legacy_materials/3_cnn_with_mnist.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "8b1f2202",
   "metadata": {},
   "source": [
    "# DL Study third week\n",
    "### MNIST dataset으로 CNN 모델 학습하기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "c26aa939",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 필요 라이브러리 및 모듈 로드\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import tensorflow as tf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "46c93276",
   "metadata": {},
   "outputs": [],
   "source": [
    "# mnist dataset 로드\n",
    "mnist = tf.keras.datasets.mnist\n",
    "\n",
    "(X_train, y_train), (X_test, y_test) = mnist.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "637bcc76",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The dimension of train data : (60000, 28, 28) (60000,)\n",
      "The dimension of test data : (10000, 28, 28) (10000,)\n"
     ]
    }
   ],
   "source": [
    "# dataset의 차원 확인하기\n",
    "print('The dimension of train data :', X_train.shape, y_train.shape)\n",
    "print('The dimension of test data :', X_test.shape, y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "25cb2c51",
   "metadata": {},
   "outputs": [],
   "source": [
    "# label data를 one-hot encoding으로 변환 \n",
    "y_train = tf.keras.utils.to_categorical(y_train)\n",
    "y_test = tf.keras.utils.to_categorical(y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "75aaff6a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# normalization\n",
    "X_train = X_train/255.0\n",
    "X_test = X_test/255.0"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a1ec70e3",
   "metadata": {},
   "source": [
    "# CNN model 구축  \n",
    "- CNN을 다루기 위해 필요한 용어 파악하기!  \n",
    "    - 1. convolution  \n",
    "    - 2. channel  \n",
    "    - 3. filter or kernel  \n",
    "    - 4. stride  \n",
    "    - 5. padding  \n",
    "    - 6. pooling\n",
    "    \n",
    "- 1. convolution  \n",
    "    - 합성곱 연산을 의미 (행렬곱이 아님을 주의!)\n",
    "    \n",
    "<img src = \"http://deeplearning.stanford.edu/wiki/images/6/6c/Convolution_schematic.gif\">\n",
    "\n",
    "- 2. channel  \n",
    "    - 컬러 사진은 RGB 3 channel로 이루어짐  \n",
    "\n",
    "<img src = \"https://taewanmerepo.github.io/2018/01/cnn/channel.jpg\" width=500>  \n",
    "\n",
    "- 3. filter or kernel  \n",
    "    - 이미지의 특징을 찾아내는 역할  \n",
    "    - 일반적으로 (3,3) 정사각형 행렬로 정의됨  \n",
    "    - CNN에서 학습의 대상!  \n",
    "    - filter는 지정된 간격(stride)으로 이동하면서 이미지와 합성곱하면서 feature map 생성  \n",
    "<img src = \"https://taewanmerepo.github.io/2018/01/cnn/conv.png\">  \n",
    "\n",
    "- 4. stride  \n",
    "    - filter를 얼만큼 순회할지 결정하는 파라미터  \n",
    "    - stride=1로 설정하면 1칸씩 이동하면서 합성곱을 수행하게 됨  \n",
    "<img src = \"https://taewanmerepo.github.io/2018/01/cnn/filter.jpg\">\n",
    "\n",
    "- 5. padding  \n",
    "    - convolution layer에서 filter와 stride로 인해 featrue map의 크기는 input image 보다 크기가 작음  \n",
    "    - convolution layer에서 output의 크기가 줄어드는 것을 방지하는 것이 padding  \n",
    "    - padding은 input data의 외곽에 지정된 픽셀만큼 특정 값으로 채움을 의미  \n",
    "    - 보통 0으로 채움  \n",
    "<img src = \"https://taewanmerepo.github.io/2018/01/cnn/padding.png\">\n",
    "\n",
    "- 6. pooling  \n",
    "    - pooling layer는 convolution layer의 output을 입력으로 받아 크기를 줄여주는 역할  \n",
    "    - 또는 특정 feature를 강조하는 용도로 사용됨!  \n",
    "    - pooling layer로는 max pooling, average pooling 등이 존재  \n",
    "<img src = \"https://taewanmerepo.github.io/2018/02/cnn/maxpulling.png\">  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "cd028432",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 모델 구축하기 전에 이미지 shape 변경\n",
    "X_train = X_train.reshape(-1, 28, 28, 1)\n",
    "X_test = X_test.reshape(-1, 28, 28, 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "f05e46d1",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# CNN 구축\n",
    "model = tf.keras.models.Sequential([\n",
    "    ################# first convolution layer #################\n",
    "    tf.keras.layers.Conv2D(10, kernel_size=(3,3), strides=1, padding='same', activation='relu',\n",
    "                           input_shape=(28,28,1)),\n",
    "    ################# first pooling layer #################\n",
    "    tf.keras.layers.MaxPooling2D(pool_size=(2,2)),\n",
    "    \n",
    "    ################# second convolution layer #################\n",
    "    tf.keras.layers.Conv2D(20, kernel_size=(3,3), strides=1, padding='same', activation='relu'),\n",
    "    ################# second pooling layer #################\n",
    "    tf.keras.layers.MaxPooling2D(pool_size=(2,2)),\n",
    "    \n",
    "    ################# fully connected layer #################\n",
    "    tf.keras.layers.Flatten(),\n",
    "    tf.keras.layers.Dense(100, activation='relu'),\n",
    "    tf.keras.layers.Dense(10, activation='softmax'),\n",
    "    #########################################################\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "e4bdac02",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv2d (Conv2D)             (None, 28, 28, 10)        100       \n",
      "                                                                 \n",
      " max_pooling2d (MaxPooling2D  (None, 14, 14, 10)       0         \n",
      " )                                                               \n",
      "                                                                 \n",
      " conv2d_1 (Conv2D)           (None, 14, 14, 20)        1820      \n",
      "                                                                 \n",
      " max_pooling2d_1 (MaxPooling  (None, 7, 7, 20)         0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " flatten (Flatten)           (None, 980)               0         \n",
      "                                                                 \n",
      " dense (Dense)               (None, 100)               98100     \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 10)                1010      \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 101,030\n",
      "Trainable params: 101,030\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "2aaac1cd",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 학습하기 전 모델 compile하기\n",
    "optimizer = tf.keras.optimizers.SGD(learning_rate=0.01)\n",
    "model.compile(optimizer=optimizer, \n",
    "              loss='categorical_crossentropy',\n",
    "              metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "bb866d5a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "1875/1875 [==============================] - 10s 4ms/step - loss: 0.6006 - accuracy: 0.8257\n",
      "Epoch 2/10\n",
      "1875/1875 [==============================] - 6s 3ms/step - loss: 0.1603 - accuracy: 0.9520\n",
      "Epoch 3/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.1084 - accuracy: 0.9671\n",
      "Epoch 4/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.0848 - accuracy: 0.9742\n",
      "Epoch 5/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.0721 - accuracy: 0.9781\n",
      "Epoch 6/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.0625 - accuracy: 0.9809\n",
      "Epoch 7/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.0560 - accuracy: 0.9833\n",
      "Epoch 8/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.0505 - accuracy: 0.9839\n",
      "Epoch 9/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.0465 - accuracy: 0.9856\n",
      "Epoch 10/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.0426 - accuracy: 0.9869\n"
     ]
    }
   ],
   "source": [
    "# 모델 학습\n",
    "history = model.fit(X_train, y_train, epochs=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "71fc8145",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# loss 및 accuracy 그래프 살펴보기\n",
    "plt.figure(figsize=(10,7))\n",
    "plt.plot(history.history['loss'], label='loss')\n",
    "plt.plot(history.history['accuracy'], label='accuracy')\n",
    "plt.legend(loc='best')\n",
    "plt.show()\n",
    "# loss는 0에 가까울수록, 수렴할수록 잘 훈련되었다는 증거!\n",
    "# accuracy의 경우 1에 가까울수록 잘 훈련되었다는 증거!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "16b2c3ba",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313/313 - 1s - loss: 0.0470 - accuracy: 0.9838 - 921ms/epoch - 3ms/step\n",
      "Error: 0.047042183578014374, Accuracy: 98.38%\n"
     ]
    }
   ],
   "source": [
    "# 모델 평가\n",
    "test_loss, test_acc = model.evaluate(X_test, y_test, verbose=2)\n",
    "print(f'Error: {test_loss}, Accuracy: {test_acc*100:.2f}%')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "b06b0cd1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# test data에 대해 model prediction 진행\n",
    "prediction = model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "2c56fe31",
   "metadata": {},
   "outputs": [],
   "source": [
    "# one-hot encoding label -> integer label\n",
    "prediction_label = np.argmax(prediction, axis=1)\n",
    "test_label = np.argmax(y_test, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "bc34e95c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# prediction_label과 test_label이 같지 않은 인덱스들 뽑기\n",
    "false_index = np.where(prediction_label!=test_label)[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "6512d0a2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1800 with 25 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 잘못 분류된 이미지와 라벨과 score 살펴보기\n",
    "plt.figure(figsize=(20,25))\n",
    "for i in range(25):\n",
    "    plt.subplot(5,5,i+1)\n",
    "    plt.imshow(X_test[false_index[i]], cmap='gray')\n",
    "    plt.title(f'score: {prediction[false_index[i]][prediction_label[false_index[i]]]:.2f}'\n",
    "              f'\\nlabel: {test_label[false_index[i]]},'\n",
    "              f' predicted: {prediction_label[false_index[i]]}\\n',\n",
    "              fontsize=15)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4b3f1291",
   "metadata": {},
   "source": [
    "# Feature Map 살펴보기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "5030c749",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv2d_input (InputLayer)   [(None, 28, 28, 1)]       0         \n",
      "                                                                 \n",
      " conv2d (Conv2D)             (None, 28, 28, 10)        100       \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 100\n",
      "Trainable params: 100\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# 첫 번째 convolution layer output의 feature map 살펴보기\n",
    "extract_fmap1 = tf.keras.models.Model(inputs=model.inputs,\n",
    "                                      outputs=model.layers[0].output)\n",
    "extract_fmap1.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "a58d6725",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "TensorShape([10000, 28, 28, 10])"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# feature map 추출하기\n",
    "fmap1 = extract_fmap1(X_test)\n",
    "fmap1.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "70b0f383",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,5))\n",
    "for i in range(10):\n",
    "    plt.subplot(2,5,i+1)\n",
    "    plt.imshow(fmap1[0,:,:,i])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "645aacdf",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_1\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv2d_input (InputLayer)   [(None, 28, 28, 1)]       0         \n",
      "                                                                 \n",
      " conv2d (Conv2D)             (None, 28, 28, 10)        100       \n",
      "                                                                 \n",
      " max_pooling2d (MaxPooling2D  (None, 14, 14, 10)       0         \n",
      " )                                                               \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 100\n",
      "Trainable params: 100\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# 첫 번째 max pooling layer output의 feature map 살펴보기\n",
    "extract_fmap2 = tf.keras.models.Model(inputs=model.inputs,\n",
    "                                      outputs=model.layers[1].output)\n",
    "extract_fmap2.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "71b6727e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "TensorShape([10000, 14, 14, 10])"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# feature map 추출하기\n",
    "fmap2 = extract_fmap2(X_test)\n",
    "fmap2.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "d7c541a1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,5))\n",
    "for i in range(10):\n",
    "    plt.subplot(2,5,i+1)\n",
    "    plt.imshow(fmap2[0,:,:,i])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "2600304d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_2\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv2d_input (InputLayer)   [(None, 28, 28, 1)]       0         \n",
      "                                                                 \n",
      " conv2d (Conv2D)             (None, 28, 28, 10)        100       \n",
      "                                                                 \n",
      " max_pooling2d (MaxPooling2D  (None, 14, 14, 10)       0         \n",
      " )                                                               \n",
      "                                                                 \n",
      " conv2d_1 (Conv2D)           (None, 14, 14, 20)        1820      \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 1,920\n",
      "Trainable params: 1,920\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# 두 번째 convolution layer output의 feature map 살펴보기\n",
    "extract_fmap3 = tf.keras.models.Model(inputs=model.inputs,\n",
    "                                      outputs=model.layers[2].output)\n",
    "extract_fmap3.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "b7364e92",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "TensorShape([10000, 14, 14, 20])"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# feature map 추출하기\n",
    "fmap3 = extract_fmap3(X_test)\n",
    "fmap3.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "9954db44",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 20 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,10))\n",
    "for i in range(20):\n",
    "    plt.subplot(4,5,i+1)\n",
    "    plt.imshow(fmap3[0,:,:,i])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "d7b50b31",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_3\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv2d_input (InputLayer)   [(None, 28, 28, 1)]       0         \n",
      "                                                                 \n",
      " conv2d (Conv2D)             (None, 28, 28, 10)        100       \n",
      "                                                                 \n",
      " max_pooling2d (MaxPooling2D  (None, 14, 14, 10)       0         \n",
      " )                                                               \n",
      "                                                                 \n",
      " conv2d_1 (Conv2D)           (None, 14, 14, 20)        1820      \n",
      "                                                                 \n",
      " max_pooling2d_1 (MaxPooling  (None, 7, 7, 20)         0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 1,920\n",
      "Trainable params: 1,920\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# 두 번째 max pooling layer output의 feature map 살펴보기\n",
    "extract_fmap4 = tf.keras.models.Model(inputs=model.inputs,\n",
    "                                      outputs=model.layers[3].output)\n",
    "extract_fmap4.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "cdeb72e9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "TensorShape([10000, 7, 7, 20])"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# feature map 추출하기\n",
    "fmap4 = extract_fmap4(X_test)\n",
    "fmap4.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "8bcaaa59",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 20 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,10))\n",
    "for i in range(20):\n",
    "    plt.subplot(4,5,i+1)\n",
    "    plt.imshow(fmap4[0,:,:,i])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "26e563c0",
   "metadata": {},
   "source": [
    "# Homework  \n",
    "- 본인이 하고 싶은 dataset으로 CNN model 학습시키기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "703b5d22",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}