44d6dSpE6Wlpbz55pt178nNzWXXrl1ht/XKK69QUVHBnj17ePnllznuuONcf6YdO3bQO9AT45///Gfd8xMnTuQvf/lL3ePt27dz9NFH87///Y9169YBDVvsLly4EICFCxfWvR5KpM83ePBgNm3axOeffw7Arl276v58pk2bxg033MDYsWNdXRG4wQp6EFbQLS2ZqVOnsnjxYi6++OK65y655BLmz5/PmDFjmD59epOTNVxzzTXs3r2bYcOGcffddzNu3DhAZx8aOXIkhx9+OFdeeWWD1rtXX301kyZNqhsUdRg1ahSXX34548aN48gjj2TatGmMHDnS9ee54447uPDCCznuuOMaxOdvu+02tm/fztChQxk+fDhz5szhgAMO4LHHHuO8885j+PDhdW1vzz//fLZt28aIESN45JFHODRCOXakz9e2bVtefPFFrr/+eoYPH86pp55a5+WPHj2avLy8hPZMl2iXNclkzJgxpqn81FRjDNx1F0yZYnuiWFLHV199xZAhQ7w2w5JiNm3axIQJE/j6669p1Sq8bx3u3BCRBcaYMeHWtx56ECJw221WzC0WS3J5+umnOfLII7nrrrsiink82CwXi8ViSTHf+c53+M53vpPw7VoP3WLxAV6FPi3+JZ5zwgq6xeIx7du3Z+vWrVbULXUYY9i6dSvt27eP6X025GKxeEyfPn0oKiqirKzMa1MsPqJ9+/b0iXEKMivoFovHtGnTpq5C0WJpDjbkYrFYLBmCFXSLxWLJEKygWywWS4bgWaWoiJQB0RtDRKY7sCWB5iQSa1t8WNviw9oWH+ls20HGmAPCveCZoDcHEZkfqfTVa6xt8WFtiw9rW3xkqm025GKxWCwZghV0i8ViyRDSVdAf89qAKFjb4sPaFh/WtvjISNvSMoZusVgslsakq4dusVgslhCsoFssFkuGkHaCLiKnicgKEVktIrd4bU8wIlIoIl+KyCIR8XQ6JhH5h4hsFpGlQc91FZHZIrIqsPRksr0Itt0hIhsDx26RiJzukW19RWSOiHwlIstE5EeB5z0/dlFs8/zYiUh7EflMRBYHbPtN4PkBIvJp4Li9KCJtfWTbUyKyLui4jUi1bUE2thaRL0RkZuBxfMfNGJM2N6A1sAYYCLQFFgMFXtsVZF8h0N1rOwK2HA+MApYGPXc3cEvg/i3An3xk2x3ATT44br2AUYH7ucBKoMAPxy6KbZ4fO0CAnMD9NsCnwFHADODiwPN/A67xkW1PARd4fc4F7PoJ8BwwM/A4ruOWbh76OGC1MWatMWY/8AJwtsc2+RJjzAfAtpCnzwac6c//CZyTUqMCRLDNFxhjio0xCwP3dwFfAb3xwbGLYpvnGGV34GGbwM0AJwH/Cjzv1XGLZJsvEJE+wBnAE4HHQpzHLd0EvTewIehxET45oQMY4G0RWSAiV3ttTBh6GmOKQcUB6OGxPaFcJyJLAiEZT8JBwYhIf2Ak6tH56tiF2AY+OHaBsMEiYDMwG72aLjfGVAdW8ez3GmqbMcY5bncFjtsDItLOC9uAB4GfA7WBx92I87ilm6BLmOd8808LHGuMGQVMAn4oIsd7bVAa8QhwMDACKAbu89IYEckB/g3caIzZ6aUtoYSxzRfHzhhTY4wZAfRBr6aHhFsttVYFdhpim4gMBX4BDAbGAl2Bm1Ntl4icCWw2xiwIfjrMqq6OW7oJehHQN+hxH2CTR7Y0whizKbDcDLyMntR+olREegEElps9tqcOY0xp4EdXCzyOh8dORNqggjndGPOfwNO+OHbhbPPTsQvYUw68j8apO4uIM5GO57/XINtOC4SwjDFmH/Ak3hy3Y4HJIlKIhpBPQj32uI5bugn658AhgRHgtsDFwGse2wSAiHQUkVznPjARWBr9XSnnNeC7gfvfBV710JYGOGIZ4Fw8OnaB+OXfga+MMfcHveT5sYtkmx+OnYgcICKdA/ezgVPQGP8c4ILAal4dt3C2fR30By1ojDrlx80Y8wtjTB9jTH9Uz94zxlxCvMfN69HdOEaDT0dH99cAv/TaniC7BqJZN4uBZV7bBjyPXn5XoVc230Njc+8CqwLLrj6y7RngS2AJKp69PLJtPHp5uwRYFLid7odjF8U2z48dMAz4ImDDUuDXgecHAp8Bq4GXgHY+su29wHFbCjxLIBPGqxswgfosl7iOmy39t1gslgwh3UIuFovFYomAFXSLxWLJEKygWywWS4ZgBd1isVgyBCvoFovFkiFYQbdYLJYMwQq6xWKxZAj/D0dxhFTpoys2AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "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": [
    "acc = history.history['accuracy']\n",
    "val_acc = history.history['val_accuracy']\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "epochs = range(len(acc))\n",
    "\n",
    "plt.plot(epochs, acc, 'r', label='Training accuracy')\n",
    "plt.plot(epochs, val_acc, 'b', label='Validation accuracy')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.legend()\n",
    "plt.figure()\n",
    "\n",
    "plt.plot(epochs, loss, 'r', label='Training Loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation Loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test accuracy: 98.4104871749878\n"
     ]
    }
   ],
   "source": [
    "print(\"test accuracy: \"+ str(model.evaluate_generator(testdata_generator)[1]*100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}