cat_dog_train_cnn.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Categorical Perception: A Groundwork for Deep Learning\n",
    "#### Laurent Bonnasse-Gahot & Jean-Pierre Nadal\n",
    "Python notebook for training a deep convolutional neural network to classify natural images of cats and dogs (using the Kaggle Dogs vs. Cats database), used in Figure 7 (Categorical perception of a cat/dog circular continuum), Section 3.3.1."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import random\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "\n",
    "# for reproducibility\n",
    "SEED = 1\n",
    "os.environ['PYTHONHASHSEED'] = str(SEED)\n",
    "os.environ['TF_DETERMINISTIC_OPS'] = '1'\n",
    "np.random.seed(SEED)\n",
    "random.seed(SEED)\n",
    "tf.random.set_seed(SEED)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# bits of code partly taken from the \"Image classification from scratch\" tutorial\n",
    "# https://keras.io/examples/vision/image_classification_from_scratch/\n",
    "\n",
    "# data available at:\n",
    "# https://download.microsoft.com/download/3/E/1/3E1C3F21-ECDB-4869-8368-6DEBA77B919F/kagglecatsanddogs_3367a.zip"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "if not os.path.exists('model'):\n",
    "    os.makedirs('model')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Deleted 0 images\n"
     ]
    }
   ],
   "source": [
    "from tensorflow import keras\n",
    "from tensorflow.keras import layers\n",
    "\n",
    "#Filter out corrupted images\n",
    "num_skipped = 0\n",
    "for folder_name in (\"Cat\", \"Dog\"):\n",
    "    folder_path = os.path.join(\"PetImages\", folder_name)\n",
    "    for fname in os.listdir(folder_path):\n",
    "        fpath = os.path.join(folder_path, fname)\n",
    "        try:\n",
    "            fobj = open(fpath, \"rb\")\n",
    "            is_jfif = tf.compat.as_bytes(\"JFIF\") in fobj.peek(10)\n",
    "        finally:\n",
    "            fobj.close()\n",
    "\n",
    "        if not is_jfif:\n",
    "            num_skipped += 1\n",
    "            # Delete corrupted image\n",
    "            os.remove(fpath)\n",
    "\n",
    "print(\"Deleted %d images\" % num_skipped)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 23410 files belonging to 2 classes.\n",
      "Using 18728 files for training.\n",
      "Found 23410 files belonging to 2 classes.\n",
      "Using 4682 files for validation.\n"
     ]
    }
   ],
   "source": [
    "image_size = (180, 180)\n",
    "batch_size = 32\n",
    "\n",
    "train_ds = tf.keras.preprocessing.image_dataset_from_directory(\n",
    "    \"PetImages\",\n",
    "    validation_split=0.2,\n",
    "    subset=\"training\",\n",
    "    seed=1337,\n",
    "    image_size=image_size,\n",
    "    batch_size=batch_size,\n",
    ")\n",
    "val_ds = tf.keras.preprocessing.image_dataset_from_directory(\n",
    "    \"PetImages\",\n",
    "    validation_split=0.2,\n",
    "    subset=\"validation\",\n",
    "    seed=1337,\n",
    "    image_size=image_size,\n",
    "    batch_size=batch_size,\n",
    ")\n",
    "\n",
    "data_augmentation = keras.Sequential(\n",
    "    [\n",
    "        layers.experimental.preprocessing.RandomFlip(\"horizontal\"),\n",
    "        layers.experimental.preprocessing.RandomRotation(0.1),\n",
    "    ]\n",
    ")\n",
    "\n",
    "train_ds = train_ds.prefetch(buffer_size=32)\n",
    "val_ds = val_ds.prefetch(buffer_size=32)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.keras.layers import Dense, GaussianDropout, Flatten, Conv2D, MaxPooling2D, GlobalAveragePooling2D\n",
    "\n",
    "def make_model(input_shape):\n",
    "    inputs = keras.Input(shape=input_shape)\n",
    "    x = data_augmentation(inputs)  # Apply random data augmentation   \n",
    "    x = layers.experimental.preprocessing.Rescaling(scale=1./127.5, offset=-1)(x)\n",
    "    x = Conv2D(64, kernel_size=(3, 3), activation='relu', name='conv2d_1')(x)\n",
    "    x = Conv2D(64, kernel_size=(3, 3), activation='relu', name='conv2d_2')(x)\n",
    "    x = MaxPooling2D(pool_size=(2, 2), name='max_pooling2d_1')(x)\n",
    "    x = GaussianDropout(0.1)(x)\n",
    "    x = Conv2D(64, kernel_size=(3, 3), activation='relu', name='conv2d_3')(x)\n",
    "    x = Conv2D(64, kernel_size=(3, 3), activation='relu', name='conv2d_4')(x)\n",
    "    x = MaxPooling2D(pool_size=(2, 2), name='max_pooling2d_2')(x)\n",
    "    x = GaussianDropout(0.2)(x)\n",
    "    x = Conv2D(64, kernel_size=(3, 3), activation='relu', name='conv2d_5')(x)\n",
    "    x = Conv2D(64, kernel_size=(3, 3), activation='relu', name='conv2d_6')(x)\n",
    "    x = MaxPooling2D(pool_size=(2, 2), name='max_pooling2d_3')(x)  \n",
    "    x = GlobalAveragePooling2D()(x)\n",
    "    x = GaussianDropout(0.4)(x)\n",
    "    outputs = Dense(1, activation='sigmoid')(x)\n",
    "    return keras.Model(inputs, outputs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--------------- trial #0 ---------------\n",
      "Epoch 1/100\n",
      "586/586 [==============================] - 85s 139ms/step - loss: 0.6833 - accuracy: 0.5425 - val_loss: 0.6619 - val_accuracy: 0.6151\n",
      "Epoch 2/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.6586 - accuracy: 0.6043 - val_loss: 0.6147 - val_accuracy: 0.6561\n",
      "Epoch 3/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.6165 - accuracy: 0.6604 - val_loss: 0.5914 - val_accuracy: 0.6766\n",
      "Epoch 4/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.5873 - accuracy: 0.6896 - val_loss: 0.5743 - val_accuracy: 0.7082\n",
      "Epoch 5/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.5648 - accuracy: 0.7084 - val_loss: 0.5398 - val_accuracy: 0.7354\n",
      "Epoch 6/100\n",
      "586/586 [==============================] - 81s 138ms/step - loss: 0.5392 - accuracy: 0.7345 - val_loss: 0.5272 - val_accuracy: 0.7364\n",
      "Epoch 7/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.5334 - accuracy: 0.7409 - val_loss: 0.4783 - val_accuracy: 0.7813\n",
      "Epoch 8/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.4878 - accuracy: 0.7720 - val_loss: 0.4389 - val_accuracy: 0.8056\n",
      "Epoch 9/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.4496 - accuracy: 0.7970 - val_loss: 0.3959 - val_accuracy: 0.8338\n",
      "Epoch 10/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.4168 - accuracy: 0.8156 - val_loss: 0.3524 - val_accuracy: 0.8539\n",
      "Epoch 11/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.3809 - accuracy: 0.8365 - val_loss: 0.3259 - val_accuracy: 0.8603\n",
      "Epoch 12/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.3479 - accuracy: 0.8512 - val_loss: 0.3123 - val_accuracy: 0.8721\n",
      "Epoch 13/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.3291 - accuracy: 0.8593 - val_loss: 0.2837 - val_accuracy: 0.8810\n",
      "Epoch 14/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.3113 - accuracy: 0.8684 - val_loss: 0.2752 - val_accuracy: 0.8836\n",
      "Epoch 15/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2985 - accuracy: 0.8670 - val_loss: 0.2502 - val_accuracy: 0.8938\n",
      "Epoch 16/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2791 - accuracy: 0.8837 - val_loss: 0.2769 - val_accuracy: 0.8825\n",
      "Epoch 17/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2682 - accuracy: 0.8827 - val_loss: 0.2545 - val_accuracy: 0.8936\n",
      "Epoch 18/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2576 - accuracy: 0.8906 - val_loss: 0.2226 - val_accuracy: 0.9056\n",
      "Epoch 19/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2487 - accuracy: 0.8965 - val_loss: 0.2816 - val_accuracy: 0.8759\n",
      "Epoch 20/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2448 - accuracy: 0.8971 - val_loss: 0.2062 - val_accuracy: 0.9135\n",
      "Epoch 21/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2304 - accuracy: 0.9033 - val_loss: 0.2022 - val_accuracy: 0.9171\n",
      "Epoch 22/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2235 - accuracy: 0.9034 - val_loss: 0.2194 - val_accuracy: 0.9035\n",
      "Epoch 23/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2256 - accuracy: 0.9051 - val_loss: 0.1827 - val_accuracy: 0.9242\n",
      "Epoch 24/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2165 - accuracy: 0.9107 - val_loss: 0.1834 - val_accuracy: 0.9244\n",
      "Epoch 25/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.2140 - accuracy: 0.9133 - val_loss: 0.2015 - val_accuracy: 0.9158\n",
      "Epoch 26/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2036 - accuracy: 0.9160 - val_loss: 0.2111 - val_accuracy: 0.9161\n",
      "Epoch 27/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.2083 - accuracy: 0.9110 - val_loss: 0.1749 - val_accuracy: 0.9255\n",
      "Epoch 28/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2099 - accuracy: 0.9121 - val_loss: 0.1771 - val_accuracy: 0.9304\n",
      "Epoch 29/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1970 - accuracy: 0.9177 - val_loss: 0.1760 - val_accuracy: 0.9255\n",
      "Epoch 30/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1983 - accuracy: 0.9160 - val_loss: 0.1935 - val_accuracy: 0.9176\n",
      "Epoch 31/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1962 - accuracy: 0.9180 - val_loss: 0.1681 - val_accuracy: 0.9289\n",
      "Epoch 32/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1878 - accuracy: 0.9218 - val_loss: 0.2573 - val_accuracy: 0.8906\n",
      "Epoch 33/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1959 - accuracy: 0.9196 - val_loss: 0.1653 - val_accuracy: 0.9342\n",
      "Epoch 34/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1871 - accuracy: 0.9226 - val_loss: 0.2202 - val_accuracy: 0.9131\n",
      "Epoch 35/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1816 - accuracy: 0.9232 - val_loss: 0.1805 - val_accuracy: 0.9297\n",
      "Epoch 36/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1747 - accuracy: 0.9271 - val_loss: 0.2331 - val_accuracy: 0.9135\n",
      "Epoch 37/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1796 - accuracy: 0.9253 - val_loss: 0.1748 - val_accuracy: 0.9276\n",
      "Epoch 38/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1812 - accuracy: 0.9256 - val_loss: 0.1848 - val_accuracy: 0.9308\n",
      "Epoch 39/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1673 - accuracy: 0.9330 - val_loss: 0.1531 - val_accuracy: 0.9391\n",
      "Epoch 40/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1685 - accuracy: 0.9298 - val_loss: 0.1759 - val_accuracy: 0.9278\n",
      "Epoch 41/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1635 - accuracy: 0.9312 - val_loss: 0.1667 - val_accuracy: 0.9364\n",
      "Epoch 42/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1609 - accuracy: 0.9337 - val_loss: 0.1742 - val_accuracy: 0.9310\n",
      "Epoch 43/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1638 - accuracy: 0.9338 - val_loss: 0.1764 - val_accuracy: 0.9304\n",
      "Epoch 44/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1652 - accuracy: 0.9298 - val_loss: 0.1793 - val_accuracy: 0.9329\n",
      "Epoch 45/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.1531 - accuracy: 0.9361 - val_loss: 0.2111 - val_accuracy: 0.9092\n",
      "Epoch 46/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1581 - accuracy: 0.9334 - val_loss: 0.1527 - val_accuracy: 0.9423\n",
      "Epoch 47/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1678 - accuracy: 0.9338 - val_loss: 0.1618 - val_accuracy: 0.9368\n",
      "Epoch 48/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1632 - accuracy: 0.9304 - val_loss: 0.1485 - val_accuracy: 0.9447\n",
      "Epoch 49/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1476 - accuracy: 0.9423 - val_loss: 0.1447 - val_accuracy: 0.9443\n",
      "Epoch 50/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1510 - accuracy: 0.9396 - val_loss: 0.1534 - val_accuracy: 0.9372\n",
      "Epoch 51/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1526 - accuracy: 0.9375 - val_loss: 0.1528 - val_accuracy: 0.9366\n",
      "Epoch 52/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1520 - accuracy: 0.9387 - val_loss: 0.2077 - val_accuracy: 0.9205\n",
      "Epoch 53/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1506 - accuracy: 0.9387 - val_loss: 0.2198 - val_accuracy: 0.9173\n",
      "Epoch 54/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1415 - accuracy: 0.9424 - val_loss: 0.1588 - val_accuracy: 0.9391\n",
      "Epoch 55/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1469 - accuracy: 0.9375 - val_loss: 0.1801 - val_accuracy: 0.9366\n",
      "Epoch 56/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1465 - accuracy: 0.9396 - val_loss: 0.1842 - val_accuracy: 0.9336\n",
      "Epoch 57/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1383 - accuracy: 0.9433 - val_loss: 0.1859 - val_accuracy: 0.9299\n",
      "Epoch 58/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1354 - accuracy: 0.9443 - val_loss: 0.1445 - val_accuracy: 0.9440\n",
      "Epoch 59/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1503 - accuracy: 0.9394 - val_loss: 0.1562 - val_accuracy: 0.9404\n",
      "Epoch 60/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1355 - accuracy: 0.9437 - val_loss: 0.1550 - val_accuracy: 0.9430\n",
      "Epoch 61/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1245 - accuracy: 0.9472 - val_loss: 0.1490 - val_accuracy: 0.9428\n",
      "Epoch 62/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1376 - accuracy: 0.9472 - val_loss: 0.1531 - val_accuracy: 0.9447\n",
      "Epoch 63/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1284 - accuracy: 0.9462 - val_loss: 0.1630 - val_accuracy: 0.9378\n",
      "Epoch 64/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1374 - accuracy: 0.9428 - val_loss: 0.1493 - val_accuracy: 0.9432\n",
      "Epoch 65/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1313 - accuracy: 0.9467 - val_loss: 0.1457 - val_accuracy: 0.9472\n",
      "Epoch 66/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1286 - accuracy: 0.9487 - val_loss: 0.1541 - val_accuracy: 0.9472\n",
      "Epoch 67/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1315 - accuracy: 0.9443 - val_loss: 0.1581 - val_accuracy: 0.9411\n",
      "Epoch 68/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1190 - accuracy: 0.9513 - val_loss: 0.1478 - val_accuracy: 0.9492\n",
      "Epoch 69/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1332 - accuracy: 0.9463 - val_loss: 0.1495 - val_accuracy: 0.9485\n",
      "Epoch 70/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1302 - accuracy: 0.9482 - val_loss: 0.1605 - val_accuracy: 0.9436\n",
      "Epoch 71/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1204 - accuracy: 0.9513 - val_loss: 0.1493 - val_accuracy: 0.9460\n",
      "Epoch 72/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1281 - accuracy: 0.9492 - val_loss: 0.1667 - val_accuracy: 0.9438\n",
      "Epoch 73/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1253 - accuracy: 0.9497 - val_loss: 0.1641 - val_accuracy: 0.9466\n",
      "Epoch 74/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1235 - accuracy: 0.9502 - val_loss: 0.2039 - val_accuracy: 0.9295\n",
      "Epoch 75/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1242 - accuracy: 0.9481 - val_loss: 0.1665 - val_accuracy: 0.9376\n",
      "Epoch 76/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1248 - accuracy: 0.9495 - val_loss: 0.1797 - val_accuracy: 0.9391\n",
      "Epoch 77/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1270 - accuracy: 0.9482 - val_loss: 0.1430 - val_accuracy: 0.9477\n",
      "Epoch 78/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1238 - accuracy: 0.9482 - val_loss: 0.1546 - val_accuracy: 0.9524\n",
      "Epoch 79/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1153 - accuracy: 0.9537 - val_loss: 0.1651 - val_accuracy: 0.9451\n",
      "Epoch 80/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1256 - accuracy: 0.9502 - val_loss: 0.1611 - val_accuracy: 0.9419\n",
      "Epoch 81/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1132 - accuracy: 0.9529 - val_loss: 0.1500 - val_accuracy: 0.9421\n",
      "Epoch 82/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1162 - accuracy: 0.9530 - val_loss: 0.1472 - val_accuracy: 0.9411\n",
      "Epoch 83/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1227 - accuracy: 0.9528 - val_loss: 0.1610 - val_accuracy: 0.9432\n",
      "Epoch 84/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1208 - accuracy: 0.9520 - val_loss: 0.1654 - val_accuracy: 0.9468\n",
      "Epoch 85/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1162 - accuracy: 0.9556 - val_loss: 0.1393 - val_accuracy: 0.9468\n",
      "Epoch 86/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1115 - accuracy: 0.9538 - val_loss: 0.1446 - val_accuracy: 0.9483\n",
      "Epoch 87/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1151 - accuracy: 0.9524 - val_loss: 0.1360 - val_accuracy: 0.9502\n",
      "Epoch 88/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1089 - accuracy: 0.9571 - val_loss: 0.1326 - val_accuracy: 0.9534\n",
      "Epoch 89/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1161 - accuracy: 0.9541 - val_loss: 0.1396 - val_accuracy: 0.9468\n",
      "Epoch 90/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1091 - accuracy: 0.9565 - val_loss: 0.1316 - val_accuracy: 0.9541\n",
      "Epoch 91/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1196 - accuracy: 0.9518 - val_loss: 0.1675 - val_accuracy: 0.9445\n",
      "Epoch 92/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1111 - accuracy: 0.9549 - val_loss: 0.1398 - val_accuracy: 0.9534\n",
      "Epoch 93/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1040 - accuracy: 0.9593 - val_loss: 0.1430 - val_accuracy: 0.9502\n",
      "Epoch 94/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1048 - accuracy: 0.9588 - val_loss: 0.1464 - val_accuracy: 0.9475\n",
      "Epoch 95/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1072 - accuracy: 0.9536 - val_loss: 0.1464 - val_accuracy: 0.9449\n",
      "Epoch 96/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1105 - accuracy: 0.9527 - val_loss: 0.1543 - val_accuracy: 0.9507\n",
      "Epoch 97/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1082 - accuracy: 0.9573 - val_loss: 0.1416 - val_accuracy: 0.9502\n",
      "Epoch 98/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1055 - accuracy: 0.9570 - val_loss: 0.1454 - val_accuracy: 0.9466\n",
      "Epoch 99/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1076 - accuracy: 0.9554 - val_loss: 0.1489 - val_accuracy: 0.9528\n",
      "Epoch 100/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1075 - accuracy: 0.9571 - val_loss: 0.1293 - val_accuracy: 0.9472\n",
      "--------------- trial #1 ---------------\n",
      "Epoch 1/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.6885 - accuracy: 0.5163 - val_loss: 0.6464 - val_accuracy: 0.6482\n",
      "Epoch 2/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.6529 - accuracy: 0.6138 - val_loss: 0.6186 - val_accuracy: 0.6544\n",
      "Epoch 3/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.6199 - accuracy: 0.6586 - val_loss: 0.5812 - val_accuracy: 0.7095\n",
      "Epoch 4/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.5913 - accuracy: 0.6892 - val_loss: 0.5600 - val_accuracy: 0.7217\n",
      "Epoch 5/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.5752 - accuracy: 0.7038 - val_loss: 0.5383 - val_accuracy: 0.7294\n",
      "Epoch 6/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.5421 - accuracy: 0.7315 - val_loss: 0.4698 - val_accuracy: 0.7792\n",
      "Epoch 7/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.5107 - accuracy: 0.7531 - val_loss: 0.5359 - val_accuracy: 0.7373\n",
      "Epoch 8/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.4699 - accuracy: 0.7833 - val_loss: 0.3832 - val_accuracy: 0.8347\n",
      "Epoch 9/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.4297 - accuracy: 0.8066 - val_loss: 0.3344 - val_accuracy: 0.8565\n",
      "Epoch 10/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.3808 - accuracy: 0.8334 - val_loss: 0.3045 - val_accuracy: 0.8684\n",
      "Epoch 11/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "586/586 [==============================] - 84s 141ms/step - loss: 0.3475 - accuracy: 0.8495 - val_loss: 0.3042 - val_accuracy: 0.8644\n",
      "Epoch 12/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.3230 - accuracy: 0.8612 - val_loss: 0.2814 - val_accuracy: 0.8765\n",
      "Epoch 13/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.3003 - accuracy: 0.8727 - val_loss: 0.2534 - val_accuracy: 0.8924\n",
      "Epoch 14/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2899 - accuracy: 0.8803 - val_loss: 0.2211 - val_accuracy: 0.9065\n",
      "Epoch 15/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2784 - accuracy: 0.8812 - val_loss: 0.2424 - val_accuracy: 0.8983\n",
      "Epoch 16/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2703 - accuracy: 0.8863 - val_loss: 0.1977 - val_accuracy: 0.9186\n",
      "Epoch 17/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2537 - accuracy: 0.8935 - val_loss: 0.2128 - val_accuracy: 0.9073\n",
      "Epoch 18/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2504 - accuracy: 0.8947 - val_loss: 0.2278 - val_accuracy: 0.8966\n",
      "Epoch 19/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2415 - accuracy: 0.8965 - val_loss: 0.1914 - val_accuracy: 0.9203\n",
      "Epoch 20/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2480 - accuracy: 0.8994 - val_loss: 0.1790 - val_accuracy: 0.9191\n",
      "Epoch 21/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2278 - accuracy: 0.9044 - val_loss: 0.2142 - val_accuracy: 0.9073\n",
      "Epoch 22/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2219 - accuracy: 0.9063 - val_loss: 0.1815 - val_accuracy: 0.9242\n",
      "Epoch 23/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2150 - accuracy: 0.9093 - val_loss: 0.1781 - val_accuracy: 0.9242\n",
      "Epoch 24/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2076 - accuracy: 0.9130 - val_loss: 0.1788 - val_accuracy: 0.9246\n",
      "Epoch 25/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2083 - accuracy: 0.9136 - val_loss: 0.1798 - val_accuracy: 0.9280\n",
      "Epoch 26/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1983 - accuracy: 0.9199 - val_loss: 0.1626 - val_accuracy: 0.9312\n",
      "Epoch 27/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1952 - accuracy: 0.9206 - val_loss: 0.2127 - val_accuracy: 0.9118\n",
      "Epoch 28/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1951 - accuracy: 0.9181 - val_loss: 0.1670 - val_accuracy: 0.9299\n",
      "Epoch 29/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1840 - accuracy: 0.9252 - val_loss: 0.1675 - val_accuracy: 0.9321\n",
      "Epoch 30/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1913 - accuracy: 0.9204 - val_loss: 0.2012 - val_accuracy: 0.9154\n",
      "Epoch 31/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.1850 - accuracy: 0.9247 - val_loss: 0.1716 - val_accuracy: 0.9331\n",
      "Epoch 32/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1824 - accuracy: 0.9287 - val_loss: 0.1702 - val_accuracy: 0.9280\n",
      "Epoch 33/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1763 - accuracy: 0.9290 - val_loss: 0.1810 - val_accuracy: 0.9263\n",
      "Epoch 34/100\n",
      "586/586 [==============================] - 82s 138ms/step - loss: 0.1695 - accuracy: 0.9274 - val_loss: 0.1512 - val_accuracy: 0.9400\n",
      "Epoch 35/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1794 - accuracy: 0.9264 - val_loss: 0.1536 - val_accuracy: 0.9385\n",
      "Epoch 36/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1749 - accuracy: 0.9304 - val_loss: 0.1821 - val_accuracy: 0.9280\n",
      "Epoch 37/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1779 - accuracy: 0.9273 - val_loss: 0.1681 - val_accuracy: 0.9368\n",
      "Epoch 38/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1671 - accuracy: 0.9310 - val_loss: 0.1531 - val_accuracy: 0.9425\n",
      "Epoch 39/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1646 - accuracy: 0.9318 - val_loss: 0.1553 - val_accuracy: 0.9391\n",
      "Epoch 40/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1557 - accuracy: 0.9396 - val_loss: 0.1661 - val_accuracy: 0.9387\n",
      "Epoch 41/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1634 - accuracy: 0.9342 - val_loss: 0.1603 - val_accuracy: 0.9393\n",
      "Epoch 42/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1592 - accuracy: 0.9343 - val_loss: 0.1355 - val_accuracy: 0.9492\n",
      "Epoch 43/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1557 - accuracy: 0.9371 - val_loss: 0.1563 - val_accuracy: 0.9396\n",
      "Epoch 44/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1651 - accuracy: 0.9294 - val_loss: 0.1658 - val_accuracy: 0.9334\n",
      "Epoch 45/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1558 - accuracy: 0.9380 - val_loss: 0.1452 - val_accuracy: 0.9445\n",
      "Epoch 46/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1518 - accuracy: 0.9399 - val_loss: 0.1335 - val_accuracy: 0.9530\n",
      "Epoch 47/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1439 - accuracy: 0.9413 - val_loss: 0.1938 - val_accuracy: 0.9250\n",
      "Epoch 48/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1502 - accuracy: 0.9391 - val_loss: 0.2000 - val_accuracy: 0.9270\n",
      "Epoch 49/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1486 - accuracy: 0.9403 - val_loss: 0.1501 - val_accuracy: 0.9421\n",
      "Epoch 50/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.1440 - accuracy: 0.9410 - val_loss: 0.1469 - val_accuracy: 0.9464\n",
      "Epoch 51/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1524 - accuracy: 0.9400 - val_loss: 0.1393 - val_accuracy: 0.9460\n",
      "Epoch 52/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1452 - accuracy: 0.9397 - val_loss: 0.1387 - val_accuracy: 0.9507\n",
      "Epoch 53/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1442 - accuracy: 0.9402 - val_loss: 0.1418 - val_accuracy: 0.9472\n",
      "Epoch 54/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1334 - accuracy: 0.9455 - val_loss: 0.1378 - val_accuracy: 0.9485\n",
      "Epoch 55/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1369 - accuracy: 0.9457 - val_loss: 0.1625 - val_accuracy: 0.9385\n",
      "Epoch 56/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1440 - accuracy: 0.9424 - val_loss: 0.1396 - val_accuracy: 0.9434\n",
      "Epoch 57/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1363 - accuracy: 0.9462 - val_loss: 0.1539 - val_accuracy: 0.9432\n",
      "Epoch 58/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1428 - accuracy: 0.9403 - val_loss: 0.1400 - val_accuracy: 0.9451\n",
      "Epoch 59/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1367 - accuracy: 0.9451 - val_loss: 0.1501 - val_accuracy: 0.9398\n",
      "Epoch 60/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1318 - accuracy: 0.9484 - val_loss: 0.1534 - val_accuracy: 0.9415\n",
      "Epoch 61/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1283 - accuracy: 0.9478 - val_loss: 0.1663 - val_accuracy: 0.9396\n",
      "Epoch 62/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1337 - accuracy: 0.9459 - val_loss: 0.1584 - val_accuracy: 0.9404\n",
      "Epoch 63/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1316 - accuracy: 0.9466 - val_loss: 0.1665 - val_accuracy: 0.9485\n",
      "Epoch 64/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1309 - accuracy: 0.9484 - val_loss: 0.1318 - val_accuracy: 0.9534\n",
      "Epoch 65/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1448 - accuracy: 0.9417 - val_loss: 0.1458 - val_accuracy: 0.9500\n",
      "Epoch 66/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1244 - accuracy: 0.9490 - val_loss: 0.1508 - val_accuracy: 0.9451\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 67/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1311 - accuracy: 0.9466 - val_loss: 0.1627 - val_accuracy: 0.9361\n",
      "Epoch 68/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1285 - accuracy: 0.9474 - val_loss: 0.1516 - val_accuracy: 0.9466\n",
      "Epoch 69/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1287 - accuracy: 0.9462 - val_loss: 0.1396 - val_accuracy: 0.9449\n",
      "Epoch 70/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1285 - accuracy: 0.9480 - val_loss: 0.1395 - val_accuracy: 0.9462\n",
      "Epoch 71/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1264 - accuracy: 0.9484 - val_loss: 0.1389 - val_accuracy: 0.9487\n",
      "Epoch 72/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1339 - accuracy: 0.9444 - val_loss: 0.1443 - val_accuracy: 0.9530\n",
      "Epoch 73/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1195 - accuracy: 0.9520 - val_loss: 0.1284 - val_accuracy: 0.9475\n",
      "Epoch 74/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1198 - accuracy: 0.9534 - val_loss: 0.1332 - val_accuracy: 0.9515\n",
      "Epoch 75/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1337 - accuracy: 0.9468 - val_loss: 0.1360 - val_accuracy: 0.9513\n",
      "Epoch 76/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1124 - accuracy: 0.9547 - val_loss: 0.1668 - val_accuracy: 0.9451\n",
      "Epoch 77/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1234 - accuracy: 0.9489 - val_loss: 0.1388 - val_accuracy: 0.9481\n",
      "Epoch 78/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1199 - accuracy: 0.9520 - val_loss: 0.1323 - val_accuracy: 0.9545\n",
      "Epoch 79/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1215 - accuracy: 0.9521 - val_loss: 0.1847 - val_accuracy: 0.9479\n",
      "Epoch 80/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1218 - accuracy: 0.9538 - val_loss: 0.1370 - val_accuracy: 0.9554\n",
      "Epoch 81/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1217 - accuracy: 0.9527 - val_loss: 0.1318 - val_accuracy: 0.9530\n",
      "Epoch 82/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1145 - accuracy: 0.9544 - val_loss: 0.1409 - val_accuracy: 0.9524\n",
      "Epoch 83/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1099 - accuracy: 0.9550 - val_loss: 0.1334 - val_accuracy: 0.9507\n",
      "Epoch 84/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1103 - accuracy: 0.9583 - val_loss: 0.1339 - val_accuracy: 0.9549\n",
      "Epoch 85/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1206 - accuracy: 0.9524 - val_loss: 0.1494 - val_accuracy: 0.9457\n",
      "Epoch 86/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1084 - accuracy: 0.9560 - val_loss: 0.1382 - val_accuracy: 0.9498\n",
      "Epoch 87/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1154 - accuracy: 0.9534 - val_loss: 0.1275 - val_accuracy: 0.9526\n",
      "Epoch 88/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.1088 - accuracy: 0.9539 - val_loss: 0.1400 - val_accuracy: 0.9519\n",
      "Epoch 89/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1068 - accuracy: 0.9572 - val_loss: 0.1357 - val_accuracy: 0.9530\n",
      "Epoch 90/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1063 - accuracy: 0.9560 - val_loss: 0.1262 - val_accuracy: 0.9551\n",
      "Epoch 91/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1141 - accuracy: 0.9553 - val_loss: 0.1345 - val_accuracy: 0.9534\n",
      "Epoch 92/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1107 - accuracy: 0.9565 - val_loss: 0.1431 - val_accuracy: 0.9532\n",
      "Epoch 93/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1145 - accuracy: 0.9552 - val_loss: 0.1421 - val_accuracy: 0.9440\n",
      "Epoch 94/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1028 - accuracy: 0.9598 - val_loss: 0.1349 - val_accuracy: 0.9558\n",
      "Epoch 95/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1098 - accuracy: 0.9556 - val_loss: 0.1446 - val_accuracy: 0.9541\n",
      "Epoch 96/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1085 - accuracy: 0.9569 - val_loss: 0.1335 - val_accuracy: 0.9541\n",
      "Epoch 97/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1120 - accuracy: 0.9547 - val_loss: 0.1586 - val_accuracy: 0.9462\n",
      "Epoch 98/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1042 - accuracy: 0.9574 - val_loss: 0.1318 - val_accuracy: 0.9537\n",
      "Epoch 99/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1037 - accuracy: 0.9601 - val_loss: 0.1376 - val_accuracy: 0.9537\n",
      "Epoch 100/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1069 - accuracy: 0.9571 - val_loss: 0.1348 - val_accuracy: 0.9566\n",
      "--------------- trial #2 ---------------\n",
      "Epoch 1/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.6869 - accuracy: 0.5348 - val_loss: 0.6699 - val_accuracy: 0.6249\n",
      "Epoch 2/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.6674 - accuracy: 0.5957 - val_loss: 0.6525 - val_accuracy: 0.5927\n",
      "Epoch 3/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.6320 - accuracy: 0.6495 - val_loss: 0.5917 - val_accuracy: 0.6948\n",
      "Epoch 4/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.6002 - accuracy: 0.6843 - val_loss: 0.5602 - val_accuracy: 0.7202\n",
      "Epoch 5/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.5706 - accuracy: 0.7064 - val_loss: 0.5811 - val_accuracy: 0.6760\n",
      "Epoch 6/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.5521 - accuracy: 0.7252 - val_loss: 0.5101 - val_accuracy: 0.7657\n",
      "Epoch 7/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.5293 - accuracy: 0.7425 - val_loss: 0.5163 - val_accuracy: 0.7518\n",
      "Epoch 8/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.5043 - accuracy: 0.7568 - val_loss: 0.4435 - val_accuracy: 0.8069\n",
      "Epoch 9/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.4620 - accuracy: 0.7866 - val_loss: 0.3990 - val_accuracy: 0.8304\n",
      "Epoch 10/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.4347 - accuracy: 0.8064 - val_loss: 0.3787 - val_accuracy: 0.8372\n",
      "Epoch 11/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.3921 - accuracy: 0.8234 - val_loss: 0.3462 - val_accuracy: 0.8424\n",
      "Epoch 12/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.3647 - accuracy: 0.8405 - val_loss: 0.3090 - val_accuracy: 0.8584\n",
      "Epoch 13/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.3336 - accuracy: 0.8542 - val_loss: 0.2963 - val_accuracy: 0.8748\n",
      "Epoch 14/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.3212 - accuracy: 0.8608 - val_loss: 0.2819 - val_accuracy: 0.8755\n",
      "Epoch 15/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.3117 - accuracy: 0.8671 - val_loss: 0.2627 - val_accuracy: 0.8825\n",
      "Epoch 16/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2930 - accuracy: 0.8732 - val_loss: 0.2377 - val_accuracy: 0.9037\n",
      "Epoch 17/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.2821 - accuracy: 0.8814 - val_loss: 0.2470 - val_accuracy: 0.8966\n",
      "Epoch 18/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2740 - accuracy: 0.8838 - val_loss: 0.2637 - val_accuracy: 0.9011\n",
      "Epoch 19/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2703 - accuracy: 0.8848 - val_loss: 0.2179 - val_accuracy: 0.9103\n",
      "Epoch 20/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2547 - accuracy: 0.8930 - val_loss: 0.2967 - val_accuracy: 0.8723\n",
      "Epoch 21/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2502 - accuracy: 0.8910 - val_loss: 0.2097 - val_accuracy: 0.9148\n",
      "Epoch 22/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2373 - accuracy: 0.9004 - val_loss: 0.1998 - val_accuracy: 0.9154\n",
      "Epoch 23/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2449 - accuracy: 0.8999 - val_loss: 0.1879 - val_accuracy: 0.9216\n",
      "Epoch 24/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2283 - accuracy: 0.9061 - val_loss: 0.1803 - val_accuracy: 0.9233\n",
      "Epoch 25/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2198 - accuracy: 0.9106 - val_loss: 0.2292 - val_accuracy: 0.8998\n",
      "Epoch 26/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2181 - accuracy: 0.9066 - val_loss: 0.2335 - val_accuracy: 0.9018\n",
      "Epoch 27/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2148 - accuracy: 0.9128 - val_loss: 0.1831 - val_accuracy: 0.9212\n",
      "Epoch 28/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2040 - accuracy: 0.9147 - val_loss: 0.1639 - val_accuracy: 0.9346\n",
      "Epoch 29/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1931 - accuracy: 0.9208 - val_loss: 0.1786 - val_accuracy: 0.9259\n",
      "Epoch 30/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2002 - accuracy: 0.9179 - val_loss: 0.1733 - val_accuracy: 0.9278\n",
      "Epoch 31/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1928 - accuracy: 0.9214 - val_loss: 0.1671 - val_accuracy: 0.9302\n",
      "Epoch 32/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1898 - accuracy: 0.9241 - val_loss: 0.1582 - val_accuracy: 0.9357\n",
      "Epoch 33/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1892 - accuracy: 0.9239 - val_loss: 0.1561 - val_accuracy: 0.9344\n",
      "Epoch 34/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1848 - accuracy: 0.9219 - val_loss: 0.1777 - val_accuracy: 0.9233\n",
      "Epoch 35/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1813 - accuracy: 0.9246 - val_loss: 0.1642 - val_accuracy: 0.9366\n",
      "Epoch 36/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1810 - accuracy: 0.9244 - val_loss: 0.1588 - val_accuracy: 0.9319\n",
      "Epoch 37/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1708 - accuracy: 0.9293 - val_loss: 0.1664 - val_accuracy: 0.9361\n",
      "Epoch 38/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1684 - accuracy: 0.9304 - val_loss: 0.1661 - val_accuracy: 0.9306\n",
      "Epoch 39/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1746 - accuracy: 0.9250 - val_loss: 0.1565 - val_accuracy: 0.9376\n",
      "Epoch 40/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1600 - accuracy: 0.9366 - val_loss: 0.1579 - val_accuracy: 0.9340\n",
      "Epoch 41/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1697 - accuracy: 0.9300 - val_loss: 0.1500 - val_accuracy: 0.9383\n",
      "Epoch 42/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1630 - accuracy: 0.9318 - val_loss: 0.1442 - val_accuracy: 0.9404\n",
      "Epoch 43/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1682 - accuracy: 0.9284 - val_loss: 0.1531 - val_accuracy: 0.9425\n",
      "Epoch 44/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.1656 - accuracy: 0.9304 - val_loss: 0.1359 - val_accuracy: 0.9462\n",
      "Epoch 45/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1553 - accuracy: 0.9394 - val_loss: 0.1530 - val_accuracy: 0.9406\n",
      "Epoch 46/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1484 - accuracy: 0.9396 - val_loss: 0.1657 - val_accuracy: 0.9374\n",
      "Epoch 47/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1546 - accuracy: 0.9346 - val_loss: 0.1439 - val_accuracy: 0.9432\n",
      "Epoch 48/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1483 - accuracy: 0.9376 - val_loss: 0.1529 - val_accuracy: 0.9383\n",
      "Epoch 49/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1542 - accuracy: 0.9373 - val_loss: 0.1457 - val_accuracy: 0.9453\n",
      "Epoch 50/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1466 - accuracy: 0.9416 - val_loss: 0.1633 - val_accuracy: 0.9423\n",
      "Epoch 51/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1530 - accuracy: 0.9393 - val_loss: 0.1407 - val_accuracy: 0.9481\n",
      "Epoch 52/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.1486 - accuracy: 0.9399 - val_loss: 0.1888 - val_accuracy: 0.9295\n",
      "Epoch 53/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1525 - accuracy: 0.9385 - val_loss: 0.1961 - val_accuracy: 0.9137\n",
      "Epoch 54/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1499 - accuracy: 0.9391 - val_loss: 0.1370 - val_accuracy: 0.9462\n",
      "Epoch 55/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1517 - accuracy: 0.9384 - val_loss: 0.1390 - val_accuracy: 0.9487\n",
      "Epoch 56/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1502 - accuracy: 0.9395 - val_loss: 0.1304 - val_accuracy: 0.9502\n",
      "Epoch 57/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1417 - accuracy: 0.9393 - val_loss: 0.1428 - val_accuracy: 0.9466\n",
      "Epoch 58/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1482 - accuracy: 0.9391 - val_loss: 0.1649 - val_accuracy: 0.9355\n",
      "Epoch 59/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1307 - accuracy: 0.9461 - val_loss: 0.1828 - val_accuracy: 0.9216\n",
      "Epoch 60/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1327 - accuracy: 0.9450 - val_loss: 0.1415 - val_accuracy: 0.9438\n",
      "Epoch 61/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.1437 - accuracy: 0.9389 - val_loss: 0.1326 - val_accuracy: 0.9532\n",
      "Epoch 62/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1414 - accuracy: 0.9412 - val_loss: 0.1551 - val_accuracy: 0.9425\n",
      "Epoch 63/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1409 - accuracy: 0.9431 - val_loss: 0.1407 - val_accuracy: 0.9445\n",
      "Epoch 64/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1358 - accuracy: 0.9443 - val_loss: 0.1534 - val_accuracy: 0.9451\n",
      "Epoch 65/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1386 - accuracy: 0.9409 - val_loss: 0.1678 - val_accuracy: 0.9383\n",
      "Epoch 66/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1337 - accuracy: 0.9456 - val_loss: 0.1505 - val_accuracy: 0.9453\n",
      "Epoch 67/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1275 - accuracy: 0.9457 - val_loss: 0.1530 - val_accuracy: 0.9415\n",
      "Epoch 68/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1283 - accuracy: 0.9473 - val_loss: 0.1384 - val_accuracy: 0.9496\n",
      "Epoch 69/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1311 - accuracy: 0.9486 - val_loss: 0.1772 - val_accuracy: 0.9404\n",
      "Epoch 70/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1255 - accuracy: 0.9502 - val_loss: 0.1431 - val_accuracy: 0.9519\n",
      "Epoch 71/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1311 - accuracy: 0.9483 - val_loss: 0.1319 - val_accuracy: 0.9511\n",
      "Epoch 72/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1251 - accuracy: 0.9470 - val_loss: 0.1444 - val_accuracy: 0.9528\n",
      "Epoch 73/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1265 - accuracy: 0.9480 - val_loss: 0.1928 - val_accuracy: 0.9440\n",
      "Epoch 74/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1227 - accuracy: 0.9497 - val_loss: 0.1505 - val_accuracy: 0.9470\n",
      "Epoch 75/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1285 - accuracy: 0.9466 - val_loss: 0.1500 - val_accuracy: 0.9464\n",
      "Epoch 76/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1211 - accuracy: 0.9504 - val_loss: 0.1572 - val_accuracy: 0.9430\n",
      "Epoch 77/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1239 - accuracy: 0.9509 - val_loss: 0.1567 - val_accuracy: 0.9421\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 78/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1194 - accuracy: 0.9522 - val_loss: 0.1528 - val_accuracy: 0.9494\n",
      "Epoch 79/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1247 - accuracy: 0.9493 - val_loss: 0.1518 - val_accuracy: 0.9470\n",
      "Epoch 80/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1200 - accuracy: 0.9513 - val_loss: 0.1629 - val_accuracy: 0.9462\n",
      "Epoch 81/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1176 - accuracy: 0.9526 - val_loss: 0.1372 - val_accuracy: 0.9515\n",
      "Epoch 82/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1209 - accuracy: 0.9525 - val_loss: 0.1295 - val_accuracy: 0.9483\n",
      "Epoch 83/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1173 - accuracy: 0.9532 - val_loss: 0.1546 - val_accuracy: 0.9524\n",
      "Epoch 84/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1176 - accuracy: 0.9529 - val_loss: 0.1488 - val_accuracy: 0.9511\n",
      "Epoch 85/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1181 - accuracy: 0.9556 - val_loss: 0.1337 - val_accuracy: 0.9541\n",
      "Epoch 86/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1181 - accuracy: 0.9546 - val_loss: 0.1571 - val_accuracy: 0.9496\n",
      "Epoch 87/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1195 - accuracy: 0.9498 - val_loss: 0.1528 - val_accuracy: 0.9534\n",
      "Epoch 88/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1211 - accuracy: 0.9548 - val_loss: 0.1398 - val_accuracy: 0.9487\n",
      "Epoch 89/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1140 - accuracy: 0.9563 - val_loss: 0.1730 - val_accuracy: 0.9438\n",
      "Epoch 90/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1120 - accuracy: 0.9560 - val_loss: 0.1367 - val_accuracy: 0.9545\n",
      "Epoch 91/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1116 - accuracy: 0.9555 - val_loss: 0.1731 - val_accuracy: 0.9387\n",
      "Epoch 92/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1150 - accuracy: 0.9527 - val_loss: 0.1552 - val_accuracy: 0.9455\n",
      "Epoch 93/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1137 - accuracy: 0.9533 - val_loss: 0.2203 - val_accuracy: 0.9374\n",
      "Epoch 94/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1063 - accuracy: 0.9563 - val_loss: 0.1368 - val_accuracy: 0.9517\n",
      "Epoch 95/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1166 - accuracy: 0.9524 - val_loss: 0.1564 - val_accuracy: 0.9468\n",
      "Epoch 96/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1083 - accuracy: 0.9569 - val_loss: 0.1522 - val_accuracy: 0.9378\n",
      "Epoch 97/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1139 - accuracy: 0.9533 - val_loss: 0.1358 - val_accuracy: 0.9558\n",
      "Epoch 98/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1064 - accuracy: 0.9572 - val_loss: 0.1302 - val_accuracy: 0.9569\n",
      "Epoch 99/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1016 - accuracy: 0.9604 - val_loss: 0.1424 - val_accuracy: 0.9517\n",
      "Epoch 100/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1077 - accuracy: 0.9545 - val_loss: 0.1841 - val_accuracy: 0.9447\n",
      "--------------- trial #3 ---------------\n",
      "Epoch 1/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.6897 - accuracy: 0.5201 - val_loss: 0.6602 - val_accuracy: 0.5921\n",
      "Epoch 2/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.6546 - accuracy: 0.6132 - val_loss: 0.6130 - val_accuracy: 0.6544\n",
      "Epoch 3/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.6149 - accuracy: 0.6634 - val_loss: 0.6031 - val_accuracy: 0.6589\n",
      "Epoch 4/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.5848 - accuracy: 0.6986 - val_loss: 0.5345 - val_accuracy: 0.7352\n",
      "Epoch 5/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.5563 - accuracy: 0.7215 - val_loss: 0.5211 - val_accuracy: 0.7576\n",
      "Epoch 6/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.5302 - accuracy: 0.7431 - val_loss: 0.4778 - val_accuracy: 0.7796\n",
      "Epoch 7/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.4975 - accuracy: 0.7659 - val_loss: 0.4501 - val_accuracy: 0.8003\n",
      "Epoch 8/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.4637 - accuracy: 0.7841 - val_loss: 0.4023 - val_accuracy: 0.8323\n",
      "Epoch 9/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.4221 - accuracy: 0.8128 - val_loss: 0.3731 - val_accuracy: 0.8419\n",
      "Epoch 10/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.4065 - accuracy: 0.8198 - val_loss: 0.3466 - val_accuracy: 0.8541\n",
      "Epoch 11/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.3709 - accuracy: 0.8350 - val_loss: 0.3767 - val_accuracy: 0.8298\n",
      "Epoch 12/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.3557 - accuracy: 0.8464 - val_loss: 0.3083 - val_accuracy: 0.8738\n",
      "Epoch 13/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.3328 - accuracy: 0.8605 - val_loss: 0.2929 - val_accuracy: 0.8761\n",
      "Epoch 14/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.3190 - accuracy: 0.8631 - val_loss: 0.2643 - val_accuracy: 0.8902\n",
      "Epoch 15/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2970 - accuracy: 0.8730 - val_loss: 0.2632 - val_accuracy: 0.8932\n",
      "Epoch 16/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.2890 - accuracy: 0.8759 - val_loss: 0.2926 - val_accuracy: 0.8791\n",
      "Epoch 17/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2752 - accuracy: 0.8812 - val_loss: 0.2527 - val_accuracy: 0.8953\n",
      "Epoch 18/100\n",
      "586/586 [==============================] - 82s 138ms/step - loss: 0.2718 - accuracy: 0.8819 - val_loss: 0.2600 - val_accuracy: 0.8938\n",
      "Epoch 19/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2613 - accuracy: 0.8877 - val_loss: 0.2402 - val_accuracy: 0.9000\n",
      "Epoch 20/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2450 - accuracy: 0.8946 - val_loss: 0.2096 - val_accuracy: 0.9141\n",
      "Epoch 21/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2428 - accuracy: 0.8976 - val_loss: 0.2141 - val_accuracy: 0.9126\n",
      "Epoch 22/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2378 - accuracy: 0.9009 - val_loss: 0.2009 - val_accuracy: 0.9195\n",
      "Epoch 23/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2412 - accuracy: 0.9003 - val_loss: 0.2004 - val_accuracy: 0.9165\n",
      "Epoch 24/100\n",
      "586/586 [==============================] - 82s 138ms/step - loss: 0.2282 - accuracy: 0.9044 - val_loss: 0.2395 - val_accuracy: 0.9013\n",
      "Epoch 25/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2208 - accuracy: 0.9093 - val_loss: 0.2024 - val_accuracy: 0.9214\n",
      "Epoch 26/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2183 - accuracy: 0.9087 - val_loss: 0.1752 - val_accuracy: 0.9284\n",
      "Epoch 27/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2138 - accuracy: 0.9121 - val_loss: 0.1993 - val_accuracy: 0.9176\n",
      "Epoch 28/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2042 - accuracy: 0.9128 - val_loss: 0.2039 - val_accuracy: 0.9199\n",
      "Epoch 29/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1997 - accuracy: 0.9150 - val_loss: 0.1822 - val_accuracy: 0.9317\n",
      "Epoch 30/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2085 - accuracy: 0.9131 - val_loss: 0.1714 - val_accuracy: 0.9329\n",
      "Epoch 31/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1959 - accuracy: 0.9184 - val_loss: 0.2024 - val_accuracy: 0.9201\n",
      "Epoch 32/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1887 - accuracy: 0.9227 - val_loss: 0.1841 - val_accuracy: 0.9295\n",
      "Epoch 33/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1976 - accuracy: 0.9214 - val_loss: 0.1988 - val_accuracy: 0.9205\n",
      "Epoch 34/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1855 - accuracy: 0.9222 - val_loss: 0.1740 - val_accuracy: 0.9317\n",
      "Epoch 35/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1848 - accuracy: 0.9233 - val_loss: 0.2243 - val_accuracy: 0.9120\n",
      "Epoch 36/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1812 - accuracy: 0.9257 - val_loss: 0.1720 - val_accuracy: 0.9329\n",
      "Epoch 37/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1858 - accuracy: 0.9228 - val_loss: 0.1638 - val_accuracy: 0.9387\n",
      "Epoch 38/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1773 - accuracy: 0.9297 - val_loss: 0.1764 - val_accuracy: 0.9304\n",
      "Epoch 39/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1741 - accuracy: 0.9298 - val_loss: 0.1623 - val_accuracy: 0.9355\n",
      "Epoch 40/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1773 - accuracy: 0.9263 - val_loss: 0.1684 - val_accuracy: 0.9383\n",
      "Epoch 41/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1663 - accuracy: 0.9304 - val_loss: 0.1662 - val_accuracy: 0.9323\n",
      "Epoch 42/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1742 - accuracy: 0.9288 - val_loss: 0.1692 - val_accuracy: 0.9402\n",
      "Epoch 43/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1688 - accuracy: 0.9313 - val_loss: 0.1605 - val_accuracy: 0.9359\n",
      "Epoch 44/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1605 - accuracy: 0.9341 - val_loss: 0.1838 - val_accuracy: 0.9267\n",
      "Epoch 45/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1578 - accuracy: 0.9396 - val_loss: 0.1664 - val_accuracy: 0.9331\n",
      "Epoch 46/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1637 - accuracy: 0.9321 - val_loss: 0.1580 - val_accuracy: 0.9368\n",
      "Epoch 47/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1611 - accuracy: 0.9363 - val_loss: 0.1573 - val_accuracy: 0.9368\n",
      "Epoch 48/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1633 - accuracy: 0.9337 - val_loss: 0.1531 - val_accuracy: 0.9417\n",
      "Epoch 49/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1553 - accuracy: 0.9364 - val_loss: 0.1536 - val_accuracy: 0.9404\n",
      "Epoch 50/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1594 - accuracy: 0.9374 - val_loss: 0.1592 - val_accuracy: 0.9374\n",
      "Epoch 51/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1568 - accuracy: 0.9351 - val_loss: 0.1653 - val_accuracy: 0.9408\n",
      "Epoch 52/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1483 - accuracy: 0.9407 - val_loss: 0.1400 - val_accuracy: 0.9453\n",
      "Epoch 53/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1578 - accuracy: 0.9356 - val_loss: 0.1568 - val_accuracy: 0.9428\n",
      "Epoch 54/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1578 - accuracy: 0.9368 - val_loss: 0.1964 - val_accuracy: 0.9357\n",
      "Epoch 55/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1399 - accuracy: 0.9428 - val_loss: 0.1449 - val_accuracy: 0.9432\n",
      "Epoch 56/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1455 - accuracy: 0.9427 - val_loss: 0.1631 - val_accuracy: 0.9391\n",
      "Epoch 57/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1451 - accuracy: 0.9400 - val_loss: 0.1374 - val_accuracy: 0.9455\n",
      "Epoch 58/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1483 - accuracy: 0.9398 - val_loss: 0.1459 - val_accuracy: 0.9419\n",
      "Epoch 59/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1430 - accuracy: 0.9430 - val_loss: 0.1527 - val_accuracy: 0.9462\n",
      "Epoch 60/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1459 - accuracy: 0.9397 - val_loss: 0.1480 - val_accuracy: 0.9391\n",
      "Epoch 61/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1398 - accuracy: 0.9417 - val_loss: 0.1537 - val_accuracy: 0.9436\n",
      "Epoch 62/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1441 - accuracy: 0.9395 - val_loss: 0.2168 - val_accuracy: 0.9261\n",
      "Epoch 63/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1359 - accuracy: 0.9444 - val_loss: 0.1545 - val_accuracy: 0.9419\n",
      "Epoch 64/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1349 - accuracy: 0.9468 - val_loss: 0.1617 - val_accuracy: 0.9408\n",
      "Epoch 65/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1389 - accuracy: 0.9420 - val_loss: 0.1546 - val_accuracy: 0.9432\n",
      "Epoch 66/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1358 - accuracy: 0.9443 - val_loss: 0.1456 - val_accuracy: 0.9430\n",
      "Epoch 67/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1340 - accuracy: 0.9473 - val_loss: 0.1553 - val_accuracy: 0.9421\n",
      "Epoch 68/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1317 - accuracy: 0.9459 - val_loss: 0.2085 - val_accuracy: 0.9227\n",
      "Epoch 69/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1352 - accuracy: 0.9442 - val_loss: 0.1479 - val_accuracy: 0.9406\n",
      "Epoch 70/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1306 - accuracy: 0.9462 - val_loss: 0.1366 - val_accuracy: 0.9515\n",
      "Epoch 71/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1319 - accuracy: 0.9462 - val_loss: 0.1703 - val_accuracy: 0.9378\n",
      "Epoch 72/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1357 - accuracy: 0.9470 - val_loss: 0.1567 - val_accuracy: 0.9449\n",
      "Epoch 73/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1281 - accuracy: 0.9469 - val_loss: 0.1897 - val_accuracy: 0.9376\n",
      "Epoch 74/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1328 - accuracy: 0.9422 - val_loss: 0.1440 - val_accuracy: 0.9468\n",
      "Epoch 75/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1308 - accuracy: 0.9478 - val_loss: 0.1398 - val_accuracy: 0.9483\n",
      "Epoch 76/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1277 - accuracy: 0.9473 - val_loss: 0.1464 - val_accuracy: 0.9487\n",
      "Epoch 77/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1239 - accuracy: 0.9522 - val_loss: 0.1774 - val_accuracy: 0.9430\n",
      "Epoch 78/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1294 - accuracy: 0.9498 - val_loss: 0.1543 - val_accuracy: 0.9477\n",
      "Epoch 79/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1191 - accuracy: 0.9501 - val_loss: 0.1355 - val_accuracy: 0.9509\n",
      "Epoch 80/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1259 - accuracy: 0.9499 - val_loss: 0.1358 - val_accuracy: 0.9507\n",
      "Epoch 81/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1185 - accuracy: 0.9530 - val_loss: 0.1369 - val_accuracy: 0.9532\n",
      "Epoch 82/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1175 - accuracy: 0.9526 - val_loss: 0.1649 - val_accuracy: 0.9451\n",
      "Epoch 83/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1213 - accuracy: 0.9506 - val_loss: 0.1385 - val_accuracy: 0.9532\n",
      "Epoch 84/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1177 - accuracy: 0.9530 - val_loss: 0.1371 - val_accuracy: 0.9511\n",
      "Epoch 85/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1192 - accuracy: 0.9521 - val_loss: 0.1405 - val_accuracy: 0.9526\n",
      "Epoch 86/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1191 - accuracy: 0.9531 - val_loss: 0.1427 - val_accuracy: 0.9468\n",
      "Epoch 87/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1229 - accuracy: 0.9513 - val_loss: 0.1468 - val_accuracy: 0.9511\n",
      "Epoch 88/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1130 - accuracy: 0.9547 - val_loss: 0.1511 - val_accuracy: 0.9500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 89/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1171 - accuracy: 0.9554 - val_loss: 0.1662 - val_accuracy: 0.9423\n",
      "Epoch 90/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1136 - accuracy: 0.9563 - val_loss: 0.1397 - val_accuracy: 0.9504\n",
      "Epoch 91/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1181 - accuracy: 0.9495 - val_loss: 0.1451 - val_accuracy: 0.9515\n",
      "Epoch 92/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1097 - accuracy: 0.9547 - val_loss: 0.1433 - val_accuracy: 0.9443\n",
      "Epoch 93/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1114 - accuracy: 0.9538 - val_loss: 0.1362 - val_accuracy: 0.9539\n",
      "Epoch 94/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1120 - accuracy: 0.9561 - val_loss: 0.1390 - val_accuracy: 0.9526\n",
      "Epoch 95/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1188 - accuracy: 0.9530 - val_loss: 0.1420 - val_accuracy: 0.9526\n",
      "Epoch 96/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1106 - accuracy: 0.9530 - val_loss: 0.1577 - val_accuracy: 0.9443\n",
      "Epoch 97/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1127 - accuracy: 0.9551 - val_loss: 0.1366 - val_accuracy: 0.9470\n",
      "Epoch 98/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1199 - accuracy: 0.9544 - val_loss: 0.1519 - val_accuracy: 0.9464\n",
      "Epoch 99/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1112 - accuracy: 0.9558 - val_loss: 0.1658 - val_accuracy: 0.9470\n",
      "Epoch 100/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1071 - accuracy: 0.9585 - val_loss: 0.1241 - val_accuracy: 0.9511\n",
      "--------------- trial #4 ---------------\n",
      "Epoch 1/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.6873 - accuracy: 0.5272 - val_loss: 0.6689 - val_accuracy: 0.5720\n",
      "Epoch 2/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.6686 - accuracy: 0.5752 - val_loss: 0.6404 - val_accuracy: 0.6130\n",
      "Epoch 3/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.6431 - accuracy: 0.6093 - val_loss: 0.6115 - val_accuracy: 0.6719\n",
      "Epoch 4/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.6154 - accuracy: 0.6587 - val_loss: 0.5715 - val_accuracy: 0.7074\n",
      "Epoch 5/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.5873 - accuracy: 0.6929 - val_loss: 0.5532 - val_accuracy: 0.7144\n",
      "Epoch 6/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.5735 - accuracy: 0.7052 - val_loss: 0.5301 - val_accuracy: 0.7341\n",
      "Epoch 7/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.5533 - accuracy: 0.7208 - val_loss: 0.4968 - val_accuracy: 0.7663\n",
      "Epoch 8/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.5056 - accuracy: 0.7608 - val_loss: 0.4760 - val_accuracy: 0.7764\n",
      "Epoch 9/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.4899 - accuracy: 0.7701 - val_loss: 0.4297 - val_accuracy: 0.8086\n",
      "Epoch 10/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.4500 - accuracy: 0.8004 - val_loss: 0.3873 - val_accuracy: 0.8338\n",
      "Epoch 11/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.4231 - accuracy: 0.8116 - val_loss: 0.3554 - val_accuracy: 0.8447\n",
      "Epoch 12/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.3934 - accuracy: 0.8267 - val_loss: 0.3945 - val_accuracy: 0.8191\n",
      "Epoch 13/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.3633 - accuracy: 0.8410 - val_loss: 0.3019 - val_accuracy: 0.8695\n",
      "Epoch 14/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.3482 - accuracy: 0.8484 - val_loss: 0.4665 - val_accuracy: 0.7834\n",
      "Epoch 15/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.3233 - accuracy: 0.8623 - val_loss: 0.3012 - val_accuracy: 0.8678\n",
      "Epoch 16/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.3068 - accuracy: 0.8681 - val_loss: 0.2909 - val_accuracy: 0.8738\n",
      "Epoch 17/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2895 - accuracy: 0.8818 - val_loss: 0.2537 - val_accuracy: 0.8936\n",
      "Epoch 18/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2791 - accuracy: 0.8804 - val_loss: 0.2303 - val_accuracy: 0.8996\n",
      "Epoch 19/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2644 - accuracy: 0.8855 - val_loss: 0.2149 - val_accuracy: 0.9071\n",
      "Epoch 20/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2523 - accuracy: 0.8971 - val_loss: 0.2272 - val_accuracy: 0.9045\n",
      "Epoch 21/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2472 - accuracy: 0.8969 - val_loss: 0.1934 - val_accuracy: 0.9199\n",
      "Epoch 22/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2399 - accuracy: 0.8989 - val_loss: 0.2153 - val_accuracy: 0.9050\n",
      "Epoch 23/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2271 - accuracy: 0.9035 - val_loss: 0.2329 - val_accuracy: 0.9082\n",
      "Epoch 24/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2260 - accuracy: 0.9047 - val_loss: 0.2058 - val_accuracy: 0.9120\n",
      "Epoch 25/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2277 - accuracy: 0.9012 - val_loss: 0.1718 - val_accuracy: 0.9329\n",
      "Epoch 26/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.2203 - accuracy: 0.9083 - val_loss: 0.2020 - val_accuracy: 0.9133\n",
      "Epoch 27/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2121 - accuracy: 0.9112 - val_loss: 0.2167 - val_accuracy: 0.9124\n",
      "Epoch 28/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2088 - accuracy: 0.9165 - val_loss: 0.1813 - val_accuracy: 0.9280\n",
      "Epoch 29/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2068 - accuracy: 0.9143 - val_loss: 0.1802 - val_accuracy: 0.9289\n",
      "Epoch 30/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1955 - accuracy: 0.9224 - val_loss: 0.1940 - val_accuracy: 0.9203\n",
      "Epoch 31/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1890 - accuracy: 0.9212 - val_loss: 0.1841 - val_accuracy: 0.9276\n",
      "Epoch 32/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1928 - accuracy: 0.9194 - val_loss: 0.1663 - val_accuracy: 0.9353\n",
      "Epoch 33/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1971 - accuracy: 0.9183 - val_loss: 0.1801 - val_accuracy: 0.9270\n",
      "Epoch 34/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1869 - accuracy: 0.9223 - val_loss: 0.1672 - val_accuracy: 0.9342\n",
      "Epoch 35/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1893 - accuracy: 0.9199 - val_loss: 0.1742 - val_accuracy: 0.9297\n",
      "Epoch 36/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1862 - accuracy: 0.9239 - val_loss: 0.1713 - val_accuracy: 0.9280\n",
      "Epoch 37/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1792 - accuracy: 0.9244 - val_loss: 0.1538 - val_accuracy: 0.9400\n",
      "Epoch 38/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1797 - accuracy: 0.9282 - val_loss: 0.1974 - val_accuracy: 0.9203\n",
      "Epoch 39/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1708 - accuracy: 0.9302 - val_loss: 0.1543 - val_accuracy: 0.9383\n",
      "Epoch 40/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1791 - accuracy: 0.9246 - val_loss: 0.1676 - val_accuracy: 0.9351\n",
      "Epoch 41/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1765 - accuracy: 0.9283 - val_loss: 0.1643 - val_accuracy: 0.9378\n",
      "Epoch 42/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1724 - accuracy: 0.9311 - val_loss: 0.1726 - val_accuracy: 0.9302\n",
      "Epoch 43/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1675 - accuracy: 0.9304 - val_loss: 0.1666 - val_accuracy: 0.9353\n",
      "Epoch 44/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1644 - accuracy: 0.9332 - val_loss: 0.1679 - val_accuracy: 0.9370\n",
      "Epoch 45/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1708 - accuracy: 0.9242 - val_loss: 0.1751 - val_accuracy: 0.9357\n",
      "Epoch 46/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1671 - accuracy: 0.9289 - val_loss: 0.1575 - val_accuracy: 0.9419\n",
      "Epoch 47/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1581 - accuracy: 0.9346 - val_loss: 0.1665 - val_accuracy: 0.9370\n",
      "Epoch 48/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1601 - accuracy: 0.9367 - val_loss: 0.1523 - val_accuracy: 0.9391\n",
      "Epoch 49/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1597 - accuracy: 0.9334 - val_loss: 0.1821 - val_accuracy: 0.9306\n",
      "Epoch 50/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1556 - accuracy: 0.9367 - val_loss: 0.1487 - val_accuracy: 0.9451\n",
      "Epoch 51/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1515 - accuracy: 0.9381 - val_loss: 0.1595 - val_accuracy: 0.9334\n",
      "Epoch 52/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1539 - accuracy: 0.9360 - val_loss: 0.1376 - val_accuracy: 0.9481\n",
      "Epoch 53/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1507 - accuracy: 0.9394 - val_loss: 0.1469 - val_accuracy: 0.9453\n",
      "Epoch 54/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1448 - accuracy: 0.9394 - val_loss: 0.1524 - val_accuracy: 0.9421\n",
      "Epoch 55/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1459 - accuracy: 0.9374 - val_loss: 0.1374 - val_accuracy: 0.9457\n",
      "Epoch 56/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1473 - accuracy: 0.9389 - val_loss: 0.1357 - val_accuracy: 0.9481\n",
      "Epoch 57/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1508 - accuracy: 0.9402 - val_loss: 0.1549 - val_accuracy: 0.9396\n",
      "Epoch 58/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1473 - accuracy: 0.9373 - val_loss: 0.1568 - val_accuracy: 0.9455\n",
      "Epoch 59/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1442 - accuracy: 0.9409 - val_loss: 0.2094 - val_accuracy: 0.9280\n",
      "Epoch 60/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1457 - accuracy: 0.9404 - val_loss: 0.1414 - val_accuracy: 0.9438\n",
      "Epoch 61/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1360 - accuracy: 0.9446 - val_loss: 0.1559 - val_accuracy: 0.9408\n",
      "Epoch 62/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1416 - accuracy: 0.9413 - val_loss: 0.1563 - val_accuracy: 0.9415\n",
      "Epoch 63/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1380 - accuracy: 0.9429 - val_loss: 0.1628 - val_accuracy: 0.9400\n",
      "Epoch 64/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1386 - accuracy: 0.9444 - val_loss: 0.1539 - val_accuracy: 0.9408\n",
      "Epoch 65/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1380 - accuracy: 0.9412 - val_loss: 0.1484 - val_accuracy: 0.9370\n",
      "Epoch 66/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1464 - accuracy: 0.9384 - val_loss: 0.1948 - val_accuracy: 0.9387\n",
      "Epoch 67/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1337 - accuracy: 0.9450 - val_loss: 0.1344 - val_accuracy: 0.9468\n",
      "Epoch 68/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1361 - accuracy: 0.9462 - val_loss: 0.1556 - val_accuracy: 0.9464\n",
      "Epoch 69/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1283 - accuracy: 0.9496 - val_loss: 0.1552 - val_accuracy: 0.9438\n",
      "Epoch 70/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.1316 - accuracy: 0.9457 - val_loss: 0.1504 - val_accuracy: 0.9485\n",
      "Epoch 71/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1233 - accuracy: 0.9479 - val_loss: 0.1471 - val_accuracy: 0.9423\n",
      "Epoch 72/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1296 - accuracy: 0.9469 - val_loss: 0.1739 - val_accuracy: 0.9331\n",
      "Epoch 73/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1332 - accuracy: 0.9444 - val_loss: 0.1888 - val_accuracy: 0.9400\n",
      "Epoch 74/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1230 - accuracy: 0.9509 - val_loss: 0.1582 - val_accuracy: 0.9498\n",
      "Epoch 75/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1223 - accuracy: 0.9515 - val_loss: 0.1405 - val_accuracy: 0.9490\n",
      "Epoch 76/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1242 - accuracy: 0.9506 - val_loss: 0.1736 - val_accuracy: 0.9398\n",
      "Epoch 77/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1147 - accuracy: 0.9542 - val_loss: 0.2136 - val_accuracy: 0.9389\n",
      "Epoch 78/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1242 - accuracy: 0.9480 - val_loss: 0.2026 - val_accuracy: 0.9308\n",
      "Epoch 79/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1223 - accuracy: 0.9520 - val_loss: 0.1407 - val_accuracy: 0.9472\n",
      "Epoch 80/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1170 - accuracy: 0.9540 - val_loss: 0.1544 - val_accuracy: 0.9383\n",
      "Epoch 81/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1250 - accuracy: 0.9467 - val_loss: 0.1352 - val_accuracy: 0.9530\n",
      "Epoch 82/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1187 - accuracy: 0.9502 - val_loss: 0.1365 - val_accuracy: 0.9522\n",
      "Epoch 83/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1222 - accuracy: 0.9506 - val_loss: 0.1600 - val_accuracy: 0.9453\n",
      "Epoch 84/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1253 - accuracy: 0.9505 - val_loss: 0.1555 - val_accuracy: 0.9475\n",
      "Epoch 85/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1213 - accuracy: 0.9505 - val_loss: 0.1269 - val_accuracy: 0.9494\n",
      "Epoch 86/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1218 - accuracy: 0.9501 - val_loss: 0.1391 - val_accuracy: 0.9539\n",
      "Epoch 87/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1206 - accuracy: 0.9518 - val_loss: 0.1529 - val_accuracy: 0.9479\n",
      "Epoch 88/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1196 - accuracy: 0.9514 - val_loss: 0.1216 - val_accuracy: 0.9566\n",
      "Epoch 89/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1250 - accuracy: 0.9480 - val_loss: 0.1538 - val_accuracy: 0.9507\n",
      "Epoch 90/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.1218 - accuracy: 0.9515 - val_loss: 0.1460 - val_accuracy: 0.9487\n",
      "Epoch 91/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1190 - accuracy: 0.9511 - val_loss: 0.2091 - val_accuracy: 0.9415\n",
      "Epoch 92/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1281 - accuracy: 0.9476 - val_loss: 0.1288 - val_accuracy: 0.9569\n",
      "Epoch 93/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1201 - accuracy: 0.9510 - val_loss: 0.1418 - val_accuracy: 0.9543\n",
      "Epoch 94/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1149 - accuracy: 0.9531 - val_loss: 0.1394 - val_accuracy: 0.9500\n",
      "Epoch 95/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1182 - accuracy: 0.9510 - val_loss: 0.1391 - val_accuracy: 0.9519\n",
      "Epoch 96/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1178 - accuracy: 0.9548 - val_loss: 0.1604 - val_accuracy: 0.9470\n",
      "Epoch 97/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1076 - accuracy: 0.9568 - val_loss: 0.1240 - val_accuracy: 0.9509\n",
      "Epoch 98/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1181 - accuracy: 0.9531 - val_loss: 0.1363 - val_accuracy: 0.9500\n",
      "Epoch 99/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1166 - accuracy: 0.9546 - val_loss: 0.1373 - val_accuracy: 0.9588\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 100/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1133 - accuracy: 0.9532 - val_loss: 0.1446 - val_accuracy: 0.9517\n",
      "--------------- trial #5 ---------------\n",
      "Epoch 1/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.6811 - accuracy: 0.5542 - val_loss: 0.6180 - val_accuracy: 0.6593\n",
      "Epoch 2/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.6447 - accuracy: 0.6378 - val_loss: 0.5853 - val_accuracy: 0.7025\n",
      "Epoch 3/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.6027 - accuracy: 0.6763 - val_loss: 0.6316 - val_accuracy: 0.6604\n",
      "Epoch 4/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.5802 - accuracy: 0.6960 - val_loss: 0.5178 - val_accuracy: 0.7576\n",
      "Epoch 5/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.5397 - accuracy: 0.7341 - val_loss: 0.4827 - val_accuracy: 0.7779\n",
      "Epoch 6/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.5186 - accuracy: 0.7483 - val_loss: 0.4495 - val_accuracy: 0.7960\n",
      "Epoch 7/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.4760 - accuracy: 0.7792 - val_loss: 0.4035 - val_accuracy: 0.8281\n",
      "Epoch 8/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.4392 - accuracy: 0.8037 - val_loss: 0.4355 - val_accuracy: 0.7973\n",
      "Epoch 9/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.4204 - accuracy: 0.8101 - val_loss: 0.3965 - val_accuracy: 0.8332\n",
      "Epoch 10/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.3751 - accuracy: 0.8347 - val_loss: 0.3664 - val_accuracy: 0.8422\n",
      "Epoch 11/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.3330 - accuracy: 0.8535 - val_loss: 0.2942 - val_accuracy: 0.8757\n",
      "Epoch 12/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.3143 - accuracy: 0.8668 - val_loss: 0.2450 - val_accuracy: 0.8985\n",
      "Epoch 13/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.2923 - accuracy: 0.8760 - val_loss: 0.2505 - val_accuracy: 0.8966\n",
      "Epoch 14/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2758 - accuracy: 0.8843 - val_loss: 0.2280 - val_accuracy: 0.9075\n",
      "Epoch 15/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2643 - accuracy: 0.8909 - val_loss: 0.2299 - val_accuracy: 0.9067\n",
      "Epoch 16/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2549 - accuracy: 0.8971 - val_loss: 0.2464 - val_accuracy: 0.9007\n",
      "Epoch 17/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2442 - accuracy: 0.8973 - val_loss: 0.2889 - val_accuracy: 0.8783\n",
      "Epoch 18/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.2437 - accuracy: 0.8991 - val_loss: 0.3963 - val_accuracy: 0.8507\n",
      "Epoch 19/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2513 - accuracy: 0.8944 - val_loss: 0.2051 - val_accuracy: 0.9195\n",
      "Epoch 20/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2191 - accuracy: 0.9081 - val_loss: 0.1871 - val_accuracy: 0.9248\n",
      "Epoch 21/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2208 - accuracy: 0.9071 - val_loss: 0.1896 - val_accuracy: 0.9233\n",
      "Epoch 22/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2192 - accuracy: 0.9095 - val_loss: 0.1982 - val_accuracy: 0.9199\n",
      "Epoch 23/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2070 - accuracy: 0.9127 - val_loss: 0.1822 - val_accuracy: 0.9255\n",
      "Epoch 24/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2049 - accuracy: 0.9118 - val_loss: 0.1774 - val_accuracy: 0.9276\n",
      "Epoch 25/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2071 - accuracy: 0.9154 - val_loss: 0.1816 - val_accuracy: 0.9299\n",
      "Epoch 26/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1878 - accuracy: 0.9240 - val_loss: 0.1784 - val_accuracy: 0.9274\n",
      "Epoch 27/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1881 - accuracy: 0.9217 - val_loss: 0.1696 - val_accuracy: 0.9291\n",
      "Epoch 28/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1868 - accuracy: 0.9247 - val_loss: 0.1548 - val_accuracy: 0.9406\n",
      "Epoch 29/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1827 - accuracy: 0.9264 - val_loss: 0.1629 - val_accuracy: 0.9357\n",
      "Epoch 30/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1826 - accuracy: 0.9263 - val_loss: 0.1495 - val_accuracy: 0.9406\n",
      "Epoch 31/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1667 - accuracy: 0.9300 - val_loss: 0.1524 - val_accuracy: 0.9451\n",
      "Epoch 32/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1737 - accuracy: 0.9286 - val_loss: 0.1535 - val_accuracy: 0.9440\n",
      "Epoch 33/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1713 - accuracy: 0.9311 - val_loss: 0.1585 - val_accuracy: 0.9374\n",
      "Epoch 34/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1750 - accuracy: 0.9284 - val_loss: 0.1674 - val_accuracy: 0.9351\n",
      "Epoch 35/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1614 - accuracy: 0.9358 - val_loss: 0.1594 - val_accuracy: 0.9378\n",
      "Epoch 36/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1619 - accuracy: 0.9325 - val_loss: 0.1518 - val_accuracy: 0.9391\n",
      "Epoch 37/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1607 - accuracy: 0.9355 - val_loss: 0.1603 - val_accuracy: 0.9374\n",
      "Epoch 38/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1688 - accuracy: 0.9319 - val_loss: 0.1462 - val_accuracy: 0.9423\n",
      "Epoch 39/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1574 - accuracy: 0.9348 - val_loss: 0.1518 - val_accuracy: 0.9453\n",
      "Epoch 40/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1612 - accuracy: 0.9362 - val_loss: 0.1594 - val_accuracy: 0.9391\n",
      "Epoch 41/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1570 - accuracy: 0.9353 - val_loss: 0.1588 - val_accuracy: 0.9383\n",
      "Epoch 42/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1513 - accuracy: 0.9364 - val_loss: 0.1453 - val_accuracy: 0.9462\n",
      "Epoch 43/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1458 - accuracy: 0.9424 - val_loss: 0.1448 - val_accuracy: 0.9466\n",
      "Epoch 44/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1469 - accuracy: 0.9397 - val_loss: 0.1361 - val_accuracy: 0.9509\n",
      "Epoch 45/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1468 - accuracy: 0.9413 - val_loss: 0.1418 - val_accuracy: 0.9492\n",
      "Epoch 46/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1406 - accuracy: 0.9440 - val_loss: 0.1481 - val_accuracy: 0.9462\n",
      "Epoch 47/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1409 - accuracy: 0.9444 - val_loss: 0.1487 - val_accuracy: 0.9483\n",
      "Epoch 48/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1388 - accuracy: 0.9442 - val_loss: 0.1498 - val_accuracy: 0.9507\n",
      "Epoch 49/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1413 - accuracy: 0.9424 - val_loss: 0.1744 - val_accuracy: 0.9385\n",
      "Epoch 50/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1416 - accuracy: 0.9431 - val_loss: 0.1494 - val_accuracy: 0.9470\n",
      "Epoch 51/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1359 - accuracy: 0.9464 - val_loss: 0.1445 - val_accuracy: 0.9519\n",
      "Epoch 52/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1305 - accuracy: 0.9494 - val_loss: 0.1375 - val_accuracy: 0.9485\n",
      "Epoch 53/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1331 - accuracy: 0.9452 - val_loss: 0.1412 - val_accuracy: 0.9472\n",
      "Epoch 54/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1317 - accuracy: 0.9469 - val_loss: 0.1368 - val_accuracy: 0.9522\n",
      "Epoch 55/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1387 - accuracy: 0.9420 - val_loss: 0.1407 - val_accuracy: 0.9462\n",
      "Epoch 56/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1382 - accuracy: 0.9427 - val_loss: 0.1353 - val_accuracy: 0.9513\n",
      "Epoch 57/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1281 - accuracy: 0.9502 - val_loss: 0.1430 - val_accuracy: 0.9502\n",
      "Epoch 58/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1292 - accuracy: 0.9455 - val_loss: 0.1884 - val_accuracy: 0.9385\n",
      "Epoch 59/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1313 - accuracy: 0.9455 - val_loss: 0.1437 - val_accuracy: 0.9470\n",
      "Epoch 60/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1324 - accuracy: 0.9482 - val_loss: 0.1515 - val_accuracy: 0.9519\n",
      "Epoch 61/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1270 - accuracy: 0.9491 - val_loss: 0.2181 - val_accuracy: 0.9252\n",
      "Epoch 62/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1211 - accuracy: 0.9517 - val_loss: 0.1287 - val_accuracy: 0.9545\n",
      "Epoch 63/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1235 - accuracy: 0.9493 - val_loss: 0.1572 - val_accuracy: 0.9509\n",
      "Epoch 64/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1209 - accuracy: 0.9507 - val_loss: 0.1479 - val_accuracy: 0.9504\n",
      "Epoch 65/100\n",
      "586/586 [==============================] - 82s 140ms/step - loss: 0.1205 - accuracy: 0.9532 - val_loss: 0.1488 - val_accuracy: 0.9522\n",
      "Epoch 66/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1211 - accuracy: 0.9509 - val_loss: 0.1426 - val_accuracy: 0.9475\n",
      "Epoch 67/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1243 - accuracy: 0.9503 - val_loss: 0.1417 - val_accuracy: 0.9496\n",
      "Epoch 68/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1201 - accuracy: 0.9511 - val_loss: 0.1362 - val_accuracy: 0.9509\n",
      "Epoch 69/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1203 - accuracy: 0.9507 - val_loss: 0.1354 - val_accuracy: 0.9537\n",
      "Epoch 70/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1159 - accuracy: 0.9526 - val_loss: 0.1446 - val_accuracy: 0.9519\n",
      "Epoch 71/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1171 - accuracy: 0.9531 - val_loss: 0.1431 - val_accuracy: 0.9519\n",
      "Epoch 72/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1189 - accuracy: 0.9548 - val_loss: 0.1334 - val_accuracy: 0.9569\n",
      "Epoch 73/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1112 - accuracy: 0.9540 - val_loss: 0.1358 - val_accuracy: 0.9534\n",
      "Epoch 74/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1093 - accuracy: 0.9575 - val_loss: 0.1435 - val_accuracy: 0.9566\n",
      "Epoch 75/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1096 - accuracy: 0.9538 - val_loss: 0.1332 - val_accuracy: 0.9549\n",
      "Epoch 76/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1128 - accuracy: 0.9552 - val_loss: 0.1379 - val_accuracy: 0.9496\n",
      "Epoch 77/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1077 - accuracy: 0.9554 - val_loss: 0.1510 - val_accuracy: 0.9479\n",
      "Epoch 78/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1089 - accuracy: 0.9562 - val_loss: 0.1423 - val_accuracy: 0.9438\n",
      "Epoch 79/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1152 - accuracy: 0.9543 - val_loss: 0.1351 - val_accuracy: 0.9522\n",
      "Epoch 80/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1052 - accuracy: 0.9576 - val_loss: 0.1419 - val_accuracy: 0.9566\n",
      "Epoch 81/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1081 - accuracy: 0.9567 - val_loss: 0.1361 - val_accuracy: 0.9554\n",
      "Epoch 82/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1090 - accuracy: 0.9557 - val_loss: 0.1333 - val_accuracy: 0.9562\n",
      "Epoch 83/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1069 - accuracy: 0.9574 - val_loss: 0.1345 - val_accuracy: 0.9537\n",
      "Epoch 84/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1066 - accuracy: 0.9564 - val_loss: 0.1440 - val_accuracy: 0.9513\n",
      "Epoch 85/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1064 - accuracy: 0.9552 - val_loss: 0.1291 - val_accuracy: 0.9564\n",
      "Epoch 86/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1114 - accuracy: 0.9539 - val_loss: 0.1319 - val_accuracy: 0.9556\n",
      "Epoch 87/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1102 - accuracy: 0.9552 - val_loss: 0.1315 - val_accuracy: 0.9532\n",
      "Epoch 88/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1013 - accuracy: 0.9581 - val_loss: 0.1287 - val_accuracy: 0.9528\n",
      "Epoch 89/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.0974 - accuracy: 0.9624 - val_loss: 0.1332 - val_accuracy: 0.9571\n",
      "Epoch 90/100\n",
      "586/586 [==============================] - 82s 138ms/step - loss: 0.0988 - accuracy: 0.9597 - val_loss: 0.1406 - val_accuracy: 0.9543\n",
      "Epoch 91/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1044 - accuracy: 0.9591 - val_loss: 0.1396 - val_accuracy: 0.9526\n",
      "Epoch 92/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1087 - accuracy: 0.9550 - val_loss: 0.1477 - val_accuracy: 0.9537\n",
      "Epoch 93/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1050 - accuracy: 0.9618 - val_loss: 0.1415 - val_accuracy: 0.9490\n",
      "Epoch 94/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.0992 - accuracy: 0.9594 - val_loss: 0.1522 - val_accuracy: 0.9573\n",
      "Epoch 95/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.0975 - accuracy: 0.9625 - val_loss: 0.1426 - val_accuracy: 0.9577\n",
      "Epoch 96/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.0982 - accuracy: 0.9580 - val_loss: 0.1404 - val_accuracy: 0.9532\n",
      "Epoch 97/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1028 - accuracy: 0.9588 - val_loss: 0.1332 - val_accuracy: 0.9584\n",
      "Epoch 98/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1015 - accuracy: 0.9590 - val_loss: 0.1470 - val_accuracy: 0.9498\n",
      "Epoch 99/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1020 - accuracy: 0.9579 - val_loss: 0.1424 - val_accuracy: 0.9539\n",
      "Epoch 100/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.0992 - accuracy: 0.9599 - val_loss: 0.1345 - val_accuracy: 0.9551\n",
      "--------------- trial #6 ---------------\n",
      "Epoch 1/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.6916 - accuracy: 0.5132 - val_loss: 0.6577 - val_accuracy: 0.6034\n",
      "Epoch 2/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.6675 - accuracy: 0.5809 - val_loss: 0.6280 - val_accuracy: 0.6480\n",
      "Epoch 3/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.6354 - accuracy: 0.6372 - val_loss: 0.6017 - val_accuracy: 0.6803\n",
      "Epoch 4/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.6083 - accuracy: 0.6667 - val_loss: 0.5962 - val_accuracy: 0.6848\n",
      "Epoch 5/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.5983 - accuracy: 0.6783 - val_loss: 0.5714 - val_accuracy: 0.6905\n",
      "Epoch 6/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.5686 - accuracy: 0.7040 - val_loss: 0.5298 - val_accuracy: 0.7437\n",
      "Epoch 7/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.5499 - accuracy: 0.7224 - val_loss: 0.5124 - val_accuracy: 0.7659\n",
      "Epoch 8/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.5300 - accuracy: 0.7405 - val_loss: 0.4923 - val_accuracy: 0.7687\n",
      "Epoch 9/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.5058 - accuracy: 0.7596 - val_loss: 0.4677 - val_accuracy: 0.7864\n",
      "Epoch 10/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "586/586 [==============================] - 82s 139ms/step - loss: 0.4810 - accuracy: 0.7751 - val_loss: 0.4048 - val_accuracy: 0.8279\n",
      "Epoch 11/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.4472 - accuracy: 0.7987 - val_loss: 0.4028 - val_accuracy: 0.8272\n",
      "Epoch 12/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.4305 - accuracy: 0.8049 - val_loss: 0.3381 - val_accuracy: 0.8571\n",
      "Epoch 13/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.4043 - accuracy: 0.8238 - val_loss: 0.3374 - val_accuracy: 0.8541\n",
      "Epoch 14/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.3752 - accuracy: 0.8371 - val_loss: 0.3175 - val_accuracy: 0.8727\n",
      "Epoch 15/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.3575 - accuracy: 0.8437 - val_loss: 0.3153 - val_accuracy: 0.8648\n",
      "Epoch 16/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.3434 - accuracy: 0.8542 - val_loss: 0.2554 - val_accuracy: 0.8919\n",
      "Epoch 17/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.3142 - accuracy: 0.8631 - val_loss: 0.2550 - val_accuracy: 0.8909\n",
      "Epoch 18/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.3102 - accuracy: 0.8674 - val_loss: 0.2448 - val_accuracy: 0.8962\n",
      "Epoch 19/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2884 - accuracy: 0.8771 - val_loss: 0.2334 - val_accuracy: 0.9035\n",
      "Epoch 20/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2780 - accuracy: 0.8806 - val_loss: 0.2302 - val_accuracy: 0.9018\n",
      "Epoch 21/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2806 - accuracy: 0.8826 - val_loss: 0.2584 - val_accuracy: 0.8915\n",
      "Epoch 22/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2663 - accuracy: 0.8879 - val_loss: 0.2597 - val_accuracy: 0.8840\n",
      "Epoch 23/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.2564 - accuracy: 0.8879 - val_loss: 0.2154 - val_accuracy: 0.9077\n",
      "Epoch 24/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2640 - accuracy: 0.8898 - val_loss: 0.2473 - val_accuracy: 0.8913\n",
      "Epoch 25/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.2433 - accuracy: 0.8988 - val_loss: 0.2068 - val_accuracy: 0.9148\n",
      "Epoch 26/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2543 - accuracy: 0.8949 - val_loss: 0.1952 - val_accuracy: 0.9193\n",
      "Epoch 27/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2350 - accuracy: 0.9065 - val_loss: 0.1921 - val_accuracy: 0.9242\n",
      "Epoch 28/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.2282 - accuracy: 0.9053 - val_loss: 0.2195 - val_accuracy: 0.9018\n",
      "Epoch 29/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2337 - accuracy: 0.9007 - val_loss: 0.2143 - val_accuracy: 0.9092\n",
      "Epoch 30/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2381 - accuracy: 0.9010 - val_loss: 0.2162 - val_accuracy: 0.9126\n",
      "Epoch 31/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.2377 - accuracy: 0.8994 - val_loss: 0.1954 - val_accuracy: 0.9161\n",
      "Epoch 32/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2143 - accuracy: 0.9115 - val_loss: 0.1935 - val_accuracy: 0.9186\n",
      "Epoch 33/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2160 - accuracy: 0.9122 - val_loss: 0.2308 - val_accuracy: 0.9075\n",
      "Epoch 34/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2074 - accuracy: 0.9148 - val_loss: 0.1780 - val_accuracy: 0.9259\n",
      "Epoch 35/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1990 - accuracy: 0.9192 - val_loss: 0.1727 - val_accuracy: 0.9302\n",
      "Epoch 36/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2036 - accuracy: 0.9165 - val_loss: 0.1742 - val_accuracy: 0.9327\n",
      "Epoch 37/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1961 - accuracy: 0.9188 - val_loss: 0.1706 - val_accuracy: 0.9291\n",
      "Epoch 38/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1885 - accuracy: 0.9197 - val_loss: 0.1749 - val_accuracy: 0.9284\n",
      "Epoch 39/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1854 - accuracy: 0.9210 - val_loss: 0.2351 - val_accuracy: 0.9024\n",
      "Epoch 40/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1906 - accuracy: 0.9224 - val_loss: 0.1820 - val_accuracy: 0.9229\n",
      "Epoch 41/100\n",
      "586/586 [==============================] - 82s 138ms/step - loss: 0.1868 - accuracy: 0.9220 - val_loss: 0.1609 - val_accuracy: 0.9383\n",
      "Epoch 42/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1952 - accuracy: 0.9187 - val_loss: 0.1749 - val_accuracy: 0.9338\n",
      "Epoch 43/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1861 - accuracy: 0.9227 - val_loss: 0.1680 - val_accuracy: 0.9334\n",
      "Epoch 44/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1849 - accuracy: 0.9241 - val_loss: 0.2345 - val_accuracy: 0.9148\n",
      "Epoch 45/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1827 - accuracy: 0.9228 - val_loss: 0.1528 - val_accuracy: 0.9393\n",
      "Epoch 46/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1785 - accuracy: 0.9281 - val_loss: 0.1617 - val_accuracy: 0.9398\n",
      "Epoch 47/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1759 - accuracy: 0.9263 - val_loss: 0.1658 - val_accuracy: 0.9334\n",
      "Epoch 48/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1846 - accuracy: 0.9246 - val_loss: 0.1548 - val_accuracy: 0.9385\n",
      "Epoch 49/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1759 - accuracy: 0.9269 - val_loss: 0.1553 - val_accuracy: 0.9374\n",
      "Epoch 50/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1758 - accuracy: 0.9262 - val_loss: 0.1768 - val_accuracy: 0.9327\n",
      "Epoch 51/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1616 - accuracy: 0.9341 - val_loss: 0.1490 - val_accuracy: 0.9421\n",
      "Epoch 52/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1771 - accuracy: 0.9240 - val_loss: 0.1576 - val_accuracy: 0.9413\n",
      "Epoch 53/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1728 - accuracy: 0.9260 - val_loss: 0.1570 - val_accuracy: 0.9391\n",
      "Epoch 54/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1758 - accuracy: 0.9288 - val_loss: 0.1592 - val_accuracy: 0.9391\n",
      "Epoch 55/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1712 - accuracy: 0.9294 - val_loss: 0.1797 - val_accuracy: 0.9319\n",
      "Epoch 56/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1696 - accuracy: 0.9292 - val_loss: 0.1702 - val_accuracy: 0.9331\n",
      "Epoch 57/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1665 - accuracy: 0.9300 - val_loss: 0.1461 - val_accuracy: 0.9451\n",
      "Epoch 58/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1682 - accuracy: 0.9274 - val_loss: 0.1590 - val_accuracy: 0.9391\n",
      "Epoch 59/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1608 - accuracy: 0.9329 - val_loss: 0.1416 - val_accuracy: 0.9453\n",
      "Epoch 60/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1651 - accuracy: 0.9320 - val_loss: 0.1824 - val_accuracy: 0.9310\n",
      "Epoch 61/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1503 - accuracy: 0.9393 - val_loss: 0.1514 - val_accuracy: 0.9413\n",
      "Epoch 62/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1515 - accuracy: 0.9356 - val_loss: 0.1644 - val_accuracy: 0.9357\n",
      "Epoch 63/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1526 - accuracy: 0.9348 - val_loss: 0.1591 - val_accuracy: 0.9457\n",
      "Epoch 64/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1572 - accuracy: 0.9353 - val_loss: 0.1581 - val_accuracy: 0.9385\n",
      "Epoch 65/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1518 - accuracy: 0.9347 - val_loss: 0.1640 - val_accuracy: 0.9374\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 66/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1472 - accuracy: 0.9423 - val_loss: 0.1708 - val_accuracy: 0.9361\n",
      "Epoch 67/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1545 - accuracy: 0.9355 - val_loss: 0.1663 - val_accuracy: 0.9385\n",
      "Epoch 68/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1470 - accuracy: 0.9402 - val_loss: 0.1541 - val_accuracy: 0.9413\n",
      "Epoch 69/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1479 - accuracy: 0.9395 - val_loss: 0.1909 - val_accuracy: 0.9434\n",
      "Epoch 70/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1543 - accuracy: 0.9387 - val_loss: 0.1927 - val_accuracy: 0.9361\n",
      "Epoch 71/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1549 - accuracy: 0.9374 - val_loss: 0.1390 - val_accuracy: 0.9462\n",
      "Epoch 72/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1520 - accuracy: 0.9372 - val_loss: 0.1643 - val_accuracy: 0.9402\n",
      "Epoch 73/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1479 - accuracy: 0.9389 - val_loss: 0.1466 - val_accuracy: 0.9421\n",
      "Epoch 74/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1433 - accuracy: 0.9418 - val_loss: 0.1692 - val_accuracy: 0.9389\n",
      "Epoch 75/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1501 - accuracy: 0.9377 - val_loss: 0.1471 - val_accuracy: 0.9453\n",
      "Epoch 76/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1427 - accuracy: 0.9421 - val_loss: 0.1585 - val_accuracy: 0.9421\n",
      "Epoch 77/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1450 - accuracy: 0.9393 - val_loss: 0.1580 - val_accuracy: 0.9391\n",
      "Epoch 78/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1446 - accuracy: 0.9396 - val_loss: 0.1524 - val_accuracy: 0.9428\n",
      "Epoch 79/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1429 - accuracy: 0.9397 - val_loss: 0.1692 - val_accuracy: 0.9438\n",
      "Epoch 80/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1424 - accuracy: 0.9433 - val_loss: 0.1393 - val_accuracy: 0.9481\n",
      "Epoch 81/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1361 - accuracy: 0.9461 - val_loss: 0.1520 - val_accuracy: 0.9364\n",
      "Epoch 82/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1346 - accuracy: 0.9446 - val_loss: 0.1680 - val_accuracy: 0.9430\n",
      "Epoch 83/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1372 - accuracy: 0.9456 - val_loss: 0.1335 - val_accuracy: 0.9481\n",
      "Epoch 84/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1418 - accuracy: 0.9422 - val_loss: 0.1331 - val_accuracy: 0.9455\n",
      "Epoch 85/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1418 - accuracy: 0.9449 - val_loss: 0.1321 - val_accuracy: 0.9507\n",
      "Epoch 86/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1365 - accuracy: 0.9431 - val_loss: 0.1504 - val_accuracy: 0.9421\n",
      "Epoch 87/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1357 - accuracy: 0.9432 - val_loss: 0.1313 - val_accuracy: 0.9511\n",
      "Epoch 88/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1383 - accuracy: 0.9413 - val_loss: 0.1868 - val_accuracy: 0.9387\n",
      "Epoch 89/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1332 - accuracy: 0.9433 - val_loss: 0.1766 - val_accuracy: 0.9364\n",
      "Epoch 90/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1323 - accuracy: 0.9468 - val_loss: 0.1434 - val_accuracy: 0.9466\n",
      "Epoch 91/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1318 - accuracy: 0.9454 - val_loss: 0.1421 - val_accuracy: 0.9408\n",
      "Epoch 92/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1340 - accuracy: 0.9443 - val_loss: 0.1818 - val_accuracy: 0.9374\n",
      "Epoch 93/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1294 - accuracy: 0.9483 - val_loss: 0.1480 - val_accuracy: 0.9453\n",
      "Epoch 94/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1330 - accuracy: 0.9480 - val_loss: 0.1488 - val_accuracy: 0.9432\n",
      "Epoch 95/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1298 - accuracy: 0.9498 - val_loss: 0.1537 - val_accuracy: 0.9432\n",
      "Epoch 96/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1295 - accuracy: 0.9482 - val_loss: 0.1416 - val_accuracy: 0.9498\n",
      "Epoch 97/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1355 - accuracy: 0.9457 - val_loss: 0.1990 - val_accuracy: 0.9430\n",
      "Epoch 98/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1325 - accuracy: 0.9482 - val_loss: 0.1258 - val_accuracy: 0.9556\n",
      "Epoch 99/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1253 - accuracy: 0.9484 - val_loss: 0.1632 - val_accuracy: 0.9419\n",
      "Epoch 100/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1271 - accuracy: 0.9477 - val_loss: 0.1428 - val_accuracy: 0.9475\n",
      "--------------- trial #7 ---------------\n",
      "Epoch 1/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.6859 - accuracy: 0.5397 - val_loss: 0.6640 - val_accuracy: 0.5906\n",
      "Epoch 2/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.6584 - accuracy: 0.5953 - val_loss: 0.6142 - val_accuracy: 0.6578\n",
      "Epoch 3/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.6240 - accuracy: 0.6485 - val_loss: 0.5890 - val_accuracy: 0.6939\n",
      "Epoch 4/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.6050 - accuracy: 0.6674 - val_loss: 0.5780 - val_accuracy: 0.7055\n",
      "Epoch 5/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.5799 - accuracy: 0.6984 - val_loss: 0.5576 - val_accuracy: 0.7247\n",
      "Epoch 6/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.5610 - accuracy: 0.7140 - val_loss: 0.5437 - val_accuracy: 0.7281\n",
      "Epoch 7/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.5309 - accuracy: 0.7345 - val_loss: 0.4772 - val_accuracy: 0.7766\n",
      "Epoch 8/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.4998 - accuracy: 0.7612 - val_loss: 0.4207 - val_accuracy: 0.8150\n",
      "Epoch 9/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.4620 - accuracy: 0.7888 - val_loss: 0.3737 - val_accuracy: 0.8402\n",
      "Epoch 10/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.4124 - accuracy: 0.8175 - val_loss: 0.3815 - val_accuracy: 0.8345\n",
      "Epoch 11/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.3608 - accuracy: 0.8459 - val_loss: 0.2965 - val_accuracy: 0.8770\n",
      "Epoch 12/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.3243 - accuracy: 0.8611 - val_loss: 0.2508 - val_accuracy: 0.8968\n",
      "Epoch 13/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2956 - accuracy: 0.8754 - val_loss: 0.2599 - val_accuracy: 0.8975\n",
      "Epoch 14/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2827 - accuracy: 0.8819 - val_loss: 0.2871 - val_accuracy: 0.8746\n",
      "Epoch 15/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2794 - accuracy: 0.8867 - val_loss: 0.2888 - val_accuracy: 0.8678\n",
      "Epoch 16/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2582 - accuracy: 0.8917 - val_loss: 0.2734 - val_accuracy: 0.8757\n",
      "Epoch 17/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2483 - accuracy: 0.8987 - val_loss: 0.1991 - val_accuracy: 0.9167\n",
      "Epoch 18/100\n",
      "586/586 [==============================] - 82s 138ms/step - loss: 0.2365 - accuracy: 0.9027 - val_loss: 0.2446 - val_accuracy: 0.8960\n",
      "Epoch 19/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.2396 - accuracy: 0.9010 - val_loss: 0.1819 - val_accuracy: 0.9257\n",
      "Epoch 20/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2241 - accuracy: 0.9067 - val_loss: 0.1939 - val_accuracy: 0.9184\n",
      "Epoch 21/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2218 - accuracy: 0.9099 - val_loss: 0.1957 - val_accuracy: 0.9165\n",
      "Epoch 22/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2191 - accuracy: 0.9145 - val_loss: 0.1864 - val_accuracy: 0.9227\n",
      "Epoch 23/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.2029 - accuracy: 0.9160 - val_loss: 0.1765 - val_accuracy: 0.9259\n",
      "Epoch 24/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2032 - accuracy: 0.9162 - val_loss: 0.1807 - val_accuracy: 0.9261\n",
      "Epoch 25/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2041 - accuracy: 0.9172 - val_loss: 0.2210 - val_accuracy: 0.9124\n",
      "Epoch 26/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1908 - accuracy: 0.9242 - val_loss: 0.2080 - val_accuracy: 0.9158\n",
      "Epoch 27/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1987 - accuracy: 0.9208 - val_loss: 0.1707 - val_accuracy: 0.9280\n",
      "Epoch 28/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1900 - accuracy: 0.9213 - val_loss: 0.1819 - val_accuracy: 0.9265\n",
      "Epoch 29/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1791 - accuracy: 0.9255 - val_loss: 0.1717 - val_accuracy: 0.9312\n",
      "Epoch 30/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1775 - accuracy: 0.9300 - val_loss: 0.1777 - val_accuracy: 0.9291\n",
      "Epoch 31/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1784 - accuracy: 0.9268 - val_loss: 0.1553 - val_accuracy: 0.9389\n",
      "Epoch 32/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1743 - accuracy: 0.9291 - val_loss: 0.2416 - val_accuracy: 0.9037\n",
      "Epoch 33/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1774 - accuracy: 0.9290 - val_loss: 0.1778 - val_accuracy: 0.9336\n",
      "Epoch 34/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1711 - accuracy: 0.9322 - val_loss: 0.2003 - val_accuracy: 0.9252\n",
      "Epoch 35/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1717 - accuracy: 0.9300 - val_loss: 0.1639 - val_accuracy: 0.9383\n",
      "Epoch 36/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1669 - accuracy: 0.9325 - val_loss: 0.1639 - val_accuracy: 0.9329\n",
      "Epoch 37/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1576 - accuracy: 0.9363 - val_loss: 0.1600 - val_accuracy: 0.9359\n",
      "Epoch 38/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1602 - accuracy: 0.9358 - val_loss: 0.1664 - val_accuracy: 0.9274\n",
      "Epoch 39/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1583 - accuracy: 0.9374 - val_loss: 0.1864 - val_accuracy: 0.9297\n",
      "Epoch 40/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1532 - accuracy: 0.9356 - val_loss: 0.1456 - val_accuracy: 0.9430\n",
      "Epoch 41/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1750 - accuracy: 0.9289 - val_loss: 0.1475 - val_accuracy: 0.9430\n",
      "Epoch 42/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1539 - accuracy: 0.9415 - val_loss: 0.1499 - val_accuracy: 0.9398\n",
      "Epoch 43/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1569 - accuracy: 0.9354 - val_loss: 0.1764 - val_accuracy: 0.9295\n",
      "Epoch 44/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1554 - accuracy: 0.9373 - val_loss: 0.1622 - val_accuracy: 0.9323\n",
      "Epoch 45/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1548 - accuracy: 0.9394 - val_loss: 0.1594 - val_accuracy: 0.9398\n",
      "Epoch 46/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1490 - accuracy: 0.9398 - val_loss: 0.1382 - val_accuracy: 0.9460\n",
      "Epoch 47/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1400 - accuracy: 0.9426 - val_loss: 0.1454 - val_accuracy: 0.9404\n",
      "Epoch 48/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1459 - accuracy: 0.9409 - val_loss: 0.1746 - val_accuracy: 0.9366\n",
      "Epoch 49/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1486 - accuracy: 0.9397 - val_loss: 0.1390 - val_accuracy: 0.9455\n",
      "Epoch 50/100\n",
      "586/586 [==============================] - 79s 133ms/step - loss: 0.1332 - accuracy: 0.9440 - val_loss: 0.1437 - val_accuracy: 0.9440\n",
      "Epoch 51/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1408 - accuracy: 0.9431 - val_loss: 0.1606 - val_accuracy: 0.9359\n",
      "Epoch 52/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1409 - accuracy: 0.9438 - val_loss: 0.1503 - val_accuracy: 0.9378\n",
      "Epoch 53/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.1457 - accuracy: 0.9377 - val_loss: 0.1520 - val_accuracy: 0.9415\n",
      "Epoch 54/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1299 - accuracy: 0.9475 - val_loss: 0.1373 - val_accuracy: 0.9436\n",
      "Epoch 55/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1435 - accuracy: 0.9404 - val_loss: 0.1464 - val_accuracy: 0.9404\n",
      "Epoch 56/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1355 - accuracy: 0.9417 - val_loss: 0.1506 - val_accuracy: 0.9419\n",
      "Epoch 57/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1393 - accuracy: 0.9430 - val_loss: 0.1675 - val_accuracy: 0.9355\n",
      "Epoch 58/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1343 - accuracy: 0.9453 - val_loss: 0.1488 - val_accuracy: 0.9406\n",
      "Epoch 59/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1263 - accuracy: 0.9499 - val_loss: 0.1334 - val_accuracy: 0.9468\n",
      "Epoch 60/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1309 - accuracy: 0.9475 - val_loss: 0.1520 - val_accuracy: 0.9408\n",
      "Epoch 61/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1281 - accuracy: 0.9481 - val_loss: 0.1781 - val_accuracy: 0.9383\n",
      "Epoch 62/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1365 - accuracy: 0.9455 - val_loss: 0.1350 - val_accuracy: 0.9462\n",
      "Epoch 63/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1305 - accuracy: 0.9480 - val_loss: 0.1563 - val_accuracy: 0.9445\n",
      "Epoch 64/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1265 - accuracy: 0.9518 - val_loss: 0.2158 - val_accuracy: 0.9321\n",
      "Epoch 65/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1165 - accuracy: 0.9546 - val_loss: 0.1274 - val_accuracy: 0.9466\n",
      "Epoch 66/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1293 - accuracy: 0.9469 - val_loss: 0.1602 - val_accuracy: 0.9479\n",
      "Epoch 67/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1225 - accuracy: 0.9502 - val_loss: 0.1377 - val_accuracy: 0.9487\n",
      "Epoch 68/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1200 - accuracy: 0.9522 - val_loss: 0.1424 - val_accuracy: 0.9475\n",
      "Epoch 69/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1220 - accuracy: 0.9519 - val_loss: 0.1335 - val_accuracy: 0.9487\n",
      "Epoch 70/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1219 - accuracy: 0.9499 - val_loss: 0.1564 - val_accuracy: 0.9370\n",
      "Epoch 71/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1158 - accuracy: 0.9514 - val_loss: 0.1386 - val_accuracy: 0.9490\n",
      "Epoch 72/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1201 - accuracy: 0.9496 - val_loss: 0.1347 - val_accuracy: 0.9485\n",
      "Epoch 73/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1229 - accuracy: 0.9523 - val_loss: 0.1946 - val_accuracy: 0.9306\n",
      "Epoch 74/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1177 - accuracy: 0.9522 - val_loss: 0.1580 - val_accuracy: 0.9477\n",
      "Epoch 75/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1172 - accuracy: 0.9512 - val_loss: 0.1315 - val_accuracy: 0.9504\n",
      "Epoch 76/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1102 - accuracy: 0.9563 - val_loss: 0.1351 - val_accuracy: 0.9477\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 77/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1135 - accuracy: 0.9553 - val_loss: 0.1622 - val_accuracy: 0.9485\n",
      "Epoch 78/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1164 - accuracy: 0.9558 - val_loss: 0.1376 - val_accuracy: 0.9528\n",
      "Epoch 79/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1133 - accuracy: 0.9566 - val_loss: 0.1442 - val_accuracy: 0.9436\n",
      "Epoch 80/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1191 - accuracy: 0.9530 - val_loss: 0.1377 - val_accuracy: 0.9511\n",
      "Epoch 81/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1076 - accuracy: 0.9574 - val_loss: 0.1350 - val_accuracy: 0.9481\n",
      "Epoch 82/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1097 - accuracy: 0.9546 - val_loss: 0.1242 - val_accuracy: 0.9522\n",
      "Epoch 83/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1128 - accuracy: 0.9541 - val_loss: 0.1504 - val_accuracy: 0.9449\n",
      "Epoch 84/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1196 - accuracy: 0.9514 - val_loss: 0.1298 - val_accuracy: 0.9549\n",
      "Epoch 85/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1088 - accuracy: 0.9539 - val_loss: 0.1284 - val_accuracy: 0.9490\n",
      "Epoch 86/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1102 - accuracy: 0.9549 - val_loss: 0.1192 - val_accuracy: 0.9532\n",
      "Epoch 87/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1101 - accuracy: 0.9562 - val_loss: 0.1450 - val_accuracy: 0.9511\n",
      "Epoch 88/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1080 - accuracy: 0.9556 - val_loss: 0.1355 - val_accuracy: 0.9483\n",
      "Epoch 89/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1131 - accuracy: 0.9539 - val_loss: 0.1528 - val_accuracy: 0.9466\n",
      "Epoch 90/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1191 - accuracy: 0.9517 - val_loss: 0.1773 - val_accuracy: 0.9383\n",
      "Epoch 91/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1073 - accuracy: 0.9557 - val_loss: 0.1446 - val_accuracy: 0.9483\n",
      "Epoch 92/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1035 - accuracy: 0.9610 - val_loss: 0.1410 - val_accuracy: 0.9496\n",
      "Epoch 93/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1011 - accuracy: 0.9584 - val_loss: 0.1469 - val_accuracy: 0.9504\n",
      "Epoch 94/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1052 - accuracy: 0.9584 - val_loss: 0.1217 - val_accuracy: 0.9560\n",
      "Epoch 95/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1017 - accuracy: 0.9602 - val_loss: 0.1191 - val_accuracy: 0.9524\n",
      "Epoch 96/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1027 - accuracy: 0.9571 - val_loss: 0.1333 - val_accuracy: 0.9549\n",
      "Epoch 97/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1028 - accuracy: 0.9590 - val_loss: 0.1450 - val_accuracy: 0.9547\n",
      "Epoch 98/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1098 - accuracy: 0.9575 - val_loss: 0.1551 - val_accuracy: 0.9502\n",
      "Epoch 99/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1082 - accuracy: 0.9519 - val_loss: 0.1331 - val_accuracy: 0.9524\n",
      "Epoch 100/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1066 - accuracy: 0.9569 - val_loss: 0.1499 - val_accuracy: 0.9511\n",
      "--------------- trial #8 ---------------\n",
      "Epoch 1/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.6812 - accuracy: 0.5479 - val_loss: 0.6592 - val_accuracy: 0.6307\n",
      "Epoch 2/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.6498 - accuracy: 0.6217 - val_loss: 0.6157 - val_accuracy: 0.6694\n",
      "Epoch 3/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.6177 - accuracy: 0.6609 - val_loss: 0.5691 - val_accuracy: 0.7134\n",
      "Epoch 4/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.5925 - accuracy: 0.6930 - val_loss: 0.5556 - val_accuracy: 0.7315\n",
      "Epoch 5/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.5719 - accuracy: 0.7112 - val_loss: 0.5737 - val_accuracy: 0.6944\n",
      "Epoch 6/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.5577 - accuracy: 0.7208 - val_loss: 0.5222 - val_accuracy: 0.7606\n",
      "Epoch 7/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.5325 - accuracy: 0.7365 - val_loss: 0.5417 - val_accuracy: 0.7431\n",
      "Epoch 8/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.5089 - accuracy: 0.7551 - val_loss: 0.4458 - val_accuracy: 0.8069\n",
      "Epoch 9/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.4698 - accuracy: 0.7790 - val_loss: 0.4376 - val_accuracy: 0.8123\n",
      "Epoch 10/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.4477 - accuracy: 0.7953 - val_loss: 0.3739 - val_accuracy: 0.8413\n",
      "Epoch 11/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.4200 - accuracy: 0.8128 - val_loss: 0.3739 - val_accuracy: 0.8398\n",
      "Epoch 12/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.3923 - accuracy: 0.8300 - val_loss: 0.3699 - val_accuracy: 0.8387\n",
      "Epoch 13/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.3641 - accuracy: 0.8451 - val_loss: 0.3011 - val_accuracy: 0.8723\n",
      "Epoch 14/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.3437 - accuracy: 0.8495 - val_loss: 0.2686 - val_accuracy: 0.8870\n",
      "Epoch 15/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.3224 - accuracy: 0.8609 - val_loss: 0.2615 - val_accuracy: 0.8921\n",
      "Epoch 16/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.3151 - accuracy: 0.8682 - val_loss: 0.2754 - val_accuracy: 0.8847\n",
      "Epoch 17/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2862 - accuracy: 0.8802 - val_loss: 0.2437 - val_accuracy: 0.8992\n",
      "Epoch 18/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2830 - accuracy: 0.8777 - val_loss: 0.2350 - val_accuracy: 0.9030\n",
      "Epoch 19/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.2731 - accuracy: 0.8835 - val_loss: 0.2397 - val_accuracy: 0.9003\n",
      "Epoch 20/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2610 - accuracy: 0.8900 - val_loss: 0.2186 - val_accuracy: 0.9133\n",
      "Epoch 21/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2440 - accuracy: 0.8972 - val_loss: 0.2559 - val_accuracy: 0.8904\n",
      "Epoch 22/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2514 - accuracy: 0.8934 - val_loss: 0.2164 - val_accuracy: 0.9090\n",
      "Epoch 23/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2446 - accuracy: 0.8993 - val_loss: 0.2395 - val_accuracy: 0.9030\n",
      "Epoch 24/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2441 - accuracy: 0.8981 - val_loss: 0.2651 - val_accuracy: 0.8857\n",
      "Epoch 25/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2328 - accuracy: 0.9007 - val_loss: 0.1993 - val_accuracy: 0.9161\n",
      "Epoch 26/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.2289 - accuracy: 0.8995 - val_loss: 0.1968 - val_accuracy: 0.9223\n",
      "Epoch 27/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2228 - accuracy: 0.9068 - val_loss: 0.2312 - val_accuracy: 0.9035\n",
      "Epoch 28/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2192 - accuracy: 0.9083 - val_loss: 0.1695 - val_accuracy: 0.9331\n",
      "Epoch 29/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2105 - accuracy: 0.9113 - val_loss: 0.1829 - val_accuracy: 0.9304\n",
      "Epoch 30/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2058 - accuracy: 0.9108 - val_loss: 0.1719 - val_accuracy: 0.9331\n",
      "Epoch 31/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.1997 - accuracy: 0.9168 - val_loss: 0.1667 - val_accuracy: 0.9323\n",
      "Epoch 32/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1984 - accuracy: 0.9179 - val_loss: 0.1766 - val_accuracy: 0.9302\n",
      "Epoch 33/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2004 - accuracy: 0.9170 - val_loss: 0.1756 - val_accuracy: 0.9297\n",
      "Epoch 34/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1925 - accuracy: 0.9209 - val_loss: 0.1693 - val_accuracy: 0.9306\n",
      "Epoch 35/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1911 - accuracy: 0.9205 - val_loss: 0.1750 - val_accuracy: 0.9280\n",
      "Epoch 36/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1919 - accuracy: 0.9195 - val_loss: 0.1745 - val_accuracy: 0.9289\n",
      "Epoch 37/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1854 - accuracy: 0.9232 - val_loss: 0.2111 - val_accuracy: 0.9111\n",
      "Epoch 38/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1834 - accuracy: 0.9263 - val_loss: 0.1611 - val_accuracy: 0.9370\n",
      "Epoch 39/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1779 - accuracy: 0.9285 - val_loss: 0.1602 - val_accuracy: 0.9404\n",
      "Epoch 40/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1830 - accuracy: 0.9283 - val_loss: 0.2100 - val_accuracy: 0.9144\n",
      "Epoch 41/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1808 - accuracy: 0.9261 - val_loss: 0.1682 - val_accuracy: 0.9325\n",
      "Epoch 42/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1777 - accuracy: 0.9260 - val_loss: 0.2359 - val_accuracy: 0.9097\n",
      "Epoch 43/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1828 - accuracy: 0.9276 - val_loss: 0.2047 - val_accuracy: 0.9163\n",
      "Epoch 44/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1717 - accuracy: 0.9259 - val_loss: 0.1585 - val_accuracy: 0.9364\n",
      "Epoch 45/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1677 - accuracy: 0.9313 - val_loss: 0.1636 - val_accuracy: 0.9342\n",
      "Epoch 46/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1677 - accuracy: 0.9295 - val_loss: 0.1644 - val_accuracy: 0.9361\n",
      "Epoch 47/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1683 - accuracy: 0.9294 - val_loss: 0.1981 - val_accuracy: 0.9240\n",
      "Epoch 48/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1700 - accuracy: 0.9303 - val_loss: 0.1553 - val_accuracy: 0.9370\n",
      "Epoch 49/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1698 - accuracy: 0.9303 - val_loss: 0.1554 - val_accuracy: 0.9408\n",
      "Epoch 50/100\n",
      "586/586 [==============================] - 81s 137ms/step - loss: 0.1614 - accuracy: 0.9325 - val_loss: 0.1532 - val_accuracy: 0.9430\n",
      "Epoch 51/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1614 - accuracy: 0.9335 - val_loss: 0.2114 - val_accuracy: 0.9139\n",
      "Epoch 52/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1630 - accuracy: 0.9341 - val_loss: 0.1880 - val_accuracy: 0.9231\n",
      "Epoch 53/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1554 - accuracy: 0.9345 - val_loss: 0.1493 - val_accuracy: 0.9413\n",
      "Epoch 54/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1621 - accuracy: 0.9321 - val_loss: 0.1523 - val_accuracy: 0.9421\n",
      "Epoch 55/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1564 - accuracy: 0.9394 - val_loss: 0.1477 - val_accuracy: 0.9432\n",
      "Epoch 56/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1526 - accuracy: 0.9365 - val_loss: 0.1542 - val_accuracy: 0.9404\n",
      "Epoch 57/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1511 - accuracy: 0.9378 - val_loss: 0.1470 - val_accuracy: 0.9408\n",
      "Epoch 58/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1449 - accuracy: 0.9406 - val_loss: 0.1428 - val_accuracy: 0.9419\n",
      "Epoch 59/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1477 - accuracy: 0.9408 - val_loss: 0.1432 - val_accuracy: 0.9434\n",
      "Epoch 60/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1478 - accuracy: 0.9397 - val_loss: 0.1643 - val_accuracy: 0.9368\n",
      "Epoch 61/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1460 - accuracy: 0.9397 - val_loss: 0.1546 - val_accuracy: 0.9400\n",
      "Epoch 62/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1420 - accuracy: 0.9422 - val_loss: 0.1574 - val_accuracy: 0.9404\n",
      "Epoch 63/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1462 - accuracy: 0.9403 - val_loss: 0.1402 - val_accuracy: 0.9481\n",
      "Epoch 64/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1429 - accuracy: 0.9436 - val_loss: 0.1473 - val_accuracy: 0.9483\n",
      "Epoch 65/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1346 - accuracy: 0.9461 - val_loss: 0.1513 - val_accuracy: 0.9423\n",
      "Epoch 66/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1423 - accuracy: 0.9406 - val_loss: 0.1436 - val_accuracy: 0.9440\n",
      "Epoch 67/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1357 - accuracy: 0.9465 - val_loss: 0.1351 - val_accuracy: 0.9511\n",
      "Epoch 68/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1313 - accuracy: 0.9478 - val_loss: 0.1461 - val_accuracy: 0.9443\n",
      "Epoch 69/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1408 - accuracy: 0.9427 - val_loss: 0.1569 - val_accuracy: 0.9378\n",
      "Epoch 70/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1436 - accuracy: 0.9431 - val_loss: 0.1864 - val_accuracy: 0.9325\n",
      "Epoch 71/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1375 - accuracy: 0.9450 - val_loss: 0.1682 - val_accuracy: 0.9402\n",
      "Epoch 72/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1371 - accuracy: 0.9449 - val_loss: 0.1600 - val_accuracy: 0.9415\n",
      "Epoch 73/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1422 - accuracy: 0.9455 - val_loss: 0.1409 - val_accuracy: 0.9511\n",
      "Epoch 74/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1290 - accuracy: 0.9469 - val_loss: 0.1788 - val_accuracy: 0.9383\n",
      "Epoch 75/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1313 - accuracy: 0.9469 - val_loss: 0.1472 - val_accuracy: 0.9445\n",
      "Epoch 76/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1276 - accuracy: 0.9448 - val_loss: 0.2445 - val_accuracy: 0.9171\n",
      "Epoch 77/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1318 - accuracy: 0.9449 - val_loss: 0.1556 - val_accuracy: 0.9466\n",
      "Epoch 78/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.1298 - accuracy: 0.9482 - val_loss: 0.1363 - val_accuracy: 0.9530\n",
      "Epoch 79/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1263 - accuracy: 0.9498 - val_loss: 0.1518 - val_accuracy: 0.9472\n",
      "Epoch 80/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1314 - accuracy: 0.9468 - val_loss: 0.1406 - val_accuracy: 0.9451\n",
      "Epoch 81/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1342 - accuracy: 0.9454 - val_loss: 0.1482 - val_accuracy: 0.9500\n",
      "Epoch 82/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1227 - accuracy: 0.9520 - val_loss: 0.1437 - val_accuracy: 0.9443\n",
      "Epoch 83/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1289 - accuracy: 0.9452 - val_loss: 0.1358 - val_accuracy: 0.9504\n",
      "Epoch 84/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1215 - accuracy: 0.9511 - val_loss: 0.1406 - val_accuracy: 0.9449\n",
      "Epoch 85/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1277 - accuracy: 0.9486 - val_loss: 0.1978 - val_accuracy: 0.9393\n",
      "Epoch 86/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1205 - accuracy: 0.9493 - val_loss: 0.1506 - val_accuracy: 0.9447\n",
      "Epoch 87/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1234 - accuracy: 0.9507 - val_loss: 0.1710 - val_accuracy: 0.9393\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 88/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1208 - accuracy: 0.9504 - val_loss: 0.1601 - val_accuracy: 0.9494\n",
      "Epoch 89/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1293 - accuracy: 0.9466 - val_loss: 0.1471 - val_accuracy: 0.9468\n",
      "Epoch 90/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1229 - accuracy: 0.9510 - val_loss: 0.1518 - val_accuracy: 0.9494\n",
      "Epoch 91/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1194 - accuracy: 0.9507 - val_loss: 0.1506 - val_accuracy: 0.9417\n",
      "Epoch 92/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1186 - accuracy: 0.9520 - val_loss: 0.1737 - val_accuracy: 0.9393\n",
      "Epoch 93/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1221 - accuracy: 0.9501 - val_loss: 0.1466 - val_accuracy: 0.9447\n",
      "Epoch 94/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1180 - accuracy: 0.9522 - val_loss: 0.1522 - val_accuracy: 0.9502\n",
      "Epoch 95/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1121 - accuracy: 0.9560 - val_loss: 0.1363 - val_accuracy: 0.9541\n",
      "Epoch 96/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1162 - accuracy: 0.9522 - val_loss: 0.1552 - val_accuracy: 0.9447\n",
      "Epoch 97/100\n",
      "586/586 [==============================] - 82s 139ms/step - loss: 0.1224 - accuracy: 0.9500 - val_loss: 0.1524 - val_accuracy: 0.9453\n",
      "Epoch 98/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1075 - accuracy: 0.9551 - val_loss: 0.1363 - val_accuracy: 0.9494\n",
      "Epoch 99/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1140 - accuracy: 0.9513 - val_loss: 0.2193 - val_accuracy: 0.9325\n",
      "Epoch 100/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1216 - accuracy: 0.9522 - val_loss: 0.1550 - val_accuracy: 0.9483\n",
      "--------------- trial #9 ---------------\n",
      "Epoch 1/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.6804 - accuracy: 0.5499 - val_loss: 0.6548 - val_accuracy: 0.6433\n",
      "Epoch 2/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.6364 - accuracy: 0.6406 - val_loss: 0.5905 - val_accuracy: 0.7038\n",
      "Epoch 3/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.6032 - accuracy: 0.6796 - val_loss: 0.5695 - val_accuracy: 0.7076\n",
      "Epoch 4/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.5765 - accuracy: 0.6971 - val_loss: 0.5486 - val_accuracy: 0.7360\n",
      "Epoch 5/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.5630 - accuracy: 0.7149 - val_loss: 0.5379 - val_accuracy: 0.7399\n",
      "Epoch 6/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.5316 - accuracy: 0.7405 - val_loss: 0.6537 - val_accuracy: 0.6801\n",
      "Epoch 7/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.5068 - accuracy: 0.7579 - val_loss: 0.4549 - val_accuracy: 0.7965\n",
      "Epoch 8/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.4801 - accuracy: 0.7774 - val_loss: 0.4469 - val_accuracy: 0.8016\n",
      "Epoch 9/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.4542 - accuracy: 0.7932 - val_loss: 0.5039 - val_accuracy: 0.7740\n",
      "Epoch 10/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.4500 - accuracy: 0.7995 - val_loss: 0.3890 - val_accuracy: 0.8328\n",
      "Epoch 11/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.4091 - accuracy: 0.8185 - val_loss: 0.3825 - val_accuracy: 0.8385\n",
      "Epoch 12/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.3786 - accuracy: 0.8390 - val_loss: 0.3309 - val_accuracy: 0.8511\n",
      "Epoch 13/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.3540 - accuracy: 0.8507 - val_loss: 0.3140 - val_accuracy: 0.8684\n",
      "Epoch 14/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.3353 - accuracy: 0.8576 - val_loss: 0.2687 - val_accuracy: 0.8902\n",
      "Epoch 15/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.3191 - accuracy: 0.8645 - val_loss: 0.2665 - val_accuracy: 0.8902\n",
      "Epoch 16/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.3040 - accuracy: 0.8686 - val_loss: 0.2972 - val_accuracy: 0.8738\n",
      "Epoch 17/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2782 - accuracy: 0.8848 - val_loss: 0.2428 - val_accuracy: 0.8975\n",
      "Epoch 18/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2748 - accuracy: 0.8852 - val_loss: 0.2642 - val_accuracy: 0.8917\n",
      "Epoch 19/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2630 - accuracy: 0.8931 - val_loss: 0.2330 - val_accuracy: 0.9005\n",
      "Epoch 20/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2533 - accuracy: 0.8921 - val_loss: 0.2569 - val_accuracy: 0.8979\n",
      "Epoch 21/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2432 - accuracy: 0.8990 - val_loss: 0.2309 - val_accuracy: 0.9026\n",
      "Epoch 22/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.2416 - accuracy: 0.9021 - val_loss: 0.2736 - val_accuracy: 0.8926\n",
      "Epoch 23/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2252 - accuracy: 0.9057 - val_loss: 0.2198 - val_accuracy: 0.9099\n",
      "Epoch 24/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.2269 - accuracy: 0.9078 - val_loss: 0.1884 - val_accuracy: 0.9233\n",
      "Epoch 25/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.2151 - accuracy: 0.9101 - val_loss: 0.2432 - val_accuracy: 0.8977\n",
      "Epoch 26/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2127 - accuracy: 0.9124 - val_loss: 0.1934 - val_accuracy: 0.9208\n",
      "Epoch 27/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.2028 - accuracy: 0.9151 - val_loss: 0.1806 - val_accuracy: 0.9259\n",
      "Epoch 28/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2054 - accuracy: 0.9175 - val_loss: 0.1894 - val_accuracy: 0.9223\n",
      "Epoch 29/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.2072 - accuracy: 0.9167 - val_loss: 0.1954 - val_accuracy: 0.9173\n",
      "Epoch 30/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.2005 - accuracy: 0.9189 - val_loss: 0.1736 - val_accuracy: 0.9276\n",
      "Epoch 31/100\n",
      "586/586 [==============================] - 82s 138ms/step - loss: 0.2072 - accuracy: 0.9151 - val_loss: 0.1729 - val_accuracy: 0.9336\n",
      "Epoch 32/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1842 - accuracy: 0.9255 - val_loss: 0.1761 - val_accuracy: 0.9299\n",
      "Epoch 33/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1858 - accuracy: 0.9240 - val_loss: 0.1923 - val_accuracy: 0.9210\n",
      "Epoch 34/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1921 - accuracy: 0.9223 - val_loss: 0.1945 - val_accuracy: 0.9238\n",
      "Epoch 35/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1784 - accuracy: 0.9273 - val_loss: 0.1687 - val_accuracy: 0.9329\n",
      "Epoch 36/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1729 - accuracy: 0.9302 - val_loss: 0.2001 - val_accuracy: 0.9144\n",
      "Epoch 37/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1857 - accuracy: 0.9249 - val_loss: 0.1627 - val_accuracy: 0.9308\n",
      "Epoch 38/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1672 - accuracy: 0.9322 - val_loss: 0.1739 - val_accuracy: 0.9297\n",
      "Epoch 39/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1773 - accuracy: 0.9273 - val_loss: 0.1613 - val_accuracy: 0.9338\n",
      "Epoch 40/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1635 - accuracy: 0.9329 - val_loss: 0.1907 - val_accuracy: 0.9193\n",
      "Epoch 41/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1697 - accuracy: 0.9305 - val_loss: 0.1988 - val_accuracy: 0.9203\n",
      "Epoch 42/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1741 - accuracy: 0.9295 - val_loss: 0.1564 - val_accuracy: 0.9387\n",
      "Epoch 43/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1632 - accuracy: 0.9333 - val_loss: 0.2084 - val_accuracy: 0.9195\n",
      "Epoch 44/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1638 - accuracy: 0.9333 - val_loss: 0.1773 - val_accuracy: 0.9263\n",
      "Epoch 45/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1568 - accuracy: 0.9371 - val_loss: 0.1593 - val_accuracy: 0.9372\n",
      "Epoch 46/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1578 - accuracy: 0.9351 - val_loss: 0.1865 - val_accuracy: 0.9299\n",
      "Epoch 47/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1646 - accuracy: 0.9322 - val_loss: 0.1507 - val_accuracy: 0.9396\n",
      "Epoch 48/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1549 - accuracy: 0.9370 - val_loss: 0.1577 - val_accuracy: 0.9357\n",
      "Epoch 49/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1536 - accuracy: 0.9400 - val_loss: 0.1556 - val_accuracy: 0.9436\n",
      "Epoch 50/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1583 - accuracy: 0.9341 - val_loss: 0.1480 - val_accuracy: 0.9411\n",
      "Epoch 51/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1476 - accuracy: 0.9399 - val_loss: 0.1496 - val_accuracy: 0.9402\n",
      "Epoch 52/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1503 - accuracy: 0.9429 - val_loss: 0.1512 - val_accuracy: 0.9419\n",
      "Epoch 53/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1482 - accuracy: 0.9383 - val_loss: 0.1471 - val_accuracy: 0.9417\n",
      "Epoch 54/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1408 - accuracy: 0.9454 - val_loss: 0.1559 - val_accuracy: 0.9370\n",
      "Epoch 55/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1479 - accuracy: 0.9408 - val_loss: 0.1574 - val_accuracy: 0.9423\n",
      "Epoch 56/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1447 - accuracy: 0.9431 - val_loss: 0.1753 - val_accuracy: 0.9396\n",
      "Epoch 57/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1376 - accuracy: 0.9424 - val_loss: 0.1641 - val_accuracy: 0.9400\n",
      "Epoch 58/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1558 - accuracy: 0.9372 - val_loss: 0.1453 - val_accuracy: 0.9453\n",
      "Epoch 59/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1390 - accuracy: 0.9445 - val_loss: 0.1403 - val_accuracy: 0.9468\n",
      "Epoch 60/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1434 - accuracy: 0.9438 - val_loss: 0.1468 - val_accuracy: 0.9419\n",
      "Epoch 61/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1420 - accuracy: 0.9467 - val_loss: 0.1496 - val_accuracy: 0.9389\n",
      "Epoch 62/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1390 - accuracy: 0.9443 - val_loss: 0.1526 - val_accuracy: 0.9483\n",
      "Epoch 63/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1370 - accuracy: 0.9448 - val_loss: 0.1395 - val_accuracy: 0.9466\n",
      "Epoch 64/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1315 - accuracy: 0.9467 - val_loss: 0.1327 - val_accuracy: 0.9502\n",
      "Epoch 65/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1288 - accuracy: 0.9485 - val_loss: 0.1485 - val_accuracy: 0.9443\n",
      "Epoch 66/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1255 - accuracy: 0.9481 - val_loss: 0.1490 - val_accuracy: 0.9434\n",
      "Epoch 67/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1293 - accuracy: 0.9484 - val_loss: 0.1466 - val_accuracy: 0.9423\n",
      "Epoch 68/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1312 - accuracy: 0.9456 - val_loss: 0.1460 - val_accuracy: 0.9468\n",
      "Epoch 69/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1224 - accuracy: 0.9493 - val_loss: 0.1564 - val_accuracy: 0.9398\n",
      "Epoch 70/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1257 - accuracy: 0.9490 - val_loss: 0.1435 - val_accuracy: 0.9487\n",
      "Epoch 71/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1314 - accuracy: 0.9493 - val_loss: 0.1491 - val_accuracy: 0.9440\n",
      "Epoch 72/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1220 - accuracy: 0.9500 - val_loss: 0.1741 - val_accuracy: 0.9344\n",
      "Epoch 73/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1249 - accuracy: 0.9506 - val_loss: 0.1421 - val_accuracy: 0.9524\n",
      "Epoch 74/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1206 - accuracy: 0.9507 - val_loss: 0.1316 - val_accuracy: 0.9530\n",
      "Epoch 75/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1205 - accuracy: 0.9514 - val_loss: 0.1445 - val_accuracy: 0.9449\n",
      "Epoch 76/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1200 - accuracy: 0.9500 - val_loss: 0.1383 - val_accuracy: 0.9530\n",
      "Epoch 77/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1198 - accuracy: 0.9523 - val_loss: 0.1441 - val_accuracy: 0.9477\n",
      "Epoch 78/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1240 - accuracy: 0.9470 - val_loss: 0.1866 - val_accuracy: 0.9361\n",
      "Epoch 79/100\n",
      "586/586 [==============================] - 84s 142ms/step - loss: 0.1237 - accuracy: 0.9498 - val_loss: 0.1358 - val_accuracy: 0.9496\n",
      "Epoch 80/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1168 - accuracy: 0.9526 - val_loss: 0.1349 - val_accuracy: 0.9558\n",
      "Epoch 81/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1167 - accuracy: 0.9510 - val_loss: 0.1519 - val_accuracy: 0.9451\n",
      "Epoch 82/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1235 - accuracy: 0.9499 - val_loss: 0.1430 - val_accuracy: 0.9487\n",
      "Epoch 83/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1180 - accuracy: 0.9524 - val_loss: 0.1359 - val_accuracy: 0.9511\n",
      "Epoch 84/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1111 - accuracy: 0.9536 - val_loss: 0.1798 - val_accuracy: 0.9457\n",
      "Epoch 85/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1160 - accuracy: 0.9554 - val_loss: 0.1291 - val_accuracy: 0.9562\n",
      "Epoch 86/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1162 - accuracy: 0.9537 - val_loss: 0.1725 - val_accuracy: 0.9455\n",
      "Epoch 87/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1041 - accuracy: 0.9592 - val_loss: 0.1516 - val_accuracy: 0.9466\n",
      "Epoch 88/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1188 - accuracy: 0.9513 - val_loss: 0.1451 - val_accuracy: 0.9496\n",
      "Epoch 89/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1158 - accuracy: 0.9545 - val_loss: 0.1355 - val_accuracy: 0.9530\n",
      "Epoch 90/100\n",
      "586/586 [==============================] - 84s 143ms/step - loss: 0.1164 - accuracy: 0.9519 - val_loss: 0.1674 - val_accuracy: 0.9402\n",
      "Epoch 91/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1060 - accuracy: 0.9574 - val_loss: 0.1565 - val_accuracy: 0.9417\n",
      "Epoch 92/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1207 - accuracy: 0.9513 - val_loss: 0.1468 - val_accuracy: 0.9498\n",
      "Epoch 93/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1194 - accuracy: 0.9523 - val_loss: 0.1407 - val_accuracy: 0.9487\n",
      "Epoch 94/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1135 - accuracy: 0.9550 - val_loss: 0.1368 - val_accuracy: 0.9519\n",
      "Epoch 95/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1056 - accuracy: 0.9585 - val_loss: 0.1490 - val_accuracy: 0.9551\n",
      "Epoch 96/100\n",
      "586/586 [==============================] - 85s 143ms/step - loss: 0.1148 - accuracy: 0.9539 - val_loss: 0.1463 - val_accuracy: 0.9549\n",
      "Epoch 97/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1120 - accuracy: 0.9539 - val_loss: 0.1554 - val_accuracy: 0.9400\n",
      "Epoch 98/100\n",
      "586/586 [==============================] - 84s 141ms/step - loss: 0.1106 - accuracy: 0.9556 - val_loss: 0.1879 - val_accuracy: 0.9481\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 99/100\n",
      "586/586 [==============================] - 83s 141ms/step - loss: 0.1045 - accuracy: 0.9588 - val_loss: 0.1422 - val_accuracy: 0.9517\n",
      "Epoch 100/100\n",
      "586/586 [==============================] - 83s 140ms/step - loss: 0.1116 - accuracy: 0.9538 - val_loss: 0.1348 - val_accuracy: 0.9517\n"
     ]
    }
   ],
   "source": [
    "n_trials = 10\n",
    "\n",
    "epochs = 100\n",
    "\n",
    "for trial in range(n_trials):\n",
    "    print('--------------- trial #{} ---------------'.format(trial))\n",
    "    model = make_model(input_shape=image_size + (3,))\n",
    "\n",
    "    model.compile(\n",
    "        optimizer='adam',\n",
    "        loss=\"binary_crossentropy\",\n",
    "        metrics=[\"accuracy\"],\n",
    "    )\n",
    "\n",
    "    model.save('model/model_cat_dog_cnn_init_{}.h5'.format(trial))\n",
    "    model.fit(\n",
    "        train_ds, epochs=epochs, validation_data=val_ds,\n",
    "    )\n",
    "    model.save('model/model_cat_dog_cnn_{}.h5'.format(trial))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "cpdl",
   "language": "python",
   "name": "cpdl"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}