Day_071-使用損失函數_HW.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "cj4tE8e8iUtO"
   },
   "source": [
    "#   \n",
    "\n",
    "# 作業目標:\n",
    "\n",
    "    1. 藉由固定的 dataset, 來驗證不同loss function\n",
    "    2. Dataset 的特性跟我們選用的loss function 對accrancy 的影響\n",
    "    \n",
    "    \n",
    "# 作業重點: \n",
    "    請分別選用 \"MSE\", \"binary _crossentropy\"\n",
    "    查看Train/test accurancy and loss rate\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "R51J4KyyiUsL"
   },
   "source": [
    "# 導入必要的函數"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Qd68OQjciUsM",
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "from keras.datasets import cifar10\n",
    "import numpy as np\n",
    "np.random.seed(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "GxKuzQLQiUsP"
   },
   "source": [
    "# 資料準備"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "MIbFZdTCiUsQ"
   },
   "outputs": [],
   "source": [
    "#取得Keras Dataset\n",
    "(x_img_train,y_label_train),(x_img_test,y_label_test)=cifar10.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "J8CJVjNOiUsS"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train data: images: (50000, 32, 32, 3)  labels: (50000, 1)\n",
      "test  data: images: (10000, 32, 32, 3)  labels: (10000, 1)\n"
     ]
    }
   ],
   "source": [
    "#確認 CIFAR10 Dataset 資料維度\n",
    "print(\"train data:\",'images:',x_img_train.shape,\n",
    "      \" labels:\",y_label_train.shape) \n",
    "print(\"test  data:\",'images:',x_img_test.shape ,\n",
    "      \" labels:\",y_label_test.shape) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "V3P6P4DXiUsV"
   },
   "outputs": [],
   "source": [
    "#資料正規化\n",
    "x_img_train_normalize = x_img_train.astype('float32') / 255.0\n",
    "x_img_test_normalize = x_img_test.astype('float32') / 255.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "vsDyjKYliUsX"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10000, 10)"
      ]
     },
     "execution_count": 78,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#針對Label 做 ONE HOT ENCODE\n",
    "from keras.utils import np_utils\n",
    "y_label_train_OneHot = np_utils.to_categorical(y_label_train)\n",
    "y_label_test_OneHot = np_utils.to_categorical(y_label_test)\n",
    "y_label_test_OneHot.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "RutW2OtgiUsZ"
   },
   "source": [
    "# 建立模型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "aHz592aXiUsa"
   },
   "outputs": [],
   "source": [
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Dropout, Activation, Flatten\n",
    "from keras.layers import Conv2D, MaxPooling2D, ZeroPadding2D"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "hslqrIp0iUse"
   },
   "outputs": [],
   "source": [
    "model = Sequential()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Hqe35hs2iUsi"
   },
   "outputs": [],
   "source": [
    "#卷積層1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "ED7dqe1YiUsm"
   },
   "outputs": [],
   "source": [
    "model.add(Conv2D(filters=32,kernel_size=(3,3),\n",
    "                 input_shape=(32, 32,3), \n",
    "                 activation='relu', \n",
    "                 padding='same'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "n-GRKmPViUso"
   },
   "outputs": [],
   "source": [
    "model.add(Dropout(rate=0.25))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "GlWnnD1uiUsw"
   },
   "outputs": [],
   "source": [
    "model.add(MaxPooling2D(pool_size=(2, 2)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "nawlkt55iUsy"
   },
   "outputs": [],
   "source": [
    "#卷積層2與池化層2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "xxyETiKNiUs1"
   },
   "outputs": [],
   "source": [
    "model.add(Conv2D(filters=64, kernel_size=(3, 3), \n",
    "                 activation='relu', padding='same'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "RqcKWtqGiUs2"
   },
   "outputs": [],
   "source": [
    "model.add(Dropout(0.25))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "BUOOzlqfiUs4"
   },
   "outputs": [],
   "source": [
    "model.add(MaxPooling2D(pool_size=(2, 2)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "LpkppB2siUs7"
   },
   "outputs": [],
   "source": [
    "#建立神經網路(平坦層、隱藏層、輸出層)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "UQOfU54_iUs_"
   },
   "outputs": [],
   "source": [
    "model.add(Flatten())\n",
    "model.add(Dropout(rate=0.25))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "_NV-8sq4iUtD"
   },
   "outputs": [],
   "source": [
    "model.add(Dense(1024, activation='relu'))\n",
    "model.add(Dropout(rate=0.25))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "qVK77krOiUtF"
   },
   "outputs": [],
   "source": [
    "model.add(Dense(10, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [],
   "source": [
    "# #卷積層3與池化層3\n",
    "# model.add(Conv2D(filters=128, kernel_size=(3, 3), \n",
    "#                  activation='relu', padding='same'))\n",
    "# model.add(MaxPooling2D(pool_size=(2, 2)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [],
   "source": [
    "# #卷積層4與池化層4\n",
    "# model.add(Conv2D(filters=256, kernel_size=(3, 3), \n",
    "#                  activation='relu', padding='same'))\n",
    "# model.add(MaxPooling2D(pool_size=(2, 2)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [],
   "source": [
    "# #建立神經網路(平坦層、隱藏層、輸出層)\n",
    "# model.add(Flatten())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [],
   "source": [
    "# #建立全網路連接層\n",
    "# model.add(Dense(512, activation='relu'))\n",
    "# model.add(Dense(512, activation='relu'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [],
   "source": [
    "#建立輸出層\n",
    "model.add(Dense(10, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "ZxzRIjR0iUtH"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_3\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_8 (Conv2D)            (None, 32, 32, 32)        896       \n",
      "_________________________________________________________________\n",
      "dropout_8 (Dropout)          (None, 32, 32, 32)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_8 (MaxPooling2 (None, 16, 16, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_9 (Conv2D)            (None, 16, 16, 64)        18496     \n",
      "_________________________________________________________________\n",
      "dropout_9 (Dropout)          (None, 16, 16, 64)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_9 (MaxPooling2 (None, 8, 8, 64)          0         \n",
      "_________________________________________________________________\n",
      "flatten_3 (Flatten)          (None, 4096)              0         \n",
      "_________________________________________________________________\n",
      "dropout_10 (Dropout)         (None, 4096)              0         \n",
      "_________________________________________________________________\n",
      "dense_7 (Dense)              (None, 1024)              4195328   \n",
      "_________________________________________________________________\n",
      "dropout_11 (Dropout)         (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dense_8 (Dense)              (None, 10)                10250     \n",
      "_________________________________________________________________\n",
      "dense_9 (Dense)              (None, 10)                110       \n",
      "=================================================================\n",
      "Total params: 4,225,080\n",
      "Trainable params: 4,225,080\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "#檢查model 的STACK\n",
    "print(model.summary())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "QJCtpJM0iUtJ"
   },
   "source": [
    "# 載入之前訓練的模型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "JU3A6NzQiUtK",
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "載入模型失敗!開始訓練一個新模型\n"
     ]
    }
   ],
   "source": [
    "try:\n",
    "    model.load_weights(\"SaveModel/cifarCnnModel.h5\")\n",
    "    print(\"載入模型成功!繼續訓練模型\")\n",
    "except :    \n",
    "    print(\"載入模型失敗!開始訓練一個新模型\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "qBbCuziziUtM"
   },
   "source": [
    "# 訓練模型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Q6BiBcseiUtP"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/12\n",
      "293/293 [==============================] - 39s 134ms/step - loss: 0.0901 - accuracy: 0.0980 - val_loss: 0.0901 - val_accuracy: 0.1025\n",
      "Epoch 2/12\n",
      "293/293 [==============================] - 37s 128ms/step - loss: 0.0901 - accuracy: 0.0969 - val_loss: 0.0901 - val_accuracy: 0.1025\n",
      "Epoch 3/12\n",
      "293/293 [==============================] - 38s 131ms/step - loss: 0.0901 - accuracy: 0.1006 - val_loss: 0.0901 - val_accuracy: 0.1025\n",
      "Epoch 4/12\n",
      "293/293 [==============================] - 38s 129ms/step - loss: 0.0901 - accuracy: 0.0982 - val_loss: 0.0901 - val_accuracy: 0.1025\n",
      "Epoch 5/12\n",
      "293/293 [==============================] - 38s 129ms/step - loss: 0.0901 - accuracy: 0.1015 - val_loss: 0.0901 - val_accuracy: 0.1025\n",
      "Epoch 6/12\n",
      "293/293 [==============================] - 38s 129ms/step - loss: 0.0901 - accuracy: 0.1023 - val_loss: 0.0901 - val_accuracy: 0.1025\n",
      "Epoch 7/12\n",
      "293/293 [==============================] - 38s 129ms/step - loss: 0.0901 - accuracy: 0.1001 - val_loss: 0.0901 - val_accuracy: 0.1025\n",
      "Epoch 8/12\n",
      "293/293 [==============================] - 38s 129ms/step - loss: 0.0901 - accuracy: 0.1013 - val_loss: 0.0901 - val_accuracy: 0.1025\n",
      "Epoch 9/12\n",
      "293/293 [==============================] - 38s 129ms/step - loss: 0.0901 - accuracy: 0.1013 - val_loss: 0.0901 - val_accuracy: 0.1025\n",
      "Epoch 10/12\n",
      "293/293 [==============================] - 38s 131ms/step - loss: 0.0901 - accuracy: 0.1039 - val_loss: 0.0901 - val_accuracy: 0.1025\n",
      "Epoch 11/12\n",
      "293/293 [==============================] - 40s 136ms/step - loss: 0.0901 - accuracy: 0.1020 - val_loss: 0.0901 - val_accuracy: 0.1025\n",
      "Epoch 12/12\n",
      "293/293 [==============================] - 39s 133ms/step - loss: 0.0901 - accuracy: 0.1022 - val_loss: 0.0901 - val_accuracy: 0.1025\n"
     ]
    }
   ],
   "source": [
    "model.compile(loss='MSE', optimizer='sgd', metrics=['accuracy'])\n",
    "\n",
    "'''\n",
    "作業:\n",
    "請分別選用 \"MSE\", \"binary _crossentropy\"\n",
    "查看Train/test accurancy and loss rate\n",
    "'''\n",
    "#模型訓練, \"Train_History\" 把訓練過程所得到的數值存起來\n",
    "train_history=model.fit(x_img_train_normalize, y_label_train_OneHot,\n",
    "                        validation_split=0.25,\n",
    "                        epochs=12, batch_size=128, verbose=1)        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "4-GdDgmQiUtS"
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "#定義一個繪圖函數\n",
    "def show_train_history(train_acc,test_acc):\n",
    "    plt.plot(train_history.history[train_acc])\n",
    "    plt.plot(train_history.history[test_acc])\n",
    "    plt.title('Train History')\n",
    "    plt.ylabel('Accuracy')\n",
    "    plt.xlabel('Epoch')\n",
    "    plt.legend(['train_acc', 'val_acc'], loc='upper left')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_train_history('accuracy','val_accuracy')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/12\n",
      "293/293 [==============================] - 40s 138ms/step - loss: 0.3253 - accuracy: 0.1017 - val_loss: 0.3253 - val_accuracy: 0.1025\n",
      "Epoch 2/12\n",
      "293/293 [==============================] - 39s 134ms/step - loss: 0.3252 - accuracy: 0.1036 - val_loss: 0.3252 - val_accuracy: 0.1025\n",
      "Epoch 3/12\n",
      "293/293 [==============================] - 39s 133ms/step - loss: 0.3250 - accuracy: 0.1025 - val_loss: 0.3251 - val_accuracy: 0.1025\n",
      "Epoch 4/12\n",
      "293/293 [==============================] - 69s 237ms/step - loss: 0.3249 - accuracy: 0.1017 - val_loss: 0.3249 - val_accuracy: 0.1025\n",
      "Epoch 5/12\n",
      "293/293 [==============================] - 73s 248ms/step - loss: 0.3247 - accuracy: 0.1023 - val_loss: 0.3248 - val_accuracy: 0.1026\n",
      "Epoch 6/12\n",
      "293/293 [==============================] - 73s 248ms/step - loss: 0.3246 - accuracy: 0.1030 - val_loss: 0.3247 - val_accuracy: 0.1025\n",
      "Epoch 7/12\n",
      "293/293 [==============================] - 71s 242ms/step - loss: 0.3244 - accuracy: 0.1049 - val_loss: 0.3246 - val_accuracy: 0.1025\n",
      "Epoch 8/12\n",
      "293/293 [==============================] - 74s 252ms/step - loss: 0.3243 - accuracy: 0.1053 - val_loss: 0.3245 - val_accuracy: 0.1028\n",
      "Epoch 9/12\n",
      "293/293 [==============================] - 74s 252ms/step - loss: 0.3241 - accuracy: 0.1043 - val_loss: 0.3244 - val_accuracy: 0.1026\n",
      "Epoch 10/12\n",
      "293/293 [==============================] - 71s 242ms/step - loss: 0.3240 - accuracy: 0.1063 - val_loss: 0.3242 - val_accuracy: 0.1030\n",
      "Epoch 11/12\n",
      "293/293 [==============================] - 73s 248ms/step - loss: 0.3238 - accuracy: 0.1060 - val_loss: 0.3241 - val_accuracy: 0.1034\n",
      "Epoch 12/12\n",
      "293/293 [==============================] - 75s 256ms/step - loss: 0.3236 - accuracy: 0.1095 - val_loss: 0.3240 - val_accuracy: 0.1066\n"
     ]
    }
   ],
   "source": [
    "model.compile(loss='binary_crossentropy', optimizer='sgd', metrics=['accuracy'])\n",
    "\n",
    "'''\n",
    "作業:\n",
    "請分別選用 \"MSE\", \"binary _crossentropy\"\n",
    "查看Train/test accurancy and loss rate\n",
    "'''\n",
    "#模型訓練, \"Train_History\" 把訓練過程所得到的數值存起來\n",
    "train_history=model.fit(x_img_train_normalize, y_label_train_OneHot,\n",
    "                        validation_split=0.25,\n",
    "                        epochs=12, batch_size=128, verbose=1)        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAY4AAAEWCAYAAABxMXBSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nOzdd3xV9fnA8c9DBmGvhL33BiEMUUFBlgvrKoqIbS0/V7VatVqttu62to6qKCruighisSIBmW4ZgiTMEFaADBJCwghZz++Pc6PXcAMXck/uvcnzfr3ui5xzz/ec5yCeJ+c7RVUxxhhj/FUj2AEYY4wJL5Y4jDHGnBJLHMYYY06JJQ5jjDGnxBKHMcaYU2KJwxhjzCmxxGHMKRKRCBE5JCJtXTp/RxE55Ma5jQkESxymyvM85Es/JSJy1Gt70qmeT1WLVbWuqu46jVg6i8hxg6dE5B0R+Yvn/CmqWtePc90gIstONQZjKioy2AEY4zbvh7CI7ABuUNXPyjteRCJVtagyYgum6nKfJvDsjcNUeyLyqIi8LyLviUgecK2InCki34hIjojsE5HnRCTKc3ykiKiItPdsv+P5/lMRyRORr0WkQwXi+dlbiYj8RkR2eM6dIiITRaQP8DxwjufNab/n2IaeeDI9Ze4TEfF8d4OIrPDEmg086rm/Hl7XaiEiR0SkyenGb6o+SxzGOH4B/AdoALwPFAG3A7HAWcA44P9OUP4a4M9AY2AX8EggghKR+sC/gNGqWs8Tyw+quh64FfjcU20W6ynyIlAb6AiMBH4DXOd1ymHARiAO+CswC7i2zH0kqGpWIOI3VZMlDmMcX6jqx6paoqpHVXWlqn6rqkWqmgJMB0acoPxsVV2lqoXAu0D/E13M85v+jx/gqhMcrkBvEYlR1X2quqGcc0Z5znOvquZ54n4amOx12C5VneZppzkKvAlcU/pW4jn27RPFbowlDmMcu703RKS7iHwiImkikgs8jPP2UZ40r5+PACds3FbVht4fnN/8fR2XC1wN3AKkicj/RKRrOadtCkQAO7327QRaeW3/7D5V9Uuct6uzRaQ30Bb45ESxG2OJwxhH2Z5OLwOJQGdVrQ88CMhxpSqBqn6qqucDLYBkT2xwfMwZQDHQzmtfW2CP9+l8XOItnOqqycAsVT0WiLhN1WWJwxjf6gEHgcOexuMTtW+4xtNYfbGI1AYKgMM4yQEgHWhd2mjvqSabDTwuInU9DfR3AO+c5DJvA1fgtG+85cJtmCrGEocxvv0BmALk4fyG/36Q4ogA7gb2AVk4jdu3er5bBGwF0kWktKrsZpwEsx1YjtOGccJkoKo7gPVAgap+FeD4TRUktpCTMUZE3gJSVPUvwY7FhD4bAGhMNSciHYEJQJ9gx2LCg1VVGVONicgTwDrg8dOZQsVUT1ZVZYwx5pTYG4cxxphTUi3aOGJjY7V9+/bBDsMYY8LK6tWr96tqXNn91SJxtG/fnlWrVgU7DGOMCSsistPXfquqMsYYc0oscRhjjDklljiMMcackmrRxuFLYWEhqamp5OfnBzuUsBQTE0Pr1q2JiooKdijGmEpWbRNHamoq9erVo3379vy0FIHxh6qSlZVFamoqHTqc9kJ3xpgwVW2rqvLz82nSpIkljdMgIjRp0sTe1oyppqpt4gAsaVSA/d0ZU31V68RhjDFV1ZpdB3hhaTJ5+YUBP7clDmOMqYI+XJPKC0uTiYoI/GPeEkeQ5OTk8OKLL55yuQsuuICcnBwXIjLGVBUlJcrCpHRGdI0jJioi4Oe3xBEk5SWO4uJiH0f/ZP78+TRs2NCtsIwxVcC61Bwy8o4xplczV85fbbvjevvrx0ls2Jsb0HP2bFmfhy7uVe739957L9u2baN///5ERUVRt25dWrRowdq1a9mwYQOXXnopu3fvJj8/n9tvv52pU6cCP827dejQIcaPH8/ZZ5/NV199RatWrfjvf/9LrVq1fF7vlVdeYfr06RQUFNC5c2fefvttateuTXp6OjfeeCMpKSkATJs2jWHDhvHWW2/x1FNPISL07duXt99+O6B/P8YY9yQkpRNZQxjZzZ3EYW8cQfLkk0/SqVMn1q5dyz/+8Q++++47HnvsMTZs2ADAjBkzWL16NatWreK5554jKyvruHNs3bqVW265haSkJBo2bMicOXPKvd5ll13GypUrWbduHT169OC1114D4LbbbmPEiBGsW7eONWvW0KtXL5KSknjsscdYsmQJ69at49lnn3XnL8EY44qFG9I4s1MTGtR2Z4CuvXHACd8MKsvgwYN/NpjuueeeY+7cuQDs3r2brVu30qRJk5+V6dChA/379wdg4MCB7Nixo9zzJyYm8sADD5CTk8OhQ4cYO3YsAEuWLOGtt94CICIiggYNGvDWW29xxRVXEBsbC0Djxo0Ddp/GGHclZ+SRknmYXw1r79o1LHGEiDp16vz487Jly/jss8/4+uuvqV27Nueee67PwXY1a9b88eeIiAiOHj1a7vmvv/56PvroI/r168cbb7zBsmXLyj1WVW2chjFhKiEpHYDRPZu7dg2rqgqSevXqkZeX5/O7gwcP0qhRI2rXrs2mTZv45ptvKny9vLw8WrRoQWFhIe++++6P+0eNGsW0adMAp2E+NzeXUaNGMWvWrB+rx7Kzsyt8fWNM5ViYlEa/Ng1p3iDGtWtY4giSJk2acNZZZ9G7d2/uvvvun303btw4ioqK6Nu3L3/+858ZOnRoha/3yCOPMGTIEEaPHk337t1/3P/ss8+ydOlS+vTpw8CBA0lKSqJXr17cf//9jBgxgn79+nHnnXdW+PrGGPftO3iUdakHGetSb6pSoqrunVxkHPAsEAG8qqpPlvl+OPAM0BeYqKqzvb6bAjzg2XxUVd/07P8lcL/nnJ+o6j0niyM+Pl7LrgC4ceNGevTocbq3ZrC/Q2NCzVtf7+DB/ybx2Z0j6Ny0boXPJyKrVTW+7H7X3jhEJAJ4ARgP9ASuFpGeZQ7bBVwP/KdM2cbAQ8AQYDDwkIg0EpEmwD+AUaraC2gmIqPcugdjjAknCUlpdIqrE5CkcSJuVlUNBpJVNUVVC4CZwATvA1R1h6r+AJSUKTsWWKSq2ap6AFgEjAM6AltUNdNz3GfA5S7eQ9i55ZZb6N+//88+r7/+erDDMsa47OCRQr5JyWZML/caxUu52auqFbDbazsV5w3idMu2AhYA3UWkvWffpUC0rxOIyFRgKkDbtm1PIezw9sILLwQ7BGNMECzelE5xiTK2EhKHm28cvvpz+tug4rOs5+3jJuB94HNgB1Dk6wSqOl1V41U1Pi4uzs/LGmNMeFqYlE7z+jH0bdXA9Wu5mThSgTZe262BvRUtq6ofq+oQVT0T2AxsDUCsxhgTtvILi1m+JZPRPZtRo4b7Y7DcTBwrgS4i0kFEooGJwDw/yyYAYzwN4o2AMZ59iEhTz5+NgJuBVwMeuTHGhJEVWzI5WlhcKdVU4GLiUNUi4FacB/5GYJaqJonIwyJyCYCIDBKRVOBK4GURSfKUzQYewUk+K4GHPfsAnhWRDcCXwJOqusWtezDGmHCwcEM69WMiGdKxcqYHcnXKEVWdD8wvs+9Br59X4lRD+So7A5jhY//VAQ4zLNStW5dDhw4FOwxjTIgpKi5h8cZ0RvVo5sqiTb7YyHFjjAljK3cc4MCRQsb0dHe0uDeb5BDg03shbX1gz9m8D4x/styv//jHP9KuXTtuvvlmAP7yl78gIqxYsYIDBw5QWFjIo48+yoQJE8o9R6lDhw4xYcIEn+V8ratR3hocxpjwk5CURs3IGozoVnm9Ry1xBMnEiRP5/e9//2PimDVrFgsWLOCOO+6gfv367N+/n6FDh3LJJZecdKbamJgY5s6de1y5DRs28Nhjj/Hll18SGxv742SFpWtwzJ07l+LiYqsCMyZMqSqLNqRzTpdYakdX3uPcEgec8M3ALWeccQYZGRns3buXzMxMGjVqRIsWLbjjjjtYsWIFNWrUYM+ePaSnp9O8+Yl7Sqgqf/rTn44rt2TJEp/ravhag8MYE36S9uayJ+cot5/fpVKva4kjiK644gpmz55NWloaEydO5N133yUzM5PVq1cTFRVF+/btfa7DUVZ55WxdDWOqtoSkNGoInN+j8to3wBrHg2rixInMnDmT2bNnc8UVV3Dw4EGaNm1KVFQUS5cuZefOnX6dp7xy5a2r4WsNDmNM+FmYlM6g9o1pXMfnzEuuscQRRL169SIvL49WrVrRokULJk2axKpVq4iPj+fdd9/92boZJ1JeufLW1fC1BocxJrzs2H+Yzel5lTboz5ur63GECluPwx32d2hM8ExfsY3H52/i83vOo03j2q5co9LX4zDGGOOehKR0erWs71rSOBFrHA8j69evZ/LkyT/bV7NmTb799tsgRWSMCYaMvHzW7DrA70d1Dcr1q3XiCLdeR3369GHt2rXBDgNw/u6MMcHx2YYMVGFs78rtTVWq2lZVxcTEkJWVZQ/A06CqZGVlERMTE+xQjKmWFm5Io23j2nRrVi8o16+2bxytW7cmNTWVzMzMkx9sjhMTE0Pr1j7npzTGuCgvv5CvkrOYMqxd0GpMqm3iiIqKokOHDsEOwxhjTsnSzZkUFJcEpRtuqWpbVWWMMeFoYVIasXWjOaNto6DFYInDGGPCxLGiYpZtdpaIjaiEJWLLY4nDGGPCxFfbsjh0rIgxPYNXTQUuJw4RGScim0UkWUTu9fH9cBFZIyJFInJFme+miMhWz2eK1/6rRWS9iPwgIgtEJNbNezDGmFCxMCmNOtERDOvcJKhxuJY4RCQCeAEYD/QErhaRnmUO2wVcD/ynTNnGwEPAEGAw8JCINBKRSOBZ4DxV7Qv8gLOuuTHGVGnFJc7aG+d2b0rNyIigxuLmG8dgIFlVU1S1AJgJ/Gw5O1Xdoao/ACVlyo4FFqlqtqoeABYB4wDxfOqI0w+tPrDXxXswxpiQ8P2uA+w/VBDU3lSl3EwcrYDdXtupnn2nXVZVC4GbgPU4CaMn8JqvE4jIVBFZJSKrbKyGMSbcJSSlERUhnFeJS8SWx83E4avJ399h2j7LikgUTuI4A2iJU1V1n68TqOp0VY1X1fi4uOD/RRtjzOlSVRZuSGdYp1jqxUQFOxxXE0cq0MZruzX+VyuVV7Y/gKpuU2eukFnAsIqHaowxoWtzeh47s46ERDUVuJs4VgJdRKSDiEQDE4F5fpZNAMZ4GsQbAWM8+/YAPUWk9BViNLAxwHEbY0xIWZiUjgic37NpsEMBXJxyRFWLRORWnAd+BDBDVZNE5GFglarOE5FBwFygEXCxiPxVVXuparaIPIKTfAAeVtVsABH5K7BCRAqBnTi9sowxpspKSEpjQNtGNK0XGhOLujpXlarOB+aX2feg188rcaqhfJWdAczwsf8l4KXARmqMMaEp9cARkvbmct94/5aSrgw2ctwYY0LYwqR0AMaESPsGWOIwxpiQlpCURtdmdekQWyfYofzIEocxxoSo7MMFrNyRHTK9qUpZ4jDGmBD12cZ0ShRLHMYYY/yzMCmdVg1r0atl/WCH8jOWOIwxJgQdKSji863O2hvBWiK2PJY4jDEmBK3YksmxohLG9GoW7FCOY4nDGGNCUEJSOg1rRzG4feNgh3IcSxzGGBNiCotLWLwxnVHdmxEZEXqP6dCLyBhjqrlvU7LJzS9ibAhWU4ElDmOMCTkJSWnERNXgnC6huSSEJQ5jjAkhJZ4lYkd0jaNWdHCXiC2PJQ5jjAkhP+w5SFpufsgN+vNmicMYY0LIwqQ0ImoII7uHxtobvljiMMaYEJKQlMbQjo1pWDs62KGUyxKHMcaEiOSMQ2zLPMyYnqFbTQUuJw4RGScim0UkWUTu9fH9cBFZIyJFInJFme+miMhWz2eKZ189EVnr9dkvIs+4eQ/GGFNZFm5IA2B0z9DshlvKtRUARSQCeAFnXfBUYKWIzFPVDV6H7cJZ+vWuMmUbAw8B8YACqz1lDwD9vY5bDXzo1j0YY0xlSkhKp2/rBrRsWCvYoZyQm28cg4FkVU1R1QJgJjDB+wBV3aGqPwAlZcqOBRaparYnWSwCxnkfICJdgKbA527dgDHGVJa0g/ms250T0r2pSrmZOFoBu722Uz37AlX2auB9VVVfJxCRqSKySkRWZWZm+nlZY4wJjkWeaqoxIV5NBe4mDl/zAPt8yJ9m2YnAe+WdQFWnq2q8qsbHxYXm6EtjjCm1cEM6HWPr0Llp3WCHclJuJo5UoI3XdmtgbyDKikg/IFJVV1c0SGOMCbaDRwr5elsWY3o1D7m1N3xxM3GsBLqISAcRicZ5Q5jnZ9kEYIyINBKRRsAYz75SV3OCtw1jjAknSzdnUFSiIbn2hi+uJQ5VLQJuxXngbwRmqWqSiDwsIpcAiMggEUkFrgReFpEkT9ls4BGc5LMSeNizr9RVWOIwxlQRCUlpNK1Xk/6tGwY7FL+41h0XQFXnA/PL7HvQ6+eVONVQvsrOAGaU813HAIZpjDFBk19YzLLNmVw2oBU1aoR+NRXYyHFjjAmqL7bu52hhcVh0wy1licMYY4IoISmNejGRDO3YJNih+M0ShzHGBElRcQmfbUxnZPemREeGz+M4fCI1xpgqZtXOAxw4UhjykxqWZYnDGGOCZGFSOtGRNTi3mwuDlHP3QVoi+J5co0IscRhjTBCoKglJaZzTOZY6NV3o4Lr6DXjpbDgc+CmXLHEYY0wQbNiXy56co+4M+lOFxNnQ/myoG/iVBC1xGGNMECQkpVND4PweLiSOfesgKxn6XHHyY0+DJQ5jjAmChUlpxLdrTJO6NQN/8sTZUCMKelwS+HNjicMYYyrdzqzDbErLc6eaqqQEEudC51FQu3Hgz48fiUNEbvVMNGiMMVXK80u28odZ6/jkh33k5hdW2nUXJqUDuDNafPe3kJsKvd2ppgL/5qpqjrPs6xqcuaMSyls8yRhjwkXS3oM8tXALURHCnDWpRNYQBndozMjuTRnZvSkd49xbF2PhhjR6tKhPm8a1A3/yxNkQWQu6jQ/8uT1OmjhU9QER+TPO1Oa/Ap4XkVnAa6q6zbXIjDHGRX9fsJkGtaJYdte5JGceYvHGDJZsSufRTzby6Ccb6RBbh/O6NWVUj6YMat84YCO7M/OOsWrnAW4b2SUg5/uZ4iJI+gi6jYOa7iU+vzoPq6qKSBqQBhQBjYDZIrJIVe9xLTpjjHHB19uyWL4lkz9d0J1GdaIZVKcxg9o35t7x3dmdfYSlmzNYvDGDd77dyYwvt1O3ZiTndIllZPemnNe9KbEVaNBevDEdVZeqqbYvgyP7Xa2mAj8Sh4jcBkwB9gOvAneraqGI1AC2ApY4jDFhQ1V5csEmWjSI4boz2x/3fZvGtbnuzPZcd2Z7jhQU8WVyFks2pbNkUwafJqYhAv1aN/yxSqtXy/qntGpfQlIabRrXokeLegG8K4/ED6FmA+gyOvDn9uLPG0cscJmq7vTeqaolInKRO2EZY4w7EpLSWbc7h79d3oeYqIgTHls7OpLRPZsxumczVJWkvbks2ZTB4k0ZPP3ZFv61aAvN68dwnieJnNW5CbWjy3+s5uUX8mVyFpPPbBf4JWIL82Hjx04X3EgXuvh68SdxzAd+XH1PROoBPVX1W1Xd6FpkxhgTYEXFJfwjYROd4upw+QCfa8iVS0To3aoBvVs14LZRXcjMO8ayzRks2ZTBvLV7eO+7XURH1mBYpyZOlVa3psc1fi/fkklBcYk71VTJi+BYLvS5PPDnLsOfxDENGOC1fdjHPp9EZBzwLBABvKqqT5b5fjjwDNAXmKiqs72+mwI84Nl8VFXf9OyPBp4HzgVKgPtVdY4f92GMqebmrEllW+ZhXrp2IJERFWvsjqtXkyvj23BlfBsKikpYuSP7xwb2B/+bBCTRrVk9zuvuNLCf0aYhCUnpNKkTzcB2LoxwWD8b6sRB++GBP3cZ/iQO8e5+66mi8qdtJAJ4ARgNpOJ06Z2nqhu8DtsFXA/cVaZsY+AhIB5QYLWn7AHgfiBDVbt62lncGeFijKlS8guLeXrRVvq3acjYAA+8i46swVmdYzmrcywPXtyTlMxDTpXWxgxe/TyFl5Zvo2HtKI4WFHNp/1ZEBHqJ2GN5sGUBnDEZIlxdERzwL3GkeBrIp3m2bwZS/Cg3GEhW1RQAEZkJTAB+TByqusPzXUmZsmOBRaqa7fl+ETAOeA/4NdDdU74Ep9HeGGNO6K2vd5CWm8/Tv+wf+PaFMjrG1aVjXF1uOKcjufmFfL5lP4s3pfP9rhyuGtQm8BfcNB+K8l2bm6osfxLHjcBzONVGCiwGpvpRrhWw22s7FRjiZ1y+yrYSkYae7UdE5FxgG3CrqqaXPYGITC2Ns23btn5e1hhTFR08WsgLS7cxomscZ3aq3CVa68dEcWHfFlzYt4V7F0mcAw3aQOvB7l3Dy0kr+VQ1Q1UnqmpTVW2mqteoaoYf5/aV0v0dcV5e2UigNfClqg4AvgaeKifu6aoar6rxcXEuLJJijAkbLy/fxsGjhdwzrluwQwm8I9mwbTH0vgxqVM70g/60VcQAvwF6ATGl+1X11ycpmgp4v5O1Bvb6GVcqTuO3d9llQBZwBJjr2f+BJzZjjPEpPTefGV9u55J+LenVskGwwwm8Df+FkiLo7X5vqlL+pKe3cearGgssx3mI5/lRbiXQRUQ6eHpCTQTm+RlXAjBGRBp5Jlgcw09zZH3MT0llFF5tJsYYU9Zzi7dSVKz8YUzXYIfijsQ50KQLNO9baZf0J3F0VtU/A4c9XWIvBPqcrJCqFgG34iSBjcAsVU0SkYdF5BIAERkkIqnAlcDLIpLkKZsNPIKTfFYCD5c2lAN/BP4iIj8Ak4E/+H+7xpjqZPv+w8xcuZtrhrSlXZM6wQ4n8HL3wo4vnEZxlxv8vfnTOF4613COiPTGma+qvT8nV9X5OAMIvfc96PXzSpw3GF9lZ+DMxlt2/07A/Y7Kxpiw99TCzdSMrMHv3JhQMBQkfQSo63NTleVP4pjuqS56AKeqqS7wZ1ejMsaYClqfepBPftjH70Z2Jq6eu1NwBE3ibGjRD2I7V+plT5g4PAPscj0D71YAHSslKmOMqaC/J2yiUe0opg6voo+t7BTYsxpGP1zplz5hG4dngN2tlRSLMcYExJfJ+/l8635uOa8z9WKigh2OOxI9My31uqzSL+1P4/giEblLRNqISOPSj+uRGWPMaVBV/rZgE60a1uLaoe2CHY571s+BtmdCQxdGop+EP20cpeM1bvHap1i1lTEmBH2amMYPqQf5xxV9TzptethKT4LMjXCBz/HPrvNn6dgOlRGIMcZUVFFxCU8lbKZrs7pcdorTpoeVxDkgEdDrF0G5vD8jx6/ztV9V3wp8OMYYc/pmrUolZf9hXrkuPvAz0IYKVSdxdDwX6sQGJQR/qqoGef0cgzNaew1gicMYEzKOFhTzzGdbGNiuEef3aBrscNyzZzUc2AHDg7dqtz9VVb/z3haRBjjTkBhjTMh446sdZOQd4/lrBrg+bXpQrZ8NETWhR/BW7j6dqRSPAFV0GKYxJhwdPFLItGXJjOzelMEdqnCnz5JiSPoQuoyGmOBN2OhPG8fH/DQdeg2gJzDLzaCMMeZUvLg8mbxjRdw9tgpOm+5t55dwKL3SFmwqjz9tHN79vYqAnaqa6lI8xpgQoapsTs/j+105jO7ZjNi6oTltx76DR3njyx1c2r8VPVrUD3Y47lo/G6LrQpexQQ3Dn8SxC9inqvkAIlJLRNqXLvtqjKk6SkqUdak5LEhKIyExjR1ZRwCYtmwbb/56MB1iQ2+G2ecWb6VElTtHV9Fp00sVFThrb3S7AKJrBzUUfxLHB8Awr+1iz75Bvg83xoSTouISvtuRTUJiGglJ6aTl5hNZQxjWOZapwzvRsmEMd85ax+XTvuK1KfGc0bZRsEP+0bbMQ8xalcrkoe1o0zi4D1PXbVsC+TlBr6YC/xJHpKoWlG6oaoFnYSZjTJg6VlTMl8n7WZCYxqIN6Rw4UkhMVA1GdI3jj727MbJbMxrU/mmOpzk3DWPKjO+4+pVveP7qAZzfs1kQo//JUwmbiYmswa0jK3d22KBInA21GkHH84IdiV+JI1NELlHVeQAiMgHY725YxphAO3ysiGWbM1mQlMbSTRkcOlZEvZqRjOrRlHG9mzO8axy1o30/EjrE1mHOTcP4zZsrmfr2Kh65tDeThgR3Hqi1u3P4NDGN20d1Cdn2l4ApOAKb5kPfKyEy+L+3+5M4bgTeFZHnPdupgM/R5GWJyDjgWSACeFVVnyzz/XDgGaAvMFFVZ3t9NwVnDRCARz2rDyIiy4AWwFHPd2NUNcOfeIypbnKOFPDZxgwWJKaxYmsmBUUlNKkTzcX9WjC2V3OGdYolOtK/Xvlx9Woyc+pQbnl3DffPTWRfTj5/GNM1KGMmVJW/fbqJJnWi+W1VnTbd25YFUHi40hdsKo8/AwC3AUNFpC4gqurPeuOISATwAjAaJ9msFJF5quq9Rvgu4HrgrjJlGwMPAfE4XYFXe8oe8BwySVVX+ROHMdVNRm4+CRvSSUhM4+uULIpLlJYNYpg0pC3jejUnvn3j056Oo3Z0JK9cF88DHyXy/NJk9h3M58nL+xAVcTpDwk7f51v383VKFg9d3JO6Nf35/TfMJc6Bus2h3bCTH1sJ/BnH8Tjwd1XN8Ww3Av6gqg+cuCSDgWRVTfGUmwlMAH5MHKU9s0SkpEzZscCi0nXGRWQRMA54z497Mqba2ZV1hAVJ+1iQmMaaXTkAdIytw/8N78i43s3p06pBwN4MIiNq8MRlfWjRoBZPf7aFjLx8pl07sNIe4CUlzrTprRvV4pohbSvlmkF1NAe2LoRBN0CN0Jjt15//0uNV9U+lG6p6QEQu4KdqpPK0AnZ7bacCQ/yMy1fZVl7br4tIMTAHpxpLKUNEpgJTAdq2rQb/uEy1oqpsST/EgsQ0FiSlsXFfLgC9WtbnD6O7Mq53czo3retaNZKIcPv5XWjRIIb75q7nly9/zevXD6Jp/RhXruftf+v3kbQ3l39d1Y+akaHxIHXVpv9BcUHIVFOBf4kjQkRqquoxcMZxAP60RPn6F3vcA6djDZwAACAASURBVP40yk5S1T0iUg8ncUzGx4SLqjodmA4QHx/v73WNCXnvr9zFS8tT2L7/MCIwsG0jHriwB2N7Na/0LqlXDWpDXP2a3PLuGn7x4le8+evBdG5a17XrFRaX8M+Fm+nevB4T+rc6eYGqYP1saNQBWg0IdiQ/8qdi8h1gsYj8RkR+AywC3vSjXCrgvTRVa2Cvn3GVW1ZV93j+zAP+g1MlZky1kJxxiD/NTaReTCSPXtqbb+8bxeybhnHDOR2DNo7hvG5NmTl1KMeKirnipa9YtSPbtWvNXLmbnVlHuGdct6o7bbq3Q5mwfTn0vhxCaOLGkyYOVf078CjQA2eeqgWAP/3wVgJdRKSDZ9zHRGCen3ElAGNEpJGnTWUMkCAikSISCyAiUcBFQKKf5zQm7D356UZqRUXw+vWDuHZou0qpGvJH39YN+fCms2hUO5pJr37LgsS0gF/jSEERzy3eyuD2jTmvWxWeNt3bho9AS0Ji0J83f7tCpAElwOU463FsPFkBVS0CbsVJAhuBWaqaJCIPi8glACIySERSgSuBl0UkyVM2G3gEJ/msBB727KuJk0B+ANYCe4BX/L1ZY8LZV9v289nGDG4+rxNNQnDcQtsmtZlz0zB6tqzPTe+u5s2vdgT0/DO+2E5m3jH+OL5b1Z423dv62dC0JzTtEexIfqbcNg4R6YrzlnA1kAW8j9Md1+9hi6o6H5hfZt+DXj+vxKmG8lV2BjCjzL7DwEB/r29MVVFSojw+fyOtGtbi12eF7mrOjetE858bhvK7977noXlJ7DuYzz1ju1GjgtVKBw4X8PLyFM7v0YyB7arwtOnecnbD7m9g5J+DHclxTvTGsQnn7eJiVT1bVf+NM0+VMaaSfbR2D4l7crl7bDdiokK7J1Gt6AheunYAk4a05aXl27hz1loKisr2uD81Ly5L5nBBEfeMq+LTpntLnOP82fvy4Mbhw4l6VV2O88axVEQWADPx3dvJGOOi/MJi/pGwmT6tGnBJv5bBDscvkRE1ePTS3rRsWIt/JGwm89Axpl07kPoxUScvXMaenKO8+fVOLhvQmq7N6rkQbYhKnA2t4qFx6L1hlvvGoapzVfWXQHdgGXAH0ExEponImEqKz5hq77UvtrPvYD73X9ijwlU+lUlEuOW8zvzzyn58m5LNVS99TdrB/FM+zzOLtoDC78+vRguPZm6BtPUh1yheyp9eVYdV9V1VvQinPWItcK/rkRlj2H/oGNOWbWN0z2YM7dgk2OGclssHtmbG9YPYnX2Ey178ki3pfs1aBMDW9DzmrEll8pntaN2oik+b7i1xDiDQ89JgR+LTKU0wo6rZqvqyqo50KyBjzE+e+WwLRwuLuXd892CHUiHDu8Yx68YzKSxRrpj2Fd+mZPlV7h8Jm6kdHckt51WDadNLqTrVVO3Phvotgh2NT5U7M5kxxm/JGXm8991uJg1pS6c490ZjV5ZeLRsw9+ZhxNWryeTXvuOTH/ad8PjVOw+wcEM6U4d3pHGd4E8lXmn2rYOs5JCtpgJLHMaErCfmb6J2VAS3j6o6dfutGzljPfq2bsCt763htS+2+zxO1ZnIMLZuTX5zdug1DrsqcTbUiIIelwQ7knJZ4jAmBH2VvJ/FmzK4+bzOITnYryIa1o7mnRuGMLZncx753wYe/d8GSkp+Pp3csi2ZfLc9m9tGdaZOdZg2vVRJCSTOhc6joHbojlexxGFMiCkpUR7zDPb71Vntgx2OK2KiInhh0gCuH9aeV7/Yzm0zv+dYkTNMrKRE+fuCzbRtXJuJg6rZzNa7v4Xc1JAcu+GtGqVyY8LD3O/3kLQ3l2cn9g/5wX4VEVFDeOjinrRsGMPj8zeRmXeM6ZPjWbo5g437nPv3d3XCKiNxNkTWgm4XBDuSE7LEYUwIOVpQzFMLN9O3dQMu7hseg/0qQkSYOrwTzerHcNcH67jy5a84UlBMzxb1q8X9/0xxESR9BN3GQc3Q7gxhicOYEPLaFynsO5jPM7/sH1aD/SpqQv9WxNWryf+9tZq8Y0W88ave1er+Adi+DI7sD6kFm8pjicOYEJGZ5wz2G9OzGUPCdLBfRQzrFMvcW4axZmcOI7rGBTucyrd+DtRsAF1GBzuSk7LEYUyIeOazLRwrKgn7wX4V0blpPTo3rUbzUZUqzHeWiO1xCUSGfi+6atbyZExo2pqex3vf7WLSkLZ0rAKD/cwpSl4Ex3Kh92XBjsQvljiMCQFPfLqJOtGR3H5+12CHYoJh/WyoEwcdRgQ7Er9Y4jAmyL5M3s+STRncMrJz9ZpawziO5cGWBc6EhhHh0XrgauIQkXEisllEkkXkuBl1RWS4iKwRkSIRuaLMd1NEZKvnM8VH2XkiYuuNm7BWXKI89okz2O/6Ye2DHY4Jhk3zoSg/pOemKsu1xCEiEcALwHigJ3C1iPQsc9gu4HrgP2XKNgYeAoYAg4GHRKSR1/eXAYfcit2YyjL3+z1s2JfLPeNCf2U/45LEOdCgDbQeHOxI/ObmG8dgIFlVU1S1AGcFwQneB6jqDlX9ASi7ruRYYJFnGvcDwCJgHICI1AXuBB51MXZjXHe0oJinEjbTr5oM9jM+HMmGbYuh1y+gRvi0HLgZaStgt9d2qmdfRcs+AvwTOHKiE4jIVBFZJSKrMjMz/bysMZXn1c9TSMvN5/4Le1a/wW7GseG/UFIUVtVU4G7i8PV/gvrY53dZEekPdFbVuSc7gapOV9V4VY2Pi6uGg4lMSMvIy2fa8m2M7dWMwR1CdxZU47LEOdCkCzTvG+xITombiSMVaOO13RrYW8GyZwIDRWQH8AXQVUSWVThSYyrZM59tpaCohD+Oq76D/aq93L2w4wvnbUPC643TzcSxEugiIh1EJBqYCMzzs2wCMEZEGnkaxccACao6TVVbqmp74Gxgi6qe60LsxrhmS3oeM7/bxbVD29lgv+osaS6gIT+Fui+uJQ5VLQJuxUkCG4FZqpokIg+LyCUAIjJIRFKBK4GXRSTJUzYbpy1jpefzsGefMWHvifkbqVMzktuq0Mp+5jQkznGqqGLD79+Bq6NNVHU+ML/Mvge9fl6JUw3lq+wMYMYJzr0D6B2QQI2pJF9s3c/SzZncN767DfarzrJTYM9qGP1wsCM5LeHT/8uYMFfsWdmvdaNaTLHBftVb4hznz17hMTdVWZY4jKkkH65JZeO+XO4Z190G+1V36+dA2zOhYZuTHxuCLHEYUwlKV/br16YhF/dtEexwTDClJ0HmxrBsFC9liSNEFBSV8OZXO8jIzQ92KMYFr3yeQnruMR64sAcSZl0vTYAlzgGJcCY1DFOWOELE619u56F5SUx85RtLHlVMRl4+Ly3fxrhezRnU3gb7VWuqTuLoOALqhu/AZEscISAjL59/L0mmb+sGpB3M5+pXviEz71iwwzIB8vQiz2C/aryyn/HYsxoO7AiLdcVPxBJHCPjHgs0cKyrmmV/25/XrB7E3J59rXvmG/YcseYS7zWl5vL9yF5PPbEeH2DrBDscE2/rZEFETelwU7EgqxBJHkK3bncMHq1P51Vkd6BhXlyEdm/D6rwaReuAo17zyDVmWPMLaE596BvuNDL9BXibADmc5o8W7jIaYBsGOpkIscQSRqvLXj5OIrRvN70Z2/nH/0I5NeO36eHZlH2HSq9+SfbggiFGa0/X51kyWbc7kdyM708gG+1VfmVvg49/D0z3hUBoM/FWwI6owSxxB9N+1e1mzK4d7xnanXkzUz74b1imW16YMYvv+w1zzyjccsOQRVkpX9rPBftWUKmxbCu9eCS8MgrX/gb5Xwc3fQJfzgx1dhVniCJLDx4p44tON9GnVgCsG+px1hbM6x/LqlHhS9h9m0qvfknPEkke4mLMmlU1pefxxXHdqRtpgv2qj6Bh8/w5MOwvevhT2fg/n/gnuSIJL/g1NewQ7woCwxBEk05ZtIz33GH+55MSL+JzTJY5XrosnOfOQJY8wcaSgiH8u3Ez/Ng25yAb7VQ+H98Oyv8HTveG/tzj7JrzoJIxz/xjWXW99scQRBLuyjjD98xQm9G/JwHYn79c/omsc0ycPZGv6ISa/9h0HjxRWQpTmdL2yYrsN9qsuMjbBvN/Bv3rCssehZX+Y/BHc9CWcMQkiawY7QldY4giCx+dvJEKEe0+hX/+53Zry0uQBbErL5boZ33LwqCWPUJSRm8/LK7Yxvndz4m2wX9WkCsmL4e3L4MUh8MMs6H8N3PIdTPoAOp0XdgsznSpLHJXsq+T9LEhK4+ZzO9GiQa1TKjuyezOmTRrIhn25XDfjO3LzLXmEmn8t2kJhsa3sVyUV5sOat+DFM+GdyyA9EUY+AHdsgIufgbhuwY6w0ljiqERFxSX89eMNtG5Ui98O73ha5zi/ZzNeuGYASXsOMmXGd+RZ8ggZm9JymbVqN5OHtqe9DfarOg5lwNIn4OleTrVUjUi49CX4/XoYfjfUaRLsCCudq4lDRMaJyGYRSRaRe318P1xE1ohIkYhcUea7KSKy1fOZ4rV/gYisE5EkEXlJRMKmy8p73+1ic3oe91/Qo0LTao/p1ZznrxnA+tSDXP/6Sg4dKwpglOZ0PTF/E3VrRv5sTI4JY+kbnIbup3vB8iehdTxM+Rhu/Bz6X11l2y/84Vri8DzQXwDGAz2Bq0WkZ5nDdgHXA/8pU7Yx8BAwBBgMPORZexzgKlXth7P6XxzOsrMhL+dIAf9ctIUzOzZhXO/mFT7fuN7N+ffVZ7B2dw7Xz/jOkkeQrdiSyfItmfxuZBcb7BfOSkpg6yJ461KYdqazbsYZk+HWVXDN+9BheJVvv/CHm0vHDgaSVTUFQERmAhOADaUHeJZ/RURKypQdCywqXWdcRBYB44D3VDXXK/ZoQF28h4B5etEWco8W8uDFPQPW02Z8nxY8p3DbzO/59esref1Xg6hT09XVgI0PxSXK4/M30qZxLa4b1i7Y4ZjTUXgU1s2Eb6bB/s1QrwWMetAZ5V3bOjmU5eZTphWw22s7FecN4nTLtirdEJEEnMT0KTC7YmG6b3NaHu98u4tJQ9rRo0X9gJ77wr4tKFHl9pnf8+s3nORRO9qSR2U5VlTM3z7dzKa0PJ6/5gwb7BducvfCqtdh1WtwJAta9INfTIdev4BIe3Msj5tPGF+/Vvv7dnDCsqo6VkRigHeBkcCi404gMhWYCtC2bVs/Lxt4pfNR1a0ZyZ2ju7pyjYv7taRElTveX8tv3ljFjOsHUSvaHmBu+yE1h7s/+IHN6XlcPbgNF/axwX4hr6QE9n0PWxbClgWwby0g0O0COPNmaHeWVUX5wc3EkQp4L6jbGth7CmXPLVN2mfcBqpovIvNwqr+OSxyqOh2YDhAfHx+06qyFG9L5alsWf72kl6t13xP6t0IV7py1lt+8uZLXpljycMuxomKeW7yVl5anEFs3mtevH8R53ZsGOyxTnmN5zrxRWxOchHE4AxBoM9ipjup5KTTpFOwow4qbiWMl0EVEOgB7gInANX6WTQAe92oQHwPcJyJ1gXqquk9EIoELgM8DHHfA5BcW8+gnG+jarC6Thrj/1nPpGa0oUeUPH6zjt2+t4tUp8RXqvWWO90NqDnd9sI4t6Ye4cmBrHrioJw1qRZ28oKlc2SmwJcH57PgCSgqhZgPoPAq6joPO51fLbrSB4lriUNUiEbkVJwlEADNUNUlEHgZWqeo8ERkEzAUaAReLyF9VtZeqZovIIzjJB+Bhz75mwDwRqek55xLgJbfuoaJe+2I7u7OP8u4NQ4iMqJwhM5cNaE2Jwt2zneTxynWWPALhWFExz362lZdXpBBXtyav/2oQ53WrJm8Z+bkQXQdqhPC/o+JC2PWNU/20dSHs3+Lsj+0KQ290kkWbIRBhST4QRDUsOiVVSHx8vK5atapSr5l2MJ+R/1zG2Z1jmX5dfKVeG2DWyt3cM+cHRnSN4+XJAy15VMDa3Tnc/cE6tmYc4qp45y2jfkwVfgCVlMDeNc5DeEsCpP3grFrXuKNTpdOks+fj+blOXHDaBQ5nQfIiJ8bkxXDsIEREO+0UXcdB1zFOzOa0ichqVT3uAWbdb1zy9wWbKCpW7r8wONMoXzWoDSWq3Pvhem56ZzUvTR5oPX5OUX5hMc98tpXpK7bRrH4Mb/xqEOdW1beM/FzYtsTzEF4EhzNBakCboc604AV5kJUC+7c6x5R4zVhQs36ZhNLZk2Q6Q0wAexGqQnrST28Vu78DFOo2g54XO8mi47lQs17grml8ssThgjW7DvDh93u4+dxOtGsSvKknJg5uS4nCn+au5+Z31vDitQMsefhp7W6nLSM54xATB7XhTxf2qHpvGVnbPG8VC2DnV1BSBDENnaVNu4x12gN8jWEoLoKDu53yWcnOJ3sb7P7WWVPbu/NknaY/fzsp/bNRB4iKOXmMhUdh+wpPnAshN9XZ3/IMGPFH6DoWWvSHGjZ7UmWyqqoAKylRfvHil+w7mM+Su86lbggMyHvnm5088FEi5/doxouTBhAdaf+Tlcf7LaN5/RieuLwvI7pWkbUUigpg19eeRuMFzsMeIK6H8wDuOhZaD4aICvybLcyHA9t/SihZ235KMIczvA4UaNimzFtKJyex1IhwRm9vXQgpy6HoKETVcWad7ToWuoyBehWffcGcnFVVVZI5a1JZl3qQf13VLySSBsC1Q9tRosqD/03i1v+s4YVJA4iqpMZ6V5WUQO6enz+ksj0PqZxdThVG2SqUJp2hYVufjaTf7zrAXR+sY1vmYa4e3Ib7LqgCbxmHMj3tAAsgeYlT5RRREzqcA0Nvch7CjQI42j0qxlnlztdKd/kHf0ok2V5vK+tmwrHc449v2A4GTnFibH92tZ4bKtSExpOtisjLL+RvC5yV3y7t3+rkBSrRdWe2p6RE+cvHG/jdf77n39ecER7JQ9UZ0ftjcij9pDgPn6L8n46Nqu0kihb9oPtFzqymWcmQ+CHk5/x0XI1I56HkSSSFDTvywY6aPL9OkXoteOvXgxkerm8Zqk5jdukAtz2rAXWm0Oh9macdYITTS6qyxTSAVgOcT9mYD2f+lPwLDjtvF7FdbTBeiLLEEUAvLN3G/kPHeHVK/AmXgw2W68/qQLHCI//bwO0zv+fZiSGUPI7ledWZl6k7zz/403E1Ip368SadnYeLdw+fei3Kf9AcyfaRfLZRkrKcqOJ8rgGuqQlaUgtZ3AnW+nhTCdU5iwoOO1U6pQPc8vYC4jygz/uTU73TvG/oPoRFoG5T59NuWLCjMX6wxBEgO/YfZsYX27l8QGv6t2kY7HDK9ZuzO6CqPPrJRuB7nrqyX+XNbVV0DA7sOO7hTVYyHEr/+bEN2jjJoM+VP08ODdqeXh187cZQe7AzWhinLeNfi7bw2q5k+tQ/ysNnx9A3JhMpjSc9CTZ94jQYl4ppWCaZdPqpB1HNuqf/93I6cnb9NMBt+wooPgbR9TztAOOcBu66VbQHmAk6axw/kdfGQtZWvw7Nyy+isLiEhrWjCcGXjeMcLSzmSEExESLUjYkk0u2gVZ3qIvWaCLl27PEP4SadoXEHiDq11RFPxeqdB7h79jpSMg9zzZC23De+O/V8tWUUFzoPaF/VZKW9e0rFNKy8AXJaAkcPOD837ugZszAW2g6ziflMQFnj+OnofD40733Sw/bmHOWzjRkMaNuQxq0aVEJgFVcLyDmYz5fJ+8k/XEz/tg3p1aKBu7UZtRpDbBcnSTTuBLUq980sv7CYfy7czKtfbKdlg1q885shnN0ltvwCEVGehNYJZ6Z/LwVHnGktSpNJ3j5XYz9Oow5Owoi1RaNM5bM3jgoqLC5h/LOfU1hcwsI7hofdOImcIwXc9+F6Pk1MY1inJvzrqv40b+BH//ows3pnNnd/8AMp+w8zaUhb7rugR8j0ejMmVJX3xhEiLaPh6+2vd5KccYgHLuwZdkkDoGHtaF6cNIC/Xd6H73flMO7ZFSxITAt2WAFztKCYR/+3gSte+ppjRSW8e8MQHvtFH0saxlSA/d9TAVmHjvH0Z1s4p0ss5/cI34ZIEeGXg9oyqH1jbp+5lhvfWc3Vg9vw54t6hvWiUKt2ZHP37B/Yvv8wk4e244/ju1vCMCYA7P+iCvjnoi0cKSjmwYsCtxxsMHWMq8ucm4bxr0VbeHnFNr7dns1zE8+gd5i025TKOVLAc4uTef2r7bRqWIv//HYIwzqdoC3DGHNKLHGcpqS9B3nvu11MObM9XZpVnUnVoiNrcO/47gzvGsud76/jFy9+yR/GdGPqOR1DcmyKt+zDBbz2RQpvfrWTQ8eKmDy0HfeO727rsBsTYPZ/1GlQVR7+eAMNa0Vxx/nuLAcbbMM6xfLp7edw34frefLTTXy+NZN/XhmaDef7Dx3jlc9TePvrnRwtLObCPi343cgudGtedRK6MaHEEsdpmL8+jW+3Z/Popb1pUDvM5zI6gUZ1opl27QBmrdrNX+ZtYNyzK/jb5X0Z2ys0JpjLyMtn+vIU3vl2JwVFJVzSryW3juxM56aWMIxxkyWOU5RfWMzj8zfSvXk9rh7s/nKwwVa24fz/3l7N1YPb8ueLegSt4TztYD4vLd/Ge9/toqhEmdC/Jbec15lOcZU8etuYasrV7rgiMk5ENotIsojc6+P74SKyRkSKROSKMt9NEZGtns8Uz77aIvKJiGwSkSQRedLN+H15eXkKe3KO8pdLehER4nX+gVTacH7jiE7MXLmLi/79BYl7Dp68YADtzTnKnz9KZPjfl/LONzuZ0L8li+8cwb+u6m9Jw5hK5NqvjCISAbwAjAZSgZUiMk9VN3gdtgu4HrirTNnGwENAPM6qMKtFZB5wDHhKVZeKSDSwWETGq+qnbt2Ht705R5m2PJkL+7RgaMfqt9C9r4bzu8Z047cuN5zvzj7CtOXb+GDVbgCuGNiGm8/tRJvGtV27pjGmfG7WNQwGklU1BUBEZgITgB8Th6ru8HxXUqbsWGCRqmZ7vl8EjFPV94ClnrIFIrIGaO3iPfzME59uQhXuu6B7ZV0yJHk3nD/x6SZWuNRwvjPrMC8u3cacNanUEOGXg9pw07mdadXQvXmsjDEn52biaAXs9tpOBYZUoOzPFrgQkYbAxcCzvk4gIlOBqQBt21a8LeK77dl8vG4vt43qQutG9ptuacP5+yt389ePA9twvn3/YZ5fksxHa/cQUUO4dmg7/m9ER1o0sIRhTChwM3H4qrvwd2KsE5YVkUjgPeC50jea4w5WnQ5MB2euKj+v61NxifLXj5No0SCGG0d0rMipqhQRYeLgtgzuEJiG8+SMQ7ywNJn/rt1DdGQNrh/Wnv8b3pGm9UOvC7Ax1ZmbiSMVaOO13RrYewplzy1TdpnX9nRgq6o+U4H4/PbBqt0k7c3l2Yn9w3oKDreUNpz/c9Fmpq9I4dvtWac04nxLeh7/XpLM/37YS0xkBL89pyM3nNORuHq2VKgxocjNp+BKoIuIdAD2ABOBa/wsmwA8LiKNPNtjgPsARORRoAFwQ2DD9S03v5B/JGwmvl0jLunXsjIuGZaiI2tw3/gejOgSx52znIbzu8d244azy28437gvl38v2cr89WnUiY7gxhGduOHsDjSpawnDmFDmWuJQ1SIRuRUnCUQAM1Q1SUQeBlap6jwRGQTMBRoBF4vIX1W1l6pmi8gjOMkH4GHPvtbA/cAmYI1nfqjnVfVVt+7juc+2kn2kgDcvGVwl5qNy27DOPzWcPz5/Eyu27OefV/WjmVd1U+Kegzy3eCsLN6RTr2YkvxvZmV+f1YFGdWwRImPCga3HcQLbMg8x9ukVXDGwNU9e3teFyKouVf2x4TwmqgZPXt6X5vVjeG7xVhZvyqB+TCS/PrsDvxrWoUqPvjcmnNkKgKfhkf9toFZUBHeN7RbsUMJOacP5oA6N+b2n4RygQa0o/jC6K1POak99X8u1GmNCniWOchQVl9CtWT3O7RpHrNW5n7ZOnobzN77aDsA1Q9rZmhjGhDmrqjLGGOOTLR1rjDEmICxxGGOMOSWWOIwxxpwSSxzGGGNOiSUOY4wxp8QShzHGmFNiicMYY8wpscRhjDHmlFSLAYAikgnsPM3iscD+AIYTSqryvUHVvj+7t/AVTvfXTlXjyu6sFomjIkRkla+Rk1VBVb43qNr3Z/cWvqrC/VlVlTHGmFNiicMYY8wpscRxctODHYCLqvK9QdW+P7u38BX292dtHMYYY06JvXEYY4w5JZY4jDHGnBJLHOUQkXEisllEkkXk3mDHE0gi0kZElorIRhFJEpHbgx1ToIlIhIh8LyL/C3YsgSQiDUVktohs8vz3OzPYMQWSiNzh+TeZKCLviUhMsGM6XSIyQ0QyRCTRa19jEVkkIls9fzYKZoynyxKHDyISAbwAjAd6AleLSM/gRhVQRcAfVLUHMBS4pYrdH8DtwMZgB+GCZ4EFqtod6EcVukcRaQXcBsSram8gApgY3Kgq5A1gXJl99wKLVbULsNizHXYscfg2GEhW1RRVLQBmAhOCHFPAqOo+VV3j+TkP5+HTKrhRBY6ItAYuBF4NdiyBJCL1geHAawCqWqCqOcGNKuAigVoiEgnUBvYGOZ7TpqorgOwyuycAb3p+fhO4tFKDChBLHL61AnZ7badShR6s3kSkPXAG8G1wIwmoZ4B7gJJgBxJgHYFM4HVPNdyrIlIn2EEFiqruAZ4CdgH7gIOqujC4UQVcM1XdB84vcEDTIMdzWixx+CY+9lW5fssiUheYA/xeVXODHU8giMhFQIaqrg52LC6IBAYA01T1DOAwYVrV4Yunvn8C0AFoCdQRkWuDG5XxxRKHb6lAG6/t1oTxK7MvIhKFkzTeVdUPgx1PAJ0FXCIiO3CqGEeKyDvBDSlgUoFUVS19O5yNk0iqivOB7aqaqaqFwIfAsCDHFGjpItICwPNnRpDjOS2WOHxbCXQRkQ4iEo3TQDcvyDEFjIgITj35/7d3Py82hTxpAQAAAoNJREFUhXEcxz+fkEayMSU1MQuyUFhIykb8CdIkaZKV8mMlsbGxYGMhNpSFmpSUWChEKdFY+Bl2EooyC0lJTB+L81xumJoznev48X7V6T4993b7nu6t7/Oc55zv8zTJkbbjaVKSfUkGkgyq+t2uJ/knRq1J3kh6aXtJ6Vov6UmLITXthaTVtmeV/+h6/UOL/8VFScOlPSzpQouxTNn0tgP4EyX5YnuHpMuq7uw4leRxy2E1aY2kLZIe2b5f+vYnudRiTJicnZJGyoDmmaStLcfTmCSjts9Juqvqzr97+ovLc9g+I2mtpH7bryQdkHRI0lnb21Qlyo3tRTh1lBwBANTCpSoAQC0kDgBALSQOAEAtJA4AQC0kDgBALSQOoAG2x23f7zoae6Lb9mB3hVWgbTzHATTjY5IVbQcB/A7MOIAesv3c9mHbd8qxqPQvtH3N9sPyuqD0z7N93vaDcnRKbkyzfbLsVXHFdl9rJ4X/HokDaEbfD5eqhrree59klaRjqir3qrRPJ1kmaUTS0dJ/VNKNJMtV1aHqVCxYLOl4kqWS3kna0OPzASbEk+NAA2x/SDL7F/3PJa1L8qwUlnyTZK7tMUnzk3wu/a+T9Nt+K2kgyaeu7xiUdLVs/iPbeyXNSHKw92cG/IwZB9B7maA90Wd+5VNXe1ysT6JFJA6g94a6Xm+X9i193xZ1s6SbpX1N0nbp277pc35XkMBkMWoBmtHXVWlYqvYF79ySO9P2qKqB2qbSt0vSKdt7VO3q16lyu1vSiVI9dVxVEnnd8+iBGljjAHqorHGsTDLWdixAU7hUBQCohRkHAKAWZhwAgFpIHACAWkgcAIBaSBwAgFpIHACAWr4CW6pORPLj7IMAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_train_history('accuracy','val_accuracy')"
   ]
  }
 ],
 "metadata": {
  "colab": {
   "name": "Day71-使用損失函數_HW.ipynb",
   "provenance": [],
   "version": "0.3.2"
  },
  "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}