015_alexnet.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "import matplotlib.pyplot as plt\n",
    "import os\n",
    "import time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "(train_images, train_labels), (test_images, test_labels) = keras.datasets.cifar10.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "CLASS_NAMES= ['airplane', 'automobile', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "validation_images, validation_labels = train_images[:5000], train_labels[:5000]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_ds = tf.data.Dataset.from_tensor_slices((train_images, train_labels))\n",
    "test_ds = tf.data.Dataset.from_tensor_slices((test_images, test_labels))\n",
    "validation_ds = tf.data.Dataset.from_tensor_slices((validation_images, validation_labels))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1440 with 5 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(20,20))\n",
    "for i, (image, label) in enumerate(train_ds.take(5)):\n",
    "    ax = plt.subplot(5,5,i+1)\n",
    "    plt.imshow(image)\n",
    "    plt.title(CLASS_NAMES[label.numpy()[0]])\n",
    "    plt.axis('off')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def augment_images(image, label):\n",
    "    # Normalize images to have a mean of 0 and standard deviation of 1\n",
    "    image = tf.image.per_image_standardization(image)\n",
    "    # Resize images from 32x32 to 277x277\n",
    "    image = tf.image.resize(image, (227,227))\n",
    "    return image, label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_ds_size = len(list(train_ds))\n",
    "train_ds_size = len(list(test_ds))\n",
    "validation_ds_size = len(list(validation_ds))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From C:\\Users\\LoveShark\\anaconda3\\envs\\tensorflowgpu2\\lib\\site-packages\\tensorflow_core\\python\\ops\\image_ops_impl.py:1556: div (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Deprecated in favor of operator or tf.math.divide.\n"
     ]
    }
   ],
   "source": [
    "train_ds = (train_ds\n",
    "                  .map(augment_images)\n",
    "                  .shuffle(buffer_size=train_ds_size)\n",
    "                  .batch(batch_size=32, drop_remainder=True))\n",
    "\n",
    "test_ds = (test_ds\n",
    "                  .map(augment_images)\n",
    "                  .shuffle(buffer_size=train_ds_size)\n",
    "                  .batch(batch_size=32, drop_remainder=True))\n",
    "\n",
    "validation_ds = (validation_ds\n",
    "                  .map(augment_images)\n",
    "                  .shuffle(buffer_size=train_ds_size)\n",
    "                  .batch(batch_size=32, drop_remainder=True))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = keras.models.Sequential([\n",
    "    keras.layers.Conv2D(filters=96, kernel_size=(11,11), strides=(4,4), activation='relu', input_shape=(227,227,3)),\n",
    "    keras.layers.BatchNormalization(),\n",
    "    keras.layers.MaxPool2D(pool_size=(3,3), strides=(2,2)),\n",
    "    keras.layers.Conv2D(filters=256, kernel_size=(5,5), strides=(1,1), activation='relu', padding=\"same\"),\n",
    "    keras.layers.BatchNormalization(),\n",
    "    keras.layers.MaxPool2D(pool_size=(3,3), strides=(2,2)),\n",
    "    keras.layers.Conv2D(filters=384, kernel_size=(3,3), strides=(1,1), activation='relu', padding=\"same\"),\n",
    "    keras.layers.BatchNormalization(),\n",
    "    keras.layers.Conv2D(filters=384, kernel_size=(1,1), strides=(1,1), activation='relu', padding=\"same\"),\n",
    "    keras.layers.BatchNormalization(),\n",
    "    keras.layers.Conv2D(filters=256, kernel_size=(1,1), strides=(1,1), activation='relu', padding=\"same\"),\n",
    "    keras.layers.BatchNormalization(),\n",
    "    keras.layers.MaxPool2D(pool_size=(3,3), strides=(2,2)),\n",
    "    keras.layers.Flatten(),\n",
    "    keras.layers.Dense(4096, activation='relu'),\n",
    "    keras.layers.Dropout(0.5),\n",
    "    keras.layers.Dense(4096, activation='relu'),\n",
    "    keras.layers.Dropout(0.5),\n",
    "    keras.layers.Dense(10, activation='softmax')\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "root_logdir = os.path.join(os.curdir, \"logs\\\\fit\\\\\")\n",
    "\n",
    "def get_run_logdir():\n",
    "    run_id = time.strftime(\"run_%Y_%m_%d-%H_%M_%S\")\n",
    "    return os.path.join(root_logdir, run_id)\n",
    "\n",
    "run_logdir = get_run_logdir()\n",
    "tensorboard_cb = keras.callbacks.TensorBoard(run_logdir)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d (Conv2D)              (None, 55, 55, 96)        34944     \n",
      "_________________________________________________________________\n",
      "batch_normalization (BatchNo (None, 55, 55, 96)        384       \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 27, 27, 96)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 27, 27, 256)       614656    \n",
      "_________________________________________________________________\n",
      "batch_normalization_1 (Batch (None, 27, 27, 256)       1024      \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 13, 13, 256)       0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 13, 13, 384)       885120    \n",
      "_________________________________________________________________\n",
      "batch_normalization_2 (Batch (None, 13, 13, 384)       1536      \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 13, 13, 384)       147840    \n",
      "_________________________________________________________________\n",
      "batch_normalization_3 (Batch (None, 13, 13, 384)       1536      \n",
      "_________________________________________________________________\n",
      "conv2d_4 (Conv2D)            (None, 13, 13, 256)       98560     \n",
      "_________________________________________________________________\n",
      "batch_normalization_4 (Batch (None, 13, 13, 256)       1024      \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 6, 6, 256)         0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 9216)              0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 4096)              37752832  \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 4096)              0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 4096)              16781312  \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 4096)              0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 10)                40970     \n",
      "=================================================================\n",
      "Total params: 56,361,738\n",
      "Trainable params: 56,358,986\n",
      "Non-trainable params: 2,752\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.compile(loss='sparse_categorical_crossentropy', optimizer=tf.optimizers.SGD(lr=0.001), metrics=['accuracy'])\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train for 1562 steps, validate for 156 steps\n",
      "Epoch 1/50\n",
      "   1/1562 [..............................] - ETA: 3:05:44 - loss: 5.6104 - accuracy: 0.0625WARNING:tensorflow:Method (on_train_batch_end) is slow compared to the batch update (0.168881). Check your callbacks.\n",
      "1562/1562 [==============================] - 42s 27ms/step - loss: 2.0966 - accuracy: 0.3251 - val_loss: 1.4436 - val_accuracy: 0.4920\n",
      "Epoch 2/50\n",
      "1562/1562 [==============================] - 39s 25ms/step - loss: 1.5864 - accuracy: 0.4382 - val_loss: 1.2939 - val_accuracy: 0.5447\n",
      "Epoch 3/50\n",
      "1562/1562 [==============================] - 39s 25ms/step - loss: 1.4391 - accuracy: 0.4889 - val_loss: 1.1749 - val_accuracy: 0.5859\n",
      "Epoch 4/50\n",
      "1562/1562 [==============================] - 39s 25ms/step - loss: 1.3278 - accuracy: 0.5307 - val_loss: 1.0841 - val_accuracy: 0.6228\n",
      "Epoch 5/50\n",
      "1562/1562 [==============================] - 39s 25ms/step - loss: 1.2349 - accuracy: 0.5630 - val_loss: 1.0094 - val_accuracy: 0.6569\n",
      "Epoch 6/50\n",
      "1562/1562 [==============================] - 40s 25ms/step - loss: 1.1657 - accuracy: 0.5876 - val_loss: 0.9599 - val_accuracy: 0.6851\n",
      "Epoch 7/50\n",
      "1562/1562 [==============================] - 39s 25ms/step - loss: 1.1054 - accuracy: 0.6128 - val_loss: 0.9102 - val_accuracy: 0.6937\n",
      "Epoch 8/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 1.0477 - accuracy: 0.6285 - val_loss: 0.8584 - val_accuracy: 0.7109\n",
      "Epoch 9/50\n",
      "1562/1562 [==============================] - 39s 25ms/step - loss: 1.0026 - accuracy: 0.6461 - val_loss: 0.8392 - val_accuracy: 0.7137\n",
      "Epoch 10/50\n",
      "1562/1562 [==============================] - 39s 25ms/step - loss: 0.9601 - accuracy: 0.6627 - val_loss: 0.7684 - val_accuracy: 0.7398\n",
      "Epoch 11/50\n",
      "1562/1562 [==============================] - 40s 25ms/step - loss: 0.9175 - accuracy: 0.6771 - val_loss: 0.7683 - val_accuracy: 0.7476\n",
      "Epoch 12/50\n",
      "1562/1562 [==============================] - 40s 25ms/step - loss: 0.8827 - accuracy: 0.6914 - val_loss: 0.7012 - val_accuracy: 0.7702\n",
      "Epoch 13/50\n",
      "1562/1562 [==============================] - 40s 25ms/step - loss: 0.8465 - accuracy: 0.7035 - val_loss: 0.6496 - val_accuracy: 0.7903\n",
      "Epoch 14/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.8129 - accuracy: 0.7160 - val_loss: 0.6137 - val_accuracy: 0.7991\n",
      "Epoch 15/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.7832 - accuracy: 0.7250 - val_loss: 0.6181 - val_accuracy: 0.7957\n",
      "Epoch 16/50\n",
      "1562/1562 [==============================] - 40s 25ms/step - loss: 0.7527 - accuracy: 0.7371 - val_loss: 0.6102 - val_accuracy: 0.7953\n",
      "Epoch 17/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.7193 - accuracy: 0.7470 - val_loss: 0.5236 - val_accuracy: 0.8327\n",
      "Epoch 18/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.6898 - accuracy: 0.7559 - val_loss: 0.5091 - val_accuracy: 0.8425\n",
      "Epoch 19/50\n",
      "1562/1562 [==============================] - 40s 25ms/step - loss: 0.6620 - accuracy: 0.7677 - val_loss: 0.4824 - val_accuracy: 0.8468\n",
      "Epoch 20/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.6370 - accuracy: 0.7766 - val_loss: 0.4491 - val_accuracy: 0.8620\n",
      "Epoch 21/50\n",
      "1562/1562 [==============================] - 41s 26ms/step - loss: 0.6120 - accuracy: 0.7850 - val_loss: 0.4212 - val_accuracy: 0.8694\n",
      "Epoch 22/50\n",
      "1562/1562 [==============================] - 41s 26ms/step - loss: 0.5846 - accuracy: 0.7943 - val_loss: 0.4091 - val_accuracy: 0.8746\n",
      "Epoch 23/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.5561 - accuracy: 0.8070 - val_loss: 0.3737 - val_accuracy: 0.8872\n",
      "Epoch 24/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.5314 - accuracy: 0.8150 - val_loss: 0.3808 - val_accuracy: 0.8810\n",
      "Epoch 25/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.5107 - accuracy: 0.8197 - val_loss: 0.3246 - val_accuracy: 0.9048\n",
      "Epoch 26/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.4833 - accuracy: 0.8304 - val_loss: 0.3085 - val_accuracy: 0.9115\n",
      "Epoch 27/50\n",
      "1562/1562 [==============================] - 41s 26ms/step - loss: 0.4595 - accuracy: 0.8425 - val_loss: 0.2992 - val_accuracy: 0.9111\n",
      "Epoch 28/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.4395 - accuracy: 0.8467 - val_loss: 0.2566 - val_accuracy: 0.9305\n",
      "Epoch 29/50\n",
      "1562/1562 [==============================] - 41s 26ms/step - loss: 0.4157 - accuracy: 0.8563 - val_loss: 0.2482 - val_accuracy: 0.9339\n",
      "Epoch 30/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.3930 - accuracy: 0.8629 - val_loss: 0.2129 - val_accuracy: 0.9449\n",
      "Epoch 31/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.3727 - accuracy: 0.8705 - val_loss: 0.1999 - val_accuracy: 0.9525\n",
      "Epoch 32/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.3584 - accuracy: 0.8751 - val_loss: 0.1791 - val_accuracy: 0.9593\n",
      "Epoch 33/50\n",
      "1562/1562 [==============================] - 41s 26ms/step - loss: 0.3387 - accuracy: 0.8830 - val_loss: 0.1770 - val_accuracy: 0.9557\n",
      "Epoch 34/50\n",
      "1562/1562 [==============================] - 41s 26ms/step - loss: 0.3189 - accuracy: 0.8905 - val_loss: 0.1613 - val_accuracy: 0.9643\n",
      "Epoch 35/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.3036 - accuracy: 0.8969 - val_loss: 0.1421 - val_accuracy: 0.9681\n",
      "Epoch 36/50\n",
      "1562/1562 [==============================] - 41s 26ms/step - loss: 0.2784 - accuracy: 0.9039 - val_loss: 0.1290 - val_accuracy: 0.9736\n",
      "Epoch 37/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.2626 - accuracy: 0.9080 - val_loss: 0.1148 - val_accuracy: 0.9762\n",
      "Epoch 38/50\n",
      "1562/1562 [==============================] - 41s 26ms/step - loss: 0.2521 - accuracy: 0.9145 - val_loss: 0.0937 - val_accuracy: 0.9828\n",
      "Epoch 39/50\n",
      "1562/1562 [==============================] - 42s 27ms/step - loss: 0.2387 - accuracy: 0.9190 - val_loss: 0.1045 - val_accuracy: 0.9768\n",
      "Epoch 40/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.2215 - accuracy: 0.9247 - val_loss: 0.0850 - val_accuracy: 0.9860\n",
      "Epoch 41/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.2124 - accuracy: 0.9274 - val_loss: 0.0750 - val_accuracy: 0.9862\n",
      "Epoch 42/50\n",
      "1562/1562 [==============================] - 41s 26ms/step - loss: 0.1980 - accuracy: 0.9335 - val_loss: 0.0680 - val_accuracy: 0.9896\n",
      "Epoch 43/50\n",
      "1562/1562 [==============================] - 41s 26ms/step - loss: 0.1906 - accuracy: 0.9350 - val_loss: 0.0616 - val_accuracy: 0.9912\n",
      "Epoch 44/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.1769 - accuracy: 0.9410 - val_loss: 0.0508 - val_accuracy: 0.9922\n",
      "Epoch 45/50\n",
      "1562/1562 [==============================] - 41s 26ms/step - loss: 0.1648 - accuracy: 0.9455 - val_loss: 0.0485 - val_accuracy: 0.9936\n",
      "Epoch 46/50\n",
      "1562/1562 [==============================] - 41s 26ms/step - loss: 0.1571 - accuracy: 0.9487 - val_loss: 0.0435 - val_accuracy: 0.9952\n",
      "Epoch 47/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.1514 - accuracy: 0.9501 - val_loss: 0.0395 - val_accuracy: 0.9950\n",
      "Epoch 48/50\n",
      "1562/1562 [==============================] - 41s 26ms/step - loss: 0.1402 - accuracy: 0.9535 - val_loss: 0.0274 - val_accuracy: 0.9984\n",
      "Epoch 49/50\n",
      "1562/1562 [==============================] - 40s 26ms/step - loss: 0.1357 - accuracy: 0.9549 - val_loss: 0.0308 - val_accuracy: 0.9966\n",
      "Epoch 50/50\n",
      "1562/1562 [==============================] - 42s 27ms/step - loss: 0.1269 - accuracy: 0.9596 - val_loss: 0.0251 - val_accuracy: 0.9976\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x2de3aaa0ec8>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(train_ds,\n",
    "          epochs=50,\n",
    "          validation_data=validation_ds,\n",
    "          validation_freq=1,\n",
    "          callbacks=[tensorboard_cb])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "312/312 [==============================] - 8s 27ms/step - loss: 0.9814 - accuracy: 0.7439\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.9813630809673132, 0.7438902]"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate(test_ds)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Keras Class API of AlexNet\n",
    "\n",
    "class CustomAlexNet(keras.Model):\n",
    "    def __init__(self):\n",
    "        super(CustomAlexNet, self).__init__()\n",
    "        self.conv1 = keras.layers.Conv2D(filters=96, kernel_size=(11,11), strides=(4,4), activation='relu', input_shape=(227,227,3))\n",
    "        self.conv2 = keras.layers.Conv2D(filters=256, kernel_size=(5,5), strides=(1,1), activation='relu', padding=\"same\")\n",
    "        self.conv3 = keras.layers.Conv2D(filters=384, kernel_size=(3,3), strides=(1,1), activation='relu', padding=\"same\")\n",
    "        self.conv4 = keras.layers.Conv2D(filters=384, kernel_size=(3,3), strides=(1,1), activation='relu', padding=\"same\" )\n",
    "        self.conv5 = keras.layers.Conv2D(filters=256, kernel_size=(3,3), strides=(1,1), activation='relu', padding=\"same\")\n",
    "        self.bn = keras.layers.BatchNormalization()\n",
    "        self.maxpool = keras.layers.MaxPool2D(pool_size=(3,3), strides=(2,2))\n",
    "        self.dropout = keras.layers.Dropout(0.5)\n",
    "        self.flatten_layer = keras.layers.Flatten()\n",
    "        self.dense_layer = keras.layers.Dense(units=4096, activation='relu')\n",
    "        self.output_layer = keras.layers.Dense(units=10, activation='softmax')\n",
    "    \n",
    "    def call(self, x):\n",
    "        x = self.conv1(x)\n",
    "        x = self.bn(x)\n",
    "        x = self.maxpool(x)\n",
    "        x = self.conv2(x)\n",
    "        x = self.bn(x)\n",
    "        x = self.maxpool(x)\n",
    "        x = self.conv3(x)\n",
    "        x = self.conv4(x)\n",
    "        x = self.conv5(x)\n",
    "        x = self.maxpool(x)\n",
    "        x = self.dense_layer(x)\n",
    "        x = self.dropout(x)\n",
    "        x = self.dense_layer(x)\n",
    "        x = self.dropout(x)\n",
    "        x = self.output_layer(x)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = CustomAlexNet()\n",
    "model.compile(loss='sparse_categorical_crossentropy', optimizer=tf.optimizers.SGD(lr=0.001), metrics=['accuracy'])\n",
    "model.fit(train_ds,\n",
    "          epochs=30,\n",
    "          validation_data=validation_ds,\n",
    "          validation_freq=1,\n",
    "          callbacks=[tensorboard_cb])"
   ]
  },
  {
   "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": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}