cat-vs-dogv1-ipynb.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.015012,
     "end_time": "2021-02-28T18:18:51.446384",
     "exception": false,
     "start_time": "2021-02-28T18:18:51.431372",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## importing libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-28T18:18:51.481246Z",
     "iopub.status.busy": "2021-02-28T18:18:51.480628Z",
     "iopub.status.idle": "2021-02-28T18:18:56.611851Z",
     "shell.execute_reply": "2021-02-28T18:18:56.610820Z"
    },
    "papermill": {
     "duration": 5.150326,
     "end_time": "2021-02-28T18:18:56.611986",
     "exception": false,
     "start_time": "2021-02-28T18:18:51.461660",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import keras\n",
    "from keras.preprocessing.image import load_img\n",
    "import matplotlib.pyplot as plt\n",
    "import random\n",
    "from sklearn.model_selection import train_test_split\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "import tensorflow as tf\n",
    "tf.config.run_functions_eagerly(True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.014586,
     "end_time": "2021-02-28T18:18:56.641284",
     "exception": false,
     "start_time": "2021-02-28T18:18:56.626698",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## extracting files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-28T18:18:56.674313Z",
     "iopub.status.busy": "2021-02-28T18:18:56.673787Z",
     "iopub.status.idle": "2021-02-28T18:19:14.812755Z",
     "shell.execute_reply": "2021-02-28T18:19:14.811763Z"
    },
    "papermill": {
     "duration": 18.157307,
     "end_time": "2021-02-28T18:19:14.812921",
     "exception": false,
     "start_time": "2021-02-28T18:18:56.655614",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 7.02 s, sys: 2.21 s, total: 9.23 s\n",
      "Wall time: 18.1 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "import zipfile\n",
    "with zipfile.ZipFile(\"../input/dogs-vs-cats/train.zip\",'r') as z:\n",
    "    z.extractall(\".\")\n",
    "with zipfile.ZipFile(\"../input/dogs-vs-cats/test1.zip\",'r') as z:\n",
    "    z.extractall(\".\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-28T18:19:14.858686Z",
     "iopub.status.busy": "2021-02-28T18:19:14.857877Z",
     "iopub.status.idle": "2021-02-28T18:19:14.905142Z",
     "shell.execute_reply": "2021-02-28T18:19:14.904698Z"
    },
    "papermill": {
     "duration": 0.076683,
     "end_time": "2021-02-28T18:19:14.905284",
     "exception": false,
     "start_time": "2021-02-28T18:19:14.828601",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "filenames = os.listdir('/kaggle/working/train')\n",
    "categories = []\n",
    "for filename in filenames:\n",
    "    category = filename.split('.')[0]\n",
    "    if category == 'dog':\n",
    "        categories.append(str(1))\n",
    "    else:\n",
    "        categories.append(str(0))\n",
    "\n",
    "df = pd.DataFrame({\n",
    "    'filename': filenames,\n",
    "    'category': categories\n",
    "})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.014726,
     "end_time": "2021-02-28T18:19:14.935045",
     "exception": false,
     "start_time": "2021-02-28T18:19:14.920319",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## splitting of dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-28T18:19:14.973034Z",
     "iopub.status.busy": "2021-02-28T18:19:14.972545Z",
     "iopub.status.idle": "2021-02-28T18:19:14.980918Z",
     "shell.execute_reply": "2021-02-28T18:19:14.980529Z"
    },
    "papermill": {
     "duration": 0.031241,
     "end_time": "2021-02-28T18:19:14.981023",
     "exception": false,
     "start_time": "2021-02-28T18:19:14.949782",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "train_data, valid_data = train_test_split(df, test_size=0.2, random_state = 42)\n",
    "train_data = train_data.reset_index(drop=True)\n",
    "valid_data = valid_data.reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.015048,
     "end_time": "2021-02-28T18:19:15.011071",
     "exception": false,
     "start_time": "2021-02-28T18:19:14.996023",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## generating training data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-28T18:19:15.044867Z",
     "iopub.status.busy": "2021-02-28T18:19:15.044370Z",
     "iopub.status.idle": "2021-02-28T18:19:15.048072Z",
     "shell.execute_reply": "2021-02-28T18:19:15.047682Z"
    },
    "papermill": {
     "duration": 0.022248,
     "end_time": "2021-02-28T18:19:15.048174",
     "exception": false,
     "start_time": "2021-02-28T18:19:15.025926",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "train_data_gen = ImageDataGenerator(\n",
    "    rotation_range=30,\n",
    "    rescale=1./255,\n",
    "    shear_range=0.1,\n",
    "    zoom_range=0.2,\n",
    "    horizontal_flip=True,\n",
    "    width_shift_range=0.2,\n",
    "    height_shift_range=0.2\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-28T18:19:15.114476Z",
     "iopub.status.busy": "2021-02-28T18:19:15.104210Z",
     "iopub.status.idle": "2021-02-28T18:19:15.419145Z",
     "shell.execute_reply": "2021-02-28T18:19:15.418551Z"
    },
    "papermill": {
     "duration": 0.355912,
     "end_time": "2021-02-28T18:19:15.419324",
     "exception": false,
     "start_time": "2021-02-28T18:19:15.063412",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 20000 validated image filenames belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "train_generator = train_data_gen.flow_from_dataframe(\n",
    "    train_data, \n",
    "    \"/kaggle/working/train/\", \n",
    "    x_col='filename',\n",
    "    y_col='category',\n",
    "    target_size=(224,224),\n",
    "    class_mode='categorical',\n",
    "    batch_size=20\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.01535,
     "end_time": "2021-02-28T18:19:15.451029",
     "exception": false,
     "start_time": "2021-02-28T18:19:15.435679",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## generating validation data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-28T18:19:15.501129Z",
     "iopub.status.busy": "2021-02-28T18:19:15.500398Z",
     "iopub.status.idle": "2021-02-28T18:19:15.539116Z",
     "shell.execute_reply": "2021-02-28T18:19:15.539695Z"
    },
    "papermill": {
     "duration": 0.073417,
     "end_time": "2021-02-28T18:19:15.539853",
     "exception": false,
     "start_time": "2021-02-28T18:19:15.466436",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 5000 validated image filenames belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "valid_data_gen = ImageDataGenerator(rescale=1./255)\n",
    "\n",
    "valid_generator = valid_data_gen.flow_from_dataframe(valid_data, \n",
    "    \"/kaggle/working/train/\", \n",
    "    x_col='filename',\n",
    "    y_col='category',\n",
    "    target_size=(224,224),\n",
    "    class_mode='categorical',\n",
    "    batch_size=20\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.015715,
     "end_time": "2021-02-28T18:19:15.571819",
     "exception": false,
     "start_time": "2021-02-28T18:19:15.556104",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# *CNN Model*(ResNet)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-28T18:19:15.619531Z",
     "iopub.status.busy": "2021-02-28T18:19:15.616186Z",
     "iopub.status.idle": "2021-02-28T18:19:19.336108Z",
     "shell.execute_reply": "2021-02-28T18:19:19.328761Z"
    },
    "papermill": {
     "duration": 3.752781,
     "end_time": "2021-02-28T18:19:19.340480",
     "exception": false,
     "start_time": "2021-02-28T18:19:15.587699",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "class ResidualUnit(keras.layers.Layer):\n",
    "    def __init__(self, filters, strides=1, activation_func=\"relu\", **kwargs):\n",
    "        super().__init__(**kwargs)\n",
    "        self.filters = filters\n",
    "        self.strides = strides\n",
    "        self.activation_func = activation_func\n",
    "        self.activation = keras.activations.get(self.activation_func)\n",
    "        self.main_layers = [\n",
    "                            keras.layers.Conv2D(self.filters, 3, strides=self.strides, padding=\"same\", use_bias=False),\n",
    "                            keras.layers.BatchNormalization(),\n",
    "                            self.activation,\n",
    "                            keras.layers.Conv2D(self.filters, 3, strides=1, padding=\"same\", use_bias=False),\n",
    "                            keras.layers.BatchNormalization()\n",
    "        ]\n",
    "        self.skip_layers = []\n",
    "        if self.strides>1:\n",
    "            self.skip_layers = [\n",
    "                            keras.layers.Conv2D(self.filters, 1, strides=self.strides, padding=\"same\", use_bias=False),\n",
    "                            keras.layers.BatchNormalization() \n",
    "            ]\n",
    "            \n",
    "    def call(self, inputs):\n",
    "        Z = inputs\n",
    "        for layer in self.main_layers:\n",
    "            Z = layer(Z)\n",
    "        skip_Z = inputs\n",
    "        for layer in self.skip_layers:\n",
    "            skip_Z = layer(skip_Z)\n",
    "        return self.activation(Z+skip_Z)\n",
    "    \n",
    "    def get_config(self):\n",
    "\n",
    "        config = super().get_config().copy()\n",
    "        config.update({\n",
    "            'filters': self.filters,\n",
    "            'strides': self.strides,\n",
    "            'activation_func': self.activation_func\n",
    "        })\n",
    "        return config\n",
    "\n",
    "model1 = keras.models.Sequential()\n",
    "model1.add(keras.layers.Conv2D(filters=64, kernel_size=7, strides=2, padding='same', use_bias=False, input_shape=(224,224,3)))\n",
    "model1.add(keras.layers.BatchNormalization())\n",
    "model1.add(keras.layers.Activation(\"relu\"))\n",
    "model1.add(keras.layers.MaxPooling2D(pool_size=3, strides=1, padding=\"same\"))\n",
    "prev_filters = 64\n",
    "for filters in [64]*3+[128]*4+[256]*6+[512]*3:\n",
    "    strides = 1 if filters == prev_filters else 2\n",
    "    model1.add(ResidualUnit(filters, strides=strides))\n",
    "    prev_filters=filters\n",
    "model1.add(keras.layers.GlobalAvgPool2D())\n",
    "model1.add(keras.layers.Flatten())\n",
    "model1.add(keras.layers.Dense(units=256, activation='relu'))\n",
    "model1.add(keras.layers.Dropout(0.25))                        \n",
    "model1.add(keras.layers.Dense(units=128, activation='relu'))\n",
    "model1.add(keras.layers.Dropout(0.15))\n",
    "model1.add(keras.layers.Dense(units=64, activation='relu'))\n",
    "model1.add(keras.layers.Dropout(0.10)) \n",
    "model1.add(keras.layers.Dense(units=32, activation='relu'))\n",
    "model1.add(keras.layers.Dropout(0.5)) \n",
    "model1.add(keras.layers.Dense(units=2, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-28T18:19:19.504578Z",
     "iopub.status.busy": "2021-02-28T18:19:19.503720Z",
     "iopub.status.idle": "2021-02-28T18:19:19.545342Z",
     "shell.execute_reply": "2021-02-28T18:19:19.546529Z"
    },
    "papermill": {
     "duration": 0.104305,
     "end_time": "2021-02-28T18:19:19.546774",
     "exception": false,
     "start_time": "2021-02-28T18:19:19.442469",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d (Conv2D)              (None, 112, 112, 64)      9408      \n",
      "_________________________________________________________________\n",
      "batch_normalization (BatchNo (None, 112, 112, 64)      256       \n",
      "_________________________________________________________________\n",
      "activation (Activation)      (None, 112, 112, 64)      0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 112, 112, 64)      0         \n",
      "_________________________________________________________________\n",
      "residual_unit (ResidualUnit) (None, 112, 112, 64)      74240     \n",
      "_________________________________________________________________\n",
      "residual_unit_1 (ResidualUni (None, 112, 112, 64)      74240     \n",
      "_________________________________________________________________\n",
      "residual_unit_2 (ResidualUni (None, 112, 112, 64)      74240     \n",
      "_________________________________________________________________\n",
      "residual_unit_3 (ResidualUni (None, 56, 56, 128)       230912    \n",
      "_________________________________________________________________\n",
      "residual_unit_4 (ResidualUni (None, 56, 56, 128)       295936    \n",
      "_________________________________________________________________\n",
      "residual_unit_5 (ResidualUni (None, 56, 56, 128)       295936    \n",
      "_________________________________________________________________\n",
      "residual_unit_6 (ResidualUni (None, 56, 56, 128)       295936    \n",
      "_________________________________________________________________\n",
      "residual_unit_7 (ResidualUni (None, 28, 28, 256)       920576    \n",
      "_________________________________________________________________\n",
      "residual_unit_8 (ResidualUni (None, 28, 28, 256)       1181696   \n",
      "_________________________________________________________________\n",
      "residual_unit_9 (ResidualUni (None, 28, 28, 256)       1181696   \n",
      "_________________________________________________________________\n",
      "residual_unit_10 (ResidualUn (None, 28, 28, 256)       1181696   \n",
      "_________________________________________________________________\n",
      "residual_unit_11 (ResidualUn (None, 28, 28, 256)       1181696   \n",
      "_________________________________________________________________\n",
      "residual_unit_12 (ResidualUn (None, 28, 28, 256)       1181696   \n",
      "_________________________________________________________________\n",
      "residual_unit_13 (ResidualUn (None, 14, 14, 512)       3676160   \n",
      "_________________________________________________________________\n",
      "residual_unit_14 (ResidualUn (None, 14, 14, 512)       4722688   \n",
      "_________________________________________________________________\n",
      "residual_unit_15 (ResidualUn (None, 14, 14, 512)       4722688   \n",
      "_________________________________________________________________\n",
      "global_average_pooling2d (Gl (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 256)               131328    \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 64)                8256      \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 32)                2080      \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 32)                0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 2)                 66        \n",
      "=================================================================\n",
      "Total params: 21,476,322\n",
      "Trainable params: 21,459,298\n",
      "Non-trainable params: 17,024\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model1.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-28T18:19:19.622833Z",
     "iopub.status.busy": "2021-02-28T18:19:19.622053Z",
     "iopub.status.idle": "2021-02-28T18:19:19.645843Z",
     "shell.execute_reply": "2021-02-28T18:19:19.644762Z"
    },
    "papermill": {
     "duration": 0.066386,
     "end_time": "2021-02-28T18:19:19.645998",
     "exception": false,
     "start_time": "2021-02-28T18:19:19.579612",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "model1.compile(optimizer='adam', loss= 'categorical_crossentropy', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.026195,
     "end_time": "2021-02-28T18:19:19.698125",
     "exception": false,
     "start_time": "2021-02-28T18:19:19.671930",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## training model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-28T18:19:19.758814Z",
     "iopub.status.busy": "2021-02-28T18:19:19.758067Z",
     "iopub.status.idle": "2021-02-28T23:34:23.375415Z",
     "shell.execute_reply": "2021-02-28T23:34:23.374914Z"
    },
    "papermill": {
     "duration": 18903.650846,
     "end_time": "2021-02-28T23:34:23.375561",
     "exception": false,
     "start_time": "2021-02-28T18:19:19.724715",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.7/site-packages/tensorflow/python/data/ops/dataset_ops.py:3504: UserWarning: Even though the tf.config.experimental_run_functions_eagerly option is set, this option does not apply to tf.data functions. tf.data functions are still traced and executed as graphs.\n",
      "  \"Even though the tf.config.experimental_run_functions_eagerly \"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/150\n",
      "1000/1000 [==============================] - 385s 381ms/step - loss: 0.7224 - accuracy: 0.5195 - val_loss: 0.6946 - val_accuracy: 0.4904\n",
      "Epoch 2/150\n",
      "1000/1000 [==============================] - 384s 384ms/step - loss: 0.6915 - accuracy: 0.5285 - val_loss: 0.6909 - val_accuracy: 0.5414\n",
      "Epoch 3/150\n",
      "1000/1000 [==============================] - 381s 381ms/step - loss: 0.6864 - accuracy: 0.5429 - val_loss: 1.0559 - val_accuracy: 0.5344\n",
      "Epoch 4/150\n",
      "1000/1000 [==============================] - 387s 387ms/step - loss: 0.6715 - accuracy: 0.5709 - val_loss: 0.6702 - val_accuracy: 0.5690\n",
      "Epoch 5/150\n",
      "1000/1000 [==============================] - 389s 389ms/step - loss: 0.6577 - accuracy: 0.6144 - val_loss: 0.6752 - val_accuracy: 0.5640\n",
      "Epoch 6/150\n",
      "1000/1000 [==============================] - 388s 387ms/step - loss: 0.6467 - accuracy: 0.6193 - val_loss: 0.6673 - val_accuracy: 0.5988\n",
      "Epoch 7/150\n",
      "1000/1000 [==============================] - 401s 401ms/step - loss: 0.6525 - accuracy: 0.6168 - val_loss: 0.6427 - val_accuracy: 0.6184\n",
      "Epoch 8/150\n",
      "1000/1000 [==============================] - 408s 408ms/step - loss: 0.6393 - accuracy: 0.6363 - val_loss: 0.6881 - val_accuracy: 0.5602\n",
      "Epoch 9/150\n",
      "1000/1000 [==============================] - 412s 412ms/step - loss: 0.6316 - accuracy: 0.6449 - val_loss: 0.6395 - val_accuracy: 0.6288\n",
      "Epoch 10/150\n",
      "1000/1000 [==============================] - 416s 416ms/step - loss: 0.6183 - accuracy: 0.6730 - val_loss: 0.5919 - val_accuracy: 0.6742\n",
      "Epoch 11/150\n",
      "1000/1000 [==============================] - 419s 419ms/step - loss: 0.5984 - accuracy: 0.6910 - val_loss: 0.7105 - val_accuracy: 0.6688\n",
      "Epoch 12/150\n",
      "1000/1000 [==============================] - 422s 421ms/step - loss: 0.5872 - accuracy: 0.6990 - val_loss: 0.6364 - val_accuracy: 0.6576\n",
      "Epoch 13/150\n",
      "1000/1000 [==============================] - 422s 422ms/step - loss: 0.5541 - accuracy: 0.7274 - val_loss: 0.4981 - val_accuracy: 0.7624\n",
      "Epoch 14/150\n",
      "1000/1000 [==============================] - 412s 412ms/step - loss: 0.5375 - accuracy: 0.7382 - val_loss: 0.5297 - val_accuracy: 0.7490\n",
      "Epoch 15/150\n",
      "1000/1000 [==============================] - 414s 414ms/step - loss: 0.5198 - accuracy: 0.7523 - val_loss: 0.4754 - val_accuracy: 0.7688\n",
      "Epoch 16/150\n",
      "1000/1000 [==============================] - 415s 415ms/step - loss: 0.5007 - accuracy: 0.7670 - val_loss: 0.4433 - val_accuracy: 0.7996\n",
      "Epoch 17/150\n",
      "1000/1000 [==============================] - 421s 421ms/step - loss: 0.4906 - accuracy: 0.7768 - val_loss: 0.5509 - val_accuracy: 0.7400\n",
      "Epoch 18/150\n",
      "1000/1000 [==============================] - 419s 419ms/step - loss: 0.4716 - accuracy: 0.7880 - val_loss: 0.6658 - val_accuracy: 0.6536\n",
      "Epoch 19/150\n",
      "1000/1000 [==============================] - 419s 419ms/step - loss: 0.4366 - accuracy: 0.8052 - val_loss: 0.3622 - val_accuracy: 0.8468\n",
      "Epoch 20/150\n",
      "1000/1000 [==============================] - 421s 421ms/step - loss: 0.4144 - accuracy: 0.8197 - val_loss: 0.4664 - val_accuracy: 0.8124\n",
      "Epoch 21/150\n",
      "1000/1000 [==============================] - 429s 429ms/step - loss: 0.3888 - accuracy: 0.8346 - val_loss: 0.3646 - val_accuracy: 0.8470\n",
      "Epoch 22/150\n",
      "1000/1000 [==============================] - 426s 426ms/step - loss: 0.3574 - accuracy: 0.8504 - val_loss: 0.3899 - val_accuracy: 0.8022\n",
      "Epoch 23/150\n",
      "1000/1000 [==============================] - 423s 423ms/step - loss: 0.3312 - accuracy: 0.8621 - val_loss: 0.3720 - val_accuracy: 0.8424\n",
      "Epoch 24/150\n",
      "1000/1000 [==============================] - 421s 420ms/step - loss: 0.3297 - accuracy: 0.8606 - val_loss: 0.2634 - val_accuracy: 0.8998\n",
      "Epoch 25/150\n",
      "1000/1000 [==============================] - 429s 429ms/step - loss: 0.3161 - accuracy: 0.8713 - val_loss: 0.3127 - val_accuracy: 0.8530\n",
      "Epoch 26/150\n",
      "1000/1000 [==============================] - 433s 432ms/step - loss: 0.2897 - accuracy: 0.8801 - val_loss: 0.3945 - val_accuracy: 0.8540\n",
      "Epoch 27/150\n",
      "1000/1000 [==============================] - 413s 413ms/step - loss: 0.2758 - accuracy: 0.8934 - val_loss: 0.4168 - val_accuracy: 0.8190\n",
      "Epoch 28/150\n",
      "1000/1000 [==============================] - 408s 408ms/step - loss: 0.2602 - accuracy: 0.8964 - val_loss: 0.2387 - val_accuracy: 0.9096\n",
      "Epoch 29/150\n",
      "1000/1000 [==============================] - 402s 402ms/step - loss: 0.2470 - accuracy: 0.9004 - val_loss: 0.2183 - val_accuracy: 0.9104\n",
      "Epoch 30/150\n",
      "1000/1000 [==============================] - 404s 404ms/step - loss: 0.2344 - accuracy: 0.9075 - val_loss: 0.2715 - val_accuracy: 0.8786\n",
      "Epoch 31/150\n",
      "1000/1000 [==============================] - 413s 413ms/step - loss: 0.2247 - accuracy: 0.9092 - val_loss: 0.4332 - val_accuracy: 0.8284\n",
      "Epoch 32/150\n",
      "1000/1000 [==============================] - 416s 416ms/step - loss: 0.2242 - accuracy: 0.9070 - val_loss: 0.1832 - val_accuracy: 0.9328\n",
      "Epoch 33/150\n",
      "1000/1000 [==============================] - 421s 421ms/step - loss: 0.2104 - accuracy: 0.9188 - val_loss: 0.2070 - val_accuracy: 0.9150\n",
      "Epoch 34/150\n",
      "1000/1000 [==============================] - 421s 421ms/step - loss: 0.2045 - accuracy: 0.9205 - val_loss: 0.2346 - val_accuracy: 0.9130\n",
      "Epoch 35/150\n",
      "1000/1000 [==============================] - 418s 418ms/step - loss: 0.1976 - accuracy: 0.9232 - val_loss: 0.8239 - val_accuracy: 0.8030\n",
      "Epoch 36/150\n",
      "1000/1000 [==============================] - 405s 405ms/step - loss: 0.1872 - accuracy: 0.9260 - val_loss: 0.1356 - val_accuracy: 0.9454\n",
      "Epoch 37/150\n",
      "1000/1000 [==============================] - 408s 408ms/step - loss: 0.1781 - accuracy: 0.9290 - val_loss: 0.1463 - val_accuracy: 0.9460\n",
      "Epoch 38/150\n",
      "1000/1000 [==============================] - 411s 411ms/step - loss: 0.1742 - accuracy: 0.9327 - val_loss: 0.1388 - val_accuracy: 0.9492\n",
      "Epoch 39/150\n",
      "1000/1000 [==============================] - 417s 417ms/step - loss: 0.1622 - accuracy: 0.9346 - val_loss: 0.3465 - val_accuracy: 0.8622\n",
      "Epoch 40/150\n",
      "1000/1000 [==============================] - 415s 415ms/step - loss: 0.1673 - accuracy: 0.9353 - val_loss: 0.2378 - val_accuracy: 0.9026\n",
      "Epoch 41/150\n",
      "1000/1000 [==============================] - 405s 405ms/step - loss: 0.1640 - accuracy: 0.9363 - val_loss: 0.1602 - val_accuracy: 0.9418\n",
      "Epoch 42/150\n",
      "1000/1000 [==============================] - 403s 403ms/step - loss: 0.1586 - accuracy: 0.9410 - val_loss: 0.1619 - val_accuracy: 0.9376\n",
      "Epoch 43/150\n",
      "1000/1000 [==============================] - 404s 404ms/step - loss: 0.1456 - accuracy: 0.9400 - val_loss: 0.1534 - val_accuracy: 0.9368\n",
      "Epoch 44/150\n",
      "1000/1000 [==============================] - 398s 398ms/step - loss: 0.1472 - accuracy: 0.9406 - val_loss: 0.3467 - val_accuracy: 0.9494\n",
      "Epoch 45/150\n",
      "1000/1000 [==============================] - 404s 403ms/step - loss: 0.1371 - accuracy: 0.9441 - val_loss: 0.2440 - val_accuracy: 0.9404\n",
      "Epoch 46/150\n",
      "1000/1000 [==============================] - 403s 403ms/step - loss: 0.1377 - accuracy: 0.9449 - val_loss: 0.1851 - val_accuracy: 0.9184\n"
     ]
    }
   ],
   "source": [
    "checkpoint_cb = keras.callbacks.ModelCheckpoint(\"test_model_resnet.h5\", save_best_only=True)\n",
    "early_stopping_cb = keras.callbacks.EarlyStopping(patience=10, restore_best_weights=True)\n",
    "history = model1.fit(train_generator, epochs=150, validation_data=valid_generator, callbacks=[checkpoint_cb, early_stopping_cb])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-28T23:34:47.854473Z",
     "iopub.status.busy": "2021-02-28T23:34:47.853613Z",
     "iopub.status.idle": "2021-02-28T23:34:49.020103Z",
     "shell.execute_reply": "2021-02-28T23:34:49.018405Z"
    },
    "papermill": {
     "duration": 13.59594,
     "end_time": "2021-02-28T23:34:49.020267",
     "exception": false,
     "start_time": "2021-02-28T23:34:35.424327",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "model = keras.models.load_model(\"test_model_resnet.h5\", custom_objects={'ResidualUnit': ResidualUnit})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-28T23:35:13.348485Z",
     "iopub.status.busy": "2021-02-28T23:35:13.346597Z",
     "iopub.status.idle": "2021-02-28T23:35:13.571200Z",
     "shell.execute_reply": "2021-02-28T23:35:13.570770Z"
    },
    "papermill": {
     "duration": 12.538733,
     "end_time": "2021-02-28T23:35:13.571352",
     "exception": false,
     "start_time": "2021-02-28T23:35:01.032619",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Training and validation loss')"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAEICAYAAABPgw/pAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/Il7ecAAAACXBIWXMAAAsTAAALEwEAmpwYAABCpElEQVR4nO3dd3hUVfrA8e+bSW8kJAECSagBDCIISBFEQKQoir2tfdW1r6u7q24v7q7u2n6ra1t7r6wdRURARJTeew81CS2F9PP748yQIZkkk2SSSWbez/PwzMydO/eeXODNmfee8x4xxqCUUqrtC/F3A5RSSvmGBnSllAoQGtCVUipAaEBXSqkAoQFdKaUChAZ0pZQKEBrQlUciMl1ErvH1vv4kIttEZHwzHNeISC/n82dE5Pfe7NuI8/xERGY0tp11HHeMiGT7+riq5YX6uwHKd0SkwO1lNFACVDhf/8wY84a3xzLGTG6OfQOdMeZmXxxHRLoBW4EwY0y589hvAF7/HargowE9gBhjYl3PRWQbcIMxZmb1/UQk1BUklFKBQ1MuQcD1lVpE7hWRvcBLIpIoIp+KSI6IHHQ+T3P7zGwRucH5/FoRmSciDzv33Soikxu5b3cRmSsi+SIyU0T+IyKv19Jub9r4VxH5znm8GSKS7Pb+VSKyXUTyROS3dVyf4SKyV0QcbtvOF5EVzudDReR7ETkkIntE5EkRCa/lWC+LyANur3/l/MxuEbm+2r5ni8hSETkiIjtF5E9ub891Ph4SkQIRGeG6tm6fP1VEForIYefjqd5em7qIyAnOzx8SkdUicq7be2eJyBrnMXeJyC+d25Odfz+HROSAiHwrIhpfWphe8ODRCWgPdAVuwv7dv+R8nQEcBZ6s4/PDgPVAMvBP4AURkUbs+ybwI5AE/Am4qo5zetPGK4DrgA5AOOAKMFnA087jd3aeLw0PjDELgEJgXLXjvul8XgH8wvnzjADOAG6to9042zDJ2Z4zgUygev6+ELgaSADOBm4RkfOc7412PiYYY2KNMd9XO3Z74DPg386f7VHgMxFJqvYz1Lg29bQ5DPgEmOH83B3AGyLSx7nLC9j0XRxwIjDLuf0eIBtIAToCvwG0rkgL04AePCqBPxpjSowxR40xecaYD4wxRcaYfOBvwOl1fH67Mea/xpgK4BUgFfsf1+t9RSQDOAX4gzGm1BgzD/i4thN62caXjDEbjDFHgXeBgc7tFwGfGmPmGmNKgN87r0Ft3gIuBxCROOAs5zaMMYuNMQuMMeXGmG3Asx7a4cklzvatMsYUYn+Buf98s40xK40xlcaYFc7zeXNcsL8ANhpjXnO26y1gHXCO2z61XZu6DAdigQedf0ezgE9xXhugDMgSkXhjzEFjzBK37alAV2NMmTHmW6OFolqcBvTgkWOMKXa9EJFoEXnWmZI4gv2Kn+Cedqhmr+uJMabI+TS2gft2Bg64bQPYWVuDvWzjXrfnRW5t6ux+bGdAzavtXNje+AUiEgFcACwxxmx3tqO3M52w19mOv2N76/U5rg3A9mo/3zAR+caZUjoM3OzlcV3H3l5t23agi9vr2q5NvW02xrj/8nM/7oXYX3bbRWSOiIxwbv8XsAmYISJbROQ+734M5Usa0INH9d7SPUAfYJgxJp6qr/i1pVF8YQ/QXkSi3bal17F/U9q4x/3YznMm1bazMWYNNnBN5vh0C9jUzTog09mO3zSmDdi0kbs3sd9Q0o0x7YBn3I5bX+92NzYV5S4D2OVFu+o7bnq1/Pex4xpjFhpjpmLTMR9ie/4YY/KNMfcYY3pgvyXcLSJnNLEtqoE0oAevOGxO+pAzH/vH5j6hs8e7CPiTiIQ7e3fn1PGRprTxfWCKiIxy3sD8C/X/e38TuBP7i+O9au04AhSISF/gFi/b8C5wrYhkOX+hVG9/HPYbS7GIDMX+InHJwaaIetRy7M+B3iJyhYiEisilQBY2PdIUP2Bz+78WkTARGYP9O3rb+Xf2ExFpZ4wpw16TCgARmSIivZz3SlzbKzyeQTUbDejB63EgCsgFFgBftNB5f4K9sZgHPAC8gx0v78njNLKNxpjVwG3YIL0HOIi9aVeXt4AxwCxjTK7b9l9ig20+8F9nm71pw3TnzzALm46YVW2XW4G/iEg+8AecvV3nZ4uw9wy+c44cGV7t2HnAFOy3mDzg18CUau1uMGNMKXAu9ptKLvAUcLUxZp1zl6uAbc7U083Alc7tmcBMoAD4HnjKGDO7KW1RDSd630L5k4i8A6wzxjT7NwSlAp320FWLEpFTRKSniIQ4h/VNxeZilVJNpDNFVUvrBEzD3qDMBm4xxiz1b5OUCgyaclFKqQChKRellAoQfku5JCcnm27duvnr9Eop1SYtXrw41xiT4uk9vwX0bt26sWjRIn+dXiml2iQRqT5D+BhNuSilVIDQgK6UUgFCA7pSSgUIDehKKRUgNKArpVSA0ICulFIBQgO6UkoFiOAJ6DsWwN6V/m6FUko1m+AJ6J/+Amb+2d+tUEqpZhM81RYL9oEjzN+tUEqpZhMcAb2iHIoOQFh0/fsqpVQbFRwpl6MHAAOFTVqdSymlWrXgCOiFOfax/CiUFvq3LUop1UyCK6CD9tKVUgErSAK6WxAv0oCulApMQRLQ3Xvoef5rh1JKNaPgCOgF+6ueaw9dKRWg6g3oIvKiiOwXkVW1vC8i8m8R2SQiK0RkkO+b2USFORAe63yuAV0pFZi86aG/DEyq4/3JQKbzz03A001vlo8V5kJiN3CEaw9dKRWw6g3oxpi5wIE6dpkKvGqsBUCCiKT6qoE+UZgDMSkQnaw5dKVUwPJFDr0LsNPtdbZzWw0icpOILBKRRTk5OZ52aR6ugB6TpD10pVTA8kVAFw/bjKcdjTHPGWOGGGOGpKSk+ODUXirMhdgOzh66BnSlVGDyRUDPBtLdXqcBu31wXN8oLYSyQohJtn+0h66UClC+COgfA1c7R7sMBw4bY/b44Li+4eqRaw5dKRXg6q22KCJvAWOAZBHJBv4IhAEYY54BPgfOAjYBRcB1zdXYRnFNKnLl0EvzobwEQiP82y6llPKxegO6Mebyet43wG0+a5GvHQvoybaHDlCUB/Gd/dcmpZRqBoE/U/S4HrozoOuNUaVUAAqegB7t3kPXgK6UCjxBENBz7bT/8GiITnJu0xujSqnAEwQBPacq1RKjPXSlVOAKkoDewT6PTABxaA5dKRWQgiCg59obogAhIRDdXnvoSqmAFAQB3S3lAjr9XykVsAI7oFdWHt9DB+f0f70pqpQKPIEd0I8eBFNxfECPTtIeulIqIAV2QHefJeqiBbqUUgEqSAK6ew892fbcK8r90yallGomwRfQXb31o3UtwqSUUm1PgAd0t9K5Lsdmi2raRSkVWAI8oOeAOMeeu8S4VVxUSqkAEvgBPToJQhxV27RAl1IqQAV+QI+ptnapplyUUgEqCAJ68vHbXOkXTbkopQJMEAT0aj10R5gt0qU9dKVUgAnwgJ5bM6CDTi5SSgWkwA3oZcVQcqRmygW0QJdSKiAFbkAv8jAG3UULdCmlAlDgBnRPs0RdtECXUioABXBAd/XQO9R8z9VDr6xs2TYppVQzCuCA7qHSokt0si2rW3yoRZuklFLNKXADesF++1hbDh00j66UCiiBG9ALcyA0CsJjar6ns0WVUgEogAO6cwy6SM33YrSei1Iq8ARwQPcw7d/FVaBLe+hKqQAS4AHdQ/4cqlIumkNXSgWQAA7otUz7BwiLhPBYDehKqYASmAHdmLpTLqCTi5RSAScwA3rxYagsg1gPk4pctECXUr5zaAd8fCdUlPm7JUEtMAN6XdP+XbRAl1K+s+lrWPIKHNjq75YEtQAP6HWkXLRAl1K+U5JvH0sL/NuOIOdVQBeRSSKyXkQ2ich9Ht5vJyKfiMhyEVktItf5vqkN4FUP3ZlDN6Zl2qRUIHMF8rIi/7YjyNUb0EXEAfwHmAxkAZeLSFa13W4D1hhjBgBjgEdEJNzHbfWeNwE9JhkqSrRHoZQvHOuhF/q3HUHOmx76UGCTMWaLMaYUeBuYWm0fA8SJiACxwAGg3KctbQhXbtw13twTnVyklO+UHLGP2kHyK28Cehdgp9vrbOc2d08CJwC7gZXAz40xNWrTishNIrJIRBbl5OQ0ssleKMyBqES7fmhttECXUr5T4gzkpZpy8SdvArqHYihUTzxPBJYBnYGBwJMiEl/jQ8Y8Z4wZYowZkpJSRzqkqeqaJeqiPXSlfEdTLq2CNwE9G0h3e52G7Ym7uw6YZqxNwFagr2+a2Ah1zRJ1iXFN/9eArlST6SiXVsGbgL4QyBSR7s4bnZcBH1fbZwdwBoCIdAT6AFt82dAG8aqHriV0lfIZVyDXHrpf1RvQjTHlwO3Al8Ba4F1jzGoRuVlEbnbu9lfgVBFZCXwN3GuM8V+kLNhff0APjwVHhPbQlfIFVw9dhy36Vag3OxljPgc+r7btGbfnu4EJvm1aI5WX2qXl6gvoIs7JRQdapFlKBTQd5dIqBN5MUdeolbpmibpogS6lms4YvSnaSgReQPdmUpGLFuhSqunKjoJrlLIOW/Sr4A7oWqBLqaZz9c5Be+h+FoAB3Rmgve6h68QipZrkuICuOXR/CsCA7kWlRZfoJPsPsKy4edukVCArdQb00EjtoftZYAZ0RzhEtqt/32PT/zXtolSjuXrosR112KKfBWZAj0mxwxLro9P/lWo6V0CPS9WUi58FaED3It0C2kNXyhdchbniOmrKxc8CNKB7WfjrWA9db4wq1WiuSUVxqVBZbif3Kb8IwIDuRWEuFy3QpVTTuefQQdMufhRYAd2YhqVcItqBODSHrlRTlBZASGhVwTtNu/hNYAX00gIoL/a+hx4SYv8Rag9dqcYrybfF7iJi7WsN6H4TWAG9IbNEXWKS23YO/cAW2Lfa361QwawkHyLiISzGvi7TgO4vARbQGzBL1KWt99A/vA1enQrlJf5uiQpWJfkQEQfhzoCuPXS/aXMBvbisggOFtdxFb2wPva1O/y8+Ajt/sD/36g/93RoVrErybbpFA7rftbmA/u26PZzywJdc/Mx8np2zmS05bnfUC/bbxwb10Ntwga7t88FUQGgU/Pisv1ujgtWxHrrm0P3NqwUuWpMBxT+wIfIm9u9PYvvu9iydmcT8qFSSOvfklJB1JIP3o1xc+xYfgooycIQ1U6ubyZbZtn7G2N/AV7+H7MWQNtjfrVLBpiQfErtCeLR9rQHdb9pcQO+Q0RdG3kHq4Wza5+3gxANbiCqej2ObrcecZ9rxs/8upm9qHH07xXNCahy9O8YRF1lLsHYNtSo6YGe6NSdjYNEL0O00SOnT9ONtmQ0ZI2DwtTDnIdtLT3uu6cdVqiFKC2zvXFMuftfmAjod+9k/QITzD5UV5Odms3zVKhbvh5BDwkfLdvN68Y5jH0tvH8W4Ph24bWwvOsRHVh3Pffp/cwf0pa/BZ/dAUi+4+TsIi6z/M7XJ3ws5a2HAZRAZDwOvgEUvwYQHILaD79qsVH2qj3LRgO43bS+gexLiIK5DV0aN68oo4OeAMYbdh4tZt+cI6/bmszL7MG/8sIN3Fu3kmlO7cfPoniTGhLdcga4DW2D6fZCUCXkbbY96/B8bf7wtc+xjjzH2cehN8ONzsPhlOP3XTW2tUt6prLQ99Ig4CA2HkDAdtuhHgRHQPRARuiRE0SUhijNOsD3v7XmFPD5zI8/N3cKbC3Zw4+ge3NAnkWiwPfSjhyBvE+RucP7ZCPl7IHUAdBsFXUc1rhdfWQH/u9nOprvqfzD7Qfju/6DfefbYjbFlNkS1h04n2dfJmdBzHCx6EUb9ou3dD1Btk2uav2tSUXiM9tD9KGADuiddk2J47NKB3Hx6Tx79aj2PfrWBj747yteA+fA2pPxo1c4hodC+p01frHjPBkqA5N42uHc7DbqfXlUPpi7fPW6HF17wX0hIh4kPwKav4KPb4MZvGh58jbEBvcfpdrary9CfwVuXwtqP4cQLG3ZMpRrDVcclIs4+hsdqQPejoAroLn06xfHsVUNYvvMQD3+5jte3nUFkZTkV7TPp2KM//QecQlJa76pAW1EOe5fDtnn2jyvAh0XDOf+Gky6u/WR7lsM3/4B+50N/535RiXDWw/DuVTD/33DaPQ37AXI3Qv7uqnSLS+aZkNgNfniubQT0nA3w1R/goherRkiotqVGQI/R4lx+FJQB3WVAegKv3TCcpTteYPqqvXy5ei/b5xfB/G2cnHGICVmdGH9CB3qkxOLoMhi6DIaRP7cBfs9ymPFbmHYDZC+0NyNDw48/QVkxTLvJjqQ5+9HjF93IOhdOOBdmPwR9z4GU3t43fMts+1g9oIc44JQbbbv2LG98OqelLHsdNkyH/Wt1uGVbdSzlEm8fw6OhVFct8pegDuguJ2ckcnJGIvdP7svG/QXMWL2XL1fv46Ev1vHQF+sIcwhpidF0TYqma/touibF0DUpjbSJb5Kx5EGifnwW9iyDi1+B+NSqA3/9F8hZB1d+ANHta574rIdh6xz4+A64bvrx6ZO6bJlte+KJ3Tz8MFfCN3+zN0in/qfhF6MlbZplH/N3AxrQ2yRXLXTXpCJNufiVBnQ3IkLvjnbc+u3jMtl96CjzNuayNa+QHXlFbMsrZNG2gxSUlLt96nSmOML5587nKH5sOP+XeD97EoYwMXodF676j+0x9xrv+YRxHWHiP+CjW+349KE31t/IinLY9i2ceIHn96MS4KRLYflbcOZfPf8iaQ3y98G+lc7ne/3bFtV4nlIuR3b7rz1BTgN6HTonRHHJKenHbTPGcKCwlO0Hith18CgHCkvJK8zkhdyhXLLlN/zhwP28UngFI4o/Z7NJ5Y87zuachTs4q3+q58lNA6+Ale/BzD9B74mQkFF3o3Yvtb2i6ukWd0NvgsUvwZJX7IiX1siVNgINAG1Z9YAeFq0LRfuRBvQGEhGSYiNIio1gUEai2zu9oXgMfHgL1697HRPi4P2BL7F7k3DvByv548ermdivExcOSmNkr2QcIeI6IJzzf/DUCPj0F/CT9+te4HrLbECg2+ja9+mYZUfhLHwBRtwBjlb417x5lr234IjQHnpb5lpP9Liboppy8ZdW+D+9DYuMh0tfh8UvIeFxXHzSVC4yhmU7D/HBkmw+Xrabj5btJr19FNeM6MbFQ9JpFxVm62Cc8Qf44l6bKhl4Re3n2DIbUk+qf7jksJ/BO1fam44nnOPTH7PJjLEBvcdYOLTdmUNXbZKrh6459FahzVVbbPVEYMj1x4YyiggnZyTywHn9+fG343ni8pPpFB/JA5+tZcQ/vub3H65i0/4CmyZJHw5f3AdH9ng+dmmhHc9eV7rFpfdkiOnQOsvq7lsFhfvtRKi4TtpDb8tKjtgCca4RXq5hi8b4t11BSgN6C4oMc3DOgM68d/OpfHrHKCafmMo7C3cy/tE5XP3yIr7v/2dMeYlNvXj6D7H9e6gs8y6gO0Kh6wjY+aPPf44m2+wc3dJzLMR1rv0XmGr9XIW5XMKjwVTqgit+ogHdT07s0o5HLhnA/PvHcc+ZvVm35wiXT8vjkfJLYMN0ln7232qjaYAt39icc8YI706SPgwO72h9Nx03z4KUEyC+s+2hlxzWr+ltlasWuovWRPcrzaH7WXJsBHeckcnPTu/JN+v389WqVJavXUD3hX9iwvdxZPbsyfisjkzI6kjHLXMgYxiERXl38PTh9nHnj7ZuTGtQWmS/abiGaMZ3to/5eyGpp//apRqnRkB3VVws8K4shvIp7aG3EuGhIUzs14mHLx1Ev1teI95Rxkspb7Itt4Dff7iKqQ99CPtWUtl9jPcH7dTf5jd3/tBMrW6E7fOhosSmW8D20KH1fYtQ3ikp8BzQdeiiX3gV0EVkkoisF5FNInJfLfuMEZFlIrJaROb4tpnBJbRjX0LO+B19Ds1l9qQ8ZvxiNDd0sbXd71/Wnu15Xn6dDQ235QpaU0DfPMuZNjrVvo5z66GrtqfkyPEBXWui+1W9AV1EHMB/gMlAFnC5iGRV2ycBeAo41xjTD6ijWpXyyojbocsQZPqv6B1zlJ923k5pWDxfHOjE5P/7ltcXbMd4M5Igfait61J2tP59W8LmWfZmrasYl6tUgg5dbJvqSrmoFudND30osMkYs8UYUwq8DUytts8VwDRjzA4AY8x+3zYzCIU44LynbM75s7uRrXMI7zWG6XeNYVBGIr/7cBXXvrSQvYeL6z5O+jCoLIddS1qm3XU5stuustRzXNU21+LCOtKlbSqtJeWiPXS/8CagdwF2ur3Odm5z1xtIFJHZIrJYRK72dCARuUlEFonIopycnMa1OJik9IEx98HaT+DwTugxhs4JUbx6/VD+MrUfP2zNY+Ljc3n7xx2UlFd4PkbaUPvYGtIum7+xjz3POH57XKpdSES1PSX51YYtugK65tD9wZuA7mkeevXv+qHYcnlnAxOB34tIjXqwxpjnjDFDjDFDUlJSGtzYoHTqndD5ZPvcOf48JES4ekQ3Pr/zNHqmxHDftJWMfPAbHp+5gZz8auN/Y5LssnetIqB/bSc7OdeEPSaukwb0tqi8FMqLq0rngqZc/MybgJ4NuFeoSgOqJzyzgS+MMYXGmFxgLtDKi3G3EY5QuORVW++l2rC+HimxfHDLqbz206H07xLP4zM3MvLBWdzz7nJW7TpctWPGMBvQ/Tl7r7LS9tB7jqtZqya+swb0tqi0Wh0X0JSLn3kT0BcCmSLSXUTCgcuAj6vt8xFwmoiEikg0MAxY69umBrGEDBh8rce3RITTMlN46bqhfH3P6Vw2NJ3pq/Yw5Yl5XPLM9yzbecjm0Y8etOul+sve5XD0wPH5c5e4VDvKRaeLty3HKi26pVzCdNiiP9Ub0I0x5cDtwJfYIP2uMWa1iNwsIjc791kLfAGsAH4EnjfGrGq+ZitPeqbE8pepJ/L9/Wfw27NOYMeBIq747wKW0cfusGOB/xrnPt2/urhUqCiFogMt2ybVNNVL54L9RumI0JSLn3g1U9QY8znwebVtz1R7/S/gX75rmmqsdlFh3Di6B1MHduaK53/g8v/lsiIqgbCdP8Cgq/zTqM3f2IlOsR1qvuc+dFFnF7YdngI6aAldP9KZogGsQ3wkb980nK5Jccwr7k7h5vn+aUhJgf124CndAraHDjp0sa1x9cLDqwd0LaHrLxrQA1xybARv3jic7dH9iTmymTnL1rd8I7bNs1Ui6wvoemO0bXGtJ1qjhx6tAd1PNKAHgfYx4Vx4nl2D9PX33+OLVS08zX7zLAiNqioWVl1sR/uoAb1t0ZRLq6MBPUjE9RiGCQllYvwObntzCZ+uaMGp9ptnQbdREBbp+f3QcIhJ0YDe1lRffs5FA7rfaEAPFuHRSKeTOC9pJ4MyErjzraU8OH0d+cVlzXveg9shb6Pn0S3u4lI1h97WVF9+ziU8Fso0oPuDBvRgkj6M0D1Lefnqkzn/5DSembOZsQ/P4Z2FO6iobKYx4Btn2MfMCXXvF5eqBbraGte0/5BqYSRMc+j+ogE9mKQPhfKjxBxcwyOXDOCj20bSNSmaez9YyblPzuOHLXm+P+eGLyGxOyT1qnu/+FQtodvWlObXTLeAplz8SAN6MEkfZh+d64wOSE/g/ZtH8O/LT+ZgYSmXPreAW99YzM4DPprlV1oE276F3hNrTvevLi4VCnNsfRDVNlQvzOWiwxb9RgN6MGnXBdqlH1eoS0Q4d0Bnvr5nDHef2Zsd65by3mN38cj0VRRWX9O0obbOtcWb6ku3QNXQxYJ9TTunajnVa6G7uIYtaimHFqcBPdikD4UdNQt1RYU7uPPEMj6O/Tt3O94hf96znPHIHD5cusu7hTQ82filre3RbVT9+x5bW1RvjLYZtQb0GMC0nkVVgogG9GCTPtzefDycffz2fWvglSmEhIZD2in8LuZDesYUc9c7y7jome9ZmX3Y8/FqYwxsmGFL/oZG1L+/a21RDehtR/X1RF1caRhNu7Q4DejBJt3Dghf71sAr54AjHK75FM59gtCyQl7vMZN/XngS2/MKOfc/87j3/RU1663XZv8aOJINvb1It0DV2qI6dLHtqLOHjhbo8gMN6MGm44k2DeIK6PvX2mAeEmqDeXIv6HACnHIDsuRlLkk/xKxfjuGGUd35YEk24x6ezfPfbqGsorLu82z40j56kz8HiG5vf6Ho0MW2o/oC0S5hzvVitYRui9OAHmwcoZA22AZ092B+7Wc2mLuMvR8iE2D6fcRHhPLbs7P48hejGdQ1kQc+W8ukx+fy7cY6lhHcOMNWV3Tlxusj4ly5SIcutgnG2B54baNcQFMufqABPRilD4O9q2wwF0fNYA4QlQjjfgvb58GajwBbb/3l607hhWuGUF5puOqFH7np1UXsyKvWEys6YH9hZE5sWLviUu1C0qr1Ky+2i49ryqVV0YAejNKHg6lwBvNPawZzl8HX2RTNjN8fG7EgIpxxQkdm/GI0v57Uh3mbchn/2BwembGeolLnMMfNs8BU2vHnDRGnk4vajNoKc4EuFO1HGtCDUffRcPq9zp55Zu37hThg0oNweAfMf+K4tyJCHdw6phez7hnDWSd24olZmxj54Cwe+2oDJWumQ3QSdBncsHbp2qJtx7GAHl/zPV1X1G80oAej0HAY+5vae+buup8GWVPh20drDnUEOrWL5PHLTmbaracyuGsiT3y9nqI1X7I8Ygg7D3k5IsYlrpP9ml58pGGfUy3P03qiLppy8RsN6Kp+Z/4VMPDVH2vdZVBGIs9fcwpzLo8lUfJ5Kac3Yx6ezZ1vLWX1bi/HsLuGLmrapfXzKuWiPfSWpgFd1S+xK5x6J6x6H7Z/X+eu6bnfgji4747buH5kN75eu4+z/z2PW99YzPa8ev6DH5tcpDdGW73SWmqhgw5b9CMN6Mo7o+6C+C7w6V12FEttNsyA9GF06pjKb8/OYv79Z3DX+Ey+WZfD+Efn8LfP1nD4aC012OO1h95mHKuF7iGghzjsClWacmlxGtCVd8Jj4Lyn4MAWeO08OHqw5j6Hd8G+lcfNDm0XFcZd43sz+1djOP/kLjw/bytj/vUNL3+3tebkJFcPXYcutn61rSfqoiV0/UIDuvJejzFw6Ru2VMBrF0Bxtdz4scUsag5X7BgfyT8vGsCnd4zihNR4/vTJGiY+NpeZa9yqK4bHQES7+nvob14K0+9r2s+imqauHDo4A7qmXFqaBnTVML0nwKWvwd4V8PpFVf+xwQb0dum2dEAt+nVuxxs3DOOFa4YgAje8uohHZqyvqugYX8/KRUf2wIYvYPmbUNHMy+ep2pUUgIRAWJTn98NjNOXiBxrQVcP1mQwXvQS7FsMbF9v/3GXFsGW2rd1Sz2IWrslJX9w1mkuHpPPErE3cP20l5RWVNu1SV4Gujc4aMcWHYft3vvuZVMO4CnPV9netKRe/0ICuGifrXLjoBbv60ZuXwqav7KiGBswODXOE8OCF/bl9bC/eXriTW99YQkVsPbNF10+3N2dDo2Dd5z74QVSjlOR7nlTkogHdL0L93QDVhvU7HyorYNqNsGsRhEZCt9MadAgR4ZcT+5AcG86fP13DR+0N5x/di1RW1lx8uLTIfgsYdA0c3gnrPoPJD9W/vJ3yvdJalp9zCYuBwtyWa48CtIeumqr/RXDe01BeAt1Pt8uPNcK1I7vzf5edzIojUUhlOTn7as5KZctsWxSqz2Toe7att753RdParxqntlroLppD9wsN6KrpBlwGN8yEKY816TDnDujMJWPtAhy/fulLtuRUCwgbptuv+V1HQu9J9qacpl38w6uArimXlqYBXflG2hC7CHUTZfXuA0C78lwufHo+C7bk2TcqK2H9F9DrDFuLJibZlgFe91mTzxmwts2DFybab0++Vtvycy46bNEvNKCr1iU+FYDfjU6kfUw4Vz7/A2/9uAN2L4XC/dB7ctW+fc6yE5kObvdTY1u5NR/DzgVwYKvvj12S77kwl0t4DJQV2l/EqsVoQFetS0wHkBCSK/OYdutITu2VzP3TVjL301cw4oDMM6v27Xu2fVw/3T9tbe32LLePh5rhF543o1xA67m0MA3oqnVxhNqgnr+HdlFhvHjNEK4f2Z2U3bNYH57FYXH7mp/UE1L6wrpP/dfehirIgezFzX+eygrYu9I+P7TDx8eutDc860u5gObRW5gGdNX6xKceW+gi1BHCH0bFcELITv5XdBLnP/Xd8TdL+5wF2+fXXTCstVj3OTw1DF6cAEcPNe+58jbblAf4vodeVgiY+octHttXtRSvArqITBKR9SKySURqLaIhIqeISIWIXOS7JqqgU30pug1fAHDWhddxqKiM8/7zHbPWOWvA9J1il9Pb+JUfGuql0kL45Ofw9uV2ZE5luV2guzm50i2OcN/30Our4wLaQ/eTegO6iDiA/wCTgSzgchHJqmW/h4Avfd1IFWSqLxa9fjokZTJg4Cl8dNtIOidEcf3Li7jl9cXsjukLsZ1ab9pl1xJ4djQsfgVG/hx+6ixgtm9V855373JwREDGCN/fNNaA3mp500MfCmwyxmwxxpQCbwNTPex3B/ABsN+H7VPBKC4Vjh6w9WGKj9jhd30mAZDePpqPbh/Jryb2Yda6/Yx/7FvWxI/EbPra7t9aVFbA3H/BC2faBbav+RjO/AskdrcVJfevad7z71kOHfvZ+ww+76HXsbiFiysdowG9RXkT0LsAO91eZzu3HSMiXYDzgWfqOpCI3CQii0RkUU5OTkPbqoKFc+giBXth89dQWWZz5U4RoQ5uG9uLmXefzogeSfxzW0+krJB137eSXnpJPrw8BWY9ACecC7d8ZxfmBlumoGM/W4K4uRhjA3rqAEjIsL8c3atiNlV9tdChasawBvQW5U1A91Qow1R7/ThwrzGmoq4DGWOeM8YMMcYMSUlJ8bKJKujEOQP6kT023RKVCGlDa+yW3j6aF649hSsvv5oiIlky43XufncZBwtLax4zbzN8+VvIWd/MjQeWvg475sO5T8JFL9r2u+uYZXvopvp/Ix85tN1Wo3QFdIBDO+v+TENoyqXV8iagZwPpbq/TgOoFq4cAb4vINuAi4CkROc8XDVRByBXQD2fbGuuZE+1wxlqM759OeN+JnBu1gk+XZzPliXms2uW2+MbGmfDfsfD9k/DUCPj8V803KsYYWPIadD4ZBl3luXBYhyzbyz3swyDrznVDNPUkSOhmn/typEtd64m6HEu5aD2XluRNQF8IZIpIdxEJBy4DPnbfwRjT3RjTzRjTDXgfuNUY86GvG6uChCvlsuZDu9SdM39el9CsKcSW5fHZBdFUGsOFT8/nwyXZ8O2j8MZFduGNG2fBkOtg4fPw74Hw/X+g3ENvvil2L4H9q+Hkq2rfp+OJ9rG50i57loM4oEM/tx66D/Poda0n6qILRftFvQHdGFMO3I4dvbIWeNcYs1pEbhaRm5u7gSoIRSbYUrzrp0NIGPQ8o/7PZJ4J4iDzwBw+uWMUQ7tEEvq/n8LXf6Yy6zw7uqTLYDj7Ebhlvn3+5W/gqeF2fLiv0h9LX7dtP/HC2vdxrei0f7VvzlndnuX2HGGRtuZNWLSPA7orh17XOHTNofuDV+PQjTGfG2N6G2N6GmP+5tz2jDGmxk1QY8y1xpj3fd1QFUREbNrFVEC3URBZxxRzl6hEu+/6z0ku3c2r5jec5fiRv5ddzhUHbyKv1C1l0+EEuHIaXPGeXaH+7cvh3aubHtRLi2Dl+5A1FaISat8vMh7aZcC+ZgjoxsDuZTZ/DvZaJmTAwW2+O0dJgR0SGRpR+z4hIXZykQb0FqUzRVXr5Mqju41uqVffsyF3Azw7Gjmym5Ar36fPBb9j6c7DnPPEPFZmu+XVRez6qLfMh+G3wdqPIW9T09q89hPbe60r3eLSXCNd8vdAUW5VQAcb0H2dcqmrd+6iJXRbnAZ01Tq58uhe5M+P6XOWzR23S4ebvoFeZ3Dh4DTev/lURIQLn5nP2z/uqFqQGsARBsN+Zp9vnNG0Ni99DRK72Xrt9emYBXkbfV/ado9zwY9mD+h15M9dwqM1oLcwDeiqdep3Pgy9qeqmnjcS0uHWBXaxjfY9jm3un9aOj28fybDu7blv2krueXc5RaXlVZ9L7GqLfDUloB/YAtu+hZOvrLl0nicdsmwJgNwNjT+nJ3uWA1J14xXsNSw+ZIcy+kJ9hblcwmM1oLcwDeiqdTrhHDjrXw3/XEpvj8vgJcVG8PJ1Q7n7zN78b9kuzn3yOzbsc5tskzkBtn3X+Ak4S9+wdVoG/sS7/Tv2s4++TrvsWQ7JmcenRBK62kdf9dJL8use4eKiy9C1OA3oKmg4QoQ7z8jkjZ8O41BRGVOf/I4PFjvXLs2cYGekbpnT8ANXVsCyN6HXeIjv7N1nknrZwlm+HumyZzl0Oun4bb4eulhyxMseekzzDFusrLQLnvhbSX5ViquV0ICugs6pvZL5/M5RDEhvxz3vLefe91dQnHqKXbChMWmXzbMgf7dNt3jLEQbJfXw70qUw1y6c7Z4/h+bpoXsT0MOaKYe+YTo8NwZ2/uj7YzfE/CfshLXCXP+2w40GdBWUOsRH8vpPh3HHuF68s2gn5z79A4c6j7JleBs6fHHJqxCdfPzyeN7omOXblMuxGaLVAnp0e5vP9lXVxZICL0e5xDZPysX1S9DfK1XtXWXvg2xoPQVmNaCroBXqCOGeCX145fqhHD5axt83pEH+bkp2NeBrdGGuDSwDLrOLVzdEx362Z++rMgTuU/7ducait3QPvbmGLeZutI/+roGf66wLtP5z/7bDjQZ0FfRO753CjF+cTruT7Jj3V1/9Lz9syfPuwyvesbn3hqRbXDo4b4z6qpTunuU2vVK9GBj4LqBXlEP50brXE3UJj7aTrXzNNV9g30o4vMv3x/dGeYkd2RQSalNuZUf9045qNKArBbSLCuO3l46loP2JDCtfyKXPLeD3H66ioKS89g+5CnF1GVI1nb8hOjrXifFV2mXviprpFhdfBfRSLyotuoTH2uBfWWcR1oYxxgb0bqfZ15v81EvP2wSmEk66zN74bczN9GagAV0pN7EnnkV/s4Fbh7Xn9R+2M+HROXy2Yg/lFZU1d961BHLW2qqKjRGXauvW+GKkS/Fh22OsNaB3hZLDtthZUxwrzOXlTFHwbdqlYL8dZdN3ii2f4K+0S846+3jK9XYI5/rP/NOOajSgK+UucwJiKvl1r118cMupxESEctubSzj9X7N5evbm42utL33VjuTod0HjziXOCUC+6KHvXWkfUwd6ft9XQxe9qYXu4grovhy6mOfMnyf3sgXZtsz2/Wxbb+RssPMOOmRB5nhY/4UdTulnGtCVctdlEEQnwcYZDMpI5Iu7RvPsVYPpmhTNQ1+sY/g/vube91ewftNmWPkBZJ3nXfGw2rgWu2hqMKjthqiLzwK6F7XQXcKaoYfuuiGalGnnDpQWwPb5vju+t3LW2TIPYVHQ52wo3A+7Frd8O6qpfdUApYJRiMNOENo0EyorcIQ4mNivExP7dWL93nxe+X4b05ZkM2r54/RwFLMk7RqGGoN4WsjCGx2ybFA6vMMGiMbas9ymcGI7eH4/0Udj0RvTQ/fl0MW8TbY8cbt0WxrYEWHTLj3H+u4c3shZb+cRgO2hi8OmXdJPadl2VKM9dKWqy5wARXk1ZiP26RTH38/vz+KLyznHsYAXQy/m0g/yuOy5BSze3sjctK8Wu3CtIVqbyAQ7MqWpY9G9WU/UpTly6HmboH1PWy8nPMaWTG5qUbWGqiiz7UhxBvSoROg20v/j4tGArlRNPcfZ/KinCSPFR4iZeS90yOK6Xz3GX6b2Y3NOIRc+PZ8bX110fH0Yb3Toax+bMmO0tNAW+aoroPtqLLo3y8+5HAvoPsyh5260+XOX3hNtXv3AFt+doz4Httqhqil9q7b1OdumYfI2t1w7PNCArlR10e3totSeen5f/wWO7IZznyA8IpKrR3Rjzq/G8MsJvVmwOY+Jj8/lnneXk33QyyAWEWdHoDRlpMu+1XYIXV0BHXwT0Bs1ysVHKZfyUrtQR1Jm1bZe4+3jxpm+OYc3XBOKXD10gD7OWcJ+nmSkAV0pTzLPhD3LIH9f1bYdC+x6pMNuhrQhxzbHRIRy+7hM5v56LDee1oNPVuxm7MOz+enLC5m2JJsjxWV1n6upI11qm/JfXUJXu1h0U1ZmalQO3Ucpl4Pb7CpWSW499KSe9vXGFpx+7xqymNy7altiV/v36Oe0iwZ0pTzJnGAfXRNXykvg4zuhXRqM+53HjyTGhPObs05g9i/HcO2p3Vi75wh3v7ucIX+dyQ2vLOLDpbvI9xTcO2bZnGxZcePaume5HZkT36Xu/RIybG+5KWPRS/Lt6JUQR/37unrxvhq2eGzIYubx2zMnwNZvm2dWqic56+1N2er1bPpMhh3fQ6GXs4ybgQZ0pTzp1N+OGnGlXeY9Zr9qT3ms3sJUnROi+O3ZWcy7dxzTbj2Vq0Z0ZdWuw9z1zjIGPzCT295cws4DbsGnQ5btebq+yjeU64ZofSNtjg1dbMKNUW+XnwO3haJ9lHI5NmSx1/HbM8+EihK7wEhLyFl3fLrFpc9ZNvXV0jdp3WhAV8oTERsoNn9jJ+3MfRj6X2K3eSkkRBiUkcjvp2Qx/75xfHDLCK4YmsHsdfs587E5PDtnM2UVlU1b7KK8BPavrVkD3RNfDF30tjAX2DHaiO9SLnmbICal5gLcXUfabw0tEUgrK+wvFvcboi6dT7adAD/OGtWArlRtMifYYXqvX2SD2KR/NPpQISHC4K7t+dO5/fjq7tMZ1SuFf0xfx7lPfseyoiQ7nnrfqoYfePMsO+Ki88n179su3T42ZehiQwK6iG+XocvbdPwNUZfQCOgxBjbMaNr9AW8c2gHlxZ576CI27bJpVuPTZ02kAV2p2vQYAyFhULDXBvOYZJ8ctnNCFP+9ejDPXDmYA4UlnP/MD+wO70r53gaOdCkvgS9/Y4Ncn7Pq3z8qASLbNa2H7u16oi6+XCi6+pBFd5ln2slZOY1MW3nLdfxkDwEd7PDFskLYOrd521ELDehK1SYiDrKm2kJQJ13q00OLCJNO7MTMu0/n6uFd+b6gIwe3LuOtH3ew/4iXvbvvn7Tjryc/5H0t9qYOXfR2PVEXX9VEP3oQinI999ChKhXW3GkX1wiXlN6e3+9+mv1W4qe0i079V6ouF73QrIePiwzjz1NPZFfYaFJ+/JZ/TvuO+4mnb6c4Tu+dwujeKQzplkhEaLVRJYezbV6/7xTodYb3J0zoWlVPvDG8XU/UxVcBPdfZ5uo3RF3apdn68htnwMg7m36+2uSsh9hOnmvOg03/9DrDDl88+zE7o7UFaQ9dqVagS+/BAEy7KIF7J/UlITqMF7/byk+e/4GBf/6K619eyEfLdtmbqAAzfmdHVEz8e8NOlNDV9tAbm2suaWjKJdamIJqqtiGL7npPsMMGi480/Xy1yV3vOX/urs/ZULDPLwtZa0BXqjVwjnTpXr6dW8b05O2bRrD0DxN4/uohXDwkjQ378vn528s47aFv+Ph/b8Hq/8Gou6tGrngrIcOOCy9qxFhpYxo2bBF8t1B03ia7OlBdBcwyJ9g1Prd80/TzeWKM7aHXF9Azz6wq1tXCNKAr1RrEdoSo9seNdImNCGV8Vkf+MvVE5v5qLC9eO4TM5Aj6LH2AnaYDfzk4nq25DQyWx4YuNmKkS3mJHVHjl5TLRhvMHWG175M2FCLaNV8e/cgue1O4voAe3d4W61r+TouPdtGArlRrIAKdB8Kaj2DZmzVSIiEhwri+HXmt/wr6hGTzVcbPeW3RPsY9MvvYLNTcAi8WenBNLmrM0MVjhbkaUP/dV8MWaxuy6M4RCr3G2eGLFXUsHdhYx26IehiDXt2ou+FIti0V0YI0oCvVWpz9iJ01+uEt8PoFNYNu/j6Y/Q/oNZ7rr7+N7+4dx+1je7Fs50HuemcZQx6YyTlPzOPhL9ezcNsBz8vmucaiN3SkS2UFzPmnfR6X6v3nfNFDr6ywVQxrG7Lo7sSL7GITzVHbJWeDffQmoPccCz3GwrcP2+UBW4iOclGqtWjfA66bDotegJl/gqdGwBm/h6E32dopM/9kV5ef9BCI0CE+knsm9OEX43uzavdh5qzPYe7GHJ6es5knv9lEXEQog7slkhAVRkxEKLERoUSHh/Kz0Hbs3LCGQ2kHGNI1sf7FOUqLYNqNsO5TGH5bVWVBb/hiHPrhnXZqf309dIDek+wvnEUvQt+zm3be6nLW2Zo53s5HGP8neO50+O7f9u+xBWhAV6o1CQmBoTfawPTpL+CL+2DVBzD4Wlj+Joy8q0ZPNSREOCktgZPSErjjjEwOHy1j/qZc5mzIYXn2YTbnFFBUUkFBSTkl5ZWMDW/Pga3ruPaZ78lKjef6Ud05Z0BqzaGRAIW58NZlkL0IJj0Iw29p2M8THmuDcUW5TYk0Rn1DFt05QmHQNTDnIVu3vH33xp3TE/dVirzReaBdb3bBU/bvNK6T79pSCw3oSrVGCenwk/dg5Xsw/V746Dbb8xz9q3o/2i4qjMn9U5ncv2ZqpKyiEt55C3LW8eCI/rwwbyu/fG85D32xjquGd+UnwzJIio2wO+dthtcvhPw9cOlrcMI5Df85ji0UXQiOdg3/PFSNm69ryKK7QVfD3H/CkldsL9kXjLE99H7nN+xz434Haz+26aopj/qmLXXQHLpSrZUInHQJ3L4Qht0CFzzXsCGDHoQ5QghL6kpY/k4uOyWdGb8YzavXDyUrNZ5Hv9rAqQ/O4r4PVrBh0deY58fbiUTXfNq4YA5uFRebkHbJ22hHr8SkeLd/uy7QezIsec0uiuELBfuh+JB3+XN3ST3tN4Ylr7TIakZeBXQRmSQi60Vkk4jc5+H9n4jICuef+SJST6V9pZTXYpJh8oPQfbRvjpfYzRaYKsxBRBjdO4VXrh/KzLtHc8GgNAqXTSPjk0vJLo7g392f5oeyHlRUNnIikqsmelMCuquGS0MW4h5yvS0VsO6Txp/3uDZ4WKXIW6ffC45wmPWAb9pSh3pTLiLiAP4DnAlkAwtF5GNjjHutz63A6caYgyIyGXgOGNYcDVZKNZH70MXYDsc290qK5B+x74DjCfISB/JI3B/4fHkpjy5eQHJsOGdmdWJCv45ktI8mLjKU+MgwIkJD6r6p6otl6PI2QbfTGvaZnuPsz7noJTjxwsaf2yWnCQE9riMMv9WOeBl5p3eVMRvJmxz6UGCTMWYLgIi8DUwFjgV0Y8x8t/0XAGm+bKRSyofcF7pIP8U+P5wN710H2T/CKTeQNOFvPB4Wyd9Kyvlm/X6+WLWXj5ft4q0fjx/uGOYQ4iLDiIsMpWNcJJcNTeecAZ0Jczi//Dd1oejSQjuhx5shi+5CQmDwdfD1n+1ww9qKaXkrZ50df9+QIZvuRt5pR97M/DNc/WHT2lIHbwJ6F2Cn2+ts6u59/xTwuLCeiNwE3ASQkZHhZROVUj5VfSz6hhnwv5vsSJSLXjyuRxsTEcqUkzoz5aTOFJdVsGjbQXILSsgvLuNIcTn5xeXkF5eRX1zOGueSew9/uZ7rR3XnsqEZxDZ1XVHXDVFvhixWd/KV8M3fYfFLTaplD1RN+W9I2sddZDs47R6Y8VvYMtuWZm4G3gR0Tz+Bx4SaiIzFBvRRnt43xjyHTccwZMiQZq5Er5TyKCIWopPhwGb46o/w3eN2yb2LX7E38WoRGeZgVGbtY7ArKw2zN+zn2TlbeOCztfz76438/KQKfgqNT7nkNWDIYnWxHezN3GVvwBl/cK6g1Eg5623xr6Y45QZY8LSdT3DjN43/5VAHb26KZgPpbq/TgN3VdxKRk4DnganGGP+tkqqUql9CBix93QbzwdfBT2fWGcy94SpP8M7PRvC/W09lZK9kXlq4H4C35q9j5pp9FJdVNOyguZsAaXzbhlxvZ2qu/l/jPg9QdMDOPm3oCJfqwiJh7G9sFcY1HzbtWLXwpoe+EMgUke7ALuAy4Ar3HUQkA5gGXGWM2eDzViqlfKtTf8jdAOf8H/S/yOeHPzkjkaevHMz2nSnwAmzbvZ/7X11EdLiDMX1SmNivE2P7diA+0nOxraLScg4VlZGauxFpl9743nW3UZDc2+avB15R//6e1LdKUUMMuMzWdynY3/RjeVBvQDfGlIvI7cCXgAN40RizWkRudr7/DPAHIAl4ynnHu9wYM6RZWqyUarpJ/4AJf7W53WbUtZMdRfOrsWmMTB3Kl6v3MmPNPj5fuZcwhzC8RxKd20WRV1hCbkEpuQUl5BWUctTZk58etYTwuFSO7DjIwPSE+ssUVCdie+lf3Ad7VkCqF4tpV3esKJcPAnqIA274utkWvvBqpqgx5nPg82rbnnF7fgNwg2+bppRqNq6blc0tNALEQWh5EaOdKzD9deqJLN15iBmr9/LVmn2s35tPUmwEybHhdE+OISkmnKTYCGLCQ+j+1W7ePZjJH56aT1piFOcM6MyUk1LJSo2nvNKwPa+ITfsL2JxTwGbnY3FZJYO6JnBKt/YM7d6etAGX2bz14pdgymMN/xlyN9gJUu3S69/XG824ipFO/VdKNR8RZwndqmGLISHC4K6JDO6ayP1nnVD7Z4/sgRlHuXjSWKIjB/DJ8t08N3cLT8/eTHJsBIeKSil3m/CU2i6SnimxJMYIn67Yw1s/2sF5ndtF8ljMGAYtfZudJ99L984dG9bTz1ln0zYtvJxcY2hAV0o1r/Doxo1ycS47F9WpDxf1TOOiwWkcKCxl+qo9LNx6gC6JUfRMiaVXh1h6pMQSG1EVzioqDev35rNw2wF+3HaApzefzssVX/Dd07fxUWRXstoLPRMM6VEVRFQW2WGVKX1s2dv0YfabhUvOepuLbwM0oCulmldja6J7GLLYPiacnwzryk+G1b30niNEyOocT1bneK45tRumciAlz7zDVftnQjmwH4r3hZFPFAccMYRGRJG04UtC5j1KZWgklemn4ug1FskYbic2+SJ/3gI0oCulmldjA3ruJgiNgvguTW6ChIQQccPnUJgDEfGUh8awZm8R327I5duNOSzdeYjIyiKGh6xhVPkqTtu8kl5bZx37/N9+rGTOojkUl1VSUl5BcVklxWUVJESHcfPpPbliWIbn8sMtTAO6Uqp5hcXYhamrqyiHnQtAQqDzIDtO213eRts791XuOjzm2M3gUGBQRgSDMhL5+fhMikrL2XO4mH1HxpCTX8I3R0r4PHc7Sfu/J75gKzsShtEjPIbIsBAiQh32MczBiuxD/PmTNTz/7VbuGp/JBYPScIT4fsKQtzSgK6WaV3iMrXwIUFkJO76H1dPs+qmFOXa7Ixy6DIaup0LGqZA+1FZZ7DywRZoYHR5Kz5RYeqa4lyfuAYwFoLbiwcYYvt2Yy7++XM+v3l/Bs3O38MsJvZnYr1PDh1j6gAZ0pVTzCo+Bfavgi/th9YeQv9umUnpPsCv6OMJg+3wb6Oc9DuYR22s3ldD/Yn+3vk6u8sOnZSYzfdVeHp6xnptfX8KAtHZcNjSDDnERJMdGkBQbTnJsBJFhzZuW0YCulGpeke3sqkcLn4deZ8KJf7VL7Lkv1uFa/7OkALIX2uC+dyVkneufNjeQiHBW/1QmZHVk2tJd/N/Mjdw/bWWN/WLCHSTFRnDV8K7cOLqH79thjH9qZA0ZMsQsWrTIL+dWSrWgg9th12Jbozwqwd+taRHlFZXsOVxMXmEpec7Zr7mF9jGvoISxfTswdWDjbvaKyOLaZuJrD10p1bwSu9o/QSTUEUJ6+2jS20e36Hlb/9QnpZRSXtGArpRSAUIDulJKBQgN6EopFSA0oCulVIDQgK6UUgFCA7pSSgUIDehKKRUg/DZTVERygO2N/HgykOvD5rR1ej2Op9ejil6L4wXC9ehqjEnx9IbfAnpTiMgiXYS6il6P4+n1qKLX4niBfj005aKUUgFCA7pSSgWIthrQn/N3A1oZvR7H0+tRRa/F8QL6erTJHLpSSqma2moPXSmlVDUa0JVSKkC0uYAuIpNEZL2IbBKR+/zdnpYmIi+KyH4RWeW2rb2IfCUiG52Pif5sY0sRkXQR+UZE1orIahH5uXN7sF6PSBH5UUSWO6/Hn53bg/J6AIiIQ0SWisinztcBfS3aVEAXEQfwH2AykAVcLiJZ/m1Vi3sZmFRt233A18aYTOBr5+tgUA7cY4w5ARgO3Ob89xCs16MEGGeMGQAMBCaJyHCC93oA/BxY6/Y6oK9FmwrowFBgkzFmizGmFHgbmOrnNrUoY8xc4EC1zVOBV5zPXwHOa8k2+YsxZo8xZonzeT72P24Xgvd6GGNMgfNlmPOPIUivh4ikAWcDz7ttDuhr0dYCehdgp9vrbOe2YNfRGLMHbJADOvi5PS1ORLoBJwM/EMTXw5liWAbsB74yxgTz9Xgc+DVQ6bYtoK9FWwvo4mGbjrsMciISC3wA3GWMOeLv9viTMabCGDMQSAOGisiJfm6SX4jIFGC/MWaxv9vSktpaQM8G0t1epwG7/dSW1mSfiKQCOB/3+7k9LUZEwrDB/A1jzDTn5qC9Hi7GmEPAbOz9lmC8HiOBc0VkGzY1O05EXifAr0VbC+gLgUwR6S4i4cBlwMd+blNr8DFwjfP5NcBHfmxLixERAV4A1hpjHnV7K1ivR4qIJDifRwHjgXUE4fUwxtxvjEkzxnTDxolZxpgrCfBr0eZmiorIWdjcmAN40RjzN/+2qGWJyFvAGGwZ0H3AH4EPgXeBDGAHcLExpvqN04AjIqOAb4GVVOVJf4PNowfj9TgJe6PPge2svWuM+YuIJBGE18NFRMYAvzTGTAn0a9HmArpSSinP2lrKRSmlVC00oCulVIDQgK6UUgFCA7pSSgUIDehKKRUgNKArpVSA0ICulFIB4v8BFYrrQZLLPhcAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#-----------------------------------------------------------\n",
    "# Retrieve a list of list results on training and test data\n",
    "# sets for each training epoch\n",
    "#-----------------------------------------------------------\n",
    "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)) # Get number of epochs\n",
    "\n",
    "#------------------------------------------------\n",
    "# Plot training and validation accuracy per epoch\n",
    "#------------------------------------------------\n",
    "plt.plot  ( epochs, acc )\n",
    "plt.plot  ( epochs, val_acc )\n",
    "plt.title ('Training and validation accuracy')\n",
    "plt.figure()\n",
    "\n",
    "#------------------------------------------------\n",
    "# Plot training and validation loss per epoch\n",
    "#------------------------------------------------\n",
    "plt.plot  ( epochs,     loss )\n",
    "plt.plot  ( epochs, val_loss )\n",
    "plt.title ('Training and validation loss'   )"
   ]
  }
 ],
 "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": "3.8.5"
  },
  "papermill": {
   "default_parameters": {},
   "duration": 19291.732353,
   "end_time": "2021-02-28T23:40:18.478474",
   "environment_variables": {},
   "exception": null,
   "input_path": "__notebook__.ipynb",
   "output_path": "__notebook__.ipynb",
   "parameters": {},
   "start_time": "2021-02-28T18:18:46.746121",
   "version": "2.2.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}