Captcha_Solver_Alphanumerical.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Captcha_Solver_Alphanumerical.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "6Tzheb9xoLiD",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "2a196451-10bd-491a-fa67-2d3872867c3c"
      },
      "source": [
        "%pip install captcha \n",
        "%pip install opencv-python\n",
        "!apt update && apt install -y libsm6 libxext6\n",
        "!pip install opencv-python-headless\n",
        "!pip install opencv-contrib-python-headless\n",
        "%pip install keras\n",
        "\n",
        "\n",
        "import argparse\n",
        "import json\n",
        "import string\n",
        "import os\n",
        "import shutil\n",
        "import uuid\n",
        "from captcha.image import ImageCaptcha\n",
        "\n",
        "import itertools\n",
        "\n",
        "import os\n",
        "import cv2\n",
        "import numpy as np\n",
        "from random import random, randint, choices\n",
        "\n",
        "import keras\n",
        "from keras.models import Sequential, Model\n",
        "from keras.layers import Dense, Dropout, Activation, Flatten\n",
        "from keras.layers import Conv2D, MaxPooling2D, Input\n",
        "import matplotlib.pyplot as plt"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting captcha\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/90/fe/d4ddf1e6576073b5eaea76e9b2afa022c626212a30c871968480be3ccb7b/captcha-0.3-py3-none-any.whl (101kB)\n",
            "\r\u001b[K     |███▎                            | 10kB 23.9MB/s eta 0:00:01\r\u001b[K     |██████▌                         | 20kB 3.0MB/s eta 0:00:01\r\u001b[K     |█████████▊                      | 30kB 4.1MB/s eta 0:00:01\r\u001b[K     |█████████████                   | 40kB 4.4MB/s eta 0:00:01\r\u001b[K     |████████████████▏               | 51kB 3.6MB/s eta 0:00:01\r\u001b[K     |███████████████████▍            | 61kB 4.1MB/s eta 0:00:01\r\u001b[K     |██████████████████████▋         | 71kB 4.3MB/s eta 0:00:01\r\u001b[K     |█████████████████████████▉      | 81kB 4.7MB/s eta 0:00:01\r\u001b[K     |█████████████████████████████   | 92kB 5.0MB/s eta 0:00:01\r\u001b[K     |████████████████████████████████| 102kB 3.7MB/s \n",
            "\u001b[?25hRequirement already satisfied: Pillow in /usr/local/lib/python3.6/dist-packages (from captcha) (7.0.0)\n",
            "Installing collected packages: captcha\n",
            "Successfully installed captcha-0.3\n",
            "Requirement already satisfied: opencv-python in /usr/local/lib/python3.6/dist-packages (4.1.2.30)\n",
            "Requirement already satisfied: numpy>=1.11.3 in /usr/local/lib/python3.6/dist-packages (from opencv-python) (1.18.5)\n",
            "Get:1 https://cloud.r-project.org/bin/linux/ubuntu bionic-cran35/ InRelease [3,626 B]\n",
            "Ign:2 https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64  InRelease\n",
            "Ign:3 https://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1804/x86_64  InRelease\n",
            "Hit:4 https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64  Release\n",
            "Hit:5 https://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1804/x86_64  Release\n",
            "Get:6 http://security.ubuntu.com/ubuntu bionic-security InRelease [88.7 kB]\n",
            "Hit:7 http://ppa.launchpad.net/graphics-drivers/ppa/ubuntu bionic InRelease\n",
            "Hit:8 http://archive.ubuntu.com/ubuntu bionic InRelease\n",
            "Get:10 http://archive.ubuntu.com/ubuntu bionic-updates InRelease [88.7 kB]\n",
            "Get:11 http://ppa.launchpad.net/marutter/c2d4u3.5/ubuntu bionic InRelease [15.4 kB]\n",
            "Get:13 http://archive.ubuntu.com/ubuntu bionic-backports InRelease [74.6 kB]\n",
            "Get:14 http://ppa.launchpad.net/marutter/c2d4u3.5/ubuntu bionic/main Sources [1,853 kB]\n",
            "Get:15 http://archive.ubuntu.com/ubuntu bionic-updates/main amd64 Packages [1,334 kB]\n",
            "Get:16 http://archive.ubuntu.com/ubuntu bionic-updates/multiverse amd64 Packages [27.1 kB]\n",
            "Get:17 http://archive.ubuntu.com/ubuntu bionic-updates/universe amd64 Packages [1,413 kB]\n",
            "Get:18 http://archive.ubuntu.com/ubuntu bionic-updates/restricted amd64 Packages [116 kB]\n",
            "Get:19 http://ppa.launchpad.net/marutter/c2d4u3.5/ubuntu bionic/main amd64 Packages [894 kB]\n",
            "Fetched 5,909 kB in 4s (1,633 kB/s)\n",
            "Reading package lists... Done\n",
            "Building dependency tree       \n",
            "Reading state information... Done\n",
            "38 packages can be upgraded. Run 'apt list --upgradable' to see them.\n",
            "Reading package lists... Done\n",
            "Building dependency tree       \n",
            "Reading state information... Done\n",
            "libsm6 is already the newest version (2:1.2.2-1).\n",
            "libxext6 is already the newest version (2:1.3.3-1).\n",
            "The following package was automatically installed and is no longer required:\n",
            "  libnvidia-common-440\n",
            "Use 'apt autoremove' to remove it.\n",
            "0 upgraded, 0 newly installed, 0 to remove and 38 not upgraded.\n",
            "Collecting opencv-python-headless\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/17/e4/a98a3c3098ea55b6ae193a1cd19a221dc3c1bde87a36db5550addc879d36/opencv_python_headless-4.3.0.36-cp36-cp36m-manylinux2014_x86_64.whl (36.4MB)\n",
            "\u001b[K     |████████████████████████████████| 36.4MB 83kB/s \n",
            "\u001b[?25hRequirement already satisfied: numpy>=1.11.3 in /usr/local/lib/python3.6/dist-packages (from opencv-python-headless) (1.18.5)\n",
            "Installing collected packages: opencv-python-headless\n",
            "Successfully installed opencv-python-headless-4.3.0.36\n",
            "Collecting opencv-contrib-python-headless\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/a1/90/b2982491b86fc8f39153162da5775081361ce731156df9a248c55396c5c8/opencv_contrib_python_headless-4.3.0.36-cp36-cp36m-manylinux2014_x86_64.whl (42.6MB)\n",
            "\u001b[K     |████████████████████████████████| 42.6MB 77kB/s \n",
            "\u001b[?25hRequirement already satisfied: numpy>=1.11.3 in /usr/local/lib/python3.6/dist-packages (from opencv-contrib-python-headless) (1.18.5)\n",
            "Installing collected packages: opencv-contrib-python-headless\n",
            "Successfully installed opencv-contrib-python-headless-4.3.0.36\n",
            "Requirement already satisfied: keras in /usr/local/lib/python3.6/dist-packages (2.3.1)\n",
            "Requirement already satisfied: six>=1.9.0 in /usr/local/lib/python3.6/dist-packages (from keras) (1.15.0)\n",
            "Requirement already satisfied: numpy>=1.9.1 in /usr/local/lib/python3.6/dist-packages (from keras) (1.18.5)\n",
            "Requirement already satisfied: keras-preprocessing>=1.0.5 in /usr/local/lib/python3.6/dist-packages (from keras) (1.1.2)\n",
            "Requirement already satisfied: scipy>=0.14 in /usr/local/lib/python3.6/dist-packages (from keras) (1.4.1)\n",
            "Requirement already satisfied: h5py in /usr/local/lib/python3.6/dist-packages (from keras) (2.10.0)\n",
            "Requirement already satisfied: pyyaml in /usr/local/lib/python3.6/dist-packages (from keras) (3.13)\n",
            "Requirement already satisfied: keras-applications>=1.0.6 in /usr/local/lib/python3.6/dist-packages (from keras) (1.0.8)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yuX06GVdok0-",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "alphabet_all = list('qwertyupasdfghjkzxcvbnm23456789QWERTYUPKJHGFDSAZXCVBNM')\n",
        "alphabet = list('qwertyupasdfghjkzxcvbnm23456789')#QWERTYUIOPLKJHGFDSAZXCVBNM')\n",
        "num_alphabet = len(alphabet)\n",
        "def _gen_captcha(img_dir, num_of_letters, num_of_repetition, width, height):\n",
        "    if os.path.exists(img_dir):\n",
        "        shutil.rmtree(img_dir)\n",
        "    if not os.path.exists(img_dir):\n",
        "        os.makedirs(img_dir)\n",
        "\n",
        "    image = ImageCaptcha(width=width, height=height)\n",
        "\n",
        "    for counter in range(num_of_repetition):\n",
        "        i = choices(alphabet_all, k=5)\n",
        "        captcha = ''.join(i)\n",
        "        fn = os.path.join(img_dir, '%s_%s.png' % (captcha, uuid.uuid4()))\n",
        "        image.write(captcha, fn)\n",
        "\n",
        "\n",
        "def gen_dataset(path, num_of_repetition, num_of_letters, width, height):\n",
        "    _gen_captcha(os.path.join(path, 'data'), num_of_letters, num_of_repetition, width, height)\n",
        "    print('Finished Data Generation')"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MQYOsJyioy5g",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "BATCH_SIZE = 128\n",
        "NUM_OF_LETTERS = 5\n",
        "EPOCHS = 50\n",
        "IMG_ROW, IMG_COLS = 50, 135\n",
        "\n",
        "# Non-configs\n",
        "PATH = os.getcwd()\n",
        "DATA_PATH = os.path.join(PATH, 'train')"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ItSQ3uTvo5YW",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def load_data(path, test_split=0.1):\n",
        "    print ('loading dataset...')\n",
        "    y_train = []\n",
        "    y_test = []\n",
        "    x_train = []\n",
        "    x_test = []\n",
        "\n",
        "    # r=root, d=directories, f = files\n",
        "    counter = 0\n",
        "    for r, d, f in os.walk(path):\n",
        "        for fl in f:\n",
        "            if '.png' in fl:\n",
        "                flr = fl.split('_')[0]\n",
        "                counter += 1\n",
        "                label = np.zeros((NUM_OF_LETTERS, num_alphabet))\n",
        "                for i in range(NUM_OF_LETTERS):\n",
        "                    label[i, alphabet.index(flr[i].lower())] = 1\n",
        "#                 label = np.zeros((50, 1))\n",
        "#                 for i in range(5):\n",
        "#                     label[i*5+int(flr[i])] = 1\n",
        "\n",
        "                img = cv2.imread(os.path.join(r, fl))\n",
        "                img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)\n",
        "                img = cv2.resize(img, (int(135/2), int(50/2)), interpolation=cv2.INTER_AREA)\n",
        "                img = np.reshape(img, (img.shape[0], img.shape[1], 1))\n",
        "\n",
        "                if random() < test_split:\n",
        "                    y_test.append(label)\n",
        "                    x_test.append(img)\n",
        "                else:\n",
        "                    y_train.append(label)\n",
        "                    x_train.append(img)\n",
        "\n",
        "    print('dataset size:', counter, '(train=%d, test=%d)' % (len(y_train), len(y_test)))\n",
        "    return np.array(x_train), np.array(y_train), np.array(x_test), np.array(y_test)"
      ],
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bCqtzWWvo798",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "07d26802-aedc-4f24-8ff6-e8458f0c2125"
      },
      "source": [
        "if not os.path.exists(DATA_PATH):\n",
        "    print('Generating Dataset')\n",
        "    gen_dataset(DATA_PATH, 700*1000 , NUM_OF_LETTERS, IMG_COLS, IMG_ROW)"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Generating Dataset\n",
            "Finished Data Generation\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AenCMkmco-jv",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "77cd1965-6095-4341-a0c2-3259788fa71a"
      },
      "source": [
        "x_train, y_train, x_test, y_test = load_data(DATA_PATH)\n",
        "\n",
        "x_train = x_train.astype('float32')\n",
        "x_test = x_test.astype('float32')\n",
        "x_train /= 255\n",
        "x_test /= 255"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "loading dataset...\n",
            "dataset size: 10000 (train=8953, test=1047)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zqV-rssUuhIy",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 88
        },
        "outputId": "cebc9f7e-f5a8-42a8-882d-80e1b7b0b131"
      },
      "source": [
        "print(x_train.shape)\n",
        "print(y_train.shape)\n",
        "print(x_test.shape)\n",
        "print(y_test.shape)"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(8953, 25, 67, 1)\n",
            "(8953, 5, 31)\n",
            "(1047, 25, 67, 1)\n",
            "(1047, 5, 31)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eASawvTyvMaL",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "s_train = []\n",
        "s_test = []\n",
        "for i in range(NUM_OF_LETTERS):\n",
        "    s_train.append(y_train[:, i, :])\n",
        "    s_test.append(y_test[:, i, :])"
      ],
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DCom2NFKvVF0",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "save_dir = os.path.join(PATH, 'saved_models')\n",
        "model_name = 'keras_cifar10_trained_model.h5'"
      ],
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XS0vWhEuvXgo",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "input_layer = Input((25, 67, 1))\n",
        "x = Conv2D(filters=32, kernel_size=(5, 5), padding='same', activation='relu')(input_layer)\n",
        "x = MaxPooling2D(pool_size=(2, 2))(x)\n",
        "\n",
        "x = Conv2D(filters=48, kernel_size=(5, 5), padding='same', activation='relu')(x)\n",
        "x = MaxPooling2D(pool_size=(2, 2))(x)\n",
        "\n",
        "x = Conv2D(filters=64, kernel_size=(5, 5), padding='same', activation='relu')(x)\n",
        "x = MaxPooling2D(pool_size=(2, 2))(x)\n",
        "\n",
        "x = Dropout(0.3)(x)\n",
        "x = Flatten()(x)\n",
        "x = Dense(512, activation='relu')(x)\n",
        "x = Dropout(0.3)(x)\n",
        "\n",
        "out = [Dense(num_alphabet, name='digit%d' % i, activation='softmax')(x) for i in range(NUM_OF_LETTERS)]\n",
        "# out = Dense(num_alphabet*5, activation='sigmoid')(x)\n",
        "\n",
        "model = Model(inputs=input_layer, outputs=out)"
      ],
      "execution_count": 10,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_1aTZ32pvZpl",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# model_path = os.path.join(save_dir, model_name)\n",
        "# model = keras.models.load_model(model_path)"
      ],
      "execution_count": 11,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mvyFxZTlvfKR",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 728
        },
        "outputId": "e7e68721-8767-4bca-f1fb-7d02e7b68dc0"
      },
      "source": [
        "# initiate Adam optimizer\n",
        "\n",
        "model.compile(loss='binary_crossentropy',\n",
        "              optimizer='adam',\n",
        "              metrics=['accuracy'])\n",
        "\n",
        "\n",
        "model.summary() "
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"model_1\"\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_1 (InputLayer)            (None, 25, 67, 1)    0                                            \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_1 (Conv2D)               (None, 25, 67, 32)   832         input_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2D)  (None, 12, 33, 32)   0           conv2d_1[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_2 (Conv2D)               (None, 12, 33, 48)   38448       max_pooling2d_1[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2D)  (None, 6, 16, 48)    0           conv2d_2[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_3 (Conv2D)               (None, 6, 16, 64)    76864       max_pooling2d_2[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "max_pooling2d_3 (MaxPooling2D)  (None, 3, 8, 64)     0           conv2d_3[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "dropout_1 (Dropout)             (None, 3, 8, 64)     0           max_pooling2d_3[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "flatten_1 (Flatten)             (None, 1536)         0           dropout_1[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "dense_1 (Dense)                 (None, 512)          786944      flatten_1[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "dropout_2 (Dropout)             (None, 512)          0           dense_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "digit0 (Dense)                  (None, 31)           15903       dropout_2[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "digit1 (Dense)                  (None, 31)           15903       dropout_2[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "digit2 (Dense)                  (None, 31)           15903       dropout_2[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "digit3 (Dense)                  (None, 31)           15903       dropout_2[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "digit4 (Dense)                  (None, 31)           15903       dropout_2[0][0]                  \n",
            "==================================================================================================\n",
            "Total params: 982,603\n",
            "Trainable params: 982,603\n",
            "Non-trainable params: 0\n",
            "__________________________________________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xNW40AXivcK7",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "hist_train_loss_digit = {i:[] for i in range(5)}\n",
        "hist_test_loss_digit = {i:[] for i in range(5)}\n",
        "\n",
        "hist_train_acc_digit = {i:[] for i in range(5)}\n",
        "hist_test_acc_digit = {i:[] for i in range(5)}\n",
        "\n",
        "hist_train_loss = []\n",
        "hist_test_loss = []\n",
        "\n",
        "hist_train_acc = []\n",
        "hist_test_acc = []"
      ],
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VFeIxaatvhap",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "digit_acc = [[] for _ in range(NUM_OF_LETTERS)]\n",
        "val_digit_acc = [[] for _ in range(NUM_OF_LETTERS)]\n",
        "loss = []\n",
        "val_loss = []"
      ],
      "execution_count": 14,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IQTy-mCRvjIz",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "ed453b3f-f6a8-40f5-932a-50064fbac4cb"
      },
      "source": [
        "history = model.fit(x_train, s_train,\n",
        "                    batch_size=BATCH_SIZE,\n",
        "                    epochs=EPOCHS,\n",
        "                    verbose=1,\n",
        "                    validation_data=(x_test, s_test)\n",
        "                   )"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 8953 samples, validate on 1047 samples\n",
            "Epoch 1/50\n",
            "8953/8953 [==============================] - 9s 1ms/step - loss: 0.7088 - digit0_loss: 0.1418 - digit1_loss: 0.1418 - digit2_loss: 0.1418 - digit3_loss: 0.1417 - digit4_loss: 0.1418 - digit0_accuracy: 0.9677 - digit1_accuracy: 0.9677 - digit2_accuracy: 0.9677 - digit3_accuracy: 0.9677 - digit4_accuracy: 0.9677 - val_loss: 0.7077 - val_digit0_loss: 0.1417 - val_digit1_loss: 0.1417 - val_digit2_loss: 0.1411 - val_digit3_loss: 0.1419 - val_digit4_loss: 0.1416 - val_digit0_accuracy: 0.9677 - val_digit1_accuracy: 0.9677 - val_digit2_accuracy: 0.9677 - val_digit3_accuracy: 0.9677 - val_digit4_accuracy: 0.9677\n",
            "Epoch 2/50\n",
            "8953/8953 [==============================] - 1s 158us/step - loss: 0.7073 - digit0_loss: 0.1415 - digit1_loss: 0.1416 - digit2_loss: 0.1415 - digit3_loss: 0.1414 - digit4_loss: 0.1414 - digit0_accuracy: 0.9677 - digit1_accuracy: 0.9677 - digit2_accuracy: 0.9677 - digit3_accuracy: 0.9677 - digit4_accuracy: 0.9677 - val_loss: 0.7075 - val_digit0_loss: 0.1416 - val_digit1_loss: 0.1416 - val_digit2_loss: 0.1412 - val_digit3_loss: 0.1418 - val_digit4_loss: 0.1416 - val_digit0_accuracy: 0.9677 - val_digit1_accuracy: 0.9677 - val_digit2_accuracy: 0.9677 - val_digit3_accuracy: 0.9677 - val_digit4_accuracy: 0.9677\n",
            "Epoch 3/50\n",
            "8953/8953 [==============================] - 1s 152us/step - loss: 0.7071 - digit0_loss: 0.1414 - digit1_loss: 0.1415 - digit2_loss: 0.1414 - digit3_loss: 0.1414 - digit4_loss: 0.1413 - digit0_accuracy: 0.9677 - digit1_accuracy: 0.9677 - digit2_accuracy: 0.9677 - digit3_accuracy: 0.9677 - digit4_accuracy: 0.9677 - val_loss: 0.7072 - val_digit0_loss: 0.1415 - val_digit1_loss: 0.1416 - val_digit2_loss: 0.1411 - val_digit3_loss: 0.1418 - val_digit4_loss: 0.1415 - val_digit0_accuracy: 0.9677 - val_digit1_accuracy: 0.9677 - val_digit2_accuracy: 0.9677 - val_digit3_accuracy: 0.9677 - val_digit4_accuracy: 0.9677\n",
            "Epoch 4/50\n",
            "8953/8953 [==============================] - 1s 150us/step - loss: 0.7062 - digit0_loss: 0.1409 - digit1_loss: 0.1415 - digit2_loss: 0.1414 - digit3_loss: 0.1412 - digit4_loss: 0.1412 - digit0_accuracy: 0.9677 - digit1_accuracy: 0.9677 - digit2_accuracy: 0.9677 - digit3_accuracy: 0.9677 - digit4_accuracy: 0.9677 - val_loss: 0.7043 - val_digit0_loss: 0.1396 - val_digit1_loss: 0.1414 - val_digit2_loss: 0.1410 - val_digit3_loss: 0.1415 - val_digit4_loss: 0.1411 - val_digit0_accuracy: 0.9677 - val_digit1_accuracy: 0.9677 - val_digit2_accuracy: 0.9677 - val_digit3_accuracy: 0.9677 - val_digit4_accuracy: 0.9677\n",
            "Epoch 5/50\n",
            "8953/8953 [==============================] - 1s 150us/step - loss: 0.6975 - digit0_loss: 0.1371 - digit1_loss: 0.1407 - digit2_loss: 0.1405 - digit3_loss: 0.1402 - digit4_loss: 0.1390 - digit0_accuracy: 0.9677 - digit1_accuracy: 0.9677 - digit2_accuracy: 0.9677 - digit3_accuracy: 0.9677 - digit4_accuracy: 0.9677 - val_loss: 0.6844 - val_digit0_loss: 0.1338 - val_digit1_loss: 0.1380 - val_digit2_loss: 0.1392 - val_digit3_loss: 0.1388 - val_digit4_loss: 0.1346 - val_digit0_accuracy: 0.9677 - val_digit1_accuracy: 0.9677 - val_digit2_accuracy: 0.9677 - val_digit3_accuracy: 0.9677 - val_digit4_accuracy: 0.9677\n",
            "Epoch 6/50\n",
            "8953/8953 [==============================] - 1s 155us/step - loss: 0.6734 - digit0_loss: 0.1296 - digit1_loss: 0.1374 - digit2_loss: 0.1374 - digit3_loss: 0.1365 - digit4_loss: 0.1323 - digit0_accuracy: 0.9677 - digit1_accuracy: 0.9677 - digit2_accuracy: 0.9677 - digit3_accuracy: 0.9677 - digit4_accuracy: 0.9678 - val_loss: 0.6492 - val_digit0_loss: 0.1227 - val_digit1_loss: 0.1334 - val_digit2_loss: 0.1347 - val_digit3_loss: 0.1335 - val_digit4_loss: 0.1252 - val_digit0_accuracy: 0.9680 - val_digit1_accuracy: 0.9677 - val_digit2_accuracy: 0.9678 - val_digit3_accuracy: 0.9677 - val_digit4_accuracy: 0.9678\n",
            "Epoch 7/50\n",
            "8953/8953 [==============================] - 1s 150us/step - loss: 0.6364 - digit0_loss: 0.1185 - digit1_loss: 0.1323 - digit2_loss: 0.1326 - digit3_loss: 0.1311 - digit4_loss: 0.1219 - digit0_accuracy: 0.9680 - digit1_accuracy: 0.9677 - digit2_accuracy: 0.9677 - digit3_accuracy: 0.9677 - digit4_accuracy: 0.9678 - val_loss: 0.6166 - val_digit0_loss: 0.1107 - val_digit1_loss: 0.1308 - val_digit2_loss: 0.1307 - val_digit3_loss: 0.1285 - val_digit4_loss: 0.1162 - val_digit0_accuracy: 0.9687 - val_digit1_accuracy: 0.9678 - val_digit2_accuracy: 0.9678 - val_digit3_accuracy: 0.9678 - val_digit4_accuracy: 0.9680\n",
            "Epoch 8/50\n",
            "8953/8953 [==============================] - 1s 151us/step - loss: 0.5902 - digit0_loss: 0.1054 - digit1_loss: 0.1254 - digit2_loss: 0.1267 - digit3_loss: 0.1239 - digit4_loss: 0.1087 - digit0_accuracy: 0.9690 - digit1_accuracy: 0.9679 - digit2_accuracy: 0.9679 - digit3_accuracy: 0.9680 - digit4_accuracy: 0.9686 - val_loss: 0.5526 - val_digit0_loss: 0.0936 - val_digit1_loss: 0.1225 - val_digit2_loss: 0.1211 - val_digit3_loss: 0.1195 - val_digit4_loss: 0.0977 - val_digit0_accuracy: 0.9709 - val_digit1_accuracy: 0.9682 - val_digit2_accuracy: 0.9679 - val_digit3_accuracy: 0.9682 - val_digit4_accuracy: 0.9701\n",
            "Epoch 9/50\n",
            "8953/8953 [==============================] - 1s 153us/step - loss: 0.5408 - digit0_loss: 0.0925 - digit1_loss: 0.1177 - digit2_loss: 0.1197 - digit3_loss: 0.1158 - digit4_loss: 0.0950 - digit0_accuracy: 0.9711 - digit1_accuracy: 0.9684 - digit2_accuracy: 0.9682 - digit3_accuracy: 0.9684 - digit4_accuracy: 0.9704 - val_loss: 0.5047 - val_digit0_loss: 0.0818 - val_digit1_loss: 0.1145 - val_digit2_loss: 0.1146 - val_digit3_loss: 0.1112 - val_digit4_loss: 0.0844 - val_digit0_accuracy: 0.9728 - val_digit1_accuracy: 0.9687 - val_digit2_accuracy: 0.9684 - val_digit3_accuracy: 0.9692 - val_digit4_accuracy: 0.9724\n",
            "Epoch 10/50\n",
            "8953/8953 [==============================] - 1s 152us/step - loss: 0.4955 - digit0_loss: 0.0820 - digit1_loss: 0.1100 - digit2_loss: 0.1128 - digit3_loss: 0.1074 - digit4_loss: 0.0833 - digit0_accuracy: 0.9731 - digit1_accuracy: 0.9690 - digit2_accuracy: 0.9687 - digit3_accuracy: 0.9693 - digit4_accuracy: 0.9727 - val_loss: 0.4704 - val_digit0_loss: 0.0748 - val_digit1_loss: 0.1089 - val_digit2_loss: 0.1090 - val_digit3_loss: 0.1053 - val_digit4_loss: 0.0745 - val_digit0_accuracy: 0.9740 - val_digit1_accuracy: 0.9693 - val_digit2_accuracy: 0.9690 - val_digit3_accuracy: 0.9693 - val_digit4_accuracy: 0.9751\n",
            "Epoch 11/50\n",
            "8953/8953 [==============================] - 1s 150us/step - loss: 0.4607 - digit0_loss: 0.0743 - digit1_loss: 0.1036 - digit2_loss: 0.1072 - digit3_loss: 0.1018 - digit4_loss: 0.0738 - digit0_accuracy: 0.9751 - digit1_accuracy: 0.9698 - digit2_accuracy: 0.9693 - digit3_accuracy: 0.9699 - digit4_accuracy: 0.9749 - val_loss: 0.4454 - val_digit0_loss: 0.0697 - val_digit1_loss: 0.1036 - val_digit2_loss: 0.1058 - val_digit3_loss: 0.0999 - val_digit4_loss: 0.0685 - val_digit0_accuracy: 0.9757 - val_digit1_accuracy: 0.9704 - val_digit2_accuracy: 0.9699 - val_digit3_accuracy: 0.9700 - val_digit4_accuracy: 0.9774\n",
            "Epoch 12/50\n",
            "8953/8953 [==============================] - 1s 152us/step - loss: 0.4331 - digit0_loss: 0.0689 - digit1_loss: 0.0984 - digit2_loss: 0.1023 - digit3_loss: 0.0962 - digit4_loss: 0.0673 - digit0_accuracy: 0.9765 - digit1_accuracy: 0.9706 - digit2_accuracy: 0.9700 - digit3_accuracy: 0.9707 - digit4_accuracy: 0.9768 - val_loss: 0.4241 - val_digit0_loss: 0.0647 - val_digit1_loss: 0.0991 - val_digit2_loss: 0.1023 - val_digit3_loss: 0.0967 - val_digit4_loss: 0.0624 - val_digit0_accuracy: 0.9778 - val_digit1_accuracy: 0.9708 - val_digit2_accuracy: 0.9696 - val_digit3_accuracy: 0.9705 - val_digit4_accuracy: 0.9784\n",
            "Epoch 13/50\n",
            "8953/8953 [==============================] - 1s 152us/step - loss: 0.4067 - digit0_loss: 0.0628 - digit1_loss: 0.0932 - digit2_loss: 0.0981 - digit3_loss: 0.0909 - digit4_loss: 0.0618 - digit0_accuracy: 0.9784 - digit1_accuracy: 0.9714 - digit2_accuracy: 0.9707 - digit3_accuracy: 0.9718 - digit4_accuracy: 0.9785 - val_loss: 0.4044 - val_digit0_loss: 0.0598 - val_digit1_loss: 0.0947 - val_digit2_loss: 0.0990 - val_digit3_loss: 0.0927 - val_digit4_loss: 0.0592 - val_digit0_accuracy: 0.9791 - val_digit1_accuracy: 0.9723 - val_digit2_accuracy: 0.9705 - val_digit3_accuracy: 0.9712 - val_digit4_accuracy: 0.9798\n",
            "Epoch 14/50\n",
            "8953/8953 [==============================] - 1s 151us/step - loss: 0.3838 - digit0_loss: 0.0586 - digit1_loss: 0.0888 - digit2_loss: 0.0946 - digit3_loss: 0.0861 - digit4_loss: 0.0558 - digit0_accuracy: 0.9798 - digit1_accuracy: 0.9725 - digit2_accuracy: 0.9711 - digit3_accuracy: 0.9727 - digit4_accuracy: 0.9805 - val_loss: 0.3923 - val_digit0_loss: 0.0586 - val_digit1_loss: 0.0920 - val_digit2_loss: 0.0966 - val_digit3_loss: 0.0906 - val_digit4_loss: 0.0562 - val_digit0_accuracy: 0.9801 - val_digit1_accuracy: 0.9720 - val_digit2_accuracy: 0.9705 - val_digit3_accuracy: 0.9717 - val_digit4_accuracy: 0.9805\n",
            "Epoch 15/50\n",
            "8953/8953 [==============================] - 1s 153us/step - loss: 0.3668 - digit0_loss: 0.0547 - digit1_loss: 0.0852 - digit2_loss: 0.0910 - digit3_loss: 0.0828 - digit4_loss: 0.0531 - digit0_accuracy: 0.9811 - digit1_accuracy: 0.9730 - digit2_accuracy: 0.9720 - digit3_accuracy: 0.9736 - digit4_accuracy: 0.9814 - val_loss: 0.3768 - val_digit0_loss: 0.0551 - val_digit1_loss: 0.0882 - val_digit2_loss: 0.0953 - val_digit3_loss: 0.0872 - val_digit4_loss: 0.0523 - val_digit0_accuracy: 0.9810 - val_digit1_accuracy: 0.9731 - val_digit2_accuracy: 0.9715 - val_digit3_accuracy: 0.9720 - val_digit4_accuracy: 0.9819\n",
            "Epoch 16/50\n",
            "8953/8953 [==============================] - 1s 150us/step - loss: 0.3502 - digit0_loss: 0.0519 - digit1_loss: 0.0817 - digit2_loss: 0.0877 - digit3_loss: 0.0792 - digit4_loss: 0.0497 - digit0_accuracy: 0.9816 - digit1_accuracy: 0.9738 - digit2_accuracy: 0.9727 - digit3_accuracy: 0.9745 - digit4_accuracy: 0.9827 - val_loss: 0.3682 - val_digit0_loss: 0.0534 - val_digit1_loss: 0.0862 - val_digit2_loss: 0.0941 - val_digit3_loss: 0.0851 - val_digit4_loss: 0.0507 - val_digit0_accuracy: 0.9816 - val_digit1_accuracy: 0.9735 - val_digit2_accuracy: 0.9717 - val_digit3_accuracy: 0.9725 - val_digit4_accuracy: 0.9823\n",
            "Epoch 17/50\n",
            "8953/8953 [==============================] - 1s 149us/step - loss: 0.3353 - digit0_loss: 0.0491 - digit1_loss: 0.0789 - digit2_loss: 0.0842 - digit3_loss: 0.0763 - digit4_loss: 0.0468 - digit0_accuracy: 0.9828 - digit1_accuracy: 0.9746 - digit2_accuracy: 0.9735 - digit3_accuracy: 0.9751 - digit4_accuracy: 0.9835 - val_loss: 0.3617 - val_digit0_loss: 0.0514 - val_digit1_loss: 0.0848 - val_digit2_loss: 0.0924 - val_digit3_loss: 0.0846 - val_digit4_loss: 0.0494 - val_digit0_accuracy: 0.9822 - val_digit1_accuracy: 0.9741 - val_digit2_accuracy: 0.9715 - val_digit3_accuracy: 0.9727 - val_digit4_accuracy: 0.9828\n",
            "Epoch 18/50\n",
            "8953/8953 [==============================] - 1s 152us/step - loss: 0.3212 - digit0_loss: 0.0460 - digit1_loss: 0.0753 - digit2_loss: 0.0814 - digit3_loss: 0.0729 - digit4_loss: 0.0456 - digit0_accuracy: 0.9837 - digit1_accuracy: 0.9755 - digit2_accuracy: 0.9743 - digit3_accuracy: 0.9759 - digit4_accuracy: 0.9839 - val_loss: 0.3566 - val_digit0_loss: 0.0507 - val_digit1_loss: 0.0823 - val_digit2_loss: 0.0913 - val_digit3_loss: 0.0831 - val_digit4_loss: 0.0502 - val_digit0_accuracy: 0.9831 - val_digit1_accuracy: 0.9751 - val_digit2_accuracy: 0.9721 - val_digit3_accuracy: 0.9731 - val_digit4_accuracy: 0.9830\n",
            "Epoch 19/50\n",
            "8953/8953 [==============================] - 1s 150us/step - loss: 0.3088 - digit0_loss: 0.0438 - digit1_loss: 0.0729 - digit2_loss: 0.0786 - digit3_loss: 0.0704 - digit4_loss: 0.0432 - digit0_accuracy: 0.9844 - digit1_accuracy: 0.9761 - digit2_accuracy: 0.9748 - digit3_accuracy: 0.9766 - digit4_accuracy: 0.9848 - val_loss: 0.3498 - val_digit0_loss: 0.0502 - val_digit1_loss: 0.0808 - val_digit2_loss: 0.0893 - val_digit3_loss: 0.0812 - val_digit4_loss: 0.0488 - val_digit0_accuracy: 0.9824 - val_digit1_accuracy: 0.9749 - val_digit2_accuracy: 0.9723 - val_digit3_accuracy: 0.9735 - val_digit4_accuracy: 0.9829\n",
            "Epoch 20/50\n",
            "8953/8953 [==============================] - 1s 152us/step - loss: 0.3005 - digit0_loss: 0.0427 - digit1_loss: 0.0704 - digit2_loss: 0.0772 - digit3_loss: 0.0691 - digit4_loss: 0.0412 - digit0_accuracy: 0.9847 - digit1_accuracy: 0.9769 - digit2_accuracy: 0.9751 - digit3_accuracy: 0.9772 - digit4_accuracy: 0.9852 - val_loss: 0.3419 - val_digit0_loss: 0.0481 - val_digit1_loss: 0.0789 - val_digit2_loss: 0.0883 - val_digit3_loss: 0.0804 - val_digit4_loss: 0.0469 - val_digit0_accuracy: 0.9836 - val_digit1_accuracy: 0.9756 - val_digit2_accuracy: 0.9726 - val_digit3_accuracy: 0.9736 - val_digit4_accuracy: 0.9839\n",
            "Epoch 21/50\n",
            "8953/8953 [==============================] - 1s 151us/step - loss: 0.2884 - digit0_loss: 0.0410 - digit1_loss: 0.0677 - digit2_loss: 0.0737 - digit3_loss: 0.0666 - digit4_loss: 0.0393 - digit0_accuracy: 0.9855 - digit1_accuracy: 0.9776 - digit2_accuracy: 0.9761 - digit3_accuracy: 0.9779 - digit4_accuracy: 0.9859 - val_loss: 0.3391 - val_digit0_loss: 0.0485 - val_digit1_loss: 0.0782 - val_digit2_loss: 0.0879 - val_digit3_loss: 0.0789 - val_digit4_loss: 0.0460 - val_digit0_accuracy: 0.9838 - val_digit1_accuracy: 0.9757 - val_digit2_accuracy: 0.9727 - val_digit3_accuracy: 0.9743 - val_digit4_accuracy: 0.9843\n",
            "Epoch 22/50\n",
            "8953/8953 [==============================] - 1s 151us/step - loss: 0.2780 - digit0_loss: 0.0392 - digit1_loss: 0.0656 - digit2_loss: 0.0721 - digit3_loss: 0.0637 - digit4_loss: 0.0374 - digit0_accuracy: 0.9860 - digit1_accuracy: 0.9777 - digit2_accuracy: 0.9765 - digit3_accuracy: 0.9786 - digit4_accuracy: 0.9866 - val_loss: 0.3351 - val_digit0_loss: 0.0454 - val_digit1_loss: 0.0783 - val_digit2_loss: 0.0873 - val_digit3_loss: 0.0782 - val_digit4_loss: 0.0460 - val_digit0_accuracy: 0.9842 - val_digit1_accuracy: 0.9762 - val_digit2_accuracy: 0.9729 - val_digit3_accuracy: 0.9745 - val_digit4_accuracy: 0.9847\n",
            "Epoch 23/50\n",
            "8953/8953 [==============================] - 1s 151us/step - loss: 0.2697 - digit0_loss: 0.0374 - digit1_loss: 0.0639 - digit2_loss: 0.0699 - digit3_loss: 0.0621 - digit4_loss: 0.0364 - digit0_accuracy: 0.9872 - digit1_accuracy: 0.9785 - digit2_accuracy: 0.9771 - digit3_accuracy: 0.9791 - digit4_accuracy: 0.9869 - val_loss: 0.3349 - val_digit0_loss: 0.0460 - val_digit1_loss: 0.0775 - val_digit2_loss: 0.0865 - val_digit3_loss: 0.0785 - val_digit4_loss: 0.0472 - val_digit0_accuracy: 0.9844 - val_digit1_accuracy: 0.9764 - val_digit2_accuracy: 0.9732 - val_digit3_accuracy: 0.9750 - val_digit4_accuracy: 0.9835\n",
            "Epoch 24/50\n",
            "8953/8953 [==============================] - 1s 152us/step - loss: 0.2628 - digit0_loss: 0.0365 - digit1_loss: 0.0622 - digit2_loss: 0.0680 - digit3_loss: 0.0606 - digit4_loss: 0.0354 - digit0_accuracy: 0.9870 - digit1_accuracy: 0.9789 - digit2_accuracy: 0.9775 - digit3_accuracy: 0.9793 - digit4_accuracy: 0.9871 - val_loss: 0.3268 - val_digit0_loss: 0.0454 - val_digit1_loss: 0.0757 - val_digit2_loss: 0.0839 - val_digit3_loss: 0.0776 - val_digit4_loss: 0.0440 - val_digit0_accuracy: 0.9843 - val_digit1_accuracy: 0.9766 - val_digit2_accuracy: 0.9735 - val_digit3_accuracy: 0.9752 - val_digit4_accuracy: 0.9848\n",
            "Epoch 25/50\n",
            "8953/8953 [==============================] - 1s 151us/step - loss: 0.2554 - digit0_loss: 0.0353 - digit1_loss: 0.0609 - digit2_loss: 0.0669 - digit3_loss: 0.0578 - digit4_loss: 0.0345 - digit0_accuracy: 0.9871 - digit1_accuracy: 0.9794 - digit2_accuracy: 0.9778 - digit3_accuracy: 0.9802 - digit4_accuracy: 0.9876 - val_loss: 0.3272 - val_digit0_loss: 0.0454 - val_digit1_loss: 0.0770 - val_digit2_loss: 0.0842 - val_digit3_loss: 0.0767 - val_digit4_loss: 0.0441 - val_digit0_accuracy: 0.9844 - val_digit1_accuracy: 0.9756 - val_digit2_accuracy: 0.9736 - val_digit3_accuracy: 0.9753 - val_digit4_accuracy: 0.9838\n",
            "Epoch 26/50\n",
            "8953/8953 [==============================] - 1s 150us/step - loss: 0.2450 - digit0_loss: 0.0333 - digit1_loss: 0.0580 - digit2_loss: 0.0650 - digit3_loss: 0.0565 - digit4_loss: 0.0322 - digit0_accuracy: 0.9882 - digit1_accuracy: 0.9800 - digit2_accuracy: 0.9782 - digit3_accuracy: 0.9806 - digit4_accuracy: 0.9886 - val_loss: 0.3219 - val_digit0_loss: 0.0452 - val_digit1_loss: 0.0755 - val_digit2_loss: 0.0836 - val_digit3_loss: 0.0756 - val_digit4_loss: 0.0432 - val_digit0_accuracy: 0.9842 - val_digit1_accuracy: 0.9763 - val_digit2_accuracy: 0.9736 - val_digit3_accuracy: 0.9750 - val_digit4_accuracy: 0.9850\n",
            "Epoch 27/50\n",
            "8953/8953 [==============================] - 1s 152us/step - loss: 0.2371 - digit0_loss: 0.0329 - digit1_loss: 0.0559 - digit2_loss: 0.0624 - digit3_loss: 0.0549 - digit4_loss: 0.0311 - digit0_accuracy: 0.9883 - digit1_accuracy: 0.9808 - digit2_accuracy: 0.9793 - digit3_accuracy: 0.9809 - digit4_accuracy: 0.9887 - val_loss: 0.3215 - val_digit0_loss: 0.0442 - val_digit1_loss: 0.0738 - val_digit2_loss: 0.0828 - val_digit3_loss: 0.0770 - val_digit4_loss: 0.0429 - val_digit0_accuracy: 0.9847 - val_digit1_accuracy: 0.9764 - val_digit2_accuracy: 0.9738 - val_digit3_accuracy: 0.9750 - val_digit4_accuracy: 0.9844\n",
            "Epoch 28/50\n",
            "8953/8953 [==============================] - 1s 151us/step - loss: 0.2344 - digit0_loss: 0.0322 - digit1_loss: 0.0553 - digit2_loss: 0.0619 - digit3_loss: 0.0545 - digit4_loss: 0.0305 - digit0_accuracy: 0.9883 - digit1_accuracy: 0.9808 - digit2_accuracy: 0.9791 - digit3_accuracy: 0.9811 - digit4_accuracy: 0.9891 - val_loss: 0.3256 - val_digit0_loss: 0.0448 - val_digit1_loss: 0.0757 - val_digit2_loss: 0.0846 - val_digit3_loss: 0.0759 - val_digit4_loss: 0.0437 - val_digit0_accuracy: 0.9845 - val_digit1_accuracy: 0.9760 - val_digit2_accuracy: 0.9734 - val_digit3_accuracy: 0.9757 - val_digit4_accuracy: 0.9839\n",
            "Epoch 29/50\n",
            "8953/8953 [==============================] - 1s 150us/step - loss: 0.2281 - digit0_loss: 0.0312 - digit1_loss: 0.0549 - digit2_loss: 0.0602 - digit3_loss: 0.0518 - digit4_loss: 0.0299 - digit0_accuracy: 0.9887 - digit1_accuracy: 0.9808 - digit2_accuracy: 0.9796 - digit3_accuracy: 0.9821 - digit4_accuracy: 0.9892 - val_loss: 0.3217 - val_digit0_loss: 0.0451 - val_digit1_loss: 0.0737 - val_digit2_loss: 0.0837 - val_digit3_loss: 0.0752 - val_digit4_loss: 0.0435 - val_digit0_accuracy: 0.9842 - val_digit1_accuracy: 0.9768 - val_digit2_accuracy: 0.9735 - val_digit3_accuracy: 0.9748 - val_digit4_accuracy: 0.9847\n",
            "Epoch 30/50\n",
            "8953/8953 [==============================] - 1s 151us/step - loss: 0.2211 - digit0_loss: 0.0310 - digit1_loss: 0.0521 - digit2_loss: 0.0579 - digit3_loss: 0.0509 - digit4_loss: 0.0292 - digit0_accuracy: 0.9886 - digit1_accuracy: 0.9821 - digit2_accuracy: 0.9802 - digit3_accuracy: 0.9823 - digit4_accuracy: 0.9893 - val_loss: 0.3220 - val_digit0_loss: 0.0466 - val_digit1_loss: 0.0752 - val_digit2_loss: 0.0830 - val_digit3_loss: 0.0751 - val_digit4_loss: 0.0433 - val_digit0_accuracy: 0.9843 - val_digit1_accuracy: 0.9764 - val_digit2_accuracy: 0.9729 - val_digit3_accuracy: 0.9751 - val_digit4_accuracy: 0.9852\n",
            "Epoch 31/50\n",
            "8953/8953 [==============================] - 1s 149us/step - loss: 0.2145 - digit0_loss: 0.0291 - digit1_loss: 0.0507 - digit2_loss: 0.0568 - digit3_loss: 0.0497 - digit4_loss: 0.0283 - digit0_accuracy: 0.9895 - digit1_accuracy: 0.9823 - digit2_accuracy: 0.9805 - digit3_accuracy: 0.9829 - digit4_accuracy: 0.9895 - val_loss: 0.3160 - val_digit0_loss: 0.0436 - val_digit1_loss: 0.0726 - val_digit2_loss: 0.0817 - val_digit3_loss: 0.0757 - val_digit4_loss: 0.0423 - val_digit0_accuracy: 0.9848 - val_digit1_accuracy: 0.9768 - val_digit2_accuracy: 0.9739 - val_digit3_accuracy: 0.9755 - val_digit4_accuracy: 0.9849\n",
            "Epoch 32/50\n",
            "8953/8953 [==============================] - 1s 151us/step - loss: 0.2077 - digit0_loss: 0.0283 - digit1_loss: 0.0492 - digit2_loss: 0.0545 - digit3_loss: 0.0481 - digit4_loss: 0.0277 - digit0_accuracy: 0.9896 - digit1_accuracy: 0.9830 - digit2_accuracy: 0.9814 - digit3_accuracy: 0.9829 - digit4_accuracy: 0.9898 - val_loss: 0.3195 - val_digit0_loss: 0.0459 - val_digit1_loss: 0.0735 - val_digit2_loss: 0.0822 - val_digit3_loss: 0.0752 - val_digit4_loss: 0.0431 - val_digit0_accuracy: 0.9840 - val_digit1_accuracy: 0.9769 - val_digit2_accuracy: 0.9736 - val_digit3_accuracy: 0.9760 - val_digit4_accuracy: 0.9851\n",
            "Epoch 33/50\n",
            "8953/8953 [==============================] - 1s 151us/step - loss: 0.2046 - digit0_loss: 0.0275 - digit1_loss: 0.0490 - digit2_loss: 0.0540 - digit3_loss: 0.0479 - digit4_loss: 0.0262 - digit0_accuracy: 0.9899 - digit1_accuracy: 0.9828 - digit2_accuracy: 0.9812 - digit3_accuracy: 0.9830 - digit4_accuracy: 0.9906 - val_loss: 0.3159 - val_digit0_loss: 0.0434 - val_digit1_loss: 0.0734 - val_digit2_loss: 0.0818 - val_digit3_loss: 0.0739 - val_digit4_loss: 0.0427 - val_digit0_accuracy: 0.9849 - val_digit1_accuracy: 0.9766 - val_digit2_accuracy: 0.9738 - val_digit3_accuracy: 0.9753 - val_digit4_accuracy: 0.9848\n",
            "Epoch 34/50\n",
            "8953/8953 [==============================] - 1s 152us/step - loss: 0.1976 - digit0_loss: 0.0269 - digit1_loss: 0.0465 - digit2_loss: 0.0518 - digit3_loss: 0.0459 - digit4_loss: 0.0266 - digit0_accuracy: 0.9902 - digit1_accuracy: 0.9837 - digit2_accuracy: 0.9823 - digit3_accuracy: 0.9838 - digit4_accuracy: 0.9903 - val_loss: 0.3167 - val_digit0_loss: 0.0448 - val_digit1_loss: 0.0725 - val_digit2_loss: 0.0814 - val_digit3_loss: 0.0746 - val_digit4_loss: 0.0425 - val_digit0_accuracy: 0.9849 - val_digit1_accuracy: 0.9769 - val_digit2_accuracy: 0.9736 - val_digit3_accuracy: 0.9758 - val_digit4_accuracy: 0.9850\n",
            "Epoch 35/50\n",
            "8953/8953 [==============================] - 1s 151us/step - loss: 0.1961 - digit0_loss: 0.0269 - digit1_loss: 0.0470 - digit2_loss: 0.0513 - digit3_loss: 0.0446 - digit4_loss: 0.0262 - digit0_accuracy: 0.9901 - digit1_accuracy: 0.9837 - digit2_accuracy: 0.9821 - digit3_accuracy: 0.9842 - digit4_accuracy: 0.9902 - val_loss: 0.3137 - val_digit0_loss: 0.0442 - val_digit1_loss: 0.0716 - val_digit2_loss: 0.0821 - val_digit3_loss: 0.0740 - val_digit4_loss: 0.0418 - val_digit0_accuracy: 0.9853 - val_digit1_accuracy: 0.9773 - val_digit2_accuracy: 0.9730 - val_digit3_accuracy: 0.9755 - val_digit4_accuracy: 0.9856\n",
            "Epoch 36/50\n",
            "8953/8953 [==============================] - 1s 153us/step - loss: 0.1916 - digit0_loss: 0.0264 - digit1_loss: 0.0457 - digit2_loss: 0.0502 - digit3_loss: 0.0442 - digit4_loss: 0.0252 - digit0_accuracy: 0.9904 - digit1_accuracy: 0.9837 - digit2_accuracy: 0.9824 - digit3_accuracy: 0.9844 - digit4_accuracy: 0.9908 - val_loss: 0.3130 - val_digit0_loss: 0.0438 - val_digit1_loss: 0.0731 - val_digit2_loss: 0.0820 - val_digit3_loss: 0.0741 - val_digit4_loss: 0.0416 - val_digit0_accuracy: 0.9849 - val_digit1_accuracy: 0.9770 - val_digit2_accuracy: 0.9734 - val_digit3_accuracy: 0.9758 - val_digit4_accuracy: 0.9854\n",
            "Epoch 37/50\n",
            "8953/8953 [==============================] - 1s 152us/step - loss: 0.1851 - digit0_loss: 0.0251 - digit1_loss: 0.0449 - digit2_loss: 0.0485 - digit3_loss: 0.0426 - digit4_loss: 0.0240 - digit0_accuracy: 0.9908 - digit1_accuracy: 0.9841 - digit2_accuracy: 0.9831 - digit3_accuracy: 0.9848 - digit4_accuracy: 0.9914 - val_loss: 0.3149 - val_digit0_loss: 0.0455 - val_digit1_loss: 0.0720 - val_digit2_loss: 0.0825 - val_digit3_loss: 0.0733 - val_digit4_loss: 0.0424 - val_digit0_accuracy: 0.9850 - val_digit1_accuracy: 0.9771 - val_digit2_accuracy: 0.9736 - val_digit3_accuracy: 0.9755 - val_digit4_accuracy: 0.9854\n",
            "Epoch 38/50\n",
            "8953/8953 [==============================] - 1s 150us/step - loss: 0.1817 - digit0_loss: 0.0249 - digit1_loss: 0.0429 - digit2_loss: 0.0475 - digit3_loss: 0.0422 - digit4_loss: 0.0242 - digit0_accuracy: 0.9907 - digit1_accuracy: 0.9846 - digit2_accuracy: 0.9835 - digit3_accuracy: 0.9852 - digit4_accuracy: 0.9911 - val_loss: 0.3156 - val_digit0_loss: 0.0449 - val_digit1_loss: 0.0732 - val_digit2_loss: 0.0830 - val_digit3_loss: 0.0732 - val_digit4_loss: 0.0423 - val_digit0_accuracy: 0.9849 - val_digit1_accuracy: 0.9763 - val_digit2_accuracy: 0.9739 - val_digit3_accuracy: 0.9758 - val_digit4_accuracy: 0.9853\n",
            "Epoch 39/50\n",
            "8953/8953 [==============================] - 1s 151us/step - loss: 0.1781 - digit0_loss: 0.0237 - digit1_loss: 0.0425 - digit2_loss: 0.0470 - digit3_loss: 0.0411 - digit4_loss: 0.0236 - digit0_accuracy: 0.9913 - digit1_accuracy: 0.9849 - digit2_accuracy: 0.9836 - digit3_accuracy: 0.9853 - digit4_accuracy: 0.9915 - val_loss: 0.3146 - val_digit0_loss: 0.0441 - val_digit1_loss: 0.0730 - val_digit2_loss: 0.0825 - val_digit3_loss: 0.0744 - val_digit4_loss: 0.0414 - val_digit0_accuracy: 0.9857 - val_digit1_accuracy: 0.9765 - val_digit2_accuracy: 0.9734 - val_digit3_accuracy: 0.9757 - val_digit4_accuracy: 0.9856\n",
            "Epoch 40/50\n",
            "8953/8953 [==============================] - 1s 150us/step - loss: 0.1734 - digit0_loss: 0.0241 - digit1_loss: 0.0411 - digit2_loss: 0.0464 - digit3_loss: 0.0392 - digit4_loss: 0.0226 - digit0_accuracy: 0.9911 - digit1_accuracy: 0.9854 - digit2_accuracy: 0.9838 - digit3_accuracy: 0.9861 - digit4_accuracy: 0.9917 - val_loss: 0.3118 - val_digit0_loss: 0.0426 - val_digit1_loss: 0.0727 - val_digit2_loss: 0.0817 - val_digit3_loss: 0.0733 - val_digit4_loss: 0.0420 - val_digit0_accuracy: 0.9857 - val_digit1_accuracy: 0.9773 - val_digit2_accuracy: 0.9739 - val_digit3_accuracy: 0.9758 - val_digit4_accuracy: 0.9856\n",
            "Epoch 41/50\n",
            "8953/8953 [==============================] - 1s 151us/step - loss: 0.1701 - digit0_loss: 0.0232 - digit1_loss: 0.0404 - digit2_loss: 0.0443 - digit3_loss: 0.0394 - digit4_loss: 0.0227 - digit0_accuracy: 0.9916 - digit1_accuracy: 0.9856 - digit2_accuracy: 0.9843 - digit3_accuracy: 0.9857 - digit4_accuracy: 0.9915 - val_loss: 0.3150 - val_digit0_loss: 0.0455 - val_digit1_loss: 0.0727 - val_digit2_loss: 0.0819 - val_digit3_loss: 0.0743 - val_digit4_loss: 0.0418 - val_digit0_accuracy: 0.9851 - val_digit1_accuracy: 0.9774 - val_digit2_accuracy: 0.9741 - val_digit3_accuracy: 0.9756 - val_digit4_accuracy: 0.9855\n",
            "Epoch 42/50\n",
            "8953/8953 [==============================] - 1s 152us/step - loss: 0.1655 - digit0_loss: 0.0223 - digit1_loss: 0.0400 - digit2_loss: 0.0432 - digit3_loss: 0.0386 - digit4_loss: 0.0213 - digit0_accuracy: 0.9918 - digit1_accuracy: 0.9858 - digit2_accuracy: 0.9848 - digit3_accuracy: 0.9861 - digit4_accuracy: 0.9922 - val_loss: 0.3151 - val_digit0_loss: 0.0470 - val_digit1_loss: 0.0724 - val_digit2_loss: 0.0821 - val_digit3_loss: 0.0733 - val_digit4_loss: 0.0423 - val_digit0_accuracy: 0.9845 - val_digit1_accuracy: 0.9773 - val_digit2_accuracy: 0.9737 - val_digit3_accuracy: 0.9761 - val_digit4_accuracy: 0.9859\n",
            "Epoch 43/50\n",
            "8953/8953 [==============================] - 1s 154us/step - loss: 0.1641 - digit0_loss: 0.0226 - digit1_loss: 0.0381 - digit2_loss: 0.0432 - digit3_loss: 0.0378 - digit4_loss: 0.0223 - digit0_accuracy: 0.9917 - digit1_accuracy: 0.9863 - digit2_accuracy: 0.9846 - digit3_accuracy: 0.9864 - digit4_accuracy: 0.9919 - val_loss: 0.3135 - val_digit0_loss: 0.0436 - val_digit1_loss: 0.0725 - val_digit2_loss: 0.0828 - val_digit3_loss: 0.0736 - val_digit4_loss: 0.0425 - val_digit0_accuracy: 0.9854 - val_digit1_accuracy: 0.9773 - val_digit2_accuracy: 0.9736 - val_digit3_accuracy: 0.9758 - val_digit4_accuracy: 0.9854\n",
            "Epoch 44/50\n",
            "8953/8953 [==============================] - 1s 155us/step - loss: 0.1586 - digit0_loss: 0.0218 - digit1_loss: 0.0377 - digit2_loss: 0.0411 - digit3_loss: 0.0369 - digit4_loss: 0.0211 - digit0_accuracy: 0.9919 - digit1_accuracy: 0.9864 - digit2_accuracy: 0.9854 - digit3_accuracy: 0.9868 - digit4_accuracy: 0.9923 - val_loss: 0.3177 - val_digit0_loss: 0.0453 - val_digit1_loss: 0.0743 - val_digit2_loss: 0.0821 - val_digit3_loss: 0.0746 - val_digit4_loss: 0.0426 - val_digit0_accuracy: 0.9854 - val_digit1_accuracy: 0.9767 - val_digit2_accuracy: 0.9741 - val_digit3_accuracy: 0.9759 - val_digit4_accuracy: 0.9855\n",
            "Epoch 45/50\n",
            "8953/8953 [==============================] - 1s 153us/step - loss: 0.1582 - digit0_loss: 0.0223 - digit1_loss: 0.0373 - digit2_loss: 0.0417 - digit3_loss: 0.0359 - digit4_loss: 0.0209 - digit0_accuracy: 0.9917 - digit1_accuracy: 0.9867 - digit2_accuracy: 0.9851 - digit3_accuracy: 0.9871 - digit4_accuracy: 0.9923 - val_loss: 0.3151 - val_digit0_loss: 0.0461 - val_digit1_loss: 0.0720 - val_digit2_loss: 0.0823 - val_digit3_loss: 0.0752 - val_digit4_loss: 0.0405 - val_digit0_accuracy: 0.9845 - val_digit1_accuracy: 0.9769 - val_digit2_accuracy: 0.9737 - val_digit3_accuracy: 0.9759 - val_digit4_accuracy: 0.9857\n",
            "Epoch 46/50\n",
            "8953/8953 [==============================] - 1s 150us/step - loss: 0.1559 - digit0_loss: 0.0214 - digit1_loss: 0.0375 - digit2_loss: 0.0402 - digit3_loss: 0.0360 - digit4_loss: 0.0206 - digit0_accuracy: 0.9919 - digit1_accuracy: 0.9865 - digit2_accuracy: 0.9855 - digit3_accuracy: 0.9871 - digit4_accuracy: 0.9925 - val_loss: 0.3186 - val_digit0_loss: 0.0460 - val_digit1_loss: 0.0733 - val_digit2_loss: 0.0821 - val_digit3_loss: 0.0748 - val_digit4_loss: 0.0423 - val_digit0_accuracy: 0.9842 - val_digit1_accuracy: 0.9778 - val_digit2_accuracy: 0.9744 - val_digit3_accuracy: 0.9756 - val_digit4_accuracy: 0.9853\n",
            "Epoch 47/50\n",
            "8953/8953 [==============================] - 1s 152us/step - loss: 0.1531 - digit0_loss: 0.0212 - digit1_loss: 0.0356 - digit2_loss: 0.0403 - digit3_loss: 0.0353 - digit4_loss: 0.0207 - digit0_accuracy: 0.9924 - digit1_accuracy: 0.9870 - digit2_accuracy: 0.9856 - digit3_accuracy: 0.9871 - digit4_accuracy: 0.9923 - val_loss: 0.3161 - val_digit0_loss: 0.0461 - val_digit1_loss: 0.0733 - val_digit2_loss: 0.0823 - val_digit3_loss: 0.0745 - val_digit4_loss: 0.0419 - val_digit0_accuracy: 0.9845 - val_digit1_accuracy: 0.9775 - val_digit2_accuracy: 0.9736 - val_digit3_accuracy: 0.9754 - val_digit4_accuracy: 0.9855\n",
            "Epoch 48/50\n",
            "8953/8953 [==============================] - 1s 149us/step - loss: 0.1508 - digit0_loss: 0.0209 - digit1_loss: 0.0357 - digit2_loss: 0.0393 - digit3_loss: 0.0347 - digit4_loss: 0.0201 - digit0_accuracy: 0.9924 - digit1_accuracy: 0.9873 - digit2_accuracy: 0.9859 - digit3_accuracy: 0.9876 - digit4_accuracy: 0.9926 - val_loss: 0.3141 - val_digit0_loss: 0.0451 - val_digit1_loss: 0.0727 - val_digit2_loss: 0.0814 - val_digit3_loss: 0.0744 - val_digit4_loss: 0.0424 - val_digit0_accuracy: 0.9849 - val_digit1_accuracy: 0.9772 - val_digit2_accuracy: 0.9743 - val_digit3_accuracy: 0.9759 - val_digit4_accuracy: 0.9852\n",
            "Epoch 49/50\n",
            "8953/8953 [==============================] - 1s 151us/step - loss: 0.1494 - digit0_loss: 0.0205 - digit1_loss: 0.0354 - digit2_loss: 0.0386 - digit3_loss: 0.0341 - digit4_loss: 0.0208 - digit0_accuracy: 0.9924 - digit1_accuracy: 0.9872 - digit2_accuracy: 0.9862 - digit3_accuracy: 0.9875 - digit4_accuracy: 0.9923 - val_loss: 0.3186 - val_digit0_loss: 0.0454 - val_digit1_loss: 0.0759 - val_digit2_loss: 0.0821 - val_digit3_loss: 0.0759 - val_digit4_loss: 0.0408 - val_digit0_accuracy: 0.9843 - val_digit1_accuracy: 0.9768 - val_digit2_accuracy: 0.9736 - val_digit3_accuracy: 0.9751 - val_digit4_accuracy: 0.9853\n",
            "Epoch 50/50\n",
            "8953/8953 [==============================] - 1s 150us/step - loss: 0.1465 - digit0_loss: 0.0202 - digit1_loss: 0.0348 - digit2_loss: 0.0379 - digit3_loss: 0.0336 - digit4_loss: 0.0200 - digit0_accuracy: 0.9927 - digit1_accuracy: 0.9874 - digit2_accuracy: 0.9865 - digit3_accuracy: 0.9879 - digit4_accuracy: 0.9927 - val_loss: 0.3141 - val_digit0_loss: 0.0450 - val_digit1_loss: 0.0741 - val_digit2_loss: 0.0806 - val_digit3_loss: 0.0751 - val_digit4_loss: 0.0402 - val_digit0_accuracy: 0.9844 - val_digit1_accuracy: 0.9770 - val_digit2_accuracy: 0.9742 - val_digit3_accuracy: 0.9754 - val_digit4_accuracy: 0.9854\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PK5FjnNp63eL",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "digit_acc = [[] for _ in range(NUM_OF_LETTERS)]\n",
        "val_digit_acc = [[] for _ in range(NUM_OF_LETTERS)]\n",
        "loss = []\n",
        "val_loss = []\n",
        "\n",
        "\n",
        "def plot_diagram(digit_acc_now, val_digit_acc_now, loss_now, val_loss_now):\n",
        "    global digit_acc, val_digit_acc, loss, val_loss\n",
        "    \n",
        "    \n",
        "    for i in range(NUM_OF_LETTERS):\n",
        "        digit_acc[i].extend(digit_acc_now[i])\n",
        "        val_digit_acc[i].extend(val_digit_acc_now[i])\n",
        "    loss.extend(loss_now)\n",
        "    val_loss.extend(val_loss_now)\n",
        "    \n",
        "    for i in range(NUM_OF_LETTERS):\n",
        "        s = {0:'First', 1:'Second', 2:'Third', 3:'Fourth', 4:'Fifth'}[i]\n",
        "        # plt.plot(val_digit_acc[i], label='%s Digit Train' % s)\n",
        "        plt.plot(digit_acc[i], label='%s Digit Test' % s)\n",
        "\n",
        "    plt.title('Model accuracy')\n",
        "    plt.ylabel('Accuracy')\n",
        "    plt.xlabel('Epoch')\n",
        "    plt.legend()\n",
        "    plt.show()\n",
        "\n",
        "    for i in range(NUM_OF_LETTERS):\n",
        "        s = {0:'First', 1:'Second', 2:'Third', 3:'Fourth', 4:'Fifth'}[i]\n",
        "        plt.plot(val_digit_acc[i], label='%s Digit Train' % s)\n",
        "        # plt.plot(digit_acc[i], label='%s Digit Test' % s)\n",
        "\n",
        "    plt.title('Model accuracy')\n",
        "    plt.ylabel('Accuracy')\n",
        "    plt.xlabel('Epoch')\n",
        "    plt.legend()\n",
        "    plt.show()\n",
        "\n",
        "\n",
        "\n",
        "    # Plot training & validation loss values\n",
        "    \n",
        "    plt.plot(val_loss, label='Train')\n",
        "    plt.plot(loss, label='Test')\n",
        "    plt.title('Model loss')\n",
        "    plt.ylabel('Loss')\n",
        "    plt.xlabel('Epoch')\n",
        "    plt.legend()\n",
        "    plt.show()"
      ],
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gmz3c1Ve8xG1",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 851
        },
        "outputId": "746b74b1-02a1-43eb-d7f5-a094234f55e3"
      },
      "source": [
        "plot_diagram(\n",
        "    \n",
        "    [history.history['digit%d_accuracy' % i] for i in range(NUM_OF_LETTERS)],\n",
        "    [history.history['val_digit%d_accuracy' % i] for i in range(NUM_OF_LETTERS)],\n",
        "    history.history['loss'],\n",
        "    history.history['val_loss'],\n",
        ")"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3dd3xV9f348df73uwNSRhJgIS9CYiKirhbrILWCYqjah2tq9qhbX+ttbW1/ba1Ym0rrjqqiBv3RFEB2Rtkj0BCBmRB9n3//jgHiBgwQG5Ocu/7+fA87j2fc+6976PXvO/nfJaoKsYYY8KXz+sAjDHGeMsSgTHGhDlLBMYYE+YsERhjTJizRGCMMWHOEoExxoQ5SwTGNIOIZIuIikhEM869WkQ+P9r3Maa1WCIwIUdENolIrYikHVC+yP0jnO1NZMa0TZYITKjaCEzcuyMiQ4A478Ixpu2yRGBC1TPAlY32rwKebnyCiCSLyNMiUiQim0Xk1yLic4/5ReSvIlIsIhuAc5p47eMiki8i20TkDyLiP9wgRSRDRKaLyE4RWSciP2x07DgRmS8i5SKyQ0T+7pbHiMizIlIiIqUiMk9EOh/uZxuzlyUCE6rmAEkiMsD9Az0BePaAcx4CkoGewCk4ieMH7rEfAucCw4GRwEUHvPa/QD3Q2z3nO8B1RxDnVCAPyHA/448icrp77EHgQVVNAnoB09zyq9y4uwGpwI1A1RF8tjGAJQIT2vbWCs4CVgHb9h5olBzuVtUKVd0E/A24wj3lEuAfqrpVVXcCf2r02s7A94DbVXW3qhYCD7jv12wi0g04CfiFqlar6mLgMfbXZOqA3iKSpqqVqjqnUXkq0FtVG1R1gaqWH85nG9OYJQITyp4BLgOu5oDbQkAaEAlsblS2Gch0n2cAWw84tlcP97X57q2ZUuARoNNhxpcB7FTVioPEcC3QF1jt3v45t9F1vQdMFZHtIvIXEYk8zM82Zh9LBCZkqepmnEbj7wGvHHC4GOeXdY9GZd3ZX2vIx7n10vjYXluBGiBNVVPcLUlVBx1miNuBjiKS2FQMqrpWVSfiJJg/Ay+JSLyq1qnq71R1IHAizi2sKzHmCFkiMKHuWuB0Vd3duFBVG3Duud8nIoki0gO4g/3tCNOAW0UkS0Q6AHc1em0+8D7wNxFJEhGfiPQSkVMOJzBV3QrMAv7kNgAPdeN9FkBEJolIuqoGgFL3ZQEROU1Ehri3t8pxElrgcD7bmMYsEZiQpqrrVXX+QQ7fAuwGNgCfA88BT7jHHsW5/bIEWMg3axRXAlHASmAX8BLQ9QhCnAhk49QOXgV+q6ofusfGAitEpBKn4XiCqlYBXdzPK8dp+/gU53aRMUdEbGEaY4wJb1YjMMaYMGeJwBhjwpwlAmOMCXOWCIwxJsy1u6lw09LSNDs72+swjDGmXVmwYEGxqqY3dazdJYLs7Gzmzz9Yb0BjjDFNEZHNBztmt4aMMSbMWSIwxpgwZ4nAGGPCXLtrIzDGmMNVV1dHXl4e1dXVXocSdDExMWRlZREZ2fwJaS0RGGNCXl5eHomJiWRnZyMiXocTNKpKSUkJeXl55OTkNPt1dmvIGBPyqqurSU1NDekkACAipKamHnbNxxKBMSYshHoS2OtIrjNsbg3tWfc5Des/QePTkbhUJL4TJKThT+iMLy4Zv8+H3ydh82Uxxpi9gpoIRGQszjzqfuAxVb3/gOMPAKe5u3FAJ1VNCUYsS2e/z6j1DzZ5rEGFeiKowUcDPhrwE3Af63zRBKISkOhEouKSiE1IIT4xBV+PUTDkYvDbCoHGmEMrKSnhjDPOAKCgoAC/3096ujPId+7cuURFRR30tfPnz+fpp59m8uTJQYsvaOsRuKsnrcFZODwPmAdMVNWVBzn/FmC4ql5zqPcdOXKkHsnI4lX55SzZXERE9S6ianYSVV1CVM1OomtLiKotRQL1EKhHtAECDYjWQ6CBuqpKGqrL8dXtJp4qEqgiRfbQQSrQlO7IyXfCsMsg4uD/IY0x3lq1ahUDBgzwOgwA7rnnHhISEvjpT3+6r6y+vp6IiJb7Xd7U9YrIAlUd2dT5wawRHAesU9UNbhBTgfNwVnRqykTgt8EKZkDXJAZ0TTri11fVNrCusJKFOyr4fG0RpUvf4o/1b9P1jdvg0/+D0bfD8CsgMqYFozbGhKqrr76amJgYFi1axEknncSECRO47bbbqK6uJjY2lieffJJ+/frxySef8Ne//pU333yTe+65hy1btrBhwwa2bNnC7bffzq233nrUsQQzEWTiLPK9Vx5wfFMnuuvF5gAfH+T49cD1AN27d2/qlKCLjfIzJCuZIVnJXDAik7sj/ZwwL5fHR1dwxo4n4e2fwmd/gzN+A7mXeRKjMebb/e6NFazcXt6i7zkwI4nfjht02K/Ly8tj1qxZ+P1+ysvL+eyzz4iIiODDDz/kl7/8JS+//PI3XrN69WpmzJhBRUUF/fr146abbjqsMQNNaSuNxROAl9wFxb9BVacAU8C5NdSagTVFRPjD+YPZUV7ND78QHpn0NGed9hV8/Ad47SZI6AS9z/Q6TGNMG3fxxRfj9/sBKCsr46qrrmLt2rWICHV1dU2+5pxzziE6Opro6Gg6derEjh07yMrKOqo4gpkItgHdGu1nuWVNmQD8OIixtLgIv4+HLx/BxClzuGXqIp774ShGXDUdHj0dXr0RbvwCEjt7HaYx5gBH8ss9WOLj4/c9/3//7/9x2mmn8eqrr7Jp0yZOPfXUJl8THR2977nf76e+vv6o4wjmOIJ5QB8RyRGRKJw/9tMPPElE+gMdgNlBjCUo4qIiePzqY+mcFMO1/53HhtIGuOgJqKmEV6+HQMDrEI0x7URZWRmZmZkA/Pe//23Vzw5aIlDVeuBm4D1gFTBNVVeIyL0iMr7RqROAqRqs7ktBlpYQzVM/OA6fCFc9OZfC2Bw4+37Y8AnMarq7qjHGHOjnP/85d999N8OHD2+RX/mHI2jdR4PlSLuPBtviraVMnDKH/l0TeeXGE5CXfgCr3oBr3oNux3odnjFhrS11H20Nh9t91KaYaCG53VL43fhBLNpSyvurCmHcg5CcCS9fA1WlXodnjDEHZYmgBV0wIpPs1Dge/HAtGpMMFz4BZdvgjdugndW8jDHhwxJBC4rw+7jl9D6szC/n/ZU7nFtCp/8aVr4GC5/yOjxjjGmSJYIWdl5uBjlp8fzjw7UEAgon3Q49T4V3fuHUDowxpo2xRNDCnFpBb1btrRX4fDBuMjTUwex/eh2eMcZ8gyWCIBg/LIOeafH848M1Tq2gQw8YegnMfxJ2F3sdnjHGfI0lgiCI8Pu45YzerC6o4P2VBU7h6J9AfTXM+be3wRljWl1JSQm5ubnk5ubSpUsXMjMz9+3X1tZ+6+s/+eQTZs2aFbT4LBEEyfhhmW6twG0rSO8HA8bB3Eehuszr8IwxrSg1NZXFixezePFibrzxRn7yk5/s2z/UWgR7WSJop/w+4dYz+rC6oIL3Vri1gpPvhJoymPeYt8EZYzy3YMECTjnlFI455hi++93vkp+fD8DkyZMZOHAgQ4cOZcKECWzatIn//Oc/PPDAA+Tm5vLZZ5+1eCxtZfbRkDRuWAaTP17Lgx+t5buDuuDLyHVmJZ39Lzj+JoiK8zpEY8LPO3dBwbKWfc8uQ5ypZZpJVbnlllt4/fXXSU9P54UXXuBXv/oVTzzxBPfffz8bN24kOjqa0tJSUlJSuPHGG7+xmE1LshpBEPl9wm1ureDdfbWCn8KeYlj4tLfBGWM8U1NTw/LlyznrrLPIzc3lD3/4A3l5eQAMHTqUyy+/nGeffbZFVy07FKsRBNm5QzOY/NFaHvxwLWMHdcHX4wTofiLMmgwjr7ElLo1pbYfxyz1YVJVBgwYxe/Y3J11+6623mDlzJm+88Qb33Xcfy5a1cO2lCVYjCDK/T7jtzL58taOC6Uu2O4Vj7oTybbB0qrfBGWM8ER0dTVFR0b5EUFdXx4oVKwgEAmzdupXTTjuNP//5z5SVlVFZWUliYiIVFRVBi8cSQSs4d0hXBmcm8df3v6KmvgF6nQFdc+HzByDQ5KJsxpgQ5vP5eOmll/jFL37BsGHDyM3NZdasWTQ0NDBp0iSGDBnC8OHDufXWW0lJSWHcuHG8+uqrQWsstmmoW8nna4uZ9PiX/PqcAVx3ck9YOR2mXQEXPg5DLvI6PGNCmk1DbdNQtwmj+6Rxcp80/jljHWVVddD/XEjrC5/93WYmNcZ4yhJBK7rr7P6UVdXxn0/XO3MQjb4DClfAug+9Ds0YE8YsEbSiQRnJnJ+byROfbyS/rAoGXwhxqdaV1JhW0N5ugx+pI7lOSwSt7I6z+qIKD3ywxuk6OnQCfPUO7C7xOjRjQlZMTAwlJSUhnwxUlZKSEmJiYg7rdTaOoJV16xjHFSf04MkvNnLdyT3pO3wSzHkYlk2DUTd5HZ4xISkrK4u8vDyKioq8DiXoYmJiyMrKOqzXWCLwwM2n9Wba/K38+Z3VPH71sZAxAhY+A8ffCCJeh2dMyImMjCQnJ8frMNosuzXkgQ7xUdx0ai8+Wl3IlxtKYPgkp9E4f7HXoRljwpAlAo9cc1IOXZJi+NM7q9HBF0BEDCx61uuwjDFhyBKBR2Ii/dxxVl8Wby1l5tZ6GDAelr0IdVVeh2aMCTOWCDw0PjeDmEgfM1YXwvDLnQVrVr/ldVjGmDBjicBDMZF+RvVMZeaaIsgeA8nd7faQMabVBTURiMhYEflKRNaJyF0HOecSEVkpIitE5LlgxtMWjemTzobi3WwtrXZqBRs+gdItXodljAkjQUsEIuIHHgbOBgYCE0Vk4AHn9AHuBk5S1UHA7cGKp60a0zcNgM/WFkPuZU7h4uc9jMgYE26CWSM4DlinqhtUtRaYCpx3wDk/BB5W1V0AqloYxHjapF7pCWQkxzi3h1K6Q89TYPGzEAh4HZoxJkwEMxFkAlsb7ee5ZY31BfqKyBciMkdExjb1RiJyvYjMF5H5oTYyUEQY0zedL9YXU98QgOFXOLeGNrX8nOPGGNMUrxuLI4A+wKnAROBREUk58CRVnaKqI1V1ZHp6eiuHGHxj+qZTUV3P4q2l0P8ciEm2RmNjTKsJZiLYBnRrtJ/lljWWB0xX1TpV3QiswUkMYeWkXmn4BGauLYbIWBh8Eaya7nQnNcaYIAtmIpgH9BGRHBGJAiYA0w845zWc2gAikoZzq2hDEGNqk5LjIhnWLcVpJwBnyon6alj+sreBGWPCQtASgarWAzcD7wGrgGmqukJE7hWR8e5p7wElIrISmAH8TFXDcj7mMX3SWZpXSumeWsgYDukDYMkLXodljAkDQW0jUNW3VbWvqvZS1fvcst+o6nT3uarqHao6UFWHqOrUYMbTlo3pm0ZA4Yt1Jc4MpEMvga1zYNcmr0MzxoQ4rxuLjWtYVgqJMRH7bw/tXdB+2YveBWWMCQuWCNqICL+P0b3TmLm2yFlFKaU7dD8Rlk6zxe2NMUFliaANGdM3nfyyatYVVjoFQy+B4jWQv8TbwIwxIc0SQRtych9nuomZa4udgoHngS/SqRUYY0yQWCJoQ7I6xNEzPX5/O0FcR+j7XVj+EgQavA3OGBOyLBG0MWP6pPPlxhKq69w//EMuhsodsPFTbwMzxoQsSwRtzJi+aVTXBZi/aZdT0HcsRCfBUus9ZIwJDksEbcyonqlE+X3MXOveHoqMgYHjnSknavd4G5wxJiRZImhj4qIiGJndYX87AcDQS6G2Eta8411gxpiQZYmgDRrTN53VBRXsKK92CnqMhsQM6z1kjAkKSwRt0L5upHtrBT6fM9J43YewOyynYjLGBJElgjZoYNckuibH8M7ygv2FQy+BQD2seMW7wIwxIckSQRskIpw/PJNP1xRRVFHjFHYeDJ0G2txDxpgWZ4mgjbpgeCYNAeX1xe5aPiLOmIKtX8LOjd4GZ4wJKZYI2qg+nRMZmpXMKwsbLeo25GLncdlL3gRljAlJlgjasAtHZLEyv5xV+eVOQUo3pwfRkudsRlJjTIuxRNCGjRuWQaRfeGVh3v7C4ZNg5wbY/IV3gRljQoolgjasY3wUp/XrxGuLt1PfEHAKB57nTDmx8GlvgzPGhAxLBG3cBSOyKKqo4bN17tTUUXFOW8HK16Fql7fBGWNCgiWCNu60/umkxEV+vdF4xJVQX22NxsaYFmGJoI2LjvAzflgG768ooLy6zinMyIUuQ2HBU9ZobIw5apYI2oELRmRRUx/g7aX5+wtHXAk7lkH+Yu8CM8aEBEsE7cCwrGR6pcd/c0xBRAwsfMa7wIwxIcESQTsgIlwwIou5m3aypcRdkyA2xelBtOxFW6fAGHNULBG0E98fnokIvLKo0ZiCEVdCTbnTg8gYY46QJYJ2IiMllhN7pfLKwm3o3gbiHidBx542psAYc1SCmghEZKyIfCUi60TkriaOXy0iRSKy2N2uC2Y87d0Fw7PYsnMP8ze74wdEnFrBlllQvNbb4Iwx7VbQEoGI+IGHgbOBgcBEERnYxKkvqGquuz0WrHhCwdjBXYiL8vPi/K37C4ddBuK3WoEx5ogFs0ZwHLBOVTeoai0wFTgviJ8X8uKjIzh/eCavLtpG3i63gTixM/Q7G5Y8D/W13gZojGmXgpkIMoFGP13Jc8sOdKGILBWRl0SkWxDjCQk/Pq03gvDPj9ftLxx+BewugjXveheYMabd8rqx+A0gW1WHAh8ATzV1kohcLyLzRWR+UVFRqwbY1mSmxHLZ8d15cUEem4p3O4W9z4TErnZ7yBhzRIKZCLYBjX/hZ7ll+6hqiaq6azHyGHBMU2+kqlNUdaSqjkxPTw9KsO3Jj07tRaRfmPyR20Dsj3Aajdd9CDtWeBucMabdCWYimAf0EZEcEYkCJgDTG58gIl0b7Y4HVgUxnpDRKSmGK0/I5rXF21hXWOEUHn8jRCXAJ/d7G5wxpt0JWiJQ1XrgZuA9nD/w01R1hYjcKyLj3dNuFZEVIrIEuBW4OljxhJobxvQkNtLPAx+6tYK4jnDCj2DVdMhf6m1wxph2RbSdzV45cuRInT9/vtdhtAl/e/8rHvp4HW/fejIDM5KgqhT+MRSyT4KJz3sdnjGmDRGRBao6sqljXjcWm6Nw3eieJMZE8MCHa5yC2BQ48Wb46m3YttDb4Iwx7YYlgnYsOS6S60/uyQcrd7Bka6lTePyNENsBPvmTt8EZY9oNSwTt3A9G59AhLpK/f+DWCmKS4MRbYe37sHWet8EZY9oFSwTtXEJ0BDee0otP1xQxb9NOp/C46yEuFT75o7fBGWPaBUsEIeDKE7JJS4jm/977ypmZNDoBTrod1n8Mm2d7HZ4xpo2zRBACYqP83HFWX+Zu3MkzczY7hcdeB/GdrFZgjPlWlghCxMTjunFqv3Tue2sVa3dUQFQcjP4JbJwJGz/zOjxjTBtmiSBEiAh/uWgo8dER3Dp1MTX1DTDyB5DQxelB1M7GixhjWo8lghDSKTGGv1w4lFX55fzt/TUQGQtjfgqbv4BVb3gdnjGmjbJEEGLOHNiZy4/vzpSZG/hiXTEc8wPoMhTe/pkz8tgYYw5giSAE/fqcgfRMj+fOaUsorQnA+Iec9Qo++I3XoRlj2iBLBCEoNsrP5AnDKdldw92vLEO7DoMTfgwLn7KGY2PMNzQrEYhIvIj43Od9RWS8iEQGNzRzNAZnJnPHWf14Z3kBLy7Ig1Pvhg7Z8MZtUFfldXjGmDakuTWCmUCMiGQC7wNXAP8NVlCmZVw/piejenbknukrWLurAcY9CDvXw6d/8To0Y0wb0txEIKq6B7gA+JeqXgwMCl5YpiX4fcIDl+YSFxXBdU/PZ1fnEyF3EnzxIBQs8zo8Y0wb0exEICInAJcDb7ll/uCEZFpS1+RYHrniGPJLq7npfwuoO/NeZxGb6bdAQ73X4Rlj2oDmJoLbgbuBV91VxnoCM4IXlmlJx/TowP0XDmHOhp389oPt6Nl/ge2L4Mt/ex2aMaYNiGjOSar6KfApgNtoXKyqtwYzMNOyLhiRxZodlfzn0/X06zScq/qeDR/fB/3PgY49vQ7PGOOh5vYaek5EkkQkHlgOrBSRnwU3NNPSfv7dfpw5oBP3vrWKLwf9EvyR8PotEAh4HZoxxkPNvTU0UFXLgfOBd4AcnJ5Dph3x+YR/TBhOn04JXPdqPoUn/gY2fw7zHvM6NGOMh5qbCCLdcQPnA9NVtQ6wWczaoYToCB69ciRRfh+Xzu1NXc7p8OFvYecGr0MzxnikuYngEWATEA/MFJEeQHmwgjLB1a1jHP+54hi2lVZzS+UPUF8EvH6z3SIyJkw1KxGo6mRVzVTV76ljM3BakGMzQXRsdkf+dskw3t3q5+mkG5wZSudO8TosY4wHmttYnCwifxeR+e72N5zagWnHxg3L4DfnDuS3W3P5KukE9MN7oGS912EZY1pZc28NPQFUAJe4WznwZLCCMq3nmtE53DCmF1cWXk6t+uH1H9stImPCTHMTQS9V/a2qbnC33wHW+TxE/GJsf04cPoRfVk2CLbPhy/94HZIxphU1NxFUicjovTsichJgU1iGCJ9P+POFQyns+X0+Coyg4cPfQfE6r8MyxrSS5iaCG4GHRWSTiGwC/gnc8G0vEpGxIvKViKwTkbsOcd6FIqIiMrKZ8ZgWFhXh499XjOTp1NupqPdT+dxVUFftdVjGmFbQ3F5DS1R1GDAUGKqqw4HTD/UaEfEDDwNnAwOBiSIysInzEoHbgC8PM3bTwhKiI/jrNWP5v9jbSNi5nCWP3kBtvbUXGBPqDmuFMlUtd0cYA9zxLacfB6xz2xRqganAeU2c93vgz4D9/GwD0hOjuev2O/g47XKGFb7GQ3+/l9UFNmTEmFB2NEtVyrcczwS2NtrPc8v2v4HICKCbqr7FIYjI9Xu7rhYVFR1RsKb5EmMiOf2myezsNIof73mYnz/0HP/5dD0NARtMbkwoOppEcFR/FdxZTP8O3PmtH6Q6RVVHqurI9PT0o/lY01z+CDpe+QxRCR15LPZB/vXOAiZMmc2Wkj1eR2aMaWGHTAQiUiEi5U1sFUDGt7z3NqBbo/0st2yvRGAw8InbAD0KmG4Nxm1IQid8lzxFeqCQd7OnsrqgnHH//Jy5G3d6HZkxpgUdMhGoaqKqJjWxJarqt61lMA/oIyI5IhIFTACmN3rvMlVNU9VsVc0G5gDjVXX+UV6TaUndRyFn/Z6Mgo/47KTlpCZEMemxL3ljyXavIzPGtJCjuTV0SKpaD9wMvAesAqa5q5vdKyLjg/W5JghG3QQDzydl1n1MP0cZ1i2ZW55fxCOfrkfV2g2Mae+kvf2PPHLkSJ0/3yoNra6mAqacBrsLqZn4Mnd+4efNpflcMaoH94wfhN/3bX0HjDFeEpEFqtrkrfeg1QhMiIlOhEkvQ0wK0f/7PpNPrOGGU3ryzJzN3PDMfPbU1nsdoTHmCFkiMM3XoQf84B1I7Izvfxdyd/8ifn/eID5eXciF/57Np2uK7FaRMe2QJQJzeJIz4eq3IaU7/O9irkhbx6NXjqR0Ty1XPTGX8/81i49W7bCEYEw7YonAHL7EznD1W5DWB6ZO5AxZwCc/O5U/fn8IJZU1XPvUfM596HPeXV5AwAahGdPmWWOxOXJVu+CZC6BgKVz4GAz6PnUNAV5btI2HZ6xjU8ke+nVOZMJx3Rg3LIO0hGivIzYmbB2qsdgSgTk61eXwv4shby6M/ycMvxyA+oYAby7N59HPNrBiezl+n3Bq33S+PyKTMwd0JibS73HgxoQXSwQmuGp3wwuTYP3H8N0/wQk/+trhrwoqeGVRHq8v2k5BeTWJ0RF8b0hXxg7uwqieqcRGWVIwJtgsEZjgq6+Bl6+DVdPhlF/AqXeDfH1sQUNAmb2+hFcW5fHu8gL21DYQHeHjhF6pnN6/E6f160S3jnEeXYAxoc0SgWkdDfXw5m2w6Fk47gYYez/4mu6PUF3XwJcbdzJjdSEzvipkszuZXa/0eM4e3JWLR2bRIzW+NaM3JqRZIjCtRxXe/zXM/icMm+i0G/i/bVoq2Fi8mxmrC/l4dSGz1hcTUBjVsyOXjOzG2YO72u0jY46SJQLTulRh5l9hxh+g3zlw0eMQGdvsl+eXVfHKwm1Mm7+VzSV7SIyOYFxuBhOP7c6QrOQgBm5M6LJEYLzx5RR45+eQeQxMfB4SOh3WywMBZe6mnUybt5W3l+dTXRfguJyOXH9yT07v3wmfzW9kTLNZIjDeWfUGvHI9xKXCZS9A50FH9Dbl1XVMm7eVJz7fyPayanqmx3Pt6BwuHJFlXVGNaQZLBMZb2xfD8xOcGUwvehL6fueI36quIcDby5zxCcu3ldMxPopLRnZjcGYSPdMSyEmLt/YEY5pgicB4r3w7PHcp7Fju9CY6/oajejtVZc6GnTz22QY+/qqQvV9jEchIjqVnejy9OyWQ2y2FY3p0IDMlFhG7lWTClyUC0zbU7nZuE61+E479oZMQmtGj6NtU1TawsXg364sq2VDkPhZXsr5wN1V1DQB0SYrhmB4dOKZHB0b06EBqfBR+n+D3CSLgF+d5UkyktT2YkGSJwLQdgQB8dA988SDkjIGL/gvxqUH5qPqGAKsLKliwede+bVtp1SFfk9UhlmtH53Dpsd2Iizr6JGVMW2GJwLQ9i5+DN26HhM5w6TOQkdsqH5tfVsXiLaVU1NQTCCgNqgQCSkCd9od3lhewYPMukmMjuWJUD646MZv0RJssz7R/lghM27RtIbxwBewphnGTYdilXkcEwILNO3nk0w18sGoHkX4fF47IZNzQDBCoa1DqGwLUNQSoa1Ai/T5G9exISlyU12Ebc0iWCEzbVVkEL14Nmz+H42+C7/we/JFeRwXAhqJKHv1sIy8vzKO2PnDQ8/w+4ZgeHThzQCfOGNCZXukJzXr/6roGlmwtZf7mXSzasgtVyO2WwvDuHRjaLZmkmLbx78GEBksEpm1rqIMPfgNz/gU9RsPF/4WEdK+j2qe4soZV+eVE+HxERQgRPh+Rfh+RfqGsqo4ZXxXy0apCVhdUAJCTFs+p/dJJS4gmwidEuOdG+Hz4fbBmRyXzN+9ixbYy6t2Fe3qmx/VpKkUAABSlSURBVOMTYV1hJeD0fuqdnsDw7in07pSAKvtuY9UHnMeoCB+n9O3E4Mwk6xFlvpUlAtM+LJ0G02+B6CQ47+GjGm/ghbxde/h4tZMUZq8vobah6VpEdISPYVkpjOjRgZFuL6aO8c6tpbKqOpbmlbJoSymLtuxi0dZSSvfUHfJzs1PjOGdoV84ZksGArolfSwq19QHW7KhgxfYyVm4vJyEmgiGZKQzNSqZrcowlkIMIBLTZvccKy6tZkV9OTISf7LQ4OifGHHHPs60797BmRwWDM5PpnBRzRO9xMJYITPuxYwW8/EMoXAEjr4Xv/AGi2t/U1IGAUhcIUN+g1Dfsf17XEKBzUgxREc1bJVZVKa+uJ8Lt6ur3CX4RfD5h1+5a3l9ZwJtL85m1voSGgNIzPZ6zBnamvKqOZdvK+KqggroG5//x+Cg/NfWBfbWQtIQohmQmMyQrhS5JMeyprWdPbQO7a+vZU+M8CkKvTvH07ZRI386JZHWIPaI/cg0BZXdtPVW1DVTVNrCntoGqOuc5QGJMhLtFkhgTcUSjxXfXOPEDKIr7DwCxUX4SoyOaTHyqSt6uKuZu3Mm8TTuZu2knm4p309Udj5KTFk92ajw56fFkpsSyuWQPy7aVsdzdCitqvvZ+MZE+slOd1/RIi6N7xzi6JsfQNTmWrskxJMdG7otje2kVs9eXMHtDCbPXl3ytV1vPtHhG9UrlhJ6pjOqZetSdFiwRmPalrho+/j3MfhhSe8EFU5z5isxBlVTW8O6KAt5cks+cjSUkxUQyJDOZQZlJDM5IZkhmMt07xlHbEGBVfjnLtpWxNK+MZXllrC2soPHS0lERPuKj/MRFRVAfCLCjfP8futhIP707JdArPZ6k2EhiI/3ERPqJjfITG+knOsLHrj11FJRVUVBeTUF5DTvKqimqrKHhMNavjorw0SEukpy0eHqlJ9Az3fnMXukJdE2OYeuuKlbnl7Mqv5xVBRWsyi8nb9ehuwZHRfhIi48iNSGatATnsbqugXmbdu67xqSYCEZmd6Rv50QKyqrYWLybDUW7qaip/9p7+QR6pSe4/46TGZyRRF2DsrFkN5uK3a1kN1t27tmXiPeKifTRNTmWhoCyZacz/XpKXCSjclI5oVcq/boksiyvjNkbSpi7cSeV7mf36ZTAnd/py9jBXZv977ExSwSmfdo4E169CSoLnMVuRt/RIgPQQl1VbQMxkb5m3/bZU1tPWVUdcVERxEX5ifR/vbZSVlXHusIK1uyoZM2OCtbuqGRj8e59v/BrmmhIT4yOoEtyDF2SY+icFEOXpBhS4iKJjfIT5yaN2KgIYt1f/hXVdVRU11Pe6LG4opaNxZWsK6ykvLr+G58Bzh/knLR4BnRNon+XRJJj3QZ2EYT9ayPtqWmgeHcNxRW1lOyuobiyhpLKWnwijOjRgeOyO3BsTkf6dkr8Ro1HVSnZXcvG4t3k7dpDtw5xDMxIatY4k4aAUlhRTX5ZNQVl1WwvraKgrJr88moaGpRjczpyQs9U+nf55ueCMxZm+fbyfbWGa0fncErfI2s/s0Rg2q+qUnj7p7DsRegyFMb8DPqfe9AFb0zrCwSU6nrnNk91fYCU2Ejio1suYe/9Q7x31Pi2XVV07xhH/67O7SqbdLB5PEsEIjIWeBDwA4+p6v0HHL8R+DHQAFQC16vqykO9pyWCMLX8Zfjo97BrI6T2gZNug6GXQoT13zemOTxJBCLiB9YAZwF5wDxgYuM/9CKSpKrl7vPxwI9Udeyh3tcSQRgLNMDK1+HzB6BgKSRmwAk/hmOuguhEr6Mzpk07VCIIZv36OGCdqm5Q1VpgKnBe4xP2JgFXPPsb+Y35Jp8fBl8AN8yESa84Dcnv/woeHAYrXvM6OmParWAmgkxga6P9PLfsa0TkxyKyHvgLcGtTbyQi14vIfBGZX1RUFJRgTTsiAr3PgKvfhOs+gpTu8OJVTrfTqlKvozOm3fG8xU1VH1bVXsAvgF8f5JwpqjpSVUemp7edEaemDcgaCdd+AKfe7bQj/PtEWD/D66iMaVeCmQi2Ad0a7We5ZQczFTg/iPGYUOWPhFPvgus+hKh4eOZ8ePvnULvH68iMaReCmQjmAX1EJEdEooAJwPTGJ4hIn0a75wBrgxiPCXWZI5z2g+NvgrmPwCNjnLEIxphDCloiUNV64GbgPWAVME1VV4jIvW4PIYCbRWSFiCwG7gCuClY8JkxExsLZ98OV06GhBp4aB9OuhNItXkdmTJtlA8pM6Kqrgln/hM/+BiicdLsz/qAdzl1kzNHyqvuoMd6KjIVTfgY3z4N+34NP74eHj3O6mrazH0DGBJMlAhP6UrrBxU/C1W9BTLLT1fTR02HVG84aysaEOUsEJnxkj4brP3WWxazaCS9Mgn+NgsXPO4vjGBOmLBGY8OKPcKakuHkBXPg4+CLgtRth8gj4cop1OTVhyRKBCU/+CBhyEdz0BUx8AZK6wjs/g7/3h7fuhO2LvY7QmFZjk7ub8CYC/cZC3+/Cltkw/0lY+AzMe8yZ9nrElU7CiO3gdaTGBI11HzXmQFW7YNlLsPBpZ5bTiBgYeB4c+0NnSgtb59e0Q7YwjTFHavtiWPQMLHkBaisgYzgcdz0MugAiW3ZxcWOCyRKBMUerpgKWvgBzH4Wi1RDb0bltdOy1zuynxrRxNqDMmKMVnQjHXgc/mgNXvQHZJ8GsyfBgLrx7N1SXeR2hMUfMEoExh0MEcsbApc/CbUth+CSY82946BinkdkGqJl2yBKBMUcqpRuMnwzXfwIde8L0m+Gx02HrPK8jM+awWCIw5mhl5MI178EFj0J5Pjx+Jrx6I+zc6HVkxjSLJQJjWoIIDL0EbpkPo3/irJb20AhnGostX9okd6ZNs0RgTEuKToQz74HbljhTXm+cCU98Bx47E5a/Ag31XkdozDdY91FjgqmmEpY8D3P+BTs3QHJ36HMWpPVxt76QlAU++01mgutQ3Udtigljgik6AY77IYy8Bta8C3OnwPKXvt7dNCIWUns7o5aHXAzdT7DEYFqVJQJjWoPPD/3PcTZV2F0MJWuheA0Uu49LX4AFT0JyNxh8IQy9FDoP9DpyEwYsERjT2kQgId3Zepy4v7x2N6x+20kIsx6CL/4BnQc7jdDDLnPONyYIrI3AmLaosghWvArLpkHePGfdhH7fc9ZS6Hm63Toyh83mGjKmPSv6ypkJdfFzzspqyd1hxBXOqOakDK+jM+2EJQJjQkF9Dax+ExY8BRs/BfFBr9Nh2ESn7SEy1usITRtmvYaMCQUR0U4j8uALna6oi/4HS6bCy9dCdDIM/j7kXg5Zx9qaCeawWI3AmPYsEIBNM2Hx87BqOtTtcbqi5l7m1BTs1pFx2a0hY8JBTQWsfN1pS9j8hXPrqPdZTltC37EQEeV1hMZDlgiMCTcl62Hx/5ykUJEPcakwdAL0/Y7TJTU+zesITSvzLBGIyFjgQcAPPKaq9x9w/A7gOqAeKAKuUdXNh3pPSwTGHIZAA6z/2Ol19NU7EKhzyhM6Q+dB7jYYuo+CDtmehmqCy5PGYhHxAw8DZwF5wDwRma6qKxudtggYqap7ROQm4C/ApcGKyZiw4/M7cxv1OQuqdjlrMO9Y4W7L4ctHoKHWOTdnDIy4CgaMcxqmTdgIZq+h44B1qroBQESmAucB+xKBqs5odP4cYFIQ4zEmvMV2gF6nOdteDXVQsg5WvenUGl6+1lmPedhEZ/Baej/v4jWtJpiJIBPY2mg/Dzj+EOdfC7zT1AERuR64HqB7d1so3JgW44+ETgOc7eQ7YcPHzjiFuY/AnIehyxBndtT4NIhP378ldoaM4c6026bdaxPjCERkEjASOKWp46o6BZgCThtBK4ZmTPjw+aD3mc5WWeg0NG+YAWVbYftCZ6I8bdh/vvidZJBzMmSf7LQzRMV7F785YsFMBNuAbo32s9yyrxGRM4FfAaeoak0Q4zHGNFdCJxh9u7PtFQhAdamTEMq2wOZZsPEzZ4K8zx8AX6QzlfaAcc502gmdvIvfHJag9RoSkQhgDXAGTgKYB1ymqisanTMceAkYq6prm/O+1mvImDamphK2zHEGtq2fAQVLndpC7zNh2ARnsrzIGK+jDHue9BpS1XoRuRl4D6f76BOqukJE7gXmq+p04P+ABOBFcYbEb1HV8cGKyRgTBNEJ0OdMZzsLKFwNS6fCkhdg7Xv7p7/oOxa6DoPErjYFRhtjA8qMMcERaHDWbF4ydf/0F+A0NncZ6iSFrkOha64zhsGSQ1DZpHPGmNbn8+/vrlr7dyhYBvlLIH+p8zhrMgTqnXNjUpyG58wRzmPGcEjKtOTQSiwRGGOCLyre6VXUfdT+svoaKFzpJIVtC2H7Ivjiwf3JIaGzM8it56mQcwqkdGvqnU0LsERgjPFGRPT+X//HXO2U1VU5o563L4KtX8KGT2HZi86xjr3cpDDGqTkkd7MaQwuxNgJjTNulCoWrnIV4NnwCm76A2grnWEwydB7iDHrrMth57DTQGSRnvsHaCIwx7ZMIdB7obKNucqbEyF/idFEtWAYFy2HhU/sboqMSoMeJTq0hZ4yTKGx9529licAY03743UFrWY1+2AYaYOdGKFji1Bg2zoS17zvHYjs4o567DHGm4o5Pg7i0/Y+xHSxRYInAGNPe+fyQ1tvZBl/olJVvd0Y9b5zpbKumN/3amGRnfEP/c6DXGc6YiDBkicAYE3qSMmDYpc4GTg+lPSXO9Bh7imF3ifOYvxTWvAtLXwB/tNPVdV9SSHRqIL5IJ9mEcMO0JQJjTOiLiHaSQ1NrODfUw9Y5sPotWP2mkxia4osEf5TzHh2yoWOO89ghGzrkOPuRsUG8iOCxRGCMCW/+CMge7Wzf/aOzYM+WOU4tIlDnJIpAndNQXV/jzMa6a5PTvbWmvNEbiTPWIa0vpPaBNHdL7Q0JXdp0W4QlAmOM2UvE7Y465NvPVXVWfdu10WmsLlkPxWucbfOs/T2ZwLnt1KGHU3PYW4tIznIasOM6OosBxXaAiKhgXdkhWSIwxpgjIeL8EY/rCJnHfP1YIAAV252ksHODU4PYu22etX8sxIGiEp0eTWl9IL2/s3XqD2n9gtqQbYnAGGNams/n/OJPzoJep3/92N6aRNlW2LMTqna6j7ucx8odTgLZ8Mn+9aQBUrrDGb+FIRe1eLiWCIwxpjU1rkkcSkO9U4MoWuVM7V20ypm5NQgsERhjTFvkj9g/PmLAuKB+VNttxjbGGNMqLBEYY0yYs0RgjDFhzhKBMcaEOUsExhgT5iwRGGNMmLNEYIwxYc4SgTHGhLl2t2axiBQBm4/w5WlAcQuG016E63VD+F67XXd4ac5191DVJocmt7tEcDREZP7BFm8OZeF63RC+127XHV6O9rrt1pAxxoQ5SwTGGBPmwi0RTPE6AI+E63VD+F67XXd4OarrDqs2AmOMMd8UbjUCY4wxB7BEYIwxYS5sEoGIjBWRr0RknYjc5XU8wSIiT4hIoYgsb1TWUUQ+EJG17mMHL2MMBhHpJiIzRGSliKwQkdvc8pC+dhGJEZG5IrLEve7fueU5IvKl+31/QUS8WRU9yETELyKLRORNdz/kr1tENonIMhFZLCLz3bKj+p6HRSIQET/wMHA2MBCYKCIDvY0qaP4LjD2g7C7gI1XtA3zk7oeaeuBOVR0IjAJ+7P43DvVrrwFOV9VhQC4wVkRGAX8GHlDV3sAu4FoPYwym24BVjfbD5bpPU9XcRmMHjup7HhaJADgOWKeqG1S1FpgKnOdxTEGhqjOBnQcUnwc85T5/Cji/VYNqBaqar6oL3ecVOH8cMgnxa1dHpbsb6W4KnA685JaH3HUDiEgWcA7wmLsvhMF1H8RRfc/DJRFkAlsb7ee5ZeGis6rmu88LgM5eBhNsIpINDAe+JAyu3b09shgoBD4A1gOlqlrvnhKq3/d/AD8HAu5+KuFx3Qq8LyILROR6t+yovue2eH2YUVUVkZDtMywiCcDLwO2qWu78SHSE6rWragOQKyIpwKtAf49DCjoRORcoVNUFInKq1/G0stGquk1EOgEfiMjqxgeP5HseLjWCbUC3RvtZblm42CEiXQHcx0KP4wkKEYnESQL/U9VX3OKwuHYAVS0FZgAnACkisveHXih+308CxovIJpxbvacDDxL6142qbnMfC3ES/3Ec5fc8XBLBPKCP26MgCpgATPc4ptY0HbjKfX4V8LqHsQSFe3/4cWCVqv690aGQvnYRSXdrAohILHAWTvvIDOAi97SQu25VvVtVs1Q1G+f/549V9XJC/LpFJF5EEvc+B74DLOcov+dhM7JYRL6Hc0/RDzyhqvd5HFJQiMjzwKk409LuAH4LvAZMA7rjTOF9iaoe2KDcronIaOAzYBn77xn/EqedIGSvXUSG4jQO+nF+2E1T1XtFpCfOL+WOwCJgkqrWeBdp8Li3hn6qqueG+nW71/equxsBPKeq94lIKkfxPQ+bRGCMMaZp4XJryBhjzEFYIjDGmDBnicAYY8KcJQJjjAlzlgiMMSbMWSIw5gAi0uDO7Lh3a7GJ6kQku/HMsMa0BTbFhDHfVKWquV4HYUxrsRqBMc3kzgP/F3cu+Lki0tstzxaRj0VkqYh8JCLd3fLOIvKqu1bAEhE50X0rv4g86q4f8L47ItgYz1giMOabYg+4NXRpo2NlqjoE+CfOSHWAh4CnVHUo8D9gsls+GfjUXStgBLDCLe8DPKyqg4BS4MIgX48xh2Qji405gIhUqmpCE+WbcBaB2eBOcFegqqkiUgx0VdU6tzxfVdNEpAjIajzFgTtF9gfuAiKIyC+ASFX9Q/CvzJimWY3AmMOjB3l+OBrPfdOAtdUZj1kiMObwXNrocbb7fBbODJgAl+NMfgfOkoE3wb7FY5JbK0hjDof9EjHmm2LdFb/2eldV93Yh7SAiS3F+1U90y24BnhSRnwFFwA/c8tuAKSJyLc4v/5uAfIxpY6yNwJhmctsIRqpqsdexGNOS7NaQMcaEOasRGGNMmLMagTHGhDlLBMYYE+YsERhjTJizRGCMMWHOEoExxoS5/w80vjuHgjjqSAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MQIbUVBY6-SG",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "26fe21c3-aeb8-42a5-b5bf-4058be42d516"
      },
      "source": [
        "# Save model and weights\n",
        "if not os.path.isdir(save_dir):\n",
        "    os.makedirs(save_dir)\n",
        "model_path = os.path.join(save_dir, model_name)\n",
        "model.save(model_path)\n",
        "print('Saved trained model at %s ' % model_path)"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Saved trained model at /content/saved_models/keras_cifar10_trained_model.h5 \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "x8cshRKD7Bak",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 70
        },
        "outputId": "e09d242b-a39f-4d4f-f289-5d5c8e0fe62e"
      },
      "source": [
        "# Score trained model.\n",
        "scores = model.evaluate(x_train, s_train, verbose=1)\n",
        "print('Train loss:     %f' % np.mean(scores[0:5]))\n",
        "acc = 1.\n",
        "for i in range(5):\n",
        "    acc *= scores[6+i]\n",
        "print('Train accuracy: %.2f' % (acc * 100.))"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "8953/8953 [==============================] - 2s 169us/step\n",
            "Train loss:     0.014435\n",
            "Train accuracy: 99.20\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yfQ_5Ago7EWh",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 70
        },
        "outputId": "19d26d76-4ef6-4d67-f0da-b56442076319"
      },
      "source": [
        "scores = model.evaluate(x_test, s_test, verbose=1)\n",
        "print('Test loss:     %f' % np.mean(scores[0:5]))\n",
        "acc = 1.\n",
        "for i in range(5):\n",
        "    acc *= scores[6+i]\n",
        "print('Test accuracy: %.2f' % (acc * 100.))"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "1047/1047 [==============================] - 0s 160us/step\n",
            "Test loss:     0.117577\n",
            "Test accuracy: 90.06\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}