nsl_adversarial_classifier_cloud (2).ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "nsl_adversarial_classifier_cloud.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "cWNa__TuP9oW",
        "colab_type": "code",
        "outputId": "da03b91b-2186-4dea-f038-0f785c0d843d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 408
        }
      },
      "source": [
        "!nvidia-smi # P100\n",
        "!git clone https://github.com/klinime/Tiny_Imagenet_200.git"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Thu May 14 00:33:23 2020       \n",
            "+-----------------------------------------------------------------------------+\n",
            "| NVIDIA-SMI 440.82       Driver Version: 418.67       CUDA Version: 10.1     |\n",
            "|-------------------------------+----------------------+----------------------+\n",
            "| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n",
            "| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n",
            "|===============================+======================+======================|\n",
            "|   0  Tesla P100-PCIE...  Off  | 00000000:00:04.0 Off |                    0 |\n",
            "| N/A   32C    P0    26W / 250W |      0MiB / 16280MiB |      0%      Default |\n",
            "+-------------------------------+----------------------+----------------------+\n",
            "                                                                               \n",
            "+-----------------------------------------------------------------------------+\n",
            "| Processes:                                                       GPU Memory |\n",
            "|  GPU       PID   Type   Process name                             Usage      |\n",
            "|=============================================================================|\n",
            "|  No running processes found                                                 |\n",
            "+-----------------------------------------------------------------------------+\n",
            "Cloning into 'Tiny_Imagenet_200'...\n",
            "remote: Enumerating objects: 120574, done.\u001b[K\n",
            "remote: Total 120574 (delta 0), reused 0 (delta 0), pack-reused 120574\u001b[K\n",
            "Receiving objects: 100% (120574/120574), 212.26 MiB | 33.51 MiB/s, done.\n",
            "Resolving deltas: 100% (1108/1108), done.\n",
            "Checking out files: 100% (120204/120204), done.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "b9SqhiptHl4c",
        "colab_type": "code",
        "outputId": "cb51b66d-f033-4d0c-e5b9-68955756005c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 136
        }
      },
      "source": [
        "!git clone https://github.com/klinime/Tiny_Imagenet_200.git\n",
        "!pip install neural_structured_learning"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "fatal: destination path 'Tiny_Imagenet_200' already exists and is not an empty directory.\n",
            "Requirement already satisfied: neural_structured_learning in /usr/local/lib/python3.6/dist-packages (1.1.0)\n",
            "Requirement already satisfied: scipy in /usr/local/lib/python3.6/dist-packages (from neural_structured_learning) (1.4.1)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from neural_structured_learning) (1.12.0)\n",
            "Requirement already satisfied: absl-py in /usr/local/lib/python3.6/dist-packages (from neural_structured_learning) (0.9.0)\n",
            "Requirement already satisfied: attrs in /usr/local/lib/python3.6/dist-packages (from neural_structured_learning) (19.3.0)\n",
            "Requirement already satisfied: numpy>=1.13.3 in /usr/local/lib/python3.6/dist-packages (from scipy->neural_structured_learning) (1.18.4)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Em6b5N7kB1Za",
        "colab_type": "code",
        "outputId": "bf6d5848-9926-421c-cec4-b8e68713d483",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "import numpy as np\n",
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "from tensorflow.keras import backend as K\n",
        "import neural_structured_learning as nsl\n",
        "\n",
        "from google.colab import drive\n",
        "drive.mount('/content/gdrive')\n",
        "import pathlib\n",
        "import os\n",
        "\n",
        "from tensorflow.python.util import deprecation\n",
        "deprecation._PRINT_DEPRECATION_WARNINGS = False\n",
        "\n",
        "PATH = '/content/gdrive/My Drive/CS182_Project/'"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/gdrive; to attempt to forcibly remount, call drive.mount(\"/content/gdrive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OWCcwpZWWHc7",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Data augmentation based on:\n",
        "# https://www.wouterbulten.nl/blog/tech/data-augmentation-using-tensorflow-data-dataset/\n",
        "\n",
        "def hflip(img: tf.Tensor) -> tf.Tensor:\n",
        "  return tf.image.random_flip_left_right(img)\n",
        "\n",
        "# interpolation pads 0 instead, undesired so not used\n",
        "# import tensorflow_addons as tfa\n",
        "# def random_rotate(img: tf.Tensor) -> tf.Tensor:\n",
        "#   return tfa.image.rotate(img, tf.random.uniform([], -0.5, 0.5))\n",
        "\n",
        "def random_crop(img: tf.Tensor) -> tf.Tensor:\n",
        "  return tf.image.random_crop(tf.image.resize(\n",
        "    img, tf.shape(img)[:2]+16), tf.shape(img))\n",
        "\n",
        "def color_jitter(img: tf.Tensor) -> tf.Tensor:\n",
        "  img = tf.image.random_hue(img, 0.04)\n",
        "  img = tf.image.random_saturation(img, 0.4, 1.6)\n",
        "  img = tf.image.random_brightness(img, 0.08)\n",
        "  img = tf.image.random_contrast(img, 0.7, 1.3)\n",
        "  return img\n",
        "\n",
        "def noise_salt_pepper(img: tf.Tensor) -> tf.Tensor:\n",
        "  mask_img = K.random_binomial(shape=tf.shape(img), p=0.01)\n",
        "  mask_noise = K.random_binomial(shape=tf.shape(img), p=0.5)\n",
        "  out = img * (1 - mask_img) + mask_noise * mask_img\n",
        "  # Issue: https://github.com/tensorflow/tensorflow/issues/24520\n",
        "  out.set_shape([None, None, None])\n",
        "  return out\n",
        "\n",
        "def noise_gaussian(img: tf.Tensor) -> tf.Tensor:\n",
        "  out = img + tf.random.normal(tf.shape(img), 0, tf.sqrt(0.01))\n",
        "  # Issue: https://github.com/tensorflow/tensorflow/issues/24520\n",
        "  out.set_shape([None, None, None])\n",
        "  return out\n",
        "\n",
        "def noise_poisson(img: tf.Tensor) -> tf.Tensor:\n",
        "  # idk what im doing for this one, but it kinda works\n",
        "  vals = tf.cast(tf.math.reduce_prod(tf.shape(img)), tf.float32)\n",
        "  vals = tf.math.exp(tf.math.floor(tf.math.log(vals)))\n",
        "  out = tf.squeeze(tf.random.poisson([1], img * vals)) / vals\n",
        "  # Issue: https://github.com/tensorflow/tensorflow/issues/24520\n",
        "  out.set_shape([None, None, None])\n",
        "  return out\n",
        "\n",
        "# Image loading based on:\n",
        "# https://www.tensorflow.org/tutorials/load_data/images\n",
        "\n",
        "AUTOTUNE = tf.data.experimental.AUTOTUNE\n",
        "\n",
        "def load_train_data(folder):\n",
        "  train_dir = pathlib.Path(folder)\n",
        "  train_image_count = len(list(train_dir.glob('**/*.JPEG')))\n",
        "  train_list_ds = tf.data.Dataset.list_files(str(train_dir/'*/images/*'))\n",
        "  print('Discovered {} images for training data.'.format(train_image_count))\n",
        "\n",
        "  class_names = np.array(sorted(\n",
        "      [item.name for item in pathlib.Path(train_dir).glob('*')]))\n",
        "  def process_path(file_path):\n",
        "    # convert the path to a list of path components\n",
        "    parts = tf.strings.split(file_path, os.path.sep)\n",
        "    # third to last is the class-directory\n",
        "    label = parts[-3] == class_names\n",
        "    \n",
        "    # encode label from string to int index in class_names\n",
        "    label = tf.argmax(tf.dtypes.cast(label, tf.int32), axis=0)\n",
        "    # load the raw data from the file as a string\n",
        "    img = tf.io.read_file(file_path)\n",
        "    # convert the compressed string to a 3D uint8 tensor\n",
        "    img = tf.image.decode_jpeg(img, channels=3)\n",
        "    # use 'convert_image_dtype' to convert to floats in the [0, 1] range\n",
        "    img = tf.image.convert_image_dtype(img, tf.float32)\n",
        "    return img, label\n",
        "  \n",
        "  train_labeled_ds = train_list_ds.map(process_path, num_parallel_calls=AUTOTUNE)\n",
        "  return class_names, train_labeled_ds, train_image_count\n",
        "\n",
        "def load_val_data(folder, class_names):\n",
        "  val_dir = pathlib.Path(folder)\n",
        "  val_image_count = len(list(val_dir.glob('*/*.JPEG')))\n",
        "  val_path = str(val_dir/'images')\n",
        "  val_list_ds = tf.data.Dataset.from_tensor_slices(\n",
        "      sorted(['{}/{}'.format(val_path, item.name)\n",
        "              for item in val_dir.glob('images/*')],\n",
        "             key=lambda x: int(x[x.rfind('_')+1:x.find('.')])))\n",
        "  val_label_ds = tf.data.TextLineDataset(str(val_dir/'val_annotations.txt'))\n",
        "  val_label_ds = val_label_ds.map(\n",
        "      lambda x: tf.argmax(tf.dtypes.cast(\n",
        "          tf.strings.split(x, '\\t')[1] == class_names, tf.int32), axis=0), \n",
        "      num_parallel_calls=AUTOTUNE, deterministic=True)\n",
        "  print('Discovered {} images for validation data.'.format(val_image_count))\n",
        "\n",
        "  def process_images(file_path):\n",
        "    img = tf.io.read_file(file_path)\n",
        "    img = tf.image.decode_jpeg(img, channels=3)\n",
        "    img = tf.image.convert_image_dtype(img, tf.float32)\n",
        "    return img\n",
        "  \n",
        "  val_img_ds = val_list_ds.map(\n",
        "      process_images, num_parallel_calls=AUTOTUNE, deterministic=True)\n",
        "  val_labeled_ds = tf.data.Dataset.zip((val_img_ds, val_label_ds))\n",
        "  return val_labeled_ds, val_image_count\n",
        "\n",
        "def prepare_for_training(ds, batch_size, cache=True, shuffle_buffer_size=1000):\n",
        "  # use '.cache(filename)' to cache preprocessing work for big datasets\n",
        "  if cache:\n",
        "    if isinstance(cache, str):\n",
        "      ds = ds.cache(cache)\n",
        "    else:\n",
        "      ds = ds.cache()\n",
        "  for augment in [hflip, random_crop, color_jitter]:\n",
        "    ds = ds.map(lambda img, label: (tf.cond(tf.random.uniform([], 0, 1) < 0.5, \n",
        "        lambda: augment(img), lambda: img), label), num_parallel_calls=AUTOTUNE)\n",
        "  for augment in [noise_salt_pepper, noise_gaussian, noise_poisson]:\n",
        "    ds = ds.map(lambda img, label: (tf.cond(tf.random.uniform([], 0, 1) < 0.2, \n",
        "        lambda: augment(img), lambda: img), label), num_parallel_calls=AUTOTUNE)\n",
        "  ds = ds.map(lambda img, label: (tf.clip_by_value(img, 0, 1), label), \n",
        "              num_parallel_calls=AUTOTUNE)\n",
        "  ds = ds.map(lambda img, label: (tf.image.per_image_standardization(img), label),\n",
        "              num_parallel_calls=AUTOTUNE)\n",
        "  \n",
        "  ds = ds.shuffle(buffer_size=shuffle_buffer_size).repeat().batch(batch_size)\n",
        "  # 'prefetch' fetches batches in the background while the model is training\n",
        "  ds = ds.prefetch(buffer_size=AUTOTUNE)\n",
        "  return ds\n",
        "\n",
        "def prepare_for_validation(ds, batch_size, cache=True, shuffle_buffer_size=1000):\n",
        "  if cache:\n",
        "    if isinstance(cache, str):\n",
        "      ds = ds.cache(cache)\n",
        "    else:\n",
        "      ds = ds.cache()\n",
        "  ds = ds.map(lambda img, label: (tf.clip_by_value(img, 0, 1), label), \n",
        "              num_parallel_calls=AUTOTUNE)\n",
        "  ds = ds.map(lambda img, label: (tf.image.per_image_standardization(img), label),\n",
        "              num_parallel_calls=AUTOTUNE)\n",
        "  ds = ds.shuffle(buffer_size=shuffle_buffer_size).batch(batch_size)\n",
        "  ds = ds.prefetch(buffer_size=AUTOTUNE)\n",
        "  return ds\n",
        "\n",
        "def prepare_for_testing(folder, class_names, batch_size):\n",
        "  test_dir = pathlib.Path(folder)\n",
        "  test_image_count = len(list(test_dir.glob('*/*.JPEG')))\n",
        "  test_path = str(test_dir/'images')\n",
        "  test_list_ds = tf.data.Dataset.from_tensor_slices(\n",
        "      sorted(['{}/{}'.format(test_path, item.name)\n",
        "              for item in test_dir.glob('images/*')],\n",
        "             key=lambda x: int(x[x.rfind('_')+1:x.find('.')])))\n",
        "  print('Discovered {} images for test data.'.format(test_image_count))\n",
        "\n",
        "  def process_images(file_path):\n",
        "    img = tf.io.read_file(file_path)\n",
        "    img = tf.image.decode_jpeg(img, channels=3)\n",
        "    img = tf.image.convert_image_dtype(img, tf.float32)\n",
        "    return img\n",
        "  \n",
        "  ds = test_list_ds.map(process_images, num_parallel_calls=AUTOTUNE, deterministic=True)\n",
        "  ds = ds.map(lambda img: tf.clip_by_value(img, 0, 1), \n",
        "              num_parallel_calls=AUTOTUNE, deterministic=True)\n",
        "  ds = ds.map(lambda img: tf.image.per_image_standardization(img),\n",
        "              num_parallel_calls=AUTOTUNE, deterministic=True)\n",
        "  ds = ds.batch(batch_size)\n",
        "  ds = ds.prefetch(buffer_size=AUTOTUNE)\n",
        "  return ds\n",
        "\n",
        "def prepare_data(params, folder):\n",
        "  class_names, train_data, train_count = load_train_data(folder + 'train/')\n",
        "  val_data, val_count = load_val_data(folder + 'val/', class_names)\n",
        "  batch_size = params['batch_size']\n",
        "  val_ratio = 10\n",
        "  train_data = prepare_for_training(train_data, batch_size)\n",
        "  val_data = prepare_for_validation(val_data, batch_size)\n",
        "  train_data_adv = train_data.map(lambda x, y: {'image': x, 'label': y},\n",
        "                                  num_parallel_calls=AUTOTUNE)\n",
        "  val_data_adv = val_data.map(lambda x, y: {'image': x, 'label': y},\n",
        "                              num_parallel_calls=AUTOTUNE, deterministic=True)\n",
        "  steps_per_epoch = np.ceil(train_count / batch_size)\n",
        "  validation_steps = np.ceil(val_count / batch_size / val_ratio)\n",
        "  return class_names, train_data_adv, val_data_adv, steps_per_epoch, validation_steps"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3jbCW3V1rZ3s",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from tensorflow.python.keras import backend\n",
        "from tensorflow.python.keras.applications import imagenet_utils\n",
        "from tensorflow.python.keras.engine import training\n",
        "from tensorflow.python.keras.utils import data_utils\n",
        "from tensorflow.keras import regularizers\n",
        "\n",
        "BASE_WEIGHT_PATH = ('https://storage.googleapis.com/tensorflow/'\n",
        "                    'keras-applications/mobilenet_v2/')\n",
        "layers = tf.keras.layers\n",
        "\n",
        "def MobileNetV2(input_shape=None,\n",
        "                alpha=1.0,\n",
        "                include_top=True,\n",
        "                weights='imagenet',\n",
        "                pooling=None,\n",
        "                l2_reg=1e-4):\n",
        "  \n",
        "  # If input_shape is not None, assume default size\n",
        "  if backend.image_data_format() == 'channels_first':\n",
        "    rows = input_shape[1]\n",
        "    cols = input_shape[2]\n",
        "  else:\n",
        "    rows = input_shape[0]\n",
        "    cols = input_shape[1]\n",
        "\n",
        "  if rows == cols and rows in [96, 128, 160, 192, 224]:\n",
        "    default_size = rows\n",
        "  else:\n",
        "    default_size = 224\n",
        "\n",
        "  input_shape = imagenet_utils.obtain_input_shape(\n",
        "      input_shape,\n",
        "      default_size=default_size,\n",
        "      min_size=32,\n",
        "      data_format=backend.image_data_format(),\n",
        "      require_flatten=include_top,\n",
        "      weights=weights)\n",
        "\n",
        "  if backend.image_data_format() == 'channels_last':\n",
        "    row_axis, col_axis = (0, 1)\n",
        "  else:\n",
        "    row_axis, col_axis = (1, 2)\n",
        "  rows = input_shape[row_axis]\n",
        "  cols = input_shape[col_axis]\n",
        "\n",
        "  img_input = layers.Input(shape=input_shape)\n",
        "\n",
        "  channel_axis = 1 if backend.image_data_format() == 'channels_first' else -1\n",
        "\n",
        "  first_block_filters = _make_divisible(32 * alpha, 8)\n",
        "  x = layers.ZeroPadding2D(\n",
        "      padding=imagenet_utils.correct_pad(img_input, 3),\n",
        "      name='Conv1_pad')(img_input)\n",
        "  x = layers.Conv2D(\n",
        "      first_block_filters,\n",
        "      kernel_size=3,\n",
        "      strides=(2, 2),\n",
        "      padding='valid',\n",
        "      use_bias=False,\n",
        "      kernel_regularizer=regularizers.l2(l2_reg),\n",
        "      name='Conv1')(\n",
        "          x)\n",
        "  x = layers.BatchNormalization(\n",
        "      axis=channel_axis, epsilon=1e-3, momentum=0.999, name='bn_Conv1')(\n",
        "          x)\n",
        "  x = layers.ReLU(6., name='Conv1_relu')(x)\n",
        "\n",
        "  x = _inverted_res_block(\n",
        "      x, filters=16, alpha=alpha, stride=1, expansion=1, block_id=0)\n",
        "\n",
        "  x = _inverted_res_block(\n",
        "      x, filters=24, alpha=alpha, stride=2, expansion=6, block_id=1)\n",
        "  x = _inverted_res_block(\n",
        "      x, filters=24, alpha=alpha, stride=1, expansion=6, block_id=2)\n",
        "\n",
        "  x = _inverted_res_block(\n",
        "      x, filters=32, alpha=alpha, stride=2, expansion=6, block_id=3)\n",
        "  x = _inverted_res_block(\n",
        "      x, filters=32, alpha=alpha, stride=1, expansion=6, block_id=4)\n",
        "  x = _inverted_res_block(\n",
        "      x, filters=32, alpha=alpha, stride=1, expansion=6, block_id=5)\n",
        "\n",
        "  x = _inverted_res_block(\n",
        "      x, filters=64, alpha=alpha, stride=2, expansion=6, block_id=6)\n",
        "  x = _inverted_res_block(\n",
        "      x, filters=64, alpha=alpha, stride=1, expansion=6, block_id=7)\n",
        "  x = _inverted_res_block(\n",
        "      x, filters=64, alpha=alpha, stride=1, expansion=6, block_id=8)\n",
        "  x = _inverted_res_block(\n",
        "      x, filters=64, alpha=alpha, stride=1, expansion=6, block_id=9)\n",
        "\n",
        "  x = _inverted_res_block(\n",
        "      x, filters=96, alpha=alpha, stride=1, expansion=6, block_id=10)\n",
        "  x = _inverted_res_block(\n",
        "      x, filters=96, alpha=alpha, stride=1, expansion=6, block_id=11)\n",
        "  x = _inverted_res_block(\n",
        "      x, filters=96, alpha=alpha, stride=1, expansion=6, block_id=12)\n",
        "\n",
        "  x = _inverted_res_block(\n",
        "      x, filters=160, alpha=alpha, stride=2, expansion=6, block_id=13)\n",
        "  x = _inverted_res_block(\n",
        "      x, filters=160, alpha=alpha, stride=1, expansion=6, block_id=14)\n",
        "  x = _inverted_res_block(\n",
        "      x, filters=160, alpha=alpha, stride=1, expansion=6, block_id=15)\n",
        "\n",
        "  x = _inverted_res_block(\n",
        "      x, filters=320, alpha=alpha, stride=1, expansion=6, block_id=16)\n",
        "\n",
        "  # no alpha applied to last conv as stated in the paper:\n",
        "  # if the width multiplier is greater than 1 we\n",
        "  # increase the number of output channels\n",
        "  if alpha > 1.0:\n",
        "    last_block_filters = _make_divisible(1280 * alpha, 8)\n",
        "  else:\n",
        "    last_block_filters = 1280\n",
        "\n",
        "  x = layers.Conv2D(\n",
        "      last_block_filters, kernel_size=1, use_bias=False, name='Conv_1',\n",
        "      kernel_regularizer=regularizers.l2(l2_reg))(\n",
        "          x)\n",
        "  x = layers.BatchNormalization(\n",
        "      axis=channel_axis, epsilon=1e-3, momentum=0.999, name='Conv_1_bn')(\n",
        "          x)\n",
        "  x = layers.ReLU(6., name='out_relu')(x)\n",
        "\n",
        "  if pooling == 'avg':\n",
        "    x = layers.GlobalAveragePooling2D()(x)\n",
        "  elif pooling == 'max':\n",
        "    x = layers.GlobalMaxPooling2D()(x)\n",
        "\n",
        "  inputs = img_input\n",
        "\n",
        "  # Create model.\n",
        "  model = training.Model(inputs, x, name='mobilenetv2_%0.2f_%s' % (alpha, rows))\n",
        "\n",
        "  # Load weights.\n",
        "  if weights == 'imagenet':\n",
        "    if include_top:\n",
        "      model_name = ('mobilenet_v2_weights_tf_dim_ordering_tf_kernels_' +\n",
        "                    str(alpha) + '_' + str(rows) + '.h5')\n",
        "      weight_path = BASE_WEIGHT_PATH + model_name\n",
        "      weights_path = data_utils.get_file(\n",
        "          model_name, weight_path, cache_subdir='models')\n",
        "    else:\n",
        "      model_name = ('mobilenet_v2_weights_tf_dim_ordering_tf_kernels_' +\n",
        "                    str(alpha) + '_' + str(rows) + '_no_top' + '.h5')\n",
        "      weight_path = BASE_WEIGHT_PATH + model_name\n",
        "      weights_path = data_utils.get_file(\n",
        "          model_name, weight_path, cache_subdir='models')\n",
        "    model.load_weights(weights_path)\n",
        "  elif weights is not None:\n",
        "    model.load_weights(weights)\n",
        "\n",
        "  return model\n",
        "\n",
        "\n",
        "def _inverted_res_block(inputs, expansion, stride, alpha, filters, block_id):\n",
        "  \"\"\"Inverted ResNet block.\"\"\"\n",
        "  channel_axis = 1 if backend.image_data_format() == 'channels_first' else -1\n",
        "\n",
        "  in_channels = backend.int_shape(inputs)[channel_axis]\n",
        "  pointwise_conv_filters = int(filters * alpha)\n",
        "  pointwise_filters = _make_divisible(pointwise_conv_filters, 8)\n",
        "  x = inputs\n",
        "  prefix = 'block_{}_'.format(block_id)\n",
        "\n",
        "  if block_id:\n",
        "    # Expand\n",
        "    x = layers.Conv2D(\n",
        "        expansion * in_channels,\n",
        "        kernel_size=1,\n",
        "        padding='same',\n",
        "        use_bias=False,\n",
        "        activation=None,\n",
        "        kernel_regularizer=regularizers.l2(l2_reg),\n",
        "        name=prefix + 'expand')(\n",
        "            x)\n",
        "    x = layers.BatchNormalization(\n",
        "        axis=channel_axis,\n",
        "        epsilon=1e-3,\n",
        "        momentum=0.999,\n",
        "        name=prefix + 'expand_BN')(\n",
        "            x)\n",
        "    x = layers.ReLU(6., name=prefix + 'expand_relu')(x)\n",
        "  else:\n",
        "    prefix = 'expanded_conv_'\n",
        "\n",
        "  # Depthwise\n",
        "  if stride == 2:\n",
        "    x = layers.ZeroPadding2D(\n",
        "        padding=imagenet_utils.correct_pad(x, 3),\n",
        "        name=prefix + 'pad')(x)\n",
        "  x = layers.DepthwiseConv2D(\n",
        "      kernel_size=3,\n",
        "      strides=stride,\n",
        "      activation=None,\n",
        "      use_bias=False,\n",
        "      padding='same' if stride == 1 else 'valid',\n",
        "      depthwise_regularizer=regularizers.l2(l2_reg),\n",
        "      name=prefix + 'depthwise')(\n",
        "          x)\n",
        "  x = layers.BatchNormalization(\n",
        "      axis=channel_axis,\n",
        "      epsilon=1e-3,\n",
        "      momentum=0.999,\n",
        "      name=prefix + 'depthwise_BN')(\n",
        "          x)\n",
        "\n",
        "  x = layers.ReLU(6., name=prefix + 'depthwise_relu')(x)\n",
        "\n",
        "  # Project\n",
        "  x = layers.Conv2D(\n",
        "      pointwise_filters,\n",
        "      kernel_size=1,\n",
        "      padding='same',\n",
        "      use_bias=False,\n",
        "      activation=None,\n",
        "      kernel_regularizer=regularizers.l2(l2_reg),\n",
        "      name=prefix + 'project')(\n",
        "          x)\n",
        "  x = layers.BatchNormalization(\n",
        "      axis=channel_axis,\n",
        "      epsilon=1e-3,\n",
        "      momentum=0.999,\n",
        "      name=prefix + 'project_BN')(\n",
        "          x)\n",
        "\n",
        "  if in_channels == pointwise_filters and stride == 1:\n",
        "    return layers.Add(name=prefix + 'add')([inputs, x])\n",
        "  return x\n",
        "\n",
        "def _make_divisible(v, divisor, min_value=None):\n",
        "  if min_value is None:\n",
        "    min_value = divisor\n",
        "  new_v = max(min_value, int(v + divisor / 2) // divisor * divisor)\n",
        "  # Make sure that round down does not go down by more than 10%.\n",
        "  if new_v < 0.9 * v:\n",
        "    new_v += divisor\n",
        "  return new_v\n",
        "\n",
        "def create_model(name, img_dim, up_ratio, num_classes):\n",
        "  img = keras.Input(shape=(img_dim, img_dim, 3), dtype='float32', name='image')\n",
        "  upsample = keras.layers.UpSampling2D(up_ratio, name='upsample')\n",
        "  base_model = MobileNetV2(\n",
        "      input_shape=(img_dim * up_ratio, img_dim * up_ratio, 3),\n",
        "      include_top=False,\n",
        "      pooling='avg'\n",
        "  )\n",
        "  base_model.summary()\n",
        "  base_model.trainable = False\n",
        "  prediction = keras.layers.Dense(num_classes, name='prediction')\n",
        "  out = prediction(base_model(upsample(img)))\n",
        "  return keras.Model(inputs=img, outputs=out, name=name)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jNwLPeLcC9X3",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def main(params):\n",
        "  tf.random.set_seed(0) # is not deterministic though\n",
        "  classes, train_data_adv, val_data_adv, steps_per_epoch, validation_steps = \\\n",
        "      prepare_data(params, 'Tiny_Imagenet_200/data/tiny-imagenet-200/')\n",
        "  adv_config = nsl.configs.make_adv_reg_config(\n",
        "      multiplier=params['adv_multiplier'],\n",
        "      adv_step_size=params['adv_step_size'],\n",
        "      adv_grad_norm=params['adv_grad_norm']\n",
        "  )\n",
        "  base_model = create_model('base_model', params['img_dim'], 2, len(classes))\n",
        "  base_model.summary()\n",
        "  adv_model = nsl.keras.AdversarialRegularization(\n",
        "      base_model, label_keys=['label'], adv_config=adv_config)\n",
        "  adv_model.compile(optimizer=keras.optimizers.SGD(learning_rate=params['lr'], momentum=0.9),\n",
        "                    loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),\n",
        "                    metrics=['sparse_categorical_crossentropy', 'sparse_categorical_accuracy',\n",
        "                             'sparse_top_k_categorical_accuracy'])\n",
        "  \n",
        "  ckpt_callback = keras.callbacks.ModelCheckpoint(PATH + 'checkpoint/ckpt_{epoch:03d}')\n",
        "  tsb_callback = keras.callbacks.TensorBoard(log_dir=PATH+'logs')\n",
        "  stop_callback = keras.callbacks.EarlyStopping(\n",
        "      monitor='loss', min_delta=0.001, patience=2, mode='min')\n",
        "  ft_history = adv_model.fit(\n",
        "      train_data_adv,\n",
        "      epochs=params['init_epochs'],\n",
        "      steps_per_epoch=steps_per_epoch,\n",
        "      validation_data=val_data_adv,\n",
        "      validation_steps=validation_steps,\n",
        "      callbacks=[ckpt_callback, tsb_callback, stop_callback],\n",
        "      verbose=2\n",
        "  )\n",
        "  print(ft_history.history)\n",
        "  epochs_trained = len(ft_history.history['loss'])\n",
        "\n",
        "  v2_model = base_model.get_layer(name='mobilenetv2_1.00_128')\n",
        "  v2_model.trainable = True\n",
        "  finetune_layer_0 = -12\n",
        "  print('\\n\\nStart finetuning at {}'.format(v2_model.layers[finetune_layer_0].name))\n",
        "  for layer in v2_model.layers[:finetune_layer_0]:\n",
        "    layer.trainable = False\n",
        "  adv_model.compile(optimizer=keras.optimizers.SGD(learning_rate=params['lr']/10, momentum=0.9),\n",
        "                    loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),\n",
        "                    metrics=['sparse_categorical_crossentropy', 'sparse_categorical_accuracy',\n",
        "                             'sparse_top_k_categorical_accuracy'])\n",
        "  base_model.summary()\n",
        "  ft0_history = adv_model.fit(\n",
        "      train_data_adv,\n",
        "      initial_epoch=epochs_trained,\n",
        "      epochs=epochs_trained+params['ft0_epochs'],\n",
        "      steps_per_epoch=steps_per_epoch,\n",
        "      validation_data=val_data_adv,\n",
        "      validation_steps=validation_steps,\n",
        "      callbacks=[ckpt_callback, tsb_callback, stop_callback],\n",
        "      verbose=2\n",
        "  )\n",
        "  print(ft0_history.history)\n",
        "  epochs_trained += len(ft0_history.history['loss'])\n",
        "\n",
        "  finetune_layer_1 = -30\n",
        "  print('\\n\\nStart finetuning at {}'.format(v2_model.layers[finetune_layer_1].name))\n",
        "  for layer in v2_model.layers[finetune_layer_1:finetune_layer_0]:\n",
        "    layer.trainable = True\n",
        "  adv_model.compile(optimizer=keras.optimizers.SGD(learning_rate=params['lr']/100, momentum=0.9),\n",
        "                    loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),\n",
        "                    metrics=['sparse_categorical_crossentropy', 'sparse_categorical_accuracy',\n",
        "                             'sparse_top_k_categorical_accuracy'])\n",
        "  base_model.summary()\n",
        "  ft1_history = adv_model.fit(\n",
        "      train_data_adv,\n",
        "      initial_epoch=epochs_trained,\n",
        "      epochs=epochs_trained+params['ft1_epochs'],\n",
        "      steps_per_epoch=steps_per_epoch,\n",
        "      validation_data=val_data_adv,\n",
        "      validation_steps=validation_steps,\n",
        "      callbacks=[ckpt_callback, tsb_callback],\n",
        "      verbose=2\n",
        "  )\n",
        "  print(ft1_history.history)\n",
        "  epochs_trained += len(ft1_history.history['loss'])\n",
        "\n",
        "  # print('\\n\\nFinetune all')\n",
        "  # for layer in v2_model.layers[:finetune_layer_1]:\n",
        "  #   layer.trainable = True\n",
        "  # adv_model.compile(optimizer=keras.optimizers.SGD(learning_rate=params['lr']/1000, momentum=0.9),\n",
        "  #                   loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),\n",
        "  #                   metrics=['sparse_categorical_crossentropy', 'sparse_categorical_accuracy',\n",
        "  #                            'sparse_top_k_categorical_accuracy'])\n",
        "  # base_model.summary()\n",
        "  # ft2_history = adv_model.fit(\n",
        "  #     train_data_adv,\n",
        "  #     initial_epoch=epochs_trained,\n",
        "  #     epochs=epochs_trained+params['ft2_epochs'],\n",
        "  #     steps_per_epoch=steps_per_epoch,\n",
        "  #     # validation_data=val_data_adv,\n",
        "  #     # validation_steps=validation_steps,\n",
        "  #     callbacks=[ckpt_callback, tsb_callback, stop_callback],\n",
        "  #     verbose=2\n",
        "  # )\n",
        "  # print(ft2_history.history)\n",
        "\n",
        "  return adv_model"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FccmzcCMTHQr",
        "colab_type": "code",
        "outputId": "4306ece6-7faf-4caa-b56c-81b1616e8579",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "if __name__ == '__main__':\n",
        "  params = {\n",
        "      'img_dim': 64,\n",
        "      'lr': 0.01,\n",
        "      'batch_size': 128,\n",
        "      'val_ratio': 10,\n",
        "      'adv_multiplier': 0.2,\n",
        "      'adv_step_size': 0.2,\n",
        "      'adv_grad_norm': 'infinity',\n",
        "      # max number of epochs to train for each stage of finetuning\n",
        "      'init_epochs': 100, # 108s per epoch, P100 batch_size 128\n",
        "      'ft0_epochs':  100, # 111s per epoch, P100 batch_size 128\n",
        "      'ft1_epochs':  100, # 115s per epoch, P100 batch_size 128\n",
        "      # 'ft2_epochs':  100,\n",
        "  }\n",
        "  main(params)\n",
        "\n",
        "# checkpoint_v2\n",
        "# Epoch  78/144, val_loss: 3.1419 - val_sparse_categorical_accuracy: 0.4814 - val_sparse_top_k_categorical_accuracy: 0.7217\n",
        "# Epoch  90/144, val_loss: 3.1792 - val_sparse_categorical_accuracy: 0.4814 - val_sparse_top_k_categorical_accuracy: 0.7168\n",
        "\n",
        "# checkpoint_v2reg\n",
        "# Epoch  91/154, val_loss: 7.5040 - val_sparse_categorical_accuracy: 0.5000 - val_sparse_top_k_categorical_accuracy: 0.7412\n",
        "# Epoch  92/154, val_loss: 7.5210 - val_sparse_categorical_accuracy: 0.5010 - val_sparse_top_k_categorical_accuracy: 0.7285\n",
        "# Epoch 113/154, val_loss: 7.5302 - val_sparse_categorical_accuracy: 0.4814 - val_sparse_top_k_categorical_accuracy: 0.7461\n",
        "# Epoch 154/154, val_loss: 7.4702 - val_sparse_categorical_accuracy: 0.4941 - val_sparse_top_k_categorical_accuracy: 0.7432\n"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Discovered 100000 images for training data.\n",
            "Discovered 10000 images for validation data.\n",
            "Model: \"mobilenetv2_1.00_128\"\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_1 (InputLayer)            [(None, 128, 128, 3) 0                                            \n",
            "__________________________________________________________________________________________________\n",
            "Conv1_pad (ZeroPadding2D)       (None, 129, 129, 3)  0           input_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "Conv1 (Conv2D)                  (None, 64, 64, 32)   864         Conv1_pad[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "bn_Conv1 (BatchNormalization)   (None, 64, 64, 32)   128         Conv1[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "Conv1_relu (ReLU)               (None, 64, 64, 32)   0           bn_Conv1[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "expanded_conv_depthwise (Depthw (None, 64, 64, 32)   288         Conv1_relu[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "expanded_conv_depthwise_BN (Bat (None, 64, 64, 32)   128         expanded_conv_depthwise[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "expanded_conv_depthwise_relu (R (None, 64, 64, 32)   0           expanded_conv_depthwise_BN[0][0] \n",
            "__________________________________________________________________________________________________\n",
            "expanded_conv_project (Conv2D)  (None, 64, 64, 16)   512         expanded_conv_depthwise_relu[0][0\n",
            "__________________________________________________________________________________________________\n",
            "expanded_conv_project_BN (Batch (None, 64, 64, 16)   64          expanded_conv_project[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "block_1_expand (Conv2D)         (None, 64, 64, 96)   1536        expanded_conv_project_BN[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "block_1_expand_BN (BatchNormali (None, 64, 64, 96)   384         block_1_expand[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "block_1_expand_relu (ReLU)      (None, 64, 64, 96)   0           block_1_expand_BN[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_1_pad (ZeroPadding2D)     (None, 65, 65, 96)   0           block_1_expand_relu[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_1_depthwise (DepthwiseCon (None, 32, 32, 96)   864         block_1_pad[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "block_1_depthwise_BN (BatchNorm (None, 32, 32, 96)   384         block_1_depthwise[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_1_depthwise_relu (ReLU)   (None, 32, 32, 96)   0           block_1_depthwise_BN[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_1_project (Conv2D)        (None, 32, 32, 24)   2304        block_1_depthwise_relu[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "block_1_project_BN (BatchNormal (None, 32, 32, 24)   96          block_1_project[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_2_expand (Conv2D)         (None, 32, 32, 144)  3456        block_1_project_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_2_expand_BN (BatchNormali (None, 32, 32, 144)  576         block_2_expand[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "block_2_expand_relu (ReLU)      (None, 32, 32, 144)  0           block_2_expand_BN[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_2_depthwise (DepthwiseCon (None, 32, 32, 144)  1296        block_2_expand_relu[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_2_depthwise_BN (BatchNorm (None, 32, 32, 144)  576         block_2_depthwise[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_2_depthwise_relu (ReLU)   (None, 32, 32, 144)  0           block_2_depthwise_BN[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_2_project (Conv2D)        (None, 32, 32, 24)   3456        block_2_depthwise_relu[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "block_2_project_BN (BatchNormal (None, 32, 32, 24)   96          block_2_project[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_2_add (Add)               (None, 32, 32, 24)   0           block_1_project_BN[0][0]         \n",
            "                                                                 block_2_project_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_3_expand (Conv2D)         (None, 32, 32, 144)  3456        block_2_add[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "block_3_expand_BN (BatchNormali (None, 32, 32, 144)  576         block_3_expand[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "block_3_expand_relu (ReLU)      (None, 32, 32, 144)  0           block_3_expand_BN[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_3_pad (ZeroPadding2D)     (None, 33, 33, 144)  0           block_3_expand_relu[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_3_depthwise (DepthwiseCon (None, 16, 16, 144)  1296        block_3_pad[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "block_3_depthwise_BN (BatchNorm (None, 16, 16, 144)  576         block_3_depthwise[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_3_depthwise_relu (ReLU)   (None, 16, 16, 144)  0           block_3_depthwise_BN[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_3_project (Conv2D)        (None, 16, 16, 32)   4608        block_3_depthwise_relu[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "block_3_project_BN (BatchNormal (None, 16, 16, 32)   128         block_3_project[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_4_expand (Conv2D)         (None, 16, 16, 192)  6144        block_3_project_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_4_expand_BN (BatchNormali (None, 16, 16, 192)  768         block_4_expand[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "block_4_expand_relu (ReLU)      (None, 16, 16, 192)  0           block_4_expand_BN[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_4_depthwise (DepthwiseCon (None, 16, 16, 192)  1728        block_4_expand_relu[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_4_depthwise_BN (BatchNorm (None, 16, 16, 192)  768         block_4_depthwise[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_4_depthwise_relu (ReLU)   (None, 16, 16, 192)  0           block_4_depthwise_BN[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_4_project (Conv2D)        (None, 16, 16, 32)   6144        block_4_depthwise_relu[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "block_4_project_BN (BatchNormal (None, 16, 16, 32)   128         block_4_project[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_4_add (Add)               (None, 16, 16, 32)   0           block_3_project_BN[0][0]         \n",
            "                                                                 block_4_project_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_5_expand (Conv2D)         (None, 16, 16, 192)  6144        block_4_add[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "block_5_expand_BN (BatchNormali (None, 16, 16, 192)  768         block_5_expand[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "block_5_expand_relu (ReLU)      (None, 16, 16, 192)  0           block_5_expand_BN[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_5_depthwise (DepthwiseCon (None, 16, 16, 192)  1728        block_5_expand_relu[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_5_depthwise_BN (BatchNorm (None, 16, 16, 192)  768         block_5_depthwise[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_5_depthwise_relu (ReLU)   (None, 16, 16, 192)  0           block_5_depthwise_BN[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_5_project (Conv2D)        (None, 16, 16, 32)   6144        block_5_depthwise_relu[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "block_5_project_BN (BatchNormal (None, 16, 16, 32)   128         block_5_project[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_5_add (Add)               (None, 16, 16, 32)   0           block_4_add[0][0]                \n",
            "                                                                 block_5_project_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_6_expand (Conv2D)         (None, 16, 16, 192)  6144        block_5_add[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "block_6_expand_BN (BatchNormali (None, 16, 16, 192)  768         block_6_expand[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "block_6_expand_relu (ReLU)      (None, 16, 16, 192)  0           block_6_expand_BN[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_6_pad (ZeroPadding2D)     (None, 17, 17, 192)  0           block_6_expand_relu[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_6_depthwise (DepthwiseCon (None, 8, 8, 192)    1728        block_6_pad[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "block_6_depthwise_BN (BatchNorm (None, 8, 8, 192)    768         block_6_depthwise[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_6_depthwise_relu (ReLU)   (None, 8, 8, 192)    0           block_6_depthwise_BN[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_6_project (Conv2D)        (None, 8, 8, 64)     12288       block_6_depthwise_relu[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "block_6_project_BN (BatchNormal (None, 8, 8, 64)     256         block_6_project[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_7_expand (Conv2D)         (None, 8, 8, 384)    24576       block_6_project_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_7_expand_BN (BatchNormali (None, 8, 8, 384)    1536        block_7_expand[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "block_7_expand_relu (ReLU)      (None, 8, 8, 384)    0           block_7_expand_BN[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_7_depthwise (DepthwiseCon (None, 8, 8, 384)    3456        block_7_expand_relu[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_7_depthwise_BN (BatchNorm (None, 8, 8, 384)    1536        block_7_depthwise[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_7_depthwise_relu (ReLU)   (None, 8, 8, 384)    0           block_7_depthwise_BN[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_7_project (Conv2D)        (None, 8, 8, 64)     24576       block_7_depthwise_relu[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "block_7_project_BN (BatchNormal (None, 8, 8, 64)     256         block_7_project[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_7_add (Add)               (None, 8, 8, 64)     0           block_6_project_BN[0][0]         \n",
            "                                                                 block_7_project_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_8_expand (Conv2D)         (None, 8, 8, 384)    24576       block_7_add[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "block_8_expand_BN (BatchNormali (None, 8, 8, 384)    1536        block_8_expand[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "block_8_expand_relu (ReLU)      (None, 8, 8, 384)    0           block_8_expand_BN[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_8_depthwise (DepthwiseCon (None, 8, 8, 384)    3456        block_8_expand_relu[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_8_depthwise_BN (BatchNorm (None, 8, 8, 384)    1536        block_8_depthwise[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_8_depthwise_relu (ReLU)   (None, 8, 8, 384)    0           block_8_depthwise_BN[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_8_project (Conv2D)        (None, 8, 8, 64)     24576       block_8_depthwise_relu[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "block_8_project_BN (BatchNormal (None, 8, 8, 64)     256         block_8_project[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_8_add (Add)               (None, 8, 8, 64)     0           block_7_add[0][0]                \n",
            "                                                                 block_8_project_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_9_expand (Conv2D)         (None, 8, 8, 384)    24576       block_8_add[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "block_9_expand_BN (BatchNormali (None, 8, 8, 384)    1536        block_9_expand[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "block_9_expand_relu (ReLU)      (None, 8, 8, 384)    0           block_9_expand_BN[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_9_depthwise (DepthwiseCon (None, 8, 8, 384)    3456        block_9_expand_relu[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_9_depthwise_BN (BatchNorm (None, 8, 8, 384)    1536        block_9_depthwise[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_9_depthwise_relu (ReLU)   (None, 8, 8, 384)    0           block_9_depthwise_BN[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_9_project (Conv2D)        (None, 8, 8, 64)     24576       block_9_depthwise_relu[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "block_9_project_BN (BatchNormal (None, 8, 8, 64)     256         block_9_project[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_9_add (Add)               (None, 8, 8, 64)     0           block_8_add[0][0]                \n",
            "                                                                 block_9_project_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_10_expand (Conv2D)        (None, 8, 8, 384)    24576       block_9_add[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "block_10_expand_BN (BatchNormal (None, 8, 8, 384)    1536        block_10_expand[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_10_expand_relu (ReLU)     (None, 8, 8, 384)    0           block_10_expand_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_10_depthwise (DepthwiseCo (None, 8, 8, 384)    3456        block_10_expand_relu[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_10_depthwise_BN (BatchNor (None, 8, 8, 384)    1536        block_10_depthwise[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_10_depthwise_relu (ReLU)  (None, 8, 8, 384)    0           block_10_depthwise_BN[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "block_10_project (Conv2D)       (None, 8, 8, 96)     36864       block_10_depthwise_relu[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "block_10_project_BN (BatchNorma (None, 8, 8, 96)     384         block_10_project[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "block_11_expand (Conv2D)        (None, 8, 8, 576)    55296       block_10_project_BN[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_11_expand_BN (BatchNormal (None, 8, 8, 576)    2304        block_11_expand[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_11_expand_relu (ReLU)     (None, 8, 8, 576)    0           block_11_expand_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_11_depthwise (DepthwiseCo (None, 8, 8, 576)    5184        block_11_expand_relu[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_11_depthwise_BN (BatchNor (None, 8, 8, 576)    2304        block_11_depthwise[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_11_depthwise_relu (ReLU)  (None, 8, 8, 576)    0           block_11_depthwise_BN[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "block_11_project (Conv2D)       (None, 8, 8, 96)     55296       block_11_depthwise_relu[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "block_11_project_BN (BatchNorma (None, 8, 8, 96)     384         block_11_project[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "block_11_add (Add)              (None, 8, 8, 96)     0           block_10_project_BN[0][0]        \n",
            "                                                                 block_11_project_BN[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_12_expand (Conv2D)        (None, 8, 8, 576)    55296       block_11_add[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "block_12_expand_BN (BatchNormal (None, 8, 8, 576)    2304        block_12_expand[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_12_expand_relu (ReLU)     (None, 8, 8, 576)    0           block_12_expand_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_12_depthwise (DepthwiseCo (None, 8, 8, 576)    5184        block_12_expand_relu[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_12_depthwise_BN (BatchNor (None, 8, 8, 576)    2304        block_12_depthwise[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_12_depthwise_relu (ReLU)  (None, 8, 8, 576)    0           block_12_depthwise_BN[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "block_12_project (Conv2D)       (None, 8, 8, 96)     55296       block_12_depthwise_relu[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "block_12_project_BN (BatchNorma (None, 8, 8, 96)     384         block_12_project[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "block_12_add (Add)              (None, 8, 8, 96)     0           block_11_add[0][0]               \n",
            "                                                                 block_12_project_BN[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_13_expand (Conv2D)        (None, 8, 8, 576)    55296       block_12_add[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "block_13_expand_BN (BatchNormal (None, 8, 8, 576)    2304        block_13_expand[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_13_expand_relu (ReLU)     (None, 8, 8, 576)    0           block_13_expand_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_13_pad (ZeroPadding2D)    (None, 9, 9, 576)    0           block_13_expand_relu[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_13_depthwise (DepthwiseCo (None, 4, 4, 576)    5184        block_13_pad[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "block_13_depthwise_BN (BatchNor (None, 4, 4, 576)    2304        block_13_depthwise[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_13_depthwise_relu (ReLU)  (None, 4, 4, 576)    0           block_13_depthwise_BN[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "block_13_project (Conv2D)       (None, 4, 4, 160)    92160       block_13_depthwise_relu[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "block_13_project_BN (BatchNorma (None, 4, 4, 160)    640         block_13_project[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "block_14_expand (Conv2D)        (None, 4, 4, 960)    153600      block_13_project_BN[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_14_expand_BN (BatchNormal (None, 4, 4, 960)    3840        block_14_expand[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_14_expand_relu (ReLU)     (None, 4, 4, 960)    0           block_14_expand_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_14_depthwise (DepthwiseCo (None, 4, 4, 960)    8640        block_14_expand_relu[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_14_depthwise_BN (BatchNor (None, 4, 4, 960)    3840        block_14_depthwise[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_14_depthwise_relu (ReLU)  (None, 4, 4, 960)    0           block_14_depthwise_BN[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "block_14_project (Conv2D)       (None, 4, 4, 160)    153600      block_14_depthwise_relu[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "block_14_project_BN (BatchNorma (None, 4, 4, 160)    640         block_14_project[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "block_14_add (Add)              (None, 4, 4, 160)    0           block_13_project_BN[0][0]        \n",
            "                                                                 block_14_project_BN[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_15_expand (Conv2D)        (None, 4, 4, 960)    153600      block_14_add[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "block_15_expand_BN (BatchNormal (None, 4, 4, 960)    3840        block_15_expand[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_15_expand_relu (ReLU)     (None, 4, 4, 960)    0           block_15_expand_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_15_depthwise (DepthwiseCo (None, 4, 4, 960)    8640        block_15_expand_relu[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_15_depthwise_BN (BatchNor (None, 4, 4, 960)    3840        block_15_depthwise[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_15_depthwise_relu (ReLU)  (None, 4, 4, 960)    0           block_15_depthwise_BN[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "block_15_project (Conv2D)       (None, 4, 4, 160)    153600      block_15_depthwise_relu[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "block_15_project_BN (BatchNorma (None, 4, 4, 160)    640         block_15_project[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "block_15_add (Add)              (None, 4, 4, 160)    0           block_14_add[0][0]               \n",
            "                                                                 block_15_project_BN[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_16_expand (Conv2D)        (None, 4, 4, 960)    153600      block_15_add[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "block_16_expand_BN (BatchNormal (None, 4, 4, 960)    3840        block_16_expand[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_16_expand_relu (ReLU)     (None, 4, 4, 960)    0           block_16_expand_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_16_depthwise (DepthwiseCo (None, 4, 4, 960)    8640        block_16_expand_relu[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_16_depthwise_BN (BatchNor (None, 4, 4, 960)    3840        block_16_depthwise[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_16_depthwise_relu (ReLU)  (None, 4, 4, 960)    0           block_16_depthwise_BN[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "block_16_project (Conv2D)       (None, 4, 4, 320)    307200      block_16_depthwise_relu[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "block_16_project_BN (BatchNorma (None, 4, 4, 320)    1280        block_16_project[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "Conv_1 (Conv2D)                 (None, 4, 4, 1280)   409600      block_16_project_BN[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "Conv_1_bn (BatchNormalization)  (None, 4, 4, 1280)   5120        Conv_1[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "out_relu (ReLU)                 (None, 4, 4, 1280)   0           Conv_1_bn[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling2d (Globa (None, 1280)         0           out_relu[0][0]                   \n",
            "==================================================================================================\n",
            "Total params: 2,257,984\n",
            "Trainable params: 2,223,872\n",
            "Non-trainable params: 34,112\n",
            "__________________________________________________________________________________________________\n",
            "Model: \"base_model\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "image (InputLayer)           [(None, 64, 64, 3)]       0         \n",
            "_________________________________________________________________\n",
            "upsample (UpSampling2D)      (None, 128, 128, 3)       0         \n",
            "_________________________________________________________________\n",
            "mobilenetv2_1.00_128 (Model) (None, 1280)              2257984   \n",
            "_________________________________________________________________\n",
            "prediction (Dense)           (None, 200)               256200    \n",
            "=================================================================\n",
            "Total params: 2,514,184\n",
            "Trainable params: 256,200\n",
            "Non-trainable params: 2,257,984\n",
            "_________________________________________________________________\n",
            "Epoch 1/100\n",
            "782/782 - 130s - loss: 9.6381 - sparse_categorical_crossentropy: 3.9514 - sparse_categorical_crossentropy_2: 6.2041 - sparse_categorical_accuracy: 0.1868 - sparse_top_k_categorical_accuracy: 0.3775 - adversarial_loss: 5.9568 - val_loss: 8.6240 - val_sparse_categorical_crossentropy: 3.0480 - val_sparse_categorical_crossentropy_2: 4.9231 - val_sparse_categorical_accuracy: 0.3154 - val_sparse_top_k_categorical_accuracy: 0.5830 - val_adversarial_loss: 5.4032\n",
            "Epoch 2/100\n",
            "782/782 - 107s - loss: 9.0065 - sparse_categorical_crossentropy: 3.3533 - sparse_categorical_crossentropy_2: 5.2706 - sparse_categorical_accuracy: 0.2735 - sparse_top_k_categorical_accuracy: 0.5017 - adversarial_loss: 5.7892 - val_loss: 8.3820 - val_sparse_categorical_crossentropy: 2.8339 - val_sparse_categorical_crossentropy_2: 4.8547 - val_sparse_categorical_accuracy: 0.3799 - val_sparse_top_k_categorical_accuracy: 0.6182 - val_adversarial_loss: 5.2643\n",
            "Epoch 3/100\n",
            "782/782 - 108s - loss: 8.9172 - sparse_categorical_crossentropy: 3.2623 - sparse_categorical_crossentropy_2: 5.1795 - sparse_categorical_accuracy: 0.2877 - sparse_top_k_categorical_accuracy: 0.5183 - adversarial_loss: 5.7979 - val_loss: 8.3860 - val_sparse_categorical_crossentropy: 2.8296 - val_sparse_categorical_crossentropy_2: 4.8375 - val_sparse_categorical_accuracy: 0.3672 - val_sparse_top_k_categorical_accuracy: 0.6162 - val_adversarial_loss: 5.3056\n",
            "Epoch 4/100\n",
            "782/782 - 108s - loss: 8.9297 - sparse_categorical_crossentropy: 3.2719 - sparse_categorical_crossentropy_2: 5.1733 - sparse_categorical_accuracy: 0.2851 - sparse_top_k_categorical_accuracy: 0.5166 - adversarial_loss: 5.8123 - val_loss: 8.2680 - val_sparse_categorical_crossentropy: 2.7386 - val_sparse_categorical_crossentropy_2: 4.7876 - val_sparse_categorical_accuracy: 0.3867 - val_sparse_top_k_categorical_accuracy: 0.6455 - val_adversarial_loss: 5.1704\n",
            "Epoch 5/100\n",
            "782/782 - 108s - loss: 8.8790 - sparse_categorical_crossentropy: 3.2267 - sparse_categorical_crossentropy_2: 5.1470 - sparse_categorical_accuracy: 0.2935 - sparse_top_k_categorical_accuracy: 0.5266 - adversarial_loss: 5.7848 - val_loss: 8.2911 - val_sparse_categorical_crossentropy: 2.7606 - val_sparse_categorical_crossentropy_2: 4.8057 - val_sparse_categorical_accuracy: 0.3799 - val_sparse_top_k_categorical_accuracy: 0.6396 - val_adversarial_loss: 5.1759\n",
            "Epoch 6/100\n",
            "782/782 - 108s - loss: 8.7036 - sparse_categorical_crossentropy: 3.0679 - sparse_categorical_crossentropy_2: 5.0896 - sparse_categorical_accuracy: 0.3228 - sparse_top_k_categorical_accuracy: 0.5570 - adversarial_loss: 5.7018 - val_loss: 8.2105 - val_sparse_categorical_crossentropy: 2.6907 - val_sparse_categorical_crossentropy_2: 4.7032 - val_sparse_categorical_accuracy: 0.3955 - val_sparse_top_k_categorical_accuracy: 0.6533 - val_adversarial_loss: 5.1223\n",
            "Epoch 7/100\n",
            "782/782 - 108s - loss: 8.6868 - sparse_categorical_crossentropy: 3.0604 - sparse_categorical_crossentropy_2: 5.0420 - sparse_categorical_accuracy: 0.3227 - sparse_top_k_categorical_accuracy: 0.5594 - adversarial_loss: 5.6553 - val_loss: 8.2868 - val_sparse_categorical_crossentropy: 2.7408 - val_sparse_categorical_crossentropy_2: 4.7867 - val_sparse_categorical_accuracy: 0.3945 - val_sparse_top_k_categorical_accuracy: 0.6416 - val_adversarial_loss: 5.2533\n",
            "Epoch 8/100\n",
            "782/782 - 108s - loss: 8.6620 - sparse_categorical_crossentropy: 3.0372 - sparse_categorical_crossentropy_2: 5.0304 - sparse_categorical_accuracy: 0.3273 - sparse_top_k_categorical_accuracy: 0.5648 - adversarial_loss: 5.6473 - val_loss: 8.1636 - val_sparse_categorical_crossentropy: 2.6464 - val_sparse_categorical_crossentropy_2: 4.7422 - val_sparse_categorical_accuracy: 0.4053 - val_sparse_top_k_categorical_accuracy: 0.6484 - val_adversarial_loss: 5.1094\n",
            "Epoch 9/100\n",
            "782/782 - 107s - loss: 8.7757 - sparse_categorical_crossentropy: 3.1353 - sparse_categorical_crossentropy_2: 5.0615 - sparse_categorical_accuracy: 0.3086 - sparse_top_k_categorical_accuracy: 0.5430 - adversarial_loss: 5.7255 - val_loss: 8.2068 - val_sparse_categorical_crossentropy: 2.6834 - val_sparse_categorical_crossentropy_2: 4.6778 - val_sparse_categorical_accuracy: 0.4023 - val_sparse_top_k_categorical_accuracy: 0.6465 - val_adversarial_loss: 5.1402\n",
            "Epoch 10/100\n",
            "782/782 - 108s - loss: 8.7282 - sparse_categorical_crossentropy: 3.0945 - sparse_categorical_crossentropy_2: 5.0404 - sparse_categorical_accuracy: 0.3157 - sparse_top_k_categorical_accuracy: 0.5530 - adversarial_loss: 5.6919 - val_loss: 8.1841 - val_sparse_categorical_crossentropy: 2.6699 - val_sparse_categorical_crossentropy_2: 4.7509 - val_sparse_categorical_accuracy: 0.3984 - val_sparse_top_k_categorical_accuracy: 0.6484 - val_adversarial_loss: 5.0946\n",
            "{'loss': [9.638101577758789, 9.006455421447754, 8.917177200317383, 8.929708480834961, 8.87895679473877, 8.703607559204102, 8.686792373657227, 8.661980628967285, 8.77571964263916, 8.728175163269043], 'sparse_categorical_crossentropy': [3.9514315128326416, 3.353311061859131, 3.2622804641723633, 3.271940231323242, 3.226689338684082, 3.067929983139038, 3.0604231357574463, 3.0372161865234375, 3.135312557220459, 3.0944840908050537], 'sparse_categorical_crossentropy_2': [6.204130172729492, 5.2706122398376465, 5.179452896118164, 5.173294544219971, 5.146977424621582, 5.089603900909424, 5.042018413543701, 5.030432224273682, 5.061454772949219, 5.040362358093262], 'sparse_categorical_accuracy': [0.18679067492485046, 0.2734774649143219, 0.28771379590034485, 0.285126268863678, 0.2934882640838623, 0.3228000998497009, 0.3226502537727356, 0.3273257613182068, 0.308623731136322, 0.31573688983917236], 'sparse_top_k_categorical_accuracy': [0.3774576485157013, 0.50169837474823, 0.5182824730873108, 0.5166440010070801, 0.5265845060348511, 0.5570052862167358, 0.5593929886817932, 0.5648077726364136, 0.5429787635803223, 0.5530490875244141], 'adversarial_loss': [5.95680570602417, 5.789165019989014, 5.797914028167725, 5.812265396118164, 5.784789562225342, 5.701812744140625, 5.655319690704346, 5.647271156311035, 5.725475311279297, 5.691916465759277], 'val_loss': [8.623994827270508, 8.382033348083496, 8.386035919189453, 8.267996788024902, 8.291098594665527, 8.21048641204834, 8.286825180053711, 8.163565635681152, 8.206779479980469, 8.184096336364746], 'val_sparse_categorical_crossentropy': [3.048036575317383, 2.8338561058044434, 2.8296077251434326, 2.738600969314575, 2.7606143951416016, 2.6907174587249756, 2.7408461570739746, 2.646376371383667, 2.6834282875061035, 2.6698691844940186], 'val_sparse_categorical_crossentropy_2': [4.923077583312988, 4.8546833992004395, 4.837491512298584, 4.787577152252197, 4.8057050704956055, 4.703192710876465, 4.786705493927002, 4.742247104644775, 4.677759170532227, 4.7509002685546875], 'val_sparse_categorical_accuracy': [0.3154296875, 0.3798828125, 0.3671875, 0.38671875, 0.3798828125, 0.3955078125, 0.39453125, 0.4052734375, 0.40234375, 0.3984375], 'val_sparse_top_k_categorical_accuracy': [0.5830078125, 0.6181640625, 0.6162109375, 0.6455078125, 0.6396484375, 0.6533203125, 0.6416015625, 0.6484375, 0.646484375, 0.6484375], 'val_adversarial_loss': [5.403234004974365, 5.264327526092529, 5.305583953857422, 5.170426845550537, 5.175867080688477, 5.122290134429932, 5.253340721130371, 5.109393119812012, 5.140202045440674, 5.094577789306641]}\n",
            "\n",
            "\n",
            "Start finetuning at block_16_expand\n",
            "Model: \"base_model\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "image (InputLayer)           [(None, 64, 64, 3)]       0         \n",
            "_________________________________________________________________\n",
            "upsample (UpSampling2D)      (None, 128, 128, 3)       0         \n",
            "_________________________________________________________________\n",
            "mobilenetv2_1.00_128 (Model) (None, 1280)              2257984   \n",
            "_________________________________________________________________\n",
            "prediction (Dense)           (None, 200)               256200    \n",
            "=================================================================\n",
            "Total params: 2,514,184\n",
            "Trainable params: 1,142,280\n",
            "Non-trainable params: 1,371,904\n",
            "_________________________________________________________________\n",
            "Epoch 11/110\n",
            "WARNING:tensorflow:Method (on_train_batch_end) is slow compared to the batch update (0.408377). Check your callbacks.\n",
            "782/782 - 114s - loss: 8.5727 - sparse_categorical_crossentropy: 2.9334 - sparse_categorical_crossentropy_2: 4.9574 - sparse_categorical_accuracy: 0.3452 - sparse_top_k_categorical_accuracy: 0.5851 - adversarial_loss: 5.7216 - val_loss: 8.1379 - val_sparse_categorical_crossentropy: 2.5708 - val_sparse_categorical_crossentropy_2: 4.7175 - val_sparse_categorical_accuracy: 0.4170 - val_sparse_top_k_categorical_accuracy: 0.6758 - val_adversarial_loss: 5.3616\n",
            "Epoch 12/110\n",
            "782/782 - 111s - loss: 8.3672 - sparse_categorical_crossentropy: 2.7751 - sparse_categorical_crossentropy_2: 4.7760 - sparse_categorical_accuracy: 0.3730 - sparse_top_k_categorical_accuracy: 0.6179 - adversarial_loss: 5.4884 - val_loss: 8.0037 - val_sparse_categorical_crossentropy: 2.4793 - val_sparse_categorical_crossentropy_2: 4.7373 - val_sparse_categorical_accuracy: 0.4316 - val_sparse_top_k_categorical_accuracy: 0.6982 - val_adversarial_loss: 5.1515\n",
            "Epoch 13/110\n",
            "782/782 - 111s - loss: 8.2694 - sparse_categorical_crossentropy: 2.7090 - sparse_categorical_crossentropy_2: 4.7255 - sparse_categorical_accuracy: 0.3819 - sparse_top_k_categorical_accuracy: 0.6318 - adversarial_loss: 5.3335 - val_loss: 8.0292 - val_sparse_categorical_crossentropy: 2.5233 - val_sparse_categorical_crossentropy_2: 4.6957 - val_sparse_categorical_accuracy: 0.4307 - val_sparse_top_k_categorical_accuracy: 0.6787 - val_adversarial_loss: 5.0624\n",
            "Epoch 14/110\n",
            "782/782 - 111s - loss: 8.1743 - sparse_categorical_crossentropy: 2.6391 - sparse_categorical_crossentropy_2: 4.6909 - sparse_categorical_accuracy: 0.3964 - sparse_top_k_categorical_accuracy: 0.6454 - adversarial_loss: 5.2107 - val_loss: 7.9937 - val_sparse_categorical_crossentropy: 2.5123 - val_sparse_categorical_crossentropy_2: 4.6866 - val_sparse_categorical_accuracy: 0.4229 - val_sparse_top_k_categorical_accuracy: 0.6699 - val_adversarial_loss: 4.9430\n",
            "Epoch 15/110\n",
            "782/782 - 111s - loss: 8.1019 - sparse_categorical_crossentropy: 2.5872 - sparse_categorical_crossentropy_2: 4.6788 - sparse_categorical_accuracy: 0.4065 - sparse_top_k_categorical_accuracy: 0.6559 - adversarial_loss: 5.1119 - val_loss: 8.0342 - val_sparse_categorical_crossentropy: 2.5564 - val_sparse_categorical_crossentropy_2: 4.7153 - val_sparse_categorical_accuracy: 0.4111 - val_sparse_top_k_categorical_accuracy: 0.6738 - val_adversarial_loss: 4.9288\n",
            "Epoch 16/110\n",
            "782/782 - 111s - loss: 8.0384 - sparse_categorical_crossentropy: 2.5412 - sparse_categorical_crossentropy_2: 4.6564 - sparse_categorical_accuracy: 0.4141 - sparse_top_k_categorical_accuracy: 0.6646 - adversarial_loss: 5.0276 - val_loss: 7.9515 - val_sparse_categorical_crossentropy: 2.4951 - val_sparse_categorical_crossentropy_2: 4.7028 - val_sparse_categorical_accuracy: 0.4395 - val_sparse_top_k_categorical_accuracy: 0.6875 - val_adversarial_loss: 4.8253\n",
            "Epoch 17/110\n",
            "782/782 - 111s - loss: 7.9644 - sparse_categorical_crossentropy: 2.4835 - sparse_categorical_crossentropy_2: 4.6397 - sparse_categorical_accuracy: 0.4261 - sparse_top_k_categorical_accuracy: 0.6754 - adversarial_loss: 4.9492 - val_loss: 7.9342 - val_sparse_categorical_crossentropy: 2.4835 - val_sparse_categorical_crossentropy_2: 4.7291 - val_sparse_categorical_accuracy: 0.4287 - val_sparse_top_k_categorical_accuracy: 0.6807 - val_adversarial_loss: 4.8001\n",
            "Epoch 18/110\n",
            "782/782 - 111s - loss: 7.9139 - sparse_categorical_crossentropy: 2.4464 - sparse_categorical_crossentropy_2: 4.6205 - sparse_categorical_accuracy: 0.4330 - sparse_top_k_categorical_accuracy: 0.6817 - adversarial_loss: 4.8859 - val_loss: 8.0091 - val_sparse_categorical_crossentropy: 2.5676 - val_sparse_categorical_crossentropy_2: 4.6833 - val_sparse_categorical_accuracy: 0.4219 - val_sparse_top_k_categorical_accuracy: 0.6602 - val_adversarial_loss: 4.7578\n",
            "Epoch 19/110\n",
            "782/782 - 111s - loss: 7.8676 - sparse_categorical_crossentropy: 2.4113 - sparse_categorical_crossentropy_2: 4.6214 - sparse_categorical_accuracy: 0.4388 - sparse_top_k_categorical_accuracy: 0.6867 - adversarial_loss: 4.8334 - val_loss: 7.8711 - val_sparse_categorical_crossentropy: 2.4347 - val_sparse_categorical_crossentropy_2: 4.7511 - val_sparse_categorical_accuracy: 0.4385 - val_sparse_top_k_categorical_accuracy: 0.6875 - val_adversarial_loss: 4.7358\n",
            "Epoch 20/110\n",
            "782/782 - 111s - loss: 7.8463 - sparse_categorical_crossentropy: 2.3990 - sparse_categorical_crossentropy_2: 4.6215 - sparse_categorical_accuracy: 0.4399 - sparse_top_k_categorical_accuracy: 0.6894 - adversarial_loss: 4.7920 - val_loss: 7.9249 - val_sparse_categorical_crossentropy: 2.4928 - val_sparse_categorical_crossentropy_2: 4.6453 - val_sparse_categorical_accuracy: 0.4316 - val_sparse_top_k_categorical_accuracy: 0.6680 - val_adversarial_loss: 4.7174\n",
            "Epoch 21/110\n",
            "782/782 - 111s - loss: 7.8286 - sparse_categorical_crossentropy: 2.3890 - sparse_categorical_crossentropy_2: 4.6277 - sparse_categorical_accuracy: 0.4425 - sparse_top_k_categorical_accuracy: 0.6911 - adversarial_loss: 4.7566 - val_loss: 7.9418 - val_sparse_categorical_crossentropy: 2.5203 - val_sparse_categorical_crossentropy_2: 4.6076 - val_sparse_categorical_accuracy: 0.4180 - val_sparse_top_k_categorical_accuracy: 0.6650 - val_adversarial_loss: 4.6680\n",
            "Epoch 22/110\n",
            "782/782 - 111s - loss: 7.7854 - sparse_categorical_crossentropy: 2.3523 - sparse_categorical_crossentropy_2: 4.6064 - sparse_categorical_accuracy: 0.4487 - sparse_top_k_categorical_accuracy: 0.6973 - adversarial_loss: 4.7278 - val_loss: 7.9076 - val_sparse_categorical_crossentropy: 2.4901 - val_sparse_categorical_crossentropy_2: 4.7634 - val_sparse_categorical_accuracy: 0.4287 - val_sparse_top_k_categorical_accuracy: 0.6846 - val_adversarial_loss: 4.6515\n",
            "Epoch 23/110\n",
            "782/782 - 111s - loss: 7.7584 - sparse_categorical_crossentropy: 2.3360 - sparse_categorical_crossentropy_2: 4.6062 - sparse_categorical_accuracy: 0.4529 - sparse_top_k_categorical_accuracy: 0.7005 - adversarial_loss: 4.6775 - val_loss: 7.8089 - val_sparse_categorical_crossentropy: 2.3994 - val_sparse_categorical_crossentropy_2: 4.5926 - val_sparse_categorical_accuracy: 0.4355 - val_sparse_top_k_categorical_accuracy: 0.6875 - val_adversarial_loss: 4.6147\n",
            "Epoch 24/110\n",
            "782/782 - 111s - loss: 7.7289 - sparse_categorical_crossentropy: 2.3129 - sparse_categorical_crossentropy_2: 4.6005 - sparse_categorical_accuracy: 0.4565 - sparse_top_k_categorical_accuracy: 0.7038 - adversarial_loss: 4.6488 - val_loss: 7.8760 - val_sparse_categorical_crossentropy: 2.4707 - val_sparse_categorical_crossentropy_2: 4.6289 - val_sparse_categorical_accuracy: 0.4229 - val_sparse_top_k_categorical_accuracy: 0.6855 - val_adversarial_loss: 4.5968\n",
            "Epoch 25/110\n",
            "782/782 - 111s - loss: 7.7032 - sparse_categorical_crossentropy: 2.2902 - sparse_categorical_crossentropy_2: 4.5876 - sparse_categorical_accuracy: 0.4604 - sparse_top_k_categorical_accuracy: 0.7092 - adversarial_loss: 4.6373 - val_loss: 7.8473 - val_sparse_categorical_crossentropy: 2.4525 - val_sparse_categorical_crossentropy_2: 4.6144 - val_sparse_categorical_accuracy: 0.4424 - val_sparse_top_k_categorical_accuracy: 0.6855 - val_adversarial_loss: 4.5478\n",
            "Epoch 26/110\n",
            "782/782 - 111s - loss: 7.6822 - sparse_categorical_crossentropy: 2.2751 - sparse_categorical_crossentropy_2: 4.5925 - sparse_categorical_accuracy: 0.4629 - sparse_top_k_categorical_accuracy: 0.7124 - adversarial_loss: 4.6111 - val_loss: 7.7929 - val_sparse_categorical_crossentropy: 2.4041 - val_sparse_categorical_crossentropy_2: 4.5821 - val_sparse_categorical_accuracy: 0.4434 - val_sparse_top_k_categorical_accuracy: 0.6973 - val_adversarial_loss: 4.5211\n",
            "Epoch 27/110\n",
            "782/782 - 111s - loss: 7.6544 - sparse_categorical_crossentropy: 2.2520 - sparse_categorical_crossentropy_2: 4.5823 - sparse_categorical_accuracy: 0.4669 - sparse_top_k_categorical_accuracy: 0.7152 - adversarial_loss: 4.5908 - val_loss: 7.7480 - val_sparse_categorical_crossentropy: 2.3721 - val_sparse_categorical_crossentropy_2: 4.6416 - val_sparse_categorical_accuracy: 0.4443 - val_sparse_top_k_categorical_accuracy: 0.7148 - val_adversarial_loss: 4.4603\n",
            "Epoch 28/110\n",
            "782/782 - 111s - loss: 7.6317 - sparse_categorical_crossentropy: 2.2330 - sparse_categorical_crossentropy_2: 4.5819 - sparse_categorical_accuracy: 0.4721 - sparse_top_k_categorical_accuracy: 0.7190 - adversarial_loss: 4.5759 - val_loss: 7.8229 - val_sparse_categorical_crossentropy: 2.4331 - val_sparse_categorical_crossentropy_2: 4.6593 - val_sparse_categorical_accuracy: 0.4365 - val_sparse_top_k_categorical_accuracy: 0.6865 - val_adversarial_loss: 4.5332\n",
            "Epoch 29/110\n",
            "782/782 - 110s - loss: 7.6105 - sparse_categorical_crossentropy: 2.2170 - sparse_categorical_crossentropy_2: 4.5773 - sparse_categorical_accuracy: 0.4735 - sparse_top_k_categorical_accuracy: 0.7214 - adversarial_loss: 4.5533 - val_loss: 7.7489 - val_sparse_categorical_crossentropy: 2.3846 - val_sparse_categorical_crossentropy_2: 4.5643 - val_sparse_categorical_accuracy: 0.4375 - val_sparse_top_k_categorical_accuracy: 0.6963 - val_adversarial_loss: 4.4087\n",
            "Epoch 30/110\n",
            "782/782 - 111s - loss: 7.5853 - sparse_categorical_crossentropy: 2.1970 - sparse_categorical_crossentropy_2: 4.5704 - sparse_categorical_accuracy: 0.4806 - sparse_top_k_categorical_accuracy: 0.7259 - adversarial_loss: 4.5306 - val_loss: 7.6892 - val_sparse_categorical_crossentropy: 2.3331 - val_sparse_categorical_crossentropy_2: 4.6048 - val_sparse_categorical_accuracy: 0.4502 - val_sparse_top_k_categorical_accuracy: 0.7051 - val_adversarial_loss: 4.3707\n",
            "Epoch 31/110\n",
            "782/782 - 111s - loss: 7.5735 - sparse_categorical_crossentropy: 2.1871 - sparse_categorical_crossentropy_2: 4.5664 - sparse_categorical_accuracy: 0.4818 - sparse_top_k_categorical_accuracy: 0.7255 - adversarial_loss: 4.5242 - val_loss: 7.7737 - val_sparse_categorical_crossentropy: 2.4019 - val_sparse_categorical_crossentropy_2: 4.6228 - val_sparse_categorical_accuracy: 0.4512 - val_sparse_top_k_categorical_accuracy: 0.6963 - val_adversarial_loss: 4.4526\n",
            "Epoch 32/110\n",
            "782/782 - 112s - loss: 7.5353 - sparse_categorical_crossentropy: 2.1533 - sparse_categorical_crossentropy_2: 4.5622 - sparse_categorical_accuracy: 0.4896 - sparse_top_k_categorical_accuracy: 0.7325 - adversarial_loss: 4.5058 - val_loss: 7.7801 - val_sparse_categorical_crossentropy: 2.4093 - val_sparse_categorical_crossentropy_2: 4.6107 - val_sparse_categorical_accuracy: 0.4570 - val_sparse_top_k_categorical_accuracy: 0.7031 - val_adversarial_loss: 4.4512\n",
            "Epoch 33/110\n",
            "782/782 - 112s - loss: 7.5129 - sparse_categorical_crossentropy: 2.1340 - sparse_categorical_crossentropy_2: 4.5639 - sparse_categorical_accuracy: 0.4919 - sparse_top_k_categorical_accuracy: 0.7345 - adversarial_loss: 4.4935 - val_loss: 7.7484 - val_sparse_categorical_crossentropy: 2.3787 - val_sparse_categorical_crossentropy_2: 4.6650 - val_sparse_categorical_accuracy: 0.4473 - val_sparse_top_k_categorical_accuracy: 0.7080 - val_adversarial_loss: 4.4489\n",
            "Epoch 34/110\n",
            "782/782 - 112s - loss: 7.5110 - sparse_categorical_crossentropy: 2.1350 - sparse_categorical_crossentropy_2: 4.5668 - sparse_categorical_accuracy: 0.4906 - sparse_top_k_categorical_accuracy: 0.7343 - adversarial_loss: 4.4818 - val_loss: 7.6404 - val_sparse_categorical_crossentropy: 2.2798 - val_sparse_categorical_crossentropy_2: 4.6229 - val_sparse_categorical_accuracy: 0.4639 - val_sparse_top_k_categorical_accuracy: 0.7246 - val_adversarial_loss: 4.4064\n",
            "Epoch 35/110\n",
            "782/782 - 112s - loss: 7.4916 - sparse_categorical_crossentropy: 2.1179 - sparse_categorical_crossentropy_2: 4.5569 - sparse_categorical_accuracy: 0.4949 - sparse_top_k_categorical_accuracy: 0.7385 - adversarial_loss: 4.4736 - val_loss: 7.7724 - val_sparse_categorical_crossentropy: 2.3924 - val_sparse_categorical_crossentropy_2: 4.6619 - val_sparse_categorical_accuracy: 0.4482 - val_sparse_top_k_categorical_accuracy: 0.7119 - val_adversarial_loss: 4.5067\n",
            "Epoch 36/110\n",
            "782/782 - 111s - loss: 7.4632 - sparse_categorical_crossentropy: 2.0944 - sparse_categorical_crossentropy_2: 4.5558 - sparse_categorical_accuracy: 0.5007 - sparse_top_k_categorical_accuracy: 0.7421 - adversarial_loss: 4.4525 - val_loss: 7.7713 - val_sparse_categorical_crossentropy: 2.3943 - val_sparse_categorical_crossentropy_2: 4.6363 - val_sparse_categorical_accuracy: 0.4365 - val_sparse_top_k_categorical_accuracy: 0.6992 - val_adversarial_loss: 4.4949\n",
            "Epoch 37/110\n",
            "782/782 - 111s - loss: 7.4615 - sparse_categorical_crossentropy: 2.0955 - sparse_categorical_crossentropy_2: 4.5539 - sparse_categorical_accuracy: 0.4997 - sparse_top_k_categorical_accuracy: 0.7427 - adversarial_loss: 4.4418 - val_loss: 7.8181 - val_sparse_categorical_crossentropy: 2.4489 - val_sparse_categorical_crossentropy_2: 4.7205 - val_sparse_categorical_accuracy: 0.4463 - val_sparse_top_k_categorical_accuracy: 0.6885 - val_adversarial_loss: 4.4589\n",
            "Epoch 38/110\n",
            "782/782 - 111s - loss: 7.4404 - sparse_categorical_crossentropy: 2.0772 - sparse_categorical_crossentropy_2: 4.5469 - sparse_categorical_accuracy: 0.5031 - sparse_top_k_categorical_accuracy: 0.7465 - adversarial_loss: 4.4310 - val_loss: 7.7037 - val_sparse_categorical_crossentropy: 2.3474 - val_sparse_categorical_crossentropy_2: 4.6479 - val_sparse_categorical_accuracy: 0.4512 - val_sparse_top_k_categorical_accuracy: 0.7236 - val_adversarial_loss: 4.3977\n",
            "Epoch 39/110\n",
            "782/782 - 111s - loss: 7.4165 - sparse_categorical_crossentropy: 2.0548 - sparse_categorical_crossentropy_2: 4.5452 - sparse_categorical_accuracy: 0.5080 - sparse_top_k_categorical_accuracy: 0.7502 - adversarial_loss: 4.4266 - val_loss: 7.7512 - val_sparse_categorical_crossentropy: 2.3770 - val_sparse_categorical_crossentropy_2: 4.6916 - val_sparse_categorical_accuracy: 0.4502 - val_sparse_top_k_categorical_accuracy: 0.7109 - val_adversarial_loss: 4.4905\n",
            "Epoch 40/110\n",
            "782/782 - 111s - loss: 7.4110 - sparse_categorical_crossentropy: 2.0507 - sparse_categorical_crossentropy_2: 4.5559 - sparse_categorical_accuracy: 0.5088 - sparse_top_k_categorical_accuracy: 0.7512 - adversarial_loss: 4.4227 - val_loss: 7.8105 - val_sparse_categorical_crossentropy: 2.4440 - val_sparse_categorical_crossentropy_2: 4.5880 - val_sparse_categorical_accuracy: 0.4492 - val_sparse_top_k_categorical_accuracy: 0.6904 - val_adversarial_loss: 4.4554\n",
            "Epoch 41/110\n",
            "782/782 - 111s - loss: 7.4156 - sparse_categorical_crossentropy: 2.0582 - sparse_categorical_crossentropy_2: 4.5509 - sparse_categorical_accuracy: 0.5081 - sparse_top_k_categorical_accuracy: 0.7481 - adversarial_loss: 4.4114 - val_loss: 7.7607 - val_sparse_categorical_crossentropy: 2.3986 - val_sparse_categorical_crossentropy_2: 4.5664 - val_sparse_categorical_accuracy: 0.4385 - val_sparse_top_k_categorical_accuracy: 0.7021 - val_adversarial_loss: 4.4363\n",
            "Epoch 42/110\n",
            "782/782 - 111s - loss: 7.3836 - sparse_categorical_crossentropy: 2.0293 - sparse_categorical_crossentropy_2: 4.5452 - sparse_categorical_accuracy: 0.5137 - sparse_top_k_categorical_accuracy: 0.7537 - adversarial_loss: 4.3987 - val_loss: 7.7679 - val_sparse_categorical_crossentropy: 2.4076 - val_sparse_categorical_crossentropy_2: 4.6628 - val_sparse_categorical_accuracy: 0.4521 - val_sparse_top_k_categorical_accuracy: 0.7002 - val_adversarial_loss: 4.4308\n",
            "Epoch 43/110\n",
            "782/782 - 111s - loss: 7.3717 - sparse_categorical_crossentropy: 2.0211 - sparse_categorical_crossentropy_2: 4.5485 - sparse_categorical_accuracy: 0.5147 - sparse_top_k_categorical_accuracy: 0.7542 - adversarial_loss: 4.3833 - val_loss: 7.7758 - val_sparse_categorical_crossentropy: 2.4176 - val_sparse_categorical_crossentropy_2: 4.6719 - val_sparse_categorical_accuracy: 0.4268 - val_sparse_top_k_categorical_accuracy: 0.7090 - val_adversarial_loss: 4.4232\n",
            "Epoch 44/110\n",
            "782/782 - 111s - loss: 7.3574 - sparse_categorical_crossentropy: 2.0078 - sparse_categorical_crossentropy_2: 4.5429 - sparse_categorical_accuracy: 0.5177 - sparse_top_k_categorical_accuracy: 0.7582 - adversarial_loss: 4.3814 - val_loss: 7.7672 - val_sparse_categorical_crossentropy: 2.4158 - val_sparse_categorical_crossentropy_2: 4.6558 - val_sparse_categorical_accuracy: 0.4404 - val_sparse_top_k_categorical_accuracy: 0.7002 - val_adversarial_loss: 4.3916\n",
            "Epoch 45/110\n",
            "782/782 - 111s - loss: 7.3375 - sparse_categorical_crossentropy: 1.9913 - sparse_categorical_crossentropy_2: 4.5380 - sparse_categorical_accuracy: 0.5218 - sparse_top_k_categorical_accuracy: 0.7616 - adversarial_loss: 4.3672 - val_loss: 7.6899 - val_sparse_categorical_crossentropy: 2.3355 - val_sparse_categorical_crossentropy_2: 4.6020 - val_sparse_categorical_accuracy: 0.4463 - val_sparse_top_k_categorical_accuracy: 0.7148 - val_adversarial_loss: 4.4102\n",
            "Epoch 46/110\n",
            "782/782 - 111s - loss: 7.3477 - sparse_categorical_crossentropy: 2.0020 - sparse_categorical_crossentropy_2: 4.5479 - sparse_categorical_accuracy: 0.5196 - sparse_top_k_categorical_accuracy: 0.7564 - adversarial_loss: 4.3680 - val_loss: 7.8157 - val_sparse_categorical_crossentropy: 2.4604 - val_sparse_categorical_crossentropy_2: 4.6583 - val_sparse_categorical_accuracy: 0.4346 - val_sparse_top_k_categorical_accuracy: 0.6914 - val_adversarial_loss: 4.4180\n",
            "Epoch 47/110\n",
            "782/782 - 111s - loss: 7.3149 - sparse_categorical_crossentropy: 1.9723 - sparse_categorical_crossentropy_2: 4.5339 - sparse_categorical_accuracy: 0.5248 - sparse_top_k_categorical_accuracy: 0.7654 - adversarial_loss: 4.3556 - val_loss: 7.7209 - val_sparse_categorical_crossentropy: 2.3672 - val_sparse_categorical_crossentropy_2: 4.5398 - val_sparse_categorical_accuracy: 0.4453 - val_sparse_top_k_categorical_accuracy: 0.7119 - val_adversarial_loss: 4.4124\n",
            "Epoch 48/110\n",
            "782/782 - 111s - loss: 7.3171 - sparse_categorical_crossentropy: 1.9761 - sparse_categorical_crossentropy_2: 4.5366 - sparse_categorical_accuracy: 0.5231 - sparse_top_k_categorical_accuracy: 0.7631 - adversarial_loss: 4.3507 - val_loss: 7.6622 - val_sparse_categorical_crossentropy: 2.3134 - val_sparse_categorical_crossentropy_2: 4.6187 - val_sparse_categorical_accuracy: 0.4541 - val_sparse_top_k_categorical_accuracy: 0.7246 - val_adversarial_loss: 4.3908\n",
            "Epoch 49/110\n",
            "782/782 - 111s - loss: 7.2905 - sparse_categorical_crossentropy: 1.9535 - sparse_categorical_crossentropy_2: 4.5346 - sparse_categorical_accuracy: 0.5274 - sparse_top_k_categorical_accuracy: 0.7681 - adversarial_loss: 4.3338 - val_loss: 7.7222 - val_sparse_categorical_crossentropy: 2.3798 - val_sparse_categorical_crossentropy_2: 4.6037 - val_sparse_categorical_accuracy: 0.4424 - val_sparse_top_k_categorical_accuracy: 0.7070 - val_adversarial_loss: 4.3623\n",
            "Epoch 50/110\n",
            "782/782 - 111s - loss: 7.2971 - sparse_categorical_crossentropy: 1.9604 - sparse_categorical_crossentropy_2: 4.5331 - sparse_categorical_accuracy: 0.5275 - sparse_top_k_categorical_accuracy: 0.7648 - adversarial_loss: 4.3350 - val_loss: 7.7139 - val_sparse_categorical_crossentropy: 2.3713 - val_sparse_categorical_crossentropy_2: 4.5849 - val_sparse_categorical_accuracy: 0.4473 - val_sparse_top_k_categorical_accuracy: 0.7100 - val_adversarial_loss: 4.3661\n",
            "Epoch 51/110\n",
            "782/782 - 111s - loss: 7.2719 - sparse_categorical_crossentropy: 1.9384 - sparse_categorical_crossentropy_2: 4.5266 - sparse_categorical_accuracy: 0.5320 - sparse_top_k_categorical_accuracy: 0.7687 - adversarial_loss: 4.3224 - val_loss: 7.7288 - val_sparse_categorical_crossentropy: 2.3806 - val_sparse_categorical_crossentropy_2: 4.7012 - val_sparse_categorical_accuracy: 0.4463 - val_sparse_top_k_categorical_accuracy: 0.7031 - val_adversarial_loss: 4.3969\n",
            "Epoch 52/110\n",
            "782/782 - 111s - loss: 7.2551 - sparse_categorical_crossentropy: 1.9247 - sparse_categorical_crossentropy_2: 4.5251 - sparse_categorical_accuracy: 0.5366 - sparse_top_k_categorical_accuracy: 0.7704 - adversarial_loss: 4.3096 - val_loss: 7.6515 - val_sparse_categorical_crossentropy: 2.3138 - val_sparse_categorical_crossentropy_2: 4.5834 - val_sparse_categorical_accuracy: 0.4727 - val_sparse_top_k_categorical_accuracy: 0.7070 - val_adversarial_loss: 4.3472\n",
            "Epoch 53/110\n",
            "782/782 - 111s - loss: 7.2564 - sparse_categorical_crossentropy: 1.9253 - sparse_categorical_crossentropy_2: 4.5269 - sparse_categorical_accuracy: 0.5347 - sparse_top_k_categorical_accuracy: 0.7696 - adversarial_loss: 4.3164 - val_loss: 7.7372 - val_sparse_categorical_crossentropy: 2.3912 - val_sparse_categorical_crossentropy_2: 4.6538 - val_sparse_categorical_accuracy: 0.4590 - val_sparse_top_k_categorical_accuracy: 0.7100 - val_adversarial_loss: 4.3918\n",
            "Epoch 54/110\n",
            "782/782 - 111s - loss: 7.2566 - sparse_categorical_crossentropy: 1.9275 - sparse_categorical_crossentropy_2: 4.5364 - sparse_categorical_accuracy: 0.5341 - sparse_top_k_categorical_accuracy: 0.7708 - adversarial_loss: 4.3088 - val_loss: 7.7140 - val_sparse_categorical_crossentropy: 2.3696 - val_sparse_categorical_crossentropy_2: 4.5607 - val_sparse_categorical_accuracy: 0.4609 - val_sparse_top_k_categorical_accuracy: 0.7061 - val_adversarial_loss: 4.3866\n",
            "{'loss': [8.572748184204102, 8.367218017578125, 8.269430160522461, 8.174264907836914, 8.101924896240234, 8.038383483886719, 7.964371681213379, 7.913931846618652, 7.867595195770264, 7.846309185028076, 7.8285722732543945, 7.785412311553955, 7.7584357261657715, 7.72885274887085, 7.7031941413879395, 7.682199001312256, 7.654356002807617, 7.631694316864014, 7.61054801940918, 7.585294246673584, 7.57346248626709, 7.535344123840332, 7.51292085647583, 7.510989665985107, 7.49156379699707, 7.463211536407471, 7.461534023284912, 7.440444469451904, 7.416536331176758, 7.411014080047607, 7.4156389236450195, 7.3836140632629395, 7.371676445007324, 7.357356548309326, 7.337485313415527, 7.347672939300537, 7.314868927001953, 7.317071914672852, 7.29054594039917, 7.29710054397583, 7.271949768066406, 7.2551469802856445, 7.256441116333008, 7.256587982177734], 'sparse_categorical_crossentropy': [2.933403730392456, 2.775132179260254, 2.7089927196502686, 2.6390626430511475, 2.5871520042419434, 2.5411643981933594, 2.483527898788452, 2.4464340209960938, 2.4113008975982666, 2.3989715576171875, 2.3890039920806885, 2.3522942066192627, 2.336047410964966, 2.312882661819458, 2.290196180343628, 2.2751123905181885, 2.2519948482513428, 2.232980489730835, 2.2170217037200928, 2.1969690322875977, 2.1870715618133545, 2.153282642364502, 2.1339828968048096, 2.135014295578003, 2.117892026901245, 2.0943872928619385, 2.0954904556274414, 2.0771982669830322, 2.054811716079712, 2.0506927967071533, 2.058187961578369, 2.029340982437134, 2.0211009979248047, 2.007777690887451, 1.9913419485092163, 2.001997232437134, 1.9722775220870972, 1.9760578870773315, 1.9535114765167236, 1.960411787033081, 1.9383835792541504, 1.9247249364852905, 1.9252593517303467, 1.9275041818618774], 'sparse_categorical_crossentropy_2': [4.957363605499268, 4.775960445404053, 4.725543022155762, 4.690871715545654, 4.6788434982299805, 4.656364917755127, 4.639693260192871, 4.620462894439697, 4.621417045593262, 4.621495246887207, 4.627703666687012, 4.606389999389648, 4.606157302856445, 4.600514888763428, 4.587589263916016, 4.592503547668457, 4.582268714904785, 4.581936836242676, 4.577315330505371, 4.570405006408691, 4.566430568695068, 4.562232494354248, 4.563927173614502, 4.566756725311279, 4.556941986083984, 4.555811882019043, 4.553900241851807, 4.546882152557373, 4.545202255249023, 4.555947303771973, 4.550892353057861, 4.54522705078125, 4.5484843254089355, 4.542861461639404, 4.537952899932861, 4.547885894775391, 4.533884525299072, 4.536584854125977, 4.534596920013428, 4.533052444458008, 4.5266242027282715, 4.525112628936768, 4.526856899261475, 4.536373615264893], 'sparse_categorical_accuracy': [0.34517863392829895, 0.3730219006538391, 0.3819233477115631, 0.39636948704719543, 0.40654972195625305, 0.4141324460506439, 0.42609095573425293, 0.43301430344581604, 0.43877875804901123, 0.4399276673793793, 0.4424852132797241, 0.4486992359161377, 0.45289522409439087, 0.45645180344581604, 0.46039801836013794, 0.462875634431839, 0.4668518304824829, 0.47214674949645996, 0.473535418510437, 0.48060861229896545, 0.48181745409965515, 0.48962995409965515, 0.49191775918006897, 0.4905890226364136, 0.4948749244213104, 0.5006593465805054, 0.4997102916240692, 0.5030670762062073, 0.5080122947692871, 0.5087915658950806, 0.5081022381782532, 0.5136569142341614, 0.5146759152412415, 0.5177130103111267, 0.5217990875244141, 0.5196211934089661, 0.524766206741333, 0.523097813129425, 0.5273936986923218, 0.5275335907936096, 0.5319793224334717, 0.5366048812866211, 0.5347166657447815, 0.5340772867202759], 'sparse_top_k_categorical_accuracy': [0.5850683450698853, 0.6179068088531494, 0.6318134665489197, 0.645440399646759, 0.6559302806854248, 0.6645919680595398, 0.675441563129425, 0.6816756129264832, 0.6867307424545288, 0.6893782019615173, 0.6910865306854248, 0.6973305344581604, 0.7004975080490112, 0.7037843465805054, 0.709209144115448, 0.7123861312866211, 0.7152333855628967, 0.7190197706222534, 0.721357524394989, 0.7259031534194946, 0.7254635691642761, 0.7325267791748047, 0.7344748973846436, 0.7342650890350342, 0.7384810447692871, 0.7421075701713562, 0.7427269816398621, 0.7465433478355408, 0.7501698136329651, 0.751218855381012, 0.7481318116188049, 0.7537364363670349, 0.7542159557342529, 0.7582321166992188, 0.7615888714790344, 0.7563539147377014, 0.7654151916503906, 0.763107419013977, 0.7681125998497009, 0.7647657990455627, 0.7686620950698853, 0.770380437374115, 0.7696411609649658, 0.7707900404930115], 'adversarial_loss': [5.721555233001709, 5.488397598266602, 5.333466529846191, 5.21068000793457, 5.111922740936279, 5.027641773223877, 4.9492082595825195, 4.885947227478027, 4.833354949951172, 4.792004108428955, 4.756614685058594, 4.727768898010254, 4.67748498916626, 4.648802280426025, 4.6372833251953125, 4.611106872558594, 4.590818881988525, 4.575942516326904, 4.553283214569092, 4.5305633544921875, 4.524200439453125, 4.505836486816406, 4.493486404418945, 4.481846809387207, 4.473569393157959, 4.452539443969727, 4.441844463348389, 4.431008338928223, 4.426581382751465, 4.422698497772217, 4.4114460945129395, 4.398654937744141, 4.3832688331604, 4.381385326385498, 4.367199897766113, 4.367971420288086, 4.355574131011963, 4.3506927490234375, 4.333765506744385, 4.335036277770996, 4.3223876953125, 4.309637069702148, 4.316381454467773, 4.308811664581299], 'val_loss': [8.137866020202637, 8.003684997558594, 8.02920150756836, 7.993657112121582, 8.034241676330566, 7.951540470123291, 7.934216499328613, 8.009106636047363, 7.871101379394531, 7.9249138832092285, 7.94184684753418, 7.907625198364258, 7.8088765144348145, 7.875975608825684, 7.847291946411133, 7.792858123779297, 7.748037338256836, 7.8228864669799805, 7.748867034912109, 7.689157962799072, 7.773670673370361, 7.780118465423584, 7.748383522033691, 7.640393257141113, 7.772367477416992, 7.771294116973877, 7.818094730377197, 7.70371150970459, 7.75123929977417, 7.810534954071045, 7.760660171508789, 7.767948627471924, 7.775796890258789, 7.767163276672363, 7.689887046813965, 7.815738677978516, 7.720898628234863, 7.662182331085205, 7.722222328186035, 7.713908672332764, 7.728768825531006, 7.651483535766602, 7.737195014953613, 7.713985919952393], 'val_sparse_categorical_crossentropy': [2.570814371109009, 2.479304313659668, 2.523317575454712, 2.5123393535614014, 2.5564417839050293, 2.49513578414917, 2.4835453033447266, 2.5675723552703857, 2.4346654415130615, 2.4928476810455322, 2.5203373432159424, 2.4900922775268555, 2.3993959426879883, 2.4707419872283936, 2.4525225162506104, 2.404100179672241, 2.3721041679382324, 2.4330506324768066, 2.3845930099487305, 2.3331401348114014, 2.4019317626953125, 2.4093010425567627, 2.3786725997924805, 2.2798287868499756, 2.392385244369507, 2.394315242767334, 2.4489471912384033, 2.347435712814331, 2.3770315647125244, 2.4439773559570312, 2.3985581398010254, 2.4075512886047363, 2.417550802230835, 2.415846347808838, 2.335461378097534, 2.460353136062622, 2.367241621017456, 2.313441276550293, 2.3797833919525146, 2.371303081512451, 2.3806042671203613, 2.3138487339019775, 2.3912250995635986, 2.369624137878418], 'val_sparse_categorical_crossentropy_2': [4.717459201812744, 4.737258434295654, 4.695714950561523, 4.6866278648376465, 4.715278148651123, 4.702843189239502, 4.729068279266357, 4.683297157287598, 4.751060962677002, 4.645287036895752, 4.607646942138672, 4.763427734375, 4.59256649017334, 4.628941059112549, 4.614377021789551, 4.582137107849121, 4.641619682312012, 4.659272193908691, 4.564294815063477, 4.604775428771973, 4.622807502746582, 4.610738277435303, 4.665027618408203, 4.622939586639404, 4.661938190460205, 4.636314392089844, 4.720477104187012, 4.647881507873535, 4.691575050354004, 4.588021278381348, 4.566407680511475, 4.662768363952637, 4.671908378601074, 4.655767917633057, 4.6020402908325195, 4.6582536697387695, 4.539838790893555, 4.6187028884887695, 4.603734493255615, 4.584903717041016, 4.701158046722412, 4.58341646194458, 4.653829574584961, 4.560707092285156], 'val_sparse_categorical_accuracy': [0.4169921875, 0.431640625, 0.4306640625, 0.4228515625, 0.4111328125, 0.439453125, 0.4287109375, 0.421875, 0.4384765625, 0.431640625, 0.41796875, 0.4287109375, 0.435546875, 0.4228515625, 0.4423828125, 0.443359375, 0.4443359375, 0.4365234375, 0.4375, 0.4501953125, 0.451171875, 0.45703125, 0.447265625, 0.4638671875, 0.4482421875, 0.4365234375, 0.4462890625, 0.451171875, 0.4501953125, 0.44921875, 0.4384765625, 0.4521484375, 0.4267578125, 0.4404296875, 0.4462890625, 0.4345703125, 0.4453125, 0.4541015625, 0.4423828125, 0.447265625, 0.4462890625, 0.47265625, 0.458984375, 0.4609375], 'val_sparse_top_k_categorical_accuracy': [0.67578125, 0.6982421875, 0.6787109375, 0.669921875, 0.673828125, 0.6875, 0.6806640625, 0.66015625, 0.6875, 0.66796875, 0.6650390625, 0.6845703125, 0.6875, 0.685546875, 0.685546875, 0.697265625, 0.71484375, 0.6865234375, 0.6962890625, 0.705078125, 0.6962890625, 0.703125, 0.7080078125, 0.724609375, 0.7119140625, 0.69921875, 0.6884765625, 0.7236328125, 0.7109375, 0.6904296875, 0.7021484375, 0.7001953125, 0.708984375, 0.7001953125, 0.71484375, 0.69140625, 0.7119140625, 0.724609375, 0.70703125, 0.7099609375, 0.703125, 0.70703125, 0.7099609375, 0.7060546875], 'val_adversarial_loss': [5.361622333526611, 5.151512622833252, 5.062379360198975, 4.94296407699585, 4.928814888000488, 4.825291633605957, 4.800084590911865, 4.757845878601074, 4.735803604125977, 4.717382907867432, 4.668009281158447, 4.651532173156738, 4.614663124084473, 4.596803665161133, 4.547848224639893, 4.521137714385986, 4.460340976715088, 4.533179759979248, 4.408681869506836, 4.370689392089844, 4.452574729919434, 4.451236248016357, 4.448946475982666, 4.40643835067749, 4.506734371185303, 4.494912147521973, 4.458928108215332, 4.397736549377441, 4.490545749664307, 4.455433368682861, 4.436264991760254, 4.430838108062744, 4.423171520233154, 4.3915910720825195, 4.410195827484131, 4.418023586273193, 4.412403106689453, 4.390824317932129, 4.3623046875, 4.366106986999512, 4.396869659423828, 4.347175598144531, 4.391775131225586, 4.386648178100586]}\n",
            "\n",
            "\n",
            "Start finetuning at block_14_expand\n",
            "Model: \"base_model\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "image (InputLayer)           [(None, 64, 64, 3)]       0         \n",
            "_________________________________________________________________\n",
            "upsample (UpSampling2D)      (None, 128, 128, 3)       0         \n",
            "_________________________________________________________________\n",
            "mobilenetv2_1.00_128 (Model) (None, 1280)              2257984   \n",
            "_________________________________________________________________\n",
            "prediction (Dense)           (None, 200)               256200    \n",
            "=================================================================\n",
            "Total params: 2,514,184\n",
            "Trainable params: 1,782,280\n",
            "Non-trainable params: 731,904\n",
            "_________________________________________________________________\n",
            "Epoch 55/154\n",
            "WARNING:tensorflow:Method (on_train_batch_end) is slow compared to the batch update (0.455502). Check your callbacks.\n",
            "782/782 - 119s - loss: 7.6075 - sparse_categorical_crossentropy: 2.2036 - sparse_categorical_crossentropy_2: 4.6126 - sparse_categorical_accuracy: 0.4768 - sparse_top_k_categorical_accuracy: 0.7246 - adversarial_loss: 4.6850 - val_loss: 7.7611 - val_sparse_categorical_crossentropy: 2.3885 - val_sparse_categorical_crossentropy_2: 4.7182 - val_sparse_categorical_accuracy: 0.4541 - val_sparse_top_k_categorical_accuracy: 0.7051 - val_adversarial_loss: 4.5292\n",
            "Epoch 56/154\n",
            "782/782 - 116s - loss: 7.5344 - sparse_categorical_crossentropy: 2.1440 - sparse_categorical_crossentropy_2: 4.5644 - sparse_categorical_accuracy: 0.4853 - sparse_top_k_categorical_accuracy: 0.7346 - adversarial_loss: 4.6191 - val_loss: 7.8101 - val_sparse_categorical_crossentropy: 2.4479 - val_sparse_categorical_crossentropy_2: 4.7421 - val_sparse_categorical_accuracy: 0.4287 - val_sparse_top_k_categorical_accuracy: 0.7070 - val_adversarial_loss: 4.4788\n",
            "Epoch 57/154\n",
            "782/782 - 116s - loss: 7.4472 - sparse_categorical_crossentropy: 2.0745 - sparse_categorical_crossentropy_2: 4.5507 - sparse_categorical_accuracy: 0.5013 - sparse_top_k_categorical_accuracy: 0.7477 - adversarial_loss: 4.5322 - val_loss: 7.7568 - val_sparse_categorical_crossentropy: 2.4149 - val_sparse_categorical_crossentropy_2: 4.7525 - val_sparse_categorical_accuracy: 0.4414 - val_sparse_top_k_categorical_accuracy: 0.6953 - val_adversarial_loss: 4.3787\n",
            "Epoch 58/154\n",
            "782/782 - 116s - loss: 7.3924 - sparse_categorical_crossentropy: 2.0305 - sparse_categorical_crossentropy_2: 4.5442 - sparse_categorical_accuracy: 0.5080 - sparse_top_k_categorical_accuracy: 0.7549 - adversarial_loss: 4.4794 - val_loss: 7.8925 - val_sparse_categorical_crossentropy: 2.5269 - val_sparse_categorical_crossentropy_2: 4.7682 - val_sparse_categorical_accuracy: 0.4268 - val_sparse_top_k_categorical_accuracy: 0.6865 - val_adversarial_loss: 4.4985\n",
            "Epoch 59/154\n",
            "782/782 - 116s - loss: 7.3435 - sparse_categorical_crossentropy: 1.9887 - sparse_categorical_crossentropy_2: 4.5339 - sparse_categorical_accuracy: 0.5185 - sparse_top_k_categorical_accuracy: 0.7618 - adversarial_loss: 4.4449 - val_loss: 7.8395 - val_sparse_categorical_crossentropy: 2.4899 - val_sparse_categorical_crossentropy_2: 4.7094 - val_sparse_categorical_accuracy: 0.4268 - val_sparse_top_k_categorical_accuracy: 0.6885 - val_adversarial_loss: 4.4196\n",
            "Epoch 60/154\n",
            "782/782 - 116s - loss: 7.3108 - sparse_categorical_crossentropy: 1.9626 - sparse_categorical_crossentropy_2: 4.5391 - sparse_categorical_accuracy: 0.5255 - sparse_top_k_categorical_accuracy: 0.7640 - adversarial_loss: 4.4138 - val_loss: 7.7131 - val_sparse_categorical_crossentropy: 2.3837 - val_sparse_categorical_crossentropy_2: 4.6612 - val_sparse_categorical_accuracy: 0.4424 - val_sparse_top_k_categorical_accuracy: 0.7041 - val_adversarial_loss: 4.3206\n",
            "Epoch 61/154\n",
            "782/782 - 116s - loss: 7.2924 - sparse_categorical_crossentropy: 1.9477 - sparse_categorical_crossentropy_2: 4.5436 - sparse_categorical_accuracy: 0.5292 - sparse_top_k_categorical_accuracy: 0.7679 - adversarial_loss: 4.3975 - val_loss: 7.7732 - val_sparse_categorical_crossentropy: 2.4309 - val_sparse_categorical_crossentropy_2: 4.6984 - val_sparse_categorical_accuracy: 0.4346 - val_sparse_top_k_categorical_accuracy: 0.6924 - val_adversarial_loss: 4.3867\n",
            "Epoch 62/154\n",
            "782/782 - 116s - loss: 7.2672 - sparse_categorical_crossentropy: 1.9275 - sparse_categorical_crossentropy_2: 4.5242 - sparse_categorical_accuracy: 0.5330 - sparse_top_k_categorical_accuracy: 0.7714 - adversarial_loss: 4.3743 - val_loss: 7.6582 - val_sparse_categorical_crossentropy: 2.3367 - val_sparse_categorical_crossentropy_2: 4.6753 - val_sparse_categorical_accuracy: 0.4619 - val_sparse_top_k_categorical_accuracy: 0.7061 - val_adversarial_loss: 4.2837\n",
            "Epoch 63/154\n",
            "782/782 - 116s - loss: 7.2584 - sparse_categorical_crossentropy: 1.9218 - sparse_categorical_crossentropy_2: 4.5290 - sparse_categorical_accuracy: 0.5350 - sparse_top_k_categorical_accuracy: 0.7706 - adversarial_loss: 4.3600 - val_loss: 7.7505 - val_sparse_categorical_crossentropy: 2.4126 - val_sparse_categorical_crossentropy_2: 4.6702 - val_sparse_categorical_accuracy: 0.4395 - val_sparse_top_k_categorical_accuracy: 0.6963 - val_adversarial_loss: 4.3669\n",
            "Epoch 64/154\n",
            "782/782 - 116s - loss: 7.2439 - sparse_categorical_crossentropy: 1.9094 - sparse_categorical_crossentropy_2: 4.5241 - sparse_categorical_accuracy: 0.5370 - sparse_top_k_categorical_accuracy: 0.7723 - adversarial_loss: 4.3511 - val_loss: 7.7399 - val_sparse_categorical_crossentropy: 2.4101 - val_sparse_categorical_crossentropy_2: 4.6926 - val_sparse_categorical_accuracy: 0.4482 - val_sparse_top_k_categorical_accuracy: 0.6963 - val_adversarial_loss: 4.3279\n",
            "Epoch 65/154\n",
            "782/782 - 115s - loss: 7.2025 - sparse_categorical_crossentropy: 1.8736 - sparse_categorical_crossentropy_2: 4.5349 - sparse_categorical_accuracy: 0.5463 - sparse_top_k_categorical_accuracy: 0.7788 - adversarial_loss: 4.3245 - val_loss: 7.6221 - val_sparse_categorical_crossentropy: 2.2916 - val_sparse_categorical_crossentropy_2: 4.6329 - val_sparse_categorical_accuracy: 0.4727 - val_sparse_top_k_categorical_accuracy: 0.7100 - val_adversarial_loss: 4.3330\n",
            "Epoch 66/154\n",
            "782/782 - 115s - loss: 7.1957 - sparse_categorical_crossentropy: 1.8678 - sparse_categorical_crossentropy_2: 4.5287 - sparse_categorical_accuracy: 0.5472 - sparse_top_k_categorical_accuracy: 0.7810 - adversarial_loss: 4.3206 - val_loss: 7.6352 - val_sparse_categorical_crossentropy: 2.3093 - val_sparse_categorical_crossentropy_2: 4.6461 - val_sparse_categorical_accuracy: 0.4600 - val_sparse_top_k_categorical_accuracy: 0.7129 - val_adversarial_loss: 4.3118\n",
            "Epoch 67/154\n",
            "782/782 - 115s - loss: 7.1822 - sparse_categorical_crossentropy: 1.8570 - sparse_categorical_crossentropy_2: 4.5244 - sparse_categorical_accuracy: 0.5506 - sparse_top_k_categorical_accuracy: 0.7797 - adversarial_loss: 4.3089 - val_loss: 7.6583 - val_sparse_categorical_crossentropy: 2.3287 - val_sparse_categorical_crossentropy_2: 4.5873 - val_sparse_categorical_accuracy: 0.4697 - val_sparse_top_k_categorical_accuracy: 0.7090 - val_adversarial_loss: 4.3317\n",
            "Epoch 68/154\n",
            "782/782 - 115s - loss: 7.1734 - sparse_categorical_crossentropy: 1.8485 - sparse_categorical_crossentropy_2: 4.5136 - sparse_categorical_accuracy: 0.5511 - sparse_top_k_categorical_accuracy: 0.7838 - adversarial_loss: 4.3085 - val_loss: 7.6991 - val_sparse_categorical_crossentropy: 2.3660 - val_sparse_categorical_crossentropy_2: 4.6295 - val_sparse_categorical_accuracy: 0.4600 - val_sparse_top_k_categorical_accuracy: 0.6895 - val_adversarial_loss: 4.3505\n",
            "Epoch 69/154\n",
            "782/782 - 115s - loss: 7.1682 - sparse_categorical_crossentropy: 1.8445 - sparse_categorical_crossentropy_2: 4.5268 - sparse_categorical_accuracy: 0.5525 - sparse_top_k_categorical_accuracy: 0.7846 - adversarial_loss: 4.3044 - val_loss: 7.6192 - val_sparse_categorical_crossentropy: 2.3002 - val_sparse_categorical_crossentropy_2: 4.6926 - val_sparse_categorical_accuracy: 0.4688 - val_sparse_top_k_categorical_accuracy: 0.7109 - val_adversarial_loss: 4.2815\n",
            "Epoch 70/154\n",
            "782/782 - 116s - loss: 7.1627 - sparse_categorical_crossentropy: 1.8408 - sparse_categorical_crossentropy_2: 4.5193 - sparse_categorical_accuracy: 0.5521 - sparse_top_k_categorical_accuracy: 0.7833 - adversarial_loss: 4.2968 - val_loss: 7.6518 - val_sparse_categorical_crossentropy: 2.3281 - val_sparse_categorical_crossentropy_2: 4.6404 - val_sparse_categorical_accuracy: 0.4707 - val_sparse_top_k_categorical_accuracy: 0.7090 - val_adversarial_loss: 4.3067\n",
            "Epoch 71/154\n",
            "782/782 - 116s - loss: 7.1441 - sparse_categorical_crossentropy: 1.8228 - sparse_categorical_crossentropy_2: 4.5168 - sparse_categorical_accuracy: 0.5567 - sparse_top_k_categorical_accuracy: 0.7883 - adversarial_loss: 4.2955 - val_loss: 7.6740 - val_sparse_categorical_crossentropy: 2.3421 - val_sparse_categorical_crossentropy_2: 4.5637 - val_sparse_categorical_accuracy: 0.4795 - val_sparse_top_k_categorical_accuracy: 0.6953 - val_adversarial_loss: 4.3490\n",
            "Epoch 72/154\n",
            "782/782 - 115s - loss: 7.1329 - sparse_categorical_crossentropy: 1.8138 - sparse_categorical_crossentropy_2: 4.5189 - sparse_categorical_accuracy: 0.5589 - sparse_top_k_categorical_accuracy: 0.7896 - adversarial_loss: 4.2852 - val_loss: 7.6247 - val_sparse_categorical_crossentropy: 2.3030 - val_sparse_categorical_crossentropy_2: 4.5799 - val_sparse_categorical_accuracy: 0.4756 - val_sparse_top_k_categorical_accuracy: 0.7041 - val_adversarial_loss: 4.2991\n",
            "Epoch 73/154\n",
            "782/782 - 115s - loss: 7.1265 - sparse_categorical_crossentropy: 1.8098 - sparse_categorical_crossentropy_2: 4.5248 - sparse_categorical_accuracy: 0.5589 - sparse_top_k_categorical_accuracy: 0.7899 - adversarial_loss: 4.2748 - val_loss: 7.7075 - val_sparse_categorical_crossentropy: 2.3745 - val_sparse_categorical_crossentropy_2: 4.6028 - val_sparse_categorical_accuracy: 0.4600 - val_sparse_top_k_categorical_accuracy: 0.6924 - val_adversarial_loss: 4.3570\n",
            "Epoch 74/154\n",
            "782/782 - 116s - loss: 7.1234 - sparse_categorical_crossentropy: 1.8068 - sparse_categorical_crossentropy_2: 4.5096 - sparse_categorical_accuracy: 0.5590 - sparse_top_k_categorical_accuracy: 0.7910 - adversarial_loss: 4.2760 - val_loss: 7.6603 - val_sparse_categorical_crossentropy: 2.3240 - val_sparse_categorical_crossentropy_2: 4.7037 - val_sparse_categorical_accuracy: 0.4785 - val_sparse_top_k_categorical_accuracy: 0.7070 - val_adversarial_loss: 4.3753\n",
            "Epoch 75/154\n",
            "782/782 - 115s - loss: 7.1173 - sparse_categorical_crossentropy: 1.8018 - sparse_categorical_crossentropy_2: 4.5198 - sparse_categorical_accuracy: 0.5613 - sparse_top_k_categorical_accuracy: 0.7916 - adversarial_loss: 4.2715 - val_loss: 7.6722 - val_sparse_categorical_crossentropy: 2.3399 - val_sparse_categorical_crossentropy_2: 4.6070 - val_sparse_categorical_accuracy: 0.4551 - val_sparse_top_k_categorical_accuracy: 0.7070 - val_adversarial_loss: 4.3568\n",
            "Epoch 76/154\n",
            "782/782 - 115s - loss: 7.1063 - sparse_categorical_crossentropy: 1.7921 - sparse_categorical_crossentropy_2: 4.5126 - sparse_categorical_accuracy: 0.5625 - sparse_top_k_categorical_accuracy: 0.7921 - adversarial_loss: 4.2667 - val_loss: 7.6860 - val_sparse_categorical_crossentropy: 2.3581 - val_sparse_categorical_crossentropy_2: 4.6551 - val_sparse_categorical_accuracy: 0.4492 - val_sparse_top_k_categorical_accuracy: 0.6982 - val_adversarial_loss: 4.3360\n",
            "Epoch 77/154\n",
            "782/782 - 116s - loss: 7.1014 - sparse_categorical_crossentropy: 1.7875 - sparse_categorical_crossentropy_2: 4.5083 - sparse_categorical_accuracy: 0.5647 - sparse_top_k_categorical_accuracy: 0.7944 - adversarial_loss: 4.2667 - val_loss: 7.6992 - val_sparse_categorical_crossentropy: 2.3617 - val_sparse_categorical_crossentropy_2: 4.6416 - val_sparse_categorical_accuracy: 0.4580 - val_sparse_top_k_categorical_accuracy: 0.7070 - val_adversarial_loss: 4.3853\n",
            "Epoch 78/154\n",
            "782/782 - 116s - loss: 7.1065 - sparse_categorical_crossentropy: 1.7939 - sparse_categorical_crossentropy_2: 4.5186 - sparse_categorical_accuracy: 0.5628 - sparse_top_k_categorical_accuracy: 0.7931 - adversarial_loss: 4.2618 - val_loss: 7.5318 - val_sparse_categorical_crossentropy: 2.2125 - val_sparse_categorical_crossentropy_2: 4.5882 - val_sparse_categorical_accuracy: 0.4893 - val_sparse_top_k_categorical_accuracy: 0.7314 - val_adversarial_loss: 4.2959\n",
            "Epoch 79/154\n",
            "782/782 - 115s - loss: 7.0779 - sparse_categorical_crossentropy: 1.7675 - sparse_categorical_crossentropy_2: 4.5127 - sparse_categorical_accuracy: 0.5691 - sparse_top_k_categorical_accuracy: 0.7984 - adversarial_loss: 4.2516 - val_loss: 7.6738 - val_sparse_categorical_crossentropy: 2.3431 - val_sparse_categorical_crossentropy_2: 4.7231 - val_sparse_categorical_accuracy: 0.4805 - val_sparse_top_k_categorical_accuracy: 0.7168 - val_adversarial_loss: 4.3540\n",
            "Epoch 80/154\n",
            "782/782 - 115s - loss: 7.0705 - sparse_categorical_crossentropy: 1.7628 - sparse_categorical_crossentropy_2: 4.5155 - sparse_categorical_accuracy: 0.5701 - sparse_top_k_categorical_accuracy: 0.7977 - adversarial_loss: 4.2402 - val_loss: 7.6376 - val_sparse_categorical_crossentropy: 2.3141 - val_sparse_categorical_crossentropy_2: 4.5990 - val_sparse_categorical_accuracy: 0.4629 - val_sparse_top_k_categorical_accuracy: 0.7129 - val_adversarial_loss: 4.3198\n",
            "Epoch 81/154\n",
            "782/782 - 115s - loss: 7.0689 - sparse_categorical_crossentropy: 1.7615 - sparse_categorical_crossentropy_2: 4.5139 - sparse_categorical_accuracy: 0.5723 - sparse_top_k_categorical_accuracy: 0.7970 - adversarial_loss: 4.2402 - val_loss: 7.7028 - val_sparse_categorical_crossentropy: 2.3691 - val_sparse_categorical_crossentropy_2: 4.7160 - val_sparse_categorical_accuracy: 0.4688 - val_sparse_top_k_categorical_accuracy: 0.6992 - val_adversarial_loss: 4.3719\n",
            "Epoch 82/154\n",
            "782/782 - 115s - loss: 7.0703 - sparse_categorical_crossentropy: 1.7635 - sparse_categorical_crossentropy_2: 4.5074 - sparse_categorical_accuracy: 0.5692 - sparse_top_k_categorical_accuracy: 0.7974 - adversarial_loss: 4.2387 - val_loss: 7.5978 - val_sparse_categorical_crossentropy: 2.2812 - val_sparse_categorical_crossentropy_2: 4.6489 - val_sparse_categorical_accuracy: 0.4785 - val_sparse_top_k_categorical_accuracy: 0.7275 - val_adversarial_loss: 4.2879\n",
            "Epoch 83/154\n",
            "782/782 - 115s - loss: 7.0716 - sparse_categorical_crossentropy: 1.7659 - sparse_categorical_crossentropy_2: 4.5079 - sparse_categorical_accuracy: 0.5696 - sparse_top_k_categorical_accuracy: 0.7976 - adversarial_loss: 4.2343 - val_loss: 7.6717 - val_sparse_categorical_crossentropy: 2.3358 - val_sparse_categorical_crossentropy_2: 4.6867 - val_sparse_categorical_accuracy: 0.4746 - val_sparse_top_k_categorical_accuracy: 0.7080 - val_adversarial_loss: 4.3861\n",
            "Epoch 84/154\n",
            "782/782 - 115s - loss: 7.0572 - sparse_categorical_crossentropy: 1.7531 - sparse_categorical_crossentropy_2: 4.5084 - sparse_categorical_accuracy: 0.5727 - sparse_top_k_categorical_accuracy: 0.7991 - adversarial_loss: 4.2278 - val_loss: 7.6113 - val_sparse_categorical_crossentropy: 2.2838 - val_sparse_categorical_crossentropy_2: 4.6052 - val_sparse_categorical_accuracy: 0.4746 - val_sparse_top_k_categorical_accuracy: 0.7227 - val_adversarial_loss: 4.3454\n",
            "Epoch 85/154\n",
            "782/782 - 115s - loss: 7.0581 - sparse_categorical_crossentropy: 1.7538 - sparse_categorical_crossentropy_2: 4.5132 - sparse_categorical_accuracy: 0.5724 - sparse_top_k_categorical_accuracy: 0.7982 - adversarial_loss: 4.2301 - val_loss: 7.6135 - val_sparse_categorical_crossentropy: 2.2879 - val_sparse_categorical_crossentropy_2: 4.7189 - val_sparse_categorical_accuracy: 0.4854 - val_sparse_top_k_categorical_accuracy: 0.7158 - val_adversarial_loss: 4.3371\n",
            "Epoch 86/154\n",
            "782/782 - 115s - loss: 7.0401 - sparse_categorical_crossentropy: 1.7381 - sparse_categorical_crossentropy_2: 4.5119 - sparse_categorical_accuracy: 0.5789 - sparse_top_k_categorical_accuracy: 0.8006 - adversarial_loss: 4.2201 - val_loss: 7.6608 - val_sparse_categorical_crossentropy: 2.3405 - val_sparse_categorical_crossentropy_2: 4.6613 - val_sparse_categorical_accuracy: 0.4561 - val_sparse_top_k_categorical_accuracy: 0.7119 - val_adversarial_loss: 4.3124\n",
            "Epoch 87/154\n",
            "782/782 - 115s - loss: 7.0334 - sparse_categorical_crossentropy: 1.7330 - sparse_categorical_crossentropy_2: 4.5082 - sparse_categorical_accuracy: 0.5806 - sparse_top_k_categorical_accuracy: 0.8022 - adversarial_loss: 4.2133 - val_loss: 7.5875 - val_sparse_categorical_crossentropy: 2.2649 - val_sparse_categorical_crossentropy_2: 4.5277 - val_sparse_categorical_accuracy: 0.4795 - val_sparse_top_k_categorical_accuracy: 0.7275 - val_adversarial_loss: 4.3251\n",
            "Epoch 88/154\n",
            "782/782 - 115s - loss: 7.0369 - sparse_categorical_crossentropy: 1.7352 - sparse_categorical_crossentropy_2: 4.5081 - sparse_categorical_accuracy: 0.5766 - sparse_top_k_categorical_accuracy: 0.8020 - adversarial_loss: 4.2210 - val_loss: 7.6562 - val_sparse_categorical_crossentropy: 2.3156 - val_sparse_categorical_crossentropy_2: 4.6810 - val_sparse_categorical_accuracy: 0.4775 - val_sparse_top_k_categorical_accuracy: 0.7188 - val_adversarial_loss: 4.4169\n",
            "Epoch 89/154\n",
            "782/782 - 116s - loss: 7.0247 - sparse_categorical_crossentropy: 1.7246 - sparse_categorical_crossentropy_2: 4.5099 - sparse_categorical_accuracy: 0.5807 - sparse_top_k_categorical_accuracy: 0.8036 - adversarial_loss: 4.2148 - val_loss: 7.7132 - val_sparse_categorical_crossentropy: 2.3762 - val_sparse_categorical_crossentropy_2: 4.6985 - val_sparse_categorical_accuracy: 0.4512 - val_sparse_top_k_categorical_accuracy: 0.6963 - val_adversarial_loss: 4.3998\n",
            "Epoch 90/154\n",
            "782/782 - 116s - loss: 7.0413 - sparse_categorical_crossentropy: 1.7413 - sparse_categorical_crossentropy_2: 4.5092 - sparse_categorical_accuracy: 0.5764 - sparse_top_k_categorical_accuracy: 0.8000 - adversarial_loss: 4.2155 - val_loss: 7.6220 - val_sparse_categorical_crossentropy: 2.3050 - val_sparse_categorical_crossentropy_2: 4.6189 - val_sparse_categorical_accuracy: 0.4648 - val_sparse_top_k_categorical_accuracy: 0.7129 - val_adversarial_loss: 4.3019\n",
            "Epoch 91/154\n",
            "782/782 - 116s - loss: 7.0250 - sparse_categorical_crossentropy: 1.7262 - sparse_categorical_crossentropy_2: 4.5047 - sparse_categorical_accuracy: 0.5776 - sparse_top_k_categorical_accuracy: 0.8037 - adversarial_loss: 4.2113 - val_loss: 7.5040 - val_sparse_categorical_crossentropy: 2.2034 - val_sparse_categorical_crossentropy_2: 4.6276 - val_sparse_categorical_accuracy: 0.5000 - val_sparse_top_k_categorical_accuracy: 0.7412 - val_adversarial_loss: 4.2207\n",
            "Epoch 92/154\n",
            "782/782 - 116s - loss: 7.0061 - sparse_categorical_crossentropy: 1.7112 - sparse_categorical_crossentropy_2: 4.5139 - sparse_categorical_accuracy: 0.5845 - sparse_top_k_categorical_accuracy: 0.8052 - adversarial_loss: 4.1930 - val_loss: 7.5210 - val_sparse_categorical_crossentropy: 2.2018 - val_sparse_categorical_crossentropy_2: 4.5901 - val_sparse_categorical_accuracy: 0.5010 - val_sparse_top_k_categorical_accuracy: 0.7285 - val_adversarial_loss: 4.3151\n",
            "Epoch 93/154\n",
            "782/782 - 116s - loss: 6.9996 - sparse_categorical_crossentropy: 1.7059 - sparse_categorical_crossentropy_2: 4.5039 - sparse_categorical_accuracy: 0.5852 - sparse_top_k_categorical_accuracy: 0.8054 - adversarial_loss: 4.1885 - val_loss: 7.6204 - val_sparse_categorical_crossentropy: 2.2949 - val_sparse_categorical_crossentropy_2: 4.5754 - val_sparse_categorical_accuracy: 0.4844 - val_sparse_top_k_categorical_accuracy: 0.7119 - val_adversarial_loss: 4.3479\n",
            "Epoch 94/154\n",
            "782/782 - 116s - loss: 6.9996 - sparse_categorical_crossentropy: 1.7061 - sparse_categorical_crossentropy_2: 4.5100 - sparse_categorical_accuracy: 0.5848 - sparse_top_k_categorical_accuracy: 0.8066 - adversarial_loss: 4.1892 - val_loss: 7.6494 - val_sparse_categorical_crossentropy: 2.3278 - val_sparse_categorical_crossentropy_2: 4.6590 - val_sparse_categorical_accuracy: 0.4697 - val_sparse_top_k_categorical_accuracy: 0.7129 - val_adversarial_loss: 4.3306\n",
            "Epoch 95/154\n",
            "782/782 - 115s - loss: 6.9978 - sparse_categorical_crossentropy: 1.7037 - sparse_categorical_crossentropy_2: 4.5025 - sparse_categorical_accuracy: 0.5842 - sparse_top_k_categorical_accuracy: 0.8077 - adversarial_loss: 4.1936 - val_loss: 7.6569 - val_sparse_categorical_crossentropy: 2.3302 - val_sparse_categorical_crossentropy_2: 4.6329 - val_sparse_categorical_accuracy: 0.4648 - val_sparse_top_k_categorical_accuracy: 0.7090 - val_adversarial_loss: 4.3572\n",
            "Epoch 96/154\n",
            "782/782 - 115s - loss: 6.9941 - sparse_categorical_crossentropy: 1.7023 - sparse_categorical_crossentropy_2: 4.5099 - sparse_categorical_accuracy: 0.5841 - sparse_top_k_categorical_accuracy: 0.8077 - adversarial_loss: 4.1834 - val_loss: 7.6129 - val_sparse_categorical_crossentropy: 2.2896 - val_sparse_categorical_crossentropy_2: 4.6361 - val_sparse_categorical_accuracy: 0.4697 - val_sparse_top_k_categorical_accuracy: 0.7100 - val_adversarial_loss: 4.3414\n",
            "Epoch 97/154\n",
            "782/782 - 115s - loss: 6.9973 - sparse_categorical_crossentropy: 1.7044 - sparse_categorical_crossentropy_2: 4.5082 - sparse_categorical_accuracy: 0.5862 - sparse_top_k_categorical_accuracy: 0.8069 - adversarial_loss: 4.1906 - val_loss: 7.5826 - val_sparse_categorical_crossentropy: 2.2663 - val_sparse_categorical_crossentropy_2: 4.6119 - val_sparse_categorical_accuracy: 0.4658 - val_sparse_top_k_categorical_accuracy: 0.7324 - val_adversarial_loss: 4.3079\n",
            "Epoch 98/154\n",
            "782/782 - 115s - loss: 6.9862 - sparse_categorical_crossentropy: 1.6931 - sparse_categorical_crossentropy_2: 4.5046 - sparse_categorical_accuracy: 0.5874 - sparse_top_k_categorical_accuracy: 0.8088 - adversarial_loss: 4.1926 - val_loss: 7.5888 - val_sparse_categorical_crossentropy: 2.2689 - val_sparse_categorical_crossentropy_2: 4.6345 - val_sparse_categorical_accuracy: 0.4697 - val_sparse_top_k_categorical_accuracy: 0.7256 - val_adversarial_loss: 4.3270\n",
            "Epoch 99/154\n",
            "782/782 - 115s - loss: 6.9949 - sparse_categorical_crossentropy: 1.7026 - sparse_categorical_crossentropy_2: 4.5035 - sparse_categorical_accuracy: 0.5846 - sparse_top_k_categorical_accuracy: 0.8070 - adversarial_loss: 4.1902 - val_loss: 7.5852 - val_sparse_categorical_crossentropy: 2.2643 - val_sparse_categorical_crossentropy_2: 4.6648 - val_sparse_categorical_accuracy: 0.4639 - val_sparse_top_k_categorical_accuracy: 0.7207 - val_adversarial_loss: 4.3340\n",
            "Epoch 100/154\n",
            "782/782 - 115s - loss: 6.9907 - sparse_categorical_crossentropy: 1.6986 - sparse_categorical_crossentropy_2: 4.5071 - sparse_categorical_accuracy: 0.5855 - sparse_top_k_categorical_accuracy: 0.8073 - adversarial_loss: 4.1903 - val_loss: 7.6057 - val_sparse_categorical_crossentropy: 2.2812 - val_sparse_categorical_crossentropy_2: 4.6052 - val_sparse_categorical_accuracy: 0.4824 - val_sparse_top_k_categorical_accuracy: 0.7080 - val_adversarial_loss: 4.3532\n",
            "Epoch 101/154\n",
            "782/782 - 115s - loss: 6.9810 - sparse_categorical_crossentropy: 1.6908 - sparse_categorical_crossentropy_2: 4.5031 - sparse_categorical_accuracy: 0.5890 - sparse_top_k_categorical_accuracy: 0.8089 - adversarial_loss: 4.1826 - val_loss: 7.6195 - val_sparse_categorical_crossentropy: 2.2916 - val_sparse_categorical_crossentropy_2: 4.6190 - val_sparse_categorical_accuracy: 0.4697 - val_sparse_top_k_categorical_accuracy: 0.7090 - val_adversarial_loss: 4.3717\n",
            "Epoch 102/154\n",
            "782/782 - 115s - loss: 6.9765 - sparse_categorical_crossentropy: 1.6865 - sparse_categorical_crossentropy_2: 4.5071 - sparse_categorical_accuracy: 0.5891 - sparse_top_k_categorical_accuracy: 0.8105 - adversarial_loss: 4.1830 - val_loss: 7.5604 - val_sparse_categorical_crossentropy: 2.2473 - val_sparse_categorical_crossentropy_2: 4.6643 - val_sparse_categorical_accuracy: 0.4805 - val_sparse_top_k_categorical_accuracy: 0.7256 - val_adversarial_loss: 4.2995\n",
            "Epoch 103/154\n",
            "782/782 - 115s - loss: 6.9818 - sparse_categorical_crossentropy: 1.6932 - sparse_categorical_crossentropy_2: 4.5041 - sparse_categorical_accuracy: 0.5865 - sparse_top_k_categorical_accuracy: 0.8077 - adversarial_loss: 4.1771 - val_loss: 7.6163 - val_sparse_categorical_crossentropy: 2.2942 - val_sparse_categorical_crossentropy_2: 4.6851 - val_sparse_categorical_accuracy: 0.4707 - val_sparse_top_k_categorical_accuracy: 0.7129 - val_adversarial_loss: 4.3456\n",
            "Epoch 104/154\n",
            "782/782 - 115s - loss: 6.9818 - sparse_categorical_crossentropy: 1.6925 - sparse_categorical_crossentropy_2: 4.5015 - sparse_categorical_accuracy: 0.5872 - sparse_top_k_categorical_accuracy: 0.8086 - adversarial_loss: 4.1823 - val_loss: 7.6026 - val_sparse_categorical_crossentropy: 2.2807 - val_sparse_categorical_crossentropy_2: 4.6211 - val_sparse_categorical_accuracy: 0.4707 - val_sparse_top_k_categorical_accuracy: 0.7129 - val_adversarial_loss: 4.3459\n",
            "Epoch 105/154\n",
            "782/782 - 115s - loss: 6.9654 - sparse_categorical_crossentropy: 1.6791 - sparse_categorical_crossentropy_2: 4.4973 - sparse_categorical_accuracy: 0.5889 - sparse_top_k_categorical_accuracy: 0.8106 - adversarial_loss: 4.1686 - val_loss: 7.6587 - val_sparse_categorical_crossentropy: 2.3310 - val_sparse_categorical_crossentropy_2: 4.6796 - val_sparse_categorical_accuracy: 0.4600 - val_sparse_top_k_categorical_accuracy: 0.7051 - val_adversarial_loss: 4.3763\n",
            "Epoch 106/154\n",
            "782/782 - 115s - loss: 6.9606 - sparse_categorical_crossentropy: 1.6757 - sparse_categorical_crossentropy_2: 4.5051 - sparse_categorical_accuracy: 0.5919 - sparse_top_k_categorical_accuracy: 0.8124 - adversarial_loss: 4.1629 - val_loss: 7.6902 - val_sparse_categorical_crossentropy: 2.3625 - val_sparse_categorical_crossentropy_2: 4.7309 - val_sparse_categorical_accuracy: 0.4600 - val_sparse_top_k_categorical_accuracy: 0.7070 - val_adversarial_loss: 4.3777\n",
            "Epoch 107/154\n",
            "782/782 - 115s - loss: 6.9577 - sparse_categorical_crossentropy: 1.6724 - sparse_categorical_crossentropy_2: 4.5033 - sparse_categorical_accuracy: 0.5922 - sparse_top_k_categorical_accuracy: 0.8124 - adversarial_loss: 4.1666 - val_loss: 7.5636 - val_sparse_categorical_crossentropy: 2.2416 - val_sparse_categorical_crossentropy_2: 4.6022 - val_sparse_categorical_accuracy: 0.4727 - val_sparse_top_k_categorical_accuracy: 0.7275 - val_adversarial_loss: 4.3507\n",
            "Epoch 108/154\n",
            "782/782 - 115s - loss: 6.9630 - sparse_categorical_crossentropy: 1.6778 - sparse_categorical_crossentropy_2: 4.5051 - sparse_categorical_accuracy: 0.5908 - sparse_top_k_categorical_accuracy: 0.8101 - adversarial_loss: 4.1674 - val_loss: 7.5703 - val_sparse_categorical_crossentropy: 2.2398 - val_sparse_categorical_crossentropy_2: 4.5895 - val_sparse_categorical_accuracy: 0.4824 - val_sparse_top_k_categorical_accuracy: 0.7217 - val_adversarial_loss: 4.3948\n",
            "Epoch 109/154\n",
            "782/782 - 115s - loss: 6.9435 - sparse_categorical_crossentropy: 1.6615 - sparse_categorical_crossentropy_2: 4.5026 - sparse_categorical_accuracy: 0.5939 - sparse_top_k_categorical_accuracy: 0.8143 - adversarial_loss: 4.1529 - val_loss: 7.6341 - val_sparse_categorical_crossentropy: 2.3049 - val_sparse_categorical_crossentropy_2: 4.6361 - val_sparse_categorical_accuracy: 0.4541 - val_sparse_top_k_categorical_accuracy: 0.7246 - val_adversarial_loss: 4.3897\n",
            "Epoch 110/154\n",
            "782/782 - 115s - loss: 6.9534 - sparse_categorical_crossentropy: 1.6706 - sparse_categorical_crossentropy_2: 4.5067 - sparse_categorical_accuracy: 0.5932 - sparse_top_k_categorical_accuracy: 0.8128 - adversarial_loss: 4.1583 - val_loss: 7.5517 - val_sparse_categorical_crossentropy: 2.2365 - val_sparse_categorical_crossentropy_2: 4.6010 - val_sparse_categorical_accuracy: 0.4727 - val_sparse_top_k_categorical_accuracy: 0.7285 - val_adversarial_loss: 4.3211\n",
            "Epoch 111/154\n",
            "782/782 - 115s - loss: 6.9480 - sparse_categorical_crossentropy: 1.6647 - sparse_categorical_crossentropy_2: 4.5026 - sparse_categorical_accuracy: 0.5938 - sparse_top_k_categorical_accuracy: 0.8127 - adversarial_loss: 4.1621 - val_loss: 7.5606 - val_sparse_categorical_crossentropy: 2.2410 - val_sparse_categorical_crossentropy_2: 4.5750 - val_sparse_categorical_accuracy: 0.4814 - val_sparse_top_k_categorical_accuracy: 0.7266 - val_adversarial_loss: 4.3446\n",
            "Epoch 112/154\n",
            "782/782 - 115s - loss: 6.9525 - sparse_categorical_crossentropy: 1.6701 - sparse_categorical_crossentropy_2: 4.5062 - sparse_categorical_accuracy: 0.5929 - sparse_top_k_categorical_accuracy: 0.8127 - adversarial_loss: 4.1593 - val_loss: 7.6279 - val_sparse_categorical_crossentropy: 2.3041 - val_sparse_categorical_crossentropy_2: 4.6043 - val_sparse_categorical_accuracy: 0.4688 - val_sparse_top_k_categorical_accuracy: 0.7129 - val_adversarial_loss: 4.3671\n",
            "Epoch 113/154\n",
            "782/782 - 115s - loss: 6.9443 - sparse_categorical_crossentropy: 1.6629 - sparse_categorical_crossentropy_2: 4.5080 - sparse_categorical_accuracy: 0.5938 - sparse_top_k_categorical_accuracy: 0.8131 - adversarial_loss: 4.1556 - val_loss: 7.5302 - val_sparse_categorical_crossentropy: 2.2208 - val_sparse_categorical_crossentropy_2: 4.5900 - val_sparse_categorical_accuracy: 0.4814 - val_sparse_top_k_categorical_accuracy: 0.7461 - val_adversarial_loss: 4.2966\n",
            "Epoch 114/154\n",
            "782/782 - 115s - loss: 6.9330 - sparse_categorical_crossentropy: 1.6533 - sparse_categorical_crossentropy_2: 4.5037 - sparse_categorical_accuracy: 0.5958 - sparse_top_k_categorical_accuracy: 0.8155 - adversarial_loss: 4.1485 - val_loss: 7.5329 - val_sparse_categorical_crossentropy: 2.2205 - val_sparse_categorical_crossentropy_2: 4.6501 - val_sparse_categorical_accuracy: 0.4941 - val_sparse_top_k_categorical_accuracy: 0.7354 - val_adversarial_loss: 4.3124\n",
            "Epoch 115/154\n",
            "782/782 - 115s - loss: 6.9403 - sparse_categorical_crossentropy: 1.6609 - sparse_categorical_crossentropy_2: 4.5049 - sparse_categorical_accuracy: 0.5926 - sparse_top_k_categorical_accuracy: 0.8143 - adversarial_loss: 4.1486 - val_loss: 7.5459 - val_sparse_categorical_crossentropy: 2.2403 - val_sparse_categorical_crossentropy_2: 4.5906 - val_sparse_categorical_accuracy: 0.4707 - val_sparse_top_k_categorical_accuracy: 0.7197 - val_adversarial_loss: 4.2800\n",
            "Epoch 116/154\n",
            "782/782 - 115s - loss: 6.9317 - sparse_categorical_crossentropy: 1.6518 - sparse_categorical_crossentropy_2: 4.5026 - sparse_categorical_accuracy: 0.5965 - sparse_top_k_categorical_accuracy: 0.8160 - adversarial_loss: 4.1523 - val_loss: 7.5165 - val_sparse_categorical_crossentropy: 2.2097 - val_sparse_categorical_crossentropy_2: 4.6526 - val_sparse_categorical_accuracy: 0.4990 - val_sparse_top_k_categorical_accuracy: 0.7314 - val_adversarial_loss: 4.2878\n",
            "Epoch 117/154\n",
            "782/782 - 115s - loss: 6.9215 - sparse_categorical_crossentropy: 1.6449 - sparse_categorical_crossentropy_2: 4.4998 - sparse_categorical_accuracy: 0.5985 - sparse_top_k_categorical_accuracy: 0.8153 - adversarial_loss: 4.1373 - val_loss: 7.5610 - val_sparse_categorical_crossentropy: 2.2447 - val_sparse_categorical_crossentropy_2: 4.6176 - val_sparse_categorical_accuracy: 0.4697 - val_sparse_top_k_categorical_accuracy: 0.7256 - val_adversarial_loss: 4.3362\n",
            "Epoch 118/154\n",
            "782/782 - 115s - loss: 6.9043 - sparse_categorical_crossentropy: 1.6294 - sparse_categorical_crossentropy_2: 4.4921 - sparse_categorical_accuracy: 0.6014 - sparse_top_k_categorical_accuracy: 0.8187 - adversarial_loss: 4.1297 - val_loss: 7.6118 - val_sparse_categorical_crossentropy: 2.2828 - val_sparse_categorical_crossentropy_2: 4.6215 - val_sparse_categorical_accuracy: 0.4629 - val_sparse_top_k_categorical_accuracy: 0.7305 - val_adversarial_loss: 4.4015\n",
            "Epoch 119/154\n",
            "782/782 - 115s - loss: 6.9169 - sparse_categorical_crossentropy: 1.6418 - sparse_categorical_crossentropy_2: 4.4995 - sparse_categorical_accuracy: 0.6010 - sparse_top_k_categorical_accuracy: 0.8155 - adversarial_loss: 4.1326 - val_loss: 7.5671 - val_sparse_categorical_crossentropy: 2.2509 - val_sparse_categorical_crossentropy_2: 4.5024 - val_sparse_categorical_accuracy: 0.4736 - val_sparse_top_k_categorical_accuracy: 0.7217 - val_adversarial_loss: 4.3388\n",
            "Epoch 120/154\n",
            "782/782 - 115s - loss: 6.9176 - sparse_categorical_crossentropy: 1.6424 - sparse_categorical_crossentropy_2: 4.5018 - sparse_categorical_accuracy: 0.5994 - sparse_top_k_categorical_accuracy: 0.8153 - adversarial_loss: 4.1344 - val_loss: 7.5503 - val_sparse_categorical_crossentropy: 2.2336 - val_sparse_categorical_crossentropy_2: 4.5451 - val_sparse_categorical_accuracy: 0.4707 - val_sparse_top_k_categorical_accuracy: 0.7334 - val_adversarial_loss: 4.3423\n",
            "Epoch 121/154\n",
            "782/782 - 115s - loss: 6.9174 - sparse_categorical_crossentropy: 1.6432 - sparse_categorical_crossentropy_2: 4.5087 - sparse_categorical_accuracy: 0.5990 - sparse_top_k_categorical_accuracy: 0.8170 - adversarial_loss: 4.1311 - val_loss: 7.6048 - val_sparse_categorical_crossentropy: 2.2850 - val_sparse_categorical_crossentropy_2: 4.6289 - val_sparse_categorical_accuracy: 0.4629 - val_sparse_top_k_categorical_accuracy: 0.7197 - val_adversarial_loss: 4.3591\n",
            "Epoch 122/154\n",
            "782/782 - 116s - loss: 6.9011 - sparse_categorical_crossentropy: 1.6273 - sparse_categorical_crossentropy_2: 4.4958 - sparse_categorical_accuracy: 0.6037 - sparse_top_k_categorical_accuracy: 0.8186 - adversarial_loss: 4.1304 - val_loss: 7.5363 - val_sparse_categorical_crossentropy: 2.2154 - val_sparse_categorical_crossentropy_2: 4.5308 - val_sparse_categorical_accuracy: 0.4805 - val_sparse_top_k_categorical_accuracy: 0.7383 - val_adversarial_loss: 4.3662\n",
            "Epoch 123/154\n",
            "782/782 - 116s - loss: 6.9023 - sparse_categorical_crossentropy: 1.6285 - sparse_categorical_crossentropy_2: 4.4920 - sparse_categorical_accuracy: 0.6016 - sparse_top_k_categorical_accuracy: 0.8179 - adversarial_loss: 4.1315 - val_loss: 7.6290 - val_sparse_categorical_crossentropy: 2.3075 - val_sparse_categorical_crossentropy_2: 4.6229 - val_sparse_categorical_accuracy: 0.4658 - val_sparse_top_k_categorical_accuracy: 0.7207 - val_adversarial_loss: 4.3712\n",
            "Epoch 124/154\n",
            "782/782 - 116s - loss: 6.9097 - sparse_categorical_crossentropy: 1.6368 - sparse_categorical_crossentropy_2: 4.4992 - sparse_categorical_accuracy: 0.6010 - sparse_top_k_categorical_accuracy: 0.8177 - adversarial_loss: 4.1287 - val_loss: 7.5948 - val_sparse_categorical_crossentropy: 2.2736 - val_sparse_categorical_crossentropy_2: 4.6417 - val_sparse_categorical_accuracy: 0.4697 - val_sparse_top_k_categorical_accuracy: 0.7324 - val_adversarial_loss: 4.3707\n",
            "Epoch 125/154\n",
            "782/782 - 115s - loss: 6.8922 - sparse_categorical_crossentropy: 1.6201 - sparse_categorical_crossentropy_2: 4.4969 - sparse_categorical_accuracy: 0.6042 - sparse_top_k_categorical_accuracy: 0.8204 - adversarial_loss: 4.1259 - val_loss: 7.6323 - val_sparse_categorical_crossentropy: 2.3077 - val_sparse_categorical_crossentropy_2: 4.6270 - val_sparse_categorical_accuracy: 0.4639 - val_sparse_top_k_categorical_accuracy: 0.7158 - val_adversarial_loss: 4.3895\n",
            "Epoch 126/154\n",
            "782/782 - 115s - loss: 6.8895 - sparse_categorical_crossentropy: 1.6190 - sparse_categorical_crossentropy_2: 4.4981 - sparse_categorical_accuracy: 0.6051 - sparse_top_k_categorical_accuracy: 0.8202 - adversarial_loss: 4.1192 - val_loss: 7.5378 - val_sparse_categorical_crossentropy: 2.2281 - val_sparse_categorical_crossentropy_2: 4.5961 - val_sparse_categorical_accuracy: 0.4883 - val_sparse_top_k_categorical_accuracy: 0.7197 - val_adversarial_loss: 4.3164\n",
            "Epoch 127/154\n",
            "782/782 - 115s - loss: 6.9014 - sparse_categorical_crossentropy: 1.6304 - sparse_categorical_crossentropy_2: 4.4995 - sparse_categorical_accuracy: 0.6017 - sparse_top_k_categorical_accuracy: 0.8177 - adversarial_loss: 4.1236 - val_loss: 7.5792 - val_sparse_categorical_crossentropy: 2.2578 - val_sparse_categorical_crossentropy_2: 4.7111 - val_sparse_categorical_accuracy: 0.4775 - val_sparse_top_k_categorical_accuracy: 0.7295 - val_adversarial_loss: 4.3762\n",
            "Epoch 128/154\n",
            "782/782 - 116s - loss: 6.8967 - sparse_categorical_crossentropy: 1.6266 - sparse_categorical_crossentropy_2: 4.5006 - sparse_categorical_accuracy: 0.6036 - sparse_top_k_categorical_accuracy: 0.8199 - adversarial_loss: 4.1202 - val_loss: 7.5472 - val_sparse_categorical_crossentropy: 2.2372 - val_sparse_categorical_crossentropy_2: 4.5912 - val_sparse_categorical_accuracy: 0.4668 - val_sparse_top_k_categorical_accuracy: 0.7236 - val_adversarial_loss: 4.3201\n",
            "Epoch 129/154\n",
            "782/782 - 115s - loss: 6.8850 - sparse_categorical_crossentropy: 1.6176 - sparse_categorical_crossentropy_2: 4.4977 - sparse_categorical_accuracy: 0.6045 - sparse_top_k_categorical_accuracy: 0.8210 - adversarial_loss: 4.1080 - val_loss: 7.5116 - val_sparse_categorical_crossentropy: 2.2075 - val_sparse_categorical_crossentropy_2: 4.5993 - val_sparse_categorical_accuracy: 0.4814 - val_sparse_top_k_categorical_accuracy: 0.7158 - val_adversarial_loss: 4.2926\n",
            "Epoch 130/154\n",
            "782/782 - 115s - loss: 6.8813 - sparse_categorical_crossentropy: 1.6142 - sparse_categorical_crossentropy_2: 4.4944 - sparse_categorical_accuracy: 0.6056 - sparse_top_k_categorical_accuracy: 0.8203 - adversarial_loss: 4.1081 - val_loss: 7.6055 - val_sparse_categorical_crossentropy: 2.2888 - val_sparse_categorical_crossentropy_2: 4.6829 - val_sparse_categorical_accuracy: 0.4561 - val_sparse_top_k_categorical_accuracy: 0.7129 - val_adversarial_loss: 4.3571\n",
            "Epoch 131/154\n",
            "782/782 - 115s - loss: 6.8798 - sparse_categorical_crossentropy: 1.6124 - sparse_categorical_crossentropy_2: 4.5014 - sparse_categorical_accuracy: 0.6065 - sparse_top_k_categorical_accuracy: 0.8207 - adversarial_loss: 4.1112 - val_loss: 7.5912 - val_sparse_categorical_crossentropy: 2.2742 - val_sparse_categorical_crossentropy_2: 4.5903 - val_sparse_categorical_accuracy: 0.4600 - val_sparse_top_k_categorical_accuracy: 0.7285 - val_adversarial_loss: 4.3594\n",
            "Epoch 132/154\n",
            "782/782 - 114s - loss: 6.8783 - sparse_categorical_crossentropy: 1.6126 - sparse_categorical_crossentropy_2: 4.4973 - sparse_categorical_accuracy: 0.6057 - sparse_top_k_categorical_accuracy: 0.8216 - adversarial_loss: 4.1038 - val_loss: 7.5427 - val_sparse_categorical_crossentropy: 2.2337 - val_sparse_categorical_crossentropy_2: 4.5596 - val_sparse_categorical_accuracy: 0.4766 - val_sparse_top_k_categorical_accuracy: 0.7256 - val_adversarial_loss: 4.3210\n",
            "Epoch 133/154\n",
            "782/782 - 115s - loss: 6.8770 - sparse_categorical_crossentropy: 1.6107 - sparse_categorical_crossentropy_2: 4.4933 - sparse_categorical_accuracy: 0.6052 - sparse_top_k_categorical_accuracy: 0.8214 - adversarial_loss: 4.1082 - val_loss: 7.6067 - val_sparse_categorical_crossentropy: 2.2783 - val_sparse_categorical_crossentropy_2: 4.5958 - val_sparse_categorical_accuracy: 0.4658 - val_sparse_top_k_categorical_accuracy: 0.7129 - val_adversarial_loss: 4.4196\n",
            "Epoch 134/154\n",
            "782/782 - 115s - loss: 6.8682 - sparse_categorical_crossentropy: 1.6025 - sparse_categorical_crossentropy_2: 4.4970 - sparse_categorical_accuracy: 0.6081 - sparse_top_k_categorical_accuracy: 0.8235 - adversarial_loss: 4.1068 - val_loss: 7.6117 - val_sparse_categorical_crossentropy: 2.2974 - val_sparse_categorical_crossentropy_2: 4.5931 - val_sparse_categorical_accuracy: 0.4824 - val_sparse_top_k_categorical_accuracy: 0.7197 - val_adversarial_loss: 4.3499\n",
            "Epoch 135/154\n",
            "782/782 - 114s - loss: 6.8763 - sparse_categorical_crossentropy: 1.6102 - sparse_categorical_crossentropy_2: 4.5002 - sparse_categorical_accuracy: 0.6056 - sparse_top_k_categorical_accuracy: 0.8217 - adversarial_loss: 4.1096 - val_loss: 7.5874 - val_sparse_categorical_crossentropy: 2.2777 - val_sparse_categorical_crossentropy_2: 4.6112 - val_sparse_categorical_accuracy: 0.4766 - val_sparse_top_k_categorical_accuracy: 0.7207 - val_adversarial_loss: 4.3287\n",
            "Epoch 136/154\n",
            "782/782 - 114s - loss: 6.8655 - sparse_categorical_crossentropy: 1.6004 - sparse_categorical_crossentropy_2: 4.4950 - sparse_categorical_accuracy: 0.6101 - sparse_top_k_categorical_accuracy: 0.8222 - adversarial_loss: 4.1066 - val_loss: 7.5536 - val_sparse_categorical_crossentropy: 2.2391 - val_sparse_categorical_crossentropy_2: 4.5592 - val_sparse_categorical_accuracy: 0.4580 - val_sparse_top_k_categorical_accuracy: 0.7285 - val_adversarial_loss: 4.3542\n",
            "Epoch 137/154\n",
            "782/782 - 114s - loss: 6.8638 - sparse_categorical_crossentropy: 1.5995 - sparse_categorical_crossentropy_2: 4.4972 - sparse_categorical_accuracy: 0.6094 - sparse_top_k_categorical_accuracy: 0.8227 - adversarial_loss: 4.1039 - val_loss: 7.5228 - val_sparse_categorical_crossentropy: 2.2165 - val_sparse_categorical_crossentropy_2: 4.6185 - val_sparse_categorical_accuracy: 0.4883 - val_sparse_top_k_categorical_accuracy: 0.7402 - val_adversarial_loss: 4.3141\n",
            "Epoch 138/154\n",
            "782/782 - 114s - loss: 6.8699 - sparse_categorical_crossentropy: 1.6060 - sparse_categorical_crossentropy_2: 4.4972 - sparse_categorical_accuracy: 0.6082 - sparse_top_k_categorical_accuracy: 0.8224 - adversarial_loss: 4.1036 - val_loss: 7.5762 - val_sparse_categorical_crossentropy: 2.2580 - val_sparse_categorical_crossentropy_2: 4.5745 - val_sparse_categorical_accuracy: 0.4678 - val_sparse_top_k_categorical_accuracy: 0.7266 - val_adversarial_loss: 4.3754\n",
            "Epoch 139/154\n",
            "782/782 - 114s - loss: 6.8620 - sparse_categorical_crossentropy: 1.5985 - sparse_categorical_crossentropy_2: 4.5024 - sparse_categorical_accuracy: 0.6102 - sparse_top_k_categorical_accuracy: 0.8228 - adversarial_loss: 4.1029 - val_loss: 7.5963 - val_sparse_categorical_crossentropy: 2.2772 - val_sparse_categorical_crossentropy_2: 4.6002 - val_sparse_categorical_accuracy: 0.4697 - val_sparse_top_k_categorical_accuracy: 0.7305 - val_adversarial_loss: 4.3812\n",
            "Epoch 140/154\n",
            "782/782 - 114s - loss: 6.8559 - sparse_categorical_crossentropy: 1.5941 - sparse_categorical_crossentropy_2: 4.4924 - sparse_categorical_accuracy: 0.6105 - sparse_top_k_categorical_accuracy: 0.8234 - adversarial_loss: 4.0958 - val_loss: 7.5890 - val_sparse_categorical_crossentropy: 2.2713 - val_sparse_categorical_crossentropy_2: 4.5807 - val_sparse_categorical_accuracy: 0.4814 - val_sparse_top_k_categorical_accuracy: 0.7256 - val_adversarial_loss: 4.3756\n",
            "Epoch 141/154\n",
            "782/782 - 114s - loss: 6.8489 - sparse_categorical_crossentropy: 1.5879 - sparse_categorical_crossentropy_2: 4.4929 - sparse_categorical_accuracy: 0.6108 - sparse_top_k_categorical_accuracy: 0.8263 - adversarial_loss: 4.0930 - val_loss: 7.4958 - val_sparse_categorical_crossentropy: 2.1826 - val_sparse_categorical_crossentropy_2: 4.6020 - val_sparse_categorical_accuracy: 0.4824 - val_sparse_top_k_categorical_accuracy: 0.7432 - val_adversarial_loss: 4.3547\n",
            "Epoch 142/154\n",
            "782/782 - 115s - loss: 6.8468 - sparse_categorical_crossentropy: 1.5867 - sparse_categorical_crossentropy_2: 4.4861 - sparse_categorical_accuracy: 0.6118 - sparse_top_k_categorical_accuracy: 0.8246 - adversarial_loss: 4.0898 - val_loss: 7.5815 - val_sparse_categorical_crossentropy: 2.2693 - val_sparse_categorical_crossentropy_2: 4.6282 - val_sparse_categorical_accuracy: 0.4746 - val_sparse_top_k_categorical_accuracy: 0.7275 - val_adversarial_loss: 4.3513\n",
            "Epoch 143/154\n",
            "782/782 - 115s - loss: 6.8475 - sparse_categorical_crossentropy: 1.5884 - sparse_categorical_crossentropy_2: 4.4958 - sparse_categorical_accuracy: 0.6117 - sparse_top_k_categorical_accuracy: 0.8238 - adversarial_loss: 4.0865 - val_loss: 7.5319 - val_sparse_categorical_crossentropy: 2.2293 - val_sparse_categorical_crossentropy_2: 4.6224 - val_sparse_categorical_accuracy: 0.4707 - val_sparse_top_k_categorical_accuracy: 0.7354 - val_adversarial_loss: 4.3047\n",
            "Epoch 144/154\n",
            "782/782 - 115s - loss: 6.8369 - sparse_categorical_crossentropy: 1.5777 - sparse_categorical_crossentropy_2: 4.4931 - sparse_categorical_accuracy: 0.6132 - sparse_top_k_categorical_accuracy: 0.8274 - adversarial_loss: 4.0882 - val_loss: 7.6250 - val_sparse_categorical_crossentropy: 2.3019 - val_sparse_categorical_crossentropy_2: 4.5628 - val_sparse_categorical_accuracy: 0.4521 - val_sparse_top_k_categorical_accuracy: 0.7139 - val_adversarial_loss: 4.4080\n",
            "Epoch 145/154\n",
            "782/782 - 115s - loss: 6.8387 - sparse_categorical_crossentropy: 1.5804 - sparse_categorical_crossentropy_2: 4.5006 - sparse_categorical_accuracy: 0.6141 - sparse_top_k_categorical_accuracy: 0.8256 - adversarial_loss: 4.0851 - val_loss: 7.5302 - val_sparse_categorical_crossentropy: 2.2271 - val_sparse_categorical_crossentropy_2: 4.6192 - val_sparse_categorical_accuracy: 0.4756 - val_sparse_top_k_categorical_accuracy: 0.7246 - val_adversarial_loss: 4.3100\n",
            "Epoch 146/154\n",
            "782/782 - 115s - loss: 6.8425 - sparse_categorical_crossentropy: 1.5841 - sparse_categorical_crossentropy_2: 4.4952 - sparse_categorical_accuracy: 0.6133 - sparse_top_k_categorical_accuracy: 0.8250 - adversarial_loss: 4.0870 - val_loss: 7.5011 - val_sparse_categorical_crossentropy: 2.2014 - val_sparse_categorical_crossentropy_2: 4.5593 - val_sparse_categorical_accuracy: 0.4785 - val_sparse_top_k_categorical_accuracy: 0.7285 - val_adversarial_loss: 4.2944\n",
            "Epoch 147/154\n",
            "782/782 - 115s - loss: 6.8372 - sparse_categorical_crossentropy: 1.5792 - sparse_categorical_crossentropy_2: 4.4925 - sparse_categorical_accuracy: 0.6135 - sparse_top_k_categorical_accuracy: 0.8259 - adversarial_loss: 4.0862 - val_loss: 7.5300 - val_sparse_categorical_crossentropy: 2.2244 - val_sparse_categorical_crossentropy_2: 4.5908 - val_sparse_categorical_accuracy: 0.4717 - val_sparse_top_k_categorical_accuracy: 0.7324 - val_adversarial_loss: 4.3251\n",
            "Epoch 148/154\n",
            "782/782 - 115s - loss: 6.8322 - sparse_categorical_crossentropy: 1.5751 - sparse_categorical_crossentropy_2: 4.4946 - sparse_categorical_accuracy: 0.6156 - sparse_top_k_categorical_accuracy: 0.8256 - adversarial_loss: 4.0834 - val_loss: 7.5454 - val_sparse_categorical_crossentropy: 2.2246 - val_sparse_categorical_crossentropy_2: 4.5716 - val_sparse_categorical_accuracy: 0.4756 - val_sparse_top_k_categorical_accuracy: 0.7344 - val_adversarial_loss: 4.4024\n",
            "Epoch 149/154\n",
            "782/782 - 115s - loss: 6.8275 - sparse_categorical_crossentropy: 1.5718 - sparse_categorical_crossentropy_2: 4.4955 - sparse_categorical_accuracy: 0.6148 - sparse_top_k_categorical_accuracy: 0.8275 - adversarial_loss: 4.0779 - val_loss: 7.4659 - val_sparse_categorical_crossentropy: 2.1635 - val_sparse_categorical_crossentropy_2: 4.5683 - val_sparse_categorical_accuracy: 0.4844 - val_sparse_top_k_categorical_accuracy: 0.7383 - val_adversarial_loss: 4.3120\n",
            "Epoch 150/154\n",
            "782/782 - 115s - loss: 6.8373 - sparse_categorical_crossentropy: 1.5810 - sparse_categorical_crossentropy_2: 4.4945 - sparse_categorical_accuracy: 0.6111 - sparse_top_k_categorical_accuracy: 0.8266 - adversarial_loss: 4.0823 - val_loss: 7.5503 - val_sparse_categorical_crossentropy: 2.2449 - val_sparse_categorical_crossentropy_2: 4.6418 - val_sparse_categorical_accuracy: 0.4883 - val_sparse_top_k_categorical_accuracy: 0.7207 - val_adversarial_loss: 4.3282\n",
            "Epoch 151/154\n",
            "782/782 - 115s - loss: 6.8190 - sparse_categorical_crossentropy: 1.5650 - sparse_categorical_crossentropy_2: 4.4879 - sparse_categorical_accuracy: 0.6162 - sparse_top_k_categorical_accuracy: 0.8289 - adversarial_loss: 4.0717 - val_loss: 7.4629 - val_sparse_categorical_crossentropy: 2.1591 - val_sparse_categorical_crossentropy_2: 4.5906 - val_sparse_categorical_accuracy: 0.4863 - val_sparse_top_k_categorical_accuracy: 0.7373 - val_adversarial_loss: 4.3216\n",
            "Epoch 152/154\n",
            "782/782 - 114s - loss: 6.8149 - sparse_categorical_crossentropy: 1.5611 - sparse_categorical_crossentropy_2: 4.4885 - sparse_categorical_accuracy: 0.6176 - sparse_top_k_categorical_accuracy: 0.8285 - adversarial_loss: 4.0725 - val_loss: 7.5796 - val_sparse_categorical_crossentropy: 2.2628 - val_sparse_categorical_crossentropy_2: 4.5675 - val_sparse_categorical_accuracy: 0.4717 - val_sparse_top_k_categorical_accuracy: 0.7158 - val_adversarial_loss: 4.3879\n",
            "Epoch 153/154\n",
            "782/782 - 115s - loss: 6.8097 - sparse_categorical_crossentropy: 1.5571 - sparse_categorical_crossentropy_2: 4.4889 - sparse_categorical_accuracy: 0.6184 - sparse_top_k_categorical_accuracy: 0.8298 - adversarial_loss: 4.0677 - val_loss: 7.5297 - val_sparse_categorical_crossentropy: 2.2291 - val_sparse_categorical_crossentropy_2: 4.6092 - val_sparse_categorical_accuracy: 0.4844 - val_sparse_top_k_categorical_accuracy: 0.7266 - val_adversarial_loss: 4.3083\n",
            "Epoch 154/154\n",
            "782/782 - 115s - loss: 6.8177 - sparse_categorical_crossentropy: 1.5650 - sparse_categorical_crossentropy_2: 4.4905 - sparse_categorical_accuracy: 0.6172 - sparse_top_k_categorical_accuracy: 0.8291 - adversarial_loss: 4.0697 - val_loss: 7.4702 - val_sparse_categorical_crossentropy: 2.1711 - val_sparse_categorical_crossentropy_2: 4.5580 - val_sparse_categorical_accuracy: 0.4941 - val_sparse_top_k_categorical_accuracy: 0.7432 - val_adversarial_loss: 4.3022\n",
            "{'loss': [7.607454776763916, 7.5344157218933105, 7.44722318649292, 7.392447471618652, 7.343456745147705, 7.3107781410217285, 7.292405128479004, 7.267248630523682, 7.258369445800781, 7.243922710418701, 7.2024760246276855, 7.195651531219482, 7.182215690612793, 7.1733598709106445, 7.168249130249023, 7.162722587585449, 7.144147872924805, 7.132863521575928, 7.126498699188232, 7.123446464538574, 7.117282390594482, 7.106267929077148, 7.1013665199279785, 7.1065239906311035, 7.077850341796875, 7.070512294769287, 7.068944931030273, 7.070338726043701, 7.071625709533691, 7.05722713470459, 7.058084964752197, 7.0401082038879395, 7.03339958190918, 7.036855220794678, 7.0246500968933105, 7.041295528411865, 7.024977684020996, 7.0060529708862305, 6.999626636505127, 6.999627113342285, 6.997832775115967, 6.994109630584717, 6.997340679168701, 6.986176490783691, 6.99489164352417, 6.99066162109375, 6.980993270874023, 6.976491451263428, 6.981761455535889, 6.981786727905273, 6.965369701385498, 6.960561752319336, 6.957676410675049, 6.962958812713623, 6.9435272216796875, 6.953381538391113, 6.947988510131836, 6.952548503875732, 6.944303512573242, 6.933012962341309, 6.940320014953613, 6.931742191314697, 6.921483993530273, 6.9042649269104, 6.916939735412598, 6.917567729949951, 6.91741943359375, 6.901109218597412, 6.902294635772705, 6.909712791442871, 6.892217636108398, 6.889467239379883, 6.901426315307617, 6.896693706512451, 6.884953022003174, 6.881303310394287, 6.879826545715332, 6.87825870513916, 6.876965045928955, 6.868236064910889, 6.876254558563232, 6.865506172180176, 6.8637824058532715, 6.86993408203125, 6.862031936645508, 6.855922698974609, 6.848930835723877, 6.846827507019043, 6.847501277923584, 6.836891174316406, 6.838675022125244, 6.842483043670654, 6.837188243865967, 6.83220100402832, 6.827518939971924, 6.837324142456055, 6.818997383117676, 6.814874649047852, 6.809715270996094, 6.817724227905273], 'sparse_categorical_crossentropy': [2.2035655975341797, 2.1439874172210693, 2.074460744857788, 2.0305376052856445, 1.9887341260910034, 1.9625709056854248, 1.947736144065857, 1.9275225400924683, 1.9217731952667236, 1.9094176292419434, 1.873565435409546, 1.8678117990493774, 1.8569992780685425, 1.8485182523727417, 1.844509243965149, 1.8407903909683228, 1.8227729797363281, 1.8138271570205688, 1.8098233938217163, 1.806816577911377, 1.8018466234207153, 1.7920795679092407, 1.7874696254730225, 1.7938892841339111, 1.7675368785858154, 1.762768268585205, 1.761484146118164, 1.7634813785552979, 1.7659189701080322, 1.753115177154541, 1.753791093826294, 1.7381032705307007, 1.7330447435379028, 1.7352343797683716, 1.7245615720748901, 1.7413469552993774, 1.7261720895767212, 1.7111872434616089, 1.7059361934661865, 1.7060778141021729, 1.7036927938461304, 1.702296257019043, 1.7043724060058594, 1.6930915117263794, 1.7025783061981201, 1.6986114978790283, 1.6907691955566406, 1.6864656209945679, 1.6931979656219482, 1.6924630403518677, 1.6790910959243774, 1.6756995916366577, 1.672350525856018, 1.6777697801589966, 1.6615086793899536, 1.6705598831176758, 1.6647111177444458, 1.6700994968414307, 1.662886619567871, 1.6532856225967407, 1.6608620882034302, 1.6518231630325317, 1.6448564529418945, 1.629438877105713, 1.6418288946151733, 1.642361044883728, 1.6431655883789062, 1.6272658109664917, 1.6285102367401123, 1.6367753744125366, 1.6201399564743042, 1.6190063953399658, 1.6303690671920776, 1.6265864372253418, 1.6175633668899536, 1.6141935586929321, 1.6123698949813843, 1.6125684976577759, 1.6106773614883423, 1.6024986505508423, 1.6102479696273804, 1.6003696918487549, 1.5994881391525269, 1.6059614419937134, 1.5984876155853271, 1.5940743684768677, 1.5879278182983398, 1.5867455005645752, 1.588364839553833, 1.5776945352554321, 1.5803687572479248, 1.5840892791748047, 1.579221248626709, 1.575085997581482, 1.5717734098434448, 1.580994725227356, 1.5650492906570435, 1.5610584020614624, 1.5571414232254028, 1.5650155544281006], 'sparse_categorical_crossentropy_2': [4.612550258636475, 4.564403057098389, 4.550710201263428, 4.544210910797119, 4.533928871154785, 4.539067268371582, 4.5436110496521, 4.524194717407227, 4.529013156890869, 4.524127006530762, 4.534859657287598, 4.528696537017822, 4.524359226226807, 4.513555526733398, 4.526833534240723, 4.5193190574646, 4.516759395599365, 4.518898010253906, 4.524831771850586, 4.509633541107178, 4.519842147827148, 4.512578010559082, 4.508338928222656, 4.518624782562256, 4.512697696685791, 4.51547384262085, 4.513866901397705, 4.507386207580566, 4.507912635803223, 4.5083513259887695, 4.5132012367248535, 4.51188325881958, 4.508157253265381, 4.508138656616211, 4.509906768798828, 4.509241580963135, 4.504735469818115, 4.51389217376709, 4.503856658935547, 4.509976863861084, 4.502501010894775, 4.509894371032715, 4.508191108703613, 4.504602909088135, 4.503477573394775, 4.507051944732666, 4.503136157989502, 4.507071495056152, 4.50408411026001, 4.5015411376953125, 4.497293949127197, 4.505054950714111, 4.503296375274658, 4.5050554275512695, 4.502584934234619, 4.506677150726318, 4.502587795257568, 4.506157875061035, 4.507986068725586, 4.5036516189575195, 4.5049333572387695, 4.502618312835693, 4.499820709228516, 4.492100238800049, 4.499486923217773, 4.5018439292907715, 4.508671760559082, 4.4958109855651855, 4.491975784301758, 4.499181270599365, 4.496922492980957, 4.498081684112549, 4.499450206756592, 4.500609874725342, 4.49765157699585, 4.49439001083374, 4.501410007476807, 4.497322082519531, 4.493265151977539, 4.496960163116455, 4.500227928161621, 4.495041847229004, 4.497154712677002, 4.497215270996094, 4.502374172210693, 4.49243688583374, 4.492887496948242, 4.486122131347656, 4.495786666870117, 4.493062496185303, 4.500633239746094, 4.495197772979736, 4.492486476898193, 4.494556903839111, 4.495453834533691, 4.494515419006348, 4.487929821014404, 4.488494396209717, 4.488880634307861, 4.490512371063232], 'sparse_categorical_accuracy': [0.47675231099128723, 0.48527413606643677, 0.5012587904930115, 0.5079823136329651, 0.5184822678565979, 0.5255154967308044, 0.5292219519615173, 0.5330482721328735, 0.5350064039230347, 0.5369845032691956, 0.5463455319404602, 0.5471647381782532, 0.5505814552307129, 0.5510609745979309, 0.5525495409965515, 0.5520600080490112, 0.5567455291748047, 0.5589234232902527, 0.5588734745979309, 0.5589833855628967, 0.5613111257553101, 0.5625199675559998, 0.564737856388092, 0.5627996921539307, 0.5691435933113098, 0.5701326727867126, 0.5723105669021606, 0.5691735744476318, 0.5696131587028503, 0.5726802349090576, 0.5724204778671265, 0.5789242386817932, 0.5805726647377014, 0.5766264200210571, 0.5807124972343445, 0.5763766765594482, 0.5775855183601379, 0.5844589471817017, 0.5852481722831726, 0.5848285555839539, 0.5841591954231262, 0.5841492414474487, 0.5861572623252869, 0.5874060988426208, 0.5845687985420227, 0.5855178833007812, 0.5890145301818848, 0.5890644788742065, 0.5864570140838623, 0.5871763229370117, 0.5888546705245972, 0.5918717980384827, 0.5921714901924133, 0.5908228158950806, 0.5939098596572876, 0.59324049949646, 0.59375, 0.592870831489563, 0.5938399434089661, 0.5958279967308044, 0.5925810933113098, 0.5965473055839539, 0.5984655022621155, 0.6014326214790344, 0.6010230183601379, 0.5993745923042297, 0.598984956741333, 0.603710412979126, 0.6015725135803223, 0.6010230183601379, 0.6041600108146667, 0.6051090955734253, 0.6017223596572876, 0.6036105155944824, 0.6045296788215637, 0.6056185960769653, 0.6065477132797241, 0.6056685447692871, 0.6051590442657471, 0.6081361770629883, 0.6056086421012878, 0.6101342439651489, 0.6093649864196777, 0.6082060933113098, 0.6101842522621155, 0.6105139255523682, 0.6107836365699768, 0.6118026971817017, 0.6117327213287354, 0.6132113337516785, 0.614070475101471, 0.613341212272644, 0.6135309934616089, 0.6155990362167358, 0.6147897839546204, 0.611103355884552, 0.6161984205245972, 0.6176270842552185, 0.6183564066886902, 0.6172074675559998], 'sparse_top_k_categorical_accuracy': [0.7246443629264832, 0.7346347570419312, 0.7476722598075867, 0.7549252510070801, 0.7618186473846436, 0.7640065550804138, 0.767862856388092, 0.7713794708251953, 0.7706301808357239, 0.7722985744476318, 0.7788123488426208, 0.7809602618217468, 0.7797414660453796, 0.783767580986023, 0.7845867872238159, 0.7833380103111267, 0.7883431911468506, 0.7895520329475403, 0.789911687374115, 0.7909606695175171, 0.7915500998497009, 0.792119562625885, 0.7943973541259766, 0.7931485772132874, 0.7983535528182983, 0.7977042198181152, 0.7969748973846436, 0.7974144816398621, 0.7976242899894714, 0.7990928888320923, 0.798203706741333, 0.8006014227867126, 0.8022198677062988, 0.8019701242446899, 0.8035985231399536, 0.7999719977378845, 0.8036984205245972, 0.8052369952201843, 0.8054267764091492, 0.8066456317901611, 0.8076746463775635, 0.8077345490455627, 0.80689537525177, 0.8088335394859314, 0.8070352673530579, 0.8073349595069885, 0.8089234232902527, 0.8104619383811951, 0.807714581489563, 0.8086336851119995, 0.8105618357658386, 0.8123900890350342, 0.8124100565910339, 0.8101223111152649, 0.8143182396888733, 0.812839686870575, 0.8127297759056091, 0.8127297759056091, 0.8131294250488281, 0.8154571652412415, 0.8142982721328735, 0.8159866333007812, 0.8153272867202759, 0.8187240362167358, 0.8155470490455627, 0.8153472542762756, 0.8170256614685059, 0.8186340928077698, 0.8179147839546204, 0.8176950216293335, 0.8204423785209656, 0.8202025890350342, 0.8177449703216553, 0.8199328780174255, 0.8210217952728271, 0.8203324675559998, 0.8206621408462524, 0.8215513229370117, 0.8213714957237244, 0.8235194087028503, 0.8216512203216553, 0.8222106695175171, 0.8227401971817017, 0.8223505616188049, 0.8227801322937012, 0.8233895301818848, 0.8262767791748047, 0.824588418006897, 0.8238391280174255, 0.8274056911468506, 0.8256273865699768, 0.8249779939651489, 0.8259271383285522, 0.8255574703216553, 0.8274656534194946, 0.8266164660453796, 0.8289342522621155, 0.828494668006897, 0.8298133611679077, 0.8291040658950806], 'adversarial_loss': [4.685000896453857, 4.619104862213135, 4.532217979431152, 4.479434967041016, 4.4449357986450195, 4.413814067840576, 4.397525310516357, 4.374263286590576, 4.360037326812744, 4.351068496704102, 4.324509143829346, 4.320579528808594, 4.308920383453369, 4.308496952056885, 4.30440092086792, 4.2967939376831055, 4.295477390289307, 4.2851972579956055, 4.274838447570801, 4.276041030883789, 4.271488189697266, 4.26669454574585, 4.266684532165527, 4.2618021965026855, 4.2516398429870605, 4.240194320678711, 4.240198135375977, 4.23865270614624, 4.234347343444824, 4.227771759033203, 4.2301154136657715, 4.220107078552246, 4.213282585144043, 4.221049785614014, 4.2148027420043945, 4.215523719787598, 4.2112507820129395, 4.192956924438477, 4.188534736633301, 4.189243793487549, 4.1936140060424805, 4.1833977699279785, 4.190616130828857, 4.1926422119140625, 4.190203666687012, 4.190306186676025, 4.182619571685791, 4.1830153465271, 4.177121162414551, 4.182348251342773, 4.168550968170166, 4.162889003753662, 4.16660213470459, 4.167369842529297, 4.1529364585876465, 4.158333778381348, 4.162065505981445, 4.159321308135986, 4.155588626861572, 4.148542404174805, 4.148638725280762, 4.15232515335083, 4.137302398681641, 4.129700660705566, 4.132553577423096, 4.134440898895264, 4.131087303161621, 4.130423069000244, 4.1315460205078125, 4.128749370574951, 4.125863075256348, 4.119168281555176, 4.1235737800598145, 4.120218753814697, 4.108044624328613, 4.108067035675049, 4.111199855804443, 4.1037702560424805, 4.108194828033447, 4.106837272644043, 4.109603404998779, 4.106644630432129, 4.103865146636963, 4.103607654571533, 4.102899074554443, 4.095820426940918, 4.093018054962158, 4.0898332595825195, 4.086478233337402, 4.088185787200928, 4.08514928817749, 4.086974143981934, 4.08624267578125, 4.083399772644043, 4.0779314041137695, 4.082263946533203, 4.071715354919434, 4.072493076324463, 4.067688941955566, 4.069725513458252], 'val_loss': [7.761130332946777, 7.810141563415527, 7.756776809692383, 7.892454147338867, 7.839459419250488, 7.713130950927734, 7.773242950439453, 7.658208847045898, 7.750454425811768, 7.739865303039551, 7.622087478637695, 7.635181903839111, 7.658319473266602, 7.699062824249268, 7.619222164154053, 7.651819705963135, 7.674038410186768, 7.624720573425293, 7.707508563995361, 7.660290241241455, 7.6722493171691895, 7.685980319976807, 7.699208736419678, 7.53177547454834, 7.673772811889648, 7.637578010559082, 7.702773094177246, 7.597776412963867, 7.671709060668945, 7.611298561096191, 7.613450527191162, 7.660796642303467, 7.587474346160889, 7.656217575073242, 7.713181972503662, 7.622042655944824, 7.503955364227295, 7.520960807800293, 7.620360851287842, 7.649442672729492, 7.656921863555908, 7.612896919250488, 7.5825700759887695, 7.588757514953613, 7.585234642028809, 7.605703830718994, 7.619532585144043, 7.560427665710449, 7.616315841674805, 7.602553367614746, 7.658680438995361, 7.69016695022583, 7.563589096069336, 7.57032585144043, 7.6340837478637695, 7.551692485809326, 7.560609817504883, 7.6279497146606445, 7.530228614807129, 7.5328755378723145, 7.545942783355713, 7.516536712646484, 7.560956954956055, 7.611806392669678, 7.567100524902344, 7.550269603729248, 7.604751110076904, 7.536289691925049, 7.629044532775879, 7.594783306121826, 7.632322311401367, 7.537805557250977, 7.5792436599731445, 7.547191619873047, 7.511608123779297, 7.605530261993408, 7.591187477111816, 7.542722702026367, 7.606704235076904, 7.611652851104736, 7.5873565673828125, 7.553603172302246, 7.522757053375244, 7.576235294342041, 7.596291542053223, 7.5889692306518555, 7.495789527893066, 7.581526279449463, 7.53192663192749, 7.6249566078186035, 7.530193328857422, 7.50111198425293, 7.529998302459717, 7.545395374298096, 7.465881824493408, 7.550297260284424, 7.462879180908203, 7.579553604125977, 7.529653549194336, 7.4702019691467285], 'val_sparse_categorical_crossentropy': [2.388537645339966, 2.4479258060455322, 2.414877414703369, 2.5268683433532715, 2.4899396896362305, 2.383697032928467, 2.430891752243042, 2.336747884750366, 2.4126358032226562, 2.4101316928863525, 2.2916202545166016, 2.309252977371216, 2.328695297241211, 2.365954637527466, 2.30021333694458, 2.3280622959136963, 2.3420910835266113, 2.303049325942993, 2.374546766281128, 2.323958158493042, 2.3398919105529785, 2.358083486557007, 2.3617348670959473, 2.2124733924865723, 2.3431198596954346, 2.3140525817871094, 2.3691318035125732, 2.281205654144287, 2.3357834815979004, 2.283801794052124, 2.287895917892456, 2.340477466583252, 2.2649075984954834, 2.3155765533447266, 2.3762333393096924, 2.304957151412964, 2.203401565551758, 2.2018203735351562, 2.294940710067749, 2.327772617340088, 2.330205202102661, 2.2896268367767334, 2.2662932872772217, 2.2689459323883057, 2.2643015384674072, 2.281217575073242, 2.29162859916687, 2.247250556945801, 2.2941954135894775, 2.280651807785034, 2.3309922218322754, 2.362473726272583, 2.2415988445281982, 2.239793062210083, 2.304861307144165, 2.2364704608917236, 2.2409586906433105, 2.304081916809082, 2.220754623413086, 2.220510959625244, 2.2403476238250732, 2.2096776962280273, 2.2446975708007812, 2.282756805419922, 2.250885248184204, 2.2336344718933105, 2.2850422859191895, 2.215444564819336, 2.307467222213745, 2.273594856262207, 2.3076584339141846, 2.228050947189331, 2.2578125, 2.2372498512268066, 2.2074546813964844, 2.2887587547302246, 2.274233818054199, 2.233727216720581, 2.278264045715332, 2.297438621520996, 2.277658224105835, 2.239102840423584, 2.2165493965148926, 2.2580463886260986, 2.277221441268921, 2.2713160514831543, 2.1825978755950928, 2.269291639328003, 2.2292885780334473, 2.301936388015747, 2.2270631790161133, 2.20137357711792, 2.224390745162964, 2.2246227264404297, 2.163455009460449, 2.244924306869507, 2.1591029167175293, 2.2627997398376465, 2.2290868759155273, 2.171143054962158], 'val_sparse_categorical_crossentropy_2': [4.71815299987793, 4.742135047912598, 4.752484321594238, 4.768160343170166, 4.709384441375732, 4.661199569702148, 4.69838809967041, 4.675281524658203, 4.670245170593262, 4.692590713500977, 4.632898807525635, 4.646143913269043, 4.5872673988342285, 4.62945032119751, 4.692627429962158, 4.6403608322143555, 4.563697814941406, 4.579889297485352, 4.602789402008057, 4.703657627105713, 4.607027053833008, 4.655128479003906, 4.641644477844238, 4.588247299194336, 4.72312593460083, 4.598952770233154, 4.716031074523926, 4.64886474609375, 4.686679363250732, 4.605246067047119, 4.718912124633789, 4.661323547363281, 4.527681350708008, 4.680962562561035, 4.698493003845215, 4.618869304656982, 4.627613067626953, 4.590142250061035, 4.575366020202637, 4.658977031707764, 4.632893085479736, 4.636102676391602, 4.611905574798584, 4.63453483581543, 4.664838790893555, 4.605198383331299, 4.619019985198975, 4.664309501647949, 4.685132026672363, 4.621124744415283, 4.679596900939941, 4.7309250831604, 4.602170944213867, 4.589500427246094, 4.636073589324951, 4.600983619689941, 4.575023174285889, 4.604315280914307, 4.589974880218506, 4.6500563621521, 4.590578079223633, 4.652641296386719, 4.617644786834717, 4.62147331237793, 4.502403259277344, 4.5450944900512695, 4.628942966461182, 4.530795097351074, 4.622862815856934, 4.64168119430542, 4.626984119415283, 4.596057415008545, 4.711093902587891, 4.59122371673584, 4.599311828613281, 4.682923316955566, 4.590320110321045, 4.5596208572387695, 4.59579610824585, 4.593064308166504, 4.611205101013184, 4.5591583251953125, 4.618547439575195, 4.574487209320068, 4.600153923034668, 4.580684661865234, 4.60198974609375, 4.628169536590576, 4.622376918792725, 4.562788009643555, 4.619246482849121, 4.559298515319824, 4.590754508972168, 4.571585178375244, 4.568279266357422, 4.641763210296631, 4.590607166290283, 4.567532539367676, 4.609151363372803, 4.558012962341309], 'val_sparse_categorical_accuracy': [0.4541015625, 0.4287109375, 0.44140625, 0.4267578125, 0.4267578125, 0.4423828125, 0.4345703125, 0.4619140625, 0.439453125, 0.4482421875, 0.47265625, 0.4599609375, 0.4697265625, 0.4599609375, 0.46875, 0.470703125, 0.4794921875, 0.4755859375, 0.4599609375, 0.478515625, 0.455078125, 0.44921875, 0.4580078125, 0.4892578125, 0.48046875, 0.462890625, 0.46875, 0.478515625, 0.474609375, 0.474609375, 0.4853515625, 0.4560546875, 0.4794921875, 0.4775390625, 0.451171875, 0.46484375, 0.5, 0.5009765625, 0.484375, 0.4697265625, 0.46484375, 0.4697265625, 0.4658203125, 0.4697265625, 0.4638671875, 0.482421875, 0.4697265625, 0.48046875, 0.470703125, 0.470703125, 0.4599609375, 0.4599609375, 0.47265625, 0.482421875, 0.4541015625, 0.47265625, 0.4814453125, 0.46875, 0.4814453125, 0.494140625, 0.470703125, 0.4990234375, 0.4697265625, 0.462890625, 0.4736328125, 0.470703125, 0.462890625, 0.48046875, 0.4658203125, 0.4697265625, 0.4638671875, 0.48828125, 0.4775390625, 0.466796875, 0.4814453125, 0.4560546875, 0.4599609375, 0.4765625, 0.4658203125, 0.482421875, 0.4765625, 0.4580078125, 0.48828125, 0.4677734375, 0.4697265625, 0.4814453125, 0.482421875, 0.474609375, 0.470703125, 0.4521484375, 0.4755859375, 0.478515625, 0.4716796875, 0.4755859375, 0.484375, 0.48828125, 0.486328125, 0.4716796875, 0.484375, 0.494140625], 'val_sparse_top_k_categorical_accuracy': [0.705078125, 0.70703125, 0.6953125, 0.6865234375, 0.6884765625, 0.7041015625, 0.6923828125, 0.7060546875, 0.6962890625, 0.6962890625, 0.7099609375, 0.712890625, 0.708984375, 0.689453125, 0.7109375, 0.708984375, 0.6953125, 0.7041015625, 0.6923828125, 0.70703125, 0.70703125, 0.6982421875, 0.70703125, 0.7314453125, 0.716796875, 0.712890625, 0.69921875, 0.7275390625, 0.7080078125, 0.72265625, 0.7158203125, 0.7119140625, 0.7275390625, 0.71875, 0.6962890625, 0.712890625, 0.7412109375, 0.728515625, 0.7119140625, 0.712890625, 0.708984375, 0.7099609375, 0.732421875, 0.7255859375, 0.720703125, 0.7080078125, 0.708984375, 0.7255859375, 0.712890625, 0.712890625, 0.705078125, 0.70703125, 0.7275390625, 0.7216796875, 0.724609375, 0.728515625, 0.7265625, 0.712890625, 0.74609375, 0.7353515625, 0.7197265625, 0.7314453125, 0.7255859375, 0.73046875, 0.7216796875, 0.7333984375, 0.7197265625, 0.73828125, 0.720703125, 0.732421875, 0.7158203125, 0.7197265625, 0.7294921875, 0.7236328125, 0.7158203125, 0.712890625, 0.728515625, 0.7255859375, 0.712890625, 0.7197265625, 0.720703125, 0.728515625, 0.740234375, 0.7265625, 0.73046875, 0.7255859375, 0.7431640625, 0.7275390625, 0.7353515625, 0.7138671875, 0.724609375, 0.728515625, 0.732421875, 0.734375, 0.73828125, 0.720703125, 0.7373046875, 0.7158203125, 0.7265625, 0.7431640625], 'val_adversarial_loss': [4.529231071472168, 4.478791236877441, 4.378655433654785, 4.498525619506836, 4.419639587402344, 4.32064962387085, 4.3866777420043945, 4.2836689949035645, 4.3668975830078125, 4.32790994644165, 4.333014965057373, 4.311760902404785, 4.3316802978515625, 4.35053825378418, 4.2814788818359375, 4.306654930114746, 4.34904146194458, 4.299095630645752, 4.356985569000244, 4.375273704528809, 4.356832981109619, 4.335964202880859, 4.385280132293701, 4.295858383178711, 4.354043006896973, 4.319840431213379, 4.371851921081543, 4.28792667388916, 4.386133670806885, 4.345422267913818, 4.337139129638672, 4.312387466430664, 4.3250532150268555, 4.4168548583984375, 4.399820804595947, 4.3019304275512695, 4.220700740814209, 4.315059185028076, 4.347883701324463, 4.330560207366943, 4.3572187423706055, 4.341411113739014, 4.307868480682373, 4.326968193054199, 4.333995819091797, 4.353185653686523, 4.3716936111450195, 4.2994866371154785, 4.3456220626831055, 4.345944404602051, 4.37630033493042, 4.377748966217041, 4.350651264190674, 4.394783973693848, 4.389652252197266, 4.321066379547119, 4.344631671905518, 4.367128372192383, 4.296577453613281, 4.312445640563965, 4.280017375946045, 4.287752151489258, 4.336170196533203, 4.4015326499938965, 4.338775634765625, 4.342286586761475, 4.359071731567383, 4.366163730621338, 4.371234893798828, 4.370703220367432, 4.389491081237793, 4.3163557052612305, 4.376152515411377, 4.32011604309082, 4.292580604553223, 4.357084274291992, 4.359399318695068, 4.321016311645508, 4.419648170471191, 4.349924564361572, 4.328749656677246, 4.354169845581055, 4.314109802246094, 4.375421524047852, 4.381234169006348, 4.375553131103516, 4.354650020599365, 4.351267337799072, 4.304685592651367, 4.407998085021973, 4.309950351715088, 4.294391632080078, 4.3251423835754395, 4.402368545532227, 4.312043190002441, 4.328169822692871, 4.321584701538086, 4.387868404388428, 4.308332920074463, 4.302192211151123]}\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "joUGMLLljevM",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}