selectmodel(2).ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## import library"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras import layers, models\n",
    "import numpy as np \n",
    "import matplotlib.pyplot as plt "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Get MNIST Data. \n",
    "### MNIST data loacted in tensorflow > keras > datasets > mnist \n",
    "### Split data to (train images, train labels) and (test images, test labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "mnist = keras.datasets.mnist\n",
    "(train_images, train_labels), (test_images, test_labels) = mnist.load_data()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### There are Total 60000 Train images and Train labels. (6000 images for single class)\n",
    "### Shape of single image is 28 x 28 (pixel)\n",
    "### "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of Train images : (60000, 28, 28)\n",
      "Shape of Train labels :  (60000,)\n",
      "\n",
      "Shape of Test images :  (10000, 28, 28)\n",
      "Shape of Test labels :  (10000,)\n"
     ]
    }
   ],
   "source": [
    "print('Shape of Train images :',train_images.shape)\n",
    "print('Shape of Train labels : ', train_labels.shape)\n",
    "print('\\nShape of Test images : ', test_images.shape)\n",
    "print(\"Shape of Test labels : \",test_labels.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train labels :  [5 0 4 ... 5 6 8]\n"
     ]
    }
   ],
   "source": [
    "print('Train labels : ',train_labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Plot first train image. \n",
    "### when value is close to 0 : dark \n",
    "### when value is close to 255 : white"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0  51 159 253\n",
      "  159  50   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0  48 238 252 252\n",
      "  252 237   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0  54 227 253 252 239\n",
      "  233 252  57   6   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0  10  60 224 252 253 252 202\n",
      "   84 252 253 122   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0 163 252 252 252 253 252 252\n",
      "   96 189 253 167   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0  51 238 253 253 190 114 253 228\n",
      "   47  79 255 168   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0  48 238 252 252 179  12  75 121  21\n",
      "    0   0 253 243  50   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0  38 165 253 233 208  84   0   0   0   0\n",
      "    0   0 253 252 165   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   7 178 252 240  71  19  28   0   0   0   0\n",
      "    0   0 253 252 195   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0  57 252 252  63   0   0   0   0   0   0   0\n",
      "    0   0 253 252 195   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0 198 253 190   0   0   0   0   0   0   0   0\n",
      "    0   0 255 253 196   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0  76 246 252 112   0   0   0   0   0   0   0   0\n",
      "    0   0 253 252 148   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0  85 252 230  25   0   0   0   0   0   0   0   0\n",
      "    7 135 253 186  12   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0  85 252 223   0   0   0   0   0   0   0   0   7\n",
      "  131 252 225  71   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0  85 252 145   0   0   0   0   0   0   0  48 165\n",
      "  252 173   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0  86 253 225   0   0   0   0   0   0 114 238 253\n",
      "  162   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0  85 252 249 146  48  29  85 178 225 253 223 167\n",
      "   56   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0  85 252 252 252 229 215 252 252 252 196 130   0\n",
      "    0   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0  28 199 252 252 253 252 252 233 145   0   0   0\n",
      "    0   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0  25 128 252 253 252 141  37   0   0   0   0\n",
      "    0   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   0   0   0   0   0   0   0   0   0]]\n"
     ]
    }
   ],
   "source": [
    "print(train_images[1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Plot First 10 Train images and Corresponding labels  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "First 10 Train images in MNIST dataset\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 10 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Train labels match with Train label sequentialy\n",
      " [5 0 4 1 9 2 1 3 1 4]\n"
     ]
    }
   ],
   "source": [
    "print('First 10 Train images in MNIST dataset\\n')\n",
    "for i in range(10):\n",
    "    plt.subplot(1, 10, i+1)\n",
    "    plt.xticks([])\n",
    "    plt.yticks([])\n",
    "    plt.imshow(train_images[i])\n",
    "plt.show()\n",
    "print('\\nTrain labels match with Train label sequentialy\\n',train_labels[:10])\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Important\n",
    "### Change data shape (60000 x 28 x 28) to (60000 x 28 x 28 x 1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_images = tf.reshape(train_images, [-1, 28, 28, 1])\n",
    "test_images = tf.reshape(test_images, [-1, 28, 28, 1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Select one convolution model below \n",
    "## There are 3 example models. \n",
    "## 3, 5, 7 layer each \n",
    "## MODEL 1 : 3 Layers with 1 Convolution layer  \n",
    "## MODEL 2 : 5 Layers with 2 Convolution layer \n",
    "## MODEL 3 : 7 Layers with 4 Convolution layer "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "def select_model(model_number):\n",
    "    if model_number == 1:\n",
    "        model = keras.models.Sequential([\n",
    "                    keras.layers.Conv2D(32, (3,3), activation = 'relu', input_shape = (28, 28,1)),  # layer 1 \n",
    "                    keras.layers.MaxPool2D((2,2)),                                                  # layer 2 \n",
    "                    keras.layers.Flatten(),\n",
    "                    keras.layers.Dense(10, activation = 'softmax')])                                # layer 3\n",
    "\n",
    "    if model_number == 2:\n",
    "        model = keras.models.Sequential([\n",
    "                    keras.layers.Conv2D(32, (3,3), activation = 'relu', input_shape=(28,28,1)),     # layer 1 \n",
    "                    keras.layers.MaxPool2D((2,2)),                                                  # layer 2\n",
    "                    keras.layers.Conv2D(64, (3,3), activation = 'relu'),                            # layer 3 \n",
    "                    keras.layers.MaxPool2D((2,2)),                                                  # layer 4\n",
    "                    keras.layers.Flatten(),\n",
    "                    keras.layers.Dense(10, activation = 'softmax')])                                # layer 5\n",
    "                    \n",
    "    if model_number == 3: \n",
    "        model = keras.models.Sequential([\n",
    "                    keras.layers.Conv2D(32, (3,3), activation = 'relu', input_shape = (28, 28,1)),  # layer 1\n",
    "                    keras.layers.MaxPool2D((2,2)),                                                  # layer 2\n",
    "                    keras.layers.Conv2D(64, (3,3), activation = 'relu'),                            # layer 3\n",
    "                    keras.layers.Conv2D(64, (3,3), activation = 'relu'),                            # layer 4\n",
    "                    keras.layers.MaxPool2D((2,2)),                                                  # layer 5\n",
    "                    keras.layers.Conv2D(128, (3,3), activation = 'relu'),                           # layer 6\n",
    "                    keras.layers.Flatten(),\n",
    "                    keras.layers.Dense(10, activation = 'softmax')])                                # layer 7\n",
    "    \n",
    "    return model \n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = select_model(2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## If you want to see information of model, model.summary() will help\n",
    "### summary() is also built in function "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_2\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv2d_4 (Conv2D)           (None, 26, 26, 32)        320       \n",
      "                                                                 \n",
      " max_pooling2d_4 (MaxPooling  (None, 13, 13, 32)       0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " conv2d_5 (Conv2D)           (None, 11, 11, 64)        18496     \n",
      "                                                                 \n",
      " max_pooling2d_5 (MaxPooling  (None, 5, 5, 64)         0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " flatten_2 (Flatten)         (None, 1600)              0         \n",
      "                                                                 \n",
      " dense_2 (Dense)             (None, 10)                16010     \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 34,826\n",
      "Trainable params: 34,826\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Components in training step \n",
    "### Optimizer, Loss function, accuracy metrics "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(\n",
    "    optimizer = 'adam',\n",
    "    loss = 'sparse_categorical_crossentropy',\n",
    "    metrics = ['accuracy']\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training Step \n",
    "## Training for 5 epochs. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "1875/1875 [==============================] - 54s 28ms/step - loss: 0.3717 - accuracy: 0.9449\n",
      "Epoch 2/5\n",
      "1875/1875 [==============================] - 51s 27ms/step - loss: 0.0723 - accuracy: 0.9790\n",
      "Epoch 3/5\n",
      "1875/1875 [==============================] - 43s 23ms/step - loss: 0.0563 - accuracy: 0.9825\n",
      "Epoch 4/5\n",
      "1875/1875 [==============================] - 43s 23ms/step - loss: 0.0473 - accuracy: 0.9852\n",
      "Epoch 5/5\n",
      "1875/1875 [==============================] - 43s 23ms/step - loss: 0.0418 - accuracy: 0.9875\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x21490ef62b0>"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(train_images, train_labels,  epochs = 5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Test Step \n",
    "## Perform Test with Test data "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313/313 - 2s - loss: 0.0546 - accuracy: 0.9849 - 2s/epoch - 8ms/step\n",
      "\n",
      "Test loss :  0.054558057337999344\n",
      "Test accuracy : 0.9848999977111816\n"
     ]
    }
   ],
   "source": [
    "test_loss, accuracy = model.evaluate(test_images, test_labels, verbose = 2)\n",
    "print('\\nTest loss : ', test_loss)\n",
    "print('Test accuracy :', accuracy)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Before prediction, change test image's type to float 32. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313/313 [==============================] - 3s 8ms/step\n"
     ]
    }
   ],
   "source": [
    "test_images = tf.cast(test_images, tf.float32)\n",
    "pred = model.predict(test_images)\n",
    "Number = [0,1,2,3,4,5,6,7,8,9]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Prediction :  (10000, 10)\n",
      "Test labels :  (10000,)\n"
     ]
    }
   ],
   "source": [
    "print('Prediction : ', pred.shape)\n",
    "print('Test labels : ', test_labels.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Functions for plot images, probability"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_image(i, predictions_array, true_label, img):\n",
    "  predictions_array, true_label, img = predictions_array[i], true_label[i], img[i]\n",
    "  plt.grid(False)\n",
    "  plt.xticks([])\n",
    "  plt.yticks([])\n",
    "\n",
    "  plt.imshow(img, cmap=plt.cm.binary)\n",
    "\n",
    "  predicted_label = np.argmax(predictions_array)\n",
    "  if predicted_label == true_label:\n",
    "    color = 'blue'\n",
    "  else:\n",
    "    color = 'red'\n",
    "\n",
    "  plt.xlabel(\"{} {:2.0f}% ({})\".format(Number[predicted_label],\n",
    "                                100*np.max(predictions_array),\n",
    "                                Number[true_label]),\n",
    "                                color=color)\n",
    "\n",
    "def plot_value_array(i, predictions_array, true_label):\n",
    "  predictions_array, true_label = predictions_array[i], true_label[i]\n",
    "  plt.grid(False)\n",
    "  plt.xticks([])\n",
    "  plt.yticks([])\n",
    "  thisplot = plt.bar(range(10), predictions_array, color=\"#777777\")\n",
    "  plt.ylim([0, 1])\n",
    "  predicted_label = np.argmax(predictions_array)\n",
    "  plt.xticks(Number)\n",
    "\n",
    "  thisplot[predicted_label].set_color('red')\n",
    "  thisplot[true_label].set_color('blue')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "(train_images, train_labels), (test_images, test_labels) = mnist.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x216 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "i = 1\n",
    "plt.figure(figsize=(6,3))\n",
    "plt.subplot(1,2,1)\n",
    "plot_image(i, pred, test_labels, test_images)\n",
    "plt.subplot(1,2,2)\n",
    "plot_value_array(i, pred,  test_labels)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x720 with 30 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "num_rows = 5\n",
    "num_cols = 3\n",
    "num_images = num_rows*num_cols\n",
    "plt.figure(figsize=(2*2*num_cols, 2*num_rows))\n",
    "for i in range(num_images):\n",
    "  plt.subplot(num_rows, 2*num_cols, 2*i+1)\n",
    "  plot_image(i, pred, test_labels, test_images)\n",
    "  plt.subplot(num_rows, 2*num_cols, 2*i+2)\n",
    "  plot_value_array(i, pred, test_labels)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Plot images and probability that model predicted wrong"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "def error_mnist(prediction_array, true_label):\n",
    "    error_index = []\n",
    "    \n",
    "    for i in range(true_label.shape[0]):\n",
    "        if np.argmax(prediction_array[i]) != true_label[i]:\n",
    "            error_index.append(i)\n",
    "    return error_index\n",
    "\n",
    "# change num_cols, num_rows if you want to see more result.  \n",
    "def plot_error(index, prediction_array, true_label):\n",
    "    num_cols = 5\n",
    "    num_rows = 5\n",
    "    plt.figure(figsize=(2*2*num_cols, 2*num_rows))\n",
    "\n",
    "    assert len(index) < num_cols * num_rows\n",
    "    for i in range(len(index)):\n",
    "        plt.subplot(num_rows, 2*num_cols, 2*i+1)\n",
    "        idx = index[i]\n",
    "        plt.imshow(test_images[idx])\n",
    "        plt.subplot(num_rows, 2*num_cols, 2*i+2)\n",
    "        plt.bar(range(10), prediction_array[idx])\n",
    "        plt.xticks(Number)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Find index of wrong prediction\n",
    "## Plot first 10 wrong predicted images and probability"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[115, 149, 259, 290, 320, 321, 445, 449, 583, 611]\n"
     ]
    }
   ],
   "source": [
    "index = error_mnist(pred, test_labels)\n",
    "index_slice = index[:10]\n",
    "print(index[:10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 20 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_error(index_slice, pred, test_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'DONE' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-23-87f6c984fac4>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[1;32m----> 1\u001b[1;33m \u001b[0mDONE\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m: name 'DONE' is not defined"
     ]
    }
   ],
   "source": [
    "DONE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "file_extension": ".py",
  "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"
  },
  "mimetype": "text/x-python",
  "name": "python",
  "npconvert_exporter": "python",
  "pygments_lexer": "ipython3",
  "version": 3
 },
 "nbformat": 4,
 "nbformat_minor": 2
}