x1001000/cifar10_part1_training_on_colab.ipynb

## cifar10_part1_training_on_colab.ipynb
{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "cifar10_part1_training_on_colab.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true,
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/gist/x1001000/41c3a70b24d9c555d2f53f73199f3c57/cifar10_part1_training_on_colab.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "metadata": {
        "id": "L5DG-6wxa1uC",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Data"
      ]
    },
    {
      "metadata": {
        "id": "BLYIPvYpa52i",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- 資料都是數據化的多維陣列  \n",
        "資料常要視覺化以幫助理解\n",
        "\n",
        "1. 匯入多維陣列運算要用的函式庫 numpy，以下簡稱為 np  \n",
        "匯入作圖要用的函式庫 matplotlib.pyplot，以下簡稱為 plt"
      ]
    },
    {
      "metadata": {
        "id": "9CNV6V0IZbXq",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import matplotlib.pyplot as plt"
      ],
      "execution_count": 1,
      "outputs": []
    },
    {
      "metadata": {
        "id": "eK86NyZPa_VK",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- [CIFAR-10](https://www.cs.toronto.edu/~kriz/cifar.html) 資料集有10類各6000張32x32的彩圖，預先被切成50000筆訓練圖片及其類別，和10000筆測試圖片及其類別\n",
        "\n",
        "4. x aka feature, image 32x32x3 in this case  \n",
        "y aka label, classification 0~9 in this case\n",
        "\n",
        "5. x 是四維陣列，陣列中有50000/10000個三維(三個數字)陣列  \n",
        "y 是二維陣列，陣列中有50000/10000個一維(一個數字)陣列，一維陣列又稱向量，而這個向量是個一維(數字為1)向量，就是包含一個純量的向量"
      ]
    },
    {
      "metadata": {
        "id": "gJQVlj2UZn45",
        "colab_type": "code",
        "outputId": "0c89ff94-cc25-45aa-a021-08574faceef3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 136
        }
      },
      "cell_type": "code",
      "source": [
        "from keras.datasets import cifar10\n",
        "#data = cifar10.load_data()\n",
        "(x_train, y_train), (x_test, y_test) = cifar10.load_data()\n",
        "print(x_train.shape)\n",
        "print(y_train.shape)\n",
        "print(x_test.shape)\n",
        "print(y_test.shape)"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz\n",
            "170500096/170498071 [==============================] - 9s 0us/step\n",
            "(50000, 32, 32, 3)\n",
            "(50000, 1)\n",
            "(10000, 32, 32, 3)\n",
            "(10000, 1)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "4vlVGDywK6_y",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- 準備一個字典，方便將類別的數字翻成英文"
      ]
    },
    {
      "metadata": {
        "id": "GYEim37OHwDT",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "thing = {0:\"airplane\", 1:\"automobile\", 2:\"bird\", 3:\"cat\", 4:\"deer\", 5:\"dog\", 6:\"frog\", 7:\"horse\", 8:\"ship\", 9:\"truck\"}"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "metadata": {
        "id": "ysWimfgyx6ZQ",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- 看一下 training data 的 5566~5571 筆的圖片及其類別\n",
        "\n",
        "8. 5566可以換個數字試試"
      ]
    },
    {
      "metadata": {
        "id": "cCOWVOifA7SQ",
        "colab_type": "code",
        "outputId": "8e21d191-af83-4f3e-b04b-525c41c505ae",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 340
        }
      },
      "cell_type": "code",
      "source": [
        "for i in range(6):\n",
        "  ax = plt.subplot(2, 3, 1+i)\n",
        "  ax.imshow(x_train[5566+i])\n",
        "  ax.set_title(thing[y_train[5566+i][0]])\n",
        "  ax.set_xticks([])\n",
        "  ax.set_yticks([])"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAcwAAAFDCAYAAAC3PT6LAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nCOHBXziT5lfhBmCGAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f02cfb53c88>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "k9oyCAqZ-YHE",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Normalization"
      ]
    },
    {
      "metadata": {
        "id": "R-oODbaBMg_h",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- training data 的第5566筆的第0列的第0~31個像素的RGB值"
      ]
    },
    {
      "metadata": {
        "id": "gqaJyTIhQmNC",
        "colab_type": "code",
        "outputId": "59f38a30-668c-42a3-cec1-040f8e9f9dea",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 561
        }
      },
      "cell_type": "code",
      "source": [
        "x_train[5566, 0, :, :]"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[253, 253, 253],\n",
              "       [253, 253, 253],\n",
              "       [254, 254, 254],\n",
              "       [254, 254, 254],\n",
              "       [254, 254, 254],\n",
              "       [253, 253, 253],\n",
              "       [252, 252, 252],\n",
              "       [251, 251, 251],\n",
              "       [251, 251, 251],\n",
              "       [252, 251, 251],\n",
              "       [251, 252, 252],\n",
              "       [246, 254, 253],\n",
              "       [243, 253, 252],\n",
              "       [241, 253, 251],\n",
              "       [239, 253, 250],\n",
              "       [237, 252, 249],\n",
              "       [234, 252, 250],\n",
              "       [231, 253, 254],\n",
              "       [229, 253, 254],\n",
              "       [226, 252, 253],\n",
              "       [222, 251, 252],\n",
              "       [220, 251, 250],\n",
              "       [218, 249, 250],\n",
              "       [214, 245, 250],\n",
              "       [210, 241, 249],\n",
              "       [207, 237, 250],\n",
              "       [205, 234, 251],\n",
              "       [201, 230, 251],\n",
              "       [198, 228, 249],\n",
              "       [196, 227, 246],\n",
              "       [192, 224, 243],\n",
              "       [190, 221, 240]], dtype=uint8)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 5
        }
      ]
    },
    {
      "metadata": {
        "id": "hIwNjNTyVrOL",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- Normalization就是把0~255變成0~1"
      ]
    },
    {
      "metadata": {
        "id": "aTLvQnUlfSua",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "x_train = x_train/255\n",
        "x_test = x_test/255"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "metadata": {
        "id": "YyfSya0sWNJk",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- Normalized training data 的第5566筆的第0列的第0~31個像素的RGB值"
      ]
    },
    {
      "metadata": {
        "id": "OUa476ynhVUI",
        "colab_type": "code",
        "outputId": "cdad363c-6c97-454b-883d-ebb759e091cb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 561
        }
      },
      "cell_type": "code",
      "source": [
        "x_train[5566, 0, :, :]"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[0.99215686, 0.99215686, 0.99215686],\n",
              "       [0.99215686, 0.99215686, 0.99215686],\n",
              "       [0.99607843, 0.99607843, 0.99607843],\n",
              "       [0.99607843, 0.99607843, 0.99607843],\n",
              "       [0.99607843, 0.99607843, 0.99607843],\n",
              "       [0.99215686, 0.99215686, 0.99215686],\n",
              "       [0.98823529, 0.98823529, 0.98823529],\n",
              "       [0.98431373, 0.98431373, 0.98431373],\n",
              "       [0.98431373, 0.98431373, 0.98431373],\n",
              "       [0.98823529, 0.98431373, 0.98431373],\n",
              "       [0.98431373, 0.98823529, 0.98823529],\n",
              "       [0.96470588, 0.99607843, 0.99215686],\n",
              "       [0.95294118, 0.99215686, 0.98823529],\n",
              "       [0.94509804, 0.99215686, 0.98431373],\n",
              "       [0.9372549 , 0.99215686, 0.98039216],\n",
              "       [0.92941176, 0.98823529, 0.97647059],\n",
              "       [0.91764706, 0.98823529, 0.98039216],\n",
              "       [0.90588235, 0.99215686, 0.99607843],\n",
              "       [0.89803922, 0.99215686, 0.99607843],\n",
              "       [0.88627451, 0.98823529, 0.99215686],\n",
              "       [0.87058824, 0.98431373, 0.98823529],\n",
              "       [0.8627451 , 0.98431373, 0.98039216],\n",
              "       [0.85490196, 0.97647059, 0.98039216],\n",
              "       [0.83921569, 0.96078431, 0.98039216],\n",
              "       [0.82352941, 0.94509804, 0.97647059],\n",
              "       [0.81176471, 0.92941176, 0.98039216],\n",
              "       [0.80392157, 0.91764706, 0.98431373],\n",
              "       [0.78823529, 0.90196078, 0.98431373],\n",
              "       [0.77647059, 0.89411765, 0.97647059],\n",
              "       [0.76862745, 0.89019608, 0.96470588],\n",
              "       [0.75294118, 0.87843137, 0.95294118],\n",
              "       [0.74509804, 0.86666667, 0.94117647]])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "metadata": {
        "id": "FPHSz-xuhZ3n",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# One-hot encoding"
      ]
    },
    {
      "metadata": {
        "id": "hHl2Yb0RZIbk",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- training data 的第0筆是個一維向量(一對中括號間有一個數字)"
      ]
    },
    {
      "metadata": {
        "id": "QjX8E6cI_ig0",
        "colab_type": "code",
        "outputId": "4f79dd53-37d4-4b3a-b3cc-e041f5ce3728",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "cell_type": "code",
      "source": [
        "y_train[0]"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([6], dtype=uint8)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "metadata": {
        "id": "YvSe0et7ZkWv",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- 獨熱編碼就是把一維向量變成獨熱的N維向量(一對中括號間有N個數字，只有一個是1，其他是0)"
      ]
    },
    {
      "metadata": {
        "id": "U9w9-mKv-3I1",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "from keras.utils import np_utils\n",
        "y_train_onehot = np_utils.to_categorical(y_train)\n",
        "y_test_onehot = np_utils.to_categorical(y_test)"
      ],
      "execution_count": 9,
      "outputs": []
    },
    {
      "metadata": {
        "id": "XbULrz8SyNda",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- 獨熱編碼的 training data 的第0筆是個10維向量(一對中括號間有10個數字)"
      ]
    },
    {
      "metadata": {
        "id": "1rBd8dLrAGpc",
        "colab_type": "code",
        "outputId": "820b23c2-4c30-4dbe-b75c-632c72bbab58",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "cell_type": "code",
      "source": [
        "y_train_onehot[0]"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([0., 0., 0., 0., 0., 0., 1., 0., 0., 0.], dtype=float32)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        }
      ]
    },
    {
      "metadata": {
        "id": "xaXGVtb2FaCQ",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Model"
      ]
    },
    {
      "metadata": {
        "id": "K27ixz0BBeDZ",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- 初學者用高階的Keras函式庫(之後再用TensorFlow函式庫)  \n",
        "初學者用順序式模型(之後再用函數式模型)"
      ]
    },
    {
      "metadata": {
        "id": "ub7Iwo_0A-I1",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "from keras.models import Sequential\n",
        "model = Sequential()"
      ],
      "execution_count": 11,
      "outputs": []
    },
    {
      "metadata": {
        "id": "B5hRX0xgL7WV",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- 匯入要用的神經網路層\n",
        "\n",
        "17. 卷積層 => 池化層 => 卷積層 => 池化層 => 卷積層 => 池化層 => 攤平層 => 全連接層 => 全連接層  \n",
        "放棄層藉由隨機放棄更新一部份神經元，避免模型過度擬合(overfitting)訓練資料，反而降低理解力  \n",
        "試試註解掉放棄層，看看有何不同\n",
        "\n",
        "18. 層就是一排神經元，輸入上一層的輸出\n",
        "\n",
        "19. 卷積層建構函數的引數：  \n",
        "filters是濾鏡組數，就是特徵圖數，就是神經元數，就是線性函數的個數  \n",
        "kernel_size是濾鏡的形狀，Conv2D的kernel_size就會是2D的  \n",
        "padding='same'，讓卷積結果圖不會小一圈  \n",
        "input_shape是輸入這一層眾神經元們的陣列形狀，只有第一層要填  \n",
        "activation='relu'，relu是REctified Linear Unit的縮寫，y=f(x)=max(0,x)，y等於0和x中較大者，換句話說，當x>0時y等於x，否則y等於0\n",
        "\n",
        "20. 池化層建構函數的引數：  \n",
        "pool_size就是從多大的池子取一個值，池化目的是壓縮資料量\n",
        "\n",
        "21. 放棄層建構函數的引數：  \n",
        "rate就是隨機放棄更新的比例\n",
        "\n",
        "22. 攤平層把上一層輸出的所有特徵圖像素排成一直線，以便銜接全連接層，再銜接至降至10個神經元的全連接層，輸出10維向量\n",
        "\n",
        "23. 輸出層的activation='softmax'，使10個輸出值的和，為機率的和，為1"
      ]
    },
    {
      "metadata": {
        "id": "YO1ldsVPF1mB",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "from keras.layers import Conv2D, MaxPooling2D, Dropout, Flatten, Dense\n",
        "\n",
        "model.add(Conv2D(filters=32,\n",
        "                 kernel_size=(3,3),\n",
        "                 padding='same',\n",
        "                 input_shape=(32,32,3), \n",
        "                 activation='relu'))\n",
        "model.add(MaxPooling2D(pool_size=(2,2)))\n",
        "model.add(Dropout(rate=0.25))\n",
        "\n",
        "model.add(Conv2D(filters=64,\n",
        "                 kernel_size=(3,3),\n",
        "                 padding='same',\n",
        "                 activation='relu'))\n",
        "model.add(MaxPooling2D(pool_size=(2,2)))\n",
        "model.add(Dropout(rate=0.25))\n",
        "\n",
        "model.add(Conv2D(filters=128,\n",
        "                 kernel_size=(3,3),\n",
        "                 padding='same',\n",
        "                 activation='relu'))\n",
        "model.add(MaxPooling2D(pool_size=(2,2)))\n",
        "model.add(Dropout(rate=0.25))\n",
        "\n",
        "model.add(Flatten())\n",
        "model.add(Dense(512, activation='relu'))\n",
        "model.add(Dropout(rate=0.25))\n",
        "\n",
        "model.add(Dense(10, activation='softmax'))"
      ],
      "execution_count": 12,
      "outputs": []
    },
    {
      "metadata": {
        "id": "GUM7CwStj7sG",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- 列印神經網路架構\n",
        "\n",
        "25. 模型參數Param個數：  \n",
        "( 3x3濾鏡形狀 x 3面RGB + 1個bias ) x 32面特徵圖 = 896  \n",
        "( 3x3濾鏡形狀 x 32面特徵圖 + 1個bias ) x 64面特徵圖 = 18496  \n",
        "( 3x3濾鏡形狀 x 64面特徵圖 + 1個bias ) x 128面特徵圖 = 73856  \n",
        "( 2048個輸入 + 1個bias ) x 512個神經元 = 1049088  \n",
        "( 512個輸入 + 1個bias ) x 10個神經元 = 5130"
      ]
    },
    {
      "metadata": {
        "id": "jMCSNBfij6zK",
        "colab_type": "code",
        "outputId": "36123ce0-46fc-467f-b8cf-5fc45cf6f10f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 578
        }
      },
      "cell_type": "code",
      "source": [
        "model.summary()"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_1 (Conv2D)            (None, 32, 32, 32)        896       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 16, 16, 32)        0         \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 16, 16, 32)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 16, 16, 64)        18496     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 8, 8, 64)          0         \n",
            "_________________________________________________________________\n",
            "dropout_2 (Dropout)          (None, 8, 8, 64)          0         \n",
            "_________________________________________________________________\n",
            "conv2d_3 (Conv2D)            (None, 8, 8, 128)         73856     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_3 (MaxPooling2 (None, 4, 4, 128)         0         \n",
            "_________________________________________________________________\n",
            "dropout_3 (Dropout)          (None, 4, 4, 128)         0         \n",
            "_________________________________________________________________\n",
            "flatten_1 (Flatten)          (None, 2048)              0         \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 512)               1049088   \n",
            "_________________________________________________________________\n",
            "dropout_4 (Dropout)          (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 10)                5130      \n",
            "=================================================================\n",
            "Total params: 1,147,466\n",
            "Trainable params: 1,147,466\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "99AZyEjOymo3",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Train"
      ]
    },
    {
      "metadata": {
        "id": "j7SuXq9svRZL",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- [Before training a model, you need to configure the learning process, which is done via the compile method. It receives three arguments:](https://keras.io/getting-started/sequential-model-guide/)\n",
        ">1. loss='categorical_crossentropy'，針對分類問題的loss算法\n",
        ">2. optimizer='adam'，[Adam — latest trends in deep learning optimization](https://towardsdatascience.com/adam-latest-trends-in-deep-learning-optimization-6be9a291375c)  \n",
        ">3. metrics=['accuracy']，訓練過程顯示準確率"
      ]
    },
    {
      "metadata": {
        "id": "oBLPWY1UwhlT",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer='adam',\n",
        "              metrics=['accuracy'])"
      ],
      "execution_count": 14,
      "outputs": []
    },
    {
      "metadata": {
        "id": "HNXO4jKQ0b7L",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- 用訓練資料的feature(x_train)和label(y_train)讓模型去學習(fit)  \n",
        "batch_size = 學習(更新Param)一波的樣本數  \n",
        "epochs = 學習訓練資料全部樣本的回合數  \n",
        "validation_split = 切多少比例的訓練資料，不做訓練，做驗證"
      ]
    },
    {
      "metadata": {
        "id": "7iFyQA_Jytkx",
        "colab_type": "code",
        "outputId": "4bdfadf5-165f-46da-9ce2-d2ca8472d08d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 734
        }
      },
      "cell_type": "code",
      "source": [
        "learning = model.fit(x_train,\n",
        "                     y_train_onehot,\n",
        "                     batch_size=128,\n",
        "                     epochs=20,\n",
        "                     validation_split=0.2)"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 40000 samples, validate on 10000 samples\n",
            "Epoch 1/20\n",
            "40000/40000 [==============================] - 13s 336us/step - loss: 1.6934 - acc: 0.3787 - val_loss: 1.3531 - val_acc: 0.5242\n",
            "Epoch 2/20\n",
            "40000/40000 [==============================] - 9s 228us/step - loss: 1.2846 - acc: 0.5408 - val_loss: 1.1291 - val_acc: 0.6062\n",
            "Epoch 3/20\n",
            "40000/40000 [==============================] - 9s 229us/step - loss: 1.1153 - acc: 0.6023 - val_loss: 0.9813 - val_acc: 0.6553\n",
            "Epoch 4/20\n",
            "40000/40000 [==============================] - 9s 228us/step - loss: 1.0032 - acc: 0.6451 - val_loss: 0.8939 - val_acc: 0.6892\n",
            "Epoch 5/20\n",
            "40000/40000 [==============================] - 9s 226us/step - loss: 0.9330 - acc: 0.6724 - val_loss: 0.8816 - val_acc: 0.6945\n",
            "Epoch 6/20\n",
            "40000/40000 [==============================] - 9s 227us/step - loss: 0.8754 - acc: 0.6934 - val_loss: 0.8052 - val_acc: 0.7208\n",
            "Epoch 7/20\n",
            "40000/40000 [==============================] - 9s 227us/step - loss: 0.8171 - acc: 0.7110 - val_loss: 0.7810 - val_acc: 0.7318\n",
            "Epoch 8/20\n",
            "40000/40000 [==============================] - 9s 223us/step - loss: 0.7678 - acc: 0.7292 - val_loss: 0.7583 - val_acc: 0.7420\n",
            "Epoch 9/20\n",
            "40000/40000 [==============================] - 9s 221us/step - loss: 0.7258 - acc: 0.7446 - val_loss: 0.7330 - val_acc: 0.7441\n",
            "Epoch 10/20\n",
            "40000/40000 [==============================] - 9s 221us/step - loss: 0.6935 - acc: 0.7538 - val_loss: 0.7080 - val_acc: 0.7533\n",
            "Epoch 11/20\n",
            "40000/40000 [==============================] - 9s 222us/step - loss: 0.6574 - acc: 0.7672 - val_loss: 0.6754 - val_acc: 0.7667\n",
            "Epoch 12/20\n",
            "40000/40000 [==============================] - 9s 221us/step - loss: 0.6216 - acc: 0.7794 - val_loss: 0.7191 - val_acc: 0.7515\n",
            "Epoch 13/20\n",
            "40000/40000 [==============================] - 9s 221us/step - loss: 0.5932 - acc: 0.7896 - val_loss: 0.6825 - val_acc: 0.7661\n",
            "Epoch 14/20\n",
            "40000/40000 [==============================] - 9s 225us/step - loss: 0.5767 - acc: 0.7944 - val_loss: 0.6942 - val_acc: 0.7610\n",
            "Epoch 15/20\n",
            "40000/40000 [==============================] - 9s 221us/step - loss: 0.5436 - acc: 0.8066 - val_loss: 0.6676 - val_acc: 0.7737\n",
            "Epoch 16/20\n",
            "40000/40000 [==============================] - 9s 221us/step - loss: 0.5249 - acc: 0.8140 - val_loss: 0.6667 - val_acc: 0.7757\n",
            "Epoch 17/20\n",
            "40000/40000 [==============================] - 9s 221us/step - loss: 0.5070 - acc: 0.8187 - val_loss: 0.6748 - val_acc: 0.7707\n",
            "Epoch 18/20\n",
            "40000/40000 [==============================] - 9s 221us/step - loss: 0.4714 - acc: 0.8319 - val_loss: 0.6667 - val_acc: 0.7797\n",
            "Epoch 19/20\n",
            "40000/40000 [==============================] - 9s 220us/step - loss: 0.4634 - acc: 0.8337 - val_loss: 0.6761 - val_acc: 0.7777\n",
            "Epoch 20/20\n",
            "40000/40000 [==============================] - 9s 220us/step - loss: 0.4396 - acc: 0.8418 - val_loss: 0.6669 - val_acc: 0.7796\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "oaXQtjBRFSza",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- 畫出學習歷程的模型推論準確率  \n",
        "藍色曲線是模型對訓練資料的推論準確率  \n",
        "綠色曲線是模型對驗證資料的推論準確率  \n",
        "約在第10學習週期開始過度擬合"
      ]
    },
    {
      "metadata": {
        "id": "FC7y_zB-3YGO",
        "colab_type": "code",
        "outputId": "43409c00-a2d5-4524-a72e-68822f6fbc67",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 378
        }
      },
      "cell_type": "code",
      "source": [
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.plot(learning.history['acc'])\n",
        "plt.plot(learning.history['val_acc'])\n",
        "plt.legend(['training data learned', 'training data not learned'])"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f02c5606940>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 16
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAFYCAYAAAB6RnQAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4yLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvNQv5yAAAIABJREFUeJzs3Xd829W9//GXhmXJljwl27Edxzte\n2YskJISQFMoKUEaAAiUUWigU2tKW5gdNbwdtKe2FS1vaS4FyW0agJEChEEYSyCQ7eGR4x9vykpdk\na3x/fyhxMBk2seQhf56Phx+W9JXP9xzZ5M053/M9R6UoioIQQgghxgz1SFdACCGEEF+OhLcQQggx\nxkh4CyGEEGOMhLcQQggxxkh4CyGEEGOMhLcQQggxxmhHugKDZbV2+LS8yMgQWlu7fVrmaBCI7QrE\nNkFgtkvaNHYEYrsCsU0Wi+m0r4/bnrdWqxnpKvhFILYrENsEgdkuadPYEYjtCsQ2ncm4DW8hhBBi\nrJLwFkIIIcYYCW8hhBBijJHwFkIIIcYYCW8hhBBijJHwFkIIIcYYCW8hhBBijJHwHqLNmz8a9Huf\nfPL31NbWnPH4Qw993xdVAuDhh3/Evn17znh869aPcTqdgyrr2muvoLt7ZBY+uOOOW6iurh6Rcwsh\nxGgl4T0EdXW1fPjhhkG///77f0B8fMIZj//mN3/wRbUG5ZVXXhx0eAshhBhdxszyqKPRH/7wWw4d\nKuT555/B4/FQW1tDXV0tTzzxZ379659jtTZit9tZteouFi5cxL333sX3v/8jNm36iK6uTo4dq6Sm\npprvfvcHzJ+/kMsuu4h33vmIe++9izlz5rFv3x7a2tr47W//G7PZzM9//gj19XVMmTKVjRs/ZP36\n//Srz4svvsDmzR9iNsfQ1dUFQGNjA7/4xU8BcLlcPPzwf5Gff5CiogIefPC7PPnk0/zlL09RVFRI\nb28vV131Na644qrTtrepycqvf/0LXC4narWaH//4EeLi4nj55X+yefNHeDwe5s9fyKpVd/Hss3/t\n+zxWrbqLN974FyqVmsrKcpYsuYhVq+6ivLyM//7vx1CpVISEhLB69c8wmUw88cTvKCjIJylpEi6X\n/A+GEEJ8UcCE96sbS9h9uHHQ79doVLjdylnfMycrhuuXpp/x+I033sK6da9y++138uyzf8XlcvLn\nP/+N1tYW5s49j69+9XJqaqp55JGHWLhwUb+fbWxs4PHH/4edO7fz5puvM3/+wn7HQ0NDefLJp3n6\n6af45JONxMcn0tvbw//+79/Ztm0Lr776cr/3d3R0sH79v3j//Q3U17dy/fXeAG5ubuL22+9k5szZ\nvP32m6xb9xr33fc9/va3v/D44/+Dx+MhLi6e++77Pj09Dq6//qozhvczzzzNypU3M2fOPHbs2MoL\nL/yNH//4YQD+/Oe/oVaruf76Fdxww00AfZ/Hvn17KCoq5KWXXsfj8XDddVewatVdPPHE7/jhD1cz\ncWIS69a9xrp1r7J48YXk53/GM8+8gNXayMqVV5/1dySEECOtvauX0hobPS4387JjUalUfj9nwIT3\naJCdnQuAyRTGoUOFvPXWOlQqNe3ttlPeO3XqdABiYmLo7Ow85fi0aTP6jttsNiory5kyZRoA8+cv\nRKPpv4ZvTU0VKSmpBAcHExISyuTJ2QBERUXzxBOP8+yzf6Wjo73v9ROCg4Npb7fx7W+vQqvV0tbW\nesb2FRR8xrFjlbzwwrN4PB4iIiIB0Ov13HvvXWg0Gtra2mhvb+/3eQBMnpyFXq/vV15RUSG//e0v\nAXA6nWRn51BRUUZOTh5qtZrY2LizXmYQQojh5vZ4qG7sorTWRmmNjZIaG9Y2R9/x7ElRhIfq/F6P\ngAnv65emn7WX/EUWi8nnO5UFBQUB8MEH79He3s6f/vQ32tvb+eY3bznlvZ8PX0U5dQTgi8cVRUGt\n9r6mUqlO+T87RVFQqdSfe+4B4Nln/8q8eedx1VXXsmnTh2zfvrXfz+3fv5d9+/bwxz/+L1qtluXL\n+48QfJ5WG8QvfvFbzGZz32v19XWsXfsizz33IiEhIdxyy/WnfB5fbM8Jer2ep576a7+2bNz4IWr1\nyecej+eM9RFCCH/rtDv7Qrq0xkZ5XQc9Tnff8ZBgLVNSo0lLCCMvJXpYghsCKLxHglqtxu12n/J6\nW1sbEybEo1ar+fjjjT6ZGJaQkNg3s33Xrp2nnDchIZHKynJ6e3vp6urkyJFDfXVJSEhEURS2bv0Y\nt9sbhiqVt+42WxsxMbFotdq+406ns1/wnpCTk8eWLZu5+upr2bt3N83NzSQlTSIyMpKQkBCOHDlM\nfX39oNubnp7Bzp3bmT9/IR9+uIGIiEiSkibx6qsvoSgKDQ311NXVDuVjE0KIQfN4FGqbuijp61W3\n09DS/06beHMoafFhpCeEk5YQTlx0COphGCb/IgnvIZg0KYUjRw7zP//ze0JDjX2vL1mylIce+j5F\nRQVcdtmVxMTE8PzzzwzpXAsWLOKdd97i7rvvYMaMWYSFhfc7HhYWzle/ejkrV67EYokjK8s7ZL1i\nxTX893//jri4eK699gYee+xX7Nq1kxkzZnLPPXfwu989yYsvvsC9997FokUXsGDB+Tz++K/5yU9+\nekod7rjjLh599L/48MMNqFQqVq9eQ2xsHAZDCHffvYopU6azYsU1/P73v2Xq1GkDtun++x/kscd+\nxYsvvoBOF8zPfvZLwsLCSU1N41vfup2JE5PIyMgc0ucmhBBn0u1wUlbb3terLqtrx95zsmNkCNaQ\nmxxJWkI46QnhpMaHEaI/tWMzElTK6cZsRyFfD3H7Y9jcn9rbbezbt4clSy7Cam3k/vvv5qWXXj/l\nfWOtXYMRiG2CwGyXtGnsCJR2udweHL1uHL0uQox66urbjz/3vtbjPPn45Otu6pq6qG3q4vMBGBsV\nQnpCmDes48OJN4f2u4w3EiwW02lfl573GBESEsrGjR/y0kv/QFE83Hef7xZ0EUKI0cbl9lBU0cKB\n4iY6up3e8D0exD2fC2fXAHcNnUlwkIbJSRF9veq0hHCMhtHRqx4MCe8xQqvV8vOf/3qkqyGEEH6j\nKAoV9R1sL6hn16EGOrr7z59Rq1TodRr0wRrCQnXERBoIDtJ4X9NpiQjXg9uDXqchWKc9/vrJ497X\nvY+NBi0a9dhdp0zCWwghxIiyttnZWVjPjsIG6o9PEDMagrhoZiLzcmOJiTRg0GnQatRnvYc6UC4F\nDIaEtxBCiGHX5XCy+3AjOwrqKa72roURpFUzNzuG83LjyEuJQqsZuz1jf5PwFkIIMSycLg+flTaz\ns7Ceg6VNuNwKKiArKYL5eXHMyowhRC+xNBjyKQkhhPAbRVEoqbGxo6Ce3Ycb6XK4AEgwhzI/L47z\ncmKJCtMPUIr4IhmTGKLxsCXol3HgwD5aW1v6vbZv3x4efvhHPj/XYJSVlXDvvXeNyLmFGM/qW7pZ\n/0kZP/7LDn79z31sPlCLVqPmK3Mm8rPb5/DzO+Zy6XmTJLjPkfS8h+DElqBLllw0qPfff/8Pznp8\nuLcEnTlzzmlXUhuKd955ixtv/DqRkVE+LVcIMfq1d/eyq6iBHYUNlNd59zgIDtIwPzeO+Xmx5EyK\nGvH7pgOFhPcQjJctQW+44SpWrLiGbdu20Nvby5NP/hmdLpjHHvsVtbU19Pb28s1vfhuVCrZs2Ux5\neRm//OVjxMXFnfKZffzxRl555Z9oNFomT87mvvu+R1dXJ//1Xw9jt9txOBx873s/JCcnj5Urr+a8\n8xaSmBjHkSMlmM0Wjhw5RENDPT/96S+ZPDmL119/lQ8/fA+VSs2iRUu48cav09jYwCOPPERQUBDp\n6bJCmxC+5Oh10dTmwGqzY21z0NRmp8nmfV7f3I3bo6BSQV5KFPPz4piZYSFYd+reBmJoAia815W8\nzf7G/EG/X6NW4fac/eb+GTFTuCb98jMeHy9bgrrdbpKSkrnppltZs+Yn7Nmzm66uTnQ6HX/84//S\n1GTl3nu/xSuvrCM9PZPvf/9Hpw3u7u5uXnjhWf7yl+fR6XQ88shDfPbZASIiIrn88qtYvHgJe/fu\n5sUXX+BXv/odLpeL885bwBVXXMwDD/yA3t5e/vCHP/LGG//ivffewWQysXnzR/z5z88CcPfdd3Dh\nhctYt24tF130Fa6//kb++c+/U1Jy9Ky/ZyHESS63h+bjYXwipJvaHDQdD+tO++kvt+l1GibFmZib\nFcO8nFjCjcHDXPPxJWDCezQI5C1BT9THYont2/hkxoxZAJjNFnS6oNO28/PKy8toaKjn+9+/F4Cu\nrk7q6+tJTU3nhRf+xssv/wOn09lv69CcnJPbin6+DkVFhRw6VEh1dRX33fctALq7u6ivr6WiopwL\nL1wGwIwZs9m5c/tZ6yXEeGPr6qWxo5niimasx3vOTW12rDYHbR09nK5bo9WoiA43kBxnwhxhwBKu\nxxxhwByuxxJhIFSvHZZ9rIVXwIT3NemXn7WX/EWyJajXYLcEPbW+qn71djqd/c5/OkFB3qHyP/zh\nj/1ef+65/8VsjuGRR37B4cNF/PGPT/Qd02pPv62ooihotUHMn7+QH/3o//Ur78UXX+iry4nPQYjx\nqqO7l8r6DsrrO6ioa6eivoPWjp5T3qcCIsOCyZgYgeV4IJsj9JjDDVgiDIQbdSOye5Y4vYAJ75Ew\n3rYE/bzs7Bz27dvDsmUX09BQj1qtxmQynfEzAUhKSqaiopzW1hYiI6N49tm/cuWVV2OztZGWlgHA\nxx9vwuVyDeozmTw5m6effgqHw0FwcDBPPvl77r77XpKSJnH4cBFZWdlnnXEvRKDpdjipqO/wfh0P\n6iabo997Iow6pqebSU+KJESnxhLuDenoML0sijKGSHgPwXjbEvTzLrroK+zfv5f77vsWLpeTH/5w\nNQDTp8/k4Yd/zK9//XtSU9P6/Yxer+f++3/Agw/ej04XREbGZMxmC5dcchm//OUaNm36kK997Xo+\n/PB93nnnrQE/k7i4OK6//ka+8507UavVLF68hOBgPddddyOPPPIQn3yyqe9/CoQINPYeF8caOiiv\n66Ci3hvUja32fu8xhQQxJTWa5DgTKRPCmBRnItLkvRY9npYSDUSyJegYIVuCBlabIDDbJW3yjx6n\nm6qGTsrr26k4Htb1zd39rk2H6rUkx5lInhDm/R4XRlRY8BmvQ4+GdvlaoLbpdKTnPUbIlqBCjB+K\nolBj7WLX4UYOFDdR09TJ57tZep13O8vkuDCSJ3gD2xKulwlj44hfw/vRRx/l4MGDqFQqVq9ezdSp\nU/uOvfjii7z11luo1Wry8vL4f//v/52lJCFbggoR+Gqauth9qIHdhxupa/burhWkVZOeEH4yqONM\nxEaFyOSxcc5v4b1r1y4qKytZu3YtpaWlrF69mrVr1wLQ2dnJs88+y/vvv49Wq2XVqlUcOHCA6dOn\n+6s6QggxKtU1d7H7cCO7DzVS0+RdXClIq2bWZAtzsmKYlmaWRU7EKfwW3jt27GDZMu+9tmlpadhs\nNjo7OzEajQQFBREUFER3dzchISHY7XbCw8MHKFEIIQJDQ2s3uw81sutQI9VW7zoPWo2aGRlm5mTH\nMD3djF4nVzXFmfntr6OpqYnc3JMLbERFRWG1WjEajQQHB/Od73yHZcuWERwczGWXXUZKSoq/qiKE\nECOusc3OnsON7DrUwLEGb2Br1Cqmp5uZkxXD9AwzhmAJbDE4w/aX8vlJ7Z2dnfz1r3/lvffew2g0\nctttt3H48GGysrLO+PORkSFotb4dOjrTLL6xLhDbFYhtgsBsl7TppMaWbrYerGXLwRpKqtoAb2DP\nzo7l/GnxzMubgNHg282Bvgz5XY1dfgvvmJgYmpqa+p43NjZisVgAKC0tZeLEiURFeXeemj17NgUF\nBWcN79bWbp/WLxBvKYDAbFcgtgkCs13SJmhpd3h72IcbKav17qylVqnITYliblYMMzItfYFt73Rg\n73ScrTi/kd/V2DDst4otXLiQp556ipUrV1JYWEhMTAxGo3chk4SEBEpLS3E4HOj1egoKCrjgggv8\nVRUhhPA5l9tDS0ePd23wNu+mHUer2iip8a7xr1JB9qRI5mbHMDPTgilEN8I1FoHEb+E9c+ZMcnNz\nWblyJSqVijVr1rBu3TpMJhPLly/njjvu4NZbb0Wj0TBjxgxmz57tr6oIIcSXpigK7d1OmrtbKC5v\nxmpz9Avqlg4HX1ziSgVkJUUwJzuWWZkWwkIlsIV/yAprASYQ2xWIbYLAbNdYa9Pp9qa2fm5/6l7n\n6Te2iTDq+nbWskQYjm/eoWeCOZSwMdLDHmu/q8EI1DadjkxtFEKMCx6PQkV9BwXlzRRVtFLX3EVH\n9+k3DTIEa4iLDMEcYSBpQhihOg2WCG9AR4fp0QXJfddiZEl4CyECVltnD4XlLeSXeQO70+4Na5UK\nYiIMJMWavKF8fG9qy/EtMD+/N3Ug9ubE2CfhLYQIGE6Xh5LqNgrKW8gva+lbAAUg0hTMoqkTyEuN\nJic5klD9yN2iJcRQSXgLIca0htZuCspaKChr5vCxNnqc3v3ktRo1ucmR5KZEMyU1inhzqGzcIQKG\nhLcQYkyx97g4fKyVgnJvYFvbTt4nHRcVQl5qFHkp0UxOiiBYrk2LACXhLYQY1RRFoaqxk/yyZgrL\nWyiutuH2eG+SMQRrmJlpIS8liryUKMwRhhGurRDDQ8JbCDHqeDwKR6ra2H24kf1Hrdi6evuOTYoz\nMeV47zo1PgytRj2CNRW+1uPupdxWSXFrKcVtZVR2VKNWqTFogtFr9eg1evTaE4+9308cM9vCcdnB\noNX3O67XBmPQ6NGoTz8S4/a4cbh7cLh66HH34HA7cLh6+r/mcnifH3/cc/yYw91Dj6sHu9tBqDaE\nB2ffi0Gr9/vnJOEthBgVPB6F4uo2dh1uZO/hRtqP38ZlNAQxPzeOvNQocpOjZOGTAONw9VBmq6C4\nrYzi1jIqO6rwKN7761WoiDfGoVGpcbh66HbZaXG04fSc/ha/gQSptX3hr0BfCDs9rnOuv06jQ68J\nxqDVE2WIRKsanks1Et5CiBHjURRKqm3sPtTIniONfT1soyGIJdPjmZMVQ2ZSBBq19K7PRlEUXB4X\nDncPdpfjZM/R5Tj+/HM9R5cDu6uHKJMJIybMhui+L53G/zPw7S4HpW3llLSVc7StlKqOmr6wVqvU\nTDQlkBGRSkZEKmkRyRi0p14KOdHWE2080a6gEBWNrW39XnMc7xWfeOxwez8TFSpCg0KI1kcRrA3G\noAkmWBvc11sP1px4/IXXtCdfD9YEo1aNzN+mhLcQYlh5FIXSmpOB3dZ5MrAXT4tnTnYMWRLYKIpC\nbVc9Rc1H6HJ2nzaATjx2uHpwK+4vd4KGU1+KCA7HbIjCbIjGcjzQLce/QoJCzqkd3U47pbZyilvL\nKG4ro6qjBgXvnAW1Ss0kUyIZkWmkR6SSFj4J/SCGnLVqLUa1FmNQaL/XLRYT1pDxcU++hLcQwu88\nikJZbXtfYLd29AAQqteyaOqE44EdKdevAWt3M3saDrCn8QD1XadJWLzDySd6gSadiRiD+ZRrvHqN\n/vi13+DjQ8V6DMcfB2t0BJvUHK09RlN3M1Z7M0127/fStgpK2spPOWeI1nBKqJsN0VhCognTmfp6\noF3ObkravEFd0lpGdWddX1hrVBpSwid5e9aRqaSETUKvDfbfhxnAJLyFEH6hKApldScDu6XdG9gh\nwVrOn+IN7OxJEtgAtp529jYeZE/DASrbqwBv73K6JY8ZMVOJ1kd5g/d4QOs0uiEP11oiTRhdEae8\n7vS4aLG3YLWfDHVvsLdQ21nHsY7qU34mSK0l2hCNGhV1XQ19Ya1VaUiLSCYjIpX0iFRSwyeh08ic\nBV+Q8BZC+IyiKJQfD+zdhxtpbvfeg20I1rIwL4452THkJEdJYAPdzm72W/PZ03CQ4tZSFBRUqMiO\nymR27HSmWXJPe73X34LUWmJDY4gNjTnlmEfxYOtpx2pvOh7sLf0C3ulxkR6RQkZkGhkRqSSHJQ3L\ndfTxSMJbCDFkNdZO3i/Mp/BYMy21BkCFXqdhfq43sHOTowjSjp7AVhRlRFZb63H3kt9UxJ6GAxQ1\nH+m7Tp0aPolZsdOZGTOVMN3pd5EaDdQqNZH6CCL1EWRGpvc7pigKCsqITeAabyS8hRDnpKO7l82F\nJWyt2kN7cBlqQzckQnh8GDOjZnFlziLC9caRrmY/tZ31bK39lN31+1CgbzKWJcR8/LEZS0g0piCj\nz8Ld5XFxqOUoexoO8FlTEb1u7wS9BOMEZsdMZ1bsNKINUT4510hSqVSokOVnh4uEtxBi0JwuD3uL\n6/igZBe17iOow1ogAjSKhkn6LBJiwvi0eh872zaxb+dW5sTNYHHCAhJN8SNW5163k/2Nn7G1didl\ntkoAwnQmQrSGM17DDdbosBjMmA3RTIqeQCgmb7AbogkPDhuwd+lRPJS0lbOn4QAHGvPpcnUDYNZH\nMTtxOrNipxNvjPN9Y8W4IeEthDgrRVEorbXxbtFeDrcXoITVoQr1oAai1fFcMGkuCybOwKA1YLGY\nKKu5hB11u9lSs5NttbvYVruL1PBkLkiYz/SYKWjVw/PPTl1XA1trdvJp/T7sLnvf9eTzE85jSnQ2\nGrUGj+KhrceGtfvkbOsT13Mbu61Ud9ZywJrfr9wgtfb4bGszZkNUX2/dYjDT5exiT8MB9jV+RluP\nDfD+j8KFieczO246k0wTZXMU4RMS3kKI02q2OXg/v4hdDfvoCa1EpeuBSDAoYcy2zGB5+nzMpxnu\nNemMfGXShSxLuoDC5sN8XL2dQy1HKbNVYCr5Nwvj53F+/Dwi9afOdB6qXreTA9Z8ttbspNRW0Vef\niyctZUH83FPqq1apidJHEqWPZDKnXsNt7+3AGdxNcW1Vv3Bv7G6m7gy3cQEYtAYWTJjD7NgZZESm\nynVg4XMS3kKIPvYeF9sPV7K5fDdNmhLUoe0QCRpFx2TjNC7OWEB6RPKgeo9qlZop5hymmHNo7Lay\npWYnO+r28F7FR7xfuYmp5hwWJcxncmT6kHuj9V0NbK39lE/r9tLtsgOQFZnB+QnnMdWcc8Y1rc9G\npVIRHhyGxZKAmf5D3Iqi0OXs7uulW+3NWLubUalghmUK2dGTCRqmEQYxPslflxDjnMejkF9h5f3D\nuynvOQQmK6owBbWiIj4ohaUp85g9YQpBQ7jlJybEwtcyruCK1IvZ3bCfT6p3cMBawAFrAbEhMSxO\nmM+8CbO+1IYOTreT/dZ8ttV+2reoiCnI2+tfGD8XsyH6nOs7EJVKhVEXilEXSkr4JL+dR4gzkfAW\nYpyqtnbyXsFBPms5iMtUjUrvQqUHk8rMgvhZXJgyD5POt7PFdRodC+PnsWDCXMrbK/mkegf7Gj/j\nteI3ebPsXebGzeSChAVnncxV39XIttpP+bR+L11O70SwrMgMFibMY6o5Z9iuqQsxkuSvXIhxwt7j\n4nBVC3uOHeVoWzGd+krU+m6IBJ1iYErELC7OmD8sM8NVKhWp4cmkhifztYwr2Fa7i601O/u+0iNS\nWJywgOmWPDRqDU6Pi4ON+Wyt/ZTitjIAjEGhLE9awoL4ucSEmP1eZyFGEwlvIQJUT6+bI9Ut7K4s\npritFBt1qIytqDTuvtu7kvVZfCV9PrnmzHO6LuwLJp2RS5KXsjzpAgqaD/FJ9Q4OtxZT0lZOuM5E\nVlQmBc2H+nrZmZHpnB8/j2mWXOlli3FL/vKFCBC9TjdHq1vZc6yYIy1ltFGLytjiDetwUAOhRJJs\nSmZ2QjZTYrK+1DVmf9OoNUyz5DHNkkdDVyOf1OxgZ91ePq3fizEolGVJF7Awfi4xIZaRrqoQI07C\nW4gxyulyU1zdxp5jJRxpLqVFqfH2rLWuvrA2EE6KKYWZE7LIjckY1Utvfl5saAzXZa7gitRLqO2q\nY6IpUWZvC/E58l+DEGOE0+WhrNbG7ooSDreU0OypRWVs7hfWesJIDk1mZrw3rCOCw0e62kOi1waT\nGp480tUQYtSR8BZilPIoHqqb29h2uJyiphKa3dVgbEEV5IQwb1gHKyYmHQ/rPEuGXxY+EUKMPhLe\nQviB2+PG7nJ4v9x2HC4HdlfP8e/eL3Wth+b29r7XHO6Tx7p67TiV3pMFhoEK0ClGkkIymTkhi7yY\nTKINkSPWRiHEyJHwFmIIPIqHqo4aipqPUNRyFKu9CYfLgdPj+tJlqVGjQYfLqcbdG4ziDiUkyEBi\nZCTT4zOZGjuZaH2krI0thJDwFuLLsvV0cLjlKEUtRzjUcrTvFiYVKsyGKCKDIzBo9Ri0evSa499P\nPNcGY9AaMGj0TLBE4ejw0GpzszO/mV0FTXQ5PWg1as7LiWXpnASS48JGuLVCiNFIwluIAbg8Lspt\nlRS1HKWo+QjVnbV9xyKCw1kwYQ7Z0ZPJikwnJChkUGW6PR5Ka7p4Y1Mxh4+1ARAdFswVCxNZNHUC\nphCdX9oihAgMEt5CnEazvYWiliMUNR/laGsJDncPAFqVhqzIDLKjM8mJmsyE0NgvNYzd3tXLxwdr\n2by/htYOb5k5yZFcNDORaelm1GoZEhdCDEzCWwig191LcVsZh5q9w+EN3da+YzEGM/OiZ5ETNZmM\nyDSCNV++V1xW285He6vZfbgBl1shWKfh8oUpnJcdQ7w51JdNEUKMAxLeYlxSFIX67kaKmr3XrYvb\nynAdn2Sm0+iYYs4mJ2oy2VGTsYSc2+5UTpebXYca2bivmvK6DgDiokK4aFYiC/LiSEqMxGrt8Fmb\nhBDjh4S3GDecHhdHW0spaCoiv+kQrT1tfccSjBPIiZpMTnQmKeHJQ1rNq6Xdwab9NXx8oJZOu9O7\nx3OGmaWzEsmZJLPFhRBDJ+EtAlpHbycFzYcpaCriUMtRetzee6cNWgOzYqaRHT2Z7CjfrER2tKqN\nD3ZXsa/YiqJAqF7LV+clceGMBMwRhiGXL4QQJ0h4i4ByYjg831pEfnMR5bZjKCiA99p1njmbqeYc\nUsOTfbaL1tGqNt7YUtY3azwp1shFsxKZlx2LLmhkduoSQgQ2CW8x5rk9bkrayslvLiLfWkSTowXw\n3nedGp7MlOOBHRsa49PzHq35stz1AAAgAElEQVRq482t5RyqbAUgLyWKyxckk5EYLkPjQgi/kvAW\nY1K3s5vC5iPkNxVR1HIEu8sBgF4TzAzLFKaYc8iNzsKo8/1M7pJqG29sLaOowhvauSlRrDg/hfSE\nsb0JiBBi7JDwFmNGY3cT+U1F5DcVUWqrwKN4AIjSRzI3biZTzDlkRKSi9dPWkSU1Nt7cWk5hubdn\nn5scyYrzU0lPlNAWQgwvCW8xqjXbW9hw8EM+PXaAhu7GvtcnhU1kqjmHKeYc4kPj/DpMXVpr480t\n5RQcD+3sSZGsOD+FzImyg5cQYmRIeItRa3f9fl45sg6Hu4cgdRBTzDlMNeeQG51NeLDJ7+cvq23n\nza3l5Jc1AxLaQojRQ8JbjDoOl4NXj77Jp/V70Wl03DX7ZrJDc9Bpgobl/OV13tD+rNQb2llJEaw4\nP4XJSbL9phBidJDwFqNKZXsVzxe+hNXeTJIpgdtzbyJ3UuqwrERWXtfOW1vLOXg8tDMnRnDV+Slk\nTZLQFkKMLhLeYlTwKB4+OvYJb5W9h0fxsCzpAq5Ivdhvk88+r6K+nbe2VnCgpAmAzMRwVixKJSsp\nQm75EkKMShLeYsTZetr5v6K1HG4tJkxn4tacG8iOyvT7eSvrO3hza3lfaKcnhnPV+SlkyxKmQohR\nTsJbjKj8piL+eeg1Op1d5EVn8fXs6zHpjH49Z0V9O//eVsH+4uOhnRDOikUpsu64EGLMkPAWI8Lp\ndrK+9D98XL0NrVrLdRkruCBxgV/Ds7i6jX9vr6CgzHvLV1p8GCsWpZCbHCWhLYQYUyS8xbCr72rg\nucKXqOmsIy4khlV5N5NgnOCXcymKQlFlK+9sr+hbe3zyxAguX5gsPW0hxJgl4S2GjaIobKv9lH8V\n/xunx8nC+Hlcm3EFOo3OL+c6WNLM2zsqKKttByAvNYrL5yfLfdpCiDHPr+H96KOPcvDgQVQqFatX\nr2bq1KkANDQ08OCDD/a9r6qqih/84AdcccUV/qyOGEFdzm5eOvwvDlgLCNEa+EbOSqbHTPH5eTwe\nhb1Hrby9vYKqxk7Au5f25QuSSZkQ5vPzCSHESPBbeO/atYvKykrWrl1LaWkpq1evZu3atQDExsby\nj3/8AwCXy8Utt9zC0qVL/VUVMcKKW0v5e9ErtPXYSI9I4Rs5NxKp923v1+X28GlRA+/sqKS+pRuV\nCublxHLZ/EkkWvw7AU4IIYab38J7x44dLFu2DIC0tDRsNhudnZ0Yjf3/IV2/fj0XX3wxoaG+3/1J\njCy3x827FR/yXsVGVCoVl6dczMXJF6JWqX12DqfLw7b8Ov6zs5ImmwONWsX5Uydw2XmTiI0K8dl5\nhBBiNPFbeDc1NZGbm9v3PCoqCqvVekp4v/baazz33HP+qoYYIc32Fv5e9DJltkqi9ZF8I/dGUsOT\nfVZ+j9PNJwdqeW/XMVo7etBq1CydmcAl85Iwhxt8dh4hhBiNhm3CmqIop7y2f/9+UlNTTwn004mM\nDEGr1fi0ThaL/ze3GAkj3a7tx/bwv3teottpZ0HSbO6adRMhuqEF6ok2dTucvLOtnDc/KcXW2Yte\np+HqJelcdUEaUWF6X1R/WI3078ofpE1jRyC2KxDbdDp+C++YmBiampr6njc2NmKxWPq9Z/Pmzcyf\nP39Q5bW2dvu0fhaLaVjWyx5uI9kuh6uH146+yc76Peg0Or6efT3nxc2iy+aii3Ovk8ViovxYCx/u\nqeLDPdV097gwBGu5fEEyy2cnYgrR4e5xYrU6fdga/wvEv0Fp09gRiO0K1Dadjt/Ce+HChTz11FOs\nXLmSwsJCYmJiTulh5+fnc+mll/qrCsJPupzdNHY30dhtxWpvorG7Cau9iYZuKz3uXpJMCXwj9yZi\nQywDFzaA9u5e3nm7kLe3ldPT68ZoCOKaxaksnZlIiF7udBRCjE9++9dv5syZ5ObmsnLlSlQqFWvW\nrGHdunWYTCaWL18OgNVqJTo62l9VEENgdzmwdjfRaG/q+97Y7X3c5Tp1FESr1hJjMDPVnMNXU5YN\neUMRl9vDR3ureWtbBfYeF+FGHVefn8IF0xMI1vn28okQQow1fu26fP5eboCsrKx+z//973/78/Ri\nAL3uXqz25r5Q9ga0lUZ7Ex29nae8X61SYzZEkRI+iZgQMxaDmZgQ71dEcLjPZpF/VtrEyx+V0NDS\nTUiwljtX5DE7I5ogH895EEKIsUrGHceh4tZSXjr8Oo32plOOqVARpY8kOyqzf0AbLETpI9Co/Reg\ntU1dvLKxmIKyFlQqWDozgasWpZKSFBVw17GEEGIoJLzHmfymIv5W8E8URSEzIg1LyIlw9n6PNkQT\nNAx7aH9el8PJm1vK2bivBo+ikJMcycqLMmRxFSGEOAMJ73FkV/0+/nHoVbQqDXdOvY2c6MkjWh+3\nx8PHB2p5Y0s5nXYnMZEGbliazvR0s2wYIoQQZyHhPU58XL2dV4++gUFr4J5pt/t0wZRzUVTRwssf\nFVNj7UKv03DdhWksmzWRIK3vVl8TQohAJeEd4BRF4b2KjbxdvgGTzsh90+/02/abg9HQ2s2rG0vY\nX9yEClg0dQLXXJBGeKjvdxYTQohAJeEdwDyKh/Ul77CxagtR+kjum/5NYnxw7/W5sPe4eHt7BR/s\nqcLlVshMDOfGZZlMihsfqyEJIYQvSXgHKLfHzUuHX2dn/R7iQmK4b8adRASHD3s9PIrCts/qeP2T\nMtq7eokOC+a6C9OZkxUj17WFEOIcSXgHIKfHxd8LX+KAtYBJponcM20VRt3w79p2tKqNlz8sprKh\nA12QmqsWpXDJ3CR0QXK/thBCDIWEd4BxOB08ffA5jrSWkBmRxrem3oZeO7wbdjTbHLy2uYRdhxoB\nmJ8by9fG6MYhQggxGkl4B5AuZzdPbP47xa0VTDXnsir3JoI0QcN2/p5eN+9+Wsm7nx7D6fKQMiGM\nm5ZlkJYw/MP1QggRyCS8A0Rbj40/HvgbdV0NzI2bydezrvPramhfVFbbzl/eLKDJ5iDcqOPaC9KY\nnxeHWq5rCyGEz0l4B4AmezP/s/8Zmh0tXJKxhMsSL/HZOuMDURSFD/ZU89qmEjweha/OS+KKhcno\ndfKnJYQQ/iL/wo5xtZ31/PHAM9h6O7g0ZTm3zbiapqZTNxXxh26Hk+f+c5h9R62EhQRx55W55CZH\nDcu5hRBiPJPwHsPKbZX8+eBzdLvsXJtxJRdOPH/Ybr8qr2vn6Te8w+RZSRHcdWUuEcbgYTm3EEKM\ndxLeY9ThlmL+mv8CLo+LW7NvYN6EWcNyXkVR+GhvNWs3eofJr1iQzJXnJ6NRy7KmQggxXCS8x6AD\njfk8X/gSAN/Mu4VpltxhOW+3w8Xz7x5i7xErppAg7rwih7yU6GE5txBCiJMkvMeYHbW7efHwv9Bp\ngvjWlG8wOSp9WM5bUe8dJre2OcicGMG3rswl0iTD5EIIMRIkvMeQjcc+4fWStwnVhnDP9FUkhyX5\n/ZyKorBxXw1rNxbjcitcvmASK85PkWFyIYQYQRLeY4CiKLxd/j7vVXxEuC6M+2bcyYTQWL+ft9vh\n4u/vHWbP4UaMhiDuuiKHvFQZJhdCiJEm4T3KeRQPrx19i09qtmM2RPPd6XcSbfD/7ViV9R08/UYB\njW12MhLD+daVubK8qRBCjBIS3qPca0ff5JOaHSQYJ/Cdad8kPNi/W2gqisLmA7W8/GExLreHS8+b\nxNWLZZhcCCFGEwnvUWxvwwE+qdlBfGgcD8z4FiFBIX49n73HxQvvHWbXIe8w+Tcvn8LUNBkmF0KI\n0UbCe5Sydjfz0uHX0Wl0fDPv634P7mMN3mHyhlY76QnhfHuFDJMLIcRoJeE9Cjk9Lp4r/CcOdw+3\nZt9AbGiM386lKAofH6zlpQ+8w+SXzEvimsWpaDUyTC6EEKOVhPco9GbJfzjWUcN5cbP9unKavcfF\n/204wqdFDYTqtdxzdR7T081+O58QQgjfkPAeZT6zFrKpeiuxITFcP/kqv52nqrGTP79RQENLN2kJ\nYXz7yjyiw2WYXAghxoIBw7u0tJS0tLThqMu41+Jo5R+HXiVIreWOvJsJ1uj8cp5DFS088a/PcLo8\nXDx3Il+7IE2GyYUQYgwZMLy/+93vEhYWxrXXXsull16KwWAYjnqNO26Pm+cLX6LbZefGydeQYJzg\nl/PUNXfxp/UFeDwK914zhZmZFr+cRwghhP8MGN7vvPMOR48e5d133+WWW24hOzub6667jqlTpw5H\n/caNt8vfp8xWyayYaSyMn+eXc3R09/LEawfp7nHxzcuzJbiFEGKMGtRYaWZmJvfffz8PPfQQpaWl\n3HPPPdx8881UVFT4uXrjQ1HzEd6v3ITZEM2NWV/zy57cTpeHp9blY21zcMWCZBbk+adnL4QQwv8G\n7HnX1NSwfv163n77bdLT0/n2t7/NokWLyM/P54c//CGvvfbacNQzYNl62nmh6BU0Kg2rcm/CoPX9\npDFFUXj+3UOUVNuYmx3DVYtSfH4OIYQQw2fA8L7lllu49tpreeGFF4iNPbkZxtSpU2XofIg8ioe/\nF71Cp7OLazOuZFLYRL+c59/bKthZ2EBaQhh3XJbtl569EEKI4TPgsPlbb71FcnJyX3C//PLLdHV1\nAfDII4/4t3YBbkPFRo62ljDFnMOSxIV+OcfOwnre2FqOOVzPfddMJUir8ct5hBBCDJ8Bw/snP/kJ\nTU1Nfc8dDgc/+tGP/Fqp8aC4tZR3yj8gMjiCW7Kv90tvuLi6jef+cwhDsIb7r5tGWKh/bj0TQggx\nvAYM77a2Nm699da+57fffjvt7e1+rVSg6+jt5PnCl1GpVKzKu4lQP6xb3thm56nX8/F44J6rppBg\nDvX5OYQQQoyMAcPb6XRSWlra97ygoACn0+nXSgUyj+Lh/w6txdbbzhUpF5Manuzzc3Q7nDz52kE6\n7U6+fnEmuSn+3/9bCCHE8BlwwtpPfvIT7rnnHjo6OnC73URFRfHYY48NR90C0saqLRQ1HyE7KpNl\nky7wefkut4c/rS+grrmbi+dOZMn0BJ+fQwghxMgaMLynTZvGhg0baG1tRaVSERERwb59+4ajbgGn\n3FbJm6XvEqYzcVvOStQq3y5JqigKT7/+GYcqW5mRYea6Jek+LV8IIcToMGB4d3Z28uabb9La2gp4\nh9Fff/11tm7d6vfKBZJuZzfPFb6Eoih8I+dGTDqjz8+xYVcV739aSVKskbuuyEWtllvChBAiEA3Y\n9XvggQc4cuQI69ato6uri02bNvGzn/1sGKoWOBRF4cXD/6LF0colyRcxOcr3PeK9R6y8tqmE6HA9\n9187jWCd3BImhBCBasDw7unp4ec//zkJCQn8+Mc/5v/+7/949913h6NuAeOTmh0csBaQEZHKpSnL\nfF5+RX07z/y7kKAgNY+smkekKdjn5xBCCDF6DGq2eXd3Nx6Ph9bWViIiIqiqqhqOugWEqo5a1hX/\nG2NQKN/IvdHn17lb2h08eXx7z29dmUtaYoRPyxdCCDH6DHjNe8WKFbz66qtcd911XHrppURFRTFp\n0qThqNuY53A5eK7gn7gUN7fm3EBEcLhvy+918eS/PsPW2cvKpenMyJBdwoQQYjwYMLxXrlzZt/rX\n/PnzaW5uJjs72+8VG+sUReGVI+tptDexLOkCcqOzfFq+x6Pw1zcLqWrsZMmMBJbP8c+66EIIIUaf\nAcdwP7+6WmxsLDk5ObKxxSDsrNvD7ob9pIQlcWXqJT4vf+3GEg6WNpObEsVNyzLkdyKEEOPIgD3v\n7OxsnnzySWbMmEFQUFDf6/Pnz/drxcayuq4G1h59A4PWwO25N6FR+3bm98Z91Xywp4p4cyh3r8hD\nq/HtdXQhhBCj24DhfejQIQD27NnT95pKpZLwPoNedy/PFvwTp8fJN3JWEm3w7dKk+WXNvPRBMWEh\nQTxw7VRC9AP+CoUQQgSYAf/l/8c//jEc9QgYrx19i7quBi5IXMD0mCk+Lbva2snTbxSgVqu472tT\nMUcYfFq+EEKIsWHA8L7ppptOez31xRdf9EuFxrI99fvZXreLicZ4rk67zKdl2zp7ePK1gzh63Xx7\nRS5pCb6duS6EEGLsGDC8H3jggb7HTqeTnTt3EhLi+y0sx7puZzcvH1lPsEbHqrybCdIEDfxDg9Tr\ndPM/r+fT3N7D1YtTmZsd67OyhRBCjD0DhvfcuXP7PV+4cCF33nnnoAp/9NFHOXjwICqVitWrVzN1\n6tS+Y3V1dXz/+9/H6XSSk5PDz3/+8y9Z9dFlW+0uHG4HV6VdSkyI7+639igKf3vnEOV17SzIi+Py\n+XKPvRBCjHcDTlOuqqrq97Vr1y7Ky8sHLHjXrl1UVlaydu1afvWrX/GrX/2q3/Hf/OY3rFq1in/9\n619oNBpqa2vPvRUjzO1xs7l6GzqNjoXx83xa9jvbK9hzuJHMiRHcdkmW3BImhBBi4J73bbfd1vdY\npVJhNBq59957Byx4x44dLFvmXcc7LS0Nm81GZ2cnRqMRj8fD3r17+cMf/gDAmjVrzrX+o8L+xs9o\n67GxJHEhIUG+m0TW0NrNv7dXEGkK5t5rphCklVvChBBCDCK8N27ciMfjQa32BofT6ex3v/eZNDU1\nkZub2/c8KioKq9WK0WikpaWF0NBQfv3rX1NYWMjs2bP5wQ9+cNbyIiND0Gp9e7+0xWIachmKovDJ\n/u2oUPG1aRdjMQ69zBOefqsQl1vhrqunkJI0+FvOfNGu0SYQ2wSB2S5p09gRiO0KxDadzoDhvWHD\nBtavX89f/vIXAG6++WZWrVrFJZd8uVXDFEXp97ihoYFbb72VhIQE7rrrLjZv3sySJUvO+POtrd1f\n6nwDsVhMWK0dQy6npK2c0tZKplnyUNv1WO1DLxPgYEkTu4sayEqKIHPC4Ovqq3aNJoHYJgjMdkmb\nxo5AbFegtul0BhyHff755/nd737X9/y5557j+eefH/CEMTExNDU19T1vbGzEYvFO5IqMjCQ+Pp6k\npCQ0Gg3z58+nuLh4wDJHo41VWwBYOnGRz8p0uty8/GExapWKm5dnynVuIYQQ/QwY3oqiYDKdTH6j\n0TioMFm4cCEbNmwAoLCwkJiYGIxGIwBarZaJEydSUVHRdzwlJeVc6j+irN3NfGYtJMmUSFp4ss/K\n3bCrisY2OxfNSiTBYvRZuUIIIQLDgMPmeXl5PPDAA8ydOxdFUdiyZQt5eXkDFjxz5kxyc3P7diVb\ns2YN69atw2QysXz5clavXs1DDz2EoihkZmaydOlSnzRoOG2u3oqCwkUTF/msd9zS7uDtHRWEhQSx\n4vyx9z80Qggh/G/A8H744Yd56623+Oyzz1CpVFx55ZWDvt794IMP9nuelXVyW8xJkybx8ssvf8nq\njh7dTjvb63YTERzOjJipA//AIK3dWEKv08PXl0+WdcuFEEKc1oDpYLfbCQoK4pFHHgHg5Zdfxm63\nExoa6vfKjWbb63bR6+7l0uRlPts17FBlK7sPN5IWH8aCKXE+KVMIIUTgGfCa949//ON+E88cDgc/\n+tGP/Fqp0c7tcbO56sSiLHMH/oFBcLk9vPTBUVTAzV/JRC2T1IQQQpzBgOHd1tbGrbfe2vf89ttv\np7293a+VGu0OWPNp7Wlj/oTZhAT5Zp33TftqqGnqYvH0eJLjwnxSphBCiMA0YHg7nU5KS0v7nufn\n5+N0Ov1aqdFMURQ+OrYFFSqWJJ7vkzJtXb28sbWMUL2Waxan+qRMIYQQgWvAa94/+clPuOeee+jo\n6MDj8RAZGcljjz02HHUblcpslVR2VDHNnEtMiNknZb6+uRR7j5uvfyUTU4jOJ2UKIYQIXAOG97Rp\n09iwYQN1dXV8+umnrF+/nrvvvputW7cOR/1GnROLslzoo0VZSmtsbM2vIynGyJLpCT4pUwghRGAb\nMLwPHDjAunXr+M9//oPH4+EXv/gFX/nKV4ajbqNOk72Zg9YCkkwJpEcM/R5sj0fhnx8cBeCm5Zmo\n1TJJTQghxMDOeM37mWee4dJLL+V73/seUVFRvP766yQlJXHZZZcNamOSQLS5ahsKCksnLvbJoixb\nPqulsr6D+bmxZE6M8EENhRBCjAdn7Hk/8cQTpKen89Of/pTzzjsPYFyvsW132dlet4uI4HBm+mBR\nlk67k9c/LiNYp+G6C9N9UEMhhBDjxRnDe/Pmzaxfv541a9bg8Xi4+uqrx/Us8221u+hx93JJ8kU+\nWZTljS1ldNqdXH9hOhHGYB/UUAghxHhxxmFzi8XCXXfdxYYNG3j00Uc5duwYNTU1fPvb3+bjjz8e\nzjqOuL5FWdRBnB8/b8jlHWvoYNP+GuKiQlg2O9EHNRRCCDGeDHifN8CcOXP4zW9+w5YtW1iyZAl/\n+tOf/F2vUeWAtYDWnjbOmzBnyIuyKIrCix8cRVHgpuUZaDWD+hUIIYQQfb5UchiNRlauXMmrr77q\nr/qMShurvIuyXDhx4ZDL2lnUQHG1jZmZFvJSon1QOyGEEOONdPsGUGarpKL9GHnmbGJCLEMqy97j\n4tVNJQRp1axcKpPUhBBCnBsJ7wFsPPYJABf5YFGWf2+vwNbZy6XnTcIcYRhyeUIIIcYnCe+zaLa3\ncMBawERjPOkRQ1tzvK65iw92V2EO1/PVeUk+qqEQQojxSML7LDZXH1+UJWloi7IoisJLHxbj9iis\nvCgDXZBv9v8WQggxPkl4n4Hd5WB77S7CdWFDXpRlf3ETheUt5KVEMSPDN5uZCCGEGL8kvM9gR+0u\nHO4eLkhcgFY94BLwZ9TrdPPKR8Vo1CpuXJYxrlepE0II4RsS3qfh9rjZVH18UZaE84ZU1rufHqPJ\n5uArcyYyITrURzUUQggxnkl4n8bBpkJaHK3MmzCb0CEsytLUZuc/OyuJMOq4fEGy7yoohBBiXJPw\nPo2Nx07s2X3+kMp5ZWMJTpeH6y9MxxB87kPvQgghxOdJeH9Bua2S8vZKppiziR3CoiwF5c3sO2ol\nMzGceTmxPqyhEEKI8U7C+ws2Vnl73UuHsCiLy+3hpQ+KUangpuWZMklNCCGET0l4f06zvYX9jfkk\nGuPJiEg753I+2FNFfUs3S2ckkhRr8mENhRBCCAnvfvoWZZm46Jx7y60dPby1rQKjIYirFqf4uIZC\nCCGEhHcf76IsuwnXmZgVO+2cy3ltcwk9vW6uXZJGqD7IhzUUQgghvCS8j9tRtxuH28HixIXnvCjL\n0ao2dhY2kBxn4vypE3xcQyGEEMJLwhvwKB42V20lSB3E+QnzzqkMRVF46YOjANz8lUzUMklNCCGE\nn0h4AwethTQ7Wpk3YRbGoHNbBa2x1c6xxk5mZlpIiw/3cQ2FEEKIkyS8gY1V3j27lyae+6IsJTU2\nALKSInxSJyGEEOJMxn14l9uOUWarJC86i9jQmHMup/R4eKcnSq9bCCGEf4378N7UtyjL4iGVU1LT\njk6rJtFi9EW1hBBCiDMa1+Hd4mhlvzWfBOMEMiPPfVEWe4+LGmsnKRPC0GrG9UcqhBBiGIzrpNlc\nvQ2P4hnSoiwAZXXtKEBaggyZCyGE8L9xG952p4NtNbsw6YzMip0+pLJKq73Xu9MSwnxRNSGEEOKs\nxm14byrfjsPt4IKEhQSd46IsJ5TUnghv6XkLIYTwv3EZ3h7Fw3+ObiRIrWVRwnlDLEuhtKadmEgD\nYSE6H9VQCCGEOLNxGd5HW0tp7GpmbtwsjLpzW5TlhLrmbuw9LtKl1y2EEGKYDG28eIyKC41hSfJ8\nlicsHXJZJ+7vliFzIYQQw2Vc9rwjgsO5Z96tRAQPPXBPrKwmPW8hhBDDZVyGty+V1tjQ6zQkmIc2\n/C6EEEIMloT3EHTandQ1d5MaH4ZaLbuICSGEGB4S3kNQduIWMdlFTAghxDCS8B6Ckpp2QDYjEUII\nMbwkvIfgxEzz1HhZWU0IIcTwkfA+Rx6PQlldO/HmUEL1QSNdHSGEEOOIhPc5qrZ20tPrJk163UII\nIYaZhPc5KpX7u4UQQowQv66w9uijj3Lw4EFUKhWrV69m6tSpfceWLl1KXFwcGo0GgMcff5zY2Fh/\nVsenTkxWk5XVhBBCDDe/hfeuXbuorKxk7dq1lJaWsnr1atauXdvvPc888wyhoWNzcZPSGhuhei1x\n0SEjXRUhhBDjjN+GzXfs2MGyZcsASEtLw2az0dnZ6a/TDav2rl4a2+ykxoejVsniLEIIIYaX38K7\nqamJyMjIvudRUVFYrdZ+71mzZg033ngjjz/+OIqi+KsqPnfyerdMVhNCCDH8hm1XsS+G83e/+10W\nLVpEeHg43/nOd9iwYQOXXHLJGX8+MjIErVbj0zpZLKZz+rnaT48BMCtnwjmX4U+jsU5DFYhtgsBs\nl7Rp7AjEdgVim07Hb+EdExNDU1NT3/PGxkYsFkvf86uuuqrv8eLFizl69OhZw7u1tdun9bNYTFit\nHef0s/nFVlQqiAzRnnMZ/jKUdo1WgdgmCMx2SZvGjkBsV6C26XT8Nmy+cOFCNmzYAEBhYSExMTEY\njUYAOjo6uOOOO+jt7QVg9+7dZGRk+KsqPuVyeyiv7yDRYsQQPC63QxdCCDHC/JY+M2fOJDc3l5Ur\nV6JSqVizZg3r1q3DZDKxfPlyFi9ezA033EBwcDA5OTln7XWPJlWNnThdHrlFTAghxIjxa9fxwQcf\n7Pc8Kyur7/Ftt93Gbbfd5s/T+0WJTFYTQggxwmSFtS/pxExz6XkLIYQYKRLeX1JpjQ1TSBAxEYaR\nrooQQohxSsL7S2jt6KG5vYe0+HBUsjiLEEKIESLh/SX0Lc6SKEPmQgghRo6E95dwYrKabAMqhBBi\nJEl4fwmlNTY0ahXJEyS8hRBCjBwJ70FyutxU1HcwMcZIcJBvl2kVQgghvgwJ70GqrO/E7VFIl1vE\nhBBCjDAJ70Eqkfu7hRBCjBIS3oN0chtQCW8hhBAjS8J7EBRFoaTGRoRRR1RY8EhXRwghxDgn4T0I\nzTYHtq5e0hNkcRYhhCQSCSsAAA9pSURBVBAjT8J7EEpq5Xq3EEKI0UPCexBKq9sBud4thBBidJDw\nHoSSWhtajYqkWNNIV0UIIYSQ8B5IT6+bqoZOkuPCCNLKxyWEEGLkSRoNoKK+HY+ikJYgS6IKIYQY\nHSS8B1Ai93cLIYQYZSS8B1Ba452sJjPNhRBCjBYS3mdxYnGW6DA9EUZZnEUIIcToIOF9Fo2tdjrt\nTtITpdcthBBi9JDwPou+zUjiZbKaEEKI0UPC+yz6NiORnrcQQohRRML7LEpqbOi0ahItxpGuihBC\nCNFHwvsM7D0uaqxdpEwIQ6uRj0kIIcToIal0BmW17SjILWJCCCFGHwnvMyiVxVmEEEKMUhLeZ3Bi\npnmqLIsqhBBilJHwPg2PolBa205spIGwEN1IV0cIIYToR8L7NOqaurD3uOR6txBCiFFJwvs0Smu9\n65nL9W4hhBCjkYT3aZRUH19ZTcJbCCHEKCThfRqltTb0Og0J5tCRror4/+3de1BU9f/H8eeyqyiI\nCrrghZ+VeHcqtSQvKWpqat9u00zJhFaDU3ltTEV0QmgauSg6mjalZlZ4Gc2YxspGp6k/rADNHEx0\nfobm/QasgPg1keX8/kD3BwKKKe6e3dfjr875nN3enz7n9OJ8ztlzRESkFoX3TcquXONs0X/p3KEl\nfn4Wd5cjIiJSi8L7JkfP6PfdIiLi2RTeN3G9SUzhLSIiHkrhfZMjp6vuNNdrQEVExFMpvKtxVlZy\n9GwpHdoGEtCsibvLERERqZPCu5rTBZe5Wu6kix6JKiIiHkzhXc2Nl5FEdND1bhER8VwK72p0s5qI\niJiBwruaI6dLCWxmo12bAHeXIiIiUi+F93Wll8u5UHyFzh1a4WfRw1lERMRzKbyvu3G9WzeriYiI\np1N4X6fr3SIiYhYK7+uOnC7BYoGH2uvMW0REPJvCG6hwVvL3uUuE21vQ3N/m7nJERERuSeENnLxQ\nxrWKSr2MRERETEHhDeSfunG9W1PmIiLi+RTewBG9BlREREykUcM7OTmZV155hfHjx7N///46t1my\nZAkTJkxozDJuK/90CUEBTbC3bu7WOkRERBqi0cJ79+7dHD9+nM2bN7Nw4UIWLlxYa5v8/Hz27NnT\nWCU0iKP0HxylV+nSsRUWPZxFRERMoNHCOysri5EjRwIQERFBSUkJZWVlNbZJTU1l5syZjVVCgxw5\nc/393ZoyFxERk2i08C4sLCQ4ONi1HBISQkFBgWs5MzOTyMhIOnbs2FglNMj/P1lN4S0iIuZw337U\nbBiG65+Li4vJzMxk3bp1nD9/vkGfDw4OwGaz3tOa7PYgjp8vw+pn4fGHO+Df5N5+v7vY7UHuLuGe\n88Y+gXf2S30yD2/slzf2qS6NFt6hoaEUFha6li9cuIDdbgcgOzsbh8PBq6++Snl5OSdOnCA5OZn5\n8+fX+30XL/73ntZntwdx5mwx+aeK6RTWgtLie/v97mK3B1FQcMndZdxT3tgn8M5+qU/m4Y398tY+\n1aXRps0HDx7Mjh07AMjLyyM0NJQWLVoAMGbMGLZv386WLVtYuXIlvXv3vmVwN5Zj5y7hrDSI6KAp\ncxERMY9GO/Pu168fvXv3Zvz48VgsFhITE8nMzCQoKIhRo0Y11r/2jhw5XXWzWpdwhbeIiJhHo17z\nnj17do3lHj161NomPDycjIyMxiyjXq43ienMW0RETMRnn7BmGAZHTpcQHORPSEt/d5cjIiLSYD4b\n3hcuXqHkcjkRHVrq4SwiImIqPhveh445AD2cRUREzMdnw/t/r4e3Hs4iIiJm47Phfei4A5vVQqcw\n3/hBv4iIeA+fDO+r5U7+PlPKg+1a0sTmk/8JRETExHwyuf4+W0plpUFEx5buLkVEROSO+WR4+/lZ\nsFkt9OnS1t2liIiI3LH79mIST9Ltf1qzNeU/OByX3V2KiIjIHfPJM28Aq9Vnuy4iIianBBMRETEZ\nhbeIiIjJKLxFRERMRuEtIiJiMgpvERERk1F4i4iImIzCW0RExGQU3iIiIiaj8BYRETEZhbeIiIjJ\nKLxFRERMxmIYhuHuIkRERKThdOYtIiJiMgpvERERk1F4i4iImIzCW0RExGQU3iIiIiaj8BYRETEZ\nm7sLuB+Sk5PJzc3FYrEwf/58HnnkEVfbb7/9xtKlS7FarQwdOpSpU6e6sdKGW7RoEXv37qWiooK3\n3nqL0aNHu9pGjBhBu3btsFqtAKSnpxMWFuauUhssJyeHd955h65duwLQrVs3EhISXO1mHKuvvvqK\nbdu2uZYPHDjAvn37XMu9e/emX79+ruXPP//cNW6e6PDhw0yZMoXXX3+dmJgYzp49S1xcHE6nE7vd\nzuLFi2natGmNz9zq+PMEdfVp3rx5VFRUYLPZWLx4MXa73bX97fZTT3Fzv+Lj48nLy6N169YAxMbG\nMmzYsBqfMdtYzZgxg4sXLwJQXFxMnz59+OCDD1zbZ2Zmsnz5cjp16gTAoEGDmDx5sltqv+cML5eT\nk2O8+eabhmEYRn5+vvHyyy/XaB87dqxx5swZw+l0GtHR0cZff/3ljjLvSFZWljFp0iTDMAzD4XAY\nUVFRNdqHDx9ulJWVuaGyu5OdnW1Mnz693nYzjlV1OTk5RlJSUo11kZGRbqrmzl2+fNmIiYkx3nvv\nPSMjI8MwDMOIj483tm/fbhiGYSxZssTYsGFDjc/c7vhzt7r6FBcXZ3z//feGYRjG+vXrjbS0tBqf\nud1+6gnq6tfcuXONn376qd7PmHGsqouPjzdyc3NrrPv666+N1NTU+1XifeX10+ZZWVmMHDkSgIiI\nCEpKSigrKwPg5MmTtGrVivbt2+Pn50dUVBRZWVnuLLdB+vfvz/LlywFo2bIlV65cwel0urmqxmXW\nsaruo48+YsqUKe4u419r2rQpa9asITQ01LUuJyeHp556CoDhw4fXGpNbHX+eoK4+JSYm8vTTTwMQ\nHBxMcXGxu8r71+rq1+2YcaxuOHr0KJcuXfK4mYLG5PXhXVhYSHBwsGs5JCSEgoICAAoKCggJCamz\nzZNZrVYCAgIA2Lp1K0OHDq011ZqYmEh0dDTp6ekYJnqIXn5+Pm+//TbR0dH8+uuvrvVmHasb9u/f\nT/v27WtMvwKUl5cza9Ysxo8fz7p169xUXcPYbDaaNWtWY92VK1dc0+Rt2rSpNSa3Ov48QV19CggI\nwGq14nQ62bhxI88++2ytz9W3n3qKuvoFsH79eiZOnMjMmTNxOBw12sw4Vjd8+eWXxMTE1Nm2e/du\nYmNjee211zh48GBjlnhf+cQ17+rMFGS38+OPP7J161Y+++yzGutnzJjBkCFDaNWqFVOnTmXHjh2M\nGTPGTVU23IMPPsi0adMYO3YsJ0+eZOLEiezcubPWNVQz2rp1Ky+++GKt9XFxcTz33HNYLBZiYmJ4\n/PHHefjhh91Q4d1ryLFlluPP6XQSFxfHgAEDGDhwYI02s+6nzz//PK1bt6Znz56sXr2alStXsmDB\ngnq3N8tYlZeXs3fvXpKSkmq1Pfroo4SEhDBs2DD27dvH3Llz+fbbb+9/kY3A68+8Q0NDKSwsdC1f\nuHDBdfZzc9v58+fvaJrJnXbt2sUnn3zCmjVrCAoKqtH2wgsv0KZNG2w2G0OHDuXw4cNuqvLOhIWF\nMW7cOCwWC506daJt27acP38eMPdYQdX0ct++fWutj46OJjAwkICAAAYMGGCasbohICCAf/75B6h7\nTG51/HmyefPm8cADDzBt2rRabbfaTz3ZwIED6dmzJ1B1U+vN+5pZx2rPnj31TpdHRES4bsrr27cv\nDofDay4xen14Dx48mB07dgCQl5dHaGgoLVq0ACA8PJyysjJOnTpFRUUFP//8M4MHD3ZnuQ1y6dIl\nFi1axKpVq1x3jlZvi42Npby8HKjasW/cFevptm3bxtq1a4GqafKioiLXXfJmHSuoCrXAwMBaZ2ZH\njx5l1qxZGIZBRUUFf/zxh2nG6oZBgwa5jq+dO3cyZMiQGu23Ov481bZt22jSpAkzZsyot72+/dST\nTZ8+nZMnTwJVf0zevK+ZcawA/vzzT3r06FFn25o1a/juu++AqjvVQ0JCPPrXHHfCJ94qlp6ezu+/\n/47FYiExMZGDBw8SFBTEqFGj2LNnD+np6QCMHj2a2NhYN1d7e5s3b2bFihU89NBDrnVPPPEE3bt3\nZ9SoUXzxxRd88803+Pv706tXLxISErBYLG6suGHKysqYPXs2paWlXLt2jWnTplFUVGTqsYKqn4ct\nW7aMTz/9FIDVq1fTv39/+vbty+LFi8nOzsbPz48RI0Z49M9YDhw4QFpaGqdPn8ZmsxEWFkZ6ejrx\n8fFcvXqVDh06kJKSQpMmTZg5cyYpKSk0a9as1vFX3/9o3aGuPhUVFeHv7+8KroiICJKSklx9qqio\nqLWfRkVFubknNdXVr5iYGFavXk3z5s0JCAggJSWFNm3amHqsVqxYwYoVK3jssccYN26ca9vJkyfz\n8ccfc+7cOebMmeP6A9kTf/72b/lEeIuIiHgTr582FxER8TYKbxEREZNReIuIiJiMwltERMRkFN4i\nIiIm43NPWBPxVadOnWLMmDG1HhYTFRXFpEmT7vr7c3JyWLZsGZs2bbrr7xKRW1N4i/iQkJAQMjIy\n3F2GiNwlhbeI0KtXL6ZMmUJOTg6XL18mNTWVbt26kZubS2pqKjabDYvFwoIFC+jSpQvHjh0jISGB\nyspK/P39SUlJAaCyspLExEQOHTpE06ZNWbVqFYGBgW7unYj30TVvEcHpdNK1a1cyMjKIjo7mww8/\nBKpenDJv3jwyMjJ44403eP/994Gqt9bFxsayYcMGXnrpJX744QcAjhw5wvTp09myZQs2m41ffvnF\nbX0S8WY68xbxIQ6HgwkTJtRYN2fOHACefPJJAPr168fatWspLS2lqKjI9TjJyMhI3n33XaDqFaeR\nkZEAPPPMM0DVNe/OnTvTtm1bANq1a0dpaWnjd0rEBym8RXzIra55V39SssViqfU8/JufpFxZWVnr\nO7zlpQ8ink7T5iICQHZ2NgB79+6le/fuBAUFYbfbyc3NBSArK4s+ffoAVWfnu3btAmD79u0sXbrU\nPUWL+CideYv4kLqmzcPDwwE4ePAgmzZtoqSkhLS0NADS0tJITU3FarXi5+dHUlISAAkJCSQkJLBx\n40ZsNhvJycmcOHHivvZFxJfprWIiQvfu3cnLy8Nm09/zImagaXMRERGT0Zm3iIiIyejMW0RExGQU\n3iIiIiaj8BYRETEZhbeIiIjJKLxFRERMRuEtIiJiMv8HhOiqzS0/c+MAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f02c5e58390>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "4lhlEGpKLKr0",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- 最後的模型推論準確率"
      ]
    },
    {
      "metadata": {
        "id": "2FHMn4tBK1OH",
        "colab_type": "code",
        "outputId": "b84cb431-c998-422f-e4aa-af07e39c7394",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "cell_type": "code",
      "source": [
        "learning.history['val_acc'][-1]"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.7796"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 17
        }
      ]
    },
    {
      "metadata": {
        "id": "0KGo4Lw9MHcc",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- 用model.evaluate函數評估模型對10000筆測試資料的推論準確率，約略同上"
      ]
    },
    {
      "metadata": {
        "id": "nOcTHXj57vLa",
        "colab_type": "code",
        "outputId": "39645e5f-7c10-41eb-9eb2-8c02c78a1c01",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "cell_type": "code",
      "source": [
        "loss, accuracy = model.evaluate(x_test, y_test_onehot)\n",
        "accuracy"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "10000/10000 [==============================] - 1s 133us/step\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.7744"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 18
        }
      ]
    },
    {
      "metadata": {
        "id": "2JBEUjMqL_jU",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Inference"
      ]
    },
    {
      "metadata": {
        "id": "BbVLNCkMfVF-",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- 以模型對測試資料取得推論為例  \n",
        "用model.predict函數取得推論，為10000筆10維向量  \n",
        "用model.predict_classes取得推論，為10000筆純量  \n",
        "查看第0筆的獨熱編碼及其類別是一致的"
      ]
    },
    {
      "metadata": {
        "id": "d5XY25-mUXH4",
        "colab_type": "code",
        "outputId": "9aa767bf-f991-47aa-af12-d72cc2497d91",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 102
        }
      },
      "cell_type": "code",
      "source": [
        "prediction = model.predict(x_test)\n",
        "classification = model.predict_classes(x_test)\n",
        "print(prediction.shape)\n",
        "print(classification.shape)\n",
        "print(prediction[0])\n",
        "print(classification[0])"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(10000, 10)\n",
            "(10000,)\n",
            "[1.7070312e-02 5.4837577e-03 7.8360259e-04 7.5729650e-01 4.2174882e-04\n",
            " 1.6835628e-01 1.5961150e-02 3.0863092e-03 4.3150582e-03 2.7225273e-02]\n",
            "3\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "sDq28SY3kCTd",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- 任意看15張圖的推論結果正確與否"
      ]
    },
    {
      "metadata": {
        "id": "M_OVSHnqGsN_",
        "colab_type": "code",
        "outputId": "55894c44-44b9-45e1-b182-dd58fef8d036",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 334
        }
      },
      "cell_type": "code",
      "source": [
        "for i in range(15):\n",
        "  ax = plt.subplot(3, 5, 1+i)\n",
        "  ax.imshow(x_test[i])\n",
        "  ax.set_title('O' if classification[5566+i] == y_test[5566+i][0] else 'X')\n",
        "  ax.set_xticks([])\n",
        "  ax.set_yticks([])"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAcwAAAE9CAYAAACGFVI6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f02c5e6e518>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "hSPrHPbuuYIP",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Save"
      ]
    },
    {
      "metadata": {
        "id": "qOpZeWSD1dqb",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- Save model to a JSON file"
      ]
    },
    {
      "metadata": {
        "id": "DO4Tfv1puTbB",
        "colab_type": "code",
        "outputId": "a94e009c-709e-4bee-96cf-cf0d31d524f1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "cell_type": "code",
      "source": [
        "!ls\n",
        "model_json = model.to_json()\n",
        "with open('cifar10_model.json', 'w') as f:\n",
        "  f.write(model_json)\n",
        "!ls"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "sample_data\n",
            "cifar10_model.json  sample_data\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "7RYJK_tf1c-C",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- Save params to a H5 file"
      ]
    },
    {
      "metadata": {
        "id": "dKAu4KVR1cdp",
        "colab_type": "code",
        "outputId": "1965b889-8047-4bca-f282-1a3213583b9e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "cell_type": "code",
      "source": [
        "!ls\n",
        "model.save_weights('cifar10_params.h5')\n",
        "!ls"
      ],
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "cifar10_model.json  sample_data\n",
            "cifar10_model.json  cifar10_params.h5  sample_data\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "UqUW0P_93dlw",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- Save params+model to a H5 file"
      ]
    },
    {
      "metadata": {
        "id": "UKHQA_gR3imH",
        "colab_type": "code",
        "outputId": "73f17288-8282-4758-cd7b-ce20fb60236f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "cell_type": "code",
      "source": [
        "!ls\n",
        "model.save('cifar10_params_model.h5')\n",
        "!ls"
      ],
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "cifar10_model.json  cifar10_params.h5  sample_data\n",
            "cifar10_model.json  cifar10_params.h5  cifar10_params_model.h5\tsample_data\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "1cao5noVlv57",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Download to local drive"
      ]
    },
    {
      "metadata": {
        "id": "HGfjM65Jv2dY",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- Colab runtime 和 Chrome 斷線一段時間會被重置，儲存在 Colab runtime 中的檔案不會被保留，而這個 ipynb(IPython/Jupyter Notebook) 還是會存在你的雲端硬碟上\n",
        "37. 點開左側面板，點到Files，在此介面操作檔案的上傳及下載\n",
        "38. 也可以用 google.colab 的 files.upload() 和 files.download()"
      ]
    },
    {
      "metadata": {
        "id": "V2Jx8jSwwGSQ",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "from google.colab import files\n",
        "files.download('cifar10_params_model.h5')"
      ],
      "execution_count": 24,
      "outputs": []
    },
    {
      "metadata": {
        "id": "YKr2whZSmOpI",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Download to Google Drive"
      ]
    },
    {
      "metadata": {
        "id": "KzUIam89mp1e",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- Colab runtime 可以掛載 Google Drive 為一個目錄"
      ]
    },
    {
      "metadata": {
        "id": "6ulTkIwtnoqJ",
        "colab_type": "code",
        "outputId": "23a483c4-e032-4ce1-a819-4fac5e9b900d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "cell_type": "code",
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/gdrive')"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Go to this URL in a browser: https://accounts.google.com/o/oauth2/auth?client_id=947318989803-6bn6qk8qdgf4n4g3pfee6491hc0brc4i.apps.googleusercontent.com&redirect_uri=urn%3Aietf%3Awg%3Aoauth%3A2.0%3Aoob&scope=email%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdocs.test%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive.photos.readonly%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fpeopleapi.readonly&response_type=code\n",
            "\n",
            "Enter your authorization code:\n",
            "··········\n",
            "Mounted at /content/gdrive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "0yzI16beohQI",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- 將要存下來的檔案備份在gdrive/My Drive目錄，即使runtime被重置，檔案還是會存在你的雲端硬碟上"
      ]
    },
    {
      "metadata": {
        "id": "agBv3OclpvoN",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "!cp cifar10* gdrive/My\\ Drive"
      ],
      "execution_count": 26,
      "outputs": []
    }
  ]
}